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,
7 ======================================================================
8 http://gcc.gnu.org/gcc-4.7/index.html
13 The [1]GNU project and the GCC developers are pleased to announce the
16 This release is a bug-fix release, containing fixes for regressions in
17 GCC 4.7.1 relative to previous releases of GCC.
22 September 20, 2012 ([2]changes)
25 June 14, 2012 ([3]changes)
28 March 22, 2012 ([4]changes)
30 References and Acknowledgements
32 GCC used to stand for the GNU C Compiler, but since the compiler
33 supports several other languages aside from C, it now stands for the
34 GNU Compiler Collection.
36 A list of [5]successful builds is updated as new information becomes
39 The GCC developers would like to thank the numerous people that have
40 contributed new features, improvements, bug fixes, and other changes as
41 well as test results to GCC. This [6]amazing group of volunteers is
42 what makes GCC successful.
44 For additional information about GCC please refer to the [7]GCC project
45 web site or contact the [8]GCC development mailing list.
47 To obtain GCC please use [9]our mirror sites or [10]our SVN server.
50 For questions related to the use of GCC, please consult these web
51 pages and the [11]GCC manuals. If that fails, the
52 [12]gcc-help@gcc.gnu.org mailing list might help. Comments on these
53 web pages and the development of GCC are welcome on our developer
54 list at [13]gcc@gcc.gnu.org. All of [14]our lists have public
57 Copyright (C) [15]Free Software Foundation, Inc. Verbatim copying and
58 distribution of this entire article is permitted in any medium,
59 provided this notice is preserved.
61 These pages are [16]maintained by the GCC team. Last modified
66 1. http://www.gnu.org/
67 2. http://gcc.gnu.org/gcc-4.7/changes.html
68 3. http://gcc.gnu.org/gcc-4.7/changes.html
69 4. http://gcc.gnu.org/gcc-4.7/changes.html
70 5. http://gcc.gnu.org/gcc-4.7/buildstat.html
71 6. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
72 7. http://gcc.gnu.org/index.html
73 8. mailto:gcc@gcc.gnu.org
74 9. http://gcc.gnu.org/mirrors.html
75 10. http://gcc.gnu.org/svn.html
76 11. http://gcc.gnu.org/onlinedocs/
77 12. mailto:gcc-help@gcc.gnu.org
78 13. mailto:gcc@gcc.gnu.org
79 14. http://gcc.gnu.org/lists.html
80 15. http://www.fsf.org/
81 16. http://gcc.gnu.org/about.html
82 17. http://validator.w3.org/check/referer
83 ======================================================================
84 http://gcc.gnu.org/gcc-4.7/changes.html
85 GCC 4.7 Release Series
86 Changes, New Features, and Fixes
90 * The -fconserve-space flag has been deprecated. The flag had no
91 effect for most targets: only targets without a global .bss section
92 and without support for switchable sections. Furthermore, the flag
93 only had an effect for G++, where it could result in wrong
94 semantics (please refer to the GCC manual for further details). The
95 flag will be removed in GCC 4.8
96 * Support for a number of older systems and recently unmaintained or
97 untested target ports of GCC has been declared obsolete in GCC 4.7.
98 Unless there is activity to revive them, the next release of GCC
99 will have their sources permanently removed.
100 All GCC ports for the following processor architectures have been
102 + picoChip (picochip-*)
103 The following ports for individual systems on particular
104 architectures have been obsoleted:
105 + IRIX 6.5 (mips-sgi-irix6.5)
106 + MIPS OpenBSD (mips*-*-openbsd*)
107 + Solaris 8 (*-*-solaris2.8). Details can be found in the
109 + Tru64 UNIX V5.1 (alpha*-dec-osf5.1*)
110 * On ARM, when compiling for ARMv6 (but not ARMv6-M), ARMv7-A,
111 ARMv7-R, or ARMv7-M, the new option -munaligned-access is active by
112 default, which for some source codes generates code that accesses
113 memory on unaligned addresses. This will require the kernel of
114 those systems to enable such accesses (controlled by CP15 register
115 c1, refer to ARM documentation). Alternatively or for compatibility
116 with kernels where unaligned accesses are not supported, all code
117 has to be compiled with -mno-unaligned-access. Linux/ARM in
118 official releases has automatically and unconditionally supported
119 unaligned accesses as emitted by GCC due to this option being
120 active, since Linux version 2.6.28.
121 * Support on ARM for the legacy floating-point accelerator (FPA) and
122 the mixed-endian floating-point format that it used has been
123 obsoleted. The ports that still use this format have been obsoleted
124 as well. Many legacy ARM ports already provide an alternative that
125 uses the VFP floating-point format. The obsolete ports will be
126 deleted in the next release.
127 The obsolete ports with alternatives are:
128 + arm*-*-rtems (use arm*-*-rtemseabi)
129 + arm*-*-linux-gnu (use arm*-*-linux-gnueabi)
130 + arm*-*-elf (use arm*-*-eabi)
131 + arm*-*-uclinux* (use arm*-*-uclinux*eabi)
132 Note, however, that these alternatives are not binary compatible
133 with their legacy counterparts (although some can support running
134 legacy applications).
135 The obsolete ports that currently lack a modern alternative are:
139 New ports that support more recent versions of the architecture are
141 * Support for the Maverick co-processor on ARM has been obsoleted.
142 Code to support it will be deleted in the next release.
143 * Support has been removed for Unix International threads on Solaris
144 2, so the --enable-threads=solaris configure option and the
145 -threads compiler option don't work any longer.
146 * Support has been removed for the Solaris BSD Compatibility Package,
147 which lives in /usr/ucbinclude and /usr/ucblib. It has been removed
148 from Solaris 11, and was only intended as a migration aid from
149 SunOS 4 to SunOS 5. The -compat-bsd compiler option is not
150 recognized any longer.
151 * The AVR port's libgcc has been improved and its multilib structure
152 has been enhanced. As a result, all objects contributing to an
153 application must either be compiled with GCC versions up to 4.6.x
154 or with GCC versions 4.7.0 or later.
155 * The ARM port's -mwords-little-endian option has been deprecated. It
156 will be removed in a future release.
157 * Support has been removed for the NetWare x86 configuration
158 obsoleted in GCC 4.6.
159 * It is no longer possible to use the "l" constraint in MIPS16 asm
161 * GCC versions 4.7.0 and 4.7.1 had changes to the C++ standard
162 library which affected the ABI in C++11 mode: a data member was
163 added to std::list changing its size and altering the definitions
164 of some member functions, and std::pair's move constructor was
165 non-trivial which altered the calling convention for functions with
166 std::pair arguments or return types. The ABI incompatibilities have
167 been fixed for GCC version 4.7.2 but as a result C++11 code
168 compiled with GCC 4.7.0 or 4.7.1 may be incompatible with C++11
169 code compiled with different GCC versions and with C++98/C++03 code
170 compiled with any version.
171 * On ARM, a bug has been fixed in GCC's implementation of the AAPCS
172 rules for the layout of vectors that could lead to wrong code being
173 generated. Vectors larger than 8 bytes in size are now by default
174 aligned to an 8-byte boundary. This is an ABI change: code that
175 makes explicit use of vector types may be incompatible with binary
176 objects built with older versions of GCC. Auto-vectorized code is
177 not affected by this change. (This change affects GCC versions
179 * More information on porting to GCC 4.7 from previous versions of
180 GCC can be found in the [2]porting guide for this release.
182 General Optimizer Improvements
184 * Support for a new parameter --param case-values-threshold=n was
185 added to allow users to control the cutoff between doing switch
186 statements as a series of if statements and using a jump table.
187 * Link-time optimization (LTO) improvements:
188 + Improved scalability and reduced memory usage. Link time
189 optimization of Firefox now requires 3GB of RAM on a 64-bit
190 system, while over 8GB was needed previously. Linking time has
191 been improved, too. The serial stage of linking Firefox has
192 been sped up by about a factor of 10.
193 + Reduced size of object files and temporary storage used during
195 + Streaming performance (both outbound and inbound) has been
197 + ld -r is now supported with LTO.
198 + Several bug fixes, especially in symbol table handling and
200 * Interprocedural optimization improvements:
201 + Heuristics now take into account that after inlining code will
202 be optimized out because of known values (or properties) of
203 function parameters. For example:
214 The call of foo will be inlined into bar even when optimizing
215 for code size. Constructs based on __builtin_constant_p are
216 now understood by the inliner and code size estimates are
217 evaluated a lot more realistically.
218 + The representation of C++ virtual thunks and aliases (both
219 implicit and defined via the alias attribute) has been
220 re-engineered. Aliases no longer pose optimization barriers
221 and calls to an alias can be inlined and otherwise optimized.
222 + The inter-procedural constant propagation pass has been
223 rewritten. It now performs generic function specialization.
224 For example when compiling the following:
230 ... do something else ...
242 GCC will now produce two copies of foo. One with flag being
243 true, while other with flag being false. This leads to
244 performance improvements previously possible only by inlining
245 all calls. Cloning causes a lot less code size growth.
246 * A string length optimization pass has been added. It attempts to
247 track string lengths and optimize various standard C string
248 functions like strlen, strchr, strcpy, strcat, stpcpy and their
249 _FORTIFY_SOURCE counterparts into faster alternatives. This pass is
250 enabled by default at -O2 or above, unless optimizing for size, and
251 can be disabled by the -fno-optimize-strlen option. The pass can
253 char *bar (const char *a)
255 size_t l = strlen (a) + 2;
256 char *p = malloc (l); if (p == NULL) return p;
257 strcpy (p, a); strcat (p, "/"); return p;
261 char *bar (const char *a)
263 size_t tmp = strlen (a);
264 char *p = malloc (tmp + 2); if (p == NULL) return p;
265 memcpy (p, a, tmp); memcpy (p + tmp, "/", 2); return p;
268 or for hosted compilations where stpcpy is available in the runtime
269 and headers provide its prototype, e.g.
270 void foo (char *a, const char *b, const char *c, const char *d)
272 strcpy (a, b); strcat (a, c); strcat (a, d);
275 can be optimized into:
276 void foo (char *a, const char *b, const char *c, const char *d)
278 strcpy (stpcpy (stpcpy (a, b), c), d);
281 New Languages and Language specific improvements
283 * Version 3.1 of the [3]OpenMP specification is now supported for the
284 C, C++, and Fortran compilers.
288 * The command-line option -feliminate-unused-debug-types has been
289 re-enabled by default, as it is for the other languages, leading to
290 a reduction in debug info size of 12.5% and more for relevant
291 cases, as well as to a small compilation speedup.
295 * A new built-in, __builtin_assume_aligned, has been added, through
296 which the compiler can be hinted about pointer alignment and can
297 use it to improve generated code.
298 * A new -Wunused-local-typedefs warning was added for C, C++,
299 Objective-C and Objective-C++. This warning diagnoses typedefs
300 locally defined in a function, and otherwise not used.
301 * A new experimental -ftrack-macro-expansion option was added for C,
302 C++, Objective-C, Objective-C++ and Fortran. It allows the compiler
303 to emit diagnostic about the current macro expansion stack when a
304 compilation error occurs in a macro expansion.
305 * Experimental support for transactional memory has been added. It
306 includes support in the compiler, as well as a supporting runtime
307 library called libitm. To compile code with transactional memory
308 constructs, use the -fgnu-tm option.
309 Support is currently available for Alpha, ARM, PowerPC, SH, SPARC,
310 and 32-bit/64-bit x86 platforms.
311 For more details on transactional memory see [4]the GCC WiKi.
312 * Support for atomic operations specifying the C++11/C11 memory model
313 has been added. These new __atomic routines replace the existing
314 __sync built-in routines.
315 Atomic support is also available for memory blocks. Lock-free
316 instructions will be used if a memory block is the same size and
317 alignment as a supported integer type. Atomic operations which do
318 not have lock-free support are left as function calls. A set of
319 library functions is available on the GCC atomic wiki in the
320 "External Atomics Library" section.
321 For more details on the memory models and features, see the
323 * When a binary operation is performed on vector types and one of the
324 operands is a uniform vector, it is possible to replace the vector
325 with the generating element. For example:
326 typedef int v4si __attribute__ ((vector_size (16)));
327 v4si res, a = {1,2,3,4};
330 res = 2 + a; /* means {2,2,2,2} + a */
331 res = a - x; /* means a - {x,x,x,x} */
335 * There is support for some more features from the C11 revision of
336 the ISO C standard. GCC now accepts the options -std=c11 and
337 -std=gnu11, in addition to the previous -std=c1x and -std=gnu1x.
338 + Unicode strings (previously supported only with options such
339 as -std=gnu11, now supported with -std=c11), and the
340 predefined macros __STDC_UTF_16__ and __STDC_UTF_32__.
341 + Nonreturning functions (_Noreturn and <stdnoreturn.h>).
342 + Alignment support (_Alignas, _Alignof, max_align_t,
344 + A built-in function __builtin_complex is provided to support C
345 library implementation of the CMPLX family of macros.
349 * G++ now accepts the -std=c++11, -std=gnu++11, and -Wc++11-compat
350 options, which are equivalent to -std=c++0x, -std=gnu++0x, and
351 -Wc++0x-compat, respectively.
352 * G++ now implements [6]C++11 extended friend syntax:
357 static const int I = 2;
367 * Thanks to Ville Voutilainen, G++ now implements [7]C++11 explicit
371 virtual void f() const final;
376 void f() const; // error: D::f attempts to override final B::f
377 void f(long) override; // error: doesn't override anything
378 void f(int) override; // ok
382 struct F: E { }; // error: deriving from final class
384 * G++ now implements [8]C++11 non-static data member initializers.
388 } a; // initializes a.i to 42
390 * Thanks to Ed Smith-Rowland, G++ now implements [9]C++11
391 user-defined literals.
393 // Not actually a good approximation. :)
394 constexpr long double operator"" _degrees (long double d) { return d * 0.0175; }
395 long double pi = 180.0_degrees;
397 * G++ now implements [10]C++11 alias-declarations.
399 template <class T> using Ptr = T*;
400 Ptr<int> ip; // decltype(ip) is int*
402 * Thanks to Ville Voutilainen and Pedro Lamarao, G++ now implements
403 [11]C++11 delegating constructors.
407 A(): A(42) { } // delegate to the A(int) constructor
410 * G++ now fully implements C++11 atomic classes rather than just
411 integer derived classes.
417 std::atomic<POD> my_atomic_POD;
419 * G++ now sets the predefined macro __cplusplus to the correct value,
420 199711L for C++98/03, and 201103L for C++11.
421 * G++ now correctly implements the two-phase lookup rules such that
422 an unqualified name used in a template must have an appropriate
423 declaration found either in scope at the point of definition of the
424 template or by argument-dependent lookup at the point of
425 instantiation. As a result, code that relies on a second
426 unqualified lookup at the point of instantiation to find functions
427 declared after the template or in dependent bases will be rejected.
428 The compiler will suggest ways to fix affected code, and using the
429 -fpermissive compiler flag will allow the code to compile with a
433 void f() { g(T()); } // error, g(int) not found by argument-dependent lookup
434 void g(int) { } // fix by moving this declaration before the declaration of f
438 // error, B::g(B) not found by argument-dependent lookup
439 void f() { g(T()); } // fix by using this->g or A::g
442 struct B { void g(B); };
450 * G++ now properly re-uses stack space allocated for temporary
451 objects when their lifetime ends, which can significantly lower
452 stack consumption for some C++ functions. As a result of this, some
453 code with undefined behavior will now break:
455 const int &f(const int &i) { return i; }
460 Here, x refers to the temporary allocated to hold the 1 argument,
461 which only lives until the end of the initialization; it
462 immediately becomes a dangling reference. So the next statement
463 re-uses the stack slot to hold the 2 argument, and users of x get
465 Note that this should not cause any change of behavior for
466 temporaries of types with non-trivial destructors, as they are
467 already destroyed at end of full-expression; the change is that now
468 the storage is released as well.
469 * A new command-line option -Wdelete-non-virtual-dtor has been added
470 to warn when delete is used to destroy an instance of a class which
471 has virtual functions and non-virtual destructor. It is unsafe to
472 delete an instance of a derived class through a pointer to a base
473 class if the base class does not have a virtual destructor. This
474 warning is enabled by -Wall.
475 * A new command-line option -Wzero-as-null-pointer-constant has been
476 added to warn when a literal '0' is used as null pointer constant.
477 It can be useful to facilitate the conversion to nullptr in C++11.
478 * As per C++98, access-declarations are now deprecated by G++.
479 Using-declarations are to be used instead. Furthermore, some
480 efforts have been made to improve the support of class scope
481 using-declarations. In particular, using-declarations referring to
482 a dependent type now work as expected ([12]bug c++/14258).
483 * The ELF symbol visibility of a template instantiation is now
484 properly constrained by the visibility of its template arguments
487 Runtime Library (libstdc++)
489 * [14]Improved experimental support for the new ISO C++ standard,
491 + using noexcept in most of the library;
492 + implementations of pointer_traits, allocator_traits and
493 scoped_allocator_adaptor;
494 + uses-allocator construction for tuple;
495 + vector meets the allocator-aware container requirements;
496 + replacing monotonic_clock with steady_clock;
497 + enabling the thread support library on most POSIX targets;
498 + many small improvements to conform to the FDIS.
499 * Added --enable-clocale=newlib configure option.
500 * Debug Mode iterators for unordered associative containers.
501 * Avoid polluting the global namespace and do not include <unistd.h>.
505 * The compile flag [15]-fstack-arrays has been added, which causes
506 all local arrays to be put on stack memory. For some programs this
507 will improve the performance significantly. If your program uses
508 very large local arrays, it is possible that you will have to
509 extend your runtime limits for stack memory.
510 * The [16]-Ofast flag now also implies [17]-fno-protect-parens and
512 * Front-end optimizations can now be selected by the
513 [19]-ffrontend-optimize option and deselected by the
514 -fno-frontend-optimize option.
515 * When front-end optimization removes a function call,
516 [20]-Wfunction-elimination warns about that.
517 * When performing front-end-optimization, the
518 [21]-faggressive-function-elimination option allows the removal of
519 duplicate function calls even for impure functions.
520 * The flag [22]-Wreal-q-constant has been added, which warns if
521 floating-point literals have been specified using q (such as
522 1.0q0); the q marker is now supported as a vendor extension to
523 denote quad precision (REAL(16) or, if not available, REAL(10)).
524 Consider using a kind parameter (such as in 1.0_qp) instead, which
525 can be obtained via [23]SELECTED_REAL_KIND.
526 * The GFORTRAN_USE_STDERR environment variable has been removed. GNU
527 Fortran now always prints error messages to standard error. If you
528 wish to redirect standard error, please consult the manual for your
529 OS, shell, batch environment etc. as appropriate.
530 * The -fdump-core option and GFORTRAN_ERROR_DUMPCORE environment
531 variable have been removed. When encountering a serious error,
532 gfortran will now always abort the program. Whether a core dump is
533 generated depends on the user environment settings; see the ulimit
534 -c setting for POSIX shells, limit coredumpsize for C shells, and
535 the [24]WER user-mode dumps settings on Windows.
536 * The [25]-fbacktrace option is now enabled by default. When
537 encountering a fatal error, gfortran will attempt to print a
538 backtrace to standard error before aborting. It can be disabled
539 with -fno-backtrace. Note: On POSIX targets with the addr2line
540 utility from GNU binutils, GNU Fortran can print a backtrace with
541 function name, file name, line number information in addition to
542 the addresses; otherwise only the addresses are printed.
544 + Generic interface names which have the same name as derived
545 types are now supported, which allows to write constructor
546 functions. Note that Fortran does not support static
547 constructor functions; only default initialization or an
548 explicit structure-constructor initialization are available.
549 + [27]Polymorphic (class) arrays are now supported.
551 + Support for the DO CONCURRENT construct has been added, which
552 allows the user to specify that individual loop iterations
553 have no interdependencies.
554 + [29]Coarrays: Full single-image support except for polymorphic
555 coarrays. Additionally, preliminary support for multiple
556 images via an MPI-based [30]coarray communication library has
557 been added. Note: The library version is not yet usable as
558 remote coarray access is not yet possible.
560 + New flag [32]-std=f2008ts permits programs that are expected
561 to conform to the Fortran 2008 standard and the draft
562 Technical Specification (TS) 29113 on Further Interoperability
564 + The OPTIONAL attribute is now allowed for dummy arguments of
566 + The RANK intrinsic has been added.
567 + The implementation of the ASYNCHRONOUS attribute in GCC is
568 compatible with the candidate draft of TS 29113 (since GCC
573 * GCC 4.7 implements the [33]Go 1 language standard. The library
574 support in 4.7.0 is not quite complete, due to release timing.
575 Release 4.7.1 includes complete support for Go 1. The Go library is
576 from the Go 1.0.1 release.
577 * Go has been tested on GNU/Linux and Solaris platforms. It may work
578 on other platforms as well.
580 New Targets and Target Specific Improvements
584 * GCC now supports the Cortex-A7 processor implementing the v7-a
585 version of the architecture using the option -mcpu=cortex-a7.
586 * The default vector size in auto-vectorization for NEON is now 128
587 bits. If vectorization fails thusly, the vectorizer tries again
589 * A new option -mvectorize-with-neon-double was added to allow users
590 to change the vector size to 64 bits.
594 * GCC now supports the XMEGA architecture. This requires GNU binutils
596 * Support for the [34]named address spaces __flash, __flash1, ...,
597 __flash5 and __memx has been added. These address spaces locate
598 read-only data in flash memory and allow reading from flash memory
599 by means of ordinary C code, i.e. without the need of (inline)
602 const __flash int values[] = { 42, 31 };
604 int add_values (const __flash int *p, int i)
606 return values[i] + *p;
609 * Support has been added for a new AVR-specific configure option
610 --with-avrlibc=yes in order to arrange for better integration of
611 [35]AVR-Libc. This configure option is supported in avr-gcc 4.7.2
612 and newer and will only take effect in non-RTEMS configurations. If
613 avr-gcc is configured for RTEMS, the option will be ignored which
614 is the same as specifying --with-avrlibc=no. See [36]PR54461 for
615 more technical details.
616 * Support for AVR-specific [37]built-in functions has been added.
617 * Support has been added for the signed and unsigned 24-bit scalar
618 integer types __int24 and __uint24.
619 * New command-line options -maccumulate-args, -mbranch-cost=cost and
620 -mstrict-X were added to allow better fine-tuning of code
622 * The command option -fdata-sections now also takes affect on the
623 section names of variables with the progmem attribute.
624 * A new inline assembler print modifier %i to print a RAM address as
625 I/O address has been added:
627 #include <avr/io.h> /* Port Definitions from AVR-LibC */
629 void set_portb (uint8_t value)
631 asm volatile ("out %0, %i1" :: "r" (value), "n" (&PORTB) : "memory");
634 The offset between an I/O address and the RAM address for that I/O
635 location is device-specific. This offset is taken into account when
636 printing a RAM address with the %i modifier so that the address is
637 suitable to be used as operand in an I/O command. The address must
638 be a constant integer known at compile time.
639 * The inline assembler constraint "R" to represent integers in the
640 range -6 ... 5 has been removed without replacement.
641 * Many optimizations to:
642 + 64-bit integer arithmetic
643 + Widening multiplication
644 + Integer division by a constant
645 + Avoid constant reloading in multi-byte instructions.
646 + Micro-optimizations for special instruction sequences.
647 + Generic built-in functions like __builtin_ffs*,
649 + If-else decision trees generated by switch instructions
650 + Merging of data located in flash memory
651 + New libgcc variants for devices with 8-bit wide stack pointer
653 * Better documentation:
654 + Handling of EIND and indirect jumps on devices with more than
655 128 KiB of program memory.
656 + Handling of the RAMPD, RAMPX, RAMPY and RAMPZ special function
658 + Function attributes OS_main and OS_task.
659 + AVR-specific built-in macros.
663 * Support has been added for the Texas Instruments C6X family of
668 * Support has been added for National Semiconductor's CR16
673 * Support has been added for Adapteva's Epiphany architecture.
677 * Support for Intel AVX2 intrinsics, built-in functions and code
678 generation is available via -mavx2.
679 * Support for Intel BMI2 intrinsics, built-in functions and code
680 generation is available via -mbmi2.
681 * Implementation and automatic generation of __builtin_clz* using the
682 lzcnt instruction is available via -mlzcnt.
683 * Support for Intel FMA3 intrinsics and code generation is available
685 * A new -mfsgsbase command-line option is available that makes GCC
686 generate new segment register read/write instructions through
688 * Support for the new Intel rdrnd instruction is available via
690 * Two additional AVX vector conversion instructions are available via
692 * Support for new Intel processor codename IvyBridge with RDRND,
693 FSGSBASE and F16C is available through -march=core-avx-i.
694 * Support for the new Intel processor codename Haswell with AVX2,
695 FMA, BMI, BMI2, LZCNT is available through -march=core-avx2.
696 * Support for new AMD family 15h processors (Piledriver core) is now
697 available through -march=bdver2 and -mtune=bdver2 options.
698 * Support for [38]the x32 psABI is now available through the -mx32
700 * Windows mingw targets are using the -mms-bitfields option by
702 * Windows x86 targets are using the __thiscall calling convention for
703 C++ class-member functions.
704 * Support for the configure option --with-threads=posix for Windows
709 * GCC now supports thread-local storage (TLS) for MIPS16. This
710 requires GNU binutils 2.22 or later.
711 * GCC can now generate code specifically for the Cavium Octeon+ and
712 Octeon2 processors. The associated command-line options are
713 -march=octeon+ and -march=octeon2 respectively. Both options
714 require GNU binutils 2.22 or later.
715 * GCC can now work around certain 24k errata, under the control of
716 the command-line option -mfix-24k. These workarounds require GNU
717 binutils 2.20 or later.
718 * 32-bit MIPS GNU/Linux targets such as mips-linux-gnu can now build
719 n32 and n64 multilibs. The result is effectively a 64-bit GNU/Linux
720 toolchain that generates 32-bit code by default. Use the
721 configure-time option --enable-targets=all to select these extra
723 * Passing -fno-delayed-branch now also stops the assembler from
724 automatically filling delay slots.
728 * Vectors of type vector long long or vector long are passed and
729 returned using the same method as other vectors with the VSX
730 instruction set. Previously GCC did not adhere to the ABI for
731 128-bit vectors with 64-bit integer base types (PR 48857). This
732 will also be fixed in the GCC 4.6.1 and 4.5.4 releases.
733 * A new option -mno-pointers-to-nested-functions was added to allow
734 AIX 32-bit/64-bit and GNU/Linux 64-bit PowerPC users to specify
735 that the compiler should not load up the chain register (r11)
736 before calling a function through a pointer. If you use this
737 option, you cannot call nested functions through a pointer, or call
738 other languages that might use the static chain.
739 * A new option msave-toc-indirect was added to allow AIX
740 32-bit/64-bit and GNU/Linux 64-bit PowerPC users control whether we
741 save the TOC in the prologue for indirect calls or generate the
742 save inline. This can speed up some programs that call through a
743 function pointer a lot, but it can slow down other functions that
744 only call through a function pointer in exceptional cases.
745 * The PowerPC port will now enable machine-specific built-in
746 functions when the user switches the target machine using the
747 #pragma GCC target or __attribute__ ((__target__ ("target"))) code
748 sequences. In addition, the target macros are updated. However, due
749 to the way the -save-temps switch is implemented, you won't see the
750 effect of these additional macros being defined in preprocessor
755 * A new option -msoft-atomic has been added. When it is specified,
756 GCC will generate GNU/Linux-compatible gUSA atomic sequences for
757 the new __atomic routines.
758 * Since it is neither supported by GAS nor officially documented,
759 code generation for little endian SH2A has been disabled.
760 Specifying -ml with -m2a* will now result in a compiler error.
761 * The defunct -mbranch-cost option has been fixed.
762 * Some improvements to the generated code of:
763 + Utilization of the tst #imm,R0 instruction.
764 + Dynamic shift instructions on SH2A.
765 + Integer absolute value calculations.
769 * The option -mflat has been reinstated. When it is specified, the
770 compiler will generate code for a single register window model.
771 This is essentially a new implementation and the corresponding
772 debugger support has been added to GDB 7.4.
773 * Support for the options -mtune=native and -mcpu=native has been
774 added on selected native platforms (GNU/Linux and Solaris).
775 * Support for the SPARC T3 (Niagara 3) processor has been added.
777 + An intrinsics header visintrin.h has been added.
778 + Builtin intrinsics for the VIS 1.0 edge handling and pixel
779 compare instructions have been added.
780 + The little-endian version of alignaddr is now supported.
781 + When possible, VIS builtins are marked const, which should
782 increase the compiler's ability to optimize VIS operations.
783 + The compiler now properly tracks the %gsr register and how it
784 behaves as an input for various VIS instructions.
785 + Akin to fzero, the compiler can now generate fone instructions
786 in order to set all of the bits of a floating-point register
788 + The documentation for the VIS intrinsics in the GCC manual has
789 been brought up to date and many inaccuracies were fixed.
790 + Intrinsics for the VIS 2.0 bmask, bshuffle, and
791 non-condition-code setting edge instructions have been added.
792 Their availability is controlled by the new -mvis2 and
793 -mno-vis2 options. They are enabled by default on
794 UltraSPARC-III and later CPUs.
795 * Support for UltraSPARC Fused Multiply-Add floating-point extensions
796 has been added. These instructions are enabled by default on SPARC
797 T3 (Niagara 3) and later CPUs.
801 * Support has been added for the Tilera TILE-Gx and TILEPro families
804 Other significant improvements
806 * A new option (-grecord-gcc-switches) was added that appends
807 compiler command-line options that might affect code generation to
808 the DW_AT_producer attribute string in the DWARF debugging
810 * GCC now supports various new GNU extensions to the DWARF debugging
811 information format, like [39]entry value and [40]call site
812 information, [41]typed DWARF stack or [42]a more compact macro
813 representation. Support for these extensions has been added to GDB
814 7.4. They can be disabled through the -gstrict-dwarf command-line
819 This is the [43]list of problem reports (PRs) from GCC's bug tracking
820 system that are known to be fixed in the 4.7.1 release. This list might
821 not be complete (that is, it is possible that some PRs that have been
822 fixed are not listed here).
824 The Go frontend in the 4.7.1 release fully supports the [44]Go 1
829 This is the [45]list of problem reports (PRs) from GCC's bug tracking
830 system that are known to be fixed in the 4.7.2 release. This list might
831 not be complete (that is, it is possible that some PRs that have been
832 fixed are not listed here).
835 For questions related to the use of GCC, please consult these web
836 pages and the [46]GCC manuals. If that fails, the
837 [47]gcc-help@gcc.gnu.org mailing list might help. Comments on these
838 web pages and the development of GCC are welcome on our developer
839 list at [48]gcc@gcc.gnu.org. All of [49]our lists have public
842 Copyright (C) [50]Free Software Foundation, Inc. Verbatim copying and
843 distribution of this entire article is permitted in any medium,
844 provided this notice is preserved.
846 These pages are [51]maintained by the GCC team. Last modified
851 1. http://gcc.gnu.org/ml/gcc-patches/2011-03/msg01263.html
852 2. http://gcc.gnu.org/gcc-4.7/porting_to.html
853 3. http://openmp.org/wp/openmp-specifications/
854 4. http://gcc.gnu.org/wiki/TransactionalMemory
855 5. http://gcc.gnu.org/wiki/Atomic/GCCMM
856 6. http://gcc.gnu.org/gcc-4.7/cxx0x_status.html
857 7. http://gcc.gnu.org/gcc-4.7/cxx0x_status.html
858 8. http://gcc.gnu.org/gcc-4.7/cxx0x_status.html
859 9. http://gcc.gnu.org/gcc-4.7/cxx0x_status.html
860 10. http://gcc.gnu.org/gcc-4.7/cxx0x_status.html
861 11. http://gcc.gnu.org/gcc-4.7/cxx0x_status.html
862 12. http://gcc.gnu.org/PR14258
863 13. http://gcc.gnu.org/PR35688
864 14. http://gcc.gnu.org/onlinedocs/gcc-4.7.1/libstdc++/manual/manual/status.html#status.iso.2011
865 15. http://gcc.gnu.org/onlinedocs/gcc-4.7.1/gfortran/Code-Gen-Options.html#index-g_t_0040code_007bfstack-arrays_007d-254
866 16. http://gcc.gnu.org/onlinedocs/gcc-4.7.1/gcc/Optimize-Options.html#index-Ofast-689
867 17. http://gcc.gnu.org/onlinedocs/gcc-4.7.1/gfortran/Code-Gen-Options.html#index-g_t_0040code_007bfno-protect-parens_007d-270
868 18. http://gcc.gnu.org/onlinedocs/gcc-4.7.1/gfortran/Code-Gen-Options.html#index-g_t_0040code_007bfstack-arrays_007d-254
869 19. http://gcc.gnu.org/onlinedocs/gcc-4.7.1/gfortran/Code-Gen-Options.html#index-g_t_0040code_007bfrontend-optimize_007d-275
870 20. http://gcc.gnu.org/onlinedocs/gcc-4.7.1/gfortran/Error-and-Warning-Options.html#index-g_t_0040code_007bWfunction-elimination_007d-170
871 21. http://gcc.gnu.org/onlinedocs/gcc-4.7.1/gfortran/Code-Gen-Options.html#index-g_t_0040code_007bfaggressive-function-elimination_007d-270
872 22. http://gcc.gnu.org/onlinedocs/gcc-4.7.1/gfortran/Error-and-Warning-Options.html#index-g_t_0040code_007bWreal-q-constant_007d-149
873 23. http://gcc.gnu.org/onlinedocs/gcc-4.7.1/gfortran/SELECTED_005fREAL_005fKIND.html
874 24. http://msdn.microsoft.com/en-us/library/bb787181%28v=vs.85%29.aspx
875 25. http://gcc.gnu.org/onlinedocs/gcc-4.7.1/gfortran/Debugging-Options.html#index-g_t_0040code_007bfno-backtrace_007d-183
876 26. http://gcc.gnu.org/wiki/Fortran2003Status
877 27. http://gcc.gnu.org/wiki/OOP
878 28. http://gcc.gnu.org/wiki/Fortran2008Status
879 29. http://gcc.gnu.org/wiki/Coarray
880 30. http://gcc.gnu.org/wiki/CoarrayLib
881 31. http://gcc.gnu.org/wiki/TS29113Status
882 32. http://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
883 33. http://weekly.golang.org/doc/go1.html
884 34. http://gcc.gnu.org/onlinedocs/gcc-4.7.1/gcc/Named-Address-Spaces.html
885 35. http://nongnu.org/avr-libc/
886 36. http://gcc.gnu.org/PR54461
887 37. http://gcc.gnu.org/onlinedocs/gcc-4.7.1/gcc/AVR-Built%5f002din-Functions.html
888 38. http://sites.google.com/site/x32abi/
889 39. http://www.dwarfstd.org/ShowIssue.php?issue=100909.1
890 40. http://www.dwarfstd.org/ShowIssue.php?issue=100909.2
891 41. http://www.dwarfstd.org/doc/040408.1.html
892 42. http://www.dwarfstd.org/ShowIssue.php?issue=110722.1
893 43. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.7.1
894 44. http://weekly.golang.org/doc/go1.html
895 45. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.7.2
896 46. http://gcc.gnu.org/onlinedocs/
897 47. mailto:gcc-help@gcc.gnu.org
898 48. mailto:gcc@gcc.gnu.org
899 49. http://gcc.gnu.org/lists.html
900 50. http://www.fsf.org/
901 51. http://gcc.gnu.org/about.html
902 52. http://validator.w3.org/check/referer
903 ======================================================================
904 http://gcc.gnu.org/gcc-4.6/index.html
905 GCC 4.6 Release Series
909 The [1]GNU project and the GCC developers are pleased to announce the
910 release of GCC 4.6.3.
912 This release is a bug-fix release, containing fixes for regressions in
913 GCC 4.6.2 relative to previous releases of GCC.
918 March 1, 2012 ([2]changes)
921 October 26, 2011 ([3]changes)
924 June 27, 2011 ([4]changes)
927 March 25, 2011 ([5]changes)
929 References and Acknowledgements
931 GCC used to stand for the GNU C Compiler, but since the compiler
932 supports several other languages aside from C, it now stands for the
933 GNU Compiler Collection.
935 A list of [6]successful builds is updated as new information becomes
938 The GCC developers would like to thank the numerous people that have
939 contributed new features, improvements, bug fixes, and other changes as
940 well as test results to GCC. This [7]amazing group of volunteers is
941 what makes GCC successful.
943 For additional information about GCC please refer to the [8]GCC project
944 web site or contact the [9]GCC development mailing list.
946 To obtain GCC please use [10]our mirror sites or [11]our SVN server.
949 For questions related to the use of GCC, please consult these web
950 pages and the [12]GCC manuals. If that fails, the
951 [13]gcc-help@gcc.gnu.org mailing list might help. Comments on these
952 web pages and the development of GCC are welcome on our developer
953 list at [14]gcc@gcc.gnu.org. All of [15]our lists have public
956 Copyright (C) [16]Free Software Foundation, Inc. Verbatim copying and
957 distribution of this entire article is permitted in any medium,
958 provided this notice is preserved.
960 These pages are [17]maintained by the GCC team. Last modified
965 1. http://www.gnu.org/
966 2. http://gcc.gnu.org/gcc-4.6/changes.html
967 3. http://gcc.gnu.org/gcc-4.6/changes.html
968 4. http://gcc.gnu.org/gcc-4.6/changes.html
969 5. http://gcc.gnu.org/gcc-4.6/changes.html
970 6. http://gcc.gnu.org/gcc-4.6/buildstat.html
971 7. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
972 8. http://gcc.gnu.org/index.html
973 9. mailto:gcc@gcc.gnu.org
974 10. http://gcc.gnu.org/mirrors.html
975 11. http://gcc.gnu.org/svn.html
976 12. http://gcc.gnu.org/onlinedocs/
977 13. mailto:gcc-help@gcc.gnu.org
978 14. mailto:gcc@gcc.gnu.org
979 15. http://gcc.gnu.org/lists.html
980 16. http://www.fsf.org/
981 17. http://gcc.gnu.org/about.html
982 18. http://validator.w3.org/check/referer
983 ======================================================================
984 http://gcc.gnu.org/gcc-4.6/changes.html
985 GCC 4.6 Release Series
986 Changes, New Features, and Fixes
990 * The options -b <machine> and -V <version> have been removed because
991 they were unreliable. Instead, users should directly run
992 <machine>-gcc when cross-compiling, or <machine>-gcc-<version> to
993 run a different version of gcc.
994 * GCC now has stricter checks for invalid command-line options. In
995 particular, when gcc was called to link object files rather than
996 compile source code, it would previously accept and ignore all
997 options starting with --, including linker options such as
998 --as-needed and --export-dynamic, although such options would
999 result in errors if any source code was compiled. Such options, if
1000 unknown to the compiler, are now rejected in all cases; if the
1001 intent was to pass them to the linker, options such as
1002 -Wl,--as-needed should be used.
1003 * Versions of the GNU C library up to and including 2.11.1 included
1004 an [1]incorrect implementation of the cproj function. GCC optimizes
1005 its builtin cproj according to the behavior specified and allowed
1006 by the ISO C99 standard. If you want to avoid discrepancies between
1007 the C library and GCC's builtin transformations when using cproj in
1008 your code, use GLIBC 2.12 or later. If you are using an older GLIBC
1009 and actually rely on the incorrect behavior of cproj, then you can
1010 disable GCC's transformations using -fno-builtin-cproj.
1011 * The C-only intermodule optimization framework (IMA, enabled by
1012 -combine) has been removed in favor of the new generic link-time
1013 optimization framework (LTO) introduced in [2]GCC 4.5.0.
1014 * GCC now ships with the LGPL-licensed libquadmath library, which
1015 provides quad-precision mathematical functions for targets with a
1016 __float128 datatype. __float128 is available for targets on 32-bit
1017 x86, x86-64 and Itanium architectures. The libquadmath library is
1018 automatically built on such targets when building the Fortran
1020 * New -Wunused-but-set-variable and -Wunused-but-set-parameter
1021 warnings were added for C, C++, Objective-C and Objective-C++.
1022 These warnings diagnose variables respective parameters which are
1023 only set in the code and never otherwise used. Usually such
1024 variables are useless and often even the value assigned to them is
1025 computed needlessly, sometimes expensively. The
1026 -Wunused-but-set-variable warning is enabled by default by -Wall
1027 flag and -Wunused-but-set-parameter by -Wall -Wextra flags.
1028 * On ARM, a bug has been fixed in GCC's implementation of the AAPCS
1029 rules for the layout of vectors that could lead to wrong code being
1030 generated. Vectors larger than 8 bytes in size are now by default
1031 aligned to an 8-byte boundary. This is an ABI change: code that
1032 makes explicit use of vector types may be incompatible with binary
1033 objects built with older versions of GCC. Auto-vectorized code is
1034 not affected by this change. (This change affects GCC versions
1035 4.6.4 and later, with the exception of versions 4.7.0 and 4.7.1.)
1036 * On AVR, variables with the progmem attribute to locate data in
1037 flash memory must be qualified as const.
1038 * Support for a number of older systems and recently unmaintained or
1039 untested target ports of GCC has been declared obsolete in GCC 4.6.
1040 Unless there is activity to revive them, the next release of GCC
1041 will have their sources permanently removed.
1042 All GCC ports for the following processor architectures have been
1044 + Argonaut ARC (arc-*)
1045 + National Semiconductor CRX (crx-*)
1046 + Motorola 68HC11 and 68HC12 (m68hc11-*-*, m6811-*-*,
1047 m68hc12-*-*, m6812-*-*)
1048 + Sunplus S+core (score-*)
1049 The following ports for individual systems on particular
1050 architectures have been obsoleted:
1051 + Interix (i[34567]86-*-interix3*)
1052 + NetWare x86 (i[3456x]86-*-netware*)
1053 + Generic ARM PE (arm-*-pe* other than arm*-wince-pe*)
1054 + MCore PE (mcore-*-pe*)
1055 + SH SymbianOS (sh*-*-symbianelf*)
1056 + GNU Hurd on Alpha and PowerPC (alpha*-*-gnu*, powerpc*-*-gnu*)
1057 + M68K uClinux old ABI (m68k-*-uclinuxoldabi*)
1058 + a.out NetBSD (arm*-*-netbsd*, i[34567]86-*-netbsd*,
1059 vax-*-netbsd*, but not *-*-netbsdelf*)
1060 The i[34567]86-*-pe alias for Cygwin targets has also been
1061 obsoleted; users should configure for i[34567]86-*-cygwin* instead.
1062 Certain configure options to control the set of libraries built
1063 with GCC on some targets have been obsoleted. On ARM targets, the
1064 options --disable-fpu, --disable-26bit, --disable-underscore,
1065 --disable-interwork, --disable-biendian and --disable-nofmult have
1066 been obsoleted. On MIPS targets, the options
1067 --disable-single-float, --disable-biendian and --disable-softfloat
1068 have been obsoleted.
1069 * Support has been removed for all the [3]configurations obsoleted in
1071 * More information on porting to GCC 4.6 from previous versions of
1072 GCC can be found in the [4]porting guide for this release.
1074 General Optimizer Improvements
1076 * A new general optimization level, -Ofast, has been introduced. It
1077 combines the existing optimization level -O3 with options that can
1078 affect standards compliance but result in better optimized code.
1079 For example, -Ofast enables -ffast-math.
1080 * Link-time optimization improvements:
1081 + The [5]Scalable Whole Program Optimizer (WHOPR) project has
1082 stabilized to the point of being usable. It has become the
1083 default mode when using the LTO optimization model. Link time
1084 optimization can now split itself into multiple parallel
1085 compilations. Parallelism is controlled with -flto=n (where n
1086 specifies the number of compilations to execute in parallel).
1087 GCC can also cooperate with a GNU make job server by
1088 specifying the -flto=jobserver option and adding + to the
1089 beginning of the Makefile rule executing the linker.
1090 Classical LTO mode can be enforced by -flto-partition=none.
1091 This may result in small code quality improvements.
1092 + A large number of bugs were fixed. GCC itself, Mozilla Firefox
1093 and other large applications can be built with LTO enabled.
1094 + The linker plugin support improvements
1095 o Linker plugin is now enabled by default when the linker
1096 is detected to have plugin support. This is the case for
1097 GNU ld 2.21.51 or newer (on ELF and Cygwin targets) and
1098 the Gold linker on ELF targets. Plugin support of the
1099 Apple linker on Darwin is not compatible with GCC. The
1100 linker plugin can also be controlled by the
1101 -fuse-linker-plugin command line option.
1102 o Resolution information from the linker plugin is used to
1103 drive whole program assumptions. Use of the linker plugin
1104 results in more aggressive optimization on binaries and
1105 on shared libraries that use the hidden visibility
1106 attribute. Consequently the use of -fwhole-program is not
1107 necessary in addition to LTO.
1108 + Hidden symbols used from non-LTO objects now have to be
1109 explicitly annotated with externally_visible when the linker
1111 + C++ inline functions and virtual tables are now privatized
1112 more aggressively, leading to better inter-procedural
1113 optimization and faster dynamic linking.
1114 + Memory usage and intermediate language streaming performance
1116 + Static constructors and destructors from individual units are
1117 inlined into a single function. This can significantly improve
1118 startup times of large C++ applications where static
1119 constructors are very common. For example, static constructors
1120 are used when including the iostream header.
1121 + Support for the Ada language has been added.
1122 * Interprocedural optimization improvements
1123 + The interprocedural framework was re-tuned for link time
1124 optimization. Several scalability issues were resolved.
1125 + Improved auto-detection of const and pure functions. Newly,
1126 noreturn functions are auto-detected.
1127 The [6]-Wsuggest-attribute=[const|pure|noreturn] flag is
1128 available that informs users when adding attributes to headers
1129 might improve code generation.
1130 + A number of inlining heuristic improvements. In particular:
1131 o Partial inlining is now supported and enabled by default
1132 at -O2 and greater. The feature can be controlled via
1134 Partial inlining splits functions with short hot path to
1135 return. This allows more aggressive inlining of the hot
1136 path leading to better performance and often to code size
1137 reductions (because cold parts of functions are not
1139 o Scalability for large compilation units was improved
1141 o Inlining of callbacks is now more aggressive.
1142 o Virtual methods are considered for inlining when the
1143 caller is inlined and devirtualization is then possible.
1144 o Inlining when optimizing for size (either in cold regions
1145 of a program or when compiling with -Os) was improved to
1146 better handle C++ programs with larger abstraction
1147 penalty, leading to smaller and faster code.
1148 + The IPA reference optimization pass detecting global variables
1149 used or modified by functions was strengthened and sped up.
1150 + Functions whose address was taken are now optimized out when
1151 all references to them are dead.
1152 + A new inter-procedural static profile estimation pass detects
1153 functions that are executed once or unlikely to be executed.
1154 Unlikely executed functions are optimized for size. Functions
1155 executed once are optimized for size except for the inner
1157 + On most targets with named section support, functions used
1158 only at startup (static constructors and main), functions used
1159 only at exit and functions detected to be cold are placed into
1160 separate text segment subsections. This extends the
1161 -freorder-functions feature and is controlled by the same
1162 switch. The goal is to improve the startup time of large C++
1164 Proper function placement requires linker support. GNU ld
1165 2.21.51 on ELF targets was updated to place those functions
1166 together within the text section leading to better code
1167 locality and faster startup times of large C++ programs. The
1168 feature is also supported in the Apple linker. Support in the
1169 gold linker is planned.
1170 * A new switch -fstack-usage has been added. It makes the compiler
1171 output stack usage information for the program, on a per-function
1172 basis, in an auxiliary file.
1173 * A new switch -fcombine-stack-adjustments has been added. It can be
1174 used to enable or disable the compiler's stack-slot combining pass
1175 which before was enabled automatically at -O1 and above, but could
1176 not be controlled on its own.
1177 * A new switch -fstrict-volatile-bitfields has been added. Using it
1178 indicates that accesses to volatile bitfields should use a single
1179 access of the width of the field's type. This option can be useful
1180 for precisely defining and accessing memory-mapped peripheral
1181 registers from C or C++.
1183 Compile time and memory usage improvements
1185 * Datastructures used by the dataflow framework in GCC were
1186 reorganized for better memory usage and more cache locality.
1187 Compile time is improved especially on units with large functions
1188 (possibly resulting from a lot of inlining) not fitting into the
1189 processor cache. The compile time of the GCC C compiler binary with
1190 link-time optimization went down by over 10% (benchmarked on x86-64
1193 New Languages and Language specific improvements
1197 * Stack checking has been improved on selected architectures (Alpha,
1198 IA-32/x86-64, RS/6000 and SPARC): it now will detect stack
1199 overflows in all cases on these architectures.
1200 * Initial support for Ada 2012 has been added.
1204 * A new warning, enabled by -Wdouble-promotion, has been added that
1205 warns about cases where a value of type float is implicitly
1206 promoted to double. This is especially helpful for CPUs that handle
1207 the former in hardware, but emulate the latter in software.
1208 * A new function attribute leaf was introduced. This attribute allows
1209 better inter-procedural optimization across calls to functions that
1210 return to the current unit only via returning or exception
1211 handling. This is the case for most library functions that have no
1213 * Support for a new data type __int128 for targets having wide enough
1214 machine-mode support.
1215 * The new function attribute callee_pop_aggregate allows to specify
1216 if the caller or callee is responsible for popping the aggregate
1217 return pointer value from the stack.
1218 * Support for selectively enabling and disabling warnings via #pragma
1219 GCC diagnostic has been added. For instance:
1220 #pragma GCC diagnostic error "-Wuninitialized"
1221 foo(a); /* error is given for this one */
1222 #pragma GCC diagnostic push
1223 #pragma GCC diagnostic ignored "-Wuninitialized"
1224 foo(b); /* no diagnostic for this one */
1225 #pragma GCC diagnostic pop
1226 foo(c); /* error is given for this one */
1227 #pragma GCC diagnostic pop
1228 foo(d); /* depends on command line options */
1230 * The -fmax-errors=N option is now supported. Using this option
1231 causes the compiler to exit after N errors have been issued.
1235 * There is now experimental support for some features from the
1236 upcoming C1X revision of the ISO C standard. This support may be
1237 selected with -std=c1x, or -std=gnu1x for C1X with GNU extensions.
1238 Note that this support is experimental and may change incompatibly
1239 in future releases for consistency with changes to the C1X standard
1240 draft. The following features are newly supported as described in
1241 the N1539 draft of C1X (with changes agreed at the March 2011 WG14
1242 meeting); some other features were already supported with no
1243 compiler changes being needed, or have some support but not in full
1244 accord with N1539 (as amended).
1245 + Static assertions (_Static_assert keyword)
1246 + Typedef redefinition
1247 + New macros in <float.h>
1248 + Anonymous structures and unions
1249 * The new -fplan9-extensions option directs the compiler to support
1250 some extensions for anonymous struct fields which are implemented
1251 by the Plan 9 compiler. A pointer to a struct may be automatically
1252 converted to a pointer to an anonymous field when calling a
1253 function, in order to make the types match. An anonymous struct
1254 field whose type is a typedef name may be referred to using the
1259 * Improved [7]experimental support for the upcoming C++0x ISO C++
1260 standard, including support for constexpr (thanks to Gabriel Dos
1261 Reis and Jason Merrill), nullptr (thanks to Magnus Fromreide),
1262 noexcept, unrestricted unions, range-based for loops (thanks to
1263 Rodrigo Rivas Costa), opaque enum declarations (thanks also to
1264 Rodrigo), implicitly deleted functions and implicit move
1266 * When an extern declaration within a function does not match a
1267 declaration in the enclosing context, G++ now properly declares the
1268 name within the namespace of the function rather than the namespace
1269 which was open just before the function definition ([8]c++/43145).
1270 * GCC now warns by default when casting integers to larger pointer
1271 types. These warnings can be disabled with the option
1272 -Wno-int-to-pointer-cast, which is now also available in C++.
1273 * G++ no longer optimizes using the assumption that a value of
1274 enumeration type will fall within the range specified by the
1275 standard, since that assumption is easily violated with a
1276 conversion from integer type ([9]c++/43680). The old behavior can
1277 be restored with -fstrict-enums.
1278 * The new -fnothrow-opt flag changes the semantics of a throw()
1279 exception specification to match the proposed semantics of the
1280 noexcept specification: just call terminate if an exception tries
1281 to propagate out of a function with such an exception
1282 specification. This dramatically reduces or eliminates the code
1283 size overhead from adding the exception specification.
1284 * The new -Wnoexcept flag will suggest adding a noexcept qualifier to
1285 a function that the compiler can tell doesn't throw if it would
1286 change the value of a noexcept expression.
1287 * The -Wshadow option now warns if a local variable or type
1288 declaration shadows another type in C++. Note that the compiler
1289 will not warn if a local variable shadows a struct/class/enum, but
1290 will warn if it shadows an explicit typedef.
1291 * When an identifier is not found in the current scope, G++ now
1292 offers suggestions about which identifier might have been intended.
1293 * G++ now issues clearer diagnostics for missing semicolons after
1294 class, struct, and union definitions.
1295 * G++ now issues clearer diagnostics for missing semicolons after
1296 class member declarations.
1297 * G++ now issues clearer diagnostics when a colon is used in a place
1298 where a double-colon was intended.
1299 * G++ no longer accepts mutable on reference members ([10]c++/33558).
1300 Use -fpermissive to allow the old, non-conforming behaviour.
1301 * A few mangling fixes have been made, to attribute const/volatile on
1302 function pointer types, decltype of a plain decl, and use of a
1303 function parameter in the declaration of another parameter. By
1304 default the compiler still uses the old mangling, but emits aliases
1305 with the new mangling on targets that support strong aliases. Users
1306 can switch over entirely to the new mangling with -fabi-version=5
1307 or -fabi-version=0. -Wabi will now warn about code that uses the
1309 * In 4.6.0 and 4.6.1 G++ no longer allows objects of const-qualified
1310 type to be default initialized unless the type has a user-declared
1311 default constructor. In 4.6.2 G++ implements the proposed
1312 resolution of [11]DR 253, so default initialization is allowed if
1313 it initializes all subobjects. Code that fails to compile can be
1314 fixed by providing an initializer e.g.
1316 struct B : A { int i; };
1318 Use -fpermissive to allow the old, non-conforming behaviour.
1320 Runtime Library (libstdc++)
1322 * [12]Improved experimental support for the upcoming ISO C++
1323 standard, C++0x, including using constexpr and nullptr.
1324 * Performance improvements to the [13]Debug Mode, thanks to Franc,ois
1326 * Atomic operations used for reference-counting are annotated so that
1327 they can be understood by race detectors such as Helgrind, see
1328 [14]Data Race Hunting.
1329 * Most libstdc++ standard headers have been changed to no longer
1330 include the cstddef header as an implementation detail. Code that
1331 relied on that header being included as side-effect of including
1332 other standard headers will need to include cstddef explicitly.
1336 * On systems supporting the libquadmath library, GNU Fortran now also
1337 supports a quad-precision, kind=16 floating-point data type
1338 (REAL(16), COMPLEX(16)). As the data type is not fully supported in
1339 hardware, calculations might be one to two orders of magnitude
1340 slower than with the 4, 8 or 10 bytes floating-point data types.
1341 This change does not affect systems which support REAL(16) in
1342 hardware nor those which do not support libquadmath.
1343 * Much improved compile time for large array constructors.
1344 * In order to reduce execution time and memory consumption, use of
1345 temporary arrays in assignment expressions is avoided for many
1346 cases. The compiler now reverses loops in order to avoid generating
1347 a temporary array where possible.
1348 * Improved diagnostics, especially with -fwhole-file.
1349 * The -fwhole-file flag is now enabled by default. This improves code
1350 generation and diagnostics. It can be disabled using the deprecated
1351 -fno-whole-file flag.
1352 * Support the generation of Makefile dependencies via the [15]-M...
1353 flags of GCC; you may need to specify the -cpp option in addition.
1354 The dependencies take modules, Fortran's include, and CPP's
1355 #include into account. Note: Using -M for the module path is no
1356 longer supported, use -J instead.
1357 * The flag -Wconversion has been modified to only issue warnings
1358 where a conversion leads to information loss. This drastically
1359 reduces the number of warnings; -Wconversion is thus now enabled
1360 with -Wall. The flag -Wconversion-extra has been added and also
1361 warns about other conversions; -Wconversion-extra typically issues
1362 a huge number of warnings, most of which can be ignored.
1363 * A new command-line option -Wunused-dummy-argument warns about
1364 unused dummy arguments and is included in -Wall. Before,
1365 -Wunused-variable also warned about unused dummy arguments.
1366 * Fortran 2003 support has been extended:
1367 + Improved support for polymorphism between libraries and
1368 programs and for complicated inheritance patterns (cf.
1369 [16]object-oriented programming).
1370 + Experimental support of the ASSOCIATE construct.
1371 + In pointer assignments it is now possible to specify the lower
1372 bounds of the pointer and, for a rank-1 or a simply contiguous
1373 data-target, to remap the bounds.
1374 + Automatic (re)allocation: In intrinsic assignments to
1375 allocatable variables the left-hand side will be automatically
1376 allocated (if unallocated) or reallocated (if the shape or
1377 type parameter is different). To avoid the small performance
1378 penalty, you can use a(:) = ... instead of a = ... for arrays
1379 and character strings - or disable the feature using -std=f95
1380 or -fno-realloc-lhs.
1381 + Deferred type parameter: For scalar allocatable and pointer
1382 variables the character length can be deferred.
1383 + Namelist variables with allocatable and pointer attribute and
1384 nonconstant length type parameter are supported.
1385 * Fortran 2008 support has been extended:
1386 + Experimental [17]coarray support (for one image only, i.e.
1387 num_images() == 1); use the [18]-fcoarray=single flag to
1389 + The STOP and the new ERROR STOP statements now support all
1390 constant expressions.
1391 + Support for the CONTIGUOUS attribute.
1392 + Support for ALLOCATE with MOLD.
1393 + Support for the STORAGE_SIZE intrinsic inquiry function.
1394 + Support of the NORM2 and PARITY intrinsic functions.
1395 + The following bit intrinsics were added: POPCNT and POPPAR for
1396 counting the number of 1 bits and returning the parity; BGE,
1397 BGT, BLE, and BLT for bitwise comparisons; DSHIFTL and DSHIFTR
1398 for combined left and right shifts, MASKL and MASKR for simple
1399 left and right justified masks, MERGE_BITS for a bitwise merge
1400 using a mask, SHIFTA, SHIFTL and SHIFTR for shift operations,
1401 and the transformational bit intrinsics IALL, IANY and
1403 + Support of the EXECUTE_COMMAND_LINE intrinsic subroutine.
1404 + Support for the IMPURE attribute for procedures, which allows
1405 for ELEMENTAL procedures without the restrictions of PURE.
1406 + Null pointers (including NULL()) and not allocated variables
1407 can be used as actual argument to optional non-pointer,
1408 non-allocatable dummy arguments, denoting an absent argument.
1409 + Non-pointer variables with TARGET attribute can be used as
1410 actual argument to POINTER dummies with INTENT(IN)
1411 + Pointers including procedure pointers and those in a derived
1412 type (pointer components) can now be initialized by a target
1413 instead of only by NULL.
1414 + The EXIT statement (with construct-name) can now be used to
1415 leave not only the DO but also the ASSOCIATE, BLOCK, IF,
1416 SELECT CASE and SELECT TYPE constructs.
1417 + Internal procedures can now be used as actual argument.
1418 + The named constants INTEGER_KINDS, LOGICAL_KINDS, REAL_KINDS
1419 and CHARACTER_KINDS of the intrinsic module ISO_FORTRAN_ENV
1420 have been added; these arrays contain the supported kind
1421 values for the respective types.
1422 + The module procedures C_SIZEOF of the intrinsic module
1423 ISO_C_BINDINGS and COMPILER_VERSION and COMPILER_OPTIONS of
1424 ISO_FORTRAN_ENV have been implemented.
1425 + Minor changes: obsolescence diagnostics for ENTRY was added
1426 for -std=f2008; a line may start with a semicolon; for
1427 internal and module procedures END can be used instead of END
1428 SUBROUTINE and END FUNCTION; SELECTED_REAL_KIND now also takes
1429 a RADIX argument; intrinsic types are supported for
1430 TYPE(intrinsic-type-spec); multiple type-bound procedures can
1431 be declared in a single PROCEDURE statement; implied-shape
1432 arrays are supported for named constants (PARAMETER). The
1433 transformational, three argument versions of BESSEL_JN and
1434 BESSEL_YN were added - the elemental, two-argument version had
1435 been added in GCC 4.4; note that the transformational
1436 functions use a recurrence algorithm.
1440 Support for the [19]Go programming language has been added to GCC. It
1441 is not enabled by default when you build GCC; use the
1442 --enable-languages configure option to build it. The driver program for
1443 compiling Go code is gccgo.
1445 Go is currently known to work on GNU/Linux and RTEMS. Solaris support
1446 is in progress. It may or may not work on other platforms.
1448 Objective-C and Objective-C++
1450 * The -fobjc-exceptions flag is now required to enable Objective-C
1451 exception and synchronization syntax (introduced by the keywords
1452 @try, @catch, @finally and @synchronized).
1453 * A number of Objective-C 2.0 features and extensions are now
1454 supported by GCC. These features are enabled by default; you can
1455 disable them by using the new -fobjc-std=objc1 command-line option.
1456 * The Objective-C 2.0 dot-syntax is now supported. It is an
1457 alternative syntax for using getters and setters; object.count is
1458 automatically converted into [object count] or [object setCount:
1459 ...] depending on context; for example if (object.count > 0) is
1460 automatically compiled into the equivalent of if ([object count] >
1461 0) while object.count = 0; is automatically compiled into the
1462 equivalent ot [object setCount: 0];. The dot-syntax can be used
1463 with instance and class objects and with any setters or getters, no
1464 matter if they are part of a declared property or not.
1465 * Objective-C 2.0 declared properties are now supported. They are
1466 declared using the new @property keyword, and are most commonly
1467 used in conjunction with the new Objective-C 2.0 dot-syntax. The
1468 nonatomic, readonly, readwrite, assign, retain, copy, setter and
1469 getter attributes are all supported. Marking declared properties
1470 with __attribute__ ((deprecated)) is supported too.
1471 * The Objective-C 2.0 @synthesize and @dynamic keywords are
1472 supported. @synthesize causes the compiler to automatically
1473 synthesize a declared property, while @dynamic is used to disable
1474 all warnings for a declared property for which no implementation is
1475 provided at compile time. Synthesizing declared properties requires
1476 runtime support in most useful cases; to be able to use it with the
1477 GNU runtime, appropriate helper functions have been added to the
1478 GNU Objective-C runtime ABI, and are implemented by the GNU
1479 Objective-C runtime library shipped with GCC.
1480 * The Objective-C 2.0 fast enumeration syntax is supported in
1481 Objective-C. This is currently not yet available in Objective-C++.
1482 Fast enumeration requires support in the runtime, and such support
1483 has been added to the GNU Objective-C runtime library (shipped with
1485 * The Objective-C 2.0 @optional keyword is supported. It allows you
1486 to mark methods or properties in a protocol as optional as opposed
1488 * The Objective-C 2.0 @package keyword is supported. It has currently
1489 the same effect as the @public keyword.
1490 * Objective-C 2.0 method attributes are supported. Currently the
1491 supported attributes are deprecated, sentinel, noreturn and format.
1492 * Objective-C 2.0 method argument attributes are supported. The most
1493 widely used attribute is unused, to mark an argument as unused in
1495 * Objective-C 2.0 class and protocol attributes are supported.
1496 Currently the only supported attribute is deprecated.
1497 * Objective-C 2.0 class extensions are supported. A class extension
1498 has the same syntax as a category declaration with no category
1499 name, and the methods and properties declared in it are added
1500 directly to the main class. It is mostly used as an alternative to
1501 a category to add methods to a class without advertising them in
1502 the public headers, with the advantage that for class extensions
1503 the compiler checks that all the privately declared methods are
1504 actually implemented.
1505 * As a result of these enhancements, GCC can now be used to build
1506 Objective-C and Objective-C++ software that uses Foundation and
1507 other important system frameworks with the NeXT runtime on Darwin 9
1508 and Darwin 10 (Mac OS X 10.5 and 10.6). Currently this is for m32
1510 * Many bugs in the compiler have been fixed in this release; in
1511 particular, LTO can now be used when compiling Objective-C and
1512 Objective-C++ and the parser is much more robust in dealing with
1515 Runtime Library (libobjc)
1517 * The GNU Objective-C runtime library now defines the macro
1518 __GNU_LIBOBJC__ (with a value that is increased at every release
1519 where there is any change to the API) in objc/objc.h, making it
1520 easy to determine if the GNU Objective-C runtime library is being
1521 used, and if so, which version. Previous versions of the GNU
1522 Objective-C runtime library (and other Objective-C runtime
1523 libraries such as the Apple one) do not define this macro.
1524 * A new Objective-C 2.0 API, almost identical to the one implemented
1525 by the Apple Objective-C runtime, has been implemented in the GNU
1526 Objective-C runtime library. The new API hides the internals of
1527 most runtime structures but provides a more extensive set of
1528 functions to operate on them. It is much easier, for example, to
1529 create or modify classes at runtime. The new API also makes it
1530 easier to port software from Apple to GNU as almost no changes
1531 should be required. The old API is still supported for backwards
1532 compatibility; including the old objc/objc-api.h header file
1533 automatically selects the old API, while including the new
1534 objc/runtime.h header file automatically selects the new API.
1535 Support for the old API is being phased out and upgrading the
1536 software to use the new API is strongly recommended. To check for
1537 the availability of the new API, the __GNU_LIBOBJC__ macro can be
1538 used as older versions of the GNU Objective-C runtime library,
1539 which do not support the new API, do not define such a macro.
1540 * Runtime support for @synchronized has been added.
1541 * Runtime support for Objective-C 2.0 synthesized property accessors
1543 * Runtime support for Objective-C 2.0 fast enumeration has been
1546 New Targets and Target Specific Improvements
1550 * GCC now supports the Cortex-M4 processor implementing the v7-em
1551 version of the architecture using the option -mcpu=cortex-m4.
1552 * Scheduling descriptions for the Cortex-M4, the Neon and the
1553 floating point units of the Cortex-A9 and a pipeline description
1554 for the Cortex-A5 have been added.
1555 * Synchronization primitives such as __sync_fetch_and_add and friends
1556 are now inlined for supported architectures rather than calling
1557 into a kernel helper function.
1558 * SSA loop prefetching is enabled by default for the Cortex-A9 at
1560 * Several improvements were committed to improve code generation for
1561 the ARM architecture including a rewritten implementation for load
1562 and store multiples.
1563 * Several enhancements were committed to improve SIMD code generation
1564 for NEON by adding support for widening instructions, misaligned
1565 loads and stores, vector conditionals and support for 64 bit
1567 * Support was added for the Faraday cores fa526, fa606te, fa626te,
1568 fmp626te, fmp626 and fa726te and can be used with the respective
1569 names as parameters to the -mcpu= option.
1570 * Basic support was added for Cortex-A15 and is available through
1572 * GCC for AAPCS configurations now more closely adheres to the AAPCS
1573 specification by enabling -fstrict-volatile-bitfields by default.
1577 * The new -fsplit-stack option permits programs to use a
1578 discontiguous stack. This is useful for threaded programs, in that
1579 it is no longer necessary to specify the maximum stack size when
1580 creating a thread. This feature is currently only implemented for
1581 32-bit and 64-bit x86 GNU/Linux targets.
1582 * Support for emitting profiler counter calls before function
1583 prologues. This is enabled via a new command-line option -mfentry.
1584 * Optimization for the Intel Core 2 processors is now available
1585 through the -march=core2 and -mtune=core2 options.
1586 * Support for Intel Core i3/i5/i7 processors is now available through
1587 the -march=corei7 and -mtune=corei7 options.
1588 * Support for Intel Core i3/i5/i7 processors with AVX is now
1589 available through the -march=corei7-avx and -mtune=corei7-avx
1591 * Support for AMD Bobcat (family 14) processors is now available
1592 through the -march=btver1 and -mtune=btver1 options.
1593 * Support for AMD Bulldozer (family 15) processors is now available
1594 through the -march=bdver1 and -mtune=bdver1 options.
1595 * The default setting (when not optimizing for size) for 32-bit
1596 GNU/Linux and Darwin x86 targets has been changed to
1597 -fomit-frame-pointer. The default can be reverted to
1598 -fno-omit-frame-pointer by configuring GCC with the
1599 --enable-frame-pointer configure option.
1600 * Darwin, FreeBSD, Solaris 2, MinGW and Cygwin now all support
1601 __float128 on 32-bit and 64-bit x86 targets.
1602 * AVX floating-point arithmetic can now be enabled by default at
1603 configure time with the new --with-fpmath=avx option.
1604 * The SSA loop prefetching pass is enabled when using -O3 when
1605 optimizing for CPUs where prefetching is beneficial (AMD CPUs newer
1607 * Support for TBM (Trailing Bit Manipulation) built-in functions and
1608 code generation is available via -mtbm.
1609 * Support for AMD's BMI (Bit Manipulation) built-in functions and
1610 code generation is available via -mbmi.
1614 * Support has been added for the Xilinx MicroBlaze softcore processor
1615 (microblaze-elf) embedded target. This configurable processor is
1616 supported on several Xilinx Spartan and Virtex FPGAs.
1620 * GCC now supports the Loongson 3A processor. Its canonical -march=
1621 and -mtune= name is loongson3a.
1625 * The inline assembly register constraint "A" has been renamed "c".
1626 This constraint is used to select a floating-point register that
1627 can be used as the destination of a multiply-accumulate
1629 * New inline assembly register constraints "A" and "D" have been
1630 added. These constraint letters resolve to all general registers
1631 when compiling for AM33, and resolve to address registers only or
1632 data registers only when compiling for MN10300.
1633 * The MDR register is represented in the compiler. One can access the
1634 register via the "z" constraint in inline assembly. It can be
1635 marked as clobbered or used as a local register variable via the
1636 "mdr" name. The compiler uses the RETF instruction if the function
1637 does not modify the MDR register, so it is important that inline
1638 assembly properly annotate any usage of the register.
1642 * GCC now supports the Applied Micro Titan processor with
1644 * The -mrecip option has been added, which indicates whether the
1645 reciprocal and reciprocal square root instructions should be used.
1646 * The -mveclibabi=mass option can be used to enable the compiler to
1647 autovectorize mathematical functions using the Mathematical
1648 Acceleration Subsystem library.
1649 * The -msingle-pic-base option has been added, which instructs the
1650 compiler to avoid loading the PIC base register in function
1651 prologues. The PIC base register must be initialized by the runtime
1653 * The -mblock-move-inline-limit option has been added, which enables
1654 the user to control the maximum size of inlined memcpy calls and
1656 * PowerPC64 GNU/Linux support for applications requiring a large TOC
1657 section has been improved. A new command-line option,
1658 -mcmodel=MODEL, controls this feature; valid values for MODEL are
1659 small, medium, or large.
1660 * The Altivec builtin functions vec_ld and vec_st have been modified
1661 to generate the Altivec memory instructions LVX and STVX, even if
1662 the -mvsx option is used. In the initial GCC 4.5 release, these
1663 builtin functions were changed to generate VSX memory reference
1664 instructions instead of Altivec memory instructions, but there are
1665 differences between the two instructions. If the VSX instruction
1666 set is available, you can now use the new builtin functions
1667 vec_vsx_ld and vec_vsx_st which always generates the VSX memory
1669 * The GCC compiler on AIX now defaults to a process layout with a
1670 larger data space allowing larger programs to be compiled.
1671 * The GCC long double type on AIX 6.1 and above has reverted to 64
1672 bit double precision, matching the AIX XL compiler default, because
1673 of missing C99 symbols required by the GCC runtime.
1674 * The default processor scheduling model and tuning for PowerPC64
1675 GNU/Linux and for AIX 6.1 and above now is POWER7.
1676 * Starting with GCC 4.6.1, vectors of type vector long long or vector
1677 long are passed and returned in the same method as other vectors
1678 with the VSX instruction set. Previously the GCC compiler did not
1679 adhere to the ABI for 128-bit vectors with 64-bit integer base
1680 types (PR 48857). This is also fixed in the GCC 4.5.4 release.
1682 S/390, zSeries and System z9/z10, IBM zEnterprise z196
1684 * Support for the zEnterprise z196 processor has been added. When
1685 using the -march=z196 option, the compiler will generate code
1686 making use of the following instruction facilities:
1687 + Conditional load/store
1689 + Floating-point-extension
1690 + Interlocked-access
1692 The -mtune=z196 option avoids the compare and branch instructions
1693 as well as the load address instruction with an index register as
1694 much as possible and performs instruction scheduling appropriate
1695 for the new out-of-order pipeline architecture.
1696 * When using the -m31 -mzarch options the generated code still
1697 conforms to the 32-bit ABI but uses the general purpose registers
1698 as 64-bit registers internally. This requires a Linux kernel saving
1699 the whole 64-bit registers when doing a context switch. Kernels
1700 providing that feature indicate that by the 'highgprs' string in
1702 * The SSA loop prefetching pass is enabled when using -O3.
1706 * GCC now supports the LEON series of SPARC V8 processors. The code
1707 generated by the compiler can either be tuned to it by means of the
1708 --with-tune=leon configure option and -mtune=leon compilation
1709 option, or the compiler can be built for the sparc-leon-{elf,linux}
1710 and sparc-leon3-{elf,linux} targets directly.
1711 * GCC has stopped sign/zero-extending parameter registers in the
1712 callee for functions taking parameters with sub-word size in 32-bit
1713 mode, since this is redundant with the specification of the ABI.
1714 GCC has never done so in 64-bit mode since this is also redundant.
1715 * The command line option -mfix-at697f has been added to enable the
1716 documented workaround for the single erratum of the Atmel AT697F
1723 * GCC now supports the Bionic C library and provides a convenient way
1724 of building native libraries and applications for the Android
1725 platform. Refer to the documentation of the -mandroid and -mbionic
1726 options for details on building native code. At the moment, Android
1727 support is enabled only for ARM.
1732 + Initial support for CFString types has been added.
1733 This allows GCC to build projects including the system Core
1734 Foundation frameworks. The GCC Objective-C family supports
1735 CFString "toll-free bridged" as per the Mac OS X system tools.
1736 CFString is also recognized in the context of format
1737 attributes and arguments (see the documentation for format
1738 attributes for limitations). At present, 8-bit character types
1740 + Object file size reduction.
1741 The Darwin zeroed memory allocators have been re-written to
1742 make more use of .zerofill sections. For non-debug code, this
1743 can reduce object file size significantly.
1745 + The -mdynamic-no-pic option has been enabled.
1746 Code supporting -mdynamic-no-pic optimization has been added
1747 and is applicable to -m32 builds. The compiler bootstrap uses
1748 the option where appropriate.
1749 + The default value for -mtune= has been changed.
1750 Since Darwin systems are primarily Xeon, Core-2 or similar the
1751 default tuning has been changed to -mtune=core2.
1752 + Enable 128-bit long double (__float128) support on Darwin.
1755 Several significant bugs have been fixed, such that GCC now
1756 produces code compatible with the Darwin64 PowerPC ABI.
1757 + libffi and boehm-gc.
1758 The Darwin ports of the libffi and boehm-gc libraries have
1759 been upgraded to include a Darwin64 implementation. This means
1760 that powerpc*-*-darwin9 platforms may now, for example, build
1761 Java applications with -m64 enabled.
1762 + Plug-in support has been enabled.
1763 + The -fsection-anchors option is now available although,
1764 presently, not heavily tested.
1770 * Support symbol versioning with the Sun linker.
1771 * Allow libstdc++ to leverage full ISO C99 support on Solaris 10+.
1772 * Support thread-local storage (TLS) with the Sun assembler on
1774 * Support TLS on Solaris 8/9 if prerequisites are met.
1775 * Support COMDAT group with the GNU assembler and recent Sun linker.
1776 * Support the Sun assembler visibility syntax.
1777 * Default Solaris 2/x86 to -march=pentium4 (Solaris 10+) resp.
1778 -march=pentiumpro (Solaris 8/9).
1779 * Don't use SSE on Solaris 8/9 x86 by default.
1780 * Enable 128-bit long double (__float128) support on Solaris 2/x86.
1784 * Change the ABI for returning 8-byte vectors like __m64 in MMX
1785 registers on Solaris 10+/x86 to match the Sun Studio 12.1+
1786 compilers. This is an incompatible change. If you use such types,
1787 you must either recompile all your code with the new compiler or
1788 use the new -mvect8-ret-in-mem option to remain compatible with
1789 previous versions of GCC and Sun Studio.
1793 * Initial support for decimal floating point.
1794 * Support for the __thiscall calling-convention.
1795 * Support for hot-patchable function prologues via the
1796 ms_hook_prologue attribute for x86_64 in addition to 32-bit x86.
1797 * Improvements of stack-probing and stack-allocation mechanisms.
1798 * Support of push/pop-macro pragma as preprocessor command.
1799 With #pragma push_macro("macro-name") the current definition of
1800 macro-name is saved and can be restored with #pragma
1801 pop_macro("macro-name") to its saved definition.
1802 * Enable 128-bit long double (__float128) support on MinGW and
1805 Other significant improvements
1807 Installation changes
1809 * An install-strip make target is provided that installs stripped
1810 executables, and may install libraries with unneeded or debugging
1812 * On Power7 systems, there is a potential problem if you build the
1813 GCC compiler with a host compiler using options that enable the VSX
1814 instruction set generation. If the host compiler has been patched
1815 so that the vec_ld and vec_st builtin functions generate Altivec
1816 memory instructions instead of VSX memory instructions, then you
1817 should be able to build the compiler with VSX instruction
1820 Changes for GCC Developers
1822 Note: these changes concern developers that develop GCC itself or
1823 software that integrates with GCC, such as plugins, and not the general
1825 * The gengtype utility, which previously was internal to the GCC
1826 build process, has been enchanced to provide GC root information
1827 for plugins as necessary.
1828 * The old GC allocation interface of ggc_alloc and friends was
1829 replaced with a type-safe alternative.
1833 This is the [20]list of problem reports (PRs) from GCC's bug tracking
1834 system that are known to be fixed in the 4.6.1 release. This list might
1835 not be complete (that is, it is possible that some PRs that have been
1836 fixed are not listed here).
1840 This is the [21]list of problem reports (PRs) from GCC's bug tracking
1841 system that are known to be fixed in the 4.6.2 release. This list might
1842 not be complete (that is, it is possible that some PRs that have been
1843 fixed are not listed here).
1847 This is the [22]list of problem reports (PRs) from GCC's bug tracking
1848 system that are known to be fixed in the 4.6.3 release. This list might
1849 not be complete (that is, it is possible that some PRs that have been
1850 fixed are not listed here).
1853 For questions related to the use of GCC, please consult these web
1854 pages and the [23]GCC manuals. If that fails, the
1855 [24]gcc-help@gcc.gnu.org mailing list might help. Comments on these
1856 web pages and the development of GCC are welcome on our developer
1857 list at [25]gcc@gcc.gnu.org. All of [26]our lists have public
1860 Copyright (C) [27]Free Software Foundation, Inc. Verbatim copying and
1861 distribution of this entire article is permitted in any medium,
1862 provided this notice is preserved.
1864 These pages are [28]maintained by the GCC team. Last modified
1869 1. http://sourceware.org/bugzilla/show_bug.cgi?id=10401
1870 2. http://gcc.gnu.org/gcc-4.5/changes.html
1871 3. http://gcc.gnu.org/gcc-4.5/changes.html#obsoleted
1872 4. http://gcc.gnu.org/gcc-4.6/porting_to.html
1873 5. http://gcc.gnu.org/projects/lto/whopr.pdf
1874 6. http://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html#Warning-Options
1875 7. http://gcc.gnu.org/gcc-4.6/cxx0x_status.html
1876 8. http://gcc.gnu.org/PR43145
1877 9. http://gcc.gnu.org/PR43680
1878 10. http://gcc.gnu.org/PR33558
1879 11. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_active.html#253
1880 12. http://gcc.gnu.org/onlinedocs/libstdc++/manual/status.html#status.iso.200x
1881 13. http://gcc.gnu.org/onlinedocs/libstdc++/manual/debug_mode.html
1882 14. http://gcc.gnu.org/onlinedocs/libstdc++/manual/debug.html#debug.races
1883 15. http://gcc.gnu.org/onlinedocs/gcc/Preprocessor-Options.html
1884 16. http://gcc.gnu.org/wiki/OOP
1885 17. http://gcc.gnu.org/wiki/Coarray
1886 18. http://gcc.gnu.org/onlinedocs/gfortran/Code-Gen-Options.html#index-g_t_0040code_007bfcoarray_007d-233
1887 19. http://golang.org/
1888 20. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.6.1
1889 21. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.6.2
1890 22. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.6.3
1891 23. http://gcc.gnu.org/onlinedocs/
1892 24. mailto:gcc-help@gcc.gnu.org
1893 25. mailto:gcc@gcc.gnu.org
1894 26. http://gcc.gnu.org/lists.html
1895 27. http://www.fsf.org/
1896 28. http://gcc.gnu.org/about.html
1897 29. http://validator.w3.org/check/referer
1898 ======================================================================
1899 http://gcc.gnu.org/gcc-4.5/index.html
1900 GCC 4.5 Release Series
1904 The [1]GNU project and the GCC developers are pleased to announce the
1905 release of GCC 4.5.4.
1907 This release is a bug-fix release, containing fixes for regressions in
1908 GCC 4.5.3 relative to previous releases of GCC.
1913 Jul 2, 2012 ([2]changes)
1916 Apr 28, 2011 ([3]changes)
1919 Dec 16, 2010 ([4]changes)
1922 Jul 31, 2010 ([5]changes)
1925 April 14, 2010 ([6]changes)
1927 References and Acknowledgements
1929 GCC used to stand for the GNU C Compiler, but since the compiler
1930 supports several other languages aside from C, it now stands for the
1931 GNU Compiler Collection.
1933 A list of [7]successful builds is updated as new information becomes
1936 The GCC developers would like to thank the numerous people that have
1937 contributed new features, improvements, bug fixes, and other changes as
1938 well as test results to GCC. This [8]amazing group of volunteers is
1939 what makes GCC successful.
1941 For additional information about GCC please refer to the [9]GCC project
1942 web site or contact the [10]GCC development mailing list.
1944 To obtain GCC please use [11]our mirror sites or [12]our SVN server.
1947 For questions related to the use of GCC, please consult these web
1948 pages and the [13]GCC manuals. If that fails, the
1949 [14]gcc-help@gcc.gnu.org mailing list might help. Comments on these
1950 web pages and the development of GCC are welcome on our developer
1951 list at [15]gcc@gcc.gnu.org. All of [16]our lists have public
1954 Copyright (C) [17]Free Software Foundation, Inc. Verbatim copying and
1955 distribution of this entire article is permitted in any medium,
1956 provided this notice is preserved.
1958 These pages are [18]maintained by the GCC team. Last modified
1963 1. http://www.gnu.org/
1964 2. http://gcc.gnu.org/gcc-4.5/changes.html
1965 3. http://gcc.gnu.org/gcc-4.5/changes.html
1966 4. http://gcc.gnu.org/gcc-4.5/changes.html
1967 5. http://gcc.gnu.org/gcc-4.5/changes.html
1968 6. http://gcc.gnu.org/gcc-4.5/changes.html
1969 7. http://gcc.gnu.org/gcc-4.5/buildstat.html
1970 8. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
1971 9. http://gcc.gnu.org/index.html
1972 10. mailto:gcc@gcc.gnu.org
1973 11. http://gcc.gnu.org/mirrors.html
1974 12. http://gcc.gnu.org/svn.html
1975 13. http://gcc.gnu.org/onlinedocs/
1976 14. mailto:gcc-help@gcc.gnu.org
1977 15. mailto:gcc@gcc.gnu.org
1978 16. http://gcc.gnu.org/lists.html
1979 17. http://www.fsf.org/
1980 18. http://gcc.gnu.org/about.html
1981 19. http://validator.w3.org/check/referer
1982 ======================================================================
1983 http://gcc.gnu.org/gcc-4.5/changes.html
1984 GCC 4.5 Release Series
1985 Changes, New Features, and Fixes
1989 * GCC now requires the [1]MPC library in order to build. See the
1990 [2]prerequisites page for version requirements.
1991 * Support for a number of older systems and recently unmaintained or
1992 untested target ports of GCC has been declared obsolete in GCC 4.5.
1993 Unless there is activity to revive them, the next release of GCC
1994 will have their sources permanently removed.
1995 The following ports for individual systems on particular
1996 architectures have been obsoleted:
1997 + IRIX releases before 6.5 (mips-sgi-irix5*,
1998 mips-sgi-irix6.[0-4])
1999 + Solaris 7 (*-*-solaris2.7)
2000 + Tru64 UNIX releases before V5.1 (alpha*-dec-osf4*,
2002 + Details for the IRIX, Solaris 7, and Tru64 UNIX obsoletions
2003 can be found in the [3]announcement.
2004 Support for the classic POWER architecture implemented in the
2005 original RIOS and RIOS2 processors of the old IBM RS/6000 product
2006 line has been obsoleted in the rs6000 port. This does not affect
2007 the new generation Power and PowerPC architectures.
2008 * Support has been removed for all the [4]configurations obsoleted in
2010 * Support has been removed for the protoize and unprotoize utilities,
2011 obsoleted in GCC 4.4.
2012 * Support has been removed for tuning for Itanium1 (Merced) variants.
2013 Note that code tuned for Itanium2 should also run correctly on
2015 * GCC now generates unwind info also for epilogues. DWARF debuginfo
2016 generated by GCC now uses more features of DWARF3 than before, and
2017 also some DWARF4 features. GDB older than 7.0 is not able to handle
2018 either of these, so to debug GCC 4.5 generated binaries or
2019 libraries GDB 7.0 or later is needed. You can disable use of DWARF4
2020 features with the -gdwarf-3 -gstrict-dwarf options, or use
2021 -gdwarf-2 -gstrict-dwarf to restrict GCC to just DWARF2, but
2022 epilogue unwind info is emitted unconditionally whenever unwind
2024 * On x86 targets, code containing floating-point calculations may run
2025 significantly slower when compiled with GCC 4.5 in strict C99
2026 conformance mode than they did with earlier GCC versions. This is
2027 due to stricter standard conformance of the compiler and can be
2028 avoided by using the option -fexcess-precision=fast; also see
2030 * The function attribute noinline no longer prevents GCC from cloning
2031 the function. A new attribute noclone has been introduced for this
2032 purpose. Cloning a function means that it is duplicated and the new
2033 copy is specialized for certain contexts (for example when a
2034 parameter is a known constant).
2036 General Optimizer Improvements
2038 * The -save-temps now takes an optional argument. The -save-temps and
2039 -save-temps=cwd switches write the temporary files in the current
2040 working directory based on the original source file. The
2041 -save-temps=obj switch will write files into the directory
2042 specified with the -o option, and the intermediate filenames are
2043 based on the output file. This will allow the user to get the
2044 compiler intermediate files when doing parallel builds without two
2045 builds of the same filename located in different directories from
2046 interfering with each other.
2047 * Debugging dumps are now created in the same directory as the object
2048 file rather than in the current working directory. This allows the
2049 user to get debugging dumps when doing parallel builds without two
2050 builds of the same filename interfering with each other.
2051 * GCC has been integrated with the [6]MPC library. This allows GCC to
2052 evaluate complex arithmetic at compile time [7]more accurately. It
2053 also allows GCC to evaluate calls to complex built-in math
2054 functions having constant arguments and replace them at compile
2055 time with their mathematically equivalent results. In doing so, GCC
2056 can generate correct results regardless of the math library
2057 implementation or floating point precision of the host platform.
2058 This also allows GCC to generate identical results regardless of
2059 whether one compiles in native or cross-compile configurations to a
2060 particular target. The following built-in functions take advantage
2061 of this new capability: cacos, cacosh, casin, casinh, catan,
2062 catanh, ccos, ccosh, cexp, clog, cpow, csin, csinh, csqrt, ctan,
2063 and ctanh. The float and long double variants of these functions
2064 (e.g. csinf and csinl) are also handled.
2065 * A new link-time optimizer has been added ([8]-flto). When this
2066 option is used, GCC generates a bytecode representation of each
2067 input file and writes it to specially-named sections in each object
2068 file. When the object files are linked together, all the function
2069 bodies are read from these named sections and instantiated as if
2070 they had been part of the same translation unit. This enables
2071 interprocedural optimizations to work across different files (and
2072 even different languages), potentially improving the performance of
2073 the generated code. To use the link-timer optimizer, -flto needs to
2074 be specified at compile time and during the final link. If the
2075 program does not require any symbols to be exported, it is possible
2076 to combine -flto and the experimental [9]-fwhopr with
2077 [10]-fwhole-program to allow the interprocedural optimizers to use
2078 more aggressive assumptions.
2079 * The automatic parallelization pass was enhanced to support
2080 parallelization of outer loops.
2081 * Automatic parallelization can be enabled as part of Graphite. In
2082 addition to -ftree-parallelize-loops=, specify
2083 -floop-parallelize-all to enable the Graphite-based optimization.
2084 * The infrastructure for optimizing based on [11]restrict qualified
2085 pointers has been rewritten and should result in code generation
2086 improvements. Optimizations based on restrict qualified pointers
2087 are now also available when using -fno-strict-aliasing.
2088 * There is a new optimization pass that attempts to change prototype
2089 of functions to avoid unused parameters, pass only relevant parts
2090 of structures and turn arguments passed by reference to arguments
2091 passed by value when possible. It is enabled by -O2 and above as
2092 well as -Os and can be manually invoked using the new command-line
2094 * GCC now optimize exception handling code. In particular cleanup
2095 regions that are proved to not have any effect are optimized out.
2097 New Languages and Language specific improvements
2101 * The -fshow-column option is now on by default. This means error
2102 messages now have a column associated with them.
2106 * Compilation of programs heavily using discriminated record types
2107 with variant parts has been sped up and generates more compact
2109 * Stack checking now works reasonably well on most plaforms. In some
2110 specific cases, stack overflows may still fail to be detected, but
2111 a compile-time warning will be issued for these cases.
2115 * If a header named in a #include directive is not found, the
2116 compiler exits immediately. This avoids a cascade of errors arising
2117 from declarations expected to be found in that header being
2119 * A new built-in function __builtin_unreachable() has been added that
2120 tells the compiler that control will never reach that point. It may
2121 be used after asm statements that terminate by transferring control
2122 elsewhere, and in other places that are known to be unreachable.
2123 * The -Wlogical-op option now warns for logical expressions such as
2124 (c == 1 && c == 2) and (c != 1 || c != 2), which are likely to be
2125 mistakes. This option is disabled by default.
2126 * An asm goto feature has been added to allow asm statements that
2128 * C++0x raw strings are supported for C++ and for C with -std=gnu99.
2129 * The deprecated attribute now takes an optional string argument, for
2130 example, __attribute__((deprecated("text string"))), that will be
2131 printed together with the deprecation warning.
2135 * The -Wenum-compare option, which warns when comparing values of
2136 different enum types, now works for C. It formerly only worked for
2137 C++. This warning is enabled by -Wall. It may be avoided by using a
2139 * The -Wcast-qual option now warns about casts which are unsafe in
2140 that they permit const-correctness to be violated without further
2141 warnings. Specifically, it warns about cases where a qualifier is
2142 added when all the lower types are not const. For example, it warns
2143 about a cast from char ** to const char **.
2144 * The -Wc++-compat option is significantly improved. It issues new
2146 + Using C++ reserved operator names as identifiers.
2147 + Conversions to enum types without explicit casts.
2148 + Using va_arg with an enum type.
2149 + Using different enum types in the two branches of ?:.
2150 + Using ++ or -- on a variable of enum type.
2151 + Using the same name as both a struct, union or enum tag and a
2152 typedef, unless the typedef refers to the tagged type itself.
2153 + Using a struct, union, or enum which is defined within another
2155 + A struct field defined using a typedef if there is a field in
2156 the struct, or an enclosing struct, whose name is the typedef
2158 + Duplicate definitions at file scope.
2159 + Uninitialized const variables.
2160 + A global variable with an anonymous struct, union, or enum
2162 + Using a string constant to initialize a char array whose size
2163 is the length of the string.
2164 * The new -Wjump-misses-init option warns about cases where a goto or
2165 switch skips the initialization of a variable. This sort of branch
2166 is an error in C++ but not in C. This warning is enabled by
2168 * GCC now ensures that a C99-conforming <stdint.h> is present on most
2169 targets, and uses information about the types in this header to
2170 implement the Fortran bindings to those types. GCC does not ensure
2171 the presence of such a header, and does not implement the Fortran
2172 bindings, on the following targets: NetBSD, VxWorks, VMS,
2173 SymbianOS, WinCE, LynxOS, Netware, QNX, Interix, TPF.
2174 * GCC now implements C90- and C99-conforming rules for constant
2175 expressions. This may cause warnings or errors for some code using
2176 expressions that can be folded to a constant but are not constant
2177 expressions as defined by ISO C.
2178 * All known target-independent C90 and C90 Amendment 1 conformance
2179 bugs, and all known target-independent C99 conformance bugs not
2180 related to floating point or extended identifiers, have been fixed.
2181 * The C decimal floating point support now includes support for the
2182 FLOAT_CONST_DECIMAL64 pragma.
2183 * The named address space feature from ISO/IEC TR 18037 is now
2184 supported. This is currently only implemented for the SPU
2189 * Improved [12]experimental support for the upcoming C++0x ISO C++
2190 standard, including support for raw strings, lambda expressions and
2191 explicit type conversion operators.
2192 * When printing the name of a class template specialization, G++ will
2193 now omit any template arguments which come from default template
2194 arguments. This behavior (and the pretty-printing of function
2195 template specializations as template signature and arguments) can
2196 be disabled with the -fno-pretty-templates option.
2197 * Access control is now applied to typedef names used in a template,
2198 which may cause G++ to reject some ill-formed code that was
2199 accepted by earlier releases. The -fno-access-control option can be
2200 used as a temporary workaround until the code is corrected.
2201 * Compilation time for code that uses templates should now scale
2202 linearly with the number of instantiations rather than
2203 quadratically, as template instantiations are now looked up using
2205 * Declarations of functions that look like builtin declarations of
2206 library functions are only considered to be redeclarations if they
2207 are declared with extern "C". This may cause problems with code
2208 that omits extern "C" on hand-written declarations of C library
2209 functions such as abort or memcpy. Such code is ill-formed, but was
2210 accepted by earlier releases.
2211 * Diagnostics that used to complain about passing non-POD types to
2212 ... or jumping past the declaration of a non-POD variable now check
2213 for triviality rather than PODness, as per C++0x.
2214 * In C++0x mode local and anonymous classes are now allowed as
2215 template arguments, and in declarations of variables and functions
2216 with linkage, so long as any such declaration that is used is also
2217 defined ([13]DR 757).
2218 * Labels may now have attributes, as has been permitted for a while
2219 in C. This is only permitted when the label definition and the
2220 attribute specifier is followed by a semicolon--i.e., the label
2221 applies to an empty statement. The only useful attribute for a
2223 * G++ now implements [14]DR 176. Previously G++ did not support using
2224 the injected-class-name of a template base class as a type name,
2225 and lookup of the name found the declaration of the template in the
2226 enclosing scope. Now lookup of the name finds the
2227 injected-class-name, which can be used either as a type or as a
2228 template, depending on whether or not the name is followed by a
2229 template argument list. As a result of this change, some code that
2230 was previously accepted may be ill-formed because
2231 1. The injected-class-name is not accessible because it's from a
2233 2. The injected-class-name cannot be used as an argument for a
2234 template template parameter.
2235 In either of these cases, the code can be fixed by adding a
2236 nested-name-specifier to explicitly name the template. The first
2237 can be worked around with -fno-access-control; the second is only
2238 rejected with -pedantic.
2239 * A new standard mangling for SIMD vector types has been added, to
2240 avoid name clashes on systems with vectors of varying length. By
2241 default the compiler still uses the old mangling, but emits aliases
2242 with the new mangling on targets that support strong aliases. Users
2243 can switch over entirely to the new mangling with -fabi-version=4
2244 or -fabi-version=0. -Wabi will now warn about code that uses the
2246 * The command-line option -ftemplate-depth-N is now written as
2247 -ftemplate-depth=N and the old form is deprecated.
2248 * Conversions between NULL and non-pointer types are now warned by
2249 default. The new option -Wno-conversion-null disables these
2250 warnings. Previously these warnings were only available when using
2251 -Wconversion explicitly.
2253 Runtime Library (libstdc++)
2255 * [15]Improved experimental support for the upcoming ISO C++
2256 standard, C++0x, including:
2257 + Support for <future>, <functional>, and <random>.
2258 + Existing facilities now exploit explicit operators and the
2259 newly implemented core C++0x features.
2260 * An experimental [16]profile mode has been added. This is an
2261 implementation of many C++ standard library constructs with an
2262 additional analysis layer that gives performance improvement advice
2263 based on recognition of suboptimal usage patterns. For example,
2268 for (int k = 0; k < 1024; ++k)
2269 v.insert(v.begin(), k);
2272 When instrumented via the profile mode, can return suggestions
2273 about the initial size and choice of the container used as follows:
2274 vector-to-list: improvement = 5: call stack = 0x804842c ...
2275 : advice = change std::vector to std::list
2276 vector-size: improvement = 3: call stack = 0x804842c ...
2277 : advice = change initial container size from 0 to 1024
2279 These constructs can be substituted for the normal libstdc++
2280 constructs on a piecemeal basis, or all existing components can be
2281 transformed via the -D_GLIBCXX_PROFILE macro.
2282 * [17]Support for decimal floating-point arithmetic (aka ISO C++ TR
2283 24733) has been added. This support is in header file
2284 <decimal/decimal>, uses namespace std::decimal, and includes
2285 classes decimal32, decimal64, and decimal128.
2286 * Sources have been audited for application of function attributes
2287 nothrow, const, pure, and noreturn.
2288 * Python pretty-printers have been added for many standard library
2289 components that simplify the internal representation and present a
2290 more intuitive view of components when used with
2291 appropriately-advanced versions of GDB. For more information,
2292 please consult the more [18]detailed description.
2293 * The default behavior for comparing typeinfo names has changed, so
2294 in <typeinfo>, __GXX_MERGED_TYPEINFO_NAMES now defaults to zero.
2295 * The new -static-libstdc++ option directs g++ to link the C++
2296 library statically, even if the default would normally be to link
2301 * The COMMON default padding has been changed - instead of adding the
2302 padding before a variable it is now added afterwards, which
2303 increases the compatibility with other vendors and helps to obtain
2304 the correct output in some cases. Cf. also the -falign-commons
2305 option ([19]added in 4.4).
2306 * The -finit-real= option now also supports the value snan for
2307 signalling not-a-number; to be effective, one additionally needs to
2308 enable trapping (e.g. via -ffpe-trap=). Note: Compile-time
2309 optimizations can turn a signalling NaN into a quiet one.
2310 * The new option -fcheck= has been added with the options bounds,
2311 array-temps, do, pointer, and recursive. The bounds and array-temps
2312 options are equivalent to -fbounds-check and
2313 -fcheck-array-temporaries. The do option checks for invalid
2314 modification of loop iteration variables, and the recursive option
2315 tests for recursive calls to subroutines/functions which are not
2316 marked as recursive. With pointer pointer association checks in
2317 calls are performed; however, neither undefined pointers nor
2318 pointers in expressions are handled. Using -fcheck=all enables all
2319 these run-time checks.
2320 * The run-time checking -fcheck=bounds now warns about invalid string
2321 lengths of character dummy arguments. Additionally, more
2322 compile-time checks have been added.
2323 * The new option [20]-fno-protect-parens has been added; if set, the
2324 compiler may reorder REAL and COMPLEX expressions without regard to
2326 * GNU Fortran no longer links against libgfortranbegin. As before,
2327 MAIN__ (assembler symbol name) is the actual Fortran main program,
2328 which is invoked by the main function. However, main is now
2329 generated and put in the same object file as MAIN__. For the time
2330 being, libgfortranbegin still exists for backward compatibility.
2331 For details see the new [21]Mixed-Language Programming chapter in
2333 * The I/O library was restructured for performance and cleaner code.
2334 * Array assignments and WHERE are now run in parallel when OpenMP's
2336 * The experimental option -fwhole-file was added. The option allows
2337 whole-file checking of procedure arguments and allows for better
2338 optimizations. It can also be used with -fwhole-program, which is
2339 now also supported in gfortran.
2340 * More Fortran 2003 and Fortran 2008 mathematical functions can now
2341 be used as initialization expressions.
2342 * Some extended attributes such as STDCALL are now supported via the
2343 [22]GCC$ compiler directive.
2344 * For Fortran 77 compatibility: If -fno-sign-zero is used, the SIGN
2345 intrinsic behaves now as if zero were always positive.
2346 * For legacy compatibiliy: On Cygwin and MinGW, the special files
2347 CONOUT$ and CONIN$ (and CONERR$ which maps to CONOUT$) are now
2349 * Fortran 2003 support has been extended:
2350 + Procedure-pointer function results and procedure-pointer
2351 components (including PASS),
2352 + allocatable scalars (experimental),
2353 + DEFERRED type-bound procedures,
2354 + the ERRMSG= argument of the ALLOCATE and DEALLOCATE statements
2355 have been implemented.
2356 + The ALLOCATE statement supports type-specs and the SOURCE=
2358 + OPERATOR(*) and ASSIGNMENT(=) are now allowed as GENERIC
2359 type-bound procedure (i.e. as type-bound operators).
2360 + Rounding (ROUND=, RZ, ...) for output is now supported.
2361 + The INT_FAST{8,16,32,64,128}_T kind type parameters of the
2362 intrinsic module ISO_C_BINDING are now supported, except for
2363 the targets listed above as ones where GCC does not have
2364 <stdint.h> type information.
2365 + Extensible derived types with type-bound procedure or
2366 procedure pointer with PASS attribute now have to use CLASS in
2367 line with the Fortran 2003 standard; the workaround to use
2368 TYPE is no longer supported.
2369 + [23]Experimental, incomplete support for polymorphism,
2370 including CLASS, SELECT TYPE and dynamic dispatch of
2371 type-bound procedure calls. Some features do not work yet such
2372 as unlimited polymorphism (CLASS(*)).
2373 * Fortran 2008 support has been extended:
2374 + The OPEN statement now supports the NEWUNIT= option, which
2375 returns a unique file unit, thus preventing inadvertent use of
2376 the same unit in different parts of the program.
2377 + Support for unlimited format items has been added.
2378 + The INT{8,16,32} and REAL{32,64,128} kind type parameters of
2379 the intrinsic module ISO_FORTRAN_ENV are now supported.
2380 + Using complex arguments with TAN, SINH, COSH, TANH, ASIN,
2381 ACOS, and ATAN is now possible; the functions ASINH, ACOSH,
2382 and ATANH have been added (for real and complex arguments) and
2383 ATAN(Y,X) is now an alias for ATAN2(Y,X).
2384 + The BLOCK construct has been implemented.
2386 New Targets and Target Specific Improvements
2390 * Full cross-toolchain support now available with GNU Binutils
2394 * GCC now supports the Cortex-M0 and Cortex-A5 processors.
2395 * GCC now supports the ARM v7E-M architecture.
2396 * GCC now supports VFPv4-based FPUs and FPUs with
2397 single-precision-only VFP.
2398 * GCC has many improvements to optimization for other ARM processors,
2399 including scheduling support for the integer pipeline on Cortex-A9.
2400 * GCC now supports the IEEE 754-2008 half-precision floating-point
2401 type, and a variant ARM-specific half-precision type. This type is
2402 specified using __fp16, with the layout determined by
2403 -mfp16-format. With appropriate -mfpu options, the Cortex-A9 and
2404 VFPv4 half-precision instructions will be used.
2405 * GCC now supports the variant of AAPCS that uses VFP registers for
2406 parameter passing and return values.
2410 * The -mno-tablejump option has been removed because it has the same
2411 effect as the -fno-jump-tables option.
2412 * Added support for these new AVR devices:
2419 * GCC now will set the default for -march= based on the configure
2421 * GCC now supports handling floating-point excess precision arising
2422 from use of the x87 floating-point unit in a way that conforms to
2423 ISO C99. This is enabled with -fexcess-precision=standard and with
2424 standards conformance options such as -std=c99, and may be disabled
2425 using -fexcess-precision=fast.
2426 * Support for the Intel Atom processor is now available through the
2427 -march=atom and -mtune=atom options.
2428 * A new -mcrc32 option is now available to enable crc32 intrinsics.
2429 * A new -mmovbe option is now available to enable GCC to use the
2430 movbe instruction to implement __builtin_bswap32 and
2432 * SSE math now can be enabled by default at configure time with the
2433 new --with-fpmath=sse option.
2434 * There is a new intrinsic header file, <x86intrin.h>. It should be
2435 included before using any IA-32/x86-64 intrinsics.
2436 * Support for the XOP, FMA4, and LWP instruction sets for the AMD
2437 Orochi processors are now available with the -mxop, -mfma4, and
2439 * The -mabm option enables GCC to use the popcnt and lzcnt
2440 instructions on AMD processors.
2441 * The -mpopcnt option enables GCC to use the popcnt instructions on
2442 both AMD and Intel processors.
2446 * GCC now supports ColdFire 51xx, 5221x, 5225x, 52274, 52277, 5301x
2448 * GCC now supports thread-local storage (TLS) on M68K and ColdFire
2453 Support has been added for the Toshiba Media embedded Processor (MeP,
2454 or mep-elf) embedded target.
2458 * GCC now supports MIPS 1004K processors.
2459 * GCC can now be configured with options --with-arch-32,
2460 --with-arch-64, --with-tune-32 and --with-tune-64 to control the
2461 default optimization separately for 32-bit and 64-bit modes.
2462 * MIPS targets now support an alternative _mcount interface, in which
2463 register $12 points to the function's save slot for register $31.
2464 This interface is selected by the -mcount-ra-address option; see
2465 the documentation for more details.
2466 * GNU/Linux targets can now generate read-only .eh_frame sections.
2467 This optimization requires GNU binutils 2.20 or above, and is only
2468 available if GCC is configured with a suitable version of binutils.
2469 * GNU/Linux targets can now attach special relocations to indirect
2470 calls, so that the linker can turn them into direct jumps or
2471 branches. This optimization requires GNU binutils 2.20 or later,
2472 and is automatically selected if GCC is configured with an
2473 appropriate version of binutils. It can be explicitly enabled or
2474 disabled using the -mrelax-pic-calls command-line option.
2475 * GCC now generates more heavily-optimized atomic operations on
2477 * MIPS targets now support the -fstack-protector option.
2478 * GCC now supports an -msynci option, which specifies that synci is
2479 enough to flush the instruction cache, without help from the
2480 operating system. GCC uses this information to optimize
2481 automatically-generated cache flush operations, such as those used
2482 for nested functions in C. There is also a --with-synci
2483 configure-time option, which makes -msynci the default.
2484 * GCC supports four new function attributes for interrupt handlers:
2485 interrupt, use_shadow_register_set, keep_interrupts_masked and
2486 use_debug_exception_return. See the documentation for more details
2487 about these attributes.
2489 RS/6000 (POWER/PowerPC)
2491 * GCC now supports the Power ISA 2.06, which includes the VSX
2492 instructions that add vector 64-bit floating point support, new
2493 population count instructions, and conversions between floating
2494 point and unsigned types.
2495 * Support for the power7 processor is now available through the
2496 -mcpu=power7 and -mtune=power7.
2497 * GCC will now vectorize loops that contain simple math functions
2498 like copysign when generating code for altivec or VSX targets.
2499 * Support for the A2 processor is now available through the -mcpu=a2
2500 and -mtune=a2 options.
2501 * Support for the 476 processor is now available through the
2502 -mcpu={476,476fp} and -mtune={476,476fp} options.
2503 * Support for the e500mc64 processor is now available through the
2504 -mcpu=e500mc64 and -mtune=e500mc64 options.
2505 * GCC can now be configured with options --with-cpu-32,
2506 --with-cpu-64, --with-tune-32 and --with-tune-64 to control the
2507 default optimization separately for 32-bit and 64-bit modes.
2508 * Starting with GCC 4.5.4, vectors of type vector long long or vector
2509 long are passed and returned in the same method as other vectors
2510 with the VSX instruction set. Previously the GCC compiler did not
2511 adhere to the ABI for 128-bit vectors with 64-bit integer base
2512 types (PR 48857). This is also fixed in the GCC 4.6.1 release.
2516 Support has been added for the Renesas RX Processor (rx-elf) target.
2520 Windows (Cygwin and MinGW)
2522 * GCC now installs all the major language runtime libraries as DLLs
2523 when configured with the --enable-shared option.
2524 * GCC now makes use of the new support for aligned common variables
2525 in versions of binutils >= 2.20 to fix bugs in the support for SSE
2527 * Improvements to the libffi support library increase the reliability
2528 of code generated by GCJ on all Windows platforms. Libgcj is
2529 enabled by default for the first time.
2530 * Libtool improvements simplify installation by placing the generated
2531 DLLs in the correct binaries directory.
2532 * Numerous other minor bugfixes and improvements, and substantial
2533 enhancements to the Fortran language support library.
2537 Other significant improvements
2541 * It is now possible to extend the compiler without having to modify
2542 its source code. A new option -fplugin=file.so tells GCC to load
2543 the shared object file.so and execute it as part of the compiler.
2544 The internal documentation describes the details on how plugins can
2545 interact with the compiler.
2547 Installation changes
2549 * The move to newer autotools changed default installation
2550 directories and switches to control them: The --with-datarootdir,
2551 --with-docdir, --with-pdfdir, and --with-htmldir switches are not
2552 used any more. Instead, you can now use --datarootdir, --docdir,
2553 --htmldir, and --pdfdir. The default installation directories have
2554 changed as follows according to the GNU Coding Standards:
2556 datarootdir read-only architecture-independent data root [PREFIX/share]
2557 localedir locale-specific message catalogs [DATAROOTDIR/locale]
2558 docdir documentation root [DATAROOTDIR/doc/PACKAGE]
2559 htmldir html documentation [DOCDIR]
2560 dvidir dvi documentation [DOCDIR]
2561 pdfdir pdf documentation [DOCDIR]
2562 psdir ps documentation [DOCDIR]
2563 The following variables have new default values:
2565 datadir read-only architecture-independent data [DATAROOTDIR]
2566 infodir info documentation [DATAROOTDIR/info]
2567 mandir man documentation [DATAROOTDIR/man]
2571 This is the [24]list of problem reports (PRs) from GCC's bug tracking
2572 system that are known to be fixed in the 4.5.1 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).
2578 * GCC's new link-time optimizer ([25]-flto) now also works on a few
2580 + Cygwin (*-cygwin*)
2582 + Darwin on x86-64 (x86_64-apple-darwin*)
2583 LTO is not enabled by default for these targets. To enable LTO, you
2584 should configure with the --enable-lto option.
2588 This is the [26]list of problem reports (PRs) from GCC's bug tracking
2589 system that are known to be fixed in the 4.5.2 release. This list might
2590 not be complete (that is, it is possible that some PRs that have been
2591 fixed are not listed here).
2595 This is the [27]list of problem reports (PRs) from GCC's bug tracking
2596 system that are known to be fixed in the 4.5.3 release. This list might
2597 not be complete (that is, it is possible that some PRs that have been
2598 fixed are not listed here).
2600 On the PowerPC compiler, the Altivec builtin functions vec_ld and
2601 vec_st have been modified to generate the Altivec memory instructions
2602 LVX and STVX, even if the -mvsx option is used. In the initial GCC 4.5
2603 release, these builtin functions were changed to generate VSX memory
2604 reference instructions instead of Altivec memory instructions, but
2605 there are differences between the two instructions. If the VSX
2606 instruction set is available, you can now use the new builtin functions
2607 vec_vsx_ld and vec_vsx_st which always generates the VSX memory
2612 This is the [28]list of problem reports (PRs) from GCC's bug tracking
2613 system that are known to be fixed in the 4.5.4 release. This list might
2614 not be complete (that is, it is possible that some PRs that have been
2615 fixed are not listed here).
2618 For questions related to the use of GCC, please consult these web
2619 pages and the [29]GCC manuals. If that fails, the
2620 [30]gcc-help@gcc.gnu.org mailing list might help. Comments on these
2621 web pages and the development of GCC are welcome on our developer
2622 list at [31]gcc@gcc.gnu.org. All of [32]our lists have public
2625 Copyright (C) [33]Free Software Foundation, Inc. Verbatim copying and
2626 distribution of this entire article is permitted in any medium,
2627 provided this notice is preserved.
2629 These pages are [34]maintained by the GCC team. Last modified
2634 1. http://www.multiprecision.org/
2635 2. http://gcc.gnu.org/install/prerequisites.html
2636 3. http://gcc.gnu.org/ml/gcc/2010-01/msg00510.html
2637 4. http://gcc.gnu.org/gcc-4.4/changes.html#obsoleted
2638 5. http://gcc.gnu.org/gcc-4.5/changes.html#x86
2639 6. http://www.multiprecision.org/
2640 7. http://gcc.gnu.org/PR30789
2641 8. http://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html#index-flto-801
2642 9. http://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html#index-fwhopr-802
2643 10. http://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html#index-fwhole-program-800
2644 11. http://gcc.gnu.org/onlinedocs/gcc/Restricted-Pointers.html
2645 12. http://gcc.gnu.org/gcc-4.5/cxx0x_status.html
2646 13. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#757
2647 14. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#176
2648 15. http://gcc.gnu.org/onlinedocs/libstdc++/manual/status.html#status.iso.200x
2649 16. http://gcc.gnu.org/onlinedocs/libstdc++/manual/profile_mode.html
2650 17. http://gcc.gnu.org/onlinedocs/libstdc++/manual/status.html#status.iso.tr24733
2651 18. http://sourceware.org/gdb/wiki/STLSupport
2652 19. http://gcc.gnu.org/gcc-4.4/changes.html
2653 20. http://gcc.gnu.org/onlinedocs/gfortran/Code-Gen-Options.html
2654 21. http://gcc.gnu.org/onlinedocs/gfortran/Mixed-Language-Programming.html
2655 22. http://gcc.gnu.org/onlinedocs/gfortran/GNU-Fortran-Compiler-Directives.html
2656 23. http://gcc.gnu.org/wiki/OOP
2657 24. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.5.1
2658 25. http://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html#index-flto-801
2659 26. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.5.2
2660 27. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.5.3
2661 28. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.5.4
2662 29. http://gcc.gnu.org/onlinedocs/
2663 30. mailto:gcc-help@gcc.gnu.org
2664 31. mailto:gcc@gcc.gnu.org
2665 32. http://gcc.gnu.org/lists.html
2666 33. http://www.fsf.org/
2667 34. http://gcc.gnu.org/about.html
2668 35. http://validator.w3.org/check/referer
2669 ======================================================================
2670 http://gcc.gnu.org/gcc-4.4/index.html
2671 GCC 4.4 Release Series
2675 The [1]GNU project and the GCC developers are pleased to announce the
2676 release of GCC 4.4.7.
2678 This release is a bug-fix release, containing fixes for regressions in
2679 GCC 4.4.6 relative to previous releases of GCC.
2684 March 13, 2012 ([2]changes)
2687 April 16, 2011 ([3]changes)
2690 October 1, 2010 ([4]changes)
2693 April 29, 2010 ([5]changes)
2696 January 21, 2010 ([6]changes)
2699 October 15, 2009 ([7]changes)
2702 July 22, 2009 ([8]changes)
2705 April 21, 2009 ([9]changes)
2707 References and Acknowledgements
2709 GCC used to stand for the GNU C Compiler, but since the compiler
2710 supports several other languages aside from C, it now stands for the
2711 GNU Compiler Collection.
2713 A list of [10]successful builds is updated as new information becomes
2716 The GCC developers would like to thank the numerous people that have
2717 contributed new features, improvements, bug fixes, and other changes as
2718 well as test results to GCC. This [11]amazing group of volunteers is
2719 what makes GCC successful.
2721 For additional information about GCC please refer to the [12]GCC
2722 project web site or contact the [13]GCC development mailing list.
2724 To obtain GCC please use [14]our mirror sites or [15]our SVN server.
2727 For questions related to the use of GCC, please consult these web
2728 pages and the [16]GCC manuals. If that fails, the
2729 [17]gcc-help@gcc.gnu.org mailing list might help. Comments on these
2730 web pages and the development of GCC are welcome on our developer
2731 list at [18]gcc@gcc.gnu.org. All of [19]our lists have public
2734 Copyright (C) [20]Free Software Foundation, Inc. Verbatim copying and
2735 distribution of this entire article is permitted in any medium,
2736 provided this notice is preserved.
2738 These pages are [21]maintained by the GCC team. Last modified
2743 1. http://www.gnu.org/
2744 2. http://gcc.gnu.org/gcc-4.4/changes.html
2745 3. http://gcc.gnu.org/gcc-4.4/changes.html
2746 4. http://gcc.gnu.org/gcc-4.4/changes.html
2747 5. http://gcc.gnu.org/gcc-4.4/changes.html
2748 6. http://gcc.gnu.org/gcc-4.4/changes.html
2749 7. http://gcc.gnu.org/gcc-4.4/changes.html
2750 8. http://gcc.gnu.org/gcc-4.4/changes.html
2751 9. http://gcc.gnu.org/gcc-4.4/changes.html
2752 10. http://gcc.gnu.org/gcc-4.4/buildstat.html
2753 11. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
2754 12. http://gcc.gnu.org/index.html
2755 13. mailto:gcc@gcc.gnu.org
2756 14. http://gcc.gnu.org/mirrors.html
2757 15. http://gcc.gnu.org/svn.html
2758 16. http://gcc.gnu.org/onlinedocs/
2759 17. mailto:gcc-help@gcc.gnu.org
2760 18. mailto:gcc@gcc.gnu.org
2761 19. http://gcc.gnu.org/lists.html
2762 20. http://www.fsf.org/
2763 21. http://gcc.gnu.org/about.html
2764 22. http://validator.w3.org/check/referer
2765 ======================================================================
2766 http://gcc.gnu.org/gcc-4.4/changes.html
2767 GCC 4.4 Release Series
2768 Changes, New Features, and Fixes
2770 The latest release in the 4.4 release series is [1]GCC 4.4.7.
2774 * __builtin_stdarg_start has been completely removed from GCC.
2775 Support for <varargs.h> had been deprecated since GCC 4.0. Use
2776 __builtin_va_start as a replacement.
2777 * Some of the errors issued by the C++ front end that could be
2778 downgraded to warnings in previous releases by using -fpermissive
2779 are now warnings by default. They can be converted into errors by
2780 using -pedantic-errors.
2781 * Use of the cpp assertion extension will now emit a warning when
2782 -Wdeprecated or -pedantic is used. This extension has been
2783 deprecated for many years, but never warned about.
2784 * Packed bit-fields of type char were not properly bit-packed on many
2785 targets prior to GCC 4.4. On these targets, the fix in GCC 4.4
2786 causes an ABI change. For example there is no longer a 4-bit
2787 padding between field a and b in this structure:
2792 } __attribute__ ((packed));
2793 There is a new warning to help identify fields that are affected:
2794 foo.c:5: note: Offset of packed bit-field 'b' has changed in GCC 4.4
2795 The warning can be disabled with -Wno-packed-bitfield-compat.
2796 * On ARM EABI targets, the C++ mangling of the va_list type has been
2797 changed to conform to the current revision of the EABI. This does
2798 not affect the libstdc++ library included with GCC.
2799 * The SCOUNT and POS bits of the MIPS DSP control register are now
2800 treated as global. Previous versions of GCC treated these fields as
2801 call-clobbered instead.
2802 * The MIPS port no longer recognizes the h asm constraint. It was
2803 necessary to remove this constraint in order to avoid generating
2804 unpredictable code sequences.
2805 One of the main uses of the h constraint was to extract the high
2806 part of a multiplication on 64-bit targets. For example:
2807 asm ("dmultu\t%1,%2" : "=h" (result) : "r" (x), "r" (y));
2808 You can now achieve the same effect using 128-bit types:
2809 typedef unsigned int uint128_t __attribute__((mode(TI)));
2810 result = ((uint128_t) x * y) >> 64;
2811 The second sequence is better in many ways. For example, if x and y
2812 are constants, the compiler can perform the multiplication at
2813 compile time. If x and y are not constants, the compiler can
2814 schedule the runtime multiplication better than it can schedule an
2816 * Support for a number of older systems and recently unmaintained or
2817 untested target ports of GCC has been declared obsolete in GCC 4.4.
2818 Unless there is activity to revive them, the next release of GCC
2819 will have their sources permanently removed.
2820 The following ports for individual systems on particular
2821 architectures have been obsoleted:
2822 + Generic a.out on IA32 and m68k (i[34567]86-*-aout*,
2824 + Generic COFF on ARM, H8300, IA32, m68k and SH (arm-*-coff*,
2825 armel-*-coff*, h8300-*-*, i[34567]86-*-coff*, m68k-*-coff*,
2826 sh-*-*). This does not affect other more specific targets
2827 using the COFF object format on those architectures, or the
2828 more specific H8300 and SH targets (h8300-*-rtems*,
2829 h8300-*-elf*, sh-*-elf*, sh-*-symbianelf*, sh-*-linux*,
2830 sh-*-netbsdelf*, sh-*-rtems*, sh-wrs-vxworks).
2831 + 2BSD on PDP-11 (pdp11-*-bsd)
2832 + AIX 4.1 and 4.2 on PowerPC (rs6000-ibm-aix4.[12]*,
2833 powerpc-ibm-aix4.[12]*)
2834 + Tuning support for Itanium1 (Merced) variants. Note that code
2835 tuned for Itanium2 should also run correctly on Itanium1.
2836 * The protoize and unprotoize utilities have been obsoleted and will
2837 be removed in GCC 4.5. These utilities have not been installed by
2838 default since GCC 3.0.
2839 * Support has been removed for all the [2]configurations obsoleted in
2841 * Unknown -Wno-* options are now silently ignored by GCC if no other
2842 diagnostics are issued. If other diagnostics are issued, then GCC
2843 warns about the unknown options.
2844 * More information on porting to GCC 4.4 from previous versions of
2845 GCC can be found in the [3]porting guide for this release.
2847 General Optimizer Improvements
2849 * A new command-line switch -findirect-inlining has been added. When
2850 turned on it allows the inliner to also inline indirect calls that
2851 are discovered to have known targets at compile time thanks to
2853 * A new command-line switch -ftree-switch-conversion has been added.
2854 This new pass turns simple initializations of scalar variables in
2855 switch statements into initializations from a static array, given
2856 that all the values are known at compile time and the ratio between
2857 the new array size and the original switch branches does not exceed
2858 the parameter --param switch-conversion-max-branch-ratio (default
2860 * A new command-line switch -ftree-builtin-call-dce has been added.
2861 This optimization eliminates unnecessary calls to certain builtin
2862 functions when the return value is not used, in cases where the
2863 calls can not be eliminated entirely because the function may set
2864 errno. This optimization is on by default at -O2 and above.
2865 * A new command-line switch -fconserve-stack directs the compiler to
2866 minimize stack usage even if it makes the generated code slower.
2867 This affects inlining decisions.
2868 * When the assembler supports it, the compiler will now emit unwind
2869 information using assembler .cfi directives. This makes it possible
2870 to use such directives in inline assembler code. The new option
2871 -fno-dwarf2-cfi-asm directs the compiler to not use .cfi
2873 * The [4]Graphite branch has been merged. This merge has brought in a
2874 new framework for loop optimizations based on a polyhedral
2875 intermediate representation. These optimizations apply to all the
2876 languages supported by GCC. The following new code transformations
2877 are available in GCC 4.4:
2878 + -floop-interchange performs loop interchange transformations
2879 on loops. Interchanging two nested loops switches the inner
2880 and outer loops. For example, given a loop like:
2883 A(J, I) = A(J, I) * C
2887 loop interchange will transform the loop as if the user had
2891 A(J, I) = A(J, I) * C
2895 which can be beneficial when N is larger than the caches,
2896 because in Fortran, the elements of an array are stored in
2897 memory contiguously by column, and the original loop iterates
2898 over rows, potentially creating at each access a cache miss.
2899 + -floop-strip-mine performs loop strip mining transformations
2900 on loops. Strip mining splits a loop into two nested loops.
2901 The outer loop has strides equal to the strip size and the
2902 inner loop has strides of the original loop within a strip.
2903 For example, given a loop like:
2908 loop strip mining will transform the loop as if the user had
2911 DO I = II, min (II + 3, N)
2916 + -floop-block performs loop blocking transformations on loops.
2917 Blocking strip mines each loop in the loop nest such that the
2918 memory accesses of the element loops fit inside caches. For
2919 example, given a loop like:
2922 A(J, I) = B(I) + C(J)
2926 loop blocking will transform the loop as if the user had
2930 DO I = II, min (II + 63, N)
2931 DO J = JJ, min (JJ + 63, M)
2932 A(J, I) = B(I) + C(J)
2938 which can be beneficial when M is larger than the caches,
2939 because the innermost loop will iterate over a smaller amount
2940 of data that can be kept in the caches.
2941 * A new register allocator has replaced the old one. It is called
2942 integrated register allocator (IRA) because coalescing, register
2943 live range splitting, and hard register preferencing are done
2944 on-the-fly during coloring. It also has better integration with the
2945 reload pass. IRA is a regional register allocator which uses modern
2946 Chaitin-Briggs coloring instead of Chow's priority coloring used in
2947 the old register allocator. More info about IRA internals and
2948 options can be found in the GCC manuals.
2949 * A new instruction scheduler and software pipeliner, based on the
2950 selective scheduling approach, has been added. The new pass
2951 performs instruction unification, register renaming, substitution
2952 through register copies, and speculation during scheduling. The
2953 software pipeliner is able to pipeline non-countable loops. The new
2954 pass is targeted at scheduling-eager in-order platforms. In GCC 4.4
2955 it is available for the Intel Itanium platform working by default
2956 as the second scheduling pass (after register allocation) at the
2957 -O3 optimization level.
2958 * When using -fprofile-generate with a multi-threaded program, the
2959 profile counts may be slightly wrong due to race conditions. The
2960 new -fprofile-correction option directs the compiler to apply
2961 heuristics to smooth out the inconsistencies. By default the
2962 compiler will give an error message when it finds an inconsistent
2964 * The new -fprofile-dir=PATH option permits setting the directory
2965 where profile data files are stored when using -fprofile-generate
2966 and friends, and the directory used when reading profile data files
2967 using -fprofile-use and friends.
2971 * The new -Wframe-larger-than=NUMBER option directs GCC to emit a
2972 warning if any stack frame is larger than NUMBER bytes. This may be
2973 used to help ensure that code fits within a limited amount of stack
2975 * The command-line option -Wlarger-than-N is now written as
2976 -Wlarger-than=N and the old form is deprecated.
2977 * The new -Wno-mudflap option disables warnings about constructs
2978 which can not be instrumented when using -fmudflap.
2980 New Languages and Language specific improvements
2982 * Version 3.0 of the [5]OpenMP specification is now supported for the
2983 C, C++, and Fortran compilers.
2984 * New character data types, per [6]TR 19769: New character types in
2985 C, are now supported for the C compiler in -std=gnu99 mode, as
2986 __CHAR16_TYPE__ and __CHAR32_TYPE__, and for the C++ compiler in
2987 -std=c++0x and -std=gnu++0x modes, as char16_t and char32_t too.
2991 * A new optimize attribute was added to allow programmers to change
2992 the optimization level and particular optimization options for an
2993 individual function. You can also change the optimization options
2994 via the GCC optimize pragma for functions defined after the pragma.
2995 The GCC push_options pragma and the GCC pop_options pragma allow
2996 you temporarily save and restore the options used. The GCC
2997 reset_options pragma restores the options to what was specified on
2999 * Uninitialized warnings do not require enabling optimization
3000 anymore, that is, -Wuninitialized can be used together with -O0.
3001 Nonetheless, the warnings given by -Wuninitialized will probably be
3002 more accurate if optimization is enabled.
3003 * -Wparentheses now warns about expressions such as (!x | y) and (!x
3004 & y). Using explicit parentheses, such as in ((!x) | y), silences
3006 * -Wsequence-point now warns within if, while,do while and for
3007 conditions, and within for begin/end expressions.
3008 * A new option -dU is available to dump definitions of preprocessor
3009 macros that are tested or expanded.
3013 * [7]Improved experimental support for the upcoming ISO C++ standard,
3014 C++0x. Including support for auto, inline namespaces, generalized
3015 initializer lists, defaulted and deleted functions, new character
3016 types, and scoped enums.
3017 * Those errors that may be downgraded to warnings to build legacy
3018 code now mention -fpermissive when -fdiagnostics-show-option is
3020 * -Wconversion now warns if the result of a static_cast to enumeral
3021 type is unspecified because the value is outside the range of the
3023 * -Wuninitialized now warns if a non-static reference or non-static
3024 const member appears in a class without constructors.
3025 * G++ now properly implements value-initialization, so objects with
3026 an initializer of () and an implicitly defined default constructor
3027 will be zero-initialized before the default constructor is called.
3029 Runtime Library (libstdc++)
3031 * [8]Improved experimental support for the upcoming ISO C++ standard,
3033 + Support for <chrono>, <condition_variable>, <cstdatomic>,
3034 <forward_list>, <initializer_list>, <mutex>, <ratio>,
3035 <system_error>, and <thread>.
3036 + unique_ptr, <algorithm> additions, exception propagation, and
3037 support for the new character types in <string> and <limits>.
3038 + Existing facilities now exploit initializer lists, defaulted
3039 and deleted functions, and the newly implemented core C++0x
3041 + Some standard containers are more efficient together with
3042 stateful allocators, i.e., no allocator is constructed on the
3043 fly at element construction time.
3044 * Experimental support for non-standard pointer types in containers.
3045 * The long standing libstdc++/30928 has been fixed for targets
3046 running glibc 2.10 or later.
3047 * As usual, many small and larger bug fixes, in particular quite a
3048 few corner cases in <locale>.
3052 * GNU Fortran now employs libcpp directly instead of using cc1 as an
3053 external preprocessor. The [9]-cpp option was added to allow manual
3054 invocation of the preprocessor without relying on filename
3056 * The [10]-Warray-temporaries option warns about array temporaries
3057 generated by the compiler, as an aid to optimization.
3058 * The [11]-fcheck-array-temporaries option has been added, printing a
3059 notification at run time, when an array temporary had to be created
3060 for an function argument. Contrary to -Warray-temporaries the
3061 warning is only printed if the array is noncontiguous.
3062 * Improved generation of DWARF debugging symbols
3063 * If using an intrinsic not part of the selected standard (via -std=
3064 and -fall-intrinsics) gfortran will now treat it as if this
3065 procedure were declared EXTERNAL and try to link to a user-supplied
3066 procedure. -Wintrinsics-std will warn whenever this happens. The
3067 now-useless option -Wnonstd-intrinsic was removed.
3068 * The flag -falign-commons has been added to control the alignment of
3069 variables in COMMON blocks, which is enabled by default in line
3070 with previous GCC version. Using -fno-align-commons one can force
3071 commons to be contiguous in memory as required by the Fortran
3072 standard, however, this slows down the memory access. The option
3073 -Walign-commons, which is enabled by default, warns when padding
3074 bytes were added for alignment. The proper solution is to sort the
3075 common objects by decreasing storage size, which avoids the
3077 * Fortran 2003 support has been extended:
3078 + Wide characters (ISO 10646, UCS-4, kind=4) and UTF-8 I/O is
3079 now supported (except internal reads from/writes to wide
3080 strings). [12]-fbackslash now supports also \unnnn and
3081 \Unnnnnnnn to enter Unicode characters.
3082 + Asynchronous I/O (implemented as synchronous I/O) and the
3083 decimal=, size=, sign=, pad=, blank=, and delim= specifiers
3084 are now supported in I/O statements.
3085 + Support for Fortran 2003 structure constructors and for array
3086 constructor with typespec has been added.
3087 + Procedure Pointers (but not yet as component in derived types
3088 and as function results) are now supported.
3089 + Abstract types, type extension, and type-bound procedures
3090 (both PROCEDURE and GENERIC but not as operators). Note: As
3091 CLASS/polymorphyic types are not implemented, type-bound
3092 procedures with PASS accept as non-standard extension TYPE
3094 * Fortran 2008 support has been added:
3095 + The -std=f2008 option and support for the file extensions
3096 .f2008 and .F2008 has been added.
3097 + The g0 format descriptor is now supported.
3098 + The Fortran 2008 mathematical intrinsics ASINH, ACOSH, ATANH,
3099 ERF, ERFC, GAMMA, LOG_GAMMA, BESSEL_*, HYPOT, and ERFC_SCALED
3100 are now available (some of them existed as GNU extension
3101 before). Note: The hyperbolic functions are not yet supporting
3102 complex arguments and the three- argument version of BESSEL_*N
3104 + The bit intrinsics LEADZ and TRAILZ have been added.
3110 * The Ada runtime now supports multilibs on many platforms including
3111 x86_64, SPARC and PowerPC. Their build is enabled by default.
3113 New Targets and Target Specific Improvements
3117 * GCC now supports optimizing for the Cortex-A9, Cortex-R4 and
3118 Cortex-R4F processors and has many other improvements to
3119 optimization for ARM processors.
3120 * GCC now supports the VFPv3 variant with 16 double-precision
3121 registers with -mfpu=vfpv3-d16. The option -mfpu=vfp3 has been
3122 renamed to -mfpu=vfpv3.
3123 * GCC now supports the -mfix-cortex-m3-ldrd option to work around an
3124 erratum on Cortex-M3 processors.
3125 * GCC now supports the __sync_* atomic operations for ARM EABI
3127 * The section anchors optimization is now enabled by default when
3129 * GCC now uses a new EABI-compatible profiling interface for EABI
3130 targets. This requires a function __gnu_mcount_nc, which is
3131 provided by GNU libc versions 2.8 and later.
3135 * The -mno-tablejump option has been deprecated because it has the
3136 same effect as the -fno-jump-tables option.
3137 * Added support for these new AVR devices:
3166 * Support for Intel AES built-in functions and code generation is
3167 available via -maes.
3168 * Support for Intel PCLMUL built-in function and code generation is
3169 available via -mpclmul.
3170 * Support for Intel AVX built-in functions and code generation is
3171 available via -mavx.
3172 * Automatically align the stack for local variables with alignment
3174 * GCC can now utilize the SVML library for vectorizing calls to a set
3175 of C99 functions if -mveclibabi=svml is specified and you link to
3176 an SVML ABI compatible library.
3177 * On x86-64, the ABI has been changed in the following cases to
3178 conform to the x86-64 ABI:
3179 + Passing/returning structures with flexible array member:
3185 + Passing/returning structures with complex float member:
3189 __complex__ float f;
3191 + Passing/returning unions with long double member:
3197 Code built with previous versions of GCC that uses any of these is
3198 not compatible with code built with GCC 4.4.0 or later.
3199 * A new target attribute was added to allow programmers to change the
3200 target options like -msse2 or -march=k8 for an individual function.
3201 You can also change the target options via the GCC target pragma
3202 for functions defined after the pragma.
3203 * GCC can now be configured with options --with-arch-32,
3204 --with-arch-64, --with-cpu-32, --with-cpu-64, --with-tune-32 and
3205 --with-tune-64 to control the default optimization separately for
3206 32-bit and 64-bit modes.
3210 * Support for __float128 (TFmode) IEEE quad type and corresponding
3211 TCmode IEEE complex quad type is available via the soft-fp library
3212 on IA-32/IA64 targets. This includes basic arithmetic operations
3213 (addition, subtraction, negation, multiplication and division) on
3214 __float128 real and TCmode complex values, the full set of IEEE
3215 comparisons between __float128 values, conversions to and from
3216 float, double and long double floating point types, as well as
3217 conversions to and from signed or unsigned integer, signed or
3218 unsigned long integer and signed or unsigned quad (TImode, IA64
3219 only) integer types. Additionally, all operations generate the full
3220 set of IEEE exceptions and support the full set of IEEE rounding
3225 * GCC now supports instruction scheduling for ColdFire V1, V3 and V4
3226 processors. (Scheduling support for ColdFire V2 processors was
3228 * GCC now supports the -mxgot option to support programs requiring
3229 many GOT entries on ColdFire.
3230 * The m68k-*-linux-gnu target now builds multilibs by default.
3234 * MIPS Technologies have extended the original MIPS SVR4 ABI to
3235 include support for procedure linkage tables (PLTs) and copy
3236 relocations. These extensions allow GNU/Linux executables to use a
3237 significantly more efficient code model than the one defined by the
3239 GCC support for this code model is available via a new command-line
3240 option, -mplt. There is also a new configure-time option,
3241 --with-mips-plt, to make -mplt the default.
3242 The new code model requires support from the assembler, the linker,
3243 and the runtime C library. This support is available in binutils
3245 * GCC can now generate MIPS16 code for 32-bit GNU/Linux executables
3246 and 32-bit GNU/Linux shared libraries. This feature requires GNU
3247 binutils 2.19 or above.
3248 * Support for RMI's XLR processor is now available through the
3249 -march=xlr and -mtune=xlr options.
3250 * 64-bit targets can now perform 128-bit multiplications inline,
3251 instead of relying on a libgcc function.
3252 * Native GNU/Linux toolchains now support -march=native and
3253 -mtune=native, which select the host processor.
3254 * GCC now supports the R10K, R12K, R14K and R16K processors. The
3255 canonical -march= and -mtune= names for these processors are
3256 r10000, r12000, r14000 and r16000 respectively.
3257 * GCC can now work around the side effects of speculative execution
3258 on R10K processors. Please see the documentation of the
3259 -mr10k-cache-barrier option for details.
3260 * Support for the MIPS64 Release 2 instruction set has been added.
3261 The option -march=mips64r2 enables generation of these
3263 * GCC now supports Cavium Networks' Octeon processor. This support is
3264 available through the -march=octeon and -mtune=octeon options.
3265 * GCC now supports STMicroelectronics' Loongson 2E/2F processors. The
3266 canonical -march= and -mtune= names for these processors are
3267 loongson2e and loongson2f.
3271 Picochip is a 16-bit processor. A typical picoChip contains over 250
3272 small cores, each with small amounts of memory. There are three
3273 processor variants (STAN, MEM and CTRL) with different instruction sets
3274 and memory configurations and they can be chosen using the -mae option.
3276 This port is intended to be a "C" only port.
3278 Power Architecture and PowerPC
3280 * GCC now supports the e300c2, e300c3 and e500mc processors.
3281 * GCC now supports Xilinx processors with a single-precision FPU.
3282 * Decimal floating point is now supported for e500 processors.
3284 S/390, zSeries and System z9/z10
3286 * Support for the IBM System z10 EC/BC processor has been added. When
3287 using the -march=z10 option, the compiler will generate code making
3288 use of instructions provided by the General-Instruction-Extension
3289 Facility and the Execute-Extension Facility.
3293 * GCC now supports the thread-local storage mechanism used on
3298 * GCC now supports thread-local storage (TLS) for Xtensa processor
3299 configurations that include the Thread Pointer option. TLS also
3300 requires support from the assembler and linker; this support is
3301 provided in the GNU binutils beginning with version 2.19.
3303 Documentation improvements
3305 Other significant improvements
3309 This is the [13]list of problem reports (PRs) from GCC's bug tracking
3310 system that are known to be fixed in the 4.4.1 release. This list might
3311 not be complete (that is, it is possible that some PRs that have been
3312 fixed are not listed here).
3316 This is the [14]list of problem reports (PRs) from GCC's bug tracking
3317 system that are known to be fixed in the 4.4.2 release. This list might
3318 not be complete (that is, it is possible that some PRs that have been
3319 fixed are not listed here).
3323 This is the [15]list of problem reports (PRs) from GCC's bug tracking
3324 system that are known to be fixed in the 4.4.3 release. This list might
3325 not be complete (that is, it is possible that some PRs that have been
3326 fixed are not listed here).
3330 This is the [16]list of problem reports (PRs) from GCC's bug tracking
3331 system that are known to be fixed in the 4.4.4 release. This list might
3332 not be complete (that is, it is possible that some PRs that have been
3333 fixed are not listed here).
3337 This is the [17]list of problem reports (PRs) from GCC's bug tracking
3338 system that are known to be fixed in the 4.4.5 release. This list might
3339 not be complete (that is, it is possible that some PRs that have been
3340 fixed are not listed here).
3344 This is the [18]list of problem reports (PRs) from GCC's bug tracking
3345 system that are known to be fixed in the 4.4.6 release. This list might
3346 not be complete (that is, it is possible that some PRs that have been
3347 fixed are not listed here).
3351 This is the [19]list of problem reports (PRs) from GCC's bug tracking
3352 system that are known to be fixed in the 4.4.7 release. This list might
3353 not be complete (that is, it is possible that some PRs that have been
3354 fixed are not listed here).
3357 For questions related to the use of GCC, please consult these web
3358 pages and the [20]GCC manuals. If that fails, the
3359 [21]gcc-help@gcc.gnu.org mailing list might help. Comments on these
3360 web pages and the development of GCC are welcome on our developer
3361 list at [22]gcc@gcc.gnu.org. All of [23]our lists have public
3364 Copyright (C) [24]Free Software Foundation, Inc. Verbatim copying and
3365 distribution of this entire article is permitted in any medium,
3366 provided this notice is preserved.
3368 These pages are [25]maintained by the GCC team. Last modified
3373 1. http://gcc.gnu.org/gcc-4.4/changes.html#4.4.7
3374 2. http://gcc.gnu.org/gcc-4.3/changes.html#obsoleted
3375 3. http://gcc.gnu.org/gcc-4.4/porting_to.html
3376 4. http://gcc.gnu.org/wiki/Graphite
3377 5. http://openmp.org/wp/openmp-specifications/
3378 6. http://www.open-std.org/jtc1/sc22/wg14/www/docs/n1040.pdf
3379 7. http://gcc.gnu.org/gcc-4.4/cxx0x_status.html
3380 8. http://gcc.gnu.org/onlinedocs/libstdc++/manual/status.html#id476343
3381 9. http://gcc.gnu.org/onlinedocs/gfortran/Preprocessing-Options.html
3382 10. http://gcc.gnu.org/onlinedocs/gfortran/Error-and-Warning-Options.html#index-g_t_0040code_007bWarray-temporaries_007d-125
3383 11. http://gcc.gnu.org/onlinedocs/gfortran/Code-Gen-Options.html#index-g_t_0040code_007bfcheck-array-temporaries_007d-221
3384 12. http://gcc.gnu.org/onlinedocs/gfortran/Fortran-Dialect-Options.html#index-g_t_0040code_007bbackslash_007d-34
3385 13. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.1
3386 14. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.2
3387 15. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.3
3388 16. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.4
3389 17. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.5
3390 18. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.6
3391 19. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.7
3392 20. http://gcc.gnu.org/onlinedocs/
3393 21. mailto:gcc-help@gcc.gnu.org
3394 22. mailto:gcc@gcc.gnu.org
3395 23. http://gcc.gnu.org/lists.html
3396 24. http://www.fsf.org/
3397 25. http://gcc.gnu.org/about.html
3398 26. http://validator.w3.org/check/referer
3399 ======================================================================
3400 http://gcc.gnu.org/gcc-4.3/index.html
3401 GCC 4.3 Release Series
3405 The [1]GNU project and the GCC developers are pleased to announce the
3406 release of GCC 4.3.6.
3408 This release is a bug-fix release, containing fixes for regressions in
3409 GCC 4.3.5 relative to previous releases of GCC.
3414 Jun 27, 2011 ([2]changes)
3417 May 22, 2010 ([3]changes)
3420 August 4, 2009 ([4]changes)
3423 January 24, 2009 ([5]changes)
3426 August 27, 2008 ([6]changes)
3429 June 6, 2008 ([7]changes)
3432 March 5, 2008 ([8]changes)
3434 References and Acknowledgements
3436 GCC used to stand for the GNU C Compiler, but since the compiler
3437 supports several other languages aside from C, it now stands for the
3438 GNU Compiler Collection.
3440 A list of [9]successful builds is updated as new information becomes
3443 The GCC developers would like to thank the numerous people that have
3444 contributed new features, improvements, bug fixes, and other changes as
3445 well as test results to GCC. This [10]amazing group of volunteers is
3446 what makes GCC successful.
3448 For additional information about GCC please refer to the [11]GCC
3449 project web site or contact the [12]GCC development mailing list.
3451 To obtain GCC please use [13]our mirror sites or [14]our SVN server.
3454 For questions related to the use of GCC, please consult these web
3455 pages and the [15]GCC manuals. If that fails, the
3456 [16]gcc-help@gcc.gnu.org mailing list might help. Comments on these
3457 web pages and the development of GCC are welcome on our developer
3458 list at [17]gcc@gcc.gnu.org. All of [18]our lists have public
3461 Copyright (C) [19]Free Software Foundation, Inc. Verbatim copying and
3462 distribution of this entire article is permitted in any medium,
3463 provided this notice is preserved.
3465 These pages are [20]maintained by the GCC team. Last modified
3470 1. http://www.gnu.org/
3471 2. http://gcc.gnu.org/gcc-4.3/changes.html
3472 3. http://gcc.gnu.org/gcc-4.3/changes.html
3473 4. http://gcc.gnu.org/gcc-4.3/changes.html
3474 5. http://gcc.gnu.org/gcc-4.3/changes.html
3475 6. http://gcc.gnu.org/gcc-4.3/changes.html
3476 7. http://gcc.gnu.org/gcc-4.3/changes.html
3477 8. http://gcc.gnu.org/gcc-4.3/changes.html
3478 9. http://gcc.gnu.org/gcc-4.3/buildstat.html
3479 10. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
3480 11. http://gcc.gnu.org/index.html
3481 12. mailto:gcc@gcc.gnu.org
3482 13. http://gcc.gnu.org/mirrors.html
3483 14. http://gcc.gnu.org/svn.html
3484 15. http://gcc.gnu.org/onlinedocs/
3485 16. mailto:gcc-help@gcc.gnu.org
3486 17. mailto:gcc@gcc.gnu.org
3487 18. http://gcc.gnu.org/lists.html
3488 19. http://www.fsf.org/
3489 20. http://gcc.gnu.org/about.html
3490 21. http://validator.w3.org/check/referer
3491 ======================================================================
3492 http://gcc.gnu.org/gcc-4.3/changes.html
3493 GCC 4.3 Release Series
3494 Changes, New Features, and Fixes
3496 The latest release in the 4.3 release series is [1]GCC 4.3.5.
3500 * GCC requires the [2]GMP and [3]MPFR libraries for building all the
3501 various front-end languages it supports. See the [4]prerequisites
3502 page for version requirements.
3503 * ColdFire targets now treat long double as having the same format as
3504 double. In earlier versions of GCC, they used the 68881 long double
3506 * The m68k-uclinux target now uses the same calling conventions as
3507 m68k-linux-gnu. You can select the original calling conventions by
3508 configuring for m68k-uclinuxoldabi instead. Note that
3509 m68k-uclinuxoldabi also retains the original 80-bit long double on
3511 * The -fforce-mem option has been removed because it has had no
3512 effect in the last few GCC releases.
3513 * The i386 -msvr3-shlib option has been removed since it is no longer
3515 * Fastcall for i386 has been changed not to pass aggregate arguments
3516 in registers, following Microsoft compilers.
3517 * Support for the AOF assembler has been removed from the ARM back
3518 end; this affects only the targets arm-semi-aof and armel-semi-aof,
3519 which are no longer recognized. We removed these targets without a
3520 deprecation period because we discovered that they have been
3521 unusable since GCC 4.0.0.
3522 * Support for the TMS320C3x/C4x processor (targets c4x-* and tic4x-*)
3523 has been removed. This support had been deprecated since GCC 4.0.0.
3524 * Support for a number of older systems and recently unmaintained or
3525 untested target ports of GCC has been declared obsolete in GCC 4.3.
3526 Unless there is activity to revive them, the next release of GCC
3527 will have their sources permanently removed.
3528 All GCC ports for the following processor architectures have been
3531 The following aliases for processor architectures have been
3532 declared obsolete. Users should use the indicated generic target
3533 names instead, with compile-time options such as -mcpu or
3534 configure-time options such as --with-cpu to control the
3535 configuration more precisely.
3536 + strongarm*-*-*, ep9312*-*-*, xscale*-*-* (use arm*-*-*
3538 + parisc*-*-* (use hppa*-*-* instead).
3539 + m680[012]0-*-* (use m68k-*-* instead).
3540 All GCC ports for the following operating systems have been
3544 + GNU/Linux using the a.out object format (*-*-linux*aout*)
3545 + GNU/Linux using version 1 of the GNU C Library
3547 + Solaris versions before Solaris 7 (*-*-solaris2.[0-6],
3548 *-*-solaris2.[0-6].*)
3549 + Miscellaneous System V (*-*-sysv*)
3550 + WindISS (*-*-windiss*)
3551 Also, those for some individual systems on particular architectures
3552 have been obsoleted:
3553 + UNICOS/mk on DEC Alpha (alpha*-*-unicosmk*)
3554 + CRIS with a.out object format (cris-*-aout)
3555 + BSD 4.3 on PA-RISC (hppa1.1-*-bsd*)
3556 + OSF/1 on PA-RISC (hppa1.1-*-osf*)
3557 + PRO on PA-RISC (hppa1.1-*-pro*)
3558 + Sequent PTX on IA32 (i[34567]86-sequent-ptx4*,
3559 i[34567]86-sequent-sysv4*)
3560 + SCO Open Server 5 on IA32 (i[34567]86-*-sco3.2v5*)
3561 + UWIN on IA32 (i[34567]86-*-uwin*) (support for UWIN as a host
3562 was previously [5]removed in 2001, leaving only the support
3563 for UWIN as a target now being deprecated)
3564 + ChorusOS on PowerPC (powerpc-*-chorusos*)
3565 + All VAX configurations apart from NetBSD and OpenBSD
3566 (vax-*-bsd*, vax-*-sysv*, vax-*-ultrix*)
3567 * The [6]-Wconversion option has been modified. Its purpose now is to
3568 warn for implicit conversions that may alter a value. This new
3569 behavior is available for both C and C++. Warnings about
3570 conversions between signed and unsigned integers can be disabled by
3571 using -Wno-sign-conversion. In C++, they are disabled by default
3572 unless -Wsign-conversion is explicitly requested. The old behavior
3573 of -Wconversion, that is, warn for prototypes causing a type
3574 conversion that is different from what would happen to the same
3575 argument in the absence of a prototype, has been moved to a new
3576 option -Wtraditional-conversion, which is only available for C.
3577 * The -m386, -m486, -mpentium and -mpentiumpro tuning options have
3578 been removed because they were deprecated for more than 3 GCC major
3579 releases. Use -mtune=i386, -mtune=i486, -mtune=pentium or
3580 -mtune=pentiumpro as a replacement.
3581 * The -funsafe-math-optimizations option now automatically turns on
3582 -fno-trapping-math in addition to -fno-signed-zeros, as it enables
3583 reassociation and thus may introduce or remove traps.
3584 * The -ftree-vectorize option is now on by default under -O3. In
3585 order to generate code for a SIMD extension, it has to be enabled
3586 as well: use -maltivec for PowerPC platforms and -msse/-msse2 for
3588 * More information on porting to GCC 4.3 from previous versions of
3589 GCC can be found in the [7]porting guide for this release.
3591 General Optimizer Improvements
3593 * The GCC middle-end has been integrated with the [8]MPFR library.
3594 This allows GCC to evaluate and replace at compile-time calls to
3595 built-in math functions having constant arguments with their
3596 mathematically equivalent results. In making use of [9]MPFR, GCC
3597 can generate correct results regardless of the math library
3598 implementation or floating point precision of the host platform.
3599 This also allows GCC to generate identical results regardless of
3600 whether one compiles in native or cross-compile configurations to a
3601 particular target. The following built-in functions take advantage
3602 of this new capability: acos, acosh, asin, asinh, atan2, atan,
3603 atanh, cbrt, cos, cosh, drem, erf, erfc, exp10, exp2, exp, expm1,
3604 fdim, fma, fmax, fmin, gamma_r, hypot, j0, j1, jn, lgamma_r, log10,
3605 log1p, log2, log, pow10, pow, remainder, remquo, sin, sincos, sinh,
3606 tan, tanh, tgamma, y0, y1 and yn. The float and long double
3607 variants of these functions (e.g. sinf and sinl) are also handled.
3608 The sqrt and cabs functions with constant arguments were already
3609 optimized in prior GCC releases. Now they also use [10]MPFR.
3610 * A new forward propagation pass on RTL was added. The new pass
3611 replaces several slower transformations, resulting in compile-time
3612 improvements as well as better code generation in some cases.
3613 * A new command-line switch -frecord-gcc-switches has been added to
3614 GCC, although it is only enabled for some targets. The switch
3615 causes the command line that was used to invoke the compiler to be
3616 recorded into the object file that is being created. The exact
3617 format of this recording is target and binary file format
3618 dependent, but it usually takes the form of a note section
3619 containing ASCII text. The switch is related to the -fverbose-asm
3620 switch, but that one only records the information in the assembler
3621 output file as comments, so the information never reaches the
3623 * The inliner heuristic is now aware of stack frame consumption. New
3624 command-line parameters --param large-stack-frame and --param
3625 large-stack-frame-growth can be used to limit stack frame size
3626 growth caused by inlining.
3627 * During feedback directed optimizations, the expected block size the
3628 memcpy, memset and bzero functions operate on is discovered and for
3629 cases of commonly used small sizes, specialized inline code is
3631 * __builtin_expect no longer requires its argument to be a compile
3633 * Interprocedural optimization was reorganized to work on functions
3634 in SSA form. This enables more precise and cheaper dataflow
3635 analysis and makes writing interprocedural optimizations easier.
3636 The following improvements have been implemented on top of this
3638 + Pre-inline optimization: Selected local optimization passes
3639 are run before the inliner (and other interprocedural passes)
3640 are executed. This significantly improves the accuracy of code
3641 growth estimates used by the inliner and reduces the overall
3642 memory footprint for large compilation units.
3643 + Early inlining (a simple bottom-up inliner pass inlining only
3644 functions whose body is smaller than the expected call
3645 overhead) is now executed with the early optimization passes,
3646 thus inlining already optimized function bodies into an
3647 unoptimized function that is subsequently optimized by early
3648 optimizers. This enables the compiler to quickly eliminate
3649 abstraction penalty in C++ programs.
3650 + Interprocedural constant propagation now operate on SSA form
3651 increasing accuracy of the analysis.
3652 * A new internal representation for GIMPLE statements has been
3653 contributed, resulting in compile-time memory savings.
3654 * The vectorizer was enhanced to support vectorization of outer
3655 loops, intra-iteration parallelism (loop-aware SLP), vectorization
3656 of strided accesses and loops with multiple data-types. Run-time
3657 dependency testing using loop versioning was added. The cost model,
3658 turned on by -fvect-cost-model, was developed.
3660 New Languages and Language specific improvements
3662 * We have added new command-line options
3663 -finstrument-functions-exclude-function-list and
3664 -finstrument-functions-exclude-file-list. They provide more control
3665 over which functions are annotated by the -finstrument-functions
3670 * Implicit conversions between generic vector types are now only
3671 permitted when the two vectors in question have the same number of
3672 elements and compatible element types. (Note that the restriction
3673 involves compatible element types, not implicitly-convertible
3674 element types: thus, a vector type with element type int may not be
3675 implicitly converted to a vector type with element type unsigned
3676 int.) This restriction, which is in line with specifications for
3677 SIMD architectures such as AltiVec, may be relaxed using the flag
3678 -flax-vector-conversions. This flag is intended only as a
3679 compatibility measure and should not be used for new code.
3680 * -Warray-bounds has been added and is now enabled by default for
3681 -Wall . It produces warnings for array subscripts that can be
3682 determined at compile time to be always out of bounds.
3683 -Wno-array-bounds will disable the warning.
3684 * The constructor and destructor function attributes now accept
3685 optional priority arguments which control the order in which the
3686 constructor and destructor functions are run.
3687 * New [11]command-line options -Wtype-limits,
3688 -Wold-style-declaration, -Wmissing-parameter-type, -Wempty-body,
3689 -Wclobbered and -Wignored-qualifiers have been added for finer
3690 control of the diverse warnings enabled by -Wextra.
3691 * A new function attribute alloc_size has been added to mark up
3692 malloc style functions. For constant sized allocations this can be
3693 used to find out the size of the returned pointer using the
3694 __builtin_object_size() function for buffer overflow checking and
3695 similar. This supplements the already built-in malloc and calloc
3696 constant size handling.
3697 * Integer constants written in binary are now supported as a GCC
3698 extension. They consist of a prefix 0b or 0B, followed by a
3699 sequence of 0 and 1 digits.
3700 * A new predefined macro __COUNTER__ has been added. It expands to
3701 sequential integral values starting from 0. In conjunction with the
3702 ## operator, this provides a convenient means to generate unique
3704 * A new command-line option -fdirectives-only has been added. It
3705 enables a special preprocessing mode which improves the performance
3706 of applications like distcc and ccache.
3707 * Fixed-point data types and operators have been added. They are
3708 based on Chapter 4 of the Embedded-C specification (n1169.pdf).
3709 Currently, only MIPS targets are supported.
3710 * Decimal floating-point arithmetic based on draft ISO/IEC TR 24732,
3711 N1241, is now supported as a GCC extension to C for targets
3712 i[34567]86-*-linux-gnu, powerpc*-*-linux-gnu, s390*-ibm-linux-gnu,
3713 and x86_64-*-linux-gnu. The feature introduces new data types
3714 _Decimal32, _Decimal64, and _Decimal128 with constant suffixes DF,
3719 * [12]Experimental support for the upcoming ISO C++ standard, C++0x.
3720 * -Wc++0x-compat has been added and is now enabled by default for
3721 -Wall. It produces warnings for constructs whose meaning differs
3722 between ISO C++ 1998 and C++0x.
3723 * The -Wparentheses option now works for C++ as it does for C. It
3724 warns if parentheses are omitted when operators with confusing
3725 precedence are nested. It also warns about ambiguous else
3726 statements. Since -Wparentheses is enabled by -Wall, this may cause
3727 additional warnings with existing C++ code which uses -Wall. These
3728 new warnings may be disabled by using -Wall -Wno-parentheses.
3729 * The -Wmissing-declarations now works for C++ as it does for C.
3730 * The -fvisibility-ms-compat flag was added, to make it easier to
3731 port larger projects using shared libraries from Microsoft's Visual
3732 Studio to ELF and Mach-O systems.
3733 * C++ attribute handling has been overhauled for template arguments
3734 (ie dependent types). In particular, __attribute__((aligned(T)));
3735 works for C++ types.
3737 Runtime Library (libstdc++)
3739 * [13]Experimental support for the upcoming ISO C++ standard, C++0x.
3740 * Support for TR1 mathematical special functions and regular
3741 expressions. ([14]Implementation status of TR1)
3742 * Default what implementations give more elaborate exception strings
3743 for bad_cast, bad_typeid, bad_exception, and bad_alloc.
3744 * Header dependencies have been streamlined, reducing unnecessary
3745 includes and pre-processed bloat.
3746 * Variadic template implementations of items in <tuple> and
3748 * An experimental [15]parallel mode has been added. This is a
3749 parallel implementation of many C++ Standard library algorithms,
3750 like std::accumulate, std::for_each, std::transform, or std::sort,
3751 to give but four examples. These algorithms can be substituted for
3752 the normal (sequential) libstdc++ algorithms on a piecemeal basis,
3753 or all existing algorithms can be transformed via the
3754 -D_GLIBCXX_PARALLEL macro.
3755 * Debug mode versions of classes in <unordered_set> and
3757 * Formal deprecation of <ext/hash_set> and <ext/hash_map>, which are
3758 now <backward/hash_set> and <backward/hash_map>. This code:
3759 #include <ext/hash_set>
3760 __gnu_cxx::hash_set<int> s;
3762 Can be transformed (in order of preference) to:
3763 #include <tr1/unordered_set>
3764 std::tr1::unordered_set<int> s;
3767 #include <backward/hash_set>
3768 __gnu_cxx::hash_set<int> s;
3770 Similar transformations apply to __gnu_cxx::hash_map,
3771 __gnu_cxx::hash_multimap, __gnu_cxx::hash_set,
3772 __gnu_cxx::hash_multiset.
3776 * Due to the fact that the [16]GMP and [17]MPFR libraries are
3777 required for all languages, Fortran is no longer special in this
3778 regard and is available by default.
3779 * The [18]-fexternal-blas option has been added, which generates
3780 calls to BLAS routines for intrinsic matrix operations such as
3781 matmul rather than using the built-in algorithms.
3782 * Support to give a backtrace (compiler flag -fbacktrace or
3783 environment variable GFORTRAN_ERROR_BACKTRACE; on glibc systems
3784 only) or a core dump (-fdump-core, GFORTRAN_ERROR_DUMPCORE) when a
3785 run-time error occured.
3786 * GNU Fortran now defines __GFORTRAN__ when it runs the C
3788 * The [19]-finit-local-zero, -finit-real, -finit-integer,
3789 -finit-character, and -finit-logical options have been added, which
3790 can be used to initialize local variables.
3791 * The intrinsic procedures [20]GAMMA and [21]LGAMMA have been added,
3792 which calculate the Gamma function and its logarithm. Use EXTERNAL
3793 gamma if you want to use your own gamma function.
3794 * GNU Fortran now regards the backslash character as literal (as
3795 required by the Fortran 2003 standard); using [22]-fbackslash GNU
3796 Fortran interprets backslashes as C-style escape characters.
3797 * The [23]interpretation of binary, octal and hexadecimal (BOZ)
3798 literal constants has been changed. Before they were always
3799 interpreted as integer; now they are bit-wise transferred as
3800 argument of INT, REAL, DBLE and CMPLX as required by the Fortran
3801 2003 standard, and for real and complex variables in DATA
3802 statements or when directly assigned to real and complex variables.
3803 Everywhere else and especially in expressions they are still
3804 regarded as integer constants.
3805 * Fortran 2003 support has been extended:
3806 + Intrinsic statements IMPORT, PROTECTED, VALUE and VOLATILE
3808 + Intrinsic module ISO_ENV_FORTRAN
3809 + Interoperability with C (ISO C Bindings)
3810 + ABSTRACT INTERFACES and PROCEDURE statements (without POINTER
3816 * GCJ now uses the Eclipse Java compiler for its Java parsing needs.
3817 This enables the use of all 1.5 language features, and fixes most
3818 existing front end bugs.
3819 * libgcj now supports all 1.5 language features which require runtime
3820 support: foreach, enum, annotations, generics, and auto-boxing.
3821 * We've made many changes to the tools shipped with gcj.
3822 + The old jv-scan tool has been removed. This tool never really
3823 worked properly. There is no replacement.
3824 + gcjh has been rewritten. Some of its more obscure options no
3825 longer work, but are still recognized in an attempt at
3826 compatibility. gjavah is a new program with similar
3827 functionality but different command-line options.
3828 + grmic and grmiregistry have been rewritten. grmid has been
3830 + gjar replaces the old fastjar.
3831 + gjarsigner (used for signing jars), gkeytool (used for key
3832 management), gorbd (for CORBA), gserialver (computes
3833 serialization UIDs), and gtnameserv (also for CORBA) are now
3835 * The ability to dump the contents of the java run time heap to a
3836 file for off-line analysis has been added. The heap dumps may be
3837 analyzed with the new gc-analyze tool. They may be generated on
3838 out-of-memory conditions or on demand and are controlled by the new
3839 run time class gnu.gcj.util.GCInfo.
3840 * java.util.TimeZone can now read files from /usr/share/zoneinfo to
3841 provide correct, updated, timezone information. This means that
3842 packagers no longer have to update libgcj when a time zone change
3845 New Targets and Target Specific Improvements
3849 * Tuning for Intel Core 2 processors is available via -mtune=core2
3851 * Tuning for AMD Geode processors is available via -mtune=geode and
3853 * Code generation of block move (memcpy) and block set (memset) was
3854 rewritten. GCC can now pick the best algorithm (loop, unrolled
3855 loop, instruction with rep prefix or a library call) based on the
3856 size of the block being copied and the CPU being optimized for. A
3857 new option -minline-stringops-dynamically has been added. With this
3858 option string operations of unknown size are expanded such that
3859 small blocks are copied by in-line code, while for large blocks a
3860 library call is used. This results in faster code than
3861 -minline-all-stringops when the library implementation is capable
3862 of using cache hierarchy hints. The heuristic choosing the
3863 particular algorithm can be overwritten via -mstringop-strategy.
3864 Newly also memset of values different from 0 is inlined.
3865 * GCC no longer places the cld instruction before string operations.
3866 Both i386 and x86-64 ABI documents mandate the direction flag to be
3867 clear at the entry of a function. It is now invalid to set the flag
3868 in asm statement without reseting it afterward.
3869 * Support for SSSE3 built-in functions and code generation are
3870 available via -mssse3.
3871 * Support for SSE4.1 built-in functions and code generation are
3872 available via -msse4.1.
3873 * Support for SSE4.2 built-in functions and code generation are
3874 available via -msse4.2.
3875 * Both SSE4.1 and SSE4.2 support can be enabled via -msse4.
3876 * A new set of options -mpc32, -mpc64 and -mpc80 have been added to
3877 allow explicit control of x87 floating point precision.
3878 * Support for __float128 (TFmode) IEEE quad type and corresponding
3879 TCmode IEEE complex quad type is available via the soft-fp library
3880 on x86_64 targets. This includes basic arithmetic operations
3881 (addition, subtraction, negation, multiplication and division) on
3882 __float128 real and TCmode complex values, the full set of IEEE
3883 comparisons between __float128 values, conversions to and from
3884 float, double and long double floating point types, as well as
3885 conversions to and from signed or unsigned integer, signed or
3886 unsigned long integer and signed or unsigned quad (TImode) integer
3887 types. Additionally, all operations generate the full set of IEEE
3888 exceptions and support the full set of IEEE rounding modes.
3889 * GCC can now utilize the ACML library for vectorizing calls to a set
3890 of C99 functions on x86_64 if -mveclibabi=acml is specified and you
3891 link to an ACML ABI compatible library.
3895 * Compiler and Library support for Thumb-2 and the ARMv7 architecture
3902 * Compiler and Library support for the CRIS v32 architecture, as
3903 found in Axis Communications ETRAX FS and ARTPEC-3 chips, has been
3906 Configuration changes
3908 * The cris-*-elf target now includes support for CRIS v32, including
3909 libraries, through the -march=v32 option.
3910 * A new crisv32-*-elf target defaults to generate code for CRIS v32.
3911 * A new crisv32-*-linux* target defaults to generate code for CRIS
3913 * The cris-*-aout target has been obsoleted.
3915 Improved support for built-in functions
3917 * GCC can now use the lz and swapwbr instructions to implement the
3918 __builtin_clz, __builtin_ctz and __builtin_ffs family of functions.
3919 * __builtin_bswap32 is now implemented using the swapwb instruction,
3926 * Support for several new ColdFire processors has been added. You can
3927 generate code for them using the new -mcpu option.
3928 * All targets now support ColdFire processors.
3929 * m68k-uclinux targets have improved support for C++ constructors and
3930 destructors, and for shared libraries.
3931 * It is now possible to set breakpoints on the first or last line of
3932 a function, even if there are no statements on that line.
3936 * Support for sibling calls has been added.
3937 * More use is now made of the ColdFire mov3q instruction.
3938 * __builtin_clz is now implemented using the ff1 ColdFire
3939 instruction, when available.
3940 * GCC now honors the -m68010 option. 68010 code now uses clr rather
3941 than move to zero volatile memory.
3942 * 68020 targets and above can now use symbol(index.size*scale)
3943 addresses for indexed array accesses. Earlier compilers would
3944 always load the symbol into a base register first.
3946 Configuration changes
3948 * All m68k and ColdFire targets now allow the default processor to be
3949 set at configure time using --with-cpu.
3950 * A --with-arch configuration option has been added. This option
3951 allows you to restrict a target to ColdFire or non-ColdFire
3956 * An __mcfv*__ macro is now defined for all ColdFire targets.
3957 (Earlier versions of GCC only defined __mcfv4e__.)
3958 * __mcf_cpu_*, __mcf_family_* and __mcffpu__ macros have been added.
3959 * All targets now define __mc68010 and __mc68010__ when generating
3962 Command-line changes
3964 * New command-line options -march, -mcpu, -mtune and -mhard-float
3965 have been added. These options apply to both m68k and ColdFire
3967 * -mno-short, -mno-bitfield and -mno-rtd are now accepted as negative
3968 versions of -mshort, etc.
3969 * -fforce-addr has been removed. It is now ignored by the compiler.
3973 * ColdFire targets now try to maintain a 4-byte-aligned stack where
3975 * m68k-uclinux targets now try to avoid situations that lead to the
3976 load-time error: BINFMT_FLAT: reloc outside program.
3980 Changes to existing configurations
3982 * libffi and libjava now support all three GNU/Linux ABIs: o32, n32
3983 and n64. Every GNU/Linux configuration now builds these libraries
3985 * GNU/Linux configurations now generate -mno-shared code unless
3986 overridden by -fpic, -fPIC, -fpie or -fPIE.
3987 * mipsisa32*-linux-gnu configurations now generate hard-float code by
3988 default, just like other mipsisa32* and mips*-linux-gnu
3989 configurations. You can build a soft-float version of any
3990 mips*-linux-gnu configuration by passing --with-float=soft to
3992 * mips-wrs-vxworks now supports run-time processes (RTPs).
3994 Changes to existing command-line options
3996 * The -march and -mtune options no longer accept 24k as a processor
3997 name. Please use 24kc, 24kf2_1 or 24kf1_1 instead.
3998 * The -march and -mtune options now accept 24kf2_1, 24kef2_1 and
3999 34kf2_1 as synonyms for 24kf, 24kef and 34kf respectively. The
4000 options also accept 24kf1_1, 24kef1_1 and 34kf1_1 as synonyms for
4001 24kx, 24kex and 34kx.
4005 GCC now supports the following configurations:
4006 * mipsisa32r2*-linux-gnu*, which generates MIPS32 revision 2 code by
4007 default. Earlier releases also recognized this configuration, but
4008 they treated it in the same way as mipsisa32*-linux-gnu*. Note that
4009 you can customize any mips*-linux-gnu* configuration to a
4010 particular ISA or processor by passing an appropriate --with-arch
4011 option to configure.
4012 * mipsisa*-sde-elf*, which provides compatibility with MIPS
4013 Technologies' SDE toolchains. The configuration uses the SDE
4014 libraries by default, but you can use it like other newlib-based
4015 ELF configurations by passing --with-newlib to configure. It is the
4016 only configuration besides mips64vr*-elf* to build MIPS16 as well
4017 as non-MIPS16 libraries.
4018 * mipsisa*-elfoabi*, which is similar to the general mipsisa*-elf*
4019 configuration, but uses the o32 and o64 ABIs instead of the 32-bit
4020 and 64-bit forms of the EABI.
4022 New processors and application-specific extensions
4024 * Support for the SmartMIPS ASE is available through the new
4026 * Support for revision 2 of the DSP ASE is available through the new
4027 -mdspr2 option. A new preprocessor macro called __mips_dsp_rev
4028 indicates the revision of the ASE in use.
4029 * Support for the 4KS and 74K families of processors is available
4030 through the -march and -mtune options.
4032 Improved support for built-in functions
4034 * GCC can now use load-linked, store-conditional and sync
4035 instructions to implement atomic built-in functions such as
4036 __sync_fetch_and_add. The memory reference must be 4 bytes wide for
4037 32-bit targets and either 4 or 8 bytes wide for 64-bit targets.
4038 * GCC can now use the clz and dclz instructions to implement the
4039 __builtin_ctz and __builtin_ffs families of functions.
4040 * There is a new __builtin___clear_cache function for flushing the
4041 instruction cache. GCC expands this function inline on MIPS32
4042 revision 2 targets, otherwise it calls the function specified by
4047 * GCC can now compile objects that contain a mixture of MIPS16 and
4048 non-MIPS16 code. There are two new attributes, mips16 and nomips16,
4049 for specifying which mode a function should use.
4050 * A new option called -minterlink-mips16 makes non-MIPS16 code
4051 link-compatible with MIPS16 code.
4052 * After many bug fixes, the long-standing MIPS16 -mhard-float support
4053 should now work fairly reliably.
4054 * GCC can now use the MIPS16e save and restore instructions.
4055 * -fsection-anchors now works in MIPS16 mode. MIPS16 code compiled
4056 with -G0 -fsection-anchors is often smaller than code compiled with
4057 -G8. However, please note that you must usually compile all objects
4058 in your application with the same -G option; see the documentation
4060 * A new option called-mcode-readable specifies which instructions are
4061 allowed to load from the code segment. -mcode-readable=yes is the
4062 default and says that any instruction may load from the code
4063 segment. The other alternatives are -mcode-readable=pcrel, which
4064 says that only PC-relative MIPS16 instructions may load from the
4065 code segment, and -mcode-readable=no, which says that no
4066 instruction may do so. Please see the documentation for more
4067 details, including example uses.
4069 Small-data improvements
4071 There are three new options for controlling small data:
4072 * -mno-extern-sdata, which disables small-data accesses for
4073 externally-defined variables. Code compiled with -Gn
4074 -mno-extern-sdata will be link-compatible with any -G setting
4075 between -G0 and -Gn inclusive.
4076 * -mno-local-sdata, which disables the use of small-data sections for
4077 data that is not externally visible. This option can be a useful
4078 way of reducing small-data usage in less performance-critical parts
4080 * -mno-gpopt, which disables the use of the $gp register while still
4081 honoring the -G limit when placing externally-visible data. This
4082 option implies -mno-extern-sdata and -mno-local-sdata and it can be
4083 useful in situations where $gp does not necessarily hold the
4086 Miscellaneous improvements
4088 * There is a new option called -mbranch-cost for tweaking the
4089 perceived cost of branches.
4090 * If GCC is configured to use a version of GAS that supports the
4091 .gnu_attribute directive, it will use that directive to record
4092 certain properties of the output code. .gnu_attribute is new to GAS
4094 * There are two new function attributes, near and far, for overriding
4095 the command-line setting of -mlong-calls on a function-by-function
4097 * -mfp64, which previously required a 64-bit target, now works with
4098 MIPS32 revision 2 targets as well. The mipsisa*-elfoabi* and
4099 mipsisa*-sde-elf* configurations provide suitable library support.
4100 * GCC now recognizes the -mdmx and -mmt options and passes them down
4101 to the assembler. It does nothing else with the options at present.
4103 SPU (Synergistic Processor Unit) of the Cell Broadband Engine Architecture
4106 * Support has been added for this new architecture.
4108 RS6000 (POWER/PowerPC)
4110 * Support for the PowerPC 750CL paired-single instructions has been
4111 added with a new powerpc-*-linux*paired* target configuration. It
4112 is enabled by an associated -mpaired option and can be accessed
4113 using new built-in functions.
4114 * Support for auto-detecting architecture and system configuration to
4115 auto-select processor optimization tuning.
4116 * Support for VMX on AIX 5.3 has been added.
4117 * Support for AIX Version 6.1 has been added.
4119 S/390, zSeries and System z9
4121 * Support for the IBM System z9 EC/BC processor (z9 GA3) has been
4122 added. When using the -march=z9-ec option, the compiler will
4123 generate code making use of instructions provided by the decimal
4124 floating point facility and the floating point conversion facility
4125 (pfpo). Besides the instructions used to implement decimal floating
4126 point operations these facilities also contain instructions to move
4127 between general purpose and floating point registers and to modify
4128 and copy the sign-bit of floating point values.
4129 * When the -march=z9-ec option is used the new
4130 -mhard-dfp/-mno-hard-dfp options can be used to specify whether the
4131 decimal floating point hardware instructions will be used or not.
4132 If none of them is given the hardware support is enabled by
4134 * The -mstack-guard option can now be omitted when using stack
4135 checking via -mstack-size in order to let GCC choose a sensible
4136 stack guard value according to the frame size of each function.
4137 * Various changes to improve performance of generated code have been
4138 implemented, including:
4139 + The condition code set by an add logical with carry
4140 instruction is now available for overflow checks like: a + b +
4142 + The test data class instruction is now used to implement
4143 sign-bit and infinity checks of binary and decimal floating
4148 * Support for the Sun UltraSPARC T2 (Niagara 2) processor has been
4153 * Stack unwinding for exception handling now uses by default a
4154 specialized version of DWARF unwinding. This is not
4155 binary-compatible with the setjmp/longjmp (sjlj) unwinding used for
4156 Xtensa with previous versions of GCC.
4157 * For Xtensa processors that include the Conditional Store option,
4158 the built-in functions for atomic memory access are now implemented
4159 using S32C1I instructions.
4160 * If the Xtensa NSA option is available, GCC will use it to implement
4161 the __builtin_ctz and __builtin_clz functions.
4163 Documentation improvements
4165 * Existing libstdc++ documentation has been edited and restructured
4166 into a single DocBook XML manual. The results can be viewed online
4169 Other significant improvements
4171 * The compiler's --help command-line option has been extended so that
4172 it now takes an optional set of arguments. These arguments restrict
4173 the information displayed to specific classes of command-line
4174 options, and possibly only a subset of those options. It is also
4175 now possible to replace the descriptive text associated with each
4176 displayed option with an indication of its current value, or for
4177 binary options, whether it has been enabled or disabled.
4178 Here are some examples. The following will display all the options
4179 controlling warning messages:
4182 Whereas this will display all the undocumented, target specific
4184 --help=target,undocumented
4186 This sequence of commands will display the binary optimizations
4187 that are enabled by -O3:
4188 gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts
4189 gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts
4190 diff /tmp/O2-opts /tmp/O3-opts | grep enabled
4192 * The configure options --with-pkgversion and --with-bugurl have been
4193 added. These allow distributors of GCC to include a
4194 distributor-specific string in manuals and --version output and to
4195 specify the URL for reporting bugs in their versions of GCC.
4199 This is the [25]list of problem reports (PRs) from GCC's bug tracking
4200 system that are known to be fixed in the 4.3.1 release. This list might
4201 not be complete (that is, it is possible that some PRs that have been
4202 fixed are not listed here).
4204 Target Specific Changes
4210 * Starting with GCC 4.3.1, decimal floating point variables are
4211 aligned to their natural boundaries when they are passed on the
4214 Command-line changes
4216 * Starting with GCC 4.3.1, the -mcld option has been added to
4217 automatically generate a cld instruction in the prologue of
4218 functions that use string instructions. This option is used for
4219 backward compatibility on some operating systems and can be enabled
4220 by default for 32-bit x86 targets by configuring GCC with the
4221 --enable-cld configure option.
4225 This is the [26]list of problem reports (PRs) from GCC's bug tracking
4226 system that are known to be fixed in the 4.3.2 release. This list might
4227 not be complete (that is, it is possible that some PRs that have been
4228 fixed are not listed here).
4232 This is the [27]list of problem reports (PRs) from GCC's bug tracking
4233 system that are known to be fixed in the 4.3.3 release. This list might
4234 not be complete (that is, it is possible that some PRs that have been
4235 fixed are not listed here).
4239 This is the [28]list of problem reports (PRs) from GCC's bug tracking
4240 system that are known to be fixed in the 4.3.4 release. This list might
4241 not be complete (that is, it is possible that some PRs that have been
4242 fixed are not listed here).
4246 This is the [29]list of problem reports (PRs) from GCC's bug tracking
4247 system that are known to be fixed in the 4.3.5 release. This list might
4248 not be complete (that is, it is possible that some PRs that have been
4249 fixed are not listed here).
4253 This is the [30]list of problem reports (PRs) from GCC's bug tracking
4254 system that are known to be fixed in the 4.3.6 release. This list might
4255 not be complete (that is, it is possible that some PRs that have been
4256 fixed are not listed here).
4259 For questions related to the use of GCC, please consult these web
4260 pages and the [31]GCC manuals. If that fails, the
4261 [32]gcc-help@gcc.gnu.org mailing list might help. Comments on these
4262 web pages and the development of GCC are welcome on our developer
4263 list at [33]gcc@gcc.gnu.org. All of [34]our lists have public
4266 Copyright (C) [35]Free Software Foundation, Inc. Verbatim copying and
4267 distribution of this entire article is permitted in any medium,
4268 provided this notice is preserved.
4270 These pages are [36]maintained by the GCC team. Last modified
4275 1. http://gcc.gnu.org/gcc-4.3/changes.html#4.3.5
4276 2. http://gmplib.org/
4277 3. http://www.mpfr.org/
4278 4. http://gcc.gnu.org/install/prerequisites.html
4279 5. http://gcc.gnu.org/ml/gcc-announce/2001/msg00000.html
4280 6. http://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html#Warning-Options
4281 7. http://gcc.gnu.org/gcc-4.3/porting_to.html
4282 8. http://www.mpfr.org/
4283 9. http://www.mpfr.org/
4284 10. http://www.mpfr.org/
4285 11. http://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html
4286 12. http://gcc.gnu.org/gcc-4.3/cxx0x_status.html
4287 13. http://gcc.gnu.org/gcc-4.3/cxx0x_status.html
4288 14. http://gcc.gnu.org/onlinedocs/libstdc++/manual/bk01pt01ch01.html#m anual.intro.status.standard.tr1
4289 15. http://gcc.gnu.org/onlinedocs/libstdc++/manual/parallel_mode.html
4290 16. http://gmplib.org/
4291 17. http://www.mpfr.org/
4292 18. http://gcc.gnu.org/onlinedocs/gfortran/Code-Gen-Options.html#Code-Gen-Options
4293 19. http://gcc.gnu.org/onlinedocs/gfortran/Code-Gen-Options.html#index-g_t_0040code_007bfinit-local-zero_007d-167
4294 20. http://gcc.gnu.org/onlinedocs/gcc-4.3.0/gfortran/GAMMA.html
4295 21. http://gcc.gnu.org/onlinedocs/gcc-4.3.0/gfortran/LGAMMA.html
4296 22. http://gcc.gnu.org/onlinedocs/gfortran/Fortran-Dialect-Options.html
4297 23. http://gcc.gnu.org/onlinedocs/gfortran/BOZ-literal-constants.html
4298 24. http://gcc.gnu.org/onlinedocs/libstdc++/
4299 25. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.3.1
4300 26. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.3.2
4301 27. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.3.3
4302 28. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.3.4
4303 29. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.3.5
4304 30. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.3.6
4305 31. http://gcc.gnu.org/onlinedocs/
4306 32. mailto:gcc-help@gcc.gnu.org
4307 33. mailto:gcc@gcc.gnu.org
4308 34. http://gcc.gnu.org/lists.html
4309 35. http://www.fsf.org/
4310 36. http://gcc.gnu.org/about.html
4311 37. http://validator.w3.org/check/referer
4312 ======================================================================
4313 http://gcc.gnu.org/gcc-4.2/index.html
4314 GCC 4.2 Release Series
4318 The [1]GNU project and the GCC developers are pleased to announce the
4319 release of GCC 4.2.4.
4321 This release is a bug-fix release, containing fixes for regressions in
4322 GCC 4.2.3 relative to previous releases of GCC.
4327 May 19, 2008 ([2]changes)
4330 February 1, 2008 ([3]changes)
4333 October 7, 2007 ([4]changes)
4336 July 18, 2007 ([5]changes)
4339 May 13, 2007 ([6]changes)
4341 References and Acknowledgements
4343 GCC used to stand for the GNU C Compiler, but since the compiler
4344 supports several other languages aside from C, it now stands for the
4345 GNU Compiler Collection.
4347 A list of [7]successful builds is updated as new information becomes
4350 The GCC developers would like to thank the numerous people that have
4351 contributed new features, improvements, bug fixes, and other changes as
4352 well as test results to GCC. This [8]amazing group of volunteers is
4353 what makes GCC successful.
4355 For additional information about GCC please refer to the [9]GCC project
4356 web site or contact the [10]GCC development mailing list.
4358 To obtain GCC please use [11]our mirror sites or [12]our SVN server.
4361 For questions related to the use of GCC, please consult these web
4362 pages and the [13]GCC manuals. If that fails, the
4363 [14]gcc-help@gcc.gnu.org mailing list might help. Comments on these
4364 web pages and the development of GCC are welcome on our developer
4365 list at [15]gcc@gcc.gnu.org. All of [16]our lists have public
4368 Copyright (C) [17]Free Software Foundation, Inc. Verbatim copying and
4369 distribution of this entire article is permitted in any medium,
4370 provided this notice is preserved.
4372 These pages are [18]maintained by the GCC team. Last modified
4377 1. http://www.gnu.org/
4378 2. http://gcc.gnu.org/gcc-4.2/changes.html
4379 3. http://gcc.gnu.org/gcc-4.2/changes.html
4380 4. http://gcc.gnu.org/gcc-4.2/changes.html
4381 5. http://gcc.gnu.org/gcc-4.2/changes.html
4382 6. http://gcc.gnu.org/gcc-4.2/changes.html
4383 7. http://gcc.gnu.org/gcc-4.2/buildstat.html
4384 8. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
4385 9. http://gcc.gnu.org/index.html
4386 10. mailto:gcc@gcc.gnu.org
4387 11. http://gcc.gnu.org/mirrors.html
4388 12. http://gcc.gnu.org/svn.html
4389 13. http://gcc.gnu.org/onlinedocs/
4390 14. mailto:gcc-help@gcc.gnu.org
4391 15. mailto:gcc@gcc.gnu.org
4392 16. http://gcc.gnu.org/lists.html
4393 17. http://www.fsf.org/
4394 18. http://gcc.gnu.org/about.html
4395 19. http://validator.w3.org/check/referer
4396 ======================================================================
4397 http://gcc.gnu.org/gcc-4.2/changes.html
4398 GCC 4.2 Release Series
4399 Changes, New Features, and Fixes
4403 * GCC no longer accepts the -fshared-data option. This option has had
4404 no effect in any GCC 4 release; the targets to which the option
4405 used to apply had been removed before GCC 4.0.
4407 General Optimizer Improvements
4409 * New command-line options specify the possible relationships among
4410 parameters and between parameters and global data. For example,
4411 -fargument-noalias-anything specifies that arguments do not alias
4413 Each language will automatically use whatever option is required by
4414 the language standard. You should not need to use these options
4417 New Languages and Language specific improvements
4419 * [1]OpenMP is now supported for the C, C++ and Fortran compilers.
4420 * New command-line options -fstrict-overflow and -Wstrict-overflow
4421 have been added. -fstrict-overflow tells the compiler that it may
4422 assume that the program follows the strict signed overflow
4423 semantics permitted for the language: for C and C++ this means that
4424 the compiler may assume that signed overflow does not occur. For
4425 example, a loop like
4426 for (i = 1; i > 0; i *= 2)
4428 is presumably intended to continue looping until i overflows. With
4429 -fstrict-overflow, the compiler may assume that signed overflow
4430 will not occur, and transform this into an infinite loop.
4431 -fstrict-overflow is turned on by default at -O2, and may be
4432 disabled via -fno-strict-overflow. The -Wstrict-overflow option may
4433 be used to warn about cases where the compiler assumes that signed
4434 overflow will not occur. It takes five different levels:
4435 -Wstrict-overflow=1 to 5. See the [2]documentation for details.
4436 -Wstrict-overflow=1 is enabled by -Wall.
4437 * The new command-line option -fno-toplevel-reorder directs GCC to
4438 emit top-level functions, variables, and asm statements in the same
4439 order that they appear in the input file. This is intended to
4440 support existing code which relies on a particular ordering (for
4441 example, code which uses top-level asm statements to switch
4442 sections). For new code, it is generally better to use function and
4443 variable attributes. The -fno-toplevel-reorder option may be used
4444 for most cases which currently use -fno-unit-at-a-time. The
4445 -fno-unit-at-a-time option will be removed in some future version
4446 of GCC. If you know of a case which requires -fno-unit-at-a-time
4447 which is not fixed by -fno-toplevel-reorder, please open a bug
4452 * The pragma redefine_extname will now macro expand its tokens for
4453 compatibility with SunPRO.
4454 * In the next release of GCC, 4.3, -std=c99 or -std=gnu99 will direct
4455 GCC to handle inline functions as specified in the C99 standard. In
4456 preparation for this, GCC 4.2 will warn about any use of non-static
4457 inline functions in gnu99 or c99 mode. This new warning may be
4458 disabled with the new gnu_inline function attribute or the new
4459 -fgnu89-inline command-line option. Also, GCC 4.2 and later will
4460 define one of the preprocessor macros __GNUC_GNU_INLINE__ or
4461 __GNUC_STDC_INLINE__ to indicate the semantics of inline functions
4462 in the current compilation.
4463 * A new command-line option -Waddress has been added to warn about
4464 suspicious uses of memory addresses as, for example, using the
4465 address of a function in a conditional expression, and comparisons
4466 against the memory address of a string literal. This warning is
4471 * C++ visibility handling has been overhauled.
4472 Restricted visiblity is propagated from classes to members, from
4473 functions to local statics, and from templates and template
4474 arguments to instantiations, unless the latter has explicitly
4475 declared visibility.
4476 The visibility attribute for a class must come between the
4477 class-key and the name, not after the closing brace.
4478 Attributes are now allowed for enums and elaborated-type-specifiers
4479 that only declare a type.
4480 Members of the anonymous namespace are now local to a particular
4481 translation unit, along with any other declarations which use them,
4482 though they are still treated as having external linkage for
4484 * The (undocumented) extension which permitted templates with default
4485 arguments to be bound to template template parameters with fewer
4486 parameters has been removed. For example:
4487 template <template <typename> class C>
4488 void f(C<double>) {}
4490 template <typename T, typename U = int>
4493 template void f(S<double>);
4495 is no longer accepted by G++. The reason this code is not accepted
4496 is that S is a template with two parameters; therefore, it cannot
4497 be bound to C which has only one parameter.
4498 * The <?, >?, <?=, and >?= operators, deprecated in previous GCC
4499 releases, have been removed.
4500 * The command-line option -fconst-strings, deprecated in previous GCC
4501 releases, has been removed.
4502 * The configure variable enable-__cxa_atexit is now enabled by
4503 default for more targets. Enabling this variable is necessary in
4504 order for static destructors to be executed in the correct order,
4505 but it depends upon the presence of a non-standard C library in the
4506 target library in order to work. The variable is now enabled for
4507 more targets which are known to have suitable C libraries.
4508 * -Wextra will produce warnings for if statements with a semicolon as
4509 the only body, to catch code like:
4514 To suppress the warning in valid cases, use { } instead.
4515 * The C++ frontend now also produces strict aliasing warnings when
4516 -fstrict-aliasing -Wstrict-aliasing is in effect.
4518 Runtime Library (libstdc++)
4520 * Added support for TR1 <random>, <complex>, and C compatibility
4521 headers. In addition, a lock-free version of shared_ptr was
4522 contributed as part of Phillip Jordan's Google Summer of Code
4523 project on lock-free containers. ([3]Implementation status of TR1)
4524 * In association with the Summer of Code work on lock-free
4525 containers, the interface for atomic builtins was adjusted,
4526 creating simpler alternatives for non-threaded code paths. Also,
4527 usage was consolidated and all elements were moved from namespace
4528 std to namespace__gnu_cxx. Affected interfaces are the functions
4529 __exchange_and_add, __atomic_add, and the objects __mutex,
4530 __recursive_mutex, and __scoped_lock.
4531 * Support for versioning weak symbol names via namespace association
4532 was added. However, as this changes the names of exported symbols,
4533 this is turned off by default in the current ABI. Intrepid users
4534 can enable this feature by using
4535 --enable-symvers=gnu-versioned-namespace during configuration.
4536 * Revised, simplified, and expanded policy-based associative
4537 containers, including data types for tree and trie forms
4538 (basic_tree, tree, trie), lists (list_update), and both
4539 collision-chaining and probing hash-based containers
4540 (basic_hash_table, cc_hash_table, gp_hash_table). More details per
4541 the [4]documentation.
4542 * The implementation of the debug mode was modified, whereby the
4543 debug namespaces were nested inside of namespace std and namespace
4544 __gnu_cxx in order to resolve some long standing corner cases
4545 involving name lookup. Debug functionality from the policy-based
4546 data structures was consolidated and enabled with the single macro,
4547 _GLIBCXX_DEBUG. See PR 26142 for more information.
4548 * Added extensions for type traits: __conditional_type,
4549 __numeric_traits, __add_unsigned, __removed_unsigned, __enable_if.
4550 * Added a typelist implementation for compile-time meta-programming.
4551 Elements for typelist construction and operation can be found
4552 within namespace __gnu_cxx::typelist.
4553 * Added a new allocator, __gnu_cxx::throw_allocator, for testing
4555 * Enabled library-wide visibility control, allowing -fvisibility to
4557 * Consolidated all nested namespaces and the conversion of
4558 __gnu_internal implementation-private details to anonymous
4559 namespaces whenever possible.
4560 * Implemented LWG resolutions DR 431 and DR 538.
4564 * Support for allocatable components has been added (TR 15581 and
4566 * Support for the Fortran 2003 streaming IO extension has been added.
4567 * The GNU Fortran compiler now uses 4-byte record markers by default
4568 for unformatted files to be compatible with g77 and most other
4569 compilers. The implementation allows for records greater than 2 GB
4570 and is compatible with several other compilers. Older versions of
4571 gfortran used 8-byte record markers by default (on most systems).
4572 In order to change the length of the record markers, e.g. to read
4573 unformatted files created by older gfortran versions, the
4574 [5]-frecord-marker=8 option can be used.
4578 * A new command-line option -static-libgcj has been added for targets
4579 that use a linker compatible with GNU Binutils. As its name
4580 implies, this causes libgcj to be linked statically. In some cases
4581 this causes the resulting executable to start faster and use less
4582 memory than if the shared version of libgcj were used. However
4583 caution should be used as it can also cause essential parts of the
4584 library to be omitted. Some of these issues are discussed in:
4585 [6]http://gcc.gnu.org/wiki/Statically_linking_libgcj
4586 * fastjar is no longer bundled with GCC. To build libgcj, you will
4587 need either InfoZIP (both zip and unzip) or an external jar
4588 program. In the former case, the GCC build will install a jar shell
4589 script that is based on InfoZIP and provides the same functionality
4592 New Targets and Target Specific Improvements
4596 * -mtune=generic can now be used to generate code running well on
4597 common x86 chips. This includes AMD Athlon, AMD Opteron, Intel
4598 Pentium-M, Intel Pentium 4 and Intel Core 2.
4599 * -mtune=native and -march=native will produce code optimized for the
4600 host architecture as detected using the cpuid instruction.
4601 * Added a new command-line option -fstackrealign and and
4602 __attribute__ ((force_align_arg_pointer)) to realign the stack at
4603 runtime. This allows functions compiled with a vector-aligned stack
4604 to be invoked from legacy objects that keep only word-alignment.
4608 * The default CPU setting has been changed from V7 to V9 in 32-bit
4609 mode on Solaris 7 and above. This is already the case in 64-bit
4610 mode. It can be overridden by specifying --with-cpu at configure
4612 * Back-end support of built-in functions for atomic memory access has
4614 * Support for the Sun UltraSPARC T1 (Niagara) processor has been
4619 * Various bug fixes have made some functions (notably, functions
4620 returning structures) incompatible with previous releases.
4621 Recompiling all libraries is recommended. Note that code quality
4622 has considerably improved since 4.1, making a recompile even more
4627 * Added support for the Broadcom SB-1A core.
4631 * Added support for IA-64 data and control speculation. By default
4632 speculation is enabled only during second scheduler pass. A number
4633 of machine flags was introduced to control the usage of speculation
4634 for both scheduler passes.
4638 * Added Java language support (libffi and libjava) for 32-bit HP-UX
4643 Documentation improvements
4647 * A make pdf target has been added to the top-level makefile,
4648 enabling automated production of PDF documentation files.
4649 (Front-ends external to GCC should modify their Make-lang.in file
4650 to add a lang.pdf: target.)
4652 Other significant improvements
4654 Build system improvements
4656 * All the components of the compiler are now bootstrapped by default.
4657 This improves the resilience to bugs in the system compiler or
4658 binary compatibility problems, as well as providing better testing
4659 of GCC 4.2 itself. In addition, if you build the compiler from a
4660 combined tree, the assembler, linker, etc. will also be
4661 bootstrapped (i.e. built with themselves).
4662 You can disable this behavior, and go back to the pre-GCC 4.2 set
4663 up, by configuring GCC with --disable-bootstrap.
4664 * The rules that configure follows to find target tools resemble more
4665 closely the locations that the built compiler will search. In
4666 addition, you can use the new configure option --with-target-tools
4667 to specify where to find the target tools used during the build,
4668 without affecting what the built compiler will use.
4669 This can be especially useful when building packages of GCC. For
4670 example, you may want to build GCC with GNU as or ld, even if the
4671 resulting compiler to work with the native assembler and linker. To
4672 do so, you can use --with-target-tools to point to the native
4675 Incompatible changes to the build system
4677 * Front-ends external to GCC should modify their Make-lang.in file to
4678 replace double-colon rules (e.g. dvi::) with normal rules (like
4679 lang.dvi:). Front-end makefile hooks do not use double-colon rules
4681 * Up to GCC 4.1, a popular way to specify the target tools used
4682 during the build was to create directories named gas, binutils,
4683 etc. in the build tree, and create links to the tools from there.
4684 This does not work any more when the compiler is bootstrapped. The
4685 new configure option --with-target-tools provides a better way to
4686 achieve the same effect, and works for all native and cross
4690 For questions related to the use of GCC, please consult these web
4691 pages and the [7]GCC manuals. If that fails, the
4692 [8]gcc-help@gcc.gnu.org mailing list might help. Comments on these
4693 web pages and the development of GCC are welcome on our developer
4694 list at [9]gcc@gcc.gnu.org. All of [10]our lists have public
4697 Copyright (C) [11]Free Software Foundation, Inc. Verbatim copying and
4698 distribution of this entire article is permitted in any medium,
4699 provided this notice is preserved.
4701 These pages are [12]maintained by the GCC team. Last modified
4706 1. http://gcc.gnu.org/projects/gomp/
4707 2. http://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html
4708 3. http://gcc.gnu.org/onlinedocs/libstdc++/manual/bk01pt01ch01.html#manual.intro.status.standard.tr1
4709 4. http://gcc.gnu.org/onlinedocs/libstdc++/ext/pb_ds/index.html
4710 5. http://gcc.gnu.org/onlinedocs/gfortran/Runtime-Options.html
4711 6. http://gcc.gnu.org/wiki/Statically_linking_libgcj
4712 7. http://gcc.gnu.org/onlinedocs/
4713 8. mailto:gcc-help@gcc.gnu.org
4714 9. mailto:gcc@gcc.gnu.org
4715 10. http://gcc.gnu.org/lists.html
4716 11. http://www.fsf.org/
4717 12. http://gcc.gnu.org/about.html
4718 13. http://validator.w3.org/check/referer
4719 ======================================================================
4720 http://gcc.gnu.org/gcc-4.1/index.html
4721 GCC 4.1 Release Series
4725 The [1]GNU project and the GCC developers are pleased to announce the
4726 release of GCC 4.1.2.
4728 This release is a bug-fix release, containing fixes for regressions in
4729 GCC 4.1.1 relative to previous releases of GCC.
4734 February 13, 2007 ([2]changes)
4737 May 24, 2006 ([3]changes)
4740 February 28, 2006 ([4]changes)
4742 References and Acknowledgements
4744 GCC used to stand for the GNU C Compiler, but since the compiler
4745 supports several other languages aside from C, it now stands for the
4746 GNU Compiler Collection.
4748 A list of [5]successful builds is updated as new information becomes
4751 The GCC developers would like to thank the numerous people that have
4752 contributed new features, improvements, bug fixes, and other changes as
4753 well as test results to GCC. This [6]amazing group of volunteers is
4754 what makes GCC successful.
4756 For additional information about GCC please refer to the [7]GCC project
4757 web site or contact the [8]GCC development mailing list.
4759 To obtain GCC please use [9]our mirror sites or [10]our SVN server.
4762 For questions related to the use of GCC, please consult these web
4763 pages and the [11]GCC manuals. If that fails, the
4764 [12]gcc-help@gcc.gnu.org mailing list might help. Comments on these
4765 web pages and the development of GCC are welcome on our developer
4766 list at [13]gcc@gcc.gnu.org. All of [14]our lists have public
4769 Copyright (C) [15]Free Software Foundation, Inc. Verbatim copying and
4770 distribution of this entire article is permitted in any medium,
4771 provided this notice is preserved.
4773 These pages are [16]maintained by the GCC team. Last modified
4778 1. http://www.gnu.org/
4779 2. http://gcc.gnu.org/gcc-4.1/changes.html#4.1.2
4780 3. http://gcc.gnu.org/gcc-4.1/changes.html
4781 4. http://gcc.gnu.org/gcc-4.1/changes.html
4782 5. http://gcc.gnu.org/gcc-4.1/buildstat.html
4783 6. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
4784 7. http://gcc.gnu.org/index.html
4785 8. mailto:gcc@gcc.gnu.org
4786 9. http://gcc.gnu.org/mirrors.html
4787 10. http://gcc.gnu.org/svn.html
4788 11. http://gcc.gnu.org/onlinedocs/
4789 12. mailto:gcc-help@gcc.gnu.org
4790 13. mailto:gcc@gcc.gnu.org
4791 14. http://gcc.gnu.org/lists.html
4792 15. http://www.fsf.org/
4793 16. http://gcc.gnu.org/about.html
4794 17. http://validator.w3.org/check/referer
4795 ======================================================================
4796 http://gcc.gnu.org/gcc-4.1/changes.html
4797 GCC 4.1 Release Series
4798 Changes, New Features, and Fixes
4800 The latest release in the 4.1 release series is [1]GCC 4.1.2.
4804 General Optimizer Improvements
4806 * GCC now has infrastructure for inter-procedural optimizations and
4807 the following inter-procedural optimizations are implemented:
4808 + Profile guided inlining. When doing profile feedback guided
4809 optimization, GCC can now use the profile to make better
4810 informed decisions on whether inlining of a function is
4811 profitable or not. This means that GCC will no longer inline
4812 functions at call sites that are not executed very often, and
4813 that functions at hot call sites are more likely to be
4815 A new parameter min-inline-recursive-probability is also now
4816 available to throttle recursive inlining of functions with
4817 small average recursive depths.
4818 + Discovery of pure and const functions, a form of side-effects
4819 analysis. While older GCC releases could also discover such
4820 special functions, the new IPA-based pass runs earlier so that
4821 the results are available to more optimizers. The pass is also
4822 simply more powerful than the old one.
4823 + Analysis of references to static variables and type escape
4824 analysis, also forms of side-effects analysis. The results of
4825 these passes allow the compiler to be less conservative about
4826 call-clobbered variables and references. This results in more
4827 redundant loads being eliminated and in making static
4828 variables candidates for register promotion.
4829 + Improvement of RTL-based alias analysis. The results of type
4830 escape analysis are fed to the RTL type-based alias analyzer,
4831 allowing it to disambiguate more memory references.
4832 + Interprocedural constant propagation and function versioning.
4833 This pass looks for functions that are always called with the
4834 same constant value for one or more of the function arguments,
4835 and propagates those constants into those functions.
4836 + GCC will now eliminate static variables whose usage was
4838 + -fwhole-program --combine can now be used to make all
4839 functions in program static allowing whole program
4840 optimization. As an exception, the main function and all
4841 functions marked with the new externally_visible attribute are
4842 kept global so that programs can link with runtime libraries.
4843 * GCC can now do a form of partial dead code elimination (PDCE) that
4844 allows code motion of expressions to the paths where the result of
4845 the expression is actually needed. This is not always a win, so the
4846 pass has been limited to only consider profitable cases. Here is an
4848 int foo (int *, int *);
4864 The a = b + c can be sunk to right before the printf. Normal code
4865 sinking will not do this, it will sink the first one above into the
4866 else-branch of the conditional jump, which still gives you two
4868 * GCC now has a value range propagation pass. This allows the
4869 compiler to eliminate bounds checks and branches. The results of
4870 the pass can also be used to accurately compute branch
4872 * The pass to convert PHI nodes to straight-line code (a form of
4873 if-conversion for GIMPLE) has been improved significantly. The two
4874 most significant improvements are an improved algorithm to
4875 determine the order in which the PHI nodes are considered, and an
4876 improvement that allow the pass to consider if-conversions of basic
4877 blocks with more than two predecessors.
4878 * Alias analysis improvements. GCC can now differentiate between
4879 different fields of structures in Tree-SSA's virtual operands form.
4880 This lets stores/loads from non-overlapping structure fields not
4881 conflict. A new algorithm to compute points-to sets was contributed
4882 that can allows GCC to see now that p->a and p->b, where p is a
4883 pointer to a structure, can never point to the same field.
4884 * Various enhancements to auto-vectorization:
4885 + Incrementally preserve SSA form when vectorizing.
4886 + Incrementally preserve loop-closed form when vectorizing.
4887 + Improvements to peeling for alignment: generate better code
4888 when the misalignment of an access is known at compile time,
4889 or when different accesses are known to have the same
4890 misalignment, even if the misalignment amount itself is
4892 + Consider dependence distance in the vectorizer.
4893 + Externalize generic parts of data reference analysis to make
4894 this analysis available to other passes.
4895 + Vectorization of conditional code.
4896 + Reduction support.
4897 * GCC can now partition functions in sections of hot and cold code.
4898 This can significantly improve performance due to better
4899 instruction cache locality. This feature works best together with
4900 profile feedback driven optimization.
4901 * A new pass to avoid saving of unneeded arguments to the stack in
4902 vararg functions if the compiler can prove that they will not be
4904 * Transition of basic block profiling to tree level implementation
4905 has been completed. The new implementation should be considerably
4906 more reliable (hopefully avoiding profile mismatch errors when
4907 using -fprofile-use or -fbranch-probabilities) and can be used to
4908 drive higher level optimizations, such as inlining.
4909 The -ftree-based-profiling command-line option was removed and
4910 -fprofile-use now implies disabling old RTL level loop optimizer
4911 (-fno-loop-optimize). Speculative prefetching optimization
4912 (originally enabled by -fspeculative-prefetching) was removed.
4914 New Languages and Language specific improvements
4918 * The old Bison-based C and Objective-C parser has been replaced by a
4919 new, faster hand-written recursive-descent parser.
4923 * The build infrastructure for the Ada runtime library and tools has
4924 been changed to be better integrated with the rest of the build
4925 infrastructure of GCC. This should make doing cross builds of Ada a
4930 * ARM-style name-injection of friend declarations is no longer the
4931 default. For example:
4937 will not be accepted; instead a declaration of f will need to be
4938 present outside of the scope of S. The new -ffriend-injection
4939 option will enable the old behavior.
4940 * The (undocumented) extension which permitted templates with default
4941 arguments to be bound to template template parameters with fewer
4942 parameters has been deprecated, and will be removed in the next
4943 major release of G++. For example:
4944 template <template <typename> class C>
4945 void f(C<double>) {}
4947 template <typename T, typename U = int>
4950 template void f(S<double>);
4952 makes use of the deprecated extension. The reason this code is not
4953 valid ISO C++ is that S is a template with two parameters;
4954 therefore, it cannot be bound to C which has only one parameter.
4956 Runtime Library (libstdc++)
4958 * Optimization work:
4959 + A new implementation of std::search_n is provided, better
4960 performing in case of random access iterators.
4961 + Added further efficient specializations of istream functions,
4962 i.e., character array and string extractors.
4963 + Other smaller improvements throughout.
4964 * Policy-based associative containers, designed for high-performance,
4965 flexibility and semantic safety are delivered in ext/pb_assoc.
4966 * A versatile string class, __gnu_cxx::__versa_string, providing
4967 facilities conforming to the standard requirements for
4968 basic_string, is delivered in <ext/vstring.h>. In particular:
4969 + Two base classes are provided: the default one avoids
4970 reference counting and is optimized for short strings; the
4971 alternate one, still uses it while improving in a few low
4972 level areas (e.g., alignment). See vstring_fwd.h for some
4974 + Various algorithms have been rewritten (e.g., replace), the
4975 code streamlined and simple optimizations added.
4976 + Option 3 of DR 431 is implemented for both available bases,
4977 thus improving the support for stateful allocators.
4978 * As usual, many bugs have been fixed (e.g., libstdc++/13583,
4979 libstdc++/23953) and LWG resolutions put into effect for the first
4980 time (e.g., DR 280, DR 464, N1780 recommendations for DR 233, TR1
4981 Issue 6.19). The implementation status of TR1 is now tracked in the
4986 * A new language front end for Objective-C++ has been added. This
4987 language allows users to mix the object oriented features of
4988 Objective-C with those of C++.
4992 * Core library (libgcj) updates based on GNU Classpath 0.15 - 0.19
4993 features (plus some 0.20 bug-fixes)
4995 o The java.net.HttpURLConnection implementation no longer
4996 buffers the entire response body in memory. This means
4997 that response bodies larger than available memory can now
5000 o NIO FileChannel.map implementation, fast bulk put
5001 implementation for DirectByteBuffer (speeds up this
5003 o FileChannel.lock() and FileChannel.force() implemented.
5005 o gnu.xml fix for nodes created outside a namespace
5007 o Add support for output indenting and
5008 cdata-section-elements output instruction in
5010 o xml.xpath corrections for cases where elements/attributes
5011 might have been created in non-namespace-aware mode.
5012 Corrections to handling of XSL variables and minor
5013 conformance updates.
5015 o GNU JAWT implementation, the AWT Native Interface, which
5016 allows direct access to native screen resources from
5017 within a Canvas's paint method. GNU Classpath Examples
5018 comes with a Demo, see libjava/classpath/examples/README.
5019 o awt.datatransfer updated to 1.5 with support for
5020 FlavorEvents. The gtk+ awt peers now allow copy/paste of
5021 text, images, URIs/files and serialized objects with
5022 other applications and tracking clipboard change events
5023 with gtk+ 2.6 (for gtk+ 2.4 only text and serialized
5024 objects are supported). A GNU Classpath Examples
5025 datatransfer Demo was added to show the new
5027 o Split gtk+ awt peers event handling in two threads and
5028 improve gdk lock handling (solves several awt lock ups).
5029 o Speed up awt Image loading.
5030 o Better gtk+ scrollbar peer implementation when using gtk+
5032 o Handle image loading errors correctly for gdkpixbuf and
5034 o Better handle GDK lock. Properly prefix gtkpeer native
5036 o GdkGraphics2D has been updated to use Cairo 0.5.x or
5038 o BufferedImage and GtkImage rewrites. All image drawing
5039 operations should now work correctly (flipping requires
5041 o Future Graphics2D, image and text work is documented at:
5042 [2]http://developer.classpath.org/mediation/ClasspathGrap
5044 o When gtk+ 2.6 or higher is installed the default log
5045 handler will produce stack traces whenever a WARNING,
5046 CRITICAL or ERROR message is produced.
5048 o The RepaintManager has been reworked for more efficient
5049 painting, especially for large GUIs.
5050 o The layout manager OverlayLayout has been implemented,
5051 the BoxLayout has been rewritten to make use of the
5052 SizeRequirements utility class and caching for more
5054 o Improved accessibility support.
5055 o Significant progress has been made in the implementation
5056 of the javax.swing.plaf.metal package, with most UI
5057 delegates in a working state now. Please test this with
5058 your own applications and provide feedback that will help
5059 us to improve this package.
5060 o The GUI demo (gnu.classpath.examples.swing.Demo) has been
5061 extended to highlight various features in our Free Swing
5062 implementation. And it includes a look and feel switcher
5063 for Metal (default), Ocean and GNU themes.
5064 o The javax.swing.plaf.multi package is now implemented.
5065 o Editing and several key actions for JTree and JTable were
5067 o Lots of icons and look and feel improvements for Free
5068 Swing basic and metal themes were added. Try running the
5069 GNU Classpath Swing Demo in examples
5070 (gnu.classpath.examples.swing.Demo) with:
5071 -Dswing.defaultlaf=javax.swing.plaf.basic.BasicLookAndFee
5073 -Dswing.defaultlaf=javax.swing.plaf.metal.MetalLookAndFee
5075 o Start of styled text capabilites for java.swing.text.
5076 o DefaultMutableTreeNode pre-order, post-order, depth-first
5077 and breadth-first traversal enumerations implemented.
5078 o JInternalFrame colors and titlebar draw properly.
5079 o JTree is working up to par (icons, selection and keyboard
5081 o JMenus were made more compatible in visual and
5082 programmatic behavior.
5083 o JTable changeSelection and multiple selections
5085 o JButton and JToggleButton change states work properly
5087 o JFileChooser fixes.
5088 o revalidate() and repaint() fixes which make Free Swing
5089 much more responsive.
5090 o MetalIconFactory implemented.
5091 o Free Swing Top-Level Compatibility. JFrame, JDialog,
5092 JApplet, JInternalFrame, and JWindow are now 1.5
5093 compatible in the sense that you can call add() and
5094 setLayout() directly on them, which will have the same
5095 effect as calling getContentPane().add() and
5096 getContentPane().setLayout().
5097 o The JTree interface has been completed. JTrees now
5098 recognizes mouse clicks and selections work.
5099 o BoxLayout works properly now.
5100 o Fixed GrayFilter to actually work.
5101 o Metal SplitPane implemented.
5102 o Lots of Free Swing text and editor stuff work now.
5103 + Free RMI and Corba
5104 o Andrew Watson, Vice President and Technical Director of
5105 the Object Management Group, has officially assigned us
5106 20 bit Vendor Minor Code Id: 0x47430 ("GC") that will
5107 mark remote classpath-specific system exceptions.
5108 Obtaining the VMCID means that GNU Classpath now is a
5109 recogniseable type of node in a highly interoperable
5111 o GNU Classpath now includes the first working draft to
5112 support the RMI over IIOP protocol. The current
5113 implementation is capable of remote invocations,
5114 transferring various Serializables and Externalizables
5115 via RMI-IIOP protocol. It can flatten graphs and, at
5116 least for the simple cases, is interoperable with 1.5
5118 o org.omg.PortableInterceptor and related functionality in
5119 other packages is now implemented:
5120 # The sever and client interceptors work as required
5122 # The IOR interceptor works as needed for 1.5.
5123 o The org.omg.DynamicAny package is completed and passes
5125 o The Portable Object Adapter should now support the output
5126 of the recent IDL to java compilers. These compilers now
5127 generate servants and not CORBA objects as before, making
5128 the output depend on the existing POA implementation.
5129 Completing POA means that such code can already be tried
5130 to run on Classpath. Our POA is tested for the following
5132 # POA converts servant to the CORBA object.
5133 # Servant provides to the CORBA object.
5134 # POA activates new CORBA object with the given Object
5135 Id (byte array) that is later accessible for the
5137 # During the first call, the ServantActivator provides
5138 servant for this and all subsequent calls on the
5140 # During each call, the ServantLocator provides
5141 servant for this call only.
5142 # ServantLocator or ServantActivator forwards call to
5144 # POA has a single servant, responsible for all
5146 # POA has a default servant, but some objects are
5147 explicitly connected to they specific servants.
5148 The POA is verified using tests from the former
5150 o The CORBA implementation is now a working prototype that
5151 should support features up to 1.3 inclusive. We invite
5152 groups writing CORBA dependent applications to try
5153 Classpath implementation, reporting any possible bugs.
5154 The CORBA prototype is interoperable with Sun's
5155 implementation v 1.4, transferring object references,
5156 primitive types, narrow and wide strings, arrays,
5157 structures, trees, abstract interfaces and value types
5158 (feature of CORBA 2.3) between these two platforms.
5159 Remote exceptions are transferred and handled correctly.
5160 The stringified object references (IORs) from various
5161 sources are parsed as required. The transient (for
5162 current session) and permanent (till jre restart)
5163 redirections work. Both Little and Big Endian encoded
5164 messages are accepted. The implementation is verified
5165 using tests from the former cost.omg.org. The current
5166 release includes working examples (see the examples
5167 directory), demonstrating the client-server
5168 communication, using either CORBA Request or IDL-based
5169 stub (usually generated by a IDL to java compiler). These
5170 examples also show how to use the Classpath CORBA naming
5171 service. The IDL to java compiler is not yet written, but
5172 as our library must be compatible, it naturally accepts
5173 the output of other idlj implementations.
5175 o Updated TimeZone data against Olson tzdata2005l.
5176 o Make zip and jar packages UTF-8 clean.
5177 o "native" code builds and compiles (warning free) on
5179 o java.util.logging.FileHandler now rotates files.
5180 o Start of a generic JDWP framework in gnu/classpath/jdwp.
5181 This is unfinished, but feedback (at classpath@gnu.org)
5182 from runtime hackers is greatly appreciated. Although
5183 most of the work is currently being done around gcj/gij
5184 we want this framework to be as VM neutral as possible.
5185 Early design is described in:
5186 [3]http://gcc.gnu.org/ml/java/2005-05/msg00260.html
5187 o QT4 AWT peers, enable by giving configure
5188 --enable-qt-peer. Included, but not ready for production
5189 yet. They are explicitly disabled and not supported. But
5190 if you want to help with the development of these new
5191 features we are interested in feedback. You will have to
5192 explicitly enable them to try them out (and they will
5193 most likely contain bugs).
5194 o Documentation fixes all over the place. See
5195 [4]http://developer.classpath.org/doc/
5197 New Targets and Target Specific Improvements
5201 * The x86-64 medium model (that allows building applications whose
5202 data segment exceeds 4GB) was redesigned to match latest ABI draft.
5203 New implementation split large datastructures into separate segment
5204 improving performance of accesses to small datastructures and also
5205 allows linking of small model libraries into medium model programs
5206 as long as the libraries are not accessing the large datastructures
5207 directly. Medium model is also supported in position independent
5209 The ABI change results in partial incompatibility among medium
5210 model objects. Linking medium model libraries (or objects) compiled
5211 with new compiler into medium model program compiled with older
5212 will likely result in exceeding ranges of relocations.
5213 Binutils 2.16.91 or newer are required for compiling medium model
5216 RS6000 (POWER/PowerPC)
5218 * The AltiVec vector primitives in <altivec.h> are now implemented in
5219 a way that puts a smaller burden on the preprocessor, instead
5220 processing the "overloading" in the front ends. This should benefit
5221 compilation speed on AltiVec vector code.
5222 * AltiVec initializers now are generated more efficiently.
5223 * The popcountb instruction available on POWER5 now is generated.
5224 * The floating point round to integer instructions available on
5225 POWER5+ now is generated.
5226 * Floating point divides can be synthesized using the floating point
5227 reciprocal estimate instructions.
5228 * Double precision floating point constants are initialized as single
5229 precision values if they can be represented exactly.
5231 S/390, zSeries and System z9
5233 * Support for the IBM System z9 109 processor has been added. When
5234 using the -march=z9-109 option, the compiler will generate code
5235 making use of instructions provided by the extended immediate
5237 * Support for 128-bit IEEE floating point has been added. When using
5238 the -mlong-double-128 option, the compiler will map the long double
5239 data type to 128-bit IEEE floating point. Using this option
5240 constitutes an ABI change, and requires glibc support.
5241 * Various changes to improve performance of generated code have been
5242 implemented, including:
5243 + In functions that do not require a literal pool, register %r13
5244 (which is traditionally reserved as literal pool pointer), can
5245 now be freely used for other purposes by the compiler.
5246 + More precise tracking of register use allows the compiler to
5247 generate more efficient function prolog and epilog code in
5249 + The SEARCH STRING, COMPARE LOGICAL STRING, and MOVE STRING
5250 instructions are now used to implement C string functions.
5251 + The MOVE CHARACTER instruction with single byte overlap is now
5252 used to implement the memset function with non-zero fill byte.
5253 + The LOAD ZERO instructions are now used where appropriate.
5254 + The INSERT CHARACTERS UNDER MASK, STORE CHARACTERS UNDER MASK,
5255 and INSERT IMMEDIATE instructions are now used more frequently
5256 to optimize bitfield operations.
5257 + The BRANCH ON COUNT instruction is now used more frequently.
5258 In particular, the fact that a loop contains a subroutine call
5259 no longer prevents the compiler from using this instruction.
5260 + The compiler is now aware that all shift and rotate
5261 instructions implicitly truncate the shift count to six bits.
5262 * Back-end support for the following generic features has been
5264 + The full set of [5]built-in functions for atomic memory
5266 + The -fstack-protector feature.
5267 + The optimization pass avoiding unnecessary stores of incoming
5268 argument registers in functions with variable argument list.
5272 * The default code model in 64-bit mode has been changed from
5273 Medium/Anywhere to Medium/Middle on Solaris.
5274 * TLS support is disabled by default on Solaris prior to release 10.
5275 It can be enabled on TLS-capable Solaris 9 versions (4/04 release
5276 and later) by specifying --enable-tls at configure time.
5280 * Support has been added for this new architecture.
5284 Documentation improvements
5286 Other significant improvements
5288 * GCC can now emit code for protecting applications from
5289 stack-smashing attacks. The protection is realized by buffer
5290 overflow detection and reordering of stack variables to avoid
5292 * Some built-in functions have been fortified to protect them against
5293 various buffer overflow (and format string) vulnerabilities.
5294 Compared to the mudflap bounds checking feature, the safe builtins
5295 have far smaller overhead. This means that programs built using
5296 safe builtins should not experience any measurable slowdown.
5300 This is the [6]list of problem reports (PRs) from GCC's bug tracking
5301 system that are known to be fixed in the 4.1.2 release. This list might
5302 not be complete (that is, it is possible that some PRs that have been
5303 fixed are not listed here).
5305 When generating code for a shared library, GCC now recognizes that
5306 global functions may be replaced when the program runs. Therefore, it
5307 is now more conservative in deducing information from the bodies of
5308 functions. For example, in this example:
5313 cout << "Exception";
5317 G++ would previously have optimized away the catch clause, since it
5318 would have concluded that f cannot throw exceptions. Because users may
5319 replace f with another function in the main body of the program, this
5320 optimization is unsafe, and is no longer performed. If you wish G++ to
5321 continue to optimize as before, you must add a throw() clause to the
5322 declaration of f to make clear that it does not throw exceptions.
5325 For questions related to the use of GCC, please consult these web
5326 pages and the [7]GCC manuals. If that fails, the
5327 [8]gcc-help@gcc.gnu.org mailing list might help. Comments on these
5328 web pages and the development of GCC are welcome on our developer
5329 list at [9]gcc@gcc.gnu.org. All of [10]our lists have public
5332 Copyright (C) [11]Free Software Foundation, Inc. Verbatim copying and
5333 distribution of this entire article is permitted in any medium,
5334 provided this notice is preserved.
5336 These pages are [12]maintained by the GCC team. Last modified
5341 1. http://gcc.gnu.org/gcc-4.1/changes.html#4.1.2
5342 2. http://developer.classpath.org/mediation/ClasspathGraphicsImagesText
5343 3. http://gcc.gnu.org/ml/java/2005-05/msg00260.html
5344 4. http://developer.classpath.org/doc/
5345 5. http://gcc.gnu.org/onlinedocs/gcc-4.1.0/gcc/Atomic-Builtins.html
5346 6. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.1.2
5347 7. http://gcc.gnu.org/onlinedocs/
5348 8. mailto:gcc-help@gcc.gnu.org
5349 9. mailto:gcc@gcc.gnu.org
5350 10. http://gcc.gnu.org/lists.html
5351 11. http://www.fsf.org/
5352 12. http://gcc.gnu.org/about.html
5353 13. http://validator.w3.org/check/referer
5354 ======================================================================
5355 http://gcc.gnu.org/gcc-4.0/index.html
5356 GCC 4.0 Release Series
5360 The [1]GNU project and the GCC developers are pleased to announce the
5361 release of GCC 4.0.4.
5363 This release is a bug-fix release, containing fixes for regressions in
5364 GCC 4.0.3 relative to previous releases of GCC.
5369 January 31, 2007 ([2]changes)
5372 March 10, 2006 ([3]changes)
5375 September 28, 2005 ([4]changes)
5378 July 7, 2005 ([5]changes)
5381 April 20, 2005 ([6]changes)
5383 References and Acknowledgements
5385 GCC used to stand for the GNU C Compiler, but since the compiler
5386 supports several other languages aside from C, it now stands for the
5387 GNU Compiler Collection.
5389 A list of [7]successful builds is updated as new information becomes
5392 The GCC developers would like to thank the numerous people that have
5393 contributed new features, improvements, bug fixes, and other changes as
5394 well as test results to GCC. This [8]amazing group of volunteers is
5395 what makes GCC successful.
5397 For additional information about GCC please refer to the [9]GCC project
5398 web site or contact the [10]GCC development mailing list.
5400 To obtain GCC please use [11]our mirror sites, or [12]our SVN server.
5403 For questions related to the use of GCC, please consult these web
5404 pages and the [13]GCC manuals. If that fails, the
5405 [14]gcc-help@gcc.gnu.org mailing list might help. Comments on these
5406 web pages and the development of GCC are welcome on our developer
5407 list at [15]gcc@gcc.gnu.org. All of [16]our lists have public
5410 Copyright (C) [17]Free Software Foundation, Inc. Verbatim copying and
5411 distribution of this entire article is permitted in any medium,
5412 provided this notice is preserved.
5414 These pages are [18]maintained by the GCC team. Last modified
5419 1. http://www.gnu.org/
5420 2. http://gcc.gnu.org/gcc-4.0/changes.html#4.0.4
5421 3. http://gcc.gnu.org/gcc-4.0/changes.html#4.0.3
5422 4. http://gcc.gnu.org/gcc-4.0/changes.html#4.0.2
5423 5. http://gcc.gnu.org/gcc-4.0/changes.html#4.0.1
5424 6. http://gcc.gnu.org/gcc-4.0/changes.html
5425 7. http://gcc.gnu.org/gcc-4.0/buildstat.html
5426 8. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
5427 9. http://gcc.gnu.org/index.html
5428 10. mailto:gcc@gcc.gnu.org
5429 11. http://gcc.gnu.org/mirrors.html
5430 12. http://gcc.gnu.org/svn.html
5431 13. http://gcc.gnu.org/onlinedocs/
5432 14. mailto:gcc-help@gcc.gnu.org
5433 15. mailto:gcc@gcc.gnu.org
5434 16. http://gcc.gnu.org/lists.html
5435 17. http://www.fsf.org/
5436 18. http://gcc.gnu.org/about.html
5437 19. http://validator.w3.org/check/referer
5438 ======================================================================
5439 http://gcc.gnu.org/gcc-4.0/changes.html
5440 GCC 4.0 Release Series
5441 Changes, New Features, and Fixes
5443 The latest release in the 4.0 release series is [1]GCC 4.0.4.
5447 * GCC now generates location lists by default when compiling with
5448 debug info and optimization.
5449 + GDB 6.0 and older crashes when it sees location lists. GDB 6.1
5450 or later is needed to debug binaries containing location
5452 + When you are trying to view a value of a variable in a part of
5453 a function where it has no location (for example when the
5454 variable is no longer used and thus its location was used for
5455 something else) GDB will say that it is not available.
5456 You can disable generating location lists by -fno-var-tracking.
5457 * GCC no longer accepts the -fwritable-strings option. Use named
5458 character arrays when you need a writable string.
5459 * The options -freduce-all-givs and -fmove-all-movables have been
5460 discontinued. They were used to circumvent a shortcoming in the
5461 heuristics of the old loop optimization code with respect to common
5462 Fortran constructs. The new (tree) loop optimizer works differently
5463 and doesn't need those work-arounds.
5464 * The graph-coloring register allocator, formerly enabled by the
5465 option -fnew-ra, has been discontinued.
5466 * -I- has been deprecated. -iquote is meant to replace the need for
5468 * The MIPS -membedded-pic and -mrnames options have been removed.
5469 * All MIPS targets now require the GNU assembler. In particular, IRIX
5470 configurations can no longer use the MIPSpro assemblers, although
5471 they do still support the MIPSpro linkers.
5472 * The SPARC option -mflat has been removed.
5473 * English-language diagnostic messages will now use Unicode quotation
5474 marks in UTF-8 locales. (Non-English messages already used the
5475 quotes appropriate for the language in previous releases.) If your
5476 terminal does not support UTF-8 but you are using a UTF-8 locale
5477 (such locales are the default on many GNU/Linux systems) then you
5478 should set LC_CTYPE=C in the environment to disable that locale.
5479 Programs that parse diagnostics and expect plain ASCII
5480 English-language messages should set LC_ALL=C. See [2]Markus Kuhn's
5481 explanation of Unicode quotation marks for more information.
5482 * The specs file is no longer installed on most platforms. Most users
5483 will be totally unaffected. However, if you are accustomed to
5484 editing the specs file yourself, you will now have to use the
5485 -dumpspecs option to generate the specs file, and then edit the
5488 General Optimizer Improvements
5490 * The [3]tree ssa branch has been merged. This merge has brought in a
5491 completely new optimization framework based on a higher level
5492 intermediate representation than the existing RTL representation.
5493 Numerous new code transformations based on the new framework are
5494 available in GCC 4.0, including:
5495 + Scalar replacement of aggregates
5496 + Constant propagation
5497 + Value range propagation
5498 + Partial redundancy elimination
5499 + Load and store motion
5500 + Strength reduction
5501 + Dead store elimination
5502 + Dead and unreachable code elimination
5503 + [4]Autovectorization
5505 + Tail recursion by accumulation
5506 Many of these passes outperform their counterparts from previous
5508 * [5]Swing Modulo Scheduling (SMS). An RTL level instruction
5509 scheduling optimization intended for loops that perform heavy
5512 New Languages and Language specific improvements
5516 * The sentinel attribute has been added to GCC. This function
5517 attribute allows GCC to warn when variadic functions such as execl
5518 are not NULL terminated. See the GCC manual for a complete
5519 description of its behavior.
5520 * Given __attribute__((alias("target"))) it is now an error if target
5521 is not a symbol, defined in the same translation unit. This also
5522 applies to aliases created by #pragma weak alias=target. This is
5523 because it's meaningless to define an alias to an undefined symbol.
5524 On Solaris, the native assembler would have caught this error, but
5529 * The -Wstrict-aliasing=2 option has been added. This warning catches
5530 all unsafe cases, but it may also give a warning for some cases
5532 * The cast-as-lvalue, conditional-expression-as-lvalue and
5533 compound-expression-as-lvalue extensions, which were deprecated in
5534 3.3.4 and 3.4, have been removed.
5535 * The -fwritable-strings option, which was deprecated in 3.4, has
5537 * #pragma pack() semantics have been brought closer to those used by
5538 other compilers. This also applies to C++.
5539 * Taking the address of a variable with register storage is invalid
5540 in C. GCC now issues an error instead of a warning.
5541 * Arrays of incomplete element type are invalid in C. GCC now issues
5542 an error for such arrays. Declarations such as extern struct s x[];
5543 (where struct s has not been defined) can be moved after the
5544 definition of struct s. Function parameters declared as arrays of
5545 incomplete type can instead be declared as pointers.
5549 * When compiling without optimizations (-O0), the C++ frontend is
5550 much faster than in any previous versions of GCC. Independent
5551 testers have measured speed-ups up to 25% in real-world production
5552 code, compared to the 3.4 family (which was already the fastest
5553 version to date). Upgrading from older versions might show even
5554 bigger improvements.
5555 * ELF visibility attributes can now be applied to a class type, so
5556 that it affects every member function of a class at once, without
5557 having to specify each individually:
5558 class __attribute__ ((visibility("hidden"))) Foo
5563 The syntax is deliberately similar to the __declspec() system used
5564 by Microsoft Windows based compilers, allowing cross-platform
5565 projects to easily reuse their existing macro system for denoting
5566 exports and imports. By explicitly marking internal classes never
5567 used outside a binary as hidden, one can completely avoid PLT
5568 indirection overheads during their usage by the compiler. You can
5569 find out more about the advantages of this at
5570 [6]http://www.akkadia.org/drepper/dsohowto.pdf
5571 * The -fvisibility-inlines-hidden option has been added which marks
5572 all inlineable functions as having hidden ELF visibility, thus
5573 removing their symbol and typeinfo from the exported symbol table
5574 of the output ELF binary. Using this option can reduce the exported
5575 symbol count of template-heavy code by up to 40% with no code
5576 change at all, thus notably improving link and load times for the
5577 binary as well as a reduction in size of up to 10%. Also, check the
5578 new [7]-fvisibility option.
5579 * The compiler now uses the library interface specified by the [8]C++
5580 ABI for thread-safe initialization of function-scope static
5581 variables. Most users should leave this alone, but embedded
5582 programmers may want to disable this by specifying
5583 -fno-threadsafe-statics for a small savings in code size.
5584 * Taking the address of an explicit register variable is no longer
5585 supported. Note that C++ allows taking the address of variables
5586 with register storage so this will continue to compile with a
5587 warning. For example, assuming that r0 is a machine register:
5588 register int foo asm ("r0");
5590 &foo; // error, no longer accepted
5591 &bar; // OK, with a warning
5592 * G++ has an undocumented extension to virtual function covariancy
5593 rules that allowed the overrider to return a type that was
5594 implicitly convertable to the overridden function's return type.
5595 For instance a function returning void * could be overridden by a
5596 function returning T *. This is now deprecated and will be removed
5597 in a future release.
5598 * The G++ minimum and maximum operators (<? and >?) and their
5599 compound forms (<?=) and >?=) have been deprecated and will be
5600 removed in a future version. Code using these operators should be
5601 modified to use std::min and std::max instead.
5602 * Declaration of nested classes of class templates as friends are
5604 template <typename T> struct A {
5608 template <typename T> friend class A<T>::B;
5610 This complements the feature member functions of class templates as
5611 friends introduced in GCC 3.4.0.
5612 * When declaring a friend class using an unqualified name, classes
5613 outside the innermost non-class scope are not searched:
5617 friend class A; // Refer to N::A which has not been declared yet
5618 // because name outside namespace N are not searched
5619 friend class ::A; // Refer to ::A
5622 Hiding the friend name until declaration is still not implemented.
5623 * Friends of classes defined outside their namespace are correctly
5629 friend class B; // Refer to N::B in GCC 4.0.0
5630 // but ::B in earlier versions of GCC
5633 Runtime Library (libstdc++)
5635 * Optimization work:
5636 + Added efficient specializations of istream functions for char
5638 + Further performance tuning of strings, in particular wrt
5639 single-char append and getline.
5640 + iter_swap - and therefore most of the mutating algorithms -
5641 now makes an unqualified call to swap when the value_type of
5642 the two iterators is the same.
5643 * A large subset of the features in Technical Report 1 (TR1 for
5644 short) is experimentally delivered (i.e., no guarantees about the
5645 implementation are provided. In particular it is not promised that
5646 the library will remain link-compatible when code using TR1 is
5648 + General utilities such as reference_wrapper and shared_ptr.
5649 + Function objects, i.e., result_of, mem_fn, bind, function.
5650 + Support for metaprogramming.
5651 + New containers such as tuple, array, unordered_set,
5652 unordered_map, unordered_multiset, unordered_multimap.
5653 * As usual, many bugs have been fixed and LWG resolutions implemented
5654 for the first time (e.g., DR 409).
5658 * In order to prevent naming conflicts with other implementations of
5659 these tools, some GCJ binaries have been renamed:
5660 + rmic is now grmic,
5661 + rmiregistry is now grmiregistry, and
5662 + jar is now fastjar.
5663 In particular, these names were problematic for the jpackage.org
5664 packaging conventions which install symlinks in /usr/bin that point
5665 to the preferred versions of these tools.
5666 * The -findirect-dispatch argument to the compiler now works and
5667 generates code following a new "binary compatibility" ABI. Code
5668 compiled this way follows the binary compatibility rules of the
5669 Java Language Specification.
5670 * libgcj now has support for using GCJ as a JIT, using the
5671 gnu.gcj.jit family of system properties.
5672 * libgcj can now find a shared library corresponding to the bytecode
5673 representation of a class. See the documentation for the new
5674 gcj-dbtool program, and the new gnu.gcj.precompiled.db.path system
5676 * There have been many improvements to the class library. Here are
5678 + Much more of AWT and Swing exist.
5679 + Many new packages and classes were added, including
5680 java.util.regex, java.net.URI, javax.crypto,
5681 javax.crypto.interfaces, javax.crypto.spec, javax.net,
5682 javax.net.ssl, javax.security.auth,
5683 javax.security.auth.callback, javax.security.auth.login,
5684 javax.security.auth.x500, javax.security.sasl, org.ietf.jgss,
5685 javax.imageio, javax.imageio.event, javax.imageio.spi,
5686 javax.print, javax.print.attribute,
5687 javax.print.attribute.standard, javax.print.event, and
5689 + Updated SAX and DOM, and imported GNU JAXP
5693 * A new [9]Fortran front end has replaced the aging GNU Fortran 77
5694 front end. The new front end supports Fortran 90 and Fortran 95. It
5695 may not yet be as stable as the old Fortran front end.
5699 * Ada (with tasking and Zero Cost Exceptions) is now available on
5700 many more targets, including but not limited to: alpha-linux,
5701 hppa-hpux, hppa-linux, powerpc-darwin, powerpc-linux, s390-linux,
5702 s390x-linux, sparc-linux.
5703 * Some of the new Ada 2005 features are now implemented like
5704 Wide_Wide_Character and Ada.Containers.
5705 * Many bugs have been fixed, tools and documentation improved.
5706 * To compile Ada from the sources, install an older working Ada
5707 compiler and then use --enable-languages=ada at configuration time,
5708 since the Ada frontend is not currently activated by default. See
5709 the [10]Installing GCC for details.
5711 New Targets and Target Specific Improvements
5715 * The frame layout has changed. In the new layout, the prologue of a
5716 function first saves registers and then allocate space for locals,
5717 resulting in an 1% improvement on code size.
5719 IA-32/x86-64 (AMD64)
5721 * The acos, asin, drem, exp10, exp2, expm1, fmod, ilogb, log10,
5722 log1p, log2, logb and tan mathematical builtins (and their float
5723 and long double variants) are now implemented as inline x87
5724 intrinsics when using -ffast-math.
5725 * The ceil, floor, nearbyint, rint and trunc mathematical builtins
5726 (and their float and long double variants) are now implemented as
5727 inline x87 intrinsics when using -ffast-math.
5728 * The x87's fsincos instruction is now used automatically with
5729 -ffast-math when calculating both the sin and cos of the same
5731 * Instruction selection for multiplication and division by constants
5736 * Floating point division, integer division and sqrt are now inlined,
5737 resulting in significant performance improvements on some codes.
5741 * Division by zero checks now use conditional traps if the target
5742 processor supports them. This decreases code size by one word per
5743 division operation. The old behavior (branch and break) can be
5744 obtained either at configure time by passing --with-divide=breaks
5745 to configure or at runtime by passing -mdivide-breaks to GCC.
5746 * Support for MIPS64 paired-single instructions has been added. It is
5747 enabled by -mpaired-single and can be accessed using both the
5748 target-independent vector extensions and new MIPS-specific built-in
5750 * Support for the MIPS-3D ASE has been added. It is enabled by
5751 -mips3d and provides new MIPS-3D-specific built-in functions.
5752 * The -mexplicit-relocs option now supports static n64 code (as is
5753 used, for example, in 64-bit linux kernels). -mexplicit-relocs
5754 should now be feature-complete and is enabled by default when GCC
5755 is configured to use a compatible assembler.
5756 * Support for the NEC VR4130 series has been added. This support
5757 includes the use of VR-specific instructions and a new VR4130
5758 scheduler. Full VR4130 support can be selected with -march=vr4130
5759 while code for any ISA can be tuned for the VR4130 using
5760 -mtune=vr4130. There is also a new -mvr4130-align option that
5761 produces better schedules at the cost of increased code size.
5762 * Support for the Broadcom SB-1 has been extended. There is now an
5763 SB-1 scheduler as well as support for the SB-1-specific
5764 paired-single instructions. Full SB-1 support can be selected with
5765 -march=sb1 while code for any ISA can be optimized for the SB-1
5767 * The compiler can now work around errata in R4000, R4400, VR4120 and
5768 VR4130 processors. These workarounds are enabled by -mfix-r4000,
5769 -mfix-r4400, -mfix-vr4120 and -mfix-vr4130 respectively. The VR4120
5770 and VR4130 workarounds need binutils 2.16 or above.
5771 * IRIX shared libraries are now installed into the standard library
5772 directories: o32 libraries go into lib/, n32 libraries go into
5773 lib32/ and n64 libraries go into lib64/.
5774 * The compiler supports a new -msym32 option. It can be used to
5775 optimize n64 code in which all symbols are known to have 32-bit
5780 * New command-line options help to generate code intended to run in
5781 an environment where stack space is restricted, e.g. Linux kernel
5783 + -mwarn-framesize and -mwarn-dynamicstack trigger compile-time
5784 warnings for single functions that require large or dynamic
5786 + -mstack-size and -mstack-guard generate code that checks for
5787 stack overflow at run time.
5788 + -mpacked-stack generates code that reduces the stack frame
5789 size of many functions by reusing unneeded parts of the stack
5791 * The -msoft-float option now ensures that generated code never
5792 accesses floating point registers.
5793 * The s390x-ibm-tpf target now fully supports C++, including
5794 exceptions and threads.
5795 * Various changes to improve performance of the generated code have
5796 been implemented, including:
5797 + GCC now uses sibling calls where possible.
5798 + Condition code handling has been optimized, allowing GCC to
5799 omit redundant comparisons in certain cases.
5800 + The cost function guiding many optimizations has been refined
5801 to more accurately represent the z900 and z990 processors.
5802 + The ADD LOGICAL WITH CARRY and SUBTRACT LOGICAL WITH BORROW
5803 instructions are now used to avoid conditional branches in
5805 + The back end now uses the LEGITIMIZE_RELOAD_ADDRESS feature to
5806 optimize address arithmetic required to access large stack
5808 + GCC now makes more efficient use of memory-to-memory type
5809 instructions (MVC, CLC, ...).
5810 + More precise tracking of special register use allows better
5811 instruction scheduling, in particular of the function prologue
5812 and epilogue sequences.
5813 + The Java front end now generates inline code to implement
5814 integer division, instead of calling library routines.
5818 * The options -mv8, -msparclite, -mcypress, -msupersparc, -mf930 and
5819 -mf934 have been removed. They have been replaced with -mcpu=xxx.
5820 * The internal model used to estimate the relative cost of each
5821 instruction has been updated. It is expected to give better results
5822 on recent UltraSPARC processors.
5823 * Code generation for function prologues and epilogues has been
5824 improved, resulting in better scheduling and allowing multiple exit
5825 points in functions.
5826 * Support for Sun's Visual Instruction Set (VIS) has been enhanced.
5827 It is enabled by -mvis and provides new built-in functions for VIS
5828 instructions on UltraSPARC processors.
5829 * The option -mapp-regs has been turned on by default on Solaris too.
5833 * Novell NetWare (on ix86, no other hardware platform was ever really
5834 supported by this OS) has been re-enabled and the ABI supported by
5835 GCC has been brought into sync with that of MetroWerks CodeWarrior
5836 (the ABI previously supported was that of some Unix systems, which
5837 NetWare never tried to support).
5841 Support for a number of older systems has been declared obsolete in GCC
5842 4.0. Unless there is activity to revive them, the next release of GCC
5843 will have their sources permanently removed.
5845 All GCC ports for the following processor architectures have been
5849 * National Semiconductor NS32K
5850 * Texas Instruments TMS320C[34]x
5852 Also, those for some individual systems have been obsoleted:
5854 + SPARClite-based systems (sparclite-*-coff, sparclite-*-elf,
5856 + OpenBSD 32-bit (sparc-*-openbsd*)
5858 Documentation improvements
5860 Other significant improvements
5862 * Location lists are now generated by default when compiling with
5863 debug info and optimization. Location lists provide more accurate
5864 debug info about locations of variables and they allow debugging
5865 code compiled with -fomit-frame-pointer.
5866 * The -fvisibility option has been added which allows the default ELF
5867 visibility of all symbols to be set per compilation and the new
5868 #pragma GCC visibility preprocessor command allows the setting of
5869 default ELF visibility for a region of code. Using
5870 -fvisibility=hidden especially in combination with the new
5871 -fvisibility-inlines-hidden can yield substantial improvements in
5872 output binary quality including avoiding PLT indirection overheads,
5873 reduction of the exported symbol count by up to 60% (with resultant
5874 improvements to link and load times), better scope for the
5875 optimizer to improve code and up to a 20% reduction in binary size.
5876 Using these options correctly yields a binary with a similar symbol
5877 count to a Windows DLL.
5878 Perhaps more importantly, this new feature finally allows (with
5879 careful planning) complete avoidance of symbol clashes when
5880 manually loading shared objects with RTLD_GLOBAL, thus finally
5881 solving problems many projects such as python were forced to use
5882 RTLD_LOCAL for (with its resulting issues for C++ correctness). You
5883 can find more information about using these options at
5884 [11]http://gcc.gnu.org/wiki/Visibility.
5885 __________________________________________________________________
5889 This is the [12]list of problem reports (PRs) from GCC's bug tracking
5890 system that are known to be fixed in the 4.0.1 release. This list might
5891 not be complete (that is, it is possible that some PRs that have been
5892 fixed are not listed here).
5896 This is the [13]list of problem reports (PRs) from GCC's bug tracking
5897 system that are known to be fixed in the 4.0.2 release. This list might
5898 not be complete (that is, it is possible that some PRs that have been
5899 fixed are not listed here).
5901 Unfortunately, due to a release engineering failure, this release has a
5902 regression on Solaris that will affect some C++ programs. We suggest
5903 that Solaris users apply a [14]patch that corrects the problem. Users
5904 who do not wish to apply the patch should explicitly link C++ programs
5905 with the -pthreads option, even if they do not use threads. This
5906 problem has been corrected in the current 4.0 branch sources and will
5907 not be present in GCC 4.0.3.
5911 Starting with this release, the function getcontext is recognized by
5912 the compiler as having the same semantics as the setjmp function. In
5913 particular, the compiler will ensure that all registers are dead before
5914 calling such a function and will emit a warning about the variables
5915 that may be clobbered after the second return from the function.
5919 This is the [15]list of problem reports (PRs) from GCC's bug tracking
5920 system that are known to be fixed in the 4.0.4 release. This list might
5921 not be complete (that is, it is possible that some PRs that have been
5922 fixed are not listed here).
5924 The 4.0.4 release is provided for those that require a high degree of
5925 binary compatibility with previous 4.0.x releases. For most users, the
5926 GCC team recommends that version 4.1.1 or later be used instead."
5929 For questions related to the use of GCC, please consult these web
5930 pages and the [16]GCC manuals. If that fails, the
5931 [17]gcc-help@gcc.gnu.org mailing list might help. Comments on these
5932 web pages and the development of GCC are welcome on our developer
5933 list at [18]gcc@gcc.gnu.org. All of [19]our lists have public
5936 Copyright (C) [20]Free Software Foundation, Inc. Verbatim copying and
5937 distribution of this entire article is permitted in any medium,
5938 provided this notice is preserved.
5940 These pages are [21]maintained by the GCC team. Last modified
5945 1. http://gcc.gnu.org/gcc-4.0/changes.html#4.0.4
5946 2. http://www.cl.cam.ac.uk/~mgk25/ucs/quotes.html
5947 3. http://gcc.gnu.org/projects/tree-ssa/
5948 4. http://gcc.gnu.org/projects/tree-ssa/vectorization.html
5949 5. http://gcc.gnu.org/news/sms.html
5950 6. http://www.akkadia.org/drepper/dsohowto.pdf
5951 7. http://gcc.gnu.org/gcc-4.0/changes.html#visibility
5952 8. http://mentorembedded.github.com/cxx-abi/
5953 9. http://gcc.gnu.org/fortran/
5954 10. http://gcc.gnu.org/install/
5955 11. http://gcc.gnu.org/wiki/Visibility
5956 12. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.0.1
5957 13. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.0.2
5958 14. http://gcc.gnu.org/ml/gcc-cvs/2005-09/msg00984.html
5959 15. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.0.4
5960 16. http://gcc.gnu.org/onlinedocs/
5961 17. mailto:gcc-help@gcc.gnu.org
5962 18. mailto:gcc@gcc.gnu.org
5963 19. http://gcc.gnu.org/lists.html
5964 20. http://www.fsf.org/
5965 21. http://gcc.gnu.org/about.html
5966 22. http://validator.w3.org/check/referer
5967 ======================================================================
5968 http://gcc.gnu.org/gcc-3.4/index.html
5969 GCC 3.4 Release Series
5973 The [1]GNU project and the GCC developers are pleased to announce the
5974 release of GCC 3.4.6.
5976 This release is a bug-fix release, containing fixes for regressions in
5977 GCC 3.4.4 relative to previous releases of GCC. This is the last of the
5980 The GCC 3.4 release series includes numerous [2]new features,
5981 improvements, bug fixes, and other changes, thanks to an [3]amazing
5982 group of volunteers.
5987 March 6, 2006 ([4]changes)
5990 November 30, 2005 ([5]changes)
5993 May 18, 2005 ([6]changes)
5996 November 4, 2004 ([7]changes)
5999 September 6, 2004 ([8]changes)
6002 July 1, 2004 ([9]changes)
6005 April 18, 2004 ([10]changes)
6007 References and Acknowledgements
6009 GCC used to stand for the GNU C Compiler, but since the compiler
6010 supports several other languages aside from C, it now stands for the
6011 GNU Compiler Collection.
6013 A list of [11]successful builds is updated as new information becomes
6016 The GCC developers would like to thank the numerous people that have
6017 contributed new features, improvements, bug fixes, and other changes as
6018 well as test results to GCC. This [12]amazing group of volunteers is
6019 what makes GCC successful.
6021 For additional information about GCC please refer to the [13]GCC
6022 project web site or contact the [14]GCC development mailing list.
6024 To obtain GCC please use [15]our mirror sites, or [16]our SVN server.
6027 For questions related to the use of GCC, please consult these web
6028 pages and the [17]GCC manuals. If that fails, the
6029 [18]gcc-help@gcc.gnu.org mailing list might help. Comments on these
6030 web pages and the development of GCC are welcome on our developer
6031 list at [19]gcc@gcc.gnu.org. All of [20]our lists have public
6034 Copyright (C) [21]Free Software Foundation, Inc. Verbatim copying and
6035 distribution of this entire article is permitted in any medium,
6036 provided this notice is preserved.
6038 These pages are [22]maintained by the GCC team. Last modified
6043 1. http://www.gnu.org/
6044 2. http://gcc.gnu.org/gcc-3.4/changes.html
6045 3. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
6046 4. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.6
6047 5. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.5
6048 6. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.4
6049 7. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.3
6050 8. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.2
6051 9. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.1
6052 10. http://gcc.gnu.org/gcc-3.4/changes.html
6053 11. http://gcc.gnu.org/gcc-3.4/buildstat.html
6054 12. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
6055 13. http://gcc.gnu.org/index.html
6056 14. mailto:gcc@gcc.gnu.org
6057 15. http://gcc.gnu.org/mirrors.html
6058 16. http://gcc.gnu.org/svn.html
6059 17. http://gcc.gnu.org/onlinedocs/
6060 18. mailto:gcc-help@gcc.gnu.org
6061 19. mailto:gcc@gcc.gnu.org
6062 20. http://gcc.gnu.org/lists.html
6063 21. http://www.fsf.org/
6064 22. http://gcc.gnu.org/about.html
6065 23. http://validator.w3.org/check/referer
6066 ======================================================================
6067 http://gcc.gnu.org/gcc-3.4/changes.html
6068 GCC 3.4 Release Series
6069 Changes, New Features, and Fixes
6071 The final release in the 3.4 release series is [1]GCC 3.4.6. The series
6074 GCC 3.4 has [2]many improvements in the C++ frontend. Before reporting
6075 a bug, please make sure it's really GCC, and not your code, that is
6080 * GNU Make is now required to build GCC.
6081 * With -nostdinc the preprocessor used to ignore both standard
6082 include paths and include paths contained in environment variables.
6083 It was neither documented nor intended that environment variable
6084 paths be ignored, so this has been corrected.
6085 * GCC no longer accepts the options -fvolatile, -fvolatile-global and
6086 -fvolatile-static. It is unlikely that they worked correctly in any
6088 * GCC no longer ships <varargs.h>. Use <stdarg.h> instead.
6089 * Support for all the systems [3]obsoleted in GCC 3.3 has been
6090 removed from GCC 3.4. See below for a [4]list of systems which are
6091 obsoleted in this release.
6092 * GCC now requires an ISO C90 (ANSI C89) C compiler to build. K&R C
6093 compilers will not work.
6094 * The implementation of the [5]MIPS ABIs has changed. As a result,
6095 the code generated for certain MIPS targets will not be binary
6096 compatible with earlier releases.
6097 * In previous releases, the MIPS port had a fake "hilo" register with
6098 the user-visible name accum. This register has been removed.
6099 * The implementation of the [6]SPARC ABIs has changed. As a result,
6100 the code generated will not be binary compatible with earlier
6101 releases in certain cases.
6102 * The configure option --enable-threads=pthreads has been removed;
6103 use --enable-threads=posix instead, which should have the same
6105 * Code size estimates used by inlining heuristics for C, Objective-C,
6106 C++ and Java have been redesigned significantly. As a result the
6107 parameters of -finline-insns, --param max-inline-insns-single and
6108 --param max-inline-insns-auto need to be reconsidered.
6109 * --param max-inline-slope and --param min-inline-insns have been
6110 removed; they are not needed for the new bottom-up inlining
6112 * The new unit-at-a-time compilation scheme has several compatibility
6114 + The order in which functions, variables, and top-level asm
6115 statements are emitted may have changed. Code relying on some
6116 particular ordering needs to be updated. The majority of such
6117 top-level asm statements can be replaced by section
6119 + Unreferenced static variables and functions are removed. This
6120 may result in undefined references when an asm statement
6121 refers to the variable/function directly. In that case either
6122 the variable/function shall be listed in asm statement operand
6123 or in the case of top-level asm statements the attribute used
6124 shall be used to force function/variable to be always output
6125 and considered as a possibly used by unknown code.
6126 For variables the attribute is accepted only by GCC 3.4 and
6127 newer, while for earlier versions it is sufficient to use
6128 unused to silence warnings about the variables not being
6129 referenced. To keep code portable across different GCC
6130 versions, you can use appropriate preprocessor conditionals.
6131 + Static functions now can use non-standard passing conventions
6132 that may break asm statements calling functions directly.
6133 Again the attribute used shall be used to prevent this
6135 As a temporary workaround, -fno-unit-at-a-time can be used, but
6136 this scheme may not be supported by future releases of GCC.
6137 * GCC 3.4 automatically places zero-initialized variables in the .bss
6138 section on some operating systems. Versions of GNU Emacs up to (and
6139 including) 21.3 will not work correctly when using this
6140 optimization; you can use -fno-zero-initialized-in-bss to disable
6142 * If GCC 3.4 is configured with --enable-threads=posix (the default
6143 on most targets that support pthreads) then _REENTRANT will be
6144 defined unconditionally by some libstdc++ headers. C++ code which
6145 relies on that macro to detect whether multi-threaded code is being
6146 compiled might change in meaning, possibly resulting in linker
6147 errors for single-threaded programs. Affected users of [7]Boost
6148 should compile single-threaded code with -DBOOST_DISABLE_THREADS.
6149 See Bugzilla for [8]more information.
6151 General Optimizer Improvements
6153 * Usability of the profile feedback and coverage testing has been
6155 + Performance of profiled programs has been improved by faster
6156 profile merging code.
6157 + Better use of the profile feedback for optimization (loop
6158 unrolling and loop peeling).
6159 + File locking support allowing fork() calls and parallel runs
6160 of profiled programs.
6161 + Coverage file format has been redesigned.
6162 + gcov coverage tool has been improved.
6163 + make profiledbootstrap available to build a faster compiler.
6164 Experiments made on i386 hardware showed an 11% speedup on -O0
6165 and a 7.5% speedup on -O2 compilation of a [9]large C++
6167 + New value profiling pass enabled via -fprofile-values
6168 + New value profile transformations pass enabled via -fvpt aims
6169 to optimize some code sequences by exploiting knowledge about
6170 value ranges or other properties of the operands. At the
6171 moment a conversion of expensive divisions into cheaper
6172 operations has been implemented.
6173 + New -fprofile-generate and -fprofile-use command-line options
6174 to simplify the use of profile feedback.
6175 * A new unit-at-a-time compilation scheme for C, Objective-C, C++ and
6176 Java which is enabled via -funit-at-a-time (and implied by -O2). In
6177 this scheme a whole file is parsed first and optimized later. The
6178 following basic inter-procedural optimizations are implemented:
6179 + Removal of unreachable functions and variables
6180 + Discovery of local functions (functions with static linkage
6181 whose address is never taken)
6182 + On i386, these local functions use register parameter passing
6184 + Reordering of functions in topological order of the call graph
6185 to enable better propagation of optimizing hints (such as the
6186 stack alignments needed by functions) in the back end.
6187 + Call graph based out-of-order inlining heuristics which allows
6188 to limit overall compilation unit growth (--param
6189 inline-unit-growth).
6190 Overall, the unit-at-a-time scheme produces a 1.3% improvement for
6191 the SPECint2000 benchmark on the i386 architecture (AMD Athlon
6193 * More realistic code size estimates used by inlining for C,
6194 Objective-C, C++ and Java. The growth of large functions can now be
6195 limited via --param large-function-insns and --param
6196 large-function-growth.
6197 * A new cfg-level loop optimizer pass replaces the old loop unrolling
6198 pass and adds two other loop transformations -- loop peeling and
6199 loop unswitching -- and also uses the profile feedback to limit
6200 code growth. (The three optimizations are enabled by
6201 -funroll-loops, -fpeel-loops and -funswitch-loops flags,
6203 The old loop unroller still can be enabled by -fold-unroll-loops
6204 and may produce better code in some cases, especially when the
6205 webizer optimization pass is not run.
6206 * A new web construction pass enabled via -fweb (and implied by -O3)
6207 improves the quality of register allocation, CSE, first scheduling
6208 pass and some other optimization passes by avoiding re-use of
6209 pseudo registers with non-overlapping live ranges. The pass almost
6210 always improves code quality but does make debugging difficult and
6211 thus is not enabled by default by -O2
6212 The pass is especially effective as cleanup after code duplication
6213 passes, such as the loop unroller or the tracer.
6214 * Experimental implementations of superblock or trace scheduling in
6215 the second scheduling pass can be enabled via
6216 -fsched2-use-superblocks and -fsched2-use-traces, respectively.
6218 New Languages and Language specific improvements
6222 * The Ada front end has been updated to include numerous bug fixes
6223 and enhancements. These include:
6224 + Improved project file support
6225 + Additional set of warnings about potential wrong code
6226 + Improved error messages
6227 + Improved code generation
6228 + Improved cross reference information
6230 + Better run-time check elimination
6231 + Better error recovery
6232 + More efficient implementation of unbounded strings
6233 + Added features in GNAT.Sockets, GNAT.OS_Lib, GNAT.Debug_Pools,
6235 + New GNAT.xxxx packages (e.g. GNAT.Strings,
6236 GNAT.Exception_Action)
6238 + New -gnatS switch replacing gnatpsta
6239 + Implementation of new Ada features (in particular limited
6240 with, limited aggregates)
6244 * Precompiled headers are now supported. Precompiled headers can
6245 dramatically speed up compilation of some projects. There are some
6246 known defects in the current precompiled header implementation that
6247 will result in compiler crashes in relatively rare situations.
6248 Therefore, precompiled headers should be considered a "technology
6249 preview" in this release. Read the manual for details about how to
6250 use precompiled headers.
6251 * File handling in the preprocessor has been rewritten. GCC no longer
6252 gets confused by symlinks and hardlinks, and now has a correct
6253 implementation of #import and #pragma once. These two directives
6254 have therefore been un-deprecated.
6255 * The undocumented extension that allowed C programs to have a label
6256 at the end of a compound statement, which has been deprecated since
6257 GCC 3.0, has been removed.
6258 * The cast-as-lvalue extension has been removed for C++ and
6259 deprecated for C and Objective-C. In particular, code like this:
6267 is no longer accepted for C++ and will not be accepted for C and
6268 Objective-C in a future version.
6269 * The conditional-expression-as-lvalue extension has been deprecated
6270 for C and Objective-C. In particular, code like this:
6274 will not be accepted for C and Objective-C in a future version.
6275 * The compound-expression-as-lvalue extension has been deprecated for
6276 C and Objective-C. In particular, code like this:
6280 will not be accepted for C and Objective-C in a future version. A
6281 possible non-intrusive workaround is the following:
6284 * Several [10]built-in functions such as __builtin_popcount for
6285 counting bits, finding the highest and lowest bit in a word, and
6286 parity have been added.
6287 * The -fwritable-strings option has been deprecated and will be
6289 * Many C math library functions are now recognized as built-ins and
6291 * The C, C++, and Objective-C compilers can now handle source files
6292 written in any character encoding supported by the host C library.
6293 The default input character set is taken from the current locale,
6294 and may be overridden with the -finput-charset command line option.
6295 In the future we will add support for inline encoding markers.
6299 * G++ is now much closer to full conformance to the ISO/ANSI C++
6300 standard. This means, among other things, that a lot of invalid
6301 constructs which used to be accepted in previous versions will now
6302 be rejected. It is very likely that existing C++ code will need to
6303 be fixed. This document lists some of the most common issues.
6304 * A hand-written recursive-descent C++ parser has replaced the
6305 YACC-derived C++ parser from previous GCC releases. The new parser
6306 contains much improved infrastructure needed for better parsing of
6307 C++ source codes, handling of extensions, and clean separation
6308 (where possible) between proper semantics analysis and parsing. The
6309 new parser fixes many bugs that were found in the old parser.
6310 * You must now use the typename and template keywords to disambiguate
6311 dependent names, as required by the C++ standard.
6313 typedef int mytype_t;
6316 template <class T1> struct A {
6317 template <class T2> struct B {
6321 template <int N> void bar(void)
6323 // Use 'typename' to tell the parser that T1::mytype_t names
6324 // a type. This is needed because the name is dependent (in
6325 // this case, on template parameter T1).
6326 typename T1::mytype_t x;
6331 template <class T> void template_func(void)
6333 // Use 'template' to prefix member templates within
6334 // dependent types (a has type A<T>, which depends on
6335 // the template parameter T).
6337 a.template bar<0>();
6339 // Use 'template' to tell the parser that B is a nested
6340 // template class (dependent on template parameter T), and
6341 // 'typename' because the whole A<T>::B<int> is
6342 // the name of a type (again, dependent).
6343 typename A<T>::template B<int> b;
6347 void non_template_func(void)
6349 // Outside of any template class or function, no names can be
6350 // dependent, so the use of the keyword 'typename' and 'template'
6351 // is not needed (and actually forbidden).
6357 * In a template definition, unqualified names will no longer find
6358 members of a dependent base (as specified by [temp.dep]/3 in the
6359 C++ standard). For example,
6360 template <typename T> struct B {
6368 template <typename T> struct C : B<T> {
6373 n = 0; // ::n is modified
6374 g (); // ::g is called
6377 You must make the names dependent, e.g. by prefixing them with
6378 this->. Here is the corrected definition of C<T>::h,
6379 template <typename T> void C<T>::h ()
6386 As an alternative solution (unfortunately not backwards compatible
6387 with GCC 3.3), you may use using declarations instead of this->:
6388 template <typename T> struct C : B<T> {
6401 * In templates, all non-dependent names are now looked up and bound
6402 at definition time (while parsing the code), instead of later when
6403 the template is instantiated. For instance:
6406 template <int> struct A {
6407 static void bar(void){
6416 A<0>::bar(); // Calls foo(int), used to call foo(char).
6419 * In an explicit instantiation of a class template, you must use
6420 class or struct before the template-id:
6424 template A<0>; // error, not accepted anymore
6425 template class A<0>; // OK
6426 * The "named return value" and "implicit typename" extensions have
6428 * Default arguments in function types have been deprecated and will
6430 * ARM-style name-injection of friend declarations has been deprecated
6431 and will be removed. For example: struct S { friend void f(); };
6432 void g() { f(); } will not be accepted by future versions of G++;
6433 instead a declaration of "f" will need to be present outside of the
6435 * Covariant returns are implemented for all but varadic functions
6436 that require an adjustment.
6437 * When -pedantic is used, G++ now issues errors about spurious
6438 semicolons. For example,
6439 namespace N {}; // Invalid semicolon.
6440 void f() {}; // Invalid semicolon.
6441 * G++ no longer accepts attributes for a declarator after the
6442 initializer associated with that declarator. For example,
6443 X x(1) __attribute__((...));
6444 is no longer accepted. Instead, use:
6445 X x __attribute__((...)) (1);
6446 * Inside the scope of a template class, the name of the class itself
6447 can be treated as either a class or a template. So GCC used to
6448 accept the class name as argument of type template, and template
6449 template parameter. However this is not C++ standard compliant. Now
6450 the name is not treated as a valid template template argument
6451 unless you qualify the name by its scope. For example, the code
6452 below no longer compiles.
6453 template <template <class> class TT> class X {};
6454 template <class T> class Y {
6455 X<Y> x; // Invalid, Y is always a type template parameter.
6457 The valid code for the above example is
6458 X< ::Y> x; // Valid.
6459 (Notice the space between < and : to prevent GCC to interpret this
6460 as a digraph for [.)
6461 * Friend declarations that refer to template specializations are
6462 rejected if the template has not already been declared. For
6464 template <typename T>
6466 friend void f<> (C&);
6468 is rejected. You must first declare f as a template,
6469 template <typename T>
6471 * In case of friend declarations, every name used in the friend
6472 declaration must be accessible at the point of that declaration.
6473 Previous versions of G++ used to be less strict about this and
6474 allowed friend declarations for private class members, for example.
6475 See the ISO C++ Standard Committee's [11]defect report #209 for
6477 * Declaration of member functions of class templates as friends are
6478 supported. For example,
6479 template <typename T> struct A {
6483 template <typename T> friend void A<T>::f();
6485 * You must use template <> to introduce template specializations, as
6486 required by the standard. For example,
6487 template <typename T>
6491 is rejected. You must write,
6492 template <> struct S<int> {};
6493 * G++ used to accept code like this,
6496 void f(int i = g());
6499 This behavior is not mandated by the standard. Now G++ issues an
6500 error about this code. To avoid the error, you must move the
6501 declaration of g before the declaration of f. The default arguments
6502 for g must be visible at the point where it is called.
6503 * The C++ ABI Section 3.3.3 specifications for the array construction
6504 routines __cxa_vec_new2 and __cxa_vec_new3 were changed to return
6505 NULL when the allocator argument returns NULL. These changes are
6506 incorporated into the libstdc++ runtime library.
6507 * Using a name introduced by a typedef in a friend declaration or in
6508 an explicit instantiation is now rejected, as specified by the ISO
6513 friend class B; // error, no typedef name here
6514 friend B; // error, friend always needs class/struct/enum
6515 friend class A; // OK
6518 template <int> class Q {};
6520 template class R; // error, no typedef name here
6521 template class Q<0>; // OK
6522 * When allocating an array with a new expression, GCC used to allow
6523 parentheses around the type name. This is actually ill-formed and
6525 int* a = new (int)[10]; // error, not accepted anymore
6526 int* a = new int[10]; // OK
6527 * When binding an rvalue of class type to a reference, the copy
6528 constructor of the class must be accessible. For instance, consider
6536 A(const A&); // private copy ctor
6544 foo(A()); // error, copy ctor is not accessible
6545 foo(makeA()); // error, copy ctor is not accessible
6548 foo(a1); // OK, a1 is a lvalue
6550 This might be surprising at first sight, especially since most
6551 popular compilers do not correctly implement this rule ([12]further
6553 * When forming a pointer to member or a pointer to member function,
6554 access checks for class visibility (public, protected, private) are
6555 now performed using the qualifying scope of the name itself. This
6556 is better explained with an example:
6572 &A::pub_func; // OK, pub_func is accessible through A
6573 &A::prot_func; // error, cannot access prot_func through A
6574 &A::priv_func; // error, cannot access priv_func through A
6576 &B::pub_func; // OK, pub_func is accessible through B
6577 &B::prot_func; // OK, can access prot_func through B (within B)
6578 &B::priv_func; // error, cannot access priv_func through B
6582 Runtime Library (libstdc++)
6584 * Optimization work:
6585 + Streamlined streambuf, filebuf, separate synched with C
6586 Standard I/O streambuf.
6587 + All formatted I/O now uses cached locale information.
6588 + STL optimizations (memory/speed for list, red-black trees as
6589 used by sets and maps).
6590 + More use of GCC builtins.
6591 + String optimizations (avoid contention on
6592 increment/decrement-and-test of the reference count in the
6593 empty-string object, constructor from input_iterators
6595 * Static linkage size reductions.
6596 * Large File Support (files larger than 2 GB on 32-bit systems).
6597 * Wide character and variable encoding filebuf work (UTF-8, Unicode).
6598 * Generic character traits.
6599 * Also support wchar_t specializations on Mac OS 10.3.x, FreeBSD 5.x,
6600 Solaris 2.7 and above, AIX 5.x, Irix 6.5.
6601 * The allocator class is now standard-conformant, and two additional
6602 extension allocators have been added, mt_alloc and
6604 * PCH support: -include bits/stdc++.h (2x compile speedup).
6605 * Rewrote __cxa_demangle with support for C++ style allocators.
6606 * New debug modes for STL containers and iterators.
6607 * Testsuite rewrite: five times as many tests, plus increasingly
6608 sophisticated tests, including I/O, MT, multi-locale, wide and
6610 * Use current versions of GNU "autotools" for build/configuration.
6614 * The Objective-C front end has been updated to include the numerous
6615 bug fixes and enhancements previously available only in Apple's
6616 version of GCC. These include:
6617 + Structured exception (@try... @catch... @finally, @throw) and
6618 synchronization (@synchronized) support. These are accessible
6619 via the -fobjc-exceptions switch; as of this writing, they may
6620 only be used in conjunction with -fnext-runtime on Mac OS X
6621 10.3 and later. See [13]Options Controlling Objective-C
6622 Dialect for more information.
6623 + An overhaul of @encode logic. The C99 _Bool and C++ bool type
6624 may now be encoded as 'B'. In addition, the back-end/codegen
6625 dependencies have been removed.
6626 + An overhaul of message dispatch construction, ensuring that
6627 the various receiver types (and casts thereof) are handled
6628 properly, and that correct diagnostics are issued.
6629 + Support for "Zero-Link" (-fzero-link) and "Fix-and-Continue"
6630 (-freplace-objc-classes) debugging modes, currently available
6631 on Mac OS X 10.3 and later. See [14]Options Controlling
6632 Objective-C Dialect for more information.
6633 + Access to optimized runtime entry points (-fno-nil-receivers )
6634 on the assumption that message receivers are never nil. This
6635 is currently available on Mac OS X 10.3 and later. See
6636 [15]Options Controlling Objective-C Dialect for more
6641 * Compiling a .jar file will now cause non-.class entries to be
6642 automatically compiled as resources.
6643 * libgcj has been ported to Darwin.
6644 * Jeff Sturm has adapted Jan Hubicka's call graph optimization code
6646 * libgcj has a new gcjlib URL type; this lets URLClassLoader load
6647 code from shared libraries.
6648 * libgcj has been much more completely merged with [16]GNU Classpath.
6649 * Class loading is now much more correct; in particular the caller's
6650 class loader is now used when that is required.
6651 * [17]Eclipse 2.x will run out of the box using gij.
6652 * Parts of java.nio have been implemented. Direct and indirect
6653 buffers work, as do fundamental file and socket operations.
6654 * java.awt has been improved, though it is still not ready for
6656 * The HTTP protocol handler now uses HTTP/1.1 and can handle the POST
6658 * The MinGW port has matured. Enhancements include socket timeout
6659 support, thread interruption, improved Runtime.exec() handling and
6660 support for accented characters in filenames.
6664 * Fortran improvements are listed in the [18]Fortran documentation.
6666 New Targets and Target Specific Improvements
6670 * Several [19]built-in functions have been added such as
6671 __builtin_alpha_zap to allow utilizing the more obscure
6672 instructions of the CPU.
6673 * Parameter passing of complex arguments has changed to match the
6674 [20]ABI. This change is incompatible with previous GCC versions,
6675 but does fix compatibility with the Tru64 compiler and several
6676 corner cases where GCC was incompatible with itself.
6680 * Nicolas Pitre has contributed his hand-coded floating-point support
6681 code for ARM. It is both significantly smaller and faster than the
6682 existing C-based implementation, even when building applications
6683 for Thumb. The arm-elf configuration has been converted to use the
6685 * Support for the Intel's iWMMXt architecture, a second generation
6686 XScale processor, has been added. Enabled at run time with the
6687 -mcpu=iwmmxt command line switch.
6688 * A new ARM target has been added: arm-wince-pe. This is similar to
6689 the arm-pe target, but it defaults to using the APCS32 ABI.
6690 * The existing ARM pipeline description has been converted to the use
6691 the [21]DFA processor pipeline model. There is not much change in
6692 code performance, but the description is now [22]easier to
6694 * Support for the Cirrus EP9312 Maverick floating point co-processor
6695 added. Enabled at run time with the -mcpu=ep9312 command line
6696 switch. Note however that the multilibs to support this chip are
6697 currently disabled in gcc/config/arm/t-arm-elf, so if you want to
6698 enable their production you will have to uncomment the entries in
6703 * Support for long long has been added.
6704 * Support for saveall attribute has been added.
6705 * Pavel Pisa contributed hand-written 32-bit-by-32-bit division code
6706 for H8/300H and H8S, which is much faster than the previous
6708 * A lot of small performance improvements.
6710 IA-32/AMD64 (x86-64)
6712 * Tuning for K8 (AMD Opteron/Athlon64) core is available via
6713 -march=k8 and -mcpu=k8.
6714 * Scalar SSE code generation carefully avoids reformatting penalties,
6715 hidden dependencies and minimizes the number of uops generated on
6716 both Intel and AMD CPUs.
6717 * Vector MMX and SSE operands are now passed in registers to improve
6718 performance and match the argument passing convention used by the
6719 Intel C++ Compiler. As a result it is not possible to call
6720 functions accepting vector arguments compiled by older GCC version.
6721 * Conditional jump elimination is now more aggressive on modern CPUs.
6722 * The Athlon ports has been converted to use the DFA processor
6723 pipeline description.
6724 * Optimization of indirect tail calls is now possible in a similar
6725 fashion as direct sibcall optimization.
6726 * Further small performance improvements.
6727 * -m128bit-long-double is now less buggy.
6728 * __float128 support in 64-bit compilation.
6729 * Support for data structures exceeding 2GB in 64-bit mode.
6730 * -mcpu has been renamed to -mtune.
6734 * Tuning code for the Itanium 2 processor has been added. The
6735 generation of code tuned for Itanium 2 (option -mtune=itanium2) is
6736 enabled by default now. To generate code tuned for Itanium 1 the
6737 option -mtune=itanium1 should be used.
6738 * [23]DFA processor pipeline descriptions for the IA-64 processors
6739 have been added. This resulted in about 3% improvement on the
6740 SPECInt2000 benchmark for Itanium 2.
6741 * Instruction bundling for the IA-64 processors has been rewritten
6742 using the DFA pipeline hazard recognizer. It resulted in about 60%
6743 compiler speedup on the SPECInt2000 C programs.
6747 * Support for the M32R/2 processor has been added by Renesas.
6748 * Support for an M32R GNU/Linux target and PIC code generation has
6749 been added by Renesas.
6753 * Bernardo Innocenti (Develer S.r.l.) has contributed the
6754 m68k-uclinux target, based on former work done by Paul Dale
6755 (SnapGear Inc.). Code generation for the ColdFire processors family
6756 has been enhanced and extended to support the MCF 53xx and MCF 54xx
6757 cores, integrating former work done by Peter Barada (Motorola).
6761 Processor-specific changes
6763 * Support for the RM7000 and RM9000 processors has been added. It can
6764 be selected using the -march compiler option and should work with
6765 any MIPS I (mips-*) or MIPS III (mips64-*) configuration.
6766 * Support for revision 2 of the MIPS32 ISA has been added. It can be
6767 selected with the command-line option -march=mips32r2.
6768 * There is a new option, -mfix-sb1, to work around certain SB-1
6773 * It is possible to customize GCC using the following configure-time
6775 + --with-arch, which specifies the default value of the -march
6777 + --with-tune, which specifies the default value of the -mtune
6779 + --with-abi, which specifies the default ABI.
6780 + --with-float=soft, which tells GCC to use software floating
6782 + --with-float=hard, which tells GCC to use hardware floating
6784 * A 64-bit GNU/Linux port has been added. The associated
6785 configurations are mips64-linux-gnu and mips64el-linux-gnu.
6786 * The 32-bit GNU/Linux port now supports Java.
6787 * The IRIX 6 configuration now supports the o32 ABI and will build
6788 o32 multilibs by default. This support is compatible with both
6789 binutils and the SGI tools, but note that several features,
6790 including debugging information and DWARF2 exception handling, are
6791 only available when using the GNU assembler. Use of the GNU
6792 assembler and linker (version 2.15 or above) is strongly
6794 * The IRIX 6 configuration now supports 128-bit long doubles.
6795 * There are two new RTEMS-specific configurations, mips-rtems and
6797 * There are two new *-elf configurations, mipsisa32r2-elf and
6802 * Several [24]ABI bugs have been fixed. Unfortunately, these changes
6803 will break binary compatibility with earlier releases.
6804 * GCC can now use explicit relocation operators when generating
6805 -mabicalls code. This behavior is controlled by -mexplicit-relocs
6806 and can have several performance benefits. For example:
6807 + It allows for more optimization of GOT accesses, including
6808 better scheduling and redundancy elimination.
6809 + It allows sibling calls to be implemented as jumps.
6810 + n32 and n64 leaf functions can use a call-clobbered global
6811 pointer instead of $28.
6812 + The code to set up $gp can be removed from functions that
6814 * A new option, -mxgot, allows the GOT to be bigger than 64k. This
6815 option is equivalent to the assembler's -xgot option and should be
6816 used instead of -Wa,-xgot.
6817 * Frame pointer elimination is now supported when generating 64-bit
6819 * Inline block moves have been optimized to take more account of
6820 alignment information.
6821 * Many internal changes have been made to the MIPS port, mostly aimed
6822 at reducing the reliance on assembler macros.
6826 * GCC 3.4 releases have a number of fixes for PowerPC and PowerPC64
6827 [25]ABI incompatibilities regarding the way parameters are passed
6828 during functions calls. These changes may result in incompatibility
6829 between code compiled with GCC 3.3 and GCC 3.4.
6833 * Support for shared/dylib gcc libraries has been added. It is
6834 enabled by default on powerpc-apple-darwin7.0.0 and up.
6835 * Libgcj is enabled by default. On systems older than
6836 powerpc-apple-darwin7.0.0 you need to install dlcompat.
6837 * 128-bit IBM extended precision format support added for long
6842 * By default, PowerPC64 GNU/Linux now uses natural alignment of
6843 structure elements. The old four byte alignment for double, with
6844 special rules for a struct starting with a double, can be chosen
6845 with -malign-power. This change may result in incompatibility
6846 between code compiled with GCC 3.3 and GCC 3.4.
6847 * -mabi=altivec is now the default rather than -mabi=no-altivec.
6848 * 128-bit IBM extended precision format support added for long
6853 * New command-line options allow to specify the intended execution
6854 environment for generated code:
6855 + -mesa/-mzarch allows to specify whether to generate code
6856 running in ESA/390 mode or in z/Architecture mode (this is
6857 applicable to 31-bit code only).
6858 + -march allows to specify a minimum processor architecture
6859 level (g5, g6, z900, or z990).
6860 + -mtune allows to specify which processor to tune for.
6861 * It is possible to customize GCC using the following configure-time
6863 + --with-mode, which specifies whether to default to assuming
6864 ESA/390 or z/Architecture mode.
6865 + --with-arch, which specifies the default value of the -march
6867 + --with-tune, which specifies the default value of the -mtune
6869 * Support for the z990 processor has been added, and can be selected
6870 using -march=z990 or -mtune=z990. This includes instruction
6871 scheduling tuned for the superscalar instruction pipeline of the
6872 z990 processor as well as support for all new instructions provided
6873 by the long-displacement facility.
6874 * Support to generate 31-bit code optimized for zSeries processors
6875 (running in ESA/390 or in z/Architecture mode) has been added. This
6876 can be selected using -march=z900 and -mzarch respectively.
6877 * Instruction scheduling for the z900 and z990 processors now uses
6878 the DFA pipeline hazard recognizer.
6879 * GCC no longer generates code to maintain a stack backchain,
6880 previously used to generate stack backtraces for debugging
6881 purposes. As replacement that does not incur runtime overhead,
6882 DWARF-2 call frame information is provided by GCC; this is
6883 supported by GDB 6.1. The old behavior can be restored using the
6885 * The stack frame size of functions may now exceed 2 GB in 64-bit
6887 * A port for the 64-bit IBM TPF operating system has been added; the
6888 configuration is s390x-ibm-tpf. This configuration is supported as
6889 cross-compilation target only.
6890 * Various changes to improve the generated code have been
6891 implemented, including:
6892 + GCC now uses the MULTIPLY AND ADD and MULTIPLY AND SUBTRACT
6893 instructions to significantly speed up many floating-point
6895 + GCC now uses the ADD LOGICAL WITH CARRY and SUBTRACT LOGICAL
6896 WITH BORROW instructions to speed up long long arithmetic.
6897 + GCC now uses the SEARCH STRING instruction to implement
6899 + In many cases, function call overhead for 31-bit code has been
6900 reduced by placing the literal pool after the function code
6901 instead of after the function prolog.
6902 + Register 14 is no longer reserved in 64-bit code.
6903 + Handling of global register variables has been improved.
6907 * The option -mflat is deprecated.
6908 * Support for large (> 2GB) frames has been added to the 64-bit port.
6909 * Several [26]ABI bugs have been fixed. Unfortunately, these changes
6910 will break binary compatibility with earlier releases.
6911 * The default debugging format has been switched from STABS to
6912 DWARF-2 for 32-bit code on Solaris 7 and later. DWARF-2 is already
6913 the default debugging format for 64-bit code on Solaris.
6917 * Support for the SH2E processor has been added. Enabled at run time
6918 with the -m2e command line switch, or at configure time by
6919 specifying sh2e as the machine part of the target triple.
6923 * Support for the Mitsubishi V850E1 processor has been added. This is
6924 a variant of the V850E processor with some additional debugging
6929 * Several ABI bugs have been fixed. Unfortunately, these changes
6930 break binary compatibility with earlier releases.
6931 + For big-endian processors, the padding of aggregate return
6932 values larger than a word has changed. If the size of an
6933 aggregate return value is not a multiple of 32 bits, previous
6934 versions of GCC inserted padding in the most-significant bytes
6935 of the first return value register. Aggregates larger than a
6936 word are now padded in the least-significant bytes of the last
6937 return value register used. Aggregates smaller than a word are
6938 still padded in the most-significant bytes. The return value
6939 padding has not changed for little-endian processors.
6940 + Function arguments with 16-byte alignment are now properly
6942 + The implementation of the va_list type has changed. A va_list
6943 value created by va_start from a previous release cannot be
6944 used with va_arg from this release, or vice versa.
6945 * More processor configuration options for Xtensa processors are
6947 + the ABS instruction is now optional;
6948 + the ADDX* and SUBX* instructions are now optional;
6949 + an experimental CONST16 instruction can be used to synthesize
6950 constants instead of loading them from constant pools.
6951 These and other Xtensa processor configuration options can no
6952 longer be enabled or disabled by command-line options; the
6953 processor configuration must be specified by the xtensa-config.h
6954 header file when building GCC. Additionally, the
6955 -mno-serialize-volatile option is no longer supported.
6959 Support for a number of older systems has been declared obsolete in GCC
6960 3.4. Unless there is activity to revive them, the next release of GCC
6961 will have their sources permanently removed.
6963 All configurations of the following processor architectures have been
6965 * Mitsubishi D30V, d30v-*
6966 * AT&T DSP1600 and DSP1610, dsp16xx-*
6969 Also, some individual systems have been obsoleted:
6971 + Support for generating code for operation in APCS/26 mode
6974 + "Bigfoot" port, i370-*. (The other port, s390-*, is actively
6975 maintained and supported.)
6977 + MOSS, i?86-moss-msdos and i?86-*-moss*
6978 + NCR 3000 running System V r.4, i?86-ncr-sysv4*
6979 + FreeBSD with a.out object format, i?86-*-freebsd*aout* and
6981 + GNU/Linux with a.out object format, i?86-linux*aout*
6982 + GNU/Linux with libc5, a.k.a. glibc1, i?86-linux*libc1*
6983 + Interix versions before Interix 3, i?86-*-interix
6984 + Mach microkernel, i?86-mach*
6985 + SCO UnixWare with UDK, i?86-*-udk*
6986 + Generic System V releases 1, 2, and 3, i?86-*-sysv[123]*
6987 + VSTa microkernel, i386-*-vsta
6988 * Motorola M68000 family
6989 + HPUX, m68k-hp-hpux* and m68000-hp-hpux*
6990 + NetBSD with a.out object format (before NetBSD 1.4),
6991 m68k-*-*-netbsd* except m68k-*-*-netbsdelf*
6992 + Generic System V r.4, m68k-*-sysv4*
6994 + Generic VAX, vax-*-* (This is generic VAX only; we have not
6995 obsoleted any VAX triples for specific operating systems.)
6997 Documentation improvements
6999 Other significant improvements
7001 * The build system has undergone several significant cleanups.
7002 Subdirectories will only be configured if they are being built, and
7003 all subdirectory configures are run from the make command. The top
7004 level has been autoconfiscated.
7005 * Building GCC no longer writes to its source directory. This should
7006 help those wishing to share a read-only source directory over NFS
7007 or build from a CD. The exceptions to this feature are if you
7008 configure with either --enable-maintainer-mode or
7009 --enable-generated-files-in-srcdir.
7010 * The -W warning option has been renamed to -Wextra, which is more
7011 easily understood. The older spelling will be retained for
7012 backwards compatibility.
7013 * Substantial improvements in compile time have been made,
7014 particularly for non-optimizing compilations.
7015 __________________________________________________________________
7021 A vast number of bugs have been fixed in 3.4.0, too many to publish a
7022 complete list here. [27]Follow this link to query the Bugzilla database
7023 for the list of over 900 bugs fixed in 3.4.0. This is the list of all
7024 bugs marked as resolved and fixed in 3.4.0 that are not flagged as 3.4
7026 __________________________________________________________________
7032 This section lists the problem reports (PRs) from GCC's bug tracking
7033 system that are known to be fixed in the 3.4.1 release. This list might
7034 not be complete (that is, it is possible that some PRs that have been
7035 fixed are not listed here).
7039 * [28]10129 Ada bootstrap fails on PPC-Darwin - invalid assembler
7040 emitted - PIC related
7041 * [29]14576 [ARM] ICE in libiberty when building gcc-3.4 for arm-elf
7042 * [30]14760 A bug in configure.in prevents using both
7043 --program-suffix and --program-prefix
7044 * [31]14671 [hppa64] bootstrap fails: ICE in
7045 save_call_clobbered_regs, in caller_save.c
7046 * [32]15093 [alpha][Java] make bootstrap fails to configure libffi on
7048 * [33]15178 Solaris 9/x86 fails linking after stage 3
7050 Multi-platform internal compiler errors (ICEs)
7052 * [34]12753 (preprocessor) Memory corruption in preprocessor on bad
7054 * [35]13985 ICE in gcc.c-torture/compile/930621-1.c
7055 * [36]14810 (c++) tree check failures with invalid code involving
7057 * [37]14883 (c++) ICE on invalid code, in cp_parser_lookup_name, in
7059 * [38]15044 (c++) ICE on syntax error, template header
7060 * [39]15057 (c++) Compiling of conditional value throw constructs
7061 cause a segmentation violation
7062 * [40]15064 (c++) typeid of template parameter gives ICE
7063 * [41]15142 (c++) ICE when passing a string where a char* is expected
7064 in a throw statement
7065 * [42]15159 ICE in rtl_verify_flow_info_1
7066 * [43]15165 (c++) ICE in instantiate_template
7067 * [44]15193 Unary minus using pointer to V4SF vector causes
7068 -fforce-mem to exhaust all memory
7069 * [45]15209 (c++) Runs out of memory with packed structs
7070 * [46]15227 (c++) Trouble with invalid function definition
7071 * [47]15285 (c++) instantiate_type ICE when forming pointer to
7073 * [48]15299 (c++) ICE in resolve_overloaded_unification
7074 * [49]15329 (c++) ICE on constructor of member template
7075 * [50]15550 ICE in extract_insn, in recog.c
7076 * [51]15554 (c++) ICE in tsubst_copy, in cp/pt.c
7077 * [52]15640 (c++) ICE on invalid code in arg_assoc, in
7079 * [53]15666 [unit-at-a-time] Gcc abort on valid code
7080 * [54]15696 (c++) ICE with bad pointer-to-member code
7081 * [55]15701 (c++) ICE with friends and template template parameter
7082 * [56]15761 ICE in do_SUBST, in combine.c
7083 * [57]15829 (c++) ICE on Botan-1.3.13 due to -funroll-loops
7087 * [58]14538 All RTEMS targets broken for gnat
7091 * [59]12391 missing warning about assigning to an incomplete type
7092 * [60]14649 atan(1.0) should not be a constant expression
7093 * [61]15004 [unit-at-a-time] no warning for unused paramater in
7095 * [62]15749 --pedantic-errors behaves differently from --pedantic
7096 with C-compiler on GNU/Linux
7098 C++ compiler and library
7100 * [63]10646 non-const reference is incorrectly matched in a "const T"
7101 partial specialization
7102 * [64]12077 wcin.rdbuf()->in_avail() return value too high
7103 * [65]13598 enc_filebuf doesn't work
7104 * [66]14211 const_cast returns lvalue but should be rvalue
7105 * [67]14220 num_put::do_put() undesired float/double behavior
7106 * [68]14245 problem with user-defined allocators in std::basic_string
7107 * [69]14340 libstdc++ Debug mode: failure to convert iterator to
7109 * [70]14600 __gnu_cxx::stdio_sync_filebuf should expose internal
7111 * [71]14668 no warning anymore for reevaluation of declaration
7112 * [72]14775 LFS (large file support) tests missing
7113 * [73]14821 Duplicate namespace alias declaration should not conflict
7114 * [74]14930 Friend declaration ignored
7115 * [75]14932 cannot use offsetof to get offsets of array elements in
7117 * [76]14950 [non unit-at-a-time] always_inline does not mix with
7119 * [77]14962 g++ ignores #pragma redefine_extname
7120 * [78]14975 Segfault on low-level write error during imbue
7121 * [79]15002 Linewise stream input is unusably slow (std::string slow)
7122 * [80]15025 compiler accepts redeclaration of template as
7124 * [81]15046 [arm] Math functions misdetected by cross configuration
7125 * [82]15069 a bit test on a variable of enum type is miscompiled
7126 * [83]15074 g++ -lsupc++ still links against libstdc++
7127 * [84]15083 spurious "statement has no effect" warning
7128 * [85]15096 parse error with templates and pointer to const member
7129 * [86]15287 combination of operator[] and operator .* fails in
7131 * [87]15317 __attribute__ unused in first parameter of constructor
7133 * [88]15337 sizeof on incomplete type diagnostic
7134 * [89]15361 bitset<>::_Find_next fails
7135 * [90]15412 _GLIBCXX_ symbols symbols defined and used in different
7137 * [91]15427 valid code results in incomplete type error
7138 * [92]15471 Incorrect member pointer offsets in anonymous
7140 * [93]15503 nested template problem
7141 * [94]15507 compiler hangs while laying out union
7142 * [95]15542 operator & and template definitions
7143 * [96]15565 SLES9: leading + sign for unsigned int with showpos
7144 * [97]15625 friend defined inside a template fails to find static
7146 * [98]15629 Function templates, overloads, and friend name injection
7147 * [99]15742 'noreturn' attribute ignored in method of template
7149 * [100]15775 Allocator::pointer consistently ignored
7150 * [101]15821 Duplicate namespace alias within namespace rejected
7151 * [102]15862 'enum yn' fails (confict with undeclared builtin)
7152 * [103]15875 rejects pointer to member in template
7153 * [104]15877 valid code using templates and anonymous enums is
7155 * [105]15947 Puzzling error message for wrong destructor declaration
7157 * [106]16020 cannot copy __gnu_debug::bitset
7158 * [107]16154 input iterator concept too restrictive
7159 * [108]16174 deducing top-level consts
7163 * [109]14315 Java compiler is not parallel make safe
7167 * [110]15151 [g77] incorrect logical i/o in 64-bit mode
7171 * [111]7993 private variables cannot be shadowed in subclasses
7175 * [112]15228 useless copies of floating point operands
7176 * [113]15345 [non-unit-at-a-time] unreferenced nested inline
7177 functions not optimized away
7178 * [114]15945 Incorrect floating point optimization
7179 * [115]15526 ftrapv aborts on 0 * (-1)
7180 * [116]14690 Miscompiled POOMA tests
7181 * [117]15112 GCC generates code to write to unchanging memory
7185 * [118]15067 Minor glitch in the source of cpp
7187 Main driver program bugs
7189 * [119]1963 collect2 interprets -oldstyle_liblookup as -o
7192 x86-specific (Intel/AMD)
7194 * [120]15717 Error: can't resolve `L0' {*ABS* section} - `xx' {*UND*
7199 * [121]14782 GCC produces an unaligned data access at -O2
7200 * [122]14828 FAIL: gcc.c-torture/execute/20030408-1.c execution, -O2
7201 * [123]15202 ICE in reload_cse_simplify_operands, in postreload.c
7205 * [124]14610 __float80 constants incorrectly emitted
7206 * [125]14813 init_array sections are initialized in the wrong order
7207 * [126]14857 GCC segfault on duplicated asm statement
7208 * [127]15598 Gcc 3.4 ICE on valid code
7209 * [128]15653 Gcc 3.4 ICE on valid code
7213 * [129]15189 wrong filling of delay slot with -march=mips1 -G0
7214 -mno-split-addresses -mno-explicit-relocs
7215 * [130]15331 Assembler error building gnatlib on IRIX 6.5 with GNU as
7217 * [131]16144 Bogus reference to __divdf3 when -O1
7218 * [132]16176 Miscompilation of unaligned data in MIPS backend
7222 * [133]11591 ICE in gcc.dg/altivec-5.c
7223 * [134]12028 powerpc-eabispe produces bad sCOND operation
7224 * [135]14478 rs6000 geu/ltu patterns generate incorrect code
7225 * [136]14567 long double and va_arg complex args
7226 * [137]14715 Altivec stack layout may overlap gpr save with stack
7228 * [138]14902 (libstdc++) Stream checking functions fail when -pthread
7230 * [139]14924 Compiler ICE on valid code
7231 * [140]14960 -maltivec affects vector return with -mabi=no-altivec
7232 * [141]15106 vector varargs failure passing from altivec to
7233 non-altivec code for -m32
7234 * [142]16026 ICE in function.c:4804, assign_parms, when -mpowerpc64 &
7236 * [143]15191 -maltivec -mabi=no-altivec results in mis-aligned lvx
7238 * [144]15662 Segmentation fault when an exception is thrown - even if
7239 try and catch are specified
7243 * [145]15054 Bad code due to overlapping stack temporaries
7247 * [146]15783 ICE with union assignment in 64-bit mode
7248 * [147]15626 GCC 3.4 emits "ld: warning: relocation error:
7253 * [148]14326 boehm-gc hardcodes to 3DNow! prefetch for x86_64
7254 * [149]14723 Backported -march=nocona from mainline
7255 * [150]15290 __float128 failed to pass to function properly
7257 Cygwin/Mingw32-specific
7259 * [151]15250 Option -mms-bitfields support on GCC 3.4 is not
7260 conformant to MS layout
7261 * [152]15551 -mtune=pentium4 -O2 with sjlj EH breaks stack probe
7262 worker on windows32 targets
7264 Bugs specific to embedded processors
7266 * [153]8309 [m68k] -m5200 produces erroneous SImode set of short
7268 * [154]13250 [SH] Gcc code for rotation clobbers the register, but
7269 gcc continues to use the register as if it was not clobbered
7270 * [155]13803 [coldfire] movqi operand constraints too restrictivefor
7272 * [156]14093 [SH] ICE for code when using -mhitachi option in SH
7273 * [157]14457 [m6811hc] ICE with simple c++ source
7274 * [158]14542 [m6811hc] ICE on simple source
7275 * [159]15100 [SH] cc1plus got hang-up on
7276 libstdc++-v3/testsuite/abi_check.cc
7277 * [160]15296 [CRIS] Delayed branch scheduling causing invalid code on
7279 * [161]15396 [SH] ICE with -O2 -fPIC
7280 * [162]15782 [coldfire] m68k_output_mi_thunk emits wrong code for
7283 Testsuite problems (compiler not affected)
7285 * [163]11610 libstdc++ testcases 27_io/* don't work properly remotely
7286 * [164]15488 (libstdc++) possibly insufficient file permissions for
7287 executing test suite
7288 * [165]15489 (libstdc++) testsuite_files determined incorrectly
7292 * [166]13928 (libstdc++) no whatis info in some man pages generated
7294 * [167]14150 Ada documentation out of date
7295 * [168]14949 (c++) Need to document method visibility changes
7296 * [169]15123 libstdc++-doc: Allocators.3 manpage is empty
7297 __________________________________________________________________
7303 This section lists the problem reports (PRs) from GCC's bug tracking
7304 system that are known to be fixed in the 3.4.2 release. This list might
7305 not be complete (that is, it is possible that some PRs that have been
7306 fixed are not listed here).
7308 Bootstrap failures and issues
7310 * [170]16469 [mips-sgi-irix5.3] bootstrap fails in
7311 libstdc++-v3/testsuite
7312 * [171]16344 [hppa-linux-gnu] libstdc++'s PCH built by
7313 profiledbootstrap does not work with the built compiler
7314 * [172]16842 [Solaris/x86] mkheaders can not find mkheaders.conf
7316 Multi-platform internal compiler errors (ICEs)
7318 * [173]12608 (c++) ICE: expected class 't', have 'x' (error_mark) in
7319 cp_parser_class_specifier, in cp/parser.c
7320 * [174]14492 ICE in loc_descriptor_from_tree, in dwarf2out.c
7321 * [175]15461 (c++) ICE due to NRV and inlining
7322 * [176]15890 (c++) ICE in c_expand_expr, in c-common.c
7323 * [177]16180 ICE: segmentation fault in RTL optimization
7324 * [178]16224 (c++) ICE in write_unscoped_name (template/namespace)
7325 * [179]16408 ICE: in delete_insn, in cfgrtl.c
7326 * [180]16529 (c++) ICE for: namespace-alias shall not be declared as
7327 the name of any other entity
7328 * [181]16698 (c++) ICE with exceptions and declaration of __cxa_throw
7329 * [182]16706 (c++) ICE in finish_member_declaration, in
7331 * [183]16810 (c++) Legal C++ program with cast gives ICE in
7333 * [184]16851 (c++) ICE when throwing a comma expression
7334 * [185]16870 (c++) Boost.Spirit causes ICE in tsubst, in cp/pt.c
7335 * [186]16904 (c++) ICE in finish_class_member_access_expr, in
7337 * [187]16905 (c++) ICE (segfault) with exceptions
7338 * [188]16964 (c++) ICE in cp_parser_class_specifier due to
7340 * [189]17068 (c++) ICE: tree check: expected class 'd', have 'x'
7341 (identifier_node) in dependent_template_p, in cp/pt.c
7345 * [190]16366 Preprocessor option -remap causes memory corruption
7349 * [191]15345 unreferenced nested inline functions not optimized away
7350 * [192]16590 Incorrect execution when compiling with -O2
7351 * [193]16693 Bitwise AND is lost when used within a cast to an enum
7352 of the same precision
7353 * [194]17078 Jump into if(0) substatement fails
7355 Problems in generated debug information
7357 * [195]13956 incorrect stabs for nested local variables
7361 * [196]16684 GCC should not warn about redundant redeclarations of
7364 C++ compiler and library
7366 * [197]12658 Thread safety problems in locale::global() and
7368 * [198]13092 g++ accepts invalid pointer-to-member conversion
7369 * [199]15320 Excessive memory consumption
7370 * [200]16246 Incorrect template argument deduction
7371 * [201]16273 Memory exhausted when using nested classes and virtual
7373 * [202]16401 ostringstream in gcc 3.4.x very slow for big data
7374 * [203]16411 undefined reference to
7375 __gnu_cxx::stdio_sync_filebuf<char, std::char_traits<char>
7377 * [204]16489 G++ incorrectly rejects use of a null constant integral
7378 expression as a null constant pointer
7379 * [205]16618 offsetof fails with constant member
7380 * [206]16637 syntax error reported for valid input code
7381 * [207]16717 __attribute__((constructor)) broken in C++
7382 * [208]16813 compiler error in DEBUG version of range insertion
7384 * [209]16853 pointer-to-member initialization from incompatible one
7386 * [210]16889 ambiguity is not detected
7387 * [211]16959 Segmentation fault in ios_base::sync_with_stdio
7389 Java compiler and library
7391 * [212]7587 direct threaded interpreter not thread-safe
7392 * [213]16473 ServerSocket accept() leaks file descriptors
7393 * [214]16478 Hash synchronization deadlock with finalizers
7397 * [215]10695 ICE in dwarf2out_frame_debug_expr, in dwarf2out.c
7398 * [216]16974 could not split insn (ice in final_scan_insn, in
7403 * [217]16298 ICE in output_operand
7404 * [218]17113 ICE with SSE2 intrinsics
7408 * [219]14697 libstdc++ couldn't find 32bit libgcc_s
7412 * [220]15869 [mips64] No NOP after LW (with -mips1 -O0)
7413 * [221]16325 [mips64] value profiling clobbers gp on mips
7414 * [222]16357 [mipsisa64-elf] ICE copying 7 bytes between extern
7416 * [223]16380 [mips64] Use of uninitialised register after dbra
7418 * [224]16407 [mips64] Unaligned access to local variables
7419 * [225]16643 [mips64] verify_local_live_at_start ICE after
7420 crossjumping & cfgcleanup
7424 * [226]15927 THUMB -O2: strength-reduced iteration variable ends up
7426 * [227]15948 THUMB: ICE with non-commutative cbranch
7427 * [228]17019 THUMB: bad switch statement in md code for
7428 addsi3_cbranch_scratch
7432 * [229]16130 ICE on valid code: in bundling, in config/ia64/ia64.c
7434 * [230]16142 ICE on valid code: in bundling, in config/ia64/ia64.c
7436 * [231]16278 Gcc failed to build Linux kernel with -mtune=merced
7437 * [232]16414 ICE on valid code: typo in comparison of asm_noperands
7439 * [233]16445 ICE on valid code: don't count ignored insns
7440 * [234]16490 ICE (segfault) while compiling with -fprofile-use
7441 * [235]16683 ia64 does not honor SUBTARGET_EXTRA_SPECS
7445 * [236]16195 (ppc64): Miscompilation of GCC 3.3.x by 3.4.x
7446 * [237]16239 ICE on ppc64 (mozilla 1.7 compile, -O1 -fno-exceptions
7451 * [238]16199 ICE while compiling apache 2.0.49
7452 * [239]16416 -m64 doesn't imply -mcpu=v9 anymore
7453 * [240]16430 ICE when returning non-C aggregates larger than 16 bytes
7455 Bugs specific to embedded processors
7457 * [241]16379 [m32r] can't output large model function call of memcpy
7458 * [242]17093 [m32r] ICE with -msdata=use -O0
7459 * [243]17119 [m32r] ICE at switch case 0x8000
7463 * [244]15928 libstdc++ in 3.4.x doesn't cross-compile for djgpp
7465 Alpha Tru64-specific
7467 * [245]16210 libstdc++ gratuitously omits "long long" I/O
7469 Testsuite, documentation issues (compiler is not affected):
7471 * [246]15488 (libstdc++) possibly insufficient file permissions for
7472 executing test suite
7473 * [247]16250 ada/doctools runs makeinfo even in release tarball
7474 __________________________________________________________________
7478 This is the [248]list of problem reports (PRs) from GCC's bug tracking
7479 system that are known to be fixed in the 3.4.3 release. This list might
7480 not be complete (that is, it is possible that some PRs that have been
7481 fixed are not listed here).
7485 * [249]17369 [ia64] Bootstrap failure with binutils-2.15.90.0.1.1
7486 * [250]17850 [arm-elf] bootstrap failure - libstdc++ uses strtold
7489 Internal compiler errors (ICEs) affecting multiple platforms
7491 * [251]13948 (java) GCJ segmentation fault while compiling GL4Java
7493 * [252]14492 ICE in loc_descriptor_from_tree, in dwarf2out.c
7494 * [253]16301 (c++) ICE when "strong" attribute is attached to a using
7496 * [254]16566 ICE with flexible arrays
7497 * [255]17023 ICE with nested functions in parameter declaration
7498 * [256]17027 ICE with noreturn function in loop at -O2
7499 * [257]17524 ICE in grokdeclarator, in cp/decl.c
7500 * [258]17826 (c++) ICE in cp_tree_equal
7502 C and optimization bugs
7504 * [259]15526 -ftrapv aborts on 0 * (-1)
7505 * [260]16999 #ident stopped working
7506 * [261]17503 quadratic behaviour in invalid_mode_change_p
7507 * [262]17581 Long long arithmetic fails inside a switch/case
7508 statement when compiled with -O2
7509 * [263]18129 -fwritable-strings doesn't work
7511 C++ compiler and library bugs
7513 * [264]10975 incorrect initial ostringstream::tellp()
7514 * [265]11722 Unbuffered filebuf::sgetn is slow
7515 * [266]14534 Unrecognizing static function as a template parameter
7516 when its return value is also templated
7517 * [267]15172 Copy constructor optimization in aggregate
7519 * [268]15786 Bad error message for frequently occuring error.
7520 * [269]16162 Rejects valid member-template-definition
7521 * [270]16612 empty basic_strings can't live in shared memory
7522 * [271]16715 std::basic_iostream is instantiated when used, even
7523 though instantiations are already contained in libstdc++
7524 * [272]16848 code in /ext/demangle.h appears broken
7525 * [273]17132 GCC fails to eliminate function template specialization
7526 when argument deduction fails
7527 * [274]17259 One more _S_leaf incorrectly qualified with _RopeRep::
7529 * [275]17327 use of `enumeral_type' in template type unification
7530 * [276]17393 "unused variable '._0'" warning with -Wall
7531 * [277]17501 Confusion with member templates
7532 * [278]17537 g++ not passing -lstdc++ to linker when all command line
7533 arguments are libraries
7534 * [279]17585 usage of unqualified name of static member from within
7536 * [280]17821 Poor diagnostic for using "." instead of "->"
7537 * [281]17829 wrong error: call of overloaded function is ambiguous
7538 * [282]17851 Misleading diagnostic for invalid function declarations
7539 with undeclared types
7540 * [283]17976 Destructor is called twice
7541 * [284]18020 rejects valid definition of enum value in template
7542 * [285]18093 bogus conflict in namespace aliasing
7543 * [286]18140 C++ parser bug when using >> in templates
7547 * [287]17541 data statements with double precision constants fail
7551 * [288]17853 -O2 ICE for MMX testcase
7555 * [289]17245 ICE compiling gsl-1.5 statistics/lag1.c
7559 * [290]17167 FATAL:Symbol L_foo$stub already defined.
7563 * [291]17277 could not catch an exception when specified -maix64
7567 * [292]17505 <cmath> calls acosf(), ceilf(), and other functions
7568 missing from system libraries
7572 * [293]17684 /usr/ccs/bin/ld: Can't create libgcc_s.sl
7576 * [294]17384 ICE with mode attribute on structures
7580 * [295]17770 No NOP after LWL with -mips1
7582 Other embedded target specific
7584 * [296]11476 [arc-elf] gcc ICE on newlib's vfprintf.c
7585 * [297]14064 [avr-elf] -fdata-sections triggers ICE
7586 * [298]14678 [m68hc11-elf] gcc ICE
7587 * [299]15583 [powerpc-rtems] powerpc-rtems lacks __USE_INIT_FINI__
7588 * [300]15790 [i686-coff] Alignment error building gcc with i686-coff
7590 * [301]15886 [SH] Miscompilation with -O2 -fPIC
7591 * [302]16884 [avr-elf] [fweb related] bug while initializing
7594 Bugs relating to debugger support
7596 * [303]13841 missing debug info for _Complex function arguments
7597 * [304]15860 [big-endian targets] No DW_AT_location debug info is
7598 emitted for formal arguments to a function that uses "register"
7601 Testsuite issues (compiler not affected)
7603 * [305]17465 Testsuite in libffi overrides LD_LIBRARY_PATH
7604 * [306]17469 Testsuite in libstdc++ overrides LD_LIBRARY_PATH
7605 * [307]18138 [mips-sgi-irix6.5] libgcc_s.so.1 not found by 64-bit
7610 * [308]15498 typo in gcc manual: non-existing locale example en_UK,
7612 * [309]15747 [mips-sgi-irix5.3] /bin/sh hangs during bootstrap:
7613 document broken shell
7614 * [310]16406 USE_LD_AS_NEEDED undocumented
7615 __________________________________________________________________
7619 This is the [311]list of problem reports (PRs) from GCC's bug tracking
7620 system that are known to be fixed in the 3.4.4 release. This list might
7621 not be complete (that is, it is possible that some PRs that have been
7622 fixed are not listed here).
7623 __________________________________________________________________
7627 This is the [312]list of problem reports (PRs) from GCC's bug tracking
7628 system that are known to be fixed in the 3.4.5 release. This list might
7629 not be complete (that is, it is possible that some PRs that have been
7630 fixed are not listed here).
7634 * [313]24688 sco_math fixincl breaks math.h
7638 * [314]17188 struct Foo { } redefinition
7639 * [315]20187 wrong code for ((unsigned char)(unsigned long
7640 long)((a?a:1)&(a*b)))?0:1)
7641 * [316]21873 infinite warning loop on bad array initializer
7642 * [317]21899 enum definition accepts values to be overriden
7643 * [318]22061 ICE in find_function_data, in function.c
7644 * [319]22308 Failure to diagnose violation of constraint 6.516p2
7645 * [320]22458 ICE on missing brace
7646 * [321]22589 ICE casting to long long
7647 * [322]24101 Segfault with preprocessed source
7649 C++ compiler and library bugs
7651 * [323]10611 operations on vector mode not recognized in C++
7652 * [324]13377 unexpected behavior of namespace usage directive
7653 * [325]16002 Strange error message with new parser
7654 * [326]17413 local classes as template argument
7655 * [327]17609 spurious error message after using keyword
7656 * [328]17618 ICE in cp_convert_to_pointer, in cp/cvt.c
7657 * [329]18124 ICE with invalid template template parameter
7658 * [330]18155 typedef in template declaration not rejected
7659 * [331]18177 ICE with const_cast for undeclared variable
7660 * [332]18368 C++ error message regression
7661 * [333]16378 ICE when returning a copy of a packed member
7662 * [334]18466 int ::i; accepted
7663 * [335]18512 ICE on invalid usage of template base class
7664 * [336]18454 ICE when returning undefined type
7665 * [337]18738 typename not allowed with non-dependent qualified name
7666 * [338]18803 rejects access to operator() in template
7667 * [339]19004 ICE in uses_template_parms, in cp/pt.c
7668 * [340]19208 Spurious error about variably modified type
7669 * [341]18253 bad error message / ICE for invalid template parameter
7670 * [342]19608 ICE after friend function definition in local class
7671 * [343]19884 ICE on explicit instantiation of a non-template
7673 * [344]20153 ICE when C++ template function contains anonymous union
7674 * [345]20563 Infinite loop in diagnostic (and ice after error
7676 * [346]20789 ICE with incomplete type in template
7677 * [347]21336 Internal compiler error when using custom new operators
7678 * [348]21768 ICE in error message due to violation of coding
7680 * [349]21853 constness of pointer to data member ignored
7681 * [350]21903 Default argument of template function causes a
7683 * [351]21983 multiple diagnostics
7684 * [352]21987 New testsuite failure
7685 g++.dg/warn/conversion-function-1.C
7686 * [353]22153 ICE on invalid template specialization
7687 * [354]22172 Internal compiler error, seg fault.
7688 * [355]21286 filebuf::xsgetn vs pipes
7689 * [356]22233 ICE with wrong number of template parameters
7690 * [357]22508 ICE after invalid operator new
7691 * [358]22545 ICE with pointer to class member & user defined
7693 * [359]23528 Wrong default allocator in ext/hash_map
7694 * [360]23550 char_traits requirements/1.cc test bad math
7695 * [361]23586 Bad diagnostic for invalid namespace-name
7696 * [362]23624 ICE in invert_truthvalue, in fold-const.c
7697 * [363]23639 Bad error message: not a member of '<declaration error>'
7698 * [364]23797 ICE on typename outside template
7699 * [365]23965 Bogus error message: no matching function for call to
7701 * [366]24052 &#`label_decl' not supported by dump_expr#<expression
7703 * [367]24580 virtual base class cause exception not to be caught
7705 Problems in generated debug information
7707 * [368]24267 Bad DWARF for altivec vectors
7709 Optimizations issues
7711 * [369]17810 ICE in verify_local_live_at_start
7712 * [370]17860 Wrong generated code for loop with varying bound
7713 * [371]21709 ICE on compile-time complex NaN
7714 * [372]21964 broken tail call at -O2 or more
7715 * [373]22167 Strange optimization bug when using -Os
7716 * [374]22619 Compilation failure for real_const_1.f and
7718 * [375]23241 Invalid code generated for comparison of uchar to 255
7719 * [376]23478 Miscompilation due to reloading of a var that is also
7721 * [377]24470 segmentation fault in cc1plus when compiling with -O
7722 * [378]24950 ICE in operand_subword_force
7724 Precompiled headers problems
7726 * [379]14400 Cannot compile qt-x11-free-3.3.0
7727 * [380]14940 PCH largefile test fails on various platforms
7731 * [381]20239 ICE on empty preprocessed input
7732 * [382]15220 "gcc -E -MM -MG" reports missing system headers in
7737 * [383]19275 gcc.dg/20020919-1.c fails with -fpic/-fPIC on
7742 * [384]21888 bootstrap failure with linker relaxation enabled
7746 * [385]15342 [arm-linux]: ICE in verify_local_live_at_start
7747 * [386]23985 Memory aliasing information incorrect in inlined memcpy
7751 * [387]16719 Illegal move of byte into address register causes
7756 * [388]21723 ICE while building libgfortran
7757 * [389]21841 -mhp-ld/-mgnu-ld documentation
7761 * [390]23644 IA-64 hardware models and configuration options
7763 * [391]24718 Shared libgcc not used for linking by default
7767 * [392]18421 ICE in reload_cse_simplify_operands, in postreload.c
7771 * [393]20621 ICE in change_address_1, in emit-rtl.c
7773 PowerPC and PowerPC64 specific
7775 * [394]18583 error on valid code: const
7776 __attribute__((altivec(vector__))) doesn't work in arrays
7777 * [395]20191 ICE in reload_cse_simplify_operands
7778 * [396]22083 AIX: TARGET_C99_FUNCTIONS is wrongly defined
7779 * [397]23070 CALL_V4_CLEAR_FP_ARGS flag not properly set
7780 * [398]23404 gij trashes args of functions with more than 8 fp args
7781 * [399]23539 C & C++ compiler generating misaligned references
7782 regardless of compiler flags
7783 * [400]24102 floatdisf2_internal2 broken
7784 * [401]24465 -mminimal-toc miscompilation of __thread vars
7788 * [402]19933 Problem with define of HUGE_VAL in math_c99
7789 * [403]21889 Native Solaris assembler cannot grok DTP-relative debug
7794 * [404]19300 PCH failures on sparc-linux
7795 * [405]20301 Assembler labels have a leading "-"
7796 * [406]20673 C PCH testsuite assembly comparison failure
7798 x86 and x86_64 specific
7800 * [407]18582 ICE with arrays of type V2DF
7801 * [408]19340 Compilation SEGFAULTs with -O1 -fschedule-insns2
7803 * [409]21716 ICE in reg-stack.c's swap_rtx_condition
7804 * [410]24315 amd64 fails -fpeephole2
7805 __________________________________________________________________
7809 This is the [411]list of problem reports (PRs) from GCC's bug tracking
7810 system that are known to be fixed in the 3.4.6 release. This list might
7811 not be complete (that is, it is possible that some PRs that have been
7812 fixed are not listed here).
7815 For questions related to the use of GCC, please consult these web
7816 pages and the [412]GCC manuals. If that fails, the
7817 [413]gcc-help@gcc.gnu.org mailing list might help. Comments on these
7818 web pages and the development of GCC are welcome on our developer
7819 list at [414]gcc@gcc.gnu.org. All of [415]our lists have public
7822 Copyright (C) [416]Free Software Foundation, Inc. Verbatim copying and
7823 distribution of this entire article is permitted in any medium,
7824 provided this notice is preserved.
7826 These pages are [417]maintained by the GCC team. Last modified
7831 1. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.6
7832 2. http://gcc.gnu.org/gcc-3.4/changes.html#cplusplus
7833 3. http://gcc.gnu.org/gcc-3.3/changes.html#obsolete_systems
7834 4. http://gcc.gnu.org/gcc-3.4/changes.html#obsolete_systems
7835 5. http://gcc.gnu.org/gcc-3.4/mips-abi.html
7836 6. http://gcc.gnu.org/gcc-3.4/sparc-abi.html
7837 7. http://www.boost.org/
7838 8. http://gcc.gnu.org/PR11953
7839 9. http://gcc.gnu.org/PR8361
7840 10. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Other-Builtins.html#Other%20Builtins
7841 11. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_closed.html#209
7842 12. http://gcc.gnu.org/bugs/#cxx_rvalbind
7843 13. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Objective-C-Dialect-Options.html
7844 14. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Objective-C-Dialect-Options.html
7845 15. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Objective-C-Dialect-Options.html
7846 16. http://www.gnu.org/software/classpath/
7847 17. http://www.eclipse.org/
7848 18. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/g77/News.html
7849 19. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Alpha-Built-in-Functions.html
7850 20. http://h30097.www3.hp.com/docs/base_doc/DOCUMENTATION/V51A_HTML/ARH9MBTE/DTMNPLTN.HTM#normal-argument-list-structure
7851 21. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gccint/Processor-pipeline-description.html
7852 22. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gccint/Comparison-of-the-two-descriptions.html
7853 23. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gccint/Processor-pipeline-description.html
7854 24. http://gcc.gnu.org/gcc-3.4/mips-abi.html
7855 25. http://gcc.gnu.org/gcc-3.4/powerpc-abi.html
7856 26. http://gcc.gnu.org/gcc-3.4/sparc-abi.html
7857 27. http://gcc.gnu.org/bugzilla/buglist.cgi?short_desc_type=notregexp&short_desc=%5C%5B3%5C.4.*%5BRr%5Degression&target_milestone=3.4.0&bug_status=RESOLVED&resolution=FIXED
7858 28. http://gcc.gnu.org/PR10129
7859 29. http://gcc.gnu.org/PR14576
7860 30. http://gcc.gnu.org/PR14760
7861 31. http://gcc.gnu.org/PR14671
7862 32. http://gcc.gnu.org/PR15093
7863 33. http://gcc.gnu.org/PR15178
7864 34. http://gcc.gnu.org/PR12753
7865 35. http://gcc.gnu.org/PR13985
7866 36. http://gcc.gnu.org/PR14810
7867 37. http://gcc.gnu.org/PR14883
7868 38. http://gcc.gnu.org/PR15044
7869 39. http://gcc.gnu.org/PR15057
7870 40. http://gcc.gnu.org/PR15064
7871 41. http://gcc.gnu.org/PR15142
7872 42. http://gcc.gnu.org/PR15159
7873 43. http://gcc.gnu.org/PR15165
7874 44. http://gcc.gnu.org/PR15193
7875 45. http://gcc.gnu.org/PR15209
7876 46. http://gcc.gnu.org/PR15227
7877 47. http://gcc.gnu.org/PR15285
7878 48. http://gcc.gnu.org/PR15299
7879 49. http://gcc.gnu.org/PR15329
7880 50. http://gcc.gnu.org/PR15550
7881 51. http://gcc.gnu.org/PR15554
7882 52. http://gcc.gnu.org/PR15640
7883 53. http://gcc.gnu.org/PR15666
7884 54. http://gcc.gnu.org/PR15696
7885 55. http://gcc.gnu.org/PR15701
7886 56. http://gcc.gnu.org/PR15761
7887 57. http://gcc.gnu.org/PR15829
7888 58. http://gcc.gnu.org/PR14538
7889 59. http://gcc.gnu.org/PR12391
7890 60. http://gcc.gnu.org/PR14649
7891 61. http://gcc.gnu.org/PR15004
7892 62. http://gcc.gnu.org/PR15749
7893 63. http://gcc.gnu.org/PR10646
7894 64. http://gcc.gnu.org/PR12077
7895 65. http://gcc.gnu.org/PR13598
7896 66. http://gcc.gnu.org/PR14211
7897 67. http://gcc.gnu.org/PR14220
7898 68. http://gcc.gnu.org/PR14245
7899 69. http://gcc.gnu.org/PR14340
7900 70. http://gcc.gnu.org/PR14600
7901 71. http://gcc.gnu.org/PR14668
7902 72. http://gcc.gnu.org/PR14775
7903 73. http://gcc.gnu.org/PR14821
7904 74. http://gcc.gnu.org/PR14930
7905 75. http://gcc.gnu.org/PR14932
7906 76. http://gcc.gnu.org/PR14950
7907 77. http://gcc.gnu.org/PR14962
7908 78. http://gcc.gnu.org/PR14975
7909 79. http://gcc.gnu.org/PR15002
7910 80. http://gcc.gnu.org/PR15025
7911 81. http://gcc.gnu.org/PR15046
7912 82. http://gcc.gnu.org/PR15069
7913 83. http://gcc.gnu.org/PR15074
7914 84. http://gcc.gnu.org/PR15083
7915 85. http://gcc.gnu.org/PR15096
7916 86. http://gcc.gnu.org/PR15287
7917 87. http://gcc.gnu.org/PR15317
7918 88. http://gcc.gnu.org/PR15337
7919 89. http://gcc.gnu.org/PR15361
7920 90. http://gcc.gnu.org/PR15412
7921 91. http://gcc.gnu.org/PR15427
7922 92. http://gcc.gnu.org/PR15471
7923 93. http://gcc.gnu.org/PR15503
7924 94. http://gcc.gnu.org/PR15507
7925 95. http://gcc.gnu.org/PR15542
7926 96. http://gcc.gnu.org/PR15565
7927 97. http://gcc.gnu.org/PR15625
7928 98. http://gcc.gnu.org/PR15629
7929 99. http://gcc.gnu.org/PR15742
7930 100. http://gcc.gnu.org/PR15775
7931 101. http://gcc.gnu.org/PR15821
7932 102. http://gcc.gnu.org/PR15862
7933 103. http://gcc.gnu.org/PR15875
7934 104. http://gcc.gnu.org/PR15877
7935 105. http://gcc.gnu.org/PR15947
7936 106. http://gcc.gnu.org/PR16020
7937 107. http://gcc.gnu.org/PR16154
7938 108. http://gcc.gnu.org/PR16174
7939 109. http://gcc.gnu.org/PR14315
7940 110. http://gcc.gnu.org/PR15151
7941 111. http://gcc.gnu.org/PR7993
7942 112. http://gcc.gnu.org/PR15228
7943 113. http://gcc.gnu.org/PR15345
7944 114. http://gcc.gnu.org/PR15945
7945 115. http://gcc.gnu.org/PR15526
7946 116. http://gcc.gnu.org/PR14690
7947 117. http://gcc.gnu.org/PR15112
7948 118. http://gcc.gnu.org/PR15067
7949 119. http://gcc.gnu.org/PR1963
7950 120. http://gcc.gnu.org/PR15717
7951 121. http://gcc.gnu.org/PR14782
7952 122. http://gcc.gnu.org/PR14828
7953 123. http://gcc.gnu.org/PR15202
7954 124. http://gcc.gnu.org/PR14610
7955 125. http://gcc.gnu.org/PR14813
7956 126. http://gcc.gnu.org/PR14857
7957 127. http://gcc.gnu.org/PR15598
7958 128. http://gcc.gnu.org/PR15653
7959 129. http://gcc.gnu.org/PR15189
7960 130. http://gcc.gnu.org/PR15331
7961 131. http://gcc.gnu.org/PR16144
7962 132. http://gcc.gnu.org/PR16176
7963 133. http://gcc.gnu.org/PR11591
7964 134. http://gcc.gnu.org/PR12028
7965 135. http://gcc.gnu.org/PR14478
7966 136. http://gcc.gnu.org/PR14567
7967 137. http://gcc.gnu.org/PR14715
7968 138. http://gcc.gnu.org/PR14902
7969 139. http://gcc.gnu.org/PR14924
7970 140. http://gcc.gnu.org/PR14960
7971 141. http://gcc.gnu.org/PR15106
7972 142. http://gcc.gnu.org/PR16026
7973 143. http://gcc.gnu.org/PR15191
7974 144. http://gcc.gnu.org/PR15662
7975 145. http://gcc.gnu.org/PR15054
7976 146. http://gcc.gnu.org/PR15783
7977 147. http://gcc.gnu.org/PR15626
7978 148. http://gcc.gnu.org/PR14326
7979 149. http://gcc.gnu.org/PR14723
7980 150. http://gcc.gnu.org/PR15290
7981 151. http://gcc.gnu.org/PR15250
7982 152. http://gcc.gnu.org/PR15551
7983 153. http://gcc.gnu.org/PR8309
7984 154. http://gcc.gnu.org/PR13250
7985 155. http://gcc.gnu.org/PR13803
7986 156. http://gcc.gnu.org/PR14093
7987 157. http://gcc.gnu.org/PR14457
7988 158. http://gcc.gnu.org/PR14542
7989 159. http://gcc.gnu.org/PR15100
7990 160. http://gcc.gnu.org/PR15296
7991 161. http://gcc.gnu.org/PR15396
7992 162. http://gcc.gnu.org/PR15782
7993 163. http://gcc.gnu.org/PR11610
7994 164. http://gcc.gnu.org/PR15488
7995 165. http://gcc.gnu.org/PR15489
7996 166. http://gcc.gnu.org/PR13928
7997 167. http://gcc.gnu.org/PR14150
7998 168. http://gcc.gnu.org/PR14949
7999 169. http://gcc.gnu.org/PR15123
8000 170. http://gcc.gnu.org/PR16469
8001 171. http://gcc.gnu.org/PR16344
8002 172. http://gcc.gnu.org/PR16842
8003 173. http://gcc.gnu.org/PR12608
8004 174. http://gcc.gnu.org/PR14492
8005 175. http://gcc.gnu.org/PR15461
8006 176. http://gcc.gnu.org/PR15890
8007 177. http://gcc.gnu.org/PR16180
8008 178. http://gcc.gnu.org/PR16224
8009 179. http://gcc.gnu.org/PR16408
8010 180. http://gcc.gnu.org/PR16529
8011 181. http://gcc.gnu.org/PR16698
8012 182. http://gcc.gnu.org/PR16706
8013 183. http://gcc.gnu.org/PR16810
8014 184. http://gcc.gnu.org/PR16851
8015 185. http://gcc.gnu.org/PR16870
8016 186. http://gcc.gnu.org/PR16904
8017 187. http://gcc.gnu.org/PR16905
8018 188. http://gcc.gnu.org/PR16964
8019 189. http://gcc.gnu.org/PR17068
8020 190. http://gcc.gnu.org/PR16366
8021 191. http://gcc.gnu.org/PR15345
8022 192. http://gcc.gnu.org/PR16590
8023 193. http://gcc.gnu.org/PR16693
8024 194. http://gcc.gnu.org/PR17078
8025 195. http://gcc.gnu.org/PR13956
8026 196. http://gcc.gnu.org/PR16684
8027 197. http://gcc.gnu.org/PR12658
8028 198. http://gcc.gnu.org/PR13092
8029 199. http://gcc.gnu.org/PR15320
8030 200. http://gcc.gnu.org/PR16246
8031 201. http://gcc.gnu.org/PR16273
8032 202. http://gcc.gnu.org/PR16401
8033 203. http://gcc.gnu.org/PR16411
8034 204. http://gcc.gnu.org/PR16489
8035 205. http://gcc.gnu.org/PR16618
8036 206. http://gcc.gnu.org/PR16637
8037 207. http://gcc.gnu.org/PR16717
8038 208. http://gcc.gnu.org/PR16813
8039 209. http://gcc.gnu.org/PR16853
8040 210. http://gcc.gnu.org/PR16889
8041 211. http://gcc.gnu.org/PR16959
8042 212. http://gcc.gnu.org/PR7587
8043 213. http://gcc.gnu.org/PR16473
8044 214. http://gcc.gnu.org/PR16478
8045 215. http://gcc.gnu.org/PR10695
8046 216. http://gcc.gnu.org/PR16974
8047 217. http://gcc.gnu.org/PR16298
8048 218. http://gcc.gnu.org/PR17113
8049 219. http://gcc.gnu.org/PR14697
8050 220. http://gcc.gnu.org/PR15869
8051 221. http://gcc.gnu.org/PR16325
8052 222. http://gcc.gnu.org/PR16357
8053 223. http://gcc.gnu.org/PR16380
8054 224. http://gcc.gnu.org/PR16407
8055 225. http://gcc.gnu.org/PR16643
8056 226. http://gcc.gnu.org/PR15927
8057 227. http://gcc.gnu.org/PR15948
8058 228. http://gcc.gnu.org/PR17019
8059 229. http://gcc.gnu.org/PR16130
8060 230. http://gcc.gnu.org/PR16142
8061 231. http://gcc.gnu.org/PR16278
8062 232. http://gcc.gnu.org/PR16414
8063 233. http://gcc.gnu.org/PR16445
8064 234. http://gcc.gnu.org/PR16490
8065 235. http://gcc.gnu.org/PR16683
8066 236. http://gcc.gnu.org/PR16195
8067 237. http://gcc.gnu.org/PR16239
8068 238. http://gcc.gnu.org/PR16199
8069 239. http://gcc.gnu.org/PR16416
8070 240. http://gcc.gnu.org/PR16430
8071 241. http://gcc.gnu.org/PR16379
8072 242. http://gcc.gnu.org/PR17093
8073 243. http://gcc.gnu.org/PR17119
8074 244. http://gcc.gnu.org/PR15928
8075 245. http://gcc.gnu.org/PR16210
8076 246. http://gcc.gnu.org/PR15488
8077 247. http://gcc.gnu.org/PR16250
8078 248. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.4.3
8079 249. http://gcc.gnu.org/PR17369
8080 250. http://gcc.gnu.org/PR17850
8081 251. http://gcc.gnu.org/PR13948
8082 252. http://gcc.gnu.org/PR14492
8083 253. http://gcc.gnu.org/PR16301
8084 254. http://gcc.gnu.org/PR16566
8085 255. http://gcc.gnu.org/PR17023
8086 256. http://gcc.gnu.org/PR17027
8087 257. http://gcc.gnu.org/PR17524
8088 258. http://gcc.gnu.org/PR17826
8089 259. http://gcc.gnu.org/PR15526
8090 260. http://gcc.gnu.org/PR16999
8091 261. http://gcc.gnu.org/PR17503
8092 262. http://gcc.gnu.org/PR17581
8093 263. http://gcc.gnu.org/PR18129
8094 264. http://gcc.gnu.org/PR10975
8095 265. http://gcc.gnu.org/PR11722
8096 266. http://gcc.gnu.org/PR14534
8097 267. http://gcc.gnu.org/PR15172
8098 268. http://gcc.gnu.org/PR15786
8099 269. http://gcc.gnu.org/PR16162
8100 270. http://gcc.gnu.org/PR16612
8101 271. http://gcc.gnu.org/PR16715
8102 272. http://gcc.gnu.org/PR16848
8103 273. http://gcc.gnu.org/PR17132
8104 274. http://gcc.gnu.org/PR17259
8105 275. http://gcc.gnu.org/PR17327
8106 276. http://gcc.gnu.org/PR17393
8107 277. http://gcc.gnu.org/PR17501
8108 278. http://gcc.gnu.org/PR17537
8109 279. http://gcc.gnu.org/PR17585
8110 280. http://gcc.gnu.org/PR17821
8111 281. http://gcc.gnu.org/PR17829
8112 282. http://gcc.gnu.org/PR17851
8113 283. http://gcc.gnu.org/PR17976
8114 284. http://gcc.gnu.org/PR18020
8115 285. http://gcc.gnu.org/PR18093
8116 286. http://gcc.gnu.org/PR18140
8117 287. http://gcc.gnu.org/PR17541
8118 288. http://gcc.gnu.org/PR17853
8119 289. http://gcc.gnu.org/PR17245
8120 290. http://gcc.gnu.org/PR17167
8121 291. http://gcc.gnu.org/PR17277
8122 292. http://gcc.gnu.org/PR17505
8123 293. http://gcc.gnu.org/PR17684
8124 294. http://gcc.gnu.org/PR17384
8125 295. http://gcc.gnu.org/PR17770
8126 296. http://gcc.gnu.org/PR11476
8127 297. http://gcc.gnu.org/PR14064
8128 298. http://gcc.gnu.org/PR14678
8129 299. http://gcc.gnu.org/PR15583
8130 300. http://gcc.gnu.org/PR15790
8131 301. http://gcc.gnu.org/PR15886
8132 302. http://gcc.gnu.org/PR16884
8133 303. http://gcc.gnu.org/PR13841
8134 304. http://gcc.gnu.org/PR15860
8135 305. http://gcc.gnu.org/PR17465
8136 306. http://gcc.gnu.org/PR17469
8137 307. http://gcc.gnu.org/PR18138
8138 308. http://gcc.gnu.org/PR15498
8139 309. http://gcc.gnu.org/PR15747
8140 310. http://gcc.gnu.org/PR16406
8141 311. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.4.4
8142 312. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.4.5
8143 313. http://gcc.gnu.org/PR24688
8144 314. http://gcc.gnu.org/PR17188
8145 315. http://gcc.gnu.org/PR20187
8146 316. http://gcc.gnu.org/PR21873
8147 317. http://gcc.gnu.org/PR21899
8148 318. http://gcc.gnu.org/PR22061
8149 319. http://gcc.gnu.org/PR22208
8150 320. http://gcc.gnu.org/PR22458
8151 321. http://gcc.gnu.org/PR22589
8152 322. http://gcc.gnu.org/PR24101
8153 323. http://gcc.gnu.org/PR10611
8154 324. http://gcc.gnu.org/PR13377
8155 325. http://gcc.gnu.org/PR16002
8156 326. http://gcc.gnu.org/PR17413
8157 327. http://gcc.gnu.org/PR17609
8158 328. http://gcc.gnu.org/PR17618
8159 329. http://gcc.gnu.org/PR18124
8160 330. http://gcc.gnu.org/PR18155
8161 331. http://gcc.gnu.org/PR18177
8162 332. http://gcc.gnu.org/PR18368
8163 333. http://gcc.gnu.org/PR18378
8164 334. http://gcc.gnu.org/PR18466
8165 335. http://gcc.gnu.org/PR18512
8166 336. http://gcc.gnu.org/PR18545
8167 337. http://gcc.gnu.org/PR18738
8168 338. http://gcc.gnu.org/PR18803
8169 339. http://gcc.gnu.org/PR19004
8170 340. http://gcc.gnu.org/PR19208
8171 341. http://gcc.gnu.org/PR19253
8172 342. http://gcc.gnu.org/PR19608
8173 343. http://gcc.gnu.org/PR19884
8174 344. http://gcc.gnu.org/PR20153
8175 345. http://gcc.gnu.org/PR20563
8176 346. http://gcc.gnu.org/PR20789
8177 347. http://gcc.gnu.org/PR21336
8178 348. http://gcc.gnu.org/PR21768
8179 349. http://gcc.gnu.org/PR21853
8180 350. http://gcc.gnu.org/PR21903
8181 351. http://gcc.gnu.org/PR21983
8182 352. http://gcc.gnu.org/PR21987
8183 353. http://gcc.gnu.org/PR22153
8184 354. http://gcc.gnu.org/PR22172
8185 355. http://gcc.gnu.org/PR21286
8186 356. http://gcc.gnu.org/PR22233
8187 357. http://gcc.gnu.org/PR22508
8188 358. http://gcc.gnu.org/PR22545
8189 359. http://gcc.gnu.org/PR23528
8190 360. http://gcc.gnu.org/PR23550
8191 361. http://gcc.gnu.org/PR23586
8192 362. http://gcc.gnu.org/PR23624
8193 363. http://gcc.gnu.org/PR23639
8194 364. http://gcc.gnu.org/PR23797
8195 365. http://gcc.gnu.org/PR23965
8196 366. http://gcc.gnu.org/PR24052
8197 367. http://gcc.gnu.org/PR24580
8198 368. http://gcc.gnu.org/PR24267
8199 369. http://gcc.gnu.org/PR17810
8200 370. http://gcc.gnu.org/PR17860
8201 371. http://gcc/gnu.org/PR21709
8202 372. http://gcc.gnu.org/PR21964
8203 373. http://gcc.gnu.org/PR22167
8204 374. http://gcc.gnu.org/PR22619
8205 375. http://gcc.gnu.org/PR23241
8206 376. http://gcc.gnu.org/PR23478
8207 377. http://gcc.gnu.org/PR24470
8208 378. http://gcc.gnu.org/PR24950
8209 379. http://gcc.gnu.org/PR14400
8210 380. http://gcc.gnu.org/PR14940
8211 381. http://gcc.gnu.org/PR20239
8212 382. http://gcc.gnu.org/PR15220
8213 383. http://gcc.gnu.org/PR19275
8214 384. http://gcc.gnu.org/PR21888
8215 385. http://gcc.gnu.org/PR15342
8216 386. http://gcc.gnu.org/PR23985
8217 387. http://gcc.gnu.org/PR16719
8218 388. http://gcc.gnu.org/PR21723
8219 389. http://gcc.gnu.org/PR21841
8220 390. http://gcc.gnu.org/PR23644
8221 391. http://gcc.gnu.org/PR24718
8222 392. http://gcc.gnu.org/PR18421
8223 393. http://gcc.gnu.org/PR20621
8224 394. http://gcc.gnu.org/PR18583
8225 395. http://gcc.gnu.org/PR20191
8226 396. http://gcc.gnu.org/PR22083
8227 397. http://gcc.gnu.org/PR23070
8228 398. http://gcc.gnu.org/PR23404
8229 399. http://gcc.gnu.org/PR23539
8230 400. http://gcc.gnu.org/PR24102
8231 401. http://gcc.gnu.org/PR24465
8232 402. http://gcc.gnu.org/PR19933
8233 403. http://gcc.gnu.org/PR21889
8234 404. http://gcc.gnu.org/PR19300
8235 405. http://gcc.gnu.org/PR20301
8236 406. http://gcc.gnu.org/PR20673
8237 407. http://gcc.gnu.org/PR18582
8238 408. http://gcc.gnu.org/PR19340
8239 409. http://gcc.gnu.org/PR21716
8240 410. http://gcc.gnu.org/PR24315
8241 411. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.4.6
8242 412. http://gcc.gnu.org/onlinedocs/
8243 413. mailto:gcc-help@gcc.gnu.org
8244 414. mailto:gcc@gcc.gnu.org
8245 415. http://gcc.gnu.org/lists.html
8246 416. http://www.fsf.org/
8247 417. http://gcc.gnu.org/about.html
8248 418. http://validator.w3.org/check/referer
8249 ======================================================================
8250 http://gcc.gnu.org/gcc-3.3/index.html
8251 GCC 3.3 Release Series
8255 The [1]GNU project and the GCC developers are pleased to announce the
8256 release of GCC 3.3.6.
8258 This release is a bug-fix release, containing fixes for regressions in
8259 GCC 3.3.5 relative to previous releases of GCC.
8261 This release is the last of the series 3.3.x.
8263 The GCC 3.3 release series includes numerous [2]new features,
8264 improvements, bug fixes, and other changes, thanks to an [3]amazing
8265 group of volunteers.
8270 May 3, 2005 ([4]changes)
8273 September 30, 2004 ([5]changes)
8276 May 31, 2004 ([6]changes)
8279 February 14, 2004 ([7]changes)
8282 October 16, 2003 ([8]changes)
8285 August 8, 2003 ([9]changes)
8288 May 14, 2003 ([10]changes)
8290 References and Acknowledgements
8292 GCC used to stand for the GNU C Compiler, but since the compiler
8293 supports several other languages aside from C, it now stands for the
8294 GNU Compiler Collection.
8296 A list of [11]successful builds is updated as new information becomes
8299 The GCC developers would like to thank the numerous people that have
8300 contributed new features, improvements, bug fixes, and other changes as
8301 well as test results to GCC. This [12]amazing group of volunteers is
8302 what makes GCC successful.
8304 For additional information about GCC please refer to the [13]GCC
8305 project web site or contact the [14]GCC development mailing list.
8307 To obtain GCC please use [15]our mirror sites, or our CVS server.
8310 For questions related to the use of GCC, please consult these web
8311 pages and the [16]GCC manuals. If that fails, the
8312 [17]gcc-help@gcc.gnu.org mailing list might help. Comments on these
8313 web pages and the development of GCC are welcome on our developer
8314 list at [18]gcc@gcc.gnu.org. All of [19]our lists have public
8317 Copyright (C) [20]Free Software Foundation, Inc. Verbatim copying and
8318 distribution of this entire article is permitted in any medium,
8319 provided this notice is preserved.
8321 These pages are [21]maintained by the GCC team. Last modified
8326 1. http://www.gnu.org/
8327 2. http://gcc.gnu.org/gcc-3.3/changes.html
8328 3. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
8329 4. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.6
8330 5. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.5
8331 6. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.4
8332 7. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.3
8333 8. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.2
8334 9. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.1
8335 10. http://gcc.gnu.org/gcc-3.3/changes.html
8336 11. http://gcc.gnu.org/gcc-3.3/buildstat.html
8337 12. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
8338 13. http://gcc.gnu.org/index.html
8339 14. mailto:gcc@gcc.gnu.org
8340 15. http://gcc.gnu.org/mirrors.html
8341 16. http://gcc.gnu.org/onlinedocs/
8342 17. mailto:gcc-help@gcc.gnu.org
8343 18. mailto:gcc@gcc.gnu.org
8344 19. http://gcc.gnu.org/lists.html
8345 20. http://www.fsf.org/
8346 21. http://gcc.gnu.org/about.html
8347 22. http://validator.w3.org/check/referer
8348 ======================================================================
8349 http://gcc.gnu.org/gcc-3.3/changes.html
8350 GCC 3.3 Release Series
8351 Changes, New Features, and Fixes
8353 The latest release in the 3.3 release series is [1]GCC 3.3.6.
8357 * The preprocessor no longer accepts multi-line string literals. They
8358 were deprecated in 3.0, 3.1, and 3.2.
8359 * The preprocessor no longer supports the -A- switch when appearing
8360 alone. -A- followed by an assertion is still supported.
8361 * Support for all the systems [2]obsoleted in GCC 3.1 has been
8362 removed from GCC 3.3. See below for a [3]list of systems which are
8363 obsoleted in this release.
8364 * Checking for null format arguments has been decoupled from the rest
8365 of the format checking mechanism. Programs which use the format
8366 attribute may regain this functionality by using the new [4]nonnull
8367 function attribute. Note that all functions for which GCC has a
8368 built-in format attribute, an appropriate built-in nonnull
8369 attribute is also applied.
8370 * The DWARF (version 1) debugging format has been deprecated and will
8371 be removed in a future version of GCC. Version 2 of the DWARF
8372 debugging format will continue to be supported for the foreseeable
8374 * The C and Objective-C compilers no longer accept the "Naming Types"
8375 extension (typedef foo = bar); it was already unavailable in C++.
8376 Code which uses it will need to be changed to use the "typeof"
8377 extension instead: typedef typeof(bar) foo. (We have removed this
8378 extension without a period of deprecation because it has caused the
8379 compiler to crash since version 3.0 and no one noticed until very
8380 recently. Thus we conclude it is not in widespread use.)
8381 * The -traditional C compiler option has been removed. It was
8382 deprecated in 3.1 and 3.2. (Traditional preprocessing remains
8383 available.) The <varargs.h> header, used for writing variadic
8384 functions in traditional C, still exists but will produce an error
8386 * GCC 3.3.1 automatically places zero-initialized variables in the
8387 .bss section on some operating systems. Versions of GNU Emacs up to
8388 (and including) 21.3 will not work correctly when using this
8389 optimization; you can use -fno-zero-initialized-in-bss to disable
8392 General Optimizer Improvements
8394 * A new scheme for accurately describing processor pipelines, the
8395 [5]DFA scheduler, has been added.
8396 * Pavel Nejedly, Charles University Prague, has contributed new file
8397 format used by the edge coverage profiler (-fprofile-arcs).
8398 The new format is robust and diagnoses common mistakes where
8399 profiles from different versions (or compilations) of the program
8400 are combined resulting in nonsensical profiles and slow code to
8401 produced with profile feedback. Additionally this format allows
8402 extra data to be gathered. Currently, overall statistics are
8403 produced helping optimizers to identify hot spots of a program
8404 globally replacing the old intra-procedural scheme and resulting in
8405 better code. Note that the gcov tool from older GCC versions will
8406 not be able to parse the profiles generated by GCC 3.3 and vice
8408 * Jan Hubicka, SuSE Labs, has contributed a new superblock formation
8409 pass enabled using -ftracer. This pass simplifies the control flow
8410 of functions allowing other optimizations to do better job.
8411 He also contributed the function reordering pass
8412 (-freorder-functions) to optimize function placement using profile
8415 New Languages and Language specific improvements
8419 * The preprocessor now accepts directives within macro arguments. It
8420 processes them just as if they had not been within macro arguments.
8421 * The separate ISO and traditional preprocessors have been completely
8422 removed. The front end handles either type of preprocessed output
8424 * In C99 mode preprocessor arithmetic is done in the precision of the
8425 target's intmax_t, as required by that standard.
8426 * The preprocessor can now copy comments inside macros to the output
8427 file when the macro is expanded. This feature, enabled using the
8428 -CC option, is intended for use by applications which place
8429 metadata or directives inside comments, such as lint.
8430 * The method of constructing the list of directories to be searched
8431 for header files has been revised. If a directory named by a -I
8432 option is a standard system include directory, the option is
8433 ignored to ensure that the default search order for system
8434 directories and the special treatment of system header files are
8436 * A few more [6]ISO C99 features now work correctly.
8437 * A new function attribute, nonnull, has been added which allows
8438 pointer arguments to functions to be specified as requiring a
8439 non-null value. The compiler currently uses this information to
8440 issue a warning when it detects a null value passed in such an
8442 * A new type attribute, may_alias, has been added. Accesses to
8443 objects with types with this attribute are not subjected to
8444 type-based alias analysis, but are instead assumed to be able to
8445 alias any other type of objects, just like the char type.
8449 * Type based alias analysis has been implemented for C++ aggregate
8454 * Generate an error if Objective-C objects are passed by value in
8455 function and method calls.
8456 * When -Wselector is used, check the whole list of selectors at the
8457 end of compilation, and emit a warning if a @selector() is not
8459 * Define __NEXT_RUNTIME__ when compiling for the NeXT runtime.
8460 * No longer need to include objc/objc-class.h to compile self calls
8461 in class methods (NeXT runtime only).
8462 * New -Wundeclared-selector option.
8463 * Removed selector bloating which was causing object files to be 10%
8464 bigger on average (GNU runtime only).
8465 * Using at run time @protocol() objects has been fixed in certain
8466 situations (GNU runtime only).
8467 * Type checking has been fixed and improved in many situations
8468 involving protocols.
8472 * The java.sql and javax.sql packages now implement the JDBC 3.0 (JDK
8474 * The JDK 1.4 assert facility has been implemented.
8475 * The bytecode interpreter is now direct threaded and thus faster.
8479 * Fortran improvements are listed in [7]the Fortran documentation.
8483 * Ada tasking now works with glibc 2.3.x threading libraries.
8485 New Targets and Target Specific Improvements
8487 * The following changes have been made to the HP-PA port:
8488 + The port now defaults to scheduling for the PA8000 series of
8490 + Scheduling support for the PA7300 processor has been added.
8491 + The 32-bit port now supports weak symbols under HP-UX 11.
8492 + The handling of initializers and finalizers has been improved
8493 under HP-UX 11. The 64-bit port no longer uses collect2.
8494 + Dwarf2 EH support has been added to the 32-bit GNU/Linux port.
8495 + ABI fixes to correct the passing of small structures by value.
8496 * The SPARC, HP-PA, SH4, and x86/pentium ports have been converted to
8497 use the DFA processor pipeline description.
8498 * The following NetBSD configurations for the SuperH processor family
8500 + SH3, big-endian, sh-*-netbsdelf*
8501 + SH3, little-endian, shle-*-netbsdelf*
8502 + SH5, SHmedia, big-endian, 32-bit default, sh5-*-netbsd*
8503 + SH5, SHmedia, little-endian, 32-bit default, sh5le-*-netbsd*
8504 + SH5, SHmedia, big-endian, 64-bit default, sh64-*-netbsd*
8505 + SH5, SHmedia, little-endian, 64-bit default, sh64le-*-netbsd*
8506 * The following changes have been made to the IA-32/x86-64 port:
8507 + SSE2 and 3dNOW! intrinsics are now supported.
8508 + Support for thread local storage has been added to the IA-32
8510 + The x86-64 port has been significantly improved.
8511 * The following changes have been made to the MIPS port:
8512 + All configurations now accept the -mabi switch. Note that you
8513 will need appropriate multilibs for this option to work
8515 + ELF configurations will always pass an ABI flag to the
8516 assembler, except when the MIPS EABI is selected.
8517 + -mabi=64 no longer selects MIPS IV code.
8518 + The -mcpu option, which was deprecated in 3.1 and 3.2, has
8519 been removed from this release.
8520 + -march now changes the core ISA level. In previous releases,
8521 it would change the use of processor-specific extensions, but
8522 would leave the core ISA unchanged. For example, mips64-elf
8523 -march=r8000 will now generate MIPS IV code.
8524 + Under most configurations, -mipsN now acts as a synonym for
8526 + There are some new preprocessor macros to describe the -march
8527 and -mtune settings. See the documentation of those options
8529 + Support for the NEC VR-Series processors has been added. This
8530 includes the 54xx, 5500, and 41xx series.
8531 + Support for the Sandcraft sr71k processor has been added.
8532 * The following changes have been made to the S/390 port:
8533 + Support to build the Java runtime libraries has been added.
8534 Java is now enabled by default on s390-*-linux* and
8535 s390x-*-linux* targets.
8536 + Multilib support for the s390x-*-linux* target has been added;
8537 this allows to build 31-bit binaries using the -m31 option.
8538 + Support for thread local storage has been added.
8539 + Inline assembler code may now use the 'Q' constraint to
8540 specify memory operands without index register.
8541 + Various platform-specific performance improvements have been
8542 implemented; in particular, the compiler now uses the BRANCH
8543 ON COUNT family of instructions and makes more frequent use of
8544 the TEST UNDER MASK family of instructions.
8545 * The following changes have been made to the PowerPC port:
8546 + Support for IBM Power4 processor added.
8547 + Support for Motorola e500 SPE added.
8548 + Support for AIX 5.2 added.
8549 + Function and Data sections now supported on AIX.
8550 + Sibcall optimizations added.
8551 * The support for H8 Tiny is added to the H8/300 port with -mn.
8555 Support for a number of older systems has been declared obsolete in GCC
8556 3.3. Unless there is activity to revive them, the next release of GCC
8557 will have their sources permanently removed.
8559 All configurations of the following processor architectures have been
8561 * Matsushita MN10200, mn10200-*-*
8562 * Motorola 88000, m88k-*-*
8563 * IBM ROMP, romp-*-*
8565 Also, some individual systems have been obsoleted:
8567 + Interix, alpha*-*-interix*
8568 + Linux libc1, alpha*-*-linux*libc1*
8569 + Linux ECOFF, alpha*-*-linux*ecoff*
8571 + Generic a.out, arm*-*-aout*
8572 + Conix, arm*-*-conix*
8573 + "Old ABI," arm*-*-oabi
8574 + StrongARM/COFF, strongarm-*-coff*
8576 + Generic OSF, hppa1.0-*-osf*
8577 + Generic BSD, hppa1.0-*-bsd*
8578 + HP/UX versions 7, 8, and 9, hppa1.[01]-*-hpux[789]*
8579 + HiUX, hppa*-*-hiux*
8580 + Mach Lites, hppa*-*-lites*
8582 + Windows NT 3.x, i?86-*-win32
8584 + HP systems, m68000-hp-bsd* and m68k-hp-bsd*
8585 + Sun systems, m68000-sun-sunos*, m68k-sun-sunos*, and
8587 + AT&T systems, m68000-att-sysv*
8588 + Atari systems, m68k-atari-sysv*
8589 + Motorola systems, m68k-motorola-sysv*
8590 + NCR systems, m68k-ncr-sysv*
8591 + Plexus systems, m68k-plexus-sysv*
8592 + Commodore systems, m68k-cbm-sysv*
8593 + Citicorp TTI, m68k-tti-*
8594 + Unos, m68k-crds-unos*
8595 + Concurrent RTU, m68k-ccur-rtu*
8596 + Linux a.out, m68k-*-linux*aout*
8597 + Linux libc1, m68k-*-linux*libc1*
8598 + pSOS, m68k-*-psos*
8600 + Generic ECOFF, mips*-*-ecoff*
8601 + SINIX, mips-sni-sysv4
8602 + Orion RTEMS, mips64orion-*-rtems*
8603 * National Semiconductor 32000
8604 + OpenBSD, ns32k-*-openbsd*
8605 * POWER (aka RS/6000) and PowerPC
8606 + AIX versions 1, 2, and 3, rs6000-ibm-aix[123]*
8607 + Bull BOSX, rs6000-bull-bosx
8608 + Generic Mach, rs6000-*-mach*
8609 + Generic SysV, powerpc*-*-sysv*
8610 + Linux libc1, powerpc*-*-linux*libc1*
8612 + Generic a.out, sparc-*-aout*, sparclet-*-aout*,
8613 sparclite-*-aout*, and sparc86x-*-aout*
8614 + NetBSD a.out, sparc-*-netbsd*aout*
8615 + Generic BSD, sparc-*-bsd*
8616 + ChorusOS, sparc-*-chorusos*
8617 + Linux a.out, sparc-*-linux*aout*
8618 + Linux libc1, sparc-*-linux*libc1*
8619 + LynxOS, sparc-*-lynxos*
8620 + Solaris on HAL hardware, sparc-hal-solaris2*
8621 + SunOS versions 3 and 4, sparc-*-sunos[34]*
8623 + RTEMS, v850-*-rtems*
8627 Documentation improvements
8629 Other significant improvements
8631 * Almost all front-end dependencies in the compiler have been
8632 separated out into a set of language hooks. This should make adding
8633 a new front end clearer and easier.
8634 * One effect of removing the separate preprocessor is a small
8635 increase in the robustness of the compiler in general, and the
8636 maintainability of target descriptions. Previously target-specific
8637 built-in macros and others, such as __FAST_MATH__, had to be
8638 handled with so-called specs that were hard to maintain. Often they
8639 would fail to behave properly when conflicting options were
8640 supplied on the command line, and define macros in the user's
8641 namespace even when strict ISO compliance was requested.
8642 Integrating the preprocessor has cleanly solved these issues.
8643 * The Makefile suite now supports redirection of make install by
8644 means of the variable DESTDIR.
8645 __________________________________________________________________
8649 Detailed release notes for the GCC 3.3 release follow.
8655 * [8]10140 cross compiler build failures: missing __mempcpy (DUP:
8658 Internal compiler errors (multi-platform)
8660 * [11]3581 large string causes segmentation fault in cc1
8661 * [12]4382 __builtin_{set,long}jmp with -O3 can crash the compiler
8662 * [13]5533 (c++) ICE when processing std::accumulate(begin, end,
8664 * [14]6387 -fpic -gdwarf-2 -g1 combination gives ICE in dwarf2out
8665 * [15]6412 (c++) ICE in retrieve_specialization
8666 * [16]6620 (c++) partial template specialization causes an ICE
8667 (segmentation fault)
8668 * [17]6663 (c++) ICE with attribute aligned
8669 * [18]7068 ICE with incomplete types
8670 * [19]7083 (c++) ICE using -gstabs with dodgy class derivation
8671 * [20]7647 (c++) ICE when data member has the name of the enclosing
8673 * [21]7675 ICE in fixup_var_refs_1
8674 * [22]7718 'complex' template instantiation causes ICE
8675 * [23]8116 (c++) ICE in member template function
8676 * [24]8358 (ada) Ada compiler accesses freed memory, crashes
8677 * [25]8511 (c++) ICE: (hopefully) reproducible cc1plus segmentation
8679 * [26]8564 (c++) ICE in find_function_data, in function.c
8680 * [27]8660 (c++) template overloading ICE in tsubst_expr, in cp/pt.c
8681 * [28]8766 (c++) ICE after failed initialization of static template
8683 * [29]8803 ICE in instantiate_virtual_regs_1, in function.c
8684 * [30]8846 (c++) ICE after diagnostic if fr_FR@euro locale is set
8685 * [31]8906 (c++) ICE (Segmentation fault) when parsing nested-class
8687 * [32]9216 (c++) ICE on missing template parameter
8688 * [33]9261 (c++) ICE in arg_assoc, in cp/decl2.c
8689 * [34]9263 (fortran) ICE caused by invalid PARAMETER in implied DO
8691 * [35]9429 (c++) ICE in template instantiation with a pointered new
8693 * [36]9516 Internal error when using a big array
8694 * [37]9600 (c++) ICE with typedefs in template class
8695 * [38]9629 (c++) virtual inheritance segfault
8696 * [39]9672 (c++) ICE: Error reporting routines re-entered
8697 * [40]9749 (c++) ICE in write_expression on invalid function
8699 * [41]9794 (fortran) ICE: floating point exception during constant
8701 * [42]9829 (c++) Missing colon in nested namespace usage causes ICE
8702 * [43]9916 (c++) ICE with noreturn function in ?: statement
8703 * [44]9936 ICE with local function and variable-length 2d array
8704 * [45]10262 (c++) cc1plus crashes with large generated code
8705 * [46]10278 (c++) ICE in parser for invalid code
8706 * [47]10446 (c++) ICE on definition of nonexistent member function of
8707 nested class in a class template
8708 * [48]10451 (c++) ICE in grokdeclarator on spurious mutable
8710 * [49]10506 (c++) ICE in build_new at cp/init.c with
8711 -fkeep-inline-functions and multiple inheritance
8712 * [50]10549 (c++) ICE in store_bit_field on bitfields that exceed the
8713 precision of the declared type
8717 * [51]2001 Inordinately long compile times in reload CSE regs
8718 * [52]2391 Exponential compilation time explosion in combine
8719 * [53]2960 Duplicate loop conditions even with -Os
8720 * [54]4046 redundant conditional branch
8721 * [55]6405 Loop-unrolling related performance regressions
8722 * [56]6798 very long compile time with large case-statement
8723 * [57]6871 const objects shouldn't be moved to .bss
8724 * [58]6909 problem w/ -Os on modified loop-2c.c test case
8725 * [59]7189 gcc -O2 -Wall does not print ``control reaches end of
8726 non-void function'' warning
8727 * [60]7642 optimization problem with signbit()
8728 * [61]8634 incorrect code for inlining of memcpy under -O2
8729 * [62]8750 Cygwin prolog generation erroneously emitting __alloca as
8730 regular function call
8734 * [63]2161 long if-else cascade overflows parser stack
8735 * [64]4319 short accepted on typedef'd char
8736 * [65]8602 incorrect line numbers in warning messages when using
8738 * [66]9177 -fdump-translation-unit: C front end deletes function_decl
8739 AST nodes and breaks debugging dumps
8740 * [67]9853 miscompilation of non-constant structure initializer
8742 c++ compiler and library
8744 * [68]45 legal template specialization code is rejected (DUP:
8746 * [70]764 lookup failure: friend operator and dereferencing a pointer
8747 and templates (DUP: [71]5116)
8748 * [72]2862 gcc accepts invalid explicit instantiation syntax (DUP:
8750 * [73]3663 G++ doesn't check access control during template
8752 * [74]3797 gcc fails to emit explicit specialization of a template
8754 * [75]3948 Two destructors are called when no copy destructor is
8755 defined (ABI change)
8756 * [76]4137 Conversion operator within template is not accepted
8757 * [77]4361 bogus ambiguity taking the address of a member template
8758 * [78]4802 g++ accepts illegal template code (access to private
8759 member; DUP: [79]5837)
8760 * [80]4803 inline function is used but never defined, and g++ does
8762 * [81]5094 Partial specialization cannot be friend?
8763 * [82]5730 complex<double>::norm() -- huge slowdown from egcs-2.91.66
8764 * [83]6713 Regression wrt 3.0.4: g++ -O2 leads to seg fault at run
8766 * [84]7015 certain __asm__ constructs rejected
8767 * [85]7086 compile time regression (quadratic behavior in
8769 * [86]7099 G++ doesn't set the noreturn attribute on std::exit and
8771 * [87]7247 copy constructor missing when inlining enabled (invalid
8773 * [88]7441 string array initialization compilation time regression
8774 from seconds to minutes
8775 * [89]7768 __PRETTY_FUNCTION__ for template destructor is wrong
8776 * [90]7804 bad printing of floating point constant in warning message
8777 * [91]8099 Friend classes and template specializations
8778 * [92]8117 member function pointers and multiple inheritance
8779 * [93]8205 using declaration and multiple inheritance
8780 * [94]8645 unnecessary non-zero checks in stl_tree.h
8781 * [95]8724 explicit destructor call for incomplete class allowed
8782 * [96]8805 compile time regression with many member variables
8783 * [97]8691 -O3 and -fno-implicit-templates are incompatible
8784 * [98]8700 unhelpful error message for binding temp to reference
8785 * [99]8724 explicit destructor call for incomplete class allowed
8786 * [100]8949 numeric_limits<>::denorm_min() and is_iec559 problems
8787 * [101]9016 Failure to consistently constant fold "constant" C++
8789 * [102]9053 g++ confused about ambiguity of overloaded function
8791 * [103]9152 undefined virtual thunks
8792 * [104]9182 basic_filebuf<> does not report errors in codecvt<>::out
8793 * [105]9297 data corruption due to codegen bug (when copying.)
8794 * [106]9318 i/ostream::operator>>/<<(streambuf*) broken
8795 * [107]9320 Incorrect usage of traits_type::int_type in stdio_filebuf
8796 * [108]9400 bogus -Wshadow warning: shadowed declaration of this in
8798 * [109]9424 i/ostream::operator>>/<<(streambuf*) drops characters
8799 * [110]9425 filebuf::pbackfail broken (DUP: [111]9439)
8800 * [112]9474 GCC freezes in compiling a weird code mixing <iostream>
8802 * [113]9548 Incorrect results from setf(ios::fixed) and precision(-1)
8804 * [115]9555 ostream inserters fail to set badbit on exception
8805 * [116]9561 ostream inserters rethrow exception of wrong type
8806 * [117]9563 ostream::sentry returns true after a failed preparation
8807 * [118]9582 one-definition rule violation in std::allocator
8808 * [119]9622 __PRETTY_FUNCTION__ incorrect in template destructors
8809 * [120]9683 bug in initialization chains for static const variables
8810 from template classes
8811 * [121]9791 -Woverloaded-virtual reports hiding of destructor
8812 * [122]9817 collate::compare doesn't handle nul characters
8813 * [123]9825 filebuf::sputbackc breaks sbumpc
8814 * [124]9826 operator>>(basic_istream, basic_string) fails to compile
8816 * [125]9924 Multiple using statements for builtin functions not
8818 * [126]9946 destructor is not called for temporary object
8819 * [127]9964 filebuf::close() sometimes fails to close file
8820 * [128]9988 filebuf::overflow writes EOF to file
8821 * [129]10033 optimization breaks polymorphic references w/ typeid
8823 * [130]10097 filebuf::underflow drops characters
8824 * [131]10132 filebuf destructor can throw exceptions
8825 * [132]10180 gcc fails to warn about non-inlined function
8826 * [133]10199 method parametrized by template does not work everywhere
8827 * [134]10300 use of array-new (nothrow) in segfaults on NULL return
8828 * [135]10427 Stack corruption with variable-length automatic arrays
8829 and virtual destructors
8830 * [136]10503 Compilation never stops in fixed_type_or_null
8834 * [137]5956 selectors aren't matched properly when added to the
8837 Fortran compiler and library
8839 * [138]1832 list directed i/o overflow hangs, -fbounds-check doesn't
8841 * [139]3924 g77 generates code that is rejected by GAS if COFF debug
8843 * [140]5634 doc: explain that configure --prefix=~/... does not work
8844 * [141]6367 multiple repeat counts confuse namelist read into array
8845 * [142]6491 Logical operations error on logicals when using
8847 * [143]6742 Generation of C++ Prototype for FORTRAN and extern "C"
8848 * [144]7113 Failure of g77.f-torture/execute/f90-intrinsic-bit.f -Os
8850 * [145]7236 OPEN(...,RECL=nnn,...) without ACCESS='DIRECT' should
8851 assume a direct access file
8852 * [146]7278 g77 "bug"; the executable misbehaves (with -O2
8854 * [147]7384 DATE_AND_TIME milliseconds field inactive on Windows
8855 * [148]7388 Incorrect output with 0-based array of characters
8856 * [149]8587 Double complex zero ** double precision number -> NaN
8858 * [150]9038 -ffixed-line-length-none -x f77-cpp-input gives: Warning:
8859 unknown register name line-length-none
8860 * [151]10197 Direct access files not unformatted by default
8862 Java compiler and library
8864 * [152]6005 gcj fails to build rhug on alpha
8865 * [153]6389 System.getProperty("") should always throw an
8866 IllegalArgumentException
8867 * [154]6576 java.util.ResourceBundle.getResource ignores locale
8868 * [155]6652 new java.io.File("").getCanonicalFile() throws exception
8869 * [156]7060 getMethod() doesn't search super interface
8870 * [157]7073 bytecode interpreter gives wrong answer for interface
8872 * [158]7180 possible bug in
8873 javax.naming.spi.NamingManager.getPlusPath()
8874 * [159]7416 java.security startup refs "GNU libgcj.security"
8875 * [160]7570 Runtime.exec with null envp: child doesn't inherit parent
8876 env (DUP: [161]7578)
8877 * [162]7611 Internal error while compiling libjava with -O
8878 * [163]7709 NullPointerException in _Jv_ResolvePoolEntry
8879 * [164]7766 ZipInputStream.available returns 0 immediately after
8881 * [165]7785 Calendar.getTimeInMillis/setTimeInMillis should be public
8882 * [166]7786 TimeZone.getDSTSavings() from JDK1.4 not implemented
8883 * [167]8142 '$' in class names vs. dlopen 'dynamic string tokens'
8884 * [168]8234 ZipInputStream chokes when InputStream.read() returns
8886 * [169]8415 reflection bug: exception info for Method
8887 * [170]8481 java.Random.nextInt(int) may return negative
8888 * [171]8593 Error reading GZIPped files with BufferedReader
8889 * [172]8759 java.beans.Introspector has no flushCaches() or
8890 flushFromCaches() methods
8891 * [173]8997 spin() calls Thread.sleep
8892 * [174]9253 on win32, java.io.File.listFiles("C:\\") returns pwd
8893 instead of the root content of C:
8894 * [175]9254 java::lang::Object::wait(), threads-win32.cc returns
8896 * [176]9271 Severe bias in java.security.SecureRandom
8898 Ada compiler and library
8900 * [177]6767 make gnatlib-shared fails on -laddr2line
8901 * [178]9911 gnatmake fails to link when GCC configured with
8902 --with-sjlj-exceptions=yes
8903 * [179]10020 Can't bootstrap gcc on AIX with Ada enabled
8904 * [180]10546 Ada tasking not working on Red Hat 9
8908 * [181]7029 preprocessor should ignore #warning with -M
8912 * [182]2903 [arm] Optimization bug with long long arithmetic
8913 * [183]7873 arm-linux-gcc fails when assigning address to a bit field
8917 * [184]7680 float functions undefined in math.h/cmath with #define
8920 HP-UX or HP-PA-specific
8922 * [185]8705 [HP-PA] ICE in emit_move_insn_1, in expr.c
8923 * [186]9986 [HP-UX] Incorrect transformation of fputs_unlocked to
8925 * [187]10056 [HP-PA] ICE at -O2 when building c++ code from doxygen
8929 * [188]6744 Bad assembler code generated: reference to pseudo
8931 * [189]7361 Internal compiler error in reload_cse_simplify_operands,
8936 * [190]9496 [mips-linux] bug in optimizer?
8940 * [191]7067 -Os with -mcpu=powerpc optimizes for speed (?) instead of
8942 * [192]8480 reload ICEs for LAPACK code on powerpc64-linux
8943 * [193]8784 [AIX] Internal compiler error in simplify_gen_subreg
8944 * [194]10315 [powerpc] ICE: in extract_insn, in recog.c
8948 * [195]10267 (documentation) Wrong build instructions for
8951 x86-specific (Intel/AMD)
8953 * [196]7916 ICE in instantiate_virtual_register_1
8954 * [197]7926 (c++) i486 instructions in header files make c++ programs
8956 * [198]8555 ICE in gen_split_1231
8957 * [199]8994 ICE with -O -march=pentium4
8958 * [200]9426 ICE with -fssa -funroll-loops -fprofile-arcs
8959 * [201]9806 ICE in inline assembly with -fPIC flag
8960 * [202]10077 gcc -msse2 generates movd to move dwords between xmm
8962 * [203]10233 64-bit comparison only comparing bottom 32-bits
8963 * [204]10286 type-punning doesn't work with __m64 and -O
8964 * [205]10308 [x86] ICE with -O -fgcse or -O2
8965 __________________________________________________________________
8971 This section lists the problem reports (PRs) from GCC's bug tracking
8972 system that are known to be fixed in the 3.3.1 release. This list might
8973 not be complete (that is, it is possible that some PRs that have been
8974 fixed are not listed here).
8978 * [206]11272 [Solaris] make bootstrap fails while building libstdc++
8980 Internal compiler errors (multi-platform)
8982 * [207]5754 ICE on invalid nested template class
8983 * [208]6597 ICE in set_mem_alias_set compiling Qt with -O2 on ia64
8984 and --enable-checking
8985 * [209]6949 (c++) ICE in tsubst_decl, in cp/pt.c
8986 * [210]7053 (c++) ICE when declaring a function already defined as a
8987 friend method of a template class
8988 * [211]8164 (c++) ICE when using different const expressions as
8990 * [212]8384 (c++) ICE in is_base_type, in dwarf2out.c
8991 * [213]9559 (c++) ICE with invalid initialization of a static const
8992 * [214]9649 (c++) ICE in finish_member_declaration, in cp/semantics.c
8993 when redeclaring a static member variable
8994 * [215]9864 (fortran) ICE in add_abstract_origin_attribute, in
8995 dwarfout.c with -g -O -finline-functions
8996 * [216]10432 (c++) ICE in poplevel, in cp/decl.c
8997 * [217]10475 ICE in subreg_highpart_offset for code with long long
8998 * [218]10635 (c++) ICE when dereferencing an incomplete type casted
9000 * [219]10661 (c++) ICE in instantiate_decl, in cp/pt.c while
9001 instantiating static member variables
9002 * [220]10700 ICE in copy_to_mode_reg on 64-bit targets
9003 * [221]10712 (c++) ICE in constructor_name_full, in cp/decl2.c
9004 * [222]10796 (c++) ICE when defining an enum with two values: -1 and
9006 * [223]10890 ICE in merge_assigned_reloads building Linux 2.4.2x
9008 * [224]10939 (c++) ICE with template code
9009 * [225]10956 (c++) ICE when specializing a template member function
9010 of a template class, in tsubst, in cp/pt.c
9011 * [226]11041 (c++) ICE: const myclass &x = *x; (when operator*()
9013 * [227]11059 (c++) ICE with empty union
9014 * [228]11083 (c++) ICE in commit_one_edge_insertion, in cfgrtl.c with
9015 -O2 -fnon-call-exceptions
9016 * [229]11105 (c++) ICE in mangle_conv_op_name_for_type
9017 * [230]11149 (c++) ICE on error when instantiation with call function
9019 * [231]11228 (c++) ICE on new-expression using array operator new and
9020 default-initialization
9021 * [232]11282 (c++) Infinite memory usage after syntax error
9022 * [233]11301 (fortran) ICE with -fno-globals
9023 * [234]11308 (c++) ICE when using an enum type name as if it were a
9025 * [235]11473 (c++) ICE with -gstabs when empty struct inherits from
9027 * [236]11503 (c++) ICE when instantiating template with ADDR_EXPR
9028 * [237]11513 (c++) ICE in push_template_decl_real, in cp/pt.c:
9029 template member functions
9033 * [238]11198 -O2 -frename-registers generates wrong code (aliasing
9035 * [239]11304 Wrong code production with -fomit-frame-pointer
9036 * [240]11381 volatile memory access optimized away
9037 * [241]11536 [strength-reduce] -O2 optimization produces wrong code
9038 * [242]11557 constant folding bug generates wrong code
9042 * [243]5897 No warning for statement after return
9043 * [244]11279 DWARF-2 output mishandles large enums
9047 * [245]11022 no warning for non-compatible macro redefinition
9049 C++ compiler and library
9051 * [246]2330 static_cast<>() to a private base is allowed
9052 * [247]5388 Incorrect message "operands to ?: have different types"
9053 * [248]5390 Libiberty fails to demangle multi-digit template
9055 * [249]7877 Incorrect parameter passing to specializations of member
9057 * [250]9393 Anonymous namespaces and compiling the same file twice
9058 * [251]10032 -pedantic converts some errors to warnings
9059 * [252]10468 const typeof(x) is non-const, but only in templates
9060 * [253]10527 confused error message with "new int()" parameter
9062 * [254]10679 parameter MIN_INLINE_INSNS is not honored
9063 * [255]10682 gcc chokes on a typedef for an enum inside a class
9065 * [256]10689 pow(std::complex(0),1/3) returns (nan, nan) instead of
9067 * [257]10845 template member function (with nested template as
9068 parameter) cannot be called anymore if another unrelated template
9069 member function is defined
9070 * [258]10849 Cannot define an out-of-class specialization of a
9071 private nested template class
9072 * [259]10888 Suppress -Winline warnings for system headers
9073 * [260]10929 -Winline warns about functions for which no definition
9075 * [261]10931 valid conversion static_cast<const unsigned
9076 int&>(lvalue-of-type-int) is rejected
9077 * [262]10940 Bad code with explicit specialization
9078 * [263]10968 If member function implicitly instantiated, explicit
9079 instantiation of class fails to instantiate it
9080 * [264]10990 Cannot convert with dynamic_cast<> to a private base
9081 class from within a member function
9082 * [265]11039 Bad interaction between implicit typename deprecation
9084 * [266]11062 (libstdc++) avoid __attribute__ ((unused)); say
9085 "__unused__" instead
9086 * [267]11095 C++ iostream manipulator causes segfault when called
9087 with negative argument
9088 * [268]11098 g++ doesn't emit complete debugging information for
9089 local variables in destructors
9090 * [269]11137 GNU/Linux shared library constructors not called unless
9091 there's one global object
9092 * [270]11154 spurious ambiguity report for template class
9094 * [271]11329 Compiler cannot find user defined implicit typecast
9095 * [272]11332 Spurious error with casts in ?: expression
9096 * [273]11431 static_cast behavior with subclasses when default
9097 constructor available
9098 * [274]11528 money_get facet does not accept "$.00" as valid
9099 * [275]11546 Type lookup problems in out-of-line definition of a
9100 class doubly nested from a template class
9101 * [276]11567 C++ code containing templated member function with same
9102 name as pure virtual member function results in linking failure
9103 * [277]11645 Failure to deal with using and private inheritance
9105 Java compiler and library
9107 * [278]5179 Qualified static field access doesn't initialize its
9109 * [279]8204 gcj -O2 to native reorders certain instructions
9111 * [280]10838 java.io.ObjectInputStream syntax error
9112 * [281]10886 The RMI registry that comes with GCJ does not work
9114 * [282]11349 JNDI URL context factories not located correctly
9116 x86-specific (Intel/AMD)
9118 * [283]4823 ICE on inline assembly code
9119 * [284]8878 miscompilation with -O and SSE
9120 * [285]9815 (c++ library) atomicity.h - fails to compile with -O3
9122 * [286]10402 (inline assembly) [x86] ICE in merge_assigned_reloads,
9124 * [287]10504 ICE with SSE2 code and -O3 -mcpu=pentium4 -msse2
9125 * [288]10673 ICE for x86-64 on freebsd libc vfprintf.c source
9126 * [289]11044 [x86] out of range loop instructions for FP code on K6
9127 * [290]11089 ICE: instantiate_virtual_regs_lossage while using SSE
9129 * [291]11420 [x86_64] gcc generates invalid asm code when "-O -fPIC"
9132 SPARC- or Solaris- specific
9134 * [292]9362 solaris 'as' dies when fed .s and "-gstabs"
9135 * [293]10142 [SPARC64] gcc produces wrong code when passing
9137 * [294]10663 New configure check aborts with Sun tools.
9138 * [295]10835 combinatorial explosion in scheduler on HyperSPARC
9139 * [296]10876 ICE in calculate_giv_inc when building KDE
9140 * [297]10955 wrong code at -O3 for structure argument in context of
9142 * [298]11018 -mcpu=ultrasparc busts tar-1.13.25
9143 * [299]11556 [sparc64] ICE in gen_reg_rtx() while compiling 2.6.x
9148 * [300]10907 gcc violates the ia64 ABI (GP must be preserved)
9149 * [301]11320 scheduler bug (in machine depended reorganization pass)
9150 * [302]11599 bug with conditional and __builtin_prefetch
9154 * [303]9745 [powerpc] gcc mis-compiles libmcrypt (alias problem
9156 * [304]10871 error in rs6000_stack_info save_size computation
9157 * [305]11440 gcc mis-compiles c++ code (libkhtml) with -O2, -fno-gcse
9162 * [306]7594 [m68k] ICE on legal code associated with simplify-rtx
9163 * [307]10557 [m68k] ICE in subreg_offset_representable_p
9164 * [308]11054 [m68k] ICE in reg_overlap_mentioned_p
9168 * [309]10834 [arm] GCC 3.3 still generates incorrect instructions for
9169 functions with __attribute__ ((interrupt ("IRQ")))
9170 * [310]10842 [arm] Clobbered link register is copied to pc under
9171 certain circumstances
9172 * [311]11052 [arm] noce_process_if_block() can lose REG_INC notes
9173 * [312]11183 [arm] ICE in change_address_1 (3.3) / subreg_hard_regno
9178 * [313]11084 ICE in propagate_one_insn, in flow.c
9182 * [314]10331 can't compile c++ part of gcc cross compiler for sh-elf
9183 * [315]10413 [SH] ICE in reload_cse_simplify_operands, in reload1.c
9184 * [316]11096 i686-linux to sh-linux cross compiler fails to compile
9187 GNU/Linux (or Hurd?) specific
9189 * [317]2873 Bogus fixinclude of stdio.h from glibc 2.2.3
9193 * [318]3163 configure bug: gcc/aclocal.m4 mmap test fails on UnixWare
9196 Cygwin (or mingw) specific
9198 * [319]5287 ICE with dllimport attribute
9199 * [320]10148 [MingW/CygWin] Compiler dumps core
9203 * [321]8787 GCC fails to emit .intel_syntax when invoked with
9204 -masm=intel on DJGPP
9206 Darwin (and MacOS X) specific
9208 * [322]10900 trampolines crash
9212 * [323]1607 (c++) Format attributes on methods undocumented
9213 * [324]4252 Invalid option `-fdump-translation-unit'
9214 * [325]4490 Clarify restrictions on -m96bit-long-double,
9215 -m128bit-long-double
9216 * [326]10355 document an issue with regparm attribute on some systems
9218 * [327]10726 (fortran) Documentation for function "IDate Intrinsic
9220 * [328]10805 document bug in old version of Sun assembler
9221 * [329]10815 warn against GNU binutils on AIX
9222 * [330]10877 document need for newer binutils on i?86-*-linux-gnu
9223 * [331]11280 Manual incorrect with respect to -freorder-blocks
9224 * [332]11466 Document -mlittle-endian and its restrictions for the
9227 Testsuite bugs (compiler itself is not affected)
9229 * [333]10737 newer bison causes g++.dg/parse/crash2.C to incorrectly
9231 * [334]10810 gcc-3.3 fails make check: buffer overrun in
9233 __________________________________________________________________
9239 This section lists the problem reports (PRs) from [335]GCC's bug
9240 tracking system that are known to be fixed in the 3.3.2 release. This
9241 list might not be complete (that is, it is possible that some PRs that
9242 have been fixed are not listed here).
9244 Bootstrap failures and problems
9246 * [336]8336 [SCO5] bootstrap config still tries to use COFF options
9247 * [337]9330 [alpha-osf] Bootstrap failure on Compaq Tru64 with
9248 --enable-threads=posix
9249 * [338]9631 [hppa64-linux] gcc-3.3 fails to bootstrap
9250 * [339]9877 fixincludes makes a bad sys/byteorder.h on svr5 (UnixWare
9252 * [340]11687 xstormy16-elf build fails in libf2c
9253 * [341]12263 [SGI IRIX] bootstrap fails during compile of
9254 libf2c/libI77/backspace.c
9255 * [342]12490 buffer overflow in scan-decls.c (during Solaris 9
9256 fix-header processing)
9258 Internal compiler errors (multi-platform)
9260 * [343]7277 Casting integers to vector types causes ICE
9261 * [344]7939 (c++) ICE on invalid function template specialization
9262 * [345]11063 (c++) ICE on parsing initialization list of const array
9264 * [346]11207 ICE with negative index in array element designator
9265 * [347]11522 (fortran) g77 dwarf-2 ICE in
9266 add_abstract_origin_attribute
9267 * [348]11595 (c++) ICE on duplicate label definition
9268 * [349]11646 (c++) ICE in commit_one_edge_insertion with
9269 -fnon-call-exceptions -fgcse -O
9270 * [350]11665 ICE in struct initializer when taking address
9271 * [351]11852 (c++) ICE with bad struct initializer.
9272 * [352]11878 (c++) ICE in cp_expr_size
9273 * [353]11883 ICE with any -O on mercury-generated C code
9274 * [354]11991 (c++) ICE in cxx_incomplete_type_diagnostic, in
9275 cp/typeck2.c when applying typeid operator to template template
9277 * [355]12146 ICE in lookup_template_function, in cp/pt.c
9278 * [356]12215 ICE in make_label_edge with -fnon-call-exceptions
9280 * [357]12369 (c++) ICE with templates and friends
9281 * [358]12446 ICE in emit_move_insn on complicated array reference
9282 * [359]12510 ICE in final_scan_insn
9283 * [360]12544 ICE with large parameters used in nested functions
9285 C and optimization bugs
9287 * [361]9862 spurious warnings with -W -finline-functions
9288 * [362]10962 lookup_field is a linear search on a linked list (can be
9289 slow if large struct)
9290 * [363]11370 -Wunreachable-code gives false complaints
9291 * [364]11637 invalid assembly with -fnon-call-exceptions
9292 * [365]11885 Problem with bitfields in packed structs
9293 * [366]12082 Inappropriate unreachable code warnings
9294 * [367]12180 Inline optimization fails for variadic function
9295 * [368]12340 loop unroller + gcse produces wrong code
9297 C++ compiler and library
9299 * [369]3907 nested template parameter collides with member name
9300 * [370]5293 confusing message when binding a temporary to a reference
9301 * [371]5296 [DR115] Pointers to functions and to template functions
9302 behave differently in deduction
9303 * [372]7939 ICE on function template specialization
9304 * [373]8656 Unable to assign function with __attribute__ and pointer
9305 return type to an appropriate variable
9306 * [374]10147 Confusing error message for invalid template function
9308 * [375]11400 std::search_n() makes assumptions about Size parameter
9309 * [376]11409 issues with using declarations, overloading, and
9311 * [377]11740 ctype<wchar_t>::do_is(mask, wchar_t) doesn't handle
9312 multiple bits in mask
9313 * [378]11786 operator() call on variable in other namespace not
9315 * [379]11867 static_cast ignores ambiguity
9316 * [380]11928 bug with conversion operators that are typedefs
9317 * [381]12114 Uninitialized memory accessed in dtor
9318 * [382]12163 static_cast + explicit constructor regression
9319 * [383]12181 Wrong code with comma operator and c++
9320 * [384]12236 regparm and fastcall messes up parameters
9321 * [385]12266 incorrect instantiation of unneeded template during
9323 * [386]12296 istream::peek() doesn't set eofbit
9324 * [387]12298 [sjlj exceptions] Stack unwind destroys
9325 not-yet-constructed object
9326 * [388]12369 ICE with templates and friends
9327 * [389]12337 apparently infinite loop in g++
9328 * [390]12344 stdcall attribute ignored if function returns a pointer
9329 * [391]12451 missing(late) class forward declaration in cxxabi.h
9330 * [392]12486 g++ accepts invalid use of a qualified name
9332 x86 specific (Intel/AMD)
9334 * [393]8869 [x86 MMX] ICE with const variable optimization and MMX
9336 * [394]9786 ICE in fixup_abnormal_edges with -fnon-call-exceptions
9338 * [395]11689 g++3.3 emits un-assembleable code for k6 architecture
9339 * [396]12116 [k6] Invalid assembly output values with X-MAME code
9340 * [397]12070 ICE converting between double and long double with
9345 * [398]11184 [ia64 hpux] ICE on __builtin_apply building libobjc
9346 * [399]11535 __builtin_return_address may not work on ia64
9347 * [400]11693 [ia64] ICE in gen_nop_type
9348 * [401]12224 [ia64] Thread-local storage doesn't work
9352 * [402]11087 [powerpc64-linux] GCC miscompiles raid1.c from linux
9354 * [403]11319 loop miscompiled on ppc32
9355 * [404]11949 ICE Compiler segfault with ffmpeg -maltivec code
9359 * [405]11662 wrong code for expr. with cast to long long and
9361 * [406]11965 invalid assembler code for a shift < 32 operation
9362 * [407]12301 (c++) stack corruption when a returned expression throws
9367 * [408]11717 [alpha-linux] unrecognizable insn compiling for.c of
9372 * [409]11313 problem with #pragma weak and static inline functions
9373 * [410]11712 __STDC_EXT__ not defined for C++ by default anymore?
9377 * [411]12166 Profiled programs crash if PROFDIR is set
9379 Solaris-x86 specific
9381 * [412]12101 i386 Solaris no longer works with GNU as?
9383 Miscellaneous embedded target-specific bugs
9385 * [413]10988 [m32r-elf] wrong blockmove code with -O3
9386 * [414]11805 [h8300-unknown-coff] [H8300] ICE for simple code with
9388 * [415]11902 [sh4] spec file improperly inserts rpath even when none
9390 * [416]11903 [sh4] -pthread fails to link due to error in spec file
9392 __________________________________________________________________
9398 In addition to the bug fixes documented below, this release contains
9399 few minor features such as:
9400 * Support for --with-sysroot
9401 * Support for automatic detection of executable stacks
9402 * Support for SSE3 instructions
9403 * Support for thread local storage debugging under GDB on S390
9407 This section lists the problem reports (PRs) from [417]GCC's bug
9408 tracking system that are known to be fixed in the 3.3.3 release. This
9409 list might not be complete (that is, it is possible that some PRs that
9410 have been fixed are not listed here).
9412 Bootstrap failures and issues
9414 * [418]11890 Building cross gcc-3.3.1 for sparc-sun-solaris2.6 fails
9415 * [419]12399 boehm-gc fails (when building a cross compiler): libtool
9416 unable to infer tagged configuration
9417 * [420]13068 mklibgcc.in doesn't handle multi-level multilib
9418 subdirectories properly
9420 Internal compiler errors (multi-platform)
9422 * [421]10060 ICE (stack overflow) on huge file (300k lines) due to
9423 recursive behaviour of copy_rtx_if_shared, in emit_rtl.c
9424 * [422]10555 (c++) ICE on undefined template argument
9425 * [423]10706 (c++) ICE in mangle_class_name_for_template
9426 * [424]11496 (fortran) error in flow_loops_find when -funroll-loops
9428 * [425]11741 ICE in pre_insert_copy_insn, in gcse.c
9429 * [426]12440 GCC crashes during compilation of quicktime4linux 2.0.0
9430 * [427]12632 (fortran) -fbounds-check ICE
9431 * [428]12712 (c++) ICE on short legit C++ code fragment with gcc
9433 * [429]12726 (c++) ICE (segfault) on trivial code
9434 * [430]12890 (c++) ICE on compilation of class with throwing method
9435 * [431]12900 (c++) ICE in rtl_verify_flow_info_1
9436 * [432]13060 (fortran) ICE in fixup_var_refs_1, in function.c on
9437 correct code with -O2 -fno-force-mem
9438 * [433]13289 (c++) ICE in regenerate_decl_from_template on recursive
9440 * [434]13318 ICE: floating point exception in the loop optimizer
9441 * [435]13392 (c++) ICE in convert_from_eh_region_ranges_1, in
9443 * [436]13574 (c++) invalid array default initializer in class lets
9444 gcc consume all memory and die
9445 * [437]13475 ICE on SIMD variables with partial value initialization
9446 * [438]13797 (c++) ICE on invalid template parameter
9447 * [439]13824 (java) gcj SEGV with simple .java program
9449 C and optimization bugs
9451 * [440]8776 loop invariants are not removed (most likely)
9452 * [441]10339 [sparc,ppc,ppc64] Invalid optimization: replacing
9454 * [442]11350 undefined labels with -Os -fPIC
9455 * [443]12826 Optimizer removes reference through volatile pointer
9456 * [444]12500 stabs debug info: void no longer a predefined / builtin
9458 * [445]12941 builtin-bitops-1.c miscompilation (latent bug)
9459 * [446]12953 tree inliner bug (in inline_forbidden_p) and fix
9460 * [447]13041 linux-2.6/sound/core/oss/rate.c miscompiled
9461 * [448]13507 spurious printf format warning
9462 * [449]13382 Type information for const pointer disappears during
9464 * [450]13394 noreturn attribute ignored on recursive invokation
9465 * [451]13400 Compiled code crashes storing to read-only location
9466 * [452]13521 Endless loop in calculate_global_regs_live
9468 C++ compiler and library
9470 Some of the bug fixes in this list were made to implement decisions
9471 that the ISO C++ standards committee has made concerning several defect
9472 reports (DRs). Links in the list below point to detailed discussion of
9473 the relevant defect report.
9474 * [453]2094 unimplemented: use of `ptrmem_cst' in template type
9476 * [454]2294 using declaration confusion
9477 * [455]5050 template instantiation depth exceeds limit: recursion
9479 * [456]9371 Bad exception handling in
9480 i/ostream::operator>>/<<(streambuf*)
9481 * [457]9546 bad exception handling in ostream members
9482 * [458]10081 basic_ios::_M_cache_locale leaves NULL members in the
9483 face of unknown locales
9484 * [459]10093 [460][DR 61] Setting failbit in exceptions doesn't work
9485 * [461]10095 istream::operator>>(int&) sets ios::badbit when
9486 ios::failbit is set.
9487 * [462]11554 Warning about reordering of initializers doesn't mention
9488 location of constructor
9489 * [463]12297 istream::sentry::sentry() handles eof() incorrectly.
9490 * [464]12352 Exception safety problems in src/localename.cc
9491 * [465]12438 Memory leak in locale::combine()
9492 * [466]12540 Memory leak in locale::locale(const char*)
9493 * [467]12594 DRs [468]60 [TC] and [469]63 [TC] not implemented
9494 * [470]12657 Resolution of [471]DR 292 (WP) still unimplemented
9495 * [472]12696 memory eating infinite loop in diagnostics (error
9497 * [473]12815 Code compiled with optimization behaves unexpectedly
9498 * [474]12862 Conflicts between typedefs/enums and namespace member
9500 * [475]12926 Wrong value after assignment in initialize list using
9502 * [476]12967 Resolution of [477]DR 300 [WP] still unimplemented
9503 * [478]12971 Resolution of [479]DR 328 [WP] still unimplemented
9504 * [480]13007 basic_streambuf::pubimbue, imbue wrong
9505 * [481]13009 Implicitly-defined assignment operator writes to wrong
9507 * [482]13057 regparm attribute not applied to destructor
9508 * [483]13070 -Wformat option ignored in g++
9509 * [484]13081 forward template declarations in <complex> let inlining
9511 * [485]13239 Assertion does not seem to work correctly anymore
9512 * [486]13262 "xxx is private within this context" when initializing a
9513 self-contained template class
9514 * [487]13290 simple typo in concept checking for std::generate_n
9515 * [488]13323 Template code does not compile in presence of typedef
9516 * [489]13369 __verify_grouping (and __add_grouping?) not correct
9517 * [490]13371 infinite loop with packed struct and inlining
9518 * [491]13445 Template argument replacement "dereferences" a typedef
9519 * [492]13461 Fails to access protected-ctor from public constant
9520 * [493]13462 Non-standard-conforming type set::pointer
9521 * [494]13478 gcc uses wrong constructor to initialize a const
9523 * [495]13544 "conflicting types" for enums in different scopes
9524 * [496]13650 string::compare should not (always) use
9525 traits_type::length()
9526 * [497]13683 bogus warning about passing non-PODs through ellipsis
9527 * [498]13688 Derived class is denied access to protected base class
9529 * [499]13774 Member variable cleared in virtual multiple inheritance
9531 * [500]13884 Protect sstream.tcc from extern template use
9533 Java compiler and library
9535 * [501]10746 [win32] garbage collection crash in GCJ
9537 Objective-C compiler and library
9539 * [502]11433 Crash due to dereferencing null pointer when querying
9542 Fortran compiler and library
9544 * [503]12633 logical expression gives incorrect result with
9545 -fugly-logint option
9546 * [504]13037 [gcse-lm] g77 generates incorrect code
9547 * [505]13213 Hex constant problem when compiling with -fugly-logint
9550 x86-specific (Intel/AMD)
9552 * [506]4490 ICE with -m128bit-long-double
9553 * [507]12292 [x86_64] ICE: RTL check: expected code `const_int', have
9554 `reg' in make_field_assignment, in combine.c
9555 * [508]12441 ICE: can't find a register to spill
9556 * [509]12943 array static-init failure under -fpic, -fPIC
9557 * [510]13608 Incorrect code with -O3 -ffast-math
9561 * [511]11598 testcase gcc.dg/20020118-1.c fails runtime check of
9562 __attribute__((aligned(16)))
9563 * [512]11793 ICE in extract_insn, in recog.c (const_vector's)
9564 * [513]12467 vmsumubm emitted when vmsummbm appropriate (typo in
9566 * [514]12537 g++ generates writeable text sections
9570 * [515]12496 wrong result for __atomic_add(&value, -1) when using -O0
9572 * [516]12865 mprotect call to make trampoline executable may fail
9573 * [517]13354 ICE in sparc_emit_set_const32
9577 * [518]10467 [arm] ICE in pre_insert_copy_insn,
9581 * [519]11226 ICE passing struct arg with two floats
9582 * [520]11227 ICE for _Complex float, _Complex long double args
9583 * [521]12644 GCC 3.3.2 fails to compile glibc on ia64
9584 * [522]13149 build gcc-3.3.2 1305 error:unrecognizable insn
9585 * Various fixes for libunwind
9589 * [523]12654 Incorrect comparison code generated for Alpha
9590 * [524]12965 SEGV+ICE in cc1plus on alpha-linux with -O2
9591 * [525]13031 ICE (unrecognizable insn) when building gnome-libs-1.4.2
9595 * [526]11634 [hppa] ICE in verify_local_live_at_start, in flow.c
9596 * [527]12158 [hppa] compilation does not terminate at -O1
9600 * [528]11992 Wrong built-in code for memcmp with length 1<<24: only
9601 (1<<24)-1 possible for CLCL-Instruction
9605 * [529]9365 segfault in gen_far_branch (config/sh/sh.c)
9606 * [530]10392 optimizer generates faulty array indexing
9607 * [531]11322 SH profiler outputs multiple definitions of symbol
9608 * [532]13069 gcc/config/sh/rtems.h broken
9609 * [533]13302 Putting a va_list in a struct causes seg fault
9610 * [534]13585 Incorrect optimization of call to sfunc
9611 * Fix inappropriately exported libgcc functions from the shared
9614 Other embedded target specific
9616 * [535]8916 [mcore] unsigned char assign gets hosed.
9617 * [536]11576 [h8300] ICE in change_address_1, in emit-rtl.c
9618 * [537]13122 [h8300] local variable gets corrupted by function call
9619 when -fomit-frame-pointer is given
9620 * [538]13256 [cris] strict_low_part mistreated in delay slots
9621 * [539]13373 [mcore] optimization with -frerun-cse-after-loop
9622 -fexpensive-optimizations produces wrong code on mcore
9626 * [540]12561 gcc/config/t-gnu needs updating to work with
9631 * [541]6243 testsuite fails almost all tests due to no libintl in
9632 LD_LIBRARY_PATH during test.
9633 * [542]11397 weak aliases broken on Tru64 UNIX
9637 * [543]12505 build failure due to defines of uchar in cpphash.h and
9639 * [544]13150 WEAK symbols not exported by collect2
9643 * [545]12666 fixincludes problem on IRIX 6.5.19m
9647 * [546]12969 Including sys/byteorder.h breaks configure checks
9649 Testsuite problems (compiler is not affected)
9651 * [547]10819 testsuite creates CR+LF on compiler version lines in
9653 * [548]11612 abi_check not finding correct libgcc_s.so.1
9657 * [549]13211 using -###, incorrect warnings about unused linker file
9659 __________________________________________________________________
9663 This is the [550]list of problem reports (PRs) from GCC's bug tracking
9664 system that are known to be fixed in the 3.3.4 release. This list might
9665 not be complete (that is, it is possible that some PRs that have been
9666 fixed are not listed here).
9667 __________________________________________________________________
9671 This is the [551]list of problem reports (PRs) from GCC's bug tracking
9672 system that are known to be fixed in the 3.3.5 release. This list might
9673 not be complete (that is, it is possible that some PRs that have been
9674 fixed are not listed here).
9675 __________________________________________________________________
9679 This is the [552]list of problem reports (PRs) from GCC's bug tracking
9680 system that are known to be fixed in the 3.3.6 release. This list might
9681 not be complete (that is, it is possible that some PRs that have been
9682 fixed are not listed here).
9685 For questions related to the use of GCC, please consult these web
9686 pages and the [553]GCC manuals. If that fails, the
9687 [554]gcc-help@gcc.gnu.org mailing list might help. Comments on these
9688 web pages and the development of GCC are welcome on our developer
9689 list at [555]gcc@gcc.gnu.org. All of [556]our lists have public
9692 Copyright (C) [557]Free Software Foundation, Inc. Verbatim copying and
9693 distribution of this entire article is permitted in any medium,
9694 provided this notice is preserved.
9696 These pages are [558]maintained by the GCC team. Last modified
9701 1. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.6
9702 2. http://gcc.gnu.org/gcc-3.1/changes.html#obsolete_systems
9703 3. http://gcc.gnu.org/gcc-3.3/changes.html#obsolete_systems
9704 4. http://gcc.gnu.org/gcc-3.3/changes.html#nonnull_attribute
9705 5. http://gcc.gnu.org/news/dfa.html
9706 6. http://gcc.gnu.org/gcc-3.3/c99status.html
9707 7. http://gcc.gnu.org/onlinedocs/gcc-3.3.6/g77/News.html
9708 8. http://gcc.gnu.org/PR10140
9709 9. http://gcc.gnu.org/PR10198
9710 10. http://gcc.gnu.org/PR10338
9711 11. http://gcc.gnu.org/PR3581
9712 12. http://gcc.gnu.org/PR4382
9713 13. http://gcc.gnu.org/PR5533
9714 14. http://gcc.gnu.org/PR6387
9715 15. http://gcc.gnu.org/PR6412
9716 16. http://gcc.gnu.org/PR6620
9717 17. http://gcc.gnu.org/PR6663
9718 18. http://gcc.gnu.org/PR7068
9719 19. http://gcc.gnu.org/PR7083
9720 20. http://gcc.gnu.org/PR7647
9721 21. http://gcc.gnu.org/PR7675
9722 22. http://gcc.gnu.org/PR7718
9723 23. http://gcc.gnu.org/PR8116
9724 24. http://gcc.gnu.org/PR8358
9725 25. http://gcc.gnu.org/PR8511
9726 26. http://gcc.gnu.org/PR8564
9727 27. http://gcc.gnu.org/PR8660
9728 28. http://gcc.gnu.org/PR8766
9729 29. http://gcc.gnu.org/PR8803
9730 30. http://gcc.gnu.org/PR8846
9731 31. http://gcc.gnu.org/PR8906
9732 32. http://gcc.gnu.org/PR9216
9733 33. http://gcc.gnu.org/PR9261
9734 34. http://gcc.gnu.org/PR9263
9735 35. http://gcc.gnu.org/PR9429
9736 36. http://gcc.gnu.org/PR9516
9737 37. http://gcc.gnu.org/PR9600
9738 38. http://gcc.gnu.org/PR9629
9739 39. http://gcc.gnu.org/PR9672
9740 40. http://gcc.gnu.org/PR9749
9741 41. http://gcc.gnu.org/PR9794
9742 42. http://gcc.gnu.org/PR9829
9743 43. http://gcc.gnu.org/PR9916
9744 44. http://gcc.gnu.org/PR9936
9745 45. http://gcc.gnu.org/PR10262
9746 46. http://gcc.gnu.org/PR10278
9747 47. http://gcc.gnu.org/PR10446
9748 48. http://gcc.gnu.org/PR10451
9749 49. http://gcc.gnu.org/PR10506
9750 50. http://gcc.gnu.org/PR10549
9751 51. http://gcc.gnu.org/PR2001
9752 52. http://gcc.gnu.org/PR2391
9753 53. http://gcc.gnu.org/PR2960
9754 54. http://gcc.gnu.org/PR4046
9755 55. http://gcc.gnu.org/PR6405
9756 56. http://gcc.gnu.org/PR6798
9757 57. http://gcc.gnu.org/PR6871
9758 58. http://gcc.gnu.org/PR6909
9759 59. http://gcc.gnu.org/PR7189
9760 60. http://gcc.gnu.org/PR7642
9761 61. http://gcc.gnu.org/PR8634
9762 62. http://gcc.gnu.org/PR8750
9763 63. http://gcc.gnu.org/PR2161
9764 64. http://gcc.gnu.org/PR4319
9765 65. http://gcc.gnu.org/PR8602
9766 66. http://gcc.gnu.org/PR9177
9767 67. http://gcc.gnu.org/PR9853
9768 68. http://gcc.gnu.org/PR45
9769 69. http://gcc.gnu.org/PR3784
9770 70. http://gcc.gnu.org/PR764
9771 71. http://gcc.gnu.org/PR5116
9772 72. http://gcc.gnu.org/PR2862
9773 73. http://gcc.gnu.org/PR3663
9774 74. http://gcc.gnu.org/PR3797
9775 75. http://gcc.gnu.org/PR3948
9776 76. http://gcc.gnu.org/PR4137
9777 77. http://gcc.gnu.org/PR4361
9778 78. http://gcc.gnu.org/PR4802
9779 79. http://gcc.gnu.org/PR5837
9780 80. http://gcc.gnu.org/PR4803
9781 81. http://gcc.gnu.org/PR5094
9782 82. http://gcc.gnu.org/PR5730
9783 83. http://gcc.gnu.org/PR6713
9784 84. http://gcc.gnu.org/PR7015
9785 85. http://gcc.gnu.org/PR7086
9786 86. http://gcc.gnu.org/PR7099
9787 87. http://gcc.gnu.org/PR7247
9788 88. http://gcc.gnu.org/PR7441
9789 89. http://gcc.gnu.org/PR7768
9790 90. http://gcc.gnu.org/PR7804
9791 91. http://gcc.gnu.org/PR8099
9792 92. http://gcc.gnu.org/PR8117
9793 93. http://gcc.gnu.org/PR8205
9794 94. http://gcc.gnu.org/PR8645
9795 95. http://gcc.gnu.org/PR8724
9796 96. http://gcc.gnu.org/PR8805
9797 97. http://gcc.gnu.org/PR8691
9798 98. http://gcc.gnu.org/PR8700
9799 99. http://gcc.gnu.org/PR8724
9800 100. http://gcc.gnu.org/PR8949
9801 101. http://gcc.gnu.org/PR9016
9802 102. http://gcc.gnu.org/PR9053
9803 103. http://gcc.gnu.org/PR9152
9804 104. http://gcc.gnu.org/PR9182
9805 105. http://gcc.gnu.org/PR9297
9806 106. http://gcc.gnu.org/PR9318
9807 107. http://gcc.gnu.org/PR9320
9808 108. http://gcc.gnu.org/PR9400
9809 109. http://gcc.gnu.org/PR9424
9810 110. http://gcc.gnu.org/PR9425
9811 111. http://gcc.gnu.org/PR9439
9812 112. http://gcc.gnu.org/PR9474
9813 113. http://gcc.gnu.org/PR9548
9814 114. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#231
9815 115. http://gcc.gnu.org/PR9555
9816 116. http://gcc.gnu.org/PR9561
9817 117. http://gcc.gnu.org/PR9563
9818 118. http://gcc.gnu.org/PR9582
9819 119. http://gcc.gnu.org/PR9622
9820 120. http://gcc.gnu.org/PR9683
9821 121. http://gcc.gnu.org/PR9791
9822 122. http://gcc.gnu.org/PR9817
9823 123. http://gcc.gnu.org/PR9825
9824 124. http://gcc.gnu.org/PR9826
9825 125. http://gcc.gnu.org/PR9924
9826 126. http://gcc.gnu.org/PR9946
9827 127. http://gcc.gnu.org/PR9964
9828 128. http://gcc.gnu.org/PR9988
9829 129. http://gcc.gnu.org/PR10033
9830 130. http://gcc.gnu.org/PR10097
9831 131. http://gcc.gnu.org/PR10132
9832 132. http://gcc.gnu.org/PR10180
9833 133. http://gcc.gnu.org/PR10199
9834 134. http://gcc.gnu.org/PR10300
9835 135. http://gcc.gnu.org/PR10427
9836 136. http://gcc.gnu.org/PR10503
9837 137. http://gcc.gnu.org/PR5956
9838 138. http://gcc.gnu.org/PR1832
9839 139. http://gcc.gnu.org/PR3924
9840 140. http://gcc.gnu.org/PR5634
9841 141. http://gcc.gnu.org/PR6367
9842 142. http://gcc.gnu.org/PR6491
9843 143. http://gcc.gnu.org/PR6742
9844 144. http://gcc.gnu.org/PR7113
9845 145. http://gcc.gnu.org/PR7236
9846 146. http://gcc.gnu.org/PR7278
9847 147. http://gcc.gnu.org/PR7384
9848 148. http://gcc.gnu.org/PR7388
9849 149. http://gcc.gnu.org/PR8587
9850 150. http://gcc.gnu.org/PR9038
9851 151. http://gcc.gnu.org/PR10197
9852 152. http://gcc.gnu.org/PR6005
9853 153. http://gcc.gnu.org/PR6389
9854 154. http://gcc.gnu.org/PR6576
9855 155. http://gcc.gnu.org/PR6652
9856 156. http://gcc.gnu.org/PR7060
9857 157. http://gcc.gnu.org/PR7073
9858 158. http://gcc.gnu.org/PR7180
9859 159. http://gcc.gnu.org/PR7416
9860 160. http://gcc.gnu.org/PR7570
9861 161. http://gcc.gnu.org/PR7578
9862 162. http://gcc.gnu.org/PR7611
9863 163. http://gcc.gnu.org/PR7709
9864 164. http://gcc.gnu.org/PR7766
9865 165. http://gcc.gnu.org/PR7785
9866 166. http://gcc.gnu.org/PR7786
9867 167. http://gcc.gnu.org/PR8142
9868 168. http://gcc.gnu.org/PR8234
9869 169. http://gcc.gnu.org/PR8415
9870 170. http://gcc.gnu.org/PR8481
9871 171. http://gcc.gnu.org/PR8593
9872 172. http://gcc.gnu.org/PR8759
9873 173. http://gcc.gnu.org/PR8997
9874 174. http://gcc.gnu.org/PR9253
9875 175. http://gcc.gnu.org/PR9254
9876 176. http://gcc.gnu.org/PR9271
9877 177. http://gcc.gnu.org/PR6767
9878 178. http://gcc.gnu.org/PR9911
9879 179. http://gcc.gnu.org/PR10020
9880 180. http://gcc.gnu.org/PR10546
9881 181. http://gcc.gnu.org/PR7029
9882 182. http://gcc.gnu.org/PR2903
9883 183. http://gcc.gnu.org/PR7873
9884 184. http://gcc.gnu.org/PR7680
9885 185. http://gcc.gnu.org/PR8705
9886 186. http://gcc.gnu.org/PR9986
9887 187. http://gcc.gnu.org/PR10056
9888 188. http://gcc.gnu.org/PR6744
9889 189. http://gcc.gnu.org/PR7361
9890 190. http://gcc.gnu.org/PR9496
9891 191. http://gcc.gnu.org/PR7067
9892 192. http://gcc.gnu.org/PR8480
9893 193. http://gcc.gnu.org/PR8784
9894 194. http://gcc.gnu.org/PR10315
9895 195. http://gcc.gnu.org/PR10267
9896 196. http://gcc.gnu.org/PR7916
9897 197. http://gcc.gnu.org/PR7926
9898 198. http://gcc.gnu.org/PR8555
9899 199. http://gcc.gnu.org/PR8994
9900 200. http://gcc.gnu.org/PR9426
9901 201. http://gcc.gnu.org/PR9806
9902 202. http://gcc.gnu.org/PR10077
9903 203. http://gcc.gnu.org/PR10233
9904 204. http://gcc.gnu.org/PR10286
9905 205. http://gcc.gnu.org/PR10308
9906 206. http://gcc.gnu.org/PR11272
9907 207. http://gcc.gnu.org/PR5754
9908 208. http://gcc.gnu.org/PR6597
9909 209. http://gcc.gnu.org/PR6949
9910 210. http://gcc.gnu.org/PR7053
9911 211. http://gcc.gnu.org/PR8164
9912 212. http://gcc.gnu.org/PR8384
9913 213. http://gcc.gnu.org/PR9559
9914 214. http://gcc.gnu.org/PR9649
9915 215. http://gcc.gnu.org/PR9864
9916 216. http://gcc.gnu.org/PR10432
9917 217. http://gcc.gnu.org/PR10475
9918 218. http://gcc.gnu.org/PR10635
9919 219. http://gcc.gnu.org/PR10661
9920 220. http://gcc.gnu.org/PR10700
9921 221. http://gcc.gnu.org/PR10712
9922 222. http://gcc.gnu.org/PR10796
9923 223. http://gcc.gnu.org/PR10890
9924 224. http://gcc.gnu.org/PR10939
9925 225. http://gcc.gnu.org/PR10956
9926 226. http://gcc.gnu.org/PR11041
9927 227. http://gcc.gnu.org/PR11059
9928 228. http://gcc.gnu.org/PR11083
9929 229. http://gcc.gnu.org/PR11105
9930 230. http://gcc.gnu.org/PR11149
9931 231. http://gcc.gnu.org/PR11228
9932 232. http://gcc.gnu.org/PR11282
9933 233. http://gcc.gnu.org/PR11301
9934 234. http://gcc.gnu.org/PR11308
9935 235. http://gcc.gnu.org/PR11473
9936 236. http://gcc.gnu.org/PR11503
9937 237. http://gcc.gnu.org/PR11513
9938 238. http://gcc.gnu.org/PR11198
9939 239. http://gcc.gnu.org/PR11304
9940 240. http://gcc.gnu.org/PR11381
9941 241. http://gcc.gnu.org/PR11536
9942 242. http://gcc.gnu.org/PR11557
9943 243. http://gcc.gnu.org/PR5897
9944 244. http://gcc.gnu.org/PR11279
9945 245. http://gcc.gnu.org/PR11022
9946 246. http://gcc.gnu.org/PR2330
9947 247. http://gcc.gnu.org/PR5388
9948 248. http://gcc.gnu.org/PR5390
9949 249. http://gcc.gnu.org/PR7877
9950 250. http://gcc.gnu.org/PR9393
9951 251. http://gcc.gnu.org/PR10032
9952 252. http://gcc.gnu.org/PR10468
9953 253. http://gcc.gnu.org/PR10527
9954 254. http://gcc.gnu.org/PR10679
9955 255. http://gcc.gnu.org/PR10682
9956 256. http://gcc.gnu.org/PR10689
9957 257. http://gcc.gnu.org/PR10845
9958 258. http://gcc.gnu.org/PR10849
9959 259. http://gcc.gnu.org/PR10888
9960 260. http://gcc.gnu.org/PR10929
9961 261. http://gcc.gnu.org/PR10931
9962 262. http://gcc.gnu.org/PR10940
9963 263. http://gcc.gnu.org/PR10968
9964 264. http://gcc.gnu.org/PR10990
9965 265. http://gcc.gnu.org/PR11039
9966 266. http://gcc.gnu.org/PR11062
9967 267. http://gcc.gnu.org/PR11095
9968 268. http://gcc.gnu.org/PR11098
9969 269. http://gcc.gnu.org/PR11137
9970 270. http://gcc.gnu.org/PR11154
9971 271. http://gcc.gnu.org/PR11329
9972 272. http://gcc.gnu.org/PR11332
9973 273. http://gcc.gnu.org/PR11431
9974 274. http://gcc.gnu.org/PR11528
9975 275. http://gcc.gnu.org/PR11546
9976 276. http://gcc.gnu.org/PR11567
9977 277. http://gcc.gnu.org/PR11645
9978 278. http://gcc.gnu.org/PR5179
9979 279. http://gcc.gnu.org/PR8204
9980 280. http://gcc.gnu.org/PR10838
9981 281. http://gcc.gnu.org/PR10886
9982 282. http://gcc.gnu.org/PR11349
9983 283. http://gcc.gnu.org/PR4823
9984 284. http://gcc.gnu.org/PR8878
9985 285. http://gcc.gnu.org/PR9815
9986 286. http://gcc.gnu.org/PR10402
9987 287. http://gcc.gnu.org/PR10504
9988 288. http://gcc.gnu.org/PR10673
9989 289. http://gcc.gnu.org/PR11044
9990 290. http://gcc.gnu.org/PR11089
9991 291. http://gcc.gnu.org/PR11420
9992 292. http://gcc.gnu.org/PR9362
9993 293. http://gcc.gnu.org/PR10142
9994 294. http://gcc.gnu.org/PR10663
9995 295. http://gcc.gnu.org/PR10835
9996 296. http://gcc.gnu.org/PR10876
9997 297. http://gcc.gnu.org/PR10955
9998 298. http://gcc.gnu.org/PR11018
9999 299. http://gcc.gnu.org/PR11556
10000 300. http://gcc.gnu.org/PR10907
10001 301. http://gcc.gnu.org/PR11320
10002 302. http://gcc.gnu.org/PR11599
10003 303. http://gcc.gnu.org/PR9745
10004 304. http://gcc.gnu.org/PR10871
10005 305. http://gcc.gnu.org/PR11440
10006 306. http://gcc.gnu.org/PR7594
10007 307. http://gcc.gnu.org/PR10557
10008 308. http://gcc.gnu.org/PR11054
10009 309. http://gcc.gnu.org/PR10834
10010 310. http://gcc.gnu.org/PR10842
10011 311. http://gcc.gnu.org/PR11052
10012 312. http://gcc.gnu.org/PR11183
10013 313. http://gcc.gnu.org/PR11084
10014 314. http://gcc.gnu.org/PR10331
10015 315. http://gcc.gnu.org/PR10413
10016 316. http://gcc.gnu.org/PR11096
10017 317. http://gcc.gnu.org/PR2873
10018 318. http://gcc.gnu.org/PR3163
10019 319. http://gcc.gnu.org/PR5287
10020 320. http://gcc.gnu.org/PR10148
10021 321. http://gcc.gnu.org/PR8787
10022 322. http://gcc.gnu.org/PR10900
10023 323. http://gcc.gnu.org/PR1607
10024 324. http://gcc.gnu.org/PR4252
10025 325. http://gcc.gnu.org/PR4490
10026 326. http://gcc.gnu.org/PR10355
10027 327. http://gcc.gnu.org/PR10726
10028 328. http://gcc.gnu.org/PR10805
10029 329. http://gcc.gnu.org/PR10815
10030 330. http://gcc.gnu.org/PR10877
10031 331. http://gcc.gnu.org/PR11280
10032 332. http://gcc.gnu.org/PR11466
10033 333. http://gcc.gnu.org/PR10737
10034 334. http://gcc.gnu.org/PR10810
10035 335. http://gcc.gnu.org/bugzilla/
10036 336. http://gcc.gnu.org/PR8336
10037 337. http://gcc.gnu.org/PR9330
10038 338. http://gcc.gnu.org/PR9631
10039 339. http://gcc.gnu.org/PR9877
10040 340. http://gcc.gnu.org/PR11687
10041 341. http://gcc.gnu.org/PR12263
10042 342. http://gcc.gnu.org/PR12490
10043 343. http://gcc.gnu.org/PR7277
10044 344. http://gcc.gnu.org/PR7939
10045 345. http://gcc.gnu.org/PR11063
10046 346. http://gcc.gnu.org/PR11207
10047 347. http://gcc.gnu.org/PR11522
10048 348. http://gcc.gnu.org/PR11595
10049 349. http://gcc.gnu.org/PR11646
10050 350. http://gcc.gnu.org/PR11665
10051 351. http://gcc.gnu.org/PR11852
10052 352. http://gcc.gnu.org/PR11878
10053 353. http://gcc.gnu.org/PR11883
10054 354. http://gcc.gnu.org/PR11991
10055 355. http://gcc.gnu.org/PR12146
10056 356. http://gcc.gnu.org/PR12215
10057 357. http://gcc.gnu.org/PR12369
10058 358. http://gcc.gnu.org/PR12446
10059 359. http://gcc.gnu.org/PR12510
10060 360. http://gcc.gnu.org/PR12544
10061 361. http://gcc.gnu.org/PR9862
10062 362. http://gcc.gnu.org/PR10962
10063 363. http://gcc.gnu.org/PR11370
10064 364. http://gcc.gnu.org/PR11637
10065 365. http://gcc.gnu.org/PR11885
10066 366. http://gcc.gnu.org/PR12082
10067 367. http://gcc.gnu.org/PR12180
10068 368. http://gcc.gnu.org/PR12340
10069 369. http://gcc.gnu.org/PR3907
10070 370. http://gcc.gnu.org/PR5293
10071 371. http://gcc.gnu.org/PR5296
10072 372. http://gcc.gnu.org/PR7939
10073 373. http://gcc.gnu.org/PR8656
10074 374. http://gcc.gnu.org/PR10147
10075 375. http://gcc.gnu.org/PR11400
10076 376. http://gcc.gnu.org/PR11409
10077 377. http://gcc.gnu.org/PR11740
10078 378. http://gcc.gnu.org/PR11786
10079 379. http://gcc.gnu.org/PR11867
10080 380. http://gcc.gnu.org/PR11928
10081 381. http://gcc.gnu.org/PR12114
10082 382. http://gcc.gnu.org/PR12163
10083 383. http://gcc.gnu.org/PR12181
10084 384. http://gcc.gnu.org/PR12236
10085 385. http://gcc.gnu.org/PR12266
10086 386. http://gcc.gnu.org/PR12296
10087 387. http://gcc.gnu.org/PR12298
10088 388. http://gcc.gnu.org/PR12369
10089 389. http://gcc.gnu.org/PR12337
10090 390. http://gcc.gnu.org/PR12344
10091 391. http://gcc.gnu.org/PR12451
10092 392. http://gcc.gnu.org/PR12486
10093 393. http://gcc.gnu.org/PR8869
10094 394. http://gcc.gnu.org/PR9786
10095 395. http://gcc.gnu.org/PR11689
10096 396. http://gcc.gnu.org/PR12116
10097 397. http://gcc.gnu.org/PR12070
10098 398. http://gcc.gnu.org/PR11184
10099 399. http://gcc.gnu.org/PR11535
10100 400. http://gcc.gnu.org/PR11693
10101 401. http://gcc.gnu.org/PR12224
10102 402. http://gcc.gnu.org/PR11087
10103 403. http://gcc.gnu.org/PR11319
10104 404. http://gcc.gnu.org/PR11949
10105 405. http://gcc.gnu.org/PR11662
10106 406. http://gcc.gnu.org/PR11965
10107 407. http://gcc.gnu.org/PR12301
10108 408. http://gcc.gnu.org/PR11717
10109 409. http://gcc.gnu.org/PR11313
10110 410. http://gcc.gnu.org/PR11712
10111 411. http://gcc.gnu.org/PR12166
10112 412. http://gcc.gnu.org/PR12101
10113 413. http://gcc.gnu.org/PR10988
10114 414. http://gcc.gnu.org/PR11805
10115 415. http://gcc.gnu.org/PR11902
10116 416. http://gcc.gnu.org/PR11903
10117 417. http://gcc.gnu.org/bugzilla/
10118 418. http://gcc.gnu.org/PR11890
10119 419. http://gcc.gnu.org/PR12399
10120 420. http://gcc.gnu.org/PR13068
10121 421. http://gcc.gnu.org/PR10060
10122 422. http://gcc.gnu.org/PR10555
10123 423. http://gcc.gnu.org/PR10706
10124 424. http://gcc.gnu.org/PR11496
10125 425. http://gcc.gnu.org/PR11741
10126 426. http://gcc.gnu.org/PR12440
10127 427. http://gcc.gnu.org/PR12632
10128 428. http://gcc.gnu.org/PR12712
10129 429. http://gcc.gnu.org/PR12726
10130 430. http://gcc.gnu.org/PR12890
10131 431. http://gcc.gnu.org/PR12900
10132 432. http://gcc.gnu.org/PR13060
10133 433. http://gcc.gnu.org/PR13289
10134 434. http://gcc.gnu.org/PR13318
10135 435. http://gcc.gnu.org/PR13392
10136 436. http://gcc.gnu.org/PR13574
10137 437. http://gcc.gnu.org/PR13475
10138 438. http://gcc.gnu.org/PR13797
10139 439. http://gcc.gnu.org/PR13824
10140 440. http://gcc.gnu.org/PR8776
10141 441. http://gcc.gnu.org/PR10339
10142 442. http://gcc.gnu.org/PR11350
10143 443. http://gcc.gnu.org/PR12826
10144 444. http://gcc.gnu.org/PR12500
10145 445. http://gcc.gnu.org/PR12941
10146 446. http://gcc.gnu.org/PR12953
10147 447. http://gcc.gnu.org/PR13041
10148 448. http://gcc.gnu.org/PR13507
10149 449. http://gcc.gnu.org/PR13382
10150 450. http://gcc.gnu.org/PR13394
10151 451. http://gcc.gnu.org/PR13400
10152 452. http://gcc.gnu.org/PR13521
10153 453. http://gcc.gnu.org/PR2094
10154 454. http://gcc.gnu.org/PR2294
10155 455. http://gcc.gnu.org/PR5050
10156 456. http://gcc.gnu.org/PR9371
10157 457. http://gcc.gnu.org/PR9546
10158 458. http://gcc.gnu.org/PR10081
10159 459. http://gcc.gnu.org/PR10093
10160 460. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#61
10161 461. http://gcc.gnu.org/PR10095
10162 462. http://gcc.gnu.org/PR11554
10163 463. http://gcc.gnu.org/PR12297
10164 464. http://gcc.gnu.org/PR12352
10165 465. http://gcc.gnu.org/PR12438
10166 466. http://gcc.gnu.org/PR12540
10167 467. http://gcc.gnu.org/PR12594
10168 468. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#60
10169 469. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#63
10170 470. http://gcc.gnu.org/PR12657
10171 471. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#292
10172 472. http://gcc.gnu.org/PR12696
10173 473. http://gcc.gnu.org/PR12815
10174 474. http://gcc.gnu.org/PR12862
10175 475. http://gcc.gnu.org/PR12926
10176 476. http://gcc.gnu.org/PR12967
10177 477. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html
10178 478. http://gcc.gnu.org/PR12971
10179 479. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#328
10180 480. http://gcc.gnu.org/PR13007
10181 481. http://gcc.gnu.org/PR13009
10182 482. http://gcc.gnu.org/PR13057
10183 483. http://gcc.gnu.org/PR13070
10184 484. http://gcc.gnu.org/PR13081
10185 485. http://gcc.gnu.org/PR13239
10186 486. http://gcc.gnu.org/PR13262
10187 487. http://gcc.gnu.org/PR13290
10188 488. http://gcc.gnu.org/PR13323
10189 489. http://gcc.gnu.org/PR13369
10190 490. http://gcc.gnu.org/PR13371
10191 491. http://gcc.gnu.org/PR13445
10192 492. http://gcc.gnu.org/PR13461
10193 493. http://gcc.gnu.org/PR13462
10194 494. http://gcc.gnu.org/PR13478
10195 495. http://gcc.gnu.org/PR13544
10196 496. http://gcc.gnu.org/PR13650
10197 497. http://gcc.gnu.org/PR13683
10198 498. http://gcc.gnu.org/PR13688
10199 499. http://gcc.gnu.org/PR13774
10200 500. http://gcc.gnu.org/PR13884
10201 501. http://gcc.gnu.org/PR10746
10202 502. http://gcc.gnu.org/PR11433
10203 503. http://gcc.gnu.org/PR12633
10204 504. http://gcc.gnu.org/PR13037
10205 505. http://gcc.gnu.org/PR13213
10206 506. http://gcc.gnu.org/PR4490
10207 507. http://gcc.gnu.org/PR12292
10208 508. http://gcc.gnu.org/PR12441
10209 509. http://gcc.gnu.org/PR12943
10210 510. http://gcc.gnu.org/PR13608
10211 511. http://gcc.gnu.org/PR11598
10212 512. http://gcc.gnu.org/PR11793
10213 513. http://gcc.gnu.org/PR12467
10214 514. http://gcc.gnu.org/PR12537
10215 515. http://gcc.gnu.org/PR12496
10216 516. http://gcc.gnu.org/PR12865
10217 517. http://gcc.gnu.org/PR13354
10218 518. http://gcc.gnu.org/PR10467
10219 519. http://gcc.gnu.org/PR11226
10220 520. http://gcc.gnu.org/PR11227
10221 521. http://gcc.gnu.org/PR12644
10222 522. http://gcc.gnu.org/PR13149
10223 523. http://gcc.gnu.org/PR12654
10224 524. http://gcc.gnu.org/PR12965
10225 525. http://gcc.gnu.org/PR13031
10226 526. http://gcc.gnu.org/PR11634
10227 527. http://gcc.gnu.org/PR12158
10228 528. http://gcc.gnu.org/PR11992
10229 529. http://gcc.gnu.org/PR9365
10230 530. http://gcc.gnu.org/PR10392
10231 531. http://gcc.gnu.org/PR11322
10232 532. http://gcc.gnu.org/PR13069
10233 533. http://gcc.gnu.org/PR13302
10234 534. http://gcc.gnu.org/PR13585
10235 535. http://gcc.gnu.org/PR8916
10236 536. http://gcc.gnu.org/PR11576
10237 537. http://gcc.gnu.org/PR13122
10238 538. http://gcc.gnu.org/PR13256
10239 539. http://gcc.gnu.org/PR13373
10240 540. http://gcc.gnu.org/PR12561
10241 541. http://gcc.gnu.org/PR6243
10242 542. http://gcc.gnu.org/PR11397
10243 543. http://gcc.gnu.org/PR12505
10244 544. http://gcc.gnu.org/PR13150
10245 545. http://gcc.gnu.org/PR12666
10246 546. http://gcc.gnu.org/PR12969
10247 547. http://gcc.gnu.org/PR10819
10248 548. http://gcc.gnu.org/PR11612
10249 549. http://gcc.gnu.org/PR13211
10250 550. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.3.4
10251 551. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.3.5
10252 552. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.3.6
10253 553. http://gcc.gnu.org/onlinedocs/
10254 554. mailto:gcc-help@gcc.gnu.org
10255 555. mailto:gcc@gcc.gnu.org
10256 556. http://gcc.gnu.org/lists.html
10257 557. http://www.fsf.org/
10258 558. http://gcc.gnu.org/about.html
10259 559. http://validator.w3.org/check/referer
10260 ======================================================================
10261 http://gcc.gnu.org/gcc-3.2/index.html
10262 GCC 3.2 Release Series
10266 The [1]GNU project and the GCC developers are pleased to announce the
10267 release of GCC 3.2.3.
10269 The purpose of the GCC 3.2 release series is to provide a stable
10270 platform for OS distributors to use building their next releases. A
10271 primary objective was to stabilize the C++ ABI; we believe that the
10272 interface to the compiler and the C++ standard library are now
10275 Be aware that C++ code compiled by GCC 3.2.x will (in general) not
10276 interoperate with code compiled by GCC 3.1.1 or earlier.
10278 Please refer to our [2]detailed list of news, caveats, and bug-fixes
10279 for further information.
10284 April 25, 2003 ([3]changes)
10287 February 5, 2003 ([4]changes)
10290 November 19, 2002 ([5]changes)
10293 August 14, 2002 ([6]changes)
10295 References and Acknowledgements
10297 GCC used to stand for the GNU C Compiler, but since the compiler
10298 supports several other languages aside from C, it now stands for the
10299 GNU Compiler Collection.
10301 A list of [7]successful builds is updated as new information becomes
10304 The GCC developers would like to thank the numerous people that have
10305 contributed new features, improvements, bug fixes, and other changes as
10306 well as test results to GCC. This [8]amazing group of volunteers is
10307 what makes GCC successful.
10309 For additional information about GCC please refer to the [9]GCC project
10310 web site or contact the [10]GCC development mailing list.
10312 To obtain GCC please use [11]our mirror sites, or our CVS server.
10315 For questions related to the use of GCC, please consult these web
10316 pages and the [12]GCC manuals. If that fails, the
10317 [13]gcc-help@gcc.gnu.org mailing list might help. Comments on these
10318 web pages and the development of GCC are welcome on our developer
10319 list at [14]gcc@gcc.gnu.org. All of [15]our lists have public
10322 Copyright (C) [16]Free Software Foundation, Inc. Verbatim copying and
10323 distribution of this entire article is permitted in any medium,
10324 provided this notice is preserved.
10326 These pages are [17]maintained by the GCC team. Last modified
10331 1. http://www.gnu.org/
10332 2. http://gcc.gnu.org/gcc-3.2/changes.html
10333 3. http://gcc.gnu.org/gcc-3.2/changes.html#3.2.3
10334 4. http://gcc.gnu.org/gcc-3.2/changes.html#3.2.2
10335 5. http://gcc.gnu.org/gcc-3.2/changes.html#3.2.1
10336 6. http://gcc.gnu.org/gcc-3.2/changes.html#3.2
10337 7. http://gcc.gnu.org/gcc-3.2/buildstat.html
10338 8. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
10339 9. http://gcc.gnu.org/index.html
10340 10. mailto:gcc@gcc.gnu.org
10341 11. http://gcc.gnu.org/mirrors.html
10342 12. http://gcc.gnu.org/onlinedocs/
10343 13. mailto:gcc-help@gcc.gnu.org
10344 14. mailto:gcc@gcc.gnu.org
10345 15. http://gcc.gnu.org/lists.html
10346 16. http://www.fsf.org/
10347 17. http://gcc.gnu.org/about.html
10348 18. http://validator.w3.org/check/referer
10349 ======================================================================
10350 http://gcc.gnu.org/gcc-3.2/changes.html
10351 GCC 3.2 Release Series
10352 Changes, New Features, and Fixes
10354 The latest release in the 3.2 release series is [1]GCC 3.2.3.
10356 Caveats and New Features
10360 * The C++ compiler does not correctly zero-initialize
10361 pointers-to-data members. You must explicitly initialize them. For
10362 example: int S::*m(0); will work, but depending on
10363 default-initialization to zero will not work. This bug cannot be
10364 fixed in GCC 3.2 without inducing unacceptable risks. It will be
10366 * This GCC release is based on the GCC 3.1 sourcebase, and thus has
10367 all the [2]changes in the GCC 3.1 series. In addition, GCC 3.2 has
10368 a number of C++ ABI fixes which make its C++ compiler generate
10369 binary code which is incompatible with the C++ compilers found in
10370 earlier GCC releases, including GCC 3.1 and GCC 3.1.1.
10372 Frontend Enhancements
10376 * The method of constructing the list of directories to be searched
10377 for header files has been revised. If a directory named by a -I
10378 option is a standard system include directory, the option is
10379 ignored to ensure that the default search order for system
10380 directories and the special treatment of system header files are
10382 * The C and Objective-C compilers no longer accept the "Naming Types"
10383 extension (typedef foo = bar); it was already unavailable in C++.
10384 Code which uses it will need to be changed to use the "typeof"
10385 extension instead: typedef typeof(bar) foo. (We have removed this
10386 extension without a period of deprecation because it has caused the
10387 compiler to crash since version 3.0 and no one noticed until very
10388 recently. Thus we conclude it is not in widespread use.)
10392 * GCC 3.2 fixed serveral differences between the C++ ABI implemented
10393 in GCC and the multi-vendor standard, but more have been found
10394 since the release. 3.2.1 adds a new warning, -Wabi, to warn about
10395 code which is affected by these bugs. We will fix these bugs in
10396 some future release, once we are confident that all have been
10397 found; until then, it is our intention to make changes to the ABI
10398 only if they are necessary for correct compilation of C++, as
10399 opposed to conformance to the ABI documents.
10400 * For details on how to build an ABI compliant compiler for GNU/Linux
10401 systems, check the [3]common C++ ABI page.
10403 New Targets and Target Specific Improvements
10407 * Fixed a number of bugs in SSE and MMX intrinsics.
10408 * Fixed common compiler crashes with SSE instruction set enabled
10409 (implied by -march=pentium3, pentium4, athlon-xp)
10410 * __m128 and __m128i is not 128bit aligned when used in structures.
10414 * A bug whereby the compiler could generate bad code for bzero has
10416 * ABI fixes (implying ABI incompatibilities with previous version in
10418 * Fixed prefetch code generation
10419 __________________________________________________________________
10423 3.2.3 is a bug fix release only; there are no new features that were
10424 not present in GCC 3.2.2.
10428 This section lists the problem reports (PRs) from GCC's bug tracking
10429 system that are known to be fixed in the 3.2.3 release. This list might
10430 not be complete (that is, it is possible that some PRs that have been
10431 fixed are not listed here), and some of the titles have been changed to
10432 make them more clear.
10434 Internal Compiler Errors (multi-platform)
10436 * [4]3782: (c++) -quiet -fstats produces a segmentation fault in
10438 * [5]6440: (c++) template specializations cause ICE
10439 * [6]7050: (c++) ICE on: (i ? get_string() : throw)
10440 * [7]7741: ICE on conflicting types (make_decl_rtl in varasm.c)
10441 * [8]7982: (c++) ICE due to infinite recursion (using STL set)
10442 * [9]8068: exceedingly high (infinite) memory usage
10443 * [10]8178: ICE with __builtin_ffs
10444 * [11]8396: ICE in copy_to_mode_reg, in explow.c
10445 * [12]8674: (c++) ICE in cp_expr_size, in cp/cp-lang.c
10446 * [13]9768: ICE when optimizing inline code at -O2
10447 * [14]9798: (c++) Infinite recursion (segfault) in
10448 cp/decl.c:push_using_directive with recursive using directives
10449 * [15]9799: mismatching structure initializer with nested flexible
10451 * [16]9928: ICE on duplicate enum declaration
10452 * [17]10114: ICE in mem_loc_descriptor, in dwarf2out.c (affects
10454 * [18]10352: ICE in find_reloads_toplev
10455 * [19]10336: ICE with -Wunreachable-code
10459 * [20]8224: Incorrect joining of signed and unsigned division
10460 * [21]8613: -O2 produces wrong code with builtin strlen and
10462 * [22]8828: gcc reports some code is unreachable when it is not
10463 * [23]9226: GCSE breaking argument passing
10464 * [24]9853: miscompilation of non-constant structure initializer
10465 * [25]9797: C99-style struct initializers are miscompiled
10466 * [26]9967: Some standard C function calls should not be replaced
10467 when optimizing for size
10468 * [27]10116: ce2: invalid merge of join_bb in the context of switch
10470 * [28]10171: wrong code for inlined function
10471 * [29]10175: -Wunreachable-code doesn't work for single lines
10473 C++ compiler and library:
10475 * [30]8316: Confusing diagnostic for code that misuses conversion
10477 * [31]9169: filebuf output fails if codecvt<>::out returns noconv
10478 * [32]9420: incomplete type incorrectly reported
10479 * [33]9459: typeof in return type specification of template not
10481 * [34]9507: filebuf::open handles ios_base::ate incorrectly
10482 * [35]9538: Out-of-bounds memory access in streambuf::sputbackc
10483 * [36]9602: Total confusion about template/friend/virtual/abstract
10484 * [37]9993: destructor not called for local object created within and
10485 returned from infinite loop
10486 * [38]10167: ieee_1003.1-2001 locale specialisations on a glibc-2.3.2
10489 Java compiler and library:
10491 * [39]9652: libgcj build fails on irix6.5.1[78]
10492 * [40]10144: gas on solaris complains about bad .stabs lines for
10493 java, native as unaffected
10495 x86-specific (Intel/AMD):
10497 * [41]8746: gcc miscompiles Linux kernel ppa driver on x86
10498 * [42]9888: -mcpu=k6 -Os produces out of range loop instructions
10499 * [43]9638: Cross-build for target i386-elf and i586-pc-linux-gnu
10501 * [44]9954: Cross-build for target i586-pc-linux-gnu (--with-newlib)
10506 * [45]7784: [Sparc] ICE in extract_insn, in recog.c
10507 * [46]7796: sparc extra failure with -m64 on execute/930921-1.c in
10509 * [47]8281: ICE when compiling with -O2 -fPIC for Ultrasparc
10510 * [48]8366: [Sparc] C testsuite failure with -m64 -fpic -O in
10512 * [49]8726: gcc -O2 miscompiles Samba 2.2.7 on 32-bit sparc
10513 * [50]9414: Scheduling bug on Ultrasparc
10514 * [51]10067: GCC-3.2.2 outputs invalid asm on sparc64
10518 * [52]7248: broken "inclusive or" code
10519 * [53]8343: m68k-elf/rtems ICE at instantiate_virtual_regs_1
10523 * [54]9732: Wrong code with -O2 -fPIC
10524 * [55]10073: ICE: powerpc cannot split insn
10528 * [56]7702: optimization problem on a DEC alpha under OSF1
10529 * [57]9671: gcc.3.2.2 does not build on a HP Tru64 Unix v5.1B system
10533 * [58]8694: <string> breaks <ctype.h> on HP-UX 10.20 (DUP: 9275)
10534 * [59]9953: (ada) gcc 3.2.x can't build 3.3-branch ada on HP-UX 10
10536 * [60]10271: Floating point args don't get reloaded across function
10541 * [61]6362: mips-irix6 gcc-3.1 C testsuite failure with -mips4 in
10546 * [62]10377: gcc-3.2.2 creates bad assembler code for cris
10548 Miscellaneous and minor bugs:
10550 * [63]6955: collect2 says "core dumped" when there is no core
10551 __________________________________________________________________
10555 Beginning with 3.2.2, GCC's Makefile suite supports redirection of make
10556 install by means of the DESTDIR variable. Parts of the GCC tree have
10557 featured that support long before, but now it is available even from
10560 Other than that, GCC 3.2.2 is a bug fix release only; there are no new
10561 features that were not present in GCC 3.2.1.
10565 On the following i386-based systems GCC 3.2.1 broke the C ABI wrt.
10566 functions returning structures: Cygwin, FreeBSD (GCC 3.2.1 as shipped
10567 with FreeBSD 5.0 does not have this problem), Interix, a.out-based
10568 GNU/Linux and NetBSD, OpenBSD, and Darwin. GCC 3.2.2 reverts this ABI
10569 change, and thus restores ABI-compatibility with previous releases
10570 (except GCC 3.2.1) on these platforms.
10572 This section lists the problem reports (PRs) from GCC's bug tracking
10573 system that are known to be fixed in the 3.2.2 release. This list might
10574 not be complete (that is, it is possible that some PRs that have been
10575 fixed are not listed here) and some of the titles have been changed to
10576 make them more clear.
10578 Internal Compiler Errors (multi-platform)
10580 * [64]5919: (c++) ICE when passing variable array to template
10582 * [65]7129: (c++) ICE with min/max assignment operators (<?= and >?=)
10583 * [66]7507: ICE with -O2 when address of called function is a
10584 complicated expression
10585 * [67]7622: ICE with nested inline functions if function's address is
10587 * [68]7681: (fortran) ICE in compensate_edge, in reg-stack.c (also PR
10589 * [70]8031: (c++) ICE in code comparing typeids and casting from
10591 * [71]8275: ICE in simplify_subreg
10592 * [72]8332: (c++) builtin strlen/template interaction causes ICE
10593 * [73]8372: (c++) ICE on explicit call of destructor
10594 * [74]8439: (c, not c++) empty struct causes ICE
10595 * [75]8442: (c++) ICE with nested template classes
10596 * [76]8518: ICE when compiling mplayer ("extern inline" issue)
10597 * [77]8615: (c++) ICE with out-of-range character constant template
10599 * [78]8663: (c++) ICE in cp_expr_size, at cp-lang.c:307
10600 * [79]8799: (c++) ICE: error reporting routines re-entered
10601 * [80]9328: (c++) ICE with typeof(X) for overloaded X
10602 * [81]9465: (preprocessor) cpp -traditional ICE on null bytes
10604 C++ (compiler and library) bugs
10606 * [82]47: scoping in nested classes is broken
10607 * [83]6745: problems with iostream rdbuf() member function
10608 * [84]8214: conversion from const char* const to char* sometimes
10610 * [85]8493: builtin strlen and overload resolution (same bug as
10612 * [87]8503: strange behaviour of function types
10613 * [88]8727: compiler confused by inheritance from an anonymous struct
10614 * [89]7445: poor performance of std::locale::classic() in
10615 multi-threaded applications
10616 * [90]8230: mishandling of overflow in vector<T>::resize
10617 * [91]8399: sync_with_stdio(false) breaks unformatted input
10618 * [92]8662: illegal access of private member of unnamed class is
10620 * [93]8707: "make distclean" fails in libstdc++-v3 directory
10621 * [94]8708: __USE_MALLOC doesn't work
10622 * [95]8790: Use of non-thread-safe strtok in src/localename.cc
10623 * [96]8887: Bug in date formats with --enable-clocale=generic
10624 * [97]9076: Call Frame Instructions are not handled correctly during
10626 * [98]9151: std::setprecision limited to 16 digits when outputting a
10628 * [99]9168: codecvt<char, char, mbstate_t> overwrites output buffers
10629 * [100]9269: libstdc++ headers: explicit specialization of function
10630 must precede its first use
10631 * [101]9322: return value of basic_streambuf<>::getloc affected by
10633 * [102]9433: segfault in runtime support for dynamic_cast
10635 C and optimizer bugs
10637 * [103]8032: GCC incorrectly initializes static structs that have
10639 * [104]8639: simple arithmetic expression broken
10640 * [105]8794: optimization improperly eliminates certain expressions
10641 * [106]8832: traditional "asm volatile" code is illegally optimized
10642 * [107]8988: loop optimizer bug: with -O2, code is generated that
10643 segfaults (found on i386, bug present for all platforms)
10644 * [108]9492: structure copy clobbers subsequent stores to structure
10648 * [109]9267: Objective-C parser won't build with newer bison versions
10653 * [110]8344: Ada build problem due to conflict between gcc/final.o,
10658 * [111]8524: _Pragma within macros is improperly expanded
10659 * [112]8880: __WCHAR_TYPE__ macro incorrectly set to "long int" with
10664 * [113]9090: arm ICE with >= -O2; regression from gcc-2.95
10666 x86-specific (Intel/AMD)
10668 * [114]8588: ICE in extract_insn, at recog.c:NNNN (shift instruction)
10669 * [115]8599: loop unroll bug with -march=k6-3
10670 * [116]9506: ABI breakage in structure return (affects BSD and
10671 Cygwin, but not GNU/Linux)
10673 FreeBSD 5.0 specific
10675 * [117]9484: GCC 3.2.1 Bootstrap failure on FreeBSD 5.0
10679 * [118]9292: hppa1.1-rtems configurery problems
10680 * [119]9293: [m68k-elf/rtems] config/m68k/t-crtstuff bug
10681 * [120]9295: [mips-rtems] config/mips/rtems.h init/fini issue
10682 * [121]9296: gthr-rtems regression
10683 * [122]9316: powerpc-rtems: extending multilibs
10687 * [123]9493: ICE with -O2 when building a simple function
10691 * [124]7341: hyperlink to gcov in GCC documentation doesn't work
10692 * [125]8947: Please add a warning about "-malign-double" in docs
10693 * [126]7448, [127]8882: typo cleanups
10694 __________________________________________________________________
10698 3.2.1 adds a new warning, -Wabi. This option warns when GNU C++
10699 generates code that is known not to be binary-compatible with the
10700 vendor-neutral ia32/ia64 ABI. Please consult the GCC manual, included
10701 in the distribution, for details.
10703 This release also removes an old GCC extension, "naming types", and the
10704 documentation now directs users to use a different GCC extension,
10705 __typeof__, instead. The feature had evidently been broken for a while.
10707 Otherwise, 3.2.1 is a bug fix release only; other than bug fixes and
10708 the new warning there are no new features that were not present in GCC
10711 In addition, the previous fix for [128]PR 7445 (poor performance of
10712 std::locale::classic() in multi-threaded applications) was reverted
10713 ("unfixed"), because the "fix" was not thread-safe.
10717 This section lists the problem reports (PRs) from GCC's bug tracking
10718 system that are known to be fixed in the 3.2.1 release. This list might
10719 not be complete (that is, it is possible that some PRs that have been
10720 fixed are not listed here). As you can see, the number of bug fixes is
10721 quite large, so it is strongly recommended that users of earlier GCC
10722 3.x releases upgrade to GCC 3.2.1.
10724 Internal Compiler Errors (multi-platform)
10726 * [129]2521: (c++) ICE in build_ptrmemfunc, in cp/typeck.c
10727 * [130]5661: (c++) ICE instantiating template on array of unknown
10729 * [131]6419: (c++) ICE in make_decl_rtl for "longest" attribute on
10731 * [132]6994: (c++) ICE in find_function_data
10732 * [133]7150: preprocessor: GCC -dM -E gives an ICE
10733 * [134]7160: ICE when optimizing branches without a return value
10734 * [135]7228: (c++) ICE when using member template and template
10736 * [136]7266: (c++) ICE with -pedantic on missing typename
10737 * [137]7353: ICE from use of "Naming Types" extension, see above
10738 * [138]7411: ICE in instantiate_virtual_regs_1, in function.c
10739 * [139]7478: (c++) ICE on static_cast inside template
10740 * [140]7526: preprocessor core dump when _Pragma implies #pragma
10742 * [141]7721: (c++) ICE on simple (but incorrect) template ([142]7803
10744 * [143]7754: (c++) ICE on union with template parameter
10745 * [144]7788: (c++) redeclaring a definition as an incomplete class
10747 * [145]8031: (c++) ICE in comptypes, in cp/typeck.c
10748 * [146]8055: preprocessor dies with SIG11 when building FreeBSD
10750 * [147]8067: (c++) ICE due to mishandling of __FUNCTION__ and related
10752 * [148]8134: (c++) ICE in force_store_init_value on legal code
10753 * [149]8149: (c++) ICE on incomplete type
10754 * [150]8160: (c++) ICE in build_modify_expr, in cp/typeck.c: array
10757 C++ (compiler and library) bugs
10759 * [151]5607: No pointer adjustment in covariant return types
10760 * [152]6579: Infinite loop with statement expressions in member
10762 * [153]6803: Default copy constructor bug in GCC 3.1
10763 * [154]7176: g++ confused by friend and static member with same name
10764 * [155]7188: Segfault with template class and recursive (incorrect)
10766 * [156]7306: Regression: GCC 3.x fails to compile code with virtual
10767 inheritance if a method has a variable number of arguments
10768 * [157]7461: ctype<char>::classic_table() returns offset array on
10770 * [158]7524: f(const float arg[3]) fails
10771 * [159]7584: Erroneous ambiguous base error on using declaration
10772 * [160]7676: Member template overloading problem
10773 * [161]7679: infinite loop when a right parenthesis is missing
10774 * [162]7811: default locale not taken from environment
10775 * [163]7961: compare( char *) implemented incorrectly in
10777 * [164]8071: basic_ostream::operator<<(streambuf*) loops forever if
10778 streambuf::underflow() leaves gptr() NULL (dups: [165]8127,
10780 * [167]8096: deque::at() throws std::range_error instead of
10782 * [168]8127: cout << cin.rdbuf() infinite loop
10783 * [169]8218: Excessively large memory consumed for classes with large
10785 * [170]8287: GCC 3.2: Destructor called for non-constructed local
10787 * [171]8347: empty vector range used in string construction causes
10789 * [172]8348: fail() flag is set in istringstream when eof() flag is
10791 * [173]8391: regression: infinite loop in cp/decl2.c(finish_file)
10793 C and optimizer bugs
10795 * [174]6627: -fno-align-functions doesn't seem to disable function
10797 * [175]6631: life_analysis misoptimizes code to initialize fields of
10799 * [176]7102: unsigned char division results in floating exception
10800 * [177]7120: Run once loop should *always* be unrolled
10802 * [178]7209: Bug involving array referencing and ?: operator
10803 * [179]7515: invalid inlining of global function with -O3
10804 * [180]7814: incorrect scheduling for glibc-2.2.92 strcpy test
10805 * [181]8467: bug in sibling call optimization
10809 * [182]4890: incorrect line markers from the traditional preprocessor
10810 * [183]7357: -M option omits system headers files (making it the same
10812 * [184]7358: Changes to Sun's make Dependencies
10813 * [185]7602: C++ header files found in CPLUS_INCLUDE_PATH treated as
10815 * [186]7862: Interrupting GCC -MD removes .d file but not .o
10816 * [187]8190: Failed compilation deletes -MD dependency file
10817 * [188]8524: _Pragma within macro is improperly expanded
10819 x86 specific (Intel/AMD)
10821 * [189]5351: (i686-only) function pass-by-value structure copy
10822 corrupts stack ([190]7591 is a duplicate)
10823 * [191]6845, [192]7034, [193]7124, [194]7174: ICE's with
10824 -march=pentium3/pentium2/athlon (these are all the same underlying
10825 bug, in MMX register use)
10826 * [195]7134, [196]7375, [197]7390: ICE with -march=athlon (maybe same
10828 * [198]6890: xmmintrin.h, _MM_TRANSPOSE4_PS is broken
10829 * [199]6981: wrong code in 64-bit manipulation on x86
10830 * [200]7242: GCC -mcpu=pentium[23] doesn't define __tune_pentiumpro__
10832 * [201]7396: ix86: cmpgt_ss, cmpge_ss, cmpngt_ss, and cmpnge_ss SSE
10833 intrinsics are broken
10834 * [202]7630: GCC 3.2 breaks on Mozilla 1.0's JS sources with
10836 * [203]7693: Typo in i386 mmintrin.h header
10837 * [204]7723: ICE - Pentium3 sse - GCC 3.2
10838 * [205]7951: ICE on -march=pentium4 -O2 -mfpmath=sse
10839 * [206]8146: (i686 only) gcc 3.2 miscompiles gcc 2.95.3
10843 * [207]5967: GCC bug when profiling nested functions on powerpc
10844 * [208]6984: wrong code generated with -O2, -O3, -Os for do-while
10846 * [209]7114: PowerPC: ICE building strcoll.op from glibc-2.2.5
10847 * [210]7130: miscompiled code for GCC-3.1 on
10848 powerpc-unknown-linux-gnu with -funroll-all-loops
10849 * [211]7133: PowerPC ICE: unrecognizable insn
10850 * [212]7380: ICE in extract_insn, at recog.c:2148
10851 * [213]8252: ICE on Altivec code with optimization turned on
10852 * [214]8451: Altivec ICE in GCC 3.2
10856 * [215]7250: __ashrdi3 returns wrong value on 32 bit hppa
10860 * [216]6668: when using --disable-multilib, libgcc_s.so is installed
10861 in the wrong place on sparc-solaris
10862 * [217]7151: ICE when compiling for UltraSPARC
10863 * [218]7335: SPARC: ICE in verify_wide_reg (flow.c:557) with long
10865 * [219]7842: [REGRESSION] SPARC code gen bug
10869 * [220]7856: [arm] invalid offset in constant pool reference
10870 * [221]7967: optimization produces wrong code (ARM)
10874 * [222]7374: __builtin_fabsl broken on alpha
10878 * [223]7370: ICE in fixup_var_refs_1 on s390x
10879 * [224]7409: loop optimization bug on s390x-linux-gnu
10880 * [225]8232: s390x: ICE when using bcmp with int length argument
10884 * [226]7623: SCO OpenServer build fails with machmode.def: undefined
10885 symbol: BITS_PER_UNIT
10887 m68k/Coldfire specific
10889 * [227]8314: crtbegin, crtend need to be multilib'ed for this
10894 * [228]761: Document some undocumented options
10895 * [229]5610: Fix documentation about invoking SSE instructions
10897 * [230]7484: List -Wmissing-declarations as C-only option
10898 * [231]7531: -mcmodel not documented for x86-64
10899 * [232]8120: Update documentation of bad use of ##
10900 __________________________________________________________________
10904 3.2 is a small bug fix release, but there is a change to the
10905 application binary interface (ABI), hence the change to the second part
10906 of the version number.
10908 The main purpose of the 3.2 release is to correct a couple of problems
10909 in the C++ ABI, with the intention of providing a stable interface
10910 going forward. Accordingly, 3.2 is only a small change to 3.1.1.
10916 * [233]7320: g++ 3.2 relocation problem
10917 * [234]7470: vtable: virtual function pointers not in declaration
10922 * [235]6410: Trouble with non-ASCII monetary symbols and wchar_t
10923 * [236]6503, [237]6642, [238]7186: Problems with comparing or
10924 subtracting various types of const and non-const iterators
10925 * [239]7216: ambiguity with basic_iostream::traits_type
10926 * [240]7220: problem with basic_istream::ignore(0,delimiter)
10927 * [241]7222: locale::operator==() doesn't work on std::locale("")
10928 * [242]7286: placement operator delete issue
10929 * [243]7442: cxxabi.h does not match the C++ ABI
10930 * [244]7445: poor performance of std::locale::classic() in
10931 multi-threaded applications
10935 * [245]7291: off-by-one in generated inline bzero code for x86-64
10938 For questions related to the use of GCC, please consult these web
10939 pages and the [246]GCC manuals. If that fails, the
10940 [247]gcc-help@gcc.gnu.org mailing list might help. Comments on these
10941 web pages and the development of GCC are welcome on our developer
10942 list at [248]gcc@gcc.gnu.org. All of [249]our lists have public
10945 Copyright (C) [250]Free Software Foundation, Inc. Verbatim copying and
10946 distribution of this entire article is permitted in any medium,
10947 provided this notice is preserved.
10949 These pages are [251]maintained by the GCC team. Last modified
10954 1. http://gcc.gnu.org/gcc-3.2/changes.html#3.2.3
10955 2. http://gcc.gnu.org/gcc-3.1/changes.html
10956 3. http://gcc.gnu.org/gcc-3.2/c++-abi.html
10957 4. http://gcc.gnu.org/PR3782
10958 5. http://gcc.gnu.org/PR6440
10959 6. http://gcc.gnu.org/PR7050
10960 7. http://gcc.gnu.org/PR7741
10961 8. http://gcc.gnu.org/PR7982
10962 9. http://gcc.gnu.org/PR8068
10963 10. http://gcc.gnu.org/PR8178
10964 11. http://gcc.gnu.org/PR8396
10965 12. http://gcc.gnu.org/PR8674
10966 13. http://gcc.gnu.org/PR9768
10967 14. http://gcc.gnu.org/PR9798
10968 15. http://gcc.gnu.org/PR9799
10969 16. http://gcc.gnu.org/PR9928
10970 17. http://gcc.gnu.org/PR10114
10971 18. http://gcc.gnu.org/PR10352
10972 19. http://gcc.gnu.org/PR10336
10973 20. http://gcc.gnu.org/PR8224
10974 21. http://gcc.gnu.org/PR8613
10975 22. http://gcc.gnu.org/PR8828
10976 23. http://gcc.gnu.org/PR9226
10977 24. http://gcc.gnu.org/PR9853
10978 25. http://gcc.gnu.org/PR9797
10979 26. http://gcc.gnu.org/PR9967
10980 27. http://gcc.gnu.org/PR10116
10981 28. http://gcc.gnu.org/PR10171
10982 29. http://gcc.gnu.org/PR10175
10983 30. http://gcc.gnu.org/PR8316
10984 31. http://gcc.gnu.org/PR9169
10985 32. http://gcc.gnu.org/PR9420
10986 33. http://gcc.gnu.org/PR9459
10987 34. http://gcc.gnu.org/PR9507
10988 35. http://gcc.gnu.org/PR9538
10989 36. http://gcc.gnu.org/PR9602
10990 37. http://gcc.gnu.org/PR9993
10991 38. http://gcc.gnu.org/PR10167
10992 39. http://gcc.gnu.org/PR9652
10993 40. http://gcc.gnu.org/PR10144
10994 41. http://gcc.gnu.org/PR8746
10995 42. http://gcc.gnu.org/PR9888
10996 43. http://gcc.gnu.org/PR9638
10997 44. http://gcc.gnu.org/PR9954
10998 45. http://gcc.gnu.org/PR7784
10999 46. http://gcc.gnu.org/PR7796
11000 47. http://gcc.gnu.org/PR8281
11001 48. http://gcc.gnu.org/PR8366
11002 49. http://gcc.gnu.org/PR8726
11003 50. http://gcc.gnu.org/PR9414
11004 51. http://gcc.gnu.org/PR10067
11005 52. http://gcc.gnu.org/PR7248
11006 53. http://gcc.gnu.org/PR8343
11007 54. http://gcc.gnu.org/PR9732
11008 55. http://gcc.gnu.org/PR10073
11009 56. http://gcc.gnu.org/PR7702
11010 57. http://gcc.gnu.org/PR9671
11011 58. http://gcc.gnu.org/PR8694
11012 59. http://gcc.gnu.org/PR9953
11013 60. http://gcc.gnu.org/PR10271
11014 61. http://gcc.gnu.org/PR6362
11015 62. http://gcc.gnu.org/PR10377
11016 63. http://gcc.gnu.org/PR6955
11017 64. http://gcc.gnu.org/PR5919
11018 65. http://gcc.gnu.org/PR7129
11019 66. http://gcc.gnu.org/PR7507
11020 67. http://gcc.gnu.org/PR7622
11021 68. http://gcc.gnu.org/PR7681
11022 69. http://gcc.gnu.org/PR9528
11023 70. http://gcc.gnu.org/PR8031
11024 71. http://gcc.gnu.org/PR8275
11025 72. http://gcc.gnu.org/PR8332
11026 73. http://gcc.gnu.org/PR8372
11027 74. http://gcc.gnu.org/PR8439
11028 75. http://gcc.gnu.org/PR8442
11029 76. http://gcc.gnu.org/PR8518
11030 77. http://gcc.gnu.org/PR8615
11031 78. http://gcc.gnu.org/PR8663
11032 79. http://gcc.gnu.org/PR8799
11033 80. http://gcc.gnu.org/PR9328
11034 81. http://gcc.gnu.org/PR9465
11035 82. http://gcc.gnu.org/PR47
11036 83. http://gcc.gnu.org/PR6745
11037 84. http://gcc.gnu.org/PR8214
11038 85. http://gcc.gnu.org/PR8493
11039 86. http://gcc.gnu.org/PR8332
11040 87. http://gcc.gnu.org/PR8503
11041 88. http://gcc.gnu.org/PR8727
11042 89. http://gcc.gnu.org/PR7445
11043 90. http://gcc.gnu.org/PR8230
11044 91. http://gcc.gnu.org/PR8399
11045 92. http://gcc.gnu.org/PR8662
11046 93. http://gcc.gnu.org/PR8707
11047 94. http://gcc.gnu.org/PR8708
11048 95. http://gcc.gnu.org/PR8790
11049 96. http://gcc.gnu.org/PR8887
11050 97. http://gcc.gnu.org/PR9076
11051 98. http://gcc.gnu.org/PR9151
11052 99. http://gcc.gnu.org/PR9168
11053 100. http://gcc.gnu.org/PR9269
11054 101. http://gcc.gnu.org/PR9322
11055 102. http://gcc.gnu.org/PR9433
11056 103. http://gcc.gnu.org/PR8032
11057 104. http://gcc.gnu.org/PR8639
11058 105. http://gcc.gnu.org/PR8794
11059 106. http://gcc.gnu.org/PR8832
11060 107. http://gcc.gnu.org/PR8988
11061 108. http://gcc.gnu.org/PR9492
11062 109. http://gcc.gnu.org/PR9267
11063 110. http://gcc.gnu.org/PR8344
11064 111. http://gcc.gnu.org/PR8524
11065 112. http://gcc.gnu.org/PR8880
11066 113. http://gcc.gnu.org/PR9090
11067 114. http://gcc.gnu.org/PR8588
11068 115. http://gcc.gnu.org/PR8599
11069 116. http://gcc.gnu.org/PR9506
11070 117. http://gcc.gnu.org/PR9484
11071 118. http://gcc.gnu.org/PR9292
11072 119. http://gcc.gnu.org/PR9293
11073 120. http://gcc.gnu.org/PR9295
11074 121. http://gcc.gnu.org/PR9296
11075 122. http://gcc.gnu.org/PR9316
11076 123. http://gcc.gnu.org/PR9493
11077 124. http://gcc.gnu.org/PR7341
11078 125. http://gcc.gnu.org/PR8947
11079 126. http://gcc.gnu.org/PR7448
11080 127. http://gcc.gnu.org/PR8882
11081 128. http://gcc.gnu.org/PR7445
11082 129. http://gcc.gnu.org/PR2521
11083 130. http://gcc.gnu.org/PR5661
11084 131. http://gcc.gnu.org/PR6419
11085 132. http://gcc.gnu.org/PR6994
11086 133. http://gcc.gnu.org/PR7150
11087 134. http://gcc.gnu.org/PR7160
11088 135. http://gcc.gnu.org/PR7228
11089 136. http://gcc.gnu.org/PR7266
11090 137. http://gcc.gnu.org/PR7353
11091 138. http://gcc.gnu.org/PR7411
11092 139. http://gcc.gnu.org/PR7478
11093 140. http://gcc.gnu.org/PR7526
11094 141. http://gcc.gnu.org/PR7721
11095 142. http://gcc.gnu.org/PR7803
11096 143. http://gcc.gnu.org/PR7754
11097 144. http://gcc.gnu.org/PR7788
11098 145. http://gcc.gnu.org/PR8031
11099 146. http://gcc.gnu.org/PR8055
11100 147. http://gcc.gnu.org/PR8067
11101 148. http://gcc.gnu.org/PR8134
11102 149. http://gcc.gnu.org/PR8149
11103 150. http://gcc.gnu.org/PR8160
11104 151. http://gcc.gnu.org/PR5607
11105 152. http://gcc.gnu.org/PR6579
11106 153. http://gcc.gnu.org/PR6803
11107 154. http://gcc.gnu.org/PR7176
11108 155. http://gcc.gnu.org/PR7188
11109 156. http://gcc.gnu.org/PR7306
11110 157. http://gcc.gnu.org/PR7461
11111 158. http://gcc.gnu.org/PR7524
11112 159. http://gcc.gnu.org/PR7584
11113 160. http://gcc.gnu.org/PR7676
11114 161. http://gcc.gnu.org/PR7679
11115 162. http://gcc.gnu.org/PR7811
11116 163. http://gcc.gnu.org/PR7961
11117 164. http://gcc.gnu.org/PR8071
11118 165. http://gcc.gnu.org/PR8127
11119 166. http://gcc.gnu.org/PR6745
11120 167. http://gcc.gnu.org/PR8096
11121 168. http://gcc.gnu.org/PR8127
11122 169. http://gcc.gnu.org/PR8218
11123 170. http://gcc.gnu.org/PR8287
11124 171. http://gcc.gnu.org/PR8347
11125 172. http://gcc.gnu.org/PR8348
11126 173. http://gcc.gnu.org/PR8391
11127 174. http://gcc.gnu.org/PR6627
11128 175. http://gcc.gnu.org/PR6631
11129 176. http://gcc.gnu.org/PR7102
11130 177. http://gcc.gnu.org/PR7120
11131 178. http://gcc.gnu.org/PR7209
11132 179. http://gcc.gnu.org/PR7515
11133 180. http://gcc.gnu.org/PR7814
11134 181. http://gcc.gnu.org/PR8467
11135 182. http://gcc.gnu.org/PR4890
11136 183. http://gcc.gnu.org/PR7357
11137 184. http://gcc.gnu.org/PR7358
11138 185. http://gcc.gnu.org/PR7602
11139 186. http://gcc.gnu.org/PR7862
11140 187. http://gcc.gnu.org/PR8190
11141 188. http://gcc.gnu.org/PR8524
11142 189. http://gcc.gnu.org/PR5351
11143 190. http://gcc.gnu.org/PR7591
11144 191. http://gcc.gnu.org/PR6845
11145 192. http://gcc.gnu.org/PR7034
11146 193. http://gcc.gnu.org/PR7124
11147 194. http://gcc.gnu.org/PR7174
11148 195. http://gcc.gnu.org/PR7134
11149 196. http://gcc.gnu.org/PR7375
11150 197. http://gcc.gnu.org/PR7390
11151 198. http://gcc.gnu.org/PR6890
11152 199. http://gcc.gnu.org/PR6981
11153 200. http://gcc.gnu.org/PR7242
11154 201. http://gcc.gnu.org/PR7396
11155 202. http://gcc.gnu.org/PR7630
11156 203. http://gcc.gnu.org/PR7693
11157 204. http://gcc.gnu.org/PR7723
11158 205. http://gcc.gnu.org/PR7951
11159 206. http://gcc.gnu.org/PR8146
11160 207. http://gcc.gnu.org/PR5967
11161 208. http://gcc.gnu.org/PR6984
11162 209. http://gcc.gnu.org/PR7114
11163 210. http://gcc.gnu.org/PR7130
11164 211. http://gcc.gnu.org/PR7133
11165 212. http://gcc.gnu.org/PR7380
11166 213. http://gcc.gnu.org/PR8252
11167 214. http://gcc.gnu.org/PR8451
11168 215. http://gcc.gnu.org/PR7250
11169 216. http://gcc.gnu.org/PR6668
11170 217. http://gcc.gnu.org/PR7151
11171 218. http://gcc.gnu.org/PR7335
11172 219. http://gcc.gnu.org/PR7842
11173 220. http://gcc.gnu.org/PR7856
11174 221. http://gcc.gnu.org/PR7967
11175 222. http://gcc.gnu.org/PR7374
11176 223. http://gcc.gnu.org/PR7370
11177 224. http://gcc.gnu.org/PR7409
11178 225. http://gcc.gnu.org/PR8232
11179 226. http://gcc.gnu.org/PR7623
11180 227. http://gcc.gnu.org/PR8314
11181 228. http://gcc.gnu.org/PR761
11182 229. http://gcc.gnu.org/PR5610
11183 230. http://gcc.gnu.org/PR7484
11184 231. http://gcc.gnu.org/PR7531
11185 232. http://gcc.gnu.org/PR8120
11186 233. http://gcc.gnu.org/PR7320
11187 234. http://gcc.gnu.org/PR7470
11188 235. http://gcc.gnu.org/PR6410
11189 236. http://gcc.gnu.org/PR6503
11190 237. http://gcc.gnu.org/PR6642
11191 238. http://gcc.gnu.org/PR7186
11192 239. http://gcc.gnu.org/PR7216
11193 240. http://gcc.gnu.org/PR7220
11194 241. http://gcc.gnu.org/PR7222
11195 242. http://gcc.gnu.org/PR7286
11196 243. http://gcc.gnu.org/PR7442
11197 244. http://gcc.gnu.org/PR7445
11198 245. http://gcc.gnu.org/PR7291
11199 246. http://gcc.gnu.org/onlinedocs/
11200 247. mailto:gcc-help@gcc.gnu.org
11201 248. mailto:gcc@gcc.gnu.org
11202 249. http://gcc.gnu.org/lists.html
11203 250. http://www.fsf.org/
11204 251. http://gcc.gnu.org/about.html
11205 252. http://validator.w3.org/check/referer
11206 ======================================================================
11207 http://gcc.gnu.org/gcc-3.1/index.html
11212 The [1]GNU project and the GCC developers are pleased to announce the
11213 release of GCC 3.1.1.
11215 The links below still apply to GCC 3.1.1.
11219 The [2]GNU project and the GCC developers are pleased to announce the
11220 release of GCC 3.1.
11222 GCC used to stand for the GNU C Compiler, but since the compiler
11223 supports several other languages aside from C, it now stands for the
11224 GNU Compiler Collection.
11226 A list of [3]successful builds is updated as new information becomes
11229 The GCC developers would like to thank the numerous people that have
11230 contributed [4]new features, improvements, bug fixes, and other changes
11231 as well as test results to GCC. This [5]amazing group of volunteers is
11232 what makes GCC successful.
11234 For additional information about GCC please refer to the [6]GCC project
11235 web site or contact the [7]GCC development mailing list.
11237 To obtain GCC please use [8]our mirror sites, or our CVS server.
11238 __________________________________________________________________
11241 For questions related to the use of GCC, please consult these web
11242 pages and the [9]GCC manuals. If that fails, the
11243 [10]gcc-help@gcc.gnu.org mailing list might help. Comments on these
11244 web pages and the development of GCC are welcome on our developer
11245 list at [11]gcc@gcc.gnu.org. All of [12]our lists have public
11248 Copyright (C) [13]Free Software Foundation, Inc. Verbatim copying and
11249 distribution of this entire article is permitted in any medium,
11250 provided this notice is preserved.
11252 These pages are [14]maintained by the GCC team. Last modified
11257 1. http://www.gnu.org/
11258 2. http://www.gnu.org/
11259 3. http://gcc.gnu.org/gcc-3.1/buildstat.html
11260 4. http://gcc.gnu.org/gcc-3.1/changes.html
11261 5. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
11262 6. http://gcc.gnu.org/index.html
11263 7. mailto:gcc@gcc.gnu.org
11264 8. http://gcc.gnu.org/mirrors.html
11265 9. http://gcc.gnu.org/onlinedocs/
11266 10. mailto:gcc-help@gcc.gnu.org
11267 11. mailto:gcc@gcc.gnu.org
11268 12. http://gcc.gnu.org/lists.html
11269 13. http://www.fsf.org/
11270 14. http://gcc.gnu.org/about.html
11271 15. http://validator.w3.org/check/referer
11272 ======================================================================
11273 http://gcc.gnu.org/gcc-3.1/changes.html
11274 GCC 3.1 Release Series
11275 Changes, New Features, and Fixes
11277 Additional changes in GCC 3.1.1
11279 * A bug related to how structures and unions are returned has been
11280 fixed for powerpc-*-netbsd*.
11281 * An important bug in the implementation of -fprefetch-loop-arrays
11282 has been fixed. Previously the optimization prefetched random
11283 blocks of memory for most targets except for i386.
11284 * The Java compiler now compiles Java programs much faster and also
11285 works with parallel make.
11286 * Nested functions have been fixed for mips*-*-netbsd*.
11287 * Some missing floating point support routines have beed added for
11289 * This [1]message gives additional information about the bugs fixed
11294 * The -traditional C compiler option has been deprecated and will be
11295 removed in GCC 3.3. (It remains possible to preprocess non-C code
11296 with the traditional preprocessor.)
11297 * The default debugging format for most ELF platforms (including
11298 GNU/Linux and FreeBSD; notable exception is Solaris) has changed
11299 from stabs to DWARF2. This requires GDB 5.1.1 or later.
11301 General Optimizer Improvements
11303 * Jan Hubicka, SuSE Labs, together with Richard Henderson, Red Hat,
11304 and Andreas Jaeger, SuSE Labs, has contributed [2]infrastructure
11305 for profile driven optimizations.
11306 Options -fprofile-arcs and -fbranch-probabilities can now be used
11307 to improve speed of the generated code by profiling the actual
11308 program behaviour on typical runs. In the absence of profile info
11309 the compiler attempts to guess the profile statically.
11310 * [3]SPEC2000 and SPEC95 benchmark suites are now used daily to
11311 monitor performance of the generated code.
11312 According to the SPECInt2000 results on an AMD Athlon CPU, the code
11313 generated by GCC 3.1 is 6% faster on the average (8.2% faster with
11314 profile feedback) compared to GCC 3.0. The code produced by GCC 3.0
11315 is about 2.1% faster compared to 2.95.3. Tests were done using the
11316 -O2 -march=athlon command-line options.
11317 * Alexandre Oliva, of Red Hat, has generalized the tree inlining
11318 infrastructure developed by CodeSourcery, LLC for the C++ front
11319 end, so that it is now used in the C front end too. Inlining
11320 functions as trees exposes them earlier to the compiler, giving it
11321 more opportunities for optimization.
11322 * Support for data prefetching instructions has been added to the GCC
11323 back end and several targets. A new __builtin_prefetch intrinsic is
11324 available to explicitly insert prefetch instructions and
11325 experimental support for loop array prefetching has been added (see
11326 -fprefetch-loop-array documentation).
11327 * Support for emitting debugging information for macros has been
11328 added for DWARF2. It is activated using -g3.
11330 New Languages and Language specific improvements
11334 * A few more [4]ISO C99 features.
11335 * The preprocessor is 10-50% faster than the preprocessor in GCC 3.0.
11336 * The preprocessor's symbol table has been merged with the symbol
11337 table of the C, C++ and Objective-C front ends.
11338 * The preprocessor consumes less memory than the preprocessor in GCC
11339 3.0, often significantly so. On normal input files, it typically
11340 consumes less memory than pre-3.0 cccp-based GCC, too.
11344 * -fhonor-std and -fno-honor-std have been removed. -fno-honor-std
11345 was a workaround to allow std compliant code to work with the
11346 non-std compliant libstdc++-v2. libstdc++-v3 is std compliant.
11347 * The C++ ABI has been fixed so that void (A::*)() const is mangled
11348 as "M1AKFvvE", rather than "MK1AFvvE" as before. This change only
11349 affects pointer to cv-qualified member function types.
11350 * The C++ ABI has been changed to correctly handle this code:
11352 void operator delete[] (void *, size_t);
11355 struct B : public A {
11360 The amount of storage allocated for the array will be greater than
11361 it was in 3.0, in order to store the number of elements in the
11362 array, so that the correct size can be passed to operator delete[]
11363 when the array is deleted. Previously, the value passed to operator
11364 delete[] was unpredictable.
11365 This change will only affect code that declares a two-argument
11366 operator delete[] with a second parameter of type size_t in a base
11367 class, and does not override that definition in a derived class.
11368 * The C++ ABI has been changed so that:
11370 void operator delete[] (void *, size_t);
11371 void operator delete[] (void *);
11374 does not cause unnecessary storage to be allocated when an array of
11375 A objects is allocated.
11376 This change will only affect code that declares both of these forms
11377 of operator delete[], and declared the two-argument form before the
11379 * The C++ ABI has been changed so that when a parameter is passed by
11380 value, any cleanup for that parameter is performed in the caller,
11381 as specified by the ia64 C++ ABI, rather than the called function
11382 as before. As a result, classes with a non-trivial destructor but a
11383 trivial copy constructor will be passed and returned by invisible
11384 reference, rather than by bitwise copy as before.
11385 * G++ now supports the "named return value optimization": for code
11393 G++ will allocate a in the return value slot, so that the return
11394 becomes a no-op. For this to work, all return statements in the
11395 function must return the same variable.
11396 * Improvements to the C++ library are listed in [5]the libstdc++-v3
11401 * Annoying linker warnings (due to incorrect code being generated)
11403 * If a class method cannot be found, the compiler no longer issues a
11404 warning if a corresponding instance method exists in the root
11406 * Forward @protocol declarations have been fixed.
11407 * Loading of categories has been fixed in certain situations (GNU run
11409 * The class lookup in the run-time library has been rewritten so that
11410 class method dispatch is more than twice as fast as it used to be
11411 (GNU run time only).
11415 * libgcj now includes RMI, java.lang.ref.*, javax.naming, and
11417 * Property files and other system resources can be compiled into
11418 executables which use libgcj using the new gcj --resource feature.
11419 * libgcj has been ported to more platforms. In particular there is
11420 now a mostly-functional mingw32 (Windows) target port.
11421 * JNI and CNI invocation interfaces were implemented, so gcj-compiled
11422 Java code can now be called from a C/C++ application.
11423 * gcj can now use builtin functions for certain known methods, for
11425 * gcj can now automatically remove redundant array-store checks in
11427 * The --no-store-checks optimization option was added. This can be
11428 used to omit runtime store checks for code which is known not to
11429 throw ArrayStoreException
11430 * The following third party interface standards were added to libgcj:
11431 org.w3c.dom and org.xml.sax.
11432 * java.security has been merged with GNU Classpath. The new package
11433 is now JDK 1.2 compliant, and much more complete.
11434 * A bytecode verifier was added to the libgcj interpreter.
11435 * java.lang.Character was rewritten to comply with the Unicode 3.0
11436 standard, and improve performance.
11437 * Partial support for many more locales was added to libgcj.
11438 * Socket timeouts have been implemented.
11439 * libgcj has been merged into a single shared library. There are no
11440 longer separate shared libraries for the garbage collector and
11442 * Several performance improvements were made to gcj and libgcj:
11443 + Hash synchronization (thin locks)
11444 + A special allocation path for finalizer-free objects
11445 + Thread-local allocation
11446 + Parallel GC, and other GC tweaks
11450 Fortran improvements are listed in [6]the Fortran documentation.
11454 [7]Ada Core Technologies, Inc, has contributed its GNAT Ada 95 front
11455 end and associated tools. The GNAT compiler fully implements the Ada
11456 language as defined by the ISO/IEC 8652 standard.
11458 Please note that the integration of the Ada front end is still work in
11461 New Targets and Target Specific Improvements
11463 * Hans-Peter Nilsson has contributed a port to [8]MMIX, the CPU
11464 architecture used in new editions of Donald E. Knuth's The Art of
11465 Computer Programming.
11466 * [9]Axis Communications has contributed its port to the CRIS CPU
11467 architecture, used in the ETRAX system-on-a-chip series. See
11468 [10]Axis' developer site for technical information.
11469 * Alexandre Oliva, of Red Hat, has contributed a port to the
11470 [11]SuperH SH5 64-bit RISC microprocessor architecture, extending
11471 the existing SH port.
11472 * UltraSPARC is fully supported in 64-bit mode. The option -m64
11474 * For compatibility with the Sun compiler #pragma redefine_extname
11475 has been implemented on Solaris.
11476 * The x86 back end has had some noticeable work done to it.
11477 + SuSE Labs developers Jan Hubicka, Bo Thorsen and Andreas
11478 Jaeger have contributed a port to the AMD x86-64 architecture.
11479 For more information on x86-64 see [12]http://www.x86-64.org.
11480 + The compiler now supports MMX, 3DNow!, SSE, and SSE2
11481 instructions. Options -mmmx, -m3dnow, -msse, and -msse2 will
11482 enable the respective instruction sets. Intel C++ compatible
11483 MMX/3DNow!/SSE intrinsics are implemented. SSE2 intrinsics
11484 will be added in next major release.
11485 + Following those improvements, targets for Pentium MMX, K6-2,
11486 K6-3, Pentium III, Pentium 4, and Athlon 4 Mobile/XP/MP were
11487 added. Refer to the documentation on -march= and -mcpu=
11488 options for details.
11489 + For those targets that support it, -mfpmath=sse will cause the
11490 compiler to generate SSE/SSE2 instructions for floating point
11491 math instead of x87 instructions. Usually, this will lead to
11492 quicker code -- especially on the Pentium 4. Note that only
11493 scalar floating point instructions are used and GCC does not
11494 exploit SIMD features yet.
11495 + Prefetch support has been added to the Pentium III, Pentium 4,
11496 K6-2, K6-3, and Athlon series.
11497 + Code generated for floating point to integer conversions has
11498 been improved leading to better performance of many 3D
11500 * The PowerPC back end has added 64-bit PowerPC GNU/Linux support.
11501 * C++ support for AIX has been improved.
11502 * Aldy Hernandez, of Red Hat, Inc, has contributed extensions to the
11503 PowerPC port supporting the AltiVec programming model (SIMD). The
11504 support, though presently useful, is experimental and is expected
11505 to stabilize for 3.2. The support is written to conform to
11506 Motorola's AltiVec specs. See -maltivec.
11510 Support for a number of older systems has been declared obsolete in GCC
11511 3.1. Unless there is activity to revive them, the next release of GCC
11512 will have their sources permanently removed.
11514 All configurations of the following processor architectures have been
11516 * MIL-STD-1750A, 1750a-*-*
11517 * AMD A29k, a29k-*-*
11518 * Convex, c*-convex-*
11519 * Clipper, clipper-*-*
11521 * Intel i860, i860-*-*
11522 * Sun picoJava, pj-*-* and pjl-*-*
11523 * Western Electric 32000, we32k-*-*
11525 Most configurations of the following processor architectures have been
11526 declared obsolete, but we are preserving a few systems which may have
11527 active developers. It is unlikely that the remaining systems will
11528 survive much longer unless we see definite signs of port activity.
11529 * Motorola 88000 except
11530 + Generic a.out, m88k-*-aout*
11531 + Generic SVR4, m88k-*-sysv4
11532 + OpenBSD, m88k-*-openbsd*
11534 + NetBSD, ns32k-*-netbsd*
11535 + OpenBSD, ns32k-*-openbsd*.
11537 + OpenBSD, romp-*-openbsd*.
11539 Finally, only some configurations of these processor architectures are
11542 + OSF/1, alpha*-*-osf[123]*. (Digital Unix and Tru64 Unix, aka
11543 alpha*-*-osf[45], are still supported.)
11545 + RISCiX, arm-*-riscix*.
11547 + 386BSD, i?86-*-bsd*
11548 + Chorus, i?86-*-chorusos*
11549 + DG/UX, i?86-*-dgux*
11550 + FreeBSD 1.x, i?86-*-freebsd1.*
11551 + IBM AIX, i?86-*-aix*
11552 + ISC UNIX, i?86-*-isc*
11553 + GNU/Linux with pre-BFD linker, i?86-*-linux*oldld*
11554 + NEXTstep, i?86-next-*
11555 + OSF UNIX, i?86-*-osf1* and i?86-*-osfrose*
11556 + RTEMS/coff, i?86-*-rtemscoff*
11557 + RTEMS/go32, i?86-go32-rtems*
11558 + Sequent/BSD, i?86-sequent-bsd*
11559 + Sequent/ptx before version 3, i?86-sequent-ptx[12]* and
11560 i?86-sequent-sysv3*
11561 + SunOS, i?86-*-sunos*
11563 + Altos, m68[k0]*-altos-*
11564 + Apollo, m68[k0]*-apollo-*
11565 + Apple A/UX, m68[k0]*-apple-*
11566 + Bull, m68[k0]*-bull-*
11567 + Convergent, m68[k0]*-convergent-*
11568 + Generic SVR3, m68[k0]*-*-sysv3*
11569 + ISI, m68[k0]*-isi-*
11570 + LynxOS, m68[k0]*-*-lynxos*
11571 + NEXT, m68[k0]*-next-*
11572 + RTEMS/coff, m68[k0]*-*-rtemscoff*
11573 + Sony, m68[k0]*-sony-*
11575 + DEC Ultrix, mips-*-ultrix* and mips-dec-*
11576 + Generic BSD, mips-*-bsd*
11577 + Generic System V, mips-*-sysv*
11578 + IRIX before version 5, mips-sgi-irix[1234]*
11579 + RiscOS, mips-*-riscos*
11580 + Sony, mips-sony-*
11581 + Tandem, mips-tandem-*
11583 + RTEMS/a.out, sparc-*-rtemsaout*.
11585 Documentation improvements
11587 * The old manual ("Using and Porting the GNU Compiler Collection")
11588 has been replaced by a users manual ("Using the GNU Compiler
11589 Collection") and a separate internals reference manual ("GNU
11590 Compiler Collection Internals").
11591 * More complete and much improved documentation about GCC's internal
11592 representation used by the C and C++ front ends.
11593 * Many cleanups and improvements in general.
11596 For questions related to the use of GCC, please consult these web
11597 pages and the [13]GCC manuals. If that fails, the
11598 [14]gcc-help@gcc.gnu.org mailing list might help. Comments on these
11599 web pages and the development of GCC are welcome on our developer
11600 list at [15]gcc@gcc.gnu.org. All of [16]our lists have public
11603 Copyright (C) [17]Free Software Foundation, Inc. Verbatim copying and
11604 distribution of this entire article is permitted in any medium,
11605 provided this notice is preserved.
11607 These pages are [18]maintained by the GCC team. Last modified
11612 1. http://gcc.gnu.org/ml/gcc/2002-07/msg01208.html
11613 2. http://gcc.gnu.org/news/profiledriven.html
11614 3. http://gcc.gnu.org/benchmarks/
11615 4. http://gcc.gnu.org/gcc-3.1/c99status.html
11616 5. http://gcc.gnu.org/onlinedocs/libstdc++/faq.html
11617 6. http://gcc.gnu.org/onlinedocs/gcc-3.1.1/g77/News.html
11618 7. http://www.adacore.com/
11619 8. http://www-cs-faculty.stanford.edu/~knuth/mmix.html
11620 9. http://www.axis.com/
11621 10. http://developer.axis.com/
11622 11. http://www.superh.com/
11623 12. http://www.x86-64.org/
11624 13. http://gcc.gnu.org/onlinedocs/
11625 14. mailto:gcc-help@gcc.gnu.org
11626 15. mailto:gcc@gcc.gnu.org
11627 16. http://gcc.gnu.org/lists.html
11628 17. http://www.fsf.org/
11629 18. http://gcc.gnu.org/about.html
11630 19. http://validator.w3.org/check/referer
11631 ======================================================================
11632 http://gcc.gnu.org/gcc-3.0/gcc-3.0.html
11637 The [1]GNU project and the GCC developers are pleased to announce the
11638 release of GCC 3.0.4, which is a bug-fix release for the GCC 3.0
11641 GCC used to stand for the GNU C Compiler, but since the compiler
11642 supports several other languages aside from C, it now stands for the
11643 GNU Compiler Collection.
11645 GCC 3.0.x has several new optimizations, new targets, new languages and
11646 many other new features, relative to GCC 2.95.x. See the [2]new
11647 features page for a more complete list.
11649 A list of [3]successful builds is updated as new information becomes
11652 The GCC developers would like to thank the numerous people that have
11653 contributed new features, test results, bug fixes, etc to GCC. This
11654 [4]amazing group of volunteers is what makes GCC successful.
11656 And finally, we can't in good conscience fail to mention some
11657 [5]caveats to using GCC 3.0.x.
11659 For additional information about GCC please refer to the [6]GCC project
11660 web site or contact the [7]GCC development mailing list.
11662 To obtain GCC please use [8]our mirror sites, or our CVS server.
11663 __________________________________________________________________
11665 Previous 3.0.x Releases
11667 December 20, 2001: GCC 3.0.3 has been released.
11668 October 25, 2001: GCC 3.0.2 has been released.
11669 August 20, 2001: GCC 3.0.1 has been released.
11670 June 18, 2001: GCC 3.0 has been released.
11673 For questions related to the use of GCC, please consult these web
11674 pages and the [9]GCC manuals. If that fails, the
11675 [10]gcc-help@gcc.gnu.org mailing list might help. Comments on these
11676 web pages and the development of GCC are welcome on our developer
11677 list at [11]gcc@gcc.gnu.org. All of [12]our lists have public
11680 Copyright (C) [13]Free Software Foundation, Inc. Verbatim copying and
11681 distribution of this entire article is permitted in any medium,
11682 provided this notice is preserved.
11684 These pages are [14]maintained by the GCC team. Last modified
11689 1. http://www.gnu.org/
11690 2. http://gcc.gnu.org/gcc-3.0/features.html
11691 3. http://gcc.gnu.org/gcc-3.0/buildstat.html
11692 4. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
11693 5. http://gcc.gnu.org/gcc-3.0/caveats.html
11694 6. http://gcc.gnu.org/index.html
11695 7. mailto:gcc@gcc.gnu.org
11696 8. http://gcc.gnu.org/mirrors.html
11697 9. http://gcc.gnu.org/onlinedocs/
11698 10. mailto:gcc-help@gcc.gnu.org
11699 11. mailto:gcc@gcc.gnu.org
11700 12. http://gcc.gnu.org/lists.html
11701 13. http://www.fsf.org/
11702 14. http://gcc.gnu.org/about.html
11703 15. http://validator.w3.org/check/referer
11704 ======================================================================
11705 http://gcc.gnu.org/gcc-3.0/features.html
11706 GCC 3.0 New Features
11708 Additional changes in GCC 3.0.4
11710 * GCC 3.0 now supports newer versions of the [1]NetBSD operating
11711 system, which use the ELF object file format, on x86 processors.
11712 * Correct debugging information is generated from functions that have
11713 lines from multiple files (e.g. yacc output).
11714 * A fix for whitespace handling in the -traditional preprocessor,
11715 which can affect Fortran.
11716 * Fixes to the exception handling runtime.
11717 * More fixes for bad code generation in C++.
11718 * A fix for shared library generation under AIX 4.3.
11719 * Documentation updates.
11720 * Port of GCC to Tensilica's Xtensa processor contributed.
11721 * A fix for compiling the PPC Linux kernel (FAT fs wouldn't link).
11723 Additional changes in GCC 3.0.3
11725 * A fix to correct an accidental change to the PowerPC ABI.
11726 * Fixes for bad code generation on a variety of architectures.
11727 * Improvements to the debugging information generated for C++
11729 * Fixes for bad code generation in C++.
11730 * A fix to avoid crashes in the C++ demangler.
11731 * A fix to the C++ standard library to avoid buffer overflows.
11732 * Miscellaneous improvements for a variety of architectures.
11734 Additional changes in GCC 3.0.2
11736 * Fixes for bad code generation during loop unrolling.
11737 * Fixes for bad code generation by the sibling call optimization.
11738 * Minor improvements to x86 code generation.
11739 * Implementation of function descriptors in C++ vtables for IA64.
11740 * Numerous minor bug-fixes.
11742 Additional changes in GCC 3.0.1
11744 * C++ fixes for incorrect code-generation.
11745 * Improved cross-compiling support for the C++ standard library.
11746 * Fixes for some embedded targets that worked in GCC 2.95.3, but not
11748 * Fixes for various exception-handling bugs.
11749 * A port to the S/390 architecture.
11751 General Optimizer Improvements
11753 * [2]Basic block reordering pass.
11754 * New if-conversion pass with support for conditional (predicated)
11756 * New tail call and sibling call elimination optimizations.
11757 * New register renaming pass.
11758 * New (experimental) [3]static single assignment (SSA) representation
11760 * New dead-code elimination pass implemented using the SSA
11762 * [4]Global null pointer test elimination.
11763 * [5]Global code hoisting/unification.
11764 * More builtins and optimizations for stdio.h, string.h and old BSD
11765 functions, as well as for ISO C99 functions.
11766 * New builtin __builtin_expect for giving hints to the branch
11769 New Languages and Language specific improvements
11771 * The GNU Compiler for the Java(TM) language (GCJ) is now integrated
11772 and supported, including the run-time library containing most
11773 common non-GUI Java classes, a bytecode interpreter, and the Boehm
11774 conservative garbage collector. Many bugs have been fixed. GCJ can
11775 compile Java source or Java bytecodes to either native code or Java
11776 class files, and supports native methods written in either the
11777 standard JNI or the more efficient and convenient CNI.
11778 * Here is a [6]partial list of C++ improvements, both new features
11779 and those no longer supported.
11780 * New C++ ABI. On the IA-64 platform GCC is capable of
11781 inter-operating with other IA-64 compilers.
11782 * The new ABI also significantly reduces the size of symbol and debug
11784 * New [7]C++ support library and many C++ bug fixes, vastly improving
11785 our conformance to the ISO C++ standard.
11786 * New [8]inliner for C++.
11787 * Rewritten C preprocessor, integrated into the C, C++ and Objective
11788 C compilers, with very many improvements including ISO C99 support
11789 and [9]improvements to dependency generation.
11790 * Support for more [10]ISO C99 features.
11791 * Many improvements to support for checking calls to format functions
11792 such as printf and scanf, including support for ISO C99 format
11793 features, extensions from the Single Unix Specification and GNU
11794 libc 2.2, checking of strfmon formats and features to assist in
11795 auditing for format string security bugs.
11796 * New warnings for C code that may have undefined semantics because
11797 of violations of sequence point rules in the C standard (such as a
11798 = a++;, a[n] = b[n++]; and a[i++] = i;), included in -Wall.
11799 * Additional warning option -Wfloat-equal.
11800 * Improvements to -Wtraditional.
11801 * Fortran improvements are listed in [11]the Fortran documentation.
11803 New Targets and Target Specific Improvements
11805 * New x86 back-end, generating much improved code.
11806 * Support for a generic i386-elf target contributed.
11807 * New option to emit x86 assembly code using Intel style syntax
11809 * HPUX 11 support contributed.
11810 * Improved PowerPC code generation, including scheduled prologue and
11812 * Port of GCC to Intel's IA-64 processor contributed.
11813 * Port of GCC to Motorola's MCore 210 and 340 contributed.
11814 * New unified back-end for Arm, Thumb and StrongArm contributed.
11815 * Port of GCC to Intel's XScale processor contributed.
11816 * Port of GCC to Atmel's AVR microcontrollers contributed.
11817 * Port of GCC to Mitsubishi's D30V processor contributed.
11818 * Port of GCC to Matsushita's AM33 processor (a member of the MN10300
11819 processor family) contributed.
11820 * Port of GCC to Fujitsu's FR30 processor contributed.
11821 * Port of GCC to Motorola's 68HC11 and 68HC12 processors contributed.
11822 * Port of GCC to Sun's picoJava processor core contributed.
11824 Documentation improvements
11826 * Substantially rewritten and improved C preprocessor manual.
11827 * Many improvements to other documentation.
11828 * Manpages for gcc, cpp and gcov are now generated automatically from
11829 the master Texinfo manual, eliminating the problem of manpages
11830 being out of date. (The generated manpages are only extracts from
11831 the full manual, which is provided in Texinfo form, from which
11832 info, HTML, other formats and a printed manual can be generated.)
11833 * Generated info files are included in the release tarballs alongside
11834 their Texinfo sources, avoiding problems on some platforms with
11835 building makeinfo as part of the GCC distribution.
11837 Other significant improvements
11839 * Garbage collection used internally by the compiler for most memory
11840 allocation instead of obstacks.
11841 * Lengauer and Tarjan algorithm used for computing dominators in the
11842 CFG. This algorithm can be significantly faster and more space
11843 efficient than our older algorithm.
11844 * gccbug script provided to assist in submitting bug reports to our
11845 bug tracking system. (Bug reports previously submitted directly to
11846 our mailing lists, for which you received no bug tracking number,
11847 should be submitted again using gccbug if you can reproduce the
11848 problem with GCC 3.0.)
11849 * The internal libgcc library is [12]built as a shared library on
11850 systems that support it.
11851 * Extensive testsuite included with GCC, with many new tests. In
11852 addition to tests for GCC bugs that have been fixed, many tests
11853 have been added for language features, compiler warnings and
11855 * Additional language-independent warning options -Wpacked, -Wpadded,
11856 -Wunreachable-code and -Wdisabled-optimization.
11857 * Target-independent options -falign-functions, -falign-loops and
11860 Plus a great many bug fixes and almost all the [13]features found in
11864 For questions related to the use of GCC, please consult these web
11865 pages and the [14]GCC manuals. If that fails, the
11866 [15]gcc-help@gcc.gnu.org mailing list might help. Comments on these
11867 web pages and the development of GCC are welcome on our developer
11868 list at [16]gcc@gcc.gnu.org. All of [17]our lists have public
11871 Copyright (C) [18]Free Software Foundation, Inc. Verbatim copying and
11872 distribution of this entire article is permitted in any medium,
11873 provided this notice is preserved.
11875 These pages are [19]maintained by the GCC team. Last modified
11880 1. http://www.netbsd.org/
11881 2. http://gcc.gnu.org/news/reorder.html
11882 3. http://gcc.gnu.org/news/ssa.html
11883 4. http://gcc.gnu.org/news/null.html
11884 5. http://gcc.gnu.org/news/unify.html
11885 6. http://gcc.gnu.org/gcc-3.0/c++features.html
11886 7. http://gcc.gnu.org/libstdc++/
11887 8. http://gcc.gnu.org/news/inlining.html
11888 9. http://gcc.gnu.org/news/dependencies.html
11889 10. http://gcc.gnu.org/gcc-3.0/c99status.html
11890 11. http://gcc.gnu.org/onlinedocs/gcc-3.4.6/g77/News.html
11891 12. http://gcc.gnu.org/gcc-3.0/libgcc.html
11892 13. http://gcc.gnu.org/gcc-2.95/features.html
11893 14. http://gcc.gnu.org/onlinedocs/
11894 15. mailto:gcc-help@gcc.gnu.org
11895 16. mailto:gcc@gcc.gnu.org
11896 17. http://gcc.gnu.org/lists.html
11897 18. http://www.fsf.org/
11898 19. http://gcc.gnu.org/about.html
11899 20. http://validator.w3.org/check/referer
11900 ======================================================================
11901 http://gcc.gnu.org/gcc-3.0/caveats.html
11904 * -fstrict-aliasing is now part of -O2 and higher optimization
11905 levels. This allows the compiler to assume the strictest aliasing
11906 rules applicable to the language being compiled. For C and C++,
11907 this activates optimizations based on the type of expressions. This
11908 optimization may thus break old, non-compliant code.
11909 * Enumerations are now properly promoted to int in function
11910 parameters and function returns. Normally this change is not
11911 visible, but when using -fshort-enums this is an ABI change.
11912 * The undocumented extension that allowed C programs to have a label
11913 at the end of a compound statement has been deprecated and may be
11914 removed in a future version. Programs that now generate a warning
11915 about this may be fixed by adding a null statement (a single
11916 semicolon) after the label.
11917 * The poorly documented extension that allowed string constants in C,
11918 C++ and Objective C to contain unescaped newlines has been
11919 deprecated and may be removed in a future version. Programs using
11920 this extension may be fixed in several ways: the bare newline may
11921 be replaced by \n, or preceded by \n\, or string concatenation may
11922 be used with the bare newline preceded by \n" and " placed at the
11923 start of the next line.
11924 * The Chill compiler is not included in GCC 3.0, because of the lack
11925 of a volunteer to convert it to use garbage collection.
11926 * Certain non-standard iostream methods from earlier versions of
11927 libstdc++ are not included in libstdc++ v3, i.e. filebuf::attach,
11928 ostream::form, and istream::gets.
11929 * The new C++ ABI is not yet fully supported by current (as of
11930 2001-07-01) releases and development versions of GDB, or any
11931 earlier versions. There is a problem setting breakpoints by line
11932 number, and other related issues that have been fixed in GCC 3.0
11933 but not yet handled in GDB:
11934 [1]http://gcc.gnu.org/ml/gcc-bugs/2001-06/msg00421.html
11937 For questions related to the use of GCC, please consult these web
11938 pages and the [2]GCC manuals. If that fails, the
11939 [3]gcc-help@gcc.gnu.org mailing list might help. Comments on these
11940 web pages and the development of GCC are welcome on our developer
11941 list at [4]gcc@gcc.gnu.org. All of [5]our lists have public archives.
11943 Copyright (C) [6]Free Software Foundation, Inc. Verbatim copying and
11944 distribution of this entire article is permitted in any medium,
11945 provided this notice is preserved.
11947 These pages are [7]maintained by the GCC team. Last modified
11952 1. http://gcc.gnu.org/ml/gcc-bugs/2001-06/msg00421.html
11953 2. http://gcc.gnu.org/onlinedocs/
11954 3. mailto:gcc-help@gcc.gnu.org
11955 4. mailto:gcc@gcc.gnu.org
11956 5. http://gcc.gnu.org/lists.html
11957 6. http://www.fsf.org/
11958 7. http://gcc.gnu.org/about.html
11959 8. http://validator.w3.org/check/referer
11960 ======================================================================
11961 http://gcc.gnu.org/gcc-2.95/index.html
11964 March 16, 2001: The GNU project and the GCC developers are pleased to
11965 announce the release of GCC version 2.95.3.
11979 July 31, 1999. This is the first release of GCC since the April
11980 1999 GCC/EGCS reunification and includes nearly a year's worth
11981 of new development and bugfixes.
11983 References and Acknowledgements
11985 GCC used to stand for the GNU C Compiler, but since the compiler
11986 supports several other languages aside from C, it now stands for the
11987 GNU Compiler Collection.
11989 The whole suite has been extensively [1]regression tested and
11990 [2]package tested. It should be reliable and suitable for widespread
11993 The compiler has several new optimizations, new targets, new languages
11994 and other new features. See the [3]new features page for a more
11995 complete list of new features found in the GCC 2.95 releases.
11997 The sources include installation instructions in both HTML and
11998 plaintext forms in the install directory in the distribution. However,
11999 the most up to date [4]installation instructions and [5]build/test
12000 status are on the web pages. We will update those pages as new
12001 information becomes available.
12003 The GCC developers would like to thank the numerous people that have
12004 contributed new features, test results, bugfixes, etc to GCC. This
12005 [6]amazing group of volunteers is what makes GCC successful.
12007 And finally, we can't in good conscience fail to mention some
12008 [7]caveats to using GCC 2.95.
12010 Download GCC 2.95 from one of our many [8]mirror sites.
12012 For additional information about GCC please see the [9]GCC project web
12013 server or contact the [10]GCC development mailing list.
12016 For questions related to the use of GCC, please consult these web
12017 pages and the [11]GCC manuals. If that fails, the
12018 [12]gcc-help@gcc.gnu.org mailing list might help. Comments on these
12019 web pages and the development of GCC are welcome on our developer
12020 list at [13]gcc@gcc.gnu.org. All of [14]our lists have public
12023 Copyright (C) [15]Free Software Foundation, Inc. Verbatim copying and
12024 distribution of this entire article is permitted in any medium,
12025 provided this notice is preserved.
12027 These pages are [16]maintained by the GCC team. Last modified
12032 1. http://gcc.gnu.org/gcc-2.95/regress.html
12033 2. http://gcc.gnu.org/gcc-2.95/othertest.html
12034 3. http://gcc.gnu.org/gcc-2.95/features.html
12035 4. http://gcc.gnu.org/install/
12036 5. http://gcc.gnu.org/gcc-2.95/buildstat.html
12037 6. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
12038 7. http://gcc.gnu.org/gcc-2.95/caveats.html
12039 8. http://gcc.gnu.org/mirrors.html
12040 9. http://gcc.gnu.org/index.html
12041 10. mailto:gcc@gcc.gnu.org
12042 11. http://gcc.gnu.org/onlinedocs/
12043 12. mailto:gcc-help@gcc.gnu.org
12044 13. mailto:gcc@gcc.gnu.org
12045 14. http://gcc.gnu.org/lists.html
12046 15. http://www.fsf.org/
12047 16. http://gcc.gnu.org/about.html
12048 17. http://validator.w3.org/check/referer
12049 ======================================================================
12050 http://gcc.gnu.org/gcc-2.95/features.html
12051 GCC 2.95 New Features
12053 * General Optimizer Improvements:
12054 + [1]Localized register spilling to improve speed and code
12055 density especially on small register class machines.
12056 + [2]Global CSE using lazy code motion algorithms.
12057 + [3]Improved global constant/copy propagation.
12058 + [4]Improved control flow graph analysis and manipulation.
12059 + [5]Local dead store elimination.
12060 + [6]Memory Load hoisting/store sinking in loops.
12061 + [7]Type based alias analysis is enabled by default. Note this
12062 feature will expose bugs in the Linux kernel. Please refer to
12063 the FAQ (as shipped with GCC 2.95) for additional information
12065 + Major revamp of GIV detection, combination and simplification
12066 to improve loop performance.
12067 + Major improvements to register allocation and reloading.
12068 * New Languages and Language specific improvements
12069 + [8]Many C++ improvements.
12070 + [9]Many Fortran improvements.
12071 + [10]Java front-end has been integrated. [11]runtime library is
12072 available separately.
12073 + [12]ISO C99 support
12074 + [13]Chill front-end and runtime has been integrated.
12075 + Boehm garbage collector support in libobjc.
12076 + More support for various pragmas which appear in vendor
12078 * New Targets and Target Specific Improvements
12079 + [14]SPARC backend rewrite.
12080 + -mschedule=8000 will optimize code for PA8000 class
12081 processors; -mpa-risc-2-0 will generate code for PA2.0
12083 + Various micro-optimizations for the ia32 port. K6
12085 + Compiler will attempt to align doubles in the stack on the
12087 + Alpha EV6 support
12089 + RS6000/PowerPC: -mcpu=401 was added as an alias for -mcpu=403.
12090 -mcpu=e603e was added to do -mcpu=603e and -msoft-float.
12095 + Support for new systems (OpenBSD, FreeBSD, UWIN, Interix,
12097 + vxWorks targets include support for vxWorks threads
12098 + StrongARM 110 and ARM9 support added. ARM Scheduling
12099 parameters rewritten.
12100 + Various changes to the MIPS port to avoid assembler macros,
12101 which in turn improves performance
12102 + Various performance improvements to the i960 port.
12103 + Major rewrite of ns32k port
12104 * Other significant improvements
12105 + [15]Ability to dump cfg information and display it using vcg.
12106 + The new faster scheme for fixing vendor header files is
12107 enabled by default.
12108 + Experimental internationalization support.
12109 + multibyte character support
12110 + Some compile-time speedups for pathological problems
12111 + Better support for complex types
12112 * Plus the usual mountain of bugfixes
12113 * Core compiler is based on the gcc2 development tree from Sept 30,
12114 1998, so we have all of the [16]features found in GCC 2.8.
12116 Additional Changes in GCC 2.95.1
12118 * Generic bugfixes and improvements
12119 + Various documentation fixes related to the GCC/EGCS merger.
12120 + Fix memory management bug which could lead to spurious aborts,
12121 core dumps or random parsing errors in the compiler.
12122 + Fix a couple bugs in the dwarf1 and dwarf2 debug record
12124 + Fix infinite loop in the CSE optimizer.
12125 + Avoid undefined behavior in compiler FP emulation code
12126 + Fix install problem when prefix is overridden on the make
12128 + Fix problem with unwanted installation of assert.h on some
12130 + Fix problem with finding the wrong assembler in a single tree
12132 + Avoid increasing the known alignment of a register that is
12133 already known to be a pointer.
12134 * Platform specific bugfixes and improvements
12135 + Codegen bugfix for prologue/epilogue for cpu32 target.
12136 + Fix long long code generation bug for the Coldfire target.
12137 + Fix various aborts in the SH compiler.
12138 + Fix bugs in libgcc support library for the SH.
12139 + Fix alpha ev6 code generation bug.
12140 + Fix problems with EXIT_SUCCESS/EXIT_FAILURE redefinitions on
12142 + Fix -fpic code generation bug for rs6000/ppc svr4 targets.
12143 + Fix varargs/stdarg code generation bug for rs6000/ppc svr4
12145 + Fix weak symbol handling for rs6000/ppc svr4 targets.
12146 + Fix various problems with 64bit code generation for the
12148 + Fix codegen bug which caused tetex to be mis-compiled on the
12150 + Fix compiler abort in new cfg code exposed by x86 port.
12151 + Fix out of range array reference in code convert flat
12152 registers to the x87 stacked FP register file.
12153 + Fix minor vxworks configuration bug.
12154 + Fix return type of bsearch for SunOS 4.x.
12155 * Language & Runtime specific fixes.
12156 + The G++ signature extension has been deprecated. It will be
12157 removed in the next major release of G++. Use of signatures
12158 will result in a warning from the compiler.
12159 + Several bugs relating to templates and namespaces were fixed.
12160 + A bug that caused crashes when combining templates with -g on
12161 DWARF1 platforms was fixed.
12162 + Pointers-to-members, virtual functions, and multiple
12163 inheritance should now work together correctly.
12164 + Some code-generation bugs relating to function try blocks were
12166 + G++ is a little bit more lenient with certain archaic
12167 constructs than in GCC 2.95.
12168 + Fix to prevent shared library version #s from bring truncated
12170 + Fix missing std:: in the libstdc++ library.
12171 + Fix stream locking problems in libio.
12172 + Fix problem in java compiler driver.
12174 Additional Changes in GCC 2.95.2
12176 The -fstrict-aliasing is not enabled by default for GCC 2.95.2. While
12177 the optimizations performed by -fstrict-aliasing are valid according to
12178 the C and C++ standards, the optimization have caused some problems,
12179 particularly with old non-conforming code.
12181 The GCC developers are experimenting with ways to warn users about code
12182 which violates the C/C++ standards, but those warnings are not ready
12183 for widespread use at this time. Rather than wait for those warnings
12184 the GCC developers have chosen to disable -fstrict-aliasing by default
12185 for the GCC 2.95.2 release.
12187 We strongly encourage developers to find and fix code which violates
12188 the C/C++ standards as -fstrict-aliasing may be enabled by default in
12189 future releases. Use the option -fstrict-aliasing to re-enable these
12191 * Generic bugfixes and improvements
12192 + Fix incorrectly optimized memory reference in global common
12193 subexpression elimination (GCSE) optimization pass.
12194 + Fix code generation bug in regmove.c in which it could
12195 incorrectly change a "const" value.
12196 + Fix bug in optimization of conditionals involving volatile
12198 + Avoid over-allocation of stack space for some procedures.
12199 + Fixed bug in the compiler which caused incorrect optimization
12200 of an obscure series of bit manipulations, shifts and
12202 + Fixed register allocator bug which caused teTeX to be
12203 mis-compiled on SPARC targets.
12204 + Avoid incorrect optimization of degenerate case statements for
12205 certain targets such as the ARM.
12206 + Fix out of range memory reference in the jump optimizer.
12207 + Avoid dereferencing null pointer in fix-header.
12208 + Fix test for GCC specific features so that it is possible to
12209 bootstrap with gcc-2.6.2 and older versions of GCC.
12210 + Fix typo in scheduler which could potentially cause out of
12211 range memory accesses.
12212 + Avoid incorrect loop reversal which caused incorrect code for
12213 certain loops on PowerPC targets.
12214 + Avoid incorrect optimization of switch statements on certain
12215 targets (for example the ARM).
12216 * Platform specific bugfixes and improvements
12217 + Work around bug in Sun V5.0 compilers which caused bootstrap
12218 comparison failures on SPARC targets.
12219 + Fix SPARC backend bug which caused aborts in final.c.
12220 + Fix sparc-hal-solaris2* configuration fragments.
12221 + Fix bug in sparc block profiling.
12222 + Fix obscure code generation bug for the PARISC targets.
12223 + Define __STDC_EXT__ for HPUX configurations.
12224 + Various POWERPC64 code generation bugfixes.
12225 + Fix abort for PPC targets using ELF (ex GNU/Linux).
12226 + Fix collect2 problems for AIX targets.
12227 + Correct handling of .file directive for PPC targets.
12228 + Fix bug in fix_trunc x86 patterns.
12229 + Fix x86 port to correctly pop the FP stack for functions that
12230 return structures in memory.
12231 + Fix minor bug in strlen x86 pattern.
12232 + Use stabs debugging instead of dwarf1 for x86-solaris targets.
12233 + Fix template repository code to handle leading underscore in
12235 + Fix weak/weak alias support for OpenBSD.
12236 + GNU/Linux for the ARM has C++ compatible include files.
12237 * Language & Runtime specific fixes.
12238 + Fix handling of constructor attribute in the C front-end which
12239 caused problems building the Chill runtime library on some
12241 + Fix minor problem merging type qualifiers in the C front-end.
12242 + Fix aliasing bug for pointers and references (C/C++).
12243 + Fix incorrect "non-constant initializer bug" when -traditional
12244 or -fwritable-strings is enabled.
12245 + Fix build error for Chill front-end on SunOS.
12246 + Do not complain about duplicate instantiations when using
12248 + Fix array bounds handling in C++ front-end which caused
12249 problems with dwarf debugging information in some
12251 + Fix minor namespace problem.
12252 + Fix problem linking java programs.
12254 Additional Changes in GCC 2.95.3
12256 * Generic bugfixes and improvements
12257 + Fix numerous problems that caused incorrect optimization in
12258 the register reloading code.
12259 + Fix numerous problems that caused incorrect optimization in
12260 the loop optimizer.
12261 + Fix aborts in the functions build_insn_chain and scan_loops
12262 under some circumstances.
12263 + Fix an alias analysis bug.
12264 + Fix an infinite compilation bug in the combiner.
12265 + A few problems with complex number support have been fixed.
12266 + It is no longer possible for gcc to act as a fork bomb when
12267 installed incorrectly.
12268 + The -fpack-struct option should be recognized now.
12269 + Fixed a bug that caused incorrect code to be generated due to
12270 a lost stack adjustment.
12271 * Platform specific bugfixes and improvements
12272 + Support building ARM toolchains hosted on Windows.
12273 + Fix attribute calculations in ARM toolchains.
12274 + arm-linux support has been improved.
12275 + Fix a PIC failure on sparc targets.
12276 + On ix86 targets, the regparm attribute should now work
12278 + Several updates for the h8300 port.
12279 + Fix problem building libio with glibc 2.2.
12282 For questions related to the use of GCC, please consult these web
12283 pages and the [17]GCC manuals. If that fails, the
12284 [18]gcc-help@gcc.gnu.org mailing list might help. Comments on these
12285 web pages and the development of GCC are welcome on our developer
12286 list at [19]gcc@gcc.gnu.org. All of [20]our lists have public
12289 Copyright (C) [21]Free Software Foundation, Inc. Verbatim copying and
12290 distribution of this entire article is permitted in any medium,
12291 provided this notice is preserved.
12293 These pages are [22]maintained by the GCC team. Last modified
12298 1. http://gcc.gnu.org/news/spill.html
12299 2. http://gcc.gnu.org/news/lcm.html
12300 3. http://gcc.gnu.org/news/cprop.html
12301 4. http://gcc.gnu.org/news/cfg.html
12302 5. http://gcc.gnu.org/news/dse.html
12303 6. http://gcc.gnu.org/news/hoist.html
12304 7. http://gcc.gnu.org/news/alias.html
12305 8. http://gcc.gnu.org/gcc-2.95/c++features.html
12306 9. http://gcc.gnu.org/onlinedocs/gcc-3.4.6/g77/News.html
12307 10. http://gcc.gnu.org/java/gcj-announce.txt
12308 11. http://gcc.gnu.org/news/javaannounce.html
12309 12. http://gcc.gnu.org/c99status.html
12310 13. http://gcc.gnu.org/news/chill.html
12311 14. http://gcc.gnu.org/news/sparc.html
12312 15. http://gcc.gnu.org/news/egcs-vcg.html
12313 16. http://gcc.gnu.org/egcs-1.0/features-2.8.html
12314 17. http://gcc.gnu.org/onlinedocs/
12315 18. mailto:gcc-help@gcc.gnu.org
12316 19. mailto:gcc@gcc.gnu.org
12317 20. http://gcc.gnu.org/lists.html
12318 21. http://www.fsf.org/
12319 22. http://gcc.gnu.org/about.html
12320 23. http://validator.w3.org/check/referer
12321 ======================================================================
12322 http://gcc.gnu.org/gcc-2.95/caveats.html
12325 * GCC 2.95 will issue an error for invalid asm statements that had
12326 been silently accepted by earlier versions of the compiler. This is
12327 particularly noticeable when compiling older versions of the Linux
12328 kernel (2.0.xx). Please refer to the FAQ (as shipped with GCC 2.95)
12329 for more information on this issue.
12330 * GCC 2.95 implements type based alias analysis to disambiguate
12331 memory references. Some programs, particularly the Linux kernel
12332 violate ANSI/ISO aliasing rules and therefore may not operate
12333 correctly when compiled with GCC 2.95. Please refer to the FAQ (as
12334 shipped with GCC 2.95) for more information on this issue.
12335 * GCC 2.95 has a known bug in its handling of complex variables for
12336 64bit targets. Instead of silently generating incorrect code, GCC
12337 2.95 will issue a fatal error for situations it can not handle.
12338 This primarily affects the Fortran community as Fortran makes more
12339 use of complex variables than C or C++.
12340 * GCC 2.95 has an integrated libstdc++, but does not have an
12341 integrated libg++. Furthermore old libg++ releases will not work
12342 with GCC 2.95. You can retrieve a recent copy of libg++ from the
12344 Note most C++ programs only need libstdc++.
12345 * Exception handling may not work with shared libraries, particularly
12346 on alphas, hppas, rs6000/powerpc and mips based platforms.
12347 Exception handling is known to work on x86 GNU/Linux platforms with
12349 * In general, GCC 2.95 is more rigorous about rejecting invalid C++
12350 code or deprecated C++ constructs than G++ 2.7, G++ 2.8, EGCS 1.0,
12351 or EGCS 1.1. As a result it may be necessary to fix C++ code before
12352 it will compile with GCC 2.95.
12353 * G++ is also converting toward the ISO C++ standard; as a result
12354 code which was previously valid (and thus accepted by other
12355 compilers and older versions of g++) may no longer be accepted. The
12356 flag -fpermissive may allow some non-conforming code to compile
12358 * GCC 2.95 compiled C++ code is not binary compatible with EGCS
12359 1.1.x, EGCS 1.0.x or GCC 2.8.x.
12360 * GCC 2.95 does not have changes from the GCC 2.8 tree that were made
12361 between Sept 30, 1998 and April 30, 1999 (the official end of the
12362 GCC 2.8 project). Future GCC releases will include all the changes
12363 from the defunct GCC 2.8 sources.
12366 For questions related to the use of GCC, please consult these web
12367 pages and the [2]GCC manuals. If that fails, the
12368 [3]gcc-help@gcc.gnu.org mailing list might help. Comments on these
12369 web pages and the development of GCC are welcome on our developer
12370 list at [4]gcc@gcc.gnu.org. All of [5]our lists have public archives.
12372 Copyright (C) [6]Free Software Foundation, Inc. Verbatim copying and
12373 distribution of this entire article is permitted in any medium,
12374 provided this notice is preserved.
12376 These pages are [7]maintained by the GCC team. Last modified
12381 1. ftp://gcc.gnu.org/pub/gcc/infrastructure/libg++-2.8.1.3.tar.gz
12382 2. http://gcc.gnu.org/onlinedocs/
12383 3. mailto:gcc-help@gcc.gnu.org
12384 4. mailto:gcc@gcc.gnu.org
12385 5. http://gcc.gnu.org/lists.html
12386 6. http://www.fsf.org/
12387 7. http://gcc.gnu.org/about.html
12388 8. http://validator.w3.org/check/referer
12389 ======================================================================
12390 http://gcc.gnu.org/egcs-1.1/index.html
12393 September 3, 1998: We are pleased to announce the release of EGCS 1.1.
12394 December 1, 1998: We are pleased to announce the release of EGCS 1.1.1.
12395 March 15, 1999: We are pleased to announce the release of EGCS 1.1.2.
12397 EGCS is a free software project to further the development of the GNU
12398 compilers using an open development environment.
12400 EGCS 1.1 is a major new release of the EGCS compiler system. It has
12401 been [1]extensively tested and is believed to be stable and suitable
12402 for widespread use.
12404 EGCS 1.1 is based on an June 6, 1998 snapshot of the GCC 2.8
12405 development sources; it contains all of the new features found in GCC
12406 2.8.1 as well as all new development from GCC up to June 6, 1998.
12408 EGCS 1.1 also contains many improvements and features not found in GCC
12409 or in older versions of EGCS:
12410 * Global common subexpression elimination and global constant/copy
12411 propagation (aka [2]gcse)
12412 * Ongoing improvements to the [3]alias analysis support to allow for
12413 better optimizations throughout the compiler.
12414 * Vastly improved [4]C++ compiler and integrated C++ runtime
12416 * Fixes for the /tmp symlink race security problems.
12417 * New targets including mips16, arm-thumb and 64 bit PowerPC.
12418 * Improvements to GNU Fortran (g77) compiler and runtime library made
12419 since g77 version 0.5.23.
12421 See the [5]new features page for a more complete list of new features
12422 found in EGCS 1.1 releases.
12424 EGCS 1.1.1 is a minor update to fix several serious problems in EGCS
12426 * General improvements and fixes
12427 + Avoid some stack overflows when compiling large functions.
12428 + Avoid incorrect loop invariant code motions.
12429 + Fix some core dumps on Linux kernel code.
12430 + Bring back the imake -Di386 and friends fix from EGCS 1.0.2.
12431 + Fix code generation problem in gcse.
12432 + Various documentation related fixes.
12433 * g++/libstdc++ improvements and fixes
12434 + MT safe EH fix for setjmp/longjmp based exception handling.
12435 + Fix a few bad interactions between optimization and exception
12437 + Fixes for demangling of template names starting with "__".
12438 + Fix a bug that would fail to run destructors in some cases
12440 + Fix 'new' of classes with virtual bases.
12441 + Fix crash building Qt on the Alpha.
12442 + Fix failure compiling WIFEXITED macro on GNU/Linux.
12443 + Fix some -frepo failures.
12444 * g77 and libf2c improvements and fixes
12445 + Various documentation fixes.
12446 + Avoid compiler crash on RAND intrinsic.
12447 + Fix minor bugs in makefiles exposed by BSD make programs.
12448 + Define _XOPEN_SOURCE for libI77 build to avoid potential
12449 problems on some 64-bit systems.
12450 + Fix problem with implicit endfile on rewind.
12451 + Fix spurious recursive I/O errors.
12452 * platform specific improvements and fixes
12453 + Match all versions of UnixWare7.
12454 + Do not assume x86 SVR4 or UnixWare targets can handle stabs.
12455 + Fix PPC/RS6000 LEGITIMIZE_ADDRESS macro and bug in conversion
12456 from unsigned ints to double precision floats.
12457 + Fix ARM ABI issue with NetBSD.
12458 + Fix a few arm code generation bugs.
12459 + Fixincludes will fix additional broken SCO OpenServer header
12461 + Fix a m68k backend bug which caused invalid offsets in reg+d
12463 + Fix problems with 64bit AIX 4.3 support.
12464 + Fix handling of long longs for varargs/stdarg functions on the
12466 + Minor fixes to CPP predefines for Windows.
12467 + Fix code generation problems with gpr<->fpr copies for 64bit
12469 + Fix a few coldfire code generation bugs.
12470 + Fix some more header file problems on SunOS 4.x.
12471 + Fix assert.h handling for RTEMS.
12472 + Fix Windows handling of TREE_SYMBOL_REFERENCED.
12473 + Fix x86 compiler abort in reg-stack pass.
12474 + Fix cygwin/windows problem with section attributes.
12475 + Fix Alpha code generation problem exposed by SMP Linux
12477 + Fix typo in m68k 32->64bit integer conversion.
12478 + Make sure target libraries build with -fPIC for PPC & Alpha
12481 EGCS 1.1.2 is a minor update to fix several serious problems in EGCS
12483 * General improvements and fixes
12484 + Fix bug in loop optimizer which caused the SPARC (and
12485 potentially other) ports to segfault.
12486 + Fix infinite recursion in alias analysis and combiner code.
12487 + Fix bug in regclass preferencing.
12488 + Fix incorrect loop reversal which caused incorrect code to be
12489 generated for several targets.
12490 + Fix return value for builtin memcpy.
12491 + Reduce compile time for certain loops which exposed quadratic
12492 behavior in the loop optimizer.
12493 + Fix bug which caused volatile memory to be written multiple
12494 times when only one write was needed/desired.
12495 + Fix compiler abort in caller-save.c
12496 + Fix combiner bug which caused incorrect code generation for
12497 certain division by constant operations.
12498 + Fix incorrect code generation due to a bug in range check
12500 + Fix incorrect code generation due to mis-handling of clobbered
12502 + Fix compiler abort/segfault due to incorrect register
12503 splitting when unrolling loops.
12504 + Fix code generation involving autoincremented addresses with
12506 + Work around bug in the scheduler which caused qt to be
12507 mis-compiled on some platforms.
12508 + Fix code generation problems with -fshort-enums.
12509 + Tighten security for temporary files.
12510 + Improve compile time for codes which make heavy use of
12511 overloaded functions.
12512 + Fix multiply defined constructor/destructor symbol problems.
12513 + Avoid setting bogus RPATH environment variable during
12515 + Avoid GNU-make dependencies in the texinfo subdir.
12516 + Install CPP wrapper script in $(prefix)/bin if --enable-cpp.
12517 --enable-cpp=<dirname> can be used to specify an additional
12518 install directory for the cpp wrapper script.
12519 + Fix CSE bug which caused incorrect label-label refs to appear
12521 + Avoid linking in EH routines from libgcc if they are not
12523 + Avoid obscure bug in aliasing code.
12524 + Fix bug in weak symbol handling.
12525 * Platform-specific improvements and fixes
12526 + Fix detection of PPro/PII on Unixware 7.
12527 + Fix compiler segfault when building spec99 and other programs
12529 + Fix code-generation bugs for integer and floating point
12530 conditional move instructions on the PPro/PII.
12531 + Use fixincludes to fix byteorder problems on i?86-*-sysv.
12532 + Fix build failure for the arc port.
12533 + Fix floating point format configuration for i?86-gnu port.
12534 + Fix problems with hppa1.0-hp-hpux10.20 configuration when
12535 threads are enabled.
12536 + Fix coldfire code generation bugs.
12537 + Fix "unrecognized insn" problems for Alpha and PPC ports.
12538 + Fix h8/300 code generation problem with floating point values
12540 + Fix unrecognized insn problems for the m68k port.
12541 + Fix namespace-pollution problem for the x86 port.
12542 + Fix problems with old assembler on x86 NeXT systems.
12543 + Fix PIC code-generation problems for the SPARC port.
12544 + Fix minor bug with LONG_CALLS in PowerPC SVR4 support.
12545 + Fix minor ISO namespace violation in Alpha varargs/stdarg
12547 + Fix incorrect "braf" instruction usage for the SH port.
12548 + Fix minor bug in va-sh which prevented its use with -ansi.
12549 + Fix problems recognizing and supporting FreeBSD.
12550 + Handle OpenBSD systems correctly.
12551 + Minor fixincludes fix for Digital UNIX 4.0B.
12552 + Fix problems with ctors/dtors in SCO shared libraries.
12553 + Abort instead of generating incorrect code for PPro/PII
12554 floating point conditional moves.
12555 + Avoid multiply defined symbols on GNU/Linux systems using
12557 + Fix abort in alpha compiler.
12558 * Fortran-specific fixes
12559 + Fix the IDate intrinsic (VXT) (in libg2c) so the returned year
12560 is in the documented, non-Y2K-compliant range of 0-99, instead
12561 of being returned as 100 in the year 2000.
12562 + Fix the `Date_and_Time' intrinsic (in libg2c) to return the
12563 milliseconds value properly in Values(8).
12564 + Fix the `LStat' intrinsic (in libg2c) to return device-ID
12565 information properly in SArray(7).
12567 Each release includes installation instructions in both HTML and
12568 plaintext forms (see the INSTALL directory in the toplevel directory of
12569 the distribution). However, we also keep the most up to date
12570 [6]installation instructions and [7]build/test status on our web page.
12571 We will update those pages as new information becomes available.
12573 The EGCS project would like to thank the numerous people that have
12574 contributed new features, test results, bugfixes, etc. This [8]amazing
12575 group of volunteers is what makes EGCS successful.
12577 And finally, we can't in good conscience fail to mention some
12578 [9]caveats to using EGCS 1.1.
12580 Download EGCS from egcs.cygnus.com (USA California).
12582 The EGCS 1.1 release is also available on many mirror sites.
12583 [10]Goto mirror list to find a closer site.
12586 For questions related to the use of GCC, please consult these web
12587 pages and the [11]GCC manuals. If that fails, the
12588 [12]gcc-help@gcc.gnu.org mailing list might help. Comments on these
12589 web pages and the development of GCC are welcome on our developer
12590 list at [13]gcc@gcc.gnu.org. All of [14]our lists have public
12593 Copyright (C) [15]Free Software Foundation, Inc. Verbatim copying and
12594 distribution of this entire article is permitted in any medium,
12595 provided this notice is preserved.
12597 These pages are [16]maintained by the GCC team. Last modified
12602 1. http://gcc.gnu.org/egcs-1.1/egcs-1.1-test.html
12603 2. http://gcc.gnu.org/news/gcse.html
12604 3. http://gcc.gnu.org/news/alias.html
12605 4. http://gcc.gnu.org/egcs-1.1/c++features.html
12606 5. http://gcc.gnu.org/egcs-1.1/features.html
12607 6. http://gcc.gnu.org/install/
12608 7. http://gcc.gnu.org/egcs-1.1/buildstat.html
12609 8. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
12610 9. http://gcc.gnu.org/egcs-1.1/caveats.html
12611 10. http://gcc.gnu.org/mirrors.html
12612 11. http://gcc.gnu.org/onlinedocs/
12613 12. mailto:gcc-help@gcc.gnu.org
12614 13. mailto:gcc@gcc.gnu.org
12615 14. http://gcc.gnu.org/lists.html
12616 15. http://www.fsf.org/
12617 16. http://gcc.gnu.org/about.html
12618 17. http://validator.w3.org/check/referer
12619 ======================================================================
12620 http://gcc.gnu.org/egcs-1.1/features.html
12621 EGCS 1.1 new features
12623 * Integrated GNU Fortran (g77) compiler and runtime library with
12624 improvements, based on g77 version 0.5.23.
12625 * Vast improvements in the C++ compiler; so many they have [1]page of
12627 * Compiler implements [2]global common subexpression elimination and
12628 global copy/constant propagation.
12629 * More major improvements in the [3]alias analysis code.
12630 * More major improvements in the exception handling code to improve
12631 performance, lower static overhead and provide the infrastructure
12632 for future improvements.
12633 * The infamous /tmp symlink race security problems have been fixed.
12634 * The regmove optimization pass has been nearly completely rewritten
12635 to improve performance of generated code.
12636 * The compiler now recomputes register usage information before local
12637 register allocation. By providing more accurate information to the
12638 priority based allocator, we get better register allocation.
12639 * The register reloading phase of the compiler optimizes spill code
12640 much better than in previous releases.
12641 * Some bad interactions between the register allocator and
12642 instruction scheduler have been fixed, resulting in much better
12643 code for certain programs. Additionally, we have tuned the
12644 scheduler in various ways to improve performance of generated code
12645 for some architectures.
12646 * The compiler's branch shortening algorithms have been significantly
12647 improved to work better on targets which align jump targets.
12648 * The compiler now supports -Os to prefer optimizing for code space
12649 over optimizing for code speed.
12650 * The compiler will now totally eliminate library calls which compute
12651 constant values. This primarily helps targets with no integer
12652 div/mul support and targets without floating point support.
12653 * The compiler now supports an extensive "--help" option.
12654 * cpplib has been greatly improved and may be suitable for limited
12656 * Memory footprint for the compiler has been significantly reduced
12657 for some pathological cases.
12658 * The time to build EGCS has been improved for certain targets
12659 (particularly the alpha and mips platforms).
12660 * Many infrastructure improvements throughout the compiler, plus the
12661 usual mountain of bugfixes and minor improvements.
12662 * Target dependent improvements:
12663 + SPARC port now includes V8 plus and V9 support as well as
12664 performance tuning for Ultra class machines. The SPARC port
12665 now uses the Haifa scheduler.
12666 + Alpha port has been tuned for the EV6 processor and has an
12667 optimized expansion of memcpy/bzero. The Alpha port now uses
12668 the Haifa scheduler.
12669 + RS6000/PowerPC: support for the Power64 architecture and AIX
12670 4.3. The RS6000/PowerPC port now uses the Haifa scheduler.
12671 + x86: Alignment of static store data and jump targets is per
12672 Intel recommendations now. Various improvements throughout the
12673 x86 port to improve performance on Pentium processors
12674 (including improved epilogue sequences for Pentium chips and
12675 backend improvements which should help register allocation on
12676 all x86 variants. Conditional move support has been fixed and
12677 enabled for PPro processors. The x86 port also better supports
12678 64bit operations now. Unixware 7, a System V Release 5 target,
12679 is now supported and SCO OpenServer targets can support GAS.
12680 + MIPS has improved multiply/multiply-add support and now
12681 includes mips16 ISA support.
12682 + M68k has many micro-optimizations and Coldfire fixes.
12683 * Core compiler is based on the GCC development tree from June 9,
12684 1998, so we have all of the [4]features found in GCC 2.8.
12687 For questions related to the use of GCC, please consult these web
12688 pages and the [5]GCC manuals. If that fails, the
12689 [6]gcc-help@gcc.gnu.org mailing list might help. Comments on these
12690 web pages and the development of GCC are welcome on our developer
12691 list at [7]gcc@gcc.gnu.org. All of [8]our lists have public archives.
12693 Copyright (C) [9]Free Software Foundation, Inc. Verbatim copying and
12694 distribution of this entire article is permitted in any medium,
12695 provided this notice is preserved.
12697 These pages are [10]maintained by the GCC team. Last modified
12702 1. http://gcc.gnu.org/egcs-1.1/c++features.html
12703 2. http://gcc.gnu.org/news/gcse.html
12704 3. http://gcc.gnu.org/news/alias.html
12705 4. http://gcc.gnu.org/egcs-1.0/features-2.8.html
12706 5. http://gcc.gnu.org/onlinedocs/
12707 6. mailto:gcc-help@gcc.gnu.org
12708 7. mailto:gcc@gcc.gnu.org
12709 8. http://gcc.gnu.org/lists.html
12710 9. http://www.fsf.org/
12711 10. http://gcc.gnu.org/about.html
12712 11. http://validator.w3.org/check/referer
12713 ======================================================================
12714 http://gcc.gnu.org/egcs-1.1/caveats.html
12717 * EGCS has an integrated libstdc++, but does not have an integrated
12718 libg++. Furthermore old libg++ releases will not work with EGCS; HJ
12719 Lu has made a libg++-2.8.1.2 snapshot available which may work with
12721 Note most C++ programs only need libstdc++.
12722 * Exception handling may not work with shared libraries, particularly
12723 on alphas, hppas, rs6000/powerpc and mips based platforms.
12724 Exception handling is known to work on x86-linux platforms with
12726 * Some versions of the Linux kernel have bugs which prevent them from
12727 being compiled or from running when compiled by EGCS. See the FAQ
12728 (as shipped with EGCS 1.1) for additional information.
12729 * In general, EGCS is more rigorous about rejecting invalid C++ code
12730 or deprecated C++ constructs than g++-2.7, g++-2.8 or EGCS 1.0. As
12731 a result it may be necessary to fix C++ code before it will compile
12733 * G++ is also converting toward the ISO C++ standard; as a result
12734 code which was previously valid (and thus accepted by other
12735 compilers and older versions of g++) may no longer be accepted.
12736 * EGCS 1.1 compiled C++ code is not binary compatible with EGCS 1.0.x
12737 or GCC 2.8.x due to changes necessary to support thread safe
12738 exception handling.
12741 For questions related to the use of GCC, please consult these web
12742 pages and the [1]GCC manuals. If that fails, the
12743 [2]gcc-help@gcc.gnu.org mailing list might help. Comments on these
12744 web pages and the development of GCC are welcome on our developer
12745 list at [3]gcc@gcc.gnu.org. All of [4]our lists have public archives.
12747 Copyright (C) [5]Free Software Foundation, Inc. Verbatim copying and
12748 distribution of this entire article is permitted in any medium,
12749 provided this notice is preserved.
12751 These pages are [6]maintained by the GCC team. Last modified
12756 1. http://gcc.gnu.org/onlinedocs/
12757 2. mailto:gcc-help@gcc.gnu.org
12758 3. mailto:gcc@gcc.gnu.org
12759 4. http://gcc.gnu.org/lists.html
12760 5. http://www.fsf.org/
12761 6. http://gcc.gnu.org/about.html
12762 7. http://validator.w3.org/check/referer
12763 ======================================================================
12764 http://gcc.gnu.org/egcs-1.0/index.html
12767 December 3, 1997: We are pleased to announce the release of EGCS 1.0.
12768 January 6, 1998: We are pleased to announce the release of EGCS 1.0.1.
12769 March 16, 1998: We are pleased to announce the release of EGCS 1.0.2.
12770 May 15, 1998 We are pleased to announce the release of EGCS 1.0.3.
12772 EGCS is a collaborative effort involving several groups of hackers
12773 using an open development model to accelerate development and testing
12774 of GNU compilers and runtime libraries.
12776 An important goal of EGCS is to allow wide scale testing of
12777 experimental features and optimizations; therefore, EGCS contains some
12778 features and optimizations which are still under development. However,
12779 EGCS has been carefully tested and should be comparable in quality to
12782 EGCS 1.0 is based on an August 2, 1997 snapshot of the GCC 2.8
12783 development sources; it contains nearly all of the new features found
12786 EGCS 1.0 also contains many improvements and features not found in GCC
12787 2.7 and even the GCC 2.8 series (which was released after the original
12789 * Integrated C++ runtime libraries, including support for most major
12791 * The integrated libstdc++ library includes a verbatim copy of SGI's
12793 * Integrated GNU Fortran compiler.
12794 * New instruction scheduler.
12795 * New alias analysis code.
12797 See the [1]new features page for a more complete list of new features.
12799 EGCS 1.0.1 is a minor update to the EGCS 1.0 compiler to fix a few
12800 critical bugs and add support for Red Hat 5.0 Linux. Changes since the
12802 * Add support for Red Hat 5.0 Linux and better support for Linux
12803 systems using glibc2.
12804 Many programs failed to link when compiled with EGCS 1.0 on Red Hat
12805 5.0 or on systems with newer versions of glibc2. EGCS 1.0.1 should
12806 fix these problems.
12807 * Compatibility with both EGCS 1.0 and GCC 2.8 libgcc exception
12808 handling interfaces.
12809 To avoid future compatibility problems, we strongly urge anyone who
12810 is planning on distributing shared libraries that contain C++ code
12811 to upgrade to EGCS 1.0.1 first.
12812 Soon after EGCS 1.0 was released, the GCC developers made some
12813 incompatible changes in libgcc's exception handling interfaces.
12814 These changes were needed to solve problems on some platforms. This
12815 means that GCC 2.8.0, when released, will not be seamlessly
12816 compatible with shared libraries built by EGCS 1.0. The reason is
12817 that the libgcc.a in GCC 2.8.0 will not contain a function needed
12818 by the old interface.
12819 The result of this is that there may be compatibility problems with
12820 shared libraries built by EGCS 1.0 when used with GCC 2.8.0.
12821 With EGCS 1.0.1, generated code uses the new (GCC 2.8.0) interface,
12822 and libgcc.a has the support routines for both the old and the new
12823 interfaces (so EGCS 1.0.1 and EGCS 1.0 code can be freely mixed,
12824 and EGCS 1.0.1 and GCC 2.8.0 code can be freely mixed).
12825 The maintainers of GCC 2.x have decided against including seamless
12826 support for the old interface in 2.8.0, since it was never
12827 "official", so to avoid future compatibility problems we recommend
12828 against distributing any shared libraries built by EGCS 1.0 that
12829 contain C++ code (upgrade to 1.0.1 and use that).
12830 * Various bugfixes in the x86, hppa, mips, and rs6000/ppc backends.
12831 The x86 changes fix code generation errors exposed when building
12832 glibc2 and the usual GNU/Linux dynamic linker (ld.so).
12833 The hppa change fixes a compiler abort when configured for use with
12835 The MIPS changes fix problems with the definition of LONG_MAX on
12836 newer systems, allow for command line selection of the target ABI,
12837 and fix one code generation problem.
12838 The rs6000/ppc change fixes some problems with passing structures
12839 to varargs/stdarg functions.
12840 * A few machine independent bugfixes, mostly to fix code generation
12841 errors when building Linux kernels or glibc.
12842 * Fix a few critical exception handling and template bugs in the C++
12844 * Fix Fortran namelist bug on alphas.
12845 * Fix build problems on x86-solaris systems.
12847 EGCS 1.0.2 is a minor update to the EGCS 1.0.1 compiler to fix several
12848 serious problems in EGCS 1.0.1.
12849 * General improvements and fixes
12850 + Memory consumption significantly reduced, especially for
12851 templates and inline functions.
12852 + Fix various problems with glibc2.1.
12853 + Fix loop optimization bug exposed by rs6000/ppc port.
12854 + Fix to avoid potential code generation problems in jump.c.
12855 + Fix some undefined symbol problems in dwarf1 debug support.
12856 * g++/libstdc++ improvements and fixes
12857 + libstdc++ in the EGCS release has been updated and should be
12858 link compatible with libstdc++-2.8.
12859 + Various fixes in libio/libstdc++ to work better on GNU/Linux
12861 + Fix problems with duplicate symbols on systems that do not
12862 support weak symbols.
12863 + Memory corruption bug and undefined symbols in bastring have
12865 + Various exception handling fixes.
12866 + Fix compiler abort for very long thunk names.
12867 * g77 improvements and fixes
12868 + Fix compiler crash for omitted bound in Fortran CASE
12870 + Add missing entries to g77 lang-options.
12871 + Fix problem with -fpedantic in the g77 compiler.
12872 + Fix "backspace" problem with g77 on alphas.
12873 + Fix x86 backend problem with Fortran literals and -fpic.
12874 + Fix some of the problems with negative subscripts for g77 on
12876 + Fixes for Fortran builds on cygwin32/mingw32.
12877 * platform specific improvements and fixes
12878 + Fix long double problems on x86 (exposed by glibc).
12879 + x86 ports define i386 again to keep imake happy.
12880 + Fix exception handling support on NetBSD ports.
12881 + Several changes to collect2 to fix many problems with AIX.
12882 + Define __ELF__ for GNU/Linux on rs6000.
12883 + Fix -mcall-linux problem on GNU/Linux on rs6000.
12884 + Fix stdarg/vararg problem for GNU/Linux on rs6000.
12885 + Allow autoconf to select a proper install problem on AIX 3.1.
12886 + m68k port support includes -mcpu32 option as well as cpu32
12888 + Fix stdarg bug for irix6.
12889 + Allow EGCS to build on irix5 without the gnu assembler.
12890 + Fix problem with static linking on sco5.
12891 + Fix bootstrap on sco5 with native compiler.
12892 + Fix for abort building newlib on H8 target.
12893 + Fix fixincludes handling of math.h on SunOS.
12894 + Minor fix for Motorola 3300 m68k systems.
12896 EGCS 1.0.3 is a minor update to the EGCS 1.0.2 compiler to fix a few
12897 problems reported by Red Hat for builds of Red Hat 5.1.
12898 * Generic bugfixes:
12899 + Fix a typo in the libio library which resulted in incorrect
12900 behavior of istream::get.
12901 + Fix the Fortran negative array index problem.
12902 + Fix a major problem with the ObjC runtime thread support
12904 + Reduce memory consumption of the Haifa scheduler.
12905 * Target specific bugfixes:
12906 + Fix one x86 floating point code generation bug exposed by
12908 + Fix one x86 internal compiler error exposed by glibc2 builds.
12909 + Fix profiling bugs on the Alpha.
12910 + Fix ImageMagick & emacs 20.2 build problems on the Alpha.
12911 + Fix rs6000/ppc bug when converting values from integer types
12912 to floating point types.
12914 The EGCS 1.0 releases include installation instructions in both HTML
12915 and plaintext forms (see the INSTALL directory in the toplevel
12916 directory of the distribution). However, we also keep the most up to
12917 date [2]installation instructions and [3]build/test status on our web
12918 page. We will update those pages as new information becomes available.
12920 And, we can't in good conscience fail to mention some [4]caveats to
12923 Update: Big thanks to Stanford for providing a high speed link for
12924 downloading EGCS (go.cygnus.com)!
12926 Download EGCS from ftp.cygnus.com (USA California) or go.cygnus.com
12927 (USA California -- High speed link provided by Stanford).
12929 The EGCS 1.0 release is also available many mirror sites.
12930 [5]Goto mirror list to find a closer site
12932 We'd like to thank the numerous people that have contributed new
12933 features, test results, bugfixes, etc. Unfortunately, they're far too
12934 numerous to mention by name.
12937 For questions related to the use of GCC, please consult these web
12938 pages and the [6]GCC manuals. If that fails, the
12939 [7]gcc-help@gcc.gnu.org mailing list might help. Comments on these
12940 web pages and the development of GCC are welcome on our developer
12941 list at [8]gcc@gcc.gnu.org. All of [9]our lists have public archives.
12943 Copyright (C) [10]Free Software Foundation, Inc. Verbatim copying and
12944 distribution of this entire article is permitted in any medium,
12945 provided this notice is preserved.
12947 These pages are [11]maintained by the GCC team. Last modified
12952 1. http://gcc.gnu.org/egcs-1.0/features.html
12953 2. http://gcc.gnu.org/install/
12954 3. http://gcc.gnu.org/egcs-1.0/buildstat.html
12955 4. http://gcc.gnu.org/egcs-1.0/caveats.html
12956 5. http://gcc.gnu.org/mirrors.html
12957 6. http://gcc.gnu.org/onlinedocs/
12958 7. mailto:gcc-help@gcc.gnu.org
12959 8. mailto:gcc@gcc.gnu.org
12960 9. http://gcc.gnu.org/lists.html
12961 10. http://www.fsf.org/
12962 11. http://gcc.gnu.org/about.html
12963 12. http://validator.w3.org/check/referer
12964 ======================================================================
12965 http://gcc.gnu.org/egcs-1.0/features.html
12968 * Core compiler is based on the gcc2 development tree from Aug 2,
12969 1997, so we have most of the [1]features found in GCC 2.8.
12970 * Integrated GNU Fortran compiler based on g77-0.5.22-19970929.
12971 * Vast improvements in the C++ compiler; so many they have [2]page of
12973 * Integrated C++ runtime libraries, including support for most major
12975 * New instruction scheduler from IBM Haifa which includes support for
12976 function wide instruction scheduling as well as superscalar
12978 * Significantly improved alias analysis code.
12979 * Improved register allocation for two address machines.
12980 * Significant code generation improvements for Fortran code on
12982 * Various optimizations from the g77 project as well as improved loop
12984 * Dwarf2 debug format support for some targets.
12985 * egcs libstdc++ includes the SGI STL implementation without changes.
12986 * As a result of these and other changes, egcs libstc++ is not binary
12987 compatible with previous releases of libstdc++.
12988 * Various new ports -- UltraSPARC, Irix6.2 & Irix6.3 support, The SCO
12989 Openserver 5 family (5.0.{0,2,4} and Internet FastStart 1.0 and
12990 1.1), Support for RTEMS on several embedded targets, Support for
12991 arm-linux, Mitsubishi M32R, Hitachi H8/S, Matsushita MN102 and
12992 MN103, NEC V850, Sparclet, Solaris & GNU/Linux on PowerPCs, etc.
12993 * Integrated testsuites for gcc, g++, g77, libstdc++ and libio.
12994 * RS6000/PowerPC ports generate code which can run on all
12995 RS6000/PowerPC variants by default.
12996 * -mcpu= and -march= switches for the x86 port to allow better
12997 control over how the x86 port generates code.
12998 * Includes the template repository patch (aka repo patch); note the
12999 new template code makes repo obsolete for ELF systems using gnu-ld
13001 * Plus the usual assortment of bugfixes and improvements.
13004 For questions related to the use of GCC, please consult these web
13005 pages and the [3]GCC manuals. If that fails, the
13006 [4]gcc-help@gcc.gnu.org mailing list might help. Comments on these
13007 web pages and the development of GCC are welcome on our developer
13008 list at [5]gcc@gcc.gnu.org. All of [6]our lists have public archives.
13010 Copyright (C) [7]Free Software Foundation, Inc. Verbatim copying and
13011 distribution of this entire article is permitted in any medium,
13012 provided this notice is preserved.
13014 These pages are [8]maintained by the GCC team. Last modified
13019 1. http://gcc.gnu.org/egcs-1.0/features-2.8.html
13020 2. http://gcc.gnu.org/egcs-1.0/c++features.html
13021 3. http://gcc.gnu.org/onlinedocs/
13022 4. mailto:gcc-help@gcc.gnu.org
13023 5. mailto:gcc@gcc.gnu.org
13024 6. http://gcc.gnu.org/lists.html
13025 7. http://www.fsf.org/
13026 8. http://gcc.gnu.org/about.html
13027 9. http://validator.w3.org/check/referer
13028 ======================================================================
13029 http://gcc.gnu.org/egcs-1.0/caveats.html
13032 * EGCS has an integrated libstdc++, but does not have an integrated
13033 libg++. Furthermore old libg++ releases will not work with egc; HJ
13034 Lu has made a libg++-2.8.1.2 available which may work with EGCS.
13035 Note most C++ programs only need libstdc++.
13036 * Note that using -pedantic or -Wreturn-type can cause an explosion
13037 in the amount of memory needed for template-heavy C++ code, such as
13038 code that uses STL. Also note that -Wall includes -Wreturn-type, so
13039 if you use -Wall you will need to specify -Wno-return-type to turn
13041 * Exception handling may not work with shared libraries, particularly
13042 on alphas, hppas, and mips based platforms. Exception handling is
13043 known to work on x86-linux platforms with shared libraries.
13044 * Some versions of the Linux kernel have bugs which prevent them from
13045 being compiled or from running when compiled by EGCS. See the FAQ
13046 (as shipped with EGCS 1.0) for additional information.
13047 * In general, EGCS is more rigorous about rejecting invalid C++ code
13048 or deprecated C++ constructs than G++ 2.7. As a result it may be
13049 necessary to fix C++ code before it will compile with EGCS.
13050 * G++ is also aggressively tracking the C++ standard; as a result
13051 code which was previously valid (and thus accepted by other
13052 compilers and older versions of G++) may no longer be accepted.
13053 * EGCS 1.0 may not work with Red Hat Linux 5.0 on all targets. EGCS
13054 1.0.x and later releases should work with Red Hat Linux 5.0.
13057 For questions related to the use of GCC, please consult these web
13058 pages and the [1]GCC manuals. If that fails, the
13059 [2]gcc-help@gcc.gnu.org mailing list might help. Comments on these
13060 web pages and the development of GCC are welcome on our developer
13061 list at [3]gcc@gcc.gnu.org. All of [4]our lists have public archives.
13063 Copyright (C) [5]Free Software Foundation, Inc. Verbatim copying and
13064 distribution of this entire article is permitted in any medium,
13065 provided this notice is preserved.
13067 These pages are [6]maintained by the GCC team. Last modified
13072 1. http://gcc.gnu.org/onlinedocs/
13073 2. mailto:gcc-help@gcc.gnu.org
13074 3. mailto:gcc@gcc.gnu.org
13075 4. http://gcc.gnu.org/lists.html
13076 5. http://www.fsf.org/
13077 6. http://gcc.gnu.org/about.html
13078 7. http://validator.w3.org/check/referer
13079 ======================================================================