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
10 GCC 4.7 Release Series
14 The [1]GNU project and the GCC developers are pleased to announce the
17 This release is a bug-fix release, containing fixes for regressions in
18 GCC 4.7.2 relative to previous releases of GCC.
23 April 11, 2013 ([2]changes, [3]documentation)
26 September 20, 2012 ([4]changes, [5]documentation)
29 June 14, 2012 ([6]changes, [7]documentation)
32 March 22, 2012 ([8]changes, [9]documentation)
34 References and Acknowledgements
36 GCC used to stand for the GNU C Compiler, but since the compiler
37 supports several other languages aside from C, it now stands for the
38 GNU Compiler Collection.
40 A list of [10]successful builds is updated as new information becomes
43 The GCC developers would like to thank the numerous people that have
44 contributed new features, improvements, bug fixes, and other changes as
45 well as test results to GCC. This [11]amazing group of volunteers is
46 what makes GCC successful.
48 For additional information about GCC please refer to the [12]GCC
49 project web site or contact the [13]GCC development mailing list.
51 To obtain GCC please use [14]our mirror sites or [15]our SVN server.
54 For questions related to the use of GCC, please consult these web
55 pages and the [16]GCC manuals. If that fails, the
56 [17]gcc-help@gcc.gnu.org mailing list might help. Comments on these
57 web pages and the development of GCC are welcome on our developer
58 list at [18]gcc@gcc.gnu.org. All of [19]our lists have public
61 Copyright (C) [20]Free Software Foundation, Inc. Verbatim copying and
62 distribution of this entire article is permitted in any medium,
63 provided this notice is preserved.
65 These pages are [21]maintained by the GCC team. Last modified
70 1. http://www.gnu.org/
71 2. http://gcc.gnu.org/gcc-4.7/changes.html
72 3. http://gcc.gnu.org/onlinedocs/4.7.3/
73 4. http://gcc.gnu.org/gcc-4.7/changes.html
74 5. http://gcc.gnu.org/onlinedocs/4.7.2/
75 6. http://gcc.gnu.org/gcc-4.7/changes.html
76 7. http://gcc.gnu.org/onlinedocs/4.7.1/
77 8. http://gcc.gnu.org/gcc-4.7/changes.html
78 9. http://gcc.gnu.org/onlinedocs/4.7.0/
79 10. http://gcc.gnu.org/gcc-4.7/buildstat.html
80 11. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
81 12. http://gcc.gnu.org/index.html
82 13. mailto:gcc@gcc.gnu.org
83 14. http://gcc.gnu.org/mirrors.html
84 15. http://gcc.gnu.org/svn.html
85 16. http://gcc.gnu.org/onlinedocs/
86 17. mailto:gcc-help@gcc.gnu.org
87 18. mailto:gcc@gcc.gnu.org
88 19. http://gcc.gnu.org/lists.html
89 20. http://www.fsf.org/
90 21. http://gcc.gnu.org/about.html
91 22. http://validator.w3.org/check/referer
92 ======================================================================
93 http://gcc.gnu.org/gcc-4.7/changes.html
95 GCC 4.7 Release Series
96 Changes, New Features, and Fixes
100 * The -fconserve-space flag has been deprecated. The flag had no
101 effect for most targets: only targets without a global .bss section
102 and without support for switchable sections. Furthermore, the flag
103 only had an effect for G++, where it could result in wrong
104 semantics (please refer to the GCC manual for further details). The
105 flag will be removed in GCC 4.8
106 * Support for a number of older systems and recently unmaintained or
107 untested target ports of GCC has been declared obsolete in GCC 4.7.
108 Unless there is activity to revive them, the next release of GCC
109 will have their sources permanently removed.
110 All GCC ports for the following processor architectures have been
112 + picoChip (picochip-*)
113 The following ports for individual systems on particular
114 architectures have been obsoleted:
115 + IRIX 6.5 (mips-sgi-irix6.5)
116 + MIPS OpenBSD (mips*-*-openbsd*)
117 + Solaris 8 (*-*-solaris2.8). Details can be found in the
119 + Tru64 UNIX V5.1 (alpha*-dec-osf5.1*)
120 * On ARM, when compiling for ARMv6 (but not ARMv6-M), ARMv7-A,
121 ARMv7-R, or ARMv7-M, the new option -munaligned-access is active by
122 default, which for some sources generates code that accesses memory
123 on unaligned addresses. This requires the kernel of those systems
124 to enable such accesses (controlled by CP15 register c1, refer to
125 ARM documentation). Alternatively, or for compatibility with
126 kernels where unaligned accesses are not supported, all code has to
127 be compiled with -mno-unaligned-access. Upstream Linux kernel
128 releases have automatically and unconditionally supported unaligned
129 accesses as emitted by GCC due to this option being active since
131 * Support on ARM for the legacy floating-point accelerator (FPA) and
132 the mixed-endian floating-point format that it used has been
133 obsoleted. The ports that still use this format have been obsoleted
134 as well. Many legacy ARM ports already provide an alternative that
135 uses the VFP floating-point format. The obsolete ports will be
136 deleted in the next release.
137 The obsolete ports with alternatives are:
138 + arm*-*-rtems (use arm*-*-rtemseabi)
139 + arm*-*-linux-gnu (use arm*-*-linux-gnueabi)
140 + arm*-*-elf (use arm*-*-eabi)
141 + arm*-*-uclinux* (use arm*-*-uclinux*eabi)
142 Note, however, that these alternatives are not binary compatible
143 with their legacy counterparts (although some can support running
144 legacy applications).
145 The obsolete ports that currently lack a modern alternative are:
149 New ports that support more recent versions of the architecture are
151 * Support for the Maverick co-processor on ARM has been obsoleted.
152 Code to support it will be deleted in the next release.
153 * Support has been removed for Unix International threads on Solaris
154 2, so the --enable-threads=solaris configure option and the
155 -threads compiler option don't work any longer.
156 * Support has been removed for the Solaris BSD Compatibility Package,
157 which lives in /usr/ucbinclude and /usr/ucblib. It has been removed
158 from Solaris 11, and was only intended as a migration aid from
159 SunOS 4 to SunOS 5. The -compat-bsd compiler option is not
160 recognized any longer.
161 * The AVR port's libgcc has been improved and its multilib structure
162 has been enhanced. As a result, all objects contributing to an
163 application must either be compiled with GCC versions up to 4.6.x
164 or with GCC versions 4.7.1 or later. If the compiler is used with
165 AVR Libc, you need a version that supports the new layout, i.e.
166 implements [2]#35407.
167 * The AVR port's -mshort-calls command-line option has been
168 deprecated. It will be removed in the GCC 4.8 release. See -mrelax
170 * The AVR port only references startup code that clears .bss and the
171 common section resp. initializes the .data and .rodata section
172 provided respective sections (or subsections thereof) are not
173 empty, see [3]PR18145. Applications that put all static storage
174 objects into non-standard sections and / or define all static
175 storage objects in assembler modules, must reference __do_clear_bss
176 resp. __do_copy_data by hand or undefine the symbol(s) by means of
177 -Wl,-u,__do_clear_bss resp. -Wl,-u,__do_copy_data.
178 * The ARM port's -mwords-little-endian option has been deprecated. It
179 will be removed in a future release.
180 * Support has been removed for the NetWare x86 configuration
181 obsoleted in GCC 4.6.
182 * It is no longer possible to use the "l" constraint in MIPS16 asm
184 * GCC versions 4.7.0 and 4.7.1 had changes to the C++ standard
185 library which affected the ABI in C++11 mode: a data member was
186 added to std::list changing its size and altering the definitions
187 of some member functions, and std::pair's move constructor was
188 non-trivial which altered the calling convention for functions with
189 std::pair arguments or return types. The ABI incompatibilities have
190 been fixed for GCC version 4.7.2 but as a result C++11 code
191 compiled with GCC 4.7.0 or 4.7.1 may be incompatible with C++11
192 code compiled with different GCC versions and with C++98/C++03 code
193 compiled with any version.
194 * On ARM, a bug has been fixed in GCC's implementation of the AAPCS
195 rules for the layout of vectors that could lead to wrong code being
196 generated. Vectors larger than 8 bytes in size are now by default
197 aligned to an 8-byte boundary. This is an ABI change: code that
198 makes explicit use of vector types may be incompatible with binary
199 objects built with older versions of GCC. Auto-vectorized code is
200 not affected by this change. (This change affects GCC versions
202 * More information on porting to GCC 4.7 from previous versions of
203 GCC can be found in the [4]porting guide for this release.
205 General Optimizer Improvements
207 * Support for a new parameter --param case-values-threshold=n was
208 added to allow users to control the cutoff between doing switch
209 statements as a series of if statements and using a jump table.
210 * Link-time optimization (LTO) improvements:
211 + Improved scalability and reduced memory usage. Link time
212 optimization of Firefox now requires 3GB of RAM on a 64-bit
213 system, while over 8GB was needed previously. Linking time has
214 been improved, too. The serial stage of linking Firefox has
215 been sped up by about a factor of 10.
216 + Reduced size of object files and temporary storage used during
218 + Streaming performance (both outbound and inbound) has been
220 + ld -r is now supported with LTO.
221 + Several bug fixes, especially in symbol table handling and
223 * Interprocedural optimization improvements:
224 + Heuristics now take into account that after inlining code will
225 be optimized out because of known values (or properties) of
226 function parameters. For example:
237 The call of foo will be inlined into bar even when optimizing
238 for code size. Constructs based on __builtin_constant_p are
239 now understood by the inliner and code size estimates are
240 evaluated a lot more realistically.
241 + The representation of C++ virtual thunks and aliases (both
242 implicit and defined via the alias attribute) has been
243 re-engineered. Aliases no longer pose optimization barriers
244 and calls to an alias can be inlined and otherwise optimized.
245 + The inter-procedural constant propagation pass has been
246 rewritten. It now performs generic function specialization.
247 For example when compiling the following:
253 ... do something else ...
265 GCC will now produce two copies of foo. One with flag being
266 true, while other with flag being false. This leads to
267 performance improvements previously possible only by inlining
268 all calls. Cloning causes a lot less code size growth.
269 * A string length optimization pass has been added. It attempts to
270 track string lengths and optimize various standard C string
271 functions like strlen, strchr, strcpy, strcat, stpcpy and their
272 _FORTIFY_SOURCE counterparts into faster alternatives. This pass is
273 enabled by default at -O2 or above, unless optimizing for size, and
274 can be disabled by the -fno-optimize-strlen option. The pass can
276 char *bar (const char *a)
278 size_t l = strlen (a) + 2;
279 char *p = malloc (l); if (p == NULL) return p;
280 strcpy (p, a); strcat (p, "/"); return p;
284 char *bar (const char *a)
286 size_t tmp = strlen (a);
287 char *p = malloc (tmp + 2); if (p == NULL) return p;
288 memcpy (p, a, tmp); memcpy (p + tmp, "/", 2); return p;
291 or for hosted compilations where stpcpy is available in the runtime
292 and headers provide its prototype, e.g.
293 void foo (char *a, const char *b, const char *c, const char *d)
295 strcpy (a, b); strcat (a, c); strcat (a, d);
298 can be optimized into:
299 void foo (char *a, const char *b, const char *c, const char *d)
301 strcpy (stpcpy (stpcpy (a, b), c), d);
304 New Languages and Language specific improvements
306 * Version 3.1 of the [5]OpenMP specification is now supported for the
307 C, C++, and Fortran compilers.
311 * The command-line option -feliminate-unused-debug-types has been
312 re-enabled by default, as it is for the other languages, leading to
313 a reduction in debug info size of 12.5% and more for relevant
314 cases, as well as to a small compilation speedup.
318 * A new built-in, __builtin_assume_aligned, has been added, through
319 which the compiler can be hinted about pointer alignment and can
320 use it to improve generated code.
321 * A new warning option -Wunused-local-typedefs was added for C, C++,
322 Objective-C and Objective-C++. This warning diagnoses typedefs
323 locally defined in a function, and otherwise not used.
324 * A new experimental command-line option -ftrack-macro-expansion was
325 added for C, C++, Objective-C, Objective-C++ and Fortran. It allows
326 the compiler to emit diagnostic about the current macro expansion
327 stack when a compilation error occurs in a macro expansion.
328 * Experimental support for transactional memory has been added. It
329 includes support in the compiler, as well as a supporting runtime
330 library called libitm. To compile code with transactional memory
331 constructs, use the -fgnu-tm option.
332 Support is currently available for Alpha, ARM, PowerPC, SH, SPARC,
333 and 32-bit/64-bit x86 platforms.
334 For more details on transactional memory see [6]the GCC WiKi.
335 * Support for atomic operations specifying the C++11/C11 memory model
336 has been added. These new __atomic routines replace the existing
337 __sync built-in routines.
338 Atomic support is also available for memory blocks. Lock-free
339 instructions will be used if a memory block is the same size and
340 alignment as a supported integer type. Atomic operations which do
341 not have lock-free support are left as function calls. A set of
342 library functions is available on the GCC atomic wiki in the
343 "External Atomics Library" section.
344 For more details on the memory models and features, see the
346 * When a binary operation is performed on vector types and one of the
347 operands is a uniform vector, it is possible to replace the vector
348 with the generating element. For example:
349 typedef int v4si __attribute__ ((vector_size (16)));
350 v4si res, a = {1,2,3,4};
353 res = 2 + a; /* means {2,2,2,2} + a */
354 res = a - x; /* means a - {x,x,x,x} */
358 * There is support for some more features from the C11 revision of
359 the ISO C standard. GCC now accepts the options -std=c11 and
360 -std=gnu11, in addition to the previous -std=c1x and -std=gnu1x.
361 + Unicode strings (previously supported only with options such
362 as -std=gnu11, now supported with -std=c11), and the
363 predefined macros __STDC_UTF_16__ and __STDC_UTF_32__.
364 + Nonreturning functions (_Noreturn and <stdnoreturn.h>).
365 + Alignment support (_Alignas, _Alignof, max_align_t,
367 + A built-in function __builtin_complex is provided to support C
368 library implementation of the CMPLX family of macros.
372 * G++ now accepts the -std=c++11, -std=gnu++11, and -Wc++11-compat
373 options, which are equivalent to -std=c++0x, -std=gnu++0x, and
374 -Wc++0x-compat, respectively.
375 * G++ now implements [8]C++11 extended friend syntax:
380 static const int I = 2;
390 * Thanks to Ville Voutilainen, G++ now implements [9]C++11 explicit
394 virtual void f() const final;
399 void f() const; // error: D::f attempts to override final B::f
400 void f(long) override; // error: doesn't override anything
401 void f(int) override; // ok
405 struct F: E { }; // error: deriving from final class
407 * G++ now implements [10]C++11 non-static data member initializers.
411 } a; // initializes a.i to 42
413 * Thanks to Ed Smith-Rowland, G++ now implements [11]C++11
414 user-defined literals.
416 // Not actually a good approximation. :)
417 constexpr long double operator"" _degrees (long double d) { return d * 0.0175; }
418 long double pi = 180.0_degrees;
420 * G++ now implements [12]C++11 alias-declarations.
422 template <class T> using Ptr = T*;
423 Ptr<int> ip; // decltype(ip) is int*
425 * Thanks to Ville Voutilainen and Pedro Lamarao, G++ now implements
426 [13]C++11 delegating constructors.
430 A(): A(42) { } // delegate to the A(int) constructor
433 * G++ now fully implements C++11 atomic classes rather than just
434 integer derived classes.
440 std::atomic<POD> my_atomic_POD;
442 * G++ now sets the predefined macro __cplusplus to the correct value,
443 199711L for C++98/03, and 201103L for C++11.
444 * G++ now correctly implements the two-phase lookup rules such that
445 an unqualified name used in a template must have an appropriate
446 declaration found either in scope at the point of definition of the
447 template or by argument-dependent lookup at the point of
448 instantiation. As a result, code that relies on a second
449 unqualified lookup at the point of instantiation to find functions
450 declared after the template or in dependent bases will be rejected.
451 The compiler will suggest ways to fix affected code, and using the
452 -fpermissive compiler flag will allow the code to compile with a
456 void f() { g(T()); } // error, g(int) not found by argument-dependent lookup
457 void g(int) { } // fix by moving this declaration before the declaration of f
461 // error, B::g(B) not found by argument-dependent lookup
462 void f() { g(T()); } // fix by using this->g or A::g
465 struct B { void g(B); };
473 * G++ now properly re-uses stack space allocated for temporary
474 objects when their lifetime ends, which can significantly lower
475 stack consumption for some C++ functions. As a result of this, some
476 code with undefined behavior will now break:
478 const int &f(const int &i) { return i; }
483 Here, x refers to the temporary allocated to hold the 1 argument,
484 which only lives until the end of the initialization; it
485 immediately becomes a dangling reference. So the next statement
486 re-uses the stack slot to hold the 2 argument, and users of x get
488 Note that this should not cause any change of behavior for
489 temporaries of types with non-trivial destructors, as they are
490 already destroyed at end of full-expression; the change is that now
491 the storage is released as well.
492 * A new command-line option -Wdelete-non-virtual-dtor has been added
493 to warn when delete is used to destroy an instance of a class which
494 has virtual functions and non-virtual destructor. It is unsafe to
495 delete an instance of a derived class through a pointer to a base
496 class if the base class does not have a virtual destructor. This
497 warning is enabled by -Wall.
498 * A new command-line option -Wzero-as-null-pointer-constant has been
499 added to warn when a literal '0' is used as null pointer constant.
500 It can be useful to facilitate the conversion to nullptr in C++11.
501 * As per C++98, access-declarations are now deprecated by G++.
502 Using-declarations are to be used instead. Furthermore, some
503 efforts have been made to improve the support of class scope
504 using-declarations. In particular, using-declarations referring to
505 a dependent type now work as expected ([14]bug c++/14258).
506 * The ELF symbol visibility of a template instantiation is now
507 properly constrained by the visibility of its template arguments
510 Runtime Library (libstdc++)
512 * [16]Improved experimental support for the new ISO C++ standard,
514 + using noexcept in most of the library;
515 + implementations of pointer_traits, allocator_traits and
516 scoped_allocator_adaptor;
517 + uses-allocator construction for tuple;
518 + vector meets the allocator-aware container requirements;
519 + replacing monotonic_clock with steady_clock;
520 + enabling the thread support library on most POSIX targets;
521 + many small improvements to conform to the FDIS.
522 * Added --enable-clocale=newlib configure option.
523 * Debug Mode iterators for unordered associative containers.
524 * Avoid polluting the global namespace and do not include <unistd.h>.
528 * The compile flag [17]-fstack-arrays has been added, which causes
529 all local arrays to be put on stack memory. For some programs this
530 will improve the performance significantly. If your program uses
531 very large local arrays, it is possible that you will have to
532 extend your runtime limits for stack memory.
533 * The [18]-Ofast flag now also implies [19]-fno-protect-parens and
535 * Front-end optimizations can now be selected by the
536 [21]-ffrontend-optimize option and deselected by the
537 -fno-frontend-optimize option.
538 * When front-end optimization removes a function call,
539 [22]-Wfunction-elimination warns about that.
540 * When performing front-end-optimization, the
541 [23]-faggressive-function-elimination option allows the removal of
542 duplicate function calls even for impure functions.
543 * The flag [24]-Wreal-q-constant has been added, which warns if
544 floating-point literals have been specified using q (such as
545 1.0q0); the q marker is now supported as a vendor extension to
546 denote quad precision (REAL(16) or, if not available, REAL(10)).
547 Consider using a kind parameter (such as in 1.0_qp) instead, which
548 can be obtained via [25]SELECTED_REAL_KIND.
549 * The GFORTRAN_USE_STDERR environment variable has been removed. GNU
550 Fortran now always prints error messages to standard error. If you
551 wish to redirect standard error, please consult the manual for your
552 OS, shell, batch environment etc. as appropriate.
553 * The -fdump-core option and GFORTRAN_ERROR_DUMPCORE environment
554 variable have been removed. When encountering a serious error,
555 gfortran will now always abort the program. Whether a core dump is
556 generated depends on the user environment settings; see the ulimit
557 -c setting for POSIX shells, limit coredumpsize for C shells, and
558 the [26]WER user-mode dumps settings on Windows.
559 * The [27]-fbacktrace option is now enabled by default. When
560 encountering a fatal error, gfortran will attempt to print a
561 backtrace to standard error before aborting. It can be disabled
562 with -fno-backtrace. Note: On POSIX targets with the addr2line
563 utility from GNU binutils, GNU Fortran can print a backtrace with
564 function name, file name, line number information in addition to
565 the addresses; otherwise only the addresses are printed.
567 + Generic interface names which have the same name as derived
568 types are now supported, which allows to write constructor
569 functions. Note that Fortran does not support static
570 constructor functions; only default initialization or an
571 explicit structure-constructor initialization are available.
572 + [29]Polymorphic (class) arrays are now supported.
574 + Support for the DO CONCURRENT construct has been added, which
575 allows the user to specify that individual loop iterations
576 have no interdependencies.
577 + [31]Coarrays: Full single-image support except for polymorphic
578 coarrays. Additionally, preliminary support for multiple
579 images via an MPI-based [32]coarray communication library has
580 been added. Note: The library version is not yet usable as
581 remote coarray access is not yet possible.
583 + New flag [34]-std=f2008ts permits programs that are expected
584 to conform to the Fortran 2008 standard and the draft
585 Technical Specification (TS) 29113 on Further Interoperability
587 + The OPTIONAL attribute is now allowed for dummy arguments of
589 + The RANK intrinsic has been added.
590 + The implementation of the ASYNCHRONOUS attribute in GCC is
591 compatible with the candidate draft of TS 29113 (since GCC
596 * GCC 4.7 implements the [35]Go 1 language standard. The library
597 support in 4.7.0 is not quite complete, due to release timing.
598 Release 4.7.1 includes complete support for Go 1. The Go library is
599 from the Go 1.0.1 release.
600 * Go has been tested on GNU/Linux and Solaris platforms. It may work
601 on other platforms as well.
603 New Targets and Target Specific Improvements
607 * GCC now supports the Cortex-A7 processor implementing the v7-a
608 version of the architecture using the option -mcpu=cortex-a7.
609 * The default vector size in auto-vectorization for NEON is now 128
610 bits. If vectorization fails thusly, the vectorizer tries again
612 * A new option -mvectorize-with-neon-double was added to allow users
613 to change the vector size to 64 bits.
617 * GCC now supports the XMEGA architecture. This requires GNU binutils
619 * Support for the [36]named address spaces __flash, __flash1, ...,
620 __flash5 and __memx has been added. These address spaces locate
621 read-only data in flash memory and allow reading from flash memory
622 by means of ordinary C code, i.e. without the need of (inline)
625 const __flash int values[] = { 42, 31 };
627 int add_values (const __flash int *p, int i)
629 return values[i] + *p;
632 * Support has been added for the AVR-specific configure option
633 --with-avrlibc=yes in order to arrange for better integration of
634 [37]AVR-Libc. This configure option is supported in avr-gcc 4.7.2
635 and newer and will only take effect in non-RTEMS configurations. If
636 avr-gcc is configured for RTEMS, the option will be ignored which
637 is the same as specifying --with-avrlibc=no. See [38]PR54461 for
638 more technical details.
639 * Support for AVR-specific [39]built-in functions has been added.
640 * Support has been added for the signed and unsigned 24-bit scalar
641 integer types __int24 and __uint24.
642 * New command-line options -maccumulate-args, -mbranch-cost=cost and
643 -mstrict-X were added to allow better fine-tuning of code
645 * The command option -fdata-sections now also takes affect on the
646 section names of variables with the progmem attribute.
647 * A new inline assembler print modifier %i to print a RAM address as
648 I/O address has been added:
650 #include <avr/io.h> /* Port Definitions from AVR-LibC */
652 void set_portb (uint8_t value)
654 asm volatile ("out %i0, %1" :: "n" (&PORTB), "r" (value) : "memory");
657 The offset between an I/O address and the RAM address for that I/O
658 location is device-specific. This offset is taken into account when
659 printing a RAM address with the %i modifier so that the address is
660 suitable to be used as operand in an I/O command. The address must
661 be a constant integer known at compile time.
662 * The inline assembler constraint "R" to represent integers in the
663 range -6 ... 5 has been removed without replacement.
664 * Many optimizations to:
665 + 64-bit integer arithmetic
666 + Widening multiplication
667 + Integer division by a constant
668 + Avoid constant reloading in multi-byte instructions.
669 + Micro-optimizations for special instruction sequences.
670 + Generic built-in functions like __builtin_ffs*,
672 + If-else decision trees generated by switch instructions
673 + Merging of data located in flash memory
674 + New libgcc variants for devices with 8-bit wide stack pointer
676 * Better documentation:
677 + Handling of EIND and indirect jumps on devices with more than
678 128 KiB of program memory.
679 + Handling of the RAMPD, RAMPX, RAMPY and RAMPZ special function
681 + Function attributes OS_main and OS_task.
682 + AVR-specific built-in macros.
686 * Support has been added for the Texas Instruments C6X family of
691 * Support has been added for National Semiconductor's CR16
696 * Support has been added for Adapteva's Epiphany architecture.
700 * Support for Intel AVX2 intrinsics, built-in functions and code
701 generation is available via -mavx2.
702 * Support for Intel BMI2 intrinsics, built-in functions and code
703 generation is available via -mbmi2.
704 * Implementation and automatic generation of __builtin_clz* using the
705 lzcnt instruction is available via -mlzcnt.
706 * Support for Intel FMA3 intrinsics and code generation is available
708 * A new -mfsgsbase command-line option is available that makes GCC
709 generate new segment register read/write instructions through
711 * Support for the new Intel rdrnd instruction is available via
713 * Two additional AVX vector conversion instructions are available via
715 * Support for new Intel processor codename IvyBridge with RDRND,
716 FSGSBASE and F16C is available through -march=core-avx-i.
717 * Support for the new Intel processor codename Haswell with AVX2,
718 FMA, BMI, BMI2, LZCNT is available through -march=core-avx2.
719 * Support for new AMD family 15h processors (Piledriver core) is now
720 available through -march=bdver2 and -mtune=bdver2 options.
721 * Support for [40]the x32 psABI is now available through the -mx32
723 * Windows mingw targets are using the -mms-bitfields option by
725 * Windows x86 targets are using the __thiscall calling convention for
726 C++ class-member functions.
727 * Support for the configure option --with-threads=posix for Windows
732 * GCC now supports thread-local storage (TLS) for MIPS16. This
733 requires GNU binutils 2.22 or later.
734 * GCC can now generate code specifically for the Cavium Octeon+ and
735 Octeon2 processors. The associated command-line options are
736 -march=octeon+ and -march=octeon2 respectively. Both options
737 require GNU binutils 2.22 or later.
738 * GCC can now work around certain 24k errata, under the control of
739 the command-line option -mfix-24k. These workarounds require GNU
740 binutils 2.20 or later.
741 * 32-bit MIPS GNU/Linux targets such as mips-linux-gnu can now build
742 n32 and n64 multilibs. The result is effectively a 64-bit GNU/Linux
743 toolchain that generates 32-bit code by default. Use the
744 configure-time option --enable-targets=all to select these extra
746 * Passing -fno-delayed-branch now also stops the assembler from
747 automatically filling delay slots.
751 * Vectors of type vector long long or vector long are passed and
752 returned using the same method as other vectors with the VSX
753 instruction set. Previously GCC did not adhere to the ABI for
754 128-bit vectors with 64-bit integer base types (PR 48857). This
755 will also be fixed in the GCC 4.6.1 and 4.5.4 releases.
756 * A new option -mno-pointers-to-nested-functions was added to allow
757 AIX 32-bit/64-bit and GNU/Linux 64-bit PowerPC users to specify
758 that the compiler should not load up the chain register (r11)
759 before calling a function through a pointer. If you use this
760 option, you cannot call nested functions through a pointer, or call
761 other languages that might use the static chain.
762 * A new option msave-toc-indirect was added to allow AIX
763 32-bit/64-bit and GNU/Linux 64-bit PowerPC users control whether we
764 save the TOC in the prologue for indirect calls or generate the
765 save inline. This can speed up some programs that call through a
766 function pointer a lot, but it can slow down other functions that
767 only call through a function pointer in exceptional cases.
768 * The PowerPC port will now enable machine-specific built-in
769 functions when the user switches the target machine using the
770 #pragma GCC target or __attribute__ ((__target__ ("target"))) code
771 sequences. In addition, the target macros are updated. However, due
772 to the way the -save-temps switch is implemented, you won't see the
773 effect of these additional macros being defined in preprocessor
778 * A new option -msoft-atomic has been added. When it is specified,
779 GCC will generate GNU/Linux-compatible gUSA atomic sequences for
780 the new __atomic routines.
781 * Since it is neither supported by GAS nor officially documented,
782 code generation for little endian SH2A has been disabled.
783 Specifying -ml with -m2a* will now result in a compiler error.
784 * The defunct -mbranch-cost option has been fixed.
785 * Some improvements to the generated code of:
786 + Utilization of the tst #imm,R0 instruction.
787 + Dynamic shift instructions on SH2A.
788 + Integer absolute value calculations.
789 * The -mdiv= option for targets other than SHmedia has been fixed and
794 * The option -mflat has been reinstated. When it is specified, the
795 compiler will generate code for a single register window model.
796 This is essentially a new implementation and the corresponding
797 debugger support has been added to GDB 7.4.
798 * Support for the options -mtune=native and -mcpu=native has been
799 added on selected native platforms (GNU/Linux and Solaris).
800 * Support for the SPARC T3 (Niagara 3) processor has been added.
802 + An intrinsics header visintrin.h has been added.
803 + Builtin intrinsics for the VIS 1.0 edge handling and pixel
804 compare instructions have been added.
805 + The little-endian version of alignaddr is now supported.
806 + When possible, VIS builtins are marked const, which should
807 increase the compiler's ability to optimize VIS operations.
808 + The compiler now properly tracks the %gsr register and how it
809 behaves as an input for various VIS instructions.
810 + Akin to fzero, the compiler can now generate fone instructions
811 in order to set all of the bits of a floating-point register
813 + The documentation for the VIS intrinsics in the GCC manual has
814 been brought up to date and many inaccuracies were fixed.
815 + Intrinsics for the VIS 2.0 bmask, bshuffle, and
816 non-condition-code setting edge instructions have been added.
817 Their availability is controlled by the new -mvis2 and
818 -mno-vis2 options. They are enabled by default on
819 UltraSPARC-III and later CPUs.
820 * Support for UltraSPARC Fused Multiply-Add floating-point extensions
821 has been added. These instructions are enabled by default on SPARC
822 T3 (Niagara 3) and later CPUs.
826 * Support has been added for the Tilera TILE-Gx and TILEPro families
829 Other significant improvements
831 * A new option (-grecord-gcc-switches) was added that appends
832 compiler command-line options that might affect code generation to
833 the DW_AT_producer attribute string in the DWARF debugging
835 * GCC now supports various new GNU extensions to the DWARF debugging
836 information format, like [41]entry value and [42]call site
837 information, [43]typed DWARF stack or [44]a more compact macro
838 representation. Support for these extensions has been added to GDB
839 7.4. They can be disabled through the -gstrict-dwarf command-line
844 This is the [45]list of problem reports (PRs) from GCC's bug tracking
845 system that are known to be fixed in the 4.7.1 release. This list might
846 not be complete (that is, it is possible that some PRs that have been
847 fixed are not listed here).
849 The Go frontend in the 4.7.1 release fully supports the [46]Go 1
854 This is the [47]list of problem reports (PRs) from GCC's bug tracking
855 system that are known to be fixed in the 4.7.2 release. This list might
856 not be complete (that is, it is possible that some PRs that have been
857 fixed are not listed here).
861 This is the [48]list of problem reports (PRs) from GCC's bug tracking
862 system that are known to be fixed in the 4.7.3 release. This list might
863 not be complete (that is, it is possible that some PRs that have been
864 fixed are not listed here).
867 For questions related to the use of GCC, please consult these web
868 pages and the [49]GCC manuals. If that fails, the
869 [50]gcc-help@gcc.gnu.org mailing list might help. Comments on these
870 web pages and the development of GCC are welcome on our developer
871 list at [51]gcc@gcc.gnu.org. All of [52]our lists have public
874 Copyright (C) [53]Free Software Foundation, Inc. Verbatim copying and
875 distribution of this entire article is permitted in any medium,
876 provided this notice is preserved.
878 These pages are [54]maintained by the GCC team. Last modified
883 1. http://gcc.gnu.org/ml/gcc-patches/2011-03/msg01263.html
884 2. http://savannah.nongnu.org/bugs/?35407
885 3. http://gcc.gnu.org/PR18145
886 4. http://gcc.gnu.org/gcc-4.7/porting_to.html
887 5. http://openmp.org/wp/openmp-specifications/
888 6. http://gcc.gnu.org/wiki/TransactionalMemory
889 7. http://gcc.gnu.org/wiki/Atomic/GCCMM
890 8. http://gcc.gnu.org/gcc-4.7/cxx0x_status.html
891 9. http://gcc.gnu.org/gcc-4.7/cxx0x_status.html
892 10. http://gcc.gnu.org/gcc-4.7/cxx0x_status.html
893 11. http://gcc.gnu.org/gcc-4.7/cxx0x_status.html
894 12. http://gcc.gnu.org/gcc-4.7/cxx0x_status.html
895 13. http://gcc.gnu.org/gcc-4.7/cxx0x_status.html
896 14. http://gcc.gnu.org/PR14258
897 15. http://gcc.gnu.org/PR35688
898 16. http://gcc.gnu.org/onlinedocs/gcc-4.7.1/libstdc++/manual/manual/status.html#status.iso.2011
899 17. http://gcc.gnu.org/onlinedocs/gcc-4.7.1/gfortran/Code-Gen-Options.html#index-g_t_0040code_007bfstack-arrays_007d-254
900 18. http://gcc.gnu.org/onlinedocs/gcc-4.7.1/gcc/Optimize-Options.html#index-Ofast-689
901 19. http://gcc.gnu.org/onlinedocs/gcc-4.7.1/gfortran/Code-Gen-Options.html#index-g_t_0040code_007bfno-protect-parens_007d-270
902 20. http://gcc.gnu.org/onlinedocs/gcc-4.7.1/gfortran/Code-Gen-Options.html#index-g_t_0040code_007bfstack-arrays_007d-254
903 21. http://gcc.gnu.org/onlinedocs/gcc-4.7.1/gfortran/Code-Gen-Options.html#index-g_t_0040code_007bfrontend-optimize_007d-275
904 22. http://gcc.gnu.org/onlinedocs/gcc-4.7.1/gfortran/Error-and-Warning-Options.html#index-g_t_0040code_007bWfunction-elimination_007d-170
905 23. http://gcc.gnu.org/onlinedocs/gcc-4.7.1/gfortran/Code-Gen-Options.html#index-g_t_0040code_007bfaggressive-function-elimination_007d-270
906 24. http://gcc.gnu.org/onlinedocs/gcc-4.7.1/gfortran/Error-and-Warning-Options.html#index-g_t_0040code_007bWreal-q-constant_007d-149
907 25. http://gcc.gnu.org/onlinedocs/gcc-4.7.1/gfortran/SELECTED_005fREAL_005fKIND.html
908 26. http://msdn.microsoft.com/en-us/library/bb787181%28v=vs.85%29.aspx
909 27. http://gcc.gnu.org/onlinedocs/gcc-4.7.1/gfortran/Debugging-Options.html#index-g_t_0040code_007bfno-backtrace_007d-183
910 28. http://gcc.gnu.org/wiki/Fortran2003Status
911 29. http://gcc.gnu.org/wiki/OOP
912 30. http://gcc.gnu.org/wiki/Fortran2008Status
913 31. http://gcc.gnu.org/wiki/Coarray
914 32. http://gcc.gnu.org/wiki/CoarrayLib
915 33. http://gcc.gnu.org/wiki/TS29113Status
916 34. 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
917 35. http://weekly.golang.org/doc/go1.html
918 36. http://gcc.gnu.org/onlinedocs/gcc-4.7.1/gcc/Named-Address-Spaces.html
919 37. http://nongnu.org/avr-libc/
920 38. http://gcc.gnu.org/PR54461
921 39. http://gcc.gnu.org/onlinedocs/gcc-4.7.1/gcc/AVR-Built%5f002din-Functions.html
922 40. https://sites.google.com/site/x32abi/
923 41. http://www.dwarfstd.org/ShowIssue.php?issue=100909.1
924 42. http://www.dwarfstd.org/ShowIssue.php?issue=100909.2
925 43. http://www.dwarfstd.org/doc/040408.1.html
926 44. http://www.dwarfstd.org/ShowIssue.php?issue=110722.1
927 45. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.7.1
928 46. http://weekly.golang.org/doc/go1.html
929 47. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.7.2
930 48. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.7.3
931 49. http://gcc.gnu.org/onlinedocs/
932 50. mailto:gcc-help@gcc.gnu.org
933 51. mailto:gcc@gcc.gnu.org
934 52. http://gcc.gnu.org/lists.html
935 53. http://www.fsf.org/
936 54. http://gcc.gnu.org/about.html
937 55. http://validator.w3.org/check/referer
938 ======================================================================
939 http://gcc.gnu.org/gcc-4.6/index.html
941 GCC 4.6 Release Series
945 The [1]GNU project and the GCC developers are pleased to announce the
946 release of GCC 4.6.3.
948 This release is a bug-fix release, containing fixes for regressions in
949 GCC 4.6.2 relative to previous releases of GCC.
954 March 1, 2012 ([2]changes, [3]documentation)
957 October 26, 2011 ([4]changes, [5]documentation)
960 June 27, 2011 ([6]changes, [7]documentation)
963 March 25, 2011 ([8]changes, [9]documentation)
965 References and Acknowledgements
967 GCC used to stand for the GNU C Compiler, but since the compiler
968 supports several other languages aside from C, it now stands for the
969 GNU Compiler Collection.
971 A list of [10]successful builds is updated as new information becomes
974 The GCC developers would like to thank the numerous people that have
975 contributed new features, improvements, bug fixes, and other changes as
976 well as test results to GCC. This [11]amazing group of volunteers is
977 what makes GCC successful.
979 For additional information about GCC please refer to the [12]GCC
980 project web site or contact the [13]GCC development mailing list.
982 To obtain GCC please use [14]our mirror sites or [15]our SVN server.
985 For questions related to the use of GCC, please consult these web
986 pages and the [16]GCC manuals. If that fails, the
987 [17]gcc-help@gcc.gnu.org mailing list might help. Comments on these
988 web pages and the development of GCC are welcome on our developer
989 list at [18]gcc@gcc.gnu.org. All of [19]our lists have public
992 Copyright (C) [20]Free Software Foundation, Inc. Verbatim copying and
993 distribution of this entire article is permitted in any medium,
994 provided this notice is preserved.
996 These pages are [21]maintained by the GCC team. Last modified
1001 1. http://www.gnu.org/
1002 2. http://gcc.gnu.org/gcc-4.6/changes.html
1003 3. http://gcc.gnu.org/onlinedocs/4.6.3/
1004 4. http://gcc.gnu.org/gcc-4.6/changes.html
1005 5. http://gcc.gnu.org/onlinedocs/4.6.2/
1006 6. http://gcc.gnu.org/gcc-4.6/changes.html
1007 7. http://gcc.gnu.org/onlinedocs/4.6.1/
1008 8. http://gcc.gnu.org/gcc-4.6/changes.html
1009 9. http://gcc.gnu.org/onlinedocs/4.6.0/
1010 10. http://gcc.gnu.org/gcc-4.6/buildstat.html
1011 11. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
1012 12. http://gcc.gnu.org/index.html
1013 13. mailto:gcc@gcc.gnu.org
1014 14. http://gcc.gnu.org/mirrors.html
1015 15. http://gcc.gnu.org/svn.html
1016 16. http://gcc.gnu.org/onlinedocs/
1017 17. mailto:gcc-help@gcc.gnu.org
1018 18. mailto:gcc@gcc.gnu.org
1019 19. http://gcc.gnu.org/lists.html
1020 20. http://www.fsf.org/
1021 21. http://gcc.gnu.org/about.html
1022 22. http://validator.w3.org/check/referer
1023 ======================================================================
1024 http://gcc.gnu.org/gcc-4.6/changes.html
1026 GCC 4.6 Release Series
1027 Changes, New Features, and Fixes
1031 * The options -b <machine> and -V <version> have been removed because
1032 they were unreliable. Instead, users should directly run
1033 <machine>-gcc when cross-compiling, or <machine>-gcc-<version> to
1034 run a different version of gcc.
1035 * GCC now has stricter checks for invalid command-line options. In
1036 particular, when gcc was called to link object files rather than
1037 compile source code, it would previously accept and ignore all
1038 options starting with --, including linker options such as
1039 --as-needed and --export-dynamic, although such options would
1040 result in errors if any source code was compiled. Such options, if
1041 unknown to the compiler, are now rejected in all cases; if the
1042 intent was to pass them to the linker, options such as
1043 -Wl,--as-needed should be used.
1044 * Versions of the GNU C library up to and including 2.11.1 included
1045 an [1]incorrect implementation of the cproj function. GCC optimizes
1046 its builtin cproj according to the behavior specified and allowed
1047 by the ISO C99 standard. If you want to avoid discrepancies between
1048 the C library and GCC's builtin transformations when using cproj in
1049 your code, use GLIBC 2.12 or later. If you are using an older GLIBC
1050 and actually rely on the incorrect behavior of cproj, then you can
1051 disable GCC's transformations using -fno-builtin-cproj.
1052 * The C-only intermodule optimization framework (IMA, enabled by
1053 -combine) has been removed in favor of the new generic link-time
1054 optimization framework (LTO) introduced in [2]GCC 4.5.0.
1055 * GCC now ships with the LGPL-licensed libquadmath library, which
1056 provides quad-precision mathematical functions for targets with a
1057 __float128 datatype. __float128 is available for targets on 32-bit
1058 x86, x86-64 and Itanium architectures. The libquadmath library is
1059 automatically built on such targets when building the Fortran
1061 * New -Wunused-but-set-variable and -Wunused-but-set-parameter
1062 warnings were added for C, C++, Objective-C and Objective-C++.
1063 These warnings diagnose variables respective parameters which are
1064 only set in the code and never otherwise used. Usually such
1065 variables are useless and often even the value assigned to them is
1066 computed needlessly, sometimes expensively. The
1067 -Wunused-but-set-variable warning is enabled by default by -Wall
1068 flag and -Wunused-but-set-parameter by -Wall -Wextra flags.
1069 * On ARM, a bug has been fixed in GCC's implementation of the AAPCS
1070 rules for the layout of vectors that could lead to wrong code being
1071 generated. Vectors larger than 8 bytes in size are now by default
1072 aligned to an 8-byte boundary. This is an ABI change: code that
1073 makes explicit use of vector types may be incompatible with binary
1074 objects built with older versions of GCC. Auto-vectorized code is
1075 not affected by this change. (This change affects GCC versions
1076 4.6.4 and later, with the exception of versions 4.7.0 and 4.7.1.)
1077 * On AVR, variables with the progmem attribute to locate data in
1078 flash memory must be qualified as const.
1079 * Support for a number of older systems and recently unmaintained or
1080 untested target ports of GCC has been declared obsolete in GCC 4.6.
1081 Unless there is activity to revive them, the next release of GCC
1082 will have their sources permanently removed.
1083 All GCC ports for the following processor architectures have been
1085 + Argonaut ARC (arc-*)
1086 + National Semiconductor CRX (crx-*)
1087 + Motorola 68HC11 and 68HC12 (m68hc11-*-*, m6811-*-*,
1088 m68hc12-*-*, m6812-*-*)
1089 + Sunplus S+core (score-*)
1090 The following ports for individual systems on particular
1091 architectures have been obsoleted:
1092 + Interix (i[34567]86-*-interix3*)
1093 + NetWare x86 (i[3456x]86-*-netware*)
1094 + Generic ARM PE (arm-*-pe* other than arm*-wince-pe*)
1095 + MCore PE (mcore-*-pe*)
1096 + SH SymbianOS (sh*-*-symbianelf*)
1097 + GNU Hurd on Alpha and PowerPC (alpha*-*-gnu*, powerpc*-*-gnu*)
1098 + M68K uClinux old ABI (m68k-*-uclinuxoldabi*)
1099 + a.out NetBSD (arm*-*-netbsd*, i[34567]86-*-netbsd*,
1100 vax-*-netbsd*, but not *-*-netbsdelf*)
1101 The i[34567]86-*-pe alias for Cygwin targets has also been
1102 obsoleted; users should configure for i[34567]86-*-cygwin* instead.
1103 Certain configure options to control the set of libraries built
1104 with GCC on some targets have been obsoleted. On ARM targets, the
1105 options --disable-fpu, --disable-26bit, --disable-underscore,
1106 --disable-interwork, --disable-biendian and --disable-nofmult have
1107 been obsoleted. On MIPS targets, the options
1108 --disable-single-float, --disable-biendian and --disable-softfloat
1109 have been obsoleted.
1110 * Support has been removed for all the [3]configurations obsoleted in
1112 * More information on porting to GCC 4.6 from previous versions of
1113 GCC can be found in the [4]porting guide for this release.
1115 General Optimizer Improvements
1117 * A new general optimization level, -Ofast, has been introduced. It
1118 combines the existing optimization level -O3 with options that can
1119 affect standards compliance but result in better optimized code.
1120 For example, -Ofast enables -ffast-math.
1121 * Link-time optimization improvements:
1122 + The [5]Scalable Whole Program Optimizer (WHOPR) project has
1123 stabilized to the point of being usable. It has become the
1124 default mode when using the LTO optimization model. Link time
1125 optimization can now split itself into multiple parallel
1126 compilations. Parallelism is controlled with -flto=n (where n
1127 specifies the number of compilations to execute in parallel).
1128 GCC can also cooperate with a GNU make job server by
1129 specifying the -flto=jobserver option and adding + to the
1130 beginning of the Makefile rule executing the linker.
1131 Classical LTO mode can be enforced by -flto-partition=none.
1132 This may result in small code quality improvements.
1133 + A large number of bugs were fixed. GCC itself, Mozilla Firefox
1134 and other large applications can be built with LTO enabled.
1135 + The linker plugin support improvements
1136 o Linker plugin is now enabled by default when the linker
1137 is detected to have plugin support. This is the case for
1138 GNU ld 2.21.51 or newer (on ELF and Cygwin targets) and
1139 the Gold linker on ELF targets. Plugin support of the
1140 Apple linker on Darwin is not compatible with GCC. The
1141 linker plugin can also be controlled by the
1142 -fuse-linker-plugin command line option.
1143 o Resolution information from the linker plugin is used to
1144 drive whole program assumptions. Use of the linker plugin
1145 results in more aggressive optimization on binaries and
1146 on shared libraries that use the hidden visibility
1147 attribute. Consequently the use of -fwhole-program is not
1148 necessary in addition to LTO.
1149 + Hidden symbols used from non-LTO objects now have to be
1150 explicitly annotated with externally_visible when the linker
1152 + C++ inline functions and virtual tables are now privatized
1153 more aggressively, leading to better inter-procedural
1154 optimization and faster dynamic linking.
1155 + Memory usage and intermediate language streaming performance
1157 + Static constructors and destructors from individual units are
1158 inlined into a single function. This can significantly improve
1159 startup times of large C++ applications where static
1160 constructors are very common. For example, static constructors
1161 are used when including the iostream header.
1162 + Support for the Ada language has been added.
1163 * Interprocedural optimization improvements
1164 + The interprocedural framework was re-tuned for link time
1165 optimization. Several scalability issues were resolved.
1166 + Improved auto-detection of const and pure functions. Newly,
1167 noreturn functions are auto-detected.
1168 The [6]-Wsuggest-attribute=[const|pure|noreturn] flag is
1169 available that informs users when adding attributes to headers
1170 might improve code generation.
1171 + A number of inlining heuristic improvements. In particular:
1172 o Partial inlining is now supported and enabled by default
1173 at -O2 and greater. The feature can be controlled via
1175 Partial inlining splits functions with short hot path to
1176 return. This allows more aggressive inlining of the hot
1177 path leading to better performance and often to code size
1178 reductions (because cold parts of functions are not
1180 o Scalability for large compilation units was improved
1182 o Inlining of callbacks is now more aggressive.
1183 o Virtual methods are considered for inlining when the
1184 caller is inlined and devirtualization is then possible.
1185 o Inlining when optimizing for size (either in cold regions
1186 of a program or when compiling with -Os) was improved to
1187 better handle C++ programs with larger abstraction
1188 penalty, leading to smaller and faster code.
1189 + The IPA reference optimization pass detecting global variables
1190 used or modified by functions was strengthened and sped up.
1191 + Functions whose address was taken are now optimized out when
1192 all references to them are dead.
1193 + A new inter-procedural static profile estimation pass detects
1194 functions that are executed once or unlikely to be executed.
1195 Unlikely executed functions are optimized for size. Functions
1196 executed once are optimized for size except for the inner
1198 + On most targets with named section support, functions used
1199 only at startup (static constructors and main), functions used
1200 only at exit and functions detected to be cold are placed into
1201 separate text segment subsections. This extends the
1202 -freorder-functions feature and is controlled by the same
1203 switch. The goal is to improve the startup time of large C++
1205 Proper function placement requires linker support. GNU ld
1206 2.21.51 on ELF targets was updated to place those functions
1207 together within the text section leading to better code
1208 locality and faster startup times of large C++ programs. The
1209 feature is also supported in the Apple linker. Support in the
1210 gold linker is planned.
1211 * A new switch -fstack-usage has been added. It makes the compiler
1212 output stack usage information for the program, on a per-function
1213 basis, in an auxiliary file.
1214 * A new switch -fcombine-stack-adjustments has been added. It can be
1215 used to enable or disable the compiler's stack-slot combining pass
1216 which before was enabled automatically at -O1 and above, but could
1217 not be controlled on its own.
1218 * A new switch -fstrict-volatile-bitfields has been added. Using it
1219 indicates that accesses to volatile bitfields should use a single
1220 access of the width of the field's type. This option can be useful
1221 for precisely defining and accessing memory-mapped peripheral
1222 registers from C or C++.
1224 Compile time and memory usage improvements
1226 * Datastructures used by the dataflow framework in GCC were
1227 reorganized for better memory usage and more cache locality.
1228 Compile time is improved especially on units with large functions
1229 (possibly resulting from a lot of inlining) not fitting into the
1230 processor cache. The compile time of the GCC C compiler binary with
1231 link-time optimization went down by over 10% (benchmarked on x86-64
1234 New Languages and Language specific improvements
1238 * Stack checking has been improved on selected architectures (Alpha,
1239 IA-32/x86-64, RS/6000 and SPARC): it now will detect stack
1240 overflows in all cases on these architectures.
1241 * Initial support for Ada 2012 has been added.
1245 * A new warning, enabled by -Wdouble-promotion, has been added that
1246 warns about cases where a value of type float is implicitly
1247 promoted to double. This is especially helpful for CPUs that handle
1248 the former in hardware, but emulate the latter in software.
1249 * A new function attribute leaf was introduced. This attribute allows
1250 better inter-procedural optimization across calls to functions that
1251 return to the current unit only via returning or exception
1252 handling. This is the case for most library functions that have no
1254 * Support for a new data type __int128 for targets having wide enough
1255 machine-mode support.
1256 * The new function attribute callee_pop_aggregate allows to specify
1257 if the caller or callee is responsible for popping the aggregate
1258 return pointer value from the stack.
1259 * Support for selectively enabling and disabling warnings via #pragma
1260 GCC diagnostic has been added. For instance:
1261 #pragma GCC diagnostic error "-Wuninitialized"
1262 foo(a); /* error is given for this one */
1263 #pragma GCC diagnostic push
1264 #pragma GCC diagnostic ignored "-Wuninitialized"
1265 foo(b); /* no diagnostic for this one */
1266 #pragma GCC diagnostic pop
1267 foo(c); /* error is given for this one */
1268 #pragma GCC diagnostic pop
1269 foo(d); /* depends on command line options */
1271 * The -fmax-errors=N option is now supported. Using this option
1272 causes the compiler to exit after N errors have been issued.
1276 * There is now experimental support for some features from the
1277 upcoming C1X revision of the ISO C standard. This support may be
1278 selected with -std=c1x, or -std=gnu1x for C1X with GNU extensions.
1279 Note that this support is experimental and may change incompatibly
1280 in future releases for consistency with changes to the C1X standard
1281 draft. The following features are newly supported as described in
1282 the N1539 draft of C1X (with changes agreed at the March 2011 WG14
1283 meeting); some other features were already supported with no
1284 compiler changes being needed, or have some support but not in full
1285 accord with N1539 (as amended).
1286 + Static assertions (_Static_assert keyword)
1287 + Typedef redefinition
1288 + New macros in <float.h>
1289 + Anonymous structures and unions
1290 * The new -fplan9-extensions option directs the compiler to support
1291 some extensions for anonymous struct fields which are implemented
1292 by the Plan 9 compiler. A pointer to a struct may be automatically
1293 converted to a pointer to an anonymous field when calling a
1294 function, in order to make the types match. An anonymous struct
1295 field whose type is a typedef name may be referred to using the
1300 * Improved [7]experimental support for the upcoming C++0x ISO C++
1301 standard, including support for constexpr (thanks to Gabriel Dos
1302 Reis and Jason Merrill), nullptr (thanks to Magnus Fromreide),
1303 noexcept, unrestricted unions, range-based for loops (thanks to
1304 Rodrigo Rivas Costa), opaque enum declarations (thanks also to
1305 Rodrigo), implicitly deleted functions and implicit move
1307 * When an extern declaration within a function does not match a
1308 declaration in the enclosing context, G++ now properly declares the
1309 name within the namespace of the function rather than the namespace
1310 which was open just before the function definition ([8]c++/43145).
1311 * GCC now warns by default when casting integers to larger pointer
1312 types. These warnings can be disabled with the option
1313 -Wno-int-to-pointer-cast, which is now also available in C++.
1314 * G++ no longer optimizes using the assumption that a value of
1315 enumeration type will fall within the range specified by the
1316 standard, since that assumption is easily violated with a
1317 conversion from integer type ([9]c++/43680). The old behavior can
1318 be restored with -fstrict-enums.
1319 * The new -fnothrow-opt flag changes the semantics of a throw()
1320 exception specification to match the proposed semantics of the
1321 noexcept specification: just call terminate if an exception tries
1322 to propagate out of a function with such an exception
1323 specification. This dramatically reduces or eliminates the code
1324 size overhead from adding the exception specification.
1325 * The new -Wnoexcept flag will suggest adding a noexcept qualifier to
1326 a function that the compiler can tell doesn't throw if it would
1327 change the value of a noexcept expression.
1328 * The -Wshadow option now warns if a local variable or type
1329 declaration shadows another type in C++. Note that the compiler
1330 will not warn if a local variable shadows a struct/class/enum, but
1331 will warn if it shadows an explicit typedef.
1332 * When an identifier is not found in the current scope, G++ now
1333 offers suggestions about which identifier might have been intended.
1334 * G++ now issues clearer diagnostics for missing semicolons after
1335 class, struct, and union definitions.
1336 * G++ now issues clearer diagnostics for missing semicolons after
1337 class member declarations.
1338 * G++ now issues clearer diagnostics when a colon is used in a place
1339 where a double-colon was intended.
1340 * G++ no longer accepts mutable on reference members ([10]c++/33558).
1341 Use -fpermissive to allow the old, non-conforming behaviour.
1342 * A few mangling fixes have been made, to attribute const/volatile on
1343 function pointer types, decltype of a plain decl, and use of a
1344 function parameter in the declaration of another parameter. By
1345 default the compiler still uses the old mangling, but emits aliases
1346 with the new mangling on targets that support strong aliases. Users
1347 can switch over entirely to the new mangling with -fabi-version=5
1348 or -fabi-version=0. -Wabi will now warn about code that uses the
1350 * In 4.6.0 and 4.6.1 G++ no longer allows objects of const-qualified
1351 type to be default initialized unless the type has a user-declared
1352 default constructor. In 4.6.2 G++ implements the proposed
1353 resolution of [11]DR 253, so default initialization is allowed if
1354 it initializes all subobjects. Code that fails to compile can be
1355 fixed by providing an initializer e.g.
1357 struct B : A { int i; };
1359 Use -fpermissive to allow the old, non-conforming behaviour.
1361 Runtime Library (libstdc++)
1363 * [12]Improved experimental support for the upcoming ISO C++
1364 standard, C++0x, including using constexpr and nullptr.
1365 * Performance improvements to the [13]Debug Mode, thanks to Franc,ois
1367 * Atomic operations used for reference-counting are annotated so that
1368 they can be understood by race detectors such as Helgrind, see
1369 [14]Data Race Hunting.
1370 * Most libstdc++ standard headers have been changed to no longer
1371 include the cstddef header as an implementation detail. Code that
1372 relied on that header being included as side-effect of including
1373 other standard headers will need to include cstddef explicitly.
1377 * On systems supporting the libquadmath library, GNU Fortran now also
1378 supports a quad-precision, kind=16 floating-point data type
1379 (REAL(16), COMPLEX(16)). As the data type is not fully supported in
1380 hardware, calculations might be one to two orders of magnitude
1381 slower than with the 4, 8 or 10 bytes floating-point data types.
1382 This change does not affect systems which support REAL(16) in
1383 hardware nor those which do not support libquadmath.
1384 * Much improved compile time for large array constructors.
1385 * In order to reduce execution time and memory consumption, use of
1386 temporary arrays in assignment expressions is avoided for many
1387 cases. The compiler now reverses loops in order to avoid generating
1388 a temporary array where possible.
1389 * Improved diagnostics, especially with -fwhole-file.
1390 * The -fwhole-file flag is now enabled by default. This improves code
1391 generation and diagnostics. It can be disabled using the deprecated
1392 -fno-whole-file flag.
1393 * Support the generation of Makefile dependencies via the [15]-M...
1394 flags of GCC; you may need to specify the -cpp option in addition.
1395 The dependencies take modules, Fortran's include, and CPP's
1396 #include into account. Note: Using -M for the module path is no
1397 longer supported, use -J instead.
1398 * The flag -Wconversion has been modified to only issue warnings
1399 where a conversion leads to information loss. This drastically
1400 reduces the number of warnings; -Wconversion is thus now enabled
1401 with -Wall. The flag -Wconversion-extra has been added and also
1402 warns about other conversions; -Wconversion-extra typically issues
1403 a huge number of warnings, most of which can be ignored.
1404 * A new command-line option -Wunused-dummy-argument warns about
1405 unused dummy arguments and is included in -Wall. Before,
1406 -Wunused-variable also warned about unused dummy arguments.
1407 * Fortran 2003 support has been extended:
1408 + Improved support for polymorphism between libraries and
1409 programs and for complicated inheritance patterns (cf.
1410 [16]object-oriented programming).
1411 + Experimental support of the ASSOCIATE construct.
1412 + In pointer assignments it is now possible to specify the lower
1413 bounds of the pointer and, for a rank-1 or a simply contiguous
1414 data-target, to remap the bounds.
1415 + Automatic (re)allocation: In intrinsic assignments to
1416 allocatable variables the left-hand side will be automatically
1417 allocated (if unallocated) or reallocated (if the shape or
1418 type parameter is different). To avoid the small performance
1419 penalty, you can use a(:) = ... instead of a = ... for arrays
1420 and character strings - or disable the feature using -std=f95
1421 or -fno-realloc-lhs.
1422 + Deferred type parameter: For scalar allocatable and pointer
1423 variables the character length can be deferred.
1424 + Namelist variables with allocatable and pointer attribute and
1425 nonconstant length type parameter are supported.
1426 * Fortran 2008 support has been extended:
1427 + Experimental [17]coarray support (for one image only, i.e.
1428 num_images() == 1); use the [18]-fcoarray=single flag to
1430 + The STOP and the new ERROR STOP statements now support all
1431 constant expressions.
1432 + Support for the CONTIGUOUS attribute.
1433 + Support for ALLOCATE with MOLD.
1434 + Support for the STORAGE_SIZE intrinsic inquiry function.
1435 + Support of the NORM2 and PARITY intrinsic functions.
1436 + The following bit intrinsics were added: POPCNT and POPPAR for
1437 counting the number of 1 bits and returning the parity; BGE,
1438 BGT, BLE, and BLT for bitwise comparisons; DSHIFTL and DSHIFTR
1439 for combined left and right shifts, MASKL and MASKR for simple
1440 left and right justified masks, MERGE_BITS for a bitwise merge
1441 using a mask, SHIFTA, SHIFTL and SHIFTR for shift operations,
1442 and the transformational bit intrinsics IALL, IANY and
1444 + Support of the EXECUTE_COMMAND_LINE intrinsic subroutine.
1445 + Support for the IMPURE attribute for procedures, which allows
1446 for ELEMENTAL procedures without the restrictions of PURE.
1447 + Null pointers (including NULL()) and not allocated variables
1448 can be used as actual argument to optional non-pointer,
1449 non-allocatable dummy arguments, denoting an absent argument.
1450 + Non-pointer variables with TARGET attribute can be used as
1451 actual argument to POINTER dummies with INTENT(IN)
1452 + Pointers including procedure pointers and those in a derived
1453 type (pointer components) can now be initialized by a target
1454 instead of only by NULL.
1455 + The EXIT statement (with construct-name) can now be used to
1456 leave not only the DO but also the ASSOCIATE, BLOCK, IF,
1457 SELECT CASE and SELECT TYPE constructs.
1458 + Internal procedures can now be used as actual argument.
1459 + The named constants INTEGER_KINDS, LOGICAL_KINDS, REAL_KINDS
1460 and CHARACTER_KINDS of the intrinsic module ISO_FORTRAN_ENV
1461 have been added; these arrays contain the supported kind
1462 values for the respective types.
1463 + The module procedures C_SIZEOF of the intrinsic module
1464 ISO_C_BINDINGS and COMPILER_VERSION and COMPILER_OPTIONS of
1465 ISO_FORTRAN_ENV have been implemented.
1466 + Minor changes: obsolescence diagnostics for ENTRY was added
1467 for -std=f2008; a line may start with a semicolon; for
1468 internal and module procedures END can be used instead of END
1469 SUBROUTINE and END FUNCTION; SELECTED_REAL_KIND now also takes
1470 a RADIX argument; intrinsic types are supported for
1471 TYPE(intrinsic-type-spec); multiple type-bound procedures can
1472 be declared in a single PROCEDURE statement; implied-shape
1473 arrays are supported for named constants (PARAMETER). The
1474 transformational, three argument versions of BESSEL_JN and
1475 BESSEL_YN were added - the elemental, two-argument version had
1476 been added in GCC 4.4; note that the transformational
1477 functions use a recurrence algorithm.
1481 Support for the [19]Go programming language has been added to GCC. It
1482 is not enabled by default when you build GCC; use the
1483 --enable-languages configure option to build it. The driver program for
1484 compiling Go code is gccgo.
1486 Go is currently known to work on GNU/Linux and RTEMS. Solaris support
1487 is in progress. It may or may not work on other platforms.
1489 Objective-C and Objective-C++
1491 * The -fobjc-exceptions flag is now required to enable Objective-C
1492 exception and synchronization syntax (introduced by the keywords
1493 @try, @catch, @finally and @synchronized).
1494 * A number of Objective-C 2.0 features and extensions are now
1495 supported by GCC. These features are enabled by default; you can
1496 disable them by using the new -fobjc-std=objc1 command-line option.
1497 * The Objective-C 2.0 dot-syntax is now supported. It is an
1498 alternative syntax for using getters and setters; object.count is
1499 automatically converted into [object count] or [object setCount:
1500 ...] depending on context; for example if (object.count > 0) is
1501 automatically compiled into the equivalent of if ([object count] >
1502 0) while object.count = 0; is automatically compiled into the
1503 equivalent ot [object setCount: 0];. The dot-syntax can be used
1504 with instance and class objects and with any setters or getters, no
1505 matter if they are part of a declared property or not.
1506 * Objective-C 2.0 declared properties are now supported. They are
1507 declared using the new @property keyword, and are most commonly
1508 used in conjunction with the new Objective-C 2.0 dot-syntax. The
1509 nonatomic, readonly, readwrite, assign, retain, copy, setter and
1510 getter attributes are all supported. Marking declared properties
1511 with __attribute__ ((deprecated)) is supported too.
1512 * The Objective-C 2.0 @synthesize and @dynamic keywords are
1513 supported. @synthesize causes the compiler to automatically
1514 synthesize a declared property, while @dynamic is used to disable
1515 all warnings for a declared property for which no implementation is
1516 provided at compile time. Synthesizing declared properties requires
1517 runtime support in most useful cases; to be able to use it with the
1518 GNU runtime, appropriate helper functions have been added to the
1519 GNU Objective-C runtime ABI, and are implemented by the GNU
1520 Objective-C runtime library shipped with GCC.
1521 * The Objective-C 2.0 fast enumeration syntax is supported in
1522 Objective-C. This is currently not yet available in Objective-C++.
1523 Fast enumeration requires support in the runtime, and such support
1524 has been added to the GNU Objective-C runtime library (shipped with
1526 * The Objective-C 2.0 @optional keyword is supported. It allows you
1527 to mark methods or properties in a protocol as optional as opposed
1529 * The Objective-C 2.0 @package keyword is supported. It has currently
1530 the same effect as the @public keyword.
1531 * Objective-C 2.0 method attributes are supported. Currently the
1532 supported attributes are deprecated, sentinel, noreturn and format.
1533 * Objective-C 2.0 method argument attributes are supported. The most
1534 widely used attribute is unused, to mark an argument as unused in
1536 * Objective-C 2.0 class and protocol attributes are supported.
1537 Currently the only supported attribute is deprecated.
1538 * Objective-C 2.0 class extensions are supported. A class extension
1539 has the same syntax as a category declaration with no category
1540 name, and the methods and properties declared in it are added
1541 directly to the main class. It is mostly used as an alternative to
1542 a category to add methods to a class without advertising them in
1543 the public headers, with the advantage that for class extensions
1544 the compiler checks that all the privately declared methods are
1545 actually implemented.
1546 * As a result of these enhancements, GCC can now be used to build
1547 Objective-C and Objective-C++ software that uses Foundation and
1548 other important system frameworks with the NeXT runtime on Darwin 9
1549 and Darwin 10 (OSX 10.5 and 10.6).
1550 * Many bugs in the compiler have been fixed in this release; in
1551 particular, LTO can now be used when compiling Objective-C and
1552 Objective-C++ and the parser is much more robust in dealing with
1555 Runtime Library (libobjc)
1557 * The GNU Objective-C runtime library now defines the macro
1558 __GNU_LIBOBJC__ (with a value that is increased at every release
1559 where there is any change to the API) in objc/objc.h, making it
1560 easy to determine if the GNU Objective-C runtime library is being
1561 used, and if so, which version. Previous versions of the GNU
1562 Objective-C runtime library (and other Objective-C runtime
1563 libraries such as the Apple one) do not define this macro.
1564 * A new Objective-C 2.0 API, almost identical to the one implemented
1565 by the Apple Objective-C runtime, has been implemented in the GNU
1566 Objective-C runtime library. The new API hides the internals of
1567 most runtime structures but provides a more extensive set of
1568 functions to operate on them. It is much easier, for example, to
1569 create or modify classes at runtime. The new API also makes it
1570 easier to port software from Apple to GNU as almost no changes
1571 should be required. The old API is still supported for backwards
1572 compatibility; including the old objc/objc-api.h header file
1573 automatically selects the old API, while including the new
1574 objc/runtime.h header file automatically selects the new API.
1575 Support for the old API is being phased out and upgrading the
1576 software to use the new API is strongly recommended. To check for
1577 the availability of the new API, the __GNU_LIBOBJC__ macro can be
1578 used as older versions of the GNU Objective-C runtime library,
1579 which do not support the new API, do not define such a macro.
1580 * Runtime support for @synchronized has been added.
1581 * Runtime support for Objective-C 2.0 synthesized property accessors
1583 * Runtime support for Objective-C 2.0 fast enumeration has been
1586 New Targets and Target Specific Improvements
1590 * GCC now supports the Cortex-M4 processor implementing the v7-em
1591 version of the architecture using the option -mcpu=cortex-m4.
1592 * Scheduling descriptions for the Cortex-M4, the Neon and the
1593 floating point units of the Cortex-A9 and a pipeline description
1594 for the Cortex-A5 have been added.
1595 * Synchronization primitives such as __sync_fetch_and_add and friends
1596 are now inlined for supported architectures rather than calling
1597 into a kernel helper function.
1598 * SSA loop prefetching is enabled by default for the Cortex-A9 at
1600 * Several improvements were committed to improve code generation for
1601 the ARM architecture including a rewritten implementation for load
1602 and store multiples.
1603 * Several enhancements were committed to improve SIMD code generation
1604 for NEON by adding support for widening instructions, misaligned
1605 loads and stores, vector conditionals and support for 64 bit
1607 * Support was added for the Faraday cores fa526, fa606te, fa626te,
1608 fmp626te, fmp626 and fa726te and can be used with the respective
1609 names as parameters to the -mcpu= option.
1610 * Basic support was added for Cortex-A15 and is available through
1612 * GCC for AAPCS configurations now more closely adheres to the AAPCS
1613 specification by enabling -fstrict-volatile-bitfields by default.
1617 * The new -fsplit-stack option permits programs to use a
1618 discontiguous stack. This is useful for threaded programs, in that
1619 it is no longer necessary to specify the maximum stack size when
1620 creating a thread. This feature is currently only implemented for
1621 32-bit and 64-bit x86 GNU/Linux targets.
1622 * Support for emitting profiler counter calls before function
1623 prologues. This is enabled via a new command-line option -mfentry.
1624 * Optimization for the Intel Core 2 processors is now available
1625 through the -march=core2 and -mtune=core2 options.
1626 * Support for Intel Core i3/i5/i7 processors is now available through
1627 the -march=corei7 and -mtune=corei7 options.
1628 * Support for Intel Core i3/i5/i7 processors with AVX is now
1629 available through the -march=corei7-avx and -mtune=corei7-avx
1631 * Support for AMD Bobcat (family 14) processors is now available
1632 through the -march=btver1 and -mtune=btver1 options.
1633 * Support for AMD Bulldozer (family 15) processors is now available
1634 through the -march=bdver1 and -mtune=bdver1 options.
1635 * The default setting (when not optimizing for size) for 32-bit
1636 GNU/Linux and Darwin x86 targets has been changed to
1637 -fomit-frame-pointer. The default can be reverted to
1638 -fno-omit-frame-pointer by configuring GCC with the
1639 --enable-frame-pointer configure option.
1640 * Darwin, FreeBSD, Solaris 2, MinGW and Cygwin now all support
1641 __float128 on 32-bit and 64-bit x86 targets.
1642 * AVX floating-point arithmetic can now be enabled by default at
1643 configure time with the new --with-fpmath=avx option.
1644 * The SSA loop prefetching pass is enabled when using -O3 when
1645 optimizing for CPUs where prefetching is beneficial (AMD CPUs newer
1647 * Support for TBM (Trailing Bit Manipulation) built-in functions and
1648 code generation is available via -mtbm.
1649 * Support for AMD's BMI (Bit Manipulation) built-in functions and
1650 code generation is available via -mbmi.
1654 * Support has been added for the Xilinx MicroBlaze softcore processor
1655 (microblaze-elf) embedded target. This configurable processor is
1656 supported on several Xilinx Spartan and Virtex FPGAs.
1660 * GCC now supports the Loongson 3A processor. Its canonical -march=
1661 and -mtune= name is loongson3a.
1665 * The inline assembly register constraint "A" has been renamed "c".
1666 This constraint is used to select a floating-point register that
1667 can be used as the destination of a multiply-accumulate
1669 * New inline assembly register constraints "A" and "D" have been
1670 added. These constraint letters resolve to all general registers
1671 when compiling for AM33, and resolve to address registers only or
1672 data registers only when compiling for MN10300.
1673 * The MDR register is represented in the compiler. One can access the
1674 register via the "z" constraint in inline assembly. It can be
1675 marked as clobbered or used as a local register variable via the
1676 "mdr" name. The compiler uses the RETF instruction if the function
1677 does not modify the MDR register, so it is important that inline
1678 assembly properly annotate any usage of the register.
1682 * GCC now supports the Applied Micro Titan processor with
1684 * The -mrecip option has been added, which indicates whether the
1685 reciprocal and reciprocal square root instructions should be used.
1686 * The -mveclibabi=mass option can be used to enable the compiler to
1687 autovectorize mathematical functions using the Mathematical
1688 Acceleration Subsystem library.
1689 * The -msingle-pic-base option has been added, which instructs the
1690 compiler to avoid loading the PIC base register in function
1691 prologues. The PIC base register must be initialized by the runtime
1693 * The -mblock-move-inline-limit option has been added, which enables
1694 the user to control the maximum size of inlined memcpy calls and
1696 * PowerPC64 GNU/Linux support for applications requiring a large TOC
1697 section has been improved. A new command-line option,
1698 -mcmodel=MODEL, controls this feature; valid values for MODEL are
1699 small, medium, or large.
1700 * The Altivec builtin functions vec_ld and vec_st have been modified
1701 to generate the Altivec memory instructions LVX and STVX, even if
1702 the -mvsx option is used. In the initial GCC 4.5 release, these
1703 builtin functions were changed to generate VSX memory reference
1704 instructions instead of Altivec memory instructions, but there are
1705 differences between the two instructions. If the VSX instruction
1706 set is available, you can now use the new builtin functions
1707 vec_vsx_ld and vec_vsx_st which always generates the VSX memory
1709 * The GCC compiler on AIX now defaults to a process layout with a
1710 larger data space allowing larger programs to be compiled.
1711 * The GCC long double type on AIX 6.1 and above has reverted to 64
1712 bit double precision, matching the AIX XL compiler default, because
1713 of missing C99 symbols required by the GCC runtime.
1714 * The default processor scheduling model and tuning for PowerPC64
1715 GNU/Linux and for AIX 6.1 and above now is POWER7.
1716 * Starting with GCC 4.6.1, vectors of type vector long long or vector
1717 long are passed and returned in the same method as other vectors
1718 with the VSX instruction set. Previously the GCC compiler did not
1719 adhere to the ABI for 128-bit vectors with 64-bit integer base
1720 types (PR 48857). This is also fixed in the GCC 4.5.4 release.
1722 S/390, zSeries and System z9/z10, IBM zEnterprise z196
1724 * Support for the zEnterprise z196 processor has been added. When
1725 using the -march=z196 option, the compiler will generate code
1726 making use of the following instruction facilities:
1727 + Conditional load/store
1729 + Floating-point-extension
1730 + Interlocked-access
1732 The -mtune=z196 option avoids the compare and branch instructions
1733 as well as the load address instruction with an index register as
1734 much as possible and performs instruction scheduling appropriate
1735 for the new out-of-order pipeline architecture.
1736 * When using the -m31 -mzarch options the generated code still
1737 conforms to the 32-bit ABI but uses the general purpose registers
1738 as 64-bit registers internally. This requires a Linux kernel saving
1739 the whole 64-bit registers when doing a context switch. Kernels
1740 providing that feature indicate that by the 'highgprs' string in
1742 * The SSA loop prefetching pass is enabled when using -O3.
1746 * GCC now supports the LEON series of SPARC V8 processors. The code
1747 generated by the compiler can either be tuned to it by means of the
1748 --with-tune=leon configure option and -mtune=leon compilation
1749 option, or the compiler can be built for the sparc-leon-{elf,linux}
1750 and sparc-leon3-{elf,linux} targets directly.
1751 * GCC has stopped sign/zero-extending parameter registers in the
1752 callee for functions taking parameters with sub-word size in 32-bit
1753 mode, since this is redundant with the specification of the ABI.
1754 GCC has never done so in 64-bit mode since this is also redundant.
1755 * The command line option -mfix-at697f has been added to enable the
1756 documented workaround for the single erratum of the Atmel AT697F
1763 * GCC now supports the Bionic C library and provides a convenient way
1764 of building native libraries and applications for the Android
1765 platform. Refer to the documentation of the -mandroid and -mbionic
1766 options for details on building native code. At the moment, Android
1767 support is enabled only for ARM.
1772 + Initial support for CFString types has been added.
1773 This allows GCC to build projects including the system Core
1774 Foundation frameworks. The GCC Objective-C family supports
1775 CFString "toll-free bridged" as per the Mac OS X system tools.
1776 CFString is also recognized in the context of format
1777 attributes and arguments (see the documentation for format
1778 attributes for limitations). At present, 8-bit character types
1780 + Object file size reduction.
1781 The Darwin zeroed memory allocators have been re-written to
1782 make more use of .zerofill sections. For non-debug code, this
1783 can reduce object file size significantly.
1784 + Objective-C family 64-bit support (NeXT ABI 2).
1785 Initial support has been added to support 64-bit Objective-C
1786 code using the Darwin/OS X native (NeXT) runtime. ABI version
1787 2 will be selected automatically when 64-bit code is built.
1788 + Objective-C family 32-bit ABI 1.
1789 For 32-bit code ABI 1 is also now also allowed. At present it
1790 must be selected manually using -fobjc-abi-version=1 where
1791 applicable - i.e. on Darwin 9/10 (OS X 10.5/10.6).
1793 + The -mdynamic-no-pic option has been enabled.
1794 Code supporting -mdynamic-no-pic optimization has been added
1795 and is applicable to -m32 builds. The compiler bootstrap uses
1796 the option where appropriate.
1797 + The default value for -mtune= has been changed.
1798 Since Darwin systems are primarily Xeon, Core-2 or similar the
1799 default tuning has been changed to -mtune=core2.
1800 + Enable 128-bit long double (__float128) support on Darwin.
1803 Several significant bugs have been fixed, such that GCC now
1804 produces code compatible with the Darwin64 PowerPC ABI.
1805 + libffi and boehm-gc.
1806 The Darwin ports of the libffi and boehm-gc libraries have
1807 been upgraded to include a Darwin64 implementation. This means
1808 that powerpc*-*-darwin9 platforms may now, for example, build
1809 Java applications with -m64 enabled.
1810 + Plug-in support has been enabled.
1811 + The -fsection-anchors option is now available although,
1812 presently, not heavily tested.
1818 * Support symbol versioning with the Sun linker.
1819 * Allow libstdc++ to leverage full ISO C99 support on Solaris 10+.
1820 * Support thread-local storage (TLS) with the Sun assembler on
1822 * Support TLS on Solaris 8/9 if prerequisites are met.
1823 * Support COMDAT group with the GNU assembler and recent Sun linker.
1824 * Support the Sun assembler visibility syntax.
1825 * Default Solaris 2/x86 to -march=pentium4 (Solaris 10+) resp.
1826 -march=pentiumpro (Solaris 8/9).
1827 * Don't use SSE on Solaris 8/9 x86 by default.
1828 * Enable 128-bit long double (__float128) support on Solaris 2/x86.
1832 * Change the ABI for returning 8-byte vectors like __m64 in MMX
1833 registers on Solaris 10+/x86 to match the Sun Studio 12.1+
1834 compilers. This is an incompatible change. If you use such types,
1835 you must either recompile all your code with the new compiler or
1836 use the new -mvect8-ret-in-mem option to remain compatible with
1837 previous versions of GCC and Sun Studio.
1841 * Initial support for decimal floating point.
1842 * Support for the __thiscall calling-convention.
1843 * Support for hot-patchable function prologues via the
1844 ms_hook_prologue attribute for x86_64 in addition to 32-bit x86.
1845 * Improvements of stack-probing and stack-allocation mechanisms.
1846 * Support of push/pop-macro pragma as preprocessor command.
1847 With #pragma push_macro("macro-name") the current definition of
1848 macro-name is saved and can be restored with #pragma
1849 pop_macro("macro-name") to its saved definition.
1850 * Enable 128-bit long double (__float128) support on MinGW and
1853 Other significant improvements
1855 Installation changes
1857 * An install-strip make target is provided that installs stripped
1858 executables, and may install libraries with unneeded or debugging
1860 * On Power7 systems, there is a potential problem if you build the
1861 GCC compiler with a host compiler using options that enable the VSX
1862 instruction set generation. If the host compiler has been patched
1863 so that the vec_ld and vec_st builtin functions generate Altivec
1864 memory instructions instead of VSX memory instructions, then you
1865 should be able to build the compiler with VSX instruction
1868 Changes for GCC Developers
1870 Note: these changes concern developers that develop GCC itself or
1871 software that integrates with GCC, such as plugins, and not the general
1873 * The gengtype utility, which previously was internal to the GCC
1874 build process, has been enchanced to provide GC root information
1875 for plugins as necessary.
1876 * The old GC allocation interface of ggc_alloc and friends was
1877 replaced with a type-safe alternative.
1881 This is the [20]list of problem reports (PRs) from GCC's bug tracking
1882 system that are known to be fixed in the 4.6.1 release. This list might
1883 not be complete (that is, it is possible that some PRs that have been
1884 fixed are not listed here).
1888 This is the [21]list of problem reports (PRs) from GCC's bug tracking
1889 system that are known to be fixed in the 4.6.2 release. This list might
1890 not be complete (that is, it is possible that some PRs that have been
1891 fixed are not listed here).
1895 This is the [22]list of problem reports (PRs) from GCC's bug tracking
1896 system that are known to be fixed in the 4.6.3 release. This list might
1897 not be complete (that is, it is possible that some PRs that have been
1898 fixed are not listed here).
1901 For questions related to the use of GCC, please consult these web
1902 pages and the [23]GCC manuals. If that fails, the
1903 [24]gcc-help@gcc.gnu.org mailing list might help. Comments on these
1904 web pages and the development of GCC are welcome on our developer
1905 list at [25]gcc@gcc.gnu.org. All of [26]our lists have public
1908 Copyright (C) [27]Free Software Foundation, Inc. Verbatim copying and
1909 distribution of this entire article is permitted in any medium,
1910 provided this notice is preserved.
1912 These pages are [28]maintained by the GCC team. Last modified
1917 1. http://sourceware.org/bugzilla/show_bug.cgi?id=10401
1918 2. http://gcc.gnu.org/gcc-4.5/changes.html
1919 3. http://gcc.gnu.org/gcc-4.5/changes.html#obsoleted
1920 4. http://gcc.gnu.org/gcc-4.6/porting_to.html
1921 5. http://gcc.gnu.org/projects/lto/whopr.pdf
1922 6. http://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html#Warning-Options
1923 7. http://gcc.gnu.org/gcc-4.6/cxx0x_status.html
1924 8. http://gcc.gnu.org/PR43145
1925 9. http://gcc.gnu.org/PR43680
1926 10. http://gcc.gnu.org/PR33558
1927 11. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_active.html#253
1928 12. http://gcc.gnu.org/onlinedocs/libstdc++/manual/status.html#status.iso.200x
1929 13. http://gcc.gnu.org/onlinedocs/libstdc++/manual/debug_mode.html
1930 14. http://gcc.gnu.org/onlinedocs/libstdc++/manual/debug.html#debug.races
1931 15. http://gcc.gnu.org/onlinedocs/gcc/Preprocessor-Options.html
1932 16. http://gcc.gnu.org/wiki/OOP
1933 17. http://gcc.gnu.org/wiki/Coarray
1934 18. http://gcc.gnu.org/onlinedocs/gfortran/Code-Gen-Options.html#index-g_t_0040code_007bfcoarray_007d-233
1935 19. http://golang.org/
1936 20. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.6.1
1937 21. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.6.2
1938 22. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.6.3
1939 23. http://gcc.gnu.org/onlinedocs/
1940 24. mailto:gcc-help@gcc.gnu.org
1941 25. mailto:gcc@gcc.gnu.org
1942 26. http://gcc.gnu.org/lists.html
1943 27. http://www.fsf.org/
1944 28. http://gcc.gnu.org/about.html
1945 29. http://validator.w3.org/check/referer
1946 ======================================================================
1947 http://gcc.gnu.org/gcc-4.5/index.html
1949 GCC 4.5 Release Series
1953 The [1]GNU project and the GCC developers are pleased to announce the
1954 release of GCC 4.5.4.
1956 This release is a bug-fix release, containing fixes for regressions in
1957 GCC 4.5.3 relative to previous releases of GCC.
1962 Jul 2, 2012 ([2]changes)
1965 Apr 28, 2011 ([3]changes)
1968 Dec 16, 2010 ([4]changes)
1971 Jul 31, 2010 ([5]changes)
1974 April 14, 2010 ([6]changes)
1976 References and Acknowledgements
1978 GCC used to stand for the GNU C Compiler, but since the compiler
1979 supports several other languages aside from C, it now stands for the
1980 GNU Compiler Collection.
1982 A list of [7]successful builds is updated as new information becomes
1985 The GCC developers would like to thank the numerous people that have
1986 contributed new features, improvements, bug fixes, and other changes as
1987 well as test results to GCC. This [8]amazing group of volunteers is
1988 what makes GCC successful.
1990 For additional information about GCC please refer to the [9]GCC project
1991 web site or contact the [10]GCC development mailing list.
1993 To obtain GCC please use [11]our mirror sites or [12]our SVN server.
1996 For questions related to the use of GCC, please consult these web
1997 pages and the [13]GCC manuals. If that fails, the
1998 [14]gcc-help@gcc.gnu.org mailing list might help. Comments on these
1999 web pages and the development of GCC are welcome on our developer
2000 list at [15]gcc@gcc.gnu.org. All of [16]our lists have public
2003 Copyright (C) [17]Free Software Foundation, Inc. Verbatim copying and
2004 distribution of this entire article is permitted in any medium,
2005 provided this notice is preserved.
2007 These pages are [18]maintained by the GCC team. Last modified
2012 1. http://www.gnu.org/
2013 2. http://gcc.gnu.org/gcc-4.5/changes.html
2014 3. http://gcc.gnu.org/gcc-4.5/changes.html
2015 4. http://gcc.gnu.org/gcc-4.5/changes.html
2016 5. http://gcc.gnu.org/gcc-4.5/changes.html
2017 6. http://gcc.gnu.org/gcc-4.5/changes.html
2018 7. http://gcc.gnu.org/gcc-4.5/buildstat.html
2019 8. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
2020 9. http://gcc.gnu.org/index.html
2021 10. mailto:gcc@gcc.gnu.org
2022 11. http://gcc.gnu.org/mirrors.html
2023 12. http://gcc.gnu.org/svn.html
2024 13. http://gcc.gnu.org/onlinedocs/
2025 14. mailto:gcc-help@gcc.gnu.org
2026 15. mailto:gcc@gcc.gnu.org
2027 16. http://gcc.gnu.org/lists.html
2028 17. http://www.fsf.org/
2029 18. http://gcc.gnu.org/about.html
2030 19. http://validator.w3.org/check/referer
2031 ======================================================================
2032 http://gcc.gnu.org/gcc-4.5/changes.html
2034 GCC 4.5 Release Series
2035 Changes, New Features, and Fixes
2039 * GCC now requires the [1]MPC library in order to build. See the
2040 [2]prerequisites page for version requirements.
2041 * Support for a number of older systems and recently unmaintained or
2042 untested target ports of GCC has been declared obsolete in GCC 4.5.
2043 Unless there is activity to revive them, the next release of GCC
2044 will have their sources permanently removed.
2045 The following ports for individual systems on particular
2046 architectures have been obsoleted:
2047 + IRIX releases before 6.5 (mips-sgi-irix5*,
2048 mips-sgi-irix6.[0-4])
2049 + Solaris 7 (*-*-solaris2.7)
2050 + Tru64 UNIX releases before V5.1 (alpha*-dec-osf4*,
2052 + Details for the IRIX, Solaris 7, and Tru64 UNIX obsoletions
2053 can be found in the [3]announcement.
2054 Support for the classic POWER architecture implemented in the
2055 original RIOS and RIOS2 processors of the old IBM RS/6000 product
2056 line has been obsoleted in the rs6000 port. This does not affect
2057 the new generation Power and PowerPC architectures.
2058 * Support has been removed for all the [4]configurations obsoleted in
2060 * Support has been removed for the protoize and unprotoize utilities,
2061 obsoleted in GCC 4.4.
2062 * Support has been removed for tuning for Itanium1 (Merced) variants.
2063 Note that code tuned for Itanium2 should also run correctly on
2065 * GCC now generates unwind info also for epilogues. DWARF debuginfo
2066 generated by GCC now uses more features of DWARF3 than before, and
2067 also some DWARF4 features. GDB older than 7.0 is not able to handle
2068 either of these, so to debug GCC 4.5 generated binaries or
2069 libraries GDB 7.0 or later is needed. You can disable use of DWARF4
2070 features with the -gdwarf-3 -gstrict-dwarf options, or use
2071 -gdwarf-2 -gstrict-dwarf to restrict GCC to just DWARF2, but
2072 epilogue unwind info is emitted unconditionally whenever unwind
2074 * On x86 targets, code containing floating-point calculations may run
2075 significantly slower when compiled with GCC 4.5 in strict C99
2076 conformance mode than they did with earlier GCC versions. This is
2077 due to stricter standard conformance of the compiler and can be
2078 avoided by using the option -fexcess-precision=fast; also see
2080 * The function attribute noinline no longer prevents GCC from cloning
2081 the function. A new attribute noclone has been introduced for this
2082 purpose. Cloning a function means that it is duplicated and the new
2083 copy is specialized for certain contexts (for example when a
2084 parameter is a known constant).
2086 General Optimizer Improvements
2088 * The -save-temps now takes an optional argument. The -save-temps and
2089 -save-temps=cwd switches write the temporary files in the current
2090 working directory based on the original source file. The
2091 -save-temps=obj switch will write files into the directory
2092 specified with the -o option, and the intermediate filenames are
2093 based on the output file. This will allow the user to get the
2094 compiler intermediate files when doing parallel builds without two
2095 builds of the same filename located in different directories from
2096 interfering with each other.
2097 * Debugging dumps are now created in the same directory as the object
2098 file rather than in the current working directory. This allows the
2099 user to get debugging dumps when doing parallel builds without two
2100 builds of the same filename interfering with each other.
2101 * GCC has been integrated with the [6]MPC library. This allows GCC to
2102 evaluate complex arithmetic at compile time [7]more accurately. It
2103 also allows GCC to evaluate calls to complex built-in math
2104 functions having constant arguments and replace them at compile
2105 time with their mathematically equivalent results. In doing so, GCC
2106 can generate correct results regardless of the math library
2107 implementation or floating point precision of the host platform.
2108 This also allows GCC to generate identical results regardless of
2109 whether one compiles in native or cross-compile configurations to a
2110 particular target. The following built-in functions take advantage
2111 of this new capability: cacos, cacosh, casin, casinh, catan,
2112 catanh, ccos, ccosh, cexp, clog, cpow, csin, csinh, csqrt, ctan,
2113 and ctanh. The float and long double variants of these functions
2114 (e.g. csinf and csinl) are also handled.
2115 * A new link-time optimizer has been added ([8]-flto). When this
2116 option is used, GCC generates a bytecode representation of each
2117 input file and writes it to specially-named sections in each object
2118 file. When the object files are linked together, all the function
2119 bodies are read from these named sections and instantiated as if
2120 they had been part of the same translation unit. This enables
2121 interprocedural optimizations to work across different files (and
2122 even different languages), potentially improving the performance of
2123 the generated code. To use the link-timer optimizer, -flto needs to
2124 be specified at compile time and during the final link. If the
2125 program does not require any symbols to be exported, it is possible
2126 to combine -flto and the experimental [9]-fwhopr with
2127 [10]-fwhole-program to allow the interprocedural optimizers to use
2128 more aggressive assumptions.
2129 * The automatic parallelization pass was enhanced to support
2130 parallelization of outer loops.
2131 * Automatic parallelization can be enabled as part of Graphite. In
2132 addition to -ftree-parallelize-loops=, specify
2133 -floop-parallelize-all to enable the Graphite-based optimization.
2134 * The infrastructure for optimizing based on [11]restrict qualified
2135 pointers has been rewritten and should result in code generation
2136 improvements. Optimizations based on restrict qualified pointers
2137 are now also available when using -fno-strict-aliasing.
2138 * There is a new optimization pass that attempts to change prototype
2139 of functions to avoid unused parameters, pass only relevant parts
2140 of structures and turn arguments passed by reference to arguments
2141 passed by value when possible. It is enabled by -O2 and above as
2142 well as -Os and can be manually invoked using the new command-line
2144 * GCC now optimize exception handling code. In particular cleanup
2145 regions that are proved to not have any effect are optimized out.
2147 New Languages and Language specific improvements
2151 * The -fshow-column option is now on by default. This means error
2152 messages now have a column associated with them.
2156 * Compilation of programs heavily using discriminated record types
2157 with variant parts has been sped up and generates more compact
2159 * Stack checking now works reasonably well on most plaforms. In some
2160 specific cases, stack overflows may still fail to be detected, but
2161 a compile-time warning will be issued for these cases.
2165 * If a header named in a #include directive is not found, the
2166 compiler exits immediately. This avoids a cascade of errors arising
2167 from declarations expected to be found in that header being
2169 * A new built-in function __builtin_unreachable() has been added that
2170 tells the compiler that control will never reach that point. It may
2171 be used after asm statements that terminate by transferring control
2172 elsewhere, and in other places that are known to be unreachable.
2173 * The -Wlogical-op option now warns for logical expressions such as
2174 (c == 1 && c == 2) and (c != 1 || c != 2), which are likely to be
2175 mistakes. This option is disabled by default.
2176 * An asm goto feature has been added to allow asm statements that
2178 * C++0x raw strings are supported for C++ and for C with -std=gnu99.
2179 * The deprecated attribute now takes an optional string argument, for
2180 example, __attribute__((deprecated("text string"))), that will be
2181 printed together with the deprecation warning.
2185 * The -Wenum-compare option, which warns when comparing values of
2186 different enum types, now works for C. It formerly only worked for
2187 C++. This warning is enabled by -Wall. It may be avoided by using a
2189 * The -Wcast-qual option now warns about casts which are unsafe in
2190 that they permit const-correctness to be violated without further
2191 warnings. Specifically, it warns about cases where a qualifier is
2192 added when all the lower types are not const. For example, it warns
2193 about a cast from char ** to const char **.
2194 * The -Wc++-compat option is significantly improved. It issues new
2196 + Using C++ reserved operator names as identifiers.
2197 + Conversions to enum types without explicit casts.
2198 + Using va_arg with an enum type.
2199 + Using different enum types in the two branches of ?:.
2200 + Using ++ or -- on a variable of enum type.
2201 + Using the same name as both a struct, union or enum tag and a
2202 typedef, unless the typedef refers to the tagged type itself.
2203 + Using a struct, union, or enum which is defined within another
2205 + A struct field defined using a typedef if there is a field in
2206 the struct, or an enclosing struct, whose name is the typedef
2208 + Duplicate definitions at file scope.
2209 + Uninitialized const variables.
2210 + A global variable with an anonymous struct, union, or enum
2212 + Using a string constant to initialize a char array whose size
2213 is the length of the string.
2214 * The new -Wjump-misses-init option warns about cases where a goto or
2215 switch skips the initialization of a variable. This sort of branch
2216 is an error in C++ but not in C. This warning is enabled by
2218 * GCC now ensures that a C99-conforming <stdint.h> is present on most
2219 targets, and uses information about the types in this header to
2220 implement the Fortran bindings to those types. GCC does not ensure
2221 the presence of such a header, and does not implement the Fortran
2222 bindings, on the following targets: NetBSD, VxWorks, VMS,
2223 SymbianOS, WinCE, LynxOS, Netware, QNX, Interix, TPF.
2224 * GCC now implements C90- and C99-conforming rules for constant
2225 expressions. This may cause warnings or errors for some code using
2226 expressions that can be folded to a constant but are not constant
2227 expressions as defined by ISO C.
2228 * All known target-independent C90 and C90 Amendment 1 conformance
2229 bugs, and all known target-independent C99 conformance bugs not
2230 related to floating point or extended identifiers, have been fixed.
2231 * The C decimal floating point support now includes support for the
2232 FLOAT_CONST_DECIMAL64 pragma.
2233 * The named address space feature from ISO/IEC TR 18037 is now
2234 supported. This is currently only implemented for the SPU
2239 * Improved [12]experimental support for the upcoming C++0x ISO C++
2240 standard, including support for raw strings, lambda expressions and
2241 explicit type conversion operators.
2242 * When printing the name of a class template specialization, G++ will
2243 now omit any template arguments which come from default template
2244 arguments. This behavior (and the pretty-printing of function
2245 template specializations as template signature and arguments) can
2246 be disabled with the -fno-pretty-templates option.
2247 * Access control is now applied to typedef names used in a template,
2248 which may cause G++ to reject some ill-formed code that was
2249 accepted by earlier releases. The -fno-access-control option can be
2250 used as a temporary workaround until the code is corrected.
2251 * Compilation time for code that uses templates should now scale
2252 linearly with the number of instantiations rather than
2253 quadratically, as template instantiations are now looked up using
2255 * Declarations of functions that look like builtin declarations of
2256 library functions are only considered to be redeclarations if they
2257 are declared with extern "C". This may cause problems with code
2258 that omits extern "C" on hand-written declarations of C library
2259 functions such as abort or memcpy. Such code is ill-formed, but was
2260 accepted by earlier releases.
2261 * Diagnostics that used to complain about passing non-POD types to
2262 ... or jumping past the declaration of a non-POD variable now check
2263 for triviality rather than PODness, as per C++0x.
2264 * In C++0x mode local and anonymous classes are now allowed as
2265 template arguments, and in declarations of variables and functions
2266 with linkage, so long as any such declaration that is used is also
2267 defined ([13]DR 757).
2268 * Labels may now have attributes, as has been permitted for a while
2269 in C. This is only permitted when the label definition and the
2270 attribute specifier is followed by a semicolon--i.e., the label
2271 applies to an empty statement. The only useful attribute for a
2273 * G++ now implements [14]DR 176. Previously G++ did not support using
2274 the injected-class-name of a template base class as a type name,
2275 and lookup of the name found the declaration of the template in the
2276 enclosing scope. Now lookup of the name finds the
2277 injected-class-name, which can be used either as a type or as a
2278 template, depending on whether or not the name is followed by a
2279 template argument list. As a result of this change, some code that
2280 was previously accepted may be ill-formed because
2281 1. The injected-class-name is not accessible because it's from a
2283 2. The injected-class-name cannot be used as an argument for a
2284 template template parameter.
2285 In either of these cases, the code can be fixed by adding a
2286 nested-name-specifier to explicitly name the template. The first
2287 can be worked around with -fno-access-control; the second is only
2288 rejected with -pedantic.
2289 * A new standard mangling for SIMD vector types has been added, to
2290 avoid name clashes on systems with vectors of varying length. By
2291 default the compiler still uses the old mangling, but emits aliases
2292 with the new mangling on targets that support strong aliases. Users
2293 can switch over entirely to the new mangling with -fabi-version=4
2294 or -fabi-version=0. -Wabi will now warn about code that uses the
2296 * The command-line option -ftemplate-depth-N is now written as
2297 -ftemplate-depth=N and the old form is deprecated.
2298 * Conversions between NULL and non-pointer types are now warned by
2299 default. The new option -Wno-conversion-null disables these
2300 warnings. Previously these warnings were only available when using
2301 -Wconversion explicitly.
2303 Runtime Library (libstdc++)
2305 * [15]Improved experimental support for the upcoming ISO C++
2306 standard, C++0x, including:
2307 + Support for <future>, <functional>, and <random>.
2308 + Existing facilities now exploit explicit operators and the
2309 newly implemented core C++0x features.
2310 * An experimental [16]profile mode has been added. This is an
2311 implementation of many C++ standard library constructs with an
2312 additional analysis layer that gives performance improvement advice
2313 based on recognition of suboptimal usage patterns. For example,
2318 for (int k = 0; k < 1024; ++k)
2319 v.insert(v.begin(), k);
2322 When instrumented via the profile mode, can return suggestions
2323 about the initial size and choice of the container used as follows:
2324 vector-to-list: improvement = 5: call stack = 0x804842c ...
2325 : advice = change std::vector to std::list
2326 vector-size: improvement = 3: call stack = 0x804842c ...
2327 : advice = change initial container size from 0 to 1024
2329 These constructs can be substituted for the normal libstdc++
2330 constructs on a piecemeal basis, or all existing components can be
2331 transformed via the -D_GLIBCXX_PROFILE macro.
2332 * [17]Support for decimal floating-point arithmetic (aka ISO C++ TR
2333 24733) has been added. This support is in header file
2334 <decimal/decimal>, uses namespace std::decimal, and includes
2335 classes decimal32, decimal64, and decimal128.
2336 * Sources have been audited for application of function attributes
2337 nothrow, const, pure, and noreturn.
2338 * Python pretty-printers have been added for many standard library
2339 components that simplify the internal representation and present a
2340 more intuitive view of components when used with
2341 appropriately-advanced versions of GDB. For more information,
2342 please consult the more [18]detailed description.
2343 * The default behavior for comparing typeinfo names has changed, so
2344 in <typeinfo>, __GXX_MERGED_TYPEINFO_NAMES now defaults to zero.
2345 * The new -static-libstdc++ option directs g++ to link the C++
2346 library statically, even if the default would normally be to link
2351 * The COMMON default padding has been changed - instead of adding the
2352 padding before a variable it is now added afterwards, which
2353 increases the compatibility with other vendors and helps to obtain
2354 the correct output in some cases. Cf. also the -falign-commons
2355 option ([19]added in 4.4).
2356 * The -finit-real= option now also supports the value snan for
2357 signalling not-a-number; to be effective, one additionally needs to
2358 enable trapping (e.g. via -ffpe-trap=). Note: Compile-time
2359 optimizations can turn a signalling NaN into a quiet one.
2360 * The new option -fcheck= has been added with the options bounds,
2361 array-temps, do, pointer, and recursive. The bounds and array-temps
2362 options are equivalent to -fbounds-check and
2363 -fcheck-array-temporaries. The do option checks for invalid
2364 modification of loop iteration variables, and the recursive option
2365 tests for recursive calls to subroutines/functions which are not
2366 marked as recursive. With pointer pointer association checks in
2367 calls are performed; however, neither undefined pointers nor
2368 pointers in expressions are handled. Using -fcheck=all enables all
2369 these run-time checks.
2370 * The run-time checking -fcheck=bounds now warns about invalid string
2371 lengths of character dummy arguments. Additionally, more
2372 compile-time checks have been added.
2373 * The new option [20]-fno-protect-parens has been added; if set, the
2374 compiler may reorder REAL and COMPLEX expressions without regard to
2376 * GNU Fortran no longer links against libgfortranbegin. As before,
2377 MAIN__ (assembler symbol name) is the actual Fortran main program,
2378 which is invoked by the main function. However, main is now
2379 generated and put in the same object file as MAIN__. For the time
2380 being, libgfortranbegin still exists for backward compatibility.
2381 For details see the new [21]Mixed-Language Programming chapter in
2383 * The I/O library was restructured for performance and cleaner code.
2384 * Array assignments and WHERE are now run in parallel when OpenMP's
2386 * The experimental option -fwhole-file was added. The option allows
2387 whole-file checking of procedure arguments and allows for better
2388 optimizations. It can also be used with -fwhole-program, which is
2389 now also supported in gfortran.
2390 * More Fortran 2003 and Fortran 2008 mathematical functions can now
2391 be used as initialization expressions.
2392 * Some extended attributes such as STDCALL are now supported via the
2393 [22]GCC$ compiler directive.
2394 * For Fortran 77 compatibility: If -fno-sign-zero is used, the SIGN
2395 intrinsic behaves now as if zero were always positive.
2396 * For legacy compatibiliy: On Cygwin and MinGW, the special files
2397 CONOUT$ and CONIN$ (and CONERR$ which maps to CONOUT$) are now
2399 * Fortran 2003 support has been extended:
2400 + Procedure-pointer function results and procedure-pointer
2401 components (including PASS),
2402 + allocatable scalars (experimental),
2403 + DEFERRED type-bound procedures,
2404 + the ERRMSG= argument of the ALLOCATE and DEALLOCATE statements
2405 have been implemented.
2406 + The ALLOCATE statement supports type-specs and the SOURCE=
2408 + OPERATOR(*) and ASSIGNMENT(=) are now allowed as GENERIC
2409 type-bound procedure (i.e. as type-bound operators).
2410 + Rounding (ROUND=, RZ, ...) for output is now supported.
2411 + The INT_FAST{8,16,32,64,128}_T kind type parameters of the
2412 intrinsic module ISO_C_BINDING are now supported, except for
2413 the targets listed above as ones where GCC does not have
2414 <stdint.h> type information.
2415 + Extensible derived types with type-bound procedure or
2416 procedure pointer with PASS attribute now have to use CLASS in
2417 line with the Fortran 2003 standard; the workaround to use
2418 TYPE is no longer supported.
2419 + [23]Experimental, incomplete support for polymorphism,
2420 including CLASS, SELECT TYPE and dynamic dispatch of
2421 type-bound procedure calls. Some features do not work yet such
2422 as unlimited polymorphism (CLASS(*)).
2423 * Fortran 2008 support has been extended:
2424 + The OPEN statement now supports the NEWUNIT= option, which
2425 returns a unique file unit, thus preventing inadvertent use of
2426 the same unit in different parts of the program.
2427 + Support for unlimited format items has been added.
2428 + The INT{8,16,32} and REAL{32,64,128} kind type parameters of
2429 the intrinsic module ISO_FORTRAN_ENV are now supported.
2430 + Using complex arguments with TAN, SINH, COSH, TANH, ASIN,
2431 ACOS, and ATAN is now possible; the functions ASINH, ACOSH,
2432 and ATANH have been added (for real and complex arguments) and
2433 ATAN(Y,X) is now an alias for ATAN2(Y,X).
2434 + The BLOCK construct has been implemented.
2436 New Targets and Target Specific Improvements
2440 * Full cross-toolchain support now available with GNU Binutils
2444 * GCC now supports the Cortex-M0 and Cortex-A5 processors.
2445 * GCC now supports the ARM v7E-M architecture.
2446 * GCC now supports VFPv4-based FPUs and FPUs with
2447 single-precision-only VFP.
2448 * GCC has many improvements to optimization for other ARM processors,
2449 including scheduling support for the integer pipeline on Cortex-A9.
2450 * GCC now supports the IEEE 754-2008 half-precision floating-point
2451 type, and a variant ARM-specific half-precision type. This type is
2452 specified using __fp16, with the layout determined by
2453 -mfp16-format. With appropriate -mfpu options, the Cortex-A9 and
2454 VFPv4 half-precision instructions will be used.
2455 * GCC now supports the variant of AAPCS that uses VFP registers for
2456 parameter passing and return values.
2460 * The -mno-tablejump option has been removed because it has the same
2461 effect as the -fno-jump-tables option.
2462 * Added support for these new AVR devices:
2469 * GCC now will set the default for -march= based on the configure
2471 * GCC now supports handling floating-point excess precision arising
2472 from use of the x87 floating-point unit in a way that conforms to
2473 ISO C99. This is enabled with -fexcess-precision=standard and with
2474 standards conformance options such as -std=c99, and may be disabled
2475 using -fexcess-precision=fast.
2476 * Support for the Intel Atom processor is now available through the
2477 -march=atom and -mtune=atom options.
2478 * A new -mcrc32 option is now available to enable crc32 intrinsics.
2479 * A new -mmovbe option is now available to enable GCC to use the
2480 movbe instruction to implement __builtin_bswap32 and
2482 * SSE math now can be enabled by default at configure time with the
2483 new --with-fpmath=sse option.
2484 * There is a new intrinsic header file, <x86intrin.h>. It should be
2485 included before using any IA-32/x86-64 intrinsics.
2486 * Support for the XOP, FMA4, and LWP instruction sets for the AMD
2487 Orochi processors are now available with the -mxop, -mfma4, and
2489 * The -mabm option enables GCC to use the popcnt and lzcnt
2490 instructions on AMD processors.
2491 * The -mpopcnt option enables GCC to use the popcnt instructions on
2492 both AMD and Intel processors.
2496 * GCC now supports ColdFire 51xx, 5221x, 5225x, 52274, 52277, 5301x
2498 * GCC now supports thread-local storage (TLS) on M68K and ColdFire
2503 Support has been added for the Toshiba Media embedded Processor (MeP,
2504 or mep-elf) embedded target.
2508 * GCC now supports MIPS 1004K processors.
2509 * GCC can now be configured with options --with-arch-32,
2510 --with-arch-64, --with-tune-32 and --with-tune-64 to control the
2511 default optimization separately for 32-bit and 64-bit modes.
2512 * MIPS targets now support an alternative _mcount interface, in which
2513 register $12 points to the function's save slot for register $31.
2514 This interface is selected by the -mcount-ra-address option; see
2515 the documentation for more details.
2516 * GNU/Linux targets can now generate read-only .eh_frame sections.
2517 This optimization requires GNU binutils 2.20 or above, and is only
2518 available if GCC is configured with a suitable version of binutils.
2519 * GNU/Linux targets can now attach special relocations to indirect
2520 calls, so that the linker can turn them into direct jumps or
2521 branches. This optimization requires GNU binutils 2.20 or later,
2522 and is automatically selected if GCC is configured with an
2523 appropriate version of binutils. It can be explicitly enabled or
2524 disabled using the -mrelax-pic-calls command-line option.
2525 * GCC now generates more heavily-optimized atomic operations on
2527 * MIPS targets now support the -fstack-protector option.
2528 * GCC now supports an -msynci option, which specifies that synci is
2529 enough to flush the instruction cache, without help from the
2530 operating system. GCC uses this information to optimize
2531 automatically-generated cache flush operations, such as those used
2532 for nested functions in C. There is also a --with-synci
2533 configure-time option, which makes -msynci the default.
2534 * GCC supports four new function attributes for interrupt handlers:
2535 interrupt, use_shadow_register_set, keep_interrupts_masked and
2536 use_debug_exception_return. See the documentation for more details
2537 about these attributes.
2539 RS/6000 (POWER/PowerPC)
2541 * GCC now supports the Power ISA 2.06, which includes the VSX
2542 instructions that add vector 64-bit floating point support, new
2543 population count instructions, and conversions between floating
2544 point and unsigned types.
2545 * Support for the power7 processor is now available through the
2546 -mcpu=power7 and -mtune=power7.
2547 * GCC will now vectorize loops that contain simple math functions
2548 like copysign when generating code for altivec or VSX targets.
2549 * Support for the A2 processor is now available through the -mcpu=a2
2550 and -mtune=a2 options.
2551 * Support for the 476 processor is now available through the
2552 -mcpu={476,476fp} and -mtune={476,476fp} options.
2553 * Support for the e500mc64 processor is now available through the
2554 -mcpu=e500mc64 and -mtune=e500mc64 options.
2555 * GCC can now be configured with options --with-cpu-32,
2556 --with-cpu-64, --with-tune-32 and --with-tune-64 to control the
2557 default optimization separately for 32-bit and 64-bit modes.
2558 * Starting with GCC 4.5.4, vectors of type vector long long or vector
2559 long are passed and returned in the same method as other vectors
2560 with the VSX instruction set. Previously the GCC compiler did not
2561 adhere to the ABI for 128-bit vectors with 64-bit integer base
2562 types (PR 48857). This is also fixed in the GCC 4.6.1 release.
2566 Support has been added for the Renesas RX Processor (rx-elf) target.
2570 Windows (Cygwin and MinGW)
2572 * GCC now installs all the major language runtime libraries as DLLs
2573 when configured with the --enable-shared option.
2574 * GCC now makes use of the new support for aligned common variables
2575 in versions of binutils >= 2.20 to fix bugs in the support for SSE
2577 * Improvements to the libffi support library increase the reliability
2578 of code generated by GCJ on all Windows platforms. Libgcj is
2579 enabled by default for the first time.
2580 * Libtool improvements simplify installation by placing the generated
2581 DLLs in the correct binaries directory.
2582 * Numerous other minor bugfixes and improvements, and substantial
2583 enhancements to the Fortran language support library.
2587 Other significant improvements
2591 * It is now possible to extend the compiler without having to modify
2592 its source code. A new option -fplugin=file.so tells GCC to load
2593 the shared object file.so and execute it as part of the compiler.
2594 The internal documentation describes the details on how plugins can
2595 interact with the compiler.
2597 Installation changes
2599 * The move to newer autotools changed default installation
2600 directories and switches to control them: The --with-datarootdir,
2601 --with-docdir, --with-pdfdir, and --with-htmldir switches are not
2602 used any more. Instead, you can now use --datarootdir, --docdir,
2603 --htmldir, and --pdfdir. The default installation directories have
2604 changed as follows according to the GNU Coding Standards:
2606 datarootdir read-only architecture-independent data root [PREFIX/share]
2607 localedir locale-specific message catalogs [DATAROOTDIR/locale]
2608 docdir documentation root [DATAROOTDIR/doc/PACKAGE]
2609 htmldir html documentation [DOCDIR]
2610 dvidir dvi documentation [DOCDIR]
2611 pdfdir pdf documentation [DOCDIR]
2612 psdir ps documentation [DOCDIR]
2613 The following variables have new default values:
2615 datadir read-only architecture-independent data [DATAROOTDIR]
2616 infodir info documentation [DATAROOTDIR/info]
2617 mandir man documentation [DATAROOTDIR/man]
2621 This is the [24]list of problem reports (PRs) from GCC's bug tracking
2622 system that are known to be fixed in the 4.5.1 release. This list might
2623 not be complete (that is, it is possible that some PRs that have been
2624 fixed are not listed here).
2628 * GCC's new link-time optimizer ([25]-flto) now also works on a few
2630 + Cygwin (*-cygwin*)
2632 + Darwin on x86-64 (x86_64-apple-darwin*)
2633 LTO is not enabled by default for these targets. To enable LTO, you
2634 should configure with the --enable-lto option.
2638 This is the [26]list of problem reports (PRs) from GCC's bug tracking
2639 system that are known to be fixed in the 4.5.2 release. This list might
2640 not be complete (that is, it is possible that some PRs that have been
2641 fixed are not listed here).
2645 This is the [27]list of problem reports (PRs) from GCC's bug tracking
2646 system that are known to be fixed in the 4.5.3 release. This list might
2647 not be complete (that is, it is possible that some PRs that have been
2648 fixed are not listed here).
2650 On the PowerPC compiler, the Altivec builtin functions vec_ld and
2651 vec_st have been modified to generate the Altivec memory instructions
2652 LVX and STVX, even if the -mvsx option is used. In the initial GCC 4.5
2653 release, these builtin functions were changed to generate VSX memory
2654 reference instructions instead of Altivec memory instructions, but
2655 there are differences between the two instructions. If the VSX
2656 instruction set is available, you can now use the new builtin functions
2657 vec_vsx_ld and vec_vsx_st which always generates the VSX memory
2662 This is the [28]list of problem reports (PRs) from GCC's bug tracking
2663 system that are known to be fixed in the 4.5.4 release. This list might
2664 not be complete (that is, it is possible that some PRs that have been
2665 fixed are not listed here).
2668 For questions related to the use of GCC, please consult these web
2669 pages and the [29]GCC manuals. If that fails, the
2670 [30]gcc-help@gcc.gnu.org mailing list might help. Comments on these
2671 web pages and the development of GCC are welcome on our developer
2672 list at [31]gcc@gcc.gnu.org. All of [32]our lists have public
2675 Copyright (C) [33]Free Software Foundation, Inc. Verbatim copying and
2676 distribution of this entire article is permitted in any medium,
2677 provided this notice is preserved.
2679 These pages are [34]maintained by the GCC team. Last modified
2684 1. http://www.multiprecision.org/
2685 2. http://gcc.gnu.org/install/prerequisites.html
2686 3. http://gcc.gnu.org/ml/gcc/2010-01/msg00510.html
2687 4. http://gcc.gnu.org/gcc-4.4/changes.html#obsoleted
2688 5. http://gcc.gnu.org/gcc-4.5/changes.html#x86
2689 6. http://www.multiprecision.org/
2690 7. http://gcc.gnu.org/PR30789
2691 8. http://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html#index-flto-801
2692 9. http://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html#index-fwhopr-802
2693 10. http://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html#index-fwhole-program-800
2694 11. http://gcc.gnu.org/onlinedocs/gcc/Restricted-Pointers.html
2695 12. http://gcc.gnu.org/gcc-4.5/cxx0x_status.html
2696 13. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#757
2697 14. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#176
2698 15. http://gcc.gnu.org/onlinedocs/libstdc++/manual/status.html#status.iso.200x
2699 16. http://gcc.gnu.org/onlinedocs/libstdc++/manual/profile_mode.html
2700 17. http://gcc.gnu.org/onlinedocs/libstdc++/manual/status.html#status.iso.tr24733
2701 18. http://sourceware.org/gdb/wiki/STLSupport
2702 19. http://gcc.gnu.org/gcc-4.4/changes.html
2703 20. http://gcc.gnu.org/onlinedocs/gfortran/Code-Gen-Options.html
2704 21. http://gcc.gnu.org/onlinedocs/gfortran/Mixed-Language-Programming.html
2705 22. http://gcc.gnu.org/onlinedocs/gfortran/GNU-Fortran-Compiler-Directives.html
2706 23. http://gcc.gnu.org/wiki/OOP
2707 24. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.5.1
2708 25. http://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html#index-flto-801
2709 26. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.5.2
2710 27. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.5.3
2711 28. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.5.4
2712 29. http://gcc.gnu.org/onlinedocs/
2713 30. mailto:gcc-help@gcc.gnu.org
2714 31. mailto:gcc@gcc.gnu.org
2715 32. http://gcc.gnu.org/lists.html
2716 33. http://www.fsf.org/
2717 34. http://gcc.gnu.org/about.html
2718 35. http://validator.w3.org/check/referer
2719 ======================================================================
2720 http://gcc.gnu.org/gcc-4.4/index.html
2722 GCC 4.4 Release Series
2726 The [1]GNU project and the GCC developers are pleased to announce the
2727 release of GCC 4.4.7.
2729 This release is a bug-fix release, containing fixes for regressions in
2730 GCC 4.4.6 relative to previous releases of GCC.
2735 March 13, 2012 ([2]changes)
2738 April 16, 2011 ([3]changes)
2741 October 1, 2010 ([4]changes)
2744 April 29, 2010 ([5]changes)
2747 January 21, 2010 ([6]changes)
2750 October 15, 2009 ([7]changes)
2753 July 22, 2009 ([8]changes)
2756 April 21, 2009 ([9]changes)
2758 References and Acknowledgements
2760 GCC used to stand for the GNU C Compiler, but since the compiler
2761 supports several other languages aside from C, it now stands for the
2762 GNU Compiler Collection.
2764 A list of [10]successful builds is updated as new information becomes
2767 The GCC developers would like to thank the numerous people that have
2768 contributed new features, improvements, bug fixes, and other changes as
2769 well as test results to GCC. This [11]amazing group of volunteers is
2770 what makes GCC successful.
2772 For additional information about GCC please refer to the [12]GCC
2773 project web site or contact the [13]GCC development mailing list.
2775 To obtain GCC please use [14]our mirror sites or [15]our SVN server.
2778 For questions related to the use of GCC, please consult these web
2779 pages and the [16]GCC manuals. If that fails, the
2780 [17]gcc-help@gcc.gnu.org mailing list might help. Comments on these
2781 web pages and the development of GCC are welcome on our developer
2782 list at [18]gcc@gcc.gnu.org. All of [19]our lists have public
2785 Copyright (C) [20]Free Software Foundation, Inc. Verbatim copying and
2786 distribution of this entire article is permitted in any medium,
2787 provided this notice is preserved.
2789 These pages are [21]maintained by the GCC team. Last modified
2794 1. http://www.gnu.org/
2795 2. http://gcc.gnu.org/gcc-4.4/changes.html
2796 3. http://gcc.gnu.org/gcc-4.4/changes.html
2797 4. http://gcc.gnu.org/gcc-4.4/changes.html
2798 5. http://gcc.gnu.org/gcc-4.4/changes.html
2799 6. http://gcc.gnu.org/gcc-4.4/changes.html
2800 7. http://gcc.gnu.org/gcc-4.4/changes.html
2801 8. http://gcc.gnu.org/gcc-4.4/changes.html
2802 9. http://gcc.gnu.org/gcc-4.4/changes.html
2803 10. http://gcc.gnu.org/gcc-4.4/buildstat.html
2804 11. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
2805 12. http://gcc.gnu.org/index.html
2806 13. mailto:gcc@gcc.gnu.org
2807 14. http://gcc.gnu.org/mirrors.html
2808 15. http://gcc.gnu.org/svn.html
2809 16. http://gcc.gnu.org/onlinedocs/
2810 17. mailto:gcc-help@gcc.gnu.org
2811 18. mailto:gcc@gcc.gnu.org
2812 19. http://gcc.gnu.org/lists.html
2813 20. http://www.fsf.org/
2814 21. http://gcc.gnu.org/about.html
2815 22. http://validator.w3.org/check/referer
2816 ======================================================================
2817 http://gcc.gnu.org/gcc-4.4/changes.html
2819 GCC 4.4 Release Series
2820 Changes, New Features, and Fixes
2822 The latest release in the 4.4 release series is [1]GCC 4.4.7.
2826 * __builtin_stdarg_start has been completely removed from GCC.
2827 Support for <varargs.h> had been deprecated since GCC 4.0. Use
2828 __builtin_va_start as a replacement.
2829 * Some of the errors issued by the C++ front end that could be
2830 downgraded to warnings in previous releases by using -fpermissive
2831 are now warnings by default. They can be converted into errors by
2832 using -pedantic-errors.
2833 * Use of the cpp assertion extension will now emit a warning when
2834 -Wdeprecated or -pedantic is used. This extension has been
2835 deprecated for many years, but never warned about.
2836 * Packed bit-fields of type char were not properly bit-packed on many
2837 targets prior to GCC 4.4. On these targets, the fix in GCC 4.4
2838 causes an ABI change. For example there is no longer a 4-bit
2839 padding between field a and b in this structure:
2844 } __attribute__ ((packed));
2845 There is a new warning to help identify fields that are affected:
2846 foo.c:5: note: Offset of packed bit-field 'b' has changed in GCC 4.4
2847 The warning can be disabled with -Wno-packed-bitfield-compat.
2848 * On ARM EABI targets, the C++ mangling of the va_list type has been
2849 changed to conform to the current revision of the EABI. This does
2850 not affect the libstdc++ library included with GCC.
2851 * The SCOUNT and POS bits of the MIPS DSP control register are now
2852 treated as global. Previous versions of GCC treated these fields as
2853 call-clobbered instead.
2854 * The MIPS port no longer recognizes the h asm constraint. It was
2855 necessary to remove this constraint in order to avoid generating
2856 unpredictable code sequences.
2857 One of the main uses of the h constraint was to extract the high
2858 part of a multiplication on 64-bit targets. For example:
2859 asm ("dmultu\t%1,%2" : "=h" (result) : "r" (x), "r" (y));
2860 You can now achieve the same effect using 128-bit types:
2861 typedef unsigned int uint128_t __attribute__((mode(TI)));
2862 result = ((uint128_t) x * y) >> 64;
2863 The second sequence is better in many ways. For example, if x and y
2864 are constants, the compiler can perform the multiplication at
2865 compile time. If x and y are not constants, the compiler can
2866 schedule the runtime multiplication better than it can schedule an
2868 * Support for a number of older systems and recently unmaintained or
2869 untested target ports of GCC has been declared obsolete in GCC 4.4.
2870 Unless there is activity to revive them, the next release of GCC
2871 will have their sources permanently removed.
2872 The following ports for individual systems on particular
2873 architectures have been obsoleted:
2874 + Generic a.out on IA32 and m68k (i[34567]86-*-aout*,
2876 + Generic COFF on ARM, H8300, IA32, m68k and SH (arm-*-coff*,
2877 armel-*-coff*, h8300-*-*, i[34567]86-*-coff*, m68k-*-coff*,
2878 sh-*-*). This does not affect other more specific targets
2879 using the COFF object format on those architectures, or the
2880 more specific H8300 and SH targets (h8300-*-rtems*,
2881 h8300-*-elf*, sh-*-elf*, sh-*-symbianelf*, sh-*-linux*,
2882 sh-*-netbsdelf*, sh-*-rtems*, sh-wrs-vxworks).
2883 + 2BSD on PDP-11 (pdp11-*-bsd)
2884 + AIX 4.1 and 4.2 on PowerPC (rs6000-ibm-aix4.[12]*,
2885 powerpc-ibm-aix4.[12]*)
2886 + Tuning support for Itanium1 (Merced) variants. Note that code
2887 tuned for Itanium2 should also run correctly on Itanium1.
2888 * The protoize and unprotoize utilities have been obsoleted and will
2889 be removed in GCC 4.5. These utilities have not been installed by
2890 default since GCC 3.0.
2891 * Support has been removed for all the [2]configurations obsoleted in
2893 * Unknown -Wno-* options are now silently ignored by GCC if no other
2894 diagnostics are issued. If other diagnostics are issued, then GCC
2895 warns about the unknown options.
2896 * More information on porting to GCC 4.4 from previous versions of
2897 GCC can be found in the [3]porting guide for this release.
2899 General Optimizer Improvements
2901 * A new command-line switch -findirect-inlining has been added. When
2902 turned on it allows the inliner to also inline indirect calls that
2903 are discovered to have known targets at compile time thanks to
2905 * A new command-line switch -ftree-switch-conversion has been added.
2906 This new pass turns simple initializations of scalar variables in
2907 switch statements into initializations from a static array, given
2908 that all the values are known at compile time and the ratio between
2909 the new array size and the original switch branches does not exceed
2910 the parameter --param switch-conversion-max-branch-ratio (default
2912 * A new command-line switch -ftree-builtin-call-dce has been added.
2913 This optimization eliminates unnecessary calls to certain builtin
2914 functions when the return value is not used, in cases where the
2915 calls can not be eliminated entirely because the function may set
2916 errno. This optimization is on by default at -O2 and above.
2917 * A new command-line switch -fconserve-stack directs the compiler to
2918 minimize stack usage even if it makes the generated code slower.
2919 This affects inlining decisions.
2920 * When the assembler supports it, the compiler will now emit unwind
2921 information using assembler .cfi directives. This makes it possible
2922 to use such directives in inline assembler code. The new option
2923 -fno-dwarf2-cfi-asm directs the compiler to not use .cfi
2925 * The [4]Graphite branch has been merged. This merge has brought in a
2926 new framework for loop optimizations based on a polyhedral
2927 intermediate representation. These optimizations apply to all the
2928 languages supported by GCC. The following new code transformations
2929 are available in GCC 4.4:
2930 + -floop-interchange performs loop interchange transformations
2931 on loops. Interchanging two nested loops switches the inner
2932 and outer loops. For example, given a loop like:
2935 A(J, I) = A(J, I) * C
2939 loop interchange will transform the loop as if the user had
2943 A(J, I) = A(J, I) * C
2947 which can be beneficial when N is larger than the caches,
2948 because in Fortran, the elements of an array are stored in
2949 memory contiguously by column, and the original loop iterates
2950 over rows, potentially creating at each access a cache miss.
2951 + -floop-strip-mine performs loop strip mining transformations
2952 on loops. Strip mining splits a loop into two nested loops.
2953 The outer loop has strides equal to the strip size and the
2954 inner loop has strides of the original loop within a strip.
2955 For example, given a loop like:
2960 loop strip mining will transform the loop as if the user had
2963 DO I = II, min (II + 3, N)
2968 + -floop-block performs loop blocking transformations on loops.
2969 Blocking strip mines each loop in the loop nest such that the
2970 memory accesses of the element loops fit inside caches. For
2971 example, given a loop like:
2974 A(J, I) = B(I) + C(J)
2978 loop blocking will transform the loop as if the user had
2982 DO I = II, min (II + 63, N)
2983 DO J = JJ, min (JJ + 63, M)
2984 A(J, I) = B(I) + C(J)
2990 which can be beneficial when M is larger than the caches,
2991 because the innermost loop will iterate over a smaller amount
2992 of data that can be kept in the caches.
2993 * A new register allocator has replaced the old one. It is called
2994 integrated register allocator (IRA) because coalescing, register
2995 live range splitting, and hard register preferencing are done
2996 on-the-fly during coloring. It also has better integration with the
2997 reload pass. IRA is a regional register allocator which uses modern
2998 Chaitin-Briggs coloring instead of Chow's priority coloring used in
2999 the old register allocator. More info about IRA internals and
3000 options can be found in the GCC manuals.
3001 * A new instruction scheduler and software pipeliner, based on the
3002 selective scheduling approach, has been added. The new pass
3003 performs instruction unification, register renaming, substitution
3004 through register copies, and speculation during scheduling. The
3005 software pipeliner is able to pipeline non-countable loops. The new
3006 pass is targeted at scheduling-eager in-order platforms. In GCC 4.4
3007 it is available for the Intel Itanium platform working by default
3008 as the second scheduling pass (after register allocation) at the
3009 -O3 optimization level.
3010 * When using -fprofile-generate with a multi-threaded program, the
3011 profile counts may be slightly wrong due to race conditions. The
3012 new -fprofile-correction option directs the compiler to apply
3013 heuristics to smooth out the inconsistencies. By default the
3014 compiler will give an error message when it finds an inconsistent
3016 * The new -fprofile-dir=PATH option permits setting the directory
3017 where profile data files are stored when using -fprofile-generate
3018 and friends, and the directory used when reading profile data files
3019 using -fprofile-use and friends.
3023 * The new -Wframe-larger-than=NUMBER option directs GCC to emit a
3024 warning if any stack frame is larger than NUMBER bytes. This may be
3025 used to help ensure that code fits within a limited amount of stack
3027 * The command-line option -Wlarger-than-N is now written as
3028 -Wlarger-than=N and the old form is deprecated.
3029 * The new -Wno-mudflap option disables warnings about constructs
3030 which can not be instrumented when using -fmudflap.
3032 New Languages and Language specific improvements
3034 * Version 3.0 of the [5]OpenMP specification is now supported for the
3035 C, C++, and Fortran compilers.
3036 * New character data types, per [6]TR 19769: New character types in
3037 C, are now supported for the C compiler in -std=gnu99 mode, as
3038 __CHAR16_TYPE__ and __CHAR32_TYPE__, and for the C++ compiler in
3039 -std=c++0x and -std=gnu++0x modes, as char16_t and char32_t too.
3043 * A new optimize attribute was added to allow programmers to change
3044 the optimization level and particular optimization options for an
3045 individual function. You can also change the optimization options
3046 via the GCC optimize pragma for functions defined after the pragma.
3047 The GCC push_options pragma and the GCC pop_options pragma allow
3048 you temporarily save and restore the options used. The GCC
3049 reset_options pragma restores the options to what was specified on
3051 * Uninitialized warnings do not require enabling optimization
3052 anymore, that is, -Wuninitialized can be used together with -O0.
3053 Nonetheless, the warnings given by -Wuninitialized will probably be
3054 more accurate if optimization is enabled.
3055 * -Wparentheses now warns about expressions such as (!x | y) and (!x
3056 & y). Using explicit parentheses, such as in ((!x) | y), silences
3058 * -Wsequence-point now warns within if, while,do while and for
3059 conditions, and within for begin/end expressions.
3060 * A new option -dU is available to dump definitions of preprocessor
3061 macros that are tested or expanded.
3065 * [7]Improved experimental support for the upcoming ISO C++ standard,
3066 C++0x. Including support for auto, inline namespaces, generalized
3067 initializer lists, defaulted and deleted functions, new character
3068 types, and scoped enums.
3069 * Those errors that may be downgraded to warnings to build legacy
3070 code now mention -fpermissive when -fdiagnostics-show-option is
3072 * -Wconversion now warns if the result of a static_cast to enumeral
3073 type is unspecified because the value is outside the range of the
3075 * -Wuninitialized now warns if a non-static reference or non-static
3076 const member appears in a class without constructors.
3077 * G++ now properly implements value-initialization, so objects with
3078 an initializer of () and an implicitly defined default constructor
3079 will be zero-initialized before the default constructor is called.
3081 Runtime Library (libstdc++)
3083 * [8]Improved experimental support for the upcoming ISO C++ standard,
3085 + Support for <chrono>, <condition_variable>, <cstdatomic>,
3086 <forward_list>, <initializer_list>, <mutex>, <ratio>,
3087 <system_error>, and <thread>.
3088 + unique_ptr, <algorithm> additions, exception propagation, and
3089 support for the new character types in <string> and <limits>.
3090 + Existing facilities now exploit initializer lists, defaulted
3091 and deleted functions, and the newly implemented core C++0x
3093 + Some standard containers are more efficient together with
3094 stateful allocators, i.e., no allocator is constructed on the
3095 fly at element construction time.
3096 * Experimental support for non-standard pointer types in containers.
3097 * The long standing libstdc++/30928 has been fixed for targets
3098 running glibc 2.10 or later.
3099 * As usual, many small and larger bug fixes, in particular quite a
3100 few corner cases in <locale>.
3104 * GNU Fortran now employs libcpp directly instead of using cc1 as an
3105 external preprocessor. The [9]-cpp option was added to allow manual
3106 invocation of the preprocessor without relying on filename
3108 * The [10]-Warray-temporaries option warns about array temporaries
3109 generated by the compiler, as an aid to optimization.
3110 * The [11]-fcheck-array-temporaries option has been added, printing a
3111 notification at run time, when an array temporary had to be created
3112 for an function argument. Contrary to -Warray-temporaries the
3113 warning is only printed if the array is noncontiguous.
3114 * Improved generation of DWARF debugging symbols
3115 * If using an intrinsic not part of the selected standard (via -std=
3116 and -fall-intrinsics) gfortran will now treat it as if this
3117 procedure were declared EXTERNAL and try to link to a user-supplied
3118 procedure. -Wintrinsics-std will warn whenever this happens. The
3119 now-useless option -Wnonstd-intrinsic was removed.
3120 * The flag -falign-commons has been added to control the alignment of
3121 variables in COMMON blocks, which is enabled by default in line
3122 with previous GCC version. Using -fno-align-commons one can force
3123 commons to be contiguous in memory as required by the Fortran
3124 standard, however, this slows down the memory access. The option
3125 -Walign-commons, which is enabled by default, warns when padding
3126 bytes were added for alignment. The proper solution is to sort the
3127 common objects by decreasing storage size, which avoids the
3129 * Fortran 2003 support has been extended:
3130 + Wide characters (ISO 10646, UCS-4, kind=4) and UTF-8 I/O is
3131 now supported (except internal reads from/writes to wide
3132 strings). [12]-fbackslash now supports also \unnnn and
3133 \Unnnnnnnn to enter Unicode characters.
3134 + Asynchronous I/O (implemented as synchronous I/O) and the
3135 decimal=, size=, sign=, pad=, blank=, and delim= specifiers
3136 are now supported in I/O statements.
3137 + Support for Fortran 2003 structure constructors and for array
3138 constructor with typespec has been added.
3139 + Procedure Pointers (but not yet as component in derived types
3140 and as function results) are now supported.
3141 + Abstract types, type extension, and type-bound procedures
3142 (both PROCEDURE and GENERIC but not as operators). Note: As
3143 CLASS/polymorphyic types are not implemented, type-bound
3144 procedures with PASS accept as non-standard extension TYPE
3146 * Fortran 2008 support has been added:
3147 + The -std=f2008 option and support for the file extensions
3148 .f2008 and .F2008 has been added.
3149 + The g0 format descriptor is now supported.
3150 + The Fortran 2008 mathematical intrinsics ASINH, ACOSH, ATANH,
3151 ERF, ERFC, GAMMA, LOG_GAMMA, BESSEL_*, HYPOT, and ERFC_SCALED
3152 are now available (some of them existed as GNU extension
3153 before). Note: The hyperbolic functions are not yet supporting
3154 complex arguments and the three- argument version of BESSEL_*N
3156 + The bit intrinsics LEADZ and TRAILZ have been added.
3162 * The Ada runtime now supports multilibs on many platforms including
3163 x86_64, SPARC and PowerPC. Their build is enabled by default.
3165 New Targets and Target Specific Improvements
3169 * GCC now supports optimizing for the Cortex-A9, Cortex-R4 and
3170 Cortex-R4F processors and has many other improvements to
3171 optimization for ARM processors.
3172 * GCC now supports the VFPv3 variant with 16 double-precision
3173 registers with -mfpu=vfpv3-d16. The option -mfpu=vfp3 has been
3174 renamed to -mfpu=vfpv3.
3175 * GCC now supports the -mfix-cortex-m3-ldrd option to work around an
3176 erratum on Cortex-M3 processors.
3177 * GCC now supports the __sync_* atomic operations for ARM EABI
3179 * The section anchors optimization is now enabled by default when
3181 * GCC now uses a new EABI-compatible profiling interface for EABI
3182 targets. This requires a function __gnu_mcount_nc, which is
3183 provided by GNU libc versions 2.8 and later.
3187 * The -mno-tablejump option has been deprecated because it has the
3188 same effect as the -fno-jump-tables option.
3189 * Added support for these new AVR devices:
3218 * Support for Intel AES built-in functions and code generation is
3219 available via -maes.
3220 * Support for Intel PCLMUL built-in function and code generation is
3221 available via -mpclmul.
3222 * Support for Intel AVX built-in functions and code generation is
3223 available via -mavx.
3224 * Automatically align the stack for local variables with alignment
3226 * GCC can now utilize the SVML library for vectorizing calls to a set
3227 of C99 functions if -mveclibabi=svml is specified and you link to
3228 an SVML ABI compatible library.
3229 * On x86-64, the ABI has been changed in the following cases to
3230 conform to the x86-64 ABI:
3231 + Passing/returning structures with flexible array member:
3237 + Passing/returning structures with complex float member:
3241 __complex__ float f;
3243 + Passing/returning unions with long double member:
3249 Code built with previous versions of GCC that uses any of these is
3250 not compatible with code built with GCC 4.4.0 or later.
3251 * A new target attribute was added to allow programmers to change the
3252 target options like -msse2 or -march=k8 for an individual function.
3253 You can also change the target options via the GCC target pragma
3254 for functions defined after the pragma.
3255 * GCC can now be configured with options --with-arch-32,
3256 --with-arch-64, --with-cpu-32, --with-cpu-64, --with-tune-32 and
3257 --with-tune-64 to control the default optimization separately for
3258 32-bit and 64-bit modes.
3262 * Support for __float128 (TFmode) IEEE quad type and corresponding
3263 TCmode IEEE complex quad type is available via the soft-fp library
3264 on IA-32/IA64 targets. This includes basic arithmetic operations
3265 (addition, subtraction, negation, multiplication and division) on
3266 __float128 real and TCmode complex values, the full set of IEEE
3267 comparisons between __float128 values, conversions to and from
3268 float, double and long double floating point types, as well as
3269 conversions to and from signed or unsigned integer, signed or
3270 unsigned long integer and signed or unsigned quad (TImode, IA64
3271 only) integer types. Additionally, all operations generate the full
3272 set of IEEE exceptions and support the full set of IEEE rounding
3277 * GCC now supports instruction scheduling for ColdFire V1, V3 and V4
3278 processors. (Scheduling support for ColdFire V2 processors was
3280 * GCC now supports the -mxgot option to support programs requiring
3281 many GOT entries on ColdFire.
3282 * The m68k-*-linux-gnu target now builds multilibs by default.
3286 * MIPS Technologies have extended the original MIPS SVR4 ABI to
3287 include support for procedure linkage tables (PLTs) and copy
3288 relocations. These extensions allow GNU/Linux executables to use a
3289 significantly more efficient code model than the one defined by the
3291 GCC support for this code model is available via a new command-line
3292 option, -mplt. There is also a new configure-time option,
3293 --with-mips-plt, to make -mplt the default.
3294 The new code model requires support from the assembler, the linker,
3295 and the runtime C library. This support is available in binutils
3297 * GCC can now generate MIPS16 code for 32-bit GNU/Linux executables
3298 and 32-bit GNU/Linux shared libraries. This feature requires GNU
3299 binutils 2.19 or above.
3300 * Support for RMI's XLR processor is now available through the
3301 -march=xlr and -mtune=xlr options.
3302 * 64-bit targets can now perform 128-bit multiplications inline,
3303 instead of relying on a libgcc function.
3304 * Native GNU/Linux toolchains now support -march=native and
3305 -mtune=native, which select the host processor.
3306 * GCC now supports the R10K, R12K, R14K and R16K processors. The
3307 canonical -march= and -mtune= names for these processors are
3308 r10000, r12000, r14000 and r16000 respectively.
3309 * GCC can now work around the side effects of speculative execution
3310 on R10K processors. Please see the documentation of the
3311 -mr10k-cache-barrier option for details.
3312 * Support for the MIPS64 Release 2 instruction set has been added.
3313 The option -march=mips64r2 enables generation of these
3315 * GCC now supports Cavium Networks' Octeon processor. This support is
3316 available through the -march=octeon and -mtune=octeon options.
3317 * GCC now supports STMicroelectronics' Loongson 2E/2F processors. The
3318 canonical -march= and -mtune= names for these processors are
3319 loongson2e and loongson2f.
3323 Picochip is a 16-bit processor. A typical picoChip contains over 250
3324 small cores, each with small amounts of memory. There are three
3325 processor variants (STAN, MEM and CTRL) with different instruction sets
3326 and memory configurations and they can be chosen using the -mae option.
3328 This port is intended to be a "C" only port.
3330 Power Architecture and PowerPC
3332 * GCC now supports the e300c2, e300c3 and e500mc processors.
3333 * GCC now supports Xilinx processors with a single-precision FPU.
3334 * Decimal floating point is now supported for e500 processors.
3336 S/390, zSeries and System z9/z10
3338 * Support for the IBM System z10 EC/BC processor has been added. When
3339 using the -march=z10 option, the compiler will generate code making
3340 use of instructions provided by the General-Instruction-Extension
3341 Facility and the Execute-Extension Facility.
3345 * GCC now supports the thread-local storage mechanism used on
3350 * GCC now supports thread-local storage (TLS) for Xtensa processor
3351 configurations that include the Thread Pointer option. TLS also
3352 requires support from the assembler and linker; this support is
3353 provided in the GNU binutils beginning with version 2.19.
3355 Documentation improvements
3357 Other significant improvements
3361 This is the [13]list of problem reports (PRs) from GCC's bug tracking
3362 system that are known to be fixed in the 4.4.1 release. This list might
3363 not be complete (that is, it is possible that some PRs that have been
3364 fixed are not listed here).
3368 This is the [14]list of problem reports (PRs) from GCC's bug tracking
3369 system that are known to be fixed in the 4.4.2 release. This list might
3370 not be complete (that is, it is possible that some PRs that have been
3371 fixed are not listed here).
3375 This is the [15]list of problem reports (PRs) from GCC's bug tracking
3376 system that are known to be fixed in the 4.4.3 release. This list might
3377 not be complete (that is, it is possible that some PRs that have been
3378 fixed are not listed here).
3382 This is the [16]list of problem reports (PRs) from GCC's bug tracking
3383 system that are known to be fixed in the 4.4.4 release. This list might
3384 not be complete (that is, it is possible that some PRs that have been
3385 fixed are not listed here).
3389 This is the [17]list of problem reports (PRs) from GCC's bug tracking
3390 system that are known to be fixed in the 4.4.5 release. This list might
3391 not be complete (that is, it is possible that some PRs that have been
3392 fixed are not listed here).
3396 This is the [18]list of problem reports (PRs) from GCC's bug tracking
3397 system that are known to be fixed in the 4.4.6 release. This list might
3398 not be complete (that is, it is possible that some PRs that have been
3399 fixed are not listed here).
3403 This is the [19]list of problem reports (PRs) from GCC's bug tracking
3404 system that are known to be fixed in the 4.4.7 release. This list might
3405 not be complete (that is, it is possible that some PRs that have been
3406 fixed are not listed here).
3409 For questions related to the use of GCC, please consult these web
3410 pages and the [20]GCC manuals. If that fails, the
3411 [21]gcc-help@gcc.gnu.org mailing list might help. Comments on these
3412 web pages and the development of GCC are welcome on our developer
3413 list at [22]gcc@gcc.gnu.org. All of [23]our lists have public
3416 Copyright (C) [24]Free Software Foundation, Inc. Verbatim copying and
3417 distribution of this entire article is permitted in any medium,
3418 provided this notice is preserved.
3420 These pages are [25]maintained by the GCC team. Last modified
3425 1. http://gcc.gnu.org/gcc-4.4/changes.html#4.4.7
3426 2. http://gcc.gnu.org/gcc-4.3/changes.html#obsoleted
3427 3. http://gcc.gnu.org/gcc-4.4/porting_to.html
3428 4. http://gcc.gnu.org/wiki/Graphite
3429 5. http://openmp.org/wp/openmp-specifications/
3430 6. http://www.open-std.org/jtc1/sc22/wg14/www/docs/n1040.pdf
3431 7. http://gcc.gnu.org/gcc-4.4/cxx0x_status.html
3432 8. http://gcc.gnu.org/onlinedocs/libstdc++/manual/status.html#id476343
3433 9. http://gcc.gnu.org/onlinedocs/gfortran/Preprocessing-Options.html
3434 10. http://gcc.gnu.org/onlinedocs/gfortran/Error-and-Warning-Options.html#index-g_t_0040code_007bWarray-temporaries_007d-125
3435 11. http://gcc.gnu.org/onlinedocs/gfortran/Code-Gen-Options.html#index-g_t_0040code_007bfcheck-array-temporaries_007d-221
3436 12. http://gcc.gnu.org/onlinedocs/gfortran/Fortran-Dialect-Options.html#index-g_t_0040code_007bbackslash_007d-34
3437 13. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.1
3438 14. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.2
3439 15. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.3
3440 16. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.4
3441 17. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.5
3442 18. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.6
3443 19. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.7
3444 20. http://gcc.gnu.org/onlinedocs/
3445 21. mailto:gcc-help@gcc.gnu.org
3446 22. mailto:gcc@gcc.gnu.org
3447 23. http://gcc.gnu.org/lists.html
3448 24. http://www.fsf.org/
3449 25. http://gcc.gnu.org/about.html
3450 26. http://validator.w3.org/check/referer
3451 ======================================================================
3452 http://gcc.gnu.org/gcc-4.3/index.html
3454 GCC 4.3 Release Series
3458 The [1]GNU project and the GCC developers are pleased to announce the
3459 release of GCC 4.3.6.
3461 This release is a bug-fix release, containing fixes for regressions in
3462 GCC 4.3.5 relative to previous releases of GCC.
3467 Jun 27, 2011 ([2]changes)
3470 May 22, 2010 ([3]changes)
3473 August 4, 2009 ([4]changes)
3476 January 24, 2009 ([5]changes)
3479 August 27, 2008 ([6]changes)
3482 June 6, 2008 ([7]changes)
3485 March 5, 2008 ([8]changes)
3487 References and Acknowledgements
3489 GCC used to stand for the GNU C Compiler, but since the compiler
3490 supports several other languages aside from C, it now stands for the
3491 GNU Compiler Collection.
3493 A list of [9]successful builds is updated as new information becomes
3496 The GCC developers would like to thank the numerous people that have
3497 contributed new features, improvements, bug fixes, and other changes as
3498 well as test results to GCC. This [10]amazing group of volunteers is
3499 what makes GCC successful.
3501 For additional information about GCC please refer to the [11]GCC
3502 project web site or contact the [12]GCC development mailing list.
3504 To obtain GCC please use [13]our mirror sites or [14]our SVN server.
3507 For questions related to the use of GCC, please consult these web
3508 pages and the [15]GCC manuals. If that fails, the
3509 [16]gcc-help@gcc.gnu.org mailing list might help. Comments on these
3510 web pages and the development of GCC are welcome on our developer
3511 list at [17]gcc@gcc.gnu.org. All of [18]our lists have public
3514 Copyright (C) [19]Free Software Foundation, Inc. Verbatim copying and
3515 distribution of this entire article is permitted in any medium,
3516 provided this notice is preserved.
3518 These pages are [20]maintained by the GCC team. Last modified
3523 1. http://www.gnu.org/
3524 2. http://gcc.gnu.org/gcc-4.3/changes.html
3525 3. http://gcc.gnu.org/gcc-4.3/changes.html
3526 4. http://gcc.gnu.org/gcc-4.3/changes.html
3527 5. http://gcc.gnu.org/gcc-4.3/changes.html
3528 6. http://gcc.gnu.org/gcc-4.3/changes.html
3529 7. http://gcc.gnu.org/gcc-4.3/changes.html
3530 8. http://gcc.gnu.org/gcc-4.3/changes.html
3531 9. http://gcc.gnu.org/gcc-4.3/buildstat.html
3532 10. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
3533 11. http://gcc.gnu.org/index.html
3534 12. mailto:gcc@gcc.gnu.org
3535 13. http://gcc.gnu.org/mirrors.html
3536 14. http://gcc.gnu.org/svn.html
3537 15. http://gcc.gnu.org/onlinedocs/
3538 16. mailto:gcc-help@gcc.gnu.org
3539 17. mailto:gcc@gcc.gnu.org
3540 18. http://gcc.gnu.org/lists.html
3541 19. http://www.fsf.org/
3542 20. http://gcc.gnu.org/about.html
3543 21. http://validator.w3.org/check/referer
3544 ======================================================================
3545 http://gcc.gnu.org/gcc-4.3/changes.html
3547 GCC 4.3 Release Series
3548 Changes, New Features, and Fixes
3550 The latest release in the 4.3 release series is [1]GCC 4.3.5.
3554 * GCC requires the [2]GMP and [3]MPFR libraries for building all the
3555 various front-end languages it supports. See the [4]prerequisites
3556 page for version requirements.
3557 * ColdFire targets now treat long double as having the same format as
3558 double. In earlier versions of GCC, they used the 68881 long double
3560 * The m68k-uclinux target now uses the same calling conventions as
3561 m68k-linux-gnu. You can select the original calling conventions by
3562 configuring for m68k-uclinuxoldabi instead. Note that
3563 m68k-uclinuxoldabi also retains the original 80-bit long double on
3565 * The -fforce-mem option has been removed because it has had no
3566 effect in the last few GCC releases.
3567 * The i386 -msvr3-shlib option has been removed since it is no longer
3569 * Fastcall for i386 has been changed not to pass aggregate arguments
3570 in registers, following Microsoft compilers.
3571 * Support for the AOF assembler has been removed from the ARM back
3572 end; this affects only the targets arm-semi-aof and armel-semi-aof,
3573 which are no longer recognized. We removed these targets without a
3574 deprecation period because we discovered that they have been
3575 unusable since GCC 4.0.0.
3576 * Support for the TMS320C3x/C4x processor (targets c4x-* and tic4x-*)
3577 has been removed. This support had been deprecated since GCC 4.0.0.
3578 * Support for a number of older systems and recently unmaintained or
3579 untested target ports of GCC has been declared obsolete in GCC 4.3.
3580 Unless there is activity to revive them, the next release of GCC
3581 will have their sources permanently removed.
3582 All GCC ports for the following processor architectures have been
3585 The following aliases for processor architectures have been
3586 declared obsolete. Users should use the indicated generic target
3587 names instead, with compile-time options such as -mcpu or
3588 configure-time options such as --with-cpu to control the
3589 configuration more precisely.
3590 + strongarm*-*-*, ep9312*-*-*, xscale*-*-* (use arm*-*-*
3592 + parisc*-*-* (use hppa*-*-* instead).
3593 + m680[012]0-*-* (use m68k-*-* instead).
3594 All GCC ports for the following operating systems have been
3598 + GNU/Linux using the a.out object format (*-*-linux*aout*)
3599 + GNU/Linux using version 1 of the GNU C Library
3601 + Solaris versions before Solaris 7 (*-*-solaris2.[0-6],
3602 *-*-solaris2.[0-6].*)
3603 + Miscellaneous System V (*-*-sysv*)
3604 + WindISS (*-*-windiss*)
3605 Also, those for some individual systems on particular architectures
3606 have been obsoleted:
3607 + UNICOS/mk on DEC Alpha (alpha*-*-unicosmk*)
3608 + CRIS with a.out object format (cris-*-aout)
3609 + BSD 4.3 on PA-RISC (hppa1.1-*-bsd*)
3610 + OSF/1 on PA-RISC (hppa1.1-*-osf*)
3611 + PRO on PA-RISC (hppa1.1-*-pro*)
3612 + Sequent PTX on IA32 (i[34567]86-sequent-ptx4*,
3613 i[34567]86-sequent-sysv4*)
3614 + SCO Open Server 5 on IA32 (i[34567]86-*-sco3.2v5*)
3615 + UWIN on IA32 (i[34567]86-*-uwin*) (support for UWIN as a host
3616 was previously [5]removed in 2001, leaving only the support
3617 for UWIN as a target now being deprecated)
3618 + ChorusOS on PowerPC (powerpc-*-chorusos*)
3619 + All VAX configurations apart from NetBSD and OpenBSD
3620 (vax-*-bsd*, vax-*-sysv*, vax-*-ultrix*)
3621 * The [6]-Wconversion option has been modified. Its purpose now is to
3622 warn for implicit conversions that may alter a value. This new
3623 behavior is available for both C and C++. Warnings about
3624 conversions between signed and unsigned integers can be disabled by
3625 using -Wno-sign-conversion. In C++, they are disabled by default
3626 unless -Wsign-conversion is explicitly requested. The old behavior
3627 of -Wconversion, that is, warn for prototypes causing a type
3628 conversion that is different from what would happen to the same
3629 argument in the absence of a prototype, has been moved to a new
3630 option -Wtraditional-conversion, which is only available for C.
3631 * The -m386, -m486, -mpentium and -mpentiumpro tuning options have
3632 been removed because they were deprecated for more than 3 GCC major
3633 releases. Use -mtune=i386, -mtune=i486, -mtune=pentium or
3634 -mtune=pentiumpro as a replacement.
3635 * The -funsafe-math-optimizations option now automatically turns on
3636 -fno-trapping-math in addition to -fno-signed-zeros, as it enables
3637 reassociation and thus may introduce or remove traps.
3638 * The -ftree-vectorize option is now on by default under -O3. In
3639 order to generate code for a SIMD extension, it has to be enabled
3640 as well: use -maltivec for PowerPC platforms and -msse/-msse2 for
3642 * More information on porting to GCC 4.3 from previous versions of
3643 GCC can be found in the [7]porting guide for this release.
3645 General Optimizer Improvements
3647 * The GCC middle-end has been integrated with the [8]MPFR library.
3648 This allows GCC to evaluate and replace at compile-time calls to
3649 built-in math functions having constant arguments with their
3650 mathematically equivalent results. In making use of [9]MPFR, GCC
3651 can generate correct results regardless of the math library
3652 implementation or floating point precision of the host platform.
3653 This also allows GCC to generate identical results regardless of
3654 whether one compiles in native or cross-compile configurations to a
3655 particular target. The following built-in functions take advantage
3656 of this new capability: acos, acosh, asin, asinh, atan2, atan,
3657 atanh, cbrt, cos, cosh, drem, erf, erfc, exp10, exp2, exp, expm1,
3658 fdim, fma, fmax, fmin, gamma_r, hypot, j0, j1, jn, lgamma_r, log10,
3659 log1p, log2, log, pow10, pow, remainder, remquo, sin, sincos, sinh,
3660 tan, tanh, tgamma, y0, y1 and yn. The float and long double
3661 variants of these functions (e.g. sinf and sinl) are also handled.
3662 The sqrt and cabs functions with constant arguments were already
3663 optimized in prior GCC releases. Now they also use [10]MPFR.
3664 * A new forward propagation pass on RTL was added. The new pass
3665 replaces several slower transformations, resulting in compile-time
3666 improvements as well as better code generation in some cases.
3667 * A new command-line switch -frecord-gcc-switches has been added to
3668 GCC, although it is only enabled for some targets. The switch
3669 causes the command line that was used to invoke the compiler to be
3670 recorded into the object file that is being created. The exact
3671 format of this recording is target and binary file format
3672 dependent, but it usually takes the form of a note section
3673 containing ASCII text. The switch is related to the -fverbose-asm
3674 switch, but that one only records the information in the assembler
3675 output file as comments, so the information never reaches the
3677 * The inliner heuristic is now aware of stack frame consumption. New
3678 command-line parameters --param large-stack-frame and --param
3679 large-stack-frame-growth can be used to limit stack frame size
3680 growth caused by inlining.
3681 * During feedback directed optimizations, the expected block size the
3682 memcpy, memset and bzero functions operate on is discovered and for
3683 cases of commonly used small sizes, specialized inline code is
3685 * __builtin_expect no longer requires its argument to be a compile
3687 * Interprocedural optimization was reorganized to work on functions
3688 in SSA form. This enables more precise and cheaper dataflow
3689 analysis and makes writing interprocedural optimizations easier.
3690 The following improvements have been implemented on top of this
3692 + Pre-inline optimization: Selected local optimization passes
3693 are run before the inliner (and other interprocedural passes)
3694 are executed. This significantly improves the accuracy of code
3695 growth estimates used by the inliner and reduces the overall
3696 memory footprint for large compilation units.
3697 + Early inlining (a simple bottom-up inliner pass inlining only
3698 functions whose body is smaller than the expected call
3699 overhead) is now executed with the early optimization passes,
3700 thus inlining already optimized function bodies into an
3701 unoptimized function that is subsequently optimized by early
3702 optimizers. This enables the compiler to quickly eliminate
3703 abstraction penalty in C++ programs.
3704 + Interprocedural constant propagation now operate on SSA form
3705 increasing accuracy of the analysis.
3706 * A new internal representation for GIMPLE statements has been
3707 contributed, resulting in compile-time memory savings.
3708 * The vectorizer was enhanced to support vectorization of outer
3709 loops, intra-iteration parallelism (loop-aware SLP), vectorization
3710 of strided accesses and loops with multiple data-types. Run-time
3711 dependency testing using loop versioning was added. The cost model,
3712 turned on by -fvect-cost-model, was developed.
3714 New Languages and Language specific improvements
3716 * We have added new command-line options
3717 -finstrument-functions-exclude-function-list and
3718 -finstrument-functions-exclude-file-list. They provide more control
3719 over which functions are annotated by the -finstrument-functions
3724 * Implicit conversions between generic vector types are now only
3725 permitted when the two vectors in question have the same number of
3726 elements and compatible element types. (Note that the restriction
3727 involves compatible element types, not implicitly-convertible
3728 element types: thus, a vector type with element type int may not be
3729 implicitly converted to a vector type with element type unsigned
3730 int.) This restriction, which is in line with specifications for
3731 SIMD architectures such as AltiVec, may be relaxed using the flag
3732 -flax-vector-conversions. This flag is intended only as a
3733 compatibility measure and should not be used for new code.
3734 * -Warray-bounds has been added and is now enabled by default for
3735 -Wall . It produces warnings for array subscripts that can be
3736 determined at compile time to be always out of bounds.
3737 -Wno-array-bounds will disable the warning.
3738 * The constructor and destructor function attributes now accept
3739 optional priority arguments which control the order in which the
3740 constructor and destructor functions are run.
3741 * New [11]command-line options -Wtype-limits,
3742 -Wold-style-declaration, -Wmissing-parameter-type, -Wempty-body,
3743 -Wclobbered and -Wignored-qualifiers have been added for finer
3744 control of the diverse warnings enabled by -Wextra.
3745 * A new function attribute alloc_size has been added to mark up
3746 malloc style functions. For constant sized allocations this can be
3747 used to find out the size of the returned pointer using the
3748 __builtin_object_size() function for buffer overflow checking and
3749 similar. This supplements the already built-in malloc and calloc
3750 constant size handling.
3751 * Integer constants written in binary are now supported as a GCC
3752 extension. They consist of a prefix 0b or 0B, followed by a
3753 sequence of 0 and 1 digits.
3754 * A new predefined macro __COUNTER__ has been added. It expands to
3755 sequential integral values starting from 0. In conjunction with the
3756 ## operator, this provides a convenient means to generate unique
3758 * A new command-line option -fdirectives-only has been added. It
3759 enables a special preprocessing mode which improves the performance
3760 of applications like distcc and ccache.
3761 * Fixed-point data types and operators have been added. They are
3762 based on Chapter 4 of the Embedded-C specification (n1169.pdf).
3763 Currently, only MIPS targets are supported.
3764 * Decimal floating-point arithmetic based on draft ISO/IEC TR 24732,
3765 N1241, is now supported as a GCC extension to C for targets
3766 i[34567]86-*-linux-gnu, powerpc*-*-linux-gnu, s390*-ibm-linux-gnu,
3767 and x86_64-*-linux-gnu. The feature introduces new data types
3768 _Decimal32, _Decimal64, and _Decimal128 with constant suffixes DF,
3773 * [12]Experimental support for the upcoming ISO C++ standard, C++0x.
3774 * -Wc++0x-compat has been added and is now enabled by default for
3775 -Wall. It produces warnings for constructs whose meaning differs
3776 between ISO C++ 1998 and C++0x.
3777 * The -Wparentheses option now works for C++ as it does for C. It
3778 warns if parentheses are omitted when operators with confusing
3779 precedence are nested. It also warns about ambiguous else
3780 statements. Since -Wparentheses is enabled by -Wall, this may cause
3781 additional warnings with existing C++ code which uses -Wall. These
3782 new warnings may be disabled by using -Wall -Wno-parentheses.
3783 * The -Wmissing-declarations now works for C++ as it does for C.
3784 * The -fvisibility-ms-compat flag was added, to make it easier to
3785 port larger projects using shared libraries from Microsoft's Visual
3786 Studio to ELF and Mach-O systems.
3787 * C++ attribute handling has been overhauled for template arguments
3788 (ie dependent types). In particular, __attribute__((aligned(T)));
3789 works for C++ types.
3791 Runtime Library (libstdc++)
3793 * [13]Experimental support for the upcoming ISO C++ standard, C++0x.
3794 * Support for TR1 mathematical special functions and regular
3795 expressions. ([14]Implementation status of TR1)
3796 * Default what implementations give more elaborate exception strings
3797 for bad_cast, bad_typeid, bad_exception, and bad_alloc.
3798 * Header dependencies have been streamlined, reducing unnecessary
3799 includes and pre-processed bloat.
3800 * Variadic template implementations of items in <tuple> and
3802 * An experimental [15]parallel mode has been added. This is a
3803 parallel implementation of many C++ Standard library algorithms,
3804 like std::accumulate, std::for_each, std::transform, or std::sort,
3805 to give but four examples. These algorithms can be substituted for
3806 the normal (sequential) libstdc++ algorithms on a piecemeal basis,
3807 or all existing algorithms can be transformed via the
3808 -D_GLIBCXX_PARALLEL macro.
3809 * Debug mode versions of classes in <unordered_set> and
3811 * Formal deprecation of <ext/hash_set> and <ext/hash_map>, which are
3812 now <backward/hash_set> and <backward/hash_map>. This code:
3813 #include <ext/hash_set>
3814 __gnu_cxx::hash_set<int> s;
3816 Can be transformed (in order of preference) to:
3817 #include <tr1/unordered_set>
3818 std::tr1::unordered_set<int> s;
3821 #include <backward/hash_set>
3822 __gnu_cxx::hash_set<int> s;
3824 Similar transformations apply to __gnu_cxx::hash_map,
3825 __gnu_cxx::hash_multimap, __gnu_cxx::hash_set,
3826 __gnu_cxx::hash_multiset.
3830 * Due to the fact that the [16]GMP and [17]MPFR libraries are
3831 required for all languages, Fortran is no longer special in this
3832 regard and is available by default.
3833 * The [18]-fexternal-blas option has been added, which generates
3834 calls to BLAS routines for intrinsic matrix operations such as
3835 matmul rather than using the built-in algorithms.
3836 * Support to give a backtrace (compiler flag -fbacktrace or
3837 environment variable GFORTRAN_ERROR_BACKTRACE; on glibc systems
3838 only) or a core dump (-fdump-core, GFORTRAN_ERROR_DUMPCORE) when a
3839 run-time error occured.
3840 * GNU Fortran now defines __GFORTRAN__ when it runs the C
3842 * The [19]-finit-local-zero, -finit-real, -finit-integer,
3843 -finit-character, and -finit-logical options have been added, which
3844 can be used to initialize local variables.
3845 * The intrinsic procedures [20]GAMMA and [21]LGAMMA have been added,
3846 which calculate the Gamma function and its logarithm. Use EXTERNAL
3847 gamma if you want to use your own gamma function.
3848 * GNU Fortran now regards the backslash character as literal (as
3849 required by the Fortran 2003 standard); using [22]-fbackslash GNU
3850 Fortran interprets backslashes as C-style escape characters.
3851 * The [23]interpretation of binary, octal and hexadecimal (BOZ)
3852 literal constants has been changed. Before they were always
3853 interpreted as integer; now they are bit-wise transferred as
3854 argument of INT, REAL, DBLE and CMPLX as required by the Fortran
3855 2003 standard, and for real and complex variables in DATA
3856 statements or when directly assigned to real and complex variables.
3857 Everywhere else and especially in expressions they are still
3858 regarded as integer constants.
3859 * Fortran 2003 support has been extended:
3860 + Intrinsic statements IMPORT, PROTECTED, VALUE and VOLATILE
3862 + Intrinsic module ISO_ENV_FORTRAN
3863 + Interoperability with C (ISO C Bindings)
3864 + ABSTRACT INTERFACES and PROCEDURE statements (without POINTER
3870 * GCJ now uses the Eclipse Java compiler for its Java parsing needs.
3871 This enables the use of all 1.5 language features, and fixes most
3872 existing front end bugs.
3873 * libgcj now supports all 1.5 language features which require runtime
3874 support: foreach, enum, annotations, generics, and auto-boxing.
3875 * We've made many changes to the tools shipped with gcj.
3876 + The old jv-scan tool has been removed. This tool never really
3877 worked properly. There is no replacement.
3878 + gcjh has been rewritten. Some of its more obscure options no
3879 longer work, but are still recognized in an attempt at
3880 compatibility. gjavah is a new program with similar
3881 functionality but different command-line options.
3882 + grmic and grmiregistry have been rewritten. grmid has been
3884 + gjar replaces the old fastjar.
3885 + gjarsigner (used for signing jars), gkeytool (used for key
3886 management), gorbd (for CORBA), gserialver (computes
3887 serialization UIDs), and gtnameserv (also for CORBA) are now
3889 * The ability to dump the contents of the java run time heap to a
3890 file for off-line analysis has been added. The heap dumps may be
3891 analyzed with the new gc-analyze tool. They may be generated on
3892 out-of-memory conditions or on demand and are controlled by the new
3893 run time class gnu.gcj.util.GCInfo.
3894 * java.util.TimeZone can now read files from /usr/share/zoneinfo to
3895 provide correct, updated, timezone information. This means that
3896 packagers no longer have to update libgcj when a time zone change
3899 New Targets and Target Specific Improvements
3903 * Tuning for Intel Core 2 processors is available via -mtune=core2
3905 * Tuning for AMD Geode processors is available via -mtune=geode and
3907 * Code generation of block move (memcpy) and block set (memset) was
3908 rewritten. GCC can now pick the best algorithm (loop, unrolled
3909 loop, instruction with rep prefix or a library call) based on the
3910 size of the block being copied and the CPU being optimized for. A
3911 new option -minline-stringops-dynamically has been added. With this
3912 option string operations of unknown size are expanded such that
3913 small blocks are copied by in-line code, while for large blocks a
3914 library call is used. This results in faster code than
3915 -minline-all-stringops when the library implementation is capable
3916 of using cache hierarchy hints. The heuristic choosing the
3917 particular algorithm can be overwritten via -mstringop-strategy.
3918 Newly also memset of values different from 0 is inlined.
3919 * GCC no longer places the cld instruction before string operations.
3920 Both i386 and x86-64 ABI documents mandate the direction flag to be
3921 clear at the entry of a function. It is now invalid to set the flag
3922 in asm statement without reseting it afterward.
3923 * Support for SSSE3 built-in functions and code generation are
3924 available via -mssse3.
3925 * Support for SSE4.1 built-in functions and code generation are
3926 available via -msse4.1.
3927 * Support for SSE4.2 built-in functions and code generation are
3928 available via -msse4.2.
3929 * Both SSE4.1 and SSE4.2 support can be enabled via -msse4.
3930 * A new set of options -mpc32, -mpc64 and -mpc80 have been added to
3931 allow explicit control of x87 floating point precision.
3932 * Support for __float128 (TFmode) IEEE quad type and corresponding
3933 TCmode IEEE complex quad type is available via the soft-fp library
3934 on x86_64 targets. This includes basic arithmetic operations
3935 (addition, subtraction, negation, multiplication and division) on
3936 __float128 real and TCmode complex values, the full set of IEEE
3937 comparisons between __float128 values, conversions to and from
3938 float, double and long double floating point types, as well as
3939 conversions to and from signed or unsigned integer, signed or
3940 unsigned long integer and signed or unsigned quad (TImode) integer
3941 types. Additionally, all operations generate the full set of IEEE
3942 exceptions and support the full set of IEEE rounding modes.
3943 * GCC can now utilize the ACML library for vectorizing calls to a set
3944 of C99 functions on x86_64 if -mveclibabi=acml is specified and you
3945 link to an ACML ABI compatible library.
3949 * Compiler and Library support for Thumb-2 and the ARMv7 architecture
3956 * Compiler and Library support for the CRIS v32 architecture, as
3957 found in Axis Communications ETRAX FS and ARTPEC-3 chips, has been
3960 Configuration changes
3962 * The cris-*-elf target now includes support for CRIS v32, including
3963 libraries, through the -march=v32 option.
3964 * A new crisv32-*-elf target defaults to generate code for CRIS v32.
3965 * A new crisv32-*-linux* target defaults to generate code for CRIS
3967 * The cris-*-aout target has been obsoleted.
3969 Improved support for built-in functions
3971 * GCC can now use the lz and swapwbr instructions to implement the
3972 __builtin_clz, __builtin_ctz and __builtin_ffs family of functions.
3973 * __builtin_bswap32 is now implemented using the swapwb instruction,
3980 * Support for several new ColdFire processors has been added. You can
3981 generate code for them using the new -mcpu option.
3982 * All targets now support ColdFire processors.
3983 * m68k-uclinux targets have improved support for C++ constructors and
3984 destructors, and for shared libraries.
3985 * It is now possible to set breakpoints on the first or last line of
3986 a function, even if there are no statements on that line.
3990 * Support for sibling calls has been added.
3991 * More use is now made of the ColdFire mov3q instruction.
3992 * __builtin_clz is now implemented using the ff1 ColdFire
3993 instruction, when available.
3994 * GCC now honors the -m68010 option. 68010 code now uses clr rather
3995 than move to zero volatile memory.
3996 * 68020 targets and above can now use symbol(index.size*scale)
3997 addresses for indexed array accesses. Earlier compilers would
3998 always load the symbol into a base register first.
4000 Configuration changes
4002 * All m68k and ColdFire targets now allow the default processor to be
4003 set at configure time using --with-cpu.
4004 * A --with-arch configuration option has been added. This option
4005 allows you to restrict a target to ColdFire or non-ColdFire
4010 * An __mcfv*__ macro is now defined for all ColdFire targets.
4011 (Earlier versions of GCC only defined __mcfv4e__.)
4012 * __mcf_cpu_*, __mcf_family_* and __mcffpu__ macros have been added.
4013 * All targets now define __mc68010 and __mc68010__ when generating
4016 Command-line changes
4018 * New command-line options -march, -mcpu, -mtune and -mhard-float
4019 have been added. These options apply to both m68k and ColdFire
4021 * -mno-short, -mno-bitfield and -mno-rtd are now accepted as negative
4022 versions of -mshort, etc.
4023 * -fforce-addr has been removed. It is now ignored by the compiler.
4027 * ColdFire targets now try to maintain a 4-byte-aligned stack where
4029 * m68k-uclinux targets now try to avoid situations that lead to the
4030 load-time error: BINFMT_FLAT: reloc outside program.
4034 Changes to existing configurations
4036 * libffi and libjava now support all three GNU/Linux ABIs: o32, n32
4037 and n64. Every GNU/Linux configuration now builds these libraries
4039 * GNU/Linux configurations now generate -mno-shared code unless
4040 overridden by -fpic, -fPIC, -fpie or -fPIE.
4041 * mipsisa32*-linux-gnu configurations now generate hard-float code by
4042 default, just like other mipsisa32* and mips*-linux-gnu
4043 configurations. You can build a soft-float version of any
4044 mips*-linux-gnu configuration by passing --with-float=soft to
4046 * mips-wrs-vxworks now supports run-time processes (RTPs).
4048 Changes to existing command-line options
4050 * The -march and -mtune options no longer accept 24k as a processor
4051 name. Please use 24kc, 24kf2_1 or 24kf1_1 instead.
4052 * The -march and -mtune options now accept 24kf2_1, 24kef2_1 and
4053 34kf2_1 as synonyms for 24kf, 24kef and 34kf respectively. The
4054 options also accept 24kf1_1, 24kef1_1 and 34kf1_1 as synonyms for
4055 24kx, 24kex and 34kx.
4059 GCC now supports the following configurations:
4060 * mipsisa32r2*-linux-gnu*, which generates MIPS32 revision 2 code by
4061 default. Earlier releases also recognized this configuration, but
4062 they treated it in the same way as mipsisa32*-linux-gnu*. Note that
4063 you can customize any mips*-linux-gnu* configuration to a
4064 particular ISA or processor by passing an appropriate --with-arch
4065 option to configure.
4066 * mipsisa*-sde-elf*, which provides compatibility with MIPS
4067 Technologies' SDE toolchains. The configuration uses the SDE
4068 libraries by default, but you can use it like other newlib-based
4069 ELF configurations by passing --with-newlib to configure. It is the
4070 only configuration besides mips64vr*-elf* to build MIPS16 as well
4071 as non-MIPS16 libraries.
4072 * mipsisa*-elfoabi*, which is similar to the general mipsisa*-elf*
4073 configuration, but uses the o32 and o64 ABIs instead of the 32-bit
4074 and 64-bit forms of the EABI.
4076 New processors and application-specific extensions
4078 * Support for the SmartMIPS ASE is available through the new
4080 * Support for revision 2 of the DSP ASE is available through the new
4081 -mdspr2 option. A new preprocessor macro called __mips_dsp_rev
4082 indicates the revision of the ASE in use.
4083 * Support for the 4KS and 74K families of processors is available
4084 through the -march and -mtune options.
4086 Improved support for built-in functions
4088 * GCC can now use load-linked, store-conditional and sync
4089 instructions to implement atomic built-in functions such as
4090 __sync_fetch_and_add. The memory reference must be 4 bytes wide for
4091 32-bit targets and either 4 or 8 bytes wide for 64-bit targets.
4092 * GCC can now use the clz and dclz instructions to implement the
4093 __builtin_ctz and __builtin_ffs families of functions.
4094 * There is a new __builtin___clear_cache function for flushing the
4095 instruction cache. GCC expands this function inline on MIPS32
4096 revision 2 targets, otherwise it calls the function specified by
4101 * GCC can now compile objects that contain a mixture of MIPS16 and
4102 non-MIPS16 code. There are two new attributes, mips16 and nomips16,
4103 for specifying which mode a function should use.
4104 * A new option called -minterlink-mips16 makes non-MIPS16 code
4105 link-compatible with MIPS16 code.
4106 * After many bug fixes, the long-standing MIPS16 -mhard-float support
4107 should now work fairly reliably.
4108 * GCC can now use the MIPS16e save and restore instructions.
4109 * -fsection-anchors now works in MIPS16 mode. MIPS16 code compiled
4110 with -G0 -fsection-anchors is often smaller than code compiled with
4111 -G8. However, please note that you must usually compile all objects
4112 in your application with the same -G option; see the documentation
4114 * A new option called-mcode-readable specifies which instructions are
4115 allowed to load from the code segment. -mcode-readable=yes is the
4116 default and says that any instruction may load from the code
4117 segment. The other alternatives are -mcode-readable=pcrel, which
4118 says that only PC-relative MIPS16 instructions may load from the
4119 code segment, and -mcode-readable=no, which says that no
4120 instruction may do so. Please see the documentation for more
4121 details, including example uses.
4123 Small-data improvements
4125 There are three new options for controlling small data:
4126 * -mno-extern-sdata, which disables small-data accesses for
4127 externally-defined variables. Code compiled with -Gn
4128 -mno-extern-sdata will be link-compatible with any -G setting
4129 between -G0 and -Gn inclusive.
4130 * -mno-local-sdata, which disables the use of small-data sections for
4131 data that is not externally visible. This option can be a useful
4132 way of reducing small-data usage in less performance-critical parts
4134 * -mno-gpopt, which disables the use of the $gp register while still
4135 honoring the -G limit when placing externally-visible data. This
4136 option implies -mno-extern-sdata and -mno-local-sdata and it can be
4137 useful in situations where $gp does not necessarily hold the
4140 Miscellaneous improvements
4142 * There is a new option called -mbranch-cost for tweaking the
4143 perceived cost of branches.
4144 * If GCC is configured to use a version of GAS that supports the
4145 .gnu_attribute directive, it will use that directive to record
4146 certain properties of the output code. .gnu_attribute is new to GAS
4148 * There are two new function attributes, near and far, for overriding
4149 the command-line setting of -mlong-calls on a function-by-function
4151 * -mfp64, which previously required a 64-bit target, now works with
4152 MIPS32 revision 2 targets as well. The mipsisa*-elfoabi* and
4153 mipsisa*-sde-elf* configurations provide suitable library support.
4154 * GCC now recognizes the -mdmx and -mmt options and passes them down
4155 to the assembler. It does nothing else with the options at present.
4157 SPU (Synergistic Processor Unit) of the Cell Broadband Engine Architecture
4160 * Support has been added for this new architecture.
4162 RS6000 (POWER/PowerPC)
4164 * Support for the PowerPC 750CL paired-single instructions has been
4165 added with a new powerpc-*-linux*paired* target configuration. It
4166 is enabled by an associated -mpaired option and can be accessed
4167 using new built-in functions.
4168 * Support for auto-detecting architecture and system configuration to
4169 auto-select processor optimization tuning.
4170 * Support for VMX on AIX 5.3 has been added.
4171 * Support for AIX Version 6.1 has been added.
4173 S/390, zSeries and System z9
4175 * Support for the IBM System z9 EC/BC processor (z9 GA3) has been
4176 added. When using the -march=z9-ec option, the compiler will
4177 generate code making use of instructions provided by the decimal
4178 floating point facility and the floating point conversion facility
4179 (pfpo). Besides the instructions used to implement decimal floating
4180 point operations these facilities also contain instructions to move
4181 between general purpose and floating point registers and to modify
4182 and copy the sign-bit of floating point values.
4183 * When the -march=z9-ec option is used the new
4184 -mhard-dfp/-mno-hard-dfp options can be used to specify whether the
4185 decimal floating point hardware instructions will be used or not.
4186 If none of them is given the hardware support is enabled by
4188 * The -mstack-guard option can now be omitted when using stack
4189 checking via -mstack-size in order to let GCC choose a sensible
4190 stack guard value according to the frame size of each function.
4191 * Various changes to improve performance of generated code have been
4192 implemented, including:
4193 + The condition code set by an add logical with carry
4194 instruction is now available for overflow checks like: a + b +
4196 + The test data class instruction is now used to implement
4197 sign-bit and infinity checks of binary and decimal floating
4202 * Support for the Sun UltraSPARC T2 (Niagara 2) processor has been
4207 * Stack unwinding for exception handling now uses by default a
4208 specialized version of DWARF unwinding. This is not
4209 binary-compatible with the setjmp/longjmp (sjlj) unwinding used for
4210 Xtensa with previous versions of GCC.
4211 * For Xtensa processors that include the Conditional Store option,
4212 the built-in functions for atomic memory access are now implemented
4213 using S32C1I instructions.
4214 * If the Xtensa NSA option is available, GCC will use it to implement
4215 the __builtin_ctz and __builtin_clz functions.
4217 Documentation improvements
4219 * Existing libstdc++ documentation has been edited and restructured
4220 into a single DocBook XML manual. The results can be viewed online
4223 Other significant improvements
4225 * The compiler's --help command-line option has been extended so that
4226 it now takes an optional set of arguments. These arguments restrict
4227 the information displayed to specific classes of command-line
4228 options, and possibly only a subset of those options. It is also
4229 now possible to replace the descriptive text associated with each
4230 displayed option with an indication of its current value, or for
4231 binary options, whether it has been enabled or disabled.
4232 Here are some examples. The following will display all the options
4233 controlling warning messages:
4236 Whereas this will display all the undocumented, target specific
4238 --help=target,undocumented
4240 This sequence of commands will display the binary optimizations
4241 that are enabled by -O3:
4242 gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts
4243 gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts
4244 diff /tmp/O2-opts /tmp/O3-opts | grep enabled
4246 * The configure options --with-pkgversion and --with-bugurl have been
4247 added. These allow distributors of GCC to include a
4248 distributor-specific string in manuals and --version output and to
4249 specify the URL for reporting bugs in their versions of GCC.
4253 This is the [25]list of problem reports (PRs) from GCC's bug tracking
4254 system that are known to be fixed in the 4.3.1 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).
4258 Target Specific Changes
4264 * Starting with GCC 4.3.1, decimal floating point variables are
4265 aligned to their natural boundaries when they are passed on the
4268 Command-line changes
4270 * Starting with GCC 4.3.1, the -mcld option has been added to
4271 automatically generate a cld instruction in the prologue of
4272 functions that use string instructions. This option is used for
4273 backward compatibility on some operating systems and can be enabled
4274 by default for 32-bit x86 targets by configuring GCC with the
4275 --enable-cld configure option.
4279 This is the [26]list of problem reports (PRs) from GCC's bug tracking
4280 system that are known to be fixed in the 4.3.2 release. This list might
4281 not be complete (that is, it is possible that some PRs that have been
4282 fixed are not listed here).
4286 This is the [27]list of problem reports (PRs) from GCC's bug tracking
4287 system that are known to be fixed in the 4.3.3 release. This list might
4288 not be complete (that is, it is possible that some PRs that have been
4289 fixed are not listed here).
4293 This is the [28]list of problem reports (PRs) from GCC's bug tracking
4294 system that are known to be fixed in the 4.3.4 release. This list might
4295 not be complete (that is, it is possible that some PRs that have been
4296 fixed are not listed here).
4300 This is the [29]list of problem reports (PRs) from GCC's bug tracking
4301 system that are known to be fixed in the 4.3.5 release. This list might
4302 not be complete (that is, it is possible that some PRs that have been
4303 fixed are not listed here).
4307 This is the [30]list of problem reports (PRs) from GCC's bug tracking
4308 system that are known to be fixed in the 4.3.6 release. This list might
4309 not be complete (that is, it is possible that some PRs that have been
4310 fixed are not listed here).
4313 For questions related to the use of GCC, please consult these web
4314 pages and the [31]GCC manuals. If that fails, the
4315 [32]gcc-help@gcc.gnu.org mailing list might help. Comments on these
4316 web pages and the development of GCC are welcome on our developer
4317 list at [33]gcc@gcc.gnu.org. All of [34]our lists have public
4320 Copyright (C) [35]Free Software Foundation, Inc. Verbatim copying and
4321 distribution of this entire article is permitted in any medium,
4322 provided this notice is preserved.
4324 These pages are [36]maintained by the GCC team. Last modified
4329 1. http://gcc.gnu.org/gcc-4.3/changes.html#4.3.5
4330 2. http://gmplib.org/
4331 3. http://www.mpfr.org/
4332 4. http://gcc.gnu.org/install/prerequisites.html
4333 5. http://gcc.gnu.org/ml/gcc-announce/2001/msg00000.html
4334 6. http://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html#Warning-Options
4335 7. http://gcc.gnu.org/gcc-4.3/porting_to.html
4336 8. http://www.mpfr.org/
4337 9. http://www.mpfr.org/
4338 10. http://www.mpfr.org/
4339 11. http://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html
4340 12. http://gcc.gnu.org/gcc-4.3/cxx0x_status.html
4341 13. http://gcc.gnu.org/gcc-4.3/cxx0x_status.html
4342 14. http://gcc.gnu.org/onlinedocs/libstdc++/manual/bk01pt01ch01.html#m anual.intro.status.standard.tr1
4343 15. http://gcc.gnu.org/onlinedocs/libstdc++/manual/parallel_mode.html
4344 16. http://gmplib.org/
4345 17. http://www.mpfr.org/
4346 18. http://gcc.gnu.org/onlinedocs/gfortran/Code-Gen-Options.html#Code-Gen-Options
4347 19. http://gcc.gnu.org/onlinedocs/gfortran/Code-Gen-Options.html#index-g_t_0040code_007bfinit-local-zero_007d-167
4348 20. http://gcc.gnu.org/onlinedocs/gcc-4.3.0/gfortran/GAMMA.html
4349 21. http://gcc.gnu.org/onlinedocs/gcc-4.3.0/gfortran/LGAMMA.html
4350 22. http://gcc.gnu.org/onlinedocs/gfortran/Fortran-Dialect-Options.html
4351 23. http://gcc.gnu.org/onlinedocs/gfortran/BOZ-literal-constants.html
4352 24. http://gcc.gnu.org/onlinedocs/libstdc++/
4353 25. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.3.1
4354 26. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.3.2
4355 27. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.3.3
4356 28. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.3.4
4357 29. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.3.5
4358 30. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.3.6
4359 31. http://gcc.gnu.org/onlinedocs/
4360 32. mailto:gcc-help@gcc.gnu.org
4361 33. mailto:gcc@gcc.gnu.org
4362 34. http://gcc.gnu.org/lists.html
4363 35. http://www.fsf.org/
4364 36. http://gcc.gnu.org/about.html
4365 37. http://validator.w3.org/check/referer
4366 ======================================================================
4367 http://gcc.gnu.org/gcc-4.2/index.html
4369 GCC 4.2 Release Series
4373 The [1]GNU project and the GCC developers are pleased to announce the
4374 release of GCC 4.2.4.
4376 This release is a bug-fix release, containing fixes for regressions in
4377 GCC 4.2.3 relative to previous releases of GCC.
4382 May 19, 2008 ([2]changes)
4385 February 1, 2008 ([3]changes)
4388 October 7, 2007 ([4]changes)
4391 July 18, 2007 ([5]changes)
4394 May 13, 2007 ([6]changes)
4396 References and Acknowledgements
4398 GCC used to stand for the GNU C Compiler, but since the compiler
4399 supports several other languages aside from C, it now stands for the
4400 GNU Compiler Collection.
4402 A list of [7]successful builds is updated as new information becomes
4405 The GCC developers would like to thank the numerous people that have
4406 contributed new features, improvements, bug fixes, and other changes as
4407 well as test results to GCC. This [8]amazing group of volunteers is
4408 what makes GCC successful.
4410 For additional information about GCC please refer to the [9]GCC project
4411 web site or contact the [10]GCC development mailing list.
4413 To obtain GCC please use [11]our mirror sites or [12]our SVN server.
4416 For questions related to the use of GCC, please consult these web
4417 pages and the [13]GCC manuals. If that fails, the
4418 [14]gcc-help@gcc.gnu.org mailing list might help. Comments on these
4419 web pages and the development of GCC are welcome on our developer
4420 list at [15]gcc@gcc.gnu.org. All of [16]our lists have public
4423 Copyright (C) [17]Free Software Foundation, Inc. Verbatim copying and
4424 distribution of this entire article is permitted in any medium,
4425 provided this notice is preserved.
4427 These pages are [18]maintained by the GCC team. Last modified
4432 1. http://www.gnu.org/
4433 2. http://gcc.gnu.org/gcc-4.2/changes.html
4434 3. http://gcc.gnu.org/gcc-4.2/changes.html
4435 4. http://gcc.gnu.org/gcc-4.2/changes.html
4436 5. http://gcc.gnu.org/gcc-4.2/changes.html
4437 6. http://gcc.gnu.org/gcc-4.2/changes.html
4438 7. http://gcc.gnu.org/gcc-4.2/buildstat.html
4439 8. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
4440 9. http://gcc.gnu.org/index.html
4441 10. mailto:gcc@gcc.gnu.org
4442 11. http://gcc.gnu.org/mirrors.html
4443 12. http://gcc.gnu.org/svn.html
4444 13. http://gcc.gnu.org/onlinedocs/
4445 14. mailto:gcc-help@gcc.gnu.org
4446 15. mailto:gcc@gcc.gnu.org
4447 16. http://gcc.gnu.org/lists.html
4448 17. http://www.fsf.org/
4449 18. http://gcc.gnu.org/about.html
4450 19. http://validator.w3.org/check/referer
4451 ======================================================================
4452 http://gcc.gnu.org/gcc-4.2/changes.html
4454 GCC 4.2 Release Series
4455 Changes, New Features, and Fixes
4459 * GCC no longer accepts the -fshared-data option. This option has had
4460 no effect in any GCC 4 release; the targets to which the option
4461 used to apply had been removed before GCC 4.0.
4463 General Optimizer Improvements
4465 * New command-line options specify the possible relationships among
4466 parameters and between parameters and global data. For example,
4467 -fargument-noalias-anything specifies that arguments do not alias
4469 Each language will automatically use whatever option is required by
4470 the language standard. You should not need to use these options
4473 New Languages and Language specific improvements
4475 * [1]OpenMP is now supported for the C, C++ and Fortran compilers.
4476 * New command-line options -fstrict-overflow and -Wstrict-overflow
4477 have been added. -fstrict-overflow tells the compiler that it may
4478 assume that the program follows the strict signed overflow
4479 semantics permitted for the language: for C and C++ this means that
4480 the compiler may assume that signed overflow does not occur. For
4481 example, a loop like
4482 for (i = 1; i > 0; i *= 2)
4484 is presumably intended to continue looping until i overflows. With
4485 -fstrict-overflow, the compiler may assume that signed overflow
4486 will not occur, and transform this into an infinite loop.
4487 -fstrict-overflow is turned on by default at -O2, and may be
4488 disabled via -fno-strict-overflow. The -Wstrict-overflow option may
4489 be used to warn about cases where the compiler assumes that signed
4490 overflow will not occur. It takes five different levels:
4491 -Wstrict-overflow=1 to 5. See the [2]documentation for details.
4492 -Wstrict-overflow=1 is enabled by -Wall.
4493 * The new command-line option -fno-toplevel-reorder directs GCC to
4494 emit top-level functions, variables, and asm statements in the same
4495 order that they appear in the input file. This is intended to
4496 support existing code which relies on a particular ordering (for
4497 example, code which uses top-level asm statements to switch
4498 sections). For new code, it is generally better to use function and
4499 variable attributes. The -fno-toplevel-reorder option may be used
4500 for most cases which currently use -fno-unit-at-a-time. The
4501 -fno-unit-at-a-time option will be removed in some future version
4502 of GCC. If you know of a case which requires -fno-unit-at-a-time
4503 which is not fixed by -fno-toplevel-reorder, please open a bug
4508 * The pragma redefine_extname will now macro expand its tokens for
4509 compatibility with SunPRO.
4510 * In the next release of GCC, 4.3, -std=c99 or -std=gnu99 will direct
4511 GCC to handle inline functions as specified in the C99 standard. In
4512 preparation for this, GCC 4.2 will warn about any use of non-static
4513 inline functions in gnu99 or c99 mode. This new warning may be
4514 disabled with the new gnu_inline function attribute or the new
4515 -fgnu89-inline command-line option. Also, GCC 4.2 and later will
4516 define one of the preprocessor macros __GNUC_GNU_INLINE__ or
4517 __GNUC_STDC_INLINE__ to indicate the semantics of inline functions
4518 in the current compilation.
4519 * A new command-line option -Waddress has been added to warn about
4520 suspicious uses of memory addresses as, for example, using the
4521 address of a function in a conditional expression, and comparisons
4522 against the memory address of a string literal. This warning is
4527 * C++ visibility handling has been overhauled.
4528 Restricted visiblity is propagated from classes to members, from
4529 functions to local statics, and from templates and template
4530 arguments to instantiations, unless the latter has explicitly
4531 declared visibility.
4532 The visibility attribute for a class must come between the
4533 class-key and the name, not after the closing brace.
4534 Attributes are now allowed for enums and elaborated-type-specifiers
4535 that only declare a type.
4536 Members of the anonymous namespace are now local to a particular
4537 translation unit, along with any other declarations which use them,
4538 though they are still treated as having external linkage for
4540 * The (undocumented) extension which permitted templates with default
4541 arguments to be bound to template template parameters with fewer
4542 parameters has been removed. For example:
4543 template <template <typename> class C>
4544 void f(C<double>) {}
4546 template <typename T, typename U = int>
4549 template void f(S<double>);
4551 is no longer accepted by G++. The reason this code is not accepted
4552 is that S is a template with two parameters; therefore, it cannot
4553 be bound to C which has only one parameter.
4554 * The <?, >?, <?=, and >?= operators, deprecated in previous GCC
4555 releases, have been removed.
4556 * The command-line option -fconst-strings, deprecated in previous GCC
4557 releases, has been removed.
4558 * The configure variable enable-__cxa_atexit is now enabled by
4559 default for more targets. Enabling this variable is necessary in
4560 order for static destructors to be executed in the correct order,
4561 but it depends upon the presence of a non-standard C library in the
4562 target library in order to work. The variable is now enabled for
4563 more targets which are known to have suitable C libraries.
4564 * -Wextra will produce warnings for if statements with a semicolon as
4565 the only body, to catch code like:
4570 To suppress the warning in valid cases, use { } instead.
4571 * The C++ frontend now also produces strict aliasing warnings when
4572 -fstrict-aliasing -Wstrict-aliasing is in effect.
4574 Runtime Library (libstdc++)
4576 * Added support for TR1 <random>, <complex>, and C compatibility
4577 headers. In addition, a lock-free version of shared_ptr was
4578 contributed as part of Phillip Jordan's Google Summer of Code
4579 project on lock-free containers. ([3]Implementation status of TR1)
4580 * In association with the Summer of Code work on lock-free
4581 containers, the interface for atomic builtins was adjusted,
4582 creating simpler alternatives for non-threaded code paths. Also,
4583 usage was consolidated and all elements were moved from namespace
4584 std to namespace__gnu_cxx. Affected interfaces are the functions
4585 __exchange_and_add, __atomic_add, and the objects __mutex,
4586 __recursive_mutex, and __scoped_lock.
4587 * Support for versioning weak symbol names via namespace association
4588 was added. However, as this changes the names of exported symbols,
4589 this is turned off by default in the current ABI. Intrepid users
4590 can enable this feature by using
4591 --enable-symvers=gnu-versioned-namespace during configuration.
4592 * Revised, simplified, and expanded policy-based associative
4593 containers, including data types for tree and trie forms
4594 (basic_tree, tree, trie), lists (list_update), and both
4595 collision-chaining and probing hash-based containers
4596 (basic_hash_table, cc_hash_table, gp_hash_table). More details per
4597 the [4]documentation.
4598 * The implementation of the debug mode was modified, whereby the
4599 debug namespaces were nested inside of namespace std and namespace
4600 __gnu_cxx in order to resolve some long standing corner cases
4601 involving name lookup. Debug functionality from the policy-based
4602 data structures was consolidated and enabled with the single macro,
4603 _GLIBCXX_DEBUG. See PR 26142 for more information.
4604 * Added extensions for type traits: __conditional_type,
4605 __numeric_traits, __add_unsigned, __removed_unsigned, __enable_if.
4606 * Added a typelist implementation for compile-time meta-programming.
4607 Elements for typelist construction and operation can be found
4608 within namespace __gnu_cxx::typelist.
4609 * Added a new allocator, __gnu_cxx::throw_allocator, for testing
4611 * Enabled library-wide visibility control, allowing -fvisibility to
4613 * Consolidated all nested namespaces and the conversion of
4614 __gnu_internal implementation-private details to anonymous
4615 namespaces whenever possible.
4616 * Implemented LWG resolutions DR 431 and DR 538.
4620 * Support for allocatable components has been added (TR 15581 and
4622 * Support for the Fortran 2003 streaming IO extension has been added.
4623 * The GNU Fortran compiler now uses 4-byte record markers by default
4624 for unformatted files to be compatible with g77 and most other
4625 compilers. The implementation allows for records greater than 2 GB
4626 and is compatible with several other compilers. Older versions of
4627 gfortran used 8-byte record markers by default (on most systems).
4628 In order to change the length of the record markers, e.g. to read
4629 unformatted files created by older gfortran versions, the
4630 [5]-frecord-marker=8 option can be used.
4634 * A new command-line option -static-libgcj has been added for targets
4635 that use a linker compatible with GNU Binutils. As its name
4636 implies, this causes libgcj to be linked statically. In some cases
4637 this causes the resulting executable to start faster and use less
4638 memory than if the shared version of libgcj were used. However
4639 caution should be used as it can also cause essential parts of the
4640 library to be omitted. Some of these issues are discussed in:
4641 [6]http://gcc.gnu.org/wiki/Statically_linking_libgcj
4642 * fastjar is no longer bundled with GCC. To build libgcj, you will
4643 need either InfoZIP (both zip and unzip) or an external jar
4644 program. In the former case, the GCC build will install a jar shell
4645 script that is based on InfoZIP and provides the same functionality
4648 New Targets and Target Specific Improvements
4652 * -mtune=generic can now be used to generate code running well on
4653 common x86 chips. This includes AMD Athlon, AMD Opteron, Intel
4654 Pentium-M, Intel Pentium 4 and Intel Core 2.
4655 * -mtune=native and -march=native will produce code optimized for the
4656 host architecture as detected using the cpuid instruction.
4657 * Added a new command-line option -fstackrealign and and
4658 __attribute__ ((force_align_arg_pointer)) to realign the stack at
4659 runtime. This allows functions compiled with a vector-aligned stack
4660 to be invoked from legacy objects that keep only word-alignment.
4664 * The default CPU setting has been changed from V7 to V9 in 32-bit
4665 mode on Solaris 7 and above. This is already the case in 64-bit
4666 mode. It can be overridden by specifying --with-cpu at configure
4668 * Back-end support of built-in functions for atomic memory access has
4670 * Support for the Sun UltraSPARC T1 (Niagara) processor has been
4675 * Various bug fixes have made some functions (notably, functions
4676 returning structures) incompatible with previous releases.
4677 Recompiling all libraries is recommended. Note that code quality
4678 has considerably improved since 4.1, making a recompile even more
4683 * Added support for the Broadcom SB-1A core.
4687 * Added support for IA-64 data and control speculation. By default
4688 speculation is enabled only during second scheduler pass. A number
4689 of machine flags was introduced to control the usage of speculation
4690 for both scheduler passes.
4694 * Added Java language support (libffi and libjava) for 32-bit HP-UX
4699 Documentation improvements
4703 * A make pdf target has been added to the top-level makefile,
4704 enabling automated production of PDF documentation files.
4705 (Front-ends external to GCC should modify their Make-lang.in file
4706 to add a lang.pdf: target.)
4708 Other significant improvements
4710 Build system improvements
4712 * All the components of the compiler are now bootstrapped by default.
4713 This improves the resilience to bugs in the system compiler or
4714 binary compatibility problems, as well as providing better testing
4715 of GCC 4.2 itself. In addition, if you build the compiler from a
4716 combined tree, the assembler, linker, etc. will also be
4717 bootstrapped (i.e. built with themselves).
4718 You can disable this behavior, and go back to the pre-GCC 4.2 set
4719 up, by configuring GCC with --disable-bootstrap.
4720 * The rules that configure follows to find target tools resemble more
4721 closely the locations that the built compiler will search. In
4722 addition, you can use the new configure option --with-target-tools
4723 to specify where to find the target tools used during the build,
4724 without affecting what the built compiler will use.
4725 This can be especially useful when building packages of GCC. For
4726 example, you may want to build GCC with GNU as or ld, even if the
4727 resulting compiler to work with the native assembler and linker. To
4728 do so, you can use --with-target-tools to point to the native
4731 Incompatible changes to the build system
4733 * Front-ends external to GCC should modify their Make-lang.in file to
4734 replace double-colon rules (e.g. dvi::) with normal rules (like
4735 lang.dvi:). Front-end makefile hooks do not use double-colon rules
4737 * Up to GCC 4.1, a popular way to specify the target tools used
4738 during the build was to create directories named gas, binutils,
4739 etc. in the build tree, and create links to the tools from there.
4740 This does not work any more when the compiler is bootstrapped. The
4741 new configure option --with-target-tools provides a better way to
4742 achieve the same effect, and works for all native and cross
4746 For questions related to the use of GCC, please consult these web
4747 pages and the [7]GCC manuals. If that fails, the
4748 [8]gcc-help@gcc.gnu.org mailing list might help. Comments on these
4749 web pages and the development of GCC are welcome on our developer
4750 list at [9]gcc@gcc.gnu.org. All of [10]our lists have public
4753 Copyright (C) [11]Free Software Foundation, Inc. Verbatim copying and
4754 distribution of this entire article is permitted in any medium,
4755 provided this notice is preserved.
4757 These pages are [12]maintained by the GCC team. Last modified
4762 1. http://gcc.gnu.org/projects/gomp/
4763 2. http://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html
4764 3. http://gcc.gnu.org/onlinedocs/libstdc++/manual/bk01pt01ch01.html#manual.intro.status.standard.tr1
4765 4. http://gcc.gnu.org/onlinedocs/libstdc++/ext/pb_ds/index.html
4766 5. http://gcc.gnu.org/onlinedocs/gfortran/Runtime-Options.html
4767 6. http://gcc.gnu.org/wiki/Statically_linking_libgcj
4768 7. http://gcc.gnu.org/onlinedocs/
4769 8. mailto:gcc-help@gcc.gnu.org
4770 9. mailto:gcc@gcc.gnu.org
4771 10. http://gcc.gnu.org/lists.html
4772 11. http://www.fsf.org/
4773 12. http://gcc.gnu.org/about.html
4774 13. http://validator.w3.org/check/referer
4775 ======================================================================
4776 http://gcc.gnu.org/gcc-4.1/index.html
4778 GCC 4.1 Release Series
4782 The [1]GNU project and the GCC developers are pleased to announce the
4783 release of GCC 4.1.2.
4785 This release is a bug-fix release, containing fixes for regressions in
4786 GCC 4.1.1 relative to previous releases of GCC.
4791 February 13, 2007 ([2]changes)
4794 May 24, 2006 ([3]changes)
4797 February 28, 2006 ([4]changes)
4799 References and Acknowledgements
4801 GCC used to stand for the GNU C Compiler, but since the compiler
4802 supports several other languages aside from C, it now stands for the
4803 GNU Compiler Collection.
4805 A list of [5]successful builds is updated as new information becomes
4808 The GCC developers would like to thank the numerous people that have
4809 contributed new features, improvements, bug fixes, and other changes as
4810 well as test results to GCC. This [6]amazing group of volunteers is
4811 what makes GCC successful.
4813 For additional information about GCC please refer to the [7]GCC project
4814 web site or contact the [8]GCC development mailing list.
4816 To obtain GCC please use [9]our mirror sites or [10]our SVN server.
4819 For questions related to the use of GCC, please consult these web
4820 pages and the [11]GCC manuals. If that fails, the
4821 [12]gcc-help@gcc.gnu.org mailing list might help. Comments on these
4822 web pages and the development of GCC are welcome on our developer
4823 list at [13]gcc@gcc.gnu.org. All of [14]our lists have public
4826 Copyright (C) [15]Free Software Foundation, Inc. Verbatim copying and
4827 distribution of this entire article is permitted in any medium,
4828 provided this notice is preserved.
4830 These pages are [16]maintained by the GCC team. Last modified
4835 1. http://www.gnu.org/
4836 2. http://gcc.gnu.org/gcc-4.1/changes.html#4.1.2
4837 3. http://gcc.gnu.org/gcc-4.1/changes.html
4838 4. http://gcc.gnu.org/gcc-4.1/changes.html
4839 5. http://gcc.gnu.org/gcc-4.1/buildstat.html
4840 6. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
4841 7. http://gcc.gnu.org/index.html
4842 8. mailto:gcc@gcc.gnu.org
4843 9. http://gcc.gnu.org/mirrors.html
4844 10. http://gcc.gnu.org/svn.html
4845 11. http://gcc.gnu.org/onlinedocs/
4846 12. mailto:gcc-help@gcc.gnu.org
4847 13. mailto:gcc@gcc.gnu.org
4848 14. http://gcc.gnu.org/lists.html
4849 15. http://www.fsf.org/
4850 16. http://gcc.gnu.org/about.html
4851 17. http://validator.w3.org/check/referer
4852 ======================================================================
4853 http://gcc.gnu.org/gcc-4.1/changes.html
4855 GCC 4.1 Release Series
4856 Changes, New Features, and Fixes
4858 The latest release in the 4.1 release series is [1]GCC 4.1.2.
4862 General Optimizer Improvements
4864 * GCC now has infrastructure for inter-procedural optimizations and
4865 the following inter-procedural optimizations are implemented:
4866 + Profile guided inlining. When doing profile feedback guided
4867 optimization, GCC can now use the profile to make better
4868 informed decisions on whether inlining of a function is
4869 profitable or not. This means that GCC will no longer inline
4870 functions at call sites that are not executed very often, and
4871 that functions at hot call sites are more likely to be
4873 A new parameter min-inline-recursive-probability is also now
4874 available to throttle recursive inlining of functions with
4875 small average recursive depths.
4876 + Discovery of pure and const functions, a form of side-effects
4877 analysis. While older GCC releases could also discover such
4878 special functions, the new IPA-based pass runs earlier so that
4879 the results are available to more optimizers. The pass is also
4880 simply more powerful than the old one.
4881 + Analysis of references to static variables and type escape
4882 analysis, also forms of side-effects analysis. The results of
4883 these passes allow the compiler to be less conservative about
4884 call-clobbered variables and references. This results in more
4885 redundant loads being eliminated and in making static
4886 variables candidates for register promotion.
4887 + Improvement of RTL-based alias analysis. The results of type
4888 escape analysis are fed to the RTL type-based alias analyzer,
4889 allowing it to disambiguate more memory references.
4890 + Interprocedural constant propagation and function versioning.
4891 This pass looks for functions that are always called with the
4892 same constant value for one or more of the function arguments,
4893 and propagates those constants into those functions.
4894 + GCC will now eliminate static variables whose usage was
4896 + -fwhole-program --combine can now be used to make all
4897 functions in program static allowing whole program
4898 optimization. As an exception, the main function and all
4899 functions marked with the new externally_visible attribute are
4900 kept global so that programs can link with runtime libraries.
4901 * GCC can now do a form of partial dead code elimination (PDCE) that
4902 allows code motion of expressions to the paths where the result of
4903 the expression is actually needed. This is not always a win, so the
4904 pass has been limited to only consider profitable cases. Here is an
4906 int foo (int *, int *);
4922 The a = b + c can be sunk to right before the printf. Normal code
4923 sinking will not do this, it will sink the first one above into the
4924 else-branch of the conditional jump, which still gives you two
4926 * GCC now has a value range propagation pass. This allows the
4927 compiler to eliminate bounds checks and branches. The results of
4928 the pass can also be used to accurately compute branch
4930 * The pass to convert PHI nodes to straight-line code (a form of
4931 if-conversion for GIMPLE) has been improved significantly. The two
4932 most significant improvements are an improved algorithm to
4933 determine the order in which the PHI nodes are considered, and an
4934 improvement that allow the pass to consider if-conversions of basic
4935 blocks with more than two predecessors.
4936 * Alias analysis improvements. GCC can now differentiate between
4937 different fields of structures in Tree-SSA's virtual operands form.
4938 This lets stores/loads from non-overlapping structure fields not
4939 conflict. A new algorithm to compute points-to sets was contributed
4940 that can allows GCC to see now that p->a and p->b, where p is a
4941 pointer to a structure, can never point to the same field.
4942 * Various enhancements to auto-vectorization:
4943 + Incrementally preserve SSA form when vectorizing.
4944 + Incrementally preserve loop-closed form when vectorizing.
4945 + Improvements to peeling for alignment: generate better code
4946 when the misalignment of an access is known at compile time,
4947 or when different accesses are known to have the same
4948 misalignment, even if the misalignment amount itself is
4950 + Consider dependence distance in the vectorizer.
4951 + Externalize generic parts of data reference analysis to make
4952 this analysis available to other passes.
4953 + Vectorization of conditional code.
4954 + Reduction support.
4955 * GCC can now partition functions in sections of hot and cold code.
4956 This can significantly improve performance due to better
4957 instruction cache locality. This feature works best together with
4958 profile feedback driven optimization.
4959 * A new pass to avoid saving of unneeded arguments to the stack in
4960 vararg functions if the compiler can prove that they will not be
4962 * Transition of basic block profiling to tree level implementation
4963 has been completed. The new implementation should be considerably
4964 more reliable (hopefully avoiding profile mismatch errors when
4965 using -fprofile-use or -fbranch-probabilities) and can be used to
4966 drive higher level optimizations, such as inlining.
4967 The -ftree-based-profiling command-line option was removed and
4968 -fprofile-use now implies disabling old RTL level loop optimizer
4969 (-fno-loop-optimize). Speculative prefetching optimization
4970 (originally enabled by -fspeculative-prefetching) was removed.
4972 New Languages and Language specific improvements
4976 * The old Bison-based C and Objective-C parser has been replaced by a
4977 new, faster hand-written recursive-descent parser.
4981 * The build infrastructure for the Ada runtime library and tools has
4982 been changed to be better integrated with the rest of the build
4983 infrastructure of GCC. This should make doing cross builds of Ada a
4988 * ARM-style name-injection of friend declarations is no longer the
4989 default. For example:
4995 will not be accepted; instead a declaration of f will need to be
4996 present outside of the scope of S. The new -ffriend-injection
4997 option will enable the old behavior.
4998 * The (undocumented) extension which permitted templates with default
4999 arguments to be bound to template template parameters with fewer
5000 parameters has been deprecated, and will be removed in the next
5001 major release of G++. For example:
5002 template <template <typename> class C>
5003 void f(C<double>) {}
5005 template <typename T, typename U = int>
5008 template void f(S<double>);
5010 makes use of the deprecated extension. The reason this code is not
5011 valid ISO C++ is that S is a template with two parameters;
5012 therefore, it cannot be bound to C which has only one parameter.
5014 Runtime Library (libstdc++)
5016 * Optimization work:
5017 + A new implementation of std::search_n is provided, better
5018 performing in case of random access iterators.
5019 + Added further efficient specializations of istream functions,
5020 i.e., character array and string extractors.
5021 + Other smaller improvements throughout.
5022 * Policy-based associative containers, designed for high-performance,
5023 flexibility and semantic safety are delivered in ext/pb_assoc.
5024 * A versatile string class, __gnu_cxx::__versa_string, providing
5025 facilities conforming to the standard requirements for
5026 basic_string, is delivered in <ext/vstring.h>. In particular:
5027 + Two base classes are provided: the default one avoids
5028 reference counting and is optimized for short strings; the
5029 alternate one, still uses it while improving in a few low
5030 level areas (e.g., alignment). See vstring_fwd.h for some
5032 + Various algorithms have been rewritten (e.g., replace), the
5033 code streamlined and simple optimizations added.
5034 + Option 3 of DR 431 is implemented for both available bases,
5035 thus improving the support for stateful allocators.
5036 * As usual, many bugs have been fixed (e.g., libstdc++/13583,
5037 libstdc++/23953) and LWG resolutions put into effect for the first
5038 time (e.g., DR 280, DR 464, N1780 recommendations for DR 233, TR1
5039 Issue 6.19). The implementation status of TR1 is now tracked in the
5044 * A new language front end for Objective-C++ has been added. This
5045 language allows users to mix the object oriented features of
5046 Objective-C with those of C++.
5050 * Core library (libgcj) updates based on GNU Classpath 0.15 - 0.19
5051 features (plus some 0.20 bug-fixes)
5053 o The java.net.HttpURLConnection implementation no longer
5054 buffers the entire response body in memory. This means
5055 that response bodies larger than available memory can now
5058 o NIO FileChannel.map implementation, fast bulk put
5059 implementation for DirectByteBuffer (speeds up this
5061 o FileChannel.lock() and FileChannel.force() implemented.
5063 o gnu.xml fix for nodes created outside a namespace
5065 o Add support for output indenting and
5066 cdata-section-elements output instruction in
5068 o xml.xpath corrections for cases where elements/attributes
5069 might have been created in non-namespace-aware mode.
5070 Corrections to handling of XSL variables and minor
5071 conformance updates.
5073 o GNU JAWT implementation, the AWT Native Interface, which
5074 allows direct access to native screen resources from
5075 within a Canvas's paint method. GNU Classpath Examples
5076 comes with a Demo, see libjava/classpath/examples/README.
5077 o awt.datatransfer updated to 1.5 with support for
5078 FlavorEvents. The gtk+ awt peers now allow copy/paste of
5079 text, images, URIs/files and serialized objects with
5080 other applications and tracking clipboard change events
5081 with gtk+ 2.6 (for gtk+ 2.4 only text and serialized
5082 objects are supported). A GNU Classpath Examples
5083 datatransfer Demo was added to show the new
5085 o Split gtk+ awt peers event handling in two threads and
5086 improve gdk lock handling (solves several awt lock ups).
5087 o Speed up awt Image loading.
5088 o Better gtk+ scrollbar peer implementation when using gtk+
5090 o Handle image loading errors correctly for gdkpixbuf and
5092 o Better handle GDK lock. Properly prefix gtkpeer native
5094 o GdkGraphics2D has been updated to use Cairo 0.5.x or
5096 o BufferedImage and GtkImage rewrites. All image drawing
5097 operations should now work correctly (flipping requires
5099 o Future Graphics2D, image and text work is documented at:
5100 [2]http://developer.classpath.org/mediation/ClasspathGrap
5102 o When gtk+ 2.6 or higher is installed the default log
5103 handler will produce stack traces whenever a WARNING,
5104 CRITICAL or ERROR message is produced.
5106 o The RepaintManager has been reworked for more efficient
5107 painting, especially for large GUIs.
5108 o The layout manager OverlayLayout has been implemented,
5109 the BoxLayout has been rewritten to make use of the
5110 SizeRequirements utility class and caching for more
5112 o Improved accessibility support.
5113 o Significant progress has been made in the implementation
5114 of the javax.swing.plaf.metal package, with most UI
5115 delegates in a working state now. Please test this with
5116 your own applications and provide feedback that will help
5117 us to improve this package.
5118 o The GUI demo (gnu.classpath.examples.swing.Demo) has been
5119 extended to highlight various features in our Free Swing
5120 implementation. And it includes a look and feel switcher
5121 for Metal (default), Ocean and GNU themes.
5122 o The javax.swing.plaf.multi package is now implemented.
5123 o Editing and several key actions for JTree and JTable were
5125 o Lots of icons and look and feel improvements for Free
5126 Swing basic and metal themes were added. Try running the
5127 GNU Classpath Swing Demo in examples
5128 (gnu.classpath.examples.swing.Demo) with:
5129 -Dswing.defaultlaf=javax.swing.plaf.basic.BasicLookAndFee
5131 -Dswing.defaultlaf=javax.swing.plaf.metal.MetalLookAndFee
5133 o Start of styled text capabilites for java.swing.text.
5134 o DefaultMutableTreeNode pre-order, post-order, depth-first
5135 and breadth-first traversal enumerations implemented.
5136 o JInternalFrame colors and titlebar draw properly.
5137 o JTree is working up to par (icons, selection and keyboard
5139 o JMenus were made more compatible in visual and
5140 programmatic behavior.
5141 o JTable changeSelection and multiple selections
5143 o JButton and JToggleButton change states work properly
5145 o JFileChooser fixes.
5146 o revalidate() and repaint() fixes which make Free Swing
5147 much more responsive.
5148 o MetalIconFactory implemented.
5149 o Free Swing Top-Level Compatibility. JFrame, JDialog,
5150 JApplet, JInternalFrame, and JWindow are now 1.5
5151 compatible in the sense that you can call add() and
5152 setLayout() directly on them, which will have the same
5153 effect as calling getContentPane().add() and
5154 getContentPane().setLayout().
5155 o The JTree interface has been completed. JTrees now
5156 recognizes mouse clicks and selections work.
5157 o BoxLayout works properly now.
5158 o Fixed GrayFilter to actually work.
5159 o Metal SplitPane implemented.
5160 o Lots of Free Swing text and editor stuff work now.
5161 + Free RMI and Corba
5162 o Andrew Watson, Vice President and Technical Director of
5163 the Object Management Group, has officially assigned us
5164 20 bit Vendor Minor Code Id: 0x47430 ("GC") that will
5165 mark remote classpath-specific system exceptions.
5166 Obtaining the VMCID means that GNU Classpath now is a
5167 recogniseable type of node in a highly interoperable
5169 o GNU Classpath now includes the first working draft to
5170 support the RMI over IIOP protocol. The current
5171 implementation is capable of remote invocations,
5172 transferring various Serializables and Externalizables
5173 via RMI-IIOP protocol. It can flatten graphs and, at
5174 least for the simple cases, is interoperable with 1.5
5176 o org.omg.PortableInterceptor and related functionality in
5177 other packages is now implemented:
5178 # The sever and client interceptors work as required
5180 # The IOR interceptor works as needed for 1.5.
5181 o The org.omg.DynamicAny package is completed and passes
5183 o The Portable Object Adapter should now support the output
5184 of the recent IDL to java compilers. These compilers now
5185 generate servants and not CORBA objects as before, making
5186 the output depend on the existing POA implementation.
5187 Completing POA means that such code can already be tried
5188 to run on Classpath. Our POA is tested for the following
5190 # POA converts servant to the CORBA object.
5191 # Servant provides to the CORBA object.
5192 # POA activates new CORBA object with the given Object
5193 Id (byte array) that is later accessible for the
5195 # During the first call, the ServantActivator provides
5196 servant for this and all subsequent calls on the
5198 # During each call, the ServantLocator provides
5199 servant for this call only.
5200 # ServantLocator or ServantActivator forwards call to
5202 # POA has a single servant, responsible for all
5204 # POA has a default servant, but some objects are
5205 explicitly connected to they specific servants.
5206 The POA is verified using tests from the former
5208 o The CORBA implementation is now a working prototype that
5209 should support features up to 1.3 inclusive. We invite
5210 groups writing CORBA dependent applications to try
5211 Classpath implementation, reporting any possible bugs.
5212 The CORBA prototype is interoperable with Sun's
5213 implementation v 1.4, transferring object references,
5214 primitive types, narrow and wide strings, arrays,
5215 structures, trees, abstract interfaces and value types
5216 (feature of CORBA 2.3) between these two platforms.
5217 Remote exceptions are transferred and handled correctly.
5218 The stringified object references (IORs) from various
5219 sources are parsed as required. The transient (for
5220 current session) and permanent (till jre restart)
5221 redirections work. Both Little and Big Endian encoded
5222 messages are accepted. The implementation is verified
5223 using tests from the former cost.omg.org. The current
5224 release includes working examples (see the examples
5225 directory), demonstrating the client-server
5226 communication, using either CORBA Request or IDL-based
5227 stub (usually generated by a IDL to java compiler). These
5228 examples also show how to use the Classpath CORBA naming
5229 service. The IDL to java compiler is not yet written, but
5230 as our library must be compatible, it naturally accepts
5231 the output of other idlj implementations.
5233 o Updated TimeZone data against Olson tzdata2005l.
5234 o Make zip and jar packages UTF-8 clean.
5235 o "native" code builds and compiles (warning free) on
5237 o java.util.logging.FileHandler now rotates files.
5238 o Start of a generic JDWP framework in gnu/classpath/jdwp.
5239 This is unfinished, but feedback (at classpath@gnu.org)
5240 from runtime hackers is greatly appreciated. Although
5241 most of the work is currently being done around gcj/gij
5242 we want this framework to be as VM neutral as possible.
5243 Early design is described in:
5244 [3]http://gcc.gnu.org/ml/java/2005-05/msg00260.html
5245 o QT4 AWT peers, enable by giving configure
5246 --enable-qt-peer. Included, but not ready for production
5247 yet. They are explicitly disabled and not supported. But
5248 if you want to help with the development of these new
5249 features we are interested in feedback. You will have to
5250 explicitly enable them to try them out (and they will
5251 most likely contain bugs).
5252 o Documentation fixes all over the place. See
5253 [4]http://developer.classpath.org/doc/
5255 New Targets and Target Specific Improvements
5259 * The x86-64 medium model (that allows building applications whose
5260 data segment exceeds 4GB) was redesigned to match latest ABI draft.
5261 New implementation split large datastructures into separate segment
5262 improving performance of accesses to small datastructures and also
5263 allows linking of small model libraries into medium model programs
5264 as long as the libraries are not accessing the large datastructures
5265 directly. Medium model is also supported in position independent
5267 The ABI change results in partial incompatibility among medium
5268 model objects. Linking medium model libraries (or objects) compiled
5269 with new compiler into medium model program compiled with older
5270 will likely result in exceeding ranges of relocations.
5271 Binutils 2.16.91 or newer are required for compiling medium model
5274 RS6000 (POWER/PowerPC)
5276 * The AltiVec vector primitives in <altivec.h> are now implemented in
5277 a way that puts a smaller burden on the preprocessor, instead
5278 processing the "overloading" in the front ends. This should benefit
5279 compilation speed on AltiVec vector code.
5280 * AltiVec initializers now are generated more efficiently.
5281 * The popcountb instruction available on POWER5 now is generated.
5282 * The floating point round to integer instructions available on
5283 POWER5+ now is generated.
5284 * Floating point divides can be synthesized using the floating point
5285 reciprocal estimate instructions.
5286 * Double precision floating point constants are initialized as single
5287 precision values if they can be represented exactly.
5289 S/390, zSeries and System z9
5291 * Support for the IBM System z9 109 processor has been added. When
5292 using the -march=z9-109 option, the compiler will generate code
5293 making use of instructions provided by the extended immediate
5295 * Support for 128-bit IEEE floating point has been added. When using
5296 the -mlong-double-128 option, the compiler will map the long double
5297 data type to 128-bit IEEE floating point. Using this option
5298 constitutes an ABI change, and requires glibc support.
5299 * Various changes to improve performance of generated code have been
5300 implemented, including:
5301 + In functions that do not require a literal pool, register %r13
5302 (which is traditionally reserved as literal pool pointer), can
5303 now be freely used for other purposes by the compiler.
5304 + More precise tracking of register use allows the compiler to
5305 generate more efficient function prolog and epilog code in
5307 + The SEARCH STRING, COMPARE LOGICAL STRING, and MOVE STRING
5308 instructions are now used to implement C string functions.
5309 + The MOVE CHARACTER instruction with single byte overlap is now
5310 used to implement the memset function with non-zero fill byte.
5311 + The LOAD ZERO instructions are now used where appropriate.
5312 + The INSERT CHARACTERS UNDER MASK, STORE CHARACTERS UNDER MASK,
5313 and INSERT IMMEDIATE instructions are now used more frequently
5314 to optimize bitfield operations.
5315 + The BRANCH ON COUNT instruction is now used more frequently.
5316 In particular, the fact that a loop contains a subroutine call
5317 no longer prevents the compiler from using this instruction.
5318 + The compiler is now aware that all shift and rotate
5319 instructions implicitly truncate the shift count to six bits.
5320 * Back-end support for the following generic features has been
5322 + The full set of [5]built-in functions for atomic memory
5324 + The -fstack-protector feature.
5325 + The optimization pass avoiding unnecessary stores of incoming
5326 argument registers in functions with variable argument list.
5330 * The default code model in 64-bit mode has been changed from
5331 Medium/Anywhere to Medium/Middle on Solaris.
5332 * TLS support is disabled by default on Solaris prior to release 10.
5333 It can be enabled on TLS-capable Solaris 9 versions (4/04 release
5334 and later) by specifying --enable-tls at configure time.
5338 * Support has been added for this new architecture.
5342 Documentation improvements
5344 Other significant improvements
5346 * GCC can now emit code for protecting applications from
5347 stack-smashing attacks. The protection is realized by buffer
5348 overflow detection and reordering of stack variables to avoid
5350 * Some built-in functions have been fortified to protect them against
5351 various buffer overflow (and format string) vulnerabilities.
5352 Compared to the mudflap bounds checking feature, the safe builtins
5353 have far smaller overhead. This means that programs built using
5354 safe builtins should not experience any measurable slowdown.
5358 This is the [6]list of problem reports (PRs) from GCC's bug tracking
5359 system that are known to be fixed in the 4.1.2 release. This list might
5360 not be complete (that is, it is possible that some PRs that have been
5361 fixed are not listed here).
5363 When generating code for a shared library, GCC now recognizes that
5364 global functions may be replaced when the program runs. Therefore, it
5365 is now more conservative in deducing information from the bodies of
5366 functions. For example, in this example:
5371 cout << "Exception";
5375 G++ would previously have optimized away the catch clause, since it
5376 would have concluded that f cannot throw exceptions. Because users may
5377 replace f with another function in the main body of the program, this
5378 optimization is unsafe, and is no longer performed. If you wish G++ to
5379 continue to optimize as before, you must add a throw() clause to the
5380 declaration of f to make clear that it does not throw exceptions.
5383 For questions related to the use of GCC, please consult these web
5384 pages and the [7]GCC manuals. If that fails, the
5385 [8]gcc-help@gcc.gnu.org mailing list might help. Comments on these
5386 web pages and the development of GCC are welcome on our developer
5387 list at [9]gcc@gcc.gnu.org. All of [10]our lists have public
5390 Copyright (C) [11]Free Software Foundation, Inc. Verbatim copying and
5391 distribution of this entire article is permitted in any medium,
5392 provided this notice is preserved.
5394 These pages are [12]maintained by the GCC team. Last modified
5399 1. http://gcc.gnu.org/gcc-4.1/changes.html#4.1.2
5400 2. http://developer.classpath.org/mediation/ClasspathGraphicsImagesText
5401 3. http://gcc.gnu.org/ml/java/2005-05/msg00260.html
5402 4. http://developer.classpath.org/doc/
5403 5. http://gcc.gnu.org/onlinedocs/gcc-4.1.0/gcc/Atomic-Builtins.html
5404 6. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.1.2
5405 7. http://gcc.gnu.org/onlinedocs/
5406 8. mailto:gcc-help@gcc.gnu.org
5407 9. mailto:gcc@gcc.gnu.org
5408 10. http://gcc.gnu.org/lists.html
5409 11. http://www.fsf.org/
5410 12. http://gcc.gnu.org/about.html
5411 13. http://validator.w3.org/check/referer
5412 ======================================================================
5413 http://gcc.gnu.org/gcc-4.0/index.html
5415 GCC 4.0 Release Series
5419 The [1]GNU project and the GCC developers are pleased to announce the
5420 release of GCC 4.0.4.
5422 This release is a bug-fix release, containing fixes for regressions in
5423 GCC 4.0.3 relative to previous releases of GCC.
5428 January 31, 2007 ([2]changes)
5431 March 10, 2006 ([3]changes)
5434 September 28, 2005 ([4]changes)
5437 July 7, 2005 ([5]changes)
5440 April 20, 2005 ([6]changes)
5442 References and Acknowledgements
5444 GCC used to stand for the GNU C Compiler, but since the compiler
5445 supports several other languages aside from C, it now stands for the
5446 GNU Compiler Collection.
5448 A list of [7]successful builds is updated as new information becomes
5451 The GCC developers would like to thank the numerous people that have
5452 contributed new features, improvements, bug fixes, and other changes as
5453 well as test results to GCC. This [8]amazing group of volunteers is
5454 what makes GCC successful.
5456 For additional information about GCC please refer to the [9]GCC project
5457 web site or contact the [10]GCC development mailing list.
5459 To obtain GCC please use [11]our mirror sites, or [12]our SVN server.
5462 For questions related to the use of GCC, please consult these web
5463 pages and the [13]GCC manuals. If that fails, the
5464 [14]gcc-help@gcc.gnu.org mailing list might help. Comments on these
5465 web pages and the development of GCC are welcome on our developer
5466 list at [15]gcc@gcc.gnu.org. All of [16]our lists have public
5469 Copyright (C) [17]Free Software Foundation, Inc. Verbatim copying and
5470 distribution of this entire article is permitted in any medium,
5471 provided this notice is preserved.
5473 These pages are [18]maintained by the GCC team. Last modified
5478 1. http://www.gnu.org/
5479 2. http://gcc.gnu.org/gcc-4.0/changes.html#4.0.4
5480 3. http://gcc.gnu.org/gcc-4.0/changes.html#4.0.3
5481 4. http://gcc.gnu.org/gcc-4.0/changes.html#4.0.2
5482 5. http://gcc.gnu.org/gcc-4.0/changes.html#4.0.1
5483 6. http://gcc.gnu.org/gcc-4.0/changes.html
5484 7. http://gcc.gnu.org/gcc-4.0/buildstat.html
5485 8. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
5486 9. http://gcc.gnu.org/index.html
5487 10. mailto:gcc@gcc.gnu.org
5488 11. http://gcc.gnu.org/mirrors.html
5489 12. http://gcc.gnu.org/svn.html
5490 13. http://gcc.gnu.org/onlinedocs/
5491 14. mailto:gcc-help@gcc.gnu.org
5492 15. mailto:gcc@gcc.gnu.org
5493 16. http://gcc.gnu.org/lists.html
5494 17. http://www.fsf.org/
5495 18. http://gcc.gnu.org/about.html
5496 19. http://validator.w3.org/check/referer
5497 ======================================================================
5498 http://gcc.gnu.org/gcc-4.0/changes.html
5500 GCC 4.0 Release Series
5501 Changes, New Features, and Fixes
5503 The latest release in the 4.0 release series is [1]GCC 4.0.4.
5507 * GCC now generates location lists by default when compiling with
5508 debug info and optimization.
5509 + GDB 6.0 and older crashes when it sees location lists. GDB 6.1
5510 or later is needed to debug binaries containing location
5512 + When you are trying to view a value of a variable in a part of
5513 a function where it has no location (for example when the
5514 variable is no longer used and thus its location was used for
5515 something else) GDB will say that it is not available.
5516 You can disable generating location lists by -fno-var-tracking.
5517 * GCC no longer accepts the -fwritable-strings option. Use named
5518 character arrays when you need a writable string.
5519 * The options -freduce-all-givs and -fmove-all-movables have been
5520 discontinued. They were used to circumvent a shortcoming in the
5521 heuristics of the old loop optimization code with respect to common
5522 Fortran constructs. The new (tree) loop optimizer works differently
5523 and doesn't need those work-arounds.
5524 * The graph-coloring register allocator, formerly enabled by the
5525 option -fnew-ra, has been discontinued.
5526 * -I- has been deprecated. -iquote is meant to replace the need for
5528 * The MIPS -membedded-pic and -mrnames options have been removed.
5529 * All MIPS targets now require the GNU assembler. In particular, IRIX
5530 configurations can no longer use the MIPSpro assemblers, although
5531 they do still support the MIPSpro linkers.
5532 * The SPARC option -mflat has been removed.
5533 * English-language diagnostic messages will now use Unicode quotation
5534 marks in UTF-8 locales. (Non-English messages already used the
5535 quotes appropriate for the language in previous releases.) If your
5536 terminal does not support UTF-8 but you are using a UTF-8 locale
5537 (such locales are the default on many GNU/Linux systems) then you
5538 should set LC_CTYPE=C in the environment to disable that locale.
5539 Programs that parse diagnostics and expect plain ASCII
5540 English-language messages should set LC_ALL=C. See [2]Markus Kuhn's
5541 explanation of Unicode quotation marks for more information.
5542 * The specs file is no longer installed on most platforms. Most users
5543 will be totally unaffected. However, if you are accustomed to
5544 editing the specs file yourself, you will now have to use the
5545 -dumpspecs option to generate the specs file, and then edit the
5548 General Optimizer Improvements
5550 * The [3]tree ssa branch has been merged. This merge has brought in a
5551 completely new optimization framework based on a higher level
5552 intermediate representation than the existing RTL representation.
5553 Numerous new code transformations based on the new framework are
5554 available in GCC 4.0, including:
5555 + Scalar replacement of aggregates
5556 + Constant propagation
5557 + Value range propagation
5558 + Partial redundancy elimination
5559 + Load and store motion
5560 + Strength reduction
5561 + Dead store elimination
5562 + Dead and unreachable code elimination
5563 + [4]Autovectorization
5565 + Tail recursion by accumulation
5566 Many of these passes outperform their counterparts from previous
5568 * [5]Swing Modulo Scheduling (SMS). An RTL level instruction
5569 scheduling optimization intended for loops that perform heavy
5572 New Languages and Language specific improvements
5576 * The sentinel attribute has been added to GCC. This function
5577 attribute allows GCC to warn when variadic functions such as execl
5578 are not NULL terminated. See the GCC manual for a complete
5579 description of its behavior.
5580 * Given __attribute__((alias("target"))) it is now an error if target
5581 is not a symbol, defined in the same translation unit. This also
5582 applies to aliases created by #pragma weak alias=target. This is
5583 because it's meaningless to define an alias to an undefined symbol.
5584 On Solaris, the native assembler would have caught this error, but
5589 * The -Wstrict-aliasing=2 option has been added. This warning catches
5590 all unsafe cases, but it may also give a warning for some cases
5592 * The cast-as-lvalue, conditional-expression-as-lvalue and
5593 compound-expression-as-lvalue extensions, which were deprecated in
5594 3.3.4 and 3.4, have been removed.
5595 * The -fwritable-strings option, which was deprecated in 3.4, has
5597 * #pragma pack() semantics have been brought closer to those used by
5598 other compilers. This also applies to C++.
5599 * Taking the address of a variable with register storage is invalid
5600 in C. GCC now issues an error instead of a warning.
5601 * Arrays of incomplete element type are invalid in C. GCC now issues
5602 an error for such arrays. Declarations such as extern struct s x[];
5603 (where struct s has not been defined) can be moved after the
5604 definition of struct s. Function parameters declared as arrays of
5605 incomplete type can instead be declared as pointers.
5609 * When compiling without optimizations (-O0), the C++ frontend is
5610 much faster than in any previous versions of GCC. Independent
5611 testers have measured speed-ups up to 25% in real-world production
5612 code, compared to the 3.4 family (which was already the fastest
5613 version to date). Upgrading from older versions might show even
5614 bigger improvements.
5615 * ELF visibility attributes can now be applied to a class type, so
5616 that it affects every member function of a class at once, without
5617 having to specify each individually:
5618 class __attribute__ ((visibility("hidden"))) Foo
5623 The syntax is deliberately similar to the __declspec() system used
5624 by Microsoft Windows based compilers, allowing cross-platform
5625 projects to easily reuse their existing macro system for denoting
5626 exports and imports. By explicitly marking internal classes never
5627 used outside a binary as hidden, one can completely avoid PLT
5628 indirection overheads during their usage by the compiler. You can
5629 find out more about the advantages of this at
5630 [6]http://www.akkadia.org/drepper/dsohowto.pdf
5631 * The -fvisibility-inlines-hidden option has been added which marks
5632 all inlineable functions as having hidden ELF visibility, thus
5633 removing their symbol and typeinfo from the exported symbol table
5634 of the output ELF binary. Using this option can reduce the exported
5635 symbol count of template-heavy code by up to 40% with no code
5636 change at all, thus notably improving link and load times for the
5637 binary as well as a reduction in size of up to 10%. Also, check the
5638 new [7]-fvisibility option.
5639 * The compiler now uses the library interface specified by the [8]C++
5640 ABI for thread-safe initialization of function-scope static
5641 variables. Most users should leave this alone, but embedded
5642 programmers may want to disable this by specifying
5643 -fno-threadsafe-statics for a small savings in code size.
5644 * Taking the address of an explicit register variable is no longer
5645 supported. Note that C++ allows taking the address of variables
5646 with register storage so this will continue to compile with a
5647 warning. For example, assuming that r0 is a machine register:
5648 register int foo asm ("r0");
5650 &foo; // error, no longer accepted
5651 &bar; // OK, with a warning
5652 * G++ has an undocumented extension to virtual function covariancy
5653 rules that allowed the overrider to return a type that was
5654 implicitly convertable to the overridden function's return type.
5655 For instance a function returning void * could be overridden by a
5656 function returning T *. This is now deprecated and will be removed
5657 in a future release.
5658 * The G++ minimum and maximum operators (<? and >?) and their
5659 compound forms (<?=) and >?=) have been deprecated and will be
5660 removed in a future version. Code using these operators should be
5661 modified to use std::min and std::max instead.
5662 * Declaration of nested classes of class templates as friends are
5664 template <typename T> struct A {
5668 template <typename T> friend class A<T>::B;
5670 This complements the feature member functions of class templates as
5671 friends introduced in GCC 3.4.0.
5672 * When declaring a friend class using an unqualified name, classes
5673 outside the innermost non-class scope are not searched:
5677 friend class A; // Refer to N::A which has not been declared yet
5678 // because name outside namespace N are not searched
5679 friend class ::A; // Refer to ::A
5682 Hiding the friend name until declaration is still not implemented.
5683 * Friends of classes defined outside their namespace are correctly
5689 friend class B; // Refer to N::B in GCC 4.0.0
5690 // but ::B in earlier versions of GCC
5693 Runtime Library (libstdc++)
5695 * Optimization work:
5696 + Added efficient specializations of istream functions for char
5698 + Further performance tuning of strings, in particular wrt
5699 single-char append and getline.
5700 + iter_swap - and therefore most of the mutating algorithms -
5701 now makes an unqualified call to swap when the value_type of
5702 the two iterators is the same.
5703 * A large subset of the features in Technical Report 1 (TR1 for
5704 short) is experimentally delivered (i.e., no guarantees about the
5705 implementation are provided. In particular it is not promised that
5706 the library will remain link-compatible when code using TR1 is
5708 + General utilities such as reference_wrapper and shared_ptr.
5709 + Function objects, i.e., result_of, mem_fn, bind, function.
5710 + Support for metaprogramming.
5711 + New containers such as tuple, array, unordered_set,
5712 unordered_map, unordered_multiset, unordered_multimap.
5713 * As usual, many bugs have been fixed and LWG resolutions implemented
5714 for the first time (e.g., DR 409).
5718 * In order to prevent naming conflicts with other implementations of
5719 these tools, some GCJ binaries have been renamed:
5720 + rmic is now grmic,
5721 + rmiregistry is now grmiregistry, and
5722 + jar is now fastjar.
5723 In particular, these names were problematic for the jpackage.org
5724 packaging conventions which install symlinks in /usr/bin that point
5725 to the preferred versions of these tools.
5726 * The -findirect-dispatch argument to the compiler now works and
5727 generates code following a new "binary compatibility" ABI. Code
5728 compiled this way follows the binary compatibility rules of the
5729 Java Language Specification.
5730 * libgcj now has support for using GCJ as a JIT, using the
5731 gnu.gcj.jit family of system properties.
5732 * libgcj can now find a shared library corresponding to the bytecode
5733 representation of a class. See the documentation for the new
5734 gcj-dbtool program, and the new gnu.gcj.precompiled.db.path system
5736 * There have been many improvements to the class library. Here are
5738 + Much more of AWT and Swing exist.
5739 + Many new packages and classes were added, including
5740 java.util.regex, java.net.URI, javax.crypto,
5741 javax.crypto.interfaces, javax.crypto.spec, javax.net,
5742 javax.net.ssl, javax.security.auth,
5743 javax.security.auth.callback, javax.security.auth.login,
5744 javax.security.auth.x500, javax.security.sasl, org.ietf.jgss,
5745 javax.imageio, javax.imageio.event, javax.imageio.spi,
5746 javax.print, javax.print.attribute,
5747 javax.print.attribute.standard, javax.print.event, and
5749 + Updated SAX and DOM, and imported GNU JAXP
5753 * A new [9]Fortran front end has replaced the aging GNU Fortran 77
5754 front end. The new front end supports Fortran 90 and Fortran 95. It
5755 may not yet be as stable as the old Fortran front end.
5759 * Ada (with tasking and Zero Cost Exceptions) is now available on
5760 many more targets, including but not limited to: alpha-linux,
5761 hppa-hpux, hppa-linux, powerpc-darwin, powerpc-linux, s390-linux,
5762 s390x-linux, sparc-linux.
5763 * Some of the new Ada 2005 features are now implemented like
5764 Wide_Wide_Character and Ada.Containers.
5765 * Many bugs have been fixed, tools and documentation improved.
5766 * To compile Ada from the sources, install an older working Ada
5767 compiler and then use --enable-languages=ada at configuration time,
5768 since the Ada frontend is not currently activated by default. See
5769 the [10]Installing GCC for details.
5771 New Targets and Target Specific Improvements
5775 * The frame layout has changed. In the new layout, the prologue of a
5776 function first saves registers and then allocate space for locals,
5777 resulting in an 1% improvement on code size.
5779 IA-32/x86-64 (AMD64)
5781 * The acos, asin, drem, exp10, exp2, expm1, fmod, ilogb, log10,
5782 log1p, log2, logb and tan mathematical builtins (and their float
5783 and long double variants) are now implemented as inline x87
5784 intrinsics when using -ffast-math.
5785 * The ceil, floor, nearbyint, rint and trunc mathematical builtins
5786 (and their float and long double variants) are now implemented as
5787 inline x87 intrinsics when using -ffast-math.
5788 * The x87's fsincos instruction is now used automatically with
5789 -ffast-math when calculating both the sin and cos of the same
5791 * Instruction selection for multiplication and division by constants
5796 * Floating point division, integer division and sqrt are now inlined,
5797 resulting in significant performance improvements on some codes.
5801 * Division by zero checks now use conditional traps if the target
5802 processor supports them. This decreases code size by one word per
5803 division operation. The old behavior (branch and break) can be
5804 obtained either at configure time by passing --with-divide=breaks
5805 to configure or at runtime by passing -mdivide-breaks to GCC.
5806 * Support for MIPS64 paired-single instructions has been added. It is
5807 enabled by -mpaired-single and can be accessed using both the
5808 target-independent vector extensions and new MIPS-specific built-in
5810 * Support for the MIPS-3D ASE has been added. It is enabled by
5811 -mips3d and provides new MIPS-3D-specific built-in functions.
5812 * The -mexplicit-relocs option now supports static n64 code (as is
5813 used, for example, in 64-bit linux kernels). -mexplicit-relocs
5814 should now be feature-complete and is enabled by default when GCC
5815 is configured to use a compatible assembler.
5816 * Support for the NEC VR4130 series has been added. This support
5817 includes the use of VR-specific instructions and a new VR4130
5818 scheduler. Full VR4130 support can be selected with -march=vr4130
5819 while code for any ISA can be tuned for the VR4130 using
5820 -mtune=vr4130. There is also a new -mvr4130-align option that
5821 produces better schedules at the cost of increased code size.
5822 * Support for the Broadcom SB-1 has been extended. There is now an
5823 SB-1 scheduler as well as support for the SB-1-specific
5824 paired-single instructions. Full SB-1 support can be selected with
5825 -march=sb1 while code for any ISA can be optimized for the SB-1
5827 * The compiler can now work around errata in R4000, R4400, VR4120 and
5828 VR4130 processors. These workarounds are enabled by -mfix-r4000,
5829 -mfix-r4400, -mfix-vr4120 and -mfix-vr4130 respectively. The VR4120
5830 and VR4130 workarounds need binutils 2.16 or above.
5831 * IRIX shared libraries are now installed into the standard library
5832 directories: o32 libraries go into lib/, n32 libraries go into
5833 lib32/ and n64 libraries go into lib64/.
5834 * The compiler supports a new -msym32 option. It can be used to
5835 optimize n64 code in which all symbols are known to have 32-bit
5840 * New command-line options help to generate code intended to run in
5841 an environment where stack space is restricted, e.g. Linux kernel
5843 + -mwarn-framesize and -mwarn-dynamicstack trigger compile-time
5844 warnings for single functions that require large or dynamic
5846 + -mstack-size and -mstack-guard generate code that checks for
5847 stack overflow at run time.
5848 + -mpacked-stack generates code that reduces the stack frame
5849 size of many functions by reusing unneeded parts of the stack
5851 * The -msoft-float option now ensures that generated code never
5852 accesses floating point registers.
5853 * The s390x-ibm-tpf target now fully supports C++, including
5854 exceptions and threads.
5855 * Various changes to improve performance of the generated code have
5856 been implemented, including:
5857 + GCC now uses sibling calls where possible.
5858 + Condition code handling has been optimized, allowing GCC to
5859 omit redundant comparisons in certain cases.
5860 + The cost function guiding many optimizations has been refined
5861 to more accurately represent the z900 and z990 processors.
5862 + The ADD LOGICAL WITH CARRY and SUBTRACT LOGICAL WITH BORROW
5863 instructions are now used to avoid conditional branches in
5865 + The back end now uses the LEGITIMIZE_RELOAD_ADDRESS feature to
5866 optimize address arithmetic required to access large stack
5868 + GCC now makes more efficient use of memory-to-memory type
5869 instructions (MVC, CLC, ...).
5870 + More precise tracking of special register use allows better
5871 instruction scheduling, in particular of the function prologue
5872 and epilogue sequences.
5873 + The Java front end now generates inline code to implement
5874 integer division, instead of calling library routines.
5878 * The options -mv8, -msparclite, -mcypress, -msupersparc, -mf930 and
5879 -mf934 have been removed. They have been replaced with -mcpu=xxx.
5880 * The internal model used to estimate the relative cost of each
5881 instruction has been updated. It is expected to give better results
5882 on recent UltraSPARC processors.
5883 * Code generation for function prologues and epilogues has been
5884 improved, resulting in better scheduling and allowing multiple exit
5885 points in functions.
5886 * Support for Sun's Visual Instruction Set (VIS) has been enhanced.
5887 It is enabled by -mvis and provides new built-in functions for VIS
5888 instructions on UltraSPARC processors.
5889 * The option -mapp-regs has been turned on by default on Solaris too.
5893 * Novell NetWare (on ix86, no other hardware platform was ever really
5894 supported by this OS) has been re-enabled and the ABI supported by
5895 GCC has been brought into sync with that of MetroWerks CodeWarrior
5896 (the ABI previously supported was that of some Unix systems, which
5897 NetWare never tried to support).
5901 Support for a number of older systems has been declared obsolete in GCC
5902 4.0. Unless there is activity to revive them, the next release of GCC
5903 will have their sources permanently removed.
5905 All GCC ports for the following processor architectures have been
5909 * National Semiconductor NS32K
5910 * Texas Instruments TMS320C[34]x
5912 Also, those for some individual systems have been obsoleted:
5914 + SPARClite-based systems (sparclite-*-coff, sparclite-*-elf,
5916 + OpenBSD 32-bit (sparc-*-openbsd*)
5918 Documentation improvements
5920 Other significant improvements
5922 * Location lists are now generated by default when compiling with
5923 debug info and optimization. Location lists provide more accurate
5924 debug info about locations of variables and they allow debugging
5925 code compiled with -fomit-frame-pointer.
5926 * The -fvisibility option has been added which allows the default ELF
5927 visibility of all symbols to be set per compilation and the new
5928 #pragma GCC visibility preprocessor command allows the setting of
5929 default ELF visibility for a region of code. Using
5930 -fvisibility=hidden especially in combination with the new
5931 -fvisibility-inlines-hidden can yield substantial improvements in
5932 output binary quality including avoiding PLT indirection overheads,
5933 reduction of the exported symbol count by up to 60% (with resultant
5934 improvements to link and load times), better scope for the
5935 optimizer to improve code and up to a 20% reduction in binary size.
5936 Using these options correctly yields a binary with a similar symbol
5937 count to a Windows DLL.
5938 Perhaps more importantly, this new feature finally allows (with
5939 careful planning) complete avoidance of symbol clashes when
5940 manually loading shared objects with RTLD_GLOBAL, thus finally
5941 solving problems many projects such as python were forced to use
5942 RTLD_LOCAL for (with its resulting issues for C++ correctness). You
5943 can find more information about using these options at
5944 [11]http://gcc.gnu.org/wiki/Visibility.
5945 __________________________________________________________________
5949 This is the [12]list of problem reports (PRs) from GCC's bug tracking
5950 system that are known to be fixed in the 4.0.1 release. This list might
5951 not be complete (that is, it is possible that some PRs that have been
5952 fixed are not listed here).
5956 This is the [13]list of problem reports (PRs) from GCC's bug tracking
5957 system that are known to be fixed in the 4.0.2 release. This list might
5958 not be complete (that is, it is possible that some PRs that have been
5959 fixed are not listed here).
5961 Unfortunately, due to a release engineering failure, this release has a
5962 regression on Solaris that will affect some C++ programs. We suggest
5963 that Solaris users apply a [14]patch that corrects the problem. Users
5964 who do not wish to apply the patch should explicitly link C++ programs
5965 with the -pthreads option, even if they do not use threads. This
5966 problem has been corrected in the current 4.0 branch sources and will
5967 not be present in GCC 4.0.3.
5971 Starting with this release, the function getcontext is recognized by
5972 the compiler as having the same semantics as the setjmp function. In
5973 particular, the compiler will ensure that all registers are dead before
5974 calling such a function and will emit a warning about the variables
5975 that may be clobbered after the second return from the function.
5979 This is the [15]list of problem reports (PRs) from GCC's bug tracking
5980 system that are known to be fixed in the 4.0.4 release. This list might
5981 not be complete (that is, it is possible that some PRs that have been
5982 fixed are not listed here).
5984 The 4.0.4 release is provided for those that require a high degree of
5985 binary compatibility with previous 4.0.x releases. For most users, the
5986 GCC team recommends that version 4.1.1 or later be used instead."
5989 For questions related to the use of GCC, please consult these web
5990 pages and the [16]GCC manuals. If that fails, the
5991 [17]gcc-help@gcc.gnu.org mailing list might help. Comments on these
5992 web pages and the development of GCC are welcome on our developer
5993 list at [18]gcc@gcc.gnu.org. All of [19]our lists have public
5996 Copyright (C) [20]Free Software Foundation, Inc. Verbatim copying and
5997 distribution of this entire article is permitted in any medium,
5998 provided this notice is preserved.
6000 These pages are [21]maintained by the GCC team. Last modified
6005 1. http://gcc.gnu.org/gcc-4.0/changes.html#4.0.4
6006 2. http://www.cl.cam.ac.uk/~mgk25/ucs/quotes.html
6007 3. http://gcc.gnu.org/projects/tree-ssa/
6008 4. http://gcc.gnu.org/projects/tree-ssa/vectorization.html
6009 5. http://gcc.gnu.org/news/sms.html
6010 6. http://www.akkadia.org/drepper/dsohowto.pdf
6011 7. http://gcc.gnu.org/gcc-4.0/changes.html#visibility
6012 8. http://mentorembedded.github.com/cxx-abi/
6013 9. http://gcc.gnu.org/fortran/
6014 10. http://gcc.gnu.org/install/
6015 11. http://gcc.gnu.org/wiki/Visibility
6016 12. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.0.1
6017 13. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.0.2
6018 14. http://gcc.gnu.org/ml/gcc-cvs/2005-09/msg00984.html
6019 15. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.0.4
6020 16. http://gcc.gnu.org/onlinedocs/
6021 17. mailto:gcc-help@gcc.gnu.org
6022 18. mailto:gcc@gcc.gnu.org
6023 19. http://gcc.gnu.org/lists.html
6024 20. http://www.fsf.org/
6025 21. http://gcc.gnu.org/about.html
6026 22. http://validator.w3.org/check/referer
6027 ======================================================================
6028 http://gcc.gnu.org/gcc-3.4/index.html
6030 GCC 3.4 Release Series
6034 The [1]GNU project and the GCC developers are pleased to announce the
6035 release of GCC 3.4.6.
6037 This release is a bug-fix release, containing fixes for regressions in
6038 GCC 3.4.4 relative to previous releases of GCC. This is the last of the
6041 The GCC 3.4 release series includes numerous [2]new features,
6042 improvements, bug fixes, and other changes, thanks to an [3]amazing
6043 group of volunteers.
6048 March 6, 2006 ([4]changes)
6051 November 30, 2005 ([5]changes)
6054 May 18, 2005 ([6]changes)
6057 November 4, 2004 ([7]changes)
6060 September 6, 2004 ([8]changes)
6063 July 1, 2004 ([9]changes)
6066 April 18, 2004 ([10]changes)
6068 References and Acknowledgements
6070 GCC used to stand for the GNU C Compiler, but since the compiler
6071 supports several other languages aside from C, it now stands for the
6072 GNU Compiler Collection.
6074 A list of [11]successful builds is updated as new information becomes
6077 The GCC developers would like to thank the numerous people that have
6078 contributed new features, improvements, bug fixes, and other changes as
6079 well as test results to GCC. This [12]amazing group of volunteers is
6080 what makes GCC successful.
6082 For additional information about GCC please refer to the [13]GCC
6083 project web site or contact the [14]GCC development mailing list.
6085 To obtain GCC please use [15]our mirror sites, or [16]our SVN server.
6088 For questions related to the use of GCC, please consult these web
6089 pages and the [17]GCC manuals. If that fails, the
6090 [18]gcc-help@gcc.gnu.org mailing list might help. Comments on these
6091 web pages and the development of GCC are welcome on our developer
6092 list at [19]gcc@gcc.gnu.org. All of [20]our lists have public
6095 Copyright (C) [21]Free Software Foundation, Inc. Verbatim copying and
6096 distribution of this entire article is permitted in any medium,
6097 provided this notice is preserved.
6099 These pages are [22]maintained by the GCC team. Last modified
6104 1. http://www.gnu.org/
6105 2. http://gcc.gnu.org/gcc-3.4/changes.html
6106 3. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
6107 4. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.6
6108 5. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.5
6109 6. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.4
6110 7. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.3
6111 8. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.2
6112 9. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.1
6113 10. http://gcc.gnu.org/gcc-3.4/changes.html
6114 11. http://gcc.gnu.org/gcc-3.4/buildstat.html
6115 12. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
6116 13. http://gcc.gnu.org/index.html
6117 14. mailto:gcc@gcc.gnu.org
6118 15. http://gcc.gnu.org/mirrors.html
6119 16. http://gcc.gnu.org/svn.html
6120 17. http://gcc.gnu.org/onlinedocs/
6121 18. mailto:gcc-help@gcc.gnu.org
6122 19. mailto:gcc@gcc.gnu.org
6123 20. http://gcc.gnu.org/lists.html
6124 21. http://www.fsf.org/
6125 22. http://gcc.gnu.org/about.html
6126 23. http://validator.w3.org/check/referer
6127 ======================================================================
6128 http://gcc.gnu.org/gcc-3.4/changes.html
6130 GCC 3.4 Release Series
6131 Changes, New Features, and Fixes
6133 The final release in the 3.4 release series is [1]GCC 3.4.6. The series
6136 GCC 3.4 has [2]many improvements in the C++ frontend. Before reporting
6137 a bug, please make sure it's really GCC, and not your code, that is
6142 * GNU Make is now required to build GCC.
6143 * With -nostdinc the preprocessor used to ignore both standard
6144 include paths and include paths contained in environment variables.
6145 It was neither documented nor intended that environment variable
6146 paths be ignored, so this has been corrected.
6147 * GCC no longer accepts the options -fvolatile, -fvolatile-global and
6148 -fvolatile-static. It is unlikely that they worked correctly in any
6150 * GCC no longer ships <varargs.h>. Use <stdarg.h> instead.
6151 * Support for all the systems [3]obsoleted in GCC 3.3 has been
6152 removed from GCC 3.4. See below for a [4]list of systems which are
6153 obsoleted in this release.
6154 * GCC now requires an ISO C90 (ANSI C89) C compiler to build. K&R C
6155 compilers will not work.
6156 * The implementation of the [5]MIPS ABIs has changed. As a result,
6157 the code generated for certain MIPS targets will not be binary
6158 compatible with earlier releases.
6159 * In previous releases, the MIPS port had a fake "hilo" register with
6160 the user-visible name accum. This register has been removed.
6161 * The implementation of the [6]SPARC ABIs has changed. As a result,
6162 the code generated will not be binary compatible with earlier
6163 releases in certain cases.
6164 * The configure option --enable-threads=pthreads has been removed;
6165 use --enable-threads=posix instead, which should have the same
6167 * Code size estimates used by inlining heuristics for C, Objective-C,
6168 C++ and Java have been redesigned significantly. As a result the
6169 parameters of -finline-insns, --param max-inline-insns-single and
6170 --param max-inline-insns-auto need to be reconsidered.
6171 * --param max-inline-slope and --param min-inline-insns have been
6172 removed; they are not needed for the new bottom-up inlining
6174 * The new unit-at-a-time compilation scheme has several compatibility
6176 + The order in which functions, variables, and top-level asm
6177 statements are emitted may have changed. Code relying on some
6178 particular ordering needs to be updated. The majority of such
6179 top-level asm statements can be replaced by section
6181 + Unreferenced static variables and functions are removed. This
6182 may result in undefined references when an asm statement
6183 refers to the variable/function directly. In that case either
6184 the variable/function shall be listed in asm statement operand
6185 or in the case of top-level asm statements the attribute used
6186 shall be used to force function/variable to be always output
6187 and considered as a possibly used by unknown code.
6188 For variables the attribute is accepted only by GCC 3.4 and
6189 newer, while for earlier versions it is sufficient to use
6190 unused to silence warnings about the variables not being
6191 referenced. To keep code portable across different GCC
6192 versions, you can use appropriate preprocessor conditionals.
6193 + Static functions now can use non-standard passing conventions
6194 that may break asm statements calling functions directly.
6195 Again the attribute used shall be used to prevent this
6197 As a temporary workaround, -fno-unit-at-a-time can be used, but
6198 this scheme may not be supported by future releases of GCC.
6199 * GCC 3.4 automatically places zero-initialized variables in the .bss
6200 section on some operating systems. Versions of GNU Emacs up to (and
6201 including) 21.3 will not work correctly when using this
6202 optimization; you can use -fno-zero-initialized-in-bss to disable
6204 * If GCC 3.4 is configured with --enable-threads=posix (the default
6205 on most targets that support pthreads) then _REENTRANT will be
6206 defined unconditionally by some libstdc++ headers. C++ code which
6207 relies on that macro to detect whether multi-threaded code is being
6208 compiled might change in meaning, possibly resulting in linker
6209 errors for single-threaded programs. Affected users of [7]Boost
6210 should compile single-threaded code with -DBOOST_DISABLE_THREADS.
6211 See Bugzilla for [8]more information.
6213 General Optimizer Improvements
6215 * Usability of the profile feedback and coverage testing has been
6217 + Performance of profiled programs has been improved by faster
6218 profile merging code.
6219 + Better use of the profile feedback for optimization (loop
6220 unrolling and loop peeling).
6221 + File locking support allowing fork() calls and parallel runs
6222 of profiled programs.
6223 + Coverage file format has been redesigned.
6224 + gcov coverage tool has been improved.
6225 + make profiledbootstrap available to build a faster compiler.
6226 Experiments made on i386 hardware showed an 11% speedup on -O0
6227 and a 7.5% speedup on -O2 compilation of a [9]large C++
6229 + New value profiling pass enabled via -fprofile-values
6230 + New value profile transformations pass enabled via -fvpt aims
6231 to optimize some code sequences by exploiting knowledge about
6232 value ranges or other properties of the operands. At the
6233 moment a conversion of expensive divisions into cheaper
6234 operations has been implemented.
6235 + New -fprofile-generate and -fprofile-use command-line options
6236 to simplify the use of profile feedback.
6237 * A new unit-at-a-time compilation scheme for C, Objective-C, C++ and
6238 Java which is enabled via -funit-at-a-time (and implied by -O2). In
6239 this scheme a whole file is parsed first and optimized later. The
6240 following basic inter-procedural optimizations are implemented:
6241 + Removal of unreachable functions and variables
6242 + Discovery of local functions (functions with static linkage
6243 whose address is never taken)
6244 + On i386, these local functions use register parameter passing
6246 + Reordering of functions in topological order of the call graph
6247 to enable better propagation of optimizing hints (such as the
6248 stack alignments needed by functions) in the back end.
6249 + Call graph based out-of-order inlining heuristics which allows
6250 to limit overall compilation unit growth (--param
6251 inline-unit-growth).
6252 Overall, the unit-at-a-time scheme produces a 1.3% improvement for
6253 the SPECint2000 benchmark on the i386 architecture (AMD Athlon
6255 * More realistic code size estimates used by inlining for C,
6256 Objective-C, C++ and Java. The growth of large functions can now be
6257 limited via --param large-function-insns and --param
6258 large-function-growth.
6259 * A new cfg-level loop optimizer pass replaces the old loop unrolling
6260 pass and adds two other loop transformations -- loop peeling and
6261 loop unswitching -- and also uses the profile feedback to limit
6262 code growth. (The three optimizations are enabled by
6263 -funroll-loops, -fpeel-loops and -funswitch-loops flags,
6265 The old loop unroller still can be enabled by -fold-unroll-loops
6266 and may produce better code in some cases, especially when the
6267 webizer optimization pass is not run.
6268 * A new web construction pass enabled via -fweb (and implied by -O3)
6269 improves the quality of register allocation, CSE, first scheduling
6270 pass and some other optimization passes by avoiding re-use of
6271 pseudo registers with non-overlapping live ranges. The pass almost
6272 always improves code quality but does make debugging difficult and
6273 thus is not enabled by default by -O2
6274 The pass is especially effective as cleanup after code duplication
6275 passes, such as the loop unroller or the tracer.
6276 * Experimental implementations of superblock or trace scheduling in
6277 the second scheduling pass can be enabled via
6278 -fsched2-use-superblocks and -fsched2-use-traces, respectively.
6280 New Languages and Language specific improvements
6284 * The Ada front end has been updated to include numerous bug fixes
6285 and enhancements. These include:
6286 + Improved project file support
6287 + Additional set of warnings about potential wrong code
6288 + Improved error messages
6289 + Improved code generation
6290 + Improved cross reference information
6292 + Better run-time check elimination
6293 + Better error recovery
6294 + More efficient implementation of unbounded strings
6295 + Added features in GNAT.Sockets, GNAT.OS_Lib, GNAT.Debug_Pools,
6297 + New GNAT.xxxx packages (e.g. GNAT.Strings,
6298 GNAT.Exception_Action)
6300 + New -gnatS switch replacing gnatpsta
6301 + Implementation of new Ada features (in particular limited
6302 with, limited aggregates)
6306 * Precompiled headers are now supported. Precompiled headers can
6307 dramatically speed up compilation of some projects. There are some
6308 known defects in the current precompiled header implementation that
6309 will result in compiler crashes in relatively rare situations.
6310 Therefore, precompiled headers should be considered a "technology
6311 preview" in this release. Read the manual for details about how to
6312 use precompiled headers.
6313 * File handling in the preprocessor has been rewritten. GCC no longer
6314 gets confused by symlinks and hardlinks, and now has a correct
6315 implementation of #import and #pragma once. These two directives
6316 have therefore been un-deprecated.
6317 * The undocumented extension that allowed C programs to have a label
6318 at the end of a compound statement, which has been deprecated since
6319 GCC 3.0, has been removed.
6320 * The cast-as-lvalue extension has been removed for C++ and
6321 deprecated for C and Objective-C. In particular, code like this:
6329 is no longer accepted for C++ and will not be accepted for C and
6330 Objective-C in a future version.
6331 * The conditional-expression-as-lvalue extension has been deprecated
6332 for C and Objective-C. In particular, code like this:
6336 will not be accepted for C and Objective-C in a future version.
6337 * The compound-expression-as-lvalue extension has been deprecated for
6338 C and Objective-C. In particular, code like this:
6342 will not be accepted for C and Objective-C in a future version. A
6343 possible non-intrusive workaround is the following:
6346 * Several [10]built-in functions such as __builtin_popcount for
6347 counting bits, finding the highest and lowest bit in a word, and
6348 parity have been added.
6349 * The -fwritable-strings option has been deprecated and will be
6351 * Many C math library functions are now recognized as built-ins and
6353 * The C, C++, and Objective-C compilers can now handle source files
6354 written in any character encoding supported by the host C library.
6355 The default input character set is taken from the current locale,
6356 and may be overridden with the -finput-charset command line option.
6357 In the future we will add support for inline encoding markers.
6361 * G++ is now much closer to full conformance to the ISO/ANSI C++
6362 standard. This means, among other things, that a lot of invalid
6363 constructs which used to be accepted in previous versions will now
6364 be rejected. It is very likely that existing C++ code will need to
6365 be fixed. This document lists some of the most common issues.
6366 * A hand-written recursive-descent C++ parser has replaced the
6367 YACC-derived C++ parser from previous GCC releases. The new parser
6368 contains much improved infrastructure needed for better parsing of
6369 C++ source codes, handling of extensions, and clean separation
6370 (where possible) between proper semantics analysis and parsing. The
6371 new parser fixes many bugs that were found in the old parser.
6372 * You must now use the typename and template keywords to disambiguate
6373 dependent names, as required by the C++ standard.
6375 typedef int mytype_t;
6378 template <class T1> struct A {
6379 template <class T2> struct B {
6383 template <int N> void bar(void)
6385 // Use 'typename' to tell the parser that T1::mytype_t names
6386 // a type. This is needed because the name is dependent (in
6387 // this case, on template parameter T1).
6388 typename T1::mytype_t x;
6393 template <class T> void template_func(void)
6395 // Use 'template' to prefix member templates within
6396 // dependent types (a has type A<T>, which depends on
6397 // the template parameter T).
6399 a.template bar<0>();
6401 // Use 'template' to tell the parser that B is a nested
6402 // template class (dependent on template parameter T), and
6403 // 'typename' because the whole A<T>::B<int> is
6404 // the name of a type (again, dependent).
6405 typename A<T>::template B<int> b;
6409 void non_template_func(void)
6411 // Outside of any template class or function, no names can be
6412 // dependent, so the use of the keyword 'typename' and 'template'
6413 // is not needed (and actually forbidden).
6419 * In a template definition, unqualified names will no longer find
6420 members of a dependent base (as specified by [temp.dep]/3 in the
6421 C++ standard). For example,
6422 template <typename T> struct B {
6430 template <typename T> struct C : B<T> {
6435 n = 0; // ::n is modified
6436 g (); // ::g is called
6439 You must make the names dependent, e.g. by prefixing them with
6440 this->. Here is the corrected definition of C<T>::h,
6441 template <typename T> void C<T>::h ()
6448 As an alternative solution (unfortunately not backwards compatible
6449 with GCC 3.3), you may use using declarations instead of this->:
6450 template <typename T> struct C : B<T> {
6463 * In templates, all non-dependent names are now looked up and bound
6464 at definition time (while parsing the code), instead of later when
6465 the template is instantiated. For instance:
6468 template <int> struct A {
6469 static void bar(void){
6478 A<0>::bar(); // Calls foo(int), used to call foo(char).
6481 * In an explicit instantiation of a class template, you must use
6482 class or struct before the template-id:
6486 template A<0>; // error, not accepted anymore
6487 template class A<0>; // OK
6488 * The "named return value" and "implicit typename" extensions have
6490 * Default arguments in function types have been deprecated and will
6492 * ARM-style name-injection of friend declarations has been deprecated
6493 and will be removed. For example: struct S { friend void f(); };
6494 void g() { f(); } will not be accepted by future versions of G++;
6495 instead a declaration of "f" will need to be present outside of the
6497 * Covariant returns are implemented for all but varadic functions
6498 that require an adjustment.
6499 * When -pedantic is used, G++ now issues errors about spurious
6500 semicolons. For example,
6501 namespace N {}; // Invalid semicolon.
6502 void f() {}; // Invalid semicolon.
6503 * G++ no longer accepts attributes for a declarator after the
6504 initializer associated with that declarator. For example,
6505 X x(1) __attribute__((...));
6506 is no longer accepted. Instead, use:
6507 X x __attribute__((...)) (1);
6508 * Inside the scope of a template class, the name of the class itself
6509 can be treated as either a class or a template. So GCC used to
6510 accept the class name as argument of type template, and template
6511 template parameter. However this is not C++ standard compliant. Now
6512 the name is not treated as a valid template template argument
6513 unless you qualify the name by its scope. For example, the code
6514 below no longer compiles.
6515 template <template <class> class TT> class X {};
6516 template <class T> class Y {
6517 X<Y> x; // Invalid, Y is always a type template parameter.
6519 The valid code for the above example is
6520 X< ::Y> x; // Valid.
6521 (Notice the space between < and : to prevent GCC to interpret this
6522 as a digraph for [.)
6523 * Friend declarations that refer to template specializations are
6524 rejected if the template has not already been declared. For
6526 template <typename T>
6528 friend void f<> (C&);
6530 is rejected. You must first declare f as a template,
6531 template <typename T>
6533 * In case of friend declarations, every name used in the friend
6534 declaration must be accessible at the point of that declaration.
6535 Previous versions of G++ used to be less strict about this and
6536 allowed friend declarations for private class members, for example.
6537 See the ISO C++ Standard Committee's [11]defect report #209 for
6539 * Declaration of member functions of class templates as friends are
6540 supported. For example,
6541 template <typename T> struct A {
6545 template <typename T> friend void A<T>::f();
6547 * You must use template <> to introduce template specializations, as
6548 required by the standard. For example,
6549 template <typename T>
6553 is rejected. You must write,
6554 template <> struct S<int> {};
6555 * G++ used to accept code like this,
6558 void f(int i = g());
6561 This behavior is not mandated by the standard. Now G++ issues an
6562 error about this code. To avoid the error, you must move the
6563 declaration of g before the declaration of f. The default arguments
6564 for g must be visible at the point where it is called.
6565 * The C++ ABI Section 3.3.3 specifications for the array construction
6566 routines __cxa_vec_new2 and __cxa_vec_new3 were changed to return
6567 NULL when the allocator argument returns NULL. These changes are
6568 incorporated into the libstdc++ runtime library.
6569 * Using a name introduced by a typedef in a friend declaration or in
6570 an explicit instantiation is now rejected, as specified by the ISO
6575 friend class B; // error, no typedef name here
6576 friend B; // error, friend always needs class/struct/enum
6577 friend class A; // OK
6580 template <int> class Q {};
6582 template class R; // error, no typedef name here
6583 template class Q<0>; // OK
6584 * When allocating an array with a new expression, GCC used to allow
6585 parentheses around the type name. This is actually ill-formed and
6587 int* a = new (int)[10]; // error, not accepted anymore
6588 int* a = new int[10]; // OK
6589 * When binding an rvalue of class type to a reference, the copy
6590 constructor of the class must be accessible. For instance, consider
6598 A(const A&); // private copy ctor
6606 foo(A()); // error, copy ctor is not accessible
6607 foo(makeA()); // error, copy ctor is not accessible
6610 foo(a1); // OK, a1 is a lvalue
6612 This might be surprising at first sight, especially since most
6613 popular compilers do not correctly implement this rule ([12]further
6615 * When forming a pointer to member or a pointer to member function,
6616 access checks for class visibility (public, protected, private) are
6617 now performed using the qualifying scope of the name itself. This
6618 is better explained with an example:
6634 &A::pub_func; // OK, pub_func is accessible through A
6635 &A::prot_func; // error, cannot access prot_func through A
6636 &A::priv_func; // error, cannot access priv_func through A
6638 &B::pub_func; // OK, pub_func is accessible through B
6639 &B::prot_func; // OK, can access prot_func through B (within B)
6640 &B::priv_func; // error, cannot access priv_func through B
6644 Runtime Library (libstdc++)
6646 * Optimization work:
6647 + Streamlined streambuf, filebuf, separate synched with C
6648 Standard I/O streambuf.
6649 + All formatted I/O now uses cached locale information.
6650 + STL optimizations (memory/speed for list, red-black trees as
6651 used by sets and maps).
6652 + More use of GCC builtins.
6653 + String optimizations (avoid contention on
6654 increment/decrement-and-test of the reference count in the
6655 empty-string object, constructor from input_iterators
6657 * Static linkage size reductions.
6658 * Large File Support (files larger than 2 GB on 32-bit systems).
6659 * Wide character and variable encoding filebuf work (UTF-8, Unicode).
6660 * Generic character traits.
6661 * Also support wchar_t specializations on Mac OS 10.3.x, FreeBSD 5.x,
6662 Solaris 2.7 and above, AIX 5.x, Irix 6.5.
6663 * The allocator class is now standard-conformant, and two additional
6664 extension allocators have been added, mt_alloc and
6666 * PCH support: -include bits/stdc++.h (2x compile speedup).
6667 * Rewrote __cxa_demangle with support for C++ style allocators.
6668 * New debug modes for STL containers and iterators.
6669 * Testsuite rewrite: five times as many tests, plus increasingly
6670 sophisticated tests, including I/O, MT, multi-locale, wide and
6672 * Use current versions of GNU "autotools" for build/configuration.
6676 * The Objective-C front end has been updated to include the numerous
6677 bug fixes and enhancements previously available only in Apple's
6678 version of GCC. These include:
6679 + Structured exception (@try... @catch... @finally, @throw) and
6680 synchronization (@synchronized) support. These are accessible
6681 via the -fobjc-exceptions switch; as of this writing, they may
6682 only be used in conjunction with -fnext-runtime on Mac OS X
6683 10.3 and later. See [13]Options Controlling Objective-C
6684 Dialect for more information.
6685 + An overhaul of @encode logic. The C99 _Bool and C++ bool type
6686 may now be encoded as 'B'. In addition, the back-end/codegen
6687 dependencies have been removed.
6688 + An overhaul of message dispatch construction, ensuring that
6689 the various receiver types (and casts thereof) are handled
6690 properly, and that correct diagnostics are issued.
6691 + Support for "Zero-Link" (-fzero-link) and "Fix-and-Continue"
6692 (-freplace-objc-classes) debugging modes, currently available
6693 on Mac OS X 10.3 and later. See [14]Options Controlling
6694 Objective-C Dialect for more information.
6695 + Access to optimized runtime entry points (-fno-nil-receivers )
6696 on the assumption that message receivers are never nil. This
6697 is currently available on Mac OS X 10.3 and later. See
6698 [15]Options Controlling Objective-C Dialect for more
6703 * Compiling a .jar file will now cause non-.class entries to be
6704 automatically compiled as resources.
6705 * libgcj has been ported to Darwin.
6706 * Jeff Sturm has adapted Jan Hubicka's call graph optimization code
6708 * libgcj has a new gcjlib URL type; this lets URLClassLoader load
6709 code from shared libraries.
6710 * libgcj has been much more completely merged with [16]GNU Classpath.
6711 * Class loading is now much more correct; in particular the caller's
6712 class loader is now used when that is required.
6713 * [17]Eclipse 2.x will run out of the box using gij.
6714 * Parts of java.nio have been implemented. Direct and indirect
6715 buffers work, as do fundamental file and socket operations.
6716 * java.awt has been improved, though it is still not ready for
6718 * The HTTP protocol handler now uses HTTP/1.1 and can handle the POST
6720 * The MinGW port has matured. Enhancements include socket timeout
6721 support, thread interruption, improved Runtime.exec() handling and
6722 support for accented characters in filenames.
6726 * Fortran improvements are listed in the [18]Fortran documentation.
6728 New Targets and Target Specific Improvements
6732 * Several [19]built-in functions have been added such as
6733 __builtin_alpha_zap to allow utilizing the more obscure
6734 instructions of the CPU.
6735 * Parameter passing of complex arguments has changed to match the
6736 [20]ABI. This change is incompatible with previous GCC versions,
6737 but does fix compatibility with the Tru64 compiler and several
6738 corner cases where GCC was incompatible with itself.
6742 * Nicolas Pitre has contributed his hand-coded floating-point support
6743 code for ARM. It is both significantly smaller and faster than the
6744 existing C-based implementation, even when building applications
6745 for Thumb. The arm-elf configuration has been converted to use the
6747 * Support for the Intel's iWMMXt architecture, a second generation
6748 XScale processor, has been added. Enabled at run time with the
6749 -mcpu=iwmmxt command line switch.
6750 * A new ARM target has been added: arm-wince-pe. This is similar to
6751 the arm-pe target, but it defaults to using the APCS32 ABI.
6752 * The existing ARM pipeline description has been converted to the use
6753 the [21]DFA processor pipeline model. There is not much change in
6754 code performance, but the description is now [22]easier to
6756 * Support for the Cirrus EP9312 Maverick floating point co-processor
6757 added. Enabled at run time with the -mcpu=ep9312 command line
6758 switch. Note however that the multilibs to support this chip are
6759 currently disabled in gcc/config/arm/t-arm-elf, so if you want to
6760 enable their production you will have to uncomment the entries in
6765 * Support for long long has been added.
6766 * Support for saveall attribute has been added.
6767 * Pavel Pisa contributed hand-written 32-bit-by-32-bit division code
6768 for H8/300H and H8S, which is much faster than the previous
6770 * A lot of small performance improvements.
6772 IA-32/AMD64 (x86-64)
6774 * Tuning for K8 (AMD Opteron/Athlon64) core is available via
6775 -march=k8 and -mcpu=k8.
6776 * Scalar SSE code generation carefully avoids reformatting penalties,
6777 hidden dependencies and minimizes the number of uops generated on
6778 both Intel and AMD CPUs.
6779 * Vector MMX and SSE operands are now passed in registers to improve
6780 performance and match the argument passing convention used by the
6781 Intel C++ Compiler. As a result it is not possible to call
6782 functions accepting vector arguments compiled by older GCC version.
6783 * Conditional jump elimination is now more aggressive on modern CPUs.
6784 * The Athlon ports has been converted to use the DFA processor
6785 pipeline description.
6786 * Optimization of indirect tail calls is now possible in a similar
6787 fashion as direct sibcall optimization.
6788 * Further small performance improvements.
6789 * -m128bit-long-double is now less buggy.
6790 * __float128 support in 64-bit compilation.
6791 * Support for data structures exceeding 2GB in 64-bit mode.
6792 * -mcpu has been renamed to -mtune.
6796 * Tuning code for the Itanium 2 processor has been added. The
6797 generation of code tuned for Itanium 2 (option -mtune=itanium2) is
6798 enabled by default now. To generate code tuned for Itanium 1 the
6799 option -mtune=itanium1 should be used.
6800 * [23]DFA processor pipeline descriptions for the IA-64 processors
6801 have been added. This resulted in about 3% improvement on the
6802 SPECInt2000 benchmark for Itanium 2.
6803 * Instruction bundling for the IA-64 processors has been rewritten
6804 using the DFA pipeline hazard recognizer. It resulted in about 60%
6805 compiler speedup on the SPECInt2000 C programs.
6809 * Support for the M32R/2 processor has been added by Renesas.
6810 * Support for an M32R GNU/Linux target and PIC code generation has
6811 been added by Renesas.
6815 * Bernardo Innocenti (Develer S.r.l.) has contributed the
6816 m68k-uclinux target, based on former work done by Paul Dale
6817 (SnapGear Inc.). Code generation for the ColdFire processors family
6818 has been enhanced and extended to support the MCF 53xx and MCF 54xx
6819 cores, integrating former work done by Peter Barada (Motorola).
6823 Processor-specific changes
6825 * Support for the RM7000 and RM9000 processors has been added. It can
6826 be selected using the -march compiler option and should work with
6827 any MIPS I (mips-*) or MIPS III (mips64-*) configuration.
6828 * Support for revision 2 of the MIPS32 ISA has been added. It can be
6829 selected with the command-line option -march=mips32r2.
6830 * There is a new option, -mfix-sb1, to work around certain SB-1
6835 * It is possible to customize GCC using the following configure-time
6837 + --with-arch, which specifies the default value of the -march
6839 + --with-tune, which specifies the default value of the -mtune
6841 + --with-abi, which specifies the default ABI.
6842 + --with-float=soft, which tells GCC to use software floating
6844 + --with-float=hard, which tells GCC to use hardware floating
6846 * A 64-bit GNU/Linux port has been added. The associated
6847 configurations are mips64-linux-gnu and mips64el-linux-gnu.
6848 * The 32-bit GNU/Linux port now supports Java.
6849 * The IRIX 6 configuration now supports the o32 ABI and will build
6850 o32 multilibs by default. This support is compatible with both
6851 binutils and the SGI tools, but note that several features,
6852 including debugging information and DWARF2 exception handling, are
6853 only available when using the GNU assembler. Use of the GNU
6854 assembler and linker (version 2.15 or above) is strongly
6856 * The IRIX 6 configuration now supports 128-bit long doubles.
6857 * There are two new RTEMS-specific configurations, mips-rtems and
6859 * There are two new *-elf configurations, mipsisa32r2-elf and
6864 * Several [24]ABI bugs have been fixed. Unfortunately, these changes
6865 will break binary compatibility with earlier releases.
6866 * GCC can now use explicit relocation operators when generating
6867 -mabicalls code. This behavior is controlled by -mexplicit-relocs
6868 and can have several performance benefits. For example:
6869 + It allows for more optimization of GOT accesses, including
6870 better scheduling and redundancy elimination.
6871 + It allows sibling calls to be implemented as jumps.
6872 + n32 and n64 leaf functions can use a call-clobbered global
6873 pointer instead of $28.
6874 + The code to set up $gp can be removed from functions that
6876 * A new option, -mxgot, allows the GOT to be bigger than 64k. This
6877 option is equivalent to the assembler's -xgot option and should be
6878 used instead of -Wa,-xgot.
6879 * Frame pointer elimination is now supported when generating 64-bit
6881 * Inline block moves have been optimized to take more account of
6882 alignment information.
6883 * Many internal changes have been made to the MIPS port, mostly aimed
6884 at reducing the reliance on assembler macros.
6888 * GCC 3.4 releases have a number of fixes for PowerPC and PowerPC64
6889 [25]ABI incompatibilities regarding the way parameters are passed
6890 during functions calls. These changes may result in incompatibility
6891 between code compiled with GCC 3.3 and GCC 3.4.
6895 * Support for shared/dylib gcc libraries has been added. It is
6896 enabled by default on powerpc-apple-darwin7.0.0 and up.
6897 * Libgcj is enabled by default. On systems older than
6898 powerpc-apple-darwin7.0.0 you need to install dlcompat.
6899 * 128-bit IBM extended precision format support added for long
6904 * By default, PowerPC64 GNU/Linux now uses natural alignment of
6905 structure elements. The old four byte alignment for double, with
6906 special rules for a struct starting with a double, can be chosen
6907 with -malign-power. This change may result in incompatibility
6908 between code compiled with GCC 3.3 and GCC 3.4.
6909 * -mabi=altivec is now the default rather than -mabi=no-altivec.
6910 * 128-bit IBM extended precision format support added for long
6915 * New command-line options allow to specify the intended execution
6916 environment for generated code:
6917 + -mesa/-mzarch allows to specify whether to generate code
6918 running in ESA/390 mode or in z/Architecture mode (this is
6919 applicable to 31-bit code only).
6920 + -march allows to specify a minimum processor architecture
6921 level (g5, g6, z900, or z990).
6922 + -mtune allows to specify which processor to tune for.
6923 * It is possible to customize GCC using the following configure-time
6925 + --with-mode, which specifies whether to default to assuming
6926 ESA/390 or z/Architecture mode.
6927 + --with-arch, which specifies the default value of the -march
6929 + --with-tune, which specifies the default value of the -mtune
6931 * Support for the z990 processor has been added, and can be selected
6932 using -march=z990 or -mtune=z990. This includes instruction
6933 scheduling tuned for the superscalar instruction pipeline of the
6934 z990 processor as well as support for all new instructions provided
6935 by the long-displacement facility.
6936 * Support to generate 31-bit code optimized for zSeries processors
6937 (running in ESA/390 or in z/Architecture mode) has been added. This
6938 can be selected using -march=z900 and -mzarch respectively.
6939 * Instruction scheduling for the z900 and z990 processors now uses
6940 the DFA pipeline hazard recognizer.
6941 * GCC no longer generates code to maintain a stack backchain,
6942 previously used to generate stack backtraces for debugging
6943 purposes. As replacement that does not incur runtime overhead,
6944 DWARF-2 call frame information is provided by GCC; this is
6945 supported by GDB 6.1. The old behavior can be restored using the
6947 * The stack frame size of functions may now exceed 2 GB in 64-bit
6949 * A port for the 64-bit IBM TPF operating system has been added; the
6950 configuration is s390x-ibm-tpf. This configuration is supported as
6951 cross-compilation target only.
6952 * Various changes to improve the generated code have been
6953 implemented, including:
6954 + GCC now uses the MULTIPLY AND ADD and MULTIPLY AND SUBTRACT
6955 instructions to significantly speed up many floating-point
6957 + GCC now uses the ADD LOGICAL WITH CARRY and SUBTRACT LOGICAL
6958 WITH BORROW instructions to speed up long long arithmetic.
6959 + GCC now uses the SEARCH STRING instruction to implement
6961 + In many cases, function call overhead for 31-bit code has been
6962 reduced by placing the literal pool after the function code
6963 instead of after the function prolog.
6964 + Register 14 is no longer reserved in 64-bit code.
6965 + Handling of global register variables has been improved.
6969 * The option -mflat is deprecated.
6970 * Support for large (> 2GB) frames has been added to the 64-bit port.
6971 * Several [26]ABI bugs have been fixed. Unfortunately, these changes
6972 will break binary compatibility with earlier releases.
6973 * The default debugging format has been switched from STABS to
6974 DWARF-2 for 32-bit code on Solaris 7 and later. DWARF-2 is already
6975 the default debugging format for 64-bit code on Solaris.
6979 * Support for the SH2E processor has been added. Enabled at run time
6980 with the -m2e command line switch, or at configure time by
6981 specifying sh2e as the machine part of the target triple.
6985 * Support for the Mitsubishi V850E1 processor has been added. This is
6986 a variant of the V850E processor with some additional debugging
6991 * Several ABI bugs have been fixed. Unfortunately, these changes
6992 break binary compatibility with earlier releases.
6993 + For big-endian processors, the padding of aggregate return
6994 values larger than a word has changed. If the size of an
6995 aggregate return value is not a multiple of 32 bits, previous
6996 versions of GCC inserted padding in the most-significant bytes
6997 of the first return value register. Aggregates larger than a
6998 word are now padded in the least-significant bytes of the last
6999 return value register used. Aggregates smaller than a word are
7000 still padded in the most-significant bytes. The return value
7001 padding has not changed for little-endian processors.
7002 + Function arguments with 16-byte alignment are now properly
7004 + The implementation of the va_list type has changed. A va_list
7005 value created by va_start from a previous release cannot be
7006 used with va_arg from this release, or vice versa.
7007 * More processor configuration options for Xtensa processors are
7009 + the ABS instruction is now optional;
7010 + the ADDX* and SUBX* instructions are now optional;
7011 + an experimental CONST16 instruction can be used to synthesize
7012 constants instead of loading them from constant pools.
7013 These and other Xtensa processor configuration options can no
7014 longer be enabled or disabled by command-line options; the
7015 processor configuration must be specified by the xtensa-config.h
7016 header file when building GCC. Additionally, the
7017 -mno-serialize-volatile option is no longer supported.
7021 Support for a number of older systems has been declared obsolete in GCC
7022 3.4. Unless there is activity to revive them, the next release of GCC
7023 will have their sources permanently removed.
7025 All configurations of the following processor architectures have been
7027 * Mitsubishi D30V, d30v-*
7028 * AT&T DSP1600 and DSP1610, dsp16xx-*
7031 Also, some individual systems have been obsoleted:
7033 + Support for generating code for operation in APCS/26 mode
7036 + "Bigfoot" port, i370-*. (The other port, s390-*, is actively
7037 maintained and supported.)
7039 + MOSS, i?86-moss-msdos and i?86-*-moss*
7040 + NCR 3000 running System V r.4, i?86-ncr-sysv4*
7041 + FreeBSD with a.out object format, i?86-*-freebsd*aout* and
7043 + GNU/Linux with a.out object format, i?86-linux*aout*
7044 + GNU/Linux with libc5, a.k.a. glibc1, i?86-linux*libc1*
7045 + Interix versions before Interix 3, i?86-*-interix
7046 + Mach microkernel, i?86-mach*
7047 + SCO UnixWare with UDK, i?86-*-udk*
7048 + Generic System V releases 1, 2, and 3, i?86-*-sysv[123]*
7049 + VSTa microkernel, i386-*-vsta
7050 * Motorola M68000 family
7051 + HPUX, m68k-hp-hpux* and m68000-hp-hpux*
7052 + NetBSD with a.out object format (before NetBSD 1.4),
7053 m68k-*-*-netbsd* except m68k-*-*-netbsdelf*
7054 + Generic System V r.4, m68k-*-sysv4*
7056 + Generic VAX, vax-*-* (This is generic VAX only; we have not
7057 obsoleted any VAX triples for specific operating systems.)
7059 Documentation improvements
7061 Other significant improvements
7063 * The build system has undergone several significant cleanups.
7064 Subdirectories will only be configured if they are being built, and
7065 all subdirectory configures are run from the make command. The top
7066 level has been autoconfiscated.
7067 * Building GCC no longer writes to its source directory. This should
7068 help those wishing to share a read-only source directory over NFS
7069 or build from a CD. The exceptions to this feature are if you
7070 configure with either --enable-maintainer-mode or
7071 --enable-generated-files-in-srcdir.
7072 * The -W warning option has been renamed to -Wextra, which is more
7073 easily understood. The older spelling will be retained for
7074 backwards compatibility.
7075 * Substantial improvements in compile time have been made,
7076 particularly for non-optimizing compilations.
7077 __________________________________________________________________
7083 A vast number of bugs have been fixed in 3.4.0, too many to publish a
7084 complete list here. [27]Follow this link to query the Bugzilla database
7085 for the list of over 900 bugs fixed in 3.4.0. This is the list of all
7086 bugs marked as resolved and fixed in 3.4.0 that are not flagged as 3.4
7088 __________________________________________________________________
7094 This section lists the problem reports (PRs) from GCC's bug tracking
7095 system that are known to be fixed in the 3.4.1 release. This list might
7096 not be complete (that is, it is possible that some PRs that have been
7097 fixed are not listed here).
7101 * [28]10129 Ada bootstrap fails on PPC-Darwin - invalid assembler
7102 emitted - PIC related
7103 * [29]14576 [ARM] ICE in libiberty when building gcc-3.4 for arm-elf
7104 * [30]14760 A bug in configure.in prevents using both
7105 --program-suffix and --program-prefix
7106 * [31]14671 [hppa64] bootstrap fails: ICE in
7107 save_call_clobbered_regs, in caller_save.c
7108 * [32]15093 [alpha][Java] make bootstrap fails to configure libffi on
7110 * [33]15178 Solaris 9/x86 fails linking after stage 3
7112 Multi-platform internal compiler errors (ICEs)
7114 * [34]12753 (preprocessor) Memory corruption in preprocessor on bad
7116 * [35]13985 ICE in gcc.c-torture/compile/930621-1.c
7117 * [36]14810 (c++) tree check failures with invalid code involving
7119 * [37]14883 (c++) ICE on invalid code, in cp_parser_lookup_name, in
7121 * [38]15044 (c++) ICE on syntax error, template header
7122 * [39]15057 (c++) Compiling of conditional value throw constructs
7123 cause a segmentation violation
7124 * [40]15064 (c++) typeid of template parameter gives ICE
7125 * [41]15142 (c++) ICE when passing a string where a char* is expected
7126 in a throw statement
7127 * [42]15159 ICE in rtl_verify_flow_info_1
7128 * [43]15165 (c++) ICE in instantiate_template
7129 * [44]15193 Unary minus using pointer to V4SF vector causes
7130 -fforce-mem to exhaust all memory
7131 * [45]15209 (c++) Runs out of memory with packed structs
7132 * [46]15227 (c++) Trouble with invalid function definition
7133 * [47]15285 (c++) instantiate_type ICE when forming pointer to
7135 * [48]15299 (c++) ICE in resolve_overloaded_unification
7136 * [49]15329 (c++) ICE on constructor of member template
7137 * [50]15550 ICE in extract_insn, in recog.c
7138 * [51]15554 (c++) ICE in tsubst_copy, in cp/pt.c
7139 * [52]15640 (c++) ICE on invalid code in arg_assoc, in
7141 * [53]15666 [unit-at-a-time] Gcc abort on valid code
7142 * [54]15696 (c++) ICE with bad pointer-to-member code
7143 * [55]15701 (c++) ICE with friends and template template parameter
7144 * [56]15761 ICE in do_SUBST, in combine.c
7145 * [57]15829 (c++) ICE on Botan-1.3.13 due to -funroll-loops
7149 * [58]14538 All RTEMS targets broken for gnat
7153 * [59]12391 missing warning about assigning to an incomplete type
7154 * [60]14649 atan(1.0) should not be a constant expression
7155 * [61]15004 [unit-at-a-time] no warning for unused paramater in
7157 * [62]15749 --pedantic-errors behaves differently from --pedantic
7158 with C-compiler on GNU/Linux
7160 C++ compiler and library
7162 * [63]10646 non-const reference is incorrectly matched in a "const T"
7163 partial specialization
7164 * [64]12077 wcin.rdbuf()->in_avail() return value too high
7165 * [65]13598 enc_filebuf doesn't work
7166 * [66]14211 const_cast returns lvalue but should be rvalue
7167 * [67]14220 num_put::do_put() undesired float/double behavior
7168 * [68]14245 problem with user-defined allocators in std::basic_string
7169 * [69]14340 libstdc++ Debug mode: failure to convert iterator to
7171 * [70]14600 __gnu_cxx::stdio_sync_filebuf should expose internal
7173 * [71]14668 no warning anymore for reevaluation of declaration
7174 * [72]14775 LFS (large file support) tests missing
7175 * [73]14821 Duplicate namespace alias declaration should not conflict
7176 * [74]14930 Friend declaration ignored
7177 * [75]14932 cannot use offsetof to get offsets of array elements in
7179 * [76]14950 [non unit-at-a-time] always_inline does not mix with
7181 * [77]14962 g++ ignores #pragma redefine_extname
7182 * [78]14975 Segfault on low-level write error during imbue
7183 * [79]15002 Linewise stream input is unusably slow (std::string slow)
7184 * [80]15025 compiler accepts redeclaration of template as
7186 * [81]15046 [arm] Math functions misdetected by cross configuration
7187 * [82]15069 a bit test on a variable of enum type is miscompiled
7188 * [83]15074 g++ -lsupc++ still links against libstdc++
7189 * [84]15083 spurious "statement has no effect" warning
7190 * [85]15096 parse error with templates and pointer to const member
7191 * [86]15287 combination of operator[] and operator .* fails in
7193 * [87]15317 __attribute__ unused in first parameter of constructor
7195 * [88]15337 sizeof on incomplete type diagnostic
7196 * [89]15361 bitset<>::_Find_next fails
7197 * [90]15412 _GLIBCXX_ symbols symbols defined and used in different
7199 * [91]15427 valid code results in incomplete type error
7200 * [92]15471 Incorrect member pointer offsets in anonymous
7202 * [93]15503 nested template problem
7203 * [94]15507 compiler hangs while laying out union
7204 * [95]15542 operator & and template definitions
7205 * [96]15565 SLES9: leading + sign for unsigned int with showpos
7206 * [97]15625 friend defined inside a template fails to find static
7208 * [98]15629 Function templates, overloads, and friend name injection
7209 * [99]15742 'noreturn' attribute ignored in method of template
7211 * [100]15775 Allocator::pointer consistently ignored
7212 * [101]15821 Duplicate namespace alias within namespace rejected
7213 * [102]15862 'enum yn' fails (confict with undeclared builtin)
7214 * [103]15875 rejects pointer to member in template
7215 * [104]15877 valid code using templates and anonymous enums is
7217 * [105]15947 Puzzling error message for wrong destructor declaration
7219 * [106]16020 cannot copy __gnu_debug::bitset
7220 * [107]16154 input iterator concept too restrictive
7221 * [108]16174 deducing top-level consts
7225 * [109]14315 Java compiler is not parallel make safe
7229 * [110]15151 [g77] incorrect logical i/o in 64-bit mode
7233 * [111]7993 private variables cannot be shadowed in subclasses
7237 * [112]15228 useless copies of floating point operands
7238 * [113]15345 [non-unit-at-a-time] unreferenced nested inline
7239 functions not optimized away
7240 * [114]15945 Incorrect floating point optimization
7241 * [115]15526 ftrapv aborts on 0 * (-1)
7242 * [116]14690 Miscompiled POOMA tests
7243 * [117]15112 GCC generates code to write to unchanging memory
7247 * [118]15067 Minor glitch in the source of cpp
7249 Main driver program bugs
7251 * [119]1963 collect2 interprets -oldstyle_liblookup as -o
7254 x86-specific (Intel/AMD)
7256 * [120]15717 Error: can't resolve `L0' {*ABS* section} - `xx' {*UND*
7261 * [121]14782 GCC produces an unaligned data access at -O2
7262 * [122]14828 FAIL: gcc.c-torture/execute/20030408-1.c execution, -O2
7263 * [123]15202 ICE in reload_cse_simplify_operands, in postreload.c
7267 * [124]14610 __float80 constants incorrectly emitted
7268 * [125]14813 init_array sections are initialized in the wrong order
7269 * [126]14857 GCC segfault on duplicated asm statement
7270 * [127]15598 Gcc 3.4 ICE on valid code
7271 * [128]15653 Gcc 3.4 ICE on valid code
7275 * [129]15189 wrong filling of delay slot with -march=mips1 -G0
7276 -mno-split-addresses -mno-explicit-relocs
7277 * [130]15331 Assembler error building gnatlib on IRIX 6.5 with GNU as
7279 * [131]16144 Bogus reference to __divdf3 when -O1
7280 * [132]16176 Miscompilation of unaligned data in MIPS backend
7284 * [133]11591 ICE in gcc.dg/altivec-5.c
7285 * [134]12028 powerpc-eabispe produces bad sCOND operation
7286 * [135]14478 rs6000 geu/ltu patterns generate incorrect code
7287 * [136]14567 long double and va_arg complex args
7288 * [137]14715 Altivec stack layout may overlap gpr save with stack
7290 * [138]14902 (libstdc++) Stream checking functions fail when -pthread
7292 * [139]14924 Compiler ICE on valid code
7293 * [140]14960 -maltivec affects vector return with -mabi=no-altivec
7294 * [141]15106 vector varargs failure passing from altivec to
7295 non-altivec code for -m32
7296 * [142]16026 ICE in function.c:4804, assign_parms, when -mpowerpc64 &
7298 * [143]15191 -maltivec -mabi=no-altivec results in mis-aligned lvx
7300 * [144]15662 Segmentation fault when an exception is thrown - even if
7301 try and catch are specified
7305 * [145]15054 Bad code due to overlapping stack temporaries
7309 * [146]15783 ICE with union assignment in 64-bit mode
7310 * [147]15626 GCC 3.4 emits "ld: warning: relocation error:
7315 * [148]14326 boehm-gc hardcodes to 3DNow! prefetch for x86_64
7316 * [149]14723 Backported -march=nocona from mainline
7317 * [150]15290 __float128 failed to pass to function properly
7319 Cygwin/Mingw32-specific
7321 * [151]15250 Option -mms-bitfields support on GCC 3.4 is not
7322 conformant to MS layout
7323 * [152]15551 -mtune=pentium4 -O2 with sjlj EH breaks stack probe
7324 worker on windows32 targets
7326 Bugs specific to embedded processors
7328 * [153]8309 [m68k] -m5200 produces erroneous SImode set of short
7330 * [154]13250 [SH] Gcc code for rotation clobbers the register, but
7331 gcc continues to use the register as if it was not clobbered
7332 * [155]13803 [coldfire] movqi operand constraints too restrictivefor
7334 * [156]14093 [SH] ICE for code when using -mhitachi option in SH
7335 * [157]14457 [m6811hc] ICE with simple c++ source
7336 * [158]14542 [m6811hc] ICE on simple source
7337 * [159]15100 [SH] cc1plus got hang-up on
7338 libstdc++-v3/testsuite/abi_check.cc
7339 * [160]15296 [CRIS] Delayed branch scheduling causing invalid code on
7341 * [161]15396 [SH] ICE with -O2 -fPIC
7342 * [162]15782 [coldfire] m68k_output_mi_thunk emits wrong code for
7345 Testsuite problems (compiler not affected)
7347 * [163]11610 libstdc++ testcases 27_io/* don't work properly remotely
7348 * [164]15488 (libstdc++) possibly insufficient file permissions for
7349 executing test suite
7350 * [165]15489 (libstdc++) testsuite_files determined incorrectly
7354 * [166]13928 (libstdc++) no whatis info in some man pages generated
7356 * [167]14150 Ada documentation out of date
7357 * [168]14949 (c++) Need to document method visibility changes
7358 * [169]15123 libstdc++-doc: Allocators.3 manpage is empty
7359 __________________________________________________________________
7365 This section lists the problem reports (PRs) from GCC's bug tracking
7366 system that are known to be fixed in the 3.4.2 release. This list might
7367 not be complete (that is, it is possible that some PRs that have been
7368 fixed are not listed here).
7370 Bootstrap failures and issues
7372 * [170]16469 [mips-sgi-irix5.3] bootstrap fails in
7373 libstdc++-v3/testsuite
7374 * [171]16344 [hppa-linux-gnu] libstdc++'s PCH built by
7375 profiledbootstrap does not work with the built compiler
7376 * [172]16842 [Solaris/x86] mkheaders can not find mkheaders.conf
7378 Multi-platform internal compiler errors (ICEs)
7380 * [173]12608 (c++) ICE: expected class 't', have 'x' (error_mark) in
7381 cp_parser_class_specifier, in cp/parser.c
7382 * [174]14492 ICE in loc_descriptor_from_tree, in dwarf2out.c
7383 * [175]15461 (c++) ICE due to NRV and inlining
7384 * [176]15890 (c++) ICE in c_expand_expr, in c-common.c
7385 * [177]16180 ICE: segmentation fault in RTL optimization
7386 * [178]16224 (c++) ICE in write_unscoped_name (template/namespace)
7387 * [179]16408 ICE: in delete_insn, in cfgrtl.c
7388 * [180]16529 (c++) ICE for: namespace-alias shall not be declared as
7389 the name of any other entity
7390 * [181]16698 (c++) ICE with exceptions and declaration of __cxa_throw
7391 * [182]16706 (c++) ICE in finish_member_declaration, in
7393 * [183]16810 (c++) Legal C++ program with cast gives ICE in
7395 * [184]16851 (c++) ICE when throwing a comma expression
7396 * [185]16870 (c++) Boost.Spirit causes ICE in tsubst, in cp/pt.c
7397 * [186]16904 (c++) ICE in finish_class_member_access_expr, in
7399 * [187]16905 (c++) ICE (segfault) with exceptions
7400 * [188]16964 (c++) ICE in cp_parser_class_specifier due to
7402 * [189]17068 (c++) ICE: tree check: expected class 'd', have 'x'
7403 (identifier_node) in dependent_template_p, in cp/pt.c
7407 * [190]16366 Preprocessor option -remap causes memory corruption
7411 * [191]15345 unreferenced nested inline functions not optimized away
7412 * [192]16590 Incorrect execution when compiling with -O2
7413 * [193]16693 Bitwise AND is lost when used within a cast to an enum
7414 of the same precision
7415 * [194]17078 Jump into if(0) substatement fails
7417 Problems in generated debug information
7419 * [195]13956 incorrect stabs for nested local variables
7423 * [196]16684 GCC should not warn about redundant redeclarations of
7426 C++ compiler and library
7428 * [197]12658 Thread safety problems in locale::global() and
7430 * [198]13092 g++ accepts invalid pointer-to-member conversion
7431 * [199]15320 Excessive memory consumption
7432 * [200]16246 Incorrect template argument deduction
7433 * [201]16273 Memory exhausted when using nested classes and virtual
7435 * [202]16401 ostringstream in gcc 3.4.x very slow for big data
7436 * [203]16411 undefined reference to
7437 __gnu_cxx::stdio_sync_filebuf<char, std::char_traits<char>
7439 * [204]16489 G++ incorrectly rejects use of a null constant integral
7440 expression as a null constant pointer
7441 * [205]16618 offsetof fails with constant member
7442 * [206]16637 syntax error reported for valid input code
7443 * [207]16717 __attribute__((constructor)) broken in C++
7444 * [208]16813 compiler error in DEBUG version of range insertion
7446 * [209]16853 pointer-to-member initialization from incompatible one
7448 * [210]16889 ambiguity is not detected
7449 * [211]16959 Segmentation fault in ios_base::sync_with_stdio
7451 Java compiler and library
7453 * [212]7587 direct threaded interpreter not thread-safe
7454 * [213]16473 ServerSocket accept() leaks file descriptors
7455 * [214]16478 Hash synchronization deadlock with finalizers
7459 * [215]10695 ICE in dwarf2out_frame_debug_expr, in dwarf2out.c
7460 * [216]16974 could not split insn (ice in final_scan_insn, in
7465 * [217]16298 ICE in output_operand
7466 * [218]17113 ICE with SSE2 intrinsics
7470 * [219]14697 libstdc++ couldn't find 32bit libgcc_s
7474 * [220]15869 [mips64] No NOP after LW (with -mips1 -O0)
7475 * [221]16325 [mips64] value profiling clobbers gp on mips
7476 * [222]16357 [mipsisa64-elf] ICE copying 7 bytes between extern
7478 * [223]16380 [mips64] Use of uninitialised register after dbra
7480 * [224]16407 [mips64] Unaligned access to local variables
7481 * [225]16643 [mips64] verify_local_live_at_start ICE after
7482 crossjumping & cfgcleanup
7486 * [226]15927 THUMB -O2: strength-reduced iteration variable ends up
7488 * [227]15948 THUMB: ICE with non-commutative cbranch
7489 * [228]17019 THUMB: bad switch statement in md code for
7490 addsi3_cbranch_scratch
7494 * [229]16130 ICE on valid code: in bundling, in config/ia64/ia64.c
7496 * [230]16142 ICE on valid code: in bundling, in config/ia64/ia64.c
7498 * [231]16278 Gcc failed to build Linux kernel with -mtune=merced
7499 * [232]16414 ICE on valid code: typo in comparison of asm_noperands
7501 * [233]16445 ICE on valid code: don't count ignored insns
7502 * [234]16490 ICE (segfault) while compiling with -fprofile-use
7503 * [235]16683 ia64 does not honor SUBTARGET_EXTRA_SPECS
7507 * [236]16195 (ppc64): Miscompilation of GCC 3.3.x by 3.4.x
7508 * [237]16239 ICE on ppc64 (mozilla 1.7 compile, -O1 -fno-exceptions
7513 * [238]16199 ICE while compiling apache 2.0.49
7514 * [239]16416 -m64 doesn't imply -mcpu=v9 anymore
7515 * [240]16430 ICE when returning non-C aggregates larger than 16 bytes
7517 Bugs specific to embedded processors
7519 * [241]16379 [m32r] can't output large model function call of memcpy
7520 * [242]17093 [m32r] ICE with -msdata=use -O0
7521 * [243]17119 [m32r] ICE at switch case 0x8000
7525 * [244]15928 libstdc++ in 3.4.x doesn't cross-compile for djgpp
7527 Alpha Tru64-specific
7529 * [245]16210 libstdc++ gratuitously omits "long long" I/O
7531 Testsuite, documentation issues (compiler is not affected):
7533 * [246]15488 (libstdc++) possibly insufficient file permissions for
7534 executing test suite
7535 * [247]16250 ada/doctools runs makeinfo even in release tarball
7536 __________________________________________________________________
7540 This is the [248]list of problem reports (PRs) from GCC's bug tracking
7541 system that are known to be fixed in the 3.4.3 release. This list might
7542 not be complete (that is, it is possible that some PRs that have been
7543 fixed are not listed here).
7547 * [249]17369 [ia64] Bootstrap failure with binutils-2.15.90.0.1.1
7548 * [250]17850 [arm-elf] bootstrap failure - libstdc++ uses strtold
7551 Internal compiler errors (ICEs) affecting multiple platforms
7553 * [251]13948 (java) GCJ segmentation fault while compiling GL4Java
7555 * [252]14492 ICE in loc_descriptor_from_tree, in dwarf2out.c
7556 * [253]16301 (c++) ICE when "strong" attribute is attached to a using
7558 * [254]16566 ICE with flexible arrays
7559 * [255]17023 ICE with nested functions in parameter declaration
7560 * [256]17027 ICE with noreturn function in loop at -O2
7561 * [257]17524 ICE in grokdeclarator, in cp/decl.c
7562 * [258]17826 (c++) ICE in cp_tree_equal
7564 C and optimization bugs
7566 * [259]15526 -ftrapv aborts on 0 * (-1)
7567 * [260]16999 #ident stopped working
7568 * [261]17503 quadratic behaviour in invalid_mode_change_p
7569 * [262]17581 Long long arithmetic fails inside a switch/case
7570 statement when compiled with -O2
7571 * [263]18129 -fwritable-strings doesn't work
7573 C++ compiler and library bugs
7575 * [264]10975 incorrect initial ostringstream::tellp()
7576 * [265]11722 Unbuffered filebuf::sgetn is slow
7577 * [266]14534 Unrecognizing static function as a template parameter
7578 when its return value is also templated
7579 * [267]15172 Copy constructor optimization in aggregate
7581 * [268]15786 Bad error message for frequently occuring error.
7582 * [269]16162 Rejects valid member-template-definition
7583 * [270]16612 empty basic_strings can't live in shared memory
7584 * [271]16715 std::basic_iostream is instantiated when used, even
7585 though instantiations are already contained in libstdc++
7586 * [272]16848 code in /ext/demangle.h appears broken
7587 * [273]17132 GCC fails to eliminate function template specialization
7588 when argument deduction fails
7589 * [274]17259 One more _S_leaf incorrectly qualified with _RopeRep::
7591 * [275]17327 use of `enumeral_type' in template type unification
7592 * [276]17393 "unused variable '._0'" warning with -Wall
7593 * [277]17501 Confusion with member templates
7594 * [278]17537 g++ not passing -lstdc++ to linker when all command line
7595 arguments are libraries
7596 * [279]17585 usage of unqualified name of static member from within
7598 * [280]17821 Poor diagnostic for using "." instead of "->"
7599 * [281]17829 wrong error: call of overloaded function is ambiguous
7600 * [282]17851 Misleading diagnostic for invalid function declarations
7601 with undeclared types
7602 * [283]17976 Destructor is called twice
7603 * [284]18020 rejects valid definition of enum value in template
7604 * [285]18093 bogus conflict in namespace aliasing
7605 * [286]18140 C++ parser bug when using >> in templates
7609 * [287]17541 data statements with double precision constants fail
7613 * [288]17853 -O2 ICE for MMX testcase
7617 * [289]17245 ICE compiling gsl-1.5 statistics/lag1.c
7621 * [290]17167 FATAL:Symbol L_foo$stub already defined.
7625 * [291]17277 could not catch an exception when specified -maix64
7629 * [292]17505 <cmath> calls acosf(), ceilf(), and other functions
7630 missing from system libraries
7634 * [293]17684 /usr/ccs/bin/ld: Can't create libgcc_s.sl
7638 * [294]17384 ICE with mode attribute on structures
7642 * [295]17770 No NOP after LWL with -mips1
7644 Other embedded target specific
7646 * [296]11476 [arc-elf] gcc ICE on newlib's vfprintf.c
7647 * [297]14064 [avr-elf] -fdata-sections triggers ICE
7648 * [298]14678 [m68hc11-elf] gcc ICE
7649 * [299]15583 [powerpc-rtems] powerpc-rtems lacks __USE_INIT_FINI__
7650 * [300]15790 [i686-coff] Alignment error building gcc with i686-coff
7652 * [301]15886 [SH] Miscompilation with -O2 -fPIC
7653 * [302]16884 [avr-elf] [fweb related] bug while initializing
7656 Bugs relating to debugger support
7658 * [303]13841 missing debug info for _Complex function arguments
7659 * [304]15860 [big-endian targets] No DW_AT_location debug info is
7660 emitted for formal arguments to a function that uses "register"
7663 Testsuite issues (compiler not affected)
7665 * [305]17465 Testsuite in libffi overrides LD_LIBRARY_PATH
7666 * [306]17469 Testsuite in libstdc++ overrides LD_LIBRARY_PATH
7667 * [307]18138 [mips-sgi-irix6.5] libgcc_s.so.1 not found by 64-bit
7672 * [308]15498 typo in gcc manual: non-existing locale example en_UK,
7674 * [309]15747 [mips-sgi-irix5.3] /bin/sh hangs during bootstrap:
7675 document broken shell
7676 * [310]16406 USE_LD_AS_NEEDED undocumented
7677 __________________________________________________________________
7681 This is the [311]list of problem reports (PRs) from GCC's bug tracking
7682 system that are known to be fixed in the 3.4.4 release. This list might
7683 not be complete (that is, it is possible that some PRs that have been
7684 fixed are not listed here).
7685 __________________________________________________________________
7689 This is the [312]list of problem reports (PRs) from GCC's bug tracking
7690 system that are known to be fixed in the 3.4.5 release. This list might
7691 not be complete (that is, it is possible that some PRs that have been
7692 fixed are not listed here).
7696 * [313]24688 sco_math fixincl breaks math.h
7700 * [314]17188 struct Foo { } redefinition
7701 * [315]20187 wrong code for ((unsigned char)(unsigned long
7702 long)((a?a:1)&(a*b)))?0:1)
7703 * [316]21873 infinite warning loop on bad array initializer
7704 * [317]21899 enum definition accepts values to be overriden
7705 * [318]22061 ICE in find_function_data, in function.c
7706 * [319]22308 Failure to diagnose violation of constraint 6.516p2
7707 * [320]22458 ICE on missing brace
7708 * [321]22589 ICE casting to long long
7709 * [322]24101 Segfault with preprocessed source
7711 C++ compiler and library bugs
7713 * [323]10611 operations on vector mode not recognized in C++
7714 * [324]13377 unexpected behavior of namespace usage directive
7715 * [325]16002 Strange error message with new parser
7716 * [326]17413 local classes as template argument
7717 * [327]17609 spurious error message after using keyword
7718 * [328]17618 ICE in cp_convert_to_pointer, in cp/cvt.c
7719 * [329]18124 ICE with invalid template template parameter
7720 * [330]18155 typedef in template declaration not rejected
7721 * [331]18177 ICE with const_cast for undeclared variable
7722 * [332]18368 C++ error message regression
7723 * [333]16378 ICE when returning a copy of a packed member
7724 * [334]18466 int ::i; accepted
7725 * [335]18512 ICE on invalid usage of template base class
7726 * [336]18454 ICE when returning undefined type
7727 * [337]18738 typename not allowed with non-dependent qualified name
7728 * [338]18803 rejects access to operator() in template
7729 * [339]19004 ICE in uses_template_parms, in cp/pt.c
7730 * [340]19208 Spurious error about variably modified type
7731 * [341]18253 bad error message / ICE for invalid template parameter
7732 * [342]19608 ICE after friend function definition in local class
7733 * [343]19884 ICE on explicit instantiation of a non-template
7735 * [344]20153 ICE when C++ template function contains anonymous union
7736 * [345]20563 Infinite loop in diagnostic (and ice after error
7738 * [346]20789 ICE with incomplete type in template
7739 * [347]21336 Internal compiler error when using custom new operators
7740 * [348]21768 ICE in error message due to violation of coding
7742 * [349]21853 constness of pointer to data member ignored
7743 * [350]21903 Default argument of template function causes a
7745 * [351]21983 multiple diagnostics
7746 * [352]21987 New testsuite failure
7747 g++.dg/warn/conversion-function-1.C
7748 * [353]22153 ICE on invalid template specialization
7749 * [354]22172 Internal compiler error, seg fault.
7750 * [355]21286 filebuf::xsgetn vs pipes
7751 * [356]22233 ICE with wrong number of template parameters
7752 * [357]22508 ICE after invalid operator new
7753 * [358]22545 ICE with pointer to class member & user defined
7755 * [359]23528 Wrong default allocator in ext/hash_map
7756 * [360]23550 char_traits requirements/1.cc test bad math
7757 * [361]23586 Bad diagnostic for invalid namespace-name
7758 * [362]23624 ICE in invert_truthvalue, in fold-const.c
7759 * [363]23639 Bad error message: not a member of '<declaration error>'
7760 * [364]23797 ICE on typename outside template
7761 * [365]23965 Bogus error message: no matching function for call to
7763 * [366]24052 &#`label_decl' not supported by dump_expr#<expression
7765 * [367]24580 virtual base class cause exception not to be caught
7767 Problems in generated debug information
7769 * [368]24267 Bad DWARF for altivec vectors
7771 Optimizations issues
7773 * [369]17810 ICE in verify_local_live_at_start
7774 * [370]17860 Wrong generated code for loop with varying bound
7775 * [371]21709 ICE on compile-time complex NaN
7776 * [372]21964 broken tail call at -O2 or more
7777 * [373]22167 Strange optimization bug when using -Os
7778 * [374]22619 Compilation failure for real_const_1.f and
7780 * [375]23241 Invalid code generated for comparison of uchar to 255
7781 * [376]23478 Miscompilation due to reloading of a var that is also
7783 * [377]24470 segmentation fault in cc1plus when compiling with -O
7784 * [378]24950 ICE in operand_subword_force
7786 Precompiled headers problems
7788 * [379]14400 Cannot compile qt-x11-free-3.3.0
7789 * [380]14940 PCH largefile test fails on various platforms
7793 * [381]20239 ICE on empty preprocessed input
7794 * [382]15220 "gcc -E -MM -MG" reports missing system headers in
7799 * [383]19275 gcc.dg/20020919-1.c fails with -fpic/-fPIC on
7804 * [384]21888 bootstrap failure with linker relaxation enabled
7808 * [385]15342 [arm-linux]: ICE in verify_local_live_at_start
7809 * [386]23985 Memory aliasing information incorrect in inlined memcpy
7813 * [387]16719 Illegal move of byte into address register causes
7818 * [388]21723 ICE while building libgfortran
7819 * [389]21841 -mhp-ld/-mgnu-ld documentation
7823 * [390]23644 IA-64 hardware models and configuration options
7825 * [391]24718 Shared libgcc not used for linking by default
7829 * [392]18421 ICE in reload_cse_simplify_operands, in postreload.c
7833 * [393]20621 ICE in change_address_1, in emit-rtl.c
7835 PowerPC and PowerPC64 specific
7837 * [394]18583 error on valid code: const
7838 __attribute__((altivec(vector__))) doesn't work in arrays
7839 * [395]20191 ICE in reload_cse_simplify_operands
7840 * [396]22083 AIX: TARGET_C99_FUNCTIONS is wrongly defined
7841 * [397]23070 CALL_V4_CLEAR_FP_ARGS flag not properly set
7842 * [398]23404 gij trashes args of functions with more than 8 fp args
7843 * [399]23539 C & C++ compiler generating misaligned references
7844 regardless of compiler flags
7845 * [400]24102 floatdisf2_internal2 broken
7846 * [401]24465 -mminimal-toc miscompilation of __thread vars
7850 * [402]19933 Problem with define of HUGE_VAL in math_c99
7851 * [403]21889 Native Solaris assembler cannot grok DTP-relative debug
7856 * [404]19300 PCH failures on sparc-linux
7857 * [405]20301 Assembler labels have a leading "-"
7858 * [406]20673 C PCH testsuite assembly comparison failure
7860 x86 and x86_64 specific
7862 * [407]18582 ICE with arrays of type V2DF
7863 * [408]19340 Compilation SEGFAULTs with -O1 -fschedule-insns2
7865 * [409]21716 ICE in reg-stack.c's swap_rtx_condition
7866 * [410]24315 amd64 fails -fpeephole2
7867 __________________________________________________________________
7871 This is the [411]list of problem reports (PRs) from GCC's bug tracking
7872 system that are known to be fixed in the 3.4.6 release. This list might
7873 not be complete (that is, it is possible that some PRs that have been
7874 fixed are not listed here).
7877 For questions related to the use of GCC, please consult these web
7878 pages and the [412]GCC manuals. If that fails, the
7879 [413]gcc-help@gcc.gnu.org mailing list might help. Comments on these
7880 web pages and the development of GCC are welcome on our developer
7881 list at [414]gcc@gcc.gnu.org. All of [415]our lists have public
7884 Copyright (C) [416]Free Software Foundation, Inc. Verbatim copying and
7885 distribution of this entire article is permitted in any medium,
7886 provided this notice is preserved.
7888 These pages are [417]maintained by the GCC team. Last modified
7893 1. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.6
7894 2. http://gcc.gnu.org/gcc-3.4/changes.html#cplusplus
7895 3. http://gcc.gnu.org/gcc-3.3/changes.html#obsolete_systems
7896 4. http://gcc.gnu.org/gcc-3.4/changes.html#obsolete_systems
7897 5. http://gcc.gnu.org/gcc-3.4/mips-abi.html
7898 6. http://gcc.gnu.org/gcc-3.4/sparc-abi.html
7899 7. http://www.boost.org/
7900 8. http://gcc.gnu.org/PR11953
7901 9. http://gcc.gnu.org/PR8361
7902 10. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Other-Builtins.html#Other%20Builtins
7903 11. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_closed.html#209
7904 12. http://gcc.gnu.org/bugs/#cxx_rvalbind
7905 13. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Objective-C-Dialect-Options.html
7906 14. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Objective-C-Dialect-Options.html
7907 15. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Objective-C-Dialect-Options.html
7908 16. http://www.gnu.org/software/classpath/
7909 17. http://www.eclipse.org/
7910 18. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/g77/News.html
7911 19. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Alpha-Built-in-Functions.html
7912 20. http://h30097.www3.hp.com/docs/base_doc/DOCUMENTATION/V51A_HTML/ARH9MBTE/DTMNPLTN.HTM#normal-argument-list-structure
7913 21. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gccint/Processor-pipeline-description.html
7914 22. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gccint/Comparison-of-the-two-descriptions.html
7915 23. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gccint/Processor-pipeline-description.html
7916 24. http://gcc.gnu.org/gcc-3.4/mips-abi.html
7917 25. http://gcc.gnu.org/gcc-3.4/powerpc-abi.html
7918 26. http://gcc.gnu.org/gcc-3.4/sparc-abi.html
7919 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
7920 28. http://gcc.gnu.org/PR10129
7921 29. http://gcc.gnu.org/PR14576
7922 30. http://gcc.gnu.org/PR14760
7923 31. http://gcc.gnu.org/PR14671
7924 32. http://gcc.gnu.org/PR15093
7925 33. http://gcc.gnu.org/PR15178
7926 34. http://gcc.gnu.org/PR12753
7927 35. http://gcc.gnu.org/PR13985
7928 36. http://gcc.gnu.org/PR14810
7929 37. http://gcc.gnu.org/PR14883
7930 38. http://gcc.gnu.org/PR15044
7931 39. http://gcc.gnu.org/PR15057
7932 40. http://gcc.gnu.org/PR15064
7933 41. http://gcc.gnu.org/PR15142
7934 42. http://gcc.gnu.org/PR15159
7935 43. http://gcc.gnu.org/PR15165
7936 44. http://gcc.gnu.org/PR15193
7937 45. http://gcc.gnu.org/PR15209
7938 46. http://gcc.gnu.org/PR15227
7939 47. http://gcc.gnu.org/PR15285
7940 48. http://gcc.gnu.org/PR15299
7941 49. http://gcc.gnu.org/PR15329
7942 50. http://gcc.gnu.org/PR15550
7943 51. http://gcc.gnu.org/PR15554
7944 52. http://gcc.gnu.org/PR15640
7945 53. http://gcc.gnu.org/PR15666
7946 54. http://gcc.gnu.org/PR15696
7947 55. http://gcc.gnu.org/PR15701
7948 56. http://gcc.gnu.org/PR15761
7949 57. http://gcc.gnu.org/PR15829
7950 58. http://gcc.gnu.org/PR14538
7951 59. http://gcc.gnu.org/PR12391
7952 60. http://gcc.gnu.org/PR14649
7953 61. http://gcc.gnu.org/PR15004
7954 62. http://gcc.gnu.org/PR15749
7955 63. http://gcc.gnu.org/PR10646
7956 64. http://gcc.gnu.org/PR12077
7957 65. http://gcc.gnu.org/PR13598
7958 66. http://gcc.gnu.org/PR14211
7959 67. http://gcc.gnu.org/PR14220
7960 68. http://gcc.gnu.org/PR14245
7961 69. http://gcc.gnu.org/PR14340
7962 70. http://gcc.gnu.org/PR14600
7963 71. http://gcc.gnu.org/PR14668
7964 72. http://gcc.gnu.org/PR14775
7965 73. http://gcc.gnu.org/PR14821
7966 74. http://gcc.gnu.org/PR14930
7967 75. http://gcc.gnu.org/PR14932
7968 76. http://gcc.gnu.org/PR14950
7969 77. http://gcc.gnu.org/PR14962
7970 78. http://gcc.gnu.org/PR14975
7971 79. http://gcc.gnu.org/PR15002
7972 80. http://gcc.gnu.org/PR15025
7973 81. http://gcc.gnu.org/PR15046
7974 82. http://gcc.gnu.org/PR15069
7975 83. http://gcc.gnu.org/PR15074
7976 84. http://gcc.gnu.org/PR15083
7977 85. http://gcc.gnu.org/PR15096
7978 86. http://gcc.gnu.org/PR15287
7979 87. http://gcc.gnu.org/PR15317
7980 88. http://gcc.gnu.org/PR15337
7981 89. http://gcc.gnu.org/PR15361
7982 90. http://gcc.gnu.org/PR15412
7983 91. http://gcc.gnu.org/PR15427
7984 92. http://gcc.gnu.org/PR15471
7985 93. http://gcc.gnu.org/PR15503
7986 94. http://gcc.gnu.org/PR15507
7987 95. http://gcc.gnu.org/PR15542
7988 96. http://gcc.gnu.org/PR15565
7989 97. http://gcc.gnu.org/PR15625
7990 98. http://gcc.gnu.org/PR15629
7991 99. http://gcc.gnu.org/PR15742
7992 100. http://gcc.gnu.org/PR15775
7993 101. http://gcc.gnu.org/PR15821
7994 102. http://gcc.gnu.org/PR15862
7995 103. http://gcc.gnu.org/PR15875
7996 104. http://gcc.gnu.org/PR15877
7997 105. http://gcc.gnu.org/PR15947
7998 106. http://gcc.gnu.org/PR16020
7999 107. http://gcc.gnu.org/PR16154
8000 108. http://gcc.gnu.org/PR16174
8001 109. http://gcc.gnu.org/PR14315
8002 110. http://gcc.gnu.org/PR15151
8003 111. http://gcc.gnu.org/PR7993
8004 112. http://gcc.gnu.org/PR15228
8005 113. http://gcc.gnu.org/PR15345
8006 114. http://gcc.gnu.org/PR15945
8007 115. http://gcc.gnu.org/PR15526
8008 116. http://gcc.gnu.org/PR14690
8009 117. http://gcc.gnu.org/PR15112
8010 118. http://gcc.gnu.org/PR15067
8011 119. http://gcc.gnu.org/PR1963
8012 120. http://gcc.gnu.org/PR15717
8013 121. http://gcc.gnu.org/PR14782
8014 122. http://gcc.gnu.org/PR14828
8015 123. http://gcc.gnu.org/PR15202
8016 124. http://gcc.gnu.org/PR14610
8017 125. http://gcc.gnu.org/PR14813
8018 126. http://gcc.gnu.org/PR14857
8019 127. http://gcc.gnu.org/PR15598
8020 128. http://gcc.gnu.org/PR15653
8021 129. http://gcc.gnu.org/PR15189
8022 130. http://gcc.gnu.org/PR15331
8023 131. http://gcc.gnu.org/PR16144
8024 132. http://gcc.gnu.org/PR16176
8025 133. http://gcc.gnu.org/PR11591
8026 134. http://gcc.gnu.org/PR12028
8027 135. http://gcc.gnu.org/PR14478
8028 136. http://gcc.gnu.org/PR14567
8029 137. http://gcc.gnu.org/PR14715
8030 138. http://gcc.gnu.org/PR14902
8031 139. http://gcc.gnu.org/PR14924
8032 140. http://gcc.gnu.org/PR14960
8033 141. http://gcc.gnu.org/PR15106
8034 142. http://gcc.gnu.org/PR16026
8035 143. http://gcc.gnu.org/PR15191
8036 144. http://gcc.gnu.org/PR15662
8037 145. http://gcc.gnu.org/PR15054
8038 146. http://gcc.gnu.org/PR15783
8039 147. http://gcc.gnu.org/PR15626
8040 148. http://gcc.gnu.org/PR14326
8041 149. http://gcc.gnu.org/PR14723
8042 150. http://gcc.gnu.org/PR15290
8043 151. http://gcc.gnu.org/PR15250
8044 152. http://gcc.gnu.org/PR15551
8045 153. http://gcc.gnu.org/PR8309
8046 154. http://gcc.gnu.org/PR13250
8047 155. http://gcc.gnu.org/PR13803
8048 156. http://gcc.gnu.org/PR14093
8049 157. http://gcc.gnu.org/PR14457
8050 158. http://gcc.gnu.org/PR14542
8051 159. http://gcc.gnu.org/PR15100
8052 160. http://gcc.gnu.org/PR15296
8053 161. http://gcc.gnu.org/PR15396
8054 162. http://gcc.gnu.org/PR15782
8055 163. http://gcc.gnu.org/PR11610
8056 164. http://gcc.gnu.org/PR15488
8057 165. http://gcc.gnu.org/PR15489
8058 166. http://gcc.gnu.org/PR13928
8059 167. http://gcc.gnu.org/PR14150
8060 168. http://gcc.gnu.org/PR14949
8061 169. http://gcc.gnu.org/PR15123
8062 170. http://gcc.gnu.org/PR16469
8063 171. http://gcc.gnu.org/PR16344
8064 172. http://gcc.gnu.org/PR16842
8065 173. http://gcc.gnu.org/PR12608
8066 174. http://gcc.gnu.org/PR14492
8067 175. http://gcc.gnu.org/PR15461
8068 176. http://gcc.gnu.org/PR15890
8069 177. http://gcc.gnu.org/PR16180
8070 178. http://gcc.gnu.org/PR16224
8071 179. http://gcc.gnu.org/PR16408
8072 180. http://gcc.gnu.org/PR16529
8073 181. http://gcc.gnu.org/PR16698
8074 182. http://gcc.gnu.org/PR16706
8075 183. http://gcc.gnu.org/PR16810
8076 184. http://gcc.gnu.org/PR16851
8077 185. http://gcc.gnu.org/PR16870
8078 186. http://gcc.gnu.org/PR16904
8079 187. http://gcc.gnu.org/PR16905
8080 188. http://gcc.gnu.org/PR16964
8081 189. http://gcc.gnu.org/PR17068
8082 190. http://gcc.gnu.org/PR16366
8083 191. http://gcc.gnu.org/PR15345
8084 192. http://gcc.gnu.org/PR16590
8085 193. http://gcc.gnu.org/PR16693
8086 194. http://gcc.gnu.org/PR17078
8087 195. http://gcc.gnu.org/PR13956
8088 196. http://gcc.gnu.org/PR16684
8089 197. http://gcc.gnu.org/PR12658
8090 198. http://gcc.gnu.org/PR13092
8091 199. http://gcc.gnu.org/PR15320
8092 200. http://gcc.gnu.org/PR16246
8093 201. http://gcc.gnu.org/PR16273
8094 202. http://gcc.gnu.org/PR16401
8095 203. http://gcc.gnu.org/PR16411
8096 204. http://gcc.gnu.org/PR16489
8097 205. http://gcc.gnu.org/PR16618
8098 206. http://gcc.gnu.org/PR16637
8099 207. http://gcc.gnu.org/PR16717
8100 208. http://gcc.gnu.org/PR16813
8101 209. http://gcc.gnu.org/PR16853
8102 210. http://gcc.gnu.org/PR16889
8103 211. http://gcc.gnu.org/PR16959
8104 212. http://gcc.gnu.org/PR7587
8105 213. http://gcc.gnu.org/PR16473
8106 214. http://gcc.gnu.org/PR16478
8107 215. http://gcc.gnu.org/PR10695
8108 216. http://gcc.gnu.org/PR16974
8109 217. http://gcc.gnu.org/PR16298
8110 218. http://gcc.gnu.org/PR17113
8111 219. http://gcc.gnu.org/PR14697
8112 220. http://gcc.gnu.org/PR15869
8113 221. http://gcc.gnu.org/PR16325
8114 222. http://gcc.gnu.org/PR16357
8115 223. http://gcc.gnu.org/PR16380
8116 224. http://gcc.gnu.org/PR16407
8117 225. http://gcc.gnu.org/PR16643
8118 226. http://gcc.gnu.org/PR15927
8119 227. http://gcc.gnu.org/PR15948
8120 228. http://gcc.gnu.org/PR17019
8121 229. http://gcc.gnu.org/PR16130
8122 230. http://gcc.gnu.org/PR16142
8123 231. http://gcc.gnu.org/PR16278
8124 232. http://gcc.gnu.org/PR16414
8125 233. http://gcc.gnu.org/PR16445
8126 234. http://gcc.gnu.org/PR16490
8127 235. http://gcc.gnu.org/PR16683
8128 236. http://gcc.gnu.org/PR16195
8129 237. http://gcc.gnu.org/PR16239
8130 238. http://gcc.gnu.org/PR16199
8131 239. http://gcc.gnu.org/PR16416
8132 240. http://gcc.gnu.org/PR16430
8133 241. http://gcc.gnu.org/PR16379
8134 242. http://gcc.gnu.org/PR17093
8135 243. http://gcc.gnu.org/PR17119
8136 244. http://gcc.gnu.org/PR15928
8137 245. http://gcc.gnu.org/PR16210
8138 246. http://gcc.gnu.org/PR15488
8139 247. http://gcc.gnu.org/PR16250
8140 248. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.4.3
8141 249. http://gcc.gnu.org/PR17369
8142 250. http://gcc.gnu.org/PR17850
8143 251. http://gcc.gnu.org/PR13948
8144 252. http://gcc.gnu.org/PR14492
8145 253. http://gcc.gnu.org/PR16301
8146 254. http://gcc.gnu.org/PR16566
8147 255. http://gcc.gnu.org/PR17023
8148 256. http://gcc.gnu.org/PR17027
8149 257. http://gcc.gnu.org/PR17524
8150 258. http://gcc.gnu.org/PR17826
8151 259. http://gcc.gnu.org/PR15526
8152 260. http://gcc.gnu.org/PR16999
8153 261. http://gcc.gnu.org/PR17503
8154 262. http://gcc.gnu.org/PR17581
8155 263. http://gcc.gnu.org/PR18129
8156 264. http://gcc.gnu.org/PR10975
8157 265. http://gcc.gnu.org/PR11722
8158 266. http://gcc.gnu.org/PR14534
8159 267. http://gcc.gnu.org/PR15172
8160 268. http://gcc.gnu.org/PR15786
8161 269. http://gcc.gnu.org/PR16162
8162 270. http://gcc.gnu.org/PR16612
8163 271. http://gcc.gnu.org/PR16715
8164 272. http://gcc.gnu.org/PR16848
8165 273. http://gcc.gnu.org/PR17132
8166 274. http://gcc.gnu.org/PR17259
8167 275. http://gcc.gnu.org/PR17327
8168 276. http://gcc.gnu.org/PR17393
8169 277. http://gcc.gnu.org/PR17501
8170 278. http://gcc.gnu.org/PR17537
8171 279. http://gcc.gnu.org/PR17585
8172 280. http://gcc.gnu.org/PR17821
8173 281. http://gcc.gnu.org/PR17829
8174 282. http://gcc.gnu.org/PR17851
8175 283. http://gcc.gnu.org/PR17976
8176 284. http://gcc.gnu.org/PR18020
8177 285. http://gcc.gnu.org/PR18093
8178 286. http://gcc.gnu.org/PR18140
8179 287. http://gcc.gnu.org/PR17541
8180 288. http://gcc.gnu.org/PR17853
8181 289. http://gcc.gnu.org/PR17245
8182 290. http://gcc.gnu.org/PR17167
8183 291. http://gcc.gnu.org/PR17277
8184 292. http://gcc.gnu.org/PR17505
8185 293. http://gcc.gnu.org/PR17684
8186 294. http://gcc.gnu.org/PR17384
8187 295. http://gcc.gnu.org/PR17770
8188 296. http://gcc.gnu.org/PR11476
8189 297. http://gcc.gnu.org/PR14064
8190 298. http://gcc.gnu.org/PR14678
8191 299. http://gcc.gnu.org/PR15583
8192 300. http://gcc.gnu.org/PR15790
8193 301. http://gcc.gnu.org/PR15886
8194 302. http://gcc.gnu.org/PR16884
8195 303. http://gcc.gnu.org/PR13841
8196 304. http://gcc.gnu.org/PR15860
8197 305. http://gcc.gnu.org/PR17465
8198 306. http://gcc.gnu.org/PR17469
8199 307. http://gcc.gnu.org/PR18138
8200 308. http://gcc.gnu.org/PR15498
8201 309. http://gcc.gnu.org/PR15747
8202 310. http://gcc.gnu.org/PR16406
8203 311. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.4.4
8204 312. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.4.5
8205 313. http://gcc.gnu.org/PR24688
8206 314. http://gcc.gnu.org/PR17188
8207 315. http://gcc.gnu.org/PR20187
8208 316. http://gcc.gnu.org/PR21873
8209 317. http://gcc.gnu.org/PR21899
8210 318. http://gcc.gnu.org/PR22061
8211 319. http://gcc.gnu.org/PR22208
8212 320. http://gcc.gnu.org/PR22458
8213 321. http://gcc.gnu.org/PR22589
8214 322. http://gcc.gnu.org/PR24101
8215 323. http://gcc.gnu.org/PR10611
8216 324. http://gcc.gnu.org/PR13377
8217 325. http://gcc.gnu.org/PR16002
8218 326. http://gcc.gnu.org/PR17413
8219 327. http://gcc.gnu.org/PR17609
8220 328. http://gcc.gnu.org/PR17618
8221 329. http://gcc.gnu.org/PR18124
8222 330. http://gcc.gnu.org/PR18155
8223 331. http://gcc.gnu.org/PR18177
8224 332. http://gcc.gnu.org/PR18368
8225 333. http://gcc.gnu.org/PR18378
8226 334. http://gcc.gnu.org/PR18466
8227 335. http://gcc.gnu.org/PR18512
8228 336. http://gcc.gnu.org/PR18545
8229 337. http://gcc.gnu.org/PR18738
8230 338. http://gcc.gnu.org/PR18803
8231 339. http://gcc.gnu.org/PR19004
8232 340. http://gcc.gnu.org/PR19208
8233 341. http://gcc.gnu.org/PR19253
8234 342. http://gcc.gnu.org/PR19608
8235 343. http://gcc.gnu.org/PR19884
8236 344. http://gcc.gnu.org/PR20153
8237 345. http://gcc.gnu.org/PR20563
8238 346. http://gcc.gnu.org/PR20789
8239 347. http://gcc.gnu.org/PR21336
8240 348. http://gcc.gnu.org/PR21768
8241 349. http://gcc.gnu.org/PR21853
8242 350. http://gcc.gnu.org/PR21903
8243 351. http://gcc.gnu.org/PR21983
8244 352. http://gcc.gnu.org/PR21987
8245 353. http://gcc.gnu.org/PR22153
8246 354. http://gcc.gnu.org/PR22172
8247 355. http://gcc.gnu.org/PR21286
8248 356. http://gcc.gnu.org/PR22233
8249 357. http://gcc.gnu.org/PR22508
8250 358. http://gcc.gnu.org/PR22545
8251 359. http://gcc.gnu.org/PR23528
8252 360. http://gcc.gnu.org/PR23550
8253 361. http://gcc.gnu.org/PR23586
8254 362. http://gcc.gnu.org/PR23624
8255 363. http://gcc.gnu.org/PR23639
8256 364. http://gcc.gnu.org/PR23797
8257 365. http://gcc.gnu.org/PR23965
8258 366. http://gcc.gnu.org/PR24052
8259 367. http://gcc.gnu.org/PR24580
8260 368. http://gcc.gnu.org/PR24267
8261 369. http://gcc.gnu.org/PR17810
8262 370. http://gcc.gnu.org/PR17860
8263 371. http://gcc/gnu.org/PR21709
8264 372. http://gcc.gnu.org/PR21964
8265 373. http://gcc.gnu.org/PR22167
8266 374. http://gcc.gnu.org/PR22619
8267 375. http://gcc.gnu.org/PR23241
8268 376. http://gcc.gnu.org/PR23478
8269 377. http://gcc.gnu.org/PR24470
8270 378. http://gcc.gnu.org/PR24950
8271 379. http://gcc.gnu.org/PR14400
8272 380. http://gcc.gnu.org/PR14940
8273 381. http://gcc.gnu.org/PR20239
8274 382. http://gcc.gnu.org/PR15220
8275 383. http://gcc.gnu.org/PR19275
8276 384. http://gcc.gnu.org/PR21888
8277 385. http://gcc.gnu.org/PR15342
8278 386. http://gcc.gnu.org/PR23985
8279 387. http://gcc.gnu.org/PR16719
8280 388. http://gcc.gnu.org/PR21723
8281 389. http://gcc.gnu.org/PR21841
8282 390. http://gcc.gnu.org/PR23644
8283 391. http://gcc.gnu.org/PR24718
8284 392. http://gcc.gnu.org/PR18421
8285 393. http://gcc.gnu.org/PR20621
8286 394. http://gcc.gnu.org/PR18583
8287 395. http://gcc.gnu.org/PR20191
8288 396. http://gcc.gnu.org/PR22083
8289 397. http://gcc.gnu.org/PR23070
8290 398. http://gcc.gnu.org/PR23404
8291 399. http://gcc.gnu.org/PR23539
8292 400. http://gcc.gnu.org/PR24102
8293 401. http://gcc.gnu.org/PR24465
8294 402. http://gcc.gnu.org/PR19933
8295 403. http://gcc.gnu.org/PR21889
8296 404. http://gcc.gnu.org/PR19300
8297 405. http://gcc.gnu.org/PR20301
8298 406. http://gcc.gnu.org/PR20673
8299 407. http://gcc.gnu.org/PR18582
8300 408. http://gcc.gnu.org/PR19340
8301 409. http://gcc.gnu.org/PR21716
8302 410. http://gcc.gnu.org/PR24315
8303 411. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.4.6
8304 412. http://gcc.gnu.org/onlinedocs/
8305 413. mailto:gcc-help@gcc.gnu.org
8306 414. mailto:gcc@gcc.gnu.org
8307 415. http://gcc.gnu.org/lists.html
8308 416. http://www.fsf.org/
8309 417. http://gcc.gnu.org/about.html
8310 418. http://validator.w3.org/check/referer
8311 ======================================================================
8312 http://gcc.gnu.org/gcc-3.3/index.html
8314 GCC 3.3 Release Series
8318 The [1]GNU project and the GCC developers are pleased to announce the
8319 release of GCC 3.3.6.
8321 This release is a bug-fix release, containing fixes for regressions in
8322 GCC 3.3.5 relative to previous releases of GCC.
8324 This release is the last of the series 3.3.x.
8326 The GCC 3.3 release series includes numerous [2]new features,
8327 improvements, bug fixes, and other changes, thanks to an [3]amazing
8328 group of volunteers.
8333 May 3, 2005 ([4]changes)
8336 September 30, 2004 ([5]changes)
8339 May 31, 2004 ([6]changes)
8342 February 14, 2004 ([7]changes)
8345 October 16, 2003 ([8]changes)
8348 August 8, 2003 ([9]changes)
8351 May 14, 2003 ([10]changes)
8353 References and Acknowledgements
8355 GCC used to stand for the GNU C Compiler, but since the compiler
8356 supports several other languages aside from C, it now stands for the
8357 GNU Compiler Collection.
8359 A list of [11]successful builds is updated as new information becomes
8362 The GCC developers would like to thank the numerous people that have
8363 contributed new features, improvements, bug fixes, and other changes as
8364 well as test results to GCC. This [12]amazing group of volunteers is
8365 what makes GCC successful.
8367 For additional information about GCC please refer to the [13]GCC
8368 project web site or contact the [14]GCC development mailing list.
8370 To obtain GCC please use [15]our mirror sites, or our CVS server.
8373 For questions related to the use of GCC, please consult these web
8374 pages and the [16]GCC manuals. If that fails, the
8375 [17]gcc-help@gcc.gnu.org mailing list might help. Comments on these
8376 web pages and the development of GCC are welcome on our developer
8377 list at [18]gcc@gcc.gnu.org. All of [19]our lists have public
8380 Copyright (C) [20]Free Software Foundation, Inc. Verbatim copying and
8381 distribution of this entire article is permitted in any medium,
8382 provided this notice is preserved.
8384 These pages are [21]maintained by the GCC team. Last modified
8389 1. http://www.gnu.org/
8390 2. http://gcc.gnu.org/gcc-3.3/changes.html
8391 3. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
8392 4. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.6
8393 5. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.5
8394 6. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.4
8395 7. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.3
8396 8. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.2
8397 9. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.1
8398 10. http://gcc.gnu.org/gcc-3.3/changes.html
8399 11. http://gcc.gnu.org/gcc-3.3/buildstat.html
8400 12. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
8401 13. http://gcc.gnu.org/index.html
8402 14. mailto:gcc@gcc.gnu.org
8403 15. http://gcc.gnu.org/mirrors.html
8404 16. http://gcc.gnu.org/onlinedocs/
8405 17. mailto:gcc-help@gcc.gnu.org
8406 18. mailto:gcc@gcc.gnu.org
8407 19. http://gcc.gnu.org/lists.html
8408 20. http://www.fsf.org/
8409 21. http://gcc.gnu.org/about.html
8410 22. http://validator.w3.org/check/referer
8411 ======================================================================
8412 http://gcc.gnu.org/gcc-3.3/changes.html
8414 GCC 3.3 Release Series
8415 Changes, New Features, and Fixes
8417 The latest release in the 3.3 release series is [1]GCC 3.3.6.
8421 * The preprocessor no longer accepts multi-line string literals. They
8422 were deprecated in 3.0, 3.1, and 3.2.
8423 * The preprocessor no longer supports the -A- switch when appearing
8424 alone. -A- followed by an assertion is still supported.
8425 * Support for all the systems [2]obsoleted in GCC 3.1 has been
8426 removed from GCC 3.3. See below for a [3]list of systems which are
8427 obsoleted in this release.
8428 * Checking for null format arguments has been decoupled from the rest
8429 of the format checking mechanism. Programs which use the format
8430 attribute may regain this functionality by using the new [4]nonnull
8431 function attribute. Note that all functions for which GCC has a
8432 built-in format attribute, an appropriate built-in nonnull
8433 attribute is also applied.
8434 * The DWARF (version 1) debugging format has been deprecated and will
8435 be removed in a future version of GCC. Version 2 of the DWARF
8436 debugging format will continue to be supported for the foreseeable
8438 * The C and Objective-C compilers no longer accept the "Naming Types"
8439 extension (typedef foo = bar); it was already unavailable in C++.
8440 Code which uses it will need to be changed to use the "typeof"
8441 extension instead: typedef typeof(bar) foo. (We have removed this
8442 extension without a period of deprecation because it has caused the
8443 compiler to crash since version 3.0 and no one noticed until very
8444 recently. Thus we conclude it is not in widespread use.)
8445 * The -traditional C compiler option has been removed. It was
8446 deprecated in 3.1 and 3.2. (Traditional preprocessing remains
8447 available.) The <varargs.h> header, used for writing variadic
8448 functions in traditional C, still exists but will produce an error
8450 * GCC 3.3.1 automatically places zero-initialized variables in the
8451 .bss section on some operating systems. Versions of GNU Emacs up to
8452 (and including) 21.3 will not work correctly when using this
8453 optimization; you can use -fno-zero-initialized-in-bss to disable
8456 General Optimizer Improvements
8458 * A new scheme for accurately describing processor pipelines, the
8459 [5]DFA scheduler, has been added.
8460 * Pavel Nejedly, Charles University Prague, has contributed new file
8461 format used by the edge coverage profiler (-fprofile-arcs).
8462 The new format is robust and diagnoses common mistakes where
8463 profiles from different versions (or compilations) of the program
8464 are combined resulting in nonsensical profiles and slow code to
8465 produced with profile feedback. Additionally this format allows
8466 extra data to be gathered. Currently, overall statistics are
8467 produced helping optimizers to identify hot spots of a program
8468 globally replacing the old intra-procedural scheme and resulting in
8469 better code. Note that the gcov tool from older GCC versions will
8470 not be able to parse the profiles generated by GCC 3.3 and vice
8472 * Jan Hubicka, SuSE Labs, has contributed a new superblock formation
8473 pass enabled using -ftracer. This pass simplifies the control flow
8474 of functions allowing other optimizations to do better job.
8475 He also contributed the function reordering pass
8476 (-freorder-functions) to optimize function placement using profile
8479 New Languages and Language specific improvements
8483 * The preprocessor now accepts directives within macro arguments. It
8484 processes them just as if they had not been within macro arguments.
8485 * The separate ISO and traditional preprocessors have been completely
8486 removed. The front end handles either type of preprocessed output
8488 * In C99 mode preprocessor arithmetic is done in the precision of the
8489 target's intmax_t, as required by that standard.
8490 * The preprocessor can now copy comments inside macros to the output
8491 file when the macro is expanded. This feature, enabled using the
8492 -CC option, is intended for use by applications which place
8493 metadata or directives inside comments, such as lint.
8494 * The method of constructing the list of directories to be searched
8495 for header files has been revised. If a directory named by a -I
8496 option is a standard system include directory, the option is
8497 ignored to ensure that the default search order for system
8498 directories and the special treatment of system header files are
8500 * A few more [6]ISO C99 features now work correctly.
8501 * A new function attribute, nonnull, has been added which allows
8502 pointer arguments to functions to be specified as requiring a
8503 non-null value. The compiler currently uses this information to
8504 issue a warning when it detects a null value passed in such an
8506 * A new type attribute, may_alias, has been added. Accesses to
8507 objects with types with this attribute are not subjected to
8508 type-based alias analysis, but are instead assumed to be able to
8509 alias any other type of objects, just like the char type.
8513 * Type based alias analysis has been implemented for C++ aggregate
8518 * Generate an error if Objective-C objects are passed by value in
8519 function and method calls.
8520 * When -Wselector is used, check the whole list of selectors at the
8521 end of compilation, and emit a warning if a @selector() is not
8523 * Define __NEXT_RUNTIME__ when compiling for the NeXT runtime.
8524 * No longer need to include objc/objc-class.h to compile self calls
8525 in class methods (NeXT runtime only).
8526 * New -Wundeclared-selector option.
8527 * Removed selector bloating which was causing object files to be 10%
8528 bigger on average (GNU runtime only).
8529 * Using at run time @protocol() objects has been fixed in certain
8530 situations (GNU runtime only).
8531 * Type checking has been fixed and improved in many situations
8532 involving protocols.
8536 * The java.sql and javax.sql packages now implement the JDBC 3.0 (JDK
8538 * The JDK 1.4 assert facility has been implemented.
8539 * The bytecode interpreter is now direct threaded and thus faster.
8543 * Fortran improvements are listed in [7]the Fortran documentation.
8547 * Ada tasking now works with glibc 2.3.x threading libraries.
8549 New Targets and Target Specific Improvements
8551 * The following changes have been made to the HP-PA port:
8552 + The port now defaults to scheduling for the PA8000 series of
8554 + Scheduling support for the PA7300 processor has been added.
8555 + The 32-bit port now supports weak symbols under HP-UX 11.
8556 + The handling of initializers and finalizers has been improved
8557 under HP-UX 11. The 64-bit port no longer uses collect2.
8558 + Dwarf2 EH support has been added to the 32-bit GNU/Linux port.
8559 + ABI fixes to correct the passing of small structures by value.
8560 * The SPARC, HP-PA, SH4, and x86/pentium ports have been converted to
8561 use the DFA processor pipeline description.
8562 * The following NetBSD configurations for the SuperH processor family
8564 + SH3, big-endian, sh-*-netbsdelf*
8565 + SH3, little-endian, shle-*-netbsdelf*
8566 + SH5, SHmedia, big-endian, 32-bit default, sh5-*-netbsd*
8567 + SH5, SHmedia, little-endian, 32-bit default, sh5le-*-netbsd*
8568 + SH5, SHmedia, big-endian, 64-bit default, sh64-*-netbsd*
8569 + SH5, SHmedia, little-endian, 64-bit default, sh64le-*-netbsd*
8570 * The following changes have been made to the IA-32/x86-64 port:
8571 + SSE2 and 3dNOW! intrinsics are now supported.
8572 + Support for thread local storage has been added to the IA-32
8574 + The x86-64 port has been significantly improved.
8575 * The following changes have been made to the MIPS port:
8576 + All configurations now accept the -mabi switch. Note that you
8577 will need appropriate multilibs for this option to work
8579 + ELF configurations will always pass an ABI flag to the
8580 assembler, except when the MIPS EABI is selected.
8581 + -mabi=64 no longer selects MIPS IV code.
8582 + The -mcpu option, which was deprecated in 3.1 and 3.2, has
8583 been removed from this release.
8584 + -march now changes the core ISA level. In previous releases,
8585 it would change the use of processor-specific extensions, but
8586 would leave the core ISA unchanged. For example, mips64-elf
8587 -march=r8000 will now generate MIPS IV code.
8588 + Under most configurations, -mipsN now acts as a synonym for
8590 + There are some new preprocessor macros to describe the -march
8591 and -mtune settings. See the documentation of those options
8593 + Support for the NEC VR-Series processors has been added. This
8594 includes the 54xx, 5500, and 41xx series.
8595 + Support for the Sandcraft sr71k processor has been added.
8596 * The following changes have been made to the S/390 port:
8597 + Support to build the Java runtime libraries has been added.
8598 Java is now enabled by default on s390-*-linux* and
8599 s390x-*-linux* targets.
8600 + Multilib support for the s390x-*-linux* target has been added;
8601 this allows to build 31-bit binaries using the -m31 option.
8602 + Support for thread local storage has been added.
8603 + Inline assembler code may now use the 'Q' constraint to
8604 specify memory operands without index register.
8605 + Various platform-specific performance improvements have been
8606 implemented; in particular, the compiler now uses the BRANCH
8607 ON COUNT family of instructions and makes more frequent use of
8608 the TEST UNDER MASK family of instructions.
8609 * The following changes have been made to the PowerPC port:
8610 + Support for IBM Power4 processor added.
8611 + Support for Motorola e500 SPE added.
8612 + Support for AIX 5.2 added.
8613 + Function and Data sections now supported on AIX.
8614 + Sibcall optimizations added.
8615 * The support for H8 Tiny is added to the H8/300 port with -mn.
8619 Support for a number of older systems has been declared obsolete in GCC
8620 3.3. Unless there is activity to revive them, the next release of GCC
8621 will have their sources permanently removed.
8623 All configurations of the following processor architectures have been
8625 * Matsushita MN10200, mn10200-*-*
8626 * Motorola 88000, m88k-*-*
8627 * IBM ROMP, romp-*-*
8629 Also, some individual systems have been obsoleted:
8631 + Interix, alpha*-*-interix*
8632 + Linux libc1, alpha*-*-linux*libc1*
8633 + Linux ECOFF, alpha*-*-linux*ecoff*
8635 + Generic a.out, arm*-*-aout*
8636 + Conix, arm*-*-conix*
8637 + "Old ABI," arm*-*-oabi
8638 + StrongARM/COFF, strongarm-*-coff*
8640 + Generic OSF, hppa1.0-*-osf*
8641 + Generic BSD, hppa1.0-*-bsd*
8642 + HP/UX versions 7, 8, and 9, hppa1.[01]-*-hpux[789]*
8643 + HiUX, hppa*-*-hiux*
8644 + Mach Lites, hppa*-*-lites*
8646 + Windows NT 3.x, i?86-*-win32
8648 + HP systems, m68000-hp-bsd* and m68k-hp-bsd*
8649 + Sun systems, m68000-sun-sunos*, m68k-sun-sunos*, and
8651 + AT&T systems, m68000-att-sysv*
8652 + Atari systems, m68k-atari-sysv*
8653 + Motorola systems, m68k-motorola-sysv*
8654 + NCR systems, m68k-ncr-sysv*
8655 + Plexus systems, m68k-plexus-sysv*
8656 + Commodore systems, m68k-cbm-sysv*
8657 + Citicorp TTI, m68k-tti-*
8658 + Unos, m68k-crds-unos*
8659 + Concurrent RTU, m68k-ccur-rtu*
8660 + Linux a.out, m68k-*-linux*aout*
8661 + Linux libc1, m68k-*-linux*libc1*
8662 + pSOS, m68k-*-psos*
8664 + Generic ECOFF, mips*-*-ecoff*
8665 + SINIX, mips-sni-sysv4
8666 + Orion RTEMS, mips64orion-*-rtems*
8667 * National Semiconductor 32000
8668 + OpenBSD, ns32k-*-openbsd*
8669 * POWER (aka RS/6000) and PowerPC
8670 + AIX versions 1, 2, and 3, rs6000-ibm-aix[123]*
8671 + Bull BOSX, rs6000-bull-bosx
8672 + Generic Mach, rs6000-*-mach*
8673 + Generic SysV, powerpc*-*-sysv*
8674 + Linux libc1, powerpc*-*-linux*libc1*
8676 + Generic a.out, sparc-*-aout*, sparclet-*-aout*,
8677 sparclite-*-aout*, and sparc86x-*-aout*
8678 + NetBSD a.out, sparc-*-netbsd*aout*
8679 + Generic BSD, sparc-*-bsd*
8680 + ChorusOS, sparc-*-chorusos*
8681 + Linux a.out, sparc-*-linux*aout*
8682 + Linux libc1, sparc-*-linux*libc1*
8683 + LynxOS, sparc-*-lynxos*
8684 + Solaris on HAL hardware, sparc-hal-solaris2*
8685 + SunOS versions 3 and 4, sparc-*-sunos[34]*
8687 + RTEMS, v850-*-rtems*
8691 Documentation improvements
8693 Other significant improvements
8695 * Almost all front-end dependencies in the compiler have been
8696 separated out into a set of language hooks. This should make adding
8697 a new front end clearer and easier.
8698 * One effect of removing the separate preprocessor is a small
8699 increase in the robustness of the compiler in general, and the
8700 maintainability of target descriptions. Previously target-specific
8701 built-in macros and others, such as __FAST_MATH__, had to be
8702 handled with so-called specs that were hard to maintain. Often they
8703 would fail to behave properly when conflicting options were
8704 supplied on the command line, and define macros in the user's
8705 namespace even when strict ISO compliance was requested.
8706 Integrating the preprocessor has cleanly solved these issues.
8707 * The Makefile suite now supports redirection of make install by
8708 means of the variable DESTDIR.
8709 __________________________________________________________________
8713 Detailed release notes for the GCC 3.3 release follow.
8719 * [8]10140 cross compiler build failures: missing __mempcpy (DUP:
8722 Internal compiler errors (multi-platform)
8724 * [11]3581 large string causes segmentation fault in cc1
8725 * [12]4382 __builtin_{set,long}jmp with -O3 can crash the compiler
8726 * [13]5533 (c++) ICE when processing std::accumulate(begin, end,
8728 * [14]6387 -fpic -gdwarf-2 -g1 combination gives ICE in dwarf2out
8729 * [15]6412 (c++) ICE in retrieve_specialization
8730 * [16]6620 (c++) partial template specialization causes an ICE
8731 (segmentation fault)
8732 * [17]6663 (c++) ICE with attribute aligned
8733 * [18]7068 ICE with incomplete types
8734 * [19]7083 (c++) ICE using -gstabs with dodgy class derivation
8735 * [20]7647 (c++) ICE when data member has the name of the enclosing
8737 * [21]7675 ICE in fixup_var_refs_1
8738 * [22]7718 'complex' template instantiation causes ICE
8739 * [23]8116 (c++) ICE in member template function
8740 * [24]8358 (ada) Ada compiler accesses freed memory, crashes
8741 * [25]8511 (c++) ICE: (hopefully) reproducible cc1plus segmentation
8743 * [26]8564 (c++) ICE in find_function_data, in function.c
8744 * [27]8660 (c++) template overloading ICE in tsubst_expr, in cp/pt.c
8745 * [28]8766 (c++) ICE after failed initialization of static template
8747 * [29]8803 ICE in instantiate_virtual_regs_1, in function.c
8748 * [30]8846 (c++) ICE after diagnostic if fr_FR@euro locale is set
8749 * [31]8906 (c++) ICE (Segmentation fault) when parsing nested-class
8751 * [32]9216 (c++) ICE on missing template parameter
8752 * [33]9261 (c++) ICE in arg_assoc, in cp/decl2.c
8753 * [34]9263 (fortran) ICE caused by invalid PARAMETER in implied DO
8755 * [35]9429 (c++) ICE in template instantiation with a pointered new
8757 * [36]9516 Internal error when using a big array
8758 * [37]9600 (c++) ICE with typedefs in template class
8759 * [38]9629 (c++) virtual inheritance segfault
8760 * [39]9672 (c++) ICE: Error reporting routines re-entered
8761 * [40]9749 (c++) ICE in write_expression on invalid function
8763 * [41]9794 (fortran) ICE: floating point exception during constant
8765 * [42]9829 (c++) Missing colon in nested namespace usage causes ICE
8766 * [43]9916 (c++) ICE with noreturn function in ?: statement
8767 * [44]9936 ICE with local function and variable-length 2d array
8768 * [45]10262 (c++) cc1plus crashes with large generated code
8769 * [46]10278 (c++) ICE in parser for invalid code
8770 * [47]10446 (c++) ICE on definition of nonexistent member function of
8771 nested class in a class template
8772 * [48]10451 (c++) ICE in grokdeclarator on spurious mutable
8774 * [49]10506 (c++) ICE in build_new at cp/init.c with
8775 -fkeep-inline-functions and multiple inheritance
8776 * [50]10549 (c++) ICE in store_bit_field on bitfields that exceed the
8777 precision of the declared type
8781 * [51]2001 Inordinately long compile times in reload CSE regs
8782 * [52]2391 Exponential compilation time explosion in combine
8783 * [53]2960 Duplicate loop conditions even with -Os
8784 * [54]4046 redundant conditional branch
8785 * [55]6405 Loop-unrolling related performance regressions
8786 * [56]6798 very long compile time with large case-statement
8787 * [57]6871 const objects shouldn't be moved to .bss
8788 * [58]6909 problem w/ -Os on modified loop-2c.c test case
8789 * [59]7189 gcc -O2 -Wall does not print ``control reaches end of
8790 non-void function'' warning
8791 * [60]7642 optimization problem with signbit()
8792 * [61]8634 incorrect code for inlining of memcpy under -O2
8793 * [62]8750 Cygwin prolog generation erroneously emitting __alloca as
8794 regular function call
8798 * [63]2161 long if-else cascade overflows parser stack
8799 * [64]4319 short accepted on typedef'd char
8800 * [65]8602 incorrect line numbers in warning messages when using
8802 * [66]9177 -fdump-translation-unit: C front end deletes function_decl
8803 AST nodes and breaks debugging dumps
8804 * [67]9853 miscompilation of non-constant structure initializer
8806 c++ compiler and library
8808 * [68]45 legal template specialization code is rejected (DUP:
8810 * [70]764 lookup failure: friend operator and dereferencing a pointer
8811 and templates (DUP: [71]5116)
8812 * [72]2862 gcc accepts invalid explicit instantiation syntax (DUP:
8814 * [73]3663 G++ doesn't check access control during template
8816 * [74]3797 gcc fails to emit explicit specialization of a template
8818 * [75]3948 Two destructors are called when no copy destructor is
8819 defined (ABI change)
8820 * [76]4137 Conversion operator within template is not accepted
8821 * [77]4361 bogus ambiguity taking the address of a member template
8822 * [78]4802 g++ accepts illegal template code (access to private
8823 member; DUP: [79]5837)
8824 * [80]4803 inline function is used but never defined, and g++ does
8826 * [81]5094 Partial specialization cannot be friend?
8827 * [82]5730 complex<double>::norm() -- huge slowdown from egcs-2.91.66
8828 * [83]6713 Regression wrt 3.0.4: g++ -O2 leads to seg fault at run
8830 * [84]7015 certain __asm__ constructs rejected
8831 * [85]7086 compile time regression (quadratic behavior in
8833 * [86]7099 G++ doesn't set the noreturn attribute on std::exit and
8835 * [87]7247 copy constructor missing when inlining enabled (invalid
8837 * [88]7441 string array initialization compilation time regression
8838 from seconds to minutes
8839 * [89]7768 __PRETTY_FUNCTION__ for template destructor is wrong
8840 * [90]7804 bad printing of floating point constant in warning message
8841 * [91]8099 Friend classes and template specializations
8842 * [92]8117 member function pointers and multiple inheritance
8843 * [93]8205 using declaration and multiple inheritance
8844 * [94]8645 unnecessary non-zero checks in stl_tree.h
8845 * [95]8724 explicit destructor call for incomplete class allowed
8846 * [96]8805 compile time regression with many member variables
8847 * [97]8691 -O3 and -fno-implicit-templates are incompatible
8848 * [98]8700 unhelpful error message for binding temp to reference
8849 * [99]8724 explicit destructor call for incomplete class allowed
8850 * [100]8949 numeric_limits<>::denorm_min() and is_iec559 problems
8851 * [101]9016 Failure to consistently constant fold "constant" C++
8853 * [102]9053 g++ confused about ambiguity of overloaded function
8855 * [103]9152 undefined virtual thunks
8856 * [104]9182 basic_filebuf<> does not report errors in codecvt<>::out
8857 * [105]9297 data corruption due to codegen bug (when copying.)
8858 * [106]9318 i/ostream::operator>>/<<(streambuf*) broken
8859 * [107]9320 Incorrect usage of traits_type::int_type in stdio_filebuf
8860 * [108]9400 bogus -Wshadow warning: shadowed declaration of this in
8862 * [109]9424 i/ostream::operator>>/<<(streambuf*) drops characters
8863 * [110]9425 filebuf::pbackfail broken (DUP: [111]9439)
8864 * [112]9474 GCC freezes in compiling a weird code mixing <iostream>
8866 * [113]9548 Incorrect results from setf(ios::fixed) and precision(-1)
8868 * [115]9555 ostream inserters fail to set badbit on exception
8869 * [116]9561 ostream inserters rethrow exception of wrong type
8870 * [117]9563 ostream::sentry returns true after a failed preparation
8871 * [118]9582 one-definition rule violation in std::allocator
8872 * [119]9622 __PRETTY_FUNCTION__ incorrect in template destructors
8873 * [120]9683 bug in initialization chains for static const variables
8874 from template classes
8875 * [121]9791 -Woverloaded-virtual reports hiding of destructor
8876 * [122]9817 collate::compare doesn't handle nul characters
8877 * [123]9825 filebuf::sputbackc breaks sbumpc
8878 * [124]9826 operator>>(basic_istream, basic_string) fails to compile
8880 * [125]9924 Multiple using statements for builtin functions not
8882 * [126]9946 destructor is not called for temporary object
8883 * [127]9964 filebuf::close() sometimes fails to close file
8884 * [128]9988 filebuf::overflow writes EOF to file
8885 * [129]10033 optimization breaks polymorphic references w/ typeid
8887 * [130]10097 filebuf::underflow drops characters
8888 * [131]10132 filebuf destructor can throw exceptions
8889 * [132]10180 gcc fails to warn about non-inlined function
8890 * [133]10199 method parametrized by template does not work everywhere
8891 * [134]10300 use of array-new (nothrow) in segfaults on NULL return
8892 * [135]10427 Stack corruption with variable-length automatic arrays
8893 and virtual destructors
8894 * [136]10503 Compilation never stops in fixed_type_or_null
8898 * [137]5956 selectors aren't matched properly when added to the
8901 Fortran compiler and library
8903 * [138]1832 list directed i/o overflow hangs, -fbounds-check doesn't
8905 * [139]3924 g77 generates code that is rejected by GAS if COFF debug
8907 * [140]5634 doc: explain that configure --prefix=~/... does not work
8908 * [141]6367 multiple repeat counts confuse namelist read into array
8909 * [142]6491 Logical operations error on logicals when using
8911 * [143]6742 Generation of C++ Prototype for FORTRAN and extern "C"
8912 * [144]7113 Failure of g77.f-torture/execute/f90-intrinsic-bit.f -Os
8914 * [145]7236 OPEN(...,RECL=nnn,...) without ACCESS='DIRECT' should
8915 assume a direct access file
8916 * [146]7278 g77 "bug"; the executable misbehaves (with -O2
8918 * [147]7384 DATE_AND_TIME milliseconds field inactive on Windows
8919 * [148]7388 Incorrect output with 0-based array of characters
8920 * [149]8587 Double complex zero ** double precision number -> NaN
8922 * [150]9038 -ffixed-line-length-none -x f77-cpp-input gives: Warning:
8923 unknown register name line-length-none
8924 * [151]10197 Direct access files not unformatted by default
8926 Java compiler and library
8928 * [152]6005 gcj fails to build rhug on alpha
8929 * [153]6389 System.getProperty("") should always throw an
8930 IllegalArgumentException
8931 * [154]6576 java.util.ResourceBundle.getResource ignores locale
8932 * [155]6652 new java.io.File("").getCanonicalFile() throws exception
8933 * [156]7060 getMethod() doesn't search super interface
8934 * [157]7073 bytecode interpreter gives wrong answer for interface
8936 * [158]7180 possible bug in
8937 javax.naming.spi.NamingManager.getPlusPath()
8938 * [159]7416 java.security startup refs "GNU libgcj.security"
8939 * [160]7570 Runtime.exec with null envp: child doesn't inherit parent
8940 env (DUP: [161]7578)
8941 * [162]7611 Internal error while compiling libjava with -O
8942 * [163]7709 NullPointerException in _Jv_ResolvePoolEntry
8943 * [164]7766 ZipInputStream.available returns 0 immediately after
8945 * [165]7785 Calendar.getTimeInMillis/setTimeInMillis should be public
8946 * [166]7786 TimeZone.getDSTSavings() from JDK1.4 not implemented
8947 * [167]8142 '$' in class names vs. dlopen 'dynamic string tokens'
8948 * [168]8234 ZipInputStream chokes when InputStream.read() returns
8950 * [169]8415 reflection bug: exception info for Method
8951 * [170]8481 java.Random.nextInt(int) may return negative
8952 * [171]8593 Error reading GZIPped files with BufferedReader
8953 * [172]8759 java.beans.Introspector has no flushCaches() or
8954 flushFromCaches() methods
8955 * [173]8997 spin() calls Thread.sleep
8956 * [174]9253 on win32, java.io.File.listFiles("C:\\") returns pwd
8957 instead of the root content of C:
8958 * [175]9254 java::lang::Object::wait(), threads-win32.cc returns
8960 * [176]9271 Severe bias in java.security.SecureRandom
8962 Ada compiler and library
8964 * [177]6767 make gnatlib-shared fails on -laddr2line
8965 * [178]9911 gnatmake fails to link when GCC configured with
8966 --with-sjlj-exceptions=yes
8967 * [179]10020 Can't bootstrap gcc on AIX with Ada enabled
8968 * [180]10546 Ada tasking not working on Red Hat 9
8972 * [181]7029 preprocessor should ignore #warning with -M
8976 * [182]2903 [arm] Optimization bug with long long arithmetic
8977 * [183]7873 arm-linux-gcc fails when assigning address to a bit field
8981 * [184]7680 float functions undefined in math.h/cmath with #define
8984 HP-UX or HP-PA-specific
8986 * [185]8705 [HP-PA] ICE in emit_move_insn_1, in expr.c
8987 * [186]9986 [HP-UX] Incorrect transformation of fputs_unlocked to
8989 * [187]10056 [HP-PA] ICE at -O2 when building c++ code from doxygen
8993 * [188]6744 Bad assembler code generated: reference to pseudo
8995 * [189]7361 Internal compiler error in reload_cse_simplify_operands,
9000 * [190]9496 [mips-linux] bug in optimizer?
9004 * [191]7067 -Os with -mcpu=powerpc optimizes for speed (?) instead of
9006 * [192]8480 reload ICEs for LAPACK code on powerpc64-linux
9007 * [193]8784 [AIX] Internal compiler error in simplify_gen_subreg
9008 * [194]10315 [powerpc] ICE: in extract_insn, in recog.c
9012 * [195]10267 (documentation) Wrong build instructions for
9015 x86-specific (Intel/AMD)
9017 * [196]7916 ICE in instantiate_virtual_register_1
9018 * [197]7926 (c++) i486 instructions in header files make c++ programs
9020 * [198]8555 ICE in gen_split_1231
9021 * [199]8994 ICE with -O -march=pentium4
9022 * [200]9426 ICE with -fssa -funroll-loops -fprofile-arcs
9023 * [201]9806 ICE in inline assembly with -fPIC flag
9024 * [202]10077 gcc -msse2 generates movd to move dwords between xmm
9026 * [203]10233 64-bit comparison only comparing bottom 32-bits
9027 * [204]10286 type-punning doesn't work with __m64 and -O
9028 * [205]10308 [x86] ICE with -O -fgcse or -O2
9029 __________________________________________________________________
9035 This section lists the problem reports (PRs) from GCC's bug tracking
9036 system that are known to be fixed in the 3.3.1 release. This list might
9037 not be complete (that is, it is possible that some PRs that have been
9038 fixed are not listed here).
9042 * [206]11272 [Solaris] make bootstrap fails while building libstdc++
9044 Internal compiler errors (multi-platform)
9046 * [207]5754 ICE on invalid nested template class
9047 * [208]6597 ICE in set_mem_alias_set compiling Qt with -O2 on ia64
9048 and --enable-checking
9049 * [209]6949 (c++) ICE in tsubst_decl, in cp/pt.c
9050 * [210]7053 (c++) ICE when declaring a function already defined as a
9051 friend method of a template class
9052 * [211]8164 (c++) ICE when using different const expressions as
9054 * [212]8384 (c++) ICE in is_base_type, in dwarf2out.c
9055 * [213]9559 (c++) ICE with invalid initialization of a static const
9056 * [214]9649 (c++) ICE in finish_member_declaration, in cp/semantics.c
9057 when redeclaring a static member variable
9058 * [215]9864 (fortran) ICE in add_abstract_origin_attribute, in
9059 dwarfout.c with -g -O -finline-functions
9060 * [216]10432 (c++) ICE in poplevel, in cp/decl.c
9061 * [217]10475 ICE in subreg_highpart_offset for code with long long
9062 * [218]10635 (c++) ICE when dereferencing an incomplete type casted
9064 * [219]10661 (c++) ICE in instantiate_decl, in cp/pt.c while
9065 instantiating static member variables
9066 * [220]10700 ICE in copy_to_mode_reg on 64-bit targets
9067 * [221]10712 (c++) ICE in constructor_name_full, in cp/decl2.c
9068 * [222]10796 (c++) ICE when defining an enum with two values: -1 and
9070 * [223]10890 ICE in merge_assigned_reloads building Linux 2.4.2x
9072 * [224]10939 (c++) ICE with template code
9073 * [225]10956 (c++) ICE when specializing a template member function
9074 of a template class, in tsubst, in cp/pt.c
9075 * [226]11041 (c++) ICE: const myclass &x = *x; (when operator*()
9077 * [227]11059 (c++) ICE with empty union
9078 * [228]11083 (c++) ICE in commit_one_edge_insertion, in cfgrtl.c with
9079 -O2 -fnon-call-exceptions
9080 * [229]11105 (c++) ICE in mangle_conv_op_name_for_type
9081 * [230]11149 (c++) ICE on error when instantiation with call function
9083 * [231]11228 (c++) ICE on new-expression using array operator new and
9084 default-initialization
9085 * [232]11282 (c++) Infinite memory usage after syntax error
9086 * [233]11301 (fortran) ICE with -fno-globals
9087 * [234]11308 (c++) ICE when using an enum type name as if it were a
9089 * [235]11473 (c++) ICE with -gstabs when empty struct inherits from
9091 * [236]11503 (c++) ICE when instantiating template with ADDR_EXPR
9092 * [237]11513 (c++) ICE in push_template_decl_real, in cp/pt.c:
9093 template member functions
9097 * [238]11198 -O2 -frename-registers generates wrong code (aliasing
9099 * [239]11304 Wrong code production with -fomit-frame-pointer
9100 * [240]11381 volatile memory access optimized away
9101 * [241]11536 [strength-reduce] -O2 optimization produces wrong code
9102 * [242]11557 constant folding bug generates wrong code
9106 * [243]5897 No warning for statement after return
9107 * [244]11279 DWARF-2 output mishandles large enums
9111 * [245]11022 no warning for non-compatible macro redefinition
9113 C++ compiler and library
9115 * [246]2330 static_cast<>() to a private base is allowed
9116 * [247]5388 Incorrect message "operands to ?: have different types"
9117 * [248]5390 Libiberty fails to demangle multi-digit template
9119 * [249]7877 Incorrect parameter passing to specializations of member
9121 * [250]9393 Anonymous namespaces and compiling the same file twice
9122 * [251]10032 -pedantic converts some errors to warnings
9123 * [252]10468 const typeof(x) is non-const, but only in templates
9124 * [253]10527 confused error message with "new int()" parameter
9126 * [254]10679 parameter MIN_INLINE_INSNS is not honored
9127 * [255]10682 gcc chokes on a typedef for an enum inside a class
9129 * [256]10689 pow(std::complex(0),1/3) returns (nan, nan) instead of
9131 * [257]10845 template member function (with nested template as
9132 parameter) cannot be called anymore if another unrelated template
9133 member function is defined
9134 * [258]10849 Cannot define an out-of-class specialization of a
9135 private nested template class
9136 * [259]10888 Suppress -Winline warnings for system headers
9137 * [260]10929 -Winline warns about functions for which no definition
9139 * [261]10931 valid conversion static_cast<const unsigned
9140 int&>(lvalue-of-type-int) is rejected
9141 * [262]10940 Bad code with explicit specialization
9142 * [263]10968 If member function implicitly instantiated, explicit
9143 instantiation of class fails to instantiate it
9144 * [264]10990 Cannot convert with dynamic_cast<> to a private base
9145 class from within a member function
9146 * [265]11039 Bad interaction between implicit typename deprecation
9148 * [266]11062 (libstdc++) avoid __attribute__ ((unused)); say
9149 "__unused__" instead
9150 * [267]11095 C++ iostream manipulator causes segfault when called
9151 with negative argument
9152 * [268]11098 g++ doesn't emit complete debugging information for
9153 local variables in destructors
9154 * [269]11137 GNU/Linux shared library constructors not called unless
9155 there's one global object
9156 * [270]11154 spurious ambiguity report for template class
9158 * [271]11329 Compiler cannot find user defined implicit typecast
9159 * [272]11332 Spurious error with casts in ?: expression
9160 * [273]11431 static_cast behavior with subclasses when default
9161 constructor available
9162 * [274]11528 money_get facet does not accept "$.00" as valid
9163 * [275]11546 Type lookup problems in out-of-line definition of a
9164 class doubly nested from a template class
9165 * [276]11567 C++ code containing templated member function with same
9166 name as pure virtual member function results in linking failure
9167 * [277]11645 Failure to deal with using and private inheritance
9169 Java compiler and library
9171 * [278]5179 Qualified static field access doesn't initialize its
9173 * [279]8204 gcj -O2 to native reorders certain instructions
9175 * [280]10838 java.io.ObjectInputStream syntax error
9176 * [281]10886 The RMI registry that comes with GCJ does not work
9178 * [282]11349 JNDI URL context factories not located correctly
9180 x86-specific (Intel/AMD)
9182 * [283]4823 ICE on inline assembly code
9183 * [284]8878 miscompilation with -O and SSE
9184 * [285]9815 (c++ library) atomicity.h - fails to compile with -O3
9186 * [286]10402 (inline assembly) [x86] ICE in merge_assigned_reloads,
9188 * [287]10504 ICE with SSE2 code and -O3 -mcpu=pentium4 -msse2
9189 * [288]10673 ICE for x86-64 on freebsd libc vfprintf.c source
9190 * [289]11044 [x86] out of range loop instructions for FP code on K6
9191 * [290]11089 ICE: instantiate_virtual_regs_lossage while using SSE
9193 * [291]11420 [x86_64] gcc generates invalid asm code when "-O -fPIC"
9196 SPARC- or Solaris- specific
9198 * [292]9362 solaris 'as' dies when fed .s and "-gstabs"
9199 * [293]10142 [SPARC64] gcc produces wrong code when passing
9201 * [294]10663 New configure check aborts with Sun tools.
9202 * [295]10835 combinatorial explosion in scheduler on HyperSPARC
9203 * [296]10876 ICE in calculate_giv_inc when building KDE
9204 * [297]10955 wrong code at -O3 for structure argument in context of
9206 * [298]11018 -mcpu=ultrasparc busts tar-1.13.25
9207 * [299]11556 [sparc64] ICE in gen_reg_rtx() while compiling 2.6.x
9212 * [300]10907 gcc violates the ia64 ABI (GP must be preserved)
9213 * [301]11320 scheduler bug (in machine depended reorganization pass)
9214 * [302]11599 bug with conditional and __builtin_prefetch
9218 * [303]9745 [powerpc] gcc mis-compiles libmcrypt (alias problem
9220 * [304]10871 error in rs6000_stack_info save_size computation
9221 * [305]11440 gcc mis-compiles c++ code (libkhtml) with -O2, -fno-gcse
9226 * [306]7594 [m68k] ICE on legal code associated with simplify-rtx
9227 * [307]10557 [m68k] ICE in subreg_offset_representable_p
9228 * [308]11054 [m68k] ICE in reg_overlap_mentioned_p
9232 * [309]10834 [arm] GCC 3.3 still generates incorrect instructions for
9233 functions with __attribute__ ((interrupt ("IRQ")))
9234 * [310]10842 [arm] Clobbered link register is copied to pc under
9235 certain circumstances
9236 * [311]11052 [arm] noce_process_if_block() can lose REG_INC notes
9237 * [312]11183 [arm] ICE in change_address_1 (3.3) / subreg_hard_regno
9242 * [313]11084 ICE in propagate_one_insn, in flow.c
9246 * [314]10331 can't compile c++ part of gcc cross compiler for sh-elf
9247 * [315]10413 [SH] ICE in reload_cse_simplify_operands, in reload1.c
9248 * [316]11096 i686-linux to sh-linux cross compiler fails to compile
9251 GNU/Linux (or Hurd?) specific
9253 * [317]2873 Bogus fixinclude of stdio.h from glibc 2.2.3
9257 * [318]3163 configure bug: gcc/aclocal.m4 mmap test fails on UnixWare
9260 Cygwin (or mingw) specific
9262 * [319]5287 ICE with dllimport attribute
9263 * [320]10148 [MingW/CygWin] Compiler dumps core
9267 * [321]8787 GCC fails to emit .intel_syntax when invoked with
9268 -masm=intel on DJGPP
9270 Darwin (and MacOS X) specific
9272 * [322]10900 trampolines crash
9276 * [323]1607 (c++) Format attributes on methods undocumented
9277 * [324]4252 Invalid option `-fdump-translation-unit'
9278 * [325]4490 Clarify restrictions on -m96bit-long-double,
9279 -m128bit-long-double
9280 * [326]10355 document an issue with regparm attribute on some systems
9282 * [327]10726 (fortran) Documentation for function "IDate Intrinsic
9284 * [328]10805 document bug in old version of Sun assembler
9285 * [329]10815 warn against GNU binutils on AIX
9286 * [330]10877 document need for newer binutils on i?86-*-linux-gnu
9287 * [331]11280 Manual incorrect with respect to -freorder-blocks
9288 * [332]11466 Document -mlittle-endian and its restrictions for the
9291 Testsuite bugs (compiler itself is not affected)
9293 * [333]10737 newer bison causes g++.dg/parse/crash2.C to incorrectly
9295 * [334]10810 gcc-3.3 fails make check: buffer overrun in
9297 __________________________________________________________________
9303 This section lists the problem reports (PRs) from [335]GCC's bug
9304 tracking system that are known to be fixed in the 3.3.2 release. This
9305 list might not be complete (that is, it is possible that some PRs that
9306 have been fixed are not listed here).
9308 Bootstrap failures and problems
9310 * [336]8336 [SCO5] bootstrap config still tries to use COFF options
9311 * [337]9330 [alpha-osf] Bootstrap failure on Compaq Tru64 with
9312 --enable-threads=posix
9313 * [338]9631 [hppa64-linux] gcc-3.3 fails to bootstrap
9314 * [339]9877 fixincludes makes a bad sys/byteorder.h on svr5 (UnixWare
9316 * [340]11687 xstormy16-elf build fails in libf2c
9317 * [341]12263 [SGI IRIX] bootstrap fails during compile of
9318 libf2c/libI77/backspace.c
9319 * [342]12490 buffer overflow in scan-decls.c (during Solaris 9
9320 fix-header processing)
9322 Internal compiler errors (multi-platform)
9324 * [343]7277 Casting integers to vector types causes ICE
9325 * [344]7939 (c++) ICE on invalid function template specialization
9326 * [345]11063 (c++) ICE on parsing initialization list of const array
9328 * [346]11207 ICE with negative index in array element designator
9329 * [347]11522 (fortran) g77 dwarf-2 ICE in
9330 add_abstract_origin_attribute
9331 * [348]11595 (c++) ICE on duplicate label definition
9332 * [349]11646 (c++) ICE in commit_one_edge_insertion with
9333 -fnon-call-exceptions -fgcse -O
9334 * [350]11665 ICE in struct initializer when taking address
9335 * [351]11852 (c++) ICE with bad struct initializer.
9336 * [352]11878 (c++) ICE in cp_expr_size
9337 * [353]11883 ICE with any -O on mercury-generated C code
9338 * [354]11991 (c++) ICE in cxx_incomplete_type_diagnostic, in
9339 cp/typeck2.c when applying typeid operator to template template
9341 * [355]12146 ICE in lookup_template_function, in cp/pt.c
9342 * [356]12215 ICE in make_label_edge with -fnon-call-exceptions
9344 * [357]12369 (c++) ICE with templates and friends
9345 * [358]12446 ICE in emit_move_insn on complicated array reference
9346 * [359]12510 ICE in final_scan_insn
9347 * [360]12544 ICE with large parameters used in nested functions
9349 C and optimization bugs
9351 * [361]9862 spurious warnings with -W -finline-functions
9352 * [362]10962 lookup_field is a linear search on a linked list (can be
9353 slow if large struct)
9354 * [363]11370 -Wunreachable-code gives false complaints
9355 * [364]11637 invalid assembly with -fnon-call-exceptions
9356 * [365]11885 Problem with bitfields in packed structs
9357 * [366]12082 Inappropriate unreachable code warnings
9358 * [367]12180 Inline optimization fails for variadic function
9359 * [368]12340 loop unroller + gcse produces wrong code
9361 C++ compiler and library
9363 * [369]3907 nested template parameter collides with member name
9364 * [370]5293 confusing message when binding a temporary to a reference
9365 * [371]5296 [DR115] Pointers to functions and to template functions
9366 behave differently in deduction
9367 * [372]7939 ICE on function template specialization
9368 * [373]8656 Unable to assign function with __attribute__ and pointer
9369 return type to an appropriate variable
9370 * [374]10147 Confusing error message for invalid template function
9372 * [375]11400 std::search_n() makes assumptions about Size parameter
9373 * [376]11409 issues with using declarations, overloading, and
9375 * [377]11740 ctype<wchar_t>::do_is(mask, wchar_t) doesn't handle
9376 multiple bits in mask
9377 * [378]11786 operator() call on variable in other namespace not
9379 * [379]11867 static_cast ignores ambiguity
9380 * [380]11928 bug with conversion operators that are typedefs
9381 * [381]12114 Uninitialized memory accessed in dtor
9382 * [382]12163 static_cast + explicit constructor regression
9383 * [383]12181 Wrong code with comma operator and c++
9384 * [384]12236 regparm and fastcall messes up parameters
9385 * [385]12266 incorrect instantiation of unneeded template during
9387 * [386]12296 istream::peek() doesn't set eofbit
9388 * [387]12298 [sjlj exceptions] Stack unwind destroys
9389 not-yet-constructed object
9390 * [388]12369 ICE with templates and friends
9391 * [389]12337 apparently infinite loop in g++
9392 * [390]12344 stdcall attribute ignored if function returns a pointer
9393 * [391]12451 missing(late) class forward declaration in cxxabi.h
9394 * [392]12486 g++ accepts invalid use of a qualified name
9396 x86 specific (Intel/AMD)
9398 * [393]8869 [x86 MMX] ICE with const variable optimization and MMX
9400 * [394]9786 ICE in fixup_abnormal_edges with -fnon-call-exceptions
9402 * [395]11689 g++3.3 emits un-assembleable code for k6 architecture
9403 * [396]12116 [k6] Invalid assembly output values with X-MAME code
9404 * [397]12070 ICE converting between double and long double with
9409 * [398]11184 [ia64 hpux] ICE on __builtin_apply building libobjc
9410 * [399]11535 __builtin_return_address may not work on ia64
9411 * [400]11693 [ia64] ICE in gen_nop_type
9412 * [401]12224 [ia64] Thread-local storage doesn't work
9416 * [402]11087 [powerpc64-linux] GCC miscompiles raid1.c from linux
9418 * [403]11319 loop miscompiled on ppc32
9419 * [404]11949 ICE Compiler segfault with ffmpeg -maltivec code
9423 * [405]11662 wrong code for expr. with cast to long long and
9425 * [406]11965 invalid assembler code for a shift < 32 operation
9426 * [407]12301 (c++) stack corruption when a returned expression throws
9431 * [408]11717 [alpha-linux] unrecognizable insn compiling for.c of
9436 * [409]11313 problem with #pragma weak and static inline functions
9437 * [410]11712 __STDC_EXT__ not defined for C++ by default anymore?
9441 * [411]12166 Profiled programs crash if PROFDIR is set
9443 Solaris-x86 specific
9445 * [412]12101 i386 Solaris no longer works with GNU as?
9447 Miscellaneous embedded target-specific bugs
9449 * [413]10988 [m32r-elf] wrong blockmove code with -O3
9450 * [414]11805 [h8300-unknown-coff] [H8300] ICE for simple code with
9452 * [415]11902 [sh4] spec file improperly inserts rpath even when none
9454 * [416]11903 [sh4] -pthread fails to link due to error in spec file
9456 __________________________________________________________________
9462 In addition to the bug fixes documented below, this release contains
9463 few minor features such as:
9464 * Support for --with-sysroot
9465 * Support for automatic detection of executable stacks
9466 * Support for SSE3 instructions
9467 * Support for thread local storage debugging under GDB on S390
9471 This section lists the problem reports (PRs) from [417]GCC's bug
9472 tracking system that are known to be fixed in the 3.3.3 release. This
9473 list might not be complete (that is, it is possible that some PRs that
9474 have been fixed are not listed here).
9476 Bootstrap failures and issues
9478 * [418]11890 Building cross gcc-3.3.1 for sparc-sun-solaris2.6 fails
9479 * [419]12399 boehm-gc fails (when building a cross compiler): libtool
9480 unable to infer tagged configuration
9481 * [420]13068 mklibgcc.in doesn't handle multi-level multilib
9482 subdirectories properly
9484 Internal compiler errors (multi-platform)
9486 * [421]10060 ICE (stack overflow) on huge file (300k lines) due to
9487 recursive behaviour of copy_rtx_if_shared, in emit_rtl.c
9488 * [422]10555 (c++) ICE on undefined template argument
9489 * [423]10706 (c++) ICE in mangle_class_name_for_template
9490 * [424]11496 (fortran) error in flow_loops_find when -funroll-loops
9492 * [425]11741 ICE in pre_insert_copy_insn, in gcse.c
9493 * [426]12440 GCC crashes during compilation of quicktime4linux 2.0.0
9494 * [427]12632 (fortran) -fbounds-check ICE
9495 * [428]12712 (c++) ICE on short legit C++ code fragment with gcc
9497 * [429]12726 (c++) ICE (segfault) on trivial code
9498 * [430]12890 (c++) ICE on compilation of class with throwing method
9499 * [431]12900 (c++) ICE in rtl_verify_flow_info_1
9500 * [432]13060 (fortran) ICE in fixup_var_refs_1, in function.c on
9501 correct code with -O2 -fno-force-mem
9502 * [433]13289 (c++) ICE in regenerate_decl_from_template on recursive
9504 * [434]13318 ICE: floating point exception in the loop optimizer
9505 * [435]13392 (c++) ICE in convert_from_eh_region_ranges_1, in
9507 * [436]13574 (c++) invalid array default initializer in class lets
9508 gcc consume all memory and die
9509 * [437]13475 ICE on SIMD variables with partial value initialization
9510 * [438]13797 (c++) ICE on invalid template parameter
9511 * [439]13824 (java) gcj SEGV with simple .java program
9513 C and optimization bugs
9515 * [440]8776 loop invariants are not removed (most likely)
9516 * [441]10339 [sparc,ppc,ppc64] Invalid optimization: replacing
9518 * [442]11350 undefined labels with -Os -fPIC
9519 * [443]12826 Optimizer removes reference through volatile pointer
9520 * [444]12500 stabs debug info: void no longer a predefined / builtin
9522 * [445]12941 builtin-bitops-1.c miscompilation (latent bug)
9523 * [446]12953 tree inliner bug (in inline_forbidden_p) and fix
9524 * [447]13041 linux-2.6/sound/core/oss/rate.c miscompiled
9525 * [448]13507 spurious printf format warning
9526 * [449]13382 Type information for const pointer disappears during
9528 * [450]13394 noreturn attribute ignored on recursive invokation
9529 * [451]13400 Compiled code crashes storing to read-only location
9530 * [452]13521 Endless loop in calculate_global_regs_live
9532 C++ compiler and library
9534 Some of the bug fixes in this list were made to implement decisions
9535 that the ISO C++ standards committee has made concerning several defect
9536 reports (DRs). Links in the list below point to detailed discussion of
9537 the relevant defect report.
9538 * [453]2094 unimplemented: use of `ptrmem_cst' in template type
9540 * [454]2294 using declaration confusion
9541 * [455]5050 template instantiation depth exceeds limit: recursion
9543 * [456]9371 Bad exception handling in
9544 i/ostream::operator>>/<<(streambuf*)
9545 * [457]9546 bad exception handling in ostream members
9546 * [458]10081 basic_ios::_M_cache_locale leaves NULL members in the
9547 face of unknown locales
9548 * [459]10093 [460][DR 61] Setting failbit in exceptions doesn't work
9549 * [461]10095 istream::operator>>(int&) sets ios::badbit when
9550 ios::failbit is set.
9551 * [462]11554 Warning about reordering of initializers doesn't mention
9552 location of constructor
9553 * [463]12297 istream::sentry::sentry() handles eof() incorrectly.
9554 * [464]12352 Exception safety problems in src/localename.cc
9555 * [465]12438 Memory leak in locale::combine()
9556 * [466]12540 Memory leak in locale::locale(const char*)
9557 * [467]12594 DRs [468]60 [TC] and [469]63 [TC] not implemented
9558 * [470]12657 Resolution of [471]DR 292 (WP) still unimplemented
9559 * [472]12696 memory eating infinite loop in diagnostics (error
9561 * [473]12815 Code compiled with optimization behaves unexpectedly
9562 * [474]12862 Conflicts between typedefs/enums and namespace member
9564 * [475]12926 Wrong value after assignment in initialize list using
9566 * [476]12967 Resolution of [477]DR 300 [WP] still unimplemented
9567 * [478]12971 Resolution of [479]DR 328 [WP] still unimplemented
9568 * [480]13007 basic_streambuf::pubimbue, imbue wrong
9569 * [481]13009 Implicitly-defined assignment operator writes to wrong
9571 * [482]13057 regparm attribute not applied to destructor
9572 * [483]13070 -Wformat option ignored in g++
9573 * [484]13081 forward template declarations in <complex> let inlining
9575 * [485]13239 Assertion does not seem to work correctly anymore
9576 * [486]13262 "xxx is private within this context" when initializing a
9577 self-contained template class
9578 * [487]13290 simple typo in concept checking for std::generate_n
9579 * [488]13323 Template code does not compile in presence of typedef
9580 * [489]13369 __verify_grouping (and __add_grouping?) not correct
9581 * [490]13371 infinite loop with packed struct and inlining
9582 * [491]13445 Template argument replacement "dereferences" a typedef
9583 * [492]13461 Fails to access protected-ctor from public constant
9584 * [493]13462 Non-standard-conforming type set::pointer
9585 * [494]13478 gcc uses wrong constructor to initialize a const
9587 * [495]13544 "conflicting types" for enums in different scopes
9588 * [496]13650 string::compare should not (always) use
9589 traits_type::length()
9590 * [497]13683 bogus warning about passing non-PODs through ellipsis
9591 * [498]13688 Derived class is denied access to protected base class
9593 * [499]13774 Member variable cleared in virtual multiple inheritance
9595 * [500]13884 Protect sstream.tcc from extern template use
9597 Java compiler and library
9599 * [501]10746 [win32] garbage collection crash in GCJ
9601 Objective-C compiler and library
9603 * [502]11433 Crash due to dereferencing null pointer when querying
9606 Fortran compiler and library
9608 * [503]12633 logical expression gives incorrect result with
9609 -fugly-logint option
9610 * [504]13037 [gcse-lm] g77 generates incorrect code
9611 * [505]13213 Hex constant problem when compiling with -fugly-logint
9614 x86-specific (Intel/AMD)
9616 * [506]4490 ICE with -m128bit-long-double
9617 * [507]12292 [x86_64] ICE: RTL check: expected code `const_int', have
9618 `reg' in make_field_assignment, in combine.c
9619 * [508]12441 ICE: can't find a register to spill
9620 * [509]12943 array static-init failure under -fpic, -fPIC
9621 * [510]13608 Incorrect code with -O3 -ffast-math
9625 * [511]11598 testcase gcc.dg/20020118-1.c fails runtime check of
9626 __attribute__((aligned(16)))
9627 * [512]11793 ICE in extract_insn, in recog.c (const_vector's)
9628 * [513]12467 vmsumubm emitted when vmsummbm appropriate (typo in
9630 * [514]12537 g++ generates writeable text sections
9634 * [515]12496 wrong result for __atomic_add(&value, -1) when using -O0
9636 * [516]12865 mprotect call to make trampoline executable may fail
9637 * [517]13354 ICE in sparc_emit_set_const32
9641 * [518]10467 [arm] ICE in pre_insert_copy_insn,
9645 * [519]11226 ICE passing struct arg with two floats
9646 * [520]11227 ICE for _Complex float, _Complex long double args
9647 * [521]12644 GCC 3.3.2 fails to compile glibc on ia64
9648 * [522]13149 build gcc-3.3.2 1305 error:unrecognizable insn
9649 * Various fixes for libunwind
9653 * [523]12654 Incorrect comparison code generated for Alpha
9654 * [524]12965 SEGV+ICE in cc1plus on alpha-linux with -O2
9655 * [525]13031 ICE (unrecognizable insn) when building gnome-libs-1.4.2
9659 * [526]11634 [hppa] ICE in verify_local_live_at_start, in flow.c
9660 * [527]12158 [hppa] compilation does not terminate at -O1
9664 * [528]11992 Wrong built-in code for memcmp with length 1<<24: only
9665 (1<<24)-1 possible for CLCL-Instruction
9669 * [529]9365 segfault in gen_far_branch (config/sh/sh.c)
9670 * [530]10392 optimizer generates faulty array indexing
9671 * [531]11322 SH profiler outputs multiple definitions of symbol
9672 * [532]13069 gcc/config/sh/rtems.h broken
9673 * [533]13302 Putting a va_list in a struct causes seg fault
9674 * [534]13585 Incorrect optimization of call to sfunc
9675 * Fix inappropriately exported libgcc functions from the shared
9678 Other embedded target specific
9680 * [535]8916 [mcore] unsigned char assign gets hosed.
9681 * [536]11576 [h8300] ICE in change_address_1, in emit-rtl.c
9682 * [537]13122 [h8300] local variable gets corrupted by function call
9683 when -fomit-frame-pointer is given
9684 * [538]13256 [cris] strict_low_part mistreated in delay slots
9685 * [539]13373 [mcore] optimization with -frerun-cse-after-loop
9686 -fexpensive-optimizations produces wrong code on mcore
9690 * [540]12561 gcc/config/t-gnu needs updating to work with
9695 * [541]6243 testsuite fails almost all tests due to no libintl in
9696 LD_LIBRARY_PATH during test.
9697 * [542]11397 weak aliases broken on Tru64 UNIX
9701 * [543]12505 build failure due to defines of uchar in cpphash.h and
9703 * [544]13150 WEAK symbols not exported by collect2
9707 * [545]12666 fixincludes problem on IRIX 6.5.19m
9711 * [546]12969 Including sys/byteorder.h breaks configure checks
9713 Testsuite problems (compiler is not affected)
9715 * [547]10819 testsuite creates CR+LF on compiler version lines in
9717 * [548]11612 abi_check not finding correct libgcc_s.so.1
9721 * [549]13211 using -###, incorrect warnings about unused linker file
9723 __________________________________________________________________
9727 This is the [550]list of problem reports (PRs) from GCC's bug tracking
9728 system that are known to be fixed in the 3.3.4 release. This list might
9729 not be complete (that is, it is possible that some PRs that have been
9730 fixed are not listed here).
9731 __________________________________________________________________
9735 This is the [551]list of problem reports (PRs) from GCC's bug tracking
9736 system that are known to be fixed in the 3.3.5 release. This list might
9737 not be complete (that is, it is possible that some PRs that have been
9738 fixed are not listed here).
9739 __________________________________________________________________
9743 This is the [552]list of problem reports (PRs) from GCC's bug tracking
9744 system that are known to be fixed in the 3.3.6 release. This list might
9745 not be complete (that is, it is possible that some PRs that have been
9746 fixed are not listed here).
9749 For questions related to the use of GCC, please consult these web
9750 pages and the [553]GCC manuals. If that fails, the
9751 [554]gcc-help@gcc.gnu.org mailing list might help. Comments on these
9752 web pages and the development of GCC are welcome on our developer
9753 list at [555]gcc@gcc.gnu.org. All of [556]our lists have public
9756 Copyright (C) [557]Free Software Foundation, Inc. Verbatim copying and
9757 distribution of this entire article is permitted in any medium,
9758 provided this notice is preserved.
9760 These pages are [558]maintained by the GCC team. Last modified
9765 1. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.6
9766 2. http://gcc.gnu.org/gcc-3.1/changes.html#obsolete_systems
9767 3. http://gcc.gnu.org/gcc-3.3/changes.html#obsolete_systems
9768 4. http://gcc.gnu.org/gcc-3.3/changes.html#nonnull_attribute
9769 5. http://gcc.gnu.org/news/dfa.html
9770 6. http://gcc.gnu.org/gcc-3.3/c99status.html
9771 7. http://gcc.gnu.org/onlinedocs/gcc-3.3.6/g77/News.html
9772 8. http://gcc.gnu.org/PR10140
9773 9. http://gcc.gnu.org/PR10198
9774 10. http://gcc.gnu.org/PR10338
9775 11. http://gcc.gnu.org/PR3581
9776 12. http://gcc.gnu.org/PR4382
9777 13. http://gcc.gnu.org/PR5533
9778 14. http://gcc.gnu.org/PR6387
9779 15. http://gcc.gnu.org/PR6412
9780 16. http://gcc.gnu.org/PR6620
9781 17. http://gcc.gnu.org/PR6663
9782 18. http://gcc.gnu.org/PR7068
9783 19. http://gcc.gnu.org/PR7083
9784 20. http://gcc.gnu.org/PR7647
9785 21. http://gcc.gnu.org/PR7675
9786 22. http://gcc.gnu.org/PR7718
9787 23. http://gcc.gnu.org/PR8116
9788 24. http://gcc.gnu.org/PR8358
9789 25. http://gcc.gnu.org/PR8511
9790 26. http://gcc.gnu.org/PR8564
9791 27. http://gcc.gnu.org/PR8660
9792 28. http://gcc.gnu.org/PR8766
9793 29. http://gcc.gnu.org/PR8803
9794 30. http://gcc.gnu.org/PR8846
9795 31. http://gcc.gnu.org/PR8906
9796 32. http://gcc.gnu.org/PR9216
9797 33. http://gcc.gnu.org/PR9261
9798 34. http://gcc.gnu.org/PR9263
9799 35. http://gcc.gnu.org/PR9429
9800 36. http://gcc.gnu.org/PR9516
9801 37. http://gcc.gnu.org/PR9600
9802 38. http://gcc.gnu.org/PR9629
9803 39. http://gcc.gnu.org/PR9672
9804 40. http://gcc.gnu.org/PR9749
9805 41. http://gcc.gnu.org/PR9794
9806 42. http://gcc.gnu.org/PR9829
9807 43. http://gcc.gnu.org/PR9916
9808 44. http://gcc.gnu.org/PR9936
9809 45. http://gcc.gnu.org/PR10262
9810 46. http://gcc.gnu.org/PR10278
9811 47. http://gcc.gnu.org/PR10446
9812 48. http://gcc.gnu.org/PR10451
9813 49. http://gcc.gnu.org/PR10506
9814 50. http://gcc.gnu.org/PR10549
9815 51. http://gcc.gnu.org/PR2001
9816 52. http://gcc.gnu.org/PR2391
9817 53. http://gcc.gnu.org/PR2960
9818 54. http://gcc.gnu.org/PR4046
9819 55. http://gcc.gnu.org/PR6405
9820 56. http://gcc.gnu.org/PR6798
9821 57. http://gcc.gnu.org/PR6871
9822 58. http://gcc.gnu.org/PR6909
9823 59. http://gcc.gnu.org/PR7189
9824 60. http://gcc.gnu.org/PR7642
9825 61. http://gcc.gnu.org/PR8634
9826 62. http://gcc.gnu.org/PR8750
9827 63. http://gcc.gnu.org/PR2161
9828 64. http://gcc.gnu.org/PR4319
9829 65. http://gcc.gnu.org/PR8602
9830 66. http://gcc.gnu.org/PR9177
9831 67. http://gcc.gnu.org/PR9853
9832 68. http://gcc.gnu.org/PR45
9833 69. http://gcc.gnu.org/PR3784
9834 70. http://gcc.gnu.org/PR764
9835 71. http://gcc.gnu.org/PR5116
9836 72. http://gcc.gnu.org/PR2862
9837 73. http://gcc.gnu.org/PR3663
9838 74. http://gcc.gnu.org/PR3797
9839 75. http://gcc.gnu.org/PR3948
9840 76. http://gcc.gnu.org/PR4137
9841 77. http://gcc.gnu.org/PR4361
9842 78. http://gcc.gnu.org/PR4802
9843 79. http://gcc.gnu.org/PR5837
9844 80. http://gcc.gnu.org/PR4803
9845 81. http://gcc.gnu.org/PR5094
9846 82. http://gcc.gnu.org/PR5730
9847 83. http://gcc.gnu.org/PR6713
9848 84. http://gcc.gnu.org/PR7015
9849 85. http://gcc.gnu.org/PR7086
9850 86. http://gcc.gnu.org/PR7099
9851 87. http://gcc.gnu.org/PR7247
9852 88. http://gcc.gnu.org/PR7441
9853 89. http://gcc.gnu.org/PR7768
9854 90. http://gcc.gnu.org/PR7804
9855 91. http://gcc.gnu.org/PR8099
9856 92. http://gcc.gnu.org/PR8117
9857 93. http://gcc.gnu.org/PR8205
9858 94. http://gcc.gnu.org/PR8645
9859 95. http://gcc.gnu.org/PR8724
9860 96. http://gcc.gnu.org/PR8805
9861 97. http://gcc.gnu.org/PR8691
9862 98. http://gcc.gnu.org/PR8700
9863 99. http://gcc.gnu.org/PR8724
9864 100. http://gcc.gnu.org/PR8949
9865 101. http://gcc.gnu.org/PR9016
9866 102. http://gcc.gnu.org/PR9053
9867 103. http://gcc.gnu.org/PR9152
9868 104. http://gcc.gnu.org/PR9182
9869 105. http://gcc.gnu.org/PR9297
9870 106. http://gcc.gnu.org/PR9318
9871 107. http://gcc.gnu.org/PR9320
9872 108. http://gcc.gnu.org/PR9400
9873 109. http://gcc.gnu.org/PR9424
9874 110. http://gcc.gnu.org/PR9425
9875 111. http://gcc.gnu.org/PR9439
9876 112. http://gcc.gnu.org/PR9474
9877 113. http://gcc.gnu.org/PR9548
9878 114. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#231
9879 115. http://gcc.gnu.org/PR9555
9880 116. http://gcc.gnu.org/PR9561
9881 117. http://gcc.gnu.org/PR9563
9882 118. http://gcc.gnu.org/PR9582
9883 119. http://gcc.gnu.org/PR9622
9884 120. http://gcc.gnu.org/PR9683
9885 121. http://gcc.gnu.org/PR9791
9886 122. http://gcc.gnu.org/PR9817
9887 123. http://gcc.gnu.org/PR9825
9888 124. http://gcc.gnu.org/PR9826
9889 125. http://gcc.gnu.org/PR9924
9890 126. http://gcc.gnu.org/PR9946
9891 127. http://gcc.gnu.org/PR9964
9892 128. http://gcc.gnu.org/PR9988
9893 129. http://gcc.gnu.org/PR10033
9894 130. http://gcc.gnu.org/PR10097
9895 131. http://gcc.gnu.org/PR10132
9896 132. http://gcc.gnu.org/PR10180
9897 133. http://gcc.gnu.org/PR10199
9898 134. http://gcc.gnu.org/PR10300
9899 135. http://gcc.gnu.org/PR10427
9900 136. http://gcc.gnu.org/PR10503
9901 137. http://gcc.gnu.org/PR5956
9902 138. http://gcc.gnu.org/PR1832
9903 139. http://gcc.gnu.org/PR3924
9904 140. http://gcc.gnu.org/PR5634
9905 141. http://gcc.gnu.org/PR6367
9906 142. http://gcc.gnu.org/PR6491
9907 143. http://gcc.gnu.org/PR6742
9908 144. http://gcc.gnu.org/PR7113
9909 145. http://gcc.gnu.org/PR7236
9910 146. http://gcc.gnu.org/PR7278
9911 147. http://gcc.gnu.org/PR7384
9912 148. http://gcc.gnu.org/PR7388
9913 149. http://gcc.gnu.org/PR8587
9914 150. http://gcc.gnu.org/PR9038
9915 151. http://gcc.gnu.org/PR10197
9916 152. http://gcc.gnu.org/PR6005
9917 153. http://gcc.gnu.org/PR6389
9918 154. http://gcc.gnu.org/PR6576
9919 155. http://gcc.gnu.org/PR6652
9920 156. http://gcc.gnu.org/PR7060
9921 157. http://gcc.gnu.org/PR7073
9922 158. http://gcc.gnu.org/PR7180
9923 159. http://gcc.gnu.org/PR7416
9924 160. http://gcc.gnu.org/PR7570
9925 161. http://gcc.gnu.org/PR7578
9926 162. http://gcc.gnu.org/PR7611
9927 163. http://gcc.gnu.org/PR7709
9928 164. http://gcc.gnu.org/PR7766
9929 165. http://gcc.gnu.org/PR7785
9930 166. http://gcc.gnu.org/PR7786
9931 167. http://gcc.gnu.org/PR8142
9932 168. http://gcc.gnu.org/PR8234
9933 169. http://gcc.gnu.org/PR8415
9934 170. http://gcc.gnu.org/PR8481
9935 171. http://gcc.gnu.org/PR8593
9936 172. http://gcc.gnu.org/PR8759
9937 173. http://gcc.gnu.org/PR8997
9938 174. http://gcc.gnu.org/PR9253
9939 175. http://gcc.gnu.org/PR9254
9940 176. http://gcc.gnu.org/PR9271
9941 177. http://gcc.gnu.org/PR6767
9942 178. http://gcc.gnu.org/PR9911
9943 179. http://gcc.gnu.org/PR10020
9944 180. http://gcc.gnu.org/PR10546
9945 181. http://gcc.gnu.org/PR7029
9946 182. http://gcc.gnu.org/PR2903
9947 183. http://gcc.gnu.org/PR7873
9948 184. http://gcc.gnu.org/PR7680
9949 185. http://gcc.gnu.org/PR8705
9950 186. http://gcc.gnu.org/PR9986
9951 187. http://gcc.gnu.org/PR10056
9952 188. http://gcc.gnu.org/PR6744
9953 189. http://gcc.gnu.org/PR7361
9954 190. http://gcc.gnu.org/PR9496
9955 191. http://gcc.gnu.org/PR7067
9956 192. http://gcc.gnu.org/PR8480
9957 193. http://gcc.gnu.org/PR8784
9958 194. http://gcc.gnu.org/PR10315
9959 195. http://gcc.gnu.org/PR10267
9960 196. http://gcc.gnu.org/PR7916
9961 197. http://gcc.gnu.org/PR7926
9962 198. http://gcc.gnu.org/PR8555
9963 199. http://gcc.gnu.org/PR8994
9964 200. http://gcc.gnu.org/PR9426
9965 201. http://gcc.gnu.org/PR9806
9966 202. http://gcc.gnu.org/PR10077
9967 203. http://gcc.gnu.org/PR10233
9968 204. http://gcc.gnu.org/PR10286
9969 205. http://gcc.gnu.org/PR10308
9970 206. http://gcc.gnu.org/PR11272
9971 207. http://gcc.gnu.org/PR5754
9972 208. http://gcc.gnu.org/PR6597
9973 209. http://gcc.gnu.org/PR6949
9974 210. http://gcc.gnu.org/PR7053
9975 211. http://gcc.gnu.org/PR8164
9976 212. http://gcc.gnu.org/PR8384
9977 213. http://gcc.gnu.org/PR9559
9978 214. http://gcc.gnu.org/PR9649
9979 215. http://gcc.gnu.org/PR9864
9980 216. http://gcc.gnu.org/PR10432
9981 217. http://gcc.gnu.org/PR10475
9982 218. http://gcc.gnu.org/PR10635
9983 219. http://gcc.gnu.org/PR10661
9984 220. http://gcc.gnu.org/PR10700
9985 221. http://gcc.gnu.org/PR10712
9986 222. http://gcc.gnu.org/PR10796
9987 223. http://gcc.gnu.org/PR10890
9988 224. http://gcc.gnu.org/PR10939
9989 225. http://gcc.gnu.org/PR10956
9990 226. http://gcc.gnu.org/PR11041
9991 227. http://gcc.gnu.org/PR11059
9992 228. http://gcc.gnu.org/PR11083
9993 229. http://gcc.gnu.org/PR11105
9994 230. http://gcc.gnu.org/PR11149
9995 231. http://gcc.gnu.org/PR11228
9996 232. http://gcc.gnu.org/PR11282
9997 233. http://gcc.gnu.org/PR11301
9998 234. http://gcc.gnu.org/PR11308
9999 235. http://gcc.gnu.org/PR11473
10000 236. http://gcc.gnu.org/PR11503
10001 237. http://gcc.gnu.org/PR11513
10002 238. http://gcc.gnu.org/PR11198
10003 239. http://gcc.gnu.org/PR11304
10004 240. http://gcc.gnu.org/PR11381
10005 241. http://gcc.gnu.org/PR11536
10006 242. http://gcc.gnu.org/PR11557
10007 243. http://gcc.gnu.org/PR5897
10008 244. http://gcc.gnu.org/PR11279
10009 245. http://gcc.gnu.org/PR11022
10010 246. http://gcc.gnu.org/PR2330
10011 247. http://gcc.gnu.org/PR5388
10012 248. http://gcc.gnu.org/PR5390
10013 249. http://gcc.gnu.org/PR7877
10014 250. http://gcc.gnu.org/PR9393
10015 251. http://gcc.gnu.org/PR10032
10016 252. http://gcc.gnu.org/PR10468
10017 253. http://gcc.gnu.org/PR10527
10018 254. http://gcc.gnu.org/PR10679
10019 255. http://gcc.gnu.org/PR10682
10020 256. http://gcc.gnu.org/PR10689
10021 257. http://gcc.gnu.org/PR10845
10022 258. http://gcc.gnu.org/PR10849
10023 259. http://gcc.gnu.org/PR10888
10024 260. http://gcc.gnu.org/PR10929
10025 261. http://gcc.gnu.org/PR10931
10026 262. http://gcc.gnu.org/PR10940
10027 263. http://gcc.gnu.org/PR10968
10028 264. http://gcc.gnu.org/PR10990
10029 265. http://gcc.gnu.org/PR11039
10030 266. http://gcc.gnu.org/PR11062
10031 267. http://gcc.gnu.org/PR11095
10032 268. http://gcc.gnu.org/PR11098
10033 269. http://gcc.gnu.org/PR11137
10034 270. http://gcc.gnu.org/PR11154
10035 271. http://gcc.gnu.org/PR11329
10036 272. http://gcc.gnu.org/PR11332
10037 273. http://gcc.gnu.org/PR11431
10038 274. http://gcc.gnu.org/PR11528
10039 275. http://gcc.gnu.org/PR11546
10040 276. http://gcc.gnu.org/PR11567
10041 277. http://gcc.gnu.org/PR11645
10042 278. http://gcc.gnu.org/PR5179
10043 279. http://gcc.gnu.org/PR8204
10044 280. http://gcc.gnu.org/PR10838
10045 281. http://gcc.gnu.org/PR10886
10046 282. http://gcc.gnu.org/PR11349
10047 283. http://gcc.gnu.org/PR4823
10048 284. http://gcc.gnu.org/PR8878
10049 285. http://gcc.gnu.org/PR9815
10050 286. http://gcc.gnu.org/PR10402
10051 287. http://gcc.gnu.org/PR10504
10052 288. http://gcc.gnu.org/PR10673
10053 289. http://gcc.gnu.org/PR11044
10054 290. http://gcc.gnu.org/PR11089
10055 291. http://gcc.gnu.org/PR11420
10056 292. http://gcc.gnu.org/PR9362
10057 293. http://gcc.gnu.org/PR10142
10058 294. http://gcc.gnu.org/PR10663
10059 295. http://gcc.gnu.org/PR10835
10060 296. http://gcc.gnu.org/PR10876
10061 297. http://gcc.gnu.org/PR10955
10062 298. http://gcc.gnu.org/PR11018
10063 299. http://gcc.gnu.org/PR11556
10064 300. http://gcc.gnu.org/PR10907
10065 301. http://gcc.gnu.org/PR11320
10066 302. http://gcc.gnu.org/PR11599
10067 303. http://gcc.gnu.org/PR9745
10068 304. http://gcc.gnu.org/PR10871
10069 305. http://gcc.gnu.org/PR11440
10070 306. http://gcc.gnu.org/PR7594
10071 307. http://gcc.gnu.org/PR10557
10072 308. http://gcc.gnu.org/PR11054
10073 309. http://gcc.gnu.org/PR10834
10074 310. http://gcc.gnu.org/PR10842
10075 311. http://gcc.gnu.org/PR11052
10076 312. http://gcc.gnu.org/PR11183
10077 313. http://gcc.gnu.org/PR11084
10078 314. http://gcc.gnu.org/PR10331
10079 315. http://gcc.gnu.org/PR10413
10080 316. http://gcc.gnu.org/PR11096
10081 317. http://gcc.gnu.org/PR2873
10082 318. http://gcc.gnu.org/PR3163
10083 319. http://gcc.gnu.org/PR5287
10084 320. http://gcc.gnu.org/PR10148
10085 321. http://gcc.gnu.org/PR8787
10086 322. http://gcc.gnu.org/PR10900
10087 323. http://gcc.gnu.org/PR1607
10088 324. http://gcc.gnu.org/PR4252
10089 325. http://gcc.gnu.org/PR4490
10090 326. http://gcc.gnu.org/PR10355
10091 327. http://gcc.gnu.org/PR10726
10092 328. http://gcc.gnu.org/PR10805
10093 329. http://gcc.gnu.org/PR10815
10094 330. http://gcc.gnu.org/PR10877
10095 331. http://gcc.gnu.org/PR11280
10096 332. http://gcc.gnu.org/PR11466
10097 333. http://gcc.gnu.org/PR10737
10098 334. http://gcc.gnu.org/PR10810
10099 335. http://gcc.gnu.org/bugzilla/
10100 336. http://gcc.gnu.org/PR8336
10101 337. http://gcc.gnu.org/PR9330
10102 338. http://gcc.gnu.org/PR9631
10103 339. http://gcc.gnu.org/PR9877
10104 340. http://gcc.gnu.org/PR11687
10105 341. http://gcc.gnu.org/PR12263
10106 342. http://gcc.gnu.org/PR12490
10107 343. http://gcc.gnu.org/PR7277
10108 344. http://gcc.gnu.org/PR7939
10109 345. http://gcc.gnu.org/PR11063
10110 346. http://gcc.gnu.org/PR11207
10111 347. http://gcc.gnu.org/PR11522
10112 348. http://gcc.gnu.org/PR11595
10113 349. http://gcc.gnu.org/PR11646
10114 350. http://gcc.gnu.org/PR11665
10115 351. http://gcc.gnu.org/PR11852
10116 352. http://gcc.gnu.org/PR11878
10117 353. http://gcc.gnu.org/PR11883
10118 354. http://gcc.gnu.org/PR11991
10119 355. http://gcc.gnu.org/PR12146
10120 356. http://gcc.gnu.org/PR12215
10121 357. http://gcc.gnu.org/PR12369
10122 358. http://gcc.gnu.org/PR12446
10123 359. http://gcc.gnu.org/PR12510
10124 360. http://gcc.gnu.org/PR12544
10125 361. http://gcc.gnu.org/PR9862
10126 362. http://gcc.gnu.org/PR10962
10127 363. http://gcc.gnu.org/PR11370
10128 364. http://gcc.gnu.org/PR11637
10129 365. http://gcc.gnu.org/PR11885
10130 366. http://gcc.gnu.org/PR12082
10131 367. http://gcc.gnu.org/PR12180
10132 368. http://gcc.gnu.org/PR12340
10133 369. http://gcc.gnu.org/PR3907
10134 370. http://gcc.gnu.org/PR5293
10135 371. http://gcc.gnu.org/PR5296
10136 372. http://gcc.gnu.org/PR7939
10137 373. http://gcc.gnu.org/PR8656
10138 374. http://gcc.gnu.org/PR10147
10139 375. http://gcc.gnu.org/PR11400
10140 376. http://gcc.gnu.org/PR11409
10141 377. http://gcc.gnu.org/PR11740
10142 378. http://gcc.gnu.org/PR11786
10143 379. http://gcc.gnu.org/PR11867
10144 380. http://gcc.gnu.org/PR11928
10145 381. http://gcc.gnu.org/PR12114
10146 382. http://gcc.gnu.org/PR12163
10147 383. http://gcc.gnu.org/PR12181
10148 384. http://gcc.gnu.org/PR12236
10149 385. http://gcc.gnu.org/PR12266
10150 386. http://gcc.gnu.org/PR12296
10151 387. http://gcc.gnu.org/PR12298
10152 388. http://gcc.gnu.org/PR12369
10153 389. http://gcc.gnu.org/PR12337
10154 390. http://gcc.gnu.org/PR12344
10155 391. http://gcc.gnu.org/PR12451
10156 392. http://gcc.gnu.org/PR12486
10157 393. http://gcc.gnu.org/PR8869
10158 394. http://gcc.gnu.org/PR9786
10159 395. http://gcc.gnu.org/PR11689
10160 396. http://gcc.gnu.org/PR12116
10161 397. http://gcc.gnu.org/PR12070
10162 398. http://gcc.gnu.org/PR11184
10163 399. http://gcc.gnu.org/PR11535
10164 400. http://gcc.gnu.org/PR11693
10165 401. http://gcc.gnu.org/PR12224
10166 402. http://gcc.gnu.org/PR11087
10167 403. http://gcc.gnu.org/PR11319
10168 404. http://gcc.gnu.org/PR11949
10169 405. http://gcc.gnu.org/PR11662
10170 406. http://gcc.gnu.org/PR11965
10171 407. http://gcc.gnu.org/PR12301
10172 408. http://gcc.gnu.org/PR11717
10173 409. http://gcc.gnu.org/PR11313
10174 410. http://gcc.gnu.org/PR11712
10175 411. http://gcc.gnu.org/PR12166
10176 412. http://gcc.gnu.org/PR12101
10177 413. http://gcc.gnu.org/PR10988
10178 414. http://gcc.gnu.org/PR11805
10179 415. http://gcc.gnu.org/PR11902
10180 416. http://gcc.gnu.org/PR11903
10181 417. http://gcc.gnu.org/bugzilla/
10182 418. http://gcc.gnu.org/PR11890
10183 419. http://gcc.gnu.org/PR12399
10184 420. http://gcc.gnu.org/PR13068
10185 421. http://gcc.gnu.org/PR10060
10186 422. http://gcc.gnu.org/PR10555
10187 423. http://gcc.gnu.org/PR10706
10188 424. http://gcc.gnu.org/PR11496
10189 425. http://gcc.gnu.org/PR11741
10190 426. http://gcc.gnu.org/PR12440
10191 427. http://gcc.gnu.org/PR12632
10192 428. http://gcc.gnu.org/PR12712
10193 429. http://gcc.gnu.org/PR12726
10194 430. http://gcc.gnu.org/PR12890
10195 431. http://gcc.gnu.org/PR12900
10196 432. http://gcc.gnu.org/PR13060
10197 433. http://gcc.gnu.org/PR13289
10198 434. http://gcc.gnu.org/PR13318
10199 435. http://gcc.gnu.org/PR13392
10200 436. http://gcc.gnu.org/PR13574
10201 437. http://gcc.gnu.org/PR13475
10202 438. http://gcc.gnu.org/PR13797
10203 439. http://gcc.gnu.org/PR13824
10204 440. http://gcc.gnu.org/PR8776
10205 441. http://gcc.gnu.org/PR10339
10206 442. http://gcc.gnu.org/PR11350
10207 443. http://gcc.gnu.org/PR12826
10208 444. http://gcc.gnu.org/PR12500
10209 445. http://gcc.gnu.org/PR12941
10210 446. http://gcc.gnu.org/PR12953
10211 447. http://gcc.gnu.org/PR13041
10212 448. http://gcc.gnu.org/PR13507
10213 449. http://gcc.gnu.org/PR13382
10214 450. http://gcc.gnu.org/PR13394
10215 451. http://gcc.gnu.org/PR13400
10216 452. http://gcc.gnu.org/PR13521
10217 453. http://gcc.gnu.org/PR2094
10218 454. http://gcc.gnu.org/PR2294
10219 455. http://gcc.gnu.org/PR5050
10220 456. http://gcc.gnu.org/PR9371
10221 457. http://gcc.gnu.org/PR9546
10222 458. http://gcc.gnu.org/PR10081
10223 459. http://gcc.gnu.org/PR10093
10224 460. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#61
10225 461. http://gcc.gnu.org/PR10095
10226 462. http://gcc.gnu.org/PR11554
10227 463. http://gcc.gnu.org/PR12297
10228 464. http://gcc.gnu.org/PR12352
10229 465. http://gcc.gnu.org/PR12438
10230 466. http://gcc.gnu.org/PR12540
10231 467. http://gcc.gnu.org/PR12594
10232 468. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#60
10233 469. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#63
10234 470. http://gcc.gnu.org/PR12657
10235 471. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#292
10236 472. http://gcc.gnu.org/PR12696
10237 473. http://gcc.gnu.org/PR12815
10238 474. http://gcc.gnu.org/PR12862
10239 475. http://gcc.gnu.org/PR12926
10240 476. http://gcc.gnu.org/PR12967
10241 477. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html
10242 478. http://gcc.gnu.org/PR12971
10243 479. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#328
10244 480. http://gcc.gnu.org/PR13007
10245 481. http://gcc.gnu.org/PR13009
10246 482. http://gcc.gnu.org/PR13057
10247 483. http://gcc.gnu.org/PR13070
10248 484. http://gcc.gnu.org/PR13081
10249 485. http://gcc.gnu.org/PR13239
10250 486. http://gcc.gnu.org/PR13262
10251 487. http://gcc.gnu.org/PR13290
10252 488. http://gcc.gnu.org/PR13323
10253 489. http://gcc.gnu.org/PR13369
10254 490. http://gcc.gnu.org/PR13371
10255 491. http://gcc.gnu.org/PR13445
10256 492. http://gcc.gnu.org/PR13461
10257 493. http://gcc.gnu.org/PR13462
10258 494. http://gcc.gnu.org/PR13478
10259 495. http://gcc.gnu.org/PR13544
10260 496. http://gcc.gnu.org/PR13650
10261 497. http://gcc.gnu.org/PR13683
10262 498. http://gcc.gnu.org/PR13688
10263 499. http://gcc.gnu.org/PR13774
10264 500. http://gcc.gnu.org/PR13884
10265 501. http://gcc.gnu.org/PR10746
10266 502. http://gcc.gnu.org/PR11433
10267 503. http://gcc.gnu.org/PR12633
10268 504. http://gcc.gnu.org/PR13037
10269 505. http://gcc.gnu.org/PR13213
10270 506. http://gcc.gnu.org/PR4490
10271 507. http://gcc.gnu.org/PR12292
10272 508. http://gcc.gnu.org/PR12441
10273 509. http://gcc.gnu.org/PR12943
10274 510. http://gcc.gnu.org/PR13608
10275 511. http://gcc.gnu.org/PR11598
10276 512. http://gcc.gnu.org/PR11793
10277 513. http://gcc.gnu.org/PR12467
10278 514. http://gcc.gnu.org/PR12537
10279 515. http://gcc.gnu.org/PR12496
10280 516. http://gcc.gnu.org/PR12865
10281 517. http://gcc.gnu.org/PR13354
10282 518. http://gcc.gnu.org/PR10467
10283 519. http://gcc.gnu.org/PR11226
10284 520. http://gcc.gnu.org/PR11227
10285 521. http://gcc.gnu.org/PR12644
10286 522. http://gcc.gnu.org/PR13149
10287 523. http://gcc.gnu.org/PR12654
10288 524. http://gcc.gnu.org/PR12965
10289 525. http://gcc.gnu.org/PR13031
10290 526. http://gcc.gnu.org/PR11634
10291 527. http://gcc.gnu.org/PR12158
10292 528. http://gcc.gnu.org/PR11992
10293 529. http://gcc.gnu.org/PR9365
10294 530. http://gcc.gnu.org/PR10392
10295 531. http://gcc.gnu.org/PR11322
10296 532. http://gcc.gnu.org/PR13069
10297 533. http://gcc.gnu.org/PR13302
10298 534. http://gcc.gnu.org/PR13585
10299 535. http://gcc.gnu.org/PR8916
10300 536. http://gcc.gnu.org/PR11576
10301 537. http://gcc.gnu.org/PR13122
10302 538. http://gcc.gnu.org/PR13256
10303 539. http://gcc.gnu.org/PR13373
10304 540. http://gcc.gnu.org/PR12561
10305 541. http://gcc.gnu.org/PR6243
10306 542. http://gcc.gnu.org/PR11397
10307 543. http://gcc.gnu.org/PR12505
10308 544. http://gcc.gnu.org/PR13150
10309 545. http://gcc.gnu.org/PR12666
10310 546. http://gcc.gnu.org/PR12969
10311 547. http://gcc.gnu.org/PR10819
10312 548. http://gcc.gnu.org/PR11612
10313 549. http://gcc.gnu.org/PR13211
10314 550. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.3.4
10315 551. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.3.5
10316 552. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.3.6
10317 553. http://gcc.gnu.org/onlinedocs/
10318 554. mailto:gcc-help@gcc.gnu.org
10319 555. mailto:gcc@gcc.gnu.org
10320 556. http://gcc.gnu.org/lists.html
10321 557. http://www.fsf.org/
10322 558. http://gcc.gnu.org/about.html
10323 559. http://validator.w3.org/check/referer
10324 ======================================================================
10325 http://gcc.gnu.org/gcc-3.2/index.html
10327 GCC 3.2 Release Series
10331 The [1]GNU project and the GCC developers are pleased to announce the
10332 release of GCC 3.2.3.
10334 The purpose of the GCC 3.2 release series is to provide a stable
10335 platform for OS distributors to use building their next releases. A
10336 primary objective was to stabilize the C++ ABI; we believe that the
10337 interface to the compiler and the C++ standard library are now
10340 Be aware that C++ code compiled by GCC 3.2.x will (in general) not
10341 interoperate with code compiled by GCC 3.1.1 or earlier.
10343 Please refer to our [2]detailed list of news, caveats, and bug-fixes
10344 for further information.
10349 April 25, 2003 ([3]changes)
10352 February 5, 2003 ([4]changes)
10355 November 19, 2002 ([5]changes)
10358 August 14, 2002 ([6]changes)
10360 References and Acknowledgements
10362 GCC used to stand for the GNU C Compiler, but since the compiler
10363 supports several other languages aside from C, it now stands for the
10364 GNU Compiler Collection.
10366 A list of [7]successful builds is updated as new information becomes
10369 The GCC developers would like to thank the numerous people that have
10370 contributed new features, improvements, bug fixes, and other changes as
10371 well as test results to GCC. This [8]amazing group of volunteers is
10372 what makes GCC successful.
10374 For additional information about GCC please refer to the [9]GCC project
10375 web site or contact the [10]GCC development mailing list.
10377 To obtain GCC please use [11]our mirror sites, or our CVS server.
10380 For questions related to the use of GCC, please consult these web
10381 pages and the [12]GCC manuals. If that fails, the
10382 [13]gcc-help@gcc.gnu.org mailing list might help. Comments on these
10383 web pages and the development of GCC are welcome on our developer
10384 list at [14]gcc@gcc.gnu.org. All of [15]our lists have public
10387 Copyright (C) [16]Free Software Foundation, Inc. Verbatim copying and
10388 distribution of this entire article is permitted in any medium,
10389 provided this notice is preserved.
10391 These pages are [17]maintained by the GCC team. Last modified
10396 1. http://www.gnu.org/
10397 2. http://gcc.gnu.org/gcc-3.2/changes.html
10398 3. http://gcc.gnu.org/gcc-3.2/changes.html#3.2.3
10399 4. http://gcc.gnu.org/gcc-3.2/changes.html#3.2.2
10400 5. http://gcc.gnu.org/gcc-3.2/changes.html#3.2.1
10401 6. http://gcc.gnu.org/gcc-3.2/changes.html#3.2
10402 7. http://gcc.gnu.org/gcc-3.2/buildstat.html
10403 8. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
10404 9. http://gcc.gnu.org/index.html
10405 10. mailto:gcc@gcc.gnu.org
10406 11. http://gcc.gnu.org/mirrors.html
10407 12. http://gcc.gnu.org/onlinedocs/
10408 13. mailto:gcc-help@gcc.gnu.org
10409 14. mailto:gcc@gcc.gnu.org
10410 15. http://gcc.gnu.org/lists.html
10411 16. http://www.fsf.org/
10412 17. http://gcc.gnu.org/about.html
10413 18. http://validator.w3.org/check/referer
10414 ======================================================================
10415 http://gcc.gnu.org/gcc-3.2/changes.html
10417 GCC 3.2 Release Series
10418 Changes, New Features, and Fixes
10420 The latest release in the 3.2 release series is [1]GCC 3.2.3.
10422 Caveats and New Features
10426 * The C++ compiler does not correctly zero-initialize
10427 pointers-to-data members. You must explicitly initialize them. For
10428 example: int S::*m(0); will work, but depending on
10429 default-initialization to zero will not work. This bug cannot be
10430 fixed in GCC 3.2 without inducing unacceptable risks. It will be
10432 * This GCC release is based on the GCC 3.1 sourcebase, and thus has
10433 all the [2]changes in the GCC 3.1 series. In addition, GCC 3.2 has
10434 a number of C++ ABI fixes which make its C++ compiler generate
10435 binary code which is incompatible with the C++ compilers found in
10436 earlier GCC releases, including GCC 3.1 and GCC 3.1.1.
10438 Frontend Enhancements
10442 * The method of constructing the list of directories to be searched
10443 for header files has been revised. If a directory named by a -I
10444 option is a standard system include directory, the option is
10445 ignored to ensure that the default search order for system
10446 directories and the special treatment of system header files are
10448 * The C and Objective-C compilers no longer accept the "Naming Types"
10449 extension (typedef foo = bar); it was already unavailable in C++.
10450 Code which uses it will need to be changed to use the "typeof"
10451 extension instead: typedef typeof(bar) foo. (We have removed this
10452 extension without a period of deprecation because it has caused the
10453 compiler to crash since version 3.0 and no one noticed until very
10454 recently. Thus we conclude it is not in widespread use.)
10458 * GCC 3.2 fixed serveral differences between the C++ ABI implemented
10459 in GCC and the multi-vendor standard, but more have been found
10460 since the release. 3.2.1 adds a new warning, -Wabi, to warn about
10461 code which is affected by these bugs. We will fix these bugs in
10462 some future release, once we are confident that all have been
10463 found; until then, it is our intention to make changes to the ABI
10464 only if they are necessary for correct compilation of C++, as
10465 opposed to conformance to the ABI documents.
10466 * For details on how to build an ABI compliant compiler for GNU/Linux
10467 systems, check the [3]common C++ ABI page.
10469 New Targets and Target Specific Improvements
10473 * Fixed a number of bugs in SSE and MMX intrinsics.
10474 * Fixed common compiler crashes with SSE instruction set enabled
10475 (implied by -march=pentium3, pentium4, athlon-xp)
10476 * __m128 and __m128i is not 128bit aligned when used in structures.
10480 * A bug whereby the compiler could generate bad code for bzero has
10482 * ABI fixes (implying ABI incompatibilities with previous version in
10484 * Fixed prefetch code generation
10485 __________________________________________________________________
10489 3.2.3 is a bug fix release only; there are no new features that were
10490 not present in GCC 3.2.2.
10494 This section lists the problem reports (PRs) from GCC's bug tracking
10495 system that are known to be fixed in the 3.2.3 release. This list might
10496 not be complete (that is, it is possible that some PRs that have been
10497 fixed are not listed here), and some of the titles have been changed to
10498 make them more clear.
10500 Internal Compiler Errors (multi-platform)
10502 * [4]3782: (c++) -quiet -fstats produces a segmentation fault in
10504 * [5]6440: (c++) template specializations cause ICE
10505 * [6]7050: (c++) ICE on: (i ? get_string() : throw)
10506 * [7]7741: ICE on conflicting types (make_decl_rtl in varasm.c)
10507 * [8]7982: (c++) ICE due to infinite recursion (using STL set)
10508 * [9]8068: exceedingly high (infinite) memory usage
10509 * [10]8178: ICE with __builtin_ffs
10510 * [11]8396: ICE in copy_to_mode_reg, in explow.c
10511 * [12]8674: (c++) ICE in cp_expr_size, in cp/cp-lang.c
10512 * [13]9768: ICE when optimizing inline code at -O2
10513 * [14]9798: (c++) Infinite recursion (segfault) in
10514 cp/decl.c:push_using_directive with recursive using directives
10515 * [15]9799: mismatching structure initializer with nested flexible
10517 * [16]9928: ICE on duplicate enum declaration
10518 * [17]10114: ICE in mem_loc_descriptor, in dwarf2out.c (affects
10520 * [18]10352: ICE in find_reloads_toplev
10521 * [19]10336: ICE with -Wunreachable-code
10525 * [20]8224: Incorrect joining of signed and unsigned division
10526 * [21]8613: -O2 produces wrong code with builtin strlen and
10528 * [22]8828: gcc reports some code is unreachable when it is not
10529 * [23]9226: GCSE breaking argument passing
10530 * [24]9853: miscompilation of non-constant structure initializer
10531 * [25]9797: C99-style struct initializers are miscompiled
10532 * [26]9967: Some standard C function calls should not be replaced
10533 when optimizing for size
10534 * [27]10116: ce2: invalid merge of join_bb in the context of switch
10536 * [28]10171: wrong code for inlined function
10537 * [29]10175: -Wunreachable-code doesn't work for single lines
10539 C++ compiler and library:
10541 * [30]8316: Confusing diagnostic for code that misuses conversion
10543 * [31]9169: filebuf output fails if codecvt<>::out returns noconv
10544 * [32]9420: incomplete type incorrectly reported
10545 * [33]9459: typeof in return type specification of template not
10547 * [34]9507: filebuf::open handles ios_base::ate incorrectly
10548 * [35]9538: Out-of-bounds memory access in streambuf::sputbackc
10549 * [36]9602: Total confusion about template/friend/virtual/abstract
10550 * [37]9993: destructor not called for local object created within and
10551 returned from infinite loop
10552 * [38]10167: ieee_1003.1-2001 locale specialisations on a glibc-2.3.2
10555 Java compiler and library:
10557 * [39]9652: libgcj build fails on irix6.5.1[78]
10558 * [40]10144: gas on solaris complains about bad .stabs lines for
10559 java, native as unaffected
10561 x86-specific (Intel/AMD):
10563 * [41]8746: gcc miscompiles Linux kernel ppa driver on x86
10564 * [42]9888: -mcpu=k6 -Os produces out of range loop instructions
10565 * [43]9638: Cross-build for target i386-elf and i586-pc-linux-gnu
10567 * [44]9954: Cross-build for target i586-pc-linux-gnu (--with-newlib)
10572 * [45]7784: [Sparc] ICE in extract_insn, in recog.c
10573 * [46]7796: sparc extra failure with -m64 on execute/930921-1.c in
10575 * [47]8281: ICE when compiling with -O2 -fPIC for Ultrasparc
10576 * [48]8366: [Sparc] C testsuite failure with -m64 -fpic -O in
10578 * [49]8726: gcc -O2 miscompiles Samba 2.2.7 on 32-bit sparc
10579 * [50]9414: Scheduling bug on Ultrasparc
10580 * [51]10067: GCC-3.2.2 outputs invalid asm on sparc64
10584 * [52]7248: broken "inclusive or" code
10585 * [53]8343: m68k-elf/rtems ICE at instantiate_virtual_regs_1
10589 * [54]9732: Wrong code with -O2 -fPIC
10590 * [55]10073: ICE: powerpc cannot split insn
10594 * [56]7702: optimization problem on a DEC alpha under OSF1
10595 * [57]9671: gcc.3.2.2 does not build on a HP Tru64 Unix v5.1B system
10599 * [58]8694: <string> breaks <ctype.h> on HP-UX 10.20 (DUP: 9275)
10600 * [59]9953: (ada) gcc 3.2.x can't build 3.3-branch ada on HP-UX 10
10602 * [60]10271: Floating point args don't get reloaded across function
10607 * [61]6362: mips-irix6 gcc-3.1 C testsuite failure with -mips4 in
10612 * [62]10377: gcc-3.2.2 creates bad assembler code for cris
10614 Miscellaneous and minor bugs:
10616 * [63]6955: collect2 says "core dumped" when there is no core
10617 __________________________________________________________________
10621 Beginning with 3.2.2, GCC's Makefile suite supports redirection of make
10622 install by means of the DESTDIR variable. Parts of the GCC tree have
10623 featured that support long before, but now it is available even from
10626 Other than that, GCC 3.2.2 is a bug fix release only; there are no new
10627 features that were not present in GCC 3.2.1.
10631 On the following i386-based systems GCC 3.2.1 broke the C ABI wrt.
10632 functions returning structures: Cygwin, FreeBSD (GCC 3.2.1 as shipped
10633 with FreeBSD 5.0 does not have this problem), Interix, a.out-based
10634 GNU/Linux and NetBSD, OpenBSD, and Darwin. GCC 3.2.2 reverts this ABI
10635 change, and thus restores ABI-compatibility with previous releases
10636 (except GCC 3.2.1) on these platforms.
10638 This section lists the problem reports (PRs) from GCC's bug tracking
10639 system that are known to be fixed in the 3.2.2 release. This list might
10640 not be complete (that is, it is possible that some PRs that have been
10641 fixed are not listed here) and some of the titles have been changed to
10642 make them more clear.
10644 Internal Compiler Errors (multi-platform)
10646 * [64]5919: (c++) ICE when passing variable array to template
10648 * [65]7129: (c++) ICE with min/max assignment operators (<?= and >?=)
10649 * [66]7507: ICE with -O2 when address of called function is a
10650 complicated expression
10651 * [67]7622: ICE with nested inline functions if function's address is
10653 * [68]7681: (fortran) ICE in compensate_edge, in reg-stack.c (also PR
10655 * [70]8031: (c++) ICE in code comparing typeids and casting from
10657 * [71]8275: ICE in simplify_subreg
10658 * [72]8332: (c++) builtin strlen/template interaction causes ICE
10659 * [73]8372: (c++) ICE on explicit call of destructor
10660 * [74]8439: (c, not c++) empty struct causes ICE
10661 * [75]8442: (c++) ICE with nested template classes
10662 * [76]8518: ICE when compiling mplayer ("extern inline" issue)
10663 * [77]8615: (c++) ICE with out-of-range character constant template
10665 * [78]8663: (c++) ICE in cp_expr_size, at cp-lang.c:307
10666 * [79]8799: (c++) ICE: error reporting routines re-entered
10667 * [80]9328: (c++) ICE with typeof(X) for overloaded X
10668 * [81]9465: (preprocessor) cpp -traditional ICE on null bytes
10670 C++ (compiler and library) bugs
10672 * [82]47: scoping in nested classes is broken
10673 * [83]6745: problems with iostream rdbuf() member function
10674 * [84]8214: conversion from const char* const to char* sometimes
10676 * [85]8493: builtin strlen and overload resolution (same bug as
10678 * [87]8503: strange behaviour of function types
10679 * [88]8727: compiler confused by inheritance from an anonymous struct
10680 * [89]7445: poor performance of std::locale::classic() in
10681 multi-threaded applications
10682 * [90]8230: mishandling of overflow in vector<T>::resize
10683 * [91]8399: sync_with_stdio(false) breaks unformatted input
10684 * [92]8662: illegal access of private member of unnamed class is
10686 * [93]8707: "make distclean" fails in libstdc++-v3 directory
10687 * [94]8708: __USE_MALLOC doesn't work
10688 * [95]8790: Use of non-thread-safe strtok in src/localename.cc
10689 * [96]8887: Bug in date formats with --enable-clocale=generic
10690 * [97]9076: Call Frame Instructions are not handled correctly during
10692 * [98]9151: std::setprecision limited to 16 digits when outputting a
10694 * [99]9168: codecvt<char, char, mbstate_t> overwrites output buffers
10695 * [100]9269: libstdc++ headers: explicit specialization of function
10696 must precede its first use
10697 * [101]9322: return value of basic_streambuf<>::getloc affected by
10699 * [102]9433: segfault in runtime support for dynamic_cast
10701 C and optimizer bugs
10703 * [103]8032: GCC incorrectly initializes static structs that have
10705 * [104]8639: simple arithmetic expression broken
10706 * [105]8794: optimization improperly eliminates certain expressions
10707 * [106]8832: traditional "asm volatile" code is illegally optimized
10708 * [107]8988: loop optimizer bug: with -O2, code is generated that
10709 segfaults (found on i386, bug present for all platforms)
10710 * [108]9492: structure copy clobbers subsequent stores to structure
10714 * [109]9267: Objective-C parser won't build with newer bison versions
10719 * [110]8344: Ada build problem due to conflict between gcc/final.o,
10724 * [111]8524: _Pragma within macros is improperly expanded
10725 * [112]8880: __WCHAR_TYPE__ macro incorrectly set to "long int" with
10730 * [113]9090: arm ICE with >= -O2; regression from gcc-2.95
10732 x86-specific (Intel/AMD)
10734 * [114]8588: ICE in extract_insn, at recog.c:NNNN (shift instruction)
10735 * [115]8599: loop unroll bug with -march=k6-3
10736 * [116]9506: ABI breakage in structure return (affects BSD and
10737 Cygwin, but not GNU/Linux)
10739 FreeBSD 5.0 specific
10741 * [117]9484: GCC 3.2.1 Bootstrap failure on FreeBSD 5.0
10745 * [118]9292: hppa1.1-rtems configurery problems
10746 * [119]9293: [m68k-elf/rtems] config/m68k/t-crtstuff bug
10747 * [120]9295: [mips-rtems] config/mips/rtems.h init/fini issue
10748 * [121]9296: gthr-rtems regression
10749 * [122]9316: powerpc-rtems: extending multilibs
10753 * [123]9493: ICE with -O2 when building a simple function
10757 * [124]7341: hyperlink to gcov in GCC documentation doesn't work
10758 * [125]8947: Please add a warning about "-malign-double" in docs
10759 * [126]7448, [127]8882: typo cleanups
10760 __________________________________________________________________
10764 3.2.1 adds a new warning, -Wabi. This option warns when GNU C++
10765 generates code that is known not to be binary-compatible with the
10766 vendor-neutral ia32/ia64 ABI. Please consult the GCC manual, included
10767 in the distribution, for details.
10769 This release also removes an old GCC extension, "naming types", and the
10770 documentation now directs users to use a different GCC extension,
10771 __typeof__, instead. The feature had evidently been broken for a while.
10773 Otherwise, 3.2.1 is a bug fix release only; other than bug fixes and
10774 the new warning there are no new features that were not present in GCC
10777 In addition, the previous fix for [128]PR 7445 (poor performance of
10778 std::locale::classic() in multi-threaded applications) was reverted
10779 ("unfixed"), because the "fix" was not thread-safe.
10783 This section lists the problem reports (PRs) from GCC's bug tracking
10784 system that are known to be fixed in the 3.2.1 release. This list might
10785 not be complete (that is, it is possible that some PRs that have been
10786 fixed are not listed here). As you can see, the number of bug fixes is
10787 quite large, so it is strongly recommended that users of earlier GCC
10788 3.x releases upgrade to GCC 3.2.1.
10790 Internal Compiler Errors (multi-platform)
10792 * [129]2521: (c++) ICE in build_ptrmemfunc, in cp/typeck.c
10793 * [130]5661: (c++) ICE instantiating template on array of unknown
10795 * [131]6419: (c++) ICE in make_decl_rtl for "longest" attribute on
10797 * [132]6994: (c++) ICE in find_function_data
10798 * [133]7150: preprocessor: GCC -dM -E gives an ICE
10799 * [134]7160: ICE when optimizing branches without a return value
10800 * [135]7228: (c++) ICE when using member template and template
10802 * [136]7266: (c++) ICE with -pedantic on missing typename
10803 * [137]7353: ICE from use of "Naming Types" extension, see above
10804 * [138]7411: ICE in instantiate_virtual_regs_1, in function.c
10805 * [139]7478: (c++) ICE on static_cast inside template
10806 * [140]7526: preprocessor core dump when _Pragma implies #pragma
10808 * [141]7721: (c++) ICE on simple (but incorrect) template ([142]7803
10810 * [143]7754: (c++) ICE on union with template parameter
10811 * [144]7788: (c++) redeclaring a definition as an incomplete class
10813 * [145]8031: (c++) ICE in comptypes, in cp/typeck.c
10814 * [146]8055: preprocessor dies with SIG11 when building FreeBSD
10816 * [147]8067: (c++) ICE due to mishandling of __FUNCTION__ and related
10818 * [148]8134: (c++) ICE in force_store_init_value on legal code
10819 * [149]8149: (c++) ICE on incomplete type
10820 * [150]8160: (c++) ICE in build_modify_expr, in cp/typeck.c: array
10823 C++ (compiler and library) bugs
10825 * [151]5607: No pointer adjustment in covariant return types
10826 * [152]6579: Infinite loop with statement expressions in member
10828 * [153]6803: Default copy constructor bug in GCC 3.1
10829 * [154]7176: g++ confused by friend and static member with same name
10830 * [155]7188: Segfault with template class and recursive (incorrect)
10832 * [156]7306: Regression: GCC 3.x fails to compile code with virtual
10833 inheritance if a method has a variable number of arguments
10834 * [157]7461: ctype<char>::classic_table() returns offset array on
10836 * [158]7524: f(const float arg[3]) fails
10837 * [159]7584: Erroneous ambiguous base error on using declaration
10838 * [160]7676: Member template overloading problem
10839 * [161]7679: infinite loop when a right parenthesis is missing
10840 * [162]7811: default locale not taken from environment
10841 * [163]7961: compare( char *) implemented incorrectly in
10843 * [164]8071: basic_ostream::operator<<(streambuf*) loops forever if
10844 streambuf::underflow() leaves gptr() NULL (dups: [165]8127,
10846 * [167]8096: deque::at() throws std::range_error instead of
10848 * [168]8127: cout << cin.rdbuf() infinite loop
10849 * [169]8218: Excessively large memory consumed for classes with large
10851 * [170]8287: GCC 3.2: Destructor called for non-constructed local
10853 * [171]8347: empty vector range used in string construction causes
10855 * [172]8348: fail() flag is set in istringstream when eof() flag is
10857 * [173]8391: regression: infinite loop in cp/decl2.c(finish_file)
10859 C and optimizer bugs
10861 * [174]6627: -fno-align-functions doesn't seem to disable function
10863 * [175]6631: life_analysis misoptimizes code to initialize fields of
10865 * [176]7102: unsigned char division results in floating exception
10866 * [177]7120: Run once loop should *always* be unrolled
10868 * [178]7209: Bug involving array referencing and ?: operator
10869 * [179]7515: invalid inlining of global function with -O3
10870 * [180]7814: incorrect scheduling for glibc-2.2.92 strcpy test
10871 * [181]8467: bug in sibling call optimization
10875 * [182]4890: incorrect line markers from the traditional preprocessor
10876 * [183]7357: -M option omits system headers files (making it the same
10878 * [184]7358: Changes to Sun's make Dependencies
10879 * [185]7602: C++ header files found in CPLUS_INCLUDE_PATH treated as
10881 * [186]7862: Interrupting GCC -MD removes .d file but not .o
10882 * [187]8190: Failed compilation deletes -MD dependency file
10883 * [188]8524: _Pragma within macro is improperly expanded
10885 x86 specific (Intel/AMD)
10887 * [189]5351: (i686-only) function pass-by-value structure copy
10888 corrupts stack ([190]7591 is a duplicate)
10889 * [191]6845, [192]7034, [193]7124, [194]7174: ICE's with
10890 -march=pentium3/pentium2/athlon (these are all the same underlying
10891 bug, in MMX register use)
10892 * [195]7134, [196]7375, [197]7390: ICE with -march=athlon (maybe same
10894 * [198]6890: xmmintrin.h, _MM_TRANSPOSE4_PS is broken
10895 * [199]6981: wrong code in 64-bit manipulation on x86
10896 * [200]7242: GCC -mcpu=pentium[23] doesn't define __tune_pentiumpro__
10898 * [201]7396: ix86: cmpgt_ss, cmpge_ss, cmpngt_ss, and cmpnge_ss SSE
10899 intrinsics are broken
10900 * [202]7630: GCC 3.2 breaks on Mozilla 1.0's JS sources with
10902 * [203]7693: Typo in i386 mmintrin.h header
10903 * [204]7723: ICE - Pentium3 sse - GCC 3.2
10904 * [205]7951: ICE on -march=pentium4 -O2 -mfpmath=sse
10905 * [206]8146: (i686 only) gcc 3.2 miscompiles gcc 2.95.3
10909 * [207]5967: GCC bug when profiling nested functions on powerpc
10910 * [208]6984: wrong code generated with -O2, -O3, -Os for do-while
10912 * [209]7114: PowerPC: ICE building strcoll.op from glibc-2.2.5
10913 * [210]7130: miscompiled code for GCC-3.1 on
10914 powerpc-unknown-linux-gnu with -funroll-all-loops
10915 * [211]7133: PowerPC ICE: unrecognizable insn
10916 * [212]7380: ICE in extract_insn, at recog.c:2148
10917 * [213]8252: ICE on Altivec code with optimization turned on
10918 * [214]8451: Altivec ICE in GCC 3.2
10922 * [215]7250: __ashrdi3 returns wrong value on 32 bit hppa
10926 * [216]6668: when using --disable-multilib, libgcc_s.so is installed
10927 in the wrong place on sparc-solaris
10928 * [217]7151: ICE when compiling for UltraSPARC
10929 * [218]7335: SPARC: ICE in verify_wide_reg (flow.c:557) with long
10931 * [219]7842: [REGRESSION] SPARC code gen bug
10935 * [220]7856: [arm] invalid offset in constant pool reference
10936 * [221]7967: optimization produces wrong code (ARM)
10940 * [222]7374: __builtin_fabsl broken on alpha
10944 * [223]7370: ICE in fixup_var_refs_1 on s390x
10945 * [224]7409: loop optimization bug on s390x-linux-gnu
10946 * [225]8232: s390x: ICE when using bcmp with int length argument
10950 * [226]7623: SCO OpenServer build fails with machmode.def: undefined
10951 symbol: BITS_PER_UNIT
10953 m68k/Coldfire specific
10955 * [227]8314: crtbegin, crtend need to be multilib'ed for this
10960 * [228]761: Document some undocumented options
10961 * [229]5610: Fix documentation about invoking SSE instructions
10963 * [230]7484: List -Wmissing-declarations as C-only option
10964 * [231]7531: -mcmodel not documented for x86-64
10965 * [232]8120: Update documentation of bad use of ##
10966 __________________________________________________________________
10970 3.2 is a small bug fix release, but there is a change to the
10971 application binary interface (ABI), hence the change to the second part
10972 of the version number.
10974 The main purpose of the 3.2 release is to correct a couple of problems
10975 in the C++ ABI, with the intention of providing a stable interface
10976 going forward. Accordingly, 3.2 is only a small change to 3.1.1.
10982 * [233]7320: g++ 3.2 relocation problem
10983 * [234]7470: vtable: virtual function pointers not in declaration
10988 * [235]6410: Trouble with non-ASCII monetary symbols and wchar_t
10989 * [236]6503, [237]6642, [238]7186: Problems with comparing or
10990 subtracting various types of const and non-const iterators
10991 * [239]7216: ambiguity with basic_iostream::traits_type
10992 * [240]7220: problem with basic_istream::ignore(0,delimiter)
10993 * [241]7222: locale::operator==() doesn't work on std::locale("")
10994 * [242]7286: placement operator delete issue
10995 * [243]7442: cxxabi.h does not match the C++ ABI
10996 * [244]7445: poor performance of std::locale::classic() in
10997 multi-threaded applications
11001 * [245]7291: off-by-one in generated inline bzero code for x86-64
11004 For questions related to the use of GCC, please consult these web
11005 pages and the [246]GCC manuals. If that fails, the
11006 [247]gcc-help@gcc.gnu.org mailing list might help. Comments on these
11007 web pages and the development of GCC are welcome on our developer
11008 list at [248]gcc@gcc.gnu.org. All of [249]our lists have public
11011 Copyright (C) [250]Free Software Foundation, Inc. Verbatim copying and
11012 distribution of this entire article is permitted in any medium,
11013 provided this notice is preserved.
11015 These pages are [251]maintained by the GCC team. Last modified
11020 1. http://gcc.gnu.org/gcc-3.2/changes.html#3.2.3
11021 2. http://gcc.gnu.org/gcc-3.1/changes.html
11022 3. http://gcc.gnu.org/gcc-3.2/c++-abi.html
11023 4. http://gcc.gnu.org/PR3782
11024 5. http://gcc.gnu.org/PR6440
11025 6. http://gcc.gnu.org/PR7050
11026 7. http://gcc.gnu.org/PR7741
11027 8. http://gcc.gnu.org/PR7982
11028 9. http://gcc.gnu.org/PR8068
11029 10. http://gcc.gnu.org/PR8178
11030 11. http://gcc.gnu.org/PR8396
11031 12. http://gcc.gnu.org/PR8674
11032 13. http://gcc.gnu.org/PR9768
11033 14. http://gcc.gnu.org/PR9798
11034 15. http://gcc.gnu.org/PR9799
11035 16. http://gcc.gnu.org/PR9928
11036 17. http://gcc.gnu.org/PR10114
11037 18. http://gcc.gnu.org/PR10352
11038 19. http://gcc.gnu.org/PR10336
11039 20. http://gcc.gnu.org/PR8224
11040 21. http://gcc.gnu.org/PR8613
11041 22. http://gcc.gnu.org/PR8828
11042 23. http://gcc.gnu.org/PR9226
11043 24. http://gcc.gnu.org/PR9853
11044 25. http://gcc.gnu.org/PR9797
11045 26. http://gcc.gnu.org/PR9967
11046 27. http://gcc.gnu.org/PR10116
11047 28. http://gcc.gnu.org/PR10171
11048 29. http://gcc.gnu.org/PR10175
11049 30. http://gcc.gnu.org/PR8316
11050 31. http://gcc.gnu.org/PR9169
11051 32. http://gcc.gnu.org/PR9420
11052 33. http://gcc.gnu.org/PR9459
11053 34. http://gcc.gnu.org/PR9507
11054 35. http://gcc.gnu.org/PR9538
11055 36. http://gcc.gnu.org/PR9602
11056 37. http://gcc.gnu.org/PR9993
11057 38. http://gcc.gnu.org/PR10167
11058 39. http://gcc.gnu.org/PR9652
11059 40. http://gcc.gnu.org/PR10144
11060 41. http://gcc.gnu.org/PR8746
11061 42. http://gcc.gnu.org/PR9888
11062 43. http://gcc.gnu.org/PR9638
11063 44. http://gcc.gnu.org/PR9954
11064 45. http://gcc.gnu.org/PR7784
11065 46. http://gcc.gnu.org/PR7796
11066 47. http://gcc.gnu.org/PR8281
11067 48. http://gcc.gnu.org/PR8366
11068 49. http://gcc.gnu.org/PR8726
11069 50. http://gcc.gnu.org/PR9414
11070 51. http://gcc.gnu.org/PR10067
11071 52. http://gcc.gnu.org/PR7248
11072 53. http://gcc.gnu.org/PR8343
11073 54. http://gcc.gnu.org/PR9732
11074 55. http://gcc.gnu.org/PR10073
11075 56. http://gcc.gnu.org/PR7702
11076 57. http://gcc.gnu.org/PR9671
11077 58. http://gcc.gnu.org/PR8694
11078 59. http://gcc.gnu.org/PR9953
11079 60. http://gcc.gnu.org/PR10271
11080 61. http://gcc.gnu.org/PR6362
11081 62. http://gcc.gnu.org/PR10377
11082 63. http://gcc.gnu.org/PR6955
11083 64. http://gcc.gnu.org/PR5919
11084 65. http://gcc.gnu.org/PR7129
11085 66. http://gcc.gnu.org/PR7507
11086 67. http://gcc.gnu.org/PR7622
11087 68. http://gcc.gnu.org/PR7681
11088 69. http://gcc.gnu.org/PR9528
11089 70. http://gcc.gnu.org/PR8031
11090 71. http://gcc.gnu.org/PR8275
11091 72. http://gcc.gnu.org/PR8332
11092 73. http://gcc.gnu.org/PR8372
11093 74. http://gcc.gnu.org/PR8439
11094 75. http://gcc.gnu.org/PR8442
11095 76. http://gcc.gnu.org/PR8518
11096 77. http://gcc.gnu.org/PR8615
11097 78. http://gcc.gnu.org/PR8663
11098 79. http://gcc.gnu.org/PR8799
11099 80. http://gcc.gnu.org/PR9328
11100 81. http://gcc.gnu.org/PR9465
11101 82. http://gcc.gnu.org/PR47
11102 83. http://gcc.gnu.org/PR6745
11103 84. http://gcc.gnu.org/PR8214
11104 85. http://gcc.gnu.org/PR8493
11105 86. http://gcc.gnu.org/PR8332
11106 87. http://gcc.gnu.org/PR8503
11107 88. http://gcc.gnu.org/PR8727
11108 89. http://gcc.gnu.org/PR7445
11109 90. http://gcc.gnu.org/PR8230
11110 91. http://gcc.gnu.org/PR8399
11111 92. http://gcc.gnu.org/PR8662
11112 93. http://gcc.gnu.org/PR8707
11113 94. http://gcc.gnu.org/PR8708
11114 95. http://gcc.gnu.org/PR8790
11115 96. http://gcc.gnu.org/PR8887
11116 97. http://gcc.gnu.org/PR9076
11117 98. http://gcc.gnu.org/PR9151
11118 99. http://gcc.gnu.org/PR9168
11119 100. http://gcc.gnu.org/PR9269
11120 101. http://gcc.gnu.org/PR9322
11121 102. http://gcc.gnu.org/PR9433
11122 103. http://gcc.gnu.org/PR8032
11123 104. http://gcc.gnu.org/PR8639
11124 105. http://gcc.gnu.org/PR8794
11125 106. http://gcc.gnu.org/PR8832
11126 107. http://gcc.gnu.org/PR8988
11127 108. http://gcc.gnu.org/PR9492
11128 109. http://gcc.gnu.org/PR9267
11129 110. http://gcc.gnu.org/PR8344
11130 111. http://gcc.gnu.org/PR8524
11131 112. http://gcc.gnu.org/PR8880
11132 113. http://gcc.gnu.org/PR9090
11133 114. http://gcc.gnu.org/PR8588
11134 115. http://gcc.gnu.org/PR8599
11135 116. http://gcc.gnu.org/PR9506
11136 117. http://gcc.gnu.org/PR9484
11137 118. http://gcc.gnu.org/PR9292
11138 119. http://gcc.gnu.org/PR9293
11139 120. http://gcc.gnu.org/PR9295
11140 121. http://gcc.gnu.org/PR9296
11141 122. http://gcc.gnu.org/PR9316
11142 123. http://gcc.gnu.org/PR9493
11143 124. http://gcc.gnu.org/PR7341
11144 125. http://gcc.gnu.org/PR8947
11145 126. http://gcc.gnu.org/PR7448
11146 127. http://gcc.gnu.org/PR8882
11147 128. http://gcc.gnu.org/PR7445
11148 129. http://gcc.gnu.org/PR2521
11149 130. http://gcc.gnu.org/PR5661
11150 131. http://gcc.gnu.org/PR6419
11151 132. http://gcc.gnu.org/PR6994
11152 133. http://gcc.gnu.org/PR7150
11153 134. http://gcc.gnu.org/PR7160
11154 135. http://gcc.gnu.org/PR7228
11155 136. http://gcc.gnu.org/PR7266
11156 137. http://gcc.gnu.org/PR7353
11157 138. http://gcc.gnu.org/PR7411
11158 139. http://gcc.gnu.org/PR7478
11159 140. http://gcc.gnu.org/PR7526
11160 141. http://gcc.gnu.org/PR7721
11161 142. http://gcc.gnu.org/PR7803
11162 143. http://gcc.gnu.org/PR7754
11163 144. http://gcc.gnu.org/PR7788
11164 145. http://gcc.gnu.org/PR8031
11165 146. http://gcc.gnu.org/PR8055
11166 147. http://gcc.gnu.org/PR8067
11167 148. http://gcc.gnu.org/PR8134
11168 149. http://gcc.gnu.org/PR8149
11169 150. http://gcc.gnu.org/PR8160
11170 151. http://gcc.gnu.org/PR5607
11171 152. http://gcc.gnu.org/PR6579
11172 153. http://gcc.gnu.org/PR6803
11173 154. http://gcc.gnu.org/PR7176
11174 155. http://gcc.gnu.org/PR7188
11175 156. http://gcc.gnu.org/PR7306
11176 157. http://gcc.gnu.org/PR7461
11177 158. http://gcc.gnu.org/PR7524
11178 159. http://gcc.gnu.org/PR7584
11179 160. http://gcc.gnu.org/PR7676
11180 161. http://gcc.gnu.org/PR7679
11181 162. http://gcc.gnu.org/PR7811
11182 163. http://gcc.gnu.org/PR7961
11183 164. http://gcc.gnu.org/PR8071
11184 165. http://gcc.gnu.org/PR8127
11185 166. http://gcc.gnu.org/PR6745
11186 167. http://gcc.gnu.org/PR8096
11187 168. http://gcc.gnu.org/PR8127
11188 169. http://gcc.gnu.org/PR8218
11189 170. http://gcc.gnu.org/PR8287
11190 171. http://gcc.gnu.org/PR8347
11191 172. http://gcc.gnu.org/PR8348
11192 173. http://gcc.gnu.org/PR8391
11193 174. http://gcc.gnu.org/PR6627
11194 175. http://gcc.gnu.org/PR6631
11195 176. http://gcc.gnu.org/PR7102
11196 177. http://gcc.gnu.org/PR7120
11197 178. http://gcc.gnu.org/PR7209
11198 179. http://gcc.gnu.org/PR7515
11199 180. http://gcc.gnu.org/PR7814
11200 181. http://gcc.gnu.org/PR8467
11201 182. http://gcc.gnu.org/PR4890
11202 183. http://gcc.gnu.org/PR7357
11203 184. http://gcc.gnu.org/PR7358
11204 185. http://gcc.gnu.org/PR7602
11205 186. http://gcc.gnu.org/PR7862
11206 187. http://gcc.gnu.org/PR8190
11207 188. http://gcc.gnu.org/PR8524
11208 189. http://gcc.gnu.org/PR5351
11209 190. http://gcc.gnu.org/PR7591
11210 191. http://gcc.gnu.org/PR6845
11211 192. http://gcc.gnu.org/PR7034
11212 193. http://gcc.gnu.org/PR7124
11213 194. http://gcc.gnu.org/PR7174
11214 195. http://gcc.gnu.org/PR7134
11215 196. http://gcc.gnu.org/PR7375
11216 197. http://gcc.gnu.org/PR7390
11217 198. http://gcc.gnu.org/PR6890
11218 199. http://gcc.gnu.org/PR6981
11219 200. http://gcc.gnu.org/PR7242
11220 201. http://gcc.gnu.org/PR7396
11221 202. http://gcc.gnu.org/PR7630
11222 203. http://gcc.gnu.org/PR7693
11223 204. http://gcc.gnu.org/PR7723
11224 205. http://gcc.gnu.org/PR7951
11225 206. http://gcc.gnu.org/PR8146
11226 207. http://gcc.gnu.org/PR5967
11227 208. http://gcc.gnu.org/PR6984
11228 209. http://gcc.gnu.org/PR7114
11229 210. http://gcc.gnu.org/PR7130
11230 211. http://gcc.gnu.org/PR7133
11231 212. http://gcc.gnu.org/PR7380
11232 213. http://gcc.gnu.org/PR8252
11233 214. http://gcc.gnu.org/PR8451
11234 215. http://gcc.gnu.org/PR7250
11235 216. http://gcc.gnu.org/PR6668
11236 217. http://gcc.gnu.org/PR7151
11237 218. http://gcc.gnu.org/PR7335
11238 219. http://gcc.gnu.org/PR7842
11239 220. http://gcc.gnu.org/PR7856
11240 221. http://gcc.gnu.org/PR7967
11241 222. http://gcc.gnu.org/PR7374
11242 223. http://gcc.gnu.org/PR7370
11243 224. http://gcc.gnu.org/PR7409
11244 225. http://gcc.gnu.org/PR8232
11245 226. http://gcc.gnu.org/PR7623
11246 227. http://gcc.gnu.org/PR8314
11247 228. http://gcc.gnu.org/PR761
11248 229. http://gcc.gnu.org/PR5610
11249 230. http://gcc.gnu.org/PR7484
11250 231. http://gcc.gnu.org/PR7531
11251 232. http://gcc.gnu.org/PR8120
11252 233. http://gcc.gnu.org/PR7320
11253 234. http://gcc.gnu.org/PR7470
11254 235. http://gcc.gnu.org/PR6410
11255 236. http://gcc.gnu.org/PR6503
11256 237. http://gcc.gnu.org/PR6642
11257 238. http://gcc.gnu.org/PR7186
11258 239. http://gcc.gnu.org/PR7216
11259 240. http://gcc.gnu.org/PR7220
11260 241. http://gcc.gnu.org/PR7222
11261 242. http://gcc.gnu.org/PR7286
11262 243. http://gcc.gnu.org/PR7442
11263 244. http://gcc.gnu.org/PR7445
11264 245. http://gcc.gnu.org/PR7291
11265 246. http://gcc.gnu.org/onlinedocs/
11266 247. mailto:gcc-help@gcc.gnu.org
11267 248. mailto:gcc@gcc.gnu.org
11268 249. http://gcc.gnu.org/lists.html
11269 250. http://www.fsf.org/
11270 251. http://gcc.gnu.org/about.html
11271 252. http://validator.w3.org/check/referer
11272 ======================================================================
11273 http://gcc.gnu.org/gcc-3.1/index.html
11279 The [1]GNU project and the GCC developers are pleased to announce the
11280 release of GCC 3.1.1.
11282 The links below still apply to GCC 3.1.1.
11286 The [2]GNU project and the GCC developers are pleased to announce the
11287 release of GCC 3.1.
11289 GCC used to stand for the GNU C Compiler, but since the compiler
11290 supports several other languages aside from C, it now stands for the
11291 GNU Compiler Collection.
11293 A list of [3]successful builds is updated as new information becomes
11296 The GCC developers would like to thank the numerous people that have
11297 contributed [4]new features, improvements, bug fixes, and other changes
11298 as well as test results to GCC. This [5]amazing group of volunteers is
11299 what makes GCC successful.
11301 For additional information about GCC please refer to the [6]GCC project
11302 web site or contact the [7]GCC development mailing list.
11304 To obtain GCC please use [8]our mirror sites, or our CVS server.
11305 __________________________________________________________________
11308 For questions related to the use of GCC, please consult these web
11309 pages and the [9]GCC manuals. If that fails, the
11310 [10]gcc-help@gcc.gnu.org mailing list might help. Comments on these
11311 web pages and the development of GCC are welcome on our developer
11312 list at [11]gcc@gcc.gnu.org. All of [12]our lists have public
11315 Copyright (C) [13]Free Software Foundation, Inc. Verbatim copying and
11316 distribution of this entire article is permitted in any medium,
11317 provided this notice is preserved.
11319 These pages are [14]maintained by the GCC team. Last modified
11324 1. http://www.gnu.org/
11325 2. http://www.gnu.org/
11326 3. http://gcc.gnu.org/gcc-3.1/buildstat.html
11327 4. http://gcc.gnu.org/gcc-3.1/changes.html
11328 5. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
11329 6. http://gcc.gnu.org/index.html
11330 7. mailto:gcc@gcc.gnu.org
11331 8. http://gcc.gnu.org/mirrors.html
11332 9. http://gcc.gnu.org/onlinedocs/
11333 10. mailto:gcc-help@gcc.gnu.org
11334 11. mailto:gcc@gcc.gnu.org
11335 12. http://gcc.gnu.org/lists.html
11336 13. http://www.fsf.org/
11337 14. http://gcc.gnu.org/about.html
11338 15. http://validator.w3.org/check/referer
11339 ======================================================================
11340 http://gcc.gnu.org/gcc-3.1/changes.html
11342 GCC 3.1 Release Series
11343 Changes, New Features, and Fixes
11345 Additional changes in GCC 3.1.1
11347 * A bug related to how structures and unions are returned has been
11348 fixed for powerpc-*-netbsd*.
11349 * An important bug in the implementation of -fprefetch-loop-arrays
11350 has been fixed. Previously the optimization prefetched random
11351 blocks of memory for most targets except for i386.
11352 * The Java compiler now compiles Java programs much faster and also
11353 works with parallel make.
11354 * Nested functions have been fixed for mips*-*-netbsd*.
11355 * Some missing floating point support routines have beed added for
11357 * This [1]message gives additional information about the bugs fixed
11362 * The -traditional C compiler option has been deprecated and will be
11363 removed in GCC 3.3. (It remains possible to preprocess non-C code
11364 with the traditional preprocessor.)
11365 * The default debugging format for most ELF platforms (including
11366 GNU/Linux and FreeBSD; notable exception is Solaris) has changed
11367 from stabs to DWARF2. This requires GDB 5.1.1 or later.
11369 General Optimizer Improvements
11371 * Jan Hubicka, SuSE Labs, together with Richard Henderson, Red Hat,
11372 and Andreas Jaeger, SuSE Labs, has contributed [2]infrastructure
11373 for profile driven optimizations.
11374 Options -fprofile-arcs and -fbranch-probabilities can now be used
11375 to improve speed of the generated code by profiling the actual
11376 program behaviour on typical runs. In the absence of profile info
11377 the compiler attempts to guess the profile statically.
11378 * [3]SPEC2000 and SPEC95 benchmark suites are now used daily to
11379 monitor performance of the generated code.
11380 According to the SPECInt2000 results on an AMD Athlon CPU, the code
11381 generated by GCC 3.1 is 6% faster on the average (8.2% faster with
11382 profile feedback) compared to GCC 3.0. The code produced by GCC 3.0
11383 is about 2.1% faster compared to 2.95.3. Tests were done using the
11384 -O2 -march=athlon command-line options.
11385 * Alexandre Oliva, of Red Hat, has generalized the tree inlining
11386 infrastructure developed by CodeSourcery, LLC for the C++ front
11387 end, so that it is now used in the C front end too. Inlining
11388 functions as trees exposes them earlier to the compiler, giving it
11389 more opportunities for optimization.
11390 * Support for data prefetching instructions has been added to the GCC
11391 back end and several targets. A new __builtin_prefetch intrinsic is
11392 available to explicitly insert prefetch instructions and
11393 experimental support for loop array prefetching has been added (see
11394 -fprefetch-loop-array documentation).
11395 * Support for emitting debugging information for macros has been
11396 added for DWARF2. It is activated using -g3.
11398 New Languages and Language specific improvements
11402 * A few more [4]ISO C99 features.
11403 * The preprocessor is 10-50% faster than the preprocessor in GCC 3.0.
11404 * The preprocessor's symbol table has been merged with the symbol
11405 table of the C, C++ and Objective-C front ends.
11406 * The preprocessor consumes less memory than the preprocessor in GCC
11407 3.0, often significantly so. On normal input files, it typically
11408 consumes less memory than pre-3.0 cccp-based GCC, too.
11412 * -fhonor-std and -fno-honor-std have been removed. -fno-honor-std
11413 was a workaround to allow std compliant code to work with the
11414 non-std compliant libstdc++-v2. libstdc++-v3 is std compliant.
11415 * The C++ ABI has been fixed so that void (A::*)() const is mangled
11416 as "M1AKFvvE", rather than "MK1AFvvE" as before. This change only
11417 affects pointer to cv-qualified member function types.
11418 * The C++ ABI has been changed to correctly handle this code:
11420 void operator delete[] (void *, size_t);
11423 struct B : public A {
11428 The amount of storage allocated for the array will be greater than
11429 it was in 3.0, in order to store the number of elements in the
11430 array, so that the correct size can be passed to operator delete[]
11431 when the array is deleted. Previously, the value passed to operator
11432 delete[] was unpredictable.
11433 This change will only affect code that declares a two-argument
11434 operator delete[] with a second parameter of type size_t in a base
11435 class, and does not override that definition in a derived class.
11436 * The C++ ABI has been changed so that:
11438 void operator delete[] (void *, size_t);
11439 void operator delete[] (void *);
11442 does not cause unnecessary storage to be allocated when an array of
11443 A objects is allocated.
11444 This change will only affect code that declares both of these forms
11445 of operator delete[], and declared the two-argument form before the
11447 * The C++ ABI has been changed so that when a parameter is passed by
11448 value, any cleanup for that parameter is performed in the caller,
11449 as specified by the ia64 C++ ABI, rather than the called function
11450 as before. As a result, classes with a non-trivial destructor but a
11451 trivial copy constructor will be passed and returned by invisible
11452 reference, rather than by bitwise copy as before.
11453 * G++ now supports the "named return value optimization": for code
11461 G++ will allocate a in the return value slot, so that the return
11462 becomes a no-op. For this to work, all return statements in the
11463 function must return the same variable.
11464 * Improvements to the C++ library are listed in [5]the libstdc++-v3
11469 * Annoying linker warnings (due to incorrect code being generated)
11471 * If a class method cannot be found, the compiler no longer issues a
11472 warning if a corresponding instance method exists in the root
11474 * Forward @protocol declarations have been fixed.
11475 * Loading of categories has been fixed in certain situations (GNU run
11477 * The class lookup in the run-time library has been rewritten so that
11478 class method dispatch is more than twice as fast as it used to be
11479 (GNU run time only).
11483 * libgcj now includes RMI, java.lang.ref.*, javax.naming, and
11485 * Property files and other system resources can be compiled into
11486 executables which use libgcj using the new gcj --resource feature.
11487 * libgcj has been ported to more platforms. In particular there is
11488 now a mostly-functional mingw32 (Windows) target port.
11489 * JNI and CNI invocation interfaces were implemented, so gcj-compiled
11490 Java code can now be called from a C/C++ application.
11491 * gcj can now use builtin functions for certain known methods, for
11493 * gcj can now automatically remove redundant array-store checks in
11495 * The --no-store-checks optimization option was added. This can be
11496 used to omit runtime store checks for code which is known not to
11497 throw ArrayStoreException
11498 * The following third party interface standards were added to libgcj:
11499 org.w3c.dom and org.xml.sax.
11500 * java.security has been merged with GNU Classpath. The new package
11501 is now JDK 1.2 compliant, and much more complete.
11502 * A bytecode verifier was added to the libgcj interpreter.
11503 * java.lang.Character was rewritten to comply with the Unicode 3.0
11504 standard, and improve performance.
11505 * Partial support for many more locales was added to libgcj.
11506 * Socket timeouts have been implemented.
11507 * libgcj has been merged into a single shared library. There are no
11508 longer separate shared libraries for the garbage collector and
11510 * Several performance improvements were made to gcj and libgcj:
11511 + Hash synchronization (thin locks)
11512 + A special allocation path for finalizer-free objects
11513 + Thread-local allocation
11514 + Parallel GC, and other GC tweaks
11518 Fortran improvements are listed in [6]the Fortran documentation.
11522 [7]Ada Core Technologies, Inc, has contributed its GNAT Ada 95 front
11523 end and associated tools. The GNAT compiler fully implements the Ada
11524 language as defined by the ISO/IEC 8652 standard.
11526 Please note that the integration of the Ada front end is still work in
11529 New Targets and Target Specific Improvements
11531 * Hans-Peter Nilsson has contributed a port to [8]MMIX, the CPU
11532 architecture used in new editions of Donald E. Knuth's The Art of
11533 Computer Programming.
11534 * [9]Axis Communications has contributed its port to the CRIS CPU
11535 architecture, used in the ETRAX system-on-a-chip series. See
11536 [10]Axis' developer site for technical information.
11537 * Alexandre Oliva, of Red Hat, has contributed a port to the
11538 [11]SuperH SH5 64-bit RISC microprocessor architecture, extending
11539 the existing SH port.
11540 * UltraSPARC is fully supported in 64-bit mode. The option -m64
11542 * For compatibility with the Sun compiler #pragma redefine_extname
11543 has been implemented on Solaris.
11544 * The x86 back end has had some noticeable work done to it.
11545 + SuSE Labs developers Jan Hubicka, Bo Thorsen and Andreas
11546 Jaeger have contributed a port to the AMD x86-64 architecture.
11547 For more information on x86-64 see [12]http://www.x86-64.org.
11548 + The compiler now supports MMX, 3DNow!, SSE, and SSE2
11549 instructions. Options -mmmx, -m3dnow, -msse, and -msse2 will
11550 enable the respective instruction sets. Intel C++ compatible
11551 MMX/3DNow!/SSE intrinsics are implemented. SSE2 intrinsics
11552 will be added in next major release.
11553 + Following those improvements, targets for Pentium MMX, K6-2,
11554 K6-3, Pentium III, Pentium 4, and Athlon 4 Mobile/XP/MP were
11555 added. Refer to the documentation on -march= and -mcpu=
11556 options for details.
11557 + For those targets that support it, -mfpmath=sse will cause the
11558 compiler to generate SSE/SSE2 instructions for floating point
11559 math instead of x87 instructions. Usually, this will lead to
11560 quicker code -- especially on the Pentium 4. Note that only
11561 scalar floating point instructions are used and GCC does not
11562 exploit SIMD features yet.
11563 + Prefetch support has been added to the Pentium III, Pentium 4,
11564 K6-2, K6-3, and Athlon series.
11565 + Code generated for floating point to integer conversions has
11566 been improved leading to better performance of many 3D
11568 * The PowerPC back end has added 64-bit PowerPC GNU/Linux support.
11569 * C++ support for AIX has been improved.
11570 * Aldy Hernandez, of Red Hat, Inc, has contributed extensions to the
11571 PowerPC port supporting the AltiVec programming model (SIMD). The
11572 support, though presently useful, is experimental and is expected
11573 to stabilize for 3.2. The support is written to conform to
11574 Motorola's AltiVec specs. See -maltivec.
11578 Support for a number of older systems has been declared obsolete in GCC
11579 3.1. Unless there is activity to revive them, the next release of GCC
11580 will have their sources permanently removed.
11582 All configurations of the following processor architectures have been
11584 * MIL-STD-1750A, 1750a-*-*
11585 * AMD A29k, a29k-*-*
11586 * Convex, c*-convex-*
11587 * Clipper, clipper-*-*
11589 * Intel i860, i860-*-*
11590 * Sun picoJava, pj-*-* and pjl-*-*
11591 * Western Electric 32000, we32k-*-*
11593 Most configurations of the following processor architectures have been
11594 declared obsolete, but we are preserving a few systems which may have
11595 active developers. It is unlikely that the remaining systems will
11596 survive much longer unless we see definite signs of port activity.
11597 * Motorola 88000 except
11598 + Generic a.out, m88k-*-aout*
11599 + Generic SVR4, m88k-*-sysv4
11600 + OpenBSD, m88k-*-openbsd*
11602 + NetBSD, ns32k-*-netbsd*
11603 + OpenBSD, ns32k-*-openbsd*.
11605 + OpenBSD, romp-*-openbsd*.
11607 Finally, only some configurations of these processor architectures are
11610 + OSF/1, alpha*-*-osf[123]*. (Digital Unix and Tru64 Unix, aka
11611 alpha*-*-osf[45], are still supported.)
11613 + RISCiX, arm-*-riscix*.
11615 + 386BSD, i?86-*-bsd*
11616 + Chorus, i?86-*-chorusos*
11617 + DG/UX, i?86-*-dgux*
11618 + FreeBSD 1.x, i?86-*-freebsd1.*
11619 + IBM AIX, i?86-*-aix*
11620 + ISC UNIX, i?86-*-isc*
11621 + GNU/Linux with pre-BFD linker, i?86-*-linux*oldld*
11622 + NEXTstep, i?86-next-*
11623 + OSF UNIX, i?86-*-osf1* and i?86-*-osfrose*
11624 + RTEMS/coff, i?86-*-rtemscoff*
11625 + RTEMS/go32, i?86-go32-rtems*
11626 + Sequent/BSD, i?86-sequent-bsd*
11627 + Sequent/ptx before version 3, i?86-sequent-ptx[12]* and
11628 i?86-sequent-sysv3*
11629 + SunOS, i?86-*-sunos*
11631 + Altos, m68[k0]*-altos-*
11632 + Apollo, m68[k0]*-apollo-*
11633 + Apple A/UX, m68[k0]*-apple-*
11634 + Bull, m68[k0]*-bull-*
11635 + Convergent, m68[k0]*-convergent-*
11636 + Generic SVR3, m68[k0]*-*-sysv3*
11637 + ISI, m68[k0]*-isi-*
11638 + LynxOS, m68[k0]*-*-lynxos*
11639 + NEXT, m68[k0]*-next-*
11640 + RTEMS/coff, m68[k0]*-*-rtemscoff*
11641 + Sony, m68[k0]*-sony-*
11643 + DEC Ultrix, mips-*-ultrix* and mips-dec-*
11644 + Generic BSD, mips-*-bsd*
11645 + Generic System V, mips-*-sysv*
11646 + IRIX before version 5, mips-sgi-irix[1234]*
11647 + RiscOS, mips-*-riscos*
11648 + Sony, mips-sony-*
11649 + Tandem, mips-tandem-*
11651 + RTEMS/a.out, sparc-*-rtemsaout*.
11653 Documentation improvements
11655 * The old manual ("Using and Porting the GNU Compiler Collection")
11656 has been replaced by a users manual ("Using the GNU Compiler
11657 Collection") and a separate internals reference manual ("GNU
11658 Compiler Collection Internals").
11659 * More complete and much improved documentation about GCC's internal
11660 representation used by the C and C++ front ends.
11661 * Many cleanups and improvements in general.
11664 For questions related to the use of GCC, please consult these web
11665 pages and the [13]GCC manuals. If that fails, the
11666 [14]gcc-help@gcc.gnu.org mailing list might help. Comments on these
11667 web pages and the development of GCC are welcome on our developer
11668 list at [15]gcc@gcc.gnu.org. All of [16]our lists have public
11671 Copyright (C) [17]Free Software Foundation, Inc. Verbatim copying and
11672 distribution of this entire article is permitted in any medium,
11673 provided this notice is preserved.
11675 These pages are [18]maintained by the GCC team. Last modified
11680 1. http://gcc.gnu.org/ml/gcc/2002-07/msg01208.html
11681 2. http://gcc.gnu.org/news/profiledriven.html
11682 3. http://gcc.gnu.org/benchmarks/
11683 4. http://gcc.gnu.org/gcc-3.1/c99status.html
11684 5. http://gcc.gnu.org/onlinedocs/libstdc++/faq.html
11685 6. http://gcc.gnu.org/onlinedocs/gcc-3.1.1/g77/News.html
11686 7. http://www.adacore.com/
11687 8. http://www-cs-faculty.stanford.edu/~knuth/mmix.html
11688 9. http://www.axis.com/
11689 10. http://developer.axis.com/
11690 11. http://www.superh.com/
11691 12. http://www.x86-64.org/
11692 13. http://gcc.gnu.org/onlinedocs/
11693 14. mailto:gcc-help@gcc.gnu.org
11694 15. mailto:gcc@gcc.gnu.org
11695 16. http://gcc.gnu.org/lists.html
11696 17. http://www.fsf.org/
11697 18. http://gcc.gnu.org/about.html
11698 19. http://validator.w3.org/check/referer
11699 ======================================================================
11700 http://gcc.gnu.org/gcc-3.0/gcc-3.0.html
11706 The [1]GNU project and the GCC developers are pleased to announce the
11707 release of GCC 3.0.4, which is a bug-fix release for the GCC 3.0
11710 GCC used to stand for the GNU C Compiler, but since the compiler
11711 supports several other languages aside from C, it now stands for the
11712 GNU Compiler Collection.
11714 GCC 3.0.x has several new optimizations, new targets, new languages and
11715 many other new features, relative to GCC 2.95.x. See the [2]new
11716 features page for a more complete list.
11718 A list of [3]successful builds is updated as new information becomes
11721 The GCC developers would like to thank the numerous people that have
11722 contributed new features, test results, bug fixes, etc to GCC. This
11723 [4]amazing group of volunteers is what makes GCC successful.
11725 And finally, we can't in good conscience fail to mention some
11726 [5]caveats to using GCC 3.0.x.
11728 For additional information about GCC please refer to the [6]GCC project
11729 web site or contact the [7]GCC development mailing list.
11731 To obtain GCC please use [8]our mirror sites, or our CVS server.
11732 __________________________________________________________________
11734 Previous 3.0.x Releases
11736 December 20, 2001: GCC 3.0.3 has been released.
11737 October 25, 2001: GCC 3.0.2 has been released.
11738 August 20, 2001: GCC 3.0.1 has been released.
11739 June 18, 2001: GCC 3.0 has been released.
11742 For questions related to the use of GCC, please consult these web
11743 pages and the [9]GCC manuals. If that fails, the
11744 [10]gcc-help@gcc.gnu.org mailing list might help. Comments on these
11745 web pages and the development of GCC are welcome on our developer
11746 list at [11]gcc@gcc.gnu.org. All of [12]our lists have public
11749 Copyright (C) [13]Free Software Foundation, Inc. Verbatim copying and
11750 distribution of this entire article is permitted in any medium,
11751 provided this notice is preserved.
11753 These pages are [14]maintained by the GCC team. Last modified
11758 1. http://www.gnu.org/
11759 2. http://gcc.gnu.org/gcc-3.0/features.html
11760 3. http://gcc.gnu.org/gcc-3.0/buildstat.html
11761 4. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
11762 5. http://gcc.gnu.org/gcc-3.0/caveats.html
11763 6. http://gcc.gnu.org/index.html
11764 7. mailto:gcc@gcc.gnu.org
11765 8. http://gcc.gnu.org/mirrors.html
11766 9. http://gcc.gnu.org/onlinedocs/
11767 10. mailto:gcc-help@gcc.gnu.org
11768 11. mailto:gcc@gcc.gnu.org
11769 12. http://gcc.gnu.org/lists.html
11770 13. http://www.fsf.org/
11771 14. http://gcc.gnu.org/about.html
11772 15. http://validator.w3.org/check/referer
11773 ======================================================================
11774 http://gcc.gnu.org/gcc-3.0/features.html
11776 GCC 3.0 New Features
11778 Additional changes in GCC 3.0.4
11780 * GCC 3.0 now supports newer versions of the [1]NetBSD operating
11781 system, which use the ELF object file format, on x86 processors.
11782 * Correct debugging information is generated from functions that have
11783 lines from multiple files (e.g. yacc output).
11784 * A fix for whitespace handling in the -traditional preprocessor,
11785 which can affect Fortran.
11786 * Fixes to the exception handling runtime.
11787 * More fixes for bad code generation in C++.
11788 * A fix for shared library generation under AIX 4.3.
11789 * Documentation updates.
11790 * Port of GCC to Tensilica's Xtensa processor contributed.
11791 * A fix for compiling the PPC Linux kernel (FAT fs wouldn't link).
11793 Additional changes in GCC 3.0.3
11795 * A fix to correct an accidental change to the PowerPC ABI.
11796 * Fixes for bad code generation on a variety of architectures.
11797 * Improvements to the debugging information generated for C++
11799 * Fixes for bad code generation in C++.
11800 * A fix to avoid crashes in the C++ demangler.
11801 * A fix to the C++ standard library to avoid buffer overflows.
11802 * Miscellaneous improvements for a variety of architectures.
11804 Additional changes in GCC 3.0.2
11806 * Fixes for bad code generation during loop unrolling.
11807 * Fixes for bad code generation by the sibling call optimization.
11808 * Minor improvements to x86 code generation.
11809 * Implementation of function descriptors in C++ vtables for IA64.
11810 * Numerous minor bug-fixes.
11812 Additional changes in GCC 3.0.1
11814 * C++ fixes for incorrect code-generation.
11815 * Improved cross-compiling support for the C++ standard library.
11816 * Fixes for some embedded targets that worked in GCC 2.95.3, but not
11818 * Fixes for various exception-handling bugs.
11819 * A port to the S/390 architecture.
11821 General Optimizer Improvements
11823 * [2]Basic block reordering pass.
11824 * New if-conversion pass with support for conditional (predicated)
11826 * New tail call and sibling call elimination optimizations.
11827 * New register renaming pass.
11828 * New (experimental) [3]static single assignment (SSA) representation
11830 * New dead-code elimination pass implemented using the SSA
11832 * [4]Global null pointer test elimination.
11833 * [5]Global code hoisting/unification.
11834 * More builtins and optimizations for stdio.h, string.h and old BSD
11835 functions, as well as for ISO C99 functions.
11836 * New builtin __builtin_expect for giving hints to the branch
11839 New Languages and Language specific improvements
11841 * The GNU Compiler for the Java(TM) language (GCJ) is now integrated
11842 and supported, including the run-time library containing most
11843 common non-GUI Java classes, a bytecode interpreter, and the Boehm
11844 conservative garbage collector. Many bugs have been fixed. GCJ can
11845 compile Java source or Java bytecodes to either native code or Java
11846 class files, and supports native methods written in either the
11847 standard JNI or the more efficient and convenient CNI.
11848 * Here is a [6]partial list of C++ improvements, both new features
11849 and those no longer supported.
11850 * New C++ ABI. On the IA-64 platform GCC is capable of
11851 inter-operating with other IA-64 compilers.
11852 * The new ABI also significantly reduces the size of symbol and debug
11854 * New [7]C++ support library and many C++ bug fixes, vastly improving
11855 our conformance to the ISO C++ standard.
11856 * New [8]inliner for C++.
11857 * Rewritten C preprocessor, integrated into the C, C++ and Objective
11858 C compilers, with very many improvements including ISO C99 support
11859 and [9]improvements to dependency generation.
11860 * Support for more [10]ISO C99 features.
11861 * Many improvements to support for checking calls to format functions
11862 such as printf and scanf, including support for ISO C99 format
11863 features, extensions from the Single Unix Specification and GNU
11864 libc 2.2, checking of strfmon formats and features to assist in
11865 auditing for format string security bugs.
11866 * New warnings for C code that may have undefined semantics because
11867 of violations of sequence point rules in the C standard (such as a
11868 = a++;, a[n] = b[n++]; and a[i++] = i;), included in -Wall.
11869 * Additional warning option -Wfloat-equal.
11870 * Improvements to -Wtraditional.
11871 * Fortran improvements are listed in [11]the Fortran documentation.
11873 New Targets and Target Specific Improvements
11875 * New x86 back-end, generating much improved code.
11876 * Support for a generic i386-elf target contributed.
11877 * New option to emit x86 assembly code using Intel style syntax
11879 * HPUX 11 support contributed.
11880 * Improved PowerPC code generation, including scheduled prologue and
11882 * Port of GCC to Intel's IA-64 processor contributed.
11883 * Port of GCC to Motorola's MCore 210 and 340 contributed.
11884 * New unified back-end for Arm, Thumb and StrongArm contributed.
11885 * Port of GCC to Intel's XScale processor contributed.
11886 * Port of GCC to Atmel's AVR microcontrollers contributed.
11887 * Port of GCC to Mitsubishi's D30V processor contributed.
11888 * Port of GCC to Matsushita's AM33 processor (a member of the MN10300
11889 processor family) contributed.
11890 * Port of GCC to Fujitsu's FR30 processor contributed.
11891 * Port of GCC to Motorola's 68HC11 and 68HC12 processors contributed.
11892 * Port of GCC to Sun's picoJava processor core contributed.
11894 Documentation improvements
11896 * Substantially rewritten and improved C preprocessor manual.
11897 * Many improvements to other documentation.
11898 * Manpages for gcc, cpp and gcov are now generated automatically from
11899 the master Texinfo manual, eliminating the problem of manpages
11900 being out of date. (The generated manpages are only extracts from
11901 the full manual, which is provided in Texinfo form, from which
11902 info, HTML, other formats and a printed manual can be generated.)
11903 * Generated info files are included in the release tarballs alongside
11904 their Texinfo sources, avoiding problems on some platforms with
11905 building makeinfo as part of the GCC distribution.
11907 Other significant improvements
11909 * Garbage collection used internally by the compiler for most memory
11910 allocation instead of obstacks.
11911 * Lengauer and Tarjan algorithm used for computing dominators in the
11912 CFG. This algorithm can be significantly faster and more space
11913 efficient than our older algorithm.
11914 * gccbug script provided to assist in submitting bug reports to our
11915 bug tracking system. (Bug reports previously submitted directly to
11916 our mailing lists, for which you received no bug tracking number,
11917 should be submitted again using gccbug if you can reproduce the
11918 problem with GCC 3.0.)
11919 * The internal libgcc library is [12]built as a shared library on
11920 systems that support it.
11921 * Extensive testsuite included with GCC, with many new tests. In
11922 addition to tests for GCC bugs that have been fixed, many tests
11923 have been added for language features, compiler warnings and
11925 * Additional language-independent warning options -Wpacked, -Wpadded,
11926 -Wunreachable-code and -Wdisabled-optimization.
11927 * Target-independent options -falign-functions, -falign-loops and
11930 Plus a great many bug fixes and almost all the [13]features found in
11934 For questions related to the use of GCC, please consult these web
11935 pages and the [14]GCC manuals. If that fails, the
11936 [15]gcc-help@gcc.gnu.org mailing list might help. Comments on these
11937 web pages and the development of GCC are welcome on our developer
11938 list at [16]gcc@gcc.gnu.org. All of [17]our lists have public
11941 Copyright (C) [18]Free Software Foundation, Inc. Verbatim copying and
11942 distribution of this entire article is permitted in any medium,
11943 provided this notice is preserved.
11945 These pages are [19]maintained by the GCC team. Last modified
11950 1. http://www.netbsd.org/
11951 2. http://gcc.gnu.org/news/reorder.html
11952 3. http://gcc.gnu.org/news/ssa.html
11953 4. http://gcc.gnu.org/news/null.html
11954 5. http://gcc.gnu.org/news/unify.html
11955 6. http://gcc.gnu.org/gcc-3.0/c++features.html
11956 7. http://gcc.gnu.org/libstdc++/
11957 8. http://gcc.gnu.org/news/inlining.html
11958 9. http://gcc.gnu.org/news/dependencies.html
11959 10. http://gcc.gnu.org/gcc-3.0/c99status.html
11960 11. http://gcc.gnu.org/onlinedocs/gcc-3.4.6/g77/News.html
11961 12. http://gcc.gnu.org/gcc-3.0/libgcc.html
11962 13. http://gcc.gnu.org/gcc-2.95/features.html
11963 14. http://gcc.gnu.org/onlinedocs/
11964 15. mailto:gcc-help@gcc.gnu.org
11965 16. mailto:gcc@gcc.gnu.org
11966 17. http://gcc.gnu.org/lists.html
11967 18. http://www.fsf.org/
11968 19. http://gcc.gnu.org/about.html
11969 20. http://validator.w3.org/check/referer
11970 ======================================================================
11971 http://gcc.gnu.org/gcc-3.0/caveats.html
11975 * -fstrict-aliasing is now part of -O2 and higher optimization
11976 levels. This allows the compiler to assume the strictest aliasing
11977 rules applicable to the language being compiled. For C and C++,
11978 this activates optimizations based on the type of expressions. This
11979 optimization may thus break old, non-compliant code.
11980 * Enumerations are now properly promoted to int in function
11981 parameters and function returns. Normally this change is not
11982 visible, but when using -fshort-enums this is an ABI change.
11983 * The undocumented extension that allowed C programs to have a label
11984 at the end of a compound statement has been deprecated and may be
11985 removed in a future version. Programs that now generate a warning
11986 about this may be fixed by adding a null statement (a single
11987 semicolon) after the label.
11988 * The poorly documented extension that allowed string constants in C,
11989 C++ and Objective C to contain unescaped newlines has been
11990 deprecated and may be removed in a future version. Programs using
11991 this extension may be fixed in several ways: the bare newline may
11992 be replaced by \n, or preceded by \n\, or string concatenation may
11993 be used with the bare newline preceded by \n" and " placed at the
11994 start of the next line.
11995 * The Chill compiler is not included in GCC 3.0, because of the lack
11996 of a volunteer to convert it to use garbage collection.
11997 * Certain non-standard iostream methods from earlier versions of
11998 libstdc++ are not included in libstdc++ v3, i.e. filebuf::attach,
11999 ostream::form, and istream::gets.
12000 * The new C++ ABI is not yet fully supported by current (as of
12001 2001-07-01) releases and development versions of GDB, or any
12002 earlier versions. There is a problem setting breakpoints by line
12003 number, and other related issues that have been fixed in GCC 3.0
12004 but not yet handled in GDB:
12005 [1]http://gcc.gnu.org/ml/gcc-bugs/2001-06/msg00421.html
12008 For questions related to the use of GCC, please consult these web
12009 pages and the [2]GCC manuals. If that fails, the
12010 [3]gcc-help@gcc.gnu.org mailing list might help. Comments on these
12011 web pages and the development of GCC are welcome on our developer
12012 list at [4]gcc@gcc.gnu.org. All of [5]our lists have public archives.
12014 Copyright (C) [6]Free Software Foundation, Inc. Verbatim copying and
12015 distribution of this entire article is permitted in any medium,
12016 provided this notice is preserved.
12018 These pages are [7]maintained by the GCC team. Last modified
12023 1. http://gcc.gnu.org/ml/gcc-bugs/2001-06/msg00421.html
12024 2. http://gcc.gnu.org/onlinedocs/
12025 3. mailto:gcc-help@gcc.gnu.org
12026 4. mailto:gcc@gcc.gnu.org
12027 5. http://gcc.gnu.org/lists.html
12028 6. http://www.fsf.org/
12029 7. http://gcc.gnu.org/about.html
12030 8. http://validator.w3.org/check/referer
12031 ======================================================================
12032 http://gcc.gnu.org/gcc-2.95/index.html
12036 March 16, 2001: The GNU project and the GCC developers are pleased to
12037 announce the release of GCC version 2.95.3.
12051 July 31, 1999. This is the first release of GCC since the April
12052 1999 GCC/EGCS reunification and includes nearly a year's worth
12053 of new development and bugfixes.
12055 References and Acknowledgements
12057 GCC used to stand for the GNU C Compiler, but since the compiler
12058 supports several other languages aside from C, it now stands for the
12059 GNU Compiler Collection.
12061 The whole suite has been extensively [1]regression tested and
12062 [2]package tested. It should be reliable and suitable for widespread
12065 The compiler has several new optimizations, new targets, new languages
12066 and other new features. See the [3]new features page for a more
12067 complete list of new features found in the GCC 2.95 releases.
12069 The sources include installation instructions in both HTML and
12070 plaintext forms in the install directory in the distribution. However,
12071 the most up to date [4]installation instructions and [5]build/test
12072 status are on the web pages. We will update those pages as new
12073 information becomes available.
12075 The GCC developers would like to thank the numerous people that have
12076 contributed new features, test results, bugfixes, etc to GCC. This
12077 [6]amazing group of volunteers is what makes GCC successful.
12079 And finally, we can't in good conscience fail to mention some
12080 [7]caveats to using GCC 2.95.
12082 Download GCC 2.95 from one of our many [8]mirror sites.
12084 For additional information about GCC please see the [9]GCC project web
12085 server or contact the [10]GCC development mailing list.
12088 For questions related to the use of GCC, please consult these web
12089 pages and the [11]GCC manuals. If that fails, the
12090 [12]gcc-help@gcc.gnu.org mailing list might help. Comments on these
12091 web pages and the development of GCC are welcome on our developer
12092 list at [13]gcc@gcc.gnu.org. All of [14]our lists have public
12095 Copyright (C) [15]Free Software Foundation, Inc. Verbatim copying and
12096 distribution of this entire article is permitted in any medium,
12097 provided this notice is preserved.
12099 These pages are [16]maintained by the GCC team. Last modified
12104 1. http://gcc.gnu.org/gcc-2.95/regress.html
12105 2. http://gcc.gnu.org/gcc-2.95/othertest.html
12106 3. http://gcc.gnu.org/gcc-2.95/features.html
12107 4. http://gcc.gnu.org/install/
12108 5. http://gcc.gnu.org/gcc-2.95/buildstat.html
12109 6. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
12110 7. http://gcc.gnu.org/gcc-2.95/caveats.html
12111 8. http://gcc.gnu.org/mirrors.html
12112 9. http://gcc.gnu.org/index.html
12113 10. mailto:gcc@gcc.gnu.org
12114 11. http://gcc.gnu.org/onlinedocs/
12115 12. mailto:gcc-help@gcc.gnu.org
12116 13. mailto:gcc@gcc.gnu.org
12117 14. http://gcc.gnu.org/lists.html
12118 15. http://www.fsf.org/
12119 16. http://gcc.gnu.org/about.html
12120 17. http://validator.w3.org/check/referer
12121 ======================================================================
12122 http://gcc.gnu.org/gcc-2.95/features.html
12124 GCC 2.95 New Features
12126 * General Optimizer Improvements:
12127 + [1]Localized register spilling to improve speed and code
12128 density especially on small register class machines.
12129 + [2]Global CSE using lazy code motion algorithms.
12130 + [3]Improved global constant/copy propagation.
12131 + [4]Improved control flow graph analysis and manipulation.
12132 + [5]Local dead store elimination.
12133 + [6]Memory Load hoisting/store sinking in loops.
12134 + [7]Type based alias analysis is enabled by default. Note this
12135 feature will expose bugs in the Linux kernel. Please refer to
12136 the FAQ (as shipped with GCC 2.95) for additional information
12138 + Major revamp of GIV detection, combination and simplification
12139 to improve loop performance.
12140 + Major improvements to register allocation and reloading.
12141 * New Languages and Language specific improvements
12142 + [8]Many C++ improvements.
12143 + [9]Many Fortran improvements.
12144 + [10]Java front-end has been integrated. [11]runtime library is
12145 available separately.
12146 + [12]ISO C99 support
12147 + [13]Chill front-end and runtime has been integrated.
12148 + Boehm garbage collector support in libobjc.
12149 + More support for various pragmas which appear in vendor
12151 * New Targets and Target Specific Improvements
12152 + [14]SPARC backend rewrite.
12153 + -mschedule=8000 will optimize code for PA8000 class
12154 processors; -mpa-risc-2-0 will generate code for PA2.0
12156 + Various micro-optimizations for the ia32 port. K6
12158 + Compiler will attempt to align doubles in the stack on the
12160 + Alpha EV6 support
12162 + RS6000/PowerPC: -mcpu=401 was added as an alias for -mcpu=403.
12163 -mcpu=e603e was added to do -mcpu=603e and -msoft-float.
12168 + Support for new systems (OpenBSD, FreeBSD, UWIN, Interix,
12170 + vxWorks targets include support for vxWorks threads
12171 + StrongARM 110 and ARM9 support added. ARM Scheduling
12172 parameters rewritten.
12173 + Various changes to the MIPS port to avoid assembler macros,
12174 which in turn improves performance
12175 + Various performance improvements to the i960 port.
12176 + Major rewrite of ns32k port
12177 * Other significant improvements
12178 + [15]Ability to dump cfg information and display it using vcg.
12179 + The new faster scheme for fixing vendor header files is
12180 enabled by default.
12181 + Experimental internationalization support.
12182 + multibyte character support
12183 + Some compile-time speedups for pathological problems
12184 + Better support for complex types
12185 * Plus the usual mountain of bugfixes
12186 * Core compiler is based on the gcc2 development tree from Sept 30,
12187 1998, so we have all of the [16]features found in GCC 2.8.
12189 Additional Changes in GCC 2.95.1
12191 * Generic bugfixes and improvements
12192 + Various documentation fixes related to the GCC/EGCS merger.
12193 + Fix memory management bug which could lead to spurious aborts,
12194 core dumps or random parsing errors in the compiler.
12195 + Fix a couple bugs in the dwarf1 and dwarf2 debug record
12197 + Fix infinite loop in the CSE optimizer.
12198 + Avoid undefined behavior in compiler FP emulation code
12199 + Fix install problem when prefix is overridden on the make
12201 + Fix problem with unwanted installation of assert.h on some
12203 + Fix problem with finding the wrong assembler in a single tree
12205 + Avoid increasing the known alignment of a register that is
12206 already known to be a pointer.
12207 * Platform specific bugfixes and improvements
12208 + Codegen bugfix for prologue/epilogue for cpu32 target.
12209 + Fix long long code generation bug for the Coldfire target.
12210 + Fix various aborts in the SH compiler.
12211 + Fix bugs in libgcc support library for the SH.
12212 + Fix alpha ev6 code generation bug.
12213 + Fix problems with EXIT_SUCCESS/EXIT_FAILURE redefinitions on
12215 + Fix -fpic code generation bug for rs6000/ppc svr4 targets.
12216 + Fix varargs/stdarg code generation bug for rs6000/ppc svr4
12218 + Fix weak symbol handling for rs6000/ppc svr4 targets.
12219 + Fix various problems with 64bit code generation for the
12221 + Fix codegen bug which caused tetex to be mis-compiled on the
12223 + Fix compiler abort in new cfg code exposed by x86 port.
12224 + Fix out of range array reference in code convert flat
12225 registers to the x87 stacked FP register file.
12226 + Fix minor vxworks configuration bug.
12227 + Fix return type of bsearch for SunOS 4.x.
12228 * Language & Runtime specific fixes.
12229 + The G++ signature extension has been deprecated. It will be
12230 removed in the next major release of G++. Use of signatures
12231 will result in a warning from the compiler.
12232 + Several bugs relating to templates and namespaces were fixed.
12233 + A bug that caused crashes when combining templates with -g on
12234 DWARF1 platforms was fixed.
12235 + Pointers-to-members, virtual functions, and multiple
12236 inheritance should now work together correctly.
12237 + Some code-generation bugs relating to function try blocks were
12239 + G++ is a little bit more lenient with certain archaic
12240 constructs than in GCC 2.95.
12241 + Fix to prevent shared library version #s from bring truncated
12243 + Fix missing std:: in the libstdc++ library.
12244 + Fix stream locking problems in libio.
12245 + Fix problem in java compiler driver.
12247 Additional Changes in GCC 2.95.2
12249 The -fstrict-aliasing is not enabled by default for GCC 2.95.2. While
12250 the optimizations performed by -fstrict-aliasing are valid according to
12251 the C and C++ standards, the optimization have caused some problems,
12252 particularly with old non-conforming code.
12254 The GCC developers are experimenting with ways to warn users about code
12255 which violates the C/C++ standards, but those warnings are not ready
12256 for widespread use at this time. Rather than wait for those warnings
12257 the GCC developers have chosen to disable -fstrict-aliasing by default
12258 for the GCC 2.95.2 release.
12260 We strongly encourage developers to find and fix code which violates
12261 the C/C++ standards as -fstrict-aliasing may be enabled by default in
12262 future releases. Use the option -fstrict-aliasing to re-enable these
12264 * Generic bugfixes and improvements
12265 + Fix incorrectly optimized memory reference in global common
12266 subexpression elimination (GCSE) optimization pass.
12267 + Fix code generation bug in regmove.c in which it could
12268 incorrectly change a "const" value.
12269 + Fix bug in optimization of conditionals involving volatile
12271 + Avoid over-allocation of stack space for some procedures.
12272 + Fixed bug in the compiler which caused incorrect optimization
12273 of an obscure series of bit manipulations, shifts and
12275 + Fixed register allocator bug which caused teTeX to be
12276 mis-compiled on SPARC targets.
12277 + Avoid incorrect optimization of degenerate case statements for
12278 certain targets such as the ARM.
12279 + Fix out of range memory reference in the jump optimizer.
12280 + Avoid dereferencing null pointer in fix-header.
12281 + Fix test for GCC specific features so that it is possible to
12282 bootstrap with gcc-2.6.2 and older versions of GCC.
12283 + Fix typo in scheduler which could potentially cause out of
12284 range memory accesses.
12285 + Avoid incorrect loop reversal which caused incorrect code for
12286 certain loops on PowerPC targets.
12287 + Avoid incorrect optimization of switch statements on certain
12288 targets (for example the ARM).
12289 * Platform specific bugfixes and improvements
12290 + Work around bug in Sun V5.0 compilers which caused bootstrap
12291 comparison failures on SPARC targets.
12292 + Fix SPARC backend bug which caused aborts in final.c.
12293 + Fix sparc-hal-solaris2* configuration fragments.
12294 + Fix bug in sparc block profiling.
12295 + Fix obscure code generation bug for the PARISC targets.
12296 + Define __STDC_EXT__ for HPUX configurations.
12297 + Various POWERPC64 code generation bugfixes.
12298 + Fix abort for PPC targets using ELF (ex GNU/Linux).
12299 + Fix collect2 problems for AIX targets.
12300 + Correct handling of .file directive for PPC targets.
12301 + Fix bug in fix_trunc x86 patterns.
12302 + Fix x86 port to correctly pop the FP stack for functions that
12303 return structures in memory.
12304 + Fix minor bug in strlen x86 pattern.
12305 + Use stabs debugging instead of dwarf1 for x86-solaris targets.
12306 + Fix template repository code to handle leading underscore in
12308 + Fix weak/weak alias support for OpenBSD.
12309 + GNU/Linux for the ARM has C++ compatible include files.
12310 * Language & Runtime specific fixes.
12311 + Fix handling of constructor attribute in the C front-end which
12312 caused problems building the Chill runtime library on some
12314 + Fix minor problem merging type qualifiers in the C front-end.
12315 + Fix aliasing bug for pointers and references (C/C++).
12316 + Fix incorrect "non-constant initializer bug" when -traditional
12317 or -fwritable-strings is enabled.
12318 + Fix build error for Chill front-end on SunOS.
12319 + Do not complain about duplicate instantiations when using
12321 + Fix array bounds handling in C++ front-end which caused
12322 problems with dwarf debugging information in some
12324 + Fix minor namespace problem.
12325 + Fix problem linking java programs.
12327 Additional Changes in GCC 2.95.3
12329 * Generic bugfixes and improvements
12330 + Fix numerous problems that caused incorrect optimization in
12331 the register reloading code.
12332 + Fix numerous problems that caused incorrect optimization in
12333 the loop optimizer.
12334 + Fix aborts in the functions build_insn_chain and scan_loops
12335 under some circumstances.
12336 + Fix an alias analysis bug.
12337 + Fix an infinite compilation bug in the combiner.
12338 + A few problems with complex number support have been fixed.
12339 + It is no longer possible for gcc to act as a fork bomb when
12340 installed incorrectly.
12341 + The -fpack-struct option should be recognized now.
12342 + Fixed a bug that caused incorrect code to be generated due to
12343 a lost stack adjustment.
12344 * Platform specific bugfixes and improvements
12345 + Support building ARM toolchains hosted on Windows.
12346 + Fix attribute calculations in ARM toolchains.
12347 + arm-linux support has been improved.
12348 + Fix a PIC failure on sparc targets.
12349 + On ix86 targets, the regparm attribute should now work
12351 + Several updates for the h8300 port.
12352 + Fix problem building libio with glibc 2.2.
12355 For questions related to the use of GCC, please consult these web
12356 pages and the [17]GCC manuals. If that fails, the
12357 [18]gcc-help@gcc.gnu.org mailing list might help. Comments on these
12358 web pages and the development of GCC are welcome on our developer
12359 list at [19]gcc@gcc.gnu.org. All of [20]our lists have public
12362 Copyright (C) [21]Free Software Foundation, Inc. Verbatim copying and
12363 distribution of this entire article is permitted in any medium,
12364 provided this notice is preserved.
12366 These pages are [22]maintained by the GCC team. Last modified
12371 1. http://gcc.gnu.org/news/spill.html
12372 2. http://gcc.gnu.org/news/lcm.html
12373 3. http://gcc.gnu.org/news/cprop.html
12374 4. http://gcc.gnu.org/news/cfg.html
12375 5. http://gcc.gnu.org/news/dse.html
12376 6. http://gcc.gnu.org/news/hoist.html
12377 7. http://gcc.gnu.org/news/alias.html
12378 8. http://gcc.gnu.org/gcc-2.95/c++features.html
12379 9. http://gcc.gnu.org/onlinedocs/gcc-3.4.6/g77/News.html
12380 10. http://gcc.gnu.org/java/gcj-announce.txt
12381 11. http://gcc.gnu.org/news/javaannounce.html
12382 12. http://gcc.gnu.org/c99status.html
12383 13. http://gcc.gnu.org/news/chill.html
12384 14. http://gcc.gnu.org/news/sparc.html
12385 15. http://gcc.gnu.org/news/egcs-vcg.html
12386 16. http://gcc.gnu.org/egcs-1.0/features-2.8.html
12387 17. http://gcc.gnu.org/onlinedocs/
12388 18. mailto:gcc-help@gcc.gnu.org
12389 19. mailto:gcc@gcc.gnu.org
12390 20. http://gcc.gnu.org/lists.html
12391 21. http://www.fsf.org/
12392 22. http://gcc.gnu.org/about.html
12393 23. http://validator.w3.org/check/referer
12394 ======================================================================
12395 http://gcc.gnu.org/gcc-2.95/caveats.html
12399 * GCC 2.95 will issue an error for invalid asm statements that had
12400 been silently accepted by earlier versions of the compiler. This is
12401 particularly noticeable when compiling older versions of the Linux
12402 kernel (2.0.xx). Please refer to the FAQ (as shipped with GCC 2.95)
12403 for more information on this issue.
12404 * GCC 2.95 implements type based alias analysis to disambiguate
12405 memory references. Some programs, particularly the Linux kernel
12406 violate ANSI/ISO aliasing rules and therefore may not operate
12407 correctly when compiled with GCC 2.95. Please refer to the FAQ (as
12408 shipped with GCC 2.95) for more information on this issue.
12409 * GCC 2.95 has a known bug in its handling of complex variables for
12410 64bit targets. Instead of silently generating incorrect code, GCC
12411 2.95 will issue a fatal error for situations it can not handle.
12412 This primarily affects the Fortran community as Fortran makes more
12413 use of complex variables than C or C++.
12414 * GCC 2.95 has an integrated libstdc++, but does not have an
12415 integrated libg++. Furthermore old libg++ releases will not work
12416 with GCC 2.95. You can retrieve a recent copy of libg++ from the
12418 Note most C++ programs only need libstdc++.
12419 * Exception handling may not work with shared libraries, particularly
12420 on alphas, hppas, rs6000/powerpc and mips based platforms.
12421 Exception handling is known to work on x86 GNU/Linux platforms with
12423 * In general, GCC 2.95 is more rigorous about rejecting invalid C++
12424 code or deprecated C++ constructs than G++ 2.7, G++ 2.8, EGCS 1.0,
12425 or EGCS 1.1. As a result it may be necessary to fix C++ code before
12426 it will compile with GCC 2.95.
12427 * G++ is also converting toward the ISO C++ standard; as a result
12428 code which was previously valid (and thus accepted by other
12429 compilers and older versions of g++) may no longer be accepted. The
12430 flag -fpermissive may allow some non-conforming code to compile
12432 * GCC 2.95 compiled C++ code is not binary compatible with EGCS
12433 1.1.x, EGCS 1.0.x or GCC 2.8.x.
12434 * GCC 2.95 does not have changes from the GCC 2.8 tree that were made
12435 between Sept 30, 1998 and April 30, 1999 (the official end of the
12436 GCC 2.8 project). Future GCC releases will include all the changes
12437 from the defunct GCC 2.8 sources.
12440 For questions related to the use of GCC, please consult these web
12441 pages and the [2]GCC manuals. If that fails, the
12442 [3]gcc-help@gcc.gnu.org mailing list might help. Comments on these
12443 web pages and the development of GCC are welcome on our developer
12444 list at [4]gcc@gcc.gnu.org. All of [5]our lists have public archives.
12446 Copyright (C) [6]Free Software Foundation, Inc. Verbatim copying and
12447 distribution of this entire article is permitted in any medium,
12448 provided this notice is preserved.
12450 These pages are [7]maintained by the GCC team. Last modified
12455 1. ftp://gcc.gnu.org/pub/gcc/infrastructure/libg++-2.8.1.3.tar.gz
12456 2. http://gcc.gnu.org/onlinedocs/
12457 3. mailto:gcc-help@gcc.gnu.org
12458 4. mailto:gcc@gcc.gnu.org
12459 5. http://gcc.gnu.org/lists.html
12460 6. http://www.fsf.org/
12461 7. http://gcc.gnu.org/about.html
12462 8. http://validator.w3.org/check/referer
12463 ======================================================================
12464 http://gcc.gnu.org/egcs-1.1/index.html
12468 September 3, 1998: We are pleased to announce the release of EGCS 1.1.
12469 December 1, 1998: We are pleased to announce the release of EGCS 1.1.1.
12470 March 15, 1999: We are pleased to announce the release of EGCS 1.1.2.
12472 EGCS is a free software project to further the development of the GNU
12473 compilers using an open development environment.
12475 EGCS 1.1 is a major new release of the EGCS compiler system. It has
12476 been [1]extensively tested and is believed to be stable and suitable
12477 for widespread use.
12479 EGCS 1.1 is based on an June 6, 1998 snapshot of the GCC 2.8
12480 development sources; it contains all of the new features found in GCC
12481 2.8.1 as well as all new development from GCC up to June 6, 1998.
12483 EGCS 1.1 also contains many improvements and features not found in GCC
12484 or in older versions of EGCS:
12485 * Global common subexpression elimination and global constant/copy
12486 propagation (aka [2]gcse)
12487 * Ongoing improvements to the [3]alias analysis support to allow for
12488 better optimizations throughout the compiler.
12489 * Vastly improved [4]C++ compiler and integrated C++ runtime
12491 * Fixes for the /tmp symlink race security problems.
12492 * New targets including mips16, arm-thumb and 64 bit PowerPC.
12493 * Improvements to GNU Fortran (g77) compiler and runtime library made
12494 since g77 version 0.5.23.
12496 See the [5]new features page for a more complete list of new features
12497 found in EGCS 1.1 releases.
12499 EGCS 1.1.1 is a minor update to fix several serious problems in EGCS
12501 * General improvements and fixes
12502 + Avoid some stack overflows when compiling large functions.
12503 + Avoid incorrect loop invariant code motions.
12504 + Fix some core dumps on Linux kernel code.
12505 + Bring back the imake -Di386 and friends fix from EGCS 1.0.2.
12506 + Fix code generation problem in gcse.
12507 + Various documentation related fixes.
12508 * g++/libstdc++ improvements and fixes
12509 + MT safe EH fix for setjmp/longjmp based exception handling.
12510 + Fix a few bad interactions between optimization and exception
12512 + Fixes for demangling of template names starting with "__".
12513 + Fix a bug that would fail to run destructors in some cases
12515 + Fix 'new' of classes with virtual bases.
12516 + Fix crash building Qt on the Alpha.
12517 + Fix failure compiling WIFEXITED macro on GNU/Linux.
12518 + Fix some -frepo failures.
12519 * g77 and libf2c improvements and fixes
12520 + Various documentation fixes.
12521 + Avoid compiler crash on RAND intrinsic.
12522 + Fix minor bugs in makefiles exposed by BSD make programs.
12523 + Define _XOPEN_SOURCE for libI77 build to avoid potential
12524 problems on some 64-bit systems.
12525 + Fix problem with implicit endfile on rewind.
12526 + Fix spurious recursive I/O errors.
12527 * platform specific improvements and fixes
12528 + Match all versions of UnixWare7.
12529 + Do not assume x86 SVR4 or UnixWare targets can handle stabs.
12530 + Fix PPC/RS6000 LEGITIMIZE_ADDRESS macro and bug in conversion
12531 from unsigned ints to double precision floats.
12532 + Fix ARM ABI issue with NetBSD.
12533 + Fix a few arm code generation bugs.
12534 + Fixincludes will fix additional broken SCO OpenServer header
12536 + Fix a m68k backend bug which caused invalid offsets in reg+d
12538 + Fix problems with 64bit AIX 4.3 support.
12539 + Fix handling of long longs for varargs/stdarg functions on the
12541 + Minor fixes to CPP predefines for Windows.
12542 + Fix code generation problems with gpr<->fpr copies for 64bit
12544 + Fix a few coldfire code generation bugs.
12545 + Fix some more header file problems on SunOS 4.x.
12546 + Fix assert.h handling for RTEMS.
12547 + Fix Windows handling of TREE_SYMBOL_REFERENCED.
12548 + Fix x86 compiler abort in reg-stack pass.
12549 + Fix cygwin/windows problem with section attributes.
12550 + Fix Alpha code generation problem exposed by SMP Linux
12552 + Fix typo in m68k 32->64bit integer conversion.
12553 + Make sure target libraries build with -fPIC for PPC & Alpha
12556 EGCS 1.1.2 is a minor update to fix several serious problems in EGCS
12558 * General improvements and fixes
12559 + Fix bug in loop optimizer which caused the SPARC (and
12560 potentially other) ports to segfault.
12561 + Fix infinite recursion in alias analysis and combiner code.
12562 + Fix bug in regclass preferencing.
12563 + Fix incorrect loop reversal which caused incorrect code to be
12564 generated for several targets.
12565 + Fix return value for builtin memcpy.
12566 + Reduce compile time for certain loops which exposed quadratic
12567 behavior in the loop optimizer.
12568 + Fix bug which caused volatile memory to be written multiple
12569 times when only one write was needed/desired.
12570 + Fix compiler abort in caller-save.c
12571 + Fix combiner bug which caused incorrect code generation for
12572 certain division by constant operations.
12573 + Fix incorrect code generation due to a bug in range check
12575 + Fix incorrect code generation due to mis-handling of clobbered
12577 + Fix compiler abort/segfault due to incorrect register
12578 splitting when unrolling loops.
12579 + Fix code generation involving autoincremented addresses with
12581 + Work around bug in the scheduler which caused qt to be
12582 mis-compiled on some platforms.
12583 + Fix code generation problems with -fshort-enums.
12584 + Tighten security for temporary files.
12585 + Improve compile time for codes which make heavy use of
12586 overloaded functions.
12587 + Fix multiply defined constructor/destructor symbol problems.
12588 + Avoid setting bogus RPATH environment variable during
12590 + Avoid GNU-make dependencies in the texinfo subdir.
12591 + Install CPP wrapper script in $(prefix)/bin if --enable-cpp.
12592 --enable-cpp=<dirname> can be used to specify an additional
12593 install directory for the cpp wrapper script.
12594 + Fix CSE bug which caused incorrect label-label refs to appear
12596 + Avoid linking in EH routines from libgcc if they are not
12598 + Avoid obscure bug in aliasing code.
12599 + Fix bug in weak symbol handling.
12600 * Platform-specific improvements and fixes
12601 + Fix detection of PPro/PII on Unixware 7.
12602 + Fix compiler segfault when building spec99 and other programs
12604 + Fix code-generation bugs for integer and floating point
12605 conditional move instructions on the PPro/PII.
12606 + Use fixincludes to fix byteorder problems on i?86-*-sysv.
12607 + Fix build failure for the arc port.
12608 + Fix floating point format configuration for i?86-gnu port.
12609 + Fix problems with hppa1.0-hp-hpux10.20 configuration when
12610 threads are enabled.
12611 + Fix coldfire code generation bugs.
12612 + Fix "unrecognized insn" problems for Alpha and PPC ports.
12613 + Fix h8/300 code generation problem with floating point values
12615 + Fix unrecognized insn problems for the m68k port.
12616 + Fix namespace-pollution problem for the x86 port.
12617 + Fix problems with old assembler on x86 NeXT systems.
12618 + Fix PIC code-generation problems for the SPARC port.
12619 + Fix minor bug with LONG_CALLS in PowerPC SVR4 support.
12620 + Fix minor ISO namespace violation in Alpha varargs/stdarg
12622 + Fix incorrect "braf" instruction usage for the SH port.
12623 + Fix minor bug in va-sh which prevented its use with -ansi.
12624 + Fix problems recognizing and supporting FreeBSD.
12625 + Handle OpenBSD systems correctly.
12626 + Minor fixincludes fix for Digital UNIX 4.0B.
12627 + Fix problems with ctors/dtors in SCO shared libraries.
12628 + Abort instead of generating incorrect code for PPro/PII
12629 floating point conditional moves.
12630 + Avoid multiply defined symbols on GNU/Linux systems using
12632 + Fix abort in alpha compiler.
12633 * Fortran-specific fixes
12634 + Fix the IDate intrinsic (VXT) (in libg2c) so the returned year
12635 is in the documented, non-Y2K-compliant range of 0-99, instead
12636 of being returned as 100 in the year 2000.
12637 + Fix the `Date_and_Time' intrinsic (in libg2c) to return the
12638 milliseconds value properly in Values(8).
12639 + Fix the `LStat' intrinsic (in libg2c) to return device-ID
12640 information properly in SArray(7).
12642 Each release includes installation instructions in both HTML and
12643 plaintext forms (see the INSTALL directory in the toplevel directory of
12644 the distribution). However, we also keep the most up to date
12645 [6]installation instructions and [7]build/test status on our web page.
12646 We will update those pages as new information becomes available.
12648 The EGCS project would like to thank the numerous people that have
12649 contributed new features, test results, bugfixes, etc. This [8]amazing
12650 group of volunteers is what makes EGCS successful.
12652 And finally, we can't in good conscience fail to mention some
12653 [9]caveats to using EGCS 1.1.
12655 Download EGCS from egcs.cygnus.com (USA California).
12657 The EGCS 1.1 release is also available on many mirror sites.
12658 [10]Goto mirror list to find a closer site.
12661 For questions related to the use of GCC, please consult these web
12662 pages and the [11]GCC manuals. If that fails, the
12663 [12]gcc-help@gcc.gnu.org mailing list might help. Comments on these
12664 web pages and the development of GCC are welcome on our developer
12665 list at [13]gcc@gcc.gnu.org. All of [14]our lists have public
12668 Copyright (C) [15]Free Software Foundation, Inc. Verbatim copying and
12669 distribution of this entire article is permitted in any medium,
12670 provided this notice is preserved.
12672 These pages are [16]maintained by the GCC team. Last modified
12677 1. http://gcc.gnu.org/egcs-1.1/egcs-1.1-test.html
12678 2. http://gcc.gnu.org/news/gcse.html
12679 3. http://gcc.gnu.org/news/alias.html
12680 4. http://gcc.gnu.org/egcs-1.1/c++features.html
12681 5. http://gcc.gnu.org/egcs-1.1/features.html
12682 6. http://gcc.gnu.org/install/
12683 7. http://gcc.gnu.org/egcs-1.1/buildstat.html
12684 8. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
12685 9. http://gcc.gnu.org/egcs-1.1/caveats.html
12686 10. http://gcc.gnu.org/mirrors.html
12687 11. http://gcc.gnu.org/onlinedocs/
12688 12. mailto:gcc-help@gcc.gnu.org
12689 13. mailto:gcc@gcc.gnu.org
12690 14. http://gcc.gnu.org/lists.html
12691 15. http://www.fsf.org/
12692 16. http://gcc.gnu.org/about.html
12693 17. http://validator.w3.org/check/referer
12694 ======================================================================
12695 http://gcc.gnu.org/egcs-1.1/features.html
12697 EGCS 1.1 new features
12699 * Integrated GNU Fortran (g77) compiler and runtime library with
12700 improvements, based on g77 version 0.5.23.
12701 * Vast improvements in the C++ compiler; so many they have [1]page of
12703 * Compiler implements [2]global common subexpression elimination and
12704 global copy/constant propagation.
12705 * More major improvements in the [3]alias analysis code.
12706 * More major improvements in the exception handling code to improve
12707 performance, lower static overhead and provide the infrastructure
12708 for future improvements.
12709 * The infamous /tmp symlink race security problems have been fixed.
12710 * The regmove optimization pass has been nearly completely rewritten
12711 to improve performance of generated code.
12712 * The compiler now recomputes register usage information before local
12713 register allocation. By providing more accurate information to the
12714 priority based allocator, we get better register allocation.
12715 * The register reloading phase of the compiler optimizes spill code
12716 much better than in previous releases.
12717 * Some bad interactions between the register allocator and
12718 instruction scheduler have been fixed, resulting in much better
12719 code for certain programs. Additionally, we have tuned the
12720 scheduler in various ways to improve performance of generated code
12721 for some architectures.
12722 * The compiler's branch shortening algorithms have been significantly
12723 improved to work better on targets which align jump targets.
12724 * The compiler now supports -Os to prefer optimizing for code space
12725 over optimizing for code speed.
12726 * The compiler will now totally eliminate library calls which compute
12727 constant values. This primarily helps targets with no integer
12728 div/mul support and targets without floating point support.
12729 * The compiler now supports an extensive "--help" option.
12730 * cpplib has been greatly improved and may be suitable for limited
12732 * Memory footprint for the compiler has been significantly reduced
12733 for some pathological cases.
12734 * The time to build EGCS has been improved for certain targets
12735 (particularly the alpha and mips platforms).
12736 * Many infrastructure improvements throughout the compiler, plus the
12737 usual mountain of bugfixes and minor improvements.
12738 * Target dependent improvements:
12739 + SPARC port now includes V8 plus and V9 support as well as
12740 performance tuning for Ultra class machines. The SPARC port
12741 now uses the Haifa scheduler.
12742 + Alpha port has been tuned for the EV6 processor and has an
12743 optimized expansion of memcpy/bzero. The Alpha port now uses
12744 the Haifa scheduler.
12745 + RS6000/PowerPC: support for the Power64 architecture and AIX
12746 4.3. The RS6000/PowerPC port now uses the Haifa scheduler.
12747 + x86: Alignment of static store data and jump targets is per
12748 Intel recommendations now. Various improvements throughout the
12749 x86 port to improve performance on Pentium processors
12750 (including improved epilogue sequences for Pentium chips and
12751 backend improvements which should help register allocation on
12752 all x86 variants. Conditional move support has been fixed and
12753 enabled for PPro processors. The x86 port also better supports
12754 64bit operations now. Unixware 7, a System V Release 5 target,
12755 is now supported and SCO OpenServer targets can support GAS.
12756 + MIPS has improved multiply/multiply-add support and now
12757 includes mips16 ISA support.
12758 + M68k has many micro-optimizations and Coldfire fixes.
12759 * Core compiler is based on the GCC development tree from June 9,
12760 1998, so we have all of the [4]features found in GCC 2.8.
12763 For questions related to the use of GCC, please consult these web
12764 pages and the [5]GCC manuals. If that fails, the
12765 [6]gcc-help@gcc.gnu.org mailing list might help. Comments on these
12766 web pages and the development of GCC are welcome on our developer
12767 list at [7]gcc@gcc.gnu.org. All of [8]our lists have public archives.
12769 Copyright (C) [9]Free Software Foundation, Inc. Verbatim copying and
12770 distribution of this entire article is permitted in any medium,
12771 provided this notice is preserved.
12773 These pages are [10]maintained by the GCC team. Last modified
12778 1. http://gcc.gnu.org/egcs-1.1/c++features.html
12779 2. http://gcc.gnu.org/news/gcse.html
12780 3. http://gcc.gnu.org/news/alias.html
12781 4. http://gcc.gnu.org/egcs-1.0/features-2.8.html
12782 5. http://gcc.gnu.org/onlinedocs/
12783 6. mailto:gcc-help@gcc.gnu.org
12784 7. mailto:gcc@gcc.gnu.org
12785 8. http://gcc.gnu.org/lists.html
12786 9. http://www.fsf.org/
12787 10. http://gcc.gnu.org/about.html
12788 11. http://validator.w3.org/check/referer
12789 ======================================================================
12790 http://gcc.gnu.org/egcs-1.1/caveats.html
12794 * EGCS has an integrated libstdc++, but does not have an integrated
12795 libg++. Furthermore old libg++ releases will not work with EGCS; HJ
12796 Lu has made a libg++-2.8.1.2 snapshot available which may work with
12798 Note most C++ programs only need libstdc++.
12799 * Exception handling may not work with shared libraries, particularly
12800 on alphas, hppas, rs6000/powerpc and mips based platforms.
12801 Exception handling is known to work on x86-linux platforms with
12803 * Some versions of the Linux kernel have bugs which prevent them from
12804 being compiled or from running when compiled by EGCS. See the FAQ
12805 (as shipped with EGCS 1.1) for additional information.
12806 * In general, EGCS is more rigorous about rejecting invalid C++ code
12807 or deprecated C++ constructs than g++-2.7, g++-2.8 or EGCS 1.0. As
12808 a result it may be necessary to fix C++ code before it will compile
12810 * G++ is also converting toward the ISO C++ standard; as a result
12811 code which was previously valid (and thus accepted by other
12812 compilers and older versions of g++) may no longer be accepted.
12813 * EGCS 1.1 compiled C++ code is not binary compatible with EGCS 1.0.x
12814 or GCC 2.8.x due to changes necessary to support thread safe
12815 exception handling.
12818 For questions related to the use of GCC, please consult these web
12819 pages and the [1]GCC manuals. If that fails, the
12820 [2]gcc-help@gcc.gnu.org mailing list might help. Comments on these
12821 web pages and the development of GCC are welcome on our developer
12822 list at [3]gcc@gcc.gnu.org. All of [4]our lists have public archives.
12824 Copyright (C) [5]Free Software Foundation, Inc. Verbatim copying and
12825 distribution of this entire article is permitted in any medium,
12826 provided this notice is preserved.
12828 These pages are [6]maintained by the GCC team. Last modified
12833 1. http://gcc.gnu.org/onlinedocs/
12834 2. mailto:gcc-help@gcc.gnu.org
12835 3. mailto:gcc@gcc.gnu.org
12836 4. http://gcc.gnu.org/lists.html
12837 5. http://www.fsf.org/
12838 6. http://gcc.gnu.org/about.html
12839 7. http://validator.w3.org/check/referer
12840 ======================================================================
12841 http://gcc.gnu.org/egcs-1.0/index.html
12845 December 3, 1997: We are pleased to announce the release of EGCS 1.0.
12846 January 6, 1998: We are pleased to announce the release of EGCS 1.0.1.
12847 March 16, 1998: We are pleased to announce the release of EGCS 1.0.2.
12848 May 15, 1998 We are pleased to announce the release of EGCS 1.0.3.
12850 EGCS is a collaborative effort involving several groups of hackers
12851 using an open development model to accelerate development and testing
12852 of GNU compilers and runtime libraries.
12854 An important goal of EGCS is to allow wide scale testing of
12855 experimental features and optimizations; therefore, EGCS contains some
12856 features and optimizations which are still under development. However,
12857 EGCS has been carefully tested and should be comparable in quality to
12860 EGCS 1.0 is based on an August 2, 1997 snapshot of the GCC 2.8
12861 development sources; it contains nearly all of the new features found
12864 EGCS 1.0 also contains many improvements and features not found in GCC
12865 2.7 and even the GCC 2.8 series (which was released after the original
12867 * Integrated C++ runtime libraries, including support for most major
12869 * The integrated libstdc++ library includes a verbatim copy of SGI's
12871 * Integrated GNU Fortran compiler.
12872 * New instruction scheduler.
12873 * New alias analysis code.
12875 See the [1]new features page for a more complete list of new features.
12877 EGCS 1.0.1 is a minor update to the EGCS 1.0 compiler to fix a few
12878 critical bugs and add support for Red Hat 5.0 Linux. Changes since the
12880 * Add support for Red Hat 5.0 Linux and better support for Linux
12881 systems using glibc2.
12882 Many programs failed to link when compiled with EGCS 1.0 on Red Hat
12883 5.0 or on systems with newer versions of glibc2. EGCS 1.0.1 should
12884 fix these problems.
12885 * Compatibility with both EGCS 1.0 and GCC 2.8 libgcc exception
12886 handling interfaces.
12887 To avoid future compatibility problems, we strongly urge anyone who
12888 is planning on distributing shared libraries that contain C++ code
12889 to upgrade to EGCS 1.0.1 first.
12890 Soon after EGCS 1.0 was released, the GCC developers made some
12891 incompatible changes in libgcc's exception handling interfaces.
12892 These changes were needed to solve problems on some platforms. This
12893 means that GCC 2.8.0, when released, will not be seamlessly
12894 compatible with shared libraries built by EGCS 1.0. The reason is
12895 that the libgcc.a in GCC 2.8.0 will not contain a function needed
12896 by the old interface.
12897 The result of this is that there may be compatibility problems with
12898 shared libraries built by EGCS 1.0 when used with GCC 2.8.0.
12899 With EGCS 1.0.1, generated code uses the new (GCC 2.8.0) interface,
12900 and libgcc.a has the support routines for both the old and the new
12901 interfaces (so EGCS 1.0.1 and EGCS 1.0 code can be freely mixed,
12902 and EGCS 1.0.1 and GCC 2.8.0 code can be freely mixed).
12903 The maintainers of GCC 2.x have decided against including seamless
12904 support for the old interface in 2.8.0, since it was never
12905 "official", so to avoid future compatibility problems we recommend
12906 against distributing any shared libraries built by EGCS 1.0 that
12907 contain C++ code (upgrade to 1.0.1 and use that).
12908 * Various bugfixes in the x86, hppa, mips, and rs6000/ppc backends.
12909 The x86 changes fix code generation errors exposed when building
12910 glibc2 and the usual GNU/Linux dynamic linker (ld.so).
12911 The hppa change fixes a compiler abort when configured for use with
12913 The MIPS changes fix problems with the definition of LONG_MAX on
12914 newer systems, allow for command line selection of the target ABI,
12915 and fix one code generation problem.
12916 The rs6000/ppc change fixes some problems with passing structures
12917 to varargs/stdarg functions.
12918 * A few machine independent bugfixes, mostly to fix code generation
12919 errors when building Linux kernels or glibc.
12920 * Fix a few critical exception handling and template bugs in the C++
12922 * Fix Fortran namelist bug on alphas.
12923 * Fix build problems on x86-solaris systems.
12925 EGCS 1.0.2 is a minor update to the EGCS 1.0.1 compiler to fix several
12926 serious problems in EGCS 1.0.1.
12927 * General improvements and fixes
12928 + Memory consumption significantly reduced, especially for
12929 templates and inline functions.
12930 + Fix various problems with glibc2.1.
12931 + Fix loop optimization bug exposed by rs6000/ppc port.
12932 + Fix to avoid potential code generation problems in jump.c.
12933 + Fix some undefined symbol problems in dwarf1 debug support.
12934 * g++/libstdc++ improvements and fixes
12935 + libstdc++ in the EGCS release has been updated and should be
12936 link compatible with libstdc++-2.8.
12937 + Various fixes in libio/libstdc++ to work better on GNU/Linux
12939 + Fix problems with duplicate symbols on systems that do not
12940 support weak symbols.
12941 + Memory corruption bug and undefined symbols in bastring have
12943 + Various exception handling fixes.
12944 + Fix compiler abort for very long thunk names.
12945 * g77 improvements and fixes
12946 + Fix compiler crash for omitted bound in Fortran CASE
12948 + Add missing entries to g77 lang-options.
12949 + Fix problem with -fpedantic in the g77 compiler.
12950 + Fix "backspace" problem with g77 on alphas.
12951 + Fix x86 backend problem with Fortran literals and -fpic.
12952 + Fix some of the problems with negative subscripts for g77 on
12954 + Fixes for Fortran builds on cygwin32/mingw32.
12955 * platform specific improvements and fixes
12956 + Fix long double problems on x86 (exposed by glibc).
12957 + x86 ports define i386 again to keep imake happy.
12958 + Fix exception handling support on NetBSD ports.
12959 + Several changes to collect2 to fix many problems with AIX.
12960 + Define __ELF__ for GNU/Linux on rs6000.
12961 + Fix -mcall-linux problem on GNU/Linux on rs6000.
12962 + Fix stdarg/vararg problem for GNU/Linux on rs6000.
12963 + Allow autoconf to select a proper install problem on AIX 3.1.
12964 + m68k port support includes -mcpu32 option as well as cpu32
12966 + Fix stdarg bug for irix6.
12967 + Allow EGCS to build on irix5 without the gnu assembler.
12968 + Fix problem with static linking on sco5.
12969 + Fix bootstrap on sco5 with native compiler.
12970 + Fix for abort building newlib on H8 target.
12971 + Fix fixincludes handling of math.h on SunOS.
12972 + Minor fix for Motorola 3300 m68k systems.
12974 EGCS 1.0.3 is a minor update to the EGCS 1.0.2 compiler to fix a few
12975 problems reported by Red Hat for builds of Red Hat 5.1.
12976 * Generic bugfixes:
12977 + Fix a typo in the libio library which resulted in incorrect
12978 behavior of istream::get.
12979 + Fix the Fortran negative array index problem.
12980 + Fix a major problem with the ObjC runtime thread support
12982 + Reduce memory consumption of the Haifa scheduler.
12983 * Target specific bugfixes:
12984 + Fix one x86 floating point code generation bug exposed by
12986 + Fix one x86 internal compiler error exposed by glibc2 builds.
12987 + Fix profiling bugs on the Alpha.
12988 + Fix ImageMagick & emacs 20.2 build problems on the Alpha.
12989 + Fix rs6000/ppc bug when converting values from integer types
12990 to floating point types.
12992 The EGCS 1.0 releases include installation instructions in both HTML
12993 and plaintext forms (see the INSTALL directory in the toplevel
12994 directory of the distribution). However, we also keep the most up to
12995 date [2]installation instructions and [3]build/test status on our web
12996 page. We will update those pages as new information becomes available.
12998 And, we can't in good conscience fail to mention some [4]caveats to
13001 Update: Big thanks to Stanford for providing a high speed link for
13002 downloading EGCS (go.cygnus.com)!
13004 Download EGCS from ftp.cygnus.com (USA California) or go.cygnus.com
13005 (USA California -- High speed link provided by Stanford).
13007 The EGCS 1.0 release is also available many mirror sites.
13008 [5]Goto mirror list to find a closer site
13010 We'd like to thank the numerous people that have contributed new
13011 features, test results, bugfixes, etc. Unfortunately, they're far too
13012 numerous to mention by name.
13015 For questions related to the use of GCC, please consult these web
13016 pages and the [6]GCC manuals. If that fails, the
13017 [7]gcc-help@gcc.gnu.org mailing list might help. Comments on these
13018 web pages and the development of GCC are welcome on our developer
13019 list at [8]gcc@gcc.gnu.org. All of [9]our lists have public archives.
13021 Copyright (C) [10]Free Software Foundation, Inc. Verbatim copying and
13022 distribution of this entire article is permitted in any medium,
13023 provided this notice is preserved.
13025 These pages are [11]maintained by the GCC team. Last modified
13030 1. http://gcc.gnu.org/egcs-1.0/features.html
13031 2. http://gcc.gnu.org/install/
13032 3. http://gcc.gnu.org/egcs-1.0/buildstat.html
13033 4. http://gcc.gnu.org/egcs-1.0/caveats.html
13034 5. http://gcc.gnu.org/mirrors.html
13035 6. http://gcc.gnu.org/onlinedocs/
13036 7. mailto:gcc-help@gcc.gnu.org
13037 8. mailto:gcc@gcc.gnu.org
13038 9. http://gcc.gnu.org/lists.html
13039 10. http://www.fsf.org/
13040 11. http://gcc.gnu.org/about.html
13041 12. http://validator.w3.org/check/referer
13042 ======================================================================
13043 http://gcc.gnu.org/egcs-1.0/features.html
13047 * Core compiler is based on the gcc2 development tree from Aug 2,
13048 1997, so we have most of the [1]features found in GCC 2.8.
13049 * Integrated GNU Fortran compiler based on g77-0.5.22-19970929.
13050 * Vast improvements in the C++ compiler; so many they have [2]page of
13052 * Integrated C++ runtime libraries, including support for most major
13054 * New instruction scheduler from IBM Haifa which includes support for
13055 function wide instruction scheduling as well as superscalar
13057 * Significantly improved alias analysis code.
13058 * Improved register allocation for two address machines.
13059 * Significant code generation improvements for Fortran code on
13061 * Various optimizations from the g77 project as well as improved loop
13063 * Dwarf2 debug format support for some targets.
13064 * egcs libstdc++ includes the SGI STL implementation without changes.
13065 * As a result of these and other changes, egcs libstc++ is not binary
13066 compatible with previous releases of libstdc++.
13067 * Various new ports -- UltraSPARC, Irix6.2 & Irix6.3 support, The SCO
13068 Openserver 5 family (5.0.{0,2,4} and Internet FastStart 1.0 and
13069 1.1), Support for RTEMS on several embedded targets, Support for
13070 arm-linux, Mitsubishi M32R, Hitachi H8/S, Matsushita MN102 and
13071 MN103, NEC V850, Sparclet, Solaris & GNU/Linux on PowerPCs, etc.
13072 * Integrated testsuites for gcc, g++, g77, libstdc++ and libio.
13073 * RS6000/PowerPC ports generate code which can run on all
13074 RS6000/PowerPC variants by default.
13075 * -mcpu= and -march= switches for the x86 port to allow better
13076 control over how the x86 port generates code.
13077 * Includes the template repository patch (aka repo patch); note the
13078 new template code makes repo obsolete for ELF systems using gnu-ld
13080 * Plus the usual assortment of bugfixes and improvements.
13083 For questions related to the use of GCC, please consult these web
13084 pages and the [3]GCC manuals. If that fails, the
13085 [4]gcc-help@gcc.gnu.org mailing list might help. Comments on these
13086 web pages and the development of GCC are welcome on our developer
13087 list at [5]gcc@gcc.gnu.org. All of [6]our lists have public archives.
13089 Copyright (C) [7]Free Software Foundation, Inc. Verbatim copying and
13090 distribution of this entire article is permitted in any medium,
13091 provided this notice is preserved.
13093 These pages are [8]maintained by the GCC team. Last modified
13098 1. http://gcc.gnu.org/egcs-1.0/features-2.8.html
13099 2. http://gcc.gnu.org/egcs-1.0/c++features.html
13100 3. http://gcc.gnu.org/onlinedocs/
13101 4. mailto:gcc-help@gcc.gnu.org
13102 5. mailto:gcc@gcc.gnu.org
13103 6. http://gcc.gnu.org/lists.html
13104 7. http://www.fsf.org/
13105 8. http://gcc.gnu.org/about.html
13106 9. http://validator.w3.org/check/referer
13107 ======================================================================
13108 http://gcc.gnu.org/egcs-1.0/caveats.html
13112 * EGCS has an integrated libstdc++, but does not have an integrated
13113 libg++. Furthermore old libg++ releases will not work with egc; HJ
13114 Lu has made a libg++-2.8.1.2 available which may work with EGCS.
13115 Note most C++ programs only need libstdc++.
13116 * Note that using -pedantic or -Wreturn-type can cause an explosion
13117 in the amount of memory needed for template-heavy C++ code, such as
13118 code that uses STL. Also note that -Wall includes -Wreturn-type, so
13119 if you use -Wall you will need to specify -Wno-return-type to turn
13121 * Exception handling may not work with shared libraries, particularly
13122 on alphas, hppas, and mips based platforms. Exception handling is
13123 known to work on x86-linux platforms with shared libraries.
13124 * Some versions of the Linux kernel have bugs which prevent them from
13125 being compiled or from running when compiled by EGCS. See the FAQ
13126 (as shipped with EGCS 1.0) for additional information.
13127 * In general, EGCS is more rigorous about rejecting invalid C++ code
13128 or deprecated C++ constructs than G++ 2.7. As a result it may be
13129 necessary to fix C++ code before it will compile with EGCS.
13130 * G++ is also aggressively tracking the C++ standard; as a result
13131 code which was previously valid (and thus accepted by other
13132 compilers and older versions of G++) may no longer be accepted.
13133 * EGCS 1.0 may not work with Red Hat Linux 5.0 on all targets. EGCS
13134 1.0.x and later releases should work with Red Hat Linux 5.0.
13137 For questions related to the use of GCC, please consult these web
13138 pages and the [1]GCC manuals. If that fails, the
13139 [2]gcc-help@gcc.gnu.org mailing list might help. Comments on these
13140 web pages and the development of GCC are welcome on our developer
13141 list at [3]gcc@gcc.gnu.org. All of [4]our lists have public archives.
13143 Copyright (C) [5]Free Software Foundation, Inc. Verbatim copying and
13144 distribution of this entire article is permitted in any medium,
13145 provided this notice is preserved.
13147 These pages are [6]maintained by the GCC team. Last modified
13152 1. http://gcc.gnu.org/onlinedocs/
13153 2. mailto:gcc-help@gcc.gnu.org
13154 3. mailto:gcc@gcc.gnu.org
13155 4. http://gcc.gnu.org/lists.html
13156 5. http://www.fsf.org/
13157 6. http://gcc.gnu.org/about.html
13158 7. http://validator.w3.org/check/referer
13159 ======================================================================