xref: /netbsd-src/external/gpl3/gcc.old/dist/NEWS (revision b7b7574d3bf8eeb51a1fa3977b59142ec6434a55)
1This file contains information about GCC releases which has been generated
2automatically 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
4that led to GCC 3. For information on GCC 2.8.1 and older releases of GCC 2,
5see ONEWS.
6
7======================================================================
8http://gcc.gnu.org/gcc-4.5/index.html
9
10                             GCC 4.5 Release Series
11
12   Jul 2, 2012
13
14   The [1]GNU project and the GCC developers are pleased to announce the
15   release of GCC 4.5.4.
16
17   This release is a bug-fix release, containing fixes for regressions in
18   GCC 4.5.3 relative to previous releases of GCC.
19
20Release History
21
22   GCC 4.5.4
23          Jul 2, 2012 ([2]changes)
24
25   GCC 4.5.3
26          Apr 28, 2011 ([3]changes)
27
28   GCC 4.5.2
29          Dec 16, 2010 ([4]changes)
30
31   GCC 4.5.1
32          Jul 31, 2010 ([5]changes)
33
34   GCC 4.5.0
35          April 14, 2010 ([6]changes)
36
37References and Acknowledgements
38
39   GCC used to stand for the GNU C Compiler, but since the compiler
40   supports several other languages aside from C, it now stands for the
41   GNU Compiler Collection.
42
43   A list of [7]successful builds is updated as new information becomes
44   available.
45
46   The GCC developers would like to thank the numerous people that have
47   contributed new features, improvements, bug fixes, and other changes as
48   well as test results to GCC. This [8]amazing group of volunteers is
49   what makes GCC successful.
50
51   For additional information about GCC please refer to the [9]GCC project
52   web site or contact the [10]GCC development mailing list.
53
54   To obtain GCC please use [11]our mirror sites or [12]our SVN server.
55
56
57    For questions related to the use of GCC, please consult these web
58    pages and the [13]GCC manuals. If that fails, the
59    [14]gcc-help@gcc.gnu.org mailing list might help. Comments on these
60    web pages and the development of GCC are welcome on our developer
61    list at [15]gcc@gcc.gnu.org. All of [16]our lists have public
62    archives.
63
64   Copyright (C) [17]Free Software Foundation, Inc. Verbatim copying and
65   distribution of this entire article is permitted in any medium,
66   provided this notice is preserved.
67
68   These pages are [18]maintained by the GCC team. Last modified
69   2012-07-02[19].
70
71References
72
73   1. http://www.gnu.org/
74   2. http://gcc.gnu.org/gcc-4.5/changes.html
75   3. http://gcc.gnu.org/gcc-4.5/changes.html
76   4. http://gcc.gnu.org/gcc-4.5/changes.html
77   5. http://gcc.gnu.org/gcc-4.5/changes.html
78   6. http://gcc.gnu.org/gcc-4.5/changes.html
79   7. http://gcc.gnu.org/gcc-4.5/buildstat.html
80   8. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
81   9. http://gcc.gnu.org/index.html
82  10. mailto:gcc@gcc.gnu.org
83  11. http://gcc.gnu.org/mirrors.html
84  12. http://gcc.gnu.org/svn.html
85  13. http://gcc.gnu.org/onlinedocs/
86  14. mailto:gcc-help@gcc.gnu.org
87  15. mailto:gcc@gcc.gnu.org
88  16. http://gcc.gnu.org/lists.html
89  17. http://www.fsf.org/
90  18. http://gcc.gnu.org/about.html
91  19. http://validator.w3.org/check/referer
92======================================================================
93http://gcc.gnu.org/gcc-4.5/changes.html
94
95                             GCC 4.5 Release Series
96                        Changes, New Features, and Fixes
97
98Caveats
99
100     * GCC now requires the [1]MPC library in order to build. See the
101       [2]prerequisites page for version requirements.
102     * Support for a number of older systems and recently unmaintained or
103       untested target ports of GCC has been declared obsolete in GCC 4.5.
104       Unless there is activity to revive them, the next release of GCC
105       will have their sources permanently removed.
106       The following ports for individual systems on particular
107       architectures have been obsoleted:
108          + IRIX releases before 6.5 (mips-sgi-irix5*,
109            mips-sgi-irix6.[0-4])
110          + Solaris 7 (*-*-solaris2.7)
111          + Tru64 UNIX releases before V5.1 (alpha*-dec-osf4*,
112            alpha-dec-osf5.0*)
113          + Details for the IRIX, Solaris 7, and Tru64 UNIX obsoletions
114            can be found in the [3]announcement.
115       Support for the classic POWER architecture implemented in the
116       original RIOS and RIOS2 processors of the old IBM RS/6000 product
117       line has been obsoleted in the rs6000 port. This does not affect
118       the new generation Power and PowerPC architectures.
119     * Support has been removed for all the [4]configurations obsoleted in
120       GCC 4.4.
121     * Support has been removed for the protoize and unprotoize utilities,
122       obsoleted in GCC 4.4.
123     * Support has been removed for tuning for Itanium1 (Merced) variants.
124       Note that code tuned for Itanium2 should also run correctly on
125       Itanium1.
126     * GCC now generates unwind info also for epilogues. DWARF debuginfo
127       generated by GCC now uses more features of DWARF3 than before, and
128       also some DWARF4 features. GDB older than 7.0 is not able to handle
129       either of these, so to debug GCC 4.5 generated binaries or
130       libraries GDB 7.0 or later is needed. You can disable use of DWARF4
131       features with the -gdwarf-3 -gstrict-dwarf options, or use
132       -gdwarf-2 -gstrict-dwarf to restrict GCC to just DWARF2, but
133       epilogue unwind info is emitted unconditionally whenever unwind
134       info is emitted.
135     * On x86 targets, code containing floating-point calculations may run
136       significantly slower when compiled with GCC 4.5 in strict C99
137       conformance mode than they did with earlier GCC versions. This is
138       due to stricter standard conformance of the compiler and can be
139       avoided by using the option -fexcess-precision=fast; also see
140       [5]below.
141     * The function attribute noinline no longer prevents GCC from cloning
142       the function. A new attribute noclone has been introduced for this
143       purpose. Cloning a function means that it is duplicated and the new
144       copy is specialized for certain contexts (for example when a
145       parameter is a known constant).
146
147General Optimizer Improvements
148
149     * The -save-temps now takes an optional argument. The -save-temps and
150       -save-temps=cwd switches write the temporary files in the current
151       working directory based on the original source file. The
152       -save-temps=obj switch will write files into the directory
153       specified with the -o option, and the intermediate filenames are
154       based on the output file. This will allow the user to get the
155       compiler intermediate files when doing parallel builds without two
156       builds of the same filename located in different directories from
157       interfering with each other.
158     * Debugging dumps are now created in the same directory as the object
159       file rather than in the current working directory. This allows the
160       user to get debugging dumps when doing parallel builds without two
161       builds of the same filename interfering with each other.
162     * GCC has been integrated with the [6]MPC library. This allows GCC to
163       evaluate complex arithmetic at compile time [7]more accurately. It
164       also allows GCC to evaluate calls to complex built-in math
165       functions having constant arguments and replace them at compile
166       time with their mathematically equivalent results. In doing so, GCC
167       can generate correct results regardless of the math library
168       implementation or floating point precision of the host platform.
169       This also allows GCC to generate identical results regardless of
170       whether one compiles in native or cross-compile configurations to a
171       particular target. The following built-in functions take advantage
172       of this new capability: cacos, cacosh, casin, casinh, catan,
173       catanh, ccos, ccosh, cexp, clog, cpow, csin, csinh, csqrt, ctan,
174       and ctanh. The float and long double variants of these functions
175       (e.g. csinf and csinl) are also handled.
176     * A new link-time optimizer has been added ([8]-flto). When this
177       option is used, GCC generates a bytecode representation of each
178       input file and writes it to specially-named sections in each object
179       file. When the object files are linked together, all the function
180       bodies are read from these named sections and instantiated as if
181       they had been part of the same translation unit. This enables
182       interprocedural optimizations to work across different files (and
183       even different languages), potentially improving the performance of
184       the generated code. To use the link-timer optimizer, -flto needs to
185       be specified at compile time and during the final link. If the
186       program does not require any symbols to be exported, it is possible
187       to combine -flto and the experimental [9]-fwhopr with
188       [10]-fwhole-program to allow the interprocedural optimizers to use
189       more aggressive assumptions.
190     * The automatic parallelization pass was enhanced to support
191       parallelization of outer loops.
192     * Automatic parallelization can be enabled as part of Graphite. In
193       addition to -ftree-parallelize-loops=, specify
194       -floop-parallelize-all to enable the Graphite-based optimization.
195     * The infrastructure for optimizing based on [11]restrict qualified
196       pointers has been rewritten and should result in code generation
197       improvements. Optimizations based on restrict qualified pointers
198       are now also available when using -fno-strict-aliasing.
199     * There is a new optimization pass that attempts to change prototype
200       of functions to avoid unused parameters, pass only relevant parts
201       of structures and turn arguments passed by reference to arguments
202       passed by value when possible. It is enabled by -O2 and above as
203       well as -Os and can be manually invoked using the new command-line
204       switch -fipa-sra.
205     * GCC now optimize exception handling code. In particular cleanup
206       regions that are proved to not have any effect are optimized out.
207
208New Languages and Language specific improvements
209
210  All languages
211
212     * The -fshow-column option is now on by default. This means error
213       messages now have a column associated with them.
214
215  Ada
216
217     * Compilation of programs heavily using discriminated record types
218       with variant parts has been sped up and generates more compact
219       code.
220     * Stack checking now works reasonably well on most plaforms. In some
221       specific cases, stack overflows may still fail to be detected, but
222       a compile-time warning will be issued for these cases.
223
224  C family
225
226     * If a header named in a #include directive is not found, the
227       compiler exits immediately. This avoids a cascade of errors arising
228       from declarations expected to be found in that header being
229       missing.
230     * A new built-in function __builtin_unreachable() has been added that
231       tells the compiler that control will never reach that point. It may
232       be used after asm statements that terminate by transferring control
233       elsewhere, and in other places that are known to be unreachable.
234     * The -Wlogical-op option now warns for logical expressions such as
235       (c == 1 && c == 2) and (c != 1 || c != 2), which are likely to be
236       mistakes. This option is disabled by default.
237     * An asm goto feature has been added to allow asm statements that
238       jump to C labels.
239     * C++0x raw strings are supported for C++ and for C with -std=gnu99.
240     * The deprecated attribute now takes an optional string argument, for
241       example, __attribute__((deprecated("text string"))), that will be
242       printed together with the deprecation warning.
243
244  C
245
246     * The -Wenum-compare option, which warns when comparing values of
247       different enum types, now works for C. It formerly only worked for
248       C++. This warning is enabled by -Wall. It may be avoided by using a
249       type cast.
250     * The -Wcast-qual option now warns about casts which are unsafe in
251       that they permit const-correctness to be violated without further
252       warnings. Specifically, it warns about cases where a qualifier is
253       added when all the lower types are not const. For example, it warns
254       about a cast from char ** to const char **.
255     * The -Wc++-compat option is significantly improved. It issues new
256       warnings for:
257          + Using C++ reserved operator names as identifiers.
258          + Conversions to enum types without explicit casts.
259          + Using va_arg with an enum type.
260          + Using different enum types in the two branches of ?:.
261          + Using ++ or -- on a variable of enum type.
262          + Using the same name as both a struct, union or enum tag and a
263            typedef, unless the typedef refers to the tagged type itself.
264          + Using a struct, union, or enum which is defined within another
265            struct or union.
266          + A struct field defined using a typedef if there is a field in
267            the struct, or an enclosing struct, whose name is the typedef
268            name.
269          + Duplicate definitions at file scope.
270          + Uninitialized const variables.
271          + A global variable with an anonymous struct, union, or enum
272            type.
273          + Using a string constant to initialize a char array whose size
274            is the length of the string.
275     * The new -Wjump-misses-init option warns about cases where a goto or
276       switch skips the initialization of a variable. This sort of branch
277       is an error in C++ but not in C. This warning is enabled by
278       -Wc++-compat.
279     * GCC now ensures that a C99-conforming <stdint.h> is present on most
280       targets, and uses information about the types in this header to
281       implement the Fortran bindings to those types. GCC does not ensure
282       the presence of such a header, and does not implement the Fortran
283       bindings, on the following targets: NetBSD, VxWorks, VMS,
284       SymbianOS, WinCE, LynxOS, Netware, QNX, Interix, TPF.
285     * GCC now implements C90- and C99-conforming rules for constant
286       expressions. This may cause warnings or errors for some code using
287       expressions that can be folded to a constant but are not constant
288       expressions as defined by ISO C.
289     * All known target-independent C90 and C90 Amendment 1 conformance
290       bugs, and all known target-independent C99 conformance bugs not
291       related to floating point or extended identifiers, have been fixed.
292     * The C decimal floating point support now includes support for the
293       FLOAT_CONST_DECIMAL64 pragma.
294     * The named address space feature from ISO/IEC TR 18037 is now
295       supported. This is currently only implemented for the SPU
296       processor.
297
298  C++
299
300     * Improved [12]experimental support for the upcoming C++0x ISO C++
301       standard, including support for raw strings, lambda expressions and
302       explicit type conversion operators.
303     * When printing the name of a class template specialization, G++ will
304       now omit any template arguments which come from default template
305       arguments. This behavior (and the pretty-printing of function
306       template specializations as template signature and arguments) can
307       be disabled with the -fno-pretty-templates option.
308     * Access control is now applied to typedef names used in a template,
309       which may cause G++ to reject some ill-formed code that was
310       accepted by earlier releases. The -fno-access-control option can be
311       used as a temporary workaround until the code is corrected.
312     * Compilation time for code that uses templates should now scale
313       linearly with the number of instantiations rather than
314       quadratically, as template instantiations are now looked up using
315       hash tables.
316     * Declarations of functions that look like builtin declarations of
317       library functions are only considered to be redeclarations if they
318       are declared with extern "C". This may cause problems with code
319       that omits extern "C" on hand-written declarations of C library
320       functions such as abort or memcpy. Such code is ill-formed, but was
321       accepted by earlier releases.
322     * Diagnostics that used to complain about passing non-POD types to
323       ... or jumping past the declaration of a non-POD variable now check
324       for triviality rather than PODness, as per C++0x.
325     * In C++0x mode local and anonymous classes are now allowed as
326       template arguments, and in declarations of variables and functions
327       with linkage, so long as any such declaration that is used is also
328       defined ([13]DR 757).
329     * Labels may now have attributes, as has been permitted for a while
330       in C. This is only permitted when the label definition and the
331       attribute specifier is followed by a semicolon--i.e., the label
332       applies to an empty statement. The only useful attribute for a
333       label is unused.
334     * G++ now implements [14]DR 176. Previously G++ did not support using
335       the injected-class-name of a template base class as a type name,
336       and lookup of the name found the declaration of the template in the
337       enclosing scope. Now lookup of the name finds the
338       injected-class-name, which can be used either as a type or as a
339       template, depending on whether or not the name is followed by a
340       template argument list. As a result of this change, some code that
341       was previously accepted may be ill-formed because
342         1. The injected-class-name is not accessible because it's from a
343            private base, or
344         2. The injected-class-name cannot be used as an argument for a
345            template template parameter.
346       In either of these cases, the code can be fixed by adding a
347       nested-name-specifier to explicitly name the template. The first
348       can be worked around with -fno-access-control; the second is only
349       rejected with -pedantic.
350     * A new standard mangling for SIMD vector types has been added, to
351       avoid name clashes on systems with vectors of varying length. By
352       default the compiler still uses the old mangling, but emits aliases
353       with the new mangling on targets that support strong aliases. Users
354       can switch over entirely to the new mangling with -fabi-version=4
355       or -fabi-version=0. -Wabi will now warn about code that uses the
356       old mangling.
357     * The command-line option -ftemplate-depth-N is now written as
358       -ftemplate-depth=N and the old form is deprecated.
359     * Conversions between NULL and non-pointer types are now warned by
360       default. The new option -Wno-conversion-null disables these
361       warnings. Previously these warnings were only available when using
362       -Wconversion explicitly.
363
364    Runtime Library (libstdc++)
365
366     * [15]Improved experimental support for the upcoming ISO C++
367       standard, C++0x, including:
368          + Support for <future>, <functional>, and <random>.
369          + Existing facilities now exploit explicit operators and the
370            newly implemented core C++0x features.
371     * An experimental [16]profile mode has been added. This is an
372       implementation of many C++ standard library constructs with an
373       additional analysis layer that gives performance improvement advice
374       based on recognition of suboptimal usage patterns. For example,
375#include <vector>
376int main()
377{
378  std::vector<int> v;
379  for (int k = 0; k < 1024; ++k)
380    v.insert(v.begin(), k);
381}
382
383       When instrumented via the profile mode, can return suggestions
384       about the initial size and choice of the container used as follows:
385vector-to-list: improvement = 5: call stack = 0x804842c ...
386    : advice = change std::vector to std::list
387vector-size: improvement = 3: call stack = 0x804842c ...
388    : advice = change initial container size from 0 to 1024
389
390       These constructs can be substituted for the normal libstdc++
391       constructs on a piecemeal basis, or all existing components can be
392       transformed via the -D_GLIBCXX_PROFILE macro.
393     * [17]Support for decimal floating-point arithmetic (aka ISO C++ TR
394       24733) has been added. This support is in header file
395       <decimal/decimal>, uses namespace std::decimal, and includes
396       classes decimal32, decimal64, and decimal128.
397     * Sources have been audited for application of function attributes
398       nothrow, const, pure, and noreturn.
399     * Python pretty-printers have been added for many standard library
400       components that simplify the internal representation and present a
401       more intuitive view of components when used with
402       appropriately-advanced versions of GDB. For more information,
403       please consult the more [18]detailed description.
404     * The default behavior for comparing typeinfo names has changed, so
405       in <typeinfo>, __GXX_MERGED_TYPEINFO_NAMES now defaults to zero.
406     * The new -static-libstdc++ option directs g++ to link the C++
407       library statically, even if the default would normally be to link
408       it dynamically.
409
410  Fortran
411
412     * The COMMON default padding has been changed - instead of adding the
413       padding before a variable it is now added afterwards, which
414       increases the compatibility with other vendors and helps to obtain
415       the correct output in some cases. Cf. also the -falign-commons
416       option ([19]added in 4.4).
417     * The -finit-real= option now also supports the value snan for
418       signalling not-a-number; to be effective, one additionally needs to
419       enable trapping (e.g. via -ffpe-trap=). Note: Compile-time
420       optimizations can turn a signalling NaN into a quiet one.
421     * The new option -fcheck= has been added with the options bounds,
422       array-temps, do, pointer, and recursive. The bounds and array-temps
423       options are equivalent to -fbounds-check and
424       -fcheck-array-temporaries. The do option checks for invalid
425       modification of loop iteration variables, and the recursive option
426       tests for recursive calls to subroutines/functions which are not
427       marked as recursive. With pointer pointer association checks in
428       calls are performed; however, neither undefined pointers nor
429       pointers in expressions are handled. Using -fcheck=all enables all
430       these run-time checks.
431     * The run-time checking -fcheck=bounds now warns about invalid string
432       lengths of character dummy arguments. Additionally, more
433       compile-time checks have been added.
434     * The new option [20]-fno-protect-parens has been added; if set, the
435       compiler may reorder REAL and COMPLEX expressions without regard to
436       parentheses.
437     * GNU Fortran no longer links against libgfortranbegin. As before,
438       MAIN__ (assembler symbol name) is the actual Fortran main program,
439       which is invoked by the main function. However, main is now
440       generated and put in the same object file as MAIN__. For the time
441       being, libgfortranbegin still exists for backward compatibility.
442       For details see the new [21]Mixed-Language Programming chapter in
443       the manual.
444     * The I/O library was restructured for performance and cleaner code.
445     * Array assignments and WHERE are now run in parallel when OpenMP's
446       WORKSHARE is used.
447     * The experimental option -fwhole-file was added. The option allows
448       whole-file checking of procedure arguments and allows for better
449       optimizations. It can also be used with -fwhole-program, which is
450       now also supported in gfortran.
451     * More Fortran 2003 and Fortran 2008 mathematical functions can now
452       be used as initialization expressions.
453     * Some extended attributes such as STDCALL are now supported via the
454       [22]GCC$ compiler directive.
455     * For Fortran 77 compatibility: If -fno-sign-zero is used, the SIGN
456       intrinsic behaves now as if zero were always positive.
457     * For legacy compatibiliy: On Cygwin and MinGW, the special files
458       CONOUT$ and CONIN$ (and CONERR$ which maps to CONOUT$) are now
459       supported.
460     * Fortran 2003 support has been extended:
461          + Procedure-pointer function results and procedure-pointer
462            components (including PASS),
463          + allocatable scalars (experimental),
464          + DEFERRED type-bound procedures,
465          + the ERRMSG= argument of the ALLOCATE and DEALLOCATE statements
466            have been implemented.
467          + The ALLOCATE statement supports type-specs and the SOURCE=
468            argument.
469          + OPERATOR(*) and ASSIGNMENT(=) are now allowed as GENERIC
470            type-bound procedure (i.e. as type-bound operators).
471          + Rounding (ROUND=, RZ, ...) for output is now supported.
472          + The INT_FAST{8,16,32,64,128}_T kind type parameters of the
473            intrinsic module ISO_C_BINDING are now supported, except for
474            the targets listed above as ones where GCC does not have
475            <stdint.h> type information.
476          + Extensible derived types with type-bound procedure or
477            procedure pointer with PASS attribute now have to use CLASS in
478            line with the Fortran 2003 standard; the workaround to use
479            TYPE is no longer supported.
480          + [23]Experimental, incomplete support for polymorphism,
481            including CLASS, SELECT TYPE and dynamic dispatch of
482            type-bound procedure calls. Some features do not work yet such
483            as unlimited polymorphism (CLASS(*)).
484     * Fortran 2008 support has been extended:
485          + The OPEN statement now supports the NEWUNIT= option, which
486            returns a unique file unit, thus preventing inadvertent use of
487            the same unit in different parts of the program.
488          + Support for unlimited format items has been added.
489          + The INT{8,16,32} and REAL{32,64,128} kind type parameters of
490            the intrinsic module ISO_FORTRAN_ENV are now supported.
491          + Using complex arguments with TAN, SINH, COSH, TANH, ASIN,
492            ACOS, and ATAN is now possible; the functions ASINH, ACOSH,
493            and ATANH have been added (for real and complex arguments) and
494            ATAN(Y,X) is now an alias for ATAN2(Y,X).
495          + The BLOCK construct has been implemented.
496
497New Targets and Target Specific Improvements
498
499  AIX
500
501     * Full cross-toolchain support now available with GNU Binutils
502
503  ARM
504
505     * GCC now supports the Cortex-M0 and Cortex-A5 processors.
506     * GCC now supports the ARM v7E-M architecture.
507     * GCC now supports VFPv4-based FPUs and FPUs with
508       single-precision-only VFP.
509     * GCC has many improvements to optimization for other ARM processors,
510       including scheduling support for the integer pipeline on Cortex-A9.
511     * GCC now supports the IEEE 754-2008 half-precision floating-point
512       type, and a variant ARM-specific half-precision type. This type is
513       specified using __fp16, with the layout determined by
514       -mfp16-format. With appropriate -mfpu options, the Cortex-A9 and
515       VFPv4 half-precision instructions will be used.
516     * GCC now supports the variant of AAPCS that uses VFP registers for
517       parameter passing and return values.
518
519  AVR
520
521     * The -mno-tablejump option has been removed because it has the same
522       effect as the -fno-jump-tables option.
523     * Added support for these new AVR devices:
524          + ATmega8U2
525          + ATmega16U2
526          + ATmega32U2
527
528  IA-32/x86-64
529
530     * GCC now will set the default for -march= based on the configure
531       target.
532     * GCC now supports handling floating-point excess precision arising
533       from use of the x87 floating-point unit in a way that conforms to
534       ISO C99. This is enabled with -fexcess-precision=standard and with
535       standards conformance options such as -std=c99, and may be disabled
536       using -fexcess-precision=fast.
537     * Support for the Intel Atom processor is now available through the
538       -march=atom and -mtune=atom options.
539     * A new -mcrc32 option is now available to enable crc32 intrinsics.
540     * A new -mmovbe option is now available to enable GCC to use the
541       movbe instruction to implement __builtin_bswap32 and
542       __builtin_bswap64.
543     * SSE math now can be enabled by default at configure time with the
544       new --with-fpmath=sse option.
545     * There is a new intrinsic header file, <x86intrin.h>. It should be
546       included before using any IA-32/x86-64 intrinsics.
547     * Support for the XOP, FMA4, and LWP instruction sets for the AMD
548       Orochi processors are now available with the -mxop, -mfma4, and
549       -mlwp options.
550     * The -mabm option enables GCC to use the popcnt and lzcnt
551       instructions on AMD processors.
552     * The -mpopcnt option enables GCC to use the popcnt instructions on
553       both AMD and Intel processors.
554
555  M68K/ColdFire
556
557     * GCC now supports ColdFire 51xx, 5221x, 5225x, 52274, 52277, 5301x
558       and 5441x devices.
559     * GCC now supports thread-local storage (TLS) on M68K and ColdFire
560       processors.
561
562  MeP
563
564   Support has been added for the Toshiba Media embedded Processor (MeP,
565   or mep-elf) embedded target.
566
567  MIPS
568
569     * GCC now supports MIPS 1004K processors.
570     * GCC can now be configured with options --with-arch-32,
571       --with-arch-64, --with-tune-32 and --with-tune-64 to control the
572       default optimization separately for 32-bit and 64-bit modes.
573     * MIPS targets now support an alternative _mcount interface, in which
574       register $12 points to the function's save slot for register $31.
575       This interface is selected by the -mcount-ra-address option; see
576       the documentation for more details.
577     * GNU/Linux targets can now generate read-only .eh_frame sections.
578       This optimization requires GNU binutils 2.20 or above, and is only
579       available if GCC is configured with a suitable version of binutils.
580     * GNU/Linux targets can now attach special relocations to indirect
581       calls, so that the linker can turn them into direct jumps or
582       branches. This optimization requires GNU binutils 2.20 or later,
583       and is automatically selected if GCC is configured with an
584       appropriate version of binutils. It can be explicitly enabled or
585       disabled using the -mrelax-pic-calls command-line option.
586     * GCC now generates more heavily-optimized atomic operations on
587       Octeon processors.
588     * MIPS targets now support the -fstack-protector option.
589     * GCC now supports an -msynci option, which specifies that synci is
590       enough to flush the instruction cache, without help from the
591       operating system. GCC uses this information to optimize
592       automatically-generated cache flush operations, such as those used
593       for nested functions in C. There is also a --with-synci
594       configure-time option, which makes -msynci the default.
595     * GCC supports four new function attributes for interrupt handlers:
596       interrupt, use_shadow_register_set, keep_interrupts_masked and
597       use_debug_exception_return. See the documentation for more details
598       about these attributes.
599
600  RS/6000 (POWER/PowerPC)
601
602     * GCC now supports the Power ISA 2.06, which includes the VSX
603       instructions that add vector 64-bit floating point support, new
604       population count instructions, and conversions between floating
605       point and unsigned types.
606     * Support for the power7 processor is now available through the
607       -mcpu=power7 and -mtune=power7.
608     * GCC will now vectorize loops that contain simple math functions
609       like copysign when generating code for altivec or VSX targets.
610     * Support for the A2 processor is now available through the -mcpu=a2
611       and -mtune=a2 options.
612     * Support for the 476 processor is now available through the
613       -mcpu={476,476fp} and -mtune={476,476fp} options.
614     * Support for the e500mc64 processor is now available through the
615       -mcpu=e500mc64 and -mtune=e500mc64 options.
616     * GCC can now be configured with options --with-cpu-32,
617       --with-cpu-64, --with-tune-32 and --with-tune-64 to control the
618       default optimization separately for 32-bit and 64-bit modes.
619     * Starting with GCC 4.5.4, vectors of type vector long long or vector
620       long are passed and returned in the same method as other vectors
621       with the VSX instruction set. Previously the GCC compiler did not
622       adhere to the ABI for 128-bit vectors with 64-bit integer base
623       types (PR 48857). This is also fixed in the GCC 4.6.1 release.
624
625  RX
626
627   Support has been added for the Renesas RX Processor (rx-elf) target.
628
629Operating Systems
630
631  Windows (Cygwin and MinGW)
632
633     * GCC now installs all the major language runtime libraries as DLLs
634       when configured with the --enable-shared option.
635     * GCC now makes use of the new support for aligned common variables
636       in versions of binutils >= 2.20 to fix bugs in the support for SSE
637       data types.
638     * Improvements to the libffi support library increase the reliability
639       of code generated by GCJ on all Windows platforms. Libgcj is
640       enabled by default for the first time.
641     * Libtool improvements simplify installation by placing the generated
642       DLLs in the correct binaries directory.
643     * Numerous other minor bugfixes and improvements, and substantial
644       enhancements to the Fortran language support library.
645
646   >
647
648Other significant improvements
649
650  Plugins
651
652     * It is now possible to extend the compiler without having to modify
653       its source code. A new option -fplugin=file.so tells GCC to load
654       the shared object file.so and execute it as part of the compiler.
655       The internal documentation describes the details on how plugins can
656       interact with the compiler.
657
658  Installation changes
659
660     * The move to newer autotools changed default installation
661       directories and switches to control them: The --with-datarootdir,
662       --with-docdir, --with-pdfdir, and --with-htmldir switches are not
663       used any more. Instead, you can now use --datarootdir, --docdir,
664       --htmldir, and --pdfdir. The default installation directories have
665       changed as follows according to the GNU Coding Standards:
666
667       datarootdir read-only architecture-independent data root [PREFIX/share]
668       localedir   locale-specific message catalogs [DATAROOTDIR/locale]
669       docdir      documentation root [DATAROOTDIR/doc/PACKAGE]
670       htmldir     html documentation [DOCDIR]
671       dvidir      dvi documentation [DOCDIR]
672       pdfdir      pdf documentation [DOCDIR]
673       psdir       ps documentation [DOCDIR]
674       The following variables have new default values:
675
676       datadir read-only architecture-independent data [DATAROOTDIR]
677       infodir info documentation [DATAROOTDIR/info]
678       mandir  man documentation [DATAROOTDIR/man]
679
680GCC 4.5.1
681
682   This is the [24]list of problem reports (PRs) from GCC's bug tracking
683   system that are known to be fixed in the 4.5.1 release. This list might
684   not be complete (that is, it is possible that some PRs that have been
685   fixed are not listed here).
686
687  All languages
688
689     * GCC's new link-time optimizer ([25]-flto) now also works on a few
690       non-ELF targets:
691          + Cygwin (*-cygwin*)
692          + MinGW (*-mingw*)
693          + Darwin on x86-64 (x86_64-apple-darwin*)
694       LTO is not enabled by default for these targets. To enable LTO, you
695       should configure with the --enable-lto option.
696
697GCC 4.5.2
698
699   This is the [26]list of problem reports (PRs) from GCC's bug tracking
700   system that are known to be fixed in the 4.5.2 release. This list might
701   not be complete (that is, it is possible that some PRs that have been
702   fixed are not listed here).
703
704GCC 4.5.3
705
706   This is the [27]list of problem reports (PRs) from GCC's bug tracking
707   system that are known to be fixed in the 4.5.3 release. This list might
708   not be complete (that is, it is possible that some PRs that have been
709   fixed are not listed here).
710
711   On the PowerPC compiler, the Altivec builtin functions vec_ld and
712   vec_st have been modified to generate the Altivec memory instructions
713   LVX and STVX, even if the -mvsx option is used. In the initial GCC 4.5
714   release, these builtin functions were changed to generate VSX memory
715   reference instructions instead of Altivec memory instructions, but
716   there are differences between the two instructions. If the VSX
717   instruction set is available, you can now use the new builtin functions
718   vec_vsx_ld and vec_vsx_st which always generates the VSX memory
719   instructions.
720
721GCC 4.5.4
722
723   This is the [28]list of problem reports (PRs) from GCC's bug tracking
724   system that are known to be fixed in the 4.5.4 release. This list might
725   not be complete (that is, it is possible that some PRs that have been
726   fixed are not listed here).
727
728
729    For questions related to the use of GCC, please consult these web
730    pages and the [29]GCC manuals. If that fails, the
731    [30]gcc-help@gcc.gnu.org mailing list might help. Comments on these
732    web pages and the development of GCC are welcome on our developer
733    list at [31]gcc@gcc.gnu.org. All of [32]our lists have public
734    archives.
735
736   Copyright (C) [33]Free Software Foundation, Inc. Verbatim copying and
737   distribution of this entire article is permitted in any medium,
738   provided this notice is preserved.
739
740   These pages are [34]maintained by the GCC team. Last modified
741   2012-07-02[35].
742
743References
744
745   1. http://www.multiprecision.org/
746   2. http://gcc.gnu.org/install/prerequisites.html
747   3. http://gcc.gnu.org/ml/gcc/2010-01/msg00510.html
748   4. http://gcc.gnu.org/gcc-4.4/changes.html#obsoleted
749   5. http://gcc.gnu.org/gcc-4.5/changes.html#x86
750   6. http://www.multiprecision.org/
751   7. http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30789
752   8. http://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html#index-flto-801
753   9. http://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html#index-fwhopr-802
754  10. http://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html#index-fwhole-program-800
755  11. http://gcc.gnu.org/onlinedocs/gcc/Restricted-Pointers.html
756  12. http://gcc.gnu.org/gcc-4.5/cxx0x_status.html
757  13. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#757
758  14. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#176
759  15. http://gcc.gnu.org/onlinedocs/libstdc++/manual/status.html#status.iso.200x
760  16. http://gcc.gnu.org/onlinedocs/libstdc++/manual/profile_mode.html
761  17. http://gcc.gnu.org/onlinedocs/libstdc++/manual/status.html#status.iso.tr24733
762  18. http://sourceware.org/gdb/wiki/STLSupport
763  19. http://gcc.gnu.org/gcc-4.4/changes.html
764  20. http://gcc.gnu.org/onlinedocs/gfortran/Code-Gen-Options.html
765  21. http://gcc.gnu.org/onlinedocs/gfortran/Mixed-Language-Programming.html
766  22. http://gcc.gnu.org/onlinedocs/gfortran/GNU-Fortran-Compiler-Directives.html
767  23. http://gcc.gnu.org/wiki/OOP
768  24. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.5.1
769  25. http://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html#index-flto-801
770  26. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.5.2
771  27. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.5.3
772  28. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.5.4
773  29. http://gcc.gnu.org/onlinedocs/
774  30. mailto:gcc-help@gcc.gnu.org
775  31. mailto:gcc@gcc.gnu.org
776  32. http://gcc.gnu.org/lists.html
777  33. http://www.fsf.org/
778  34. http://gcc.gnu.org/about.html
779  35. http://validator.w3.org/check/referer
780======================================================================
781http://gcc.gnu.org/gcc-4.4/index.html
782
783                             GCC 4.4 Release Series
784
785   March 13, 2012
786
787   The [1]GNU project and the GCC developers are pleased to announce the
788   release of GCC 4.4.7.
789
790   This release is a bug-fix release, containing fixes for regressions in
791   GCC 4.4.6 relative to previous releases of GCC.
792
793Release History
794
795   GCC 4.4.7
796          March 13, 2012 ([2]changes)
797
798   GCC 4.4.6
799          April 16, 2011 ([3]changes)
800
801   GCC 4.4.5
802          October 1, 2010 ([4]changes)
803
804   GCC 4.4.4
805          April 29, 2010 ([5]changes)
806
807   GCC 4.4.3
808          January 21, 2010 ([6]changes)
809
810   GCC 4.4.2
811          October 15, 2009 ([7]changes)
812
813   GCC 4.4.1
814          July 22, 2009 ([8]changes)
815
816   GCC 4.4.0
817          April 21, 2009 ([9]changes)
818
819References and Acknowledgements
820
821   GCC used to stand for the GNU C Compiler, but since the compiler
822   supports several other languages aside from C, it now stands for the
823   GNU Compiler Collection.
824
825   A list of [10]successful builds is updated as new information becomes
826   available.
827
828   The GCC developers would like to thank the numerous people that have
829   contributed new features, improvements, bug fixes, and other changes as
830   well as test results to GCC. This [11]amazing group of volunteers is
831   what makes GCC successful.
832
833   For additional information about GCC please refer to the [12]GCC
834   project web site or contact the [13]GCC development mailing list.
835
836   To obtain GCC please use [14]our mirror sites or [15]our SVN server.
837
838
839    For questions related to the use of GCC, please consult these web
840    pages and the [16]GCC manuals. If that fails, the
841    [17]gcc-help@gcc.gnu.org mailing list might help. Comments on these
842    web pages and the development of GCC are welcome on our developer
843    list at [18]gcc@gcc.gnu.org. All of [19]our lists have public
844    archives.
845
846   Copyright (C) [20]Free Software Foundation, Inc. Verbatim copying and
847   distribution of this entire article is permitted in any medium,
848   provided this notice is preserved.
849
850   These pages are [21]maintained by the GCC team. Last modified
851   2012-03-13[22].
852
853References
854
855   1. http://www.gnu.org/
856   2. http://gcc.gnu.org/gcc-4.4/changes.html
857   3. http://gcc.gnu.org/gcc-4.4/changes.html
858   4. http://gcc.gnu.org/gcc-4.4/changes.html
859   5. http://gcc.gnu.org/gcc-4.4/changes.html
860   6. http://gcc.gnu.org/gcc-4.4/changes.html
861   7. http://gcc.gnu.org/gcc-4.4/changes.html
862   8. http://gcc.gnu.org/gcc-4.4/changes.html
863   9. http://gcc.gnu.org/gcc-4.4/changes.html
864  10. http://gcc.gnu.org/gcc-4.4/buildstat.html
865  11. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
866  12. http://gcc.gnu.org/index.html
867  13. mailto:gcc@gcc.gnu.org
868  14. http://gcc.gnu.org/mirrors.html
869  15. http://gcc.gnu.org/svn.html
870  16. http://gcc.gnu.org/onlinedocs/
871  17. mailto:gcc-help@gcc.gnu.org
872  18. mailto:gcc@gcc.gnu.org
873  19. http://gcc.gnu.org/lists.html
874  20. http://www.fsf.org/
875  21. http://gcc.gnu.org/about.html
876  22. http://validator.w3.org/check/referer
877======================================================================
878http://gcc.gnu.org/gcc-4.4/changes.html
879
880                             GCC 4.4 Release Series
881                        Changes, New Features, and Fixes
882
883   The latest release in the 4.4 release series is [1]GCC 4.4.7.
884
885Caveats
886
887     * __builtin_stdarg_start has been completely removed from GCC.
888       Support for <varargs.h> had been deprecated since GCC 4.0. Use
889       __builtin_va_start as a replacement.
890     * Some of the errors issued by the C++ front end that could be
891       downgraded to warnings in previous releases by using -fpermissive
892       are now warnings by default. They can be converted into errors by
893       using -pedantic-errors.
894     * Use of the cpp assertion extension will now emit a warning when
895       -Wdeprecated or -pedantic is used. This extension has been
896       deprecated for many years, but never warned about.
897     * Packed bit-fields of type char were not properly bit-packed on many
898       targets prior to GCC 4.4. On these targets, the fix in GCC 4.4
899       causes an ABI change. For example there is no longer a 4-bit
900       padding between field a and b in this structure:
901    struct foo
902    {
903      char a:4;
904      char b:8;
905    } __attribute__ ((packed));
906       There is a new warning to help identify fields that are affected:
907    foo.c:5: note: Offset of packed bit-field 'b' has changed in GCC 4.4
908       The warning can be disabled with -Wno-packed-bitfield-compat.
909     * On ARM EABI targets, the C++ mangling of the va_list type has been
910       changed to conform to the current revision of the EABI. This does
911       not affect the libstdc++ library included with GCC.
912     * The SCOUNT and POS bits of the MIPS DSP control register are now
913       treated as global. Previous versions of GCC treated these fields as
914       call-clobbered instead.
915     * The MIPS port no longer recognizes the h asm constraint. It was
916       necessary to remove this constraint in order to avoid generating
917       unpredictable code sequences.
918       One of the main uses of the h constraint was to extract the high
919       part of a multiplication on 64-bit targets. For example:
920    asm ("dmultu\t%1,%2" : "=h" (result) : "r" (x), "r" (y));
921       You can now achieve the same effect using 128-bit types:
922    typedef unsigned int uint128_t __attribute__((mode(TI)));
923    result = ((uint128_t) x * y) >> 64;
924       The second sequence is better in many ways. For example, if x and y
925       are constants, the compiler can perform the multiplication at
926       compile time. If x and y are not constants, the compiler can
927       schedule the runtime multiplication better than it can schedule an
928       asm statement.
929     * Support for a number of older systems and recently unmaintained or
930       untested target ports of GCC has been declared obsolete in GCC 4.4.
931       Unless there is activity to revive them, the next release of GCC
932       will have their sources permanently removed.
933       The following ports for individual systems on particular
934       architectures have been obsoleted:
935          + Generic a.out on IA32 and m68k (i[34567]86-*-aout*,
936            m68k-*-aout*)
937          + Generic COFF on ARM, H8300, IA32, m68k and SH (arm-*-coff*,
938            armel-*-coff*, h8300-*-*, i[34567]86-*-coff*, m68k-*-coff*,
939            sh-*-*). This does not affect other more specific targets
940            using the COFF object format on those architectures, or the
941            more specific H8300 and SH targets (h8300-*-rtems*,
942            h8300-*-elf*, sh-*-elf*, sh-*-symbianelf*, sh-*-linux*,
943            sh-*-netbsdelf*, sh-*-rtems*, sh-wrs-vxworks).
944          + 2BSD on PDP-11 (pdp11-*-bsd)
945          + AIX 4.1 and 4.2 on PowerPC (rs6000-ibm-aix4.[12]*,
946            powerpc-ibm-aix4.[12]*)
947          + Tuning support for Itanium1 (Merced) variants. Note that code
948            tuned for Itanium2 should also run correctly on Itanium1.
949     * The protoize and unprotoize utilities have been obsoleted and will
950       be removed in GCC 4.5. These utilities have not been installed by
951       default since GCC 3.0.
952     * Support has been removed for all the [2]configurations obsoleted in
953       GCC 4.3.
954     * Unknown -Wno-* options are now silently ignored by GCC if no other
955       diagnostics are issued. If other diagnostics are issued, then GCC
956       warns about the unknown options.
957     * More information on porting to GCC 4.4 from previous versions of
958       GCC can be found in the [3]porting guide for this release.
959
960General Optimizer Improvements
961
962     * A new command-line switch -findirect-inlining has been added. When
963       turned on it allows the inliner to also inline indirect calls that
964       are discovered to have known targets at compile time thanks to
965       previous inlining.
966     * A new command-line switch -ftree-switch-conversion has been added.
967       This new pass turns simple initializations of scalar variables in
968       switch statements into initializations from a static array, given
969       that all the values are known at compile time and the ratio between
970       the new array size and the original switch branches does not exceed
971       the parameter --param switch-conversion-max-branch-ratio (default
972       is eight).
973     * A new command-line switch -ftree-builtin-call-dce has been added.
974       This optimization eliminates unnecessary calls to certain builtin
975       functions when the return value is not used, in cases where the
976       calls can not be eliminated entirely because the function may set
977       errno. This optimization is on by default at -O2 and above.
978     * A new command-line switch -fconserve-stack directs the compiler to
979       minimize stack usage even if it makes the generated code slower.
980       This affects inlining decisions.
981     * When the assembler supports it, the compiler will now emit unwind
982       information using assembler .cfi directives. This makes it possible
983       to use such directives in inline assembler code. The new option
984       -fno-dwarf2-cfi-asm directs the compiler to not use .cfi
985       directives.
986     * The [4]Graphite branch has been merged. This merge has brought in a
987       new framework for loop optimizations based on a polyhedral
988       intermediate representation. These optimizations apply to all the
989       languages supported by GCC. The following new code transformations
990       are available in GCC 4.4:
991          + -floop-interchange performs loop interchange transformations
992            on loops. Interchanging two nested loops switches the inner
993            and outer loops. For example, given a loop like:
994          DO J = 1, M
995            DO I = 1, N
996              A(J, I) = A(J, I) * C
997            ENDDO
998          ENDDO
999
1000            loop interchange will transform the loop as if the user had
1001            written:
1002          DO I = 1, N
1003            DO J = 1, M
1004              A(J, I) = A(J, I) * C
1005            ENDDO
1006          ENDDO
1007
1008            which can be beneficial when N is larger than the caches,
1009            because in Fortran, the elements of an array are stored in
1010            memory contiguously by column, and the original loop iterates
1011            over rows, potentially creating at each access a cache miss.
1012          + -floop-strip-mine performs loop strip mining transformations
1013            on loops. Strip mining splits a loop into two nested loops.
1014            The outer loop has strides equal to the strip size and the
1015            inner loop has strides of the original loop within a strip.
1016            For example, given a loop like:
1017          DO I = 1, N
1018            A(I) = A(I) + C
1019          ENDDO
1020
1021            loop strip mining will transform the loop as if the user had
1022            written:
1023          DO II = 1, N, 4
1024            DO I = II, min (II + 3, N)
1025              A(I) = A(I) + C
1026            ENDDO
1027          ENDDO
1028
1029          + -floop-block performs loop blocking transformations on loops.
1030            Blocking strip mines each loop in the loop nest such that the
1031            memory accesses of the element loops fit inside caches. For
1032            example, given a loop like:
1033          DO I = 1, N
1034            DO J = 1, M
1035              A(J, I) = B(I) + C(J)
1036            ENDDO
1037          ENDDO
1038
1039            loop blocking will transform the loop as if the user had
1040            written:
1041          DO II = 1, N, 64
1042            DO JJ = 1, M, 64
1043              DO I = II, min (II + 63, N)
1044                DO J = JJ, min (JJ + 63, M)
1045                  A(J, I) = B(I) + C(J)
1046                ENDDO
1047              ENDDO
1048            ENDDO
1049          ENDDO
1050
1051            which can be beneficial when M is larger than the caches,
1052            because the innermost loop will iterate over a smaller amount
1053            of data that can be kept in the caches.
1054     * A new register allocator has replaced the old one. It is called
1055       integrated register allocator (IRA) because coalescing, register
1056       live range splitting, and hard register preferencing are done
1057       on-the-fly during coloring. It also has better integration with the
1058       reload pass. IRA is a regional register allocator which uses modern
1059       Chaitin-Briggs coloring instead of Chow's priority coloring used in
1060       the old register allocator. More info about IRA internals and
1061       options can be found in the GCC manuals.
1062     * A new instruction scheduler and software pipeliner, based on the
1063       selective scheduling approach, has been added. The new pass
1064       performs instruction unification, register renaming, substitution
1065       through register copies, and speculation during scheduling. The
1066       software pipeliner is able to pipeline non-countable loops. The new
1067       pass is targeted at scheduling-eager in-order platforms. In GCC 4.4
1068       it is available for the Intel Itanium platform working by default
1069       as the second scheduling pass (after register allocation) at the
1070       -O3 optimization level.
1071     * When using -fprofile-generate with a multi-threaded program, the
1072       profile counts may be slightly wrong due to race conditions. The
1073       new -fprofile-correction option directs the compiler to apply
1074       heuristics to smooth out the inconsistencies. By default the
1075       compiler will give an error message when it finds an inconsistent
1076       profile.
1077     * The new -fprofile-dir=PATH option permits setting the directory
1078       where profile data files are stored when using -fprofile-generate
1079       and friends, and the directory used when reading profile data files
1080       using -fprofile-use and friends.
1081
1082New warning options
1083
1084     * The new -Wframe-larger-than=NUMBER option directs GCC to emit a
1085       warning if any stack frame is larger than NUMBER bytes. This may be
1086       used to help ensure that code fits within a limited amount of stack
1087       space.
1088     * The command-line option -Wlarger-than-N is now written as
1089       -Wlarger-than=N and the old form is deprecated.
1090     * The new -Wno-mudflap option disables warnings about constructs
1091       which can not be instrumented when using -fmudflap.
1092
1093New Languages and Language specific improvements
1094
1095     * Version 3.0 of the [5]OpenMP specification is now supported for the
1096       C, C++, and Fortran compilers.
1097     * New character data types, per [6]TR 19769: New character types in
1098       C, are now supported for the C compiler in -std=gnu99 mode, as
1099       __CHAR16_TYPE__ and __CHAR32_TYPE__, and for the C++ compiler in
1100       -std=c++0x and -std=gnu++0x modes, as char16_t and char32_t too.
1101
1102  C family
1103
1104     * A new optimize attribute was added to allow programmers to change
1105       the optimization level and particular optimization options for an
1106       individual function. You can also change the optimization options
1107       via the GCC optimize pragma for functions defined after the pragma.
1108       The GCC push_options pragma and the GCC pop_options pragma allow
1109       you temporarily save and restore the options used. The GCC
1110       reset_options pragma restores the options to what was specified on
1111       the command line.
1112     * Uninitialized warnings do not require enabling optimization
1113       anymore, that is, -Wuninitialized can be used together with -O0.
1114       Nonetheless, the warnings given by -Wuninitialized will probably be
1115       more accurate if optimization is enabled.
1116     * -Wparentheses now warns about expressions such as (!x | y) and (!x
1117       & y). Using explicit parentheses, such as in ((!x) | y), silences
1118       this warning.
1119     * -Wsequence-point now warns within if, while,do while and for
1120       conditions, and within for begin/end expressions.
1121     * A new option -dU is available to dump definitions of preprocessor
1122       macros that are tested or expanded.
1123
1124  C++
1125
1126     * [7]Improved experimental support for the upcoming ISO C++ standard,
1127       C++0x. Including support for auto, inline namespaces, generalized
1128       initializer lists, defaulted and deleted functions, new character
1129       types, and scoped enums.
1130     * Those errors that may be downgraded to warnings to build legacy
1131       code now mention -fpermissive when -fdiagnostics-show-option is
1132       enabled.
1133     * -Wconversion now warns if the result of a static_cast to enumeral
1134       type is unspecified because the value is outside the range of the
1135       enumeral type.
1136     * -Wuninitialized now warns if a non-static reference or non-static
1137       const member appears in a class without constructors.
1138     * G++ now properly implements value-initialization, so objects with
1139       an initializer of () and an implicitly defined default constructor
1140       will be zero-initialized before the default constructor is called.
1141
1142    Runtime Library (libstdc++)
1143
1144     * [8]Improved experimental support for the upcoming ISO C++ standard,
1145       C++0x, including:
1146          + Support for <chrono>, <condition_variable>, <cstdatomic>,
1147            <forward_list>, <initializer_list>, <mutex>, <ratio>,
1148            <system_error>, and <thread>.
1149          + unique_ptr, <algorithm> additions, exception propagation, and
1150            support for the new character types in <string> and <limits>.
1151          + Existing facilities now exploit initializer lists, defaulted
1152            and deleted functions, and the newly implemented core C++0x
1153            features.
1154          + Some standard containers are more efficient together with
1155            stateful allocators, i.e., no allocator is constructed on the
1156            fly at element construction time.
1157     * Experimental support for non-standard pointer types in containers.
1158     * The long standing libstdc++/30928 has been fixed for targets
1159       running glibc 2.10 or later.
1160     * As usual, many small and larger bug fixes, in particular quite a
1161       few corner cases in <locale>.
1162
1163  Fortran
1164
1165     * GNU Fortran now employs libcpp directly instead of using cc1 as an
1166       external preprocessor. The [9]-cpp option was added to allow manual
1167       invocation of the preprocessor without relying on filename
1168       extensions.
1169     * The [10]-Warray-temporaries option warns about array temporaries
1170       generated by the compiler, as an aid to optimization.
1171     * The [11]-fcheck-array-temporaries option has been added, printing a
1172       notification at run time, when an array temporary had to be created
1173       for an function argument. Contrary to -Warray-temporaries the
1174       warning is only printed if the array is noncontiguous.
1175     * Improved generation of DWARF debugging symbols
1176     * If using an intrinsic not part of the selected standard (via -std=
1177       and -fall-intrinsics) gfortran will now treat it as if this
1178       procedure were declared EXTERNAL and try to link to a user-supplied
1179       procedure. -Wintrinsics-std will warn whenever this happens. The
1180       now-useless option -Wnonstd-intrinsic was removed.
1181     * The flag -falign-commons has been added to control the alignment of
1182       variables in COMMON blocks, which is enabled by default in line
1183       with previous GCC version. Using -fno-align-commons one can force
1184       commons to be contiguous in memory as required by the Fortran
1185       standard, however, this slows down the memory access. The option
1186       -Walign-commons, which is enabled by default, warns when padding
1187       bytes were added for alignment. The proper solution is to sort the
1188       common objects by decreasing storage size, which avoids the
1189       alignment problems.
1190     * Fortran 2003 support has been extended:
1191          + Wide characters (ISO 10646, UCS-4, kind=4) and UTF-8 I/O is
1192            now supported (except internal reads from/writes to wide
1193            strings). [12]-fbackslash now supports also \unnnn and
1194            \Unnnnnnnn to enter Unicode characters.
1195          + Asynchronous I/O (implemented as synchronous I/O) and the
1196            decimal=, size=, sign=, pad=, blank=, and delim= specifiers
1197            are now supported in I/O statements.
1198          + Support for Fortran 2003 structure constructors and for array
1199            constructor with typespec has been added.
1200          + Procedure Pointers (but not yet as component in derived types
1201            and as function results) are now supported.
1202          + Abstract types, type extension, and type-bound procedures
1203            (both PROCEDURE and GENERIC but not as operators). Note: As
1204            CLASS/polymorphyic types are not implemented, type-bound
1205            procedures with PASS accept as non-standard extension TYPE
1206            arguments.
1207     * Fortran 2008 support has been added:
1208          + The -std=f2008 option and support for the file extensions
1209            .f2008 and .F2008 has been added.
1210          + The g0 format descriptor is now supported.
1211          + The Fortran 2008 mathematical intrinsics ASINH, ACOSH, ATANH,
1212            ERF, ERFC, GAMMA, LOG_GAMMA, BESSEL_*, HYPOT, and ERFC_SCALED
1213            are now available (some of them existed as GNU extension
1214            before). Note: The hyperbolic functions are not yet supporting
1215            complex arguments and the three- argument version of BESSEL_*N
1216            is not available.
1217          + The bit intrinsics LEADZ and TRAILZ have been added.
1218
1219  Java (GCJ)
1220
1221  Ada
1222
1223     * The Ada runtime now supports multilibs on many platforms including
1224       x86_64, SPARC and PowerPC. Their build is enabled by default.
1225
1226New Targets and Target Specific Improvements
1227
1228  ARM
1229
1230     * GCC now supports optimizing for the Cortex-A9, Cortex-R4 and
1231       Cortex-R4F processors and has many other improvements to
1232       optimization for ARM processors.
1233     * GCC now supports the VFPv3 variant with 16 double-precision
1234       registers with -mfpu=vfpv3-d16. The option -mfpu=vfp3 has been
1235       renamed to -mfpu=vfpv3.
1236     * GCC now supports the -mfix-cortex-m3-ldrd option to work around an
1237       erratum on Cortex-M3 processors.
1238     * GCC now supports the __sync_* atomic operations for ARM EABI
1239       GNU/Linux.
1240     * The section anchors optimization is now enabled by default when
1241       optimizing for ARM.
1242     * GCC now uses a new EABI-compatible profiling interface for EABI
1243       targets. This requires a function __gnu_mcount_nc, which is
1244       provided by GNU libc versions 2.8 and later.
1245
1246  AVR
1247
1248     * The -mno-tablejump option has been deprecated because it has the
1249       same effect as the -fno-jump-tables option.
1250     * Added support for these new AVR devices:
1251          + ATA6289
1252          + ATtiny13A
1253          + ATtiny87
1254          + ATtiny167
1255          + ATtiny327
1256          + ATmega8C1
1257          + ATmega16C1
1258          + ATmega32C1
1259          + ATmega8M1
1260          + ATmega16M1
1261          + ATmega32M1
1262          + ATmega32U4
1263          + ATmega16HVB
1264          + ATmega4HVD
1265          + ATmega8HVD
1266          + ATmega64C1
1267          + ATmega64M1
1268          + ATmega16U4
1269          + ATmega32U6
1270          + ATmega128RFA1
1271          + AT90PWM81
1272          + AT90SCR100
1273          + M3000F
1274          + M3000S
1275          + M3001B
1276
1277  IA-32/x86-64
1278
1279     * Support for Intel AES built-in functions and code generation is
1280       available via -maes.
1281     * Support for Intel PCLMUL built-in function and code generation is
1282       available via -mpclmul.
1283     * Support for Intel AVX built-in functions and code generation is
1284       available via -mavx.
1285     * Automatically align the stack for local variables with alignment
1286       requirement.
1287     * GCC can now utilize the SVML library for vectorizing calls to a set
1288       of C99 functions if -mveclibabi=svml is specified and you link to
1289       an SVML ABI compatible library.
1290     * On x86-64, the ABI has been changed in the following cases to
1291       conform to the x86-64 ABI:
1292          + Passing/returning structures with flexible array member:
1293  struct foo
1294    {
1295      int i;
1296      int flex[];
1297    };
1298          + Passing/returning structures with complex float member:
1299  struct foo
1300    {
1301      int i;
1302      __complex__ float f;
1303    };
1304          + Passing/returning unions with long double member:
1305  union foo
1306    {
1307      int x;
1308      long double ld;
1309    };
1310       Code built with previous versions of GCC that uses any of these is
1311       not compatible with code built with GCC 4.4.0 or later.
1312     * A new target attribute was added to allow programmers to change the
1313       target options like -msse2 or -march=k8 for an individual function.
1314       You can also change the target options via the GCC target pragma
1315       for functions defined after the pragma.
1316     * GCC can now be configured with options --with-arch-32,
1317       --with-arch-64, --with-cpu-32, --with-cpu-64, --with-tune-32 and
1318       --with-tune-64 to control the default optimization separately for
1319       32-bit and 64-bit modes.
1320
1321  IA-32/IA64
1322
1323     * Support for __float128 (TFmode) IEEE quad type and corresponding
1324       TCmode IEEE complex quad type is available via the soft-fp library
1325       on IA-32/IA64 targets. This includes basic arithmetic operations
1326       (addition, subtraction, negation, multiplication and division) on
1327       __float128 real and TCmode complex values, the full set of IEEE
1328       comparisons between __float128 values, conversions to and from
1329       float, double and long double floating point types, as well as
1330       conversions to and from signed or unsigned integer, signed or
1331       unsigned long integer and signed or unsigned quad (TImode, IA64
1332       only) integer types. Additionally, all operations generate the full
1333       set of IEEE exceptions and support the full set of IEEE rounding
1334       modes.
1335
1336  M68K/ColdFire
1337
1338     * GCC now supports instruction scheduling for ColdFire V1, V3 and V4
1339       processors. (Scheduling support for ColdFire V2 processors was
1340       added in GCC 4.3.)
1341     * GCC now supports the -mxgot option to support programs requiring
1342       many GOT entries on ColdFire.
1343     * The m68k-*-linux-gnu target now builds multilibs by default.
1344
1345  MIPS
1346
1347     * MIPS Technologies have extended the original MIPS SVR4 ABI to
1348       include support for procedure linkage tables (PLTs) and copy
1349       relocations. These extensions allow GNU/Linux executables to use a
1350       significantly more efficient code model than the one defined by the
1351       original ABI.
1352       GCC support for this code model is available via a new command-line
1353       option, -mplt. There is also a new configure-time option,
1354       --with-mips-plt, to make -mplt the default.
1355       The new code model requires support from the assembler, the linker,
1356       and the runtime C library. This support is available in binutils
1357       2.19 and GLIBC 2.9.
1358     * GCC can now generate MIPS16 code for 32-bit GNU/Linux executables
1359       and 32-bit GNU/Linux shared libraries. This feature requires GNU
1360       binutils 2.19 or above.
1361     * Support for RMI's XLR processor is now available through the
1362       -march=xlr and -mtune=xlr options.
1363     * 64-bit targets can now perform 128-bit multiplications inline,
1364       instead of relying on a libgcc function.
1365     * Native GNU/Linux toolchains now support -march=native and
1366       -mtune=native, which select the host processor.
1367     * GCC now supports the R10K, R12K, R14K and R16K processors. The
1368       canonical -march= and -mtune= names for these processors are
1369       r10000, r12000, r14000 and r16000 respectively.
1370     * GCC can now work around the side effects of speculative execution
1371       on R10K processors. Please see the documentation of the
1372       -mr10k-cache-barrier option for details.
1373     * Support for the MIPS64 Release 2 instruction set has been added.
1374       The option -march=mips64r2 enables generation of these
1375       instructions.
1376     * GCC now supports Cavium Networks' Octeon processor. This support is
1377       available through the -march=octeon and -mtune=octeon options.
1378     * GCC now supports STMicroelectronics' Loongson 2E/2F processors. The
1379       canonical -march= and -mtune= names for these processors are
1380       loongson2e and loongson2f.
1381
1382  picochip
1383
1384   Picochip is a 16-bit processor. A typical picoChip contains over 250
1385   small cores, each with small amounts of memory. There are three
1386   processor variants (STAN, MEM and CTRL) with different instruction sets
1387   and memory configurations and they can be chosen using the -mae option.
1388
1389   This port is intended to be a "C" only port.
1390
1391  Power Architecture and PowerPC
1392
1393     * GCC now supports the e300c2, e300c3 and e500mc processors.
1394     * GCC now supports Xilinx processors with a single-precision FPU.
1395     * Decimal floating point is now supported for e500 processors.
1396
1397  S/390, zSeries and System z9/z10
1398
1399     * Support for the IBM System z10 EC/BC processor has been added. When
1400       using the -march=z10 option, the compiler will generate code making
1401       use of instructions provided by the General-Instruction-Extension
1402       Facility and the Execute-Extension Facility.
1403
1404  VxWorks
1405
1406     * GCC now supports the thread-local storage mechanism used on
1407       VxWorks.
1408
1409  Xtensa
1410
1411     * GCC now supports thread-local storage (TLS) for Xtensa processor
1412       configurations that include the Thread Pointer option. TLS also
1413       requires support from the assembler and linker; this support is
1414       provided in the GNU binutils beginning with version 2.19.
1415
1416Documentation improvements
1417
1418Other significant improvements
1419
1420GCC 4.4.1
1421
1422   This is the [13]list of problem reports (PRs) from GCC's bug tracking
1423   system that are known to be fixed in the 4.4.1 release. This list might
1424   not be complete (that is, it is possible that some PRs that have been
1425   fixed are not listed here).
1426
1427GCC 4.4.2
1428
1429   This is the [14]list of problem reports (PRs) from GCC's bug tracking
1430   system that are known to be fixed in the 4.4.2 release. This list might
1431   not be complete (that is, it is possible that some PRs that have been
1432   fixed are not listed here).
1433
1434GCC 4.4.3
1435
1436   This is the [15]list of problem reports (PRs) from GCC's bug tracking
1437   system that are known to be fixed in the 4.4.3 release. This list might
1438   not be complete (that is, it is possible that some PRs that have been
1439   fixed are not listed here).
1440
1441GCC 4.4.4
1442
1443   This is the [16]list of problem reports (PRs) from GCC's bug tracking
1444   system that are known to be fixed in the 4.4.4 release. This list might
1445   not be complete (that is, it is possible that some PRs that have been
1446   fixed are not listed here).
1447
1448GCC 4.4.5
1449
1450   This is the [17]list of problem reports (PRs) from GCC's bug tracking
1451   system that are known to be fixed in the 4.4.5 release. This list might
1452   not be complete (that is, it is possible that some PRs that have been
1453   fixed are not listed here).
1454
1455GCC 4.4.6
1456
1457   This is the [18]list of problem reports (PRs) from GCC's bug tracking
1458   system that are known to be fixed in the 4.4.6 release. This list might
1459   not be complete (that is, it is possible that some PRs that have been
1460   fixed are not listed here).
1461
1462GCC 4.4.7
1463
1464   This is the [19]list of problem reports (PRs) from GCC's bug tracking
1465   system that are known to be fixed in the 4.4.7 release. This list might
1466   not be complete (that is, it is possible that some PRs that have been
1467   fixed are not listed here).
1468
1469
1470    For questions related to the use of GCC, please consult these web
1471    pages and the [20]GCC manuals. If that fails, the
1472    [21]gcc-help@gcc.gnu.org mailing list might help. Comments on these
1473    web pages and the development of GCC are welcome on our developer
1474    list at [22]gcc@gcc.gnu.org. All of [23]our lists have public
1475    archives.
1476
1477   Copyright (C) [24]Free Software Foundation, Inc. Verbatim copying and
1478   distribution of this entire article is permitted in any medium,
1479   provided this notice is preserved.
1480
1481   These pages are [25]maintained by the GCC team. Last modified
1482   2012-03-13[26].
1483
1484References
1485
1486   1. http://gcc.gnu.org/gcc-4.4/changes.html#4.4.7
1487   2. http://gcc.gnu.org/gcc-4.3/changes.html#obsoleted
1488   3. http://gcc.gnu.org/gcc-4.4/porting_to.html
1489   4. http://gcc.gnu.org/wiki/Graphite
1490   5. http://openmp.org/wp/openmp-specifications/
1491   6. http://www.open-std.org/jtc1/sc22/wg14/www/docs/n1040.pdf
1492   7. http://gcc.gnu.org/gcc-4.4/cxx0x_status.html
1493   8. http://gcc.gnu.org/onlinedocs/libstdc++/manual/status.html#id476343
1494   9. http://gcc.gnu.org/onlinedocs/gfortran/Preprocessing-Options.html
1495  10. http://gcc.gnu.org/onlinedocs/gfortran/Error-and-Warning-Options.html#index-g_t_0040code_007bWarray-temporaries_007d-125
1496  11. http://gcc.gnu.org/onlinedocs/gfortran/Code-Gen-Options.html#index-g_t_0040code_007bfcheck-array-temporaries_007d-221
1497  12. http://gcc.gnu.org/onlinedocs/gfortran/Fortran-Dialect-Options.html#index-g_t_0040code_007bbackslash_007d-34
1498  13. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.1
1499  14. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.2
1500  15. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.3
1501  16. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.4
1502  17. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.5
1503  18. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.6
1504  19. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.4.7
1505  20. http://gcc.gnu.org/onlinedocs/
1506  21. mailto:gcc-help@gcc.gnu.org
1507  22. mailto:gcc@gcc.gnu.org
1508  23. http://gcc.gnu.org/lists.html
1509  24. http://www.fsf.org/
1510  25. http://gcc.gnu.org/about.html
1511  26. http://validator.w3.org/check/referer
1512======================================================================
1513http://gcc.gnu.org/gcc-4.3/index.html
1514
1515                             GCC 4.3 Release Series
1516
1517   Jun 27, 2011
1518
1519   The [1]GNU project and the GCC developers are pleased to announce the
1520   release of GCC 4.3.6.
1521
1522   This release is a bug-fix release, containing fixes for regressions in
1523   GCC 4.3.5 relative to previous releases of GCC.
1524
1525Release History
1526
1527   GCC 4.3.6
1528          Jun 27, 2011 ([2]changes)
1529
1530   GCC 4.3.5
1531          May 22, 2010 ([3]changes)
1532
1533   GCC 4.3.4
1534          August 4, 2009 ([4]changes)
1535
1536   GCC 4.3.3
1537          January 24, 2009 ([5]changes)
1538
1539   GCC 4.3.2
1540          August 27, 2008 ([6]changes)
1541
1542   GCC 4.3.1
1543          June 6, 2008 ([7]changes)
1544
1545   GCC 4.3.0
1546          March 5, 2008 ([8]changes)
1547
1548References and Acknowledgements
1549
1550   GCC used to stand for the GNU C Compiler, but since the compiler
1551   supports several other languages aside from C, it now stands for the
1552   GNU Compiler Collection.
1553
1554   A list of [9]successful builds is updated as new information becomes
1555   available.
1556
1557   The GCC developers would like to thank the numerous people that have
1558   contributed new features, improvements, bug fixes, and other changes as
1559   well as test results to GCC. This [10]amazing group of volunteers is
1560   what makes GCC successful.
1561
1562   For additional information about GCC please refer to the [11]GCC
1563   project web site or contact the [12]GCC development mailing list.
1564
1565   To obtain GCC please use [13]our mirror sites or [14]our SVN server.
1566
1567
1568    For questions related to the use of GCC, please consult these web
1569    pages and the [15]GCC manuals. If that fails, the
1570    [16]gcc-help@gcc.gnu.org mailing list might help. Comments on these
1571    web pages and the development of GCC are welcome on our developer
1572    list at [17]gcc@gcc.gnu.org. All of [18]our lists have public
1573    archives.
1574
1575   Copyright (C) [19]Free Software Foundation, Inc. Verbatim copying and
1576   distribution of this entire article is permitted in any medium,
1577   provided this notice is preserved.
1578
1579   These pages are [20]maintained by the GCC team. Last modified
1580   2011-06-27[21].
1581
1582References
1583
1584   1. http://www.gnu.org/
1585   2. http://gcc.gnu.org/gcc-4.3/changes.html
1586   3. http://gcc.gnu.org/gcc-4.3/changes.html
1587   4. http://gcc.gnu.org/gcc-4.3/changes.html
1588   5. http://gcc.gnu.org/gcc-4.3/changes.html
1589   6. http://gcc.gnu.org/gcc-4.3/changes.html
1590   7. http://gcc.gnu.org/gcc-4.3/changes.html
1591   8. http://gcc.gnu.org/gcc-4.3/changes.html
1592   9. http://gcc.gnu.org/gcc-4.3/buildstat.html
1593  10. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
1594  11. http://gcc.gnu.org/index.html
1595  12. mailto:gcc@gcc.gnu.org
1596  13. http://gcc.gnu.org/mirrors.html
1597  14. http://gcc.gnu.org/svn.html
1598  15. http://gcc.gnu.org/onlinedocs/
1599  16. mailto:gcc-help@gcc.gnu.org
1600  17. mailto:gcc@gcc.gnu.org
1601  18. http://gcc.gnu.org/lists.html
1602  19. http://www.fsf.org/
1603  20. http://gcc.gnu.org/about.html
1604  21. http://validator.w3.org/check/referer
1605======================================================================
1606http://gcc.gnu.org/gcc-4.3/changes.html
1607
1608                             GCC 4.3 Release Series
1609                        Changes, New Features, and Fixes
1610
1611   The latest release in the 4.3 release series is [1]GCC 4.3.5.
1612
1613Caveats
1614
1615     * GCC requires the [2]GMP and [3]MPFR libraries for building all the
1616       various front-end languages it supports. See the [4]prerequisites
1617       page for version requirements.
1618     * ColdFire targets now treat long double as having the same format as
1619       double. In earlier versions of GCC, they used the 68881 long double
1620       format instead.
1621     * The m68k-uclinux target now uses the same calling conventions as
1622       m68k-linux-gnu. You can select the original calling conventions by
1623       configuring for m68k-uclinuxoldabi instead. Note that
1624       m68k-uclinuxoldabi also retains the original 80-bit long double on
1625       ColdFire targets.
1626     * The -fforce-mem option has been removed because it has had no
1627       effect in the last few GCC releases.
1628     * The i386 -msvr3-shlib option has been removed since it is no longer
1629       used.
1630     * Fastcall for i386 has been changed not to pass aggregate arguments
1631       in registers, following Microsoft compilers.
1632     * Support for the AOF assembler has been removed from the ARM back
1633       end; this affects only the targets arm-semi-aof and armel-semi-aof,
1634       which are no longer recognized. We removed these targets without a
1635       deprecation period because we discovered that they have been
1636       unusable since GCC 4.0.0.
1637     * Support for the TMS320C3x/C4x processor (targets c4x-* and tic4x-*)
1638       has been removed. This support had been deprecated since GCC 4.0.0.
1639     * Support for a number of older systems and recently unmaintained or
1640       untested target ports of GCC has been declared obsolete in GCC 4.3.
1641       Unless there is activity to revive them, the next release of GCC
1642       will have their sources permanently removed.
1643       All GCC ports for the following processor architectures have been
1644       declared obsolete:
1645          + Morpho MT (mt-*)
1646       The following aliases for processor architectures have been
1647       declared obsolete. Users should use the indicated generic target
1648       names instead, with compile-time options such as -mcpu or
1649       configure-time options such as --with-cpu to control the
1650       configuration more precisely.
1651          + strongarm*-*-*, ep9312*-*-*, xscale*-*-* (use arm*-*-*
1652            instead).
1653          + parisc*-*-* (use hppa*-*-* instead).
1654          + m680[012]0-*-* (use m68k-*-* instead).
1655       All GCC ports for the following operating systems have been
1656       declared obsolete:
1657          + BeOS (*-*-beos*)
1658          + kaOS (*-*-kaos*)
1659          + GNU/Linux using the a.out object format (*-*-linux*aout*)
1660          + GNU/Linux using version 1 of the GNU C Library
1661            (*-*-linux*libc1*)
1662          + Solaris versions before Solaris 7 (*-*-solaris2.[0-6],
1663            *-*-solaris2.[0-6].*)
1664          + Miscellaneous System V (*-*-sysv*)
1665          + WindISS (*-*-windiss*)
1666       Also, those for some individual systems on particular architectures
1667       have been obsoleted:
1668          + UNICOS/mk on DEC Alpha (alpha*-*-unicosmk*)
1669          + CRIS with a.out object format (cris-*-aout)
1670          + BSD 4.3 on PA-RISC (hppa1.1-*-bsd*)
1671          + OSF/1 on PA-RISC (hppa1.1-*-osf*)
1672          + PRO on PA-RISC (hppa1.1-*-pro*)
1673          + Sequent PTX on IA32 (i[34567]86-sequent-ptx4*,
1674            i[34567]86-sequent-sysv4*)
1675          + SCO Open Server 5 on IA32 (i[34567]86-*-sco3.2v5*)
1676          + UWIN on IA32 (i[34567]86-*-uwin*) (support for UWIN as a host
1677            was previously [5]removed in 2001, leaving only the support
1678            for UWIN as a target now being deprecated)
1679          + ChorusOS on PowerPC (powerpc-*-chorusos*)
1680          + All VAX configurations apart from NetBSD and OpenBSD
1681            (vax-*-bsd*, vax-*-sysv*, vax-*-ultrix*)
1682     * The [6]-Wconversion option has been modified. Its purpose now is to
1683       warn for implicit conversions that may alter a value. This new
1684       behavior is available for both C and C++. Warnings about
1685       conversions between signed and unsigned integers can be disabled by
1686       using -Wno-sign-conversion. In C++, they are disabled by default
1687       unless -Wsign-conversion is explicitly requested. The old behavior
1688       of -Wconversion, that is, warn for prototypes causing a type
1689       conversion that is different from what would happen to the same
1690       argument in the absence of a prototype, has been moved to a new
1691       option -Wtraditional-conversion, which is only available for C.
1692     * The -m386, -m486, -mpentium and -mpentiumpro tuning options have
1693       been removed because they were deprecated for more than 3 GCC major
1694       releases. Use -mtune=i386, -mtune=i486, -mtune=pentium or
1695       -mtune=pentiumpro as a replacement.
1696     * The -funsafe-math-optimizations option now automatically turns on
1697       -fno-trapping-math in addition to -fno-signed-zeros, as it enables
1698       reassociation and thus may introduce or remove traps.
1699     * The -ftree-vectorize option is now on by default under -O3. In
1700       order to generate code for a SIMD extension, it has to be enabled
1701       as well: use -maltivec for PowerPC platforms and -msse/-msse2 for
1702       i?86 and x86_64.
1703     * More information on porting to GCC 4.3 from previous versions of
1704       GCC can be found in the [7]porting guide for this release.
1705
1706General Optimizer Improvements
1707
1708     * The GCC middle-end has been integrated with the [8]MPFR library.
1709       This allows GCC to evaluate and replace at compile-time calls to
1710       built-in math functions having constant arguments with their
1711       mathematically equivalent results. In making use of [9]MPFR, GCC
1712       can generate correct results regardless of the math library
1713       implementation or floating point precision of the host platform.
1714       This also allows GCC to generate identical results regardless of
1715       whether one compiles in native or cross-compile configurations to a
1716       particular target. The following built-in functions take advantage
1717       of this new capability: acos, acosh, asin, asinh, atan2, atan,
1718       atanh, cbrt, cos, cosh, drem, erf, erfc, exp10, exp2, exp, expm1,
1719       fdim, fma, fmax, fmin, gamma_r, hypot, j0, j1, jn, lgamma_r, log10,
1720       log1p, log2, log, pow10, pow, remainder, remquo, sin, sincos, sinh,
1721       tan, tanh, tgamma, y0, y1 and yn. The float and long double
1722       variants of these functions (e.g. sinf and sinl) are also handled.
1723       The sqrt and cabs functions with constant arguments were already
1724       optimized in prior GCC releases. Now they also use [10]MPFR.
1725     * A new forward propagation pass on RTL was added. The new pass
1726       replaces several slower transformations, resulting in compile-time
1727       improvements as well as better code generation in some cases.
1728     * A new command-line switch -frecord-gcc-switches has been added to
1729       GCC, although it is only enabled for some targets. The switch
1730       causes the command line that was used to invoke the compiler to be
1731       recorded into the object file that is being created. The exact
1732       format of this recording is target and binary file format
1733       dependent, but it usually takes the form of a note section
1734       containing ASCII text. The switch is related to the -fverbose-asm
1735       switch, but that one only records the information in the assembler
1736       output file as comments, so the information never reaches the
1737       object file.
1738     * The inliner heuristic is now aware of stack frame consumption. New
1739       command-line parameters --param large-stack-frame and --param
1740       large-stack-frame-growth can be used to limit stack frame size
1741       growth caused by inlining.
1742     * During feedback directed optimizations, the expected block size the
1743       memcpy, memset and bzero functions operate on is discovered and for
1744       cases of commonly used small sizes, specialized inline code is
1745       generated.
1746     * __builtin_expect no longer requires its argument to be a compile
1747       time constant.
1748     * Interprocedural optimization was reorganized to work on functions
1749       in SSA form. This enables more precise and cheaper dataflow
1750       analysis and makes writing interprocedural optimizations easier.
1751       The following improvements have been implemented on top of this
1752       framework:
1753          + Pre-inline optimization: Selected local optimization passes
1754            are run before the inliner (and other interprocedural passes)
1755            are executed. This significantly improves the accuracy of code
1756            growth estimates used by the inliner and reduces the overall
1757            memory footprint for large compilation units.
1758          + Early inlining (a simple bottom-up inliner pass inlining only
1759            functions whose body is smaller than the expected call
1760            overhead) is now executed with the early optimization passes,
1761            thus inlining already optimized function bodies into an
1762            unoptimized function that is subsequently optimized by early
1763            optimizers. This enables the compiler to quickly eliminate
1764            abstraction penalty in C++ programs.
1765          + Interprocedural constant propagation now operate on SSA form
1766            increasing accuracy of the analysis.
1767     * A new internal representation for GIMPLE statements has been
1768       contributed, resulting in compile-time memory savings.
1769     * The vectorizer was enhanced to support vectorization of outer
1770       loops, intra-iteration parallelism (loop-aware SLP), vectorization
1771       of strided accesses and loops with multiple data-types. Run-time
1772       dependency testing using loop versioning was added. The cost model,
1773       turned on by -fvect-cost-model, was developed.
1774
1775New Languages and Language specific improvements
1776
1777     * We have added new command-line options
1778       -finstrument-functions-exclude-function-list and
1779       -finstrument-functions-exclude-file-list. They provide more control
1780       over which functions are annotated by the -finstrument-functions
1781       option.
1782
1783  C family
1784
1785     * Implicit conversions between generic vector types are now only
1786       permitted when the two vectors in question have the same number of
1787       elements and compatible element types. (Note that the restriction
1788       involves compatible element types, not implicitly-convertible
1789       element types: thus, a vector type with element type int may not be
1790       implicitly converted to a vector type with element type unsigned
1791       int.) This restriction, which is in line with specifications for
1792       SIMD architectures such as AltiVec, may be relaxed using the flag
1793       -flax-vector-conversions. This flag is intended only as a
1794       compatibility measure and should not be used for new code.
1795     * -Warray-bounds has been added and is now enabled by default for
1796       -Wall . It produces warnings for array subscripts that can be
1797       determined at compile time to be always out of bounds.
1798       -Wno-array-bounds will disable the warning.
1799     * The constructor and destructor function attributes now accept
1800       optional priority arguments which control the order in which the
1801       constructor and destructor functions are run.
1802     * New [11]command-line options -Wtype-limits,
1803       -Wold-style-declaration, -Wmissing-parameter-type, -Wempty-body,
1804       -Wclobbered and -Wignored-qualifiers have been added for finer
1805       control of the diverse warnings enabled by -Wextra.
1806     * A new function attribute alloc_size has been added to mark up
1807       malloc style functions. For constant sized allocations this can be
1808       used to find out the size of the returned pointer using the
1809       __builtin_object_size() function for buffer overflow checking and
1810       similar. This supplements the already built-in malloc and calloc
1811       constant size handling.
1812     * Integer constants written in binary are now supported as a GCC
1813       extension. They consist of a prefix 0b or 0B, followed by a
1814       sequence of 0 and 1 digits.
1815     * A new predefined macro __COUNTER__ has been added. It expands to
1816       sequential integral values starting from 0. In conjunction with the
1817       ## operator, this provides a convenient means to generate unique
1818       identifiers.
1819     * A new command-line option -fdirectives-only has been added. It
1820       enables a special preprocessing mode which improves the performance
1821       of applications like distcc and ccache.
1822     * Fixed-point data types and operators have been added. They are
1823       based on Chapter 4 of the Embedded-C specification (n1169.pdf).
1824       Currently, only MIPS targets are supported.
1825     * Decimal floating-point arithmetic based on draft ISO/IEC TR 24732,
1826       N1241, is now supported as a GCC extension to C for targets
1827       i[34567]86-*-linux-gnu, powerpc*-*-linux-gnu, s390*-ibm-linux-gnu,
1828       and x86_64-*-linux-gnu. The feature introduces new data types
1829       _Decimal32, _Decimal64, and _Decimal128 with constant suffixes DF,
1830       DD, and DL.
1831
1832  C++
1833
1834     * [12]Experimental support for the upcoming ISO C++ standard, C++0x.
1835     * -Wc++0x-compat has been added and is now enabled by default for
1836       -Wall. It produces warnings for constructs whose meaning differs
1837       between ISO C++ 1998 and C++0x.
1838     * The -Wparentheses option now works for C++ as it does for C. It
1839       warns if parentheses are omitted when operators with confusing
1840       precedence are nested. It also warns about ambiguous else
1841       statements. Since -Wparentheses is enabled by -Wall, this may cause
1842       additional warnings with existing C++ code which uses -Wall. These
1843       new warnings may be disabled by using -Wall -Wno-parentheses.
1844     * The -Wmissing-declarations now works for C++ as it does for C.
1845     * The -fvisibility-ms-compat flag was added, to make it easier to
1846       port larger projects using shared libraries from Microsoft's Visual
1847       Studio to ELF and Mach-O systems.
1848     * C++ attribute handling has been overhauled for template arguments
1849       (ie dependent types). In particular, __attribute__((aligned(T)));
1850       works for C++ types.
1851
1852    Runtime Library (libstdc++)
1853
1854     * [13]Experimental support for the upcoming ISO C++ standard, C++0x.
1855     * Support for TR1 mathematical special functions and regular
1856       expressions. ([14]Implementation status of TR1)
1857     * Default what implementations give more elaborate exception strings
1858       for bad_cast, bad_typeid, bad_exception, and bad_alloc.
1859     * Header dependencies have been streamlined, reducing unnecessary
1860       includes and pre-processed bloat.
1861     * Variadic template implementations of items in <tuple> and
1862       <functional>.
1863     * An experimental [15]parallel mode has been added. This is a
1864       parallel implementation of many C++ Standard library algorithms,
1865       like std::accumulate, std::for_each, std::transform, or std::sort,
1866       to give but four examples. These algorithms can be substituted for
1867       the normal (sequential) libstdc++ algorithms on a piecemeal basis,
1868       or all existing algorithms can be transformed via the
1869       -D_GLIBCXX_PARALLEL macro.
1870     * Debug mode versions of classes in <unordered_set> and
1871       <unordered_map>.
1872     * Formal deprecation of <ext/hash_set> and <ext/hash_map>, which are
1873       now <backward/hash_set> and <backward/hash_map>. This code:
1874    #include <ext/hash_set>
1875    __gnu_cxx::hash_set<int> s;
1876
1877       Can be transformed (in order of preference) to:
1878    #include <tr1/unordered_set>
1879    std::tr1::unordered_set<int> s;
1880
1881       or
1882    #include <backward/hash_set>
1883    __gnu_cxx::hash_set<int> s;
1884
1885       Similar transformations apply to __gnu_cxx::hash_map,
1886       __gnu_cxx::hash_multimap, __gnu_cxx::hash_set,
1887       __gnu_cxx::hash_multiset.
1888
1889  Fortran
1890
1891     * Due to the fact that the [16]GMP and [17]MPFR libraries are
1892       required for all languages, Fortran is no longer special in this
1893       regard and is available by default.
1894     * The [18]-fexternal-blas option has been added, which generates
1895       calls to BLAS routines for intrinsic matrix operations such as
1896       matmul rather than using the built-in algorithms.
1897     * Support to give a backtrace (compiler flag -fbacktrace or
1898       environment variable GFORTRAN_ERROR_BACKTRACE; on glibc systems
1899       only) or a core dump (-fdump-core, GFORTRAN_ERROR_DUMPCORE) when a
1900       run-time error occured.
1901     * GNU Fortran now defines __GFORTRAN__ when it runs the C
1902       preprocessor (CPP).
1903     * The [19]-finit-local-zero, -finit-real, -finit-integer,
1904       -finit-character, and -finit-logical options have been added, which
1905       can be used to initialize local variables.
1906     * The intrinsic procedures [20]GAMMA and [21]LGAMMA have been added,
1907       which calculate the Gamma function and its logarithm. Use EXTERNAL
1908       gamma if you want to use your own gamma function.
1909     * GNU Fortran now regards the backslash character as literal (as
1910       required by the Fortran 2003 standard); using [22]-fbackslash GNU
1911       Fortran interprets backslashes as C-style escape characters.
1912     * The [23]interpretation of binary, octal and hexadecimal (BOZ)
1913       literal constants has been changed. Before they were always
1914       interpreted as integer; now they are bit-wise transferred as
1915       argument of INT, REAL, DBLE and CMPLX as required by the Fortran
1916       2003 standard, and for real and complex variables in DATA
1917       statements or when directly assigned to real and complex variables.
1918       Everywhere else and especially in expressions they are still
1919       regarded as integer constants.
1920     * Fortran 2003 support has been extended:
1921          + Intrinsic statements IMPORT, PROTECTED, VALUE and VOLATILE
1922          + Pointer intent
1923          + Intrinsic module ISO_ENV_FORTRAN
1924          + Interoperability with C (ISO C Bindings)
1925          + ABSTRACT INTERFACES and PROCEDURE statements (without POINTER
1926            attribute)
1927          + Fortran 2003 BOZ
1928
1929  Java (GCJ)
1930
1931     * GCJ now uses the Eclipse Java compiler for its Java parsing needs.
1932       This enables the use of all 1.5 language features, and fixes most
1933       existing front end bugs.
1934     * libgcj now supports all 1.5 language features which require runtime
1935       support: foreach, enum, annotations, generics, and auto-boxing.
1936     * We've made many changes to the tools shipped with gcj.
1937          + The old jv-scan tool has been removed. This tool never really
1938            worked properly. There is no replacement.
1939          + gcjh has been rewritten. Some of its more obscure options no
1940            longer work, but are still recognized in an attempt at
1941            compatibility. gjavah is a new program with similar
1942            functionality but different command-line options.
1943          + grmic and grmiregistry have been rewritten. grmid has been
1944            added.
1945          + gjar replaces the old fastjar.
1946          + gjarsigner (used for signing jars), gkeytool (used for key
1947            management), gorbd (for CORBA), gserialver (computes
1948            serialization UIDs), and gtnameserv (also for CORBA) are now
1949            installed.
1950     * The ability to dump the contents of the java run time heap to a
1951       file for off-line analysis has been added. The heap dumps may be
1952       analyzed with the new gc-analyze tool. They may be generated on
1953       out-of-memory conditions or on demand and are controlled by the new
1954       run time class gnu.gcj.util.GCInfo.
1955     * java.util.TimeZone can now read files from /usr/share/zoneinfo to
1956       provide correct, updated, timezone information. This means that
1957       packagers no longer have to update libgcj when a time zone change
1958       is published.
1959
1960New Targets and Target Specific Improvements
1961
1962  IA-32/x86-64
1963
1964     * Tuning for Intel Core 2 processors is available via -mtune=core2
1965       and -march=core2.
1966     * Tuning for AMD Geode processors is available via -mtune=geode and
1967       -march=geode.
1968     * Code generation of block move (memcpy) and block set (memset) was
1969       rewritten. GCC can now pick the best algorithm (loop, unrolled
1970       loop, instruction with rep prefix or a library call) based on the
1971       size of the block being copied and the CPU being optimized for. A
1972       new option -minline-stringops-dynamically has been added. With this
1973       option string operations of unknown size are expanded such that
1974       small blocks are copied by in-line code, while for large blocks a
1975       library call is used. This results in faster code than
1976       -minline-all-stringops when the library implementation is capable
1977       of using cache hierarchy hints. The heuristic choosing the
1978       particular algorithm can be overwritten via -mstringop-strategy.
1979       Newly also memset of values different from 0 is inlined.
1980     * GCC no longer places the cld instruction before string operations.
1981       Both i386 and x86-64 ABI documents mandate the direction flag to be
1982       clear at the entry of a function. It is now invalid to set the flag
1983       in asm statement without reseting it afterward.
1984     * Support for SSSE3 built-in functions and code generation are
1985       available via -mssse3.
1986     * Support for SSE4.1 built-in functions and code generation are
1987       available via -msse4.1.
1988     * Support for SSE4.2 built-in functions and code generation are
1989       available via -msse4.2.
1990     * Both SSE4.1 and SSE4.2 support can be enabled via -msse4.
1991     * A new set of options -mpc32, -mpc64 and -mpc80 have been added to
1992       allow explicit control of x87 floating point precision.
1993     * Support for __float128 (TFmode) IEEE quad type and corresponding
1994       TCmode IEEE complex quad type is available via the soft-fp library
1995       on x86_64 targets. This includes basic arithmetic operations
1996       (addition, subtraction, negation, multiplication and division) on
1997       __float128 real and TCmode complex values, the full set of IEEE
1998       comparisons between __float128 values, conversions to and from
1999       float, double and long double floating point types, as well as
2000       conversions to and from signed or unsigned integer, signed or
2001       unsigned long integer and signed or unsigned quad (TImode) integer
2002       types. Additionally, all operations generate the full set of IEEE
2003       exceptions and support the full set of IEEE rounding modes.
2004     * GCC can now utilize the ACML library for vectorizing calls to a set
2005       of C99 functions on x86_64 if -mveclibabi=acml is specified and you
2006       link to an ACML ABI compatible library.
2007
2008  ARM
2009
2010     * Compiler and Library support for Thumb-2 and the ARMv7 architecture
2011       has been added.
2012
2013  CRIS
2014
2015    New features
2016
2017     * Compiler and Library support for the CRIS v32 architecture, as
2018       found in Axis Communications ETRAX FS and ARTPEC-3 chips, has been
2019       added.
2020
2021    Configuration changes
2022
2023     * The cris-*-elf target now includes support for CRIS v32, including
2024       libraries, through the -march=v32 option.
2025     * A new crisv32-*-elf target defaults to generate code for CRIS v32.
2026     * A new crisv32-*-linux* target defaults to generate code for CRIS
2027       v32.
2028     * The cris-*-aout target has been obsoleted.
2029
2030    Improved support for built-in functions
2031
2032     * GCC can now use the lz and swapwbr instructions to implement the
2033       __builtin_clz, __builtin_ctz and __builtin_ffs family of functions.
2034     * __builtin_bswap32 is now implemented using the swapwb instruction,
2035       when available.
2036
2037  m68k and ColdFire
2038
2039    New features
2040
2041     * Support for several new ColdFire processors has been added. You can
2042       generate code for them using the new -mcpu option.
2043     * All targets now support ColdFire processors.
2044     * m68k-uclinux targets have improved support for C++ constructors and
2045       destructors, and for shared libraries.
2046     * It is now possible to set breakpoints on the first or last line of
2047       a function, even if there are no statements on that line.
2048
2049    Optimizations
2050
2051     * Support for sibling calls has been added.
2052     * More use is now made of the ColdFire mov3q instruction.
2053     * __builtin_clz is now implemented using the ff1 ColdFire
2054       instruction, when available.
2055     * GCC now honors the -m68010 option. 68010 code now uses clr rather
2056       than move to zero volatile memory.
2057     * 68020 targets and above can now use symbol(index.size*scale)
2058       addresses for indexed array accesses. Earlier compilers would
2059       always load the symbol into a base register first.
2060
2061    Configuration changes
2062
2063     * All m68k and ColdFire targets now allow the default processor to be
2064       set at configure time using --with-cpu.
2065     * A --with-arch configuration option has been added. This option
2066       allows you to restrict a target to ColdFire or non-ColdFire
2067       processors.
2068
2069    Preprocessor macros
2070
2071     * An __mcfv*__ macro is now defined for all ColdFire targets.
2072       (Earlier versions of GCC only defined __mcfv4e__.)
2073     * __mcf_cpu_*, __mcf_family_* and __mcffpu__ macros have been added.
2074     * All targets now define __mc68010 and __mc68010__ when generating
2075       68010 code.
2076
2077    Command-line changes
2078
2079     * New command-line options -march, -mcpu, -mtune and -mhard-float
2080       have been added. These options apply to both m68k and ColdFire
2081       targets.
2082     * -mno-short, -mno-bitfield and -mno-rtd are now accepted as negative
2083       versions of -mshort, etc.
2084     * -fforce-addr has been removed. It is now ignored by the compiler.
2085
2086    Other improvements
2087
2088     * ColdFire targets now try to maintain a 4-byte-aligned stack where
2089       possible.
2090     * m68k-uclinux targets now try to avoid situations that lead to the
2091       load-time error: BINFMT_FLAT: reloc outside program.
2092
2093  MIPS
2094
2095    Changes to existing configurations
2096
2097     * libffi and libjava now support all three GNU/Linux ABIs: o32, n32
2098       and n64. Every GNU/Linux configuration now builds these libraries
2099       by default.
2100     * GNU/Linux configurations now generate -mno-shared code unless
2101       overridden by -fpic, -fPIC, -fpie or -fPIE.
2102     * mipsisa32*-linux-gnu configurations now generate hard-float code by
2103       default, just like other mipsisa32* and mips*-linux-gnu
2104       configurations. You can build a soft-float version of any
2105       mips*-linux-gnu configuration by passing --with-float=soft to
2106       configure.
2107     * mips-wrs-vxworks now supports run-time processes (RTPs).
2108
2109    Changes to existing command-line options
2110
2111     * The -march and -mtune options no longer accept 24k as a processor
2112       name. Please use 24kc, 24kf2_1 or 24kf1_1 instead.
2113     * The -march and -mtune options now accept 24kf2_1, 24kef2_1 and
2114       34kf2_1 as synonyms for 24kf, 24kef and 34kf respectively. The
2115       options also accept 24kf1_1, 24kef1_1 and 34kf1_1 as synonyms for
2116       24kx, 24kex and 34kx.
2117
2118    New configurations
2119
2120   GCC now supports the following configurations:
2121     * mipsisa32r2*-linux-gnu*, which generates MIPS32 revision 2 code by
2122       default. Earlier releases also recognized this configuration, but
2123       they treated it in the same way as mipsisa32*-linux-gnu*. Note that
2124       you can customize any mips*-linux-gnu* configuration to a
2125       particular ISA or processor by passing an appropriate --with-arch
2126       option to configure.
2127     * mipsisa*-sde-elf*, which provides compatibility with MIPS
2128       Technologies' SDE toolchains. The configuration uses the SDE
2129       libraries by default, but you can use it like other newlib-based
2130       ELF configurations by passing --with-newlib to configure. It is the
2131       only configuration besides mips64vr*-elf* to build MIPS16 as well
2132       as non-MIPS16 libraries.
2133     * mipsisa*-elfoabi*, which is similar to the general mipsisa*-elf*
2134       configuration, but uses the o32 and o64 ABIs instead of the 32-bit
2135       and 64-bit forms of the EABI.
2136
2137    New processors and application-specific extensions
2138
2139     * Support for the SmartMIPS ASE is available through the new
2140       -msmartmips option.
2141     * Support for revision 2 of the DSP ASE is available through the new
2142       -mdspr2 option. A new preprocessor macro called __mips_dsp_rev
2143       indicates the revision of the ASE in use.
2144     * Support for the 4KS and 74K families of processors is available
2145       through the -march and -mtune options.
2146
2147    Improved support for built-in functions
2148
2149     * GCC can now use load-linked, store-conditional and sync
2150       instructions to implement atomic built-in functions such as
2151       __sync_fetch_and_add. The memory reference must be 4 bytes wide for
2152       32-bit targets and either 4 or 8 bytes wide for 64-bit targets.
2153     * GCC can now use the clz and dclz instructions to implement the
2154       __builtin_ctz and __builtin_ffs families of functions.
2155     * There is a new __builtin___clear_cache function for flushing the
2156       instruction cache. GCC expands this function inline on MIPS32
2157       revision 2 targets, otherwise it calls the function specified by
2158       -mcache-flush-func.
2159
2160    MIPS16 improvements
2161
2162     * GCC can now compile objects that contain a mixture of MIPS16 and
2163       non-MIPS16 code. There are two new attributes, mips16 and nomips16,
2164       for specifying which mode a function should use.
2165     * A new option called -minterlink-mips16 makes non-MIPS16 code
2166       link-compatible with MIPS16 code.
2167     * After many bug fixes, the long-standing MIPS16 -mhard-float support
2168       should now work fairly reliably.
2169     * GCC can now use the MIPS16e save and restore instructions.
2170     * -fsection-anchors now works in MIPS16 mode. MIPS16 code compiled
2171       with -G0 -fsection-anchors is often smaller than code compiled with
2172       -G8. However, please note that you must usually compile all objects
2173       in your application with the same -G option; see the documentation
2174       of -G for details.
2175     * A new option called-mcode-readable specifies which instructions are
2176       allowed to load from the code segment. -mcode-readable=yes is the
2177       default and says that any instruction may load from the code
2178       segment. The other alternatives are -mcode-readable=pcrel, which
2179       says that only PC-relative MIPS16 instructions may load from the
2180       code segment, and -mcode-readable=no, which says that no
2181       instruction may do so. Please see the documentation for more
2182       details, including example uses.
2183
2184    Small-data improvements
2185
2186   There are three new options for controlling small data:
2187     * -mno-extern-sdata, which disables small-data accesses for
2188       externally-defined variables. Code compiled with -Gn
2189       -mno-extern-sdata will be link-compatible with any -G setting
2190       between -G0 and -Gn inclusive.
2191     * -mno-local-sdata, which disables the use of small-data sections for
2192       data that is not externally visible. This option can be a useful
2193       way of reducing small-data usage in less performance-critical parts
2194       of an application.
2195     * -mno-gpopt, which disables the use of the $gp register while still
2196       honoring the -G limit when placing externally-visible data. This
2197       option implies -mno-extern-sdata and -mno-local-sdata and it can be
2198       useful in situations where $gp does not necessarily hold the
2199       expected value.
2200
2201    Miscellaneous improvements
2202
2203     * There is a new option called -mbranch-cost for tweaking the
2204       perceived cost of branches.
2205     * If GCC is configured to use a version of GAS that supports the
2206       .gnu_attribute directive, it will use that directive to record
2207       certain properties of the output code. .gnu_attribute is new to GAS
2208       2.18.
2209     * There are two new function attributes, near and far, for overriding
2210       the command-line setting of -mlong-calls on a function-by-function
2211       basis.
2212     * -mfp64, which previously required a 64-bit target, now works with
2213       MIPS32 revision 2 targets as well. The mipsisa*-elfoabi* and
2214       mipsisa*-sde-elf* configurations provide suitable library support.
2215     * GCC now recognizes the -mdmx and -mmt options and passes them down
2216       to the assembler. It does nothing else with the options at present.
2217
2218  SPU (Synergistic Processor Unit) of the Cell Broadband Engine Architecture
2219  (BEA)
2220
2221     * Support has been added for this new architecture.
2222
2223  RS6000 (POWER/PowerPC)
2224
2225     * Support for the PowerPC 750CL paired-single instructions has been
2226       added with a new powerpc-*-linux*paired* target configuration. It
2227       is enabled by an associated -mpaired option and can be accessed
2228       using new built-in functions.
2229     * Support for auto-detecting architecture and system configuration to
2230       auto-select processor optimization tuning.
2231     * Support for VMX on AIX 5.3 has been added.
2232     * Support for AIX Version 6.1 has been added.
2233
2234  S/390, zSeries and System z9
2235
2236     * Support for the IBM System z9 EC/BC processor (z9 GA3) has been
2237       added. When using the -march=z9-ec option, the compiler will
2238       generate code making use of instructions provided by the decimal
2239       floating point facility and the floating point conversion facility
2240       (pfpo). Besides the instructions used to implement decimal floating
2241       point operations these facilities also contain instructions to move
2242       between general purpose and floating point registers and to modify
2243       and copy the sign-bit of floating point values.
2244     * When the -march=z9-ec option is used the new
2245       -mhard-dfp/-mno-hard-dfp options can be used to specify whether the
2246       decimal floating point hardware instructions will be used or not.
2247       If none of them is given the hardware support is enabled by
2248       default.
2249     * The -mstack-guard option can now be omitted when using stack
2250       checking via -mstack-size in order to let GCC choose a sensible
2251       stack guard value according to the frame size of each function.
2252     * Various changes to improve performance of generated code have been
2253       implemented, including:
2254          + The condition code set by an add logical with carry
2255            instruction is now available for overflow checks like: a + b +
2256            carry < b.
2257          + The test data class instruction is now used to implement
2258            sign-bit and infinity checks of binary and decimal floating
2259            point numbers.
2260
2261  SPARC
2262
2263     * Support for the Sun UltraSPARC T2 (Niagara 2) processor has been
2264       added.
2265
2266  Xtensa
2267
2268     * Stack unwinding for exception handling now uses by default a
2269       specialized version of DWARF unwinding. This is not
2270       binary-compatible with the setjmp/longjmp (sjlj) unwinding used for
2271       Xtensa with previous versions of GCC.
2272     * For Xtensa processors that include the Conditional Store option,
2273       the built-in functions for atomic memory access are now implemented
2274       using S32C1I instructions.
2275     * If the Xtensa NSA option is available, GCC will use it to implement
2276       the __builtin_ctz and __builtin_clz functions.
2277
2278Documentation improvements
2279
2280     * Existing libstdc++ documentation has been edited and restructured
2281       into a single DocBook XML manual. The results can be viewed online
2282       [24]here.
2283
2284Other significant improvements
2285
2286     * The compiler's --help command-line option has been extended so that
2287       it now takes an optional set of arguments. These arguments restrict
2288       the information displayed to specific classes of command-line
2289       options, and possibly only a subset of those options. It is also
2290       now possible to replace the descriptive text associated with each
2291       displayed option with an indication of its current value, or for
2292       binary options, whether it has been enabled or disabled.
2293       Here are some examples. The following will display all the options
2294       controlling warning messages:
2295      --help=warnings
2296
2297       Whereas this will display all the undocumented, target specific
2298       options:
2299      --help=target,undocumented
2300
2301       This sequence of commands will display the binary optimizations
2302       that are enabled by -O3:
2303      gcc -c -Q -O3 --help=optimizers > /tmp/O3-opts
2304      gcc -c -Q -O2 --help=optimizers > /tmp/O2-opts
2305      diff /tmp/O2-opts /tmp/O3-opts | grep enabled
2306
2307     * The configure options --with-pkgversion and --with-bugurl have been
2308       added. These allow distributors of GCC to include a
2309       distributor-specific string in manuals and --version output and to
2310       specify the URL for reporting bugs in their versions of GCC.
2311
2312GCC 4.3.1
2313
2314   This is the [25]list of problem reports (PRs) from GCC's bug tracking
2315   system that are known to be fixed in the 4.3.1 release. This list might
2316   not be complete (that is, it is possible that some PRs that have been
2317   fixed are not listed here).
2318
2319Target Specific Changes
2320
2321  IA-32/x86-64
2322
2323    ABI changes
2324
2325     * Starting with GCC 4.3.1, decimal floating point variables are
2326       aligned to their natural boundaries when they are passed on the
2327       stack for i386.
2328
2329    Command-line changes
2330
2331     * Starting with GCC 4.3.1, the -mcld option has been added to
2332       automatically generate a cld instruction in the prologue of
2333       functions that use string instructions. This option is used for
2334       backward compatibility on some operating systems and can be enabled
2335       by default for 32-bit x86 targets by configuring GCC with the
2336       --enable-cld configure option.
2337
2338GCC 4.3.2
2339
2340   This is the [26]list of problem reports (PRs) from GCC's bug tracking
2341   system that are known to be fixed in the 4.3.2 release. This list might
2342   not be complete (that is, it is possible that some PRs that have been
2343   fixed are not listed here).
2344
2345GCC 4.3.3
2346
2347   This is the [27]list of problem reports (PRs) from GCC's bug tracking
2348   system that are known to be fixed in the 4.3.3 release. This list might
2349   not be complete (that is, it is possible that some PRs that have been
2350   fixed are not listed here).
2351
2352GCC 4.3.4
2353
2354   This is the [28]list of problem reports (PRs) from GCC's bug tracking
2355   system that are known to be fixed in the 4.3.4 release. This list might
2356   not be complete (that is, it is possible that some PRs that have been
2357   fixed are not listed here).
2358
2359GCC 4.3.5
2360
2361   This is the [29]list of problem reports (PRs) from GCC's bug tracking
2362   system that are known to be fixed in the 4.3.5 release. This list might
2363   not be complete (that is, it is possible that some PRs that have been
2364   fixed are not listed here).
2365
2366GCC 4.3.6
2367
2368   This is the [30]list of problem reports (PRs) from GCC's bug tracking
2369   system that are known to be fixed in the 4.3.6 release. This list might
2370   not be complete (that is, it is possible that some PRs that have been
2371   fixed are not listed here).
2372
2373
2374    For questions related to the use of GCC, please consult these web
2375    pages and the [31]GCC manuals. If that fails, the
2376    [32]gcc-help@gcc.gnu.org mailing list might help. Comments on these
2377    web pages and the development of GCC are welcome on our developer
2378    list at [33]gcc@gcc.gnu.org. All of [34]our lists have public
2379    archives.
2380
2381   Copyright (C) [35]Free Software Foundation, Inc. Verbatim copying and
2382   distribution of this entire article is permitted in any medium,
2383   provided this notice is preserved.
2384
2385   These pages are [36]maintained by the GCC team. Last modified
2386   2011-09-12[37].
2387
2388References
2389
2390   1. http://gcc.gnu.org/gcc-4.3/changes.html#4.3.5
2391   2. http://gmplib.org/
2392   3. http://www.mpfr.org/
2393   4. http://gcc.gnu.org/install/prerequisites.html
2394   5. http://gcc.gnu.org/ml/gcc-announce/2001/msg00000.html
2395   6. http://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html#Warning-Options
2396   7. http://gcc.gnu.org/gcc-4.3/porting_to.html
2397   8. http://www.mpfr.org/
2398   9. http://www.mpfr.org/
2399  10. http://www.mpfr.org/
2400  11. http://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html
2401  12. http://gcc.gnu.org/gcc-4.3/cxx0x_status.html
2402  13. http://gcc.gnu.org/gcc-4.3/cxx0x_status.html
2403  14. http://gcc.gnu.org/onlinedocs/libstdc++/manual/bk01pt01ch01.html#m anual.intro.status.standard.tr1
2404  15. http://gcc.gnu.org/onlinedocs/libstdc++/manual/parallel_mode.html
2405  16. http://gmplib.org/
2406  17. http://www.mpfr.org/
2407  18. http://gcc.gnu.org/onlinedocs/gfortran/Code-Gen-Options.html#Code-Gen-Options
2408  19. http://gcc.gnu.org/onlinedocs/gfortran/Code-Gen-Options.html#index-g_t_0040code_007bfinit-local-zero_007d-167
2409  20. http://gcc.gnu.org/onlinedocs/gcc-4.3.0/gfortran/GAMMA.html
2410  21. http://gcc.gnu.org/onlinedocs/gcc-4.3.0/gfortran/LGAMMA.html
2411  22. http://gcc.gnu.org/onlinedocs/gfortran/Fortran-Dialect-Options.html
2412  23. http://gcc.gnu.org/onlinedocs/gfortran/BOZ-literal-constants.html
2413  24. http://gcc.gnu.org/onlinedocs/libstdc++/
2414  25. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.3.1
2415  26. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.3.2
2416  27. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.3.3
2417  28. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.3.4
2418  29. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.3.5
2419  30. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.3.6
2420  31. http://gcc.gnu.org/onlinedocs/
2421  32. mailto:gcc-help@gcc.gnu.org
2422  33. mailto:gcc@gcc.gnu.org
2423  34. http://gcc.gnu.org/lists.html
2424  35. http://www.fsf.org/
2425  36. http://gcc.gnu.org/about.html
2426  37. http://validator.w3.org/check/referer
2427======================================================================
2428http://gcc.gnu.org/gcc-4.2/index.html
2429
2430                             GCC 4.2 Release Series
2431
2432   May 19, 2008
2433
2434   The [1]GNU project and the GCC developers are pleased to announce the
2435   release of GCC 4.2.4.
2436
2437   This release is a bug-fix release, containing fixes for regressions in
2438   GCC 4.2.3 relative to previous releases of GCC.
2439
2440Release History
2441
2442   GCC 4.2.4
2443          May 19, 2008 ([2]changes)
2444
2445   GCC 4.2.3
2446          February 1, 2008 ([3]changes)
2447
2448   GCC 4.2.2
2449          October 7, 2007 ([4]changes)
2450
2451   GCC 4.2.1
2452          July 18, 2007 ([5]changes)
2453
2454   GCC 4.2.0
2455          May 13, 2007 ([6]changes)
2456
2457References and Acknowledgements
2458
2459   GCC used to stand for the GNU C Compiler, but since the compiler
2460   supports several other languages aside from C, it now stands for the
2461   GNU Compiler Collection.
2462
2463   A list of [7]successful builds is updated as new information becomes
2464   available.
2465
2466   The GCC developers would like to thank the numerous people that have
2467   contributed new features, improvements, bug fixes, and other changes as
2468   well as test results to GCC. This [8]amazing group of volunteers is
2469   what makes GCC successful.
2470
2471   For additional information about GCC please refer to the [9]GCC project
2472   web site or contact the [10]GCC development mailing list.
2473
2474   To obtain GCC please use [11]our mirror sites or [12]our SVN server.
2475
2476
2477    For questions related to the use of GCC, please consult these web
2478    pages and the [13]GCC manuals. If that fails, the
2479    [14]gcc-help@gcc.gnu.org mailing list might help. Comments on these
2480    web pages and the development of GCC are welcome on our developer
2481    list at [15]gcc@gcc.gnu.org. All of [16]our lists have public
2482    archives.
2483
2484   Copyright (C) [17]Free Software Foundation, Inc. Verbatim copying and
2485   distribution of this entire article is permitted in any medium,
2486   provided this notice is preserved.
2487
2488   These pages are [18]maintained by the GCC team. Last modified
2489   2011-04-25[19].
2490
2491References
2492
2493   1. http://www.gnu.org/
2494   2. http://gcc.gnu.org/gcc-4.2/changes.html
2495   3. http://gcc.gnu.org/gcc-4.2/changes.html
2496   4. http://gcc.gnu.org/gcc-4.2/changes.html
2497   5. http://gcc.gnu.org/gcc-4.2/changes.html
2498   6. http://gcc.gnu.org/gcc-4.2/changes.html
2499   7. http://gcc.gnu.org/gcc-4.2/buildstat.html
2500   8. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
2501   9. http://gcc.gnu.org/index.html
2502  10. mailto:gcc@gcc.gnu.org
2503  11. http://gcc.gnu.org/mirrors.html
2504  12. http://gcc.gnu.org/svn.html
2505  13. http://gcc.gnu.org/onlinedocs/
2506  14. mailto:gcc-help@gcc.gnu.org
2507  15. mailto:gcc@gcc.gnu.org
2508  16. http://gcc.gnu.org/lists.html
2509  17. http://www.fsf.org/
2510  18. http://gcc.gnu.org/about.html
2511  19. http://validator.w3.org/check/referer
2512======================================================================
2513http://gcc.gnu.org/gcc-4.2/changes.html
2514
2515                             GCC 4.2 Release Series
2516                        Changes, New Features, and Fixes
2517
2518Caveats
2519
2520     * GCC no longer accepts the -fshared-data option. This option has had
2521       no effect in any GCC 4 release; the targets to which the option
2522       used to apply had been removed before GCC 4.0.
2523
2524General Optimizer Improvements
2525
2526     * New command-line options specify the possible relationships among
2527       parameters and between parameters and global data. For example,
2528       -fargument-noalias-anything specifies that arguments do not alias
2529       any other storage.
2530       Each language will automatically use whatever option is required by
2531       the language standard. You should not need to use these options
2532       yourself.
2533
2534New Languages and Language specific improvements
2535
2536     * [1]OpenMP is now supported for the C, C++ and Fortran compilers.
2537     * New command-line options -fstrict-overflow and -Wstrict-overflow
2538       have been added. -fstrict-overflow tells the compiler that it may
2539       assume that the program follows the strict signed overflow
2540       semantics permitted for the language: for C and C++ this means that
2541       the compiler may assume that signed overflow does not occur. For
2542       example, a loop like
2543      for (i = 1; i > 0; i *= 2)
2544
2545       is presumably intended to continue looping until i overflows. With
2546       -fstrict-overflow, the compiler may assume that signed overflow
2547       will not occur, and transform this into an infinite loop.
2548       -fstrict-overflow is turned on by default at -O2, and may be
2549       disabled via -fno-strict-overflow. The -Wstrict-overflow option may
2550       be used to warn about cases where the compiler assumes that signed
2551       overflow will not occur. It takes five different levels:
2552       -Wstrict-overflow=1 to 5. See the [2]documentation for details.
2553       -Wstrict-overflow=1 is enabled by -Wall.
2554     * The new command-line option -fno-toplevel-reorder directs GCC to
2555       emit top-level functions, variables, and asm statements in the same
2556       order that they appear in the input file. This is intended to
2557       support existing code which relies on a particular ordering (for
2558       example, code which uses top-level asm statements to switch
2559       sections). For new code, it is generally better to use function and
2560       variable attributes. The -fno-toplevel-reorder option may be used
2561       for most cases which currently use -fno-unit-at-a-time. The
2562       -fno-unit-at-a-time option will be removed in some future version
2563       of GCC. If you know of a case which requires -fno-unit-at-a-time
2564       which is not fixed by -fno-toplevel-reorder, please open a bug
2565       report.
2566
2567  C family
2568
2569     * The pragma redefine_extname will now macro expand its tokens for
2570       compatibility with SunPRO.
2571     * In the next release of GCC, 4.3, -std=c99 or -std=gnu99 will direct
2572       GCC to handle inline functions as specified in the C99 standard. In
2573       preparation for this, GCC 4.2 will warn about any use of non-static
2574       inline functions in gnu99 or c99 mode. This new warning may be
2575       disabled with the new gnu_inline function attribute or the new
2576       -fgnu89-inline command-line option. Also, GCC 4.2 and later will
2577       define one of the preprocessor macros __GNUC_GNU_INLINE__ or
2578       __GNUC_STDC_INLINE__ to indicate the semantics of inline functions
2579       in the current compilation.
2580     * A new command-line option -Waddress has been added to warn about
2581       suspicious uses of memory addresses as, for example, using the
2582       address of a function in a conditional expression, and comparisons
2583       against the memory address of a string literal. This warning is
2584       enabled by -Wall.
2585
2586  C++
2587
2588     * C++ visibility handling has been overhauled.
2589       Restricted visiblity is propagated from classes to members, from
2590       functions to local statics, and from templates and template
2591       arguments to instantiations, unless the latter has explicitly
2592       declared visibility.
2593       The visibility attribute for a class must come between the
2594       class-key and the name, not after the closing brace.
2595       Attributes are now allowed for enums and elaborated-type-specifiers
2596       that only declare a type.
2597       Members of the anonymous namespace are now local to a particular
2598       translation unit, along with any other declarations which use them,
2599       though they are still treated as having external linkage for
2600       language semantics.
2601     * The (undocumented) extension which permitted templates with default
2602       arguments to be bound to template template parameters with fewer
2603       parameters has been removed. For example:
2604        template <template <typename> class C>
2605        void f(C<double>) {}
2606
2607        template <typename T, typename U = int>
2608        struct S {};
2609
2610        template void f(S<double>);
2611
2612       is no longer accepted by G++. The reason this code is not accepted
2613       is that S is a template with two parameters; therefore, it cannot
2614       be bound to C which has only one parameter.
2615     * The <?, >?, <?=, and >?= operators, deprecated in previous GCC
2616       releases, have been removed.
2617     * The command-line option -fconst-strings, deprecated in previous GCC
2618       releases, has been removed.
2619     * The configure variable enable-__cxa_atexit is now enabled by
2620       default for more targets. Enabling this variable is necessary in
2621       order for static destructors to be executed in the correct order,
2622       but it depends upon the presence of a non-standard C library in the
2623       target library in order to work. The variable is now enabled for
2624       more targets which are known to have suitable C libraries.
2625     * -Wextra will produce warnings for if statements with a semicolon as
2626       the only body, to catch code like:
2627         if (a);
2628            return 1;
2629         return 0;
2630
2631       To suppress the warning in valid cases, use { } instead.
2632     * The C++ frontend now also produces strict aliasing warnings when
2633       -fstrict-aliasing -Wstrict-aliasing is in effect.
2634
2635    Runtime Library (libstdc++)
2636
2637     * Added support for TR1 <random>, <complex>, and C compatibility
2638       headers. In addition, a lock-free version of shared_ptr was
2639       contributed as part of Phillip Jordan's Google Summer of Code
2640       project on lock-free containers. ([3]Implementation status of TR1)
2641     * In association with the Summer of Code work on lock-free
2642       containers, the interface for atomic builtins was adjusted,
2643       creating simpler alternatives for non-threaded code paths. Also,
2644       usage was consolidated and all elements were moved from namespace
2645       std to namespace__gnu_cxx. Affected interfaces are the functions
2646       __exchange_and_add, __atomic_add, and the objects __mutex,
2647       __recursive_mutex, and __scoped_lock.
2648     * Support for versioning weak symbol names via namespace association
2649       was added. However, as this changes the names of exported symbols,
2650       this is turned off by default in the current ABI. Intrepid users
2651       can enable this feature by using
2652       --enable-symvers=gnu-versioned-namespace during configuration.
2653     * Revised, simplified, and expanded policy-based associative
2654       containers, including data types for tree and trie forms
2655       (basic_tree, tree, trie), lists (list_update), and both
2656       collision-chaining and probing hash-based containers
2657       (basic_hash_table, cc_hash_table, gp_hash_table). More details per
2658       the [4]documentation.
2659     * The implementation of the debug mode was modified, whereby the
2660       debug namespaces were nested inside of namespace std and namespace
2661       __gnu_cxx in order to resolve some long standing corner cases
2662       involving name lookup. Debug functionality from the policy-based
2663       data structures was consolidated and enabled with the single macro,
2664       _GLIBCXX_DEBUG. See PR 26142 for more information.
2665     * Added extensions for type traits: __conditional_type,
2666       __numeric_traits, __add_unsigned, __removed_unsigned, __enable_if.
2667     * Added a typelist implementation for compile-time meta-programming.
2668       Elements for typelist construction and operation can be found
2669       within namespace __gnu_cxx::typelist.
2670     * Added a new allocator, __gnu_cxx::throw_allocator, for testing
2671       exception-safety.
2672     * Enabled library-wide visibility control, allowing -fvisibility to
2673       be used.
2674     * Consolidated all nested namespaces and the conversion of
2675       __gnu_internal implementation-private details to anonymous
2676       namespaces whenever possible.
2677     * Implemented LWG resolutions DR 431 and DR 538.
2678
2679  Fortran
2680
2681     * Support for allocatable components has been added (TR 15581 and
2682       Fortran 2003).
2683     * Support for the Fortran 2003 streaming IO extension has been added.
2684     * The GNU Fortran compiler now uses 4-byte record markers by default
2685       for unformatted files to be compatible with g77 and most other
2686       compilers. The implementation allows for records greater than 2 GB
2687       and is compatible with several other compilers. Older versions of
2688       gfortran used 8-byte record markers by default (on most systems).
2689       In order to change the length of the record markers, e.g. to read
2690       unformatted files created by older gfortran versions, the
2691       [5]-frecord-marker=8 option can be used.
2692
2693  Java (GCJ)
2694
2695     * A new command-line option -static-libgcj has been added for targets
2696       that use a linker compatible with GNU Binutils. As its name
2697       implies, this causes libgcj to be linked statically. In some cases
2698       this causes the resulting executable to start faster and use less
2699       memory than if the shared version of libgcj were used. However
2700       caution should be used as it can also cause essential parts of the
2701       library to be omitted. Some of these issues are discussed in:
2702       [6]http://gcc.gnu.org/wiki/Statically_linking_libgcj
2703     * fastjar is no longer bundled with GCC. To build libgcj, you will
2704       need either InfoZIP (both zip and unzip) or an external jar
2705       program. In the former case, the GCC build will install a jar shell
2706       script that is based on InfoZIP and provides the same functionality
2707       as fastjar.
2708
2709New Targets and Target Specific Improvements
2710
2711  IA-32/x86-64
2712
2713     * -mtune=generic can now be used to generate code running well on
2714       common x86 chips. This includes AMD Athlon, AMD Opteron, Intel
2715       Pentium-M, Intel Pentium 4 and Intel Core 2.
2716     * -mtune=native and -march=native will produce code optimized for the
2717       host architecture as detected using the cpuid instruction.
2718     * Added a new command-line option -fstackrealign and and
2719       __attribute__ ((force_align_arg_pointer)) to realign the stack at
2720       runtime. This allows functions compiled with a vector-aligned stack
2721       to be invoked from legacy objects that keep only word-alignment.
2722
2723  SPARC
2724
2725     * The default CPU setting has been changed from V7 to V9 in 32-bit
2726       mode on Solaris 7 and above. This is already the case in 64-bit
2727       mode. It can be overridden by specifying --with-cpu at configure
2728       time.
2729     * Back-end support of built-in functions for atomic memory access has
2730       been implemented.
2731     * Support for the Sun UltraSPARC T1 (Niagara) processor has been
2732       added.
2733
2734  M32C
2735
2736     * Various bug fixes have made some functions (notably, functions
2737       returning structures) incompatible with previous releases.
2738       Recompiling all libraries is recommended. Note that code quality
2739       has considerably improved since 4.1, making a recompile even more
2740       beneficial.
2741
2742  MIPS
2743
2744     * Added support for the Broadcom SB-1A core.
2745
2746  IA-64
2747
2748     * Added support for IA-64 data and control speculation. By default
2749       speculation is enabled only during second scheduler pass. A number
2750       of machine flags was introduced to control the usage of speculation
2751       for both scheduler passes.
2752
2753  HPPA
2754
2755     * Added Java language support (libffi and libjava) for 32-bit HP-UX
2756       11 target.
2757
2758Obsolete Systems
2759
2760Documentation improvements
2761
2762  PDF Documentation
2763
2764     * A make pdf target has been added to the top-level makefile,
2765       enabling automated production of PDF documentation files.
2766       (Front-ends external to GCC should modify their Make-lang.in file
2767       to add a lang.pdf: target.)
2768
2769Other significant improvements
2770
2771  Build system improvements
2772
2773     * All the components of the compiler are now bootstrapped by default.
2774       This improves the resilience to bugs in the system compiler or
2775       binary compatibility problems, as well as providing better testing
2776       of GCC 4.2 itself. In addition, if you build the compiler from a
2777       combined tree, the assembler, linker, etc. will also be
2778       bootstrapped (i.e. built with themselves).
2779       You can disable this behavior, and go back to the pre-GCC 4.2 set
2780       up, by configuring GCC with --disable-bootstrap.
2781     * The rules that configure follows to find target tools resemble more
2782       closely the locations that the built compiler will search. In
2783       addition, you can use the new configure option --with-target-tools
2784       to specify where to find the target tools used during the build,
2785       without affecting what the built compiler will use.
2786       This can be especially useful when building packages of GCC. For
2787       example, you may want to build GCC with GNU as or ld, even if the
2788       resulting compiler to work with the native assembler and linker. To
2789       do so, you can use --with-target-tools to point to the native
2790       tools.
2791
2792  Incompatible changes to the build system
2793
2794     * Front-ends external to GCC should modify their Make-lang.in file to
2795       replace double-colon rules (e.g. dvi::) with normal rules (like
2796       lang.dvi:). Front-end makefile hooks do not use double-colon rules
2797       anymore.
2798     * Up to GCC 4.1, a popular way to specify the target tools used
2799       during the build was to create directories named gas, binutils,
2800       etc. in the build tree, and create links to the tools from there.
2801       This does not work any more when the compiler is bootstrapped. The
2802       new configure option --with-target-tools provides a better way to
2803       achieve the same effect, and works for all native and cross
2804       settings.
2805
2806
2807    For questions related to the use of GCC, please consult these web
2808    pages and the [7]GCC manuals. If that fails, the
2809    [8]gcc-help@gcc.gnu.org mailing list might help. Comments on these
2810    web pages and the development of GCC are welcome on our developer
2811    list at [9]gcc@gcc.gnu.org. All of [10]our lists have public
2812    archives.
2813
2814   Copyright (C) [11]Free Software Foundation, Inc. Verbatim copying and
2815   distribution of this entire article is permitted in any medium,
2816   provided this notice is preserved.
2817
2818   These pages are [12]maintained by the GCC team. Last modified
2819   2011-04-25[13].
2820
2821References
2822
2823   1. http://gcc.gnu.org/projects/gomp/
2824   2. http://gcc.gnu.org/onlinedocs/gcc/Warning-Options.html
2825   3. http://gcc.gnu.org/onlinedocs/libstdc++/manual/bk01pt01ch01.html#manual.intro.status.standard.tr1
2826   4. http://gcc.gnu.org/onlinedocs/libstdc++/ext/pb_ds/index.html
2827   5. http://gcc.gnu.org/onlinedocs/gfortran/Runtime-Options.html
2828   6. http://gcc.gnu.org/wiki/Statically_linking_libgcj
2829   7. http://gcc.gnu.org/onlinedocs/
2830   8. mailto:gcc-help@gcc.gnu.org
2831   9. mailto:gcc@gcc.gnu.org
2832  10. http://gcc.gnu.org/lists.html
2833  11. http://www.fsf.org/
2834  12. http://gcc.gnu.org/about.html
2835  13. http://validator.w3.org/check/referer
2836======================================================================
2837http://gcc.gnu.org/gcc-4.1/index.html
2838
2839                             GCC 4.1 Release Series
2840
2841   February 13, 2007
2842
2843   The [1]GNU project and the GCC developers are pleased to announce the
2844   release of GCC 4.1.2.
2845
2846   This release is a bug-fix release, containing fixes for regressions in
2847   GCC 4.1.1 relative to previous releases of GCC.
2848
2849Release History
2850
2851   GCC 4.1.2
2852          February 13, 2007 ([2]changes)
2853
2854   GCC 4.1.1
2855          May 24, 2006 ([3]changes)
2856
2857   GCC 4.1.0
2858          February 28, 2006 ([4]changes)
2859
2860References and Acknowledgements
2861
2862   GCC used to stand for the GNU C Compiler, but since the compiler
2863   supports several other languages aside from C, it now stands for the
2864   GNU Compiler Collection.
2865
2866   A list of [5]successful builds is updated as new information becomes
2867   available.
2868
2869   The GCC developers would like to thank the numerous people that have
2870   contributed new features, improvements, bug fixes, and other changes as
2871   well as test results to GCC. This [6]amazing group of volunteers is
2872   what makes GCC successful.
2873
2874   For additional information about GCC please refer to the [7]GCC project
2875   web site or contact the [8]GCC development mailing list.
2876
2877   To obtain GCC please use [9]our mirror sites or [10]our SVN server.
2878
2879
2880    For questions related to the use of GCC, please consult these web
2881    pages and the [11]GCC manuals. If that fails, the
2882    [12]gcc-help@gcc.gnu.org mailing list might help. Comments on these
2883    web pages and the development of GCC are welcome on our developer
2884    list at [13]gcc@gcc.gnu.org. All of [14]our lists have public
2885    archives.
2886
2887   Copyright (C) [15]Free Software Foundation, Inc. Verbatim copying and
2888   distribution of this entire article is permitted in any medium,
2889   provided this notice is preserved.
2890
2891   These pages are [16]maintained by the GCC team. Last modified
2892   2011-04-25[17].
2893
2894References
2895
2896   1. http://www.gnu.org/
2897   2. http://gcc.gnu.org/gcc-4.1/changes.html#4.1.2
2898   3. http://gcc.gnu.org/gcc-4.1/changes.html
2899   4. http://gcc.gnu.org/gcc-4.1/changes.html
2900   5. http://gcc.gnu.org/gcc-4.1/buildstat.html
2901   6. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
2902   7. http://gcc.gnu.org/index.html
2903   8. mailto:gcc@gcc.gnu.org
2904   9. http://gcc.gnu.org/mirrors.html
2905  10. http://gcc.gnu.org/svn.html
2906  11. http://gcc.gnu.org/onlinedocs/
2907  12. mailto:gcc-help@gcc.gnu.org
2908  13. mailto:gcc@gcc.gnu.org
2909  14. http://gcc.gnu.org/lists.html
2910  15. http://www.fsf.org/
2911  16. http://gcc.gnu.org/about.html
2912  17. http://validator.w3.org/check/referer
2913======================================================================
2914http://gcc.gnu.org/gcc-4.1/changes.html
2915
2916                             GCC 4.1 Release Series
2917                        Changes, New Features, and Fixes
2918
2919   The latest release in the 4.1 release series is [1]GCC 4.1.2.
2920
2921Caveats
2922
2923General Optimizer Improvements
2924
2925     * GCC now has infrastructure for inter-procedural optimizations and
2926       the following inter-procedural optimizations are implemented:
2927          + Profile guided inlining. When doing profile feedback guided
2928            optimization, GCC can now use the profile to make better
2929            informed decisions on whether inlining of a function is
2930            profitable or not. This means that GCC will no longer inline
2931            functions at call sites that are not executed very often, and
2932            that functions at hot call sites are more likely to be
2933            inlined.
2934            A new parameter min-inline-recursive-probability is also now
2935            available to throttle recursive inlining of functions with
2936            small average recursive depths.
2937          + Discovery of pure and const functions, a form of side-effects
2938            analysis. While older GCC releases could also discover such
2939            special functions, the new IPA-based pass runs earlier so that
2940            the results are available to more optimizers. The pass is also
2941            simply more powerful than the old one.
2942          + Analysis of references to static variables and type escape
2943            analysis, also forms of side-effects analysis. The results of
2944            these passes allow the compiler to be less conservative about
2945            call-clobbered variables and references. This results in more
2946            redundant loads being eliminated and in making static
2947            variables candidates for register promotion.
2948          + Improvement of RTL-based alias analysis. The results of type
2949            escape analysis are fed to the RTL type-based alias analyzer,
2950            allowing it to disambiguate more memory references.
2951          + Interprocedural constant propagation and function versioning.
2952            This pass looks for functions that are always called with the
2953            same constant value for one or more of the function arguments,
2954            and propagates those constants into those functions.
2955          + GCC will now eliminate static variables whose usage was
2956            optimized out.
2957          + -fwhole-program --combine can now be used to make all
2958            functions in program static allowing whole program
2959            optimization. As an exception, the main function and all
2960            functions marked with the new externally_visible attribute are
2961            kept global so that programs can link with runtime libraries.
2962     * GCC can now do a form of partial dead code elimination (PDCE) that
2963       allows code motion of expressions to the paths where the result of
2964       the expression is actually needed. This is not always a win, so the
2965       pass has been limited to only consider profitable cases. Here is an
2966       example:
2967    int foo (int *, int *);
2968    int
2969    bar (int d)
2970    {
2971      int a, b, c;
2972      b = d + 1;
2973      c = d + 2;
2974      a = b + c;
2975      if (d)
2976        {
2977          foo (&b, &c);
2978          a = b + c;
2979        }
2980      printf ("%d\n", a);
2981    }
2982
2983       The a = b + c can be sunk to right before the printf. Normal code
2984       sinking will not do this, it will sink the first one above into the
2985       else-branch of the conditional jump, which still gives you two
2986       copies of the code.
2987     * GCC now has a value range propagation pass. This allows the
2988       compiler to eliminate bounds checks and branches. The results of
2989       the pass can also be used to accurately compute branch
2990       probabilities.
2991     * The pass to convert PHI nodes to straight-line code (a form of
2992       if-conversion for GIMPLE) has been improved significantly. The two
2993       most significant improvements are an improved algorithm to
2994       determine the order in which the PHI nodes are considered, and an
2995       improvement that allow the pass to consider if-conversions of basic
2996       blocks with more than two predecessors.
2997     * Alias analysis improvements. GCC can now differentiate between
2998       different fields of structures in Tree-SSA's virtual operands form.
2999       This lets stores/loads from non-overlapping structure fields not
3000       conflict. A new algorithm to compute points-to sets was contributed
3001       that can allows GCC to see now that p->a and p->b, where p is a
3002       pointer to a structure, can never point to the same field.
3003     * Various enhancements to auto-vectorization:
3004          + Incrementally preserve SSA form when vectorizing.
3005          + Incrementally preserve loop-closed form when vectorizing.
3006          + Improvements to peeling for alignment: generate better code
3007            when the misalignment of an access is known at compile time,
3008            or when different accesses are known to have the same
3009            misalignment, even if the misalignment amount itself is
3010            unknown.
3011          + Consider dependence distance in the vectorizer.
3012          + Externalize generic parts of data reference analysis to make
3013            this analysis available to other passes.
3014          + Vectorization of conditional code.
3015          + Reduction support.
3016     * GCC can now partition functions in sections of hot and cold code.
3017       This can significantly improve performance due to better
3018       instruction cache locality. This feature works best together with
3019       profile feedback driven optimization.
3020     * A new pass to avoid saving of unneeded arguments to the stack in
3021       vararg functions if the compiler can prove that they will not be
3022       needed.
3023     * Transition of basic block profiling to tree level implementation
3024       has been completed. The new implementation should be considerably
3025       more reliable (hopefully avoiding profile mismatch errors when
3026       using -fprofile-use or -fbranch-probabilities) and can be used to
3027       drive higher level optimizations, such as inlining.
3028       The -ftree-based-profiling command-line option was removed and
3029       -fprofile-use now implies disabling old RTL level loop optimizer
3030       (-fno-loop-optimize). Speculative prefetching optimization
3031       (originally enabled by -fspeculative-prefetching) was removed.
3032
3033New Languages and Language specific improvements
3034
3035  C and Objective-C
3036
3037     * The old Bison-based C and Objective-C parser has been replaced by a
3038       new, faster hand-written recursive-descent parser.
3039
3040  Ada
3041
3042     * The build infrastructure for the Ada runtime library and tools has
3043       been changed to be better integrated with the rest of the build
3044       infrastructure of GCC. This should make doing cross builds of Ada a
3045       bit easier.
3046
3047  C++
3048
3049     * ARM-style name-injection of friend declarations is no longer the
3050       default. For example:
3051          struct S {
3052            friend void f();
3053          };
3054
3055          void g() { f(); }
3056       will not be accepted; instead a declaration of f will need to be
3057       present outside of the scope of S. The new -ffriend-injection
3058       option will enable the old behavior.
3059     * The (undocumented) extension which permitted templates with default
3060       arguments to be bound to template template parameters with fewer
3061       parameters has been deprecated, and will be removed in the next
3062       major release of G++. For example:
3063       template <template <typename> class C>
3064       void f(C<double>) {}
3065
3066       template <typename T, typename U = int>
3067       struct S {};
3068
3069       template void f(S<double>);
3070
3071       makes use of the deprecated extension. The reason this code is not
3072       valid ISO C++ is that S is a template with two parameters;
3073       therefore, it cannot be bound to C which has only one parameter.
3074
3075    Runtime Library (libstdc++)
3076
3077     * Optimization work:
3078          + A new implementation of std::search_n is provided, better
3079            performing in case of random access iterators.
3080          + Added further efficient specializations of istream functions,
3081            i.e., character array and string extractors.
3082          + Other smaller improvements throughout.
3083     * Policy-based associative containers, designed for high-performance,
3084       flexibility and semantic safety are delivered in ext/pb_assoc.
3085     * A versatile string class, __gnu_cxx::__versa_string, providing
3086       facilities conforming to the standard requirements for
3087       basic_string, is delivered in <ext/vstring.h>. In particular:
3088          + Two base classes are provided: the default one avoids
3089            reference counting and is optimized for short strings; the
3090            alternate one, still uses it while improving in a few low
3091            level areas (e.g., alignment). See vstring_fwd.h for some
3092            useful typedefs.
3093          + Various algorithms have been rewritten (e.g., replace), the
3094            code streamlined and simple optimizations added.
3095          + Option 3 of DR 431 is implemented for both available bases,
3096            thus improving the support for stateful allocators.
3097     * As usual, many bugs have been fixed (e.g., libstdc++/13583,
3098       libstdc++/23953) and LWG resolutions put into effect for the first
3099       time (e.g., DR 280, DR 464, N1780 recommendations for DR 233, TR1
3100       Issue 6.19). The implementation status of TR1 is now tracked in the
3101       docs in tr1.html.
3102
3103  Objective-C++
3104
3105     * A new language front end for Objective-C++ has been added. This
3106       language allows users to mix the object oriented features of
3107       Objective-C with those of C++.
3108
3109  Java (GCJ)
3110
3111     * Core library (libgcj) updates based on GNU Classpath 0.15 - 0.19
3112       features (plus some 0.20 bug-fixes)
3113          + Networking
3114               o The java.net.HttpURLConnection implementation no longer
3115                 buffers the entire response body in memory. This means
3116                 that response bodies larger than available memory can now
3117                 be handled.
3118          + (N)IO
3119               o NIO FileChannel.map implementation, fast bulk put
3120                 implementation for DirectByteBuffer (speeds up this
3121                 method 10x).
3122               o FileChannel.lock() and FileChannel.force() implemented.
3123          + XML
3124               o gnu.xml fix for nodes created outside a namespace
3125                 context.
3126               o Add support for output indenting and
3127                 cdata-section-elements output instruction in
3128                 xml.transform.
3129               o xml.xpath corrections for cases where elements/attributes
3130                 might have been created in non-namespace-aware mode.
3131                 Corrections to handling of XSL variables and minor
3132                 conformance updates.
3133          + AWT
3134               o GNU JAWT implementation, the AWT Native Interface, which
3135                 allows direct access to native screen resources from
3136                 within a Canvas's paint method. GNU Classpath Examples
3137                 comes with a Demo, see libjava/classpath/examples/README.
3138               o awt.datatransfer updated to 1.5 with support for
3139                 FlavorEvents. The gtk+ awt peers now allow copy/paste of
3140                 text, images, URIs/files and serialized objects with
3141                 other applications and tracking clipboard change events
3142                 with gtk+ 2.6 (for gtk+ 2.4 only text and serialized
3143                 objects are supported). A GNU Classpath Examples
3144                 datatransfer Demo was added to show the new
3145                 functionality.
3146               o Split gtk+ awt peers event handling in two threads and
3147                 improve gdk lock handling (solves several awt lock ups).
3148               o Speed up awt Image loading.
3149               o Better gtk+ scrollbar peer implementation when using gtk+
3150                 >= 2.6.
3151               o Handle image loading errors correctly for gdkpixbuf and
3152                 MediaTracker.
3153               o Better handle GDK lock. Properly prefix gtkpeer native
3154                 functions (cp_gtk).
3155               o GdkGraphics2D has been updated to use Cairo 0.5.x or
3156                 higher.
3157               o BufferedImage and GtkImage rewrites. All image drawing
3158                 operations should now work correctly (flipping requires
3159                 gtk+ >= 2.6)
3160               o Future Graphics2D, image and text work is documented at:
3161                 [2]http://developer.classpath.org/mediation/ClasspathGrap
3162                 hicsImagesText
3163               o When gtk+ 2.6 or higher is installed the default log
3164                 handler will produce stack traces whenever a WARNING,
3165                 CRITICAL or ERROR message is produced.
3166          + Free Swing
3167               o The RepaintManager has been reworked for more efficient
3168                 painting, especially for large GUIs.
3169               o The layout manager OverlayLayout has been implemented,
3170                 the BoxLayout has been rewritten to make use of the
3171                 SizeRequirements utility class and caching for more
3172                 efficient layout.
3173               o Improved accessibility support.
3174               o Significant progress has been made in the implementation
3175                 of the javax.swing.plaf.metal package, with most UI
3176                 delegates in a working state now. Please test this with
3177                 your own applications and provide feedback that will help
3178                 us to improve this package.
3179               o The GUI demo (gnu.classpath.examples.swing.Demo) has been
3180                 extended to highlight various features in our Free Swing
3181                 implementation. And it includes a look and feel switcher
3182                 for Metal (default), Ocean and GNU themes.
3183               o The javax.swing.plaf.multi package is now implemented.
3184               o Editing and several key actions for JTree and JTable were
3185                 implemented.
3186               o Lots of icons and look and feel improvements for Free
3187                 Swing basic and metal themes were added. Try running the
3188                 GNU Classpath Swing Demo in examples
3189                 (gnu.classpath.examples.swing.Demo) with:
3190                 -Dswing.defaultlaf=javax.swing.plaf.basic.BasicLookAndFee
3191                 l or
3192                 -Dswing.defaultlaf=javax.swing.plaf.metal.MetalLookAndFee
3193                 l
3194               o Start of styled text capabilites for java.swing.text.
3195               o DefaultMutableTreeNode pre-order, post-order, depth-first
3196                 and breadth-first traversal enumerations implemented.
3197               o JInternalFrame colors and titlebar draw properly.
3198               o JTree is working up to par (icons, selection and keyboard
3199                 traversal).
3200               o JMenus were made more compatible in visual and
3201                 programmatic behavior.
3202               o JTable changeSelection and multiple selections
3203                 implemented.
3204               o JButton and JToggleButton change states work properly
3205                 now.
3206               o JFileChooser fixes.
3207               o revalidate() and repaint() fixes which make Free Swing
3208                 much more responsive.
3209               o MetalIconFactory implemented.
3210               o Free Swing Top-Level Compatibility. JFrame, JDialog,
3211                 JApplet, JInternalFrame, and JWindow are now 1.5
3212                 compatible in the sense that you can call add() and
3213                 setLayout() directly on them, which will have the same
3214                 effect as calling getContentPane().add() and
3215                 getContentPane().setLayout().
3216               o The JTree interface has been completed. JTrees now
3217                 recognizes mouse clicks and selections work.
3218               o BoxLayout works properly now.
3219               o Fixed GrayFilter to actually work.
3220               o Metal SplitPane implemented.
3221               o Lots of Free Swing text and editor stuff work now.
3222          + Free RMI and Corba
3223               o Andrew Watson, Vice President and Technical Director of
3224                 the Object Management Group, has officially assigned us
3225                 20 bit Vendor Minor Code Id: 0x47430 ("GC") that will
3226                 mark remote classpath-specific system exceptions.
3227                 Obtaining the VMCID means that GNU Classpath now is a
3228                 recogniseable type of node in a highly interoperable
3229                 CORBA world.
3230               o GNU Classpath now includes the first working draft to
3231                 support the RMI over IIOP protocol. The current
3232                 implementation is capable of remote invocations,
3233                 transferring various Serializables and Externalizables
3234                 via RMI-IIOP protocol. It can flatten graphs and, at
3235                 least for the simple cases, is interoperable with 1.5
3236                 JDKs.
3237               o org.omg.PortableInterceptor and related functionality in
3238                 other packages is now implemented:
3239                    # The sever and client interceptors work as required
3240                      since 1.4.
3241                    # The IOR interceptor works as needed for 1.5.
3242               o The org.omg.DynamicAny package is completed and passes
3243                 the prepared tests.
3244               o The Portable Object Adapter should now support the output
3245                 of the recent IDL to java compilers. These compilers now
3246                 generate servants and not CORBA objects as before, making
3247                 the output depend on the existing POA implementation.
3248                 Completing POA means that such code can already be tried
3249                 to run on Classpath. Our POA is tested for the following
3250                 usager scenarios:
3251                    # POA converts servant to the CORBA object.
3252                    # Servant provides to the CORBA object.
3253                    # POA activates new CORBA object with the given Object
3254                      Id (byte array) that is later accessible for the
3255                      servant.
3256                    # During the first call, the ServantActivator provides
3257                      servant for this and all subsequent calls on the
3258                      current object.
3259                    # During each call, the ServantLocator provides
3260                      servant for this call only.
3261                    # ServantLocator or ServantActivator forwards call to
3262                      another server.
3263                    # POA has a single servant, responsible for all
3264                      objects.
3265                    # POA has a default servant, but some objects are
3266                      explicitly connected to they specific servants.
3267                 The POA is verified using tests from the former
3268                 cost.omg.org.
3269               o The CORBA implementation is now a working prototype that
3270                 should support features up to 1.3 inclusive. We invite
3271                 groups writing CORBA dependent applications to try
3272                 Classpath implementation, reporting any possible bugs.
3273                 The CORBA prototype is interoperable with Sun's
3274                 implementation v 1.4, transferring object references,
3275                 primitive types, narrow and wide strings, arrays,
3276                 structures, trees, abstract interfaces and value types
3277                 (feature of CORBA 2.3) between these two platforms.
3278                 Remote exceptions are transferred and handled correctly.
3279                 The stringified object references (IORs) from various
3280                 sources are parsed as required. The transient (for
3281                 current session) and permanent (till jre restart)
3282                 redirections work. Both Little and Big Endian encoded
3283                 messages are accepted. The implementation is verified
3284                 using tests from the former cost.omg.org. The current
3285                 release includes working examples (see the examples
3286                 directory), demonstrating the client-server
3287                 communication, using either CORBA Request or IDL-based
3288                 stub (usually generated by a IDL to java compiler). These
3289                 examples also show how to use the Classpath CORBA naming
3290                 service. The IDL to java compiler is not yet written, but
3291                 as our library must be compatible, it naturally accepts
3292                 the output of other idlj implementations.
3293          + Misc
3294               o Updated TimeZone data against Olson tzdata2005l.
3295               o Make zip and jar packages UTF-8 clean.
3296               o "native" code builds and compiles (warning free) on
3297                 Darwin and Solaris.
3298               o java.util.logging.FileHandler now rotates files.
3299               o Start of a generic JDWP framework in gnu/classpath/jdwp.
3300                 This is unfinished, but feedback (at classpath@gnu.org)
3301                 from runtime hackers is greatly appreciated. Although
3302                 most of the work is currently being done around gcj/gij
3303                 we want this framework to be as VM neutral as possible.
3304                 Early design is described in:
3305                 [3]http://gcc.gnu.org/ml/java/2005-05/msg00260.html
3306               o QT4 AWT peers, enable by giving configure
3307                 --enable-qt-peer. Included, but not ready for production
3308                 yet. They are explicitly disabled and not supported. But
3309                 if you want to help with the development of these new
3310                 features we are interested in feedback. You will have to
3311                 explicitly enable them to try them out (and they will
3312                 most likely contain bugs).
3313               o Documentation fixes all over the place. See
3314                 [4]http://developer.classpath.org/doc/
3315
3316New Targets and Target Specific Improvements
3317
3318  IA-32/x86-64
3319
3320     * The x86-64 medium model (that allows building applications whose
3321       data segment exceeds 4GB) was redesigned to match latest ABI draft.
3322       New implementation split large datastructures into separate segment
3323       improving performance of accesses to small datastructures and also
3324       allows linking of small model libraries into medium model programs
3325       as long as the libraries are not accessing the large datastructures
3326       directly. Medium model is also supported in position independent
3327       code now.
3328       The ABI change results in partial incompatibility among medium
3329       model objects. Linking medium model libraries (or objects) compiled
3330       with new compiler into medium model program compiled with older
3331       will likely result in exceeding ranges of relocations.
3332       Binutils 2.16.91 or newer are required for compiling medium model
3333       now.
3334
3335  RS6000 (POWER/PowerPC)
3336
3337     * The AltiVec vector primitives in <altivec.h> are now implemented in
3338       a way that puts a smaller burden on the preprocessor, instead
3339       processing the "overloading" in the front ends. This should benefit
3340       compilation speed on AltiVec vector code.
3341     * AltiVec initializers now are generated more efficiently.
3342     * The popcountb instruction available on POWER5 now is generated.
3343     * The floating point round to integer instructions available on
3344       POWER5+ now is generated.
3345     * Floating point divides can be synthesized using the floating point
3346       reciprocal estimate instructions.
3347     * Double precision floating point constants are initialized as single
3348       precision values if they can be represented exactly.
3349
3350  S/390, zSeries and System z9
3351
3352     * Support for the IBM System z9 109 processor has been added. When
3353       using the -march=z9-109 option, the compiler will generate code
3354       making use of instructions provided by the extended immediate
3355       facility.
3356     * Support for 128-bit IEEE floating point has been added. When using
3357       the -mlong-double-128 option, the compiler will map the long double
3358       data type to 128-bit IEEE floating point. Using this option
3359       constitutes an ABI change, and requires glibc support.
3360     * Various changes to improve performance of generated code have been
3361       implemented, including:
3362          + In functions that do not require a literal pool, register %r13
3363            (which is traditionally reserved as literal pool pointer), can
3364            now be freely used for other purposes by the compiler.
3365          + More precise tracking of register use allows the compiler to
3366            generate more efficient function prolog and epilog code in
3367            certain cases.
3368          + The SEARCH STRING, COMPARE LOGICAL STRING, and MOVE STRING
3369            instructions are now used to implement C string functions.
3370          + The MOVE CHARACTER instruction with single byte overlap is now
3371            used to implement the memset function with non-zero fill byte.
3372          + The LOAD ZERO instructions are now used where appropriate.
3373          + The INSERT CHARACTERS UNDER MASK, STORE CHARACTERS UNDER MASK,
3374            and INSERT IMMEDIATE instructions are now used more frequently
3375            to optimize bitfield operations.
3376          + The BRANCH ON COUNT instruction is now used more frequently.
3377            In particular, the fact that a loop contains a subroutine call
3378            no longer prevents the compiler from using this instruction.
3379          + The compiler is now aware that all shift and rotate
3380            instructions implicitly truncate the shift count to six bits.
3381     * Back-end support for the following generic features has been
3382       implemented:
3383          + The full set of [5]built-in functions for atomic memory
3384            access.
3385          + The -fstack-protector feature.
3386          + The optimization pass avoiding unnecessary stores of incoming
3387            argument registers in functions with variable argument list.
3388
3389  SPARC
3390
3391     * The default code model in 64-bit mode has been changed from
3392       Medium/Anywhere to Medium/Middle on Solaris.
3393     * TLS support is disabled by default on Solaris prior to release 10.
3394       It can be enabled on TLS-capable Solaris 9 versions (4/04 release
3395       and later) by specifying --enable-tls at configure time.
3396
3397  MorphoSys
3398
3399     * Support has been added for this new architecture.
3400
3401Obsolete Systems
3402
3403Documentation improvements
3404
3405Other significant improvements
3406
3407     * GCC can now emit code for protecting applications from
3408       stack-smashing attacks. The protection is realized by buffer
3409       overflow detection and reordering of stack variables to avoid
3410       pointer corruption.
3411     * Some built-in functions have been fortified to protect them against
3412       various buffer overflow (and format string) vulnerabilities.
3413       Compared to the mudflap bounds checking feature, the safe builtins
3414       have far smaller overhead. This means that programs built using
3415       safe builtins should not experience any measurable slowdown.
3416
3417GCC 4.1.2
3418
3419   This is the [6]list of problem reports (PRs) from GCC's bug tracking
3420   system that are known to be fixed in the 4.1.2 release. This list might
3421   not be complete (that is, it is possible that some PRs that have been
3422   fixed are not listed here).
3423
3424   When generating code for a shared library, GCC now recognizes that
3425   global functions may be replaced when the program runs. Therefore, it
3426   is now more conservative in deducing information from the bodies of
3427   functions. For example, in this example:
3428    void f() {}
3429    void g() {
3430     try { f(); }
3431     catch (...) {
3432       cout << "Exception";
3433     }
3434    }
3435
3436   G++ would previously have optimized away the catch clause, since it
3437   would have concluded that f cannot throw exceptions. Because users may
3438   replace f with another function in the main body of the program, this
3439   optimization is unsafe, and is no longer performed. If you wish G++ to
3440   continue to optimize as before, you must add a throw() clause to the
3441   declaration of f to make clear that it does not throw exceptions.
3442
3443
3444    For questions related to the use of GCC, please consult these web
3445    pages and the [7]GCC manuals. If that fails, the
3446    [8]gcc-help@gcc.gnu.org mailing list might help. Comments on these
3447    web pages and the development of GCC are welcome on our developer
3448    list at [9]gcc@gcc.gnu.org. All of [10]our lists have public
3449    archives.
3450
3451   Copyright (C) [11]Free Software Foundation, Inc. Verbatim copying and
3452   distribution of this entire article is permitted in any medium,
3453   provided this notice is preserved.
3454
3455   These pages are [12]maintained by the GCC team. Last modified
3456   2011-04-25[13].
3457
3458References
3459
3460   1. http://gcc.gnu.org/gcc-4.1/changes.html#4.1.2
3461   2. http://developer.classpath.org/mediation/ClasspathGraphicsImagesText
3462   3. http://gcc.gnu.org/ml/java/2005-05/msg00260.html
3463   4. http://developer.classpath.org/doc/
3464   5. http://gcc.gnu.org/onlinedocs/gcc-4.1.0/gcc/Atomic-Builtins.html
3465   6. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.1.2
3466   7. http://gcc.gnu.org/onlinedocs/
3467   8. mailto:gcc-help@gcc.gnu.org
3468   9. mailto:gcc@gcc.gnu.org
3469  10. http://gcc.gnu.org/lists.html
3470  11. http://www.fsf.org/
3471  12. http://gcc.gnu.org/about.html
3472  13. http://validator.w3.org/check/referer
3473======================================================================
3474http://gcc.gnu.org/gcc-4.0/index.html
3475
3476                             GCC 4.0 Release Series
3477
3478   January 31, 2007
3479
3480   The [1]GNU project and the GCC developers are pleased to announce the
3481   release of GCC 4.0.4.
3482
3483   This release is a bug-fix release, containing fixes for regressions in
3484   GCC 4.0.3 relative to previous releases of GCC.
3485
3486Release History
3487
3488   GCC 4.0.4
3489          January 31, 2007 ([2]changes)
3490
3491   GCC 4.0.3
3492          March 10, 2006 ([3]changes)
3493
3494   GCC 4.0.2
3495          September 28, 2005 ([4]changes)
3496
3497   GCC 4.0.1
3498          July 7, 2005 ([5]changes)
3499
3500   GCC 4.0.0
3501          April 20, 2005 ([6]changes)
3502
3503References and Acknowledgements
3504
3505   GCC used to stand for the GNU C Compiler, but since the compiler
3506   supports several other languages aside from C, it now stands for the
3507   GNU Compiler Collection.
3508
3509   A list of [7]successful builds is updated as new information becomes
3510   available.
3511
3512   The GCC developers would like to thank the numerous people that have
3513   contributed new features, improvements, bug fixes, and other changes as
3514   well as test results to GCC. This [8]amazing group of volunteers is
3515   what makes GCC successful.
3516
3517   For additional information about GCC please refer to the [9]GCC project
3518   web site or contact the [10]GCC development mailing list.
3519
3520   To obtain GCC please use [11]our mirror sites, or [12]our SVN server.
3521
3522
3523    For questions related to the use of GCC, please consult these web
3524    pages and the [13]GCC manuals. If that fails, the
3525    [14]gcc-help@gcc.gnu.org mailing list might help. Comments on these
3526    web pages and the development of GCC are welcome on our developer
3527    list at [15]gcc@gcc.gnu.org. All of [16]our lists have public
3528    archives.
3529
3530   Copyright (C) [17]Free Software Foundation, Inc. Verbatim copying and
3531   distribution of this entire article is permitted in any medium,
3532   provided this notice is preserved.
3533
3534   These pages are [18]maintained by the GCC team. Last modified
3535   2011-04-25[19].
3536
3537References
3538
3539   1. http://www.gnu.org/
3540   2. http://gcc.gnu.org/gcc-4.0/changes.html#4.0.4
3541   3. http://gcc.gnu.org/gcc-4.0/changes.html#4.0.3
3542   4. http://gcc.gnu.org/gcc-4.0/changes.html#4.0.2
3543   5. http://gcc.gnu.org/gcc-4.0/changes.html#4.0.1
3544   6. http://gcc.gnu.org/gcc-4.0/changes.html
3545   7. http://gcc.gnu.org/gcc-4.0/buildstat.html
3546   8. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
3547   9. http://gcc.gnu.org/index.html
3548  10. mailto:gcc@gcc.gnu.org
3549  11. http://gcc.gnu.org/mirrors.html
3550  12. http://gcc.gnu.org/svn.html
3551  13. http://gcc.gnu.org/onlinedocs/
3552  14. mailto:gcc-help@gcc.gnu.org
3553  15. mailto:gcc@gcc.gnu.org
3554  16. http://gcc.gnu.org/lists.html
3555  17. http://www.fsf.org/
3556  18. http://gcc.gnu.org/about.html
3557  19. http://validator.w3.org/check/referer
3558======================================================================
3559http://gcc.gnu.org/gcc-4.0/changes.html
3560
3561                             GCC 4.0 Release Series
3562                        Changes, New Features, and Fixes
3563
3564   The latest release in the 4.0 release series is [1]GCC 4.0.4.
3565
3566Caveats
3567
3568     * GCC now generates location lists by default when compiling with
3569       debug info and optimization.
3570          + GDB 6.0 and older crashes when it sees location lists. GDB 6.1
3571            or later is needed to debug binaries containing location
3572            lists.
3573          + When you are trying to view a value of a variable in a part of
3574            a function where it has no location (for example when the
3575            variable is no longer used and thus its location was used for
3576            something else) GDB will say that it is not available.
3577       You can disable generating location lists by -fno-var-tracking.
3578     * GCC no longer accepts the -fwritable-strings option. Use named
3579       character arrays when you need a writable string.
3580     * The options -freduce-all-givs and -fmove-all-movables have been
3581       discontinued. They were used to circumvent a shortcoming in the
3582       heuristics of the old loop optimization code with respect to common
3583       Fortran constructs. The new (tree) loop optimizer works differently
3584       and doesn't need those work-arounds.
3585     * The graph-coloring register allocator, formerly enabled by the
3586       option -fnew-ra, has been discontinued.
3587     * -I- has been deprecated. -iquote is meant to replace the need for
3588       this option.
3589     * The MIPS -membedded-pic and -mrnames options have been removed.
3590     * All MIPS targets now require the GNU assembler. In particular, IRIX
3591       configurations can no longer use the MIPSpro assemblers, although
3592       they do still support the MIPSpro linkers.
3593     * The SPARC option -mflat has been removed.
3594     * English-language diagnostic messages will now use Unicode quotation
3595       marks in UTF-8 locales. (Non-English messages already used the
3596       quotes appropriate for the language in previous releases.) If your
3597       terminal does not support UTF-8 but you are using a UTF-8 locale
3598       (such locales are the default on many GNU/Linux systems) then you
3599       should set LC_CTYPE=C in the environment to disable that locale.
3600       Programs that parse diagnostics and expect plain ASCII
3601       English-language messages should set LC_ALL=C. See [2]Markus Kuhn's
3602       explanation of Unicode quotation marks for more information.
3603     * The specs file is no longer installed on most platforms. Most users
3604       will be totally unaffected. However, if you are accustomed to
3605       editing the specs file yourself, you will now have to use the
3606       -dumpspecs option to generate the specs file, and then edit the
3607       resulting file.
3608
3609General Optimizer Improvements
3610
3611     * The [3]tree ssa branch has been merged. This merge has brought in a
3612       completely new optimization framework based on a higher level
3613       intermediate representation than the existing RTL representation.
3614       Numerous new code transformations based on the new framework are
3615       available in GCC 4.0, including:
3616          + Scalar replacement of aggregates
3617          + Constant propagation
3618          + Value range propagation
3619          + Partial redundancy elimination
3620          + Load and store motion
3621          + Strength reduction
3622          + Dead store elimination
3623          + Dead and unreachable code elimination
3624          + [4]Autovectorization
3625          + Loop interchange
3626          + Tail recursion by accumulation
3627       Many of these passes outperform their counterparts from previous
3628       GCC releases.
3629     * [5]Swing Modulo Scheduling (SMS). An RTL level instruction
3630       scheduling optimization intended for loops that perform heavy
3631       computations.
3632
3633New Languages and Language specific improvements
3634
3635  C family
3636
3637     * The sentinel attribute has been added to GCC. This function
3638       attribute allows GCC to warn when variadic functions such as execl
3639       are not NULL terminated. See the GCC manual for a complete
3640       description of its behavior.
3641     * Given __attribute__((alias("target"))) it is now an error if target
3642       is not a symbol, defined in the same translation unit. This also
3643       applies to aliases created by #pragma weak alias=target. This is
3644       because it's meaningless to define an alias to an undefined symbol.
3645       On Solaris, the native assembler would have caught this error, but
3646       GNU as does not.
3647
3648  C and Objective-C
3649
3650     * The -Wstrict-aliasing=2 option has been added. This warning catches
3651       all unsafe cases, but it may also give a warning for some cases
3652       that are safe.
3653     * The cast-as-lvalue, conditional-expression-as-lvalue and
3654       compound-expression-as-lvalue extensions, which were deprecated in
3655       3.3.4 and 3.4, have been removed.
3656     * The -fwritable-strings option, which was deprecated in 3.4, has
3657       been removed.
3658     * #pragma pack() semantics have been brought closer to those used by
3659       other compilers. This also applies to C++.
3660     * Taking the address of a variable with register storage is invalid
3661       in C. GCC now issues an error instead of a warning.
3662     * Arrays of incomplete element type are invalid in C. GCC now issues
3663       an error for such arrays. Declarations such as extern struct s x[];
3664       (where struct s has not been defined) can be moved after the
3665       definition of struct s. Function parameters declared as arrays of
3666       incomplete type can instead be declared as pointers.
3667
3668  C++
3669
3670     * When compiling without optimizations (-O0), the C++ frontend is
3671       much faster than in any previous versions of GCC. Independent
3672       testers have measured speed-ups up to 25% in real-world production
3673       code, compared to the 3.4 family (which was already the fastest
3674       version to date). Upgrading from older versions might show even
3675       bigger improvements.
3676     * ELF visibility attributes can now be applied to a class type, so
3677       that it affects every member function of a class at once, without
3678       having to specify each individually:
3679class __attribute__ ((visibility("hidden"))) Foo
3680{
3681   int foo1();
3682   void foo2();
3683};
3684       The syntax is deliberately similar to the __declspec() system used
3685       by Microsoft Windows based compilers, allowing cross-platform
3686       projects to easily reuse their existing macro system for denoting
3687       exports and imports. By explicitly marking internal classes never
3688       used outside a binary as hidden, one can completely avoid PLT
3689       indirection overheads during their usage by the compiler. You can
3690       find out more about the advantages of this at
3691       [6]http://www.akkadia.org/drepper/dsohowto.pdf
3692     * The -fvisibility-inlines-hidden option has been added which marks
3693       all inlineable functions as having hidden ELF visibility, thus
3694       removing their symbol and typeinfo from the exported symbol table
3695       of the output ELF binary. Using this option can reduce the exported
3696       symbol count of template-heavy code by up to 40% with no code
3697       change at all, thus notably improving link and load times for the
3698       binary as well as a reduction in size of up to 10%. Also, check the
3699       new [7]-fvisibility option.
3700     * The compiler now uses the library interface specified by the [8]C++
3701       ABI for thread-safe initialization of function-scope static
3702       variables. Most users should leave this alone, but embedded
3703       programmers may want to disable this by specifying
3704       -fno-threadsafe-statics for a small savings in code size.
3705     * Taking the address of an explicit register variable is no longer
3706       supported. Note that C++ allows taking the address of variables
3707       with register storage so this will continue to compile with a
3708       warning. For example, assuming that r0 is a machine register:
3709register int foo asm ("r0");
3710register int bar;
3711&foo; // error, no longer accepted
3712&bar; // OK, with a warning
3713     * G++ has an undocumented extension to virtual function covariancy
3714       rules that allowed the overrider to return a type that was
3715       implicitly convertable to the overridden function's return type.
3716       For instance a function returning void * could be overridden by a
3717       function returning T *. This is now deprecated and will be removed
3718       in a future release.
3719     * The G++ minimum and maximum operators (<? and >?) and their
3720       compound forms (<?=) and >?=) have been deprecated and will be
3721       removed in a future version. Code using these operators should be
3722       modified to use std::min and std::max instead.
3723     * Declaration of nested classes of class templates as friends are
3724       supported:
3725template <typename T> struct A {
3726  class B {};
3727};
3728class C {
3729  template <typename T> friend class A<T>::B;
3730};
3731       This complements the feature member functions of class templates as
3732       friends introduced in GCC 3.4.0.
3733     * When declaring a friend class using an unqualified name, classes
3734       outside the innermost non-class scope are not searched:
3735class A;
3736namespace N {
3737  class B {
3738    friend class A;   // Refer to N::A which has not been declared yet
3739                      // because name outside namespace N are not searched
3740    friend class ::A; // Refer to ::A
3741  };
3742}
3743       Hiding the friend name until declaration is still not implemented.
3744     * Friends of classes defined outside their namespace are correctly
3745       handled:
3746namespace N {
3747  class A;
3748}
3749class N::A {
3750  friend class B; // Refer to N::B in GCC 4.0.0
3751                  // but ::B in earlier versions of GCC
3752};
3753
3754    Runtime Library (libstdc++)
3755
3756     * Optimization work:
3757          + Added efficient specializations of istream functions for char
3758            and wchar_t.
3759          + Further performance tuning of strings, in particular wrt
3760            single-char append and getline.
3761          + iter_swap - and therefore most of the mutating algorithms -
3762            now makes an unqualified call to swap when the value_type of
3763            the two iterators is the same.
3764     * A large subset of the features in Technical Report 1 (TR1 for
3765       short) is experimentally delivered (i.e., no guarantees about the
3766       implementation are provided. In particular it is not promised that
3767       the library will remain link-compatible when code using TR1 is
3768       used):
3769          + General utilities such as reference_wrapper and shared_ptr.
3770          + Function objects, i.e., result_of, mem_fn, bind, function.
3771          + Support for metaprogramming.
3772          + New containers such as tuple, array, unordered_set,
3773            unordered_map, unordered_multiset, unordered_multimap.
3774     * As usual, many bugs have been fixed and LWG resolutions implemented
3775       for the first time (e.g., DR 409).
3776
3777  Java
3778
3779     * In order to prevent naming conflicts with other implementations of
3780       these tools, some GCJ binaries have been renamed:
3781          + rmic is now grmic,
3782          + rmiregistry is now grmiregistry, and
3783          + jar is now fastjar.
3784       In particular, these names were problematic for the jpackage.org
3785       packaging conventions which install symlinks in /usr/bin that point
3786       to the preferred versions of these tools.
3787     * The -findirect-dispatch argument to the compiler now works and
3788       generates code following a new "binary compatibility" ABI. Code
3789       compiled this way follows the binary compatibility rules of the
3790       Java Language Specification.
3791     * libgcj now has support for using GCJ as a JIT, using the
3792       gnu.gcj.jit family of system properties.
3793     * libgcj can now find a shared library corresponding to the bytecode
3794       representation of a class. See the documentation for the new
3795       gcj-dbtool program, and the new gnu.gcj.precompiled.db.path system
3796       property.
3797     * There have been many improvements to the class library. Here are
3798       some highlights:
3799          + Much more of AWT and Swing exist.
3800          + Many new packages and classes were added, including
3801            java.util.regex, java.net.URI, javax.crypto,
3802            javax.crypto.interfaces, javax.crypto.spec, javax.net,
3803            javax.net.ssl, javax.security.auth,
3804            javax.security.auth.callback, javax.security.auth.login,
3805            javax.security.auth.x500, javax.security.sasl, org.ietf.jgss,
3806            javax.imageio, javax.imageio.event, javax.imageio.spi,
3807            javax.print, javax.print.attribute,
3808            javax.print.attribute.standard, javax.print.event, and
3809            javax.xml
3810          + Updated SAX and DOM, and imported GNU JAXP
3811
3812  Fortran
3813
3814     * A new [9]Fortran front end has replaced the aging GNU Fortran 77
3815       front end. The new front end supports Fortran 90 and Fortran 95. It
3816       may not yet be as stable as the old Fortran front end.
3817
3818  Ada
3819
3820     * Ada (with tasking and Zero Cost Exceptions) is now available on
3821       many more targets, including but not limited to: alpha-linux,
3822       hppa-hpux, hppa-linux, powerpc-darwin, powerpc-linux, s390-linux,
3823       s390x-linux, sparc-linux.
3824     * Some of the new Ada 2005 features are now implemented like
3825       Wide_Wide_Character and Ada.Containers.
3826     * Many bugs have been fixed, tools and documentation improved.
3827     * To compile Ada from the sources, install an older working Ada
3828       compiler and then use --enable-languages=ada at configuration time,
3829       since the Ada frontend is not currently activated by default. See
3830       the [10]Installing GCC for details.
3831
3832New Targets and Target Specific Improvements
3833
3834  H8/300
3835
3836     * The frame layout has changed. In the new layout, the prologue of a
3837       function first saves registers and then allocate space for locals,
3838       resulting in an 1% improvement on code size.
3839
3840  IA-32/x86-64 (AMD64)
3841
3842     * The acos, asin, drem, exp10, exp2, expm1, fmod, ilogb, log10,
3843       log1p, log2, logb and tan mathematical builtins (and their float
3844       and long double variants) are now implemented as inline x87
3845       intrinsics when using -ffast-math.
3846     * The ceil, floor, nearbyint, rint and trunc mathematical builtins
3847       (and their float and long double variants) are now implemented as
3848       inline x87 intrinsics when using -ffast-math.
3849     * The x87's fsincos instruction is now used automatically with
3850       -ffast-math when calculating both the sin and cos of the same
3851       argument.
3852     * Instruction selection for multiplication and division by constants
3853       has been improved.
3854
3855  IA-64
3856
3857     * Floating point division, integer division and sqrt are now inlined,
3858       resulting in significant performance improvements on some codes.
3859
3860  MIPS
3861
3862     * Division by zero checks now use conditional traps if the target
3863       processor supports them. This decreases code size by one word per
3864       division operation. The old behavior (branch and break) can be
3865       obtained either at configure time by passing --with-divide=breaks
3866       to configure or at runtime by passing -mdivide-breaks to GCC.
3867     * Support for MIPS64 paired-single instructions has been added. It is
3868       enabled by -mpaired-single and can be accessed using both the
3869       target-independent vector extensions and new MIPS-specific built-in
3870       functions.
3871     * Support for the MIPS-3D ASE has been added. It is enabled by
3872       -mips3d and provides new MIPS-3D-specific built-in functions.
3873     * The -mexplicit-relocs option now supports static n64 code (as is
3874       used, for example, in 64-bit linux kernels). -mexplicit-relocs
3875       should now be feature-complete and is enabled by default when GCC
3876       is configured to use a compatible assembler.
3877     * Support for the NEC VR4130 series has been added. This support
3878       includes the use of VR-specific instructions and a new VR4130
3879       scheduler. Full VR4130 support can be selected with -march=vr4130
3880       while code for any ISA can be tuned for the VR4130 using
3881       -mtune=vr4130. There is also a new -mvr4130-align option that
3882       produces better schedules at the cost of increased code size.
3883     * Support for the Broadcom SB-1 has been extended. There is now an
3884       SB-1 scheduler as well as support for the SB-1-specific
3885       paired-single instructions. Full SB-1 support can be selected with
3886       -march=sb1 while code for any ISA can be optimized for the SB-1
3887       using -mtune=sb1.
3888     * The compiler can now work around errata in R4000, R4400, VR4120 and
3889       VR4130 processors. These workarounds are enabled by -mfix-r4000,
3890       -mfix-r4400, -mfix-vr4120 and -mfix-vr4130 respectively. The VR4120
3891       and VR4130 workarounds need binutils 2.16 or above.
3892     * IRIX shared libraries are now installed into the standard library
3893       directories: o32 libraries go into lib/, n32 libraries go into
3894       lib32/ and n64 libraries go into lib64/.
3895     * The compiler supports a new -msym32 option. It can be used to
3896       optimize n64 code in which all symbols are known to have 32-bit
3897       values.
3898
3899  S/390 and zSeries
3900
3901     * New command-line options help to generate code intended to run in
3902       an environment where stack space is restricted, e.g. Linux kernel
3903       code:
3904          + -mwarn-framesize and -mwarn-dynamicstack trigger compile-time
3905            warnings for single functions that require large or dynamic
3906            stack frames.
3907          + -mstack-size and -mstack-guard generate code that checks for
3908            stack overflow at run time.
3909          + -mpacked-stack generates code that reduces the stack frame
3910            size of many functions by reusing unneeded parts of the stack
3911            bias area.
3912     * The -msoft-float option now ensures that generated code never
3913       accesses floating point registers.
3914     * The s390x-ibm-tpf target now fully supports C++, including
3915       exceptions and threads.
3916     * Various changes to improve performance of the generated code have
3917       been implemented, including:
3918          + GCC now uses sibling calls where possible.
3919          + Condition code handling has been optimized, allowing GCC to
3920            omit redundant comparisons in certain cases.
3921          + The cost function guiding many optimizations has been refined
3922            to more accurately represent the z900 and z990 processors.
3923          + The ADD LOGICAL WITH CARRY and SUBTRACT LOGICAL WITH BORROW
3924            instructions are now used to avoid conditional branches in
3925            certain cases.
3926          + The back end now uses the LEGITIMIZE_RELOAD_ADDRESS feature to
3927            optimize address arithmetic required to access large stack
3928            frames.
3929          + GCC now makes more efficient use of memory-to-memory type
3930            instructions (MVC, CLC, ...).
3931          + More precise tracking of special register use allows better
3932            instruction scheduling, in particular of the function prologue
3933            and epilogue sequences.
3934          + The Java front end now generates inline code to implement
3935            integer division, instead of calling library routines.
3936
3937  SPARC
3938
3939     * The options -mv8, -msparclite, -mcypress, -msupersparc, -mf930 and
3940       -mf934 have been removed. They have been replaced with -mcpu=xxx.
3941     * The internal model used to estimate the relative cost of each
3942       instruction has been updated. It is expected to give better results
3943       on recent UltraSPARC processors.
3944     * Code generation for function prologues and epilogues has been
3945       improved, resulting in better scheduling and allowing multiple exit
3946       points in functions.
3947     * Support for Sun's Visual Instruction Set (VIS) has been enhanced.
3948       It is enabled by -mvis and provides new built-in functions for VIS
3949       instructions on UltraSPARC processors.
3950     * The option -mapp-regs has been turned on by default on Solaris too.
3951
3952  NetWare
3953
3954     * Novell NetWare (on ix86, no other hardware platform was ever really
3955       supported by this OS) has been re-enabled and the ABI supported by
3956       GCC has been brought into sync with that of MetroWerks CodeWarrior
3957       (the ABI previously supported was that of some Unix systems, which
3958       NetWare never tried to support).
3959
3960Obsolete Systems
3961
3962   Support for a number of older systems has been declared obsolete in GCC
3963   4.0. Unless there is activity to revive them, the next release of GCC
3964   will have their sources permanently removed.
3965
3966   All GCC ports for the following processor architectures have been
3967   declared obsolete:
3968     * Intel i860
3969     * Ubicom IP2022
3970     * National Semiconductor NS32K
3971     * Texas Instruments TMS320C[34]x
3972
3973   Also, those for some individual systems have been obsoleted:
3974     * SPARC family
3975          + SPARClite-based systems (sparclite-*-coff, sparclite-*-elf,
3976            sparc86x-*-elf)
3977          + OpenBSD 32-bit (sparc-*-openbsd*)
3978
3979Documentation improvements
3980
3981Other significant improvements
3982
3983     * Location lists are now generated by default when compiling with
3984       debug info and optimization. Location lists provide more accurate
3985       debug info about locations of variables and they allow debugging
3986       code compiled with -fomit-frame-pointer.
3987     * The -fvisibility option has been added which allows the default ELF
3988       visibility of all symbols to be set per compilation and the new
3989       #pragma GCC visibility preprocessor command allows the setting of
3990       default ELF visibility for a region of code. Using
3991       -fvisibility=hidden especially in combination with the new
3992       -fvisibility-inlines-hidden can yield substantial improvements in
3993       output binary quality including avoiding PLT indirection overheads,
3994       reduction of the exported symbol count by up to 60% (with resultant
3995       improvements to link and load times), better scope for the
3996       optimizer to improve code and up to a 20% reduction in binary size.
3997       Using these options correctly yields a binary with a similar symbol
3998       count to a Windows DLL.
3999       Perhaps more importantly, this new feature finally allows (with
4000       careful planning) complete avoidance of symbol clashes when
4001       manually loading shared objects with RTLD_GLOBAL, thus finally
4002       solving problems many projects such as python were forced to use
4003       RTLD_LOCAL for (with its resulting issues for C++ correctness). You
4004       can find more information about using these options at
4005       [11]http://gcc.gnu.org/wiki/Visibility.
4006     __________________________________________________________________
4007
4008GCC 4.0.1
4009
4010   This is the [12]list of problem reports (PRs) from GCC's bug tracking
4011   system that are known to be fixed in the 4.0.1 release. This list might
4012   not be complete (that is, it is possible that some PRs that have been
4013   fixed are not listed here).
4014
4015GCC 4.0.2
4016
4017   This is the [13]list of problem reports (PRs) from GCC's bug tracking
4018   system that are known to be fixed in the 4.0.2 release. This list might
4019   not be complete (that is, it is possible that some PRs that have been
4020   fixed are not listed here).
4021
4022   Unfortunately, due to a release engineering failure, this release has a
4023   regression on Solaris that will affect some C++ programs. We suggest
4024   that Solaris users apply a [14]patch that corrects the problem. Users
4025   who do not wish to apply the patch should explicitly link C++ programs
4026   with the -pthreads option, even if they do not use threads. This
4027   problem has been corrected in the current 4.0 branch sources and will
4028   not be present in GCC 4.0.3.
4029
4030GCC 4.0.3
4031
4032   Starting with this release, the function getcontext is recognized by
4033   the compiler as having the same semantics as the setjmp function. In
4034   particular, the compiler will ensure that all registers are dead before
4035   calling such a function and will emit a warning about the variables
4036   that may be clobbered after the second return from the function.
4037
4038GCC 4.0.4
4039
4040   This is the [15]list of problem reports (PRs) from GCC's bug tracking
4041   system that are known to be fixed in the 4.0.4 release. This list might
4042   not be complete (that is, it is possible that some PRs that have been
4043   fixed are not listed here).
4044
4045   The 4.0.4 release is provided for those that require a high degree of
4046   binary compatibility with previous 4.0.x releases. For most users, the
4047   GCC team recommends that version 4.1.1 or later be used instead."
4048
4049
4050    For questions related to the use of GCC, please consult these web
4051    pages and the [16]GCC manuals. If that fails, the
4052    [17]gcc-help@gcc.gnu.org mailing list might help. Comments on these
4053    web pages and the development of GCC are welcome on our developer
4054    list at [18]gcc@gcc.gnu.org. All of [19]our lists have public
4055    archives.
4056
4057   Copyright (C) [20]Free Software Foundation, Inc. Verbatim copying and
4058   distribution of this entire article is permitted in any medium,
4059   provided this notice is preserved.
4060
4061   These pages are [21]maintained by the GCC team. Last modified
4062   2012-02-20[22].
4063
4064References
4065
4066   1. http://gcc.gnu.org/gcc-4.0/changes.html#4.0.4
4067   2. http://www.cl.cam.ac.uk/~mgk25/ucs/quotes.html
4068   3. http://gcc.gnu.org/projects/tree-ssa/
4069   4. http://gcc.gnu.org/projects/tree-ssa/vectorization.html
4070   5. http://gcc.gnu.org/news/sms.html
4071   6. http://www.akkadia.org/drepper/dsohowto.pdf
4072   7. http://gcc.gnu.org/gcc-4.0/changes.html#visibility
4073   8. http://sourcery.mentor.com/public/cxx-abi/
4074   9. http://gcc.gnu.org/fortran/
4075  10. http://gcc.gnu.org/install/
4076  11. http://gcc.gnu.org/wiki/Visibility
4077  12. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.0.1
4078  13. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.0.2
4079  14. http://gcc.gnu.org/ml/gcc-cvs/2005-09/msg00984.html
4080  15. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.0.4
4081  16. http://gcc.gnu.org/onlinedocs/
4082  17. mailto:gcc-help@gcc.gnu.org
4083  18. mailto:gcc@gcc.gnu.org
4084  19. http://gcc.gnu.org/lists.html
4085  20. http://www.fsf.org/
4086  21. http://gcc.gnu.org/about.html
4087  22. http://validator.w3.org/check/referer
4088======================================================================
4089http://gcc.gnu.org/gcc-3.4/index.html
4090
4091                             GCC 3.4 Release Series
4092
4093   May 26, 2006
4094
4095   The [1]GNU project and the GCC developers are pleased to announce the
4096   release of GCC 3.4.6.
4097
4098   This release is a bug-fix release, containing fixes for regressions in
4099   GCC 3.4.4 relative to previous releases of GCC. This is the last of the
4100   3.4.x series.
4101
4102   The GCC 3.4 release series includes numerous [2]new features,
4103   improvements, bug fixes, and other changes, thanks to an [3]amazing
4104   group of volunteers.
4105
4106Release History
4107
4108   GCC 3.4.6
4109          March 6, 2006 ([4]changes)
4110
4111   GCC 3.4.5
4112          November 30, 2005 ([5]changes)
4113
4114   GCC 3.4.4
4115          May 18, 2005 ([6]changes)
4116
4117   GCC 3.4.3
4118          November 4, 2004 ([7]changes)
4119
4120   GCC 3.4.2
4121          September 6, 2004 ([8]changes)
4122
4123   GCC 3.4.1
4124          July 1, 2004 ([9]changes)
4125
4126   GCC 3.4.0
4127          April 18, 2004 ([10]changes)
4128
4129References and Acknowledgements
4130
4131   GCC used to stand for the GNU C Compiler, but since the compiler
4132   supports several other languages aside from C, it now stands for the
4133   GNU Compiler Collection.
4134
4135   A list of [11]successful builds is updated as new information becomes
4136   available.
4137
4138   The GCC developers would like to thank the numerous people that have
4139   contributed new features, improvements, bug fixes, and other changes as
4140   well as test results to GCC. This [12]amazing group of volunteers is
4141   what makes GCC successful.
4142
4143   For additional information about GCC please refer to the [13]GCC
4144   project web site or contact the [14]GCC development mailing list.
4145
4146   To obtain GCC please use [15]our mirror sites, or [16]our SVN server.
4147
4148
4149    For questions related to the use of GCC, please consult these web
4150    pages and the [17]GCC manuals. If that fails, the
4151    [18]gcc-help@gcc.gnu.org mailing list might help. Comments on these
4152    web pages and the development of GCC are welcome on our developer
4153    list at [19]gcc@gcc.gnu.org. All of [20]our lists have public
4154    archives.
4155
4156   Copyright (C) [21]Free Software Foundation, Inc. Verbatim copying and
4157   distribution of this entire article is permitted in any medium,
4158   provided this notice is preserved.
4159
4160   These pages are [22]maintained by the GCC team. Last modified
4161   2011-04-25[23].
4162
4163References
4164
4165   1. http://www.gnu.org/
4166   2. http://gcc.gnu.org/gcc-3.4/changes.html
4167   3. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
4168   4. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.6
4169   5. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.5
4170   6. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.4
4171   7. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.3
4172   8. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.2
4173   9. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.1
4174  10. http://gcc.gnu.org/gcc-3.4/changes.html
4175  11. http://gcc.gnu.org/gcc-3.4/buildstat.html
4176  12. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
4177  13. http://gcc.gnu.org/index.html
4178  14. mailto:gcc@gcc.gnu.org
4179  15. http://gcc.gnu.org/mirrors.html
4180  16. http://gcc.gnu.org/svn.html
4181  17. http://gcc.gnu.org/onlinedocs/
4182  18. mailto:gcc-help@gcc.gnu.org
4183  19. mailto:gcc@gcc.gnu.org
4184  20. http://gcc.gnu.org/lists.html
4185  21. http://www.fsf.org/
4186  22. http://gcc.gnu.org/about.html
4187  23. http://validator.w3.org/check/referer
4188======================================================================
4189http://gcc.gnu.org/gcc-3.4/changes.html
4190
4191                             GCC 3.4 Release Series
4192                        Changes, New Features, and Fixes
4193
4194   The final release in the 3.4 release series is [1]GCC 3.4.6. The series
4195   is now closed.
4196
4197   GCC 3.4 has [2]many improvements in the C++ frontend. Before reporting
4198   a bug, please make sure it's really GCC, and not your code, that is
4199   broken.
4200
4201Caveats
4202
4203     * GNU Make is now required to build GCC.
4204     * With -nostdinc the preprocessor used to ignore both standard
4205       include paths and include paths contained in environment variables.
4206       It was neither documented nor intended that environment variable
4207       paths be ignored, so this has been corrected.
4208     * GCC no longer accepts the options -fvolatile, -fvolatile-global and
4209       -fvolatile-static. It is unlikely that they worked correctly in any
4210       3.x release.
4211     * GCC no longer ships <varargs.h>. Use <stdarg.h> instead.
4212     * Support for all the systems [3]obsoleted in GCC 3.3 has been
4213       removed from GCC 3.4. See below for a [4]list of systems which are
4214       obsoleted in this release.
4215     * GCC now requires an ISO C90 (ANSI C89) C compiler to build. K&R C
4216       compilers will not work.
4217     * The implementation of the [5]MIPS ABIs has changed. As a result,
4218       the code generated for certain MIPS targets will not be binary
4219       compatible with earlier releases.
4220     * In previous releases, the MIPS port had a fake "hilo" register with
4221       the user-visible name accum. This register has been removed.
4222     * The implementation of the [6]SPARC ABIs has changed. As a result,
4223       the code generated will not be binary compatible with earlier
4224       releases in certain cases.
4225     * The configure option --enable-threads=pthreads has been removed;
4226       use --enable-threads=posix instead, which should have the same
4227       effect.
4228     * Code size estimates used by inlining heuristics for C, Objective-C,
4229       C++ and Java have been redesigned significantly. As a result the
4230       parameters of -finline-insns, --param max-inline-insns-single and
4231       --param max-inline-insns-auto need to be reconsidered.
4232     * --param max-inline-slope and --param min-inline-insns have been
4233       removed; they are not needed for the new bottom-up inlining
4234       heuristics.
4235     * The new unit-at-a-time compilation scheme has several compatibility
4236       issues:
4237          + The order in which functions, variables, and top-level asm
4238            statements are emitted may have changed. Code relying on some
4239            particular ordering needs to be updated. The majority of such
4240            top-level asm statements can be replaced by section
4241            attributes.
4242          + Unreferenced static variables and functions are removed. This
4243            may result in undefined references when an asm statement
4244            refers to the variable/function directly. In that case either
4245            the variable/function shall be listed in asm statement operand
4246            or in the case of top-level asm statements the attribute used
4247            shall be used to force function/variable to be always output
4248            and considered as a possibly used by unknown code.
4249            For variables the attribute is accepted only by GCC 3.4 and
4250            newer, while for earlier versions it is sufficient to use
4251            unused to silence warnings about the variables not being
4252            referenced. To keep code portable across different GCC
4253            versions, you can use appropriate preprocessor conditionals.
4254          + Static functions now can use non-standard passing conventions
4255            that may break asm statements calling functions directly.
4256            Again the attribute used shall be used to prevent this
4257            behavior.
4258       As a temporary workaround, -fno-unit-at-a-time can be used, but
4259       this scheme may not be supported by future releases of GCC.
4260     * GCC 3.4 automatically places zero-initialized variables in the .bss
4261       section on some operating systems. Versions of GNU Emacs up to (and
4262       including) 21.3 will not work correctly when using this
4263       optimization; you can use -fno-zero-initialized-in-bss to disable
4264       it.
4265     * If GCC 3.4 is configured with --enable-threads=posix (the default
4266       on most targets that support pthreads) then _REENTRANT will be
4267       defined unconditionally by some libstdc++ headers. C++ code which
4268       relies on that macro to detect whether multi-threaded code is being
4269       compiled might change in meaning, possibly resulting in linker
4270       errors for single-threaded programs. Affected users of [7]Boost
4271       should compile single-threaded code with -DBOOST_DISABLE_THREADS.
4272       See Bugzilla for [8]more information.
4273
4274General Optimizer Improvements
4275
4276     * Usability of the profile feedback and coverage testing has been
4277       improved.
4278          + Performance of profiled programs has been improved by faster
4279            profile merging code.
4280          + Better use of the profile feedback for optimization (loop
4281            unrolling and loop peeling).
4282          + File locking support allowing fork() calls and parallel runs
4283            of profiled programs.
4284          + Coverage file format has been redesigned.
4285          + gcov coverage tool has been improved.
4286          + make profiledbootstrap available to build a faster compiler.
4287            Experiments made on i386 hardware showed an 11% speedup on -O0
4288            and a 7.5% speedup on -O2 compilation of a [9]large C++
4289            testcase.
4290          + New value profiling pass enabled via -fprofile-values
4291          + New value profile transformations pass enabled via -fvpt aims
4292            to optimize some code sequences by exploiting knowledge about
4293            value ranges or other properties of the operands. At the
4294            moment a conversion of expensive divisions into cheaper
4295            operations has been implemented.
4296          + New -fprofile-generate and -fprofile-use command-line options
4297            to simplify the use of profile feedback.
4298     * A new unit-at-a-time compilation scheme for C, Objective-C, C++ and
4299       Java which is enabled via -funit-at-a-time (and implied by -O2). In
4300       this scheme a whole file is parsed first and optimized later. The
4301       following basic inter-procedural optimizations are implemented:
4302          + Removal of unreachable functions and variables
4303          + Discovery of local functions (functions with static linkage
4304            whose address is never taken)
4305          + On i386, these local functions use register parameter passing
4306            conventions.
4307          + Reordering of functions in topological order of the call graph
4308            to enable better propagation of optimizing hints (such as the
4309            stack alignments needed by functions) in the back end.
4310          + Call graph based out-of-order inlining heuristics which allows
4311            to limit overall compilation unit growth (--param
4312            inline-unit-growth).
4313       Overall, the unit-at-a-time scheme produces a 1.3% improvement for
4314       the SPECint2000 benchmark on the i386 architecture (AMD Athlon
4315       CPU).
4316     * More realistic code size estimates used by inlining for C,
4317       Objective-C, C++ and Java. The growth of large functions can now be
4318       limited via --param large-function-insns and --param
4319       large-function-growth.
4320     * A new cfg-level loop optimizer pass replaces the old loop unrolling
4321       pass and adds two other loop transformations -- loop peeling and
4322       loop unswitching -- and also uses the profile feedback to limit
4323       code growth. (The three optimizations are enabled by
4324       -funroll-loops, -fpeel-loops and -funswitch-loops flags,
4325       respectively).
4326       The old loop unroller still can be enabled by -fold-unroll-loops
4327       and may produce better code in some cases, especially when the
4328       webizer optimization pass is not run.
4329     * A new web construction pass enabled via -fweb (and implied by -O3)
4330       improves the quality of register allocation, CSE, first scheduling
4331       pass and some other optimization passes by avoiding re-use of
4332       pseudo registers with non-overlapping live ranges. The pass almost
4333       always improves code quality but does make debugging difficult and
4334       thus is not enabled by default by -O2
4335       The pass is especially effective as cleanup after code duplication
4336       passes, such as the loop unroller or the tracer.
4337     * Experimental implementations of superblock or trace scheduling in
4338       the second scheduling pass can be enabled via
4339       -fsched2-use-superblocks and -fsched2-use-traces, respectively.
4340
4341New Languages and Language specific improvements
4342
4343  Ada
4344
4345     * The Ada front end has been updated to include numerous bug fixes
4346       and enhancements. These include:
4347          + Improved project file support
4348          + Additional set of warnings about potential wrong code
4349          + Improved error messages
4350          + Improved code generation
4351          + Improved cross reference information
4352          + Improved inlining
4353          + Better run-time check elimination
4354          + Better error recovery
4355          + More efficient implementation of unbounded strings
4356          + Added features in GNAT.Sockets, GNAT.OS_Lib, GNAT.Debug_Pools,
4357            ...
4358          + New GNAT.xxxx packages (e.g. GNAT.Strings,
4359            GNAT.Exception_Action)
4360          + New pragmas
4361          + New -gnatS switch replacing gnatpsta
4362          + Implementation of new Ada features (in particular limited
4363            with, limited aggregates)
4364
4365  C/Objective-C/C++
4366
4367     * Precompiled headers are now supported. Precompiled headers can
4368       dramatically speed up compilation of some projects. There are some
4369       known defects in the current precompiled header implementation that
4370       will result in compiler crashes in relatively rare situations.
4371       Therefore, precompiled headers should be considered a "technology
4372       preview" in this release. Read the manual for details about how to
4373       use precompiled headers.
4374     * File handling in the preprocessor has been rewritten. GCC no longer
4375       gets confused by symlinks and hardlinks, and now has a correct
4376       implementation of #import and #pragma once. These two directives
4377       have therefore been un-deprecated.
4378     * The undocumented extension that allowed C programs to have a label
4379       at the end of a compound statement, which has been deprecated since
4380       GCC 3.0, has been removed.
4381     * The cast-as-lvalue extension has been removed for C++ and
4382       deprecated for C and Objective-C. In particular, code like this:
4383        int i;
4384        (char) i = 5;
4385
4386       or this:
4387        char *p;
4388        ((int *) p)++;
4389
4390       is no longer accepted for C++ and will not be accepted for C and
4391       Objective-C in a future version.
4392     * The conditional-expression-as-lvalue extension has been deprecated
4393       for C and Objective-C. In particular, code like this:
4394        int a, b, c;
4395        (a ? b : c) = 2;
4396
4397       will not be accepted for C and Objective-C in a future version.
4398     * The compound-expression-as-lvalue extension has been deprecated for
4399       C and Objective-C. In particular, code like this:
4400        int a, b;
4401        (a, b) = 2;
4402
4403       will not be accepted for C and Objective-C in a future version. A
4404       possible non-intrusive workaround is the following:
4405        (*(a, &b)) = 2;
4406
4407     * Several [10]built-in functions such as __builtin_popcount for
4408       counting bits, finding the highest and lowest bit in a word, and
4409       parity have been added.
4410     * The -fwritable-strings option has been deprecated and will be
4411       removed.
4412     * Many C math library functions are now recognized as built-ins and
4413       optimized.
4414     * The C, C++, and Objective-C compilers can now handle source files
4415       written in any character encoding supported by the host C library.
4416       The default input character set is taken from the current locale,
4417       and may be overridden with the -finput-charset command line option.
4418       In the future we will add support for inline encoding markers.
4419
4420  C++
4421
4422     * G++ is now much closer to full conformance to the ISO/ANSI C++
4423       standard. This means, among other things, that a lot of invalid
4424       constructs which used to be accepted in previous versions will now
4425       be rejected. It is very likely that existing C++ code will need to
4426       be fixed. This document lists some of the most common issues.
4427     * A hand-written recursive-descent C++ parser has replaced the
4428       YACC-derived C++ parser from previous GCC releases. The new parser
4429       contains much improved infrastructure needed for better parsing of
4430       C++ source codes, handling of extensions, and clean separation
4431       (where possible) between proper semantics analysis and parsing. The
4432       new parser fixes many bugs that were found in the old parser.
4433     * You must now use the typename and template keywords to disambiguate
4434       dependent names, as required by the C++ standard.
4435        struct K {
4436          typedef int mytype_t;
4437        };
4438
4439        template <class T1> struct A {
4440          template <class T2> struct B {
4441              void callme(void);
4442            };
4443
4444          template <int N> void bar(void)
4445          {
4446            // Use 'typename' to tell the parser that T1::mytype_t names
4447            //  a type. This is needed because the name is dependent (in
4448            //  this case, on template parameter T1).
4449            typename T1::mytype_t x;
4450            x = 0;
4451          }
4452        };
4453
4454        template <class T> void template_func(void)
4455        {
4456          // Use 'template' to prefix member templates within
4457          //  dependent types (a has type A<T>, which depends on
4458          //  the template parameter T).
4459          A<T> a;
4460          a.template bar<0>();
4461
4462          // Use 'template' to tell the parser that B is a nested
4463          //  template class (dependent on template parameter T), and
4464          //  'typename' because the whole A<T>::B<int> is
4465          //  the name of a type (again, dependent).
4466          typename A<T>::template B<int> b;
4467          b.callme();
4468        }
4469
4470        void non_template_func(void)
4471        {
4472          // Outside of any template class or function, no names can be
4473          //  dependent, so the use of the keyword 'typename' and 'template'
4474          //  is not needed (and actually forbidden).
4475          A<K> a;
4476          a.bar<0>();
4477          A<K>::B<float> b;
4478          b.callme();
4479        }
4480     * In a template definition, unqualified names will no longer find
4481       members of a dependent base (as specified by [temp.dep]/3 in the
4482       C++ standard). For example,
4483        template <typename T> struct B {
4484          int m;
4485          int n;
4486          int f ();
4487          int g ();
4488        };
4489        int n;
4490        int g ();
4491        template <typename T> struct C : B<T> {
4492          void h ()
4493          {
4494            m = 0; // error
4495            f ();  // error
4496            n = 0; // ::n is modified
4497            g ();  // ::g is called
4498          }
4499        };
4500       You must make the names dependent, e.g. by prefixing them with
4501       this->. Here is the corrected definition of C<T>::h,
4502        template <typename T> void C<T>::h ()
4503        {
4504          this->m = 0;
4505          this->f ();
4506          this->n = 0
4507          this->g ();
4508        }
4509       As an alternative solution (unfortunately not backwards compatible
4510       with GCC 3.3), you may use using declarations instead of this->:
4511        template <typename T> struct C : B<T> {
4512          using B<T>::m;
4513          using B<T>::f;
4514          using B<T>::n;
4515          using B<T>::g;
4516          void h ()
4517          {
4518            m = 0;
4519            f ();
4520            n = 0;
4521            g ();
4522          }
4523        };
4524     * In templates, all non-dependent names are now looked up and bound
4525       at definition time (while parsing the code), instead of later when
4526       the template is instantiated. For instance:
4527        void foo(int);
4528
4529        template <int> struct A {
4530          static void bar(void){
4531            foo('a');
4532          }
4533        };
4534
4535        void foo(char);
4536
4537        int main()
4538        {
4539          A<0>::bar();    // Calls foo(int), used to call foo(char).
4540        }
4541
4542     * In an explicit instantiation of a class template, you must use
4543       class or struct before the template-id:
4544        template <int N>
4545        class A {};
4546
4547        template A<0>;         // error, not accepted anymore
4548        template class A<0>;   // OK
4549     * The "named return value" and "implicit typename" extensions have
4550       been removed.
4551     * Default arguments in function types have been deprecated and will
4552       be removed.
4553     * ARM-style name-injection of friend declarations has been deprecated
4554       and will be removed. For example: struct S { friend void f(); };
4555       void g() { f(); } will not be accepted by future versions of G++;
4556       instead a declaration of "f" will need to be present outside of the
4557       scope of "S".
4558     * Covariant returns are implemented for all but varadic functions
4559       that require an adjustment.
4560     * When -pedantic is used, G++ now issues errors about spurious
4561       semicolons. For example,
4562        namespace N {}; // Invalid semicolon.
4563        void f() {}; // Invalid semicolon.
4564     * G++ no longer accepts attributes for a declarator after the
4565       initializer associated with that declarator. For example,
4566        X x(1) __attribute__((...));
4567       is no longer accepted. Instead, use:
4568        X x __attribute__((...)) (1);
4569     * Inside the scope of a template class, the name of the class itself
4570       can be treated as either a class or a template. So GCC used to
4571       accept the class name as argument of type template, and template
4572       template parameter. However this is not C++ standard compliant. Now
4573       the name is not treated as a valid template template argument
4574       unless you qualify the name by its scope. For example, the code
4575       below no longer compiles.
4576        template <template <class> class TT> class X {};
4577        template <class T> class Y {
4578          X<Y> x; // Invalid, Y is always a type template parameter.
4579        };
4580       The valid code for the above example is
4581          X< ::Y> x; // Valid.
4582       (Notice the space between < and : to prevent GCC to interpret this
4583       as a digraph for [.)
4584     * Friend declarations that refer to template specializations are
4585       rejected if the template has not already been declared. For
4586       example,
4587        template <typename T>
4588        class C {
4589          friend void f<> (C&);
4590        };
4591       is rejected. You must first declare f as a template,
4592        template <typename T>
4593        void f(T);
4594     * In case of friend declarations, every name used in the friend
4595       declaration must be accessible at the point of that declaration.
4596       Previous versions of G++ used to be less strict about this and
4597       allowed friend declarations for private class members, for example.
4598       See the ISO C++ Standard Committee's [11]defect report #209 for
4599       details.
4600     * Declaration of member functions of class templates as friends are
4601       supported. For example,
4602        template <typename T> struct A {
4603          void f();
4604        };
4605        class C {
4606          template <typename T> friend void A<T>::f();
4607        };
4608     * You must use template <> to introduce template specializations, as
4609       required by the standard. For example,
4610        template <typename T>
4611        struct S;
4612
4613        struct S<int> { };
4614       is rejected. You must write,
4615        template <> struct S<int> {};
4616     * G++ used to accept code like this,
4617        struct S {
4618          int h();
4619          void f(int i = g());
4620          int g(int i = h());
4621        };
4622       This behavior is not mandated by the standard. Now G++ issues an
4623       error about this code. To avoid the error, you must move the
4624       declaration of g before the declaration of f. The default arguments
4625       for g must be visible at the point where it is called.
4626     * The C++ ABI Section 3.3.3 specifications for the array construction
4627       routines __cxa_vec_new2 and __cxa_vec_new3 were changed to return
4628       NULL when the allocator argument returns NULL. These changes are
4629       incorporated into the libstdc++ runtime library.
4630     * Using a name introduced by a typedef in a friend declaration or in
4631       an explicit instantiation is now rejected, as specified by the ISO
4632       C++ standard.
4633        class A;
4634        typedef A B;
4635        class C {
4636          friend class B;      // error, no typedef name here
4637          friend B;            // error, friend always needs class/struct/enum
4638          friend class A;      // OK
4639        };
4640
4641        template <int> class Q {};
4642        typedef Q<0> R;
4643        template class R;      // error, no typedef name here
4644        template class Q<0>;   // OK
4645     * When allocating an array with a new expression, GCC used to allow
4646       parentheses around the type name. This is actually ill-formed and
4647       it is now rejected:
4648        int* a = new (int)[10];    // error, not accepted anymore
4649        int* a = new int[10];      // OK
4650     * When binding an rvalue of class type to a reference, the copy
4651       constructor of the class must be accessible. For instance, consider
4652       the following code:
4653        class A
4654        {
4655        public:
4656          A();
4657
4658        private:
4659          A(const A&);   // private copy ctor
4660        };
4661
4662        A makeA(void);
4663        void foo(const A&);
4664
4665        void bar(void)
4666        {
4667          foo(A());       // error, copy ctor is not accessible
4668          foo(makeA());   // error, copy ctor is not accessible
4669
4670          A a1;
4671          foo(a1);        // OK, a1 is a lvalue
4672        }
4673       This might be surprising at first sight, especially since most
4674       popular compilers do not correctly implement this rule ([12]further
4675       details).
4676     * When forming a pointer to member or a pointer to member function,
4677       access checks for class visibility (public, protected, private) are
4678       now performed using the qualifying scope of the name itself. This
4679       is better explained with an example:
4680        class A
4681        {
4682        public:
4683          void pub_func();
4684        protected:
4685          void prot_func();
4686        private:
4687          void priv_func();
4688        };
4689
4690        class B : public A
4691        {
4692        public:
4693          void foo()
4694          {
4695            &A::pub_func;   // OK, pub_func is accessible through A
4696            &A::prot_func;  // error, cannot access prot_func through A
4697            &A::priv_func;  // error, cannot access priv_func through A
4698
4699            &B::pub_func;   // OK, pub_func is accessible through B
4700            &B::prot_func;  // OK, can access prot_func through B (within B)
4701            &B::priv_func;  // error, cannot access priv_func through B
4702          }
4703        };
4704
4705    Runtime Library (libstdc++)
4706
4707     * Optimization work:
4708          + Streamlined streambuf, filebuf, separate synched with C
4709            Standard I/O streambuf.
4710          + All formatted I/O now uses cached locale information.
4711          + STL optimizations (memory/speed for list, red-black trees as
4712            used by sets and maps).
4713          + More use of GCC builtins.
4714          + String optimizations (avoid contention on
4715            increment/decrement-and-test of the reference count in the
4716            empty-string object, constructor from input_iterators
4717            speedup).
4718     * Static linkage size reductions.
4719     * Large File Support (files larger than 2 GB on 32-bit systems).
4720     * Wide character and variable encoding filebuf work (UTF-8, Unicode).
4721     * Generic character traits.
4722     * Also support wchar_t specializations on Mac OS 10.3.x, FreeBSD 5.x,
4723       Solaris 2.7 and above, AIX 5.x, Irix 6.5.
4724     * The allocator class is now standard-conformant, and two additional
4725       extension allocators have been added, mt_alloc and
4726       bitmap_allocator.
4727     * PCH support: -include bits/stdc++.h (2x compile speedup).
4728     * Rewrote __cxa_demangle with support for C++ style allocators.
4729     * New debug modes for STL containers and iterators.
4730     * Testsuite rewrite: five times as many tests, plus increasingly
4731       sophisticated tests, including I/O, MT, multi-locale, wide and
4732       narrow characters.
4733     * Use current versions of GNU "autotools" for build/configuration.
4734
4735  Objective-C
4736
4737     * The Objective-C front end has been updated to include the numerous
4738       bug fixes and enhancements previously available only in Apple's
4739       version of GCC. These include:
4740          + Structured exception (@try... @catch... @finally, @throw) and
4741            synchronization (@synchronized) support. These are accessible
4742            via the -fobjc-exceptions switch; as of this writing, they may
4743            only be used in conjunction with -fnext-runtime on Mac OS X
4744            10.3 and later. See [13]Options Controlling Objective-C
4745            Dialect for more information.
4746          + An overhaul of @encode logic. The C99 _Bool and C++ bool type
4747            may now be encoded as 'B'. In addition, the back-end/codegen
4748            dependencies have been removed.
4749          + An overhaul of message dispatch construction, ensuring that
4750            the various receiver types (and casts thereof) are handled
4751            properly, and that correct diagnostics are issued.
4752          + Support for "Zero-Link" (-fzero-link) and "Fix-and-Continue"
4753            (-freplace-objc-classes) debugging modes, currently available
4754            on Mac OS X 10.3 and later. See [14]Options Controlling
4755            Objective-C Dialect for more information.
4756          + Access to optimized runtime entry points (-fno-nil-receivers )
4757            on the assumption that message receivers are never nil. This
4758            is currently available on Mac OS X 10.3 and later. See
4759            [15]Options Controlling Objective-C Dialect for more
4760            information.
4761
4762  Java
4763
4764     * Compiling a .jar file will now cause non-.class entries to be
4765       automatically compiled as resources.
4766     * libgcj has been ported to Darwin.
4767     * Jeff Sturm has adapted Jan Hubicka's call graph optimization code
4768       to gcj.
4769     * libgcj has a new gcjlib URL type; this lets URLClassLoader load
4770       code from shared libraries.
4771     * libgcj has been much more completely merged with [16]GNU Classpath.
4772     * Class loading is now much more correct; in particular the caller's
4773       class loader is now used when that is required.
4774     * [17]Eclipse 2.x will run out of the box using gij.
4775     * Parts of java.nio have been implemented. Direct and indirect
4776       buffers work, as do fundamental file and socket operations.
4777     * java.awt has been improved, though it is still not ready for
4778       general use.
4779     * The HTTP protocol handler now uses HTTP/1.1 and can handle the POST
4780       method.
4781     * The MinGW port has matured. Enhancements include socket timeout
4782       support, thread interruption, improved Runtime.exec() handling and
4783       support for accented characters in filenames.
4784
4785  Fortran
4786
4787     * Fortran improvements are listed in the [18]Fortran documentation.
4788
4789New Targets and Target Specific Improvements
4790
4791  Alpha
4792
4793     * Several [19]built-in functions have been added such as
4794       __builtin_alpha_zap to allow utilizing the more obscure
4795       instructions of the CPU.
4796     * Parameter passing of complex arguments has changed to match the
4797       [20]ABI. This change is incompatible with previous GCC versions,
4798       but does fix compatibility with the Tru64 compiler and several
4799       corner cases where GCC was incompatible with itself.
4800
4801  ARM
4802
4803     * Nicolas Pitre has contributed his hand-coded floating-point support
4804       code for ARM. It is both significantly smaller and faster than the
4805       existing C-based implementation, even when building applications
4806       for Thumb. The arm-elf configuration has been converted to use the
4807       new code.
4808     * Support for the Intel's iWMMXt architecture, a second generation
4809       XScale processor, has been added. Enabled at run time with the
4810       -mcpu=iwmmxt command line switch.
4811     * A new ARM target has been added: arm-wince-pe. This is similar to
4812       the arm-pe target, but it defaults to using the APCS32 ABI.
4813     * The existing ARM pipeline description has been converted to the use
4814       the [21]DFA processor pipeline model. There is not much change in
4815       code performance, but the description is now [22]easier to
4816       understand.
4817     * Support for the Cirrus EP9312 Maverick floating point co-processor
4818       added. Enabled at run time with the -mcpu=ep9312 command line
4819       switch. Note however that the multilibs to support this chip are
4820       currently disabled in gcc/config/arm/t-arm-elf, so if you want to
4821       enable their production you will have to uncomment the entries in
4822       that file.
4823
4824  H8/300
4825
4826     * Support for long long has been added.
4827     * Support for saveall attribute has been added.
4828     * Pavel Pisa contributed hand-written 32-bit-by-32-bit division code
4829       for H8/300H and H8S, which is much faster than the previous
4830       implementation.
4831     * A lot of small performance improvements.
4832
4833  IA-32/AMD64 (x86-64)
4834
4835     * Tuning for K8 (AMD Opteron/Athlon64) core is available via
4836       -march=k8 and -mcpu=k8.
4837     * Scalar SSE code generation carefully avoids reformatting penalties,
4838       hidden dependencies and minimizes the number of uops generated on
4839       both Intel and AMD CPUs.
4840     * Vector MMX and SSE operands are now passed in registers to improve
4841       performance and match the argument passing convention used by the
4842       Intel C++ Compiler. As a result it is not possible to call
4843       functions accepting vector arguments compiled by older GCC version.
4844     * Conditional jump elimination is now more aggressive on modern CPUs.
4845     * The Athlon ports has been converted to use the DFA processor
4846       pipeline description.
4847     * Optimization of indirect tail calls is now possible in a similar
4848       fashion as direct sibcall optimization.
4849     * Further small performance improvements.
4850     * -m128bit-long-double is now less buggy.
4851     * __float128 support in 64-bit compilation.
4852     * Support for data structures exceeding 2GB in 64-bit mode.
4853     * -mcpu has been renamed to -mtune.
4854
4855  IA-64
4856
4857     * Tuning code for the Itanium 2 processor has been added. The
4858       generation of code tuned for Itanium 2 (option -mtune=itanium2) is
4859       enabled by default now. To generate code tuned for Itanium 1 the
4860       option -mtune=itanium1 should be used.
4861     * [23]DFA processor pipeline descriptions for the IA-64 processors
4862       have been added. This resulted in about 3% improvement on the
4863       SPECInt2000 benchmark for Itanium 2.
4864     * Instruction bundling for the IA-64 processors has been rewritten
4865       using the DFA pipeline hazard recognizer. It resulted in about 60%
4866       compiler speedup on the SPECInt2000 C programs.
4867
4868  M32R
4869
4870     * Support for the M32R/2 processor has been added by Renesas.
4871     * Support for an M32R GNU/Linux target and PIC code generation has
4872       been added by Renesas.
4873
4874  M68000
4875
4876     * Bernardo Innocenti (Develer S.r.l.) has contributed the
4877       m68k-uclinux target, based on former work done by Paul Dale
4878       (SnapGear Inc.). Code generation for the ColdFire processors family
4879       has been enhanced and extended to support the MCF 53xx and MCF 54xx
4880       cores, integrating former work done by Peter Barada (Motorola).
4881
4882  MIPS
4883
4884    Processor-specific changes
4885
4886     * Support for the RM7000 and RM9000 processors has been added. It can
4887       be selected using the -march compiler option and should work with
4888       any MIPS I (mips-*) or MIPS III (mips64-*) configuration.
4889     * Support for revision 2 of the MIPS32 ISA has been added. It can be
4890       selected with the command-line option -march=mips32r2.
4891     * There is a new option, -mfix-sb1, to work around certain SB-1
4892       errata.
4893
4894    Configuration
4895
4896     * It is possible to customize GCC using the following configure-time
4897       options:
4898          + --with-arch, which specifies the default value of the -march
4899            option.
4900          + --with-tune, which specifies the default value of the -mtune
4901            option.
4902          + --with-abi, which specifies the default ABI.
4903          + --with-float=soft, which tells GCC to use software floating
4904            point by default.
4905          + --with-float=hard, which tells GCC to use hardware floating
4906            point by default.
4907     * A 64-bit GNU/Linux port has been added. The associated
4908       configurations are mips64-linux-gnu and mips64el-linux-gnu.
4909     * The 32-bit GNU/Linux port now supports Java.
4910     * The IRIX 6 configuration now supports the o32 ABI and will build
4911       o32 multilibs by default. This support is compatible with both
4912       binutils and the SGI tools, but note that several features,
4913       including debugging information and DWARF2 exception handling, are
4914       only available when using the GNU assembler. Use of the GNU
4915       assembler and linker (version 2.15 or above) is strongly
4916       recommended.
4917     * The IRIX 6 configuration now supports 128-bit long doubles.
4918     * There are two new RTEMS-specific configurations, mips-rtems and
4919       mipsel-rtems.
4920     * There are two new *-elf configurations, mipsisa32r2-elf and
4921       mipsisa32r2el-elf.
4922
4923    General
4924
4925     * Several [24]ABI bugs have been fixed. Unfortunately, these changes
4926       will break binary compatibility with earlier releases.
4927     * GCC can now use explicit relocation operators when generating
4928       -mabicalls code. This behavior is controlled by -mexplicit-relocs
4929       and can have several performance benefits. For example:
4930          + It allows for more optimization of GOT accesses, including
4931            better scheduling and redundancy elimination.
4932          + It allows sibling calls to be implemented as jumps.
4933          + n32 and n64 leaf functions can use a call-clobbered global
4934            pointer instead of $28.
4935          + The code to set up $gp can be removed from functions that
4936            don't need it.
4937     * A new option, -mxgot, allows the GOT to be bigger than 64k. This
4938       option is equivalent to the assembler's -xgot option and should be
4939       used instead of -Wa,-xgot.
4940     * Frame pointer elimination is now supported when generating 64-bit
4941       MIPS16 code.
4942     * Inline block moves have been optimized to take more account of
4943       alignment information.
4944     * Many internal changes have been made to the MIPS port, mostly aimed
4945       at reducing the reliance on assembler macros.
4946
4947  PowerPC
4948
4949     * GCC 3.4 releases have a number of fixes for PowerPC and PowerPC64
4950       [25]ABI incompatibilities regarding the way parameters are passed
4951       during functions calls. These changes may result in incompatibility
4952       between code compiled with GCC 3.3 and GCC 3.4.
4953
4954    PowerPC Darwin
4955
4956     * Support for shared/dylib gcc libraries has been added. It is
4957       enabled by default on powerpc-apple-darwin7.0.0 and up.
4958     * Libgcj is enabled by default. On systems older than
4959       powerpc-apple-darwin7.0.0 you need to install dlcompat.
4960     * 128-bit IBM extended precision format support added for long
4961       double.
4962
4963    PowerPC64 GNU/Linux
4964
4965     * By default, PowerPC64 GNU/Linux now uses natural alignment of
4966       structure elements. The old four byte alignment for double, with
4967       special rules for a struct starting with a double, can be chosen
4968       with -malign-power. This change may result in incompatibility
4969       between code compiled with GCC 3.3 and GCC 3.4.
4970     * -mabi=altivec is now the default rather than -mabi=no-altivec.
4971     * 128-bit IBM extended precision format support added for long
4972       double.
4973
4974  S/390 and zSeries
4975
4976     * New command-line options allow to specify the intended execution
4977       environment for generated code:
4978          + -mesa/-mzarch allows to specify whether to generate code
4979            running in ESA/390 mode or in z/Architecture mode (this is
4980            applicable to 31-bit code only).
4981          + -march allows to specify a minimum processor architecture
4982            level (g5, g6, z900, or z990).
4983          + -mtune allows to specify which processor to tune for.
4984     * It is possible to customize GCC using the following configure-time
4985       options:
4986          + --with-mode, which specifies whether to default to assuming
4987            ESA/390 or z/Architecture mode.
4988          + --with-arch, which specifies the default value of the -march
4989            option.
4990          + --with-tune, which specifies the default value of the -mtune
4991            option.
4992     * Support for the z990 processor has been added, and can be selected
4993       using -march=z990 or -mtune=z990. This includes instruction
4994       scheduling tuned for the superscalar instruction pipeline of the
4995       z990 processor as well as support for all new instructions provided
4996       by the long-displacement facility.
4997     * Support to generate 31-bit code optimized for zSeries processors
4998       (running in ESA/390 or in z/Architecture mode) has been added. This
4999       can be selected using -march=z900 and -mzarch respectively.
5000     * Instruction scheduling for the z900 and z990 processors now uses
5001       the DFA pipeline hazard recognizer.
5002     * GCC no longer generates code to maintain a stack backchain,
5003       previously used to generate stack backtraces for debugging
5004       purposes. As replacement that does not incur runtime overhead,
5005       DWARF-2 call frame information is provided by GCC; this is
5006       supported by GDB 6.1. The old behavior can be restored using the
5007       -mbackchain option.
5008     * The stack frame size of functions may now exceed 2 GB in 64-bit
5009       code.
5010     * A port for the 64-bit IBM TPF operating system has been added; the
5011       configuration is s390x-ibm-tpf. This configuration is supported as
5012       cross-compilation target only.
5013     * Various changes to improve the generated code have been
5014       implemented, including:
5015          + GCC now uses the MULTIPLY AND ADD and MULTIPLY AND SUBTRACT
5016            instructions to significantly speed up many floating-point
5017            applications.
5018          + GCC now uses the ADD LOGICAL WITH CARRY and SUBTRACT LOGICAL
5019            WITH BORROW instructions to speed up long long arithmetic.
5020          + GCC now uses the SEARCH STRING instruction to implement
5021            strlen().
5022          + In many cases, function call overhead for 31-bit code has been
5023            reduced by placing the literal pool after the function code
5024            instead of after the function prolog.
5025          + Register 14 is no longer reserved in 64-bit code.
5026          + Handling of global register variables has been improved.
5027
5028  SPARC
5029
5030     * The option -mflat is deprecated.
5031     * Support for large (> 2GB) frames has been added to the 64-bit port.
5032     * Several [26]ABI bugs have been fixed. Unfortunately, these changes
5033       will break binary compatibility with earlier releases.
5034     * The default debugging format has been switched from STABS to
5035       DWARF-2 for 32-bit code on Solaris 7 and later. DWARF-2 is already
5036       the default debugging format for 64-bit code on Solaris.
5037
5038  SuperH
5039
5040     * Support for the SH2E processor has been added. Enabled at run time
5041       with the -m2e command line switch, or at configure time by
5042       specifying sh2e as the machine part of the target triple.
5043
5044  V850
5045
5046     * Support for the Mitsubishi V850E1 processor has been added. This is
5047       a variant of the V850E processor with some additional debugging
5048       instructions.
5049
5050  Xtensa
5051
5052     * Several ABI bugs have been fixed. Unfortunately, these changes
5053       break binary compatibility with earlier releases.
5054          + For big-endian processors, the padding of aggregate return
5055            values larger than a word has changed. If the size of an
5056            aggregate return value is not a multiple of 32 bits, previous
5057            versions of GCC inserted padding in the most-significant bytes
5058            of the first return value register. Aggregates larger than a
5059            word are now padded in the least-significant bytes of the last
5060            return value register used. Aggregates smaller than a word are
5061            still padded in the most-significant bytes. The return value
5062            padding has not changed for little-endian processors.
5063          + Function arguments with 16-byte alignment are now properly
5064            aligned.
5065          + The implementation of the va_list type has changed. A va_list
5066            value created by va_start from a previous release cannot be
5067            used with va_arg from this release, or vice versa.
5068     * More processor configuration options for Xtensa processors are
5069       supported:
5070          + the ABS instruction is now optional;
5071          + the ADDX* and SUBX* instructions are now optional;
5072          + an experimental CONST16 instruction can be used to synthesize
5073            constants instead of loading them from constant pools.
5074       These and other Xtensa processor configuration options can no
5075       longer be enabled or disabled by command-line options; the
5076       processor configuration must be specified by the xtensa-config.h
5077       header file when building GCC. Additionally, the
5078       -mno-serialize-volatile option is no longer supported.
5079
5080Obsolete Systems
5081
5082   Support for a number of older systems has been declared obsolete in GCC
5083   3.4. Unless there is activity to revive them, the next release of GCC
5084   will have their sources permanently removed.
5085
5086   All configurations of the following processor architectures have been
5087   declared obsolete:
5088     * Mitsubishi D30V, d30v-*
5089     * AT&T DSP1600 and DSP1610, dsp16xx-*
5090     * Intel 80960, i960
5091
5092   Also, some individual systems have been obsoleted:
5093     * ARM Family
5094          + Support for generating code for operation in APCS/26 mode
5095            (-mapcs-26).
5096     * IBM ESA/390
5097          + "Bigfoot" port, i370-*. (The other port, s390-*, is actively
5098            maintained and supported.)
5099     * Intel 386 family
5100          + MOSS, i?86-moss-msdos and i?86-*-moss*
5101          + NCR 3000 running System V r.4, i?86-ncr-sysv4*
5102          + FreeBSD with a.out object format, i?86-*-freebsd*aout* and
5103            i?86-*-freebsd2*
5104          + GNU/Linux with a.out object format, i?86-linux*aout*
5105          + GNU/Linux with libc5, a.k.a. glibc1, i?86-linux*libc1*
5106          + Interix versions before Interix 3, i?86-*-interix
5107          + Mach microkernel, i?86-mach*
5108          + SCO UnixWare with UDK, i?86-*-udk*
5109          + Generic System V releases 1, 2, and 3, i?86-*-sysv[123]*
5110          + VSTa microkernel, i386-*-vsta
5111     * Motorola M68000 family
5112          + HPUX, m68k-hp-hpux* and m68000-hp-hpux*
5113          + NetBSD with a.out object format (before NetBSD 1.4),
5114            m68k-*-*-netbsd* except m68k-*-*-netbsdelf*
5115          + Generic System V r.4, m68k-*-sysv4*
5116     * VAX
5117          + Generic VAX, vax-*-* (This is generic VAX only; we have not
5118            obsoleted any VAX triples for specific operating systems.)
5119
5120Documentation improvements
5121
5122Other significant improvements
5123
5124     * The build system has undergone several significant cleanups.
5125       Subdirectories will only be configured if they are being built, and
5126       all subdirectory configures are run from the make command. The top
5127       level has been autoconfiscated.
5128     * Building GCC no longer writes to its source directory. This should
5129       help those wishing to share a read-only source directory over NFS
5130       or build from a CD. The exceptions to this feature are if you
5131       configure with either --enable-maintainer-mode or
5132       --enable-generated-files-in-srcdir.
5133     * The -W warning option has been renamed to -Wextra, which is more
5134       easily understood. The older spelling will be retained for
5135       backwards compatibility.
5136     * Substantial improvements in compile time have been made,
5137       particularly for non-optimizing compilations.
5138     __________________________________________________________________
5139
5140GCC 3.4.0
5141
5142  Bug Fixes
5143
5144   A vast number of bugs have been fixed in 3.4.0, too many to publish a
5145   complete list here. [27]Follow this link to query the Bugzilla database
5146   for the list of over 900 bugs fixed in 3.4.0. This is the list of all
5147   bugs marked as resolved and fixed in 3.4.0 that are not flagged as 3.4
5148   regressions.
5149     __________________________________________________________________
5150
5151GCC 3.4.1
5152
5153  Bug Fixes
5154
5155   This section lists the problem reports (PRs) from GCC's bug tracking
5156   system that are known to be fixed in the 3.4.1 release. This list might
5157   not be complete (that is, it is possible that some PRs that have been
5158   fixed are not listed here).
5159
5160    Bootstrap failures
5161
5162     * [28]10129 Ada bootstrap fails on PPC-Darwin - invalid assembler
5163       emitted - PIC related
5164     * [29]14576 [ARM] ICE in libiberty when building gcc-3.4 for arm-elf
5165     * [30]14760 A bug in configure.in prevents using both
5166       --program-suffix and --program-prefix
5167     * [31]14671 [hppa64] bootstrap fails: ICE in
5168       save_call_clobbered_regs, in caller_save.c
5169     * [32]15093 [alpha][Java] make bootstrap fails to configure libffi on
5170       Alpha
5171     * [33]15178 Solaris 9/x86 fails linking after stage 3
5172
5173    Multi-platform internal compiler errors (ICEs)
5174
5175     * [34]12753 (preprocessor) Memory corruption in preprocessor on bad
5176       input
5177     * [35]13985 ICE in gcc.c-torture/compile/930621-1.c
5178     * [36]14810 (c++) tree check failures with invalid code involving
5179       templates
5180     * [37]14883 (c++) ICE on invalid code, in cp_parser_lookup_name, in
5181       cp/parser.c
5182     * [38]15044 (c++) ICE on syntax error, template header
5183     * [39]15057 (c++) Compiling of conditional value throw constructs
5184       cause a segmentation violation
5185     * [40]15064 (c++) typeid of template parameter gives ICE
5186     * [41]15142 (c++) ICE when passing a string where a char* is expected
5187       in a throw statement
5188     * [42]15159 ICE in rtl_verify_flow_info_1
5189     * [43]15165 (c++) ICE in instantiate_template
5190     * [44]15193 Unary minus using pointer to V4SF vector causes
5191       -fforce-mem to exhaust all memory
5192     * [45]15209 (c++) Runs out of memory with packed structs
5193     * [46]15227 (c++) Trouble with invalid function definition
5194     * [47]15285 (c++) instantiate_type ICE when forming pointer to
5195       template function
5196     * [48]15299 (c++) ICE in resolve_overloaded_unification
5197     * [49]15329 (c++) ICE on constructor of member template
5198     * [50]15550 ICE in extract_insn, in recog.c
5199     * [51]15554 (c++) ICE in tsubst_copy, in cp/pt.c
5200     * [52]15640 (c++) ICE on invalid code in arg_assoc, in
5201       cp/name-lookup.c
5202     * [53]15666 [unit-at-a-time] Gcc abort on valid code
5203     * [54]15696 (c++) ICE with bad pointer-to-member code
5204     * [55]15701 (c++) ICE with friends and template template parameter
5205     * [56]15761 ICE in do_SUBST, in combine.c
5206     * [57]15829 (c++) ICE on Botan-1.3.13 due to -funroll-loops
5207
5208    Ada
5209
5210     * [58]14538 All RTEMS targets broken for gnat
5211
5212    C front end
5213
5214     * [59]12391 missing warning about assigning to an incomplete type
5215     * [60]14649 atan(1.0) should not be a constant expression
5216     * [61]15004 [unit-at-a-time] no warning for unused paramater in
5217       static function
5218     * [62]15749 --pedantic-errors behaves differently from --pedantic
5219       with C-compiler on GNU/Linux
5220
5221    C++ compiler and library
5222
5223     * [63]10646 non-const reference is incorrectly matched in a "const T"
5224       partial specialization
5225     * [64]12077 wcin.rdbuf()->in_avail() return value too high
5226     * [65]13598 enc_filebuf doesn't work
5227     * [66]14211 const_cast returns lvalue but should be rvalue
5228     * [67]14220 num_put::do_put() undesired float/double behavior
5229     * [68]14245 problem with user-defined allocators in std::basic_string
5230     * [69]14340 libstdc++ Debug mode: failure to convert iterator to
5231       const_iterator
5232     * [70]14600 __gnu_cxx::stdio_sync_filebuf should expose internal
5233       FILE*
5234     * [71]14668 no warning anymore for reevaluation of declaration
5235     * [72]14775 LFS (large file support) tests missing
5236     * [73]14821 Duplicate namespace alias declaration should not conflict
5237     * [74]14930 Friend declaration ignored
5238     * [75]14932 cannot use offsetof to get offsets of array elements in
5239       g++ 3.4.0
5240     * [76]14950 [non unit-at-a-time] always_inline does not mix with
5241       templates and -O0
5242     * [77]14962 g++ ignores #pragma redefine_extname
5243     * [78]14975 Segfault on low-level write error during imbue
5244     * [79]15002 Linewise stream input is unusably slow (std::string slow)
5245     * [80]15025 compiler accepts redeclaration of template as
5246       non-template
5247     * [81]15046 [arm] Math functions misdetected by cross configuration
5248     * [82]15069 a bit test on a variable of enum type is miscompiled
5249     * [83]15074 g++ -lsupc++ still links against libstdc++
5250     * [84]15083 spurious "statement has no effect" warning
5251     * [85]15096 parse error with templates and pointer to const member
5252     * [86]15287 combination of operator[] and operator .* fails in
5253       templates
5254     * [87]15317 __attribute__ unused in first parameter of constructor
5255       gives error
5256     * [88]15337 sizeof on incomplete type diagnostic
5257     * [89]15361 bitset<>::_Find_next fails
5258     * [90]15412 _GLIBCXX_ symbols symbols defined and used in different
5259       namespaces
5260     * [91]15427 valid code results in incomplete type error
5261     * [92]15471 Incorrect member pointer offsets in anonymous
5262       structs/unions
5263     * [93]15503 nested template problem
5264     * [94]15507 compiler hangs while laying out union
5265     * [95]15542 operator & and template definitions
5266     * [96]15565 SLES9: leading + sign for unsigned int with showpos
5267     * [97]15625 friend defined inside a template fails to find static
5268       function
5269     * [98]15629 Function templates, overloads, and friend name injection
5270     * [99]15742 'noreturn' attribute ignored in method of template
5271       functions.
5272     * [100]15775 Allocator::pointer consistently ignored
5273     * [101]15821 Duplicate namespace alias within namespace rejected
5274     * [102]15862 'enum yn' fails (confict with undeclared builtin)
5275     * [103]15875 rejects pointer to member in template
5276     * [104]15877 valid code using templates and anonymous enums is
5277       rejected
5278     * [105]15947 Puzzling error message for wrong destructor declaration
5279       in template class
5280     * [106]16020 cannot copy __gnu_debug::bitset
5281     * [107]16154 input iterator concept too restrictive
5282     * [108]16174 deducing top-level consts
5283
5284    Java
5285
5286     * [109]14315 Java compiler is not parallel make safe
5287
5288    Fortran
5289
5290     * [110]15151 [g77] incorrect logical i/o in 64-bit mode
5291
5292    Objective-C
5293
5294     * [111]7993 private variables cannot be shadowed in subclasses
5295
5296    Optimization bugs
5297
5298     * [112]15228 useless copies of floating point operands
5299     * [113]15345 [non-unit-at-a-time] unreferenced nested inline
5300       functions not optimized away
5301     * [114]15945 Incorrect floating point optimization
5302     * [115]15526 ftrapv aborts on 0 * (-1)
5303     * [116]14690 Miscompiled POOMA tests
5304     * [117]15112 GCC generates code to write to unchanging memory
5305
5306    Preprocessor
5307
5308     * [118]15067 Minor glitch in the source of cpp
5309
5310    Main driver program bugs
5311
5312     * [119]1963 collect2 interprets -oldstyle_liblookup as -o
5313       ldstyle_liblookup
5314
5315    x86-specific (Intel/AMD)
5316
5317     * [120]15717 Error: can't resolve `L0' {*ABS* section} - `xx' {*UND*
5318       section}
5319
5320    HPPA-specific
5321
5322     * [121]14782 GCC produces an unaligned data access at -O2
5323     * [122]14828 FAIL: gcc.c-torture/execute/20030408-1.c execution, -O2
5324     * [123]15202 ICE in reload_cse_simplify_operands, in postreload.c
5325
5326    IA64-specific
5327
5328     * [124]14610 __float80 constants incorrectly emitted
5329     * [125]14813 init_array sections are initialized in the wrong order
5330     * [126]14857 GCC segfault on duplicated asm statement
5331     * [127]15598 Gcc 3.4 ICE on valid code
5332     * [128]15653 Gcc 3.4 ICE on valid code
5333
5334    MIPS-specific
5335
5336     * [129]15189 wrong filling of delay slot with -march=mips1 -G0
5337       -mno-split-addresses -mno-explicit-relocs
5338     * [130]15331 Assembler error building gnatlib on IRIX 6.5 with GNU as
5339       2.14.91
5340     * [131]16144 Bogus reference to __divdf3 when -O1
5341     * [132]16176 Miscompilation of unaligned data in MIPS backend
5342
5343    PowerPC-specific
5344
5345     * [133]11591 ICE in gcc.dg/altivec-5.c
5346     * [134]12028 powerpc-eabispe produces bad sCOND operation
5347     * [135]14478 rs6000 geu/ltu patterns generate incorrect code
5348     * [136]14567 long double and va_arg complex args
5349     * [137]14715 Altivec stack layout may overlap gpr save with stack
5350       temps
5351     * [138]14902 (libstdc++) Stream checking functions fail when -pthread
5352       option is used.
5353     * [139]14924 Compiler ICE on valid code
5354     * [140]14960 -maltivec affects vector return with -mabi=no-altivec
5355     * [141]15106 vector varargs failure passing from altivec to
5356       non-altivec code for -m32
5357     * [142]16026 ICE in function.c:4804, assign_parms, when -mpowerpc64 &
5358       half-word operation
5359     * [143]15191 -maltivec -mabi=no-altivec results in mis-aligned lvx
5360       and stvx
5361     * [144]15662 Segmentation fault when an exception is thrown - even if
5362       try and catch are specified
5363
5364    s390-specific
5365
5366     * [145]15054 Bad code due to overlapping stack temporaries
5367
5368    SPARC-specific
5369
5370     * [146]15783 ICE with union assignment in 64-bit mode
5371     * [147]15626 GCC 3.4 emits "ld: warning: relocation error:
5372       R_SPARC_UA32"
5373
5374    x86-64-specific
5375
5376     * [148]14326 boehm-gc hardcodes to 3DNow! prefetch for x86_64
5377     * [149]14723 Backported -march=nocona from mainline
5378     * [150]15290 __float128 failed to pass to function properly
5379
5380    Cygwin/Mingw32-specific
5381
5382     * [151]15250 Option -mms-bitfields support on GCC 3.4 is not
5383       conformant to MS layout
5384     * [152]15551 -mtune=pentium4 -O2 with sjlj EH breaks stack probe
5385       worker on windows32 targets
5386
5387    Bugs specific to embedded processors
5388
5389     * [153]8309 [m68k] -m5200 produces erroneous SImode set of short
5390       varaible on stack
5391     * [154]13250 [SH] Gcc code for rotation clobbers the register, but
5392       gcc continues to use the register as if it was not clobbered
5393     * [155]13803 [coldfire] movqi operand constraints too restrictivefor
5394       TARGET_COLDFIRE
5395     * [156]14093 [SH] ICE for code when using -mhitachi option in SH
5396     * [157]14457 [m6811hc] ICE with simple c++ source
5397     * [158]14542 [m6811hc] ICE on simple source
5398     * [159]15100 [SH] cc1plus got hang-up on
5399       libstdc++-v3/testsuite/abi_check.cc
5400     * [160]15296 [CRIS] Delayed branch scheduling causing invalid code on
5401       cris-*
5402     * [161]15396 [SH] ICE with -O2 -fPIC
5403     * [162]15782 [coldfire] m68k_output_mi_thunk emits wrong code for
5404       ColdFire
5405
5406    Testsuite problems (compiler not affected)
5407
5408     * [163]11610 libstdc++ testcases 27_io/* don't work properly remotely
5409     * [164]15488 (libstdc++) possibly insufficient file permissions for
5410       executing test suite
5411     * [165]15489 (libstdc++) testsuite_files determined incorrectly
5412
5413    Documentation bugs
5414
5415     * [166]13928 (libstdc++) no whatis info in some man pages generated
5416       by doxygen
5417     * [167]14150 Ada documentation out of date
5418     * [168]14949 (c++) Need to document method visibility changes
5419     * [169]15123 libstdc++-doc: Allocators.3 manpage is empty
5420     __________________________________________________________________
5421
5422GCC 3.4.2
5423
5424  Bug Fixes
5425
5426   This section lists the problem reports (PRs) from GCC's bug tracking
5427   system that are known to be fixed in the 3.4.2 release. This list might
5428   not be complete (that is, it is possible that some PRs that have been
5429   fixed are not listed here).
5430
5431    Bootstrap failures and issues
5432
5433     * [170]16469 [mips-sgi-irix5.3] bootstrap fails in
5434       libstdc++-v3/testsuite
5435     * [171]16344 [hppa-linux-gnu] libstdc++'s PCH built by
5436       profiledbootstrap does not work with the built compiler
5437     * [172]16842 [Solaris/x86] mkheaders can not find mkheaders.conf
5438
5439    Multi-platform internal compiler errors (ICEs)
5440
5441     * [173]12608 (c++) ICE: expected class 't', have 'x' (error_mark) in
5442       cp_parser_class_specifier, in cp/parser.c
5443     * [174]14492 ICE in loc_descriptor_from_tree, in dwarf2out.c
5444     * [175]15461 (c++) ICE due to NRV and inlining
5445     * [176]15890 (c++) ICE in c_expand_expr, in c-common.c
5446     * [177]16180 ICE: segmentation fault in RTL optimization
5447     * [178]16224 (c++) ICE in write_unscoped_name (template/namespace)
5448     * [179]16408 ICE: in delete_insn, in cfgrtl.c
5449     * [180]16529 (c++) ICE for: namespace-alias shall not be declared as
5450       the name of any other entity
5451     * [181]16698 (c++) ICE with exceptions and declaration of __cxa_throw
5452     * [182]16706 (c++) ICE in finish_member_declaration, in
5453       cp/semantics.c
5454     * [183]16810 (c++) Legal C++ program with cast gives ICE in
5455       build_ptrmemfunc
5456     * [184]16851 (c++) ICE when throwing a comma expression
5457     * [185]16870 (c++) Boost.Spirit causes ICE in tsubst, in cp/pt.c
5458     * [186]16904 (c++) ICE in finish_class_member_access_expr, in
5459       cp/typeck.c
5460     * [187]16905 (c++) ICE (segfault) with exceptions
5461     * [188]16964 (c++) ICE in cp_parser_class_specifier due to
5462       redefinition
5463     * [189]17068 (c++) ICE: tree check: expected class 'd', have 'x'
5464       (identifier_node) in dependent_template_p, in cp/pt.c
5465
5466    Preprocessor bugs
5467
5468     * [190]16366 Preprocessor option -remap causes memory corruption
5469
5470    Optimization
5471
5472     * [191]15345 unreferenced nested inline functions not optimized away
5473     * [192]16590 Incorrect execution when compiling with -O2
5474     * [193]16693 Bitwise AND is lost when used within a cast to an enum
5475       of the same precision
5476     * [194]17078 Jump into if(0) substatement fails
5477
5478    Problems in generated debug information
5479
5480     * [195]13956 incorrect stabs for nested local variables
5481
5482    C front end bugs
5483
5484     * [196]16684 GCC should not warn about redundant redeclarations of
5485       built-ins
5486
5487    C++ compiler and library
5488
5489     * [197]12658 Thread safety problems in locale::global() and
5490       locale::locale()
5491     * [198]13092 g++ accepts invalid pointer-to-member conversion
5492     * [199]15320 Excessive memory consumption
5493     * [200]16246 Incorrect template argument deduction
5494     * [201]16273 Memory exhausted when using nested classes and virtual
5495       functions
5496     * [202]16401 ostringstream in gcc 3.4.x very slow for big data
5497     * [203]16411 undefined reference to
5498       __gnu_cxx::stdio_sync_filebuf<char, std::char_traits<char>
5499       >::file()
5500     * [204]16489 G++ incorrectly rejects use of a null constant integral
5501       expression as a null constant pointer
5502     * [205]16618 offsetof fails with constant member
5503     * [206]16637 syntax error reported for valid input code
5504     * [207]16717 __attribute__((constructor)) broken in C++
5505     * [208]16813 compiler error in DEBUG version of range insertion
5506       std::map::insert
5507     * [209]16853 pointer-to-member initialization from incompatible one
5508       accepted
5509     * [210]16889 ambiguity is not detected
5510     * [211]16959 Segmentation fault in ios_base::sync_with_stdio
5511
5512    Java compiler and library
5513
5514     * [212]7587 direct threaded interpreter not thread-safe
5515     * [213]16473 ServerSocket accept() leaks file descriptors
5516     * [214]16478 Hash synchronization deadlock with finalizers
5517
5518    Alpha-specific
5519
5520     * [215]10695 ICE in dwarf2out_frame_debug_expr, in dwarf2out.c
5521     * [216]16974 could not split insn (ice in final_scan_insn, in
5522       final.c)
5523
5524    x86-specific
5525
5526     * [217]16298 ICE in output_operand
5527     * [218]17113 ICE with SSE2 intrinsics
5528
5529    x86-64 specific
5530
5531     * [219]14697 libstdc++ couldn't find 32bit libgcc_s
5532
5533    MIPS-specific
5534
5535     * [220]15869 [mips64] No NOP after LW (with -mips1 -O0)
5536     * [221]16325 [mips64] value profiling clobbers gp on mips
5537     * [222]16357 [mipsisa64-elf] ICE copying 7 bytes between extern
5538       char[]s
5539     * [223]16380 [mips64] Use of uninitialised register after dbra
5540       conversion
5541     * [224]16407 [mips64] Unaligned access to local variables
5542     * [225]16643 [mips64] verify_local_live_at_start ICE after
5543       crossjumping & cfgcleanup
5544
5545    ARM-specific
5546
5547     * [226]15927 THUMB -O2: strength-reduced iteration variable ends up
5548       off by 1
5549     * [227]15948 THUMB: ICE with non-commutative cbranch
5550     * [228]17019 THUMB: bad switch statement in md code for
5551       addsi3_cbranch_scratch
5552
5553    IA64-specific
5554
5555     * [229]16130 ICE on valid code: in bundling, in config/ia64/ia64.c
5556       (-mtune=merced)
5557     * [230]16142 ICE on valid code: in bundling, in config/ia64/ia64.c
5558       (-mtune=itanium)
5559     * [231]16278 Gcc failed to build Linux kernel with -mtune=merced
5560     * [232]16414 ICE on valid code: typo in comparison of asm_noperands
5561       result
5562     * [233]16445 ICE on valid code: don't count ignored insns
5563     * [234]16490 ICE (segfault) while compiling with -fprofile-use
5564     * [235]16683 ia64 does not honor SUBTARGET_EXTRA_SPECS
5565
5566    PowerPC-specific
5567
5568     * [236]16195 (ppc64): Miscompilation of GCC 3.3.x by 3.4.x
5569     * [237]16239 ICE on ppc64 (mozilla 1.7 compile, -O1 -fno-exceptions
5570       issue)
5571
5572    SPARC-specific
5573
5574     * [238]16199 ICE while compiling apache 2.0.49
5575     * [239]16416 -m64 doesn't imply -mcpu=v9 anymore
5576     * [240]16430 ICE when returning non-C aggregates larger than 16 bytes
5577
5578    Bugs specific to embedded processors
5579
5580     * [241]16379 [m32r] can't output large model function call of memcpy
5581     * [242]17093 [m32r] ICE with -msdata=use -O0
5582     * [243]17119 [m32r] ICE at switch case 0x8000
5583
5584    DJGPP-specific
5585
5586     * [244]15928 libstdc++ in 3.4.x doesn't cross-compile for djgpp
5587
5588    Alpha Tru64-specific
5589
5590     * [245]16210 libstdc++ gratuitously omits "long long" I/O
5591
5592    Testsuite, documentation issues (compiler is not affected):
5593
5594     * [246]15488 (libstdc++) possibly insufficient file permissions for
5595       executing test suite
5596     * [247]16250 ada/doctools runs makeinfo even in release tarball
5597     __________________________________________________________________
5598
5599GCC 3.4.3
5600
5601   This is the [248]list of problem reports (PRs) from GCC's bug tracking
5602   system that are known to be fixed in the 3.4.3 release. This list might
5603   not be complete (that is, it is possible that some PRs that have been
5604   fixed are not listed here).
5605
5606    Bootstrap failures
5607
5608     * [249]17369 [ia64] Bootstrap failure with binutils-2.15.90.0.1.1
5609     * [250]17850 [arm-elf] bootstrap failure - libstdc++ uses strtold
5610       when undeclared
5611
5612    Internal compiler errors (ICEs) affecting multiple platforms
5613
5614     * [251]13948 (java) GCJ segmentation fault while compiling GL4Java
5615       .class files
5616     * [252]14492 ICE in loc_descriptor_from_tree, in dwarf2out.c
5617     * [253]16301 (c++) ICE when "strong" attribute is attached to a using
5618       directive
5619     * [254]16566 ICE with flexible arrays
5620     * [255]17023 ICE with nested functions in parameter declaration
5621     * [256]17027 ICE with noreturn function in loop at -O2
5622     * [257]17524 ICE in grokdeclarator, in cp/decl.c
5623     * [258]17826 (c++) ICE in cp_tree_equal
5624
5625    C and optimization bugs
5626
5627     * [259]15526 -ftrapv aborts on 0 * (-1)
5628     * [260]16999 #ident stopped working
5629     * [261]17503 quadratic behaviour in invalid_mode_change_p
5630     * [262]17581 Long long arithmetic fails inside a switch/case
5631       statement when compiled with -O2
5632     * [263]18129 -fwritable-strings doesn't work
5633
5634    C++ compiler and library bugs
5635
5636     * [264]10975 incorrect initial ostringstream::tellp()
5637     * [265]11722 Unbuffered filebuf::sgetn is slow
5638     * [266]14534 Unrecognizing static function as a template parameter
5639       when its return value is also templated
5640     * [267]15172 Copy constructor optimization in aggregate
5641       initialization
5642     * [268]15786 Bad error message for frequently occuring error.
5643     * [269]16162 Rejects valid member-template-definition
5644     * [270]16612 empty basic_strings can't live in shared memory
5645     * [271]16715 std::basic_iostream is instantiated when used, even
5646       though instantiations are already contained in libstdc++
5647     * [272]16848 code in /ext/demangle.h appears broken
5648     * [273]17132 GCC fails to eliminate function template specialization
5649       when argument deduction fails
5650     * [274]17259 One more _S_leaf incorrectly qualified with _RopeRep::
5651       in ropeimpl.h
5652     * [275]17327 use of `enumeral_type' in template type unification
5653     * [276]17393 "unused variable '._0'" warning with -Wall
5654     * [277]17501 Confusion with member templates
5655     * [278]17537 g++ not passing -lstdc++ to linker when all command line
5656       arguments are libraries
5657     * [279]17585 usage of unqualified name of static member from within
5658       class not allowed
5659     * [280]17821 Poor diagnostic for using "." instead of "->"
5660     * [281]17829 wrong error: call of overloaded function is ambiguous
5661     * [282]17851 Misleading diagnostic for invalid function declarations
5662       with undeclared types
5663     * [283]17976 Destructor is called twice
5664     * [284]18020 rejects valid definition of enum value in template
5665     * [285]18093 bogus conflict in namespace aliasing
5666     * [286]18140 C++ parser bug when using >> in templates
5667
5668    Fortran
5669
5670     * [287]17541 data statements with double precision constants fail
5671
5672    x86-specific
5673
5674     * [288]17853 -O2 ICE for MMX testcase
5675
5676    SPARC-specific
5677
5678     * [289]17245 ICE compiling gsl-1.5 statistics/lag1.c
5679
5680    Darwin-specific
5681
5682     * [290]17167 FATAL:Symbol L_foo$stub already defined.
5683
5684    AIX-specific
5685
5686     * [291]17277 could not catch an exception when specified -maix64
5687
5688    Solaris-specific
5689
5690     * [292]17505 <cmath> calls acosf(), ceilf(), and other functions
5691       missing from system libraries
5692
5693    HP/UX specific:
5694
5695     * [293]17684 /usr/ccs/bin/ld: Can't create libgcc_s.sl
5696
5697    ARM-specific
5698
5699     * [294]17384 ICE with mode attribute on structures
5700
5701    MIPS-specific
5702
5703     * [295]17770 No NOP after LWL with -mips1
5704
5705    Other embedded target specific
5706
5707     * [296]11476 [arc-elf] gcc ICE on newlib's vfprintf.c
5708     * [297]14064 [avr-elf] -fdata-sections triggers ICE
5709     * [298]14678 [m68hc11-elf] gcc ICE
5710     * [299]15583 [powerpc-rtems] powerpc-rtems lacks __USE_INIT_FINI__
5711     * [300]15790 [i686-coff] Alignment error building gcc with i686-coff
5712       target
5713     * [301]15886 [SH] Miscompilation with -O2 -fPIC
5714     * [302]16884 [avr-elf] [fweb related] bug while initializing
5715       variables
5716
5717    Bugs relating to debugger support
5718
5719     * [303]13841 missing debug info for _Complex function arguments
5720     * [304]15860 [big-endian targets] No DW_AT_location debug info is
5721       emitted for formal arguments to a function that uses "register"
5722       qualifiers
5723
5724    Testsuite issues (compiler not affected)
5725
5726     * [305]17465 Testsuite in libffi overrides LD_LIBRARY_PATH
5727     * [306]17469 Testsuite in libstdc++ overrides LD_LIBRARY_PATH
5728     * [307]18138 [mips-sgi-irix6.5] libgcc_s.so.1 not found by 64-bit
5729       testsuite
5730
5731    Documentation
5732
5733     * [308]15498 typo in gcc manual: non-existing locale example en_UK,
5734       should be en_GB
5735     * [309]15747 [mips-sgi-irix5.3] /bin/sh hangs during bootstrap:
5736       document broken shell
5737     * [310]16406 USE_LD_AS_NEEDED undocumented
5738     __________________________________________________________________
5739
5740GCC 3.4.4
5741
5742   This is the [311]list of problem reports (PRs) from GCC's bug tracking
5743   system that are known to be fixed in the 3.4.4 release. This list might
5744   not be complete (that is, it is possible that some PRs that have been
5745   fixed are not listed here).
5746     __________________________________________________________________
5747
5748GCC 3.4.5
5749
5750   This is the [312]list of problem reports (PRs) from GCC's bug tracking
5751   system that are known to be fixed in the 3.4.5 release. This list might
5752   not be complete (that is, it is possible that some PRs that have been
5753   fixed are not listed here).
5754
5755    Bootstrap issues
5756
5757     * [313]24688 sco_math fixincl breaks math.h
5758
5759    C compiler bugs
5760
5761     * [314]17188 struct Foo { } redefinition
5762     * [315]20187 wrong code for ((unsigned char)(unsigned long
5763       long)((a?a:1)&(a*b)))?0:1)
5764     * [316]21873 infinite warning loop on bad array initializer
5765     * [317]21899 enum definition accepts values to be overriden
5766     * [318]22061 ICE in find_function_data, in function.c
5767     * [319]22308 Failure to diagnose violation of constraint 6.516p2
5768     * [320]22458 ICE on missing brace
5769     * [321]22589 ICE casting to long long
5770     * [322]24101 Segfault with preprocessed source
5771
5772    C++ compiler and library bugs
5773
5774     * [323]10611 operations on vector mode not recognized in C++
5775     * [324]13377 unexpected behavior of namespace usage directive
5776     * [325]16002 Strange error message with new parser
5777     * [326]17413 local classes as template argument
5778     * [327]17609 spurious error message after using keyword
5779     * [328]17618 ICE in cp_convert_to_pointer, in cp/cvt.c
5780     * [329]18124 ICE with invalid template template parameter
5781     * [330]18155 typedef in template declaration not rejected
5782     * [331]18177 ICE with const_cast for undeclared variable
5783     * [332]18368 C++ error message regression
5784     * [333]16378 ICE when returning a copy of a packed member
5785     * [334]18466 int ::i; accepted
5786     * [335]18512 ICE on invalid usage of template base class
5787     * [336]18454 ICE when returning undefined type
5788     * [337]18738 typename not allowed with non-dependent qualified name
5789     * [338]18803 rejects access to operator() in template
5790     * [339]19004 ICE in uses_template_parms, in cp/pt.c
5791     * [340]19208 Spurious error about variably modified type
5792     * [341]18253 bad error message / ICE for invalid template parameter
5793     * [342]19608 ICE after friend function definition in local class
5794     * [343]19884 ICE on explicit instantiation of a non-template
5795       constructor
5796     * [344]20153 ICE when C++ template function contains anonymous union
5797     * [345]20563 Infinite loop in diagnostic (and ice after error
5798       message)
5799     * [346]20789 ICE with incomplete type in template
5800     * [347]21336 Internal compiler error when using custom new operators
5801     * [348]21768 ICE in error message due to violation of coding
5802       conventions
5803     * [349]21853 constness of pointer to data member ignored
5804     * [350]21903 Default argument of template function causes a
5805       compile-time error
5806     * [351]21983 multiple diagnostics
5807     * [352]21987 New testsuite failure
5808       g++.dg/warn/conversion-function-1.C
5809     * [353]22153 ICE on invalid template specialization
5810     * [354]22172 Internal compiler error, seg fault.
5811     * [355]21286 filebuf::xsgetn vs pipes
5812     * [356]22233 ICE with wrong number of template parameters
5813     * [357]22508 ICE after invalid operator new
5814     * [358]22545 ICE with pointer to class member & user defined
5815       conversion operator
5816     * [359]23528 Wrong default allocator in ext/hash_map
5817     * [360]23550 char_traits requirements/1.cc test bad math
5818     * [361]23586 Bad diagnostic for invalid namespace-name
5819     * [362]23624 ICE in invert_truthvalue, in fold-const.c
5820     * [363]23639 Bad error message: not a member of '<declaration error>'
5821     * [364]23797 ICE on typename outside template
5822     * [365]23965 Bogus error message: no matching function for call to
5823       'foo(<type error>)'
5824     * [366]24052 &#`label_decl' not supported by dump_expr#<expression
5825       error>
5826     * [367]24580 virtual base class cause exception not to be caught
5827
5828    Problems in generated debug information
5829
5830     * [368]24267 Bad DWARF for altivec vectors
5831
5832    Optimizations issues
5833
5834     * [369]17810 ICE in verify_local_live_at_start
5835     * [370]17860 Wrong generated code for loop with varying bound
5836     * [371]21709 ICE on compile-time complex NaN
5837     * [372]21964 broken tail call at -O2 or more
5838     * [373]22167 Strange optimization bug when using -Os
5839     * [374]22619 Compilation failure for real_const_1.f and
5840       real_const_2.f90
5841     * [375]23241 Invalid code generated for comparison of uchar to 255
5842     * [376]23478 Miscompilation due to reloading of a var that is also
5843       used in EH pad
5844     * [377]24470 segmentation fault in cc1plus when compiling with -O
5845     * [378]24950 ICE in operand_subword_force
5846
5847    Precompiled headers problems
5848
5849     * [379]14400 Cannot compile qt-x11-free-3.3.0
5850     * [380]14940 PCH largefile test fails on various platforms
5851
5852    Preprocessor bugs
5853
5854     * [381]20239 ICE on empty preprocessed input
5855     * [382]15220 "gcc -E -MM -MG" reports missing system headers in
5856       source directory
5857
5858    Testsuite issues
5859
5860     * [383]19275 gcc.dg/20020919-1.c fails with -fpic/-fPIC on
5861       i686-pc-linux-gnu
5862
5863    Alpha specific
5864
5865     * [384]21888 bootstrap failure with linker relaxation enabled
5866
5867    ARM specific
5868
5869     * [385]15342 [arm-linux]: ICE in verify_local_live_at_start
5870     * [386]23985 Memory aliasing information incorrect in inlined memcpy
5871
5872    ColdFile specific
5873
5874     * [387]16719 Illegal move of byte into address register causes
5875       compiler to ICE
5876
5877    HPPA specific
5878
5879     * [388]21723 ICE while building libgfortran
5880     * [389]21841 -mhp-ld/-mgnu-ld documentation
5881
5882    IA-64 specific
5883
5884     * [390]23644 IA-64 hardware models and configuration options
5885       documentation error
5886     * [391]24718 Shared libgcc not used for linking by default
5887
5888    M68000 specific
5889
5890     * [392]18421 ICE in reload_cse_simplify_operands, in postreload.c
5891
5892    MIPS specific
5893
5894     * [393]20621 ICE in change_address_1, in emit-rtl.c
5895
5896    PowerPC and PowerPC64 specific
5897
5898     * [394]18583 error on valid code: const
5899       __attribute__((altivec(vector__))) doesn't work in arrays
5900     * [395]20191 ICE in reload_cse_simplify_operands
5901     * [396]22083 AIX: TARGET_C99_FUNCTIONS is wrongly defined
5902     * [397]23070 CALL_V4_CLEAR_FP_ARGS flag not properly set
5903     * [398]23404 gij trashes args of functions with more than 8 fp args
5904     * [399]23539 C & C++ compiler generating misaligned references
5905       regardless of compiler flags
5906     * [400]24102 floatdisf2_internal2 broken
5907     * [401]24465 -mminimal-toc miscompilation of __thread vars
5908
5909    Solaris specific
5910
5911     * [402]19933 Problem with define of HUGE_VAL in math_c99
5912     * [403]21889 Native Solaris assembler cannot grok DTP-relative debug
5913       symbols
5914
5915    SPARC specific
5916
5917     * [404]19300 PCH failures on sparc-linux
5918     * [405]20301 Assembler labels have a leading "-"
5919     * [406]20673 C PCH testsuite assembly comparison failure
5920
5921    x86 and x86_64 specific
5922
5923     * [407]18582 ICE with arrays of type V2DF
5924     * [408]19340 Compilation SEGFAULTs with -O1 -fschedule-insns2
5925       -fsched2-use-traces
5926     * [409]21716 ICE in reg-stack.c's swap_rtx_condition
5927     * [410]24315 amd64 fails -fpeephole2
5928     __________________________________________________________________
5929
5930GCC 3.4.6
5931
5932   This is the [411]list of problem reports (PRs) from GCC's bug tracking
5933   system that are known to be fixed in the 3.4.6 release. This list might
5934   not be complete (that is, it is possible that some PRs that have been
5935   fixed are not listed here).
5936
5937
5938    For questions related to the use of GCC, please consult these web
5939    pages and the [412]GCC manuals. If that fails, the
5940    [413]gcc-help@gcc.gnu.org mailing list might help. Comments on these
5941    web pages and the development of GCC are welcome on our developer
5942    list at [414]gcc@gcc.gnu.org. All of [415]our lists have public
5943    archives.
5944
5945   Copyright (C) [416]Free Software Foundation, Inc. Verbatim copying and
5946   distribution of this entire article is permitted in any medium,
5947   provided this notice is preserved.
5948
5949   These pages are [417]maintained by the GCC team. Last modified
5950   2012-04-24[418].
5951
5952References
5953
5954   1. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.6
5955   2. http://gcc.gnu.org/gcc-3.4/changes.html#cplusplus
5956   3. http://gcc.gnu.org/gcc-3.3/changes.html#obsolete_systems
5957   4. http://gcc.gnu.org/gcc-3.4/changes.html#obsolete_systems
5958   5. http://gcc.gnu.org/gcc-3.4/mips-abi.html
5959   6. http://gcc.gnu.org/gcc-3.4/sparc-abi.html
5960   7. http://www.boost.org/
5961   8. http://gcc.gnu.org/PR11953
5962   9. http://gcc.gnu.org/PR8361
5963  10. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Other-Builtins.html#Other%20Builtins
5964  11. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_closed.html#209
5965  12. http://gcc.gnu.org/bugs/#cxx_rvalbind
5966  13. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Objective-C-Dialect-Options.html
5967  14. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Objective-C-Dialect-Options.html
5968  15. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Objective-C-Dialect-Options.html
5969  16. http://www.gnu.org/software/classpath/
5970  17. http://www.eclipse.org/
5971  18. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/g77/News.html
5972  19. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Alpha-Built-in-Functions.html
5973  20. http://h30097.www3.hp.com/docs/base_doc/DOCUMENTATION/V51A_HTML/ARH9MBTE/DTMNPLTN.HTM#normal-argument-list-structure
5974  21. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gccint/Processor-pipeline-description.html
5975  22. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gccint/Comparison-of-the-two-descriptions.html
5976  23. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gccint/Processor-pipeline-description.html
5977  24. http://gcc.gnu.org/gcc-3.4/mips-abi.html
5978  25. http://gcc.gnu.org/gcc-3.4/powerpc-abi.html
5979  26. http://gcc.gnu.org/gcc-3.4/sparc-abi.html
5980  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
5981  28. http://gcc.gnu.org/PR10129
5982  29. http://gcc.gnu.org/PR14576
5983  30. http://gcc.gnu.org/PR14760
5984  31. http://gcc.gnu.org/PR14671
5985  32. http://gcc.gnu.org/PR15093
5986  33. http://gcc.gnu.org/PR15178
5987  34. http://gcc.gnu.org/PR12753
5988  35. http://gcc.gnu.org/PR13985
5989  36. http://gcc.gnu.org/PR14810
5990  37. http://gcc.gnu.org/PR14883
5991  38. http://gcc.gnu.org/PR15044
5992  39. http://gcc.gnu.org/PR15057
5993  40. http://gcc.gnu.org/PR15064
5994  41. http://gcc.gnu.org/PR15142
5995  42. http://gcc.gnu.org/PR15159
5996  43. http://gcc.gnu.org/PR15165
5997  44. http://gcc.gnu.org/PR15193
5998  45. http://gcc.gnu.org/PR15209
5999  46. http://gcc.gnu.org/PR15227
6000  47. http://gcc.gnu.org/PR15285
6001  48. http://gcc.gnu.org/PR15299
6002  49. http://gcc.gnu.org/PR15329
6003  50. http://gcc.gnu.org/PR15550
6004  51. http://gcc.gnu.org/PR15554
6005  52. http://gcc.gnu.org/PR15640
6006  53. http://gcc.gnu.org/PR15666
6007  54. http://gcc.gnu.org/PR15696
6008  55. http://gcc.gnu.org/PR15701
6009  56. http://gcc.gnu.org/PR15761
6010  57. http://gcc.gnu.org/PR15829
6011  58. http://gcc.gnu.org/PR14538
6012  59. http://gcc.gnu.org/PR12391
6013  60. http://gcc.gnu.org/PR14649
6014  61. http://gcc.gnu.org/PR15004
6015  62. http://gcc.gnu.org/PR15749
6016  63. http://gcc.gnu.org/PR10646
6017  64. http://gcc.gnu.org/PR12077
6018  65. http://gcc.gnu.org/PR13598
6019  66. http://gcc.gnu.org/PR14211
6020  67. http://gcc.gnu.org/PR14220
6021  68. http://gcc.gnu.org/PR14245
6022  69. http://gcc.gnu.org/PR14340
6023  70. http://gcc.gnu.org/PR14600
6024  71. http://gcc.gnu.org/PR14668
6025  72. http://gcc.gnu.org/PR14775
6026  73. http://gcc.gnu.org/PR14821
6027  74. http://gcc.gnu.org/PR14930
6028  75. http://gcc.gnu.org/PR14932
6029  76. http://gcc.gnu.org/PR14950
6030  77. http://gcc.gnu.org/PR14962
6031  78. http://gcc.gnu.org/PR14975
6032  79. http://gcc.gnu.org/PR15002
6033  80. http://gcc.gnu.org/PR15025
6034  81. http://gcc.gnu.org/PR15046
6035  82. http://gcc.gnu.org/PR15069
6036  83. http://gcc.gnu.org/PR15074
6037  84. http://gcc.gnu.org/PR15083
6038  85. http://gcc.gnu.org/PR15096
6039  86. http://gcc.gnu.org/PR15287
6040  87. http://gcc.gnu.org/PR15317
6041  88. http://gcc.gnu.org/PR15337
6042  89. http://gcc.gnu.org/PR15361
6043  90. http://gcc.gnu.org/PR15412
6044  91. http://gcc.gnu.org/PR15427
6045  92. http://gcc.gnu.org/PR15471
6046  93. http://gcc.gnu.org/PR15503
6047  94. http://gcc.gnu.org/PR15507
6048  95. http://gcc.gnu.org/PR15542
6049  96. http://gcc.gnu.org/PR15565
6050  97. http://gcc.gnu.org/PR15625
6051  98. http://gcc.gnu.org/PR15629
6052  99. http://gcc.gnu.org/PR15742
6053 100. http://gcc.gnu.org/PR15775
6054 101. http://gcc.gnu.org/PR15821
6055 102. http://gcc.gnu.org/PR15862
6056 103. http://gcc.gnu.org/PR15875
6057 104. http://gcc.gnu.org/PR15877
6058 105. http://gcc.gnu.org/PR15947
6059 106. http://gcc.gnu.org/PR16020
6060 107. http://gcc.gnu.org/PR16154
6061 108. http://gcc.gnu.org/PR16174
6062 109. http://gcc.gnu.org/PR14315
6063 110. http://gcc.gnu.org/PR15151
6064 111. http://gcc.gnu.org/PR7993
6065 112. http://gcc.gnu.org/PR15228
6066 113. http://gcc.gnu.org/PR15345
6067 114. http://gcc.gnu.org/PR15945
6068 115. http://gcc.gnu.org/PR15526
6069 116. http://gcc.gnu.org/PR14690
6070 117. http://gcc.gnu.org/PR15112
6071 118. http://gcc.gnu.org/PR15067
6072 119. http://gcc.gnu.org/PR1963
6073 120. http://gcc.gnu.org/PR15717
6074 121. http://gcc.gnu.org/PR14782
6075 122. http://gcc.gnu.org/PR14828
6076 123. http://gcc.gnu.org/PR15202
6077 124. http://gcc.gnu.org/PR14610
6078 125. http://gcc.gnu.org/PR14813
6079 126. http://gcc.gnu.org/PR14857
6080 127. http://gcc.gnu.org/PR15598
6081 128. http://gcc.gnu.org/PR15653
6082 129. http://gcc.gnu.org/PR15189
6083 130. http://gcc.gnu.org/PR15331
6084 131. http://gcc.gnu.org/PR16144
6085 132. http://gcc.gnu.org/PR16176
6086 133. http://gcc.gnu.org/PR11591
6087 134. http://gcc.gnu.org/PR12028
6088 135. http://gcc.gnu.org/PR14478
6089 136. http://gcc.gnu.org/PR14567
6090 137. http://gcc.gnu.org/PR14715
6091 138. http://gcc.gnu.org/PR14902
6092 139. http://gcc.gnu.org/PR14924
6093 140. http://gcc.gnu.org/PR14960
6094 141. http://gcc.gnu.org/PR15106
6095 142. http://gcc.gnu.org/PR16026
6096 143. http://gcc.gnu.org/PR15191
6097 144. http://gcc.gnu.org/PR15662
6098 145. http://gcc.gnu.org/PR15054
6099 146. http://gcc.gnu.org/PR15783
6100 147. http://gcc.gnu.org/PR15626
6101 148. http://gcc.gnu.org/PR14326
6102 149. http://gcc.gnu.org/PR14723
6103 150. http://gcc.gnu.org/PR15290
6104 151. http://gcc.gnu.org/PR15250
6105 152. http://gcc.gnu.org/PR15551
6106 153. http://gcc.gnu.org/PR8309
6107 154. http://gcc.gnu.org/PR13250
6108 155. http://gcc.gnu.org/PR13803
6109 156. http://gcc.gnu.org/PR14093
6110 157. http://gcc.gnu.org/PR14457
6111 158. http://gcc.gnu.org/PR14542
6112 159. http://gcc.gnu.org/PR15100
6113 160. http://gcc.gnu.org/PR15296
6114 161. http://gcc.gnu.org/PR15396
6115 162. http://gcc.gnu.org/PR15782
6116 163. http://gcc.gnu.org/PR11610
6117 164. http://gcc.gnu.org/PR15488
6118 165. http://gcc.gnu.org/PR15489
6119 166. http://gcc.gnu.org/PR13928
6120 167. http://gcc.gnu.org/PR14150
6121 168. http://gcc.gnu.org/PR14949
6122 169. http://gcc.gnu.org/PR15123
6123 170. http://gcc.gnu.org/PR16469
6124 171. http://gcc.gnu.org/PR16344
6125 172. http://gcc.gnu.org/PR16842
6126 173. http://gcc.gnu.org/PR12608
6127 174. http://gcc.gnu.org/PR14492
6128 175. http://gcc.gnu.org/PR15461
6129 176. http://gcc.gnu.org/PR15890
6130 177. http://gcc.gnu.org/PR16180
6131 178. http://gcc.gnu.org/PR16224
6132 179. http://gcc.gnu.org/PR16408
6133 180. http://gcc.gnu.org/PR16529
6134 181. http://gcc.gnu.org/PR16698
6135 182. http://gcc.gnu.org/PR16706
6136 183. http://gcc.gnu.org/PR16810
6137 184. http://gcc.gnu.org/PR16851
6138 185. http://gcc.gnu.org/PR16870
6139 186. http://gcc.gnu.org/PR16904
6140 187. http://gcc.gnu.org/PR16905
6141 188. http://gcc.gnu.org/PR16964
6142 189. http://gcc.gnu.org/PR17068
6143 190. http://gcc.gnu.org/PR16366
6144 191. http://gcc.gnu.org/PR15345
6145 192. http://gcc.gnu.org/PR16590
6146 193. http://gcc.gnu.org/PR16693
6147 194. http://gcc.gnu.org/PR17078
6148 195. http://gcc.gnu.org/PR13956
6149 196. http://gcc.gnu.org/PR16684
6150 197. http://gcc.gnu.org/PR12658
6151 198. http://gcc.gnu.org/PR13092
6152 199. http://gcc.gnu.org/PR15320
6153 200. http://gcc.gnu.org/PR16246
6154 201. http://gcc.gnu.org/PR16273
6155 202. http://gcc.gnu.org/PR16401
6156 203. http://gcc.gnu.org/PR16411
6157 204. http://gcc.gnu.org/PR16489
6158 205. http://gcc.gnu.org/PR16618
6159 206. http://gcc.gnu.org/PR16637
6160 207. http://gcc.gnu.org/PR16717
6161 208. http://gcc.gnu.org/PR16813
6162 209. http://gcc.gnu.org/PR16853
6163 210. http://gcc.gnu.org/PR16889
6164 211. http://gcc.gnu.org/PR16959
6165 212. http://gcc.gnu.org/PR7587
6166 213. http://gcc.gnu.org/PR16473
6167 214. http://gcc.gnu.org/PR16478
6168 215. http://gcc.gnu.org/PR10695
6169 216. http://gcc.gnu.org/PR16974
6170 217. http://gcc.gnu.org/PR16298
6171 218. http://gcc.gnu.org/PR17113
6172 219. http://gcc.gnu.org/PR14697
6173 220. http://gcc.gnu.org/PR15869
6174 221. http://gcc.gnu.org/PR16325
6175 222. http://gcc.gnu.org/PR16357
6176 223. http://gcc.gnu.org/PR16380
6177 224. http://gcc.gnu.org/PR16407
6178 225. http://gcc.gnu.org/PR16643
6179 226. http://gcc.gnu.org/PR15927
6180 227. http://gcc.gnu.org/PR15948
6181 228. http://gcc.gnu.org/PR17019
6182 229. http://gcc.gnu.org/PR16130
6183 230. http://gcc.gnu.org/PR16142
6184 231. http://gcc.gnu.org/PR16278
6185 232. http://gcc.gnu.org/PR16414
6186 233. http://gcc.gnu.org/PR16445
6187 234. http://gcc.gnu.org/PR16490
6188 235. http://gcc.gnu.org/PR16683
6189 236. http://gcc.gnu.org/PR16195
6190 237. http://gcc.gnu.org/PR16239
6191 238. http://gcc.gnu.org/PR16199
6192 239. http://gcc.gnu.org/PR16416
6193 240. http://gcc.gnu.org/PR16430
6194 241. http://gcc.gnu.org/PR16379
6195 242. http://gcc.gnu.org/PR17093
6196 243. http://gcc.gnu.org/PR17119
6197 244. http://gcc.gnu.org/PR15928
6198 245. http://gcc.gnu.org/PR16210
6199 246. http://gcc.gnu.org/PR15488
6200 247. http://gcc.gnu.org/PR16250
6201 248. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.4.3
6202 249. http://gcc.gnu.org/PR17369
6203 250. http://gcc.gnu.org/PR17850
6204 251. http://gcc.gnu.org/PR13948
6205 252. http://gcc.gnu.org/PR14492
6206 253. http://gcc.gnu.org/PR16301
6207 254. http://gcc.gnu.org/PR16566
6208 255. http://gcc.gnu.org/PR17023
6209 256. http://gcc.gnu.org/PR17027
6210 257. http://gcc.gnu.org/PR17524
6211 258. http://gcc.gnu.org/PR17826
6212 259. http://gcc.gnu.org/PR15526
6213 260. http://gcc.gnu.org/PR16999
6214 261. http://gcc.gnu.org/PR17503
6215 262. http://gcc.gnu.org/PR17581
6216 263. http://gcc.gnu.org/PR18129
6217 264. http://gcc.gnu.org/PR10975
6218 265. http://gcc.gnu.org/PR11722
6219 266. http://gcc.gnu.org/PR14534
6220 267. http://gcc.gnu.org/PR15172
6221 268. http://gcc.gnu.org/PR15786
6222 269. http://gcc.gnu.org/PR16162
6223 270. http://gcc.gnu.org/PR16612
6224 271. http://gcc.gnu.org/PR16715
6225 272. http://gcc.gnu.org/PR16848
6226 273. http://gcc.gnu.org/PR17132
6227 274. http://gcc.gnu.org/PR17259
6228 275. http://gcc.gnu.org/PR17327
6229 276. http://gcc.gnu.org/PR17393
6230 277. http://gcc.gnu.org/PR17501
6231 278. http://gcc.gnu.org/PR17537
6232 279. http://gcc.gnu.org/PR17585
6233 280. http://gcc.gnu.org/PR17821
6234 281. http://gcc.gnu.org/PR17829
6235 282. http://gcc.gnu.org/PR17851
6236 283. http://gcc.gnu.org/PR17976
6237 284. http://gcc.gnu.org/PR18020
6238 285. http://gcc.gnu.org/PR18093
6239 286. http://gcc.gnu.org/PR18140
6240 287. http://gcc.gnu.org/PR17541
6241 288. http://gcc.gnu.org/PR17853
6242 289. http://gcc.gnu.org/PR17245
6243 290. http://gcc.gnu.org/PR17167
6244 291. http://gcc.gnu.org/PR17277
6245 292. http://gcc.gnu.org/PR17505
6246 293. http://gcc.gnu.org/PR17684
6247 294. http://gcc.gnu.org/PR17384
6248 295. http://gcc.gnu.org/PR17770
6249 296. http://gcc.gnu.org/PR11476
6250 297. http://gcc.gnu.org/PR14064
6251 298. http://gcc.gnu.org/PR14678
6252 299. http://gcc.gnu.org/PR15583
6253 300. http://gcc.gnu.org/PR15790
6254 301. http://gcc.gnu.org/PR15886
6255 302. http://gcc.gnu.org/PR16884
6256 303. http://gcc.gnu.org/PR13841
6257 304. http://gcc.gnu.org/PR15860
6258 305. http://gcc.gnu.org/PR17465
6259 306. http://gcc.gnu.org/PR17469
6260 307. http://gcc.gnu.org/PR18138
6261 308. http://gcc.gnu.org/PR15498
6262 309. http://gcc.gnu.org/PR15747
6263 310. http://gcc.gnu.org/PR16406
6264 311. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.4.4
6265 312. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.4.5
6266 313. http://gcc.gnu.org/PR24688
6267 314. http://gcc.gnu.org/PR17188
6268 315. http://gcc.gnu.org/PR20187
6269 316. http://gcc.gnu.org/PR21873
6270 317. http://gcc.gnu.org/PR21899
6271 318. http://gcc.gnu.org/PR22061
6272 319. http://gcc.gnu.org/PR22208
6273 320. http://gcc.gnu.org/PR22458
6274 321. http://gcc.gnu.org/PR22589
6275 322. http://gcc.gnu.org/PR24101
6276 323. http://gcc.gnu.org/PR10611
6277 324. http://gcc.gnu.org/PR13377
6278 325. http://gcc.gnu.org/PR16002
6279 326. http://gcc.gnu.org/PR17413
6280 327. http://gcc.gnu.org/PR17609
6281 328. http://gcc.gnu.org/PR17618
6282 329. http://gcc.gnu.org/PR18124
6283 330. http://gcc.gnu.org/PR18155
6284 331. http://gcc.gnu.org/PR18177
6285 332. http://gcc.gnu.org/PR18368
6286 333. http://gcc.gnu.org/PR18378
6287 334. http://gcc.gnu.org/PR18466
6288 335. http://gcc.gnu.org/PR18512
6289 336. http://gcc.gnu.org/PR18545
6290 337. http://gcc.gnu.org/PR18738
6291 338. http://gcc.gnu.org/PR18803
6292 339. http://gcc.gnu.org/PR19004
6293 340. http://gcc.gnu.org/PR19208
6294 341. http://gcc.gnu.org/PR19253
6295 342. http://gcc.gnu.org/PR19608
6296 343. http://gcc.gnu.org/PR19884
6297 344. http://gcc.gnu.org/PR20153
6298 345. http://gcc.gnu.org/PR20563
6299 346. http://gcc.gnu.org/PR20789
6300 347. http://gcc.gnu.org/PR21336
6301 348. http://gcc.gnu.org/PR21768
6302 349. http://gcc.gnu.org/PR21853
6303 350. http://gcc.gnu.org/PR21903
6304 351. http://gcc.gnu.org/PR21983
6305 352. http://gcc.gnu.org/PR21987
6306 353. http://gcc.gnu.org/PR22153
6307 354. http://gcc.gnu.org/PR22172
6308 355. http://gcc.gnu.org/PR21286
6309 356. http://gcc.gnu.org/PR22233
6310 357. http://gcc.gnu.org/PR22508
6311 358. http://gcc.gnu.org/PR22545
6312 359. http://gcc.gnu.org/PR23528
6313 360. http://gcc.gnu.org/PR23550
6314 361. http://gcc.gnu.org/PR23586
6315 362. http://gcc.gnu.org/PR23624
6316 363. http://gcc.gnu.org/PR23639
6317 364. http://gcc.gnu.org/PR23797
6318 365. http://gcc.gnu.org/PR23965
6319 366. http://gcc.gnu.org/PR24052
6320 367. http://gcc.gnu.org/PR24580
6321 368. http://gcc.gnu.org/PR24267
6322 369. http://gcc.gnu.org/PR17810
6323 370. http://gcc.gnu.org/PR17860
6324 371. http://gcc/gnu.org/PR21709
6325 372. http://gcc.gnu.org/PR21964
6326 373. http://gcc.gnu.org/PR22167
6327 374. http://gcc.gnu.org/PR22619
6328 375. http://gcc.gnu.org/PR23241
6329 376. http://gcc.gnu.org/PR23478
6330 377. http://gcc.gnu.org/PR24470
6331 378. http://gcc.gnu.org/PR24950
6332 379. http://gcc.gnu.org/PR14400
6333 380. http://gcc.gnu.org/PR14940
6334 381. http://gcc.gnu.org/PR20239
6335 382. http://gcc.gnu.org/PR15220
6336 383. http://gcc.gnu.org/PR19275
6337 384. http://gcc.gnu.org/PR21888
6338 385. http://gcc.gnu.org/PR15342
6339 386. http://gcc.gnu.org/PR23985
6340 387. http://gcc.gnu.org/PR16719
6341 388. http://gcc.gnu.org/PR21723
6342 389. http://gcc.gnu.org/PR21841
6343 390. http://gcc.gnu.org/PR23644
6344 391. http://gcc.gnu.org/PR24718
6345 392. http://gcc.gnu.org/PR18421
6346 393. http://gcc.gnu.org/PR20621
6347 394. http://gcc.gnu.org/PR18583
6348 395. http://gcc.gnu.org/PR20191
6349 396. http://gcc.gnu.org/PR22083
6350 397. http://gcc.gnu.org/PR23070
6351 398. http://gcc.gnu.org/PR23404
6352 399. http://gcc.gnu.org/PR23539
6353 400. http://gcc.gnu.org/PR24102
6354 401. http://gcc.gnu.org/PR24465
6355 402. http://gcc.gnu.org/PR19933
6356 403. http://gcc.gnu.org/PR21889
6357 404. http://gcc.gnu.org/PR19300
6358 405. http://gcc.gnu.org/PR20301
6359 406. http://gcc.gnu.org/PR20673
6360 407. http://gcc.gnu.org/PR18582
6361 408. http://gcc.gnu.org/PR19340
6362 409. http://gcc.gnu.org/PR21716
6363 410. http://gcc.gnu.org/PR24315
6364 411. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.4.6
6365 412. http://gcc.gnu.org/onlinedocs/
6366 413. mailto:gcc-help@gcc.gnu.org
6367 414. mailto:gcc@gcc.gnu.org
6368 415. http://gcc.gnu.org/lists.html
6369 416. http://www.fsf.org/
6370 417. http://gcc.gnu.org/about.html
6371 418. http://validator.w3.org/check/referer
6372======================================================================
6373http://gcc.gnu.org/gcc-3.3/index.html
6374
6375                             GCC 3.3 Release Series
6376
6377   May 03, 2005
6378
6379   The [1]GNU project and the GCC developers are pleased to announce the
6380   release of GCC 3.3.6.
6381
6382   This release is a bug-fix release, containing fixes for regressions in
6383   GCC 3.3.5 relative to previous releases of GCC.
6384
6385   This release is the last of the series 3.3.x.
6386
6387   The GCC 3.3 release series includes numerous [2]new features,
6388   improvements, bug fixes, and other changes, thanks to an [3]amazing
6389   group of volunteers.
6390
6391Release History
6392
6393   GCC 3.3.6
6394          May 3, 2005 ([4]changes)
6395
6396   GCC 3.3.5
6397          September 30, 2004 ([5]changes)
6398
6399   GCC 3.3.4
6400          May 31, 2004 ([6]changes)
6401
6402   GCC 3.3.3
6403          February 14, 2004 ([7]changes)
6404
6405   GCC 3.3.2
6406          October 16, 2003 ([8]changes)
6407
6408   GCC 3.3.1
6409          August 8, 2003 ([9]changes)
6410
6411   GCC 3.3
6412          May 14, 2003 ([10]changes)
6413
6414References and Acknowledgements
6415
6416   GCC used to stand for the GNU C Compiler, but since the compiler
6417   supports several other languages aside from C, it now stands for the
6418   GNU Compiler Collection.
6419
6420   A list of [11]successful builds is updated as new information becomes
6421   available.
6422
6423   The GCC developers would like to thank the numerous people that have
6424   contributed new features, improvements, bug fixes, and other changes as
6425   well as test results to GCC. This [12]amazing group of volunteers is
6426   what makes GCC successful.
6427
6428   For additional information about GCC please refer to the [13]GCC
6429   project web site or contact the [14]GCC development mailing list.
6430
6431   To obtain GCC please use [15]our mirror sites, or our CVS server.
6432
6433
6434    For questions related to the use of GCC, please consult these web
6435    pages and the [16]GCC manuals. If that fails, the
6436    [17]gcc-help@gcc.gnu.org mailing list might help. Comments on these
6437    web pages and the development of GCC are welcome on our developer
6438    list at [18]gcc@gcc.gnu.org. All of [19]our lists have public
6439    archives.
6440
6441   Copyright (C) [20]Free Software Foundation, Inc. Verbatim copying and
6442   distribution of this entire article is permitted in any medium,
6443   provided this notice is preserved.
6444
6445   These pages are [21]maintained by the GCC team. Last modified
6446   2011-04-25[22].
6447
6448References
6449
6450   1. http://www.gnu.org/
6451   2. http://gcc.gnu.org/gcc-3.3/changes.html
6452   3. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
6453   4. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.6
6454   5. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.5
6455   6. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.4
6456   7. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.3
6457   8. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.2
6458   9. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.1
6459  10. http://gcc.gnu.org/gcc-3.3/changes.html
6460  11. http://gcc.gnu.org/gcc-3.3/buildstat.html
6461  12. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
6462  13. http://gcc.gnu.org/index.html
6463  14. mailto:gcc@gcc.gnu.org
6464  15. http://gcc.gnu.org/mirrors.html
6465  16. http://gcc.gnu.org/onlinedocs/
6466  17. mailto:gcc-help@gcc.gnu.org
6467  18. mailto:gcc@gcc.gnu.org
6468  19. http://gcc.gnu.org/lists.html
6469  20. http://www.fsf.org/
6470  21. http://gcc.gnu.org/about.html
6471  22. http://validator.w3.org/check/referer
6472======================================================================
6473http://gcc.gnu.org/gcc-3.3/changes.html
6474
6475                             GCC 3.3 Release Series
6476                        Changes, New Features, and Fixes
6477
6478   The latest release in the 3.3 release series is [1]GCC 3.3.6.
6479
6480Caveats
6481
6482     * The preprocessor no longer accepts multi-line string literals. They
6483       were deprecated in 3.0, 3.1, and 3.2.
6484     * The preprocessor no longer supports the -A- switch when appearing
6485       alone. -A- followed by an assertion is still supported.
6486     * Support for all the systems [2]obsoleted in GCC 3.1 has been
6487       removed from GCC 3.3. See below for a [3]list of systems which are
6488       obsoleted in this release.
6489     * Checking for null format arguments has been decoupled from the rest
6490       of the format checking mechanism. Programs which use the format
6491       attribute may regain this functionality by using the new [4]nonnull
6492       function attribute. Note that all functions for which GCC has a
6493       built-in format attribute, an appropriate built-in nonnull
6494       attribute is also applied.
6495     * The DWARF (version 1) debugging format has been deprecated and will
6496       be removed in a future version of GCC. Version 2 of the DWARF
6497       debugging format will continue to be supported for the foreseeable
6498       future.
6499     * The C and Objective-C compilers no longer accept the "Naming Types"
6500       extension (typedef foo = bar); it was already unavailable in C++.
6501       Code which uses it will need to be changed to use the "typeof"
6502       extension instead: typedef typeof(bar) foo. (We have removed this
6503       extension without a period of deprecation because it has caused the
6504       compiler to crash since version 3.0 and no one noticed until very
6505       recently. Thus we conclude it is not in widespread use.)
6506     * The -traditional C compiler option has been removed. It was
6507       deprecated in 3.1 and 3.2. (Traditional preprocessing remains
6508       available.) The <varargs.h> header, used for writing variadic
6509       functions in traditional C, still exists but will produce an error
6510       message if used.
6511     * GCC 3.3.1 automatically places zero-initialized variables in the
6512       .bss section on some operating systems. Versions of GNU Emacs up to
6513       (and including) 21.3 will not work correctly when using this
6514       optimization; you can use -fno-zero-initialized-in-bss to disable
6515       it.
6516
6517General Optimizer Improvements
6518
6519     * A new scheme for accurately describing processor pipelines, the
6520       [5]DFA scheduler, has been added.
6521     * Pavel Nejedly, Charles University Prague, has contributed new file
6522       format used by the edge coverage profiler (-fprofile-arcs).
6523       The new format is robust and diagnoses common mistakes where
6524       profiles from different versions (or compilations) of the program
6525       are combined resulting in nonsensical profiles and slow code to
6526       produced with profile feedback. Additionally this format allows
6527       extra data to be gathered. Currently, overall statistics are
6528       produced helping optimizers to identify hot spots of a program
6529       globally replacing the old intra-procedural scheme and resulting in
6530       better code. Note that the gcov tool from older GCC versions will
6531       not be able to parse the profiles generated by GCC 3.3 and vice
6532       versa.
6533     * Jan Hubicka, SuSE Labs, has contributed a new superblock formation
6534       pass enabled using -ftracer. This pass simplifies the control flow
6535       of functions allowing other optimizations to do better job.
6536       He also contributed the function reordering pass
6537       (-freorder-functions) to optimize function placement using profile
6538       feedback.
6539
6540New Languages and Language specific improvements
6541
6542  C/ObjC/C++
6543
6544     * The preprocessor now accepts directives within macro arguments. It
6545       processes them just as if they had not been within macro arguments.
6546     * The separate ISO and traditional preprocessors have been completely
6547       removed. The front end handles either type of preprocessed output
6548       if necessary.
6549     * In C99 mode preprocessor arithmetic is done in the precision of the
6550       target's intmax_t, as required by that standard.
6551     * The preprocessor can now copy comments inside macros to the output
6552       file when the macro is expanded. This feature, enabled using the
6553       -CC option, is intended for use by applications which place
6554       metadata or directives inside comments, such as lint.
6555     * The method of constructing the list of directories to be searched
6556       for header files has been revised. If a directory named by a -I
6557       option is a standard system include directory, the option is
6558       ignored to ensure that the default search order for system
6559       directories and the special treatment of system header files are
6560       not defeated.
6561     * A few more [6]ISO C99 features now work correctly.
6562     * A new function attribute, nonnull, has been added which allows
6563       pointer arguments to functions to be specified as requiring a
6564       non-null value. The compiler currently uses this information to
6565       issue a warning when it detects a null value passed in such an
6566       argument slot.
6567     * A new type attribute, may_alias, has been added. Accesses to
6568       objects with types with this attribute are not subjected to
6569       type-based alias analysis, but are instead assumed to be able to
6570       alias any other type of objects, just like the char type.
6571
6572  C++
6573
6574     * Type based alias analysis has been implemented for C++ aggregate
6575       types.
6576
6577  Objective-C
6578
6579     * Generate an error if Objective-C objects are passed by value in
6580       function and method calls.
6581     * When -Wselector is used, check the whole list of selectors at the
6582       end of compilation, and emit a warning if a @selector() is not
6583       known.
6584     * Define __NEXT_RUNTIME__ when compiling for the NeXT runtime.
6585     * No longer need to include objc/objc-class.h to compile self calls
6586       in class methods (NeXT runtime only).
6587     * New -Wundeclared-selector option.
6588     * Removed selector bloating which was causing object files to be 10%
6589       bigger on average (GNU runtime only).
6590     * Using at run time @protocol() objects has been fixed in certain
6591       situations (GNU runtime only).
6592     * Type checking has been fixed and improved in many situations
6593       involving protocols.
6594
6595  Java
6596
6597     * The java.sql and javax.sql packages now implement the JDBC 3.0 (JDK
6598       1.4) API.
6599     * The JDK 1.4 assert facility has been implemented.
6600     * The bytecode interpreter is now direct threaded and thus faster.
6601
6602  Fortran
6603
6604     * Fortran improvements are listed in [7]the Fortran documentation.
6605
6606  Ada
6607
6608     * Ada tasking now works with glibc 2.3.x threading libraries.
6609
6610New Targets and Target Specific Improvements
6611
6612     * The following changes have been made to the HP-PA port:
6613          + The port now defaults to scheduling for the PA8000 series of
6614            processors.
6615          + Scheduling support for the PA7300 processor has been added.
6616          + The 32-bit port now supports weak symbols under HP-UX 11.
6617          + The handling of initializers and finalizers has been improved
6618            under HP-UX 11. The 64-bit port no longer uses collect2.
6619          + Dwarf2 EH support has been added to the 32-bit GNU/Linux port.
6620          + ABI fixes to correct the passing of small structures by value.
6621     * The SPARC, HP-PA, SH4, and x86/pentium ports have been converted to
6622       use the DFA processor pipeline description.
6623     * The following NetBSD configurations for the SuperH processor family
6624       have been added:
6625          + SH3, big-endian, sh-*-netbsdelf*
6626          + SH3, little-endian, shle-*-netbsdelf*
6627          + SH5, SHmedia, big-endian, 32-bit default, sh5-*-netbsd*
6628          + SH5, SHmedia, little-endian, 32-bit default, sh5le-*-netbsd*
6629          + SH5, SHmedia, big-endian, 64-bit default, sh64-*-netbsd*
6630          + SH5, SHmedia, little-endian, 64-bit default, sh64le-*-netbsd*
6631     * The following changes have been made to the IA-32/x86-64 port:
6632          + SSE2 and 3dNOW! intrinsics are now supported.
6633          + Support for thread local storage has been added to the IA-32
6634            and x86-64 ports.
6635          + The x86-64 port has been significantly improved.
6636     * The following changes have been made to the MIPS port:
6637          + All configurations now accept the -mabi switch. Note that you
6638            will need appropriate multilibs for this option to work
6639            properly.
6640          + ELF configurations will always pass an ABI flag to the
6641            assembler, except when the MIPS EABI is selected.
6642          + -mabi=64 no longer selects MIPS IV code.
6643          + The -mcpu option, which was deprecated in 3.1 and 3.2, has
6644            been removed from this release.
6645          + -march now changes the core ISA level. In previous releases,
6646            it would change the use of processor-specific extensions, but
6647            would leave the core ISA unchanged. For example, mips64-elf
6648            -march=r8000 will now generate MIPS IV code.
6649          + Under most configurations, -mipsN now acts as a synonym for
6650            -march.
6651          + There are some new preprocessor macros to describe the -march
6652            and -mtune settings. See the documentation of those options
6653            for details.
6654          + Support for the NEC VR-Series processors has been added. This
6655            includes the 54xx, 5500, and 41xx series.
6656          + Support for the Sandcraft sr71k processor has been added.
6657     * The following changes have been made to the S/390 port:
6658          + Support to build the Java runtime libraries has been added.
6659            Java is now enabled by default on s390-*-linux* and
6660            s390x-*-linux* targets.
6661          + Multilib support for the s390x-*-linux* target has been added;
6662            this allows to build 31-bit binaries using the -m31 option.
6663          + Support for thread local storage has been added.
6664          + Inline assembler code may now use the 'Q' constraint to
6665            specify memory operands without index register.
6666          + Various platform-specific performance improvements have been
6667            implemented; in particular, the compiler now uses the BRANCH
6668            ON COUNT family of instructions and makes more frequent use of
6669            the TEST UNDER MASK family of instructions.
6670     * The following changes have been made to the PowerPC port:
6671          + Support for IBM Power4 processor added.
6672          + Support for Motorola e500 SPE added.
6673          + Support for AIX 5.2 added.
6674          + Function and Data sections now supported on AIX.
6675          + Sibcall optimizations added.
6676     * The support for H8 Tiny is added to the H8/300 port with -mn.
6677
6678Obsolete Systems
6679
6680   Support for a number of older systems has been declared obsolete in GCC
6681   3.3. Unless there is activity to revive them, the next release of GCC
6682   will have their sources permanently removed.
6683
6684   All configurations of the following processor architectures have been
6685   declared obsolete:
6686     * Matsushita MN10200, mn10200-*-*
6687     * Motorola 88000, m88k-*-*
6688     * IBM ROMP, romp-*-*
6689
6690   Also, some individual systems have been obsoleted:
6691     * Alpha
6692          + Interix, alpha*-*-interix*
6693          + Linux libc1, alpha*-*-linux*libc1*
6694          + Linux ECOFF, alpha*-*-linux*ecoff*
6695     * ARM
6696          + Generic a.out, arm*-*-aout*
6697          + Conix, arm*-*-conix*
6698          + "Old ABI," arm*-*-oabi
6699          + StrongARM/COFF, strongarm-*-coff*
6700     * HPPA (PA-RISC)
6701          + Generic OSF, hppa1.0-*-osf*
6702          + Generic BSD, hppa1.0-*-bsd*
6703          + HP/UX versions 7, 8, and 9, hppa1.[01]-*-hpux[789]*
6704          + HiUX, hppa*-*-hiux*
6705          + Mach Lites, hppa*-*-lites*
6706     * Intel 386 family
6707          + Windows NT 3.x, i?86-*-win32
6708     * MC68000 family
6709          + HP systems, m68000-hp-bsd* and m68k-hp-bsd*
6710          + Sun systems, m68000-sun-sunos*, m68k-sun-sunos*, and
6711            m68k-sun-mach*
6712          + AT&T systems, m68000-att-sysv*
6713          + Atari systems, m68k-atari-sysv*
6714          + Motorola systems, m68k-motorola-sysv*
6715          + NCR systems, m68k-ncr-sysv*
6716          + Plexus systems, m68k-plexus-sysv*
6717          + Commodore systems, m68k-cbm-sysv*
6718          + Citicorp TTI, m68k-tti-*
6719          + Unos, m68k-crds-unos*
6720          + Concurrent RTU, m68k-ccur-rtu*
6721          + Linux a.out, m68k-*-linux*aout*
6722          + Linux libc1, m68k-*-linux*libc1*
6723          + pSOS, m68k-*-psos*
6724     * MIPS
6725          + Generic ECOFF, mips*-*-ecoff*
6726          + SINIX, mips-sni-sysv4
6727          + Orion RTEMS, mips64orion-*-rtems*
6728     * National Semiconductor 32000
6729          + OpenBSD, ns32k-*-openbsd*
6730     * POWER (aka RS/6000) and PowerPC
6731          + AIX versions 1, 2, and 3, rs6000-ibm-aix[123]*
6732          + Bull BOSX, rs6000-bull-bosx
6733          + Generic Mach, rs6000-*-mach*
6734          + Generic SysV, powerpc*-*-sysv*
6735          + Linux libc1, powerpc*-*-linux*libc1*
6736     * Sun SPARC
6737          + Generic a.out, sparc-*-aout*, sparclet-*-aout*,
6738            sparclite-*-aout*, and sparc86x-*-aout*
6739          + NetBSD a.out, sparc-*-netbsd*aout*
6740          + Generic BSD, sparc-*-bsd*
6741          + ChorusOS, sparc-*-chorusos*
6742          + Linux a.out, sparc-*-linux*aout*
6743          + Linux libc1, sparc-*-linux*libc1*
6744          + LynxOS, sparc-*-lynxos*
6745          + Solaris on HAL hardware, sparc-hal-solaris2*
6746          + SunOS versions 3 and 4, sparc-*-sunos[34]*
6747     * NEC V850
6748          + RTEMS, v850-*-rtems*
6749     * VAX
6750          + VMS, vax-*-vms*
6751
6752Documentation improvements
6753
6754Other significant improvements
6755
6756     * Almost all front-end dependencies in the compiler have been
6757       separated out into a set of language hooks. This should make adding
6758       a new front end clearer and easier.
6759     * One effect of removing the separate preprocessor is a small
6760       increase in the robustness of the compiler in general, and the
6761       maintainability of target descriptions. Previously target-specific
6762       built-in macros and others, such as __FAST_MATH__, had to be
6763       handled with so-called specs that were hard to maintain. Often they
6764       would fail to behave properly when conflicting options were
6765       supplied on the command line, and define macros in the user's
6766       namespace even when strict ISO compliance was requested.
6767       Integrating the preprocessor has cleanly solved these issues.
6768     * The Makefile suite now supports redirection of make install by
6769       means of the variable DESTDIR.
6770     __________________________________________________________________
6771
6772GCC 3.3
6773
6774   Detailed release notes for the GCC 3.3 release follow.
6775
6776  Bug Fixes
6777
6778    bootstrap failures
6779
6780     * [8]10140 cross compiler build failures: missing __mempcpy (DUP:
6781       [9]10198,[10]10338)
6782
6783    Internal compiler errors (multi-platform)
6784
6785     * [11]3581 large string causes segmentation fault in cc1
6786     * [12]4382 __builtin_{set,long}jmp with -O3 can crash the compiler
6787     * [13]5533 (c++) ICE when processing std::accumulate(begin, end,
6788       init, invalid_op)
6789     * [14]6387 -fpic -gdwarf-2 -g1 combination gives ICE in dwarf2out
6790     * [15]6412 (c++) ICE in retrieve_specialization
6791     * [16]6620 (c++) partial template specialization causes an ICE
6792       (segmentation fault)
6793     * [17]6663 (c++) ICE with attribute aligned
6794     * [18]7068 ICE with incomplete types
6795     * [19]7083 (c++) ICE using -gstabs with dodgy class derivation
6796     * [20]7647 (c++) ICE when data member has the name of the enclosing
6797       class
6798     * [21]7675 ICE in fixup_var_refs_1
6799     * [22]7718 'complex' template instantiation causes ICE
6800     * [23]8116 (c++) ICE in member template function
6801     * [24]8358 (ada) Ada compiler accesses freed memory, crashes
6802     * [25]8511 (c++) ICE: (hopefully) reproducible cc1plus segmentation
6803       fault
6804     * [26]8564 (c++) ICE in find_function_data, in function.c
6805     * [27]8660 (c++) template overloading ICE in tsubst_expr, in cp/pt.c
6806     * [28]8766 (c++) ICE after failed initialization of static template
6807       variable
6808     * [29]8803 ICE in instantiate_virtual_regs_1, in function.c
6809     * [30]8846 (c++) ICE after diagnostic if fr_FR@euro locale is set
6810     * [31]8906 (c++) ICE (Segmentation fault) when parsing nested-class
6811       definition
6812     * [32]9216 (c++) ICE on missing template parameter
6813     * [33]9261 (c++) ICE in arg_assoc, in cp/decl2.c
6814     * [34]9263 (fortran) ICE caused by invalid PARAMETER in implied DO
6815       loop
6816     * [35]9429 (c++) ICE in template instantiation with a pointered new
6817       operator
6818     * [36]9516 Internal error when using a big array
6819     * [37]9600 (c++) ICE with typedefs in template class
6820     * [38]9629 (c++) virtual inheritance segfault
6821     * [39]9672 (c++) ICE: Error reporting routines re-entered
6822     * [40]9749 (c++) ICE in write_expression on invalid function
6823       prototype
6824     * [41]9794 (fortran) ICE: floating point exception during constant
6825       folding
6826     * [42]9829 (c++) Missing colon in nested namespace usage causes ICE
6827     * [43]9916 (c++) ICE with noreturn function in ?: statement
6828     * [44]9936 ICE with local function and variable-length 2d array
6829     * [45]10262 (c++) cc1plus crashes with large generated code
6830     * [46]10278 (c++) ICE in parser for invalid code
6831     * [47]10446 (c++) ICE on definition of nonexistent member function of
6832       nested class in a class template
6833     * [48]10451 (c++) ICE in grokdeclarator on spurious mutable
6834       declaration
6835     * [49]10506 (c++) ICE in build_new at cp/init.c with
6836       -fkeep-inline-functions and multiple inheritance
6837     * [50]10549 (c++) ICE in store_bit_field on bitfields that exceed the
6838       precision of the declared type
6839
6840    Optimization bugs
6841
6842     * [51]2001 Inordinately long compile times in reload CSE regs
6843     * [52]2391 Exponential compilation time explosion in combine
6844     * [53]2960 Duplicate loop conditions even with -Os
6845     * [54]4046 redundant conditional branch
6846     * [55]6405 Loop-unrolling related performance regressions
6847     * [56]6798 very long compile time with large case-statement
6848     * [57]6871 const objects shouldn't be moved to .bss
6849     * [58]6909 problem w/ -Os on modified loop-2c.c test case
6850     * [59]7189 gcc -O2 -Wall does not print ``control reaches end of
6851       non-void function'' warning
6852     * [60]7642 optimization problem with signbit()
6853     * [61]8634 incorrect code for inlining of memcpy under -O2
6854     * [62]8750 Cygwin prolog generation erroneously emitting __alloca as
6855       regular function call
6856
6857    C front end
6858
6859     * [63]2161 long if-else cascade overflows parser stack
6860     * [64]4319 short accepted on typedef'd char
6861     * [65]8602 incorrect line numbers in warning messages when using
6862       inline functions
6863     * [66]9177 -fdump-translation-unit: C front end deletes function_decl
6864       AST nodes and breaks debugging dumps
6865     * [67]9853 miscompilation of non-constant structure initializer
6866
6867    c++ compiler and library
6868
6869     * [68]45 legal template specialization code is rejected (DUP:
6870       [69]3784)
6871     * [70]764 lookup failure: friend operator and dereferencing a pointer
6872       and templates (DUP: [71]5116)
6873     * [72]2862 gcc accepts invalid explicit instantiation syntax (DUP:
6874       2863)
6875     * [73]3663 G++ doesn't check access control during template
6876       instantiation
6877     * [74]3797 gcc fails to emit explicit specialization of a template
6878       member
6879     * [75]3948 Two destructors are called when no copy destructor is
6880       defined (ABI change)
6881     * [76]4137 Conversion operator within template is not accepted
6882     * [77]4361 bogus ambiguity taking the address of a member template
6883     * [78]4802 g++ accepts illegal template code (access to private
6884       member; DUP: [79]5837)
6885     * [80]4803 inline function is used but never defined, and g++ does
6886       not object
6887     * [81]5094 Partial specialization cannot be friend?
6888     * [82]5730 complex<double>::norm() -- huge slowdown from egcs-2.91.66
6889     * [83]6713 Regression wrt 3.0.4: g++ -O2 leads to seg fault at run
6890       time
6891     * [84]7015 certain __asm__ constructs rejected
6892     * [85]7086 compile time regression (quadratic behavior in
6893       fixup_var_refs)
6894     * [86]7099 G++ doesn't set the noreturn attribute on std::exit and
6895       std::abort
6896     * [87]7247 copy constructor missing when inlining enabled (invalid
6897       optimization?)
6898     * [88]7441 string array initialization compilation time regression
6899       from seconds to minutes
6900     * [89]7768 __PRETTY_FUNCTION__ for template destructor is wrong
6901     * [90]7804 bad printing of floating point constant in warning message
6902     * [91]8099 Friend classes and template specializations
6903     * [92]8117 member function pointers and multiple inheritance
6904     * [93]8205 using declaration and multiple inheritance
6905     * [94]8645 unnecessary non-zero checks in stl_tree.h
6906     * [95]8724 explicit destructor call for incomplete class allowed
6907     * [96]8805 compile time regression with many member variables
6908     * [97]8691 -O3 and -fno-implicit-templates are incompatible
6909     * [98]8700 unhelpful error message for binding temp to reference
6910     * [99]8724 explicit destructor call for incomplete class allowed
6911     * [100]8949 numeric_limits<>::denorm_min() and is_iec559 problems
6912     * [101]9016 Failure to consistently constant fold "constant" C++
6913       objects
6914     * [102]9053 g++ confused about ambiguity of overloaded function
6915       templates
6916     * [103]9152 undefined virtual thunks
6917     * [104]9182 basic_filebuf<> does not report errors in codecvt<>::out
6918     * [105]9297 data corruption due to codegen bug (when copying.)
6919     * [106]9318 i/ostream::operator>>/<<(streambuf*) broken
6920     * [107]9320 Incorrect usage of traits_type::int_type in stdio_filebuf
6921     * [108]9400 bogus -Wshadow warning: shadowed declaration of this in
6922       local classes
6923     * [109]9424 i/ostream::operator>>/<<(streambuf*) drops characters
6924     * [110]9425 filebuf::pbackfail broken (DUP: [111]9439)
6925     * [112]9474 GCC freezes in compiling a weird code mixing <iostream>
6926       and <iostream.h>
6927     * [113]9548 Incorrect results from setf(ios::fixed) and precision(-1)
6928       [114][DR 231]
6929     * [115]9555 ostream inserters fail to set badbit on exception
6930     * [116]9561 ostream inserters rethrow exception of wrong type
6931     * [117]9563 ostream::sentry returns true after a failed preparation
6932     * [118]9582 one-definition rule violation in std::allocator
6933     * [119]9622 __PRETTY_FUNCTION__ incorrect in template destructors
6934     * [120]9683 bug in initialization chains for static const variables
6935       from template classes
6936     * [121]9791 -Woverloaded-virtual reports hiding of destructor
6937     * [122]9817 collate::compare doesn't handle nul characters
6938     * [123]9825 filebuf::sputbackc breaks sbumpc
6939     * [124]9826 operator>>(basic_istream, basic_string) fails to compile
6940       with custom traits
6941     * [125]9924 Multiple using statements for builtin functions not
6942       allowed
6943     * [126]9946 destructor is not called for temporary object
6944     * [127]9964 filebuf::close() sometimes fails to close file
6945     * [128]9988 filebuf::overflow writes EOF to file
6946     * [129]10033 optimization breaks polymorphic references w/ typeid
6947       operator
6948     * [130]10097 filebuf::underflow drops characters
6949     * [131]10132 filebuf destructor can throw exceptions
6950     * [132]10180 gcc fails to warn about non-inlined function
6951     * [133]10199 method parametrized by template does not work everywhere
6952     * [134]10300 use of array-new (nothrow) in segfaults on NULL return
6953     * [135]10427 Stack corruption with variable-length automatic arrays
6954       and virtual destructors
6955     * [136]10503 Compilation never stops in fixed_type_or_null
6956
6957    Objective-C
6958
6959     * [137]5956 selectors aren't matched properly when added to the
6960       selector table
6961
6962    Fortran compiler and library
6963
6964     * [138]1832 list directed i/o overflow hangs, -fbounds-check doesn't
6965       detect
6966     * [139]3924 g77 generates code that is rejected by GAS if COFF debug
6967       info requested
6968     * [140]5634 doc: explain that configure --prefix=~/... does not work
6969     * [141]6367 multiple repeat counts confuse namelist read into array
6970     * [142]6491 Logical operations error on logicals when using
6971       -fugly-logint
6972     * [143]6742 Generation of C++ Prototype for FORTRAN and extern "C"
6973     * [144]7113 Failure of g77.f-torture/execute/f90-intrinsic-bit.f -Os
6974       on irix6.5
6975     * [145]7236 OPEN(...,RECL=nnn,...) without ACCESS='DIRECT' should
6976       assume a direct access file
6977     * [146]7278 g77 "bug"; the executable misbehaves (with -O2
6978       -fno-automatic)
6979     * [147]7384 DATE_AND_TIME milliseconds field inactive on Windows
6980     * [148]7388 Incorrect output with 0-based array of characters
6981     * [149]8587 Double complex zero ** double precision number -> NaN
6982       instead of zero
6983     * [150]9038 -ffixed-line-length-none -x f77-cpp-input gives: Warning:
6984       unknown register name line-length-none
6985     * [151]10197 Direct access files not unformatted by default
6986
6987    Java compiler and library
6988
6989     * [152]6005 gcj fails to build rhug on alpha
6990     * [153]6389 System.getProperty("") should always throw an
6991       IllegalArgumentException
6992     * [154]6576 java.util.ResourceBundle.getResource ignores locale
6993     * [155]6652 new java.io.File("").getCanonicalFile() throws exception
6994     * [156]7060 getMethod() doesn't search super interface
6995     * [157]7073 bytecode interpreter gives wrong answer for interface
6996       getSuperclass()
6997     * [158]7180 possible bug in
6998       javax.naming.spi.NamingManager.getPlusPath()
6999     * [159]7416 java.security startup refs "GNU libgcj.security"
7000     * [160]7570 Runtime.exec with null envp: child doesn't inherit parent
7001       env (DUP: [161]7578)
7002     * [162]7611 Internal error while compiling libjava with -O
7003     * [163]7709 NullPointerException in _Jv_ResolvePoolEntry
7004     * [164]7766 ZipInputStream.available returns 0 immediately after
7005       construction
7006     * [165]7785 Calendar.getTimeInMillis/setTimeInMillis should be public
7007     * [166]7786 TimeZone.getDSTSavings() from JDK1.4 not implemented
7008     * [167]8142 '$' in class names vs. dlopen 'dynamic string tokens'
7009     * [168]8234 ZipInputStream chokes when InputStream.read() returns
7010       small chunks
7011     * [169]8415 reflection bug: exception info for Method
7012     * [170]8481 java.Random.nextInt(int) may return negative
7013     * [171]8593 Error reading GZIPped files with BufferedReader
7014     * [172]8759 java.beans.Introspector has no flushCaches() or
7015       flushFromCaches() methods
7016     * [173]8997 spin() calls Thread.sleep
7017     * [174]9253 on win32, java.io.File.listFiles("C:\\") returns pwd
7018       instead of the root content of C:
7019     * [175]9254 java::lang::Object::wait(), threads-win32.cc returns
7020       wrong return codes
7021     * [176]9271 Severe bias in java.security.SecureRandom
7022
7023    Ada compiler and library
7024
7025     * [177]6767 make gnatlib-shared fails on -laddr2line
7026     * [178]9911 gnatmake fails to link when GCC configured with
7027       --with-sjlj-exceptions=yes
7028     * [179]10020 Can't bootstrap gcc on AIX with Ada enabled
7029     * [180]10546 Ada tasking not working on Red Hat 9
7030
7031    preprocessor
7032
7033     * [181]7029 preprocessor should ignore #warning with -M
7034
7035    ARM-specific
7036
7037     * [182]2903 [arm] Optimization bug with long long arithmetic
7038     * [183]7873 arm-linux-gcc fails when assigning address to a bit field
7039
7040    FreeBSD-specific
7041
7042     * [184]7680 float functions undefined in math.h/cmath with #define
7043       _XOPEN_SOURCE
7044
7045    HP-UX or HP-PA-specific
7046
7047     * [185]8705 [HP-PA] ICE in emit_move_insn_1, in expr.c
7048     * [186]9986 [HP-UX] Incorrect transformation of fputs_unlocked to
7049       fputc_unlocked
7050     * [187]10056 [HP-PA] ICE at -O2 when building c++ code from doxygen
7051
7052    m68hc11-specific
7053
7054     * [188]6744 Bad assembler code generated: reference to pseudo
7055       register z
7056     * [189]7361 Internal compiler error in reload_cse_simplify_operands,
7057       in reload1.c
7058
7059    MIPS-specific
7060
7061     * [190]9496 [mips-linux] bug in optimizer?
7062
7063    PowerPC-specific
7064
7065     * [191]7067 -Os with -mcpu=powerpc optimizes for speed (?) instead of
7066       space
7067     * [192]8480 reload ICEs for LAPACK code on powerpc64-linux
7068     * [193]8784 [AIX] Internal compiler error in simplify_gen_subreg
7069     * [194]10315 [powerpc] ICE: in extract_insn, in recog.c
7070
7071    SPARC-specific
7072
7073     * [195]10267 (documentation) Wrong build instructions for
7074       *-*-solaris2*
7075
7076    x86-specific (Intel/AMD)
7077
7078     * [196]7916 ICE in instantiate_virtual_register_1
7079     * [197]7926 (c++) i486 instructions in header files make c++ programs
7080       crash on i386
7081     * [198]8555 ICE in gen_split_1231
7082     * [199]8994 ICE with -O -march=pentium4
7083     * [200]9426 ICE with -fssa -funroll-loops -fprofile-arcs
7084     * [201]9806 ICE in inline assembly with -fPIC flag
7085     * [202]10077 gcc -msse2 generates movd to move dwords between xmm
7086       regs
7087     * [203]10233 64-bit comparison only comparing bottom 32-bits
7088     * [204]10286 type-punning doesn't work with __m64 and -O
7089     * [205]10308 [x86] ICE with -O -fgcse or -O2
7090     __________________________________________________________________
7091
7092GCC 3.3.1
7093
7094  Bug Fixes
7095
7096   This section lists the problem reports (PRs) from GCC's bug tracking
7097   system that are known to be fixed in the 3.3.1 release. This list might
7098   not be complete (that is, it is possible that some PRs that have been
7099   fixed are not listed here).
7100
7101    Bootstrap failures
7102
7103     * [206]11272 [Solaris] make bootstrap fails while building libstdc++
7104
7105    Internal compiler errors (multi-platform)
7106
7107     * [207]5754 ICE on invalid nested template class
7108     * [208]6597 ICE in set_mem_alias_set compiling Qt with -O2 on ia64
7109       and --enable-checking
7110     * [209]6949 (c++) ICE in tsubst_decl, in cp/pt.c
7111     * [210]7053 (c++) ICE when declaring a function already defined as a
7112       friend method of a template class
7113     * [211]8164 (c++) ICE when using different const expressions as
7114       template parameter
7115     * [212]8384 (c++) ICE in is_base_type, in dwarf2out.c
7116     * [213]9559 (c++) ICE with invalid initialization of a static const
7117     * [214]9649 (c++) ICE in finish_member_declaration, in cp/semantics.c
7118       when redeclaring a static member variable
7119     * [215]9864 (fortran) ICE in add_abstract_origin_attribute, in
7120       dwarfout.c with -g -O -finline-functions
7121     * [216]10432 (c++) ICE in poplevel, in cp/decl.c
7122     * [217]10475 ICE in subreg_highpart_offset for code with long long
7123     * [218]10635 (c++) ICE when dereferencing an incomplete type casted
7124       from a void pointer
7125     * [219]10661 (c++) ICE in instantiate_decl, in cp/pt.c while
7126       instantiating static member variables
7127     * [220]10700 ICE in copy_to_mode_reg on 64-bit targets
7128     * [221]10712 (c++) ICE in constructor_name_full, in cp/decl2.c
7129     * [222]10796 (c++) ICE when defining an enum with two values: -1 and
7130       MAX_INT_64BIT
7131     * [223]10890 ICE in merge_assigned_reloads building Linux 2.4.2x
7132       sched.c
7133     * [224]10939 (c++) ICE with template code
7134     * [225]10956 (c++) ICE when specializing a template member function
7135       of a template class, in tsubst, in cp/pt.c
7136     * [226]11041 (c++) ICE: const myclass &x = *x; (when operator*()
7137       defined)
7138     * [227]11059 (c++) ICE with empty union
7139     * [228]11083 (c++) ICE in commit_one_edge_insertion, in cfgrtl.c with
7140       -O2 -fnon-call-exceptions
7141     * [229]11105 (c++) ICE in mangle_conv_op_name_for_type
7142     * [230]11149 (c++) ICE on error when instantiation with call function
7143       of a base type
7144     * [231]11228 (c++) ICE on new-expression using array operator new and
7145       default-initialization
7146     * [232]11282 (c++) Infinite memory usage after syntax error
7147     * [233]11301 (fortran) ICE with -fno-globals
7148     * [234]11308 (c++) ICE when using an enum type name as if it were a
7149       class or namespace
7150     * [235]11473 (c++) ICE with -gstabs when empty struct inherits from
7151       an empty struct
7152     * [236]11503 (c++) ICE when instantiating template with ADDR_EXPR
7153     * [237]11513 (c++) ICE in push_template_decl_real, in cp/pt.c:
7154       template member functions
7155
7156    Optimization bugs
7157
7158     * [238]11198 -O2 -frename-registers generates wrong code (aliasing
7159       problem)
7160     * [239]11304 Wrong code production with -fomit-frame-pointer
7161     * [240]11381 volatile memory access optimized away
7162     * [241]11536 [strength-reduce] -O2 optimization produces wrong code
7163     * [242]11557 constant folding bug generates wrong code
7164
7165    C front end
7166
7167     * [243]5897 No warning for statement after return
7168     * [244]11279 DWARF-2 output mishandles large enums
7169
7170    Preprocessor bugs
7171
7172     * [245]11022 no warning for non-compatible macro redefinition
7173
7174    C++ compiler and library
7175
7176     * [246]2330 static_cast<>() to a private base is allowed
7177     * [247]5388 Incorrect message "operands to ?: have different types"
7178     * [248]5390 Libiberty fails to demangle multi-digit template
7179       parameters
7180     * [249]7877 Incorrect parameter passing to specializations of member
7181       function templates
7182     * [250]9393 Anonymous namespaces and compiling the same file twice
7183     * [251]10032 -pedantic converts some errors to warnings
7184     * [252]10468 const typeof(x) is non-const, but only in templates
7185     * [253]10527 confused error message with "new int()" parameter
7186       initializer
7187     * [254]10679 parameter MIN_INLINE_INSNS is not honored
7188     * [255]10682 gcc chokes on a typedef for an enum inside a class
7189       template
7190     * [256]10689 pow(std::complex(0),1/3) returns (nan, nan) instead of
7191       0.
7192     * [257]10845 template member function (with nested template as
7193       parameter) cannot be called anymore if another unrelated template
7194       member function is defined
7195     * [258]10849 Cannot define an out-of-class specialization of a
7196       private nested template class
7197     * [259]10888 Suppress -Winline warnings for system headers
7198     * [260]10929 -Winline warns about functions for which no definition
7199       is visible
7200     * [261]10931 valid conversion static_cast<const unsigned
7201       int&>(lvalue-of-type-int) is rejected
7202     * [262]10940 Bad code with explicit specialization
7203     * [263]10968 If member function implicitly instantiated, explicit
7204       instantiation of class fails to instantiate it
7205     * [264]10990 Cannot convert with dynamic_cast<> to a private base
7206       class from within a member function
7207     * [265]11039 Bad interaction between implicit typename deprecation
7208       and friendship
7209     * [266]11062 (libstdc++) avoid __attribute__ ((unused)); say
7210       "__unused__" instead
7211     * [267]11095 C++ iostream manipulator causes segfault when called
7212       with negative argument
7213     * [268]11098 g++ doesn't emit complete debugging information for
7214       local variables in destructors
7215     * [269]11137 GNU/Linux shared library constructors not called unless
7216       there's one global object
7217     * [270]11154 spurious ambiguity report for template class
7218       specialization
7219     * [271]11329 Compiler cannot find user defined implicit typecast
7220     * [272]11332 Spurious error with casts in ?: expression
7221     * [273]11431 static_cast behavior with subclasses when default
7222       constructor available
7223     * [274]11528 money_get facet does not accept "$.00" as valid
7224     * [275]11546 Type lookup problems in out-of-line definition of a
7225       class doubly nested from a template class
7226     * [276]11567 C++ code containing templated member function with same
7227       name as pure virtual member function results in linking failure
7228     * [277]11645 Failure to deal with using and private inheritance
7229
7230    Java compiler and library
7231
7232     * [278]5179 Qualified static field access doesn't initialize its
7233       class
7234     * [279]8204 gcj -O2 to native reorders certain instructions
7235       improperly
7236     * [280]10838 java.io.ObjectInputStream syntax error
7237     * [281]10886 The RMI registry that comes with GCJ does not work
7238       correctly
7239     * [282]11349 JNDI URL context factories not located correctly
7240
7241    x86-specific (Intel/AMD)
7242
7243     * [283]4823 ICE on inline assembly code
7244     * [284]8878 miscompilation with -O and SSE
7245     * [285]9815 (c++ library) atomicity.h - fails to compile with -O3
7246       -masm=intel
7247     * [286]10402 (inline assembly) [x86] ICE in merge_assigned_reloads,
7248       in reload1.c
7249     * [287]10504 ICE with SSE2 code and -O3 -mcpu=pentium4 -msse2
7250     * [288]10673 ICE for x86-64 on freebsd libc vfprintf.c source
7251     * [289]11044 [x86] out of range loop instructions for FP code on K6
7252     * [290]11089 ICE: instantiate_virtual_regs_lossage while using SSE
7253       built-ins
7254     * [291]11420 [x86_64] gcc generates invalid asm code when "-O -fPIC"
7255       is used
7256
7257    SPARC- or Solaris- specific
7258
7259     * [292]9362 solaris 'as' dies when fed .s and "-gstabs"
7260     * [293]10142 [SPARC64] gcc produces wrong code when passing
7261       structures by value
7262     * [294]10663 New configure check aborts with Sun tools.
7263     * [295]10835 combinatorial explosion in scheduler on HyperSPARC
7264     * [296]10876 ICE in calculate_giv_inc when building KDE
7265     * [297]10955 wrong code at -O3 for structure argument in context of
7266       structure return
7267     * [298]11018 -mcpu=ultrasparc busts tar-1.13.25
7268     * [299]11556 [sparc64] ICE in gen_reg_rtx() while compiling 2.6.x
7269       Linux kernel
7270
7271    ia64 specific
7272
7273     * [300]10907 gcc violates the ia64 ABI (GP must be preserved)
7274     * [301]11320 scheduler bug (in machine depended reorganization pass)
7275     * [302]11599 bug with conditional and __builtin_prefetch
7276
7277    PowerPC specific
7278
7279     * [303]9745 [powerpc] gcc mis-compiles libmcrypt (alias problem
7280       during loop)
7281     * [304]10871 error in rs6000_stack_info save_size computation
7282     * [305]11440 gcc mis-compiles c++ code (libkhtml) with -O2, -fno-gcse
7283       cures it
7284
7285    m68k-specific
7286
7287     * [306]7594 [m68k] ICE on legal code associated with simplify-rtx
7288     * [307]10557 [m68k] ICE in subreg_offset_representable_p
7289     * [308]11054 [m68k] ICE in reg_overlap_mentioned_p
7290
7291    ARM-specific
7292
7293     * [309]10834 [arm] GCC 3.3 still generates incorrect instructions for
7294       functions with __attribute__ ((interrupt ("IRQ")))
7295     * [310]10842 [arm] Clobbered link register is copied to pc under
7296       certain circumstances
7297     * [311]11052 [arm] noce_process_if_block() can lose REG_INC notes
7298     * [312]11183 [arm] ICE in change_address_1 (3.3) / subreg_hard_regno
7299       (3.4)
7300
7301    MIPS-specific
7302
7303     * [313]11084 ICE in propagate_one_insn, in flow.c
7304
7305    SH-specific
7306
7307     * [314]10331 can't compile c++ part of gcc cross compiler for sh-elf
7308     * [315]10413 [SH] ICE in reload_cse_simplify_operands, in reload1.c
7309     * [316]11096 i686-linux to sh-linux cross compiler fails to compile
7310       C++ files
7311
7312    GNU/Linux (or Hurd?) specific
7313
7314     * [317]2873 Bogus fixinclude of stdio.h from glibc 2.2.3
7315
7316    UnixWare specific
7317
7318     * [318]3163 configure bug: gcc/aclocal.m4 mmap test fails on UnixWare
7319       7.1.1
7320
7321    Cygwin (or mingw) specific
7322
7323     * [319]5287 ICE with dllimport attribute
7324     * [320]10148 [MingW/CygWin] Compiler dumps core
7325
7326    DJGPP specific
7327
7328     * [321]8787 GCC fails to emit .intel_syntax when invoked with
7329       -masm=intel on DJGPP
7330
7331    Darwin (and MacOS X) specific
7332
7333     * [322]10900 trampolines crash
7334
7335    Documentation
7336
7337     * [323]1607 (c++) Format attributes on methods undocumented
7338     * [324]4252 Invalid option `-fdump-translation-unit'
7339     * [325]4490 Clarify restrictions on -m96bit-long-double,
7340       -m128bit-long-double
7341     * [326]10355 document an issue with regparm attribute on some systems
7342       (e.g. Solaris)
7343     * [327]10726 (fortran) Documentation for function "IDate Intrinsic
7344       (Unix)" is wrong
7345     * [328]10805 document bug in old version of Sun assembler
7346     * [329]10815 warn against GNU binutils on AIX
7347     * [330]10877 document need for newer binutils on i?86-*-linux-gnu
7348     * [331]11280 Manual incorrect with respect to -freorder-blocks
7349     * [332]11466 Document -mlittle-endian and its restrictions for the
7350       sparc64 port
7351
7352    Testsuite bugs (compiler itself is not affected)
7353
7354     * [333]10737 newer bison causes g++.dg/parse/crash2.C to incorrectly
7355       report failure
7356     * [334]10810 gcc-3.3 fails make check: buffer overrun in
7357       test_demangle.c
7358     __________________________________________________________________
7359
7360GCC 3.3.2
7361
7362  Bug Fixes
7363
7364   This section lists the problem reports (PRs) from [335]GCC's bug
7365   tracking system that are known to be fixed in the 3.3.2 release. This
7366   list might not be complete (that is, it is possible that some PRs that
7367   have been fixed are not listed here).
7368
7369    Bootstrap failures and problems
7370
7371     * [336]8336 [SCO5] bootstrap config still tries to use COFF options
7372     * [337]9330 [alpha-osf] Bootstrap failure on Compaq Tru64 with
7373       --enable-threads=posix
7374     * [338]9631 [hppa64-linux] gcc-3.3 fails to bootstrap
7375     * [339]9877 fixincludes makes a bad sys/byteorder.h on svr5 (UnixWare
7376       7.1.1)
7377     * [340]11687 xstormy16-elf build fails in libf2c
7378     * [341]12263 [SGI IRIX] bootstrap fails during compile of
7379       libf2c/libI77/backspace.c
7380     * [342]12490 buffer overflow in scan-decls.c (during Solaris 9
7381       fix-header processing)
7382
7383    Internal compiler errors (multi-platform)
7384
7385     * [343]7277 Casting integers to vector types causes ICE
7386     * [344]7939 (c++) ICE on invalid function template specialization
7387     * [345]11063 (c++) ICE on parsing initialization list of const array
7388       member
7389     * [346]11207 ICE with negative index in array element designator
7390     * [347]11522 (fortran) g77 dwarf-2 ICE in
7391       add_abstract_origin_attribute
7392     * [348]11595 (c++) ICE on duplicate label definition
7393     * [349]11646 (c++) ICE in commit_one_edge_insertion with
7394       -fnon-call-exceptions -fgcse -O
7395     * [350]11665 ICE in struct initializer when taking address
7396     * [351]11852 (c++) ICE with bad struct initializer.
7397     * [352]11878 (c++) ICE in cp_expr_size
7398     * [353]11883 ICE with any -O on mercury-generated C code
7399     * [354]11991 (c++) ICE in cxx_incomplete_type_diagnostic, in
7400       cp/typeck2.c when applying typeid operator to template template
7401       parameter
7402     * [355]12146 ICE in lookup_template_function, in cp/pt.c
7403     * [356]12215 ICE in make_label_edge with -fnon-call-exceptions
7404       -fno-gcse -O2
7405     * [357]12369 (c++) ICE with templates and friends
7406     * [358]12446 ICE in emit_move_insn on complicated array reference
7407     * [359]12510 ICE in final_scan_insn
7408     * [360]12544 ICE with large parameters used in nested functions
7409
7410    C and optimization bugs
7411
7412     * [361]9862 spurious warnings with -W -finline-functions
7413     * [362]10962 lookup_field is a linear search on a linked list (can be
7414       slow if large struct)
7415     * [363]11370 -Wunreachable-code gives false complaints
7416     * [364]11637 invalid assembly with -fnon-call-exceptions
7417     * [365]11885 Problem with bitfields in packed structs
7418     * [366]12082 Inappropriate unreachable code warnings
7419     * [367]12180 Inline optimization fails for variadic function
7420     * [368]12340 loop unroller + gcse produces wrong code
7421
7422    C++ compiler and library
7423
7424     * [369]3907 nested template parameter collides with member name
7425     * [370]5293 confusing message when binding a temporary to a reference
7426     * [371]5296 [DR115] Pointers to functions and to template functions
7427       behave differently in deduction
7428     * [372]7939 ICE on function template specialization
7429     * [373]8656 Unable to assign function with __attribute__ and pointer
7430       return type to an appropriate variable
7431     * [374]10147 Confusing error message for invalid template function
7432       argument
7433     * [375]11400 std::search_n() makes assumptions about Size parameter
7434     * [376]11409 issues with using declarations, overloading, and
7435       built-in functions
7436     * [377]11740 ctype<wchar_t>::do_is(mask, wchar_t) doesn't handle
7437       multiple bits in mask
7438     * [378]11786 operator() call on variable in other namespace not
7439       recognized
7440     * [379]11867 static_cast ignores ambiguity
7441     * [380]11928 bug with conversion operators that are typedefs
7442     * [381]12114 Uninitialized memory accessed in dtor
7443     * [382]12163 static_cast + explicit constructor regression
7444     * [383]12181 Wrong code with comma operator and c++
7445     * [384]12236 regparm and fastcall messes up parameters
7446     * [385]12266 incorrect instantiation of unneeded template during
7447       overload resolution
7448     * [386]12296 istream::peek() doesn't set eofbit
7449     * [387]12298 [sjlj exceptions] Stack unwind destroys
7450       not-yet-constructed object
7451     * [388]12369 ICE with templates and friends
7452     * [389]12337 apparently infinite loop in g++
7453     * [390]12344 stdcall attribute ignored if function returns a pointer
7454     * [391]12451 missing(late) class forward declaration in cxxabi.h
7455     * [392]12486 g++ accepts invalid use of a qualified name
7456
7457    x86 specific (Intel/AMD)
7458
7459     * [393]8869 [x86 MMX] ICE with const variable optimization and MMX
7460       builtins
7461     * [394]9786 ICE in fixup_abnormal_edges with -fnon-call-exceptions
7462       -O2
7463     * [395]11689 g++3.3 emits un-assembleable code for k6 architecture
7464     * [396]12116 [k6] Invalid assembly output values with X-MAME code
7465     * [397]12070 ICE converting between double and long double with
7466       -msoft-float
7467
7468    ia64-specific
7469
7470     * [398]11184 [ia64 hpux] ICE on __builtin_apply building libobjc
7471     * [399]11535 __builtin_return_address may not work on ia64
7472     * [400]11693 [ia64] ICE in gen_nop_type
7473     * [401]12224 [ia64] Thread-local storage doesn't work
7474
7475    PowerPC-specific
7476
7477     * [402]11087 [powerpc64-linux] GCC miscompiles raid1.c from linux
7478       kernel
7479     * [403]11319 loop miscompiled on ppc32
7480     * [404]11949 ICE Compiler segfault with ffmpeg -maltivec code
7481
7482    SPARC-specific
7483
7484     * [405]11662 wrong code for expr. with cast to long long and
7485       exclusive or
7486     * [406]11965 invalid assembler code for a shift < 32 operation
7487     * [407]12301 (c++) stack corruption when a returned expression throws
7488       an exception
7489
7490    Alpha-specific
7491
7492     * [408]11717 [alpha-linux] unrecognizable insn compiling for.c of
7493       kernel 2.4.22-pre8
7494
7495    HPUX-specific
7496
7497     * [409]11313 problem with #pragma weak and static inline functions
7498     * [410]11712 __STDC_EXT__ not defined for C++ by default anymore?
7499
7500    Solaris specific
7501
7502     * [411]12166 Profiled programs crash if PROFDIR is set
7503
7504    Solaris-x86 specific
7505
7506     * [412]12101 i386 Solaris no longer works with GNU as?
7507
7508    Miscellaneous embedded target-specific bugs
7509
7510     * [413]10988 [m32r-elf] wrong blockmove code with -O3
7511     * [414]11805 [h8300-unknown-coff] [H8300] ICE for simple code with
7512       -O2
7513     * [415]11902 [sh4] spec file improperly inserts rpath even when none
7514       needed
7515     * [416]11903 [sh4] -pthread fails to link due to error in spec file
7516       on sh4
7517     __________________________________________________________________
7518
7519GCC 3.3.3
7520
7521  Minor features
7522
7523   In addition to the bug fixes documented below, this release contains
7524   few minor features such as:
7525     * Support for --with-sysroot
7526     * Support for automatic detection of executable stacks
7527     * Support for SSE3 instructions
7528     * Support for thread local storage debugging under GDB on S390
7529
7530  Bug Fixes
7531
7532   This section lists the problem reports (PRs) from [417]GCC's bug
7533   tracking system that are known to be fixed in the 3.3.3 release. This
7534   list might not be complete (that is, it is possible that some PRs that
7535   have been fixed are not listed here).
7536
7537    Bootstrap failures and issues
7538
7539     * [418]11890 Building cross gcc-3.3.1 for sparc-sun-solaris2.6 fails
7540     * [419]12399 boehm-gc fails (when building a cross compiler): libtool
7541       unable to infer tagged configuration
7542     * [420]13068 mklibgcc.in doesn't handle multi-level multilib
7543       subdirectories properly
7544
7545    Internal compiler errors (multi-platform)
7546
7547     * [421]10060 ICE (stack overflow) on huge file (300k lines) due to
7548       recursive behaviour of copy_rtx_if_shared, in emit_rtl.c
7549     * [422]10555 (c++) ICE on undefined template argument
7550     * [423]10706 (c++) ICE in mangle_class_name_for_template
7551     * [424]11496 (fortran) error in flow_loops_find when -funroll-loops
7552       active
7553     * [425]11741 ICE in pre_insert_copy_insn, in gcse.c
7554     * [426]12440 GCC crashes during compilation of quicktime4linux 2.0.0
7555     * [427]12632 (fortran) -fbounds-check ICE
7556     * [428]12712 (c++) ICE on short legit C++ code fragment with gcc
7557       3.3.2
7558     * [429]12726 (c++) ICE (segfault) on trivial code
7559     * [430]12890 (c++) ICE on compilation of class with throwing method
7560     * [431]12900 (c++) ICE in rtl_verify_flow_info_1
7561     * [432]13060 (fortran) ICE in fixup_var_refs_1, in function.c on
7562       correct code with -O2 -fno-force-mem
7563     * [433]13289 (c++) ICE in regenerate_decl_from_template on recursive
7564       template
7565     * [434]13318 ICE: floating point exception in the loop optimizer
7566     * [435]13392 (c++) ICE in convert_from_eh_region_ranges_1, in
7567       except.c
7568     * [436]13574 (c++) invalid array default initializer in class lets
7569       gcc consume all memory and die
7570     * [437]13475 ICE on SIMD variables with partial value initialization
7571     * [438]13797 (c++) ICE on invalid template parameter
7572     * [439]13824 (java) gcj SEGV with simple .java program
7573
7574    C and optimization bugs
7575
7576     * [440]8776 loop invariants are not removed (most likely)
7577     * [441]10339 [sparc,ppc,ppc64] Invalid optimization: replacing
7578       strncmp by memcmp
7579     * [442]11350 undefined labels with -Os -fPIC
7580     * [443]12826 Optimizer removes reference through volatile pointer
7581     * [444]12500 stabs debug info: void no longer a predefined / builtin
7582       type
7583     * [445]12941 builtin-bitops-1.c miscompilation (latent bug)
7584     * [446]12953 tree inliner bug (in inline_forbidden_p) and fix
7585     * [447]13041 linux-2.6/sound/core/oss/rate.c miscompiled
7586     * [448]13507 spurious printf format warning
7587     * [449]13382 Type information for const pointer disappears during
7588       optimization.
7589     * [450]13394 noreturn attribute ignored on recursive invokation
7590     * [451]13400 Compiled code crashes storing to read-only location
7591     * [452]13521 Endless loop in calculate_global_regs_live
7592
7593    C++ compiler and library
7594
7595   Some of the bug fixes in this list were made to implement decisions
7596   that the ISO C++ standards committee has made concerning several defect
7597   reports (DRs). Links in the list below point to detailed discussion of
7598   the relevant defect report.
7599     * [453]2094 unimplemented: use of `ptrmem_cst' in template type
7600       unification
7601     * [454]2294 using declaration confusion
7602     * [455]5050 template instantiation depth exceeds limit: recursion
7603       problem?
7604     * [456]9371 Bad exception handling in
7605       i/ostream::operator>>/<<(streambuf*)
7606     * [457]9546 bad exception handling in ostream members
7607     * [458]10081 basic_ios::_M_cache_locale leaves NULL members in the
7608       face of unknown locales
7609     * [459]10093 [460][DR 61] Setting failbit in exceptions doesn't work
7610     * [461]10095 istream::operator>>(int&) sets ios::badbit when
7611       ios::failbit is set.
7612     * [462]11554 Warning about reordering of initializers doesn't mention
7613       location of constructor
7614     * [463]12297 istream::sentry::sentry() handles eof() incorrectly.
7615     * [464]12352 Exception safety problems in src/localename.cc
7616     * [465]12438 Memory leak in locale::combine()
7617     * [466]12540 Memory leak in locale::locale(const char*)
7618     * [467]12594 DRs [468]60 [TC] and [469]63 [TC] not implemented
7619     * [470]12657 Resolution of [471]DR 292 (WP) still unimplemented
7620     * [472]12696 memory eating infinite loop in diagnostics (error
7621       recovery problem)
7622     * [473]12815 Code compiled with optimization behaves unexpectedly
7623     * [474]12862 Conflicts between typedefs/enums and namespace member
7624       declarations
7625     * [475]12926 Wrong value after assignment in initialize list using
7626       bit-fields
7627     * [476]12967 Resolution of [477]DR 300 [WP] still unimplemented
7628     * [478]12971 Resolution of [479]DR 328 [WP] still unimplemented
7629     * [480]13007 basic_streambuf::pubimbue, imbue wrong
7630     * [481]13009 Implicitly-defined assignment operator writes to wrong
7631       memory
7632     * [482]13057 regparm attribute not applied to destructor
7633     * [483]13070 -Wformat option ignored in g++
7634     * [484]13081 forward template declarations in <complex> let inlining
7635       fail
7636     * [485]13239 Assertion does not seem to work correctly anymore
7637     * [486]13262 "xxx is private within this context" when initializing a
7638       self-contained template class
7639     * [487]13290 simple typo in concept checking for std::generate_n
7640     * [488]13323 Template code does not compile in presence of typedef
7641     * [489]13369 __verify_grouping (and __add_grouping?) not correct
7642     * [490]13371 infinite loop with packed struct and inlining
7643     * [491]13445 Template argument replacement "dereferences" a typedef
7644     * [492]13461 Fails to access protected-ctor from public constant
7645     * [493]13462 Non-standard-conforming type set::pointer
7646     * [494]13478 gcc uses wrong constructor to initialize a const
7647       reference
7648     * [495]13544 "conflicting types" for enums in different scopes
7649     * [496]13650 string::compare should not (always) use
7650       traits_type::length()
7651     * [497]13683 bogus warning about passing non-PODs through ellipsis
7652     * [498]13688 Derived class is denied access to protected base class
7653       member class
7654     * [499]13774 Member variable cleared in virtual multiple inheritance
7655       class
7656     * [500]13884 Protect sstream.tcc from extern template use
7657
7658    Java compiler and library
7659
7660     * [501]10746 [win32] garbage collection crash in GCJ
7661
7662    Objective-C compiler and library
7663
7664     * [502]11433 Crash due to dereferencing null pointer when querying
7665       protocol
7666
7667    Fortran compiler and library
7668
7669     * [503]12633 logical expression gives incorrect result with
7670       -fugly-logint option
7671     * [504]13037 [gcse-lm] g77 generates incorrect code
7672     * [505]13213 Hex constant problem when compiling with -fugly-logint
7673       and -ftypeless-boz
7674
7675    x86-specific (Intel/AMD)
7676
7677     * [506]4490 ICE with -m128bit-long-double
7678     * [507]12292 [x86_64] ICE: RTL check: expected code `const_int', have
7679       `reg' in make_field_assignment, in combine.c
7680     * [508]12441 ICE: can't find a register to spill
7681     * [509]12943 array static-init failure under -fpic, -fPIC
7682     * [510]13608 Incorrect code with -O3 -ffast-math
7683
7684    PowerPC-specific
7685
7686     * [511]11598 testcase gcc.dg/20020118-1.c fails runtime check of
7687       __attribute__((aligned(16)))
7688     * [512]11793 ICE in extract_insn, in recog.c (const_vector's)
7689     * [513]12467 vmsumubm emitted when vmsummbm appropriate (typo in
7690       altivec.md)
7691     * [514]12537 g++ generates writeable text sections
7692
7693    SPARC-specific
7694
7695     * [515]12496 wrong result for __atomic_add(&value, -1) when using -O0
7696       -m64
7697     * [516]12865 mprotect call to make trampoline executable may fail
7698     * [517]13354 ICE in sparc_emit_set_const32
7699
7700    ARM-specific
7701
7702     * [518]10467 [arm] ICE in pre_insert_copy_insn,
7703
7704    ia64-specific
7705
7706     * [519]11226 ICE passing struct arg with two floats
7707     * [520]11227 ICE for _Complex float, _Complex long double args
7708     * [521]12644 GCC 3.3.2 fails to compile glibc on ia64
7709     * [522]13149 build gcc-3.3.2 1305 error:unrecognizable insn
7710     * Various fixes for libunwind
7711
7712    Alpha-specific
7713
7714     * [523]12654 Incorrect comparison code generated for Alpha
7715     * [524]12965 SEGV+ICE in cc1plus on alpha-linux with -O2
7716     * [525]13031 ICE (unrecognizable insn) when building gnome-libs-1.4.2
7717
7718    HPPA-specific
7719
7720     * [526]11634 [hppa] ICE in verify_local_live_at_start, in flow.c
7721     * [527]12158 [hppa] compilation does not terminate at -O1
7722
7723    S390-specific
7724
7725     * [528]11992 Wrong built-in code for memcmp with length 1<<24: only
7726       (1<<24)-1 possible for CLCL-Instruction
7727
7728    SH-specific
7729
7730     * [529]9365 segfault in gen_far_branch (config/sh/sh.c)
7731     * [530]10392 optimizer generates faulty array indexing
7732     * [531]11322 SH profiler outputs multiple definitions of symbol
7733     * [532]13069 gcc/config/sh/rtems.h broken
7734     * [533]13302 Putting a va_list in a struct causes seg fault
7735     * [534]13585 Incorrect optimization of call to sfunc
7736     * Fix inappropriately exported libgcc functions from the shared
7737       library
7738
7739    Other embedded target specific
7740
7741     * [535]8916 [mcore] unsigned char assign gets hosed.
7742     * [536]11576 [h8300] ICE in change_address_1, in emit-rtl.c
7743     * [537]13122 [h8300] local variable gets corrupted by function call
7744       when -fomit-frame-pointer is given
7745     * [538]13256 [cris] strict_low_part mistreated in delay slots
7746     * [539]13373 [mcore] optimization with -frerun-cse-after-loop
7747       -fexpensive-optimizations produces wrong code on mcore
7748
7749    GNU HURD-specific
7750
7751     * [540]12561 gcc/config/t-gnu needs updating to work with
7752       --with-sysroot
7753
7754    Tru64 Unix specific
7755
7756     * [541]6243 testsuite fails almost all tests due to no libintl in
7757       LD_LIBRARY_PATH during test.
7758     * [542]11397 weak aliases broken on Tru64 UNIX
7759
7760    AIX-specific
7761
7762     * [543]12505 build failure due to defines of uchar in cpphash.h and
7763       sys/types.h
7764     * [544]13150 WEAK symbols not exported by collect2
7765
7766    IRIX-specific
7767
7768     * [545]12666 fixincludes problem on IRIX 6.5.19m
7769
7770    Solaris-specific
7771
7772     * [546]12969 Including sys/byteorder.h breaks configure checks
7773
7774    Testsuite problems (compiler is not affected)
7775
7776     * [547]10819 testsuite creates CR+LF on compiler version lines in
7777       test summary files
7778     * [548]11612 abi_check not finding correct libgcc_s.so.1
7779
7780    Miscellaneous
7781
7782     * [549]13211 using -###, incorrect warnings about unused linker file
7783       are produced
7784     __________________________________________________________________
7785
7786GCC 3.3.4
7787
7788   This is the [550]list of problem reports (PRs) from GCC's bug tracking
7789   system that are known to be fixed in the 3.3.4 release. This list might
7790   not be complete (that is, it is possible that some PRs that have been
7791   fixed are not listed here).
7792     __________________________________________________________________
7793
7794GCC 3.3.5
7795
7796   This is the [551]list of problem reports (PRs) from GCC's bug tracking
7797   system that are known to be fixed in the 3.3.5 release. This list might
7798   not be complete (that is, it is possible that some PRs that have been
7799   fixed are not listed here).
7800     __________________________________________________________________
7801
7802GCC 3.3.6
7803
7804   This is the [552]list of problem reports (PRs) from GCC's bug tracking
7805   system that are known to be fixed in the 3.3.6 release. This list might
7806   not be complete (that is, it is possible that some PRs that have been
7807   fixed are not listed here).
7808
7809
7810    For questions related to the use of GCC, please consult these web
7811    pages and the [553]GCC manuals. If that fails, the
7812    [554]gcc-help@gcc.gnu.org mailing list might help. Comments on these
7813    web pages and the development of GCC are welcome on our developer
7814    list at [555]gcc@gcc.gnu.org. All of [556]our lists have public
7815    archives.
7816
7817   Copyright (C) [557]Free Software Foundation, Inc. Verbatim copying and
7818   distribution of this entire article is permitted in any medium,
7819   provided this notice is preserved.
7820
7821   These pages are [558]maintained by the GCC team. Last modified
7822   2011-10-24[559].
7823
7824References
7825
7826   1. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.6
7827   2. http://gcc.gnu.org/gcc-3.1/changes.html#obsolete_systems
7828   3. http://gcc.gnu.org/gcc-3.3/changes.html#obsolete_systems
7829   4. http://gcc.gnu.org/gcc-3.3/changes.html#nonnull_attribute
7830   5. http://gcc.gnu.org/news/dfa.html
7831   6. http://gcc.gnu.org/gcc-3.3/c99status.html
7832   7. http://gcc.gnu.org/onlinedocs/gcc-3.3.6/g77/News.html
7833   8. http://gcc.gnu.org/PR10140
7834   9. http://gcc.gnu.org/PR10198
7835  10. http://gcc.gnu.org/PR10338
7836  11. http://gcc.gnu.org/PR3581
7837  12. http://gcc.gnu.org/PR4382
7838  13. http://gcc.gnu.org/PR5533
7839  14. http://gcc.gnu.org/PR6387
7840  15. http://gcc.gnu.org/PR6412
7841  16. http://gcc.gnu.org/PR6620
7842  17. http://gcc.gnu.org/PR6663
7843  18. http://gcc.gnu.org/PR7068
7844  19. http://gcc.gnu.org/PR7083
7845  20. http://gcc.gnu.org/PR7647
7846  21. http://gcc.gnu.org/PR7675
7847  22. http://gcc.gnu.org/PR7718
7848  23. http://gcc.gnu.org/PR8116
7849  24. http://gcc.gnu.org/PR8358
7850  25. http://gcc.gnu.org/PR8511
7851  26. http://gcc.gnu.org/PR8564
7852  27. http://gcc.gnu.org/PR8660
7853  28. http://gcc.gnu.org/PR8766
7854  29. http://gcc.gnu.org/PR8803
7855  30. http://gcc.gnu.org/PR8846
7856  31. http://gcc.gnu.org/PR8906
7857  32. http://gcc.gnu.org/PR9216
7858  33. http://gcc.gnu.org/PR9261
7859  34. http://gcc.gnu.org/PR9263
7860  35. http://gcc.gnu.org/PR9429
7861  36. http://gcc.gnu.org/PR9516
7862  37. http://gcc.gnu.org/PR9600
7863  38. http://gcc.gnu.org/PR9629
7864  39. http://gcc.gnu.org/PR9672
7865  40. http://gcc.gnu.org/PR9749
7866  41. http://gcc.gnu.org/PR9794
7867  42. http://gcc.gnu.org/PR9829
7868  43. http://gcc.gnu.org/PR9916
7869  44. http://gcc.gnu.org/PR9936
7870  45. http://gcc.gnu.org/PR10262
7871  46. http://gcc.gnu.org/PR10278
7872  47. http://gcc.gnu.org/PR10446
7873  48. http://gcc.gnu.org/PR10451
7874  49. http://gcc.gnu.org/PR10506
7875  50. http://gcc.gnu.org/PR10549
7876  51. http://gcc.gnu.org/PR2001
7877  52. http://gcc.gnu.org/PR2391
7878  53. http://gcc.gnu.org/PR2960
7879  54. http://gcc.gnu.org/PR4046
7880  55. http://gcc.gnu.org/PR6405
7881  56. http://gcc.gnu.org/PR6798
7882  57. http://gcc.gnu.org/PR6871
7883  58. http://gcc.gnu.org/PR6909
7884  59. http://gcc.gnu.org/PR7189
7885  60. http://gcc.gnu.org/PR7642
7886  61. http://gcc.gnu.org/PR8634
7887  62. http://gcc.gnu.org/PR8750
7888  63. http://gcc.gnu.org/PR2161
7889  64. http://gcc.gnu.org/PR4319
7890  65. http://gcc.gnu.org/PR8602
7891  66. http://gcc.gnu.org/PR9177
7892  67. http://gcc.gnu.org/PR9853
7893  68. http://gcc.gnu.org/PR45
7894  69. http://gcc.gnu.org/PR3784
7895  70. http://gcc.gnu.org/PR764
7896  71. http://gcc.gnu.org/PR5116
7897  72. http://gcc.gnu.org/PR2862
7898  73. http://gcc.gnu.org/PR3663
7899  74. http://gcc.gnu.org/PR3797
7900  75. http://gcc.gnu.org/PR3948
7901  76. http://gcc.gnu.org/PR4137
7902  77. http://gcc.gnu.org/PR4361
7903  78. http://gcc.gnu.org/PR4802
7904  79. http://gcc.gnu.org/PR5837
7905  80. http://gcc.gnu.org/PR4803
7906  81. http://gcc.gnu.org/PR5094
7907  82. http://gcc.gnu.org/PR5730
7908  83. http://gcc.gnu.org/PR6713
7909  84. http://gcc.gnu.org/PR7015
7910  85. http://gcc.gnu.org/PR7086
7911  86. http://gcc.gnu.org/PR7099
7912  87. http://gcc.gnu.org/PR7247
7913  88. http://gcc.gnu.org/PR7441
7914  89. http://gcc.gnu.org/PR7768
7915  90. http://gcc.gnu.org/PR7804
7916  91. http://gcc.gnu.org/PR8099
7917  92. http://gcc.gnu.org/PR8117
7918  93. http://gcc.gnu.org/PR8205
7919  94. http://gcc.gnu.org/PR8645
7920  95. http://gcc.gnu.org/PR8724
7921  96. http://gcc.gnu.org/PR8805
7922  97. http://gcc.gnu.org/PR8691
7923  98. http://gcc.gnu.org/PR8700
7924  99. http://gcc.gnu.org/PR8724
7925 100. http://gcc.gnu.org/PR8949
7926 101. http://gcc.gnu.org/PR9016
7927 102. http://gcc.gnu.org/PR9053
7928 103. http://gcc.gnu.org/PR9152
7929 104. http://gcc.gnu.org/PR9182
7930 105. http://gcc.gnu.org/PR9297
7931 106. http://gcc.gnu.org/PR9318
7932 107. http://gcc.gnu.org/PR9320
7933 108. http://gcc.gnu.org/PR9400
7934 109. http://gcc.gnu.org/PR9424
7935 110. http://gcc.gnu.org/PR9425
7936 111. http://gcc.gnu.org/PR9439
7937 112. http://gcc.gnu.org/PR9474
7938 113. http://gcc.gnu.org/PR9548
7939 114. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#231
7940 115. http://gcc.gnu.org/PR9555
7941 116. http://gcc.gnu.org/PR9561
7942 117. http://gcc.gnu.org/PR9563
7943 118. http://gcc.gnu.org/PR9582
7944 119. http://gcc.gnu.org/PR9622
7945 120. http://gcc.gnu.org/PR9683
7946 121. http://gcc.gnu.org/PR9791
7947 122. http://gcc.gnu.org/PR9817
7948 123. http://gcc.gnu.org/PR9825
7949 124. http://gcc.gnu.org/PR9826
7950 125. http://gcc.gnu.org/PR9924
7951 126. http://gcc.gnu.org/PR9946
7952 127. http://gcc.gnu.org/PR9964
7953 128. http://gcc.gnu.org/PR9988
7954 129. http://gcc.gnu.org/PR10033
7955 130. http://gcc.gnu.org/PR10097
7956 131. http://gcc.gnu.org/PR10132
7957 132. http://gcc.gnu.org/PR10180
7958 133. http://gcc.gnu.org/PR10199
7959 134. http://gcc.gnu.org/PR10300
7960 135. http://gcc.gnu.org/PR10427
7961 136. http://gcc.gnu.org/PR10503
7962 137. http://gcc.gnu.org/PR5956
7963 138. http://gcc.gnu.org/PR1832
7964 139. http://gcc.gnu.org/PR3924
7965 140. http://gcc.gnu.org/PR5634
7966 141. http://gcc.gnu.org/PR6367
7967 142. http://gcc.gnu.org/PR6491
7968 143. http://gcc.gnu.org/PR6742
7969 144. http://gcc.gnu.org/PR7113
7970 145. http://gcc.gnu.org/PR7236
7971 146. http://gcc.gnu.org/PR7278
7972 147. http://gcc.gnu.org/PR7384
7973 148. http://gcc.gnu.org/PR7388
7974 149. http://gcc.gnu.org/PR8587
7975 150. http://gcc.gnu.org/PR9038
7976 151. http://gcc.gnu.org/PR10197
7977 152. http://gcc.gnu.org/PR6005
7978 153. http://gcc.gnu.org/PR6389
7979 154. http://gcc.gnu.org/PR6576
7980 155. http://gcc.gnu.org/PR6652
7981 156. http://gcc.gnu.org/PR7060
7982 157. http://gcc.gnu.org/PR7073
7983 158. http://gcc.gnu.org/PR7180
7984 159. http://gcc.gnu.org/PR7416
7985 160. http://gcc.gnu.org/PR7570
7986 161. http://gcc.gnu.org/PR7578
7987 162. http://gcc.gnu.org/PR7611
7988 163. http://gcc.gnu.org/PR7709
7989 164. http://gcc.gnu.org/PR7766
7990 165. http://gcc.gnu.org/PR7785
7991 166. http://gcc.gnu.org/PR7786
7992 167. http://gcc.gnu.org/PR8142
7993 168. http://gcc.gnu.org/PR8234
7994 169. http://gcc.gnu.org/PR8415
7995 170. http://gcc.gnu.org/PR8481
7996 171. http://gcc.gnu.org/PR8593
7997 172. http://gcc.gnu.org/PR8759
7998 173. http://gcc.gnu.org/PR8997
7999 174. http://gcc.gnu.org/PR9253
8000 175. http://gcc.gnu.org/PR9254
8001 176. http://gcc.gnu.org/PR9271
8002 177. http://gcc.gnu.org/PR6767
8003 178. http://gcc.gnu.org/PR9911
8004 179. http://gcc.gnu.org/PR10020
8005 180. http://gcc.gnu.org/PR10546
8006 181. http://gcc.gnu.org/PR7029
8007 182. http://gcc.gnu.org/PR2903
8008 183. http://gcc.gnu.org/PR7873
8009 184. http://gcc.gnu.org/PR7680
8010 185. http://gcc.gnu.org/PR8705
8011 186. http://gcc.gnu.org/PR9986
8012 187. http://gcc.gnu.org/PR10056
8013 188. http://gcc.gnu.org/PR6744
8014 189. http://gcc.gnu.org/PR7361
8015 190. http://gcc.gnu.org/PR9496
8016 191. http://gcc.gnu.org/PR7067
8017 192. http://gcc.gnu.org/PR8480
8018 193. http://gcc.gnu.org/PR8784
8019 194. http://gcc.gnu.org/PR10315
8020 195. http://gcc.gnu.org/PR10267
8021 196. http://gcc.gnu.org/PR7916
8022 197. http://gcc.gnu.org/PR7926
8023 198. http://gcc.gnu.org/PR8555
8024 199. http://gcc.gnu.org/PR8994
8025 200. http://gcc.gnu.org/PR9426
8026 201. http://gcc.gnu.org/PR9806
8027 202. http://gcc.gnu.org/PR10077
8028 203. http://gcc.gnu.org/PR10233
8029 204. http://gcc.gnu.org/PR10286
8030 205. http://gcc.gnu.org/PR10308
8031 206. http://gcc.gnu.org/PR11272
8032 207. http://gcc.gnu.org/PR5754
8033 208. http://gcc.gnu.org/PR6597
8034 209. http://gcc.gnu.org/PR6949
8035 210. http://gcc.gnu.org/PR7053
8036 211. http://gcc.gnu.org/PR8164
8037 212. http://gcc.gnu.org/PR8384
8038 213. http://gcc.gnu.org/PR9559
8039 214. http://gcc.gnu.org/PR9649
8040 215. http://gcc.gnu.org/PR9864
8041 216. http://gcc.gnu.org/PR10432
8042 217. http://gcc.gnu.org/PR10475
8043 218. http://gcc.gnu.org/PR10635
8044 219. http://gcc.gnu.org/PR10661
8045 220. http://gcc.gnu.org/PR10700
8046 221. http://gcc.gnu.org/PR10712
8047 222. http://gcc.gnu.org/PR10796
8048 223. http://gcc.gnu.org/PR10890
8049 224. http://gcc.gnu.org/PR10939
8050 225. http://gcc.gnu.org/PR10956
8051 226. http://gcc.gnu.org/PR11041
8052 227. http://gcc.gnu.org/PR11059
8053 228. http://gcc.gnu.org/PR11083
8054 229. http://gcc.gnu.org/PR11105
8055 230. http://gcc.gnu.org/PR11149
8056 231. http://gcc.gnu.org/PR11228
8057 232. http://gcc.gnu.org/PR11282
8058 233. http://gcc.gnu.org/PR11301
8059 234. http://gcc.gnu.org/PR11308
8060 235. http://gcc.gnu.org/PR11473
8061 236. http://gcc.gnu.org/PR11503
8062 237. http://gcc.gnu.org/PR11513
8063 238. http://gcc.gnu.org/PR11198
8064 239. http://gcc.gnu.org/PR11304
8065 240. http://gcc.gnu.org/PR11381
8066 241. http://gcc.gnu.org/PR11536
8067 242. http://gcc.gnu.org/PR11557
8068 243. http://gcc.gnu.org/PR5897
8069 244. http://gcc.gnu.org/PR11279
8070 245. http://gcc.gnu.org/PR11022
8071 246. http://gcc.gnu.org/PR2330
8072 247. http://gcc.gnu.org/PR5388
8073 248. http://gcc.gnu.org/PR5390
8074 249. http://gcc.gnu.org/PR7877
8075 250. http://gcc.gnu.org/PR9393
8076 251. http://gcc.gnu.org/PR10032
8077 252. http://gcc.gnu.org/PR10468
8078 253. http://gcc.gnu.org/PR10527
8079 254. http://gcc.gnu.org/PR10679
8080 255. http://gcc.gnu.org/PR10682
8081 256. http://gcc.gnu.org/PR10689
8082 257. http://gcc.gnu.org/PR10845
8083 258. http://gcc.gnu.org/PR10849
8084 259. http://gcc.gnu.org/PR10888
8085 260. http://gcc.gnu.org/PR10929
8086 261. http://gcc.gnu.org/PR10931
8087 262. http://gcc.gnu.org/PR10940
8088 263. http://gcc.gnu.org/PR10968
8089 264. http://gcc.gnu.org/PR10990
8090 265. http://gcc.gnu.org/PR11039
8091 266. http://gcc.gnu.org/PR11062
8092 267. http://gcc.gnu.org/PR11095
8093 268. http://gcc.gnu.org/PR11098
8094 269. http://gcc.gnu.org/PR11137
8095 270. http://gcc.gnu.org/PR11154
8096 271. http://gcc.gnu.org/PR11329
8097 272. http://gcc.gnu.org/PR11332
8098 273. http://gcc.gnu.org/PR11431
8099 274. http://gcc.gnu.org/PR11528
8100 275. http://gcc.gnu.org/PR11546
8101 276. http://gcc.gnu.org/PR11567
8102 277. http://gcc.gnu.org/PR11645
8103 278. http://gcc.gnu.org/PR5179
8104 279. http://gcc.gnu.org/PR8204
8105 280. http://gcc.gnu.org/PR10838
8106 281. http://gcc.gnu.org/PR10886
8107 282. http://gcc.gnu.org/PR11349
8108 283. http://gcc.gnu.org/PR4823
8109 284. http://gcc.gnu.org/PR8878
8110 285. http://gcc.gnu.org/PR9815
8111 286. http://gcc.gnu.org/PR10402
8112 287. http://gcc.gnu.org/PR10504
8113 288. http://gcc.gnu.org/PR10673
8114 289. http://gcc.gnu.org/PR11044
8115 290. http://gcc.gnu.org/PR11089
8116 291. http://gcc.gnu.org/PR11420
8117 292. http://gcc.gnu.org/PR9362
8118 293. http://gcc.gnu.org/PR10142
8119 294. http://gcc.gnu.org/PR10663
8120 295. http://gcc.gnu.org/PR10835
8121 296. http://gcc.gnu.org/PR10876
8122 297. http://gcc.gnu.org/PR10955
8123 298. http://gcc.gnu.org/PR11018
8124 299. http://gcc.gnu.org/PR11556
8125 300. http://gcc.gnu.org/PR10907
8126 301. http://gcc.gnu.org/PR11320
8127 302. http://gcc.gnu.org/PR11599
8128 303. http://gcc.gnu.org/PR9745
8129 304. http://gcc.gnu.org/PR10871
8130 305. http://gcc.gnu.org/PR11440
8131 306. http://gcc.gnu.org/PR7594
8132 307. http://gcc.gnu.org/PR10557
8133 308. http://gcc.gnu.org/PR11054
8134 309. http://gcc.gnu.org/PR10834
8135 310. http://gcc.gnu.org/PR10842
8136 311. http://gcc.gnu.org/PR11052
8137 312. http://gcc.gnu.org/PR11183
8138 313. http://gcc.gnu.org/PR11084
8139 314. http://gcc.gnu.org/PR10331
8140 315. http://gcc.gnu.org/PR10413
8141 316. http://gcc.gnu.org/PR11096
8142 317. http://gcc.gnu.org/PR2873
8143 318. http://gcc.gnu.org/PR3163
8144 319. http://gcc.gnu.org/PR5287
8145 320. http://gcc.gnu.org/PR10148
8146 321. http://gcc.gnu.org/PR8787
8147 322. http://gcc.gnu.org/PR10900
8148 323. http://gcc.gnu.org/PR1607
8149 324. http://gcc.gnu.org/PR4252
8150 325. http://gcc.gnu.org/PR4490
8151 326. http://gcc.gnu.org/PR10355
8152 327. http://gcc.gnu.org/PR10726
8153 328. http://gcc.gnu.org/PR10805
8154 329. http://gcc.gnu.org/PR10815
8155 330. http://gcc.gnu.org/PR10877
8156 331. http://gcc.gnu.org/PR11280
8157 332. http://gcc.gnu.org/PR11466
8158 333. http://gcc.gnu.org/PR10737
8159 334. http://gcc.gnu.org/PR10810
8160 335. http://gcc.gnu.org/bugzilla/
8161 336. http://gcc.gnu.org/PR8336
8162 337. http://gcc.gnu.org/PR9330
8163 338. http://gcc.gnu.org/PR9631
8164 339. http://gcc.gnu.org/PR9877
8165 340. http://gcc.gnu.org/PR11687
8166 341. http://gcc.gnu.org/PR12263
8167 342. http://gcc.gnu.org/PR12490
8168 343. http://gcc.gnu.org/PR7277
8169 344. http://gcc.gnu.org/PR7939
8170 345. http://gcc.gnu.org/PR11063
8171 346. http://gcc.gnu.org/PR11207
8172 347. http://gcc.gnu.org/PR11522
8173 348. http://gcc.gnu.org/PR11595
8174 349. http://gcc.gnu.org/PR11646
8175 350. http://gcc.gnu.org/PR11665
8176 351. http://gcc.gnu.org/PR11852
8177 352. http://gcc.gnu.org/PR11878
8178 353. http://gcc.gnu.org/PR11883
8179 354. http://gcc.gnu.org/PR11991
8180 355. http://gcc.gnu.org/PR12146
8181 356. http://gcc.gnu.org/PR12215
8182 357. http://gcc.gnu.org/PR12369
8183 358. http://gcc.gnu.org/PR12446
8184 359. http://gcc.gnu.org/PR12510
8185 360. http://gcc.gnu.org/PR12544
8186 361. http://gcc.gnu.org/PR9862
8187 362. http://gcc.gnu.org/PR10962
8188 363. http://gcc.gnu.org/PR11370
8189 364. http://gcc.gnu.org/PR11637
8190 365. http://gcc.gnu.org/PR11885
8191 366. http://gcc.gnu.org/PR12082
8192 367. http://gcc.gnu.org/PR12180
8193 368. http://gcc.gnu.org/PR12340
8194 369. http://gcc.gnu.org/PR3907
8195 370. http://gcc.gnu.org/PR5293
8196 371. http://gcc.gnu.org/PR5296
8197 372. http://gcc.gnu.org/PR7939
8198 373. http://gcc.gnu.org/PR8656
8199 374. http://gcc.gnu.org/PR10147
8200 375. http://gcc.gnu.org/PR11400
8201 376. http://gcc.gnu.org/PR11409
8202 377. http://gcc.gnu.org/PR11740
8203 378. http://gcc.gnu.org/PR11786
8204 379. http://gcc.gnu.org/PR11867
8205 380. http://gcc.gnu.org/PR11928
8206 381. http://gcc.gnu.org/PR12114
8207 382. http://gcc.gnu.org/PR12163
8208 383. http://gcc.gnu.org/PR12181
8209 384. http://gcc.gnu.org/PR12236
8210 385. http://gcc.gnu.org/PR12266
8211 386. http://gcc.gnu.org/PR12296
8212 387. http://gcc.gnu.org/PR12298
8213 388. http://gcc.gnu.org/PR12369
8214 389. http://gcc.gnu.org/PR12337
8215 390. http://gcc.gnu.org/PR12344
8216 391. http://gcc.gnu.org/PR12451
8217 392. http://gcc.gnu.org/PR12486
8218 393. http://gcc.gnu.org/PR8869
8219 394. http://gcc.gnu.org/PR9786
8220 395. http://gcc.gnu.org/PR11689
8221 396. http://gcc.gnu.org/PR12116
8222 397. http://gcc.gnu.org/PR12070
8223 398. http://gcc.gnu.org/PR11184
8224 399. http://gcc.gnu.org/PR11535
8225 400. http://gcc.gnu.org/PR11693
8226 401. http://gcc.gnu.org/PR12224
8227 402. http://gcc.gnu.org/PR11087
8228 403. http://gcc.gnu.org/PR11319
8229 404. http://gcc.gnu.org/PR11949
8230 405. http://gcc.gnu.org/PR11662
8231 406. http://gcc.gnu.org/PR11965
8232 407. http://gcc.gnu.org/PR12301
8233 408. http://gcc.gnu.org/PR11717
8234 409. http://gcc.gnu.org/PR11313
8235 410. http://gcc.gnu.org/PR11712
8236 411. http://gcc.gnu.org/PR12166
8237 412. http://gcc.gnu.org/PR12101
8238 413. http://gcc.gnu.org/PR10988
8239 414. http://gcc.gnu.org/PR11805
8240 415. http://gcc.gnu.org/PR11902
8241 416. http://gcc.gnu.org/PR11903
8242 417. http://gcc.gnu.org/bugzilla/
8243 418. http://gcc.gnu.org/PR11890
8244 419. http://gcc.gnu.org/PR12399
8245 420. http://gcc.gnu.org/PR13068
8246 421. http://gcc.gnu.org/PR10060
8247 422. http://gcc.gnu.org/PR10555
8248 423. http://gcc.gnu.org/PR10706
8249 424. http://gcc.gnu.org/PR11496
8250 425. http://gcc.gnu.org/PR11741
8251 426. http://gcc.gnu.org/PR12440
8252 427. http://gcc.gnu.org/PR12632
8253 428. http://gcc.gnu.org/PR12712
8254 429. http://gcc.gnu.org/PR12726
8255 430. http://gcc.gnu.org/PR12890
8256 431. http://gcc.gnu.org/PR12900
8257 432. http://gcc.gnu.org/PR13060
8258 433. http://gcc.gnu.org/PR13289
8259 434. http://gcc.gnu.org/PR13318
8260 435. http://gcc.gnu.org/PR13392
8261 436. http://gcc.gnu.org/PR13574
8262 437. http://gcc.gnu.org/PR13475
8263 438. http://gcc.gnu.org/PR13797
8264 439. http://gcc.gnu.org/PR13824
8265 440. http://gcc.gnu.org/PR8776
8266 441. http://gcc.gnu.org/PR10339
8267 442. http://gcc.gnu.org/PR11350
8268 443. http://gcc.gnu.org/PR12826
8269 444. http://gcc.gnu.org/PR12500
8270 445. http://gcc.gnu.org/PR12941
8271 446. http://gcc.gnu.org/PR12953
8272 447. http://gcc.gnu.org/PR13041
8273 448. http://gcc.gnu.org/PR13507
8274 449. http://gcc.gnu.org/PR13382
8275 450. http://gcc.gnu.org/PR13394
8276 451. http://gcc.gnu.org/PR13400
8277 452. http://gcc.gnu.org/PR13521
8278 453. http://gcc.gnu.org/PR2094
8279 454. http://gcc.gnu.org/PR2294
8280 455. http://gcc.gnu.org/PR5050
8281 456. http://gcc.gnu.org/PR9371
8282 457. http://gcc.gnu.org/PR9546
8283 458. http://gcc.gnu.org/PR10081
8284 459. http://gcc.gnu.org/PR10093
8285 460. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#61
8286 461. http://gcc.gnu.org/PR10095
8287 462. http://gcc.gnu.org/PR11554
8288 463. http://gcc.gnu.org/PR12297
8289 464. http://gcc.gnu.org/PR12352
8290 465. http://gcc.gnu.org/PR12438
8291 466. http://gcc.gnu.org/PR12540
8292 467. http://gcc.gnu.org/PR12594
8293 468. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#60
8294 469. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#63
8295 470. http://gcc.gnu.org/PR12657
8296 471. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#292
8297 472. http://gcc.gnu.org/PR12696
8298 473. http://gcc.gnu.org/PR12815
8299 474. http://gcc.gnu.org/PR12862
8300 475. http://gcc.gnu.org/PR12926
8301 476. http://gcc.gnu.org/PR12967
8302 477. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html
8303 478. http://gcc.gnu.org/PR12971
8304 479. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#328
8305 480. http://gcc.gnu.org/PR13007
8306 481. http://gcc.gnu.org/PR13009
8307 482. http://gcc.gnu.org/PR13057
8308 483. http://gcc.gnu.org/PR13070
8309 484. http://gcc.gnu.org/PR13081
8310 485. http://gcc.gnu.org/PR13239
8311 486. http://gcc.gnu.org/PR13262
8312 487. http://gcc.gnu.org/PR13290
8313 488. http://gcc.gnu.org/PR13323
8314 489. http://gcc.gnu.org/PR13369
8315 490. http://gcc.gnu.org/PR13371
8316 491. http://gcc.gnu.org/PR13445
8317 492. http://gcc.gnu.org/PR13461
8318 493. http://gcc.gnu.org/PR13462
8319 494. http://gcc.gnu.org/PR13478
8320 495. http://gcc.gnu.org/PR13544
8321 496. http://gcc.gnu.org/PR13650
8322 497. http://gcc.gnu.org/PR13683
8323 498. http://gcc.gnu.org/PR13688
8324 499. http://gcc.gnu.org/PR13774
8325 500. http://gcc.gnu.org/PR13884
8326 501. http://gcc.gnu.org/PR10746
8327 502. http://gcc.gnu.org/PR11433
8328 503. http://gcc.gnu.org/PR12633
8329 504. http://gcc.gnu.org/PR13037
8330 505. http://gcc.gnu.org/PR13213
8331 506. http://gcc.gnu.org/PR4490
8332 507. http://gcc.gnu.org/PR12292
8333 508. http://gcc.gnu.org/PR12441
8334 509. http://gcc.gnu.org/PR12943
8335 510. http://gcc.gnu.org/PR13608
8336 511. http://gcc.gnu.org/PR11598
8337 512. http://gcc.gnu.org/PR11793
8338 513. http://gcc.gnu.org/PR12467
8339 514. http://gcc.gnu.org/PR12537
8340 515. http://gcc.gnu.org/PR12496
8341 516. http://gcc.gnu.org/PR12865
8342 517. http://gcc.gnu.org/PR13354
8343 518. http://gcc.gnu.org/PR10467
8344 519. http://gcc.gnu.org/PR11226
8345 520. http://gcc.gnu.org/PR11227
8346 521. http://gcc.gnu.org/PR12644
8347 522. http://gcc.gnu.org/PR13149
8348 523. http://gcc.gnu.org/PR12654
8349 524. http://gcc.gnu.org/PR12965
8350 525. http://gcc.gnu.org/PR13031
8351 526. http://gcc.gnu.org/PR11634
8352 527. http://gcc.gnu.org/PR12158
8353 528. http://gcc.gnu.org/PR11992
8354 529. http://gcc.gnu.org/PR9365
8355 530. http://gcc.gnu.org/PR10392
8356 531. http://gcc.gnu.org/PR11322
8357 532. http://gcc.gnu.org/PR13069
8358 533. http://gcc.gnu.org/PR13302
8359 534. http://gcc.gnu.org/PR13585
8360 535. http://gcc.gnu.org/PR8916
8361 536. http://gcc.gnu.org/PR11576
8362 537. http://gcc.gnu.org/PR13122
8363 538. http://gcc.gnu.org/PR13256
8364 539. http://gcc.gnu.org/PR13373
8365 540. http://gcc.gnu.org/PR12561
8366 541. http://gcc.gnu.org/PR6243
8367 542. http://gcc.gnu.org/PR11397
8368 543. http://gcc.gnu.org/PR12505
8369 544. http://gcc.gnu.org/PR13150
8370 545. http://gcc.gnu.org/PR12666
8371 546. http://gcc.gnu.org/PR12969
8372 547. http://gcc.gnu.org/PR10819
8373 548. http://gcc.gnu.org/PR11612
8374 549. http://gcc.gnu.org/PR13211
8375 550. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.3.4
8376 551. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.3.5
8377 552. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.3.6
8378 553. http://gcc.gnu.org/onlinedocs/
8379 554. mailto:gcc-help@gcc.gnu.org
8380 555. mailto:gcc@gcc.gnu.org
8381 556. http://gcc.gnu.org/lists.html
8382 557. http://www.fsf.org/
8383 558. http://gcc.gnu.org/about.html
8384 559. http://validator.w3.org/check/referer
8385======================================================================
8386http://gcc.gnu.org/gcc-3.2/index.html
8387
8388                             GCC 3.2 Release Series
8389
8390   April 25, 2003
8391
8392   The [1]GNU project and the GCC developers are pleased to announce the
8393   release of GCC 3.2.3.
8394
8395   The purpose of the GCC 3.2 release series is to provide a stable
8396   platform for OS distributors to use building their next releases. A
8397   primary objective was to stabilize the C++ ABI; we believe that the
8398   interface to the compiler and the C++ standard library are now
8399   relatively stable.
8400
8401   Be aware that C++ code compiled by GCC 3.2.x will (in general) not
8402   interoperate with code compiled by GCC 3.1.1 or earlier.
8403
8404   Please refer to our [2]detailed list of news, caveats, and bug-fixes
8405   for further information.
8406
8407Release History
8408
8409   GCC 3.2.3
8410          April 25, 2003 ([3]changes)
8411
8412   GCC 3.2.2
8413          February 5, 2003 ([4]changes)
8414
8415   GCC 3.2.1
8416          November 19, 2002 ([5]changes)
8417
8418   GCC 3.2
8419          August 14, 2002 ([6]changes)
8420
8421References and Acknowledgements
8422
8423   GCC used to stand for the GNU C Compiler, but since the compiler
8424   supports several other languages aside from C, it now stands for the
8425   GNU Compiler Collection.
8426
8427   A list of [7]successful builds is updated as new information becomes
8428   available.
8429
8430   The GCC developers would like to thank the numerous people that have
8431   contributed new features, improvements, bug fixes, and other changes as
8432   well as test results to GCC. This [8]amazing group of volunteers is
8433   what makes GCC successful.
8434
8435   For additional information about GCC please refer to the [9]GCC project
8436   web site or contact the [10]GCC development mailing list.
8437
8438   To obtain GCC please use [11]our mirror sites, or our CVS server.
8439
8440
8441    For questions related to the use of GCC, please consult these web
8442    pages and the [12]GCC manuals. If that fails, the
8443    [13]gcc-help@gcc.gnu.org mailing list might help. Comments on these
8444    web pages and the development of GCC are welcome on our developer
8445    list at [14]gcc@gcc.gnu.org. All of [15]our lists have public
8446    archives.
8447
8448   Copyright (C) [16]Free Software Foundation, Inc. Verbatim copying and
8449   distribution of this entire article is permitted in any medium,
8450   provided this notice is preserved.
8451
8452   These pages are [17]maintained by the GCC team. Last modified
8453   2011-04-25[18].
8454
8455References
8456
8457   1. http://www.gnu.org/
8458   2. http://gcc.gnu.org/gcc-3.2/changes.html
8459   3. http://gcc.gnu.org/gcc-3.2/changes.html#3.2.3
8460   4. http://gcc.gnu.org/gcc-3.2/changes.html#3.2.2
8461   5. http://gcc.gnu.org/gcc-3.2/changes.html#3.2.1
8462   6. http://gcc.gnu.org/gcc-3.2/changes.html#3.2
8463   7. http://gcc.gnu.org/gcc-3.2/buildstat.html
8464   8. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
8465   9. http://gcc.gnu.org/index.html
8466  10. mailto:gcc@gcc.gnu.org
8467  11. http://gcc.gnu.org/mirrors.html
8468  12. http://gcc.gnu.org/onlinedocs/
8469  13. mailto:gcc-help@gcc.gnu.org
8470  14. mailto:gcc@gcc.gnu.org
8471  15. http://gcc.gnu.org/lists.html
8472  16. http://www.fsf.org/
8473  17. http://gcc.gnu.org/about.html
8474  18. http://validator.w3.org/check/referer
8475======================================================================
8476http://gcc.gnu.org/gcc-3.2/changes.html
8477
8478                             GCC 3.2 Release Series
8479                        Changes, New Features, and Fixes
8480
8481   The latest release in the 3.2 release series is [1]GCC 3.2.3.
8482
8483Caveats and New Features
8484
8485  Caveats
8486
8487     * The C++ compiler does not correctly zero-initialize
8488       pointers-to-data members. You must explicitly initialize them. For
8489       example: int S::*m(0); will work, but depending on
8490       default-initialization to zero will not work. This bug cannot be
8491       fixed in GCC 3.2 without inducing unacceptable risks. It will be
8492       fixed in GCC 3.3.
8493     * This GCC release is based on the GCC 3.1 sourcebase, and thus has
8494       all the [2]changes in the GCC 3.1 series. In addition, GCC 3.2 has
8495       a number of C++ ABI fixes which make its C++ compiler generate
8496       binary code which is incompatible with the C++ compilers found in
8497       earlier GCC releases, including GCC 3.1 and GCC 3.1.1.
8498
8499  Frontend Enhancements
8500
8501    C/C++/Objective-C
8502
8503     * The method of constructing the list of directories to be searched
8504       for header files has been revised. If a directory named by a -I
8505       option is a standard system include directory, the option is
8506       ignored to ensure that the default search order for system
8507       directories and the special treatment of system header files are
8508       not defeated.
8509     * The C and Objective-C compilers no longer accept the "Naming Types"
8510       extension (typedef foo = bar); it was already unavailable in C++.
8511       Code which uses it will need to be changed to use the "typeof"
8512       extension instead: typedef typeof(bar) foo. (We have removed this
8513       extension without a period of deprecation because it has caused the
8514       compiler to crash since version 3.0 and no one noticed until very
8515       recently. Thus we conclude it is not in widespread use.)
8516
8517    C++
8518
8519     * GCC 3.2 fixed serveral differences between the C++ ABI implemented
8520       in GCC and the multi-vendor standard, but more have been found
8521       since the release. 3.2.1 adds a new warning, -Wabi, to warn about
8522       code which is affected by these bugs. We will fix these bugs in
8523       some future release, once we are confident that all have been
8524       found; until then, it is our intention to make changes to the ABI
8525       only if they are necessary for correct compilation of C++, as
8526       opposed to conformance to the ABI documents.
8527     * For details on how to build an ABI compliant compiler for GNU/Linux
8528       systems, check the [3]common C++ ABI page.
8529
8530  New Targets and Target Specific Improvements
8531
8532    IA-32
8533
8534     * Fixed a number of bugs in SSE and MMX intrinsics.
8535     * Fixed common compiler crashes with SSE instruction set enabled
8536       (implied by -march=pentium3, pentium4, athlon-xp)
8537     * __m128 and __m128i is not 128bit aligned when used in structures.
8538
8539    x86-64
8540
8541     * A bug whereby the compiler could generate bad code for bzero has
8542       been fixed.
8543     * ABI fixes (implying ABI incompatibilities with previous version in
8544       some corner cases)
8545     * Fixed prefetch code generation
8546     __________________________________________________________________
8547
8548GCC 3.2.3
8549
8550   3.2.3 is a bug fix release only; there are no new features that were
8551   not present in GCC 3.2.2.
8552
8553  Bug Fixes
8554
8555   This section lists the problem reports (PRs) from GCC's bug tracking
8556   system that are known to be fixed in the 3.2.3 release. This list might
8557   not be complete (that is, it is possible that some PRs that have been
8558   fixed are not listed here), and some of the titles have been changed to
8559   make them more clear.
8560
8561    Internal Compiler Errors (multi-platform)
8562
8563     * [4]3782: (c++) -quiet -fstats produces a segmentation fault in
8564       cc1plus
8565     * [5]6440: (c++) template specializations cause ICE
8566     * [6]7050: (c++) ICE on: (i ? get_string() : throw)
8567     * [7]7741: ICE on conflicting types (make_decl_rtl in varasm.c)
8568     * [8]7982: (c++) ICE due to infinite recursion (using STL set)
8569     * [9]8068: exceedingly high (infinite) memory usage
8570     * [10]8178: ICE with __builtin_ffs
8571     * [11]8396: ICE in copy_to_mode_reg, in explow.c
8572     * [12]8674: (c++) ICE in cp_expr_size, in cp/cp-lang.c
8573     * [13]9768: ICE when optimizing inline code at -O2
8574     * [14]9798: (c++) Infinite recursion (segfault) in
8575       cp/decl.c:push_using_directive with recursive using directives
8576     * [15]9799: mismatching structure initializer with nested flexible
8577       array member: ICE
8578     * [16]9928: ICE on duplicate enum declaration
8579     * [17]10114: ICE in mem_loc_descriptor, in dwarf2out.c (affects
8580       sparc, alpha)
8581     * [18]10352: ICE in find_reloads_toplev
8582     * [19]10336: ICE with -Wunreachable-code
8583
8584    C/optimizer bugs:
8585
8586     * [20]8224: Incorrect joining of signed and unsigned division
8587     * [21]8613: -O2 produces wrong code with builtin strlen and
8588       postincrements
8589     * [22]8828: gcc reports some code is unreachable when it is not
8590     * [23]9226: GCSE breaking argument passing
8591     * [24]9853: miscompilation of non-constant structure initializer
8592     * [25]9797: C99-style struct initializers are miscompiled
8593     * [26]9967: Some standard C function calls should not be replaced
8594       when optimizing for size
8595     * [27]10116: ce2: invalid merge of join_bb in the context of switch
8596       statements
8597     * [28]10171: wrong code for inlined function
8598     * [29]10175: -Wunreachable-code doesn't work for single lines
8599
8600    C++ compiler and library:
8601
8602     * [30]8316: Confusing diagnostic for code that misuses conversion
8603       operators
8604     * [31]9169: filebuf output fails if codecvt<>::out returns noconv
8605     * [32]9420: incomplete type incorrectly reported
8606     * [33]9459: typeof in return type specification of template not
8607       supported
8608     * [34]9507: filebuf::open handles ios_base::ate incorrectly
8609     * [35]9538: Out-of-bounds memory access in streambuf::sputbackc
8610     * [36]9602: Total confusion about template/friend/virtual/abstract
8611     * [37]9993: destructor not called for local object created within and
8612       returned from infinite loop
8613     * [38]10167: ieee_1003.1-2001 locale specialisations on a glibc-2.3.2
8614       system
8615
8616    Java compiler and library:
8617
8618     * [39]9652: libgcj build fails on irix6.5.1[78]
8619     * [40]10144: gas on solaris complains about bad .stabs lines for
8620       java, native as unaffected
8621
8622    x86-specific (Intel/AMD):
8623
8624     * [41]8746: gcc miscompiles Linux kernel ppa driver on x86
8625     * [42]9888: -mcpu=k6 -Os produces out of range loop instructions
8626     * [43]9638: Cross-build for target i386-elf and i586-pc-linux-gnu
8627       failed
8628     * [44]9954: Cross-build for target i586-pc-linux-gnu (--with-newlib)
8629       failed
8630
8631    SPARC-specific:
8632
8633     * [45]7784: [Sparc] ICE in extract_insn, in recog.c
8634     * [46]7796: sparc extra failure with -m64 on execute/930921-1.c in
8635       unroll.c
8636     * [47]8281: ICE when compiling with -O2 -fPIC for Ultrasparc
8637     * [48]8366: [Sparc] C testsuite failure with -m64 -fpic -O in
8638       execute/loop-2d.c
8639     * [49]8726: gcc -O2 miscompiles Samba 2.2.7 on 32-bit sparc
8640     * [50]9414: Scheduling bug on Ultrasparc
8641     * [51]10067: GCC-3.2.2 outputs invalid asm on sparc64
8642
8643    m68k-specific:
8644
8645     * [52]7248: broken "inclusive or" code
8646     * [53]8343: m68k-elf/rtems ICE at instantiate_virtual_regs_1
8647
8648    PowerPC-specific:
8649
8650     * [54]9732: Wrong code with -O2 -fPIC
8651     * [55]10073: ICE: powerpc cannot split insn
8652
8653    Alpha-specific:
8654
8655     * [56]7702: optimization problem on a DEC alpha under OSF1
8656     * [57]9671: gcc.3.2.2 does not build on a HP Tru64 Unix v5.1B system
8657
8658    HP-specific:
8659
8660     * [58]8694: <string> breaks <ctype.h> on HP-UX 10.20 (DUP: 9275)
8661     * [59]9953: (ada) gcc 3.2.x can't build 3.3-branch ada on HP-UX 10
8662       (missing symbol)
8663     * [60]10271: Floating point args don't get reloaded across function
8664       calls with -O2
8665
8666    MIPS specific:
8667
8668     * [61]6362: mips-irix6 gcc-3.1 C testsuite failure with -mips4 in
8669       compile/920501-4.c
8670
8671    CRIS specific:
8672
8673     * [62]10377: gcc-3.2.2 creates bad assembler code for cris
8674
8675    Miscellaneous and minor bugs:
8676
8677     * [63]6955: collect2 says "core dumped" when there is no core
8678     __________________________________________________________________
8679
8680GCC 3.2.2
8681
8682   Beginning with 3.2.2, GCC's Makefile suite supports redirection of make
8683   install by means of the DESTDIR variable. Parts of the GCC tree have
8684   featured that support long before, but now it is available even from
8685   the top level.
8686
8687   Other than that, GCC 3.2.2 is a bug fix release only; there are no new
8688   features that were not present in GCC 3.2.1.
8689
8690  Bug Fixes
8691
8692   On the following i386-based systems GCC 3.2.1 broke the C ABI wrt.
8693   functions returning structures: Cygwin, FreeBSD (GCC 3.2.1 as shipped
8694   with FreeBSD 5.0 does not have this problem), Interix, a.out-based
8695   GNU/Linux and NetBSD, OpenBSD, and Darwin. GCC 3.2.2 reverts this ABI
8696   change, and thus restores ABI-compatibility with previous releases
8697   (except GCC 3.2.1) on these platforms.
8698
8699   This section lists the problem reports (PRs) from GCC's bug tracking
8700   system that are known to be fixed in the 3.2.2 release. This list might
8701   not be complete (that is, it is possible that some PRs that have been
8702   fixed are not listed here) and some of the titles have been changed to
8703   make them more clear.
8704
8705    Internal Compiler Errors (multi-platform)
8706
8707     * [64]5919: (c++) ICE when passing variable array to template
8708       function
8709     * [65]7129: (c++) ICE with min/max assignment operators (<?= and >?=)
8710     * [66]7507: ICE with -O2 when address of called function is a
8711       complicated expression
8712     * [67]7622: ICE with nested inline functions if function's address is
8713       taken
8714     * [68]7681: (fortran) ICE in compensate_edge, in reg-stack.c (also PR
8715       [69]9258)
8716     * [70]8031: (c++) ICE in code comparing typeids and casting from
8717       virtual base
8718     * [71]8275: ICE in simplify_subreg
8719     * [72]8332: (c++) builtin strlen/template interaction causes ICE
8720     * [73]8372: (c++) ICE on explicit call of destructor
8721     * [74]8439: (c, not c++) empty struct causes ICE
8722     * [75]8442: (c++) ICE with nested template classes
8723     * [76]8518: ICE when compiling mplayer ("extern inline" issue)
8724     * [77]8615: (c++) ICE with out-of-range character constant template
8725       argument
8726     * [78]8663: (c++) ICE in cp_expr_size, at cp-lang.c:307
8727     * [79]8799: (c++) ICE: error reporting routines re-entered
8728     * [80]9328: (c++) ICE with typeof(X) for overloaded X
8729     * [81]9465: (preprocessor) cpp -traditional ICE on null bytes
8730
8731    C++ (compiler and library) bugs
8732
8733     * [82]47: scoping in nested classes is broken
8734     * [83]6745: problems with iostream rdbuf() member function
8735     * [84]8214: conversion from const char* const to char* sometimes
8736       accepted illegally
8737     * [85]8493: builtin strlen and overload resolution (same bug as
8738       [86]8332)
8739     * [87]8503: strange behaviour of function types
8740     * [88]8727: compiler confused by inheritance from an anonymous struct
8741     * [89]7445: poor performance of std::locale::classic() in
8742       multi-threaded applications
8743     * [90]8230: mishandling of overflow in vector<T>::resize
8744     * [91]8399: sync_with_stdio(false) breaks unformatted input
8745     * [92]8662: illegal access of private member of unnamed class is
8746       accepted
8747     * [93]8707: "make distclean" fails in libstdc++-v3 directory
8748     * [94]8708: __USE_MALLOC doesn't work
8749     * [95]8790: Use of non-thread-safe strtok in src/localename.cc
8750     * [96]8887: Bug in date formats with --enable-clocale=generic
8751     * [97]9076: Call Frame Instructions are not handled correctly during
8752       unwind operation
8753     * [98]9151: std::setprecision limited to 16 digits when outputting a
8754       double to a stream
8755     * [99]9168: codecvt<char, char, mbstate_t> overwrites output buffers
8756     * [100]9269: libstdc++ headers: explicit specialization of function
8757       must precede its first use
8758     * [101]9322: return value of basic_streambuf<>::getloc affected by
8759       locale::global
8760     * [102]9433: segfault in runtime support for dynamic_cast
8761
8762    C and optimizer bugs
8763
8764     * [103]8032: GCC incorrectly initializes static structs that have
8765       flexible arrays
8766     * [104]8639: simple arithmetic expression broken
8767     * [105]8794: optimization improperly eliminates certain expressions
8768     * [106]8832: traditional "asm volatile" code is illegally optimized
8769     * [107]8988: loop optimizer bug: with -O2, code is generated that
8770       segfaults (found on i386, bug present for all platforms)
8771     * [108]9492: structure copy clobbers subsequent stores to structure
8772
8773    Objective-C bugs
8774
8775     * [109]9267: Objective-C parser won't build with newer bison versions
8776       (e.g. 1.875)
8777
8778    Ada bugs
8779
8780     * [110]8344: Ada build problem due to conflict between gcc/final.o,
8781       gcc/ada/final.o
8782
8783    Preprocessor bugs
8784
8785     * [111]8524: _Pragma within macros is improperly expanded
8786     * [112]8880: __WCHAR_TYPE__ macro incorrectly set to "long int" with
8787       -fshort-wchar
8788
8789    ARM-specific
8790
8791     * [113]9090: arm ICE with >= -O2; regression from gcc-2.95
8792
8793    x86-specific (Intel/AMD)
8794
8795     * [114]8588: ICE in extract_insn, at recog.c:NNNN (shift instruction)
8796     * [115]8599: loop unroll bug with -march=k6-3
8797     * [116]9506: ABI breakage in structure return (affects BSD and
8798       Cygwin, but not GNU/Linux)
8799
8800    FreeBSD 5.0 specific
8801
8802     * [117]9484: GCC 3.2.1 Bootstrap failure on FreeBSD 5.0
8803
8804    RTEMS-specific
8805
8806     * [118]9292: hppa1.1-rtems configurery problems
8807     * [119]9293: [m68k-elf/rtems] config/m68k/t-crtstuff bug
8808     * [120]9295: [mips-rtems] config/mips/rtems.h init/fini issue
8809     * [121]9296: gthr-rtems regression
8810     * [122]9316: powerpc-rtems: extending multilibs
8811
8812    HP-PA specific
8813
8814     * [123]9493: ICE with -O2 when building a simple function
8815
8816    Documentation
8817
8818     * [124]7341: hyperlink to gcov in GCC documentation doesn't work
8819     * [125]8947: Please add a warning about "-malign-double" in docs
8820     * [126]7448, [127]8882: typo cleanups
8821     __________________________________________________________________
8822
8823GCC 3.2.1
8824
8825   3.2.1 adds a new warning, -Wabi. This option warns when GNU C++
8826   generates code that is known not to be binary-compatible with the
8827   vendor-neutral ia32/ia64 ABI. Please consult the GCC manual, included
8828   in the distribution, for details.
8829
8830   This release also removes an old GCC extension, "naming types", and the
8831   documentation now directs users to use a different GCC extension,
8832   __typeof__, instead. The feature had evidently been broken for a while.
8833
8834   Otherwise, 3.2.1 is a bug fix release only; other than bug fixes and
8835   the new warning there are no new features that were not present in GCC
8836   3.2.
8837
8838   In addition, the previous fix for [128]PR 7445 (poor performance of
8839   std::locale::classic() in multi-threaded applications) was reverted
8840   ("unfixed"), because the "fix" was not thread-safe.
8841
8842  Bug Fixes
8843
8844   This section lists the problem reports (PRs) from GCC's bug tracking
8845   system that are known to be fixed in the 3.2.1 release. This list might
8846   not be complete (that is, it is possible that some PRs that have been
8847   fixed are not listed here). As you can see, the number of bug fixes is
8848   quite large, so it is strongly recommended that users of earlier GCC
8849   3.x releases upgrade to GCC 3.2.1.
8850
8851    Internal Compiler Errors (multi-platform)
8852
8853     * [129]2521: (c++) ICE in build_ptrmemfunc, in cp/typeck.c
8854     * [130]5661: (c++) ICE instantiating template on array of unknown
8855       size (bad code)
8856     * [131]6419: (c++) ICE in make_decl_rtl for "longest" attribute on
8857       64-bit platforms
8858     * [132]6994: (c++) ICE in find_function_data
8859     * [133]7150: preprocessor: GCC -dM -E gives an ICE
8860     * [134]7160: ICE when optimizing branches without a return value
8861     * [135]7228: (c++) ICE when using member template and template
8862       function
8863     * [136]7266: (c++) ICE with -pedantic on missing typename
8864     * [137]7353: ICE from use of "Naming Types" extension, see above
8865     * [138]7411: ICE in instantiate_virtual_regs_1, in function.c
8866     * [139]7478: (c++) ICE on static_cast inside template
8867     * [140]7526: preprocessor core dump when _Pragma implies #pragma
8868       dependency
8869     * [141]7721: (c++) ICE on simple (but incorrect) template ([142]7803
8870       is a duplicate)
8871     * [143]7754: (c++) ICE on union with template parameter
8872     * [144]7788: (c++) redeclaring a definition as an incomplete class
8873       causes ICE
8874     * [145]8031: (c++) ICE in comptypes, in cp/typeck.c
8875     * [146]8055: preprocessor dies with SIG11 when building FreeBSD
8876       kernel
8877     * [147]8067: (c++) ICE due to mishandling of __FUNCTION__ and related
8878       variables
8879     * [148]8134: (c++) ICE in force_store_init_value on legal code
8880     * [149]8149: (c++) ICE on incomplete type
8881     * [150]8160: (c++) ICE in build_modify_expr, in cp/typeck.c: array
8882       initialization
8883
8884    C++ (compiler and library) bugs
8885
8886     * [151]5607: No pointer adjustment in covariant return types
8887     * [152]6579: Infinite loop with statement expressions in member
8888       initialization
8889     * [153]6803: Default copy constructor bug in GCC 3.1
8890     * [154]7176: g++ confused by friend and static member with same name
8891     * [155]7188: Segfault with template class and recursive (incorrect)
8892       initializer list
8893     * [156]7306: Regression: GCC 3.x fails to compile code with virtual
8894       inheritance if a method has a variable number of arguments
8895     * [157]7461: ctype<char>::classic_table() returns offset array on
8896       Cygwin
8897     * [158]7524: f(const float arg[3]) fails
8898     * [159]7584: Erroneous ambiguous base error on using declaration
8899     * [160]7676: Member template overloading problem
8900     * [161]7679: infinite loop when a right parenthesis is missing
8901     * [162]7811: default locale not taken from environment
8902     * [163]7961: compare( char *) implemented incorrectly in
8903       basic_string<>
8904     * [164]8071: basic_ostream::operator<<(streambuf*) loops forever if
8905       streambuf::underflow() leaves gptr() NULL (dups: [165]8127,
8906       [166]6745)
8907     * [167]8096: deque::at() throws std::range_error instead of
8908       std::out_of_range
8909     * [168]8127: cout << cin.rdbuf() infinite loop
8910     * [169]8218: Excessively large memory consumed for classes with large
8911       array members
8912     * [170]8287: GCC 3.2: Destructor called for non-constructed local
8913       object
8914     * [171]8347: empty vector range used in string construction causes
8915       core dump
8916     * [172]8348: fail() flag is set in istringstream when eof() flag is
8917       set
8918     * [173]8391: regression: infinite loop in cp/decl2.c(finish_file)
8919
8920    C and optimizer bugs
8921
8922     * [174]6627: -fno-align-functions doesn't seem to disable function
8923       alignment
8924     * [175]6631: life_analysis misoptimizes code to initialize fields of
8925       a structure
8926     * [176]7102: unsigned char division results in floating exception
8927     * [177]7120: Run once loop should *always* be unrolled
8928       (pessimization)
8929     * [178]7209: Bug involving array referencing and ?: operator
8930     * [179]7515: invalid inlining of global function with -O3
8931     * [180]7814: incorrect scheduling for glibc-2.2.92 strcpy test
8932     * [181]8467: bug in sibling call optimization
8933
8934    Preprocessor bugs
8935
8936     * [182]4890: incorrect line markers from the traditional preprocessor
8937     * [183]7357: -M option omits system headers files (making it the same
8938       as -MM)
8939     * [184]7358: Changes to Sun's make Dependencies
8940     * [185]7602: C++ header files found in CPLUS_INCLUDE_PATH treated as
8941       C headers
8942     * [186]7862: Interrupting GCC -MD removes .d file but not .o
8943     * [187]8190: Failed compilation deletes -MD dependency file
8944     * [188]8524: _Pragma within macro is improperly expanded
8945
8946    x86 specific (Intel/AMD)
8947
8948     * [189]5351: (i686-only) function pass-by-value structure copy
8949       corrupts stack ([190]7591 is a duplicate)
8950     * [191]6845, [192]7034, [193]7124, [194]7174: ICE's with
8951       -march=pentium3/pentium2/athlon (these are all the same underlying
8952       bug, in MMX register use)
8953     * [195]7134, [196]7375, [197]7390: ICE with -march=athlon (maybe same
8954       as above?)
8955     * [198]6890: xmmintrin.h, _MM_TRANSPOSE4_PS is broken
8956     * [199]6981: wrong code in 64-bit manipulation on x86
8957     * [200]7242: GCC -mcpu=pentium[23] doesn't define __tune_pentiumpro__
8958       macro
8959     * [201]7396: ix86: cmpgt_ss, cmpge_ss, cmpngt_ss, and cmpnge_ss SSE
8960       intrinsics are broken
8961     * [202]7630: GCC 3.2 breaks on Mozilla 1.0's JS sources with
8962       -march=pentium4
8963     * [203]7693: Typo in i386 mmintrin.h header
8964     * [204]7723: ICE - Pentium3 sse - GCC 3.2
8965     * [205]7951: ICE on -march=pentium4 -O2 -mfpmath=sse
8966     * [206]8146: (i686 only) gcc 3.2 miscompiles gcc 2.95.3
8967
8968    PowerPC specific
8969
8970     * [207]5967: GCC bug when profiling nested functions on powerpc
8971     * [208]6984: wrong code generated with -O2, -O3, -Os for do-while
8972       loop on PowerPC
8973     * [209]7114: PowerPC: ICE building strcoll.op from glibc-2.2.5
8974     * [210]7130: miscompiled code for GCC-3.1 on
8975       powerpc-unknown-linux-gnu with -funroll-all-loops
8976     * [211]7133: PowerPC ICE: unrecognizable insn
8977     * [212]7380: ICE in extract_insn, at recog.c:2148
8978     * [213]8252: ICE on Altivec code with optimization turned on
8979     * [214]8451: Altivec ICE in GCC 3.2
8980
8981    HP/PA specific
8982
8983     * [215]7250: __ashrdi3 returns wrong value on 32 bit hppa
8984
8985    SPARC specific
8986
8987     * [216]6668: when using --disable-multilib, libgcc_s.so is installed
8988       in the wrong place on sparc-solaris
8989     * [217]7151: ICE when compiling for UltraSPARC
8990     * [218]7335: SPARC: ICE in verify_wide_reg (flow.c:557) with long
8991       double and -O1
8992     * [219]7842: [REGRESSION] SPARC code gen bug
8993
8994    ARM specific
8995
8996     * [220]7856: [arm] invalid offset in constant pool reference
8997     * [221]7967: optimization produces wrong code (ARM)
8998
8999    Alpha specific
9000
9001     * [222]7374: __builtin_fabsl broken on alpha
9002
9003    IBM s390 specific
9004
9005     * [223]7370: ICE in fixup_var_refs_1 on s390x
9006     * [224]7409: loop optimization bug on s390x-linux-gnu
9007     * [225]8232: s390x: ICE when using bcmp with int length argument
9008
9009    SCO specific
9010
9011     * [226]7623: SCO OpenServer build fails with machmode.def: undefined
9012       symbol: BITS_PER_UNIT
9013
9014    m68k/Coldfire specific
9015
9016     * [227]8314: crtbegin, crtend need to be multilib'ed for this
9017       platform
9018
9019    Documentation
9020
9021     * [228]761: Document some undocumented options
9022     * [229]5610: Fix documentation about invoking SSE instructions
9023       (-mfpmath=sse)
9024     * [230]7484: List -Wmissing-declarations as C-only option
9025     * [231]7531: -mcmodel not documented for x86-64
9026     * [232]8120: Update documentation of bad use of ##
9027     __________________________________________________________________
9028
9029GCC 3.2
9030
9031   3.2 is a small bug fix release, but there is a change to the
9032   application binary interface (ABI), hence the change to the second part
9033   of the version number.
9034
9035   The main purpose of the 3.2 release is to correct a couple of problems
9036   in the C++ ABI, with the intention of providing a stable interface
9037   going forward.  Accordingly, 3.2 is only a small change to 3.1.1.
9038
9039  Bug Fixes
9040
9041    C++
9042
9043     * [233]7320: g++ 3.2 relocation problem
9044     * [234]7470: vtable: virtual function pointers not in declaration
9045       order
9046
9047    libstdc++
9048
9049     * [235]6410: Trouble with non-ASCII monetary symbols and wchar_t
9050     * [236]6503, [237]6642, [238]7186: Problems with comparing or
9051       subtracting various types of const and non-const iterators
9052     * [239]7216: ambiguity with basic_iostream::traits_type
9053     * [240]7220: problem with basic_istream::ignore(0,delimiter)
9054     * [241]7222: locale::operator==() doesn't work on std::locale("")
9055     * [242]7286: placement operator delete issue
9056     * [243]7442: cxxabi.h does not match the C++ ABI
9057     * [244]7445: poor performance of std::locale::classic() in
9058       multi-threaded applications
9059
9060    x86-64 specific
9061
9062     * [245]7291: off-by-one in generated inline bzero code for x86-64
9063
9064
9065    For questions related to the use of GCC, please consult these web
9066    pages and the [246]GCC manuals. If that fails, the
9067    [247]gcc-help@gcc.gnu.org mailing list might help. Comments on these
9068    web pages and the development of GCC are welcome on our developer
9069    list at [248]gcc@gcc.gnu.org. All of [249]our lists have public
9070    archives.
9071
9072   Copyright (C) [250]Free Software Foundation, Inc. Verbatim copying and
9073   distribution of this entire article is permitted in any medium,
9074   provided this notice is preserved.
9075
9076   These pages are [251]maintained by the GCC team. Last modified
9077   2011-10-24[252].
9078
9079References
9080
9081   1. http://gcc.gnu.org/gcc-3.2/changes.html#3.2.3
9082   2. http://gcc.gnu.org/gcc-3.1/changes.html
9083   3. http://gcc.gnu.org/gcc-3.2/c++-abi.html
9084   4. http://gcc.gnu.org/PR3782
9085   5. http://gcc.gnu.org/PR6440
9086   6. http://gcc.gnu.org/PR7050
9087   7. http://gcc.gnu.org/PR7741
9088   8. http://gcc.gnu.org/PR7982
9089   9. http://gcc.gnu.org/PR8068
9090  10. http://gcc.gnu.org/PR8178
9091  11. http://gcc.gnu.org/PR8396
9092  12. http://gcc.gnu.org/PR8674
9093  13. http://gcc.gnu.org/PR9768
9094  14. http://gcc.gnu.org/PR9798
9095  15. http://gcc.gnu.org/PR9799
9096  16. http://gcc.gnu.org/PR9928
9097  17. http://gcc.gnu.org/PR10114
9098  18. http://gcc.gnu.org/PR10352
9099  19. http://gcc.gnu.org/PR10336
9100  20. http://gcc.gnu.org/PR8224
9101  21. http://gcc.gnu.org/PR8613
9102  22. http://gcc.gnu.org/PR8828
9103  23. http://gcc.gnu.org/PR9226
9104  24. http://gcc.gnu.org/PR9853
9105  25. http://gcc.gnu.org/PR9797
9106  26. http://gcc.gnu.org/PR9967
9107  27. http://gcc.gnu.org/PR10116
9108  28. http://gcc.gnu.org/PR10171
9109  29. http://gcc.gnu.org/PR10175
9110  30. http://gcc.gnu.org/PR8316
9111  31. http://gcc.gnu.org/PR9169
9112  32. http://gcc.gnu.org/PR9420
9113  33. http://gcc.gnu.org/PR9459
9114  34. http://gcc.gnu.org/PR9507
9115  35. http://gcc.gnu.org/PR9538
9116  36. http://gcc.gnu.org/PR9602
9117  37. http://gcc.gnu.org/PR9993
9118  38. http://gcc.gnu.org/PR10167
9119  39. http://gcc.gnu.org/PR9652
9120  40. http://gcc.gnu.org/PR10144
9121  41. http://gcc.gnu.org/PR8746
9122  42. http://gcc.gnu.org/PR9888
9123  43. http://gcc.gnu.org/PR9638
9124  44. http://gcc.gnu.org/PR9954
9125  45. http://gcc.gnu.org/PR7784
9126  46. http://gcc.gnu.org/PR7796
9127  47. http://gcc.gnu.org/PR8281
9128  48. http://gcc.gnu.org/PR8366
9129  49. http://gcc.gnu.org/PR8726
9130  50. http://gcc.gnu.org/PR9414
9131  51. http://gcc.gnu.org/PR10067
9132  52. http://gcc.gnu.org/PR7248
9133  53. http://gcc.gnu.org/PR8343
9134  54. http://gcc.gnu.org/PR9732
9135  55. http://gcc.gnu.org/PR10073
9136  56. http://gcc.gnu.org/PR7702
9137  57. http://gcc.gnu.org/PR9671
9138  58. http://gcc.gnu.org/PR8694
9139  59. http://gcc.gnu.org/PR9953
9140  60. http://gcc.gnu.org/PR10271
9141  61. http://gcc.gnu.org/PR6362
9142  62. http://gcc.gnu.org/PR10377
9143  63. http://gcc.gnu.org/PR6955
9144  64. http://gcc.gnu.org/PR5919
9145  65. http://gcc.gnu.org/PR7129
9146  66. http://gcc.gnu.org/PR7507
9147  67. http://gcc.gnu.org/PR7622
9148  68. http://gcc.gnu.org/PR7681
9149  69. http://gcc.gnu.org/PR9528
9150  70. http://gcc.gnu.org/PR8031
9151  71. http://gcc.gnu.org/PR8275
9152  72. http://gcc.gnu.org/PR8332
9153  73. http://gcc.gnu.org/PR8372
9154  74. http://gcc.gnu.org/PR8439
9155  75. http://gcc.gnu.org/PR8442
9156  76. http://gcc.gnu.org/PR8518
9157  77. http://gcc.gnu.org/PR8615
9158  78. http://gcc.gnu.org/PR8663
9159  79. http://gcc.gnu.org/PR8799
9160  80. http://gcc.gnu.org/PR9328
9161  81. http://gcc.gnu.org/PR9465
9162  82. http://gcc.gnu.org/PR47
9163  83. http://gcc.gnu.org/PR6745
9164  84. http://gcc.gnu.org/PR8214
9165  85. http://gcc.gnu.org/PR8493
9166  86. http://gcc.gnu.org/PR8332
9167  87. http://gcc.gnu.org/PR8503
9168  88. http://gcc.gnu.org/PR8727
9169  89. http://gcc.gnu.org/PR7445
9170  90. http://gcc.gnu.org/PR8230
9171  91. http://gcc.gnu.org/PR8399
9172  92. http://gcc.gnu.org/PR8662
9173  93. http://gcc.gnu.org/PR8707
9174  94. http://gcc.gnu.org/PR8708
9175  95. http://gcc.gnu.org/PR8790
9176  96. http://gcc.gnu.org/PR8887
9177  97. http://gcc.gnu.org/PR9076
9178  98. http://gcc.gnu.org/PR9151
9179  99. http://gcc.gnu.org/PR9168
9180 100. http://gcc.gnu.org/PR9269
9181 101. http://gcc.gnu.org/PR9322
9182 102. http://gcc.gnu.org/PR9433
9183 103. http://gcc.gnu.org/PR8032
9184 104. http://gcc.gnu.org/PR8639
9185 105. http://gcc.gnu.org/PR8794
9186 106. http://gcc.gnu.org/PR8832
9187 107. http://gcc.gnu.org/PR8988
9188 108. http://gcc.gnu.org/PR9492
9189 109. http://gcc.gnu.org/PR9267
9190 110. http://gcc.gnu.org/PR8344
9191 111. http://gcc.gnu.org/PR8524
9192 112. http://gcc.gnu.org/PR8880
9193 113. http://gcc.gnu.org/PR9090
9194 114. http://gcc.gnu.org/PR8588
9195 115. http://gcc.gnu.org/PR8599
9196 116. http://gcc.gnu.org/PR9506
9197 117. http://gcc.gnu.org/PR9484
9198 118. http://gcc.gnu.org/PR9292
9199 119. http://gcc.gnu.org/PR9293
9200 120. http://gcc.gnu.org/PR9295
9201 121. http://gcc.gnu.org/PR9296
9202 122. http://gcc.gnu.org/PR9316
9203 123. http://gcc.gnu.org/PR9493
9204 124. http://gcc.gnu.org/PR7341
9205 125. http://gcc.gnu.org/PR8947
9206 126. http://gcc.gnu.org/PR7448
9207 127. http://gcc.gnu.org/PR8882
9208 128. http://gcc.gnu.org/PR7445
9209 129. http://gcc.gnu.org/PR2521
9210 130. http://gcc.gnu.org/PR5661
9211 131. http://gcc.gnu.org/PR6419
9212 132. http://gcc.gnu.org/PR6994
9213 133. http://gcc.gnu.org/PR7150
9214 134. http://gcc.gnu.org/PR7160
9215 135. http://gcc.gnu.org/PR7228
9216 136. http://gcc.gnu.org/PR7266
9217 137. http://gcc.gnu.org/PR7353
9218 138. http://gcc.gnu.org/PR7411
9219 139. http://gcc.gnu.org/PR7478
9220 140. http://gcc.gnu.org/PR7526
9221 141. http://gcc.gnu.org/PR7721
9222 142. http://gcc.gnu.org/PR7803
9223 143. http://gcc.gnu.org/PR7754
9224 144. http://gcc.gnu.org/PR7788
9225 145. http://gcc.gnu.org/PR8031
9226 146. http://gcc.gnu.org/PR8055
9227 147. http://gcc.gnu.org/PR8067
9228 148. http://gcc.gnu.org/PR8134
9229 149. http://gcc.gnu.org/PR8149
9230 150. http://gcc.gnu.org/PR8160
9231 151. http://gcc.gnu.org/PR5607
9232 152. http://gcc.gnu.org/PR6579
9233 153. http://gcc.gnu.org/PR6803
9234 154. http://gcc.gnu.org/PR7176
9235 155. http://gcc.gnu.org/PR7188
9236 156. http://gcc.gnu.org/PR7306
9237 157. http://gcc.gnu.org/PR7461
9238 158. http://gcc.gnu.org/PR7524
9239 159. http://gcc.gnu.org/PR7584
9240 160. http://gcc.gnu.org/PR7676
9241 161. http://gcc.gnu.org/PR7679
9242 162. http://gcc.gnu.org/PR7811
9243 163. http://gcc.gnu.org/PR7961
9244 164. http://gcc.gnu.org/PR8071
9245 165. http://gcc.gnu.org/PR8127
9246 166. http://gcc.gnu.org/PR6745
9247 167. http://gcc.gnu.org/PR8096
9248 168. http://gcc.gnu.org/PR8127
9249 169. http://gcc.gnu.org/PR8218
9250 170. http://gcc.gnu.org/PR8287
9251 171. http://gcc.gnu.org/PR8347
9252 172. http://gcc.gnu.org/PR8348
9253 173. http://gcc.gnu.org/PR8391
9254 174. http://gcc.gnu.org/PR6627
9255 175. http://gcc.gnu.org/PR6631
9256 176. http://gcc.gnu.org/PR7102
9257 177. http://gcc.gnu.org/PR7120
9258 178. http://gcc.gnu.org/PR7209
9259 179. http://gcc.gnu.org/PR7515
9260 180. http://gcc.gnu.org/PR7814
9261 181. http://gcc.gnu.org/PR8467
9262 182. http://gcc.gnu.org/PR4890
9263 183. http://gcc.gnu.org/PR7357
9264 184. http://gcc.gnu.org/PR7358
9265 185. http://gcc.gnu.org/PR7602
9266 186. http://gcc.gnu.org/PR7862
9267 187. http://gcc.gnu.org/PR8190
9268 188. http://gcc.gnu.org/PR8524
9269 189. http://gcc.gnu.org/PR5351
9270 190. http://gcc.gnu.org/PR7591
9271 191. http://gcc.gnu.org/PR6845
9272 192. http://gcc.gnu.org/PR7034
9273 193. http://gcc.gnu.org/PR7124
9274 194. http://gcc.gnu.org/PR7174
9275 195. http://gcc.gnu.org/PR7134
9276 196. http://gcc.gnu.org/PR7375
9277 197. http://gcc.gnu.org/PR7390
9278 198. http://gcc.gnu.org/PR6890
9279 199. http://gcc.gnu.org/PR6981
9280 200. http://gcc.gnu.org/PR7242
9281 201. http://gcc.gnu.org/PR7396
9282 202. http://gcc.gnu.org/PR7630
9283 203. http://gcc.gnu.org/PR7693
9284 204. http://gcc.gnu.org/PR7723
9285 205. http://gcc.gnu.org/PR7951
9286 206. http://gcc.gnu.org/PR8146
9287 207. http://gcc.gnu.org/PR5967
9288 208. http://gcc.gnu.org/PR6984
9289 209. http://gcc.gnu.org/PR7114
9290 210. http://gcc.gnu.org/PR7130
9291 211. http://gcc.gnu.org/PR7133
9292 212. http://gcc.gnu.org/PR7380
9293 213. http://gcc.gnu.org/PR8252
9294 214. http://gcc.gnu.org/PR8451
9295 215. http://gcc.gnu.org/PR7250
9296 216. http://gcc.gnu.org/PR6668
9297 217. http://gcc.gnu.org/PR7151
9298 218. http://gcc.gnu.org/PR7335
9299 219. http://gcc.gnu.org/PR7842
9300 220. http://gcc.gnu.org/PR7856
9301 221. http://gcc.gnu.org/PR7967
9302 222. http://gcc.gnu.org/PR7374
9303 223. http://gcc.gnu.org/PR7370
9304 224. http://gcc.gnu.org/PR7409
9305 225. http://gcc.gnu.org/PR8232
9306 226. http://gcc.gnu.org/PR7623
9307 227. http://gcc.gnu.org/PR8314
9308 228. http://gcc.gnu.org/PR761
9309 229. http://gcc.gnu.org/PR5610
9310 230. http://gcc.gnu.org/PR7484
9311 231. http://gcc.gnu.org/PR7531
9312 232. http://gcc.gnu.org/PR8120
9313 233. http://gcc.gnu.org/PR7320
9314 234. http://gcc.gnu.org/PR7470
9315 235. http://gcc.gnu.org/PR6410
9316 236. http://gcc.gnu.org/PR6503
9317 237. http://gcc.gnu.org/PR6642
9318 238. http://gcc.gnu.org/PR7186
9319 239. http://gcc.gnu.org/PR7216
9320 240. http://gcc.gnu.org/PR7220
9321 241. http://gcc.gnu.org/PR7222
9322 242. http://gcc.gnu.org/PR7286
9323 243. http://gcc.gnu.org/PR7442
9324 244. http://gcc.gnu.org/PR7445
9325 245. http://gcc.gnu.org/PR7291
9326 246. http://gcc.gnu.org/onlinedocs/
9327 247. mailto:gcc-help@gcc.gnu.org
9328 248. mailto:gcc@gcc.gnu.org
9329 249. http://gcc.gnu.org/lists.html
9330 250. http://www.fsf.org/
9331 251. http://gcc.gnu.org/about.html
9332 252. http://validator.w3.org/check/referer
9333======================================================================
9334http://gcc.gnu.org/gcc-3.1/index.html
9335
9336                                    GCC 3.1
9337
9338   July 27, 2002
9339
9340   The [1]GNU project and the GCC developers are pleased to announce the
9341   release of GCC 3.1.1.
9342
9343   The links below still apply to GCC 3.1.1.
9344
9345   May 15, 2002
9346
9347   The [2]GNU project and the GCC developers are pleased to announce the
9348   release of GCC 3.1.
9349
9350   GCC used to stand for the GNU C Compiler, but since the compiler
9351   supports several other languages aside from C, it now stands for the
9352   GNU Compiler Collection.
9353
9354   A list of [3]successful builds is updated as new information becomes
9355   available.
9356
9357   The GCC developers would like to thank the numerous people that have
9358   contributed [4]new features, improvements, bug fixes, and other changes
9359   as well as test results to GCC. This [5]amazing group of volunteers is
9360   what makes GCC successful.
9361
9362   For additional information about GCC please refer to the [6]GCC project
9363   web site or contact the [7]GCC development mailing list.
9364
9365   To obtain GCC please use [8]our mirror sites, or our CVS server.
9366     __________________________________________________________________
9367
9368
9369    For questions related to the use of GCC, please consult these web
9370    pages and the [9]GCC manuals. If that fails, the
9371    [10]gcc-help@gcc.gnu.org mailing list might help. Comments on these
9372    web pages and the development of GCC are welcome on our developer
9373    list at [11]gcc@gcc.gnu.org. All of [12]our lists have public
9374    archives.
9375
9376   Copyright (C) [13]Free Software Foundation, Inc. Verbatim copying and
9377   distribution of this entire article is permitted in any medium,
9378   provided this notice is preserved.
9379
9380   These pages are [14]maintained by the GCC team. Last modified
9381   2011-04-25[15].
9382
9383References
9384
9385   1. http://www.gnu.org/
9386   2. http://www.gnu.org/
9387   3. http://gcc.gnu.org/gcc-3.1/buildstat.html
9388   4. http://gcc.gnu.org/gcc-3.1/changes.html
9389   5. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
9390   6. http://gcc.gnu.org/index.html
9391   7. mailto:gcc@gcc.gnu.org
9392   8. http://gcc.gnu.org/mirrors.html
9393   9. http://gcc.gnu.org/onlinedocs/
9394  10. mailto:gcc-help@gcc.gnu.org
9395  11. mailto:gcc@gcc.gnu.org
9396  12. http://gcc.gnu.org/lists.html
9397  13. http://www.fsf.org/
9398  14. http://gcc.gnu.org/about.html
9399  15. http://validator.w3.org/check/referer
9400======================================================================
9401http://gcc.gnu.org/gcc-3.1/changes.html
9402
9403                             GCC 3.1 Release Series
9404                        Changes, New Features, and Fixes
9405
9406Additional changes in GCC 3.1.1
9407
9408     * A bug related to how structures and unions are returned has been
9409       fixed for powerpc-*-netbsd*.
9410     * An important bug in the implementation of -fprefetch-loop-arrays
9411       has been fixed. Previously the optimization prefetched random
9412       blocks of memory for most targets except for i386.
9413     * The Java compiler now compiles Java programs much faster and also
9414       works with parallel make.
9415     * Nested functions have been fixed for mips*-*-netbsd*.
9416     * Some missing floating point support routines have beed added for
9417       mips*-*-netbsd*.
9418     * This [1]message gives additional information about the bugs fixed
9419       in this release.
9420
9421Caveats
9422
9423     * The -traditional C compiler option has been deprecated and will be
9424       removed in GCC 3.3. (It remains possible to preprocess non-C code
9425       with the traditional preprocessor.)
9426     * The default debugging format for most ELF platforms (including
9427       GNU/Linux and FreeBSD; notable exception is Solaris) has changed
9428       from stabs to DWARF2. This requires GDB 5.1.1 or later.
9429
9430General Optimizer Improvements
9431
9432     * Jan Hubicka, SuSE Labs, together with Richard Henderson, Red Hat,
9433       and Andreas Jaeger, SuSE Labs, has contributed [2]infrastructure
9434       for profile driven optimizations.
9435       Options -fprofile-arcs and -fbranch-probabilities can now be used
9436       to improve speed of the generated code by profiling the actual
9437       program behaviour on typical runs. In the absence of profile info
9438       the compiler attempts to guess the profile statically.
9439     * [3]SPEC2000 and SPEC95 benchmark suites are now used daily to
9440       monitor performance of the generated code.
9441       According to the SPECInt2000 results on an AMD Athlon CPU, the code
9442       generated by GCC 3.1 is 6% faster on the average (8.2% faster with
9443       profile feedback) compared to GCC 3.0. The code produced by GCC 3.0
9444       is about 2.1% faster compared to 2.95.3. Tests were done using the
9445       -O2 -march=athlon command-line options.
9446     * Alexandre Oliva, of Red Hat, has generalized the tree inlining
9447       infrastructure developed by CodeSourcery, LLC for the C++ front
9448       end, so that it is now used in the C front end too. Inlining
9449       functions as trees exposes them earlier to the compiler, giving it
9450       more opportunities for optimization.
9451     * Support for data prefetching instructions has been added to the GCC
9452       back end and several targets. A new __builtin_prefetch intrinsic is
9453       available to explicitly insert prefetch instructions and
9454       experimental support for loop array prefetching has been added (see
9455       -fprefetch-loop-array documentation).
9456     * Support for emitting debugging information for macros has been
9457       added for DWARF2. It is activated using -g3.
9458
9459New Languages and Language specific improvements
9460
9461  C/C++
9462
9463     * A few more [4]ISO C99 features.
9464     * The preprocessor is 10-50% faster than the preprocessor in GCC 3.0.
9465     * The preprocessor's symbol table has been merged with the symbol
9466       table of the C, C++ and Objective-C front ends.
9467     * The preprocessor consumes less memory than the preprocessor in GCC
9468       3.0, often significantly so. On normal input files, it typically
9469       consumes less memory than pre-3.0 cccp-based GCC, too.
9470
9471  C++
9472
9473     * -fhonor-std and -fno-honor-std have been removed. -fno-honor-std
9474       was a workaround to allow std compliant code to work with the
9475       non-std compliant libstdc++-v2. libstdc++-v3 is std compliant.
9476     * The C++ ABI has been fixed so that void (A::*)() const is mangled
9477       as "M1AKFvvE", rather than "MK1AFvvE" as before. This change only
9478       affects pointer to cv-qualified member function types.
9479     * The C++ ABI has been changed to correctly handle this code:
9480    struct A {
9481      void operator delete[] (void *, size_t);
9482    };
9483
9484    struct B : public A {
9485    };
9486
9487    new B[10];
9488
9489       The amount of storage allocated for the array will be greater than
9490       it was in 3.0, in order to store the number of elements in the
9491       array, so that the correct size can be passed to operator delete[]
9492       when the array is deleted. Previously, the value passed to operator
9493       delete[] was unpredictable.
9494       This change will only affect code that declares a two-argument
9495       operator delete[] with a second parameter of type size_t in a base
9496       class, and does not override that definition in a derived class.
9497     * The C++ ABI has been changed so that:
9498    struct A {
9499      void operator delete[] (void *, size_t);
9500      void operator delete[] (void *);
9501    };
9502
9503       does not cause unnecessary storage to be allocated when an array of
9504       A objects is allocated.
9505       This change will only affect code that declares both of these forms
9506       of operator delete[], and declared the two-argument form before the
9507       one-argument form.
9508     * The C++ ABI has been changed so that when a parameter is passed by
9509       value, any cleanup for that parameter is performed in the caller,
9510       as specified by the ia64 C++ ABI, rather than the called function
9511       as before. As a result, classes with a non-trivial destructor but a
9512       trivial copy constructor will be passed and returned by invisible
9513       reference, rather than by bitwise copy as before.
9514     * G++ now supports the "named return value optimization": for code
9515       like
9516    A f () {
9517      A a;
9518      ...
9519      return a;
9520    }
9521
9522       G++ will allocate a in the return value slot, so that the return
9523       becomes a no-op. For this to work, all return statements in the
9524       function must return the same variable.
9525     * Improvements to the C++ library are listed in [5]the libstdc++-v3
9526       FAQ.
9527
9528  Objective-C
9529
9530     * Annoying linker warnings (due to incorrect code being generated)
9531       have been fixed.
9532     * If a class method cannot be found, the compiler no longer issues a
9533       warning if a corresponding instance method exists in the root
9534       class.
9535     * Forward @protocol declarations have been fixed.
9536     * Loading of categories has been fixed in certain situations (GNU run
9537       time only).
9538     * The class lookup in the run-time library has been rewritten so that
9539       class method dispatch is more than twice as fast as it used to be
9540       (GNU run time only).
9541
9542  Java
9543
9544     * libgcj now includes RMI, java.lang.ref.*, javax.naming, and
9545       javax.transaction.
9546     * Property files and other system resources can be compiled into
9547       executables which use libgcj using the new gcj --resource feature.
9548     * libgcj has been ported to more platforms. In particular there is
9549       now a mostly-functional mingw32 (Windows) target port.
9550     * JNI and CNI invocation interfaces were implemented, so gcj-compiled
9551       Java code can now be called from a C/C++ application.
9552     * gcj can now use builtin functions for certain known methods, for
9553       instance Math.cos.
9554     * gcj can now automatically remove redundant array-store checks in
9555       some common cases.
9556     * The --no-store-checks optimization option was added. This can be
9557       used to omit runtime store checks for code which is known not to
9558       throw ArrayStoreException
9559     * The following third party interface standards were added to libgcj:
9560       org.w3c.dom and org.xml.sax.
9561     * java.security has been merged with GNU Classpath. The new package
9562       is now JDK 1.2 compliant, and much more complete.
9563     * A bytecode verifier was added to the libgcj interpreter.
9564     * java.lang.Character was rewritten to comply with the Unicode 3.0
9565       standard, and improve performance.
9566     * Partial support for many more locales was added to libgcj.
9567     * Socket timeouts have been implemented.
9568     * libgcj has been merged into a single shared library. There are no
9569       longer separate shared libraries for the garbage collector and
9570       zlib.
9571     * Several performance improvements were made to gcj and libgcj:
9572          + Hash synchronization (thin locks)
9573          + A special allocation path for finalizer-free objects
9574          + Thread-local allocation
9575          + Parallel GC, and other GC tweaks
9576
9577  Fortran
9578
9579   Fortran improvements are listed in [6]the Fortran documentation.
9580
9581  Ada
9582
9583   [7]Ada Core Technologies, Inc, has contributed its GNAT Ada 95 front
9584   end and associated tools. The GNAT compiler fully implements the Ada
9585   language as defined by the ISO/IEC 8652 standard.
9586
9587   Please note that the integration of the Ada front end is still work in
9588   progress.
9589
9590New Targets and Target Specific Improvements
9591
9592     * Hans-Peter Nilsson has contributed a port to [8]MMIX, the CPU
9593       architecture used in new editions of Donald E. Knuth's The Art of
9594       Computer Programming.
9595     * [9]Axis Communications has contributed its port to the CRIS CPU
9596       architecture, used in the ETRAX system-on-a-chip series. See
9597       [10]Axis' developer site for technical information.
9598     * Alexandre Oliva, of Red Hat, has contributed a port to the
9599       [11]SuperH SH5 64-bit RISC microprocessor architecture, extending
9600       the existing SH port.
9601     * UltraSPARC is fully supported in 64-bit mode. The option -m64
9602       enables it.
9603     * For compatibility with the Sun compiler #pragma redefine_extname
9604       has been implemented on Solaris.
9605     * The x86 back end has had some noticeable work done to it.
9606          + SuSE Labs developers Jan Hubicka, Bo Thorsen and Andreas
9607            Jaeger have contributed a port to the AMD x86-64 architecture.
9608            For more information on x86-64 see [12]http://www.x86-64.org.
9609          + The compiler now supports MMX, 3DNow!, SSE, and SSE2
9610            instructions. Options -mmmx, -m3dnow, -msse, and -msse2 will
9611            enable the respective instruction sets. Intel C++ compatible
9612            MMX/3DNow!/SSE intrinsics are implemented. SSE2 intrinsics
9613            will be added in next major release.
9614          + Following those improvements, targets for Pentium MMX, K6-2,
9615            K6-3, Pentium III, Pentium 4, and Athlon 4 Mobile/XP/MP were
9616            added. Refer to the documentation on -march= and -mcpu=
9617            options for details.
9618          + For those targets that support it, -mfpmath=sse will cause the
9619            compiler to generate SSE/SSE2 instructions for floating point
9620            math instead of x87 instructions. Usually, this will lead to
9621            quicker code -- especially on the Pentium 4. Note that only
9622            scalar floating point instructions are used and GCC does not
9623            exploit SIMD features yet.
9624          + Prefetch support has been added to the Pentium III, Pentium 4,
9625            K6-2, K6-3, and Athlon series.
9626          + Code generated for floating point to integer conversions has
9627            been improved leading to better performance of many 3D
9628            applications.
9629     * The PowerPC back end has added 64-bit PowerPC GNU/Linux support.
9630     * C++ support for AIX has been improved.
9631     * Aldy Hernandez, of Red Hat, Inc, has contributed extensions to the
9632       PowerPC port supporting the AltiVec programming model (SIMD). The
9633       support, though presently useful, is experimental and is expected
9634       to stabilize for 3.2. The support is written to conform to
9635       Motorola's AltiVec specs. See -maltivec.
9636
9637Obsolete Systems
9638
9639   Support for a number of older systems has been declared obsolete in GCC
9640   3.1. Unless there is activity to revive them, the next release of GCC
9641   will have their sources permanently removed.
9642
9643   All configurations of the following processor architectures have been
9644   declared obsolete:
9645     * MIL-STD-1750A, 1750a-*-*
9646     * AMD A29k, a29k-*-*
9647     * Convex, c*-convex-*
9648     * Clipper, clipper-*-*
9649     * Elxsi, elxsi-*-*
9650     * Intel i860, i860-*-*
9651     * Sun picoJava, pj-*-* and pjl-*-*
9652     * Western Electric 32000, we32k-*-*
9653
9654   Most configurations of the following processor architectures have been
9655   declared obsolete, but we are preserving a few systems which may have
9656   active developers. It is unlikely that the remaining systems will
9657   survive much longer unless we see definite signs of port activity.
9658     * Motorola 88000 except
9659          + Generic a.out, m88k-*-aout*
9660          + Generic SVR4, m88k-*-sysv4
9661          + OpenBSD, m88k-*-openbsd*
9662     * NS32k except
9663          + NetBSD, ns32k-*-netbsd*
9664          + OpenBSD, ns32k-*-openbsd*.
9665     * ROMP except
9666          + OpenBSD, romp-*-openbsd*.
9667
9668   Finally, only some configurations of these processor architectures are
9669   being obsoleted.
9670     * Alpha:
9671          + OSF/1, alpha*-*-osf[123]*. (Digital Unix and Tru64 Unix, aka
9672            alpha*-*-osf[45], are still supported.)
9673     * ARM:
9674          + RISCiX, arm-*-riscix*.
9675     * i386:
9676          + 386BSD, i?86-*-bsd*
9677          + Chorus, i?86-*-chorusos*
9678          + DG/UX, i?86-*-dgux*
9679          + FreeBSD 1.x, i?86-*-freebsd1.*
9680          + IBM AIX, i?86-*-aix*
9681          + ISC UNIX, i?86-*-isc*
9682          + GNU/Linux with pre-BFD linker, i?86-*-linux*oldld*
9683          + NEXTstep, i?86-next-*
9684          + OSF UNIX, i?86-*-osf1* and i?86-*-osfrose*
9685          + RTEMS/coff, i?86-*-rtemscoff*
9686          + RTEMS/go32, i?86-go32-rtems*
9687          + Sequent/BSD, i?86-sequent-bsd*
9688          + Sequent/ptx before version 3, i?86-sequent-ptx[12]* and
9689            i?86-sequent-sysv3*
9690          + SunOS, i?86-*-sunos*
9691     * Motorola 68000:
9692          + Altos, m68[k0]*-altos-*
9693          + Apollo, m68[k0]*-apollo-*
9694          + Apple A/UX, m68[k0]*-apple-*
9695          + Bull, m68[k0]*-bull-*
9696          + Convergent, m68[k0]*-convergent-*
9697          + Generic SVR3, m68[k0]*-*-sysv3*
9698          + ISI, m68[k0]*-isi-*
9699          + LynxOS, m68[k0]*-*-lynxos*
9700          + NEXT, m68[k0]*-next-*
9701          + RTEMS/coff, m68[k0]*-*-rtemscoff*
9702          + Sony, m68[k0]*-sony-*
9703     * MIPS:
9704          + DEC Ultrix, mips-*-ultrix* and mips-dec-*
9705          + Generic BSD, mips-*-bsd*
9706          + Generic System V, mips-*-sysv*
9707          + IRIX before version 5, mips-sgi-irix[1234]*
9708          + RiscOS, mips-*-riscos*
9709          + Sony, mips-sony-*
9710          + Tandem, mips-tandem-*
9711     * SPARC:
9712          + RTEMS/a.out, sparc-*-rtemsaout*.
9713
9714Documentation improvements
9715
9716     * The old manual ("Using and Porting the GNU Compiler Collection")
9717       has been replaced by a users manual ("Using the GNU Compiler
9718       Collection") and a separate internals reference manual ("GNU
9719       Compiler Collection Internals").
9720     * More complete and much improved documentation about GCC's internal
9721       representation used by the C and C++ front ends.
9722     * Many cleanups and improvements in general.
9723
9724
9725    For questions related to the use of GCC, please consult these web
9726    pages and the [13]GCC manuals. If that fails, the
9727    [14]gcc-help@gcc.gnu.org mailing list might help. Comments on these
9728    web pages and the development of GCC are welcome on our developer
9729    list at [15]gcc@gcc.gnu.org. All of [16]our lists have public
9730    archives.
9731
9732   Copyright (C) [17]Free Software Foundation, Inc. Verbatim copying and
9733   distribution of this entire article is permitted in any medium,
9734   provided this notice is preserved.
9735
9736   These pages are [18]maintained by the GCC team. Last modified
9737   2011-04-25[19].
9738
9739References
9740
9741   1. http://gcc.gnu.org/ml/gcc/2002-07/msg01208.html
9742   2. http://gcc.gnu.org/news/profiledriven.html
9743   3. http://gcc.gnu.org/benchmarks/
9744   4. http://gcc.gnu.org/gcc-3.1/c99status.html
9745   5. http://gcc.gnu.org/onlinedocs/libstdc++/faq.html
9746   6. http://gcc.gnu.org/onlinedocs/gcc-3.1.1/g77/News.html
9747   7. http://www.adacore.com/home/
9748   8. http://www-cs-faculty.stanford.edu/~knuth/mmix.html
9749   9. http://www.axis.com/
9750  10. http://developer.axis.com/
9751  11. http://www.superh.com/
9752  12. http://www.x86-64.org/
9753  13. http://gcc.gnu.org/onlinedocs/
9754  14. mailto:gcc-help@gcc.gnu.org
9755  15. mailto:gcc@gcc.gnu.org
9756  16. http://gcc.gnu.org/lists.html
9757  17. http://www.fsf.org/
9758  18. http://gcc.gnu.org/about.html
9759  19. http://validator.w3.org/check/referer
9760======================================================================
9761http://gcc.gnu.org/gcc-3.0/gcc-3.0.html
9762
9763                                   GCC 3.0.4
9764
9765   February 20, 2002
9766
9767   The [1]GNU project and the GCC developers are pleased to announce the
9768   release of GCC 3.0.4, which is a bug-fix release for the GCC 3.0
9769   series.
9770
9771   GCC used to stand for the GNU C Compiler, but since the compiler
9772   supports several other languages aside from C, it now stands for the
9773   GNU Compiler Collection.
9774
9775   GCC 3.0.x has several new optimizations, new targets, new languages and
9776   many other new features, relative to GCC 2.95.x. See the [2]new
9777   features page for a more complete list.
9778
9779   A list of [3]successful builds is updated as new information becomes
9780   available.
9781
9782   The GCC developers would like to thank the numerous people that have
9783   contributed new features, test results, bug fixes, etc to GCC. This
9784   [4]amazing group of volunteers is what makes GCC successful.
9785
9786   And finally, we can't in good conscience fail to mention some
9787   [5]caveats to using GCC 3.0.x.
9788
9789   For additional information about GCC please refer to the [6]GCC project
9790   web site or contact the [7]GCC development mailing list.
9791
9792   To obtain GCC please use [8]our mirror sites, or our CVS server.
9793     __________________________________________________________________
9794
9795Previous 3.0.x Releases
9796
9797   December 20, 2001: GCC 3.0.3 has been released.
9798   October 25, 2001: GCC 3.0.2 has been released.
9799   August 20, 2001: GCC 3.0.1 has been released.
9800   June 18, 2001: GCC 3.0 has been released.
9801
9802
9803    For questions related to the use of GCC, please consult these web
9804    pages and the [9]GCC manuals. If that fails, the
9805    [10]gcc-help@gcc.gnu.org mailing list might help. Comments on these
9806    web pages and the development of GCC are welcome on our developer
9807    list at [11]gcc@gcc.gnu.org. All of [12]our lists have public
9808    archives.
9809
9810   Copyright (C) [13]Free Software Foundation, Inc. Verbatim copying and
9811   distribution of this entire article is permitted in any medium,
9812   provided this notice is preserved.
9813
9814   These pages are [14]maintained by the GCC team. Last modified
9815   2011-04-25[15].
9816
9817References
9818
9819   1. http://www.gnu.org/
9820   2. http://gcc.gnu.org/gcc-3.0/features.html
9821   3. http://gcc.gnu.org/gcc-3.0/buildstat.html
9822   4. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
9823   5. http://gcc.gnu.org/gcc-3.0/caveats.html
9824   6. http://gcc.gnu.org/index.html
9825   7. mailto:gcc@gcc.gnu.org
9826   8. http://gcc.gnu.org/mirrors.html
9827   9. http://gcc.gnu.org/onlinedocs/
9828  10. mailto:gcc-help@gcc.gnu.org
9829  11. mailto:gcc@gcc.gnu.org
9830  12. http://gcc.gnu.org/lists.html
9831  13. http://www.fsf.org/
9832  14. http://gcc.gnu.org/about.html
9833  15. http://validator.w3.org/check/referer
9834======================================================================
9835http://gcc.gnu.org/gcc-3.0/features.html
9836
9837                              GCC 3.0 New Features
9838
9839Additional changes in GCC 3.0.4
9840
9841     * GCC 3.0 now supports newer versions of the [1]NetBSD operating
9842       system, which use the ELF object file format, on x86 processors.
9843     * Correct debugging information is generated from functions that have
9844       lines from multiple files (e.g. yacc output).
9845     * A fix for whitespace handling in the -traditional preprocessor,
9846       which can affect Fortran.
9847     * Fixes to the exception handling runtime.
9848     * More fixes for bad code generation in C++.
9849     * A fix for shared library generation under AIX 4.3.
9850     * Documentation updates.
9851     * Port of GCC to Tensilica's Xtensa processor contributed.
9852     * A fix for compiling the PPC Linux kernel (FAT fs wouldn't link).
9853
9854Additional changes in GCC 3.0.3
9855
9856     * A fix to correct an accidental change to the PowerPC ABI.
9857     * Fixes for bad code generation on a variety of architectures.
9858     * Improvements to the debugging information generated for C++
9859       classes.
9860     * Fixes for bad code generation in C++.
9861     * A fix to avoid crashes in the C++ demangler.
9862     * A fix to the C++ standard library to avoid buffer overflows.
9863     * Miscellaneous improvements for a variety of architectures.
9864
9865Additional changes in GCC 3.0.2
9866
9867     * Fixes for bad code generation during loop unrolling.
9868     * Fixes for bad code generation by the sibling call optimization.
9869     * Minor improvements to x86 code generation.
9870     * Implementation of function descriptors in C++ vtables for IA64.
9871     * Numerous minor bug-fixes.
9872
9873Additional changes in GCC 3.0.1
9874
9875     * C++ fixes for incorrect code-generation.
9876     * Improved cross-compiling support for the C++ standard library.
9877     * Fixes for some embedded targets that worked in GCC 2.95.3, but not
9878       in GCC 3.0.
9879     * Fixes for various exception-handling bugs.
9880     * A port to the S/390 architecture.
9881
9882General Optimizer Improvements
9883
9884     * [2]Basic block reordering pass.
9885     * New if-conversion pass with support for conditional (predicated)
9886       execution.
9887     * New tail call and sibling call elimination optimizations.
9888     * New register renaming pass.
9889     * New (experimental) [3]static single assignment (SSA) representation
9890       support.
9891     * New dead-code elimination pass implemented using the SSA
9892       representation.
9893     * [4]Global null pointer test elimination.
9894     * [5]Global code hoisting/unification.
9895     * More builtins and optimizations for stdio.h, string.h and old BSD
9896       functions, as well as for ISO C99 functions.
9897     * New builtin __builtin_expect for giving hints to the branch
9898       predictor.
9899
9900New Languages and Language specific improvements
9901
9902     * The GNU Compiler for the Java(TM) language (GCJ) is now integrated
9903       and supported, including the run-time library containing most
9904       common non-GUI Java classes, a bytecode interpreter, and the Boehm
9905       conservative garbage collector. Many bugs have been fixed. GCJ can
9906       compile Java source or Java bytecodes to either native code or Java
9907       class files, and supports native methods written in either the
9908       standard JNI or the more efficient and convenient CNI.
9909     * Here is a [6]partial list of C++ improvements, both new features
9910       and those no longer supported.
9911     * New C++ ABI. On the IA-64 platform GCC is capable of
9912       inter-operating with other IA-64 compilers.
9913     * The new ABI also significantly reduces the size of symbol and debug
9914       information.
9915     * New [7]C++ support library and many C++ bug fixes, vastly improving
9916       our conformance to the ISO C++ standard.
9917     * New [8]inliner for C++.
9918     * Rewritten C preprocessor, integrated into the C, C++ and Objective
9919       C compilers, with very many improvements including ISO C99 support
9920       and [9]improvements to dependency generation.
9921     * Support for more [10]ISO C99 features.
9922     * Many improvements to support for checking calls to format functions
9923       such as printf and scanf, including support for ISO C99 format
9924       features, extensions from the Single Unix Specification and GNU
9925       libc 2.2, checking of strfmon formats and features to assist in
9926       auditing for format string security bugs.
9927     * New warnings for C code that may have undefined semantics because
9928       of violations of sequence point rules in the C standard (such as a
9929       = a++;, a[n] = b[n++]; and a[i++] = i;), included in -Wall.
9930     * Additional warning option -Wfloat-equal.
9931     * Improvements to -Wtraditional.
9932     * Fortran improvements are listed in [11]the Fortran documentation.
9933
9934New Targets and Target Specific Improvements
9935
9936     * New x86 back-end, generating much improved code.
9937     * Support for a generic i386-elf target contributed.
9938     * New option to emit x86 assembly code using Intel style syntax
9939       (-mintel-syntax).
9940     * HPUX 11 support contributed.
9941     * Improved PowerPC code generation, including scheduled prologue and
9942       epilogue.
9943     * Port of GCC to Intel's IA-64 processor contributed.
9944     * Port of GCC to Motorola's MCore 210 and 340 contributed.
9945     * New unified back-end for Arm, Thumb and StrongArm contributed.
9946     * Port of GCC to Intel's XScale processor contributed.
9947     * Port of GCC to Atmel's AVR microcontrollers contributed.
9948     * Port of GCC to Mitsubishi's D30V processor contributed.
9949     * Port of GCC to Matsushita's AM33 processor (a member of the MN10300
9950       processor family) contributed.
9951     * Port of GCC to Fujitsu's FR30 processor contributed.
9952     * Port of GCC to Motorola's 68HC11 and 68HC12 processors contributed.
9953     * Port of GCC to Sun's picoJava processor core contributed.
9954
9955Documentation improvements
9956
9957     * Substantially rewritten and improved C preprocessor manual.
9958     * Many improvements to other documentation.
9959     * Manpages for gcc, cpp and gcov are now generated automatically from
9960       the master Texinfo manual, eliminating the problem of manpages
9961       being out of date. (The generated manpages are only extracts from
9962       the full manual, which is provided in Texinfo form, from which
9963       info, HTML, other formats and a printed manual can be generated.)
9964     * Generated info files are included in the release tarballs alongside
9965       their Texinfo sources, avoiding problems on some platforms with
9966       building makeinfo as part of the GCC distribution.
9967
9968Other significant improvements
9969
9970     * Garbage collection used internally by the compiler for most memory
9971       allocation instead of obstacks.
9972     * Lengauer and Tarjan algorithm used for computing dominators in the
9973       CFG. This algorithm can be significantly faster and more space
9974       efficient than our older algorithm.
9975     * gccbug script provided to assist in submitting bug reports to our
9976       bug tracking system. (Bug reports previously submitted directly to
9977       our mailing lists, for which you received no bug tracking number,
9978       should be submitted again using gccbug if you can reproduce the
9979       problem with GCC 3.0.)
9980     * The internal libgcc library is [12]built as a shared library on
9981       systems that support it.
9982     * Extensive testsuite included with GCC, with many new tests. In
9983       addition to tests for GCC bugs that have been fixed, many tests
9984       have been added for language features, compiler warnings and
9985       builtin functions.
9986     * Additional language-independent warning options -Wpacked, -Wpadded,
9987       -Wunreachable-code and -Wdisabled-optimization.
9988     * Target-independent options -falign-functions, -falign-loops and
9989       -falign-jumps.
9990
9991   Plus a great many bug fixes and almost all the [13]features found in
9992   GCC 2.95.
9993
9994
9995    For questions related to the use of GCC, please consult these web
9996    pages and the [14]GCC manuals. If that fails, the
9997    [15]gcc-help@gcc.gnu.org mailing list might help. Comments on these
9998    web pages and the development of GCC are welcome on our developer
9999    list at [16]gcc@gcc.gnu.org. All of [17]our lists have public
10000    archives.
10001
10002   Copyright (C) [18]Free Software Foundation, Inc. Verbatim copying and
10003   distribution of this entire article is permitted in any medium,
10004   provided this notice is preserved.
10005
10006   These pages are [19]maintained by the GCC team. Last modified
10007   2011-04-25[20].
10008
10009References
10010
10011   1. http://www.netbsd.org/
10012   2. http://gcc.gnu.org/news/reorder.html
10013   3. http://gcc.gnu.org/news/ssa.html
10014   4. http://gcc.gnu.org/news/null.html
10015   5. http://gcc.gnu.org/news/unify.html
10016   6. http://gcc.gnu.org/gcc-3.0/c++features.html
10017   7. http://gcc.gnu.org/libstdc++/
10018   8. http://gcc.gnu.org/news/inlining.html
10019   9. http://gcc.gnu.org/news/dependencies.html
10020  10. http://gcc.gnu.org/gcc-3.0/c99status.html
10021  11. http://gcc.gnu.org/onlinedocs/gcc-3.4.6/g77/News.html
10022  12. http://gcc.gnu.org/gcc-3.0/libgcc.html
10023  13. http://gcc.gnu.org/gcc-2.95/features.html
10024  14. http://gcc.gnu.org/onlinedocs/
10025  15. mailto:gcc-help@gcc.gnu.org
10026  16. mailto:gcc@gcc.gnu.org
10027  17. http://gcc.gnu.org/lists.html
10028  18. http://www.fsf.org/
10029  19. http://gcc.gnu.org/about.html
10030  20. http://validator.w3.org/check/referer
10031======================================================================
10032http://gcc.gnu.org/gcc-3.0/caveats.html
10033
10034                                GCC 3.0 Caveats
10035
10036     * -fstrict-aliasing is now part of -O2 and higher optimization
10037       levels. This allows the compiler to assume the strictest aliasing
10038       rules applicable to the language being compiled. For C and C++,
10039       this activates optimizations based on the type of expressions. This
10040       optimization may thus break old, non-compliant code.
10041     * Enumerations are now properly promoted to int in function
10042       parameters and function returns. Normally this change is not
10043       visible, but when using -fshort-enums this is an ABI change.
10044     * The undocumented extension that allowed C programs to have a label
10045       at the end of a compound statement has been deprecated and may be
10046       removed in a future version. Programs that now generate a warning
10047       about this may be fixed by adding a null statement (a single
10048       semicolon) after the label.
10049     * The poorly documented extension that allowed string constants in C,
10050       C++ and Objective C to contain unescaped newlines has been
10051       deprecated and may be removed in a future version. Programs using
10052       this extension may be fixed in several ways: the bare newline may
10053       be replaced by \n, or preceded by \n\, or string concatenation may
10054       be used with the bare newline preceded by \n" and " placed at the
10055       start of the next line.
10056     * The Chill compiler is not included in GCC 3.0, because of the lack
10057       of a volunteer to convert it to use garbage collection.
10058     * Certain non-standard iostream methods from earlier versions of
10059       libstdc++ are not included in libstdc++ v3, i.e. filebuf::attach,
10060       ostream::form, and istream::gets.
10061     * The new C++ ABI is not yet fully supported by current (as of
10062       2001-07-01) releases and development versions of GDB, or any
10063       earlier versions. There is a problem setting breakpoints by line
10064       number, and other related issues that have been fixed in GCC 3.0
10065       but not yet handled in GDB:
10066       [1]http://gcc.gnu.org/ml/gcc-bugs/2001-06/msg00421.html
10067
10068
10069    For questions related to the use of GCC, please consult these web
10070    pages and the [2]GCC manuals. If that fails, the
10071    [3]gcc-help@gcc.gnu.org mailing list might help. Comments on these
10072    web pages and the development of GCC are welcome on our developer
10073    list at [4]gcc@gcc.gnu.org. All of [5]our lists have public archives.
10074
10075   Copyright (C) [6]Free Software Foundation, Inc. Verbatim copying and
10076   distribution of this entire article is permitted in any medium,
10077   provided this notice is preserved.
10078
10079   These pages are [7]maintained by the GCC team. Last modified
10080   2011-04-25[8].
10081
10082References
10083
10084   1. http://gcc.gnu.org/ml/gcc-bugs/2001-06/msg00421.html
10085   2. http://gcc.gnu.org/onlinedocs/
10086   3. mailto:gcc-help@gcc.gnu.org
10087   4. mailto:gcc@gcc.gnu.org
10088   5. http://gcc.gnu.org/lists.html
10089   6. http://www.fsf.org/
10090   7. http://gcc.gnu.org/about.html
10091   8. http://validator.w3.org/check/referer
10092======================================================================
10093http://gcc.gnu.org/gcc-2.95/index.html
10094
10095                                    GCC 2.95
10096
10097   March 16, 2001: The GNU project and the GCC developers are pleased to
10098   announce the release of GCC version 2.95.3.
10099
10100Release History
10101
10102   GCC 2.95.3
10103          March 16, 2001
10104
10105   GCC 2.95.2
10106          October 27, 1999
10107
10108   GCC 2.95.1
10109          August 19, 1999
10110
10111   GCC 2.95
10112          July 31, 1999. This is the first release of GCC since the April
10113          1999 GCC/EGCS reunification and includes nearly a year's worth
10114          of new development and bugfixes.
10115
10116References and Acknowledgements
10117
10118   GCC used to stand for the GNU C Compiler, but since the compiler
10119   supports several other languages aside from C, it now stands for the
10120   GNU Compiler Collection.
10121
10122   The whole suite has been extensively [1]regression tested and
10123   [2]package tested. It should be reliable and suitable for widespread
10124   use.
10125
10126   The compiler has several new optimizations, new targets, new languages
10127   and other new features. See the [3]new features page for a more
10128   complete list of new features found in the GCC 2.95 releases.
10129
10130   The sources include installation instructions in both HTML and
10131   plaintext forms in the install directory in the distribution. However,
10132   the most up to date [4]installation instructions and [5]build/test
10133   status are on the web pages. We will update those pages as new
10134   information becomes available.
10135
10136   The GCC developers would like to thank the numerous people that have
10137   contributed new features, test results, bugfixes, etc to GCC. This
10138   [6]amazing group of volunteers is what makes GCC successful.
10139
10140   And finally, we can't in good conscience fail to mention some
10141   [7]caveats to using GCC 2.95.
10142
10143   Download GCC 2.95 from one of our many [8]mirror sites.
10144
10145   For additional information about GCC please see the [9]GCC project web
10146   server or contact the [10]GCC development mailing list.
10147
10148
10149    For questions related to the use of GCC, please consult these web
10150    pages and the [11]GCC manuals. If that fails, the
10151    [12]gcc-help@gcc.gnu.org mailing list might help. Comments on these
10152    web pages and the development of GCC are welcome on our developer
10153    list at [13]gcc@gcc.gnu.org. All of [14]our lists have public
10154    archives.
10155
10156   Copyright (C) [15]Free Software Foundation, Inc. Verbatim copying and
10157   distribution of this entire article is permitted in any medium,
10158   provided this notice is preserved.
10159
10160   These pages are [16]maintained by the GCC team. Last modified
10161   2011-04-25[17].
10162
10163References
10164
10165   1. http://gcc.gnu.org/gcc-2.95/regress.html
10166   2. http://gcc.gnu.org/gcc-2.95/othertest.html
10167   3. http://gcc.gnu.org/gcc-2.95/features.html
10168   4. http://gcc.gnu.org/install/
10169   5. http://gcc.gnu.org/gcc-2.95/buildstat.html
10170   6. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
10171   7. http://gcc.gnu.org/gcc-2.95/caveats.html
10172   8. http://gcc.gnu.org/mirrors.html
10173   9. http://gcc.gnu.org/index.html
10174  10. mailto:gcc@gcc.gnu.org
10175  11. http://gcc.gnu.org/onlinedocs/
10176  12. mailto:gcc-help@gcc.gnu.org
10177  13. mailto:gcc@gcc.gnu.org
10178  14. http://gcc.gnu.org/lists.html
10179  15. http://www.fsf.org/
10180  16. http://gcc.gnu.org/about.html
10181  17. http://validator.w3.org/check/referer
10182======================================================================
10183http://gcc.gnu.org/gcc-2.95/features.html
10184
10185                             GCC 2.95 New Features
10186
10187     * General Optimizer Improvements:
10188          + [1]Localized register spilling to improve speed and code
10189            density especially on small register class machines.
10190          + [2]Global CSE using lazy code motion algorithms.
10191          + [3]Improved global constant/copy propagation.
10192          + [4]Improved control flow graph analysis and manipulation.
10193          + [5]Local dead store elimination.
10194          + [6]Memory Load hoisting/store sinking in loops.
10195          + [7]Type based alias analysis is enabled by default. Note this
10196            feature will expose bugs in the Linux kernel. Please refer to
10197            the FAQ (as shipped with GCC 2.95) for additional information
10198            on this issue.
10199          + Major revamp of GIV detection, combination and simplification
10200            to improve loop performance.
10201          + Major improvements to register allocation and reloading.
10202     * New Languages and Language specific improvements
10203          + [8]Many C++ improvements.
10204          + [9]Many Fortran improvements.
10205          + [10]Java front-end has been integrated. [11]runtime library is
10206            available separately.
10207          + [12]ISO C99 support
10208          + [13]Chill front-end and runtime has been integrated.
10209          + Boehm garbage collector support in libobjc.
10210          + More support for various pragmas which appear in vendor
10211            include files
10212     * New Targets and Target Specific Improvements
10213          + [14]SPARC backend rewrite.
10214          + -mschedule=8000 will optimize code for PA8000 class
10215            processors; -mpa-risc-2-0 will generate code for PA2.0
10216            processors
10217          + Various micro-optimizations for the ia32 port. K6
10218            optimizations
10219          + Compiler will attempt to align doubles in the stack on the
10220            ia32 port
10221          + Alpha EV6 support
10222          + PowerPC 750
10223          + RS6000/PowerPC: -mcpu=401 was added as an alias for -mcpu=403.
10224            -mcpu=e603e was added to do -mcpu=603e and -msoft-float.
10225          + c3x, c4x
10226          + HyperSPARC
10227          + SparcLite86x
10228          + sh4
10229          + Support for new systems (OpenBSD, FreeBSD, UWIN, Interix,
10230            arm-linux)
10231          + vxWorks targets include support for vxWorks threads
10232          + StrongARM 110 and ARM9 support added. ARM Scheduling
10233            parameters rewritten.
10234          + Various changes to the MIPS port to avoid assembler macros,
10235            which in turn improves performance
10236          + Various performance improvements to the i960 port.
10237          + Major rewrite of ns32k port
10238     * Other significant improvements
10239          + [15]Ability to dump cfg information and display it using vcg.
10240          + The new faster scheme for fixing vendor header files is
10241            enabled by default.
10242          + Experimental internationalization support.
10243          + multibyte character support
10244          + Some compile-time speedups for pathological problems
10245          + Better support for complex types
10246     * Plus the usual mountain of bugfixes
10247     * Core compiler is based on the gcc2 development tree from Sept 30,
10248       1998, so we have all of the [16]features found in GCC 2.8.
10249
10250Additional Changes in GCC 2.95.1
10251
10252     * Generic bugfixes and improvements
10253          + Various documentation fixes related to the GCC/EGCS merger.
10254          + Fix memory management bug which could lead to spurious aborts,
10255            core dumps or random parsing errors in the compiler.
10256          + Fix a couple bugs in the dwarf1 and dwarf2 debug record
10257            support.
10258          + Fix infinite loop in the CSE optimizer.
10259          + Avoid undefined behavior in compiler FP emulation code
10260          + Fix install problem when prefix is overridden on the make
10261            install command.
10262          + Fix problem with unwanted installation of assert.h on some
10263            systems.
10264          + Fix problem with finding the wrong assembler in a single tree
10265            build.
10266          + Avoid increasing the known alignment of a register that is
10267            already known to be a pointer.
10268     * Platform specific bugfixes and improvements
10269          + Codegen bugfix for prologue/epilogue for cpu32 target.
10270          + Fix long long code generation bug for the Coldfire target.
10271          + Fix various aborts in the SH compiler.
10272          + Fix bugs in libgcc support library for the SH.
10273          + Fix alpha ev6 code generation bug.
10274          + Fix problems with EXIT_SUCCESS/EXIT_FAILURE redefinitions on
10275            AIX platforms.
10276          + Fix -fpic code generation bug for rs6000/ppc svr4 targets.
10277          + Fix varargs/stdarg code generation bug for rs6000/ppc svr4
10278            targets.
10279          + Fix weak symbol handling for rs6000/ppc svr4 targets.
10280          + Fix various problems with 64bit code generation for the
10281            rs6000/ppc port.
10282          + Fix codegen bug which caused tetex to be mis-compiled on the
10283            x86.
10284          + Fix compiler abort in new cfg code exposed by x86 port.
10285          + Fix out of range array reference in code convert flat
10286            registers to the x87 stacked FP register file.
10287          + Fix minor vxworks configuration bug.
10288          + Fix return type of bsearch for SunOS 4.x.
10289     * Language & Runtime specific fixes.
10290          + The G++ signature extension has been deprecated. It will be
10291            removed in the next major release of G++. Use of signatures
10292            will result in a warning from the compiler.
10293          + Several bugs relating to templates and namespaces were fixed.
10294          + A bug that caused crashes when combining templates with -g on
10295            DWARF1 platforms was fixed.
10296          + Pointers-to-members, virtual functions, and multiple
10297            inheritance should now work together correctly.
10298          + Some code-generation bugs relating to function try blocks were
10299            fixed.
10300          + G++ is a little bit more lenient with certain archaic
10301            constructs than in GCC 2.95.
10302          + Fix to prevent shared library version #s from bring truncated
10303            to 1 digit
10304          + Fix missing std:: in the libstdc++ library.
10305          + Fix stream locking problems in libio.
10306          + Fix problem in java compiler driver.
10307
10308Additional Changes in GCC 2.95.2
10309
10310   The -fstrict-aliasing is not enabled by default for GCC 2.95.2. While
10311   the optimizations performed by -fstrict-aliasing are valid according to
10312   the C and C++ standards, the optimization have caused some problems,
10313   particularly with old non-conforming code.
10314
10315   The GCC developers are experimenting with ways to warn users about code
10316   which violates the C/C++ standards, but those warnings are not ready
10317   for widespread use at this time. Rather than wait for those warnings
10318   the GCC developers have chosen to disable -fstrict-aliasing by default
10319   for the GCC 2.95.2 release.
10320
10321   We strongly encourage developers to find and fix code which violates
10322   the C/C++ standards as -fstrict-aliasing may be enabled by default in
10323   future releases. Use the option -fstrict-aliasing to re-enable these
10324   optimizations.
10325     * Generic bugfixes and improvements
10326          + Fix incorrectly optimized memory reference in global common
10327            subexpression elimination (GCSE) optimization pass.
10328          + Fix code generation bug in regmove.c in which it could
10329            incorrectly change a "const" value.
10330          + Fix bug in optimization of conditionals involving volatile
10331            memory references.
10332          + Avoid over-allocation of stack space for some procedures.
10333          + Fixed bug in the compiler which caused incorrect optimization
10334            of an obscure series of bit manipulations, shifts and
10335            arithmetic.
10336          + Fixed register allocator bug which caused teTeX to be
10337            mis-compiled on SPARC targets.
10338          + Avoid incorrect optimization of degenerate case statements for
10339            certain targets such as the ARM.
10340          + Fix out of range memory reference in the jump optimizer.
10341          + Avoid dereferencing null pointer in fix-header.
10342          + Fix test for GCC specific features so that it is possible to
10343            bootstrap with gcc-2.6.2 and older versions of GCC.
10344          + Fix typo in scheduler which could potentially cause out of
10345            range memory accesses.
10346          + Avoid incorrect loop reversal which caused incorrect code for
10347            certain loops on PowerPC targets.
10348          + Avoid incorrect optimization of switch statements on certain
10349            targets (for example the ARM).
10350     * Platform specific bugfixes and improvements
10351          + Work around bug in Sun V5.0 compilers which caused bootstrap
10352            comparison failures on SPARC targets.
10353          + Fix SPARC backend bug which caused aborts in final.c.
10354          + Fix sparc-hal-solaris2* configuration fragments.
10355          + Fix bug in sparc block profiling.
10356          + Fix obscure code generation bug for the PARISC targets.
10357          + Define __STDC_EXT__ for HPUX configurations.
10358          + Various POWERPC64 code generation bugfixes.
10359          + Fix abort for PPC targets using ELF (ex GNU/Linux).
10360          + Fix collect2 problems for AIX targets.
10361          + Correct handling of .file directive for PPC targets.
10362          + Fix bug in fix_trunc x86 patterns.
10363          + Fix x86 port to correctly pop the FP stack for functions that
10364            return structures in memory.
10365          + Fix minor bug in strlen x86 pattern.
10366          + Use stabs debugging instead of dwarf1 for x86-solaris targets.
10367          + Fix template repository code to handle leading underscore in
10368            mangled names.
10369          + Fix weak/weak alias support for OpenBSD.
10370          + GNU/Linux for the ARM has C++ compatible include files.
10371     * Language & Runtime specific fixes.
10372          + Fix handling of constructor attribute in the C front-end which
10373            caused problems building the Chill runtime library on some
10374            targets.
10375          + Fix minor problem merging type qualifiers in the C front-end.
10376          + Fix aliasing bug for pointers and references (C/C++).
10377          + Fix incorrect "non-constant initializer bug" when -traditional
10378            or -fwritable-strings is enabled.
10379          + Fix build error for Chill front-end on SunOS.
10380          + Do not complain about duplicate instantiations when using
10381            -frepo (C++).
10382          + Fix array bounds handling in C++ front-end which caused
10383            problems with dwarf debugging information in some
10384            circumstances.
10385          + Fix minor namespace problem.
10386          + Fix problem linking java programs.
10387
10388Additional Changes in GCC 2.95.3
10389
10390     * Generic bugfixes and improvements
10391          + Fix numerous problems that caused incorrect optimization in
10392            the register reloading code.
10393          + Fix numerous problems that caused incorrect optimization in
10394            the loop optimizer.
10395          + Fix aborts in the functions build_insn_chain and scan_loops
10396            under some circumstances.
10397          + Fix an alias analysis bug.
10398          + Fix an infinite compilation bug in the combiner.
10399          + A few problems with complex number support have been fixed.
10400          + It is no longer possible for gcc to act as a fork bomb when
10401            installed incorrectly.
10402          + The -fpack-struct option should be recognized now.
10403          + Fixed a bug that caused incorrect code to be generated due to
10404            a lost stack adjustment.
10405     * Platform specific bugfixes and improvements
10406          + Support building ARM toolchains hosted on Windows.
10407          + Fix attribute calculations in ARM toolchains.
10408          + arm-linux support has been improved.
10409          + Fix a PIC failure on sparc targets.
10410          + On ix86 targets, the regparm attribute should now work
10411            reliably.
10412          + Several updates for the h8300 port.
10413          + Fix problem building libio with glibc 2.2.
10414
10415
10416    For questions related to the use of GCC, please consult these web
10417    pages and the [17]GCC manuals. If that fails, the
10418    [18]gcc-help@gcc.gnu.org mailing list might help. Comments on these
10419    web pages and the development of GCC are welcome on our developer
10420    list at [19]gcc@gcc.gnu.org. All of [20]our lists have public
10421    archives.
10422
10423   Copyright (C) [21]Free Software Foundation, Inc. Verbatim copying and
10424   distribution of this entire article is permitted in any medium,
10425   provided this notice is preserved.
10426
10427   These pages are [22]maintained by the GCC team. Last modified
10428   2011-04-25[23].
10429
10430References
10431
10432   1. http://gcc.gnu.org/news/spill.html
10433   2. http://gcc.gnu.org/news/lcm.html
10434   3. http://gcc.gnu.org/news/cprop.html
10435   4. http://gcc.gnu.org/news/cfg.html
10436   5. http://gcc.gnu.org/news/dse.html
10437   6. http://gcc.gnu.org/news/hoist.html
10438   7. http://gcc.gnu.org/news/alias.html
10439   8. http://gcc.gnu.org/gcc-2.95/c++features.html
10440   9. http://gcc.gnu.org/onlinedocs/gcc-3.4.6/g77/News.html
10441  10. http://gcc.gnu.org/java/gcj-announce.txt
10442  11. http://gcc.gnu.org/news/javaannounce.html
10443  12. http://gcc.gnu.org/c99status.html
10444  13. http://gcc.gnu.org/news/chill.html
10445  14. http://gcc.gnu.org/news/sparc.html
10446  15. http://gcc.gnu.org/news/egcs-vcg.html
10447  16. http://gcc.gnu.org/egcs-1.0/features-2.8.html
10448  17. http://gcc.gnu.org/onlinedocs/
10449  18. mailto:gcc-help@gcc.gnu.org
10450  19. mailto:gcc@gcc.gnu.org
10451  20. http://gcc.gnu.org/lists.html
10452  21. http://www.fsf.org/
10453  22. http://gcc.gnu.org/about.html
10454  23. http://validator.w3.org/check/referer
10455======================================================================
10456http://gcc.gnu.org/gcc-2.95/caveats.html
10457
10458                                GCC 2.95 Caveats
10459
10460     * GCC 2.95 will issue an error for invalid asm statements that had
10461       been silently accepted by earlier versions of the compiler. This is
10462       particularly noticeable when compiling older versions of the Linux
10463       kernel (2.0.xx). Please refer to the FAQ (as shipped with GCC 2.95)
10464       for more information on this issue.
10465     * GCC 2.95 implements type based alias analysis to disambiguate
10466       memory references. Some programs, particularly the Linux kernel
10467       violate ANSI/ISO aliasing rules and therefore may not operate
10468       correctly when compiled with GCC 2.95. Please refer to the FAQ (as
10469       shipped with GCC 2.95) for more information on this issue.
10470     * GCC 2.95 has a known bug in its handling of complex variables for
10471       64bit targets. Instead of silently generating incorrect code, GCC
10472       2.95 will issue a fatal error for situations it can not handle.
10473       This primarily affects the Fortran community as Fortran makes more
10474       use of complex variables than C or C++.
10475     * GCC 2.95 has an integrated libstdc++, but does not have an
10476       integrated libg++. Furthermore old libg++ releases will not work
10477       with GCC 2.95. You can retrieve a recent copy of libg++ from the
10478       [1]GCC ftp server.
10479       Note most C++ programs only need libstdc++.
10480     * Exception handling may not work with shared libraries, particularly
10481       on alphas, hppas, rs6000/powerpc and mips based platforms.
10482       Exception handling is known to work on x86 GNU/Linux platforms with
10483       shared libraries.
10484     * In general, GCC 2.95 is more rigorous about rejecting invalid C++
10485       code or deprecated C++ constructs than G++ 2.7, G++ 2.8, EGCS 1.0,
10486       or EGCS 1.1. As a result it may be necessary to fix C++ code before
10487       it will compile with GCC 2.95.
10488     * G++ is also converting toward the ISO C++ standard; as a result
10489       code which was previously valid (and thus accepted by other
10490       compilers and older versions of g++) may no longer be accepted. The
10491       flag -fpermissive may allow some non-conforming code to compile
10492       with GCC 2.95.
10493     * GCC 2.95 compiled C++ code is not binary compatible with EGCS
10494       1.1.x, EGCS 1.0.x or GCC 2.8.x.
10495     * GCC 2.95 does not have changes from the GCC 2.8 tree that were made
10496       between Sept 30, 1998 and April 30, 1999 (the official end of the
10497       GCC 2.8 project). Future GCC releases will include all the changes
10498       from the defunct GCC 2.8 sources.
10499
10500
10501    For questions related to the use of GCC, please consult these web
10502    pages and the [2]GCC manuals. If that fails, the
10503    [3]gcc-help@gcc.gnu.org mailing list might help. Comments on these
10504    web pages and the development of GCC are welcome on our developer
10505    list at [4]gcc@gcc.gnu.org. All of [5]our lists have public archives.
10506
10507   Copyright (C) [6]Free Software Foundation, Inc. Verbatim copying and
10508   distribution of this entire article is permitted in any medium,
10509   provided this notice is preserved.
10510
10511   These pages are [7]maintained by the GCC team. Last modified
10512   2011-04-25[8].
10513
10514References
10515
10516   1. ftp://gcc.gnu.org/pub/gcc/infrastructure/libg++-2.8.1.3.tar.gz
10517   2. http://gcc.gnu.org/onlinedocs/
10518   3. mailto:gcc-help@gcc.gnu.org
10519   4. mailto:gcc@gcc.gnu.org
10520   5. http://gcc.gnu.org/lists.html
10521   6. http://www.fsf.org/
10522   7. http://gcc.gnu.org/about.html
10523   8. http://validator.w3.org/check/referer
10524======================================================================
10525http://gcc.gnu.org/egcs-1.1/index.html
10526
10527                                    EGCS 1.1
10528
10529   September 3, 1998: We are pleased to announce the release of EGCS 1.1.
10530   December 1, 1998: We are pleased to announce the release of EGCS 1.1.1.
10531   March 15, 1999: We are pleased to announce the release of EGCS 1.1.2.
10532
10533   EGCS is a free software project to further the development of the GNU
10534   compilers using an open development environment.
10535
10536   EGCS 1.1 is a major new release of the EGCS compiler system. It has
10537   been [1]extensively tested and is believed to be stable and suitable
10538   for widespread use.
10539
10540   EGCS 1.1 is based on an June 6, 1998 snapshot of the GCC 2.8
10541   development sources; it contains all of the new features found in GCC
10542   2.8.1 as well as all new development from GCC up to June 6, 1998.
10543
10544   EGCS 1.1 also contains many improvements and features not found in GCC
10545   or in older versions of EGCS:
10546     * Global common subexpression elimination and global constant/copy
10547       propagation (aka [2]gcse)
10548     * Ongoing improvements to the [3]alias analysis support to allow for
10549       better optimizations throughout the compiler.
10550     * Vastly improved [4]C++ compiler and integrated C++ runtime
10551       libraries.
10552     * Fixes for the /tmp symlink race security problems.
10553     * New targets including mips16, arm-thumb and 64 bit PowerPC.
10554     * Improvements to GNU Fortran (g77) compiler and runtime library made
10555       since g77 version 0.5.23.
10556
10557   See the [5]new features page for a more complete list of new features
10558   found in EGCS 1.1 releases.
10559
10560   EGCS 1.1.1 is a minor update to fix several serious problems in EGCS
10561   1.1:
10562     * General improvements and fixes
10563          + Avoid some stack overflows when compiling large functions.
10564          + Avoid incorrect loop invariant code motions.
10565          + Fix some core dumps on Linux kernel code.
10566          + Bring back the imake -Di386 and friends fix from EGCS 1.0.2.
10567          + Fix code generation problem in gcse.
10568          + Various documentation related fixes.
10569     * g++/libstdc++ improvements and fixes
10570          + MT safe EH fix for setjmp/longjmp based exception handling.
10571          + Fix a few bad interactions between optimization and exception
10572            handling.
10573          + Fixes for demangling of template names starting with "__".
10574          + Fix a bug that would fail to run destructors in some cases
10575            with -O2.
10576          + Fix 'new' of classes with virtual bases.
10577          + Fix crash building Qt on the Alpha.
10578          + Fix failure compiling WIFEXITED macro on GNU/Linux.
10579          + Fix some -frepo failures.
10580     * g77 and libf2c improvements and fixes
10581          + Various documentation fixes.
10582          + Avoid compiler crash on RAND intrinsic.
10583          + Fix minor bugs in makefiles exposed by BSD make programs.
10584          + Define _XOPEN_SOURCE for libI77 build to avoid potential
10585            problems on some 64-bit systems.
10586          + Fix problem with implicit endfile on rewind.
10587          + Fix spurious recursive I/O errors.
10588     * platform specific improvements and fixes
10589          + Match all versions of UnixWare7.
10590          + Do not assume x86 SVR4 or UnixWare targets can handle stabs.
10591          + Fix PPC/RS6000 LEGITIMIZE_ADDRESS macro and bug in conversion
10592            from unsigned ints to double precision floats.
10593          + Fix ARM ABI issue with NetBSD.
10594          + Fix a few arm code generation bugs.
10595          + Fixincludes will fix additional broken SCO OpenServer header
10596            files.
10597          + Fix a m68k backend bug which caused invalid offsets in reg+d
10598            addresses.
10599          + Fix problems with 64bit AIX 4.3 support.
10600          + Fix handling of long longs for varargs/stdarg functions on the
10601            ppc.
10602          + Minor fixes to CPP predefines for Windows.
10603          + Fix code generation problems with gpr<->fpr copies for 64bit
10604            ppc.
10605          + Fix a few coldfire code generation bugs.
10606          + Fix some more header file problems on SunOS 4.x.
10607          + Fix assert.h handling for RTEMS.
10608          + Fix Windows handling of TREE_SYMBOL_REFERENCED.
10609          + Fix x86 compiler abort in reg-stack pass.
10610          + Fix cygwin/windows problem with section attributes.
10611          + Fix Alpha code generation problem exposed by SMP Linux
10612            kernels.
10613          + Fix typo in m68k 32->64bit integer conversion.
10614          + Make sure target libraries build with -fPIC for PPC & Alpha
10615            targets.
10616
10617   EGCS 1.1.2 is a minor update to fix several serious problems in EGCS
10618   1.1.1:
10619     * General improvements and fixes
10620          + Fix bug in loop optimizer which caused the SPARC (and
10621            potentially other) ports to segfault.
10622          + Fix infinite recursion in alias analysis and combiner code.
10623          + Fix bug in regclass preferencing.
10624          + Fix incorrect loop reversal which caused incorrect code to be
10625            generated for several targets.
10626          + Fix return value for builtin memcpy.
10627          + Reduce compile time for certain loops which exposed quadratic
10628            behavior in the loop optimizer.
10629          + Fix bug which caused volatile memory to be written multiple
10630            times when only one write was needed/desired.
10631          + Fix compiler abort in caller-save.c
10632          + Fix combiner bug which caused incorrect code generation for
10633            certain division by constant operations.
10634          + Fix incorrect code generation due to a bug in range check
10635            optimizations.
10636          + Fix incorrect code generation due to mis-handling of clobbered
10637            values in CSE.
10638          + Fix compiler abort/segfault due to incorrect register
10639            splitting when unrolling loops.
10640          + Fix code generation involving autoincremented addresses with
10641            ternary operators.
10642          + Work around bug in the scheduler which caused qt to be
10643            mis-compiled on some platforms.
10644          + Fix code generation problems with -fshort-enums.
10645          + Tighten security for temporary files.
10646          + Improve compile time for codes which make heavy use of
10647            overloaded functions.
10648          + Fix multiply defined constructor/destructor symbol problems.
10649          + Avoid setting bogus RPATH environment variable during
10650            bootstrap.
10651          + Avoid GNU-make dependencies in the texinfo subdir.
10652          + Install CPP wrapper script in $(prefix)/bin if --enable-cpp.
10653            --enable-cpp=<dirname> can be used to specify an additional
10654            install directory for the cpp wrapper script.
10655          + Fix CSE bug which caused incorrect label-label refs to appear
10656            on some platforms.
10657          + Avoid linking in EH routines from libgcc if they are not
10658            needed.
10659          + Avoid obscure bug in aliasing code.
10660          + Fix bug in weak symbol handling.
10661     * Platform-specific improvements and fixes
10662          + Fix detection of PPro/PII on Unixware 7.
10663          + Fix compiler segfault when building spec99 and other programs
10664            for SPARC targets.
10665          + Fix code-generation bugs for integer and floating point
10666            conditional move instructions on the PPro/PII.
10667          + Use fixincludes to fix byteorder problems on i?86-*-sysv.
10668          + Fix build failure for the arc port.
10669          + Fix floating point format configuration for i?86-gnu port.
10670          + Fix problems with hppa1.0-hp-hpux10.20 configuration when
10671            threads are enabled.
10672          + Fix coldfire code generation bugs.
10673          + Fix "unrecognized insn" problems for Alpha and PPC ports.
10674          + Fix h8/300 code generation problem with floating point values
10675            in memory.
10676          + Fix unrecognized insn problems for the m68k port.
10677          + Fix namespace-pollution problem for the x86 port.
10678          + Fix problems with old assembler on x86 NeXT systems.
10679          + Fix PIC code-generation problems for the SPARC port.
10680          + Fix minor bug with LONG_CALLS in PowerPC SVR4 support.
10681          + Fix minor ISO namespace violation in Alpha varargs/stdarg
10682            support.
10683          + Fix incorrect "braf" instruction usage for the SH port.
10684          + Fix minor bug in va-sh which prevented its use with -ansi.
10685          + Fix problems recognizing and supporting FreeBSD.
10686          + Handle OpenBSD systems correctly.
10687          + Minor fixincludes fix for Digital UNIX 4.0B.
10688          + Fix problems with ctors/dtors in SCO shared libraries.
10689          + Abort instead of generating incorrect code for PPro/PII
10690            floating point conditional moves.
10691          + Avoid multiply defined symbols on GNU/Linux systems using
10692            libc-5.4.xx.
10693          + Fix abort in alpha compiler.
10694     * Fortran-specific fixes
10695          + Fix the IDate intrinsic (VXT) (in libg2c) so the returned year
10696            is in the documented, non-Y2K-compliant range of 0-99, instead
10697            of being returned as 100 in the year 2000.
10698          + Fix the `Date_and_Time' intrinsic (in libg2c) to return the
10699            milliseconds value properly in Values(8).
10700          + Fix the `LStat' intrinsic (in libg2c) to return device-ID
10701            information properly in SArray(7).
10702
10703   Each release includes installation instructions in both HTML and
10704   plaintext forms (see the INSTALL directory in the toplevel directory of
10705   the distribution). However, we also keep the most up to date
10706   [6]installation instructions and [7]build/test status on our web page.
10707   We will update those pages as new information becomes available.
10708
10709   The EGCS project would like to thank the numerous people that have
10710   contributed new features, test results, bugfixes, etc. This [8]amazing
10711   group of volunteers is what makes EGCS successful.
10712
10713   And finally, we can't in good conscience fail to mention some
10714   [9]caveats to using EGCS 1.1.
10715
10716   Download EGCS from egcs.cygnus.com (USA California).
10717
10718   The EGCS 1.1 release is also available on many mirror sites.
10719   [10]Goto mirror list to find a closer site.
10720
10721
10722    For questions related to the use of GCC, please consult these web
10723    pages and the [11]GCC manuals. If that fails, the
10724    [12]gcc-help@gcc.gnu.org mailing list might help. Comments on these
10725    web pages and the development of GCC are welcome on our developer
10726    list at [13]gcc@gcc.gnu.org. All of [14]our lists have public
10727    archives.
10728
10729   Copyright (C) [15]Free Software Foundation, Inc. Verbatim copying and
10730   distribution of this entire article is permitted in any medium,
10731   provided this notice is preserved.
10732
10733   These pages are [16]maintained by the GCC team. Last modified
10734   2011-10-24[17].
10735
10736References
10737
10738   1. http://gcc.gnu.org/egcs-1.1/egcs-1.1-test.html
10739   2. http://gcc.gnu.org/news/gcse.html
10740   3. http://gcc.gnu.org/news/alias.html
10741   4. http://gcc.gnu.org/egcs-1.1/c++features.html
10742   5. http://gcc.gnu.org/egcs-1.1/features.html
10743   6. http://gcc.gnu.org/install/
10744   7. http://gcc.gnu.org/egcs-1.1/buildstat.html
10745   8. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
10746   9. http://gcc.gnu.org/egcs-1.1/caveats.html
10747  10. http://gcc.gnu.org/mirrors.html
10748  11. http://gcc.gnu.org/onlinedocs/
10749  12. mailto:gcc-help@gcc.gnu.org
10750  13. mailto:gcc@gcc.gnu.org
10751  14. http://gcc.gnu.org/lists.html
10752  15. http://www.fsf.org/
10753  16. http://gcc.gnu.org/about.html
10754  17. http://validator.w3.org/check/referer
10755======================================================================
10756http://gcc.gnu.org/egcs-1.1/features.html
10757
10758                             EGCS 1.1 new features
10759
10760     * Integrated GNU Fortran (g77) compiler and runtime library with
10761       improvements, based on g77 version 0.5.23.
10762     * Vast improvements in the C++ compiler; so many they have [1]page of
10763       their own!
10764     * Compiler implements [2]global common subexpression elimination and
10765       global copy/constant propagation.
10766     * More major improvements in the [3]alias analysis code.
10767     * More major improvements in the exception handling code to improve
10768       performance, lower static overhead and provide the infrastructure
10769       for future improvements.
10770     * The infamous /tmp symlink race security problems have been fixed.
10771     * The regmove optimization pass has been nearly completely rewritten
10772       to improve performance of generated code.
10773     * The compiler now recomputes register usage information before local
10774       register allocation. By providing more accurate information to the
10775       priority based allocator, we get better register allocation.
10776     * The register reloading phase of the compiler optimizes spill code
10777       much better than in previous releases.
10778     * Some bad interactions between the register allocator and
10779       instruction scheduler have been fixed, resulting in much better
10780       code for certain programs. Additionally, we have tuned the
10781       scheduler in various ways to improve performance of generated code
10782       for some architectures.
10783     * The compiler's branch shortening algorithms have been significantly
10784       improved to work better on targets which align jump targets.
10785     * The compiler now supports -Os to prefer optimizing for code space
10786       over optimizing for code speed.
10787     * The compiler will now totally eliminate library calls which compute
10788       constant values. This primarily helps targets with no integer
10789       div/mul support and targets without floating point support.
10790     * The compiler now supports an extensive "--help" option.
10791     * cpplib has been greatly improved and may be suitable for limited
10792       use.
10793     * Memory footprint for the compiler has been significantly reduced
10794       for some pathological cases.
10795     * The time to build EGCS has been improved for certain targets
10796       (particularly the alpha and mips platforms).
10797     * Many infrastructure improvements throughout the compiler, plus the
10798       usual mountain of bugfixes and minor improvements.
10799     * Target dependent improvements:
10800          + SPARC port now includes V8 plus and V9 support as well as
10801            performance tuning for Ultra class machines. The SPARC port
10802            now uses the Haifa scheduler.
10803          + Alpha port has been tuned for the EV6 processor and has an
10804            optimized expansion of memcpy/bzero. The Alpha port now uses
10805            the Haifa scheduler.
10806          + RS6000/PowerPC: support for the Power64 architecture and AIX
10807            4.3. The RS6000/PowerPC port now uses the Haifa scheduler.
10808          + x86: Alignment of static store data and jump targets is per
10809            Intel recommendations now. Various improvements throughout the
10810            x86 port to improve performance on Pentium processors
10811            (including improved epilogue sequences for Pentium chips and
10812            backend improvements which should help register allocation on
10813            all x86 variants. Conditional move support has been fixed and
10814            enabled for PPro processors. The x86 port also better supports
10815            64bit operations now. Unixware 7, a System V Release 5 target,
10816            is now supported and SCO OpenServer targets can support GAS.
10817          + MIPS has improved multiply/multiply-add support and now
10818            includes mips16 ISA support.
10819          + M68k has many micro-optimizations and Coldfire fixes.
10820     * Core compiler is based on the GCC development tree from June 9,
10821       1998, so we have all of the [4]features found in GCC 2.8.
10822
10823
10824    For questions related to the use of GCC, please consult these web
10825    pages and the [5]GCC manuals. If that fails, the
10826    [6]gcc-help@gcc.gnu.org mailing list might help. Comments on these
10827    web pages and the development of GCC are welcome on our developer
10828    list at [7]gcc@gcc.gnu.org. All of [8]our lists have public archives.
10829
10830   Copyright (C) [9]Free Software Foundation, Inc. Verbatim copying and
10831   distribution of this entire article is permitted in any medium,
10832   provided this notice is preserved.
10833
10834   These pages are [10]maintained by the GCC team. Last modified
10835   2011-04-25[11].
10836
10837References
10838
10839   1. http://gcc.gnu.org/egcs-1.1/c++features.html
10840   2. http://gcc.gnu.org/news/gcse.html
10841   3. http://gcc.gnu.org/news/alias.html
10842   4. http://gcc.gnu.org/egcs-1.0/features-2.8.html
10843   5. http://gcc.gnu.org/onlinedocs/
10844   6. mailto:gcc-help@gcc.gnu.org
10845   7. mailto:gcc@gcc.gnu.org
10846   8. http://gcc.gnu.org/lists.html
10847   9. http://www.fsf.org/
10848  10. http://gcc.gnu.org/about.html
10849  11. http://validator.w3.org/check/referer
10850======================================================================
10851http://gcc.gnu.org/egcs-1.1/caveats.html
10852
10853                                EGCS 1.1 Caveats
10854
10855     * EGCS has an integrated libstdc++, but does not have an integrated
10856       libg++. Furthermore old libg++ releases will not work with EGCS; HJ
10857       Lu has made a libg++-2.8.1.2 snapshot available which may work with
10858       EGCS.
10859       Note most C++ programs only need libstdc++.
10860     * Exception handling may not work with shared libraries, particularly
10861       on alphas, hppas, rs6000/powerpc and mips based platforms.
10862       Exception handling is known to work on x86-linux platforms with
10863       shared libraries.
10864     * Some versions of the Linux kernel have bugs which prevent them from
10865       being compiled or from running when compiled by EGCS. See the FAQ
10866       (as shipped with EGCS 1.1) for additional information.
10867     * In general, EGCS is more rigorous about rejecting invalid C++ code
10868       or deprecated C++ constructs than g++-2.7, g++-2.8 or EGCS 1.0. As
10869       a result it may be necessary to fix C++ code before it will compile
10870       with EGCS.
10871     * G++ is also converting toward the ISO C++ standard; as a result
10872       code which was previously valid (and thus accepted by other
10873       compilers and older versions of g++) may no longer be accepted.
10874     * EGCS 1.1 compiled C++ code is not binary compatible with EGCS 1.0.x
10875       or GCC 2.8.x due to changes necessary to support thread safe
10876       exception handling.
10877
10878
10879    For questions related to the use of GCC, please consult these web
10880    pages and the [1]GCC manuals. If that fails, the
10881    [2]gcc-help@gcc.gnu.org mailing list might help. Comments on these
10882    web pages and the development of GCC are welcome on our developer
10883    list at [3]gcc@gcc.gnu.org. All of [4]our lists have public archives.
10884
10885   Copyright (C) [5]Free Software Foundation, Inc. Verbatim copying and
10886   distribution of this entire article is permitted in any medium,
10887   provided this notice is preserved.
10888
10889   These pages are [6]maintained by the GCC team. Last modified
10890   2011-04-25[7].
10891
10892References
10893
10894   1. http://gcc.gnu.org/onlinedocs/
10895   2. mailto:gcc-help@gcc.gnu.org
10896   3. mailto:gcc@gcc.gnu.org
10897   4. http://gcc.gnu.org/lists.html
10898   5. http://www.fsf.org/
10899   6. http://gcc.gnu.org/about.html
10900   7. http://validator.w3.org/check/referer
10901======================================================================
10902http://gcc.gnu.org/egcs-1.0/index.html
10903
10904                                    EGCS 1.0
10905
10906   December 3, 1997: We are pleased to announce the release of EGCS 1.0.
10907   January 6, 1998: We are pleased to announce the release of EGCS 1.0.1.
10908   March 16, 1998: We are pleased to announce the release of EGCS 1.0.2.
10909   May 15, 1998 We are pleased to announce the release of EGCS 1.0.3.
10910
10911   EGCS is a collaborative effort involving several groups of hackers
10912   using an open development model to accelerate development and testing
10913   of GNU compilers and runtime libraries.
10914
10915   An important goal of EGCS is to allow wide scale testing of
10916   experimental features and optimizations; therefore, EGCS contains some
10917   features and optimizations which are still under development. However,
10918   EGCS has been carefully tested and should be comparable in quality to
10919   most GCC releases.
10920
10921   EGCS 1.0 is based on an August 2, 1997 snapshot of the GCC 2.8
10922   development sources; it contains nearly all of the new features found
10923   in GCC 2.8.
10924
10925   EGCS 1.0 also contains many improvements and features not found in GCC
10926   2.7 and even the GCC 2.8 series (which was released after the original
10927   EGCS 1.0 release).
10928     * Integrated C++ runtime libraries, including support for most major
10929       GNU/Linux systems!
10930     * The integrated libstdc++ library includes a verbatim copy of SGI's
10931       STL release.
10932     * Integrated GNU Fortran compiler.
10933     * New instruction scheduler.
10934     * New alias analysis code.
10935
10936   See the [1]new features page for a more complete list of new features.
10937
10938   EGCS 1.0.1 is a minor update to the EGCS 1.0 compiler to fix a few
10939   critical bugs and add support for Red Hat 5.0 Linux. Changes since the
10940   EGCS 1.0 release:
10941     * Add support for Red Hat 5.0 Linux and better support for Linux
10942       systems using glibc2.
10943       Many programs failed to link when compiled with EGCS 1.0 on Red Hat
10944       5.0 or on systems with newer versions of glibc2. EGCS 1.0.1 should
10945       fix these problems.
10946     * Compatibility with both EGCS 1.0 and GCC 2.8 libgcc exception
10947       handling interfaces.
10948       To avoid future compatibility problems, we strongly urge anyone who
10949       is planning on distributing shared libraries that contain C++ code
10950       to upgrade to EGCS 1.0.1 first.
10951       Soon after EGCS 1.0 was released, the GCC developers made some
10952       incompatible changes in libgcc's exception handling interfaces.
10953       These changes were needed to solve problems on some platforms. This
10954       means that GCC 2.8.0, when released, will not be seamlessly
10955       compatible with shared libraries built by EGCS 1.0. The reason is
10956       that the libgcc.a in GCC 2.8.0 will not contain a function needed
10957       by the old interface.
10958       The result of this is that there may be compatibility problems with
10959       shared libraries built by EGCS 1.0 when used with GCC 2.8.0.
10960       With EGCS 1.0.1, generated code uses the new (GCC 2.8.0) interface,
10961       and libgcc.a has the support routines for both the old and the new
10962       interfaces (so EGCS 1.0.1 and EGCS 1.0 code can be freely mixed,
10963       and EGCS 1.0.1 and GCC 2.8.0 code can be freely mixed).
10964       The maintainers of GCC 2.x have decided against including seamless
10965       support for the old interface in 2.8.0, since it was never
10966       "official", so to avoid future compatibility problems we recommend
10967       against distributing any shared libraries built by EGCS 1.0 that
10968       contain C++ code (upgrade to 1.0.1 and use that).
10969     * Various bugfixes in the x86, hppa, mips, and rs6000/ppc backends.
10970       The x86 changes fix code generation errors exposed when building
10971       glibc2 and the usual GNU/Linux dynamic linker (ld.so).
10972       The hppa change fixes a compiler abort when configured for use with
10973       RTEMS.
10974       The MIPS changes fix problems with the definition of LONG_MAX on
10975       newer systems, allow for command line selection of the target ABI,
10976       and fix one code generation problem.
10977       The rs6000/ppc change fixes some problems with passing structures
10978       to varargs/stdarg functions.
10979     * A few machine independent bugfixes, mostly to fix code generation
10980       errors when building Linux kernels or glibc.
10981     * Fix a few critical exception handling and template bugs in the C++
10982       compiler.
10983     * Fix Fortran namelist bug on alphas.
10984     * Fix build problems on x86-solaris systems.
10985
10986   EGCS 1.0.2 is a minor update to the EGCS 1.0.1 compiler to fix several
10987   serious problems in EGCS 1.0.1.
10988     * General improvements and fixes
10989          + Memory consumption significantly reduced, especially for
10990            templates and inline functions.
10991          + Fix various problems with glibc2.1.
10992          + Fix loop optimization bug exposed by rs6000/ppc port.
10993          + Fix to avoid potential code generation problems in jump.c.
10994          + Fix some undefined symbol problems in dwarf1 debug support.
10995     * g++/libstdc++ improvements and fixes
10996          + libstdc++ in the EGCS release has been updated and should be
10997            link compatible with libstdc++-2.8.
10998          + Various fixes in libio/libstdc++ to work better on GNU/Linux
10999            systems.
11000          + Fix problems with duplicate symbols on systems that do not
11001            support weak symbols.
11002          + Memory corruption bug and undefined symbols in bastring have
11003            been fixed.
11004          + Various exception handling fixes.
11005          + Fix compiler abort for very long thunk names.
11006     * g77 improvements and fixes
11007          + Fix compiler crash for omitted bound in Fortran CASE
11008            statement.
11009          + Add missing entries to g77 lang-options.
11010          + Fix problem with -fpedantic in the g77 compiler.
11011          + Fix "backspace" problem with g77 on alphas.
11012          + Fix x86 backend problem with Fortran literals and -fpic.
11013          + Fix some of the problems with negative subscripts for g77 on
11014            alphas.
11015          + Fixes for Fortran builds on cygwin32/mingw32.
11016     * platform specific improvements and fixes
11017          + Fix long double problems on x86 (exposed by glibc).
11018          + x86 ports define i386 again to keep imake happy.
11019          + Fix exception handling support on NetBSD ports.
11020          + Several changes to collect2 to fix many problems with AIX.
11021          + Define __ELF__ for GNU/Linux on rs6000.
11022          + Fix -mcall-linux problem on GNU/Linux on rs6000.
11023          + Fix stdarg/vararg problem for GNU/Linux on rs6000.
11024          + Allow autoconf to select a proper install problem on AIX 3.1.
11025          + m68k port support includes -mcpu32 option as well as cpu32
11026            multilibs.
11027          + Fix stdarg bug for irix6.
11028          + Allow EGCS to build on irix5 without the gnu assembler.
11029          + Fix problem with static linking on sco5.
11030          + Fix bootstrap on sco5 with native compiler.
11031          + Fix for abort building newlib on H8 target.
11032          + Fix fixincludes handling of math.h on SunOS.
11033          + Minor fix for Motorola 3300 m68k systems.
11034
11035   EGCS 1.0.3 is a minor update to the EGCS 1.0.2 compiler to fix a few
11036   problems reported by Red Hat for builds of Red Hat 5.1.
11037     * Generic bugfixes:
11038          + Fix a typo in the libio library which resulted in incorrect
11039            behavior of istream::get.
11040          + Fix the Fortran negative array index problem.
11041          + Fix a major problem with the ObjC runtime thread support
11042            exposed by glibc2.
11043          + Reduce memory consumption of the Haifa scheduler.
11044     * Target specific bugfixes:
11045          + Fix one x86 floating point code generation bug exposed by
11046            glibc2 builds.
11047          + Fix one x86 internal compiler error exposed by glibc2 builds.
11048          + Fix profiling bugs on the Alpha.
11049          + Fix ImageMagick & emacs 20.2 build problems on the Alpha.
11050          + Fix rs6000/ppc bug when converting values from integer types
11051            to floating point types.
11052
11053   The EGCS 1.0 releases include installation instructions in both HTML
11054   and plaintext forms (see the INSTALL directory in the toplevel
11055   directory of the distribution). However, we also keep the most up to
11056   date [2]installation instructions and [3]build/test status on our web
11057   page. We will update those pages as new information becomes available.
11058
11059   And, we can't in good conscience fail to mention some [4]caveats to
11060   using EGCS.
11061
11062   Update: Big thanks to Stanford for providing a high speed link for
11063   downloading EGCS (go.cygnus.com)!
11064
11065   Download EGCS from ftp.cygnus.com (USA California) or go.cygnus.com
11066   (USA California -- High speed link provided by Stanford).
11067
11068   The EGCS 1.0 release is also available many mirror sites.
11069   [5]Goto mirror list to find a closer site
11070
11071   We'd like to thank the numerous people that have contributed new
11072   features, test results, bugfixes, etc. Unfortunately, they're far too
11073   numerous to mention by name.
11074
11075
11076    For questions related to the use of GCC, please consult these web
11077    pages and the [6]GCC manuals. If that fails, the
11078    [7]gcc-help@gcc.gnu.org mailing list might help. Comments on these
11079    web pages and the development of GCC are welcome on our developer
11080    list at [8]gcc@gcc.gnu.org. All of [9]our lists have public archives.
11081
11082   Copyright (C) [10]Free Software Foundation, Inc. Verbatim copying and
11083   distribution of this entire article is permitted in any medium,
11084   provided this notice is preserved.
11085
11086   These pages are [11]maintained by the GCC team. Last modified
11087   2011-10-24[12].
11088
11089References
11090
11091   1. http://gcc.gnu.org/egcs-1.0/features.html
11092   2. http://gcc.gnu.org/install/
11093   3. http://gcc.gnu.org/egcs-1.0/buildstat.html
11094   4. http://gcc.gnu.org/egcs-1.0/caveats.html
11095   5. http://gcc.gnu.org/mirrors.html
11096   6. http://gcc.gnu.org/onlinedocs/
11097   7. mailto:gcc-help@gcc.gnu.org
11098   8. mailto:gcc@gcc.gnu.org
11099   9. http://gcc.gnu.org/lists.html
11100  10. http://www.fsf.org/
11101  11. http://gcc.gnu.org/about.html
11102  12. http://validator.w3.org/check/referer
11103======================================================================
11104http://gcc.gnu.org/egcs-1.0/features.html
11105
11106                               EGCS 1.0 features
11107
11108     * Core compiler is based on the gcc2 development tree from Aug 2,
11109       1997, so we have most of the [1]features found in GCC 2.8.
11110     * Integrated GNU Fortran compiler based on g77-0.5.22-19970929.
11111     * Vast improvements in the C++ compiler; so many they have [2]page of
11112       their own!
11113     * Integrated C++ runtime libraries, including support for most major
11114       GNU/Linux systems!
11115     * New instruction scheduler from IBM Haifa which includes support for
11116       function wide instruction scheduling as well as superscalar
11117       scheduling.
11118     * Significantly improved alias analysis code.
11119     * Improved register allocation for two address machines.
11120     * Significant code generation improvements for Fortran code on
11121       Alphas.
11122     * Various optimizations from the g77 project as well as improved loop
11123       optimizations.
11124     * Dwarf2 debug format support for some targets.
11125     * egcs libstdc++ includes the SGI STL implementation without changes.
11126     * As a result of these and other changes, egcs libstc++ is not binary
11127       compatible with previous releases of libstdc++.
11128     * Various new ports -- UltraSPARC, Irix6.2 & Irix6.3 support, The SCO
11129       Openserver 5 family (5.0.{0,2,4} and Internet FastStart 1.0 and
11130       1.1), Support for RTEMS on several embedded targets, Support for
11131       arm-linux, Mitsubishi M32R, Hitachi H8/S, Matsushita MN102 and
11132       MN103, NEC V850, Sparclet, Solaris & GNU/Linux on PowerPCs, etc.
11133     * Integrated testsuites for gcc, g++, g77, libstdc++ and libio.
11134     * RS6000/PowerPC ports generate code which can run on all
11135       RS6000/PowerPC variants by default.
11136     * -mcpu= and -march= switches for the x86 port to allow better
11137       control over how the x86 port generates code.
11138     * Includes the template repository patch (aka repo patch); note the
11139       new template code makes repo obsolete for ELF systems using gnu-ld
11140       such as GNU/Linux.
11141     * Plus the usual assortment of bugfixes and improvements.
11142
11143
11144    For questions related to the use of GCC, please consult these web
11145    pages and the [3]GCC manuals. If that fails, the
11146    [4]gcc-help@gcc.gnu.org mailing list might help. Comments on these
11147    web pages and the development of GCC are welcome on our developer
11148    list at [5]gcc@gcc.gnu.org. All of [6]our lists have public archives.
11149
11150   Copyright (C) [7]Free Software Foundation, Inc. Verbatim copying and
11151   distribution of this entire article is permitted in any medium,
11152   provided this notice is preserved.
11153
11154   These pages are [8]maintained by the GCC team. Last modified
11155   2011-10-24[9].
11156
11157References
11158
11159   1. http://gcc.gnu.org/egcs-1.0/features-2.8.html
11160   2. http://gcc.gnu.org/egcs-1.0/c++features.html
11161   3. http://gcc.gnu.org/onlinedocs/
11162   4. mailto:gcc-help@gcc.gnu.org
11163   5. mailto:gcc@gcc.gnu.org
11164   6. http://gcc.gnu.org/lists.html
11165   7. http://www.fsf.org/
11166   8. http://gcc.gnu.org/about.html
11167   9. http://validator.w3.org/check/referer
11168======================================================================
11169http://gcc.gnu.org/egcs-1.0/caveats.html
11170
11171                                EGCS 1.0 Caveats
11172
11173     * EGCS has an integrated libstdc++, but does not have an integrated
11174       libg++. Furthermore old libg++ releases will not work with egc; HJ
11175       Lu has made a libg++-2.8.1.2 available which may work with EGCS.
11176       Note most C++ programs only need libstdc++.
11177     * Note that using -pedantic or -Wreturn-type can cause an explosion
11178       in the amount of memory needed for template-heavy C++ code, such as
11179       code that uses STL. Also note that -Wall includes -Wreturn-type, so
11180       if you use -Wall you will need to specify -Wno-return-type to turn
11181       it off.
11182     * Exception handling may not work with shared libraries, particularly
11183       on alphas, hppas, and mips based platforms. Exception handling is
11184       known to work on x86-linux platforms with shared libraries.
11185     * Some versions of the Linux kernel have bugs which prevent them from
11186       being compiled or from running when compiled by EGCS. See the FAQ
11187       (as shipped with EGCS 1.0) for additional information.
11188     * In general, EGCS is more rigorous about rejecting invalid C++ code
11189       or deprecated C++ constructs than G++ 2.7. As a result it may be
11190       necessary to fix C++ code before it will compile with EGCS.
11191     * G++ is also aggressively tracking the C++ standard; as a result
11192       code which was previously valid (and thus accepted by other
11193       compilers and older versions of G++) may no longer be accepted.
11194     * EGCS 1.0 may not work with Red Hat Linux 5.0 on all targets. EGCS
11195       1.0.x and later releases should work with Red Hat Linux 5.0.
11196
11197
11198    For questions related to the use of GCC, please consult these web
11199    pages and the [1]GCC manuals. If that fails, the
11200    [2]gcc-help@gcc.gnu.org mailing list might help. Comments on these
11201    web pages and the development of GCC are welcome on our developer
11202    list at [3]gcc@gcc.gnu.org. All of [4]our lists have public archives.
11203
11204   Copyright (C) [5]Free Software Foundation, Inc. Verbatim copying and
11205   distribution of this entire article is permitted in any medium,
11206   provided this notice is preserved.
11207
11208   These pages are [6]maintained by the GCC team. Last modified
11209   2011-04-25[7].
11210
11211References
11212
11213   1. http://gcc.gnu.org/onlinedocs/
11214   2. mailto:gcc-help@gcc.gnu.org
11215   3. mailto:gcc@gcc.gnu.org
11216   4. http://gcc.gnu.org/lists.html
11217   5. http://www.fsf.org/
11218   6. http://gcc.gnu.org/about.html
11219   7. http://validator.w3.org/check/referer
11220======================================================================
11221