| blob | 43c93066bb29a30d7c37b44142fea43a8394a6cf |
1 This file contains information about GCC releases which has been generated
2 automatically from the online release notes. It covers releases of GCC
3 (and the former EGCS project) since EGCS 1.0, on the line of development
4 that led to GCC 3. For information on GCC 2.8.1 and older releases of GCC 2,
5 see ONEWS.
7 ======================================================================
8 http://gcc.gnu.org/gcc-4.1/index.html
10 GCC 4.1 Release Series
12 February 13, 2007
14 The [1]GNU project and the GCC developers are pleased to announce the
15 release of GCC 4.1.2.
17 This release is a bug-fix release, containing fixes for regressions in
18 GCC 4.1.1 relative to previous releases of GCC.
20 Release History
22 GCC 4.1.2
23 February 13, 2007 ([2]changes)
25 GCC 4.1.1
26 May 24, 2006 ([3]changes)
28 GCC 4.1.0
29 February 28, 2006 ([4]changes)
31 References and Acknowledgements
33 GCC used to stand for the GNU C Compiler, but since the compiler
34 supports several other languages aside from C, it now stands for the
35 GNU Compiler Collection.
37 A list of [5]successful builds is updated as new information becomes
38 available.
40 The GCC developers would like to thank the numerous people that have
41 contributed new features, improvements, bug fixes, and other changes
42 as well as test results to GCC. This [6]amazing group of volunteers is
43 what makes GCC successful.
45 For additional information about GCC please refer to the [7]GCC
46 project web site or contact the [8]GCC development mailing list.
48 To obtain GCC please use [9]our mirror sites, one of the [10]GNU
49 mirror sites, or [11]our SVN server.
51 Please send FSF & GNU inquiries & questions to [12]gnu@gnu.org. There
52 are also [13]other ways to contact the FSF.
54 These pages are maintained by [14]the GCC team.
57 For questions related to the use of GCC, please consult these web
58 pages and the [15]GCC manuals. If that fails, the
59 [16]gcc-help@gcc.gnu.org mailing list might help.
60 Please send comments on these web pages and the development of GCC
61 to our developer mailing list at [17]gcc@gnu.org or
62 [18]gcc@gcc.gnu.org. All of our lists have [19]public archives.
64 Copyright (C) Free Software Foundation, Inc., 51 Franklin St, Fifth
65 Floor, Boston, MA 02110, USA.
67 Verbatim copying and distribution of this entire article is permitted
68 in any medium, provided this notice is preserved.
69 Last modified 2007-02-14 [20]Valid XHTML 1.0
71 References
73 1. http://www.gnu.org/
74 2. http://gcc.gnu.org/gcc-4.1/changes.html#4.1.2
75 3. http://gcc.gnu.org/gcc-4.1/changes.html
76 4. http://gcc.gnu.org/gcc-4.1/changes.html
77 5. http://gcc.gnu.org/gcc-4.1/buildstat.html
78 6. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
79 7. http://gcc.gnu.org/index.html
80 8. mailto:gcc@gcc.gnu.org
81 9. http://gcc.gnu.org/mirrors.html
82 10. http://www.gnu.org/order/ftp.html
83 11. http://gcc.gnu.org/svn.html
84 12. mailto:gnu@gnu.org
85 13. http://www.gnu.org/home.html#ContactInfo
86 14. http://gcc.gnu.org/about.html
87 15. http://gcc.gnu.org/onlinedocs/
88 16. mailto:gcc-help@gcc.gnu.org
89 17. mailto:gcc@gnu.org
90 18. mailto:gcc@gcc.gnu.org
91 19. http://gcc.gnu.org/lists.html
92 20. http://validator.w3.org/check/referer
93 ======================================================================
94 http://gcc.gnu.org/gcc-4.1/changes.html
96 GCC 4.1 Release Series
97 Changes, New Features, and Fixes
99 The latest release in the 4.1 release series is [1]GCC 4.1.2.
101 Caveats
103 General Optimizer Improvements
105 * GCC now has infrastructure for inter-procedural optimizations and
106 the following inter-procedural optimizations are implemented:
107 + Profile guided inlining. When doing profile feedback guided
108 optimization, GCC can now use the profile to make better
109 informed decisions on whether inlining of a function is
110 profitable or not. This means that GCC will no longer inline
111 functions at call sites that are not executed very often, and
112 that functions at hot call sites are more likely to be
113 inlined.
114 A new parameter min-inline-recursive-probability is also now
115 available to throttle recursive inlining of functions with
116 small average recursive depths.
117 + Discovery of pure and const functions, a form of side-effects
118 analysis. While older GCC releases could also discover such
119 special functions, the new IPA-based pass runs earlier so
120 that the results are available to more optimizers. The pass
121 is also simply more powerful than the old one.
122 + Analysis of references to static variables and type escape
123 analysis, also forms of side-effects analysis. The results of
124 these passes allow the compiler to be less conservative about
125 call-clobbered variables and references. This results in more
126 redundant loads being eliminated and in making static
127 variables candidates for register promotion.
128 + Improvement of RTL-based alias analysis. The results of type
129 escape analysis are fed to the RTL type-based alias analyzer,
130 allowing it to disambiguate more memory references.
131 + Interprocedural constant propagation and function versioning.
132 This pass looks for functions that are always called with the
133 same constant value for one or more of the function
134 arguments, and propagates those constants into those
135 functions.
136 + GCC will now eliminate static variables whose usage was
137 optimized out.
138 + -fwhole-program --combine can now be used to make all
139 functions in program static allowing whole program
140 optimization. As an exception, the main function and all
141 functions marked with the new externally_visible attribute
142 are kept global so that programs can link with runtime
143 libraries.
144 * GCC can now do a form of partial dead code elimination (PDCE) that
145 allows code motion of expressions to the paths where the result of
146 the expression is actually needed. This is not always a win, so
147 the pass has been limited to only consider profitable cases. Here
148 is an example:
149 int foo (int *, int *);
150 int
151 bar (int d)
152 {
153 int a, b, c;
154 b = d + 1;
155 c = d + 2;
156 a = b + c;
157 if (d)
158 {
159 foo (&b, &c);
160 a = b + c;
161 }
162 printf ("%d\n", a);
163 }
164 The a = b + c can be sunk to right before the printf.
165 Normal code sinking will not do this, it will sink the first one
166 above into the else-branch of the conditional jump, which still
167 gives you two copies of the code.
168 * GCC now has a value range propagation pass. This allows the
169 compiler to eliminate bounds checks and branches. The results of
170 the pass can also be used to accurately compute branch
171 probabilities.
172 * The pass to convert PHI nodes to straight-line code (a form of
173 if-conversion for GIMPLE) has been improved significantly. The two
174 most significant improvements are an improved algorithm to
175 determine the order in which the PHI nodes are considered, and an
176 improvement that allow the pass to consider if-conversions of
177 basic blocks with more than two predecessors.
178 * Alias analysis improvements. GCC can now differentiate between
179 different fields of structures in Tree-SSA's virtual operands
180 form. This lets stores/loads from non-overlapping structure fields
181 not conflict. A new algorithm to compute points-to sets was
182 contributed that can allows GCC to see now that p->a and p->b,
183 where p is a pointer to a structure, can never point to the same
184 field.
185 * Various enhancements to auto-vectorization:
186 + Incrementally preserve SSA form when vectorizing.
187 + Incrementally preserve loop-closed form when vectorizing.
188 + Improvements to peeling for alignment: generate better code
189 when the misalignment of an access is known at compile time,
190 or when different accesses are known to have the same
191 misalignment, even if the misalignment amount itself is
192 unknown.
193 + Consider dependence distance in the vectorizer.
194 + Externalize generic parts of data reference analysis to make
195 this analysis available to other passes.
196 + Vectorization of conditional code.
197 + Reduction support.
198 * GCC can now partition functions in sections of hot and cold code.
199 This can significantly improve performance due to better
200 instruction cache locality. This feature works best together with
201 profile feedback driven optimization.
202 * A new pass to avoid saving of unneeded arguments to the stack in
203 vararg functions if the compiler can prove that they will not be
204 needed.
205 * Transition of basic block profiling to tree level implementation
206 has been completed. The new implementation should be considerably
207 more reliable (hopefully avoiding profile mismatch errors when
208 using -fprofile-use or -fbranch-probabilities) and can be used to
209 drive higher level optimizations, such as inlining.
210 The -ftree-based-profiling command line option was removed and
211 -fprofile-use now implies disabling old RTL level loop optimizer
212 (-fno-loop-optimize). Speculative prefetching optimization
213 (originally enabled by -fspeculative-prefetching) was removed.
215 New Languages and Language specific improvements
217 C and Objective-C
219 * The old Bison-based C and Objective-C parser has been replaced by
220 a new, faster hand-written recursive-descent parser.
222 Ada
224 * The build infrastructure for the Ada runtime library and tools has
225 been changed to be better integrated with the rest of the build
226 infrastructure of GCC. This should make doing cross builds of Ada
227 a bit easier.
229 C++
231 * ARM-style name-injection of friend declarations is no longer the
232 default. For example:
233 struct S {
234 friend void f();
235 };
237 void g() { f(); }
238 will not be accepted; instead a declaration of f will need to be
239 present outside of the scope of S. The new -ffriend-injection
240 option will enable the old behavior.
241 * The (undocumented) extension which permitted templates with
242 default arguments to be bound to template template parameters with
243 fewer parameters has been deprecated, and will be removed in the
244 next major release of G++. For example:
245 template <template <typename> class C>
246 void f(C<double>) {}
248 template <typename T, typename U = int>
249 struct S {};
251 template void f(S<double>);
253 makes use of the deprecated extension. The reason this code is not
254 valid ISO C++ is that S is a template with two parameters;
255 therefore, it cannot be bound to C which has only one parameter.
257 Runtime Library (libstdc++)
259 * Optimization work:
260 + A new implementation of std::search_n is provided, better
261 performing in case of random access iterators.
262 + Added further efficient specializations of istream functions,
263 i.e., character array and string extractors.
264 + Other smaller improvements throughout.
265 * Policy-based associative containers, designed for
266 high-performance, flexibility and semantic safety are delivered in
267 ext/pb_assoc.
268 * A versatile string class, __gnu_cxx::__versa_string, providing
269 facilities conforming to the standard requirements for
270 basic_string, is delivered in <ext/vstring.h>. In particular:
271 + Two base classes are provided: the default one avoids
272 reference counting and is optimized for short strings; the
273 alternate one, still uses it while improving in a few low
274 level areas (e.g., alignment). See vstring_fwd.h for some
275 useful typedefs.
276 + Various algorithms have been rewritten (e.g., replace), the
277 code streamlined and simple optimizations added.
278 + Option 3 of DR 431 is implemented for both available bases,
279 thus improving the support for stateful allocators.
280 * As usual, many bugs have been fixed (e.g., libstdc++/13583,
281 libstdc++/23953) and LWG resolutions put into effect for the first
282 time (e.g., DR 280, DR 464, N1780 recommendations for DR 233, TR1
283 Issue 6.19). The implementation status of TR1 is now tracked in
284 the docs in tr1.html.
286 Objective-C++
288 * A new language front end for Objective-C++ has been added. This
289 language allows users to mix the object oriented features of
290 Objective-C with those of C++.
292 Java (GCJ)
294 * Core library (libgcj) updates based on GNU Classpath 0.15 - 0.19
295 features (plus some 0.20 bug-fixes)
296 + Networking
297 o The java.net.HttpURLConnection implementation no longer
298 buffers the entire response body in memory. This means
299 that response bodies larger than available memory can
300 now be handled.
301 + (N)IO
302 o NIO FileChannel.map implementation, fast bulk put
303 implementation for DirectByteBuffer (speeds up this
304 method 10x).
305 o FileChannel.lock() and FileChannel.force() implemented.
306 + XML
307 o gnu.xml fix for nodes created outside a namespace
308 context.
309 o Add support for output indenting and
310 cdata-section-elements output instruction in
311 xml.transform.
312 o xml.xpath corrections for cases where
313 elements/attributes might have been created in
314 non-namespace-aware mode. Corrections to handling of XSL
315 variables and minor conformance updates.
316 + AWT
317 o GNU JAWT implementation, the AWT Native Interface, which
318 allows direct access to native screen resources from
319 within a Canvas's paint method. GNU Classpath Examples
320 comes with a Demo, see
321 libjava/classpath/examples/README.
322 o awt.datatransfer updated to 1.5 with support for
323 FlavorEvents. The gtk+ awt peers now allow copy/paste of
324 text, images, URIs/files and serialized objects with
325 other applications and tracking clipboard change events
326 with gtk+ 2.6 (for gtk+ 2.4 only text and serialized
327 objects are supported). A GNU Classpath Examples
328 datatransfer Demo was added to show the new
329 functionality.
330 o Split gtk+ awt peers event handling in two threads and
331 improve gdk lock handling (solves several awt lock ups).
332 o Speed up awt Image loading.
333 o Better gtk+ scrollbar peer implementation when using
334 gtk+ >= 2.6.
335 o Handle image loading errors correctly for gdkpixbuf and
336 MediaTracker.
337 o Better handle GDK lock. Properly prefix gtkpeer native
338 functions (cp_gtk).
339 o GdkGraphics2D has been updated to use Cairo 0.5.x or
340 higher.
341 o BufferedImage and GtkImage rewrites. All image drawing
342 operations should now work correctly (flipping requires
343 gtk+ >= 2.6)
344 o Future Graphics2D, image and text work is documented at:
345 [2]http://developer.classpath.org/mediation/ClasspathGra
346 phicsImagesText
347 o When gtk+ 2.6 or higher is installed the default log
348 handler will produce stack traces whenever a WARNING,
349 CRITICAL or ERROR message is produced.
350 + Free Swing
351 o The RepaintManager has been reworked for more efficient
352 painting, especially for large GUIs.
353 o The layout manager OverlayLayout has been implemented,
354 the BoxLayout has been rewritten to make use of the
355 SizeRequirements utility class and caching for more
356 efficient layout.
357 o Improved accessibility support.
358 o Significant progress has been made in the implementation
359 of the javax.swing.plaf.metal package, with most UI
360 delegates in a working state now. Please test this with
361 your own applications and provide feedback that will
362 help us to improve this package.
363 o The GUI demo (gnu.classpath.examples.swing.Demo) has
364 been extended to highlight various features in our Free
365 Swing implementation. And it includes a look and feel
366 switcher for Metal (default), Ocean and GNU themes.
367 o The javax.swing.plaf.multi package is now implemented.
368 o Editing and several key actions for JTree and JTable
369 were implemented.
370 o Lots of icons and look and feel improvements for Free
371 Swing basic and metal themes were added. Try running the
372 GNU Classpath Swing Demo in examples
373 (gnu.classpath.examples.swing.Demo) with:
374 -Dswing.defaultlaf=javax.swing.plaf.basic.BasicLookAndFe
375 el or
376 -Dswing.defaultlaf=javax.swing.plaf.metal.MetalLookAndFe
377 el
378 o Start of styled text capabilites for java.swing.text.
379 o DefaultMutableTreeNode pre-order, post-order,
380 depth-first and breadth-first traversal enumerations
381 implemented.
382 o JInternalFrame colors and titlebar draw properly.
383 o JTree is working up to par (icons, selection and
384 keyboard traversal).
385 o JMenus were made more compatible in visual and
386 programmatic behavior.
387 o JTable changeSelection and multiple selections
388 implemented.
389 o JButton and JToggleButton change states work properly
390 now.
391 o JFileChooser fixes.
392 o revalidate() and repaint() fixes which make Free Swing
393 much more responsive.
394 o MetalIconFactory implemented.
395 o Free Swing Top-Level Compatibility. JFrame, JDialog,
396 JApplet, JInternalFrame, and JWindow are now 1.5
397 compatible in the sense that you can call add() and
398 setLayout() directly on them, which will have the same
399 effect as calling getContentPane().add() and
400 getContentPane().setLayout().
401 o The JTree interface has been completed. JTrees now
402 recognizes mouse clicks and selections work.
403 o BoxLayout works properly now.
404 o Fixed GrayFilter to actually work.
405 o Metal SplitPane implemented.
406 o Lots of Free Swing text and editor stuff work now.
407 + Free RMI and Corba
408 o Andrew Watson, Vice President and Technical Director of
409 the Object Management Group, has officially assigned us
410 20 bit Vendor Minor Code Id: 0x47430 ("GC") that will
411 mark remote classpath-specific system exceptions.
412 Obtaining the VMCID means that GNU Classpath now is a
413 recogniseable type of node in a highly interoperable
414 CORBA world.
415 o GNU Classpath now includes the first working draft to
416 support the RMI over IIOP protocol. The current
417 implementation is capable of remote invocations,
418 transferring various Serializables and Externalizables
419 via RMI-IIOP protocol. It can flatten graphs and, at
420 least for the simple cases, is interoperable with 1.5
421 JDKs.
422 o org.omg.PortableInterceptor and related functionality in
423 other packages is now implemented:
424 # The sever and client interceptors work as required
425 since 1.4.
426 # The IOR interceptor works as needed for 1.5.
427 o The org.omg.DynamicAny package is completed and passes
428 the prepared tests.
429 o The Portable Object Adapter should now support the
430 output of the recent IDL to java compilers. These
431 compilers now generate servants and not CORBA objects as
432 before, making the output depend on the existing POA
433 implementation. Completing POA means that such code can
434 already be tried to run on Classpath. Our POA is tested
435 for the following usager scenarios:
436 # POA converts servant to the CORBA object.
437 # Servant provides to the CORBA object.
438 # POA activates new CORBA object with the given
439 Object Id (byte array) that is later accessible for
440 the servant.
441 # During the first call, the ServantActivator
442 provides servant for this and all subsequent calls
443 on the current object.
444 # During each call, the ServantLocator provides
445 servant for this call only.
446 # ServantLocator or ServantActivator forwards call to
447 another server.
448 # POA has a single servant, responsible for all
449 objects.
450 # POA has a default servant, but some objects are
451 explicitly connected to they specific servants.
452 The POA is verified using tests from the former
453 cost.omg.org.
454 o The CORBA implementation is now a working prototype that
455 should support features up to 1.3 inclusive. We invite
456 groups writing CORBA dependent applications to try
457 Classpath implementation, reporting any possible bugs.
458 The CORBA prototype is interoperable with Sun's
459 implementation v 1.4, transferring object references,
460 primitive types, narrow and wide strings, arrays,
461 structures, trees, abstract interfaces and value types
462 (feature of CORBA 2.3) between these two platforms.
463 Remote exceptions are transferred and handled correctly.
464 The stringified object references (IORs) from various
465 sources are parsed as required. The transient (for
466 current session) and permanent (till jre restart)
467 redirections work. Both Little and Big Endian encoded
468 messages are accepted. The implementation is verified
469 using tests from the former cost.omg.org. The current
470 release includes working examples (see the examples
471 directory), demonstrating the client-server
472 communication, using either CORBA Request or IDL-based
473 stub (usually generated by a IDL to java compiler).
474 These examples also show how to use the Classpath CORBA
475 naming service. The IDL to java compiler is not yet
476 written, but as our library must be compatible, it
477 naturally accepts the output of other idlj
478 implementations.
479 + Misc
480 o Updated TimeZone data against Olson tzdata2005l.
481 o Make zip and jar packages UTF-8 clean.
482 o "native" code builds and compiles (warning free) on
483 Darwin and Solaris.
484 o java.util.logging.FileHandler now rotates files.
485 o Start of a generic JDWP framework in gnu/classpath/jdwp.
486 This is unfinished, but feedback (at classpath@gnu.org)
487 from runtime hackers is greatly appreciated. Although
488 most of the work is currently being done around gcj/gij
489 we want this framework to be as VM neutral as possible.
490 Early design is described in:
491 [3]http://gcc.gnu.org/ml/java/2005-05/msg00260.html
492 o QT4 AWT peers, enable by giving configure
493 --enable-qt-peer. Included, but not ready for production
494 yet. They are explicitly disabled and not supported. But
495 if you want to help with the development of these new
496 features we are interested in feedback. You will have to
497 explicitly enable them to try them out (and they will
498 most likely contain bugs).
499 o Documentation fixes all over the place. See
500 [4]http://developer.classpath.org/doc/
502 New Targets and Target Specific Improvements
504 IA-32/x86-64
506 * The x86-64 medium model (that allows building applications whose
507 data segment exceeds 4GB) was redesigned to match latest ABI
508 draft. New implementation split large datastructures into separate
509 segment improving performance of accesses to small datastructures
510 and also allows linking of small model libraries into medium model
511 programs as long as the libraries are not accessing the large
512 datastructures directly. Medium model is also supported in
513 position independent code now.
514 The ABI change results in partial incompatibility among medium
515 model objects. Linking medium model libraries (or objects)
516 compiled with new compiler into medium model program compiled with
517 older will likely result in exceeding ranges of relocations.
518 Binutils 2.16.91 or newer are required for compiling medium model
519 now.
521 RS6000 (POWER/PowerPC)
523 * The AltiVec vector primitives in <altivec.h> are now implemented
524 in a way that puts a smaller burden on the preprocessor, instead
525 processing the "overloading" in the front ends. This should
526 benefit compilation speed on AltiVec vector code.
527 * AltiVec initializers now are generated more efficiently.
528 * The popcountb instruction available on POWER5 now is generated.
529 * The floating point round to integer instructions available on
530 POWER5+ now is generated.
531 * Floating point divides can be synthesized using the floating point
532 reciprocal estimate instructions.
533 * Double precision floating point constants are initialized as
534 single precision values if they can be represented exactly.
536 S/390, zSeries and System z9
538 * Support for the IBM System z9 109 processor has been added. When
539 using the -march=z9-109 option, the compiler will generate code
540 making use of instructions provided by the extended immediate
541 facility.
542 * Support for 128-bit IEEE floating point has been added. When using
543 the -mlong-double-128 option, the compiler will map the long
544 double data type to 128-bit IEEE floating point. Using this option
545 constitutes an ABI change, and requires glibc support.
546 * Various changes to improve performance of generated code have been
547 implemented, including:
548 + In functions that do not require a literal pool, register
549 %r13 (which is traditionally reserved as literal pool
550 pointer), can now be freely used for other purposes by the
551 compiler.
552 + More precise tracking of register use allows the compiler to
553 generate more efficient function prolog and epilog code in
554 certain cases.
555 + The SEARCH STRING, COMPARE LOGICAL STRING, and MOVE STRING
556 instructions are now used to implement C string functions.
557 + The MOVE CHARACTER instruction with single byte overlap is
558 now used to implement the memset function with non-zero fill
559 byte.
560 + The LOAD ZERO instructions are now used where appropriate.
561 + The INSERT CHARACTERS UNDER MASK, STORE CHARACTERS UNDER
562 MASK, and INSERT IMMEDIATE instructions are now used more
563 frequently to optimize bitfield operations.
564 + The BRANCH ON COUNT instruction is now used more frequently.
565 In particular, the fact that a loop contains a subroutine
566 call no longer prevents the compiler from using this
567 instruction.
568 + The compiler is now aware that all shift and rotate
569 instructions implicitly truncate the shift count to six bits.
570 * Back-end support for the following generic features has been
571 implemented:
572 + The full set of [5]built-in functions for atomic memory
573 access.
574 + The -fstack-protector feature.
575 + The optimization pass avoiding unnecessary stores of incoming
576 argument registers in functions with variable argument list.
578 SPARC
580 * The default code model in 64-bit mode has been changed from
581 Medium/Anywhere to Medium/Middle on Solaris.
582 * TLS support is disabled by default on Solaris prior to release 10.
583 It can be enabled on TLS-capable Solaris 9 versions (4/04 release
584 and later) by specifying --enable-tls at configure time.
586 MorphoSys
588 * Support has been added for this new architecture.
590 Obsolete Systems
592 Documentation improvements
594 Other significant improvements
596 * GCC can now emit code for protecting applications from
597 stack-smashing attacks. The protection is realized by buffer
598 overflow detection and reordering of stack variables to avoid
599 pointer corruption.
600 * Some built-in functions have been fortified to protect them
601 against various buffer overflow (and format string)
602 vulnerabilities. Compared to the mudflap bounds checking feature,
603 the safe builtins have far smaller overhead. This means that
604 programs built using safe builtins should not experience any
605 measurable slowdown.
607 GCC 4.1.2
609 This is the [6]list of problem reports (PRs) from GCC's bug tracking
610 system that are known to be fixed in the 4.1.2 release. This list
611 might not be complete (that is, it is possible that some PRs that have
612 been fixed are not listed here).
614 When generating code for a shared library, GCC now recognizes that
615 global functions may be replaced when the program runs. Therefore, it
616 is now more conservative in deducing information from the bodies of
617 functions. For example, in this example:
618 void f() {}
619 void g() {
620 try { f(); }
621 catch (...) {
622 cout << "Exception";
623 }
624 }
627 G++ would previously have optimized away the catch clause, since it
628 would have concluded that f cannot throw exceptions. Because users may
629 replace f with another function in the main body of the program, this
630 optimization is unsafe, and is no longer performed. If you wish G++ to
631 continue to optimize as before, you must add a throw() clause to the
632 declaration of f to make clear that it does not throw exceptions.
634 Please send FSF & GNU inquiries & questions to [7]gnu@gnu.org. There
635 are also [8]other ways to contact the FSF.
637 These pages are maintained by [9]the GCC team.
640 For questions related to the use of GCC, please consult these web
641 pages and the [10]GCC manuals. If that fails, the
642 [11]gcc-help@gcc.gnu.org mailing list might help.
643 Please send comments on these web pages and the development of GCC
644 to our developer mailing list at [12]gcc@gnu.org or
645 [13]gcc@gcc.gnu.org. All of our lists have [14]public archives.
647 Copyright (C) Free Software Foundation, Inc., 51 Franklin St, Fifth
648 Floor, Boston, MA 02110, USA.
650 Verbatim copying and distribution of this entire article is permitted
651 in any medium, provided this notice is preserved.
652 Last modified 2007-02-14 [15]Valid XHTML 1.0
654 References
656 1. http://gcc.gnu.org/gcc-4.1/changes.html#4.1.2
657 2. http://developer.classpath.org/mediation/ClasspathGraphicsImagesText
658 3. http://gcc.gnu.org/ml/java/2005-05/msg00260.html
659 4. http://developer.classpath.org/doc/
660 5. http://gcc.gnu.org/onlinedocs/gcc-4.1.0/gcc/Atomic-Builtins.html
661 6. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.1.2
662 7. mailto:gnu@gnu.org
663 8. http://www.gnu.org/home.html#ContactInfo
664 9. http://gcc.gnu.org/about.html
665 10. http://gcc.gnu.org/onlinedocs/
666 11. mailto:gcc-help@gcc.gnu.org
667 12. mailto:gcc@gnu.org
668 13. mailto:gcc@gcc.gnu.org
669 14. http://gcc.gnu.org/lists.html
670 15. http://validator.w3.org/check/referer
671 ======================================================================
672 http://gcc.gnu.org/gcc-4.0/index.html
674 GCC 4.0 Release Series
676 January 31, 2007
678 The [1]GNU project and the GCC developers are pleased to announce the
679 release of GCC 4.0.4.
681 This release is a bug-fix release, containing fixes for regressions in
682 GCC 4.0.3 relative to previous releases of GCC.
684 Release History
686 GCC 4.0.4
687 January 31, 2007 ([2]changes)
689 GCC 4.0.3
690 March 10, 2006 ([3]changes)
692 GCC 4.0.2
693 September 28, 2005 ([4]changes)
695 GCC 4.0.1
696 July 7, 2005 ([5]changes)
698 GCC 4.0.0
699 April 20, 2005 ([6]changes)
701 References and Acknowledgements
703 GCC used to stand for the GNU C Compiler, but since the compiler
704 supports several other languages aside from C, it now stands for the
705 GNU Compiler Collection.
707 A list of [7]successful builds is updated as new information becomes
708 available.
710 The GCC developers would like to thank the numerous people that have
711 contributed new features, improvements, bug fixes, and other changes
712 as well as test results to GCC. This [8]amazing group of volunteers is
713 what makes GCC successful.
715 For additional information about GCC please refer to the [9]GCC
716 project web site or contact the [10]GCC development mailing list.
718 To obtain GCC please use [11]our mirror sites, one of the [12]GNU
719 mirror sites, or [13]our SVN server.
721 Please send FSF & GNU inquiries & questions to [14]gnu@gnu.org. There
722 are also [15]other ways to contact the FSF.
724 These pages are maintained by [16]the GCC team.
727 For questions related to the use of GCC, please consult these web
728 pages and the [17]GCC manuals. If that fails, the
729 [18]gcc-help@gcc.gnu.org mailing list might help.
730 Please send comments on these web pages and the development of GCC
731 to our developer mailing list at [19]gcc@gnu.org or
732 [20]gcc@gcc.gnu.org. All of our lists have [21]public archives.
734 Copyright (C) Free Software Foundation, Inc., 51 Franklin St, Fifth
735 Floor, Boston, MA 02110, USA.
737 Verbatim copying and distribution of this entire article is permitted
738 in any medium, provided this notice is preserved.
739 Last modified 2007-02-03 [22]Valid XHTML 1.0
741 References
743 1. http://www.gnu.org/
744 2. http://gcc.gnu.org/gcc-4.0/changes.html#4.0.4
745 3. http://gcc.gnu.org/gcc-4.0/changes.html#4.0.3
746 4. http://gcc.gnu.org/gcc-4.0/changes.html#4.0.2
747 5. http://gcc.gnu.org/gcc-4.0/changes.html#4.0.1
748 6. http://gcc.gnu.org/gcc-4.0/changes.html
749 7. http://gcc.gnu.org/gcc-4.0/buildstat.html
750 8. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
751 9. http://gcc.gnu.org/index.html
752 10. mailto:gcc@gcc.gnu.org
753 11. http://gcc.gnu.org/mirrors.html
754 12. http://www.gnu.org/order/ftp.html
755 13. http://gcc.gnu.org/svn.html
756 14. mailto:gnu@gnu.org
757 15. http://www.gnu.org/home.html#ContactInfo
758 16. http://gcc.gnu.org/about.html
759 17. http://gcc.gnu.org/onlinedocs/
760 18. mailto:gcc-help@gcc.gnu.org
761 19. mailto:gcc@gnu.org
762 20. mailto:gcc@gcc.gnu.org
763 21. http://gcc.gnu.org/lists.html
764 22. http://validator.w3.org/check/referer
765 ======================================================================
766 http://gcc.gnu.org/gcc-4.0/changes.html
768 GCC 4.0 Release Series
769 Changes, New Features, and Fixes
771 The latest release in the 4.0 release series is [1]GCC 4.0.4.
773 Caveats
775 * GCC now generates location lists by default when compiling with
776 debug info and optimization.
777 + GDB 6.0 and older crashes when it sees location lists. GDB
778 6.1 or later is needed to debug binaries containing location
779 lists.
780 + When you are trying to view a value of a variable in a part
781 of a function where it has no location (for example when the
782 variable is no longer used and thus its location was used for
783 something else) GDB will say that it is not available.
784 You can disable generating location lists by -fno-var-tracking.
785 * GCC no longer accepts the -fwritable-strings option. Use named
786 character arrays when you need a writable string.
787 * The options -freduce-all-givs and -fmove-all-movables have been
788 discontinued. They were used to circumvent a shortcoming in the
789 heuristics of the old loop optimization code with respect to
790 common Fortran constructs. The new (tree) loop optimizer works
791 differently and doesn't need those work-arounds.
792 * The graph-coloring register allocator, formerly enabled by the
793 option -fnew-ra, has been discontinued.
794 * -I- has been deprecated. -iquote is meant to replace the need for
795 this option.
796 * The MIPS -membedded-pic and -mrnames options have been removed.
797 * All MIPS targets now require the GNU assembler. In particular,
798 IRIX configurations can no longer use the MIPSpro assemblers,
799 although they do still support the MIPSpro linkers.
800 * The SPARC option -mflat has been removed.
801 * English-language diagnostic messages will now use Unicode
802 quotation marks in UTF-8 locales. (Non-English messages already
803 used the quotes appropriate for the language in previous
804 releases.) If your terminal does not support UTF-8 but you are
805 using a UTF-8 locale (such locales are the default on many
806 GNU/Linux systems) then you should set LC_CTYPE=C in the
807 environment to disable that locale. Programs that parse
808 diagnostics and expect plain ASCII English-language messages
809 should set LC_ALL=C. See [2]Markus Kuhn's explanation of Unicode
810 quotation marks for more information.
811 * The specs file is no longer installed on most platforms. Most
812 users will be totally unaffected. However, if you are accustomed
813 to editing the specs file yourself, you will now have to use the
814 -dumpspecs option to generate the specs file, and then edit the
815 resulting file.
817 General Optimizer Improvements
819 * The [3]tree ssa branch has been merged. This merge has brought in
820 a completely new optimization framework based on a higher level
821 intermediate representation than the existing RTL representation.
822 Numerous new code transformations based on the new framework are
823 available in GCC 4.0, including:
824 + Scalar replacement of aggregates
825 + Constant propagation
826 + Value range propagation
827 + Partial redundancy elimination
828 + Load and store motion
829 + Strength reduction
830 + Dead store elimination
831 + Dead and unreachable code elimination
832 + [4]Autovectorization
833 + Loop interchange
834 + Tail recursion by accumulation
835 Many of these passes outperform their counterparts from previous
836 GCC releases.
837 * [5]Swing Modulo Scheduling (SMS). An RTL level instruction
838 scheduling optimization intended for loops that perform heavy
839 computations.
841 New Languages and Language specific improvements
843 C family
845 * The sentinel attribute has been added to GCC. This function
846 attribute allows GCC to warn when variadic functions such as execl
847 are not NULL terminated. See the GCC manual for a complete
848 description of its behavior.
849 * Given __attribute__((alias("target"))) it is now an error if
850 target is not a symbol, defined in the same translation unit. This
851 also applies to aliases created by #pragma weak alias=target. This
852 is because it's meaningless to define an alias to an undefined
853 symbol. On Solaris, the native assembler would have caught this
854 error, but GNU as does not.
856 C and Objective-C
858 * The -Wstrict-aliasing=2 option has been added. This warning
859 catches all unsafe cases, but it may also give a warning for some
860 cases that are safe.
861 * The cast-as-lvalue, conditional-expression-as-lvalue and
862 compound-expression-as-lvalue extensions, which were deprecated in
863 3.3.4 and 3.4, have been removed.
864 * The -fwritable-strings option, which was deprecated in 3.4, has
865 been removed.
866 * #pragma pack() semantics have been brought closer to those used by
867 other compilers. This also applies to C++.
868 * Taking the address of a variable with register storage is invalid
869 in C. GCC now issues an error instead of a warning.
870 * Arrays of incomplete element type are invalid in C. GCC now issues
871 an error for such arrays. Declarations such as extern struct s
872 x[]; (where struct s has not been defined) can be moved after the
873 definition of struct s. Function parameters declared as arrays of
874 incomplete type can instead be declared as pointers.
876 C++
878 * When compiling without optimizations (-O0), the C++ frontend is
879 much faster than in any previous versions of GCC. Independent
880 testers have measured speed-ups up to 25% in real-world production
881 code, compared to the 3.4 family (which was already the fastest
882 version to date). Upgrading from older versions might show even
883 bigger improvements.
884 * ELF visibility attributes can now be applied to a class type, so
885 that it affects every member function of a class at once, without
886 having to specify each individually:
887 class __attribute__ ((visibility("hidden"))) Foo
888 {
889 int foo1();
890 void foo2();
891 };
892 The syntax is deliberately similar to the __declspec() system used
893 by Microsoft Windows based compilers, allowing cross-platform
894 projects to easily reuse their existing macro system for denoting
895 exports and imports. By explicitly marking internal classes never
896 used outside a binary as hidden, one can completely avoid PLT
897 indirection overheads during their usage by the compiler. You can
898 find out more about the advantages of this at
899 [6]http://people.redhat.com/drepper/dsohowto.pdf
900 * The -fvisibility-inlines-hidden option has been added which marks
901 all inlineable functions as having hidden ELF visibility, thus
902 removing their symbol and typeinfo from the exported symbol table
903 of the output ELF binary. Using this option can reduce the
904 exported symbol count of template-heavy code by up to 40% with no
905 code change at all, thus notably improving link and load times for
906 the binary as well as a reduction in size of up to 10%. Also,
907 check the new [7]-fvisibility option.
908 * The compiler now uses the library interface specified by the
909 [8]C++ ABI for thread-safe initialization of function-scope static
910 variables. Most users should leave this alone, but embedded
911 programmers may want to disable this by specifying
912 -fno-threadsafe-statics for a small savings in code size.
913 * Taking the address of an explicit register variable is no longer
914 supported. Note that C++ allows taking the address of variables
915 with register storage so this will continue to compile with a
916 warning. For example, assuming that r0 is a machine register:
917 register int foo asm ("r0");
918 register int bar;
919 &foo; // error, no longer accepted
920 &bar; // OK, with a warning
921 * G++ has an undocumented extension to virtual function covariancy
922 rules that allowed the overrider to return a type that was
923 implicitly convertable to the overridden function's return type.
924 For instance a function returning void * could be overridden by a
925 function returning T *. This is now deprecated and will be removed
926 in a future release.
927 * The G++ minimum and maximum operators (<? and >?) and their
928 compound forms (<?=) and >?=) have been deprecated and will be
929 removed in a future version. Code using these operators should be
930 modified to use std::min and std::max instead.
931 * Declaration of nested classes of class templates as friends are
932 supported:
933 template <typename T> struct A {
934 class B {};
935 };
936 class C {
937 template <typename T> friend class A<T>::B;
938 };
939 This complements the feature member functions of class templates
940 as friends introduced in GCC 3.4.0.
941 * When declaring a friend class using an unqualified name, classes
942 outside the innermost non-class scope are not searched:
943 class A;
944 namespace N {
945 class B {
946 friend class A; // Refer to N::A which has not been declared yet
947 // because name outside namespace N are not searched
948 friend class ::A; // Refer to ::A
949 };
950 }
951 Hiding the friend name until declaration is still not implemented.
952 * Friends of classes defined outside their namespace are correctly
953 handled:
954 namespace N {
955 class A;
956 }
957 class N::A {
958 friend class B; // Refer to N::B in GCC 4.0.0
959 // but ::B in earlier versions of GCC
960 };
962 Runtime Library (libstdc++)
964 * Optimization work:
965 + Added efficient specializations of istream functions for char
966 and wchar_t.
967 + Further performance tuning of strings, in particular wrt
968 single-char append and getline.
969 + iter_swap - and therefore most of the mutating algorithms -
970 now makes an unqualified call to swap when the value_type of
971 the two iterators is the same.
972 * A large subset of the features in Technical Report 1 (TR1 for
973 short) is experimentally delivered (i.e., no guarantees about the
974 implementation are provided. In particular it is not promised that
975 the library will remain link-compatible when code using TR1 is
976 used):
977 + General utilities such as reference_wrapper and shared_ptr.
978 + Function objects, i.e., result_of, mem_fn, bind, function.
979 + Support for metaprogramming.
980 + New containers such as tuple, array, unordered_set,
981 unordered_map, unordered_multiset, unordered_multimap.
982 * As usual, many bugs have been fixed and LWG resolutions
983 implemented for the first time (e.g., DR 409).
985 Java
987 * In order to prevent naming conflicts with other implementations of
988 these tools, some GCJ binaries have been renamed:
989 + rmic is now grmic,
990 + rmiregistry is now grmiregistry, and
991 + jar is now fastjar.
992 In particular, these names were problematic for the jpackage.org
993 packaging conventions which install symlinks in /usr/bin that
994 point to the preferred versions of these tools.
995 * The -findirect-dispatch argument to the compiler now works and
996 generates code following a new "binary compatibility" ABI. Code
997 compiled this way follows the binary compatibility rules of the
998 Java Language Specification.
999 * libgcj now has support for using GCJ as a JIT, using the
1000 gnu.gcj.jit family of system properties.
1001 * libgcj can now find a shared library corresponding to the bytecode
1002 representation of a class. See the documentation for the new
1003 gcj-dbtool program, and the new gnu.gcj.precompiled.db.path system
1004 property.
1005 * There have been many improvements to the class library. Here are
1006 some highlights:
1007 + Much more of AWT and Swing exist.
1008 + Many new packages and classes were added, including
1009 java.util.regex, java.net.URI, javax.crypto,
1010 javax.crypto.interfaces, javax.crypto.spec, javax.net,
1011 javax.net.ssl, javax.security.auth,
1012 javax.security.auth.callback, javax.security.auth.login,
1013 javax.security.auth.x500, javax.security.sasl, org.ietf.jgss,
1014 javax.imageio, javax.imageio.event, javax.imageio.spi,
1015 javax.print, javax.print.attribute,
1016 javax.print.attribute.standard, javax.print.event, and
1017 javax.xml
1018 + Updated SAX and DOM, and imported GNU JAXP
1020 Fortran
1022 * A new [9]Fortran front end has replaced the aging GNU Fortran 77
1023 front end. The new front end supports Fortran 90 and Fortran 95.
1024 It may not yet be as stable as the old Fortran front end.
1026 Ada
1028 * Ada (with tasking and Zero Cost Exceptions) is now available on
1029 many more targets, including but not limited to: alpha-linux,
1030 hppa-hpux, hppa-linux, powerpc-darwin, powerpc-linux, s390-linux,
1031 s390x-linux, sparc-linux.
1032 * Some of the new Ada 2005 features are now implemented like
1033 Wide_Wide_Character and Ada.Containers.
1034 * Many bugs have been fixed, tools and documentation improved.
1035 * To compile Ada from the sources, install an older working Ada
1036 compiler and then use --enable-languages=ada at configuration
1037 time, since the Ada frontend is not currently activated by
1038 default. See the [10]Installing GCC for details.
1040 New Targets and Target Specific Improvements
1042 H8/300
1044 * The frame layout has changed. In the new layout, the prologue of a
1045 function first saves registers and then allocate space for locals,
1046 resulting in an 1% improvement on code size.
1048 IA-32/x86-64 (AMD64)
1050 * The acos, asin, drem, exp10, exp2, expm1, fmod, ilogb, log10,
1051 log1p, log2, logb and tan mathematical builtins (and their float
1052 and long double variants) are now implemented as inline x87
1053 intrinsics when using -ffast-math.
1054 * The ceil, floor, nearbyint, rint and trunc mathematical builtins
1055 (and their float and long double variants) are now implemented as
1056 inline x87 intrinsics when using -ffast-math.
1057 * The x87's fsincos instruction is now used automatically with
1058 -ffast-math when calculating both the sin and cos of the same
1059 argument.
1060 * Instruction selection for multiplication and division by constants
1061 has been improved.
1063 IA-64
1065 * Floating point division, integer division and sqrt are now
1066 inlined, resulting in significant performance improvements on some
1067 codes.
1069 MIPS
1071 * Division by zero checks now use conditional traps if the target
1072 processor supports them. This decreases code size by one word per
1073 division operation. The old behavior (branch and break) can be
1074 obtained either at configure time by passing --with-divide=breaks
1075 to configure or at runtime by passing -mdivide-breaks to GCC.
1076 * Support for MIPS64 paired-single instructions has been added. It
1077 is enabled by -mpaired-single and can be accessed using both the
1078 target-independent vector extensions and new MIPS-specific
1079 built-in functions.
1080 * Support for the MIPS-3D ASE has been added. It is enabled by
1081 -mips3d and provides new MIPS-3D-specific built-in functions.
1082 * The -mexplicit-relocs option now supports static n64 code (as is
1083 used, for example, in 64-bit linux kernels). -mexplicit-relocs
1084 should now be feature-complete and is enabled by default when GCC
1085 is configured to use a compatible assembler.
1086 * Support for the NEC VR4130 series has been added. This support
1087 includes the use of VR-specific instructions and a new VR4130
1088 scheduler. Full VR4130 support can be selected with -march=vr4130
1089 while code for any ISA can be tuned for the VR4130 using
1090 -mtune=vr4130. There is also a new -mvr4130-align option that
1091 produces better schedules at the cost of increased code size.
1092 * Support for the Broadcom SB-1 has been extended. There is now an
1093 SB-1 scheduler as well as support for the SB-1-specific
1094 paired-single instructions. Full SB-1 support can be selected with
1095 -march=sb1 while code for any ISA can be optimized for the SB-1
1096 using -mtune=sb1.
1097 * The compiler can now work around errata in R4000, R4400, VR4120
1098 and VR4130 processors. These workarounds are enabled by
1099 -mfix-r4000, -mfix-r4400, -mfix-vr4120 and -mfix-vr4130
1100 respectively. The VR4120 and VR4130 workarounds need binutils 2.16
1101 or above.
1102 * IRIX shared libraries are now installed into the standard library
1103 directories: o32 libraries go into lib/, n32 libraries go into
1104 lib32/ and n64 libraries go into lib64/.
1105 * The compiler supports a new -msym32 option. It can be used to
1106 optimize n64 code in which all symbols are known to have 32-bit
1107 values.
1109 S/390 and zSeries
1111 * New command line options help to generate code intended to run in
1112 an environment where stack space is restricted, e.g. Linux kernel
1113 code:
1114 + -mwarn-framesize and -mwarn-dynamicstack trigger compile-time
1115 warnings for single functions that require large or dynamic
1116 stack frames.
1117 + -mstack-size and -mstack-guard generate code that checks for
1118 stack overflow at run time.
1119 + -mpacked-stack generates code that reduces the stack frame
1120 size of many functions by reusing unneeded parts of the stack
1121 bias area.
1122 * The -msoft-float option now ensures that generated code never
1123 accesses floating point registers.
1124 * The s390x-ibm-tpf target now fully supports C++, including
1125 exceptions and threads.
1126 * Various changes to improve performance of the generated code have
1127 been implemented, including:
1128 + GCC now uses sibling calls where possible.
1129 + Condition code handling has been optimized, allowing GCC to
1130 omit redundant comparisons in certain cases.
1131 + The cost function guiding many optimizations has been refined
1132 to more accurately represent the z900 and z990 processors.
1133 + The ADD LOGICAL WITH CARRY and SUBTRACT LOGICAL WITH BORROW
1134 instructions are now used to avoid conditional branches in
1135 certain cases.
1136 + The back end now uses the LEGITIMIZE_RELOAD_ADDRESS feature
1137 to optimize address arithmetic required to access large stack
1138 frames.
1139 + GCC now makes more efficient use of memory-to-memory type
1140 instructions (MVC, CLC, ...).
1141 + More precise tracking of special register use allows better
1142 instruction scheduling, in particular of the function
1143 prologue and epilogue sequences.
1144 + The Java front end now generates inline code to implement
1145 integer division, instead of calling library routines.
1147 SPARC
1149 * The options -mv8, -msparclite, -mcypress, -msupersparc, -mf930 and
1150 -mf934 have been removed. They have been replaced with -mcpu=xxx.
1151 * The internal model used to estimate the relative cost of each
1152 instruction has been updated. It is expected to give better
1153 results on recent UltraSPARC processors.
1154 * Code generation for function prologues and epilogues has been
1155 improved, resulting in better scheduling and allowing multiple
1156 exit points in functions.
1157 * Support for Sun's Visual Instruction Set (VIS) has been enhanced.
1158 It is enabled by -mvis and provides new built-in functions for VIS
1159 instructions on UltraSPARC processors.
1160 * The option -mapp-regs has been turned on by default on Solaris
1161 too.
1163 NetWare
1165 * Novell NetWare (on ix86, no other hardware platform was ever
1166 really supported by this OS) has been re-enabled and the ABI
1167 supported by GCC has been brought into sync with that of
1168 MetroWerks CodeWarrior (the ABI previously supported was that of
1169 some Unix systems, which NetWare never tried to support).
1171 Obsolete Systems
1173 Support for a number of older systems has been declared obsolete in
1174 GCC 4.0. Unless there is activity to revive them, the next release of
1175 GCC will have their sources permanently removed.
1177 All GCC ports for the following processor architectures have been
1178 declared obsolete:
1179 * Intel i860
1180 * Ubicom IP2022
1181 * National Semiconductor NS32K
1182 * Texas Instruments TMS320C[34]x
1184 Also, those for some individual systems have been obsoleted:
1185 * SPARC family
1186 + SPARClite-based systems (sparclite-*-coff, sparclite-*-elf,
1187 sparc86x-*-elf)
1188 + OpenBSD 32-bit (sparc-*-openbsd*)
1190 Documentation improvements
1192 Other significant improvements
1194 * Location lists are now generated by default when compiling with
1195 debug info and optimization. Location lists provide more accurate
1196 debug info about locations of variables and they allow debugging
1197 code compiled with -fomit-frame-pointer.
1198 * The -fvisibility option has been added which allows the default
1199 ELF visibility of all symbols to be set per compilation and the
1200 new #pragma GCC visibility preprocessor command allows the setting
1201 of default ELF visibility for a region of code. Using
1202 -fvisibility=hidden especially in combination with the new
1203 -fvisibility-inlines-hidden can yield substantial improvements in
1204 output binary quality including avoiding PLT indirection
1205 overheads, reduction of the exported symbol count by up to 60%
1206 (with resultant improvements to link and load times), better scope
1207 for the optimizer to improve code and up to a 20% reduction in
1208 binary size. Using these options correctly yields a binary with a
1209 similar symbol count to a Windows DLL.
1210 Perhaps more importantly, this new feature finally allows (with
1211 careful planning) complete avoidance of symbol clashes when
1212 manually loading shared objects with RTLD_GLOBAL, thus finally
1213 solving problems many projects such as python were forced to use
1214 RTLD_LOCAL for (with its resulting issues for C++ correctness).
1215 You can find more information about using these options at
1216 [11]http://gcc.gnu.org/wiki/Visibility.
1217 _________________________________________________________________
1219 GCC 4.0.1
1221 This is the [12]list of problem reports (PRs) from GCC's bug tracking
1222 system that are known to be fixed in the 4.0.1 release. This list
1223 might not be complete (that is, it is possible that some PRs that have
1224 been fixed are not listed here).
1226 GCC 4.0.2
1228 This is the [13]list of problem reports (PRs) from GCC's bug tracking
1229 system that are known to be fixed in the 4.0.2 release. This list
1230 might not be complete (that is, it is possible that some PRs that have
1231 been fixed are not listed here).
1233 Unfortunately, due to a release engineering failure, this release has
1234 a regression on Solaris that will affect some C++ programs. We suggest
1235 that Solaris users apply a [14]patch that corrects the problem. Users
1236 who do not wish to apply the patch should explicitly link C++ programs
1237 with the -pthreads option, even if they do not use threads. This
1238 problem has been corrected in the current 4.0 branch sources and will
1239 not be present in GCC 4.0.3.
1241 GCC 4.0.3
1243 Starting with this release, the function getcontext is recognized by
1244 the compiler as having the same semantics as the setjmp function. In
1245 particular, the compiler will ensure that all registers are dead
1246 before calling such a function and will emit a warning about the
1247 variables that may be clobbered after the second return from the
1248 function.
1250 GCC 4.0.4
1252 This is the [15]list of problem reports (PRs) from GCC's bug tracking
1253 system that are known to be fixed in the 4.0.4 release. This list
1254 might not be complete (that is, it is possible that some PRs that have
1255 been fixed are not listed here).
1257 The 4.0.4 release is provided for those that require a high degree of
1258 binary compatibility with previous 4.0.x releases. For most users, the
1259 GCC team recommends that version 4.1.1 or later be used instead."
1261 Please send FSF & GNU inquiries & questions to [16]gnu@gnu.org. There
1262 are also [17]other ways to contact the FSF.
1264 These pages are maintained by [18]the GCC team.
1267 For questions related to the use of GCC, please consult these web
1268 pages and the [19]GCC manuals. If that fails, the
1269 [20]gcc-help@gcc.gnu.org mailing list might help.
1270 Please send comments on these web pages and the development of GCC
1271 to our developer mailing list at [21]gcc@gnu.org or
1272 [22]gcc@gcc.gnu.org. All of our lists have [23]public archives.
1274 Copyright (C) Free Software Foundation, Inc., 51 Franklin St, Fifth
1275 Floor, Boston, MA 02110, USA.
1277 Verbatim copying and distribution of this entire article is permitted
1278 in any medium, provided this notice is preserved.
1279 Last modified 2007-02-03 [24]Valid XHTML 1.0
1281 References
1283 1. http://gcc.gnu.org/gcc-4.0/changes.html#4.0.4
1284 2. http://www.cl.cam.ac.uk/~mgk25/ucs/quotes.html
1285 3. http://gcc.gnu.org/projects/tree-ssa/
1286 4. http://gcc.gnu.org/projects/tree-ssa/vectorization.html
1287 5. http://gcc.gnu.org/news/sms.html
1288 6. http://people.redhat.com/drepper/dsohowto.pdf
1289 7. http://gcc.gnu.org/gcc-4.0/changes.html#visibility
1290 8. http://www.codesourcery.com/cxx-abi/
1291 9. http://gcc.gnu.org/fortran/
1292 10. http://gcc.gnu.org/install/
1293 11. http://gcc.gnu.org/wiki/Visibility
1294 12. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.0.1
1295 13. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.0.2
1296 14. http://gcc.gnu.org/ml/gcc-cvs/2005-09/msg00984.html
1297 15. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=4.0.4
1298 16. mailto:gnu@gnu.org
1299 17. http://www.gnu.org/home.html#ContactInfo
1300 18. http://gcc.gnu.org/about.html
1301 19. http://gcc.gnu.org/onlinedocs/
1302 20. mailto:gcc-help@gcc.gnu.org
1303 21. mailto:gcc@gnu.org
1304 22. mailto:gcc@gcc.gnu.org
1305 23. http://gcc.gnu.org/lists.html
1306 24. http://validator.w3.org/check/referer
1307 ======================================================================
1308 http://gcc.gnu.org/gcc-3.4/index.html
1310 GCC 3.4 Release Series
1312 May 26, 2006
1314 The [1]GNU project and the GCC developers are pleased to announce the
1315 release of GCC 3.4.6.
1317 This release is a bug-fix release, containing fixes for regressions in
1318 GCC 3.4.4 relative to previous releases of GCC. This is the last of
1319 the 3.4.x series.
1321 The GCC 3.4 release series includes numerous [2]new features,
1322 improvements, bug fixes, and other changes, thanks to an [3]amazing
1323 group of volunteers.
1325 Release History
1327 GCC 3.4.6
1328 March 6, 2006 ([4]changes)
1330 GCC 3.4.5
1331 November 30, 2005 ([5]changes)
1333 GCC 3.4.4
1334 May 18, 2005 ([6]changes)
1336 GCC 3.4.3
1337 November 4, 2004 ([7]changes)
1339 GCC 3.4.2
1340 September 6, 2004 ([8]changes)
1342 GCC 3.4.1
1343 July 1, 2004 ([9]changes)
1345 GCC 3.4.0
1346 April 18, 2004 ([10]changes)
1348 References and Acknowledgements
1350 GCC used to stand for the GNU C Compiler, but since the compiler
1351 supports several other languages aside from C, it now stands for the
1352 GNU Compiler Collection.
1354 A list of [11]successful builds is updated as new information becomes
1355 available.
1357 The GCC developers would like to thank the numerous people that have
1358 contributed new features, improvements, bug fixes, and other changes
1359 as well as test results to GCC. This [12]amazing group of volunteers
1360 is what makes GCC successful.
1362 For additional information about GCC please refer to the [13]GCC
1363 project web site or contact the [14]GCC development mailing list.
1365 To obtain GCC please use [15]our mirror sites, one of the [16]GNU
1366 mirror sites, or [17]our SVN server.
1368 Please send FSF & GNU inquiries & questions to [18]gnu@gnu.org. There
1369 are also [19]other ways to contact the FSF.
1371 These pages are maintained by [20]the GCC team.
1374 For questions related to the use of GCC, please consult these web
1375 pages and the [21]GCC manuals. If that fails, the
1376 [22]gcc-help@gcc.gnu.org mailing list might help.
1377 Please send comments on these web pages and the development of GCC
1378 to our developer mailing list at [23]gcc@gnu.org or
1379 [24]gcc@gcc.gnu.org. All of our lists have [25]public archives.
1381 Copyright (C) Free Software Foundation, Inc., 51 Franklin St, Fifth
1382 Floor, Boston, MA 02110, USA.
1384 Verbatim copying and distribution of this entire article is permitted
1385 in any medium, provided this notice is preserved.
1386 Last modified 2006-06-21 [26]Valid XHTML 1.0
1388 References
1390 1. http://www.gnu.org/
1391 2. http://gcc.gnu.org/gcc-3.4/changes.html
1392 3. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
1393 4. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.6
1394 5. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.5
1395 6. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.4
1396 7. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.3
1397 8. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.2
1398 9. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.1
1399 10. http://gcc.gnu.org/gcc-3.4/changes.html
1400 11. http://gcc.gnu.org/gcc-3.4/buildstat.html
1401 12. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
1402 13. http://gcc.gnu.org/index.html
1403 14. mailto:gcc@gcc.gnu.org
1404 15. http://gcc.gnu.org/mirrors.html
1405 16. http://www.gnu.org/order/ftp.html
1406 17. http://gcc.gnu.org/svn.html
1407 18. mailto:gnu@gnu.org
1408 19. http://www.gnu.org/home.html#ContactInfo
1409 20. http://gcc.gnu.org/about.html
1410 21. http://gcc.gnu.org/onlinedocs/
1411 22. mailto:gcc-help@gcc.gnu.org
1412 23. mailto:gcc@gnu.org
1413 24. mailto:gcc@gcc.gnu.org
1414 25. http://gcc.gnu.org/lists.html
1415 26. http://validator.w3.org/check/referer
1416 ======================================================================
1417 http://gcc.gnu.org/gcc-3.4/changes.html
1419 GCC 3.4 Release Series
1420 Changes, New Features, and Fixes
1422 The final release in the 3.4 release series is [1]GCC 3.4.6. The
1423 series is now closed.
1425 GCC 3.4 has [2]many improvements in the C++ frontend. Before reporting
1426 a bug, please make sure it's really GCC, and not your code, that is
1427 broken.
1429 Caveats
1431 * GNU Make is now required to build GCC.
1432 * With -nostdinc the preprocessor used to ignore both standard
1433 include paths and include paths contained in environment
1434 variables. It was neither documented nor intended that environment
1435 variable paths be ignored, so this has been corrected.
1436 * GCC no longer accepts the options -fvolatile, -fvolatile-global
1437 and -fvolatile-static. It is unlikely that they worked correctly
1438 in any 3.x release.
1439 * GCC no longer ships <varargs.h>. Use <stdarg.h> instead.
1440 * Support for all the systems [3]obsoleted in GCC 3.3 has been
1441 removed from GCC 3.4. See below for a [4]list of systems which are
1442 obsoleted in this release.
1443 * GCC now requires an ISO C90 (ANSI C89) C compiler to build. K&R C
1444 compilers will not work.
1445 * The implementation of the [5]MIPS ABIs has changed. As a result,
1446 the code generated for certain MIPS targets will not be binary
1447 compatible with earlier releases.
1448 * In previous releases, the MIPS port had a fake "hilo" register
1449 with the user-visible name accum. This register has been removed.
1450 * The implementation of the [6]SPARC ABIs has changed. As a result,
1451 the code generated will not be binary compatible with earlier
1452 releases in certain cases.
1453 * The configure option --enable-threads=pthreads has been removed;
1454 use --enable-threads=posix instead, which should have the same
1455 effect.
1456 * Code size estimates used by inlining heuristics for C,
1457 Objective-C, C++ and Java have been redesigned significantly. As a
1458 result the parameters of -finline-insns, --param
1459 max-inline-insns-single and --param max-inline-insns-auto need to
1460 be reconsidered.
1461 * --param max-inline-slope and --param min-inline-insns have been
1462 removed; they are not needed for the new bottom-up inlining
1463 heuristics.
1464 * The new unit-at-a-time compilation scheme has several
1465 compatibility issues:
1466 + The order in which functions, variables, and top-level asm
1467 statements are emitted may have changed. Code relying on some
1468 particular ordering needs to be updated. The majority of such
1469 top-level asm statements can be replaced by section
1470 attributes.
1471 + Unreferenced static variables and functions are removed. This
1472 may result in undefined references when an asm statement
1473 refers to the variable/function directly. In that case either
1474 the variable/function shall be listed in asm statement
1475 operand or in the case of top-level asm statements the
1476 attribute used shall be used to force function/variable to be
1477 always output and considered as a possibly used by unknown
1478 code.
1479 For variables the attribute is accepted only by GCC 3.4 and
1480 newer, while for earlier versions it is sufficient to use
1481 unused to silence warnings about the variables not being
1482 referenced. To keep code portable across different GCC
1483 versions, you can use appropriate preprocessor conditionals.
1484 + Static functions now can use non-standard passing conventions
1485 that may break asm statements calling functions directly.
1486 Again the attribute used shall be used to prevent this
1487 behavior.
1488 As a temporary workaround, -fno-unit-at-a-time can be used, but
1489 this scheme may not be supported by future releases of GCC.
1490 * GCC 3.4 automatically places zero-initialized variables in the
1491 .bss section on some operating systems. Versions of GNU Emacs up
1492 to (and including) 21.3 will not work correctly when using this
1493 optimization; you can use -fno-zero-initialized-in-bss to disable
1494 it.
1495 * If GCC 3.4 is configured with --enable-threads=posix (the default
1496 on most targets that support pthreads) then _REENTRANT will be
1497 defined unconditionally by some libstdc++ headers. C++ code which
1498 relies on that macro to detect whether multi-threaded code is
1499 being compiled might change in meaning, possibly resulting in
1500 linker errors for single-threaded programs. Affected users of
1501 [7]Boost should compile single-threaded code with
1502 -DBOOST_DISABLE_THREADS. See Bugzilla for [8]more information.
1504 General Optimizer Improvements
1506 * Usability of the profile feedback and coverage testing has been
1507 improved.
1508 + Performance of profiled programs has been improved by faster
1509 profile merging code.
1510 + Better use of the profile feedback for optimization (loop
1511 unrolling and loop peeling).
1512 + File locking support allowing fork() calls and parallel runs
1513 of profiled programs.
1514 + Coverage file format has been redesigned.
1515 + gcov coverage tool has been improved.
1516 + make profiledbootstrap available to build a faster compiler.
1517 Experiments made on i386 hardware showed an 11% speedup on
1518 -O0 and a 7.5% speedup on -O2 compilation of a [9]large C++
1519 testcase.
1520 + New value profiling pass enabled via -fprofile-values
1521 + New value profile transformations pass enabled via -fvpt aims
1522 to optimize some code sequences by exploiting knowledge about
1523 value ranges or other properties of the operands. At the
1524 moment a conversion of expensive divisions into cheaper
1525 operations has been implemented.
1526 + New -fprofile-generate and -fprofile-use command line options
1527 to simplify the use of profile feedback.
1528 * A new unit-at-a-time compilation scheme for C, Objective-C, C++
1529 and Java which is enabled via -funit-at-a-time (and implied by
1530 -O2). In this scheme a whole file is parsed first and optimized
1531 later. The following basic inter-procedural optimizations are
1532 implemented:
1533 + Removal of unreachable functions and variables
1534 + Discovery of local functions (functions with static linkage
1535 whose address is never taken)
1536 + On i386, these local functions use register parameter passing
1537 conventions.
1538 + Reordering of functions in topological order of the call
1539 graph to enable better propagation of optimizing hints (such
1540 as the stack alignments needed by functions) in the back end.
1541 + Call graph based out-of-order inlining heuristics which
1542 allows to limit overall compilation unit growth (--param
1543 inline-unit-growth).
1544 Overall, the unit-at-a-time scheme produces a 1.3% improvement for
1545 the SPECint2000 benchmark on the i386 architecture (AMD Athlon
1546 CPU).
1547 * More realistic code size estimates used by inlining for C,
1548 Objective-C, C++ and Java. The growth of large functions can now
1549 be limited via --param large-function-insns and --param
1550 large-function-growth.
1551 * A new cfg-level loop optimizer pass replaces the old loop
1552 unrolling pass and adds two other loop transformations -- loop
1553 peeling and loop unswitching -- and also uses the profile feedback
1554 to limit code growth. (The three optimizations are enabled by
1555 -funroll-loops, -fpeel-loops and -funswitch-loops flags,
1556 respectively).
1557 The old loop unroller still can be enabled by -fold-unroll-loops
1558 and may produce better code in some cases, especially when the
1559 webizer optimization pass is not run.
1560 * A new web construction pass enabled via -fweb (and implied by -O3)
1561 improves the quality of register allocation, CSE, first scheduling
1562 pass and some other optimization passes by avoiding re-use of
1563 pseudo registers with non-overlapping live ranges. The pass almost
1564 always improves code quality but does make debugging difficult and
1565 thus is not enabled by default by -O2
1566 The pass is especially effective as cleanup after code duplication
1567 passes, such as the loop unroller or the tracer.
1568 * Experimental implementations of superblock or trace scheduling in
1569 the second scheduling pass can be enabled via
1570 -fsched2-use-superblocks and -fsched2-use-traces, respectively.
1572 New Languages and Language specific improvements
1574 Ada
1576 * The Ada front end has been updated to include numerous bug fixes
1577 and enhancements. These include:
1578 + Improved project file support
1579 + Additional set of warnings about potential wrong code
1580 + Improved error messages
1581 + Improved code generation
1582 + Improved cross reference information
1583 + Improved inlining
1584 + Better run-time check elimination
1585 + Better error recovery
1586 + More efficient implementation of unbounded strings
1587 + Added features in GNAT.Sockets, GNAT.OS_Lib,
1588 GNAT.Debug_Pools, ...
1589 + New GNAT.xxxx packages (e.g. GNAT.Strings,
1590 GNAT.Exception_Action)
1591 + New pragmas
1592 + New -gnatS switch replacing gnatpsta
1593 + Implementation of new Ada features (in particular limited
1594 with, limited aggregates)
1596 C/Objective-C/C++
1598 * Precompiled headers are now supported. Precompiled headers can
1599 dramatically speed up compilation of some projects. There are some
1600 known defects in the current precompiled header implementation
1601 that will result in compiler crashes in relatively rare
1602 situations. Therefore, precompiled headers should be considered a
1603 "technology preview" in this release. Read the manual for details
1604 about how to use precompiled headers.
1605 * File handling in the preprocessor has been rewritten. GCC no
1606 longer gets confused by symlinks and hardlinks, and now has a
1607 correct implementation of #import and #pragma once. These two
1608 directives have therefore been un-deprecated.
1609 * The undocumented extension that allowed C programs to have a label
1610 at the end of a compound statement, which has been deprecated
1611 since GCC 3.0, has been removed.
1612 * The cast-as-lvalue extension has been removed for C++ and
1613 deprecated for C and Objective-C. In particular, code like this:
1614 int i;
1615 (char) i = 5;
1617 or this:
1618 char *p;
1619 ((int *) p)++;
1621 is no longer accepted for C++ and will not be accepted for C and
1622 Objective-C in a future version.
1623 * The conditional-expression-as-lvalue extension has been deprecated
1624 for C and Objective-C. In particular, code like this:
1625 int a, b, c;
1626 (a ? b : c) = 2;
1628 will not be accepted for C and Objective-C in a future version.
1629 * The compound-expression-as-lvalue extension has been deprecated
1630 for C and Objective-C. In particular, code like this:
1631 int a, b;
1632 (a, b) = 2;
1634 will not be accepted for C and Objective-C in a future version. A
1635 possible non-intrusive workaround is the following:
1636 (*(a, &b)) = 2;
1638 * Several [10]built-in functions such as __builtin_popcount for
1639 counting bits, finding the highest and lowest bit in a word, and
1640 parity have been added.
1641 * The -fwritable-strings option has been deprecated and will be
1642 removed.
1643 * Many C math library functions are now recognized as built-ins and
1644 optimized.
1645 * The C, C++, and Objective-C compilers can now handle source files
1646 written in any character encoding supported by the host C library.
1647 The default input character set is taken from the current locale,
1648 and may be overridden with the -finput-charset command line
1649 option. In the future we will add support for inline encoding
1650 markers.
1652 C++
1654 * G++ is now much closer to full conformance to the ISO/ANSI C++
1655 standard. This means, among other things, that a lot of invalid
1656 constructs which used to be accepted in previous versions will now
1657 be rejected. It is very likely that existing C++ code will need to
1658 be fixed. This document lists some of the most common issues.
1659 * A hand-written recursive-descent C++ parser has replaced the
1660 YACC-derived C++ parser from previous GCC releases. The new parser
1661 contains much improved infrastructure needed for better parsing of
1662 C++ source codes, handling of extensions, and clean separation
1663 (where possible) between proper semantics analysis and parsing.
1664 The new parser fixes many bugs that were found in the old parser.
1665 * You must now use the typename and template keywords to
1666 disambiguate dependent names, as required by the C++ standard.
1667 struct K {
1668 typedef int mytype_t;
1669 };
1671 template <class T1> struct A {
1672 template <class T2> struct B {
1673 void callme(void);
1674 };
1676 template <int N> void bar(void)
1677 {
1678 // Use 'typename' to tell the parser that T1::mytype_t names
1679 // a type. This is needed because the name is dependent (in
1680 // this case, on template parameter T1).
1681 typename T1::mytype_t x;
1682 x = 0;
1683 }
1684 };
1686 template <class T> void template_func(void)
1687 {
1688 // Use 'template' to prefix member templates within
1689 // dependent types (a has type A<T>, which depends on
1690 // the template parameter T).
1691 A<T> a;
1692 a.template bar<0>();
1694 // Use 'template' to tell the parser that B is a nested
1695 // template class (dependent on template parameter T), and
1696 // 'typename' because the whole A<T>::B<int> is
1697 // the name of a type (again, dependent).
1698 typename A<T>::template B<int> b;
1699 b.callme();
1700 }
1702 void non_template_func(void)
1703 {
1704 // Outside of any template class or function, no names can be
1705 // dependent, so the use of the keyword 'typename' and 'template'
1706 // is not needed (and actually forbidden).
1707 A<K> a;
1708 a.bar<0>();
1709 A<K>::B<float> b;
1710 b.callme();
1711 }
1712 * In a template definition, unqualified names will no longer find
1713 members of a dependent base (as specified by [temp.dep]/3 in the
1714 C++ standard). For example,
1715 template <typename T> struct B {
1716 int m;
1717 int n;
1718 int f ();
1719 int g ();
1720 };
1721 int n;
1722 int g ();
1723 template <typename T> struct C : B<T> {
1724 void h ()
1725 {
1726 m = 0; // error
1727 f (); // error
1728 n = 0; // ::n is modified
1729 g (); // ::g is called
1730 }
1731 };
1732 You must make the names dependent, e.g. by prefixing them with
1733 this->. Here is the corrected definition of C<T>::h,
1734 template <typename T> void C<T>::h ()
1735 {
1736 this->m = 0;
1737 this->f ();
1738 this->n = 0
1739 this->g ();
1740 }
1741 As an alternative solution (unfortunately not backwards compatible
1742 with GCC 3.3), you may use using declarations instead of this->:
1743 template <typename T> struct C : B<T> {
1744 using B<T>::m;
1745 using B<T>::f;
1746 using B<T>::n;
1747 using B<T>::g;
1748 void h ()
1749 {
1750 m = 0;
1751 f ();
1752 n = 0;
1753 g ();
1754 }
1755 };
1756 * In templates, all non-dependent names are now looked up and bound
1757 at definition time (while parsing the code), instead of later when
1758 the template is instantiated. For instance:
1759 void foo(int);
1761 template <int> struct A {
1762 static void bar(void){
1763 foo('a');
1764 }
1765 };
1767 void foo(char);
1769 int main()
1770 {
1771 A<0>::bar(); // Calls foo(int), used to call foo(char).
1772 }
1773 * In an explicit instantiation of a class template, you must
1774 use class or struct before the template-id:
1775 template <int N>
1776 class A {};
1778 template A<0>; // error, not accepted anymore
1779 template class A<0>; // OK
1780 * The "named return value" and "implicit typename" extensions have
1781 been removed.
1782 * Default arguments in function types have been deprecated and will
1783 be removed.
1784 * ARM-style name-injection of friend declarations has been
1785 deprecated and will be removed. For example: struct S { friend
1786 void f(); }; void g() { f(); } will not be accepted by future
1787 versions of G++; instead a declaration of "f" will need to be
1788 present outside of the scope of "S".
1789 * Covariant returns are implemented for all but varadic functions
1790 that require an adjustment.
1791 * When -pedantic is used, G++ now issues errors about spurious
1792 semicolons. For example,
1793 namespace N {}; // Invalid semicolon.
1794 void f() {}; // Invalid semicolon.
1795 * G++ no longer accepts attributes for a declarator after the
1796 initializer associated with that declarator. For example,
1797 X x(1) __attribute__((...));
1798 is no longer accepted. Instead, use:
1799 X x __attribute__((...)) (1);
1800 * Inside the scope of a template class, the name of the class itself
1801 can be treated as either a class or a template. So GCC used to
1802 accept the class name as argument of type template, and template
1803 template parameter. However this is not C++ standard compliant.
1804 Now the name is not treated as a valid template template argument
1805 unless you qualify the name by its scope. For example, the code
1806 below no longer compiles.
1807 template <template <class> class TT> class X {};
1808 template <class T> class Y {
1809 X<Y> x; // Invalid, Y is always a type template parameter.
1810 };
1811 The valid code for the above example is
1812 X< ::Y> x; // Valid.
1813 (Notice the space between < and : to prevent GCC to interpret this
1814 as a digraph for [.)
1815 * Friend declarations that refer to template specializations are
1816 rejected if the template has not already been declared. For
1817 example,
1818 template <typename T>
1819 class C {
1820 friend void f<> (C&);
1821 };
1822 is rejected. You must first declare f as a template,
1823 template <typename T>
1824 void f(T);
1825 * In case of friend declarations, every name used in the friend
1826 declaration must be accessible at the point of that declaration.
1827 Previous versions of G++ used to be less strict about this and
1828 allowed friend declarations for private class members, for
1829 example. See the ISO C++ Standard Committee's [11]defect report
1830 #209 for details.
1831 * Declaration of member functions of class templates as friends are
1832 supported. For example,
1833 template <typename T> struct A {
1834 void f();
1835 };
1836 class C {
1837 template <typename T> friend void A<T>::f();
1838 };
1839 * You must use template <> to introduce template specializations, as
1840 required by the standard. For example,
1841 template <typename T>
1842 struct S;
1844 struct S<int> { };
1845 is rejected. You must write,
1846 template <> struct S<int> {};
1847 * G++ used to accept code like this,
1848 struct S {
1849 int h();
1850 void f(int i = g());
1851 int g(int i = h());
1852 };
1853 This behavior is not mandated by the standard. Now G++ issues an
1854 error about this code. To avoid the error, you must move the
1855 declaration of g before the declaration of f. The default
1856 arguments for g must be visible at the point where it is called.
1857 * The C++ ABI Section 3.3.3 specifications for the array
1858 construction routines __cxa_vec_new2 and __cxa_vec_new3 were
1859 changed to return NULL when the allocator argument returns NULL.
1860 These changes are incorporated into the libstdc++ runtime library.
1861 * Using a name introduced by a typedef in a friend declaration or in
1862 an explicit instantiation is now rejected, as specified by the ISO
1863 C++ standard.
1864 class A;
1865 typedef A B;
1866 class C {
1867 friend class B; // error, no typedef name here
1868 friend B; // error, friend always needs class/struct/enum
1869 friend class A; // OK
1870 };
1872 template <int> class Q {};
1873 typedef Q<0> R;
1874 template class R; // error, no typedef name here
1875 template class Q<0>; // OK
1876 * When allocating an array with a new expression, GCC used to allow
1877 parentheses around the type name. This is actually ill-formed and
1878 it is now rejected:
1879 int* a = new (int)[10]; // error, not accepted anymore
1880 int* a = new int[10]; // OK
1881 * When binding an rvalue of class type to a reference, the copy
1882 constructor of the class must be accessible. For instance,
1883 consider the following code:
1884 class A
1885 {
1886 public:
1887 A();
1889 private:
1890 A(const A&); // private copy ctor
1891 };
1893 A makeA(void);
1894 void foo(const A&);
1896 void bar(void)
1897 {
1898 foo(A()); // error, copy ctor is not accessible
1899 foo(makeA()); // error, copy ctor is not accessible
1901 A a1;
1902 foo(a1); // OK, a1 is a lvalue
1903 }
1904 This might be surprising at first sight, especially since most
1905 popular compilers do not correctly implement this rule
1906 ([12]further details).
1907 * When forming a pointer to member or a pointer to member function,
1908 access checks for class visibility (public, protected, private)
1909 are now performed using the qualifying scope of the name itself.
1910 This is better explained with an example:
1911 class A
1912 {
1913 public:
1914 void pub_func();
1915 protected:
1916 void prot_func();
1917 private:
1918 void priv_func();
1919 };
1921 class B : public A
1922 {
1923 public:
1924 void foo()
1925 {
1926 &A::pub_func; // OK, pub_func is accessible through A
1927 &A::prot_func; // error, cannot access prot_func through A
1928 &A::priv_func; // error, cannot access priv_func through A
1930 &B::pub_func; // OK, pub_func is accessible through B
1931 &B::prot_func; // OK, can access prot_func through B (within B)
1932 &B::priv_func; // error, cannot access priv_func through B
1933 }
1934 };
1936 Runtime Library (libstdc++)
1938 * Optimization work:
1939 + Streamlined streambuf, filebuf, separate synched with C
1940 Standard I/O streambuf.
1941 + All formatted I/O now uses cached locale information.
1942 + STL optimizations (memory/speed for list, red-black trees as
1943 used by sets and maps).
1944 + More use of GCC builtins.
1945 + String optimizations (avoid contention on
1946 increment/decrement-and-test of the reference count in the
1947 empty-string object, constructor from input_iterators
1948 speedup).
1949 * Static linkage size reductions.
1950 * Large File Support (files larger than 2 GB on 32-bit systems).
1951 * Wide character and variable encoding filebuf work (UTF-8,
1952 Unicode).
1953 * Generic character traits.
1954 * Also support wchar_t specializations on Mac OS 10.3.x, FreeBSD
1955 5.x, Solaris 2.7 and above, AIX 5.x, Irix 6.5.
1956 * The allocator class is now standard-conformant, and two additional
1957 extension allocators have been added, mt_alloc and
1958 bitmap_allocator.
1959 * PCH support: -include bits/stdc++.h (2x compile speedup).
1960 * Rewrote __cxa_demangle with support for C++ style allocators.
1961 * New debug modes for STL containers and iterators.
1962 * Testsuite rewrite: five times as many tests, plus increasingly
1963 sophisticated tests, including I/O, MT, multi-locale, wide and
1964 narrow characters.
1965 * Use current versions of GNU "autotools" for build/configuration.
1967 Objective-C
1969 * The Objective-C front end has been updated to include the numerous
1970 bug fixes and enhancements previously available only in Apple's
1971 version of GCC. These include:
1972 + Structured exception (@try... @catch... @finally, @throw) and
1973 synchronization (@synchronized) support. These are accessible
1974 via the -fobjc-exceptions switch; as of this writing, they
1975 may only be used in conjunction with -fnext-runtime on Mac OS
1976 X 10.3 and later. See [13]Options Controlling Objective-C
1977 Dialect for more information.
1978 + An overhaul of @encode logic. The C99 _Bool and C++ bool type
1979 may now be encoded as 'B'. In addition, the back-end/codegen
1980 dependencies have been removed.
1981 + An overhaul of message dispatch construction, ensuring that
1982 the various receiver types (and casts thereof) are handled
1983 properly, and that correct diagnostics are issued.
1984 + Support for "Zero-Link" (-fzero-link) and "Fix-and-Continue"
1985 (-freplace-objc-classes) debugging modes, currently available
1986 on Mac OS X 10.3 and later. See [14]Options Controlling
1987 Objective-C Dialect for more information.
1988 + Access to optimized runtime entry points (-fno-nil-receivers
1989 ) on the assumption that message receivers are never nil.
1990 This is currently available on Mac OS X 10.3 and later. See
1991 [15]Options Controlling Objective-C Dialect for more
1992 information.
1994 Java
1996 * Compiling a .jar file will now cause non-.class entries to be
1997 automatically compiled as resources.
1998 * libgcj has been ported to Darwin.
1999 * Jeff Sturm has adapted Jan Hubicka's call graph optimization code
2000 to gcj.
2001 * libgcj has a new gcjlib URL type; this lets URLClassLoader load
2002 code from shared libraries.
2003 * libgcj has been much more completely merged with [16]GNU
2004 Classpath.
2005 * Class loading is now much more correct; in particular the caller's
2006 class loader is now used when that is required.
2007 * [17]Eclipse 2.x will run out of the box using gij.
2008 * Parts of java.nio have been implemented. Direct and indirect
2009 buffers work, as do fundamental file and socket operations.
2010 * java.awt has been improved, though it is still not ready for
2011 general use.
2012 * The HTTP protocol handler now uses HTTP/1.1 and can handle the
2013 POST method.
2014 * The MinGW port has matured. Enhancements include socket timeout
2015 support, thread interruption, improved Runtime.exec() handling and
2016 support for accented characters in filenames.
2018 Fortran
2020 * Fortran improvements are listed in the [18]Fortran documentation.
2022 New Targets and Target Specific Improvements
2024 Alpha
2026 * Several [19]built-in functions have been added such as
2027 __builtin_alpha_zap to allow utilizing the more obscure
2028 instructions of the CPU.
2029 * Parameter passing of complex arguments has changed to match the
2030 [20]ABI. This change is incompatible with previous GCC versions,
2031 but does fix compatibility with the Tru64 compiler and several
2032 corner cases where GCC was incompatible with itself.
2034 ARM
2036 * Nicolas Pitre has contributed his hand-coded floating-point
2037 support code for ARM. It is both significantly smaller and faster
2038 than the existing C-based implementation, even when building
2039 applications for Thumb. The arm-elf configuration has been
2040 converted to use the new code.
2041 * Support for the Intel's iWMMXt architecture, a second generation
2042 XScale processor, has been added. Enabled at run time with the
2043 -mcpu=iwmmxt command line switch.
2044 * A new ARM target has been added: arm-wince-pe. This is similar to
2045 the arm-pe target, but it defaults to using the APCS32 ABI.
2046 * The existing ARM pipeline description has been converted to the
2047 use the [21]DFA processor pipeline model. There is not much change
2048 in code performance, but the description is now [22]easier to
2049 understand.
2050 * Support for the Cirrus EP9312 Maverick floating point co-processor
2051 added. Enabled at run time with the -mcpu=ep9312 command line
2052 switch. Note however that the multilibs to support this chip are
2053 currently disabled in gcc/config/arm/t-arm-elf, so if you want to
2054 enable their production you will have to uncomment the entries in
2055 that file.
2057 H8/300
2059 * Support for long long has been added.
2060 * Support for saveall attribute has been added.
2061 * Pavel Pisa contributed hand-written 32-bit-by-32-bit division code
2062 for H8/300H and H8S, which is much faster than the previous
2063 implementation.
2064 * A lot of small performance improvements.
2066 IA-32/AMD64 (x86-64)
2068 * Tuning for K8 (AMD Opteron/Athlon64) core is available via
2069 -march=k8 and -mcpu=k8.
2070 * Scalar SSE code generation carefully avoids reformatting
2071 penalties, hidden dependencies and minimizes the number of uops
2072 generated on both Intel and AMD CPUs.
2073 * Vector MMX and SSE operands are now passed in registers to improve
2074 performance and match the argument passing convention used by the
2075 Intel C++ Compiler. As a result it is not possible to call
2076 functions accepting vector arguments compiled by older GCC
2077 version.
2078 * Conditional jump elimination is now more aggressive on modern
2079 CPUs.
2080 * The Athlon ports has been converted to use the DFA processor
2081 pipeline description.
2082 * Optimization of indirect tail calls is now possible in a similar
2083 fashion as direct sibcall optimization.
2084 * Further small performance improvements.
2085 * -m128bit-long-double is now less buggy.
2086 * __float128 support in 64-bit compilation.
2087 * Support for data structures exceeding 2GB in 64-bit mode.
2088 * -mcpu has been renamed to -mtune.
2090 IA-64
2092 * Tuning code for the Itanium 2 processor has been added. The
2093 generation of code tuned for Itanium 2 (option -mtune=itanium2) is
2094 enabled by default now. To generate code tuned for Itanium 1 the
2095 option -mtune=itanium1 should be used.
2096 * [23]DFA processor pipeline descriptions for the IA-64 processors
2097 have been added. This resulted in about 3% improvement on the
2098 SPECInt2000 benchmark for Itanium 2.
2099 * Instruction bundling for the IA-64 processors has been rewritten
2100 using the DFA pipeline hazard recognizer. It resulted in about 60%
2101 compiler speedup on the SPECInt2000 C programs.
2103 M32R
2105 * Support for the M32R/2 processor has been added by Renesas.
2106 * Support for an M32R Linux target and PIC code generation has been
2107 added by Renesas.
2109 M68000
2111 * Bernardo Innocenti (Develer S.r.l.) has contributed the
2112 m68k-uclinux target, based on former work done by Paul Dale
2113 (SnapGear Inc.). Code generation for the ColdFire processors
2114 family has been enhanced and extended to support the MCF 53xx and
2115 MCF 54xx cores, integrating former work done by Peter Barada
2116 (Motorola).
2118 MIPS
2120 Processor-specific changes
2122 * Support for the RM7000 and RM9000 processors has been added. It
2123 can be selected using the -march compiler option and should work
2124 with any MIPS I (mips-*) or MIPS III (mips64-*) configuration.
2125 * Support for revision 2 of the MIPS32 ISA has been added. It can be
2126 selected with the command-line option -march=mips32r2.
2127 * There is a new option, -mfix-sb1, to work around certain SB-1
2128 errata.
2130 Configuration
2132 * It is possible to customize GCC using the following configure-time
2133 options:
2134 + --with-arch, which specifies the default value of the -march
2135 option.
2136 + --with-tune, which specifies the default value of the -mtune
2137 option.
2138 + --with-abi, which specifies the default ABI.
2139 + --with-float=soft, which tells GCC to use software floating
2140 point by default.
2141 + --with-float=hard, which tells GCC to use hardware floating
2142 point by default.
2143 * A 64-bit GNU/Linux port has been added. The associated
2144 configurations are mips64-linux-gnu and mips64el-linux-gnu.
2145 * The 32-bit GNU/Linux port now supports Java.
2146 * The IRIX 6 configuration now supports the o32 ABI and will build
2147 o32 multilibs by default. This support is compatible with both
2148 binutils and the SGI tools, but note that several features,
2149 including debugging information and DWARF2 exception handling, are
2150 only available when using the GNU assembler. Use of the GNU
2151 assembler and linker (version 2.15 or above) is strongly
2152 recommended.
2153 * The IRIX 6 configuration now supports 128-bit long doubles.
2154 * There are two new RTEMS-specific configurations, mips-rtems and
2155 mipsel-rtems.
2156 * There are two new *-elf configurations, mipsisa32r2-elf and
2157 mipsisa32r2el-elf.
2159 General
2161 * Several [24]ABI bugs have been fixed. Unfortunately, these changes
2162 will break binary compatibility with earlier releases.
2163 * GCC can now use explicit relocation operators when generating
2164 -mabicalls code. This behavior is controlled by -mexplicit-relocs
2165 and can have several performance benefits. For example:
2166 + It allows for more optimization of GOT accesses, including
2167 better scheduling and redundancy elimination.
2168 + It allows sibling calls to be implemented as jumps.
2169 + n32 and n64 leaf functions can use a call-clobbered global
2170 pointer instead of $28.
2171 + The code to set up $gp can be removed from functions that
2172 don't need it.
2173 * A new option, -mxgot, allows the GOT to be bigger than 64k. This
2174 option is equivalent to the assembler's -xgot option and should be
2175 used instead of -Wa,-xgot.
2176 * Frame pointer elimination is now supported when generating 64-bit
2177 MIPS16 code.
2178 * Inline block moves have been optimized to take more account of
2179 alignment information.
2180 * Many internal changes have been made to the MIPS port, mostly
2181 aimed at reducing the reliance on assembler macros.
2183 PowerPC
2185 * GCC 3.4 releases have a number of fixes for PowerPC and PowerPC64
2186 [25]ABI incompatibilities regarding the way parameters are passed
2187 during functions calls. These changes may result in
2188 incompatibility between code compiled with GCC 3.3 and GCC 3.4.
2190 PowerPC Darwin
2192 * Support for shared/dylib gcc libraries has been added. It is
2193 enabled by default on powerpc-apple-darwin7.0.0 and up.
2194 * Libgcj is enabled by default. On systems older than
2195 powerpc-apple-darwin7.0.0 you need to install [26]dlcompat.
2196 * 128-bit IBM extended precision format support added for long
2197 double.
2199 PowerPC64 GNU/Linux
2201 * By default, PowerPC64 GNU/Linux now uses natural alignment of
2202 structure elements. The old four byte alignment for double, with
2203 special rules for a struct starting with a double, can be chosen
2204 with -malign-power. This change may result in incompatibility
2205 between code compiled with GCC 3.3 and GCC 3.4.
2206 * -mabi=altivec is now the default rather than -mabi=no-altivec.
2207 * 128-bit IBM extended precision format support added for long
2208 double.
2210 S/390 and zSeries
2212 * New command-line options allow to specify the intended execution
2213 environment for generated code:
2214 + -mesa/-mzarch allows to specify whether to generate code
2215 running in ESA/390 mode or in z/Architecture mode (this is
2216 applicable to 31-bit code only).
2217 + -march allows to specify a minimum processor architecture
2218 level (g5, g6, z900, or z990).
2219 + -mtune allows to specify which processor to tune for.
2220 * It is possible to customize GCC using the following configure-time
2221 options:
2222 + --with-mode, which specifies whether to default to assuming
2223 ESA/390 or z/Architecture mode.
2224 + --with-arch, which specifies the default value of the -march
2225 option.
2226 + --with-tune, which specifies the default value of the -mtune
2227 option.
2228 * Support for the z990 processor has been added, and can be selected
2229 using -march=z990 or -mtune=z990. This includes instruction
2230 scheduling tuned for the superscalar instruction pipeline of the
2231 z990 processor as well as support for all new instructions
2232 provided by the long-displacement facility.
2233 * Support to generate 31-bit code optimized for zSeries processors
2234 (running in ESA/390 or in z/Architecture mode) has been added.
2235 This can be selected using -march=z900 and -mzarch respectively.
2236 * Instruction scheduling for the z900 and z990 processors now uses
2237 the DFA pipeline hazard recognizer.
2238 * GCC no longer generates code to maintain a stack backchain,
2239 previously used to generate stack backtraces for debugging
2240 purposes. As replacement that does not incur runtime overhead,
2241 DWARF-2 call frame information is provided by GCC; this is
2242 supported by GDB 6.1. The old behavior can be restored using the
2243 -mbackchain option.
2244 * The stack frame size of functions may now exceed 2 GB in 64-bit
2245 code.
2246 * A port for the 64-bit IBM TPF operating system has been added; the
2247 configuration is s390x-ibm-tpf. This configuration is supported as
2248 cross-compilation target only.
2249 * Various changes to improve the generated code have been
2250 implemented, including:
2251 + GCC now uses the MULTIPLY AND ADD and MULTIPLY AND SUBTRACT
2252 instructions to significantly speed up many floating-point
2253 applications.
2254 + GCC now uses the ADD LOGICAL WITH CARRY and SUBTRACT LOGICAL
2255 WITH BORROW instructions to speed up long long arithmetic.
2256 + GCC now uses the SEARCH STRING instruction to implement
2257 strlen().
2258 + In many cases, function call overhead for 31-bit code has
2259 been reduced by placing the literal pool after the function
2260 code instead of after the function prolog.
2261 + Register 14 is no longer reserved in 64-bit code.
2262 + Handling of global register variables has been improved.
2264 SPARC
2266 * The option -mflat is deprecated.
2267 * Support for large (> 2GB) frames has been added to the 64-bit
2268 port.
2269 * Several [27]ABI bugs have been fixed. Unfortunately, these changes
2270 will break binary compatibility with earlier releases.
2271 * The default debugging format has been switched from STABS to
2272 DWARF-2 for 32-bit code on Solaris 7 and later. DWARF-2 is already
2273 the default debugging format for 64-bit code on Solaris.
2275 SuperH
2277 * Support for the SH2E processor has been added. Enabled at run time
2278 with the -m2e command line switch, or at configure time by
2279 specifying sh2e as the machine part of the target triple.
2281 V850
2283 * Support for the Mitsubishi V850E1 processor has been added. This
2284 is a variant of the V850E processor with some additional debugging
2285 instructions.
2287 Xtensa
2289 * Several ABI bugs have been fixed. Unfortunately, these changes
2290 break binary compatibility with earlier releases.
2291 + For big-endian processors, the padding of aggregate return
2292 values larger than a word has changed. If the size of an
2293 aggregate return value is not a multiple of 32 bits, previous
2294 versions of GCC inserted padding in the most-significant
2295 bytes of the first return value register. Aggregates larger
2296 than a word are now padded in the least-significant bytes of
2297 the last return value register used. Aggregates smaller than
2298 a word are still padded in the most-significant bytes. The
2299 return value padding has not changed for little-endian
2300 processors.
2301 + Function arguments with 16-byte alignment are now properly
2302 aligned.
2303 + The implementation of the va_list type has changed. A va_list
2304 value created by va_start from a previous release cannot be
2305 used with va_arg from this release, or vice versa.
2306 * More processor configuration options for Xtensa processors are
2307 supported:
2308 + the ABS instruction is now optional;
2309 + the ADDX* and SUBX* instructions are now optional;
2310 + an experimental CONST16 instruction can be used to synthesize
2311 constants instead of loading them from constant pools.
2312 These and other Xtensa processor configuration options can no
2313 longer be enabled or disabled by command-line options; the
2314 processor configuration must be specified by the xtensa-config.h
2315 header file when building GCC. Additionally, the
2316 -mno-serialize-volatile option is no longer supported.
2318 Obsolete Systems
2320 Support for a number of older systems has been declared obsolete in
2321 GCC 3.4. Unless there is activity to revive them, the next release of
2322 GCC will have their sources permanently removed.
2324 All configurations of the following processor architectures have been
2325 declared obsolete:
2326 * Mitsubishi D30V, d30v-*
2327 * AT&T DSP1600 and DSP1610, dsp16xx-*
2328 * Intel 80960, i960
2330 Also, some individual systems have been obsoleted:
2331 * ARM Family
2332 + Support for generating code for operation in APCS/26 mode
2333 (-mapcs-26).
2334 * IBM ESA/390
2335 + "Bigfoot" port, i370-*. (The other port, s390-*, is actively
2336 maintained and supported.)
2337 * Intel 386 family
2338 + MOSS, i?86-moss-msdos and i?86-*-moss*
2339 + NCR 3000 running System V r.4, i?86-ncr-sysv4*
2340 + FreeBSD with a.out object format, i?86-*-freebsd*aout* and
2341 i?86-*-freebsd2*
2342 + Linux with a.out object format, i?86-linux*aout*
2343 + Linux with libc5, a.k.a. glibc1, i?86-linux*libc1*
2344 + Interix versions before Interix 3, i?86-*-interix
2345 + Mach microkernel, i?86-mach*
2346 + SCO UnixWare with UDK, i?86-*-udk*
2347 + Generic System V releases 1, 2, and 3, i?86-*-sysv[123]*
2348 + VSTa microkernel, i386-*-vsta
2349 * Motorola M68000 family
2350 + HPUX, m68k-hp-hpux* and m68000-hp-hpux*
2351 + NetBSD with a.out object format (before NetBSD 1.4),
2352 m68k-*-*-netbsd* except m68k-*-*-netbsdelf*
2353 + Generic System V r.4, m68k-*-sysv4*
2354 * VAX
2355 + Generic VAX, vax-*-* (This is generic VAX only; we have not
2356 obsoleted any VAX triples for specific operating systems.)
2358 Documentation improvements
2360 Other significant improvements
2362 * The build system has undergone several significant cleanups.
2363 Subdirectories will only be configured if they are being built,
2364 and all subdirectory configures are run from the make command. The
2365 top level has been autoconfiscated.
2366 * Building GCC no longer writes to its source directory. This should
2367 help those wishing to share a read-only source directory over NFS
2368 or build from a CD. The exceptions to this feature are if you
2369 configure with either --enable-maintainer-mode or
2370 --enable-generated-files-in-srcdir.
2371 * The -W warning option has been renamed to -Wextra, which is more
2372 easily understood. The older spelling will be retained for
2373 backwards compatibility.
2374 * Substantial improvements in compile time have been made,
2375 particularly for non-optimizing compilations.
2376 _________________________________________________________________
2378 GCC 3.4.0
2380 Bug Fixes
2382 A vast number of bugs have been fixed in 3.4.0, too many to publish a
2383 complete list here. [28]Follow this link to query the Bugzilla
2384 database for the list of over 900 bugs fixed in 3.4.0. This is the
2385 list of all bugs marked as resolved and fixed in 3.4.0 that are not
2386 flagged as 3.4 regressions.
2387 _________________________________________________________________
2389 GCC 3.4.1
2391 Bug Fixes
2393 This section lists the problem reports (PRs) from GCC's bug tracking
2394 system that are known to be fixed in the 3.4.1 release. This list
2395 might not be complete (that is, it is possible that some PRs that have
2396 been fixed are not listed here).
2398 Bootstrap failures
2400 * [29]10129 Ada bootstrap fails on PPC-Darwin - invalid assembler
2401 emitted - PIC related
2402 * [30]14576 [ARM] ICE in libiberty when building gcc-3.4 for arm-elf
2403 * [31]14760 A bug in configure.in prevents using both
2404 --program-suffix and --program-prefix
2405 * [32]14671 [hppa64] bootstrap fails: ICE in
2406 save_call_clobbered_regs, in caller_save.c
2407 * [33]15093 [alpha][Java] make bootstrap fails to configure libffi
2408 on Alpha
2409 * [34]15178 Solaris 9/x86 fails linking after stage 3
2411 Multi-platform internal compiler errors (ICEs)
2413 * [35]12753 (preprocessor) Memory corruption in preprocessor on bad
2414 input
2415 * [36]13985 ICE in gcc.c-torture/compile/930621-1.c
2416 * [37]14810 (c++) tree check failures with invalid code involving
2417 templates
2418 * [38]14883 (c++) ICE on invalid code, in cp_parser_lookup_name, in
2419 cp/parser.c
2420 * [39]15044 (c++) ICE on syntax error, template header
2421 * [40]15057 (c++) Compiling of conditional value throw constructs
2422 cause a segmentation violation
2423 * [41]15064 (c++) typeid of template parameter gives ICE
2424 * [42]15142 (c++) ICE when passing a string where a char* is
2425 expected in a throw statement
2426 * [43]15159 ICE in rtl_verify_flow_info_1
2427 * [44]15165 (c++) ICE in instantiate_template
2428 * [45]15193 Unary minus using pointer to V4SF vector causes
2429 -fforce-mem to exhaust all memory
2430 * [46]15209 (c++) Runs out of memory with packed structs
2431 * [47]15227 (c++) Trouble with invalid function definition
2432 * [48]15285 (c++) instantiate_type ICE when forming pointer to
2433 template function
2434 * [49]15299 (c++) ICE in resolve_overloaded_unification
2435 * [50]15329 (c++) ICE on constructor of member template
2436 * [51]15550 ICE in extract_insn, in recog.c
2437 * [52]15554 (c++) ICE in tsubst_copy, in cp/pt.c
2438 * [53]15640 (c++) ICE on invalid code in arg_assoc, in
2439 cp/name-lookup.c
2440 * [54]15666 [unit-at-a-time] Gcc abort on valid code
2441 * [55]15696 (c++) ICE with bad pointer-to-member code
2442 * [56]15701 (c++) ICE with friends and template template parameter
2443 * [57]15761 ICE in do_SUBST, in combine.c
2444 * [58]15829 (c++) ICE on Botan-1.3.13 due to -funroll-loops
2446 Ada
2448 * [59]14538 All RTEMS targets broken for gnat
2450 C front end
2452 * [60]12391 missing warning about assigning to an incomplete type
2453 * [61]14649 atan(1.0) should not be a constant expression
2454 * [62]15004 [unit-at-a-time] no warning for unused paramater in
2455 static function
2456 * [63]15749 --pedantic-errors behaves differently from --pedantic
2457 with C-compiler on GNU/Linux
2459 C++ compiler and library
2461 * [64]10646 non-const reference is incorrectly matched in a "const
2462 T" partial specialization
2463 * [65]12077 wcin.rdbuf()->in_avail() return value too high
2464 * [66]13598 enc_filebuf doesn't work
2465 * [67]14211 const_cast returns lvalue but should be rvalue
2466 * [68]14220 num_put::do_put() undesired float/double behavior
2467 * [69]14245 problem with user-defined allocators in
2468 std::basic_string
2469 * [70]14340 libstdc++ Debug mode: failure to convert iterator to
2470 const_iterator
2471 * [71]14600 __gnu_cxx::stdio_sync_filebuf should expose internal
2472 FILE*
2473 * [72]14668 no warning anymore for reevaluation of declaration
2474 * [73]14775 LFS (large file support) tests missing
2475 * [74]14821 Duplicate namespace alias declaration should not
2476 conflict
2477 * [75]14930 Friend declaration ignored
2478 * [76]14932 cannot use offsetof to get offsets of array elements in
2479 g++ 3.4.0
2480 * [77]14950 [non unit-at-a-time] always_inline does not mix with
2481 templates and -O0
2482 * [78]14962 g++ ignores #pragma redefine_extname
2483 * [79]14975 Segfault on low-level write error during imbue
2484 * [80]15002 Linewise stream input is unusably slow (std::string
2485 slow)
2486 * [81]15025 compiler accepts redeclaration of template as
2487 non-template
2488 * [82]15046 [arm] Math functions misdetected by cross configuration
2489 * [83]15069 a bit test on a variable of enum type is miscompiled
2490 * [84]15074 g++ -lsupc++ still links against libstdc++
2491 * [85]15083 spurious "statement has no effect" warning
2492 * [86]15096 parse error with templates and pointer to const member
2493 * [87]15287 combination of operator[] and operator .* fails in
2494 templates
2495 * [88]15317 __attribute__ unused in first parameter of constructor
2496 gives error
2497 * [89]15337 sizeof on incomplete type diagnostic
2498 * [90]15361 bitset<>::_Find_next fails
2499 * [91]15412 _GLIBCXX_ symbols symbols defined and used in different
2500 namespaces
2501 * [92]15427 valid code results in incomplete type error
2502 * [93]15471 Incorrect member pointer offsets in anonymous
2503 structs/unions
2504 * [94]15503 nested template problem
2505 * [95]15507 compiler hangs while laying out union
2506 * [96]15542 operator & and template definitions
2507 * [97]15565 SLES9: leading + sign for unsigned int with showpos
2508 * [98]15625 friend defined inside a template fails to find static
2509 function
2510 * [99]15629 Function templates, overloads, and friend name injection
2511 * [100]15742 'noreturn' attribute ignored in method of template
2512 functions.
2513 * [101]15775 Allocator::pointer consistently ignored
2514 * [102]15821 Duplicate namespace alias within namespace rejected
2515 * [103]15862 'enum yn' fails (confict with undeclared builtin)
2516 * [104]15875 rejects pointer to member in template
2517 * [105]15877 valid code using templates and anonymous enums is
2518 rejected
2519 * [106]15947 Puzzling error message for wrong destructor declaration
2520 in template class
2521 * [107]16020 cannot copy __gnu_debug::bitset
2522 * [108]16154 input iterator concept too restrictive
2523 * [109]16174 deducing top-level consts
2525 Java
2527 * [110]14315 Java compiler is not parallel make safe
2529 Fortran
2531 * [111]15151 [g77] incorrect logical i/o in 64-bit mode
2533 Objective-C
2535 * [112]7993 private variables cannot be shadowed in subclasses
2537 Optimization bugs
2539 * [113]15228 useless copies of floating point operands
2540 * [114]15345 [non-unit-at-a-time] unreferenced nested inline
2541 functions not optimized away
2542 * [115]15945 Incorrect floating point optimization
2543 * [116]15526 ftrapv aborts on 0 * (-1)
2544 * [117]14690 Miscompiled POOMA tests
2545 * [118]15112 GCC generates code to write to unchanging memory
2547 Preprocessor
2549 * [119]15067 Minor glitch in the source of cpp
2551 Main driver program bugs
2553 * [120]1963 collect2 interprets -oldstyle_liblookup as -o
2554 ldstyle_liblookup
2556 x86-specific (Intel/AMD)
2558 * [121]15717 Error: can't resolve `L0' {*ABS* section} - `xx' {*UND*
2559 section}
2561 HPPA-specific
2563 * [122]14782 GCC produces an unaligned data access at -O2
2564 * [123]14828 FAIL: gcc.c-torture/execute/20030408-1.c execution, -O2
2565 * [124]15202 ICE in reload_cse_simplify_operands, in postreload.c
2567 IA64-specific
2569 * [125]14610 __float80 constants incorrectly emitted
2570 * [126]14813 init_array sections are initialized in the wrong order
2571 * [127]14857 GCC segfault on duplicated asm statement
2572 * [128]15598 Gcc 3.4 ICE on valid code
2573 * [129]15653 Gcc 3.4 ICE on valid code
2575 MIPS-specific
2577 * [130]15189 wrong filling of delay slot with -march=mips1 -G0
2578 -mno-split-addresses -mno-explicit-relocs
2579 * [131]15331 Assembler error building gnatlib on IRIX 6.5 with GNU
2580 as 2.14.91
2581 * [132]16144 Bogus reference to __divdf3 when -O1
2582 * [133]16176 Miscompilation of unaligned data in MIPS backend
2584 PowerPC-specific
2586 * [134]11591 ICE in gcc.dg/altivec-5.c
2587 * [135]12028 powerpc-eabispe produces bad sCOND operation
2588 * [136]14478 rs6000 geu/ltu patterns generate incorrect code
2589 * [137]14567 long double and va_arg complex args
2590 * [138]14715 Altivec stack layout may overlap gpr save with stack
2591 temps
2592 * [139]14902 (libstdc++) Stream checking functions fail when
2593 -pthread option is used.
2594 * [140]14924 Compiler ICE on valid code
2595 * [141]14960 -maltivec affects vector return with -mabi=no-altivec
2596 * [142]15106 vector varargs failure passing from altivec to
2597 non-altivec code for -m32
2598 * [143]16026 ICE in function.c:4804, assign_parms, when -mpowerpc64
2599 & half-word operation
2600 * [144]15191 -maltivec -mabi=no-altivec results in mis-aligned lvx
2601 and stvx
2602 * [145]15662 Segmentation fault when an exception is thrown - even
2603 if try and catch are specified
2605 s390-specific
2607 * [146]15054 Bad code due to overlapping stack temporaries
2609 SPARC-specific
2611 * [147]15783 ICE with union assignment in 64-bit mode
2612 * [148]15626 GCC 3.4 emits "ld: warning: relocation error:
2613 R_SPARC_UA32"
2615 x86-64-specific
2617 * [149]14326 boehm-gc hardcodes to 3DNow! prefetch for x86_64
2618 * [150]14723 Backported -march=nocona from mainline
2619 * [151]15290 __float128 failed to pass to function properly
2621 Cygwin/Mingw32-specific
2623 * [152]15250 Option -mms-bitfields support on GCC 3.4 is not
2624 conformant to MS layout
2625 * [153]15551 -mtune=pentium4 -O2 with sjlj EH breaks stack probe
2626 worker on windows32 targets
2628 Bugs specific to embedded processors
2630 * [154]8309 [m68k] -m5200 produces erroneous SImode set of short
2631 varaible on stack
2632 * [155]13250 [SH] Gcc code for rotation clobbers the register, but
2633 gcc continues to use the register as if it was not clobbered
2634 * [156]13803 [coldfire] movqi operand constraints too restrictivefor
2635 TARGET_COLDFIRE
2636 * [157]14093 [SH] ICE for code when using -mhitachi option in SH
2637 * [158]14457 [m6811hc] ICE with simple c++ source
2638 * [159]14542 [m6811hc] ICE on simple source
2639 * [160]15100 [SH] cc1plus got hang-up on
2640 libstdc++-v3/testsuite/abi_check.cc
2641 * [161]15296 [CRIS] Delayed branch scheduling causing invalid code
2642 on cris-*
2643 * [162]15396 [SH] ICE with -O2 -fPIC
2644 * [163]15782 [coldfire] m68k_output_mi_thunk emits wrong code for
2645 ColdFire
2647 Testsuite problems (compiler not affected)
2649 * [164]11610 libstdc++ testcases 27_io/* don't work properly
2650 remotely
2651 * [165]15488 (libstdc++) possibly insufficient file permissions for
2652 executing test suite
2653 * [166]15489 (libstdc++) testsuite_files determined incorrectly
2655 Documentation bugs
2657 * [167]13928 (libstdc++) no whatis info in some man pages generated
2658 by doxygen
2659 * [168]14150 Ada documentation out of date
2660 * [169]14949 (c++) Need to document method visibility changes
2661 * [170]15123 libstdc++-doc: Allocators.3 manpage is empty
2662 _________________________________________________________________
2664 GCC 3.4.2
2666 Bug Fixes
2668 This section lists the problem reports (PRs) from GCC's bug tracking
2669 system that are known to be fixed in the 3.4.2 release. This list
2670 might not be complete (that is, it is possible that some PRs that have
2671 been fixed are not listed here).
2673 Bootstrap failures and issues
2675 * [171]16469 [mips-sgi-irix5.3] bootstrap fails in
2676 libstdc++-v3/testsuite
2677 * [172]16344 [hppa-linux-gnu] libstdc++'s PCH built by
2678 profiledbootstrap does not work with the built compiler
2679 * [173]16842 [Solaris/x86] mkheaders can not find mkheaders.conf
2681 Multi-platform internal compiler errors (ICEs)
2683 * [174]12608 (c++) ICE: expected class 't', have 'x' (error_mark) in
2684 cp_parser_class_specifier, in cp/parser.c
2685 * [175]14492 ICE in loc_descriptor_from_tree, in dwarf2out.c
2686 * [176]15461 (c++) ICE due to NRV and inlining
2687 * [177]15890 (c++) ICE in c_expand_expr, in c-common.c
2688 * [178]16180 ICE: segmentation fault in RTL optimization
2689 * [179]16224 (c++) ICE in write_unscoped_name (template/namespace)
2690 * [180]16408 ICE: in delete_insn, in cfgrtl.c
2691 * [181]16529 (c++) ICE for: namespace-alias shall not be declared as
2692 the name of any other entity
2693 * [182]16698 (c++) ICE with exceptions and declaration of
2694 __cxa_throw
2695 * [183]16706 (c++) ICE in finish_member_declaration, in
2696 cp/semantics.c
2697 * [184]16810 (c++) Legal C++ program with cast gives ICE in
2698 build_ptrmemfunc
2699 * [185]16851 (c++) ICE when throwing a comma expression
2700 * [186]16870 (c++) Boost.Spirit causes ICE in tsubst, in cp/pt.c
2701 * [187]16904 (c++) ICE in finish_class_member_access_expr, in
2702 cp/typeck.c
2703 * [188]16905 (c++) ICE (segfault) with exceptions
2704 * [189]16964 (c++) ICE in cp_parser_class_specifier due to
2705 redefinition
2706 * [190]17068 (c++) ICE: tree check: expected class 'd', have 'x'
2707 (identifier_node) in dependent_template_p, in cp/pt.c
2709 Preprocessor bugs
2711 * [191]16366 Preprocessor option -remap causes memory corruption
2713 Optimization
2715 * [192]15345 unreferenced nested inline functions not optimized away
2716 * [193]16590 Incorrect execution when compiling with -O2
2717 * [194]16693 Bitwise AND is lost when used within a cast to an enum
2718 of the same precision
2719 * [195]17078 Jump into if(0) substatement fails
2721 Problems in generated debug information
2723 * [196]13956 incorrect stabs for nested local variables
2725 C front end bugs
2727 * [197]16684 GCC should not warn about redundant redeclarations of
2728 built-ins
2730 C++ compiler and library
2732 * [198]12658 Thread safety problems in locale::global() and
2733 locale::locale()
2734 * [199]13092 g++ accepts invalid pointer-to-member conversion
2735 * [200]15320 Excessive memory consumption
2736 * [201]16246 Incorrect template argument deduction
2737 * [202]16273 Memory exhausted when using nested classes and virtual
2738 functions
2739 * [203]16401 ostringstream in gcc 3.4.x very slow for big data
2740 * [204]16411 undefined reference to
2741 __gnu_cxx::stdio_sync_filebuf<char, std::char_traits<char>
2742 >::file()
2743 * [205]16489 G++ incorrectly rejects use of a null constant integral
2744 expression as a null constant pointer
2745 * [206]16618 offsetof fails with constant member
2746 * [207]16637 syntax error reported for valid input code
2747 * [208]16717 __attribute__((constructor)) broken in C++
2748 * [209]16813 compiler error in DEBUG version of range insertion
2749 std::map::insert
2750 * [210]16853 pointer-to-member initialization from incompatible one
2751 accepted
2752 * [211]16889 ambiguity is not detected
2753 * [212]16959 Segmentation fault in ios_base::sync_with_stdio
2755 Java compiler and library
2757 * [213]7587 direct threaded interpreter not thread-safe
2758 * [214]16473 ServerSocket accept() leaks file descriptors
2759 * [215]16478 Hash synchronization deadlock with finalizers
2761 Alpha-specific
2763 * [216]10695 ICE in dwarf2out_frame_debug_expr, in dwarf2out.c
2764 * [217]16974 could not split insn (ice in final_scan_insn, in
2765 final.c)
2767 x86-specific
2769 * [218]16298 ICE in output_operand
2770 * [219]17113 ICE with SSE2 intrinsics
2772 x86-64 specific
2774 * [220]14697 libstdc++ couldn't find 32bit libgcc_s
2776 MIPS-specific
2778 * [221]15869 [mips64] No NOP after LW (with -mips1 -O0)
2779 * [222]16325 [mips64] value profiling clobbers gp on mips
2780 * [223]16357 [mipsisa64-elf] ICE copying 7 bytes between extern
2781 char[]s
2782 * [224]16380 [mips64] Use of uninitialised register after dbra
2783 conversion
2784 * [225]16407 [mips64] Unaligned access to local variables
2785 * [226]16643 [mips64] verify_local_live_at_start ICE after
2786 crossjumping & cfgcleanup
2788 ARM-specific
2790 * [227]15927 THUMB -O2: strength-reduced iteration variable ends up
2791 off by 1
2792 * [228]15948 THUMB: ICE with non-commutative cbranch
2793 * [229]17019 THUMB: bad switch statement in md code for
2794 addsi3_cbranch_scratch
2796 IA64-specific
2798 * [230]16130 ICE on valid code: in bundling, in config/ia64/ia64.c
2799 (-mtune=merced)
2800 * [231]16142 ICE on valid code: in bundling, in config/ia64/ia64.c
2801 (-mtune=itanium)
2802 * [232]16278 Gcc failed to build Linux kernel with -mtune=merced
2803 * [233]16414 ICE on valid code: typo in comparison of asm_noperands
2804 result
2805 * [234]16445 ICE on valid code: don't count ignored insns
2806 * [235]16490 ICE (segfault) while compiling with -fprofile-use
2807 * [236]16683 ia64 does not honor SUBTARGET_EXTRA_SPECS
2809 PowerPC-specific
2811 * [237]16195 (ppc64): Miscompilation of GCC 3.3.x by 3.4.x
2812 * [238]16239 ICE on ppc64 (mozilla 1.7 compile, -O1 -fno-exceptions
2813 issue)
2815 SPARC-specific
2817 * [239]16199 ICE while compiling apache 2.0.49
2818 * [240]16416 -m64 doesn't imply -mcpu=v9 anymore
2819 * [241]16430 ICE when returning non-C aggregates larger than 16
2820 bytes
2822 Bugs specific to embedded processors
2824 * [242]16379 [m32r] can't output large model function call of memcpy
2825 * [243]17093 [m32r] ICE with -msdata=use -O0
2826 * [244]17119 [m32r] ICE at switch case 0x8000
2828 DJGPP-specific
2830 * [245]15928 libstdc++ in 3.4.x doesn't cross-compile for djgpp
2832 Alpha Tru64-specific
2834 * [246]16210 libstdc++ gratuitously omits "long long" I/O
2836 Testsuite, documentation issues (compiler is not affected):
2838 * [247]15488 (libstdc++) possibly insufficient file permissions for
2839 executing test suite
2840 * [248]16250 ada/doctools runs makeinfo even in release tarball
2841 _________________________________________________________________
2843 GCC 3.4.3
2845 This is the [249]list of problem reports (PRs) from GCC's bug tracking
2846 system that are known to be fixed in the 3.4.3 release. This list
2847 might not be complete (that is, it is possible that some PRs that have
2848 been fixed are not listed here).
2850 Bootstrap failures
2852 * [250]17369 [ia64] Bootstrap failure with binutils-2.15.90.0.1.1
2853 * [251]17850 [arm-elf] bootstrap failure - libstdc++ uses strtold
2854 when undeclared
2856 Internal compiler errors (ICEs) affecting multiple platforms
2858 * [252]13948 (java) GCJ segmentation fault while compiling GL4Java
2859 .class files
2860 * [253]14492 ICE in loc_descriptor_from_tree, in dwarf2out.c
2861 * [254]16301 (c++) ICE when "strong" attribute is attached to a
2862 using directive
2863 * [255]16566 ICE with flexible arrays
2864 * [256]17023 ICE with nested functions in parameter declaration
2865 * [257]17027 ICE with noreturn function in loop at -O2
2866 * [258]17524 ICE in grokdeclarator, in cp/decl.c
2867 * [259]17826 (c++) ICE in cp_tree_equal
2869 C and optimization bugs
2871 * [260]15526 -ftrapv aborts on 0 * (-1)
2872 * [261]16999 #ident stopped working
2873 * [262]17503 quadratic behaviour in invalid_mode_change_p
2874 * [263]17581 Long long arithmetic fails inside a switch/case
2875 statement when compiled with -O2
2876 * [264]18129 -fwritable-strings doesn't work
2878 C++ compiler and library bugs
2880 * [265]10975 incorrect initial ostringstream::tellp()
2881 * [266]11722 Unbuffered filebuf::sgetn is slow
2882 * [267]14534 Unrecognizing static function as a template parameter
2883 when its return value is also templated
2884 * [268]15172 Copy constructor optimization in aggregate
2885 initialization
2886 * [269]15786 Bad error message for frequently occuring error.
2887 * [270]16162 Rejects valid member-template-definition
2888 * [271]16612 empty basic_strings can't live in shared memory
2889 * [272]16715 std::basic_iostream is instantiated when used, even
2890 though instantiations are already contained in libstdc++
2891 * [273]16848 code in /ext/demangle.h appears broken
2892 * [274]17132 GCC fails to eliminate function template specialization
2893 when argument deduction fails
2894 * [275]17259 One more _S_leaf incorrectly qualified with _RopeRep::
2895 in ropeimpl.h
2896 * [276]17327 use of `enumeral_type' in template type unification
2897 * [277]17393 "unused variable '._0'" warning with -Wall
2898 * [278]17501 Confusion with member templates
2899 * [279]17537 g++ not passing -lstdc++ to linker when all command
2900 line arguments are libraries
2901 * [280]17585 usage of unqualified name of static member from within
2902 class not allowed
2903 * [281]17821 Poor diagnostic for using "." instead of "->"
2904 * [282]17829 wrong error: call of overloaded function is ambiguous
2905 * [283]17851 Misleading diagnostic for invalid function declarations
2906 with undeclared types
2907 * [284]17976 Destructor is called twice
2908 * [285]18020 rejects valid definition of enum value in template
2909 * [286]18093 bogus conflict in namespace aliasing
2910 * [287]18140 C++ parser bug when using >> in templates
2912 Fortran
2914 * [288]17541 data statements with double precision constants fail
2916 x86-specific
2918 * [289]17853 -O2 ICE for MMX testcase
2920 SPARC-specific
2922 * [290]17245 ICE compiling gsl-1.5 statistics/lag1.c
2924 Darwin-specific
2926 * [291]17167 FATAL:Symbol L_foo$stub already defined.
2928 AIX-specific
2930 * [292]17277 could not catch an exception when specified -maix64
2932 Solaris-specific
2934 * [293]17505 <cmath> calls acosf(), ceilf(), and other functions
2935 missing from system libraries
2937 HP/UX specific:
2939 * [294]17684 /usr/ccs/bin/ld: Can't create libgcc_s.sl
2941 ARM-specific
2943 * [295]17384 ICE with mode attribute on structures
2945 MIPS-specific
2947 * [296]17770 No NOP after LWL with -mips1
2949 Other embedded target specific
2951 * [297]11476 [arc-elf] gcc ICE on newlib's vfprintf.c
2952 * [298]14064 [avr-elf] -fdata-sections triggers ICE
2953 * [299]14678 [m68hc11-elf] gcc ICE
2954 * [300]15583 [powerpc-rtems] powerpc-rtems lacks __USE_INIT_FINI__
2955 * [301]15790 [i686-coff] Alignment error building gcc with i686-coff
2956 target
2957 * [302]15886 [SH] Miscompilation with -O2 -fPIC
2958 * [303]16884 [avr-elf] [fweb related] bug while initializing
2959 variables
2961 Bugs relating to debugger support
2963 * [304]13841 missing debug info for _Complex function arguments
2964 * [305]15860 [big-endian targets] No DW_AT_location debug info is
2965 emitted for formal arguments to a function that uses "register"
2966 qualifiers
2968 Testsuite issues (compiler not affected)
2970 * [306]17465 Testsuite in libffi overrides LD_LIBRARY_PATH
2971 * [307]17469 Testsuite in libstdc++ overrides LD_LIBRARY_PATH
2972 * [308]18138 [mips-sgi-irix6.5] libgcc_s.so.1 not found by 64-bit
2973 testsuite
2975 Documentation
2977 * [309]15498 typo in gcc manual: non-existing locale example en_UK,
2978 should be en_GB
2979 * [310]15747 [mips-sgi-irix5.3] /bin/sh hangs during bootstrap:
2980 document broken shell
2981 * [311]16406 USE_LD_AS_NEEDED undocumented
2982 _________________________________________________________________
2984 GCC 3.4.4
2986 This is the [312]list of problem reports (PRs) from GCC's bug tracking
2987 system that are known to be fixed in the 3.4.4 release. This list
2988 might not be complete (that is, it is possible that some PRs that have
2989 been fixed are not listed here).
2990 _________________________________________________________________
2992 GCC 3.4.5
2994 This is the [313]list of problem reports (PRs) from GCC's bug tracking
2995 system that are known to be fixed in the 3.4.5 release. This list
2996 might not be complete (that is, it is possible that some PRs that have
2997 been fixed are not listed here).
2999 Bootstrap issues
3001 * [314]24688 sco_math fixincl breaks math.h
3003 C compiler bugs
3005 * [315]17188 struct Foo { } redefinition
3006 * [316]20187 wrong code for ((unsigned char)(unsigned long
3007 long)((a?a:1)&(a*b)))?0:1)
3008 * [317]21873 infinite warning loop on bad array initializer
3009 * [318]21899 enum definition accepts values to be overriden
3010 * [319]22061 ICE in find_function_data, in function.c
3011 * [320]22308 Failure to diagnose violation of constraint 6.516p2
3012 * [321]22458 ICE on missing brace
3013 * [322]22589 ICE casting to long long
3014 * [323]24101 Segfault with preprocessed source
3016 C++ compiler and library bugs
3018 * [324]10611 operations on vector mode not recognized in C++
3019 * [325]13377 unexpected behavior of namespace usage directive
3020 * [326]16002 Strange error message with new parser
3021 * [327]17413 local classes as template argument
3022 * [328]17609 spurious error message after using keyword
3023 * [329]17618 ICE in cp_convert_to_pointer, in cp/cvt.c
3024 * [330]18124 ICE with invalid template template parameter
3025 * [331]18155 typedef in template declaration not rejected
3026 * [332]18177 ICE with const_cast for undeclared variable
3027 * [333]18368 C++ error message regression
3028 * [334]16378 ICE when returning a copy of a packed member
3029 * [335]18466 int ::i; accepted
3030 * [336]18512 ICE on invalid usage of template base class
3031 * [337]18454 ICE when returning undefined type
3032 * [338]18738 typename not allowed with non-dependent qualified name
3033 * [339]18803 rejects access to operator() in template
3034 * [340]19004 ICE in uses_template_parms, in cp/pt.c
3035 * [341]19208 Spurious error about variably modified type
3036 * [342]18253 bad error message / ICE for invalid template parameter
3037 * [343]19608 ICE after friend function definition in local class
3038 * [344]19884 ICE on explicit instantiation of a non-template
3039 constructor
3040 * [345]20153 ICE when C++ template function contains anonymous union
3041 * [346]20563 Infinite loop in diagnostic (and ice after error
3042 message)
3043 * [347]20789 ICE with incomplete type in template
3044 * [348]21336 Internal compiler error when using custom new operators
3045 * [349]21768 ICE in error message due to violation of coding
3046 conventions
3047 * [350]21853 constness of pointer to data member ignored
3048 * [351]21903 Default argument of template function causes a
3049 compile-time error
3050 * [352]21983 multiple diagnostics
3051 * [353]21987 New testsuite failure
3052 g++.dg/warn/conversion-function-1.C
3053 * [354]22153 ICE on invalid template specialization
3054 * [355]22172 Internal compiler error, seg fault.
3055 * [356]21286 filebuf::xsgetn vs pipes
3056 * [357]22233 ICE with wrong number of template parameters
3057 * [358]22508 ICE after invalid operator new
3058 * [359]22545 ICE with pointer to class member & user defined
3059 conversion operator
3060 * [360]23528 Wrong default allocator in ext/hash_map
3061 * [361]23550 char_traits requirements/1.cc test bad math
3062 * [362]23586 Bad diagnostic for invalid namespace-name
3063 * [363]23624 ICE in invert_truthvalue, in fold-const.c
3064 * [364]23639 Bad error message: not a member of '<declaration
3065 error>'
3066 * [365]23797 ICE on typename outside template
3067 * [366]23965 Bogus error message: no matching function for call to
3068 'foo(<type error>)'
3069 * [367]24052 &#`label_decl' not supported by dump_expr#<expression
3070 error>
3071 * [368]24580 virtual base class cause exception not to be caught
3073 Problems in generated debug information
3075 * [369]24267 Bad DWARF for altivec vectors
3077 Optimizations issues
3079 * [370]17810 ICE in verify_local_live_at_start
3080 * [371]17860 Wrong generated code for loop with varying bound
3081 * [372]21709 ICE on compile-time complex NaN
3082 * [373]21964 broken tail call at -O2 or more
3083 * [374]22167 Strange optimization bug when using -Os
3084 * [375]22619 Compilation failure for real_const_1.f and
3085 real_const_2.f90
3086 * [376]23241 Invalid code generated for comparison of uchar to 255
3087 * [377]23478 Miscompilation due to reloading of a var that is also
3088 used in EH pad
3089 * [378]24470 segmentation fault in cc1plus when compiling with -O
3090 * [379]24950 ICE in operand_subword_force
3092 Precompiled headers problems
3094 * [380]14400 Cannot compile qt-x11-free-3.3.0
3095 * [381]14940 PCH largefile test fails on various platforms
3097 Preprocessor bugs
3099 * [382]20239 ICE on empty preprocessed input
3100 * [383]15220 "gcc -E -MM -MG" reports missing system headers in
3101 source directory
3103 Testsuite issues
3105 * [384]19275 gcc.dg/20020919-1.c fails with -fpic/-fPIC on
3106 i686-pc-linux-gnu
3108 Alpha specific
3110 * [385]21888 bootstrap failure with linker relaxation enabled
3112 ARM specific
3114 * [386]15342 [arm-linux]: ICE in verify_local_live_at_start
3115 * [387]23985 Memory aliasing information incorrect in inlined memcpy
3117 ColdFile specific
3119 * [388]16719 Illegal move of byte into address register causes
3120 compiler to ICE
3122 HPPA specific
3124 * [389]21723 ICE while building libgfortran
3125 * [390]21841 -mhp-ld/-mgnu-ld documentation
3127 IA-64 specific
3129 * [391]23644 IA-64 hardware models and configuration options
3130 documentation error
3131 * [392]24718 Shared libgcc not used for linking by default
3133 M68000 specific
3135 * [393]18421 ICE in reload_cse_simplify_operands, in postreload.c
3137 MIPS specific
3139 * [394]20621 ICE in change_address_1, in emit-rtl.c
3141 PowerPC and PowerPC64 specific
3143 * [395]18583 error on valid code: const
3144 __attribute__((altivec(vector__))) doesn't work in arrays
3145 * [396]20191 ICE in reload_cse_simplify_operands
3146 * [397]22083 AIX: TARGET_C99_FUNCTIONS is wrongly defined
3147 * [398]23070 CALL_V4_CLEAR_FP_ARGS flag not properly set
3148 * [399]23404 gij trashes args of functions with more than 8 fp args
3149 * [400]23539 C & C++ compiler generating misaligned references
3150 regardless of compiler flags
3151 * [401]24102 floatdisf2_internal2 broken
3152 * [402]24465 -mminimal-toc miscompilation of __thread vars
3154 Solaris specific
3156 * [403]19933 Problem with define of HUGE_VAL in math_c99
3157 * [404]21889 Native Solaris assembler cannot grok DTP-relative debug
3158 symbols
3160 SPARC specific
3162 * [405]19300 PCH failures on sparc-linux
3163 * [406]20301 Assembler labels have a leading "-"
3164 * [407]20673 C PCH testsuite assembly comparison failure
3166 x86 and x86_64 specific
3168 * [408]18582 ICE with arrays of type V2DF
3169 * [409]19340 Compilation SEGFAULTs with -O1 -fschedule-insns2
3170 -fsched2-use-traces
3171 * [410]21716 ICE in reg-stack.c's swap_rtx_condition
3172 * [411]24315 amd64 fails -fpeephole2
3173 _________________________________________________________________
3175 GCC 3.4.6
3177 This is the [412]list of problem reports (PRs) from GCC's bug tracking
3178 system that are known to be fixed in the 3.4.6 release. This list
3179 might not be complete (that is, it is possible that some PRs that have
3180 been fixed are not listed here).
3182 Please send FSF & GNU inquiries & questions to [413]gnu@gnu.org. There
3183 are also [414]other ways to contact the FSF.
3185 These pages are maintained by [415]the GCC team.
3188 For questions related to the use of GCC, please consult these web
3189 pages and the [416]GCC manuals. If that fails, the
3190 [417]gcc-help@gcc.gnu.org mailing list might help.
3191 Please send comments on these web pages and the development of GCC
3192 to our developer mailing list at [418]gcc@gnu.org or
3193 [419]gcc@gcc.gnu.org. All of our lists have [420]public archives.
3195 Copyright (C) Free Software Foundation, Inc., 51 Franklin St, Fifth
3196 Floor, Boston, MA 02110, USA.
3198 Verbatim copying and distribution of this entire article is permitted
3199 in any medium, provided this notice is preserved.
3200 Last modified 2006-06-21 [421]Valid XHTML 1.0
3202 References
3204 1. http://gcc.gnu.org/gcc-3.4/changes.html#3.4.6
3205 2. http://gcc.gnu.org/gcc-3.4/changes.html#cplusplus
3206 3. http://gcc.gnu.org/gcc-3.3/changes.html#obsolete_systems
3207 4. http://gcc.gnu.org/gcc-3.4/changes.html#obsolete_systems
3208 5. http://gcc.gnu.org/gcc-3.4/mips-abi.html
3209 6. http://gcc.gnu.org/gcc-3.4/sparc-abi.html
3210 7. http://www.boost.org/
3211 8. http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11953
3212 9. http://gcc.gnu.org/bugzilla/show_bug.cgi?id=8361
3213 10. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Other-Builtins.html#Other%20Builtins
3214 11. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_closed.html#209
3215 12. http://gcc.gnu.org/bugs.html#cxx_rvalbind
3216 13. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Objective-C-Dialect-Options.html
3217 14. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Objective-C-Dialect-Options.html
3218 15. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Objective-C-Dialect-Options.html
3219 16. http://www.gnu.org/software/classpath/
3220 17. http://www.eclipse.org/
3221 18. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/g77/News.html
3222 19. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gcc/Alpha-Built-in-Functions.html
3223 20. http://h30097.www3.hp.com/docs/base_doc/DOCUMENTATION/V51A_HTML/ARH9MBTE/DTMNPLTN.HTM#normal-argument-list-structure
3224 21. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gccint/Processor-pipeline-description.html
3225 22. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gccint/Comparison-of-the-two-descriptions.html
3226 23. http://gcc.gnu.org/onlinedocs/gcc-3.4.3/gccint/Processor-pipeline-description.html
3227 24. http://gcc.gnu.org/gcc-3.4/mips-abi.html
3228 25. http://gcc.gnu.org/gcc-3.4/powerpc-abi.html
3229 26. http://www.opendarwin.org/projects/dlcompat/
3230 27. http://gcc.gnu.org/gcc-3.4/sparc-abi.html
3231 28. 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
3232 29. http://gcc.gnu.org/PR10129
3233 30. http://gcc.gnu.org/PR14576
3234 31. http://gcc.gnu.org/PR14760
3235 32. http://gcc.gnu.org/PR14671
3236 33. http://gcc.gnu.org/PR15093
3237 34. http://gcc.gnu.org/PR15178
3238 35. http://gcc.gnu.org/PR12753
3239 36. http://gcc.gnu.org/PR13985
3240 37. http://gcc.gnu.org/PR14810
3241 38. http://gcc.gnu.org/PR14883
3242 39. http://gcc.gnu.org/PR15044
3243 40. http://gcc.gnu.org/PR15057
3244 41. http://gcc.gnu.org/PR15064
3245 42. http://gcc.gnu.org/PR15142
3246 43. http://gcc.gnu.org/PR15159
3247 44. http://gcc.gnu.org/PR15165
3248 45. http://gcc.gnu.org/PR15193
3249 46. http://gcc.gnu.org/PR15209
3250 47. http://gcc.gnu.org/PR15227
3251 48. http://gcc.gnu.org/PR15285
3252 49. http://gcc.gnu.org/PR15299
3253 50. http://gcc.gnu.org/PR15329
3254 51. http://gcc.gnu.org/PR15550
3255 52. http://gcc.gnu.org/PR15554
3256 53. http://gcc.gnu.org/PR15640
3257 54. http://gcc.gnu.org/PR15666
3258 55. http://gcc.gnu.org/PR15696
3259 56. http://gcc.gnu.org/PR15701
3260 57. http://gcc.gnu.org/PR15761
3261 58. http://gcc.gnu.org/PR15829
3262 59. http://gcc.gnu.org/PR14538
3263 60. http://gcc.gnu.org/PR12391
3264 61. http://gcc.gnu.org/PR14649
3265 62. http://gcc.gnu.org/PR15004
3266 63. http://gcc.gnu.org/PR15749
3267 64. http://gcc.gnu.org/PR10646
3268 65. http://gcc.gnu.org/PR12077
3269 66. http://gcc.gnu.org/PR13598
3270 67. http://gcc.gnu.org/PR14211
3271 68. http://gcc.gnu.org/PR14220
3272 69. http://gcc.gnu.org/PR14245
3273 70. http://gcc.gnu.org/PR14340
3274 71. http://gcc.gnu.org/PR14600
3275 72. http://gcc.gnu.org/PR14668
3276 73. http://gcc.gnu.org/PR14775
3277 74. http://gcc.gnu.org/PR14821
3278 75. http://gcc.gnu.org/PR14930
3279 76. http://gcc.gnu.org/PR14932
3280 77. http://gcc.gnu.org/PR14950
3281 78. http://gcc.gnu.org/PR14962
3282 79. http://gcc.gnu.org/PR14975
3283 80. http://gcc.gnu.org/PR15002
3284 81. http://gcc.gnu.org/PR15025
3285 82. http://gcc.gnu.org/PR15046
3286 83. http://gcc.gnu.org/PR15069
3287 84. http://gcc.gnu.org/PR15074
3288 85. http://gcc.gnu.org/PR15083
3289 86. http://gcc.gnu.org/PR15096
3290 87. http://gcc.gnu.org/PR15287
3291 88. http://gcc.gnu.org/PR15317
3292 89. http://gcc.gnu.org/PR15337
3293 90. http://gcc.gnu.org/PR15361
3294 91. http://gcc.gnu.org/PR15412
3295 92. http://gcc.gnu.org/PR15427
3296 93. http://gcc.gnu.org/PR15471
3297 94. http://gcc.gnu.org/PR15503
3298 95. http://gcc.gnu.org/PR15507
3299 96. http://gcc.gnu.org/PR15542
3300 97. http://gcc.gnu.org/PR15565
3301 98. http://gcc.gnu.org/PR15625
3302 99. http://gcc.gnu.org/PR15629
3303 100. http://gcc.gnu.org/PR15742
3304 101. http://gcc.gnu.org/PR15775
3305 102. http://gcc.gnu.org/PR15821
3306 103. http://gcc.gnu.org/PR15862
3307 104. http://gcc.gnu.org/PR15875
3308 105. http://gcc.gnu.org/PR15877
3309 106. http://gcc.gnu.org/PR15947
3310 107. http://gcc.gnu.org/PR16020
3311 108. http://gcc.gnu.org/PR16154
3312 109. http://gcc.gnu.org/PR16174
3313 110. http://gcc.gnu.org/PR14315
3314 111. http://gcc.gnu.org/PR15151
3315 112. http://gcc.gnu.org/PR7993
3316 113. http://gcc.gnu.org/PR15228
3317 114. http://gcc.gnu.org/PR15345
3318 115. http://gcc.gnu.org/PR15945
3319 116. http://gcc.gnu.org/PR15526
3320 117. http://gcc.gnu.org/PR14690
3321 118. http://gcc.gnu.org/PR15112
3322 119. http://gcc.gnu.org/PR15067
3323 120. http://gcc.gnu.org/PR1963
3324 121. http://gcc.gnu.org/PR15717
3325 122. http://gcc.gnu.org/PR14782
3326 123. http://gcc.gnu.org/PR14828
3327 124. http://gcc.gnu.org/PR15202
3328 125. http://gcc.gnu.org/PR14610
3329 126. http://gcc.gnu.org/PR14813
3330 127. http://gcc.gnu.org/PR14857
3331 128. http://gcc.gnu.org/PR15598
3332 129. http://gcc.gnu.org/PR15653
3333 130. http://gcc.gnu.org/PR15189
3334 131. http://gcc.gnu.org/PR15331
3335 132. http://gcc.gnu.org/PR16144
3336 133. http://gcc.gnu.org/PR16176
3337 134. http://gcc.gnu.org/PR11591
3338 135. http://gcc.gnu.org/PR12028
3339 136. http://gcc.gnu.org/PR14478
3340 137. http://gcc.gnu.org/PR14567
3341 138. http://gcc.gnu.org/PR14715
3342 139. http://gcc.gnu.org/PR14902
3343 140. http://gcc.gnu.org/PR14924
3344 141. http://gcc.gnu.org/PR14960
3345 142. http://gcc.gnu.org/PR15106
3346 143. http://gcc.gnu.org/PR16026
3347 144. http://gcc.gnu.org/PR15191
3348 145. http://gcc.gnu.org/PR15662
3349 146. http://gcc.gnu.org/PR15054
3350 147. http://gcc.gnu.org/PR15783
3351 148. http://gcc.gnu.org/PR15626
3352 149. http://gcc.gnu.org/PR14326
3353 150. http://gcc.gnu.org/PR14723
3354 151. http://gcc.gnu.org/PR15290
3355 152. http://gcc.gnu.org/PR15250
3356 153. http://gcc.gnu.org/PR15551
3357 154. http://gcc.gnu.org/PR8309
3358 155. http://gcc.gnu.org/PR13250
3359 156. http://gcc.gnu.org/PR13803
3360 157. http://gcc.gnu.org/PR14093
3361 158. http://gcc.gnu.org/PR14457
3362 159. http://gcc.gnu.org/PR14542
3363 160. http://gcc.gnu.org/PR15100
3364 161. http://gcc.gnu.org/PR15296
3365 162. http://gcc.gnu.org/PR15396
3366 163. http://gcc.gnu.org/PR15782
3367 164. http://gcc.gnu.org/PR11610
3368 165. http://gcc.gnu.org/PR15488
3369 166. http://gcc.gnu.org/PR15489
3370 167. http://gcc.gnu.org/PR13928
3371 168. http://gcc.gnu.org/PR14150
3372 169. http://gcc.gnu.org/PR14949
3373 170. http://gcc.gnu.org/PR15123
3374 171. http://gcc.gnu.org/PR16469
3375 172. http://gcc.gnu.org/PR16344
3376 173. http://gcc.gnu.org/PR16842
3377 174. http://gcc.gnu.org/PR12608
3378 175. http://gcc.gnu.org/PR14492
3379 176. http://gcc.gnu.org/PR15461
3380 177. http://gcc.gnu.org/PR15890
3381 178. http://gcc.gnu.org/PR16180
3382 179. http://gcc.gnu.org/PR16224
3383 180. http://gcc.gnu.org/PR16408
3384 181. http://gcc.gnu.org/PR16529
3385 182. http://gcc.gnu.org/PR16698
3386 183. http://gcc.gnu.org/PR16706
3387 184. http://gcc.gnu.org/PR16810
3388 185. http://gcc.gnu.org/PR16851
3389 186. http://gcc.gnu.org/PR16870
3390 187. http://gcc.gnu.org/PR16904
3391 188. http://gcc.gnu.org/PR16905
3392 189. http://gcc.gnu.org/PR16964
3393 190. http://gcc.gnu.org/PR17068
3394 191. http://gcc.gnu.org/PR16366
3395 192. http://gcc.gnu.org/PR15345
3396 193. http://gcc.gnu.org/PR16590
3397 194. http://gcc.gnu.org/PR16693
3398 195. http://gcc.gnu.org/PR17078
3399 196. http://gcc.gnu.org/PR13956
3400 197. http://gcc.gnu.org/PR16684
3401 198. http://gcc.gnu.org/PR12658
3402 199. http://gcc.gnu.org/PR13092
3403 200. http://gcc.gnu.org/PR15320
3404 201. http://gcc.gnu.org/PR16246
3405 202. http://gcc.gnu.org/PR16273
3406 203. http://gcc.gnu.org/PR16401
3407 204. http://gcc.gnu.org/PR16411
3408 205. http://gcc.gnu.org/PR16489
3409 206. http://gcc.gnu.org/PR16618
3410 207. http://gcc.gnu.org/PR16637
3411 208. http://gcc.gnu.org/PR16717
3412 209. http://gcc.gnu.org/PR16813
3413 210. http://gcc.gnu.org/PR16853
3414 211. http://gcc.gnu.org/PR16889
3415 212. http://gcc.gnu.org/PR16959
3416 213. http://gcc.gnu.org/PR7587
3417 214. http://gcc.gnu.org/PR16473
3418 215. http://gcc.gnu.org/PR16478
3419 216. http://gcc.gnu.org/PR10695
3420 217. http://gcc.gnu.org/PR16974
3421 218. http://gcc.gnu.org/PR16298
3422 219. http://gcc.gnu.org/PR17113
3423 220. http://gcc.gnu.org/PR14697
3424 221. http://gcc.gnu.org/PR15869
3425 222. http://gcc.gnu.org/PR16325
3426 223. http://gcc.gnu.org/PR16357
3427 224. http://gcc.gnu.org/PR16380
3428 225. http://gcc.gnu.org/PR16407
3429 226. http://gcc.gnu.org/PR16643
3430 227. http://gcc.gnu.org/PR15927
3431 228. http://gcc.gnu.org/PR15948
3432 229. http://gcc.gnu.org/PR17019
3433 230. http://gcc.gnu.org/PR16130
3434 231. http://gcc.gnu.org/PR16142
3435 232. http://gcc.gnu.org/PR16278
3436 233. http://gcc.gnu.org/PR16414
3437 234. http://gcc.gnu.org/PR16445
3438 235. http://gcc.gnu.org/PR16490
3439 236. http://gcc.gnu.org/PR16683
3440 237. http://gcc.gnu.org/PR16195
3441 238. http://gcc.gnu.org/PR16239
3442 239. http://gcc.gnu.org/PR16199
3443 240. http://gcc.gnu.org/PR16416
3444 241. http://gcc.gnu.org/PR16430
3445 242. http://gcc.gnu.org/PR16379
3446 243. http://gcc.gnu.org/PR17093
3447 244. http://gcc.gnu.org/PR17119
3448 245. http://gcc.gnu.org/PR15928
3449 246. http://gcc.gnu.org/PR16210
3450 247. http://gcc.gnu.org/PR15488
3451 248. http://gcc.gnu.org/PR16250
3452 249. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.4.3
3453 250. http://gcc.gnu.org/PR17369
3454 251. http://gcc.gnu.org/PR17850
3455 252. http://gcc.gnu.org/PR13948
3456 253. http://gcc.gnu.org/PR14492
3457 254. http://gcc.gnu.org/PR16301
3458 255. http://gcc.gnu.org/PR16566
3459 256. http://gcc.gnu.org/PR17023
3460 257. http://gcc.gnu.org/PR17027
3461 258. http://gcc.gnu.org/PR17524
3462 259. http://gcc.gnu.org/PR17826
3463 260. http://gcc.gnu.org/PR15526
3464 261. http://gcc.gnu.org/PR16999
3465 262. http://gcc.gnu.org/PR17503
3466 263. http://gcc.gnu.org/PR17581
3467 264. http://gcc.gnu.org/PR18129
3468 265. http://gcc.gnu.org/PR10975
3469 266. http://gcc.gnu.org/PR11722
3470 267. http://gcc.gnu.org/PR14534
3471 268. http://gcc.gnu.org/PR15172
3472 269. http://gcc.gnu.org/PR15786
3473 270. http://gcc.gnu.org/PR16162
3474 271. http://gcc.gnu.org/PR16612
3475 272. http://gcc.gnu.org/PR16715
3476 273. http://gcc.gnu.org/PR16848
3477 274. http://gcc.gnu.org/PR17132
3478 275. http://gcc.gnu.org/PR17259
3479 276. http://gcc.gnu.org/PR17327
3480 277. http://gcc.gnu.org/PR17393
3481 278. http://gcc.gnu.org/PR17501
3482 279. http://gcc.gnu.org/PR17537
3483 280. http://gcc.gnu.org/PR17585
3484 281. http://gcc.gnu.org/PR17821
3485 282. http://gcc.gnu.org/PR17829
3486 283. http://gcc.gnu.org/PR17851
3487 284. http://gcc.gnu.org/PR17976
3488 285. http://gcc.gnu.org/PR18020
3489 286. http://gcc.gnu.org/PR18093
3490 287. http://gcc.gnu.org/PR18140
3491 288. http://gcc.gnu.org/PR17541
3492 289. http://gcc.gnu.org/PR17853
3493 290. http://gcc.gnu.org/PR17245
3494 291. http://gcc.gnu.org/PR17167
3495 292. http://gcc.gnu.org/PR17277
3496 293. http://gcc.gnu.org/PR17505
3497 294. http://gcc.gnu.org/PR17684
3498 295. http://gcc.gnu.org/PR17384
3499 296. http://gcc.gnu.org/PR17770
3500 297. http://gcc.gnu.org/PR11476
3501 298. http://gcc.gnu.org/PR14064
3502 299. http://gcc.gnu.org/PR14678
3503 300. http://gcc.gnu.org/PR15583
3504 301. http://gcc.gnu.org/PR15790
3505 302. http://gcc.gnu.org/PR15886
3506 303. http://gcc.gnu.org/PR16884
3507 304. http://gcc.gnu.org/PR13841
3508 305. http://gcc.gnu.org/PR15860
3509 306. http://gcc.gnu.org/PR17465
3510 307. http://gcc.gnu.org/PR17469
3511 308. http://gcc.gnu.org/PR18138
3512 309. http://gcc.gnu.org/PR15498
3513 310. http://gcc.gnu.org/PR15747
3514 311. http://gcc.gnu.org/PR16406
3515 312. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.4.4
3516 313. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.4.5
3517 314. http://gcc.gnu.org/PR24688
3518 315. http://gcc.gnu.org/PR17188
3519 316. http://gcc.gnu.org/PR20187
3520 317. http://gcc.gnu.org/PR21873
3521 318. http://gcc.gnu.org/PR21899
3522 319. http://gcc.gnu.org/PR22061
3523 320. http://gcc.gnu.org/PR22208
3524 321. http://gcc.gnu.org/PR22458
3525 322. http://gcc.gnu.org/PR22589
3526 323. http://gcc.gnu.org/PR24101
3527 324. http://gcc.gnu.org/PR10611
3528 325. http://gcc.gnu.org/PR13377
3529 326. http://gcc.gnu.org/PR16002
3530 327. http://gcc.gnu.org/PR17413
3531 328. http://gcc.gnu.org/PR17609
3532 329. http://gcc.gnu.org/PR17618
3533 330. http://gcc.gnu.org/PR18124
3534 331. http://gcc.gnu.org/PR18155
3535 332. http://gcc.gnu.org/PR18177
3536 333. http://gcc.gnu.org/PR18368
3537 334. http://gcc.gnu.org/PR18378
3538 335. http://gcc.gnu.org/PR18466
3539 336. http://gcc.gnu.org/PR18512
3540 337. http://gcc.gnu.org/PR18545
3541 338. http://gcc.gnu.org/PR18738
3542 339. http://gcc.gnu.org/PR18803
3543 340. http://gcc.gnu.org/PR19004
3544 341. http://gcc.gnu.org/PR19208
3545 342. http://gcc.gnu.org/PR19253
3546 343. http://gcc.gnu.org/PR19608
3547 344. http://gcc.gnu.org/PR19884
3548 345. http://gcc.gnu.org/PR20153
3549 346. http://gcc.gnu.org/PR20563
3550 347. http://gcc.gnu.org/PR20789
3551 348. http://gcc.gnu.org/PR21336
3552 349. http://gcc.gnu.org/PR21768
3553 350. http://gcc.gnu.org/PR21853
3554 351. http://gcc.gnu.org/PR21903
3555 352. http://gcc.gnu.org/PR21983
3556 353. http://gcc.gnu.org/PR21987
3557 354. http://gcc.gnu.org/PR22153
3558 355. http://gcc.gnu.org/PR22172
3559 356. http://gcc.gnu.org/PR21286
3560 357. http://gcc.gnu.org/PR22233
3561 358. http://gcc.gnu.org/PR22508
3562 359. http://gcc.gnu.org/PR22545
3563 360. http://gcc.gnu.org/PR23528
3564 361. http://gcc.gnu.org/PR23550
3565 362. http://gcc.gnu.org/PR23586
3566 363. http://gcc.gnu.org/PR23624
3567 364. http://gcc.gnu.org/PR23639
3568 365. http://gcc.gnu.org/PR23797
3569 366. http://gcc.gnu.org/PR23965
3570 367. http://gcc.gnu.org/PR24052
3571 368. http://gcc.gnu.org/PR24580
3572 369. http://gcc.gnu.org/PR24267
3573 370. http://gcc.gnu.org/PR17810
3574 371. http://gcc.gnu.org/PR17860
3575 372. http://gcc/gnu.org/PR21709
3576 373. http://gcc.gnu.org/PR21964
3577 374. http://gcc.gnu.org/PR22167
3578 375. http://gcc.gnu.org/PR22619
3579 376. http://gcc.gnu.org/PR23241
3580 377. http://gcc.gnu.org/PR23478
3581 378. http://gcc.gnu.org/PR24470
3582 379. http://gcc.gnu.org/PR24950
3583 380. http://gcc.gnu.org/PR14400
3584 381. http://gcc.gnu.org/PR14940
3585 382. http://gcc.gnu.org/PR20239
3586 383. http://gcc.gnu.org/PR15220
3587 384. http://gcc.gnu.org/PR19275
3588 385. http://gcc.gnu.org/PR21888
3589 386. http://gcc.gnu.org/PR15342
3590 387. http://gcc.gnu.org/PR23985
3591 388. http://gcc.gnu.org/PR16719
3592 389. http://gcc.gnu.org/PR21723
3593 390. http://gcc.gnu.org/PR21841
3594 391. http://gcc.gnu.org/PR23644
3595 392. http://gcc.gnu.org/PR24718
3596 393. http://gcc.gnu.org/PR18421
3597 394. http://gcc.gnu.org/PR20621
3598 395. http://gcc.gnu.org/PR18583
3599 396. http://gcc.gnu.org/PR20191
3600 397. http://gcc.gnu.org/PR22083
3601 398. http://gcc.gnu.org/PR23070
3602 399. http://gcc.gnu.org/PR23404
3603 400. http://gcc.gnu.org/PR23539
3604 401. http://gcc.gnu.org/PR24102
3605 402. http://gcc.gnu.org/PR24465
3606 403. http://gcc.gnu.org/PR19933
3607 404. http://gcc.gnu.org/PR21889
3608 405. http://gcc.gnu.org/PR19300
3609 406. http://gcc.gnu.org/PR20301
3610 407. http://gcc.gnu.org/PR20673
3611 408. http://gcc.gnu.org/PR18582
3612 409. http://gcc.gnu.org/PR19340
3613 410. http://gcc.gnu.org/PR21716
3614 411. http://gcc.gnu.org/PR24315
3615 412. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.4.6
3616 413. mailto:gnu@gnu.org
3617 414. http://www.gnu.org/home.html#ContactInfo
3618 415. http://gcc.gnu.org/about.html
3619 416. http://gcc.gnu.org/onlinedocs/
3620 417. mailto:gcc-help@gcc.gnu.org
3621 418. mailto:gcc@gnu.org
3622 419. mailto:gcc@gcc.gnu.org
3623 420. http://gcc.gnu.org/lists.html
3624 421. http://validator.w3.org/check/referer
3625 ======================================================================
3626 http://gcc.gnu.org/gcc-3.3/index.html
3628 GCC 3.3 Release Series
3630 May 03, 2005
3632 The [1]GNU project and the GCC developers are pleased to announce the
3633 release of GCC 3.3.6.
3635 This release is a bug-fix release, containing fixes for regressions in
3636 GCC 3.3.5 relative to previous releases of GCC.
3638 This release is the last of the series 3.3.x.
3640 The GCC 3.3 release series includes numerous [2]new features,
3641 improvements, bug fixes, and other changes, thanks to an [3]amazing
3642 group of volunteers.
3644 Release History
3646 GCC 3.3.6
3647 May 3, 2005 ([4]changes)
3649 GCC 3.3.5
3650 September 30, 2004 ([5]changes)
3652 GCC 3.3.4
3653 May 31, 2004 ([6]changes)
3655 GCC 3.3.3
3656 February 14, 2004 ([7]changes)
3658 GCC 3.3.2
3659 October 16, 2003 ([8]changes)
3661 GCC 3.3.1
3662 August 8, 2003 ([9]changes)
3664 GCC 3.3
3665 May 14, 2003 ([10]changes)
3667 References and Acknowledgements
3669 GCC used to stand for the GNU C Compiler, but since the compiler
3670 supports several other languages aside from C, it now stands for the
3671 GNU Compiler Collection.
3673 A list of [11]successful builds is updated as new information becomes
3674 available.
3676 The GCC developers would like to thank the numerous people that have
3677 contributed new features, improvements, bug fixes, and other changes
3678 as well as test results to GCC. This [12]amazing group of volunteers
3679 is what makes GCC successful.
3681 For additional information about GCC please refer to the [13]GCC
3682 project web site or contact the [14]GCC development mailing list.
3684 To obtain GCC please use [15]our mirror sites, one of the [16]GNU
3685 mirror sites, or our CVS server.
3687 Please send FSF & GNU inquiries & questions to [17]gnu@gnu.org. There
3688 are also [18]other ways to contact the FSF.
3690 These pages are maintained by [19]the GCC team.
3693 For questions related to the use of GCC, please consult these web
3694 pages and the [20]GCC manuals. If that fails, the
3695 [21]gcc-help@gcc.gnu.org mailing list might help.
3696 Please send comments on these web pages and the development of GCC
3697 to our developer mailing list at [22]gcc@gnu.org or
3698 [23]gcc@gcc.gnu.org. All of our lists have [24]public archives.
3700 Copyright (C) Free Software Foundation, Inc., 51 Franklin St, Fifth
3701 Floor, Boston, MA 02110, USA.
3703 Verbatim copying and distribution of this entire article is permitted
3704 in any medium, provided this notice is preserved.
3705 Last modified 2006-06-21 [25]Valid XHTML 1.0
3707 References
3709 1. http://www.gnu.org/
3710 2. http://gcc.gnu.org/gcc-3.3/changes.html
3711 3. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
3712 4. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.6
3713 5. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.5
3714 6. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.4
3715 7. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.3
3716 8. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.2
3717 9. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.1
3718 10. http://gcc.gnu.org/gcc-3.3/changes.html
3719 11. http://gcc.gnu.org/gcc-3.3/buildstat.html
3720 12. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
3721 13. http://gcc.gnu.org/index.html
3722 14. mailto:gcc@gcc.gnu.org
3723 15. http://gcc.gnu.org/mirrors.html
3724 16. http://www.gnu.org/order/ftp.html
3725 17. mailto:gnu@gnu.org
3726 18. http://www.gnu.org/home.html#ContactInfo
3727 19. http://gcc.gnu.org/about.html
3728 20. http://gcc.gnu.org/onlinedocs/
3729 21. mailto:gcc-help@gcc.gnu.org
3730 22. mailto:gcc@gnu.org
3731 23. mailto:gcc@gcc.gnu.org
3732 24. http://gcc.gnu.org/lists.html
3733 25. http://validator.w3.org/check/referer
3734 ======================================================================
3735 http://gcc.gnu.org/gcc-3.3/changes.html
3737 GCC 3.3 Release Series
3738 Changes, New Features, and Fixes
3740 The latest release in the 3.3 release series is [1]GCC 3.3.6.
3742 Caveats
3744 * The preprocessor no longer accepts multi-line string literals.
3745 They were deprecated in 3.0, 3.1, and 3.2.
3746 * The preprocessor no longer supports the -A- switch when appearing
3747 alone. -A- followed by an assertion is still supported.
3748 * Support for all the systems [2]obsoleted in GCC 3.1 has been
3749 removed from GCC 3.3. See below for a [3]list of systems which are
3750 obsoleted in this release.
3751 * Checking for null format arguments has been decoupled from the
3752 rest of the format checking mechanism. Programs which use the
3753 format attribute may regain this functionality by using the new
3754 [4]nonnull function attribute. Note that all functions for which
3755 GCC has a built-in format attribute, an appropriate built-in
3756 nonnull attribute is also applied.
3757 * The DWARF (version 1) debugging format has been deprecated and
3758 will be removed in a future version of GCC. Version 2 of the DWARF
3759 debugging format will continue to be supported for the foreseeable
3760 future.
3761 * The C and Objective-C compilers no longer accept the "Naming
3762 Types" extension (typedef foo = bar); it was already unavailable
3763 in C++. Code which uses it will need to be changed to use the
3764 "typeof" extension instead: typedef typeof(bar) foo. (We have
3765 removed this extension without a period of deprecation because it
3766 has caused the compiler to crash since version 3.0 and no one
3767 noticed until very recently. Thus we conclude it is not in
3768 widespread use.)
3769 * The -traditional C compiler option has been removed. It was
3770 deprecated in 3.1 and 3.2. (Traditional preprocessing remains
3771 available.) The <varargs.h> header, used for writing variadic
3772 functions in traditional C, still exists but will produce an error
3773 message if used.
3774 * GCC 3.3.1 automatically places zero-initialized variables in the
3775 .bss section on some operating systems. Versions of GNU Emacs up
3776 to (and including) 21.3 will not work correctly when using this
3777 optimization; you can use -fno-zero-initialized-in-bss to disable
3778 it.
3780 General Optimizer Improvements
3782 * A new scheme for accurately describing processor pipelines, the
3783 [5]DFA scheduler, has been added.
3784 * Pavel Nejedly, Charles University Prague, has contributed new file
3785 format used by the edge coverage profiler (-fprofile-arcs).
3786 The new format is robust and diagnoses common mistakes where
3787 profiles from different versions (or compilations) of the program
3788 are combined resulting in nonsensical profiles and slow code to
3789 produced with profile feedback. Additionally this format allows
3790 extra data to be gathered. Currently, overall statistics are
3791 produced helping optimizers to identify hot spots of a program
3792 globally replacing the old intra-procedural scheme and resulting
3793 in better code. Note that the gcov tool from older GCC versions
3794 will not be able to parse the profiles generated by GCC 3.3 and
3795 vice versa.
3796 * Jan Hubicka, SuSE Labs, has contributed a new superblock formation
3797 pass enabled using -ftracer. This pass simplifies the control flow
3798 of functions allowing other optimizations to do better job.
3799 He also contributed the function reordering pass
3800 (-freorder-functions) to optimize function placement using profile
3801 feedback.
3803 New Languages and Language specific improvements
3805 C/ObjC/C++
3807 * The preprocessor now accepts directives within macro arguments. It
3808 processes them just as if they had not been within macro
3809 arguments.
3810 * The separate ISO and traditional preprocessors have been
3811 completely removed. The front end handles either type of
3812 preprocessed output if necessary.
3813 * In C99 mode preprocessor arithmetic is done in the precision of
3814 the target's intmax_t, as required by that standard.
3815 * The preprocessor can now copy comments inside macros to the output
3816 file when the macro is expanded. This feature, enabled using the
3817 -CC option, is intended for use by applications which place
3818 metadata or directives inside comments, such as lint.
3819 * The method of constructing the list of directories to be searched
3820 for header files has been revised. If a directory named by a -I
3821 option is a standard system include directory, the option is
3822 ignored to ensure that the default search order for system
3823 directories and the special treatment of system header files are
3824 not defeated.
3825 * A few more [6]ISO C99 features now work correctly.
3826 * A new function attribute, nonnull, has been added which allows
3827 pointer arguments to functions to be specified as requiring a
3828 non-null value. The compiler currently uses this information to
3829 issue a warning when it detects a null value passed in such an
3830 argument slot.
3831 * A new type attribute, may_alias, has been added. Accesses to
3832 objects with types with this attribute are not subjected to
3833 type-based alias analysis, but are instead assumed to be able to
3834 alias any other type of objects, just like the char type.
3836 C++
3838 * Type based alias analysis has been implemented for C++ aggregate
3839 types.
3841 Objective-C
3843 * Generate an error if Objective-C objects are passed by value in
3844 function and method calls.
3845 * When -Wselector is used, check the whole list of selectors at the
3846 end of compilation, and emit a warning if a @selector() is not
3847 known.
3848 * Define __NEXT_RUNTIME__ when compiling for the NeXT runtime.
3849 * No longer need to include objc/objc-class.h to compile self calls
3850 in class methods (NeXT runtime only).
3851 * New -Wundeclared-selector option.
3852 * Removed selector bloating which was causing object files to be 10%
3853 bigger on average (GNU runtime only).
3854 * Using at run time @protocol() objects has been fixed in certain
3855 situations (GNU runtime only).
3856 * Type checking has been fixed and improved in many situations
3857 involving protocols.
3859 Java
3861 * The java.sql and javax.sql packages now implement the JDBC 3.0
3862 (JDK 1.4) API.
3863 * The JDK 1.4 assert facility has been implemented.
3864 * The bytecode interpreter is now direct threaded and thus faster.
3866 Fortran
3868 * Fortran improvements are listed in [7]the Fortran documentation.
3870 Ada
3872 * Ada tasking now works with glibc 2.3.x threading libraries.
3874 New Targets and Target Specific Improvements
3876 * The following changes have been made to the HP-PA port:
3877 + The port now defaults to scheduling for the PA8000 series of
3878 processors.
3879 + Scheduling support for the PA7300 processor has been added.
3880 + The 32-bit port now supports weak symbols under HP-UX 11.
3881 + The handling of initializers and finalizers has been improved
3882 under HP-UX 11. The 64-bit port no longer uses collect2.
3883 + Dwarf2 EH support has been added to the 32-bit linux port.
3884 + ABI fixes to correct the passing of small structures by
3885 value.
3886 * The SPARC, HP-PA, SH4, and x86/pentium ports have been converted
3887 to use the DFA processor pipeline description.
3888 * The following NetBSD configurations for the SuperH processor
3889 family have been added:
3890 + SH3, big-endian, sh-*-netbsdelf*
3891 + SH3, little-endian, shle-*-netbsdelf*
3892 + SH5, SHmedia, big-endian, 32-bit default, sh5-*-netbsd*
3893 + SH5, SHmedia, little-endian, 32-bit default, sh5le-*-netbsd*
3894 + SH5, SHmedia, big-endian, 64-bit default, sh64-*-netbsd*
3895 + SH5, SHmedia, little-endian, 64-bit default, sh64le-*-netbsd*
3896 * The following changes have been made to the IA-32/x86-64 port:
3897 + SSE2 and 3dNOW! intrinsics are now supported.
3898 + Support for thread local storage has been added to the IA-32
3899 and x86-64 ports.
3900 + The x86-64 port has been significantly improved.
3901 * The following changes have been made to the MIPS port:
3902 + All configurations now accept the -mabi switch. Note that you
3903 will need appropriate multilibs for this option to work
3904 properly.
3905 + ELF configurations will always pass an ABI flag to the
3906 assembler, except when the MIPS EABI is selected.
3907 + -mabi=64 no longer selects MIPS IV code.
3908 + The -mcpu option, which was deprecated in 3.1 and 3.2, has
3909 been removed from this release.
3910 + -march now changes the core ISA level. In previous releases,
3911 it would change the use of processor-specific extensions, but
3912 would leave the core ISA unchanged. For example, mips64-elf
3913 -march=r8000 will now generate MIPS IV code.
3914 + Under most configurations, -mipsN now acts as a synonym for
3915 -march.
3916 + There are some new preprocessor macros to describe the -march
3917 and -mtune settings. See the documentation of those options
3918 for details.
3919 + Support for the NEC VR-Series processors has been added. This
3920 includes the 54xx, 5500, and 41xx series.
3921 + Support for the Sandcraft sr71k processor has been added.
3922 * The following changes have been made to the S/390 port:
3923 + Support to build the Java runtime libraries has been added.
3924 Java is now enabled by default on s390-*-linux* and
3925 s390x-*-linux* targets.
3926 + Multilib support for the s390x-*-linux* target has been
3927 added; this allows to build 31-bit binaries using the -m31
3928 option.
3929 + Support for thread local storage has been added.
3930 + Inline assembler code may now use the 'Q' constraint to
3931 specify memory operands without index register.
3932 + Various platform-specific performance improvements have been
3933 implemented; in particular, the compiler now uses the BRANCH
3934 ON COUNT family of instructions and makes more frequent use
3935 of the TEST UNDER MASK family of instructions.
3936 * The following changes have been made to the PowerPC port:
3937 + Support for IBM Power4 processor added.
3938 + Support for Motorola e500 SPE added.
3939 + Support for AIX 5.2 added.
3940 + Function and Data sections now supported on AIX.
3941 + Sibcall optimizations added.
3942 * The support for H8 Tiny is added to the H8/300 port with -mn.
3944 Obsolete Systems
3946 Support for a number of older systems has been declared obsolete in
3947 GCC 3.3. Unless there is activity to revive them, the next release of
3948 GCC will have their sources permanently removed.
3950 All configurations of the following processor architectures have been
3951 declared obsolete:
3952 * Matsushita MN10200, mn10200-*-*
3953 * Motorola 88000, m88k-*-*
3954 * IBM ROMP, romp-*-*
3956 Also, some individual systems have been obsoleted:
3957 * Alpha
3958 + Interix, alpha*-*-interix*
3959 + Linux libc1, alpha*-*-linux*libc1*
3960 + Linux ECOFF, alpha*-*-linux*ecoff*
3961 * ARM
3962 + Generic a.out, arm*-*-aout*
3963 + Conix, arm*-*-conix*
3964 + "Old ABI," arm*-*-oabi
3965 + StrongARM/COFF, strongarm-*-coff*
3966 * HPPA (PA-RISC)
3967 + Generic OSF, hppa1.0-*-osf*
3968 + Generic BSD, hppa1.0-*-bsd*
3969 + HP/UX versions 7, 8, and 9, hppa1.[01]-*-hpux[789]*
3970 + HiUX, hppa*-*-hiux*
3971 + Mach Lites, hppa*-*-lites*
3972 * Intel 386 family
3973 + Windows NT 3.x, i?86-*-win32
3974 * MC68000 family
3975 + HP systems, m68000-hp-bsd* and m68k-hp-bsd*
3976 + Sun systems, m68000-sun-sunos*, m68k-sun-sunos*, and
3977 m68k-sun-mach*
3978 + AT&T systems, m68000-att-sysv*
3979 + Atari systems, m68k-atari-sysv*
3980 + Motorola systems, m68k-motorola-sysv*
3981 + NCR systems, m68k-ncr-sysv*
3982 + Plexus systems, m68k-plexus-sysv*
3983 + Commodore systems, m68k-cbm-sysv*
3984 + Citicorp TTI, m68k-tti-*
3985 + Unos, m68k-crds-unos*
3986 + Concurrent RTU, m68k-ccur-rtu*
3987 + Linux a.out, m68k-*-linux*aout*
3988 + Linux libc1, m68k-*-linux*libc1*
3989 + pSOS, m68k-*-psos*
3990 * MIPS
3991 + Generic ECOFF, mips*-*-ecoff*
3992 + SINIX, mips-sni-sysv4
3993 + Orion RTEMS, mips64orion-*-rtems*
3994 * National Semiconductor 32000
3995 + OpenBSD, ns32k-*-openbsd*
3996 * POWER (aka RS/6000) and PowerPC
3997 + AIX versions 1, 2, and 3, rs6000-ibm-aix[123]*
3998 + Bull BOSX, rs6000-bull-bosx
3999 + Generic Mach, rs6000-*-mach*
4000 + Generic SysV, powerpc*-*-sysv*
4001 + Linux libc1, powerpc*-*-linux*libc1*
4002 * Sun SPARC
4003 + Generic a.out, sparc-*-aout*, sparclet-*-aout*,
4004 sparclite-*-aout*, and sparc86x-*-aout*
4005 + NetBSD a.out, sparc-*-netbsd*aout*
4006 + Generic BSD, sparc-*-bsd*
4007 + ChorusOS, sparc-*-chorusos*
4008 + Linux a.out, sparc-*-linux*aout*
4009 + Linux libc1, sparc-*-linux*libc1*
4010 + LynxOS, sparc-*-lynxos*
4011 + Solaris on HAL hardware, sparc-hal-solaris2*
4012 + SunOS versions 3 and 4, sparc-*-sunos[34]*
4013 * NEC V850
4014 + RTEMS, v850-*-rtems*
4015 * VAX
4016 + VMS, vax-*-vms*
4018 Documentation improvements
4020 Other significant improvements
4022 * Almost all front-end dependencies in the compiler have been
4023 separated out into a set of language hooks. This should make
4024 adding a new front end clearer and easier.
4025 * One effect of removing the separate preprocessor is a small
4026 increase in the robustness of the compiler in general, and the
4027 maintainability of target descriptions. Previously target-specific
4028 built-in macros and others, such as __FAST_MATH__, had to be
4029 handled with so-called specs that were hard to maintain. Often
4030 they would fail to behave properly when conflicting options were
4031 supplied on the command line, and define macros in the user's
4032 namespace even when strict ISO compliance was requested.
4033 Integrating the preprocessor has cleanly solved these issues.
4034 * The Makefile suite now supports redirection of make install by
4035 means of the variable DESTDIR.
4036 _________________________________________________________________
4038 GCC 3.3
4040 Detailed release notes for the GCC 3.3 release follow.
4042 Bug Fixes
4044 bootstrap failures
4046 * [8]10140 cross compiler build failures: missing __mempcpy (DUP:
4047 [9]10198,[10]10338)
4049 Internal compiler errors (multi-platform)
4051 * [11]3581 large string causes segmentation fault in cc1
4052 * [12]4382 __builtin_{set,long}jmp with -O3 can crash the compiler
4053 * [13]5533 (c++) ICE when processing std::accumulate(begin, end,
4054 init, invalid_op)
4055 * [14]6387 -fpic -gdwarf-2 -g1 combination gives ICE in dwarf2out
4056 * [15]6412 (c++) ICE in retrieve_specialization
4057 * [16]6620 (c++) partial template specialization causes an ICE
4058 (segmentation fault)
4059 * [17]6663 (c++) ICE with attribute aligned
4060 * [18]7068 ICE with incomplete types
4061 * [19]7083 (c++) ICE using -gstabs with dodgy class derivation
4062 * [20]7647 (c++) ICE when data member has the name of the enclosing
4063 class
4064 * [21]7675 ICE in fixup_var_refs_1
4065 * [22]7718 'complex' template instantiation causes ICE
4066 * [23]8116 (c++) ICE in member template function
4067 * [24]8358 (ada) Ada compiler accesses freed memory, crashes
4068 * [25]8511 (c++) ICE: (hopefully) reproducible cc1plus segmentation
4069 fault
4070 * [26]8564 (c++) ICE in find_function_data, in function.c
4071 * [27]8660 (c++) template overloading ICE in tsubst_expr, in cp/pt.c
4072 * [28]8766 (c++) ICE after failed initialization of static template
4073 variable
4074 * [29]8803 ICE in instantiate_virtual_regs_1, in function.c
4075 * [30]8846 (c++) ICE after diagnostic if fr_FR@euro locale is set
4076 * [31]8906 (c++) ICE (Segmentation fault) when parsing nested-class
4077 definition
4078 * [32]9216 (c++) ICE on missing template parameter
4079 * [33]9261 (c++) ICE in arg_assoc, in cp/decl2.c
4080 * [34]9263 (fortran) ICE caused by invalid PARAMETER in implied DO
4081 loop
4082 * [35]9429 (c++) ICE in template instantiation with a pointered new
4083 operator
4084 * [36]9516 Internal error when using a big array
4085 * [37]9600 (c++) ICE with typedefs in template class
4086 * [38]9629 (c++) virtual inheritance segfault
4087 * [39]9672 (c++) ICE: Error reporting routines re-entered
4088 * [40]9749 (c++) ICE in write_expression on invalid function
4089 prototype
4090 * [41]9794 (fortran) ICE: floating point exception during constant
4091 folding
4092 * [42]9829 (c++) Missing colon in nested namespace usage causes ICE
4093 * [43]9916 (c++) ICE with noreturn function in ?: statement
4094 * [44]9936 ICE with local function and variable-length 2d array
4095 * [45]10262 (c++) cc1plus crashes with large generated code
4096 * [46]10278 (c++) ICE in parser for invalid code
4097 * [47]10446 (c++) ICE on definition of nonexistent member function
4098 of nested class in a class template
4099 * [48]10451 (c++) ICE in grokdeclarator on spurious mutable
4100 declaration
4101 * [49]10506 (c++) ICE in build_new at cp/init.c with
4102 -fkeep-inline-functions and multiple inheritance
4103 * [50]10549 (c++) ICE in store_bit_field on bitfields that exceed
4104 the precision of the declared type
4106 Optimization bugs
4108 * [51]2001 Inordinately long compile times in reload CSE regs
4109 * [52]2391 Exponential compilation time explosion in combine
4110 * [53]2960 Duplicate loop conditions even with -Os
4111 * [54]4046 redundant conditional branch
4112 * [55]6405 Loop-unrolling related performance regressions
4113 * [56]6798 very long compile time with large case-statement
4114 * [57]6871 const objects shouldn't be moved to .bss
4115 * [58]6909 problem w/ -Os on modified loop-2c.c test case
4116 * [59]7189 gcc -O2 -Wall does not print ``control reaches end of
4117 non-void function'' warning
4118 * [60]7642 optimization problem with signbit()
4119 * [61]8634 incorrect code for inlining of memcpy under -O2
4120 * [62]8750 Cygwin prolog generation erroneously emitting __alloca as
4121 regular function call
4123 C front end
4125 * [63]2161 long if-else cascade overflows parser stack
4126 * [64]4319 short accepted on typedef'd char
4127 * [65]8602 incorrect line numbers in warning messages when using
4128 inline functions
4129 * [66]9177 -fdump-translation-unit: C front end deletes
4130 function_decl AST nodes and breaks debugging dumps
4131 * [67]9853 miscompilation of non-constant structure initializer
4133 c++ compiler and library
4135 * [68]45 legal template specialization code is rejected (DUP:
4136 [69]3784)
4137 * [70]764 lookup failure: friend operator and dereferencing a
4138 pointer and templates (DUP: [71]5116)
4139 * [72]2862 gcc accepts invalid explicit instantiation syntax (DUP:
4140 2863)
4141 * [73]3663 G++ doesn't check access control during template
4142 instantiation
4143 * [74]3797 gcc fails to emit explicit specialization of a template
4144 member
4145 * [75]3948 Two destructors are called when no copy destructor is
4146 defined (ABI change)
4147 * [76]4137 Conversion operator within template is not accepted
4148 * [77]4361 bogus ambiguity taking the address of a member template
4149 * [78]4802 g++ accepts illegal template code (access to private
4150 member; DUP: [79]5837)
4151 * [80]4803 inline function is used but never defined, and g++ does
4152 not object
4153 * [81]5094 Partial specialization cannot be friend?
4154 * [82]5730 complex<double>::norm() -- huge slowdown from
4155 egcs-2.91.66
4156 * [83]6713 Regression wrt 3.0.4: g++ -O2 leads to seg fault at run
4157 time
4158 * [84]7015 certain __asm__ constructs rejected
4159 * [85]7086 compile time regression (quadratic behavior in
4160 fixup_var_refs)
4161 * [86]7099 G++ doesn't set the noreturn attribute on std::exit and
4162 std::abort
4163 * [87]7247 copy constructor missing when inlining enabled (invalid
4164 optimization?)
4165 * [88]7441 string array initialization compilation time regression
4166 from seconds to minutes
4167 * [89]7768 __PRETTY_FUNCTION__ for template destructor is wrong
4168 * [90]7804 bad printing of floating point constant in warning
4169 message
4170 * [91]8099 Friend classes and template specializations
4171 * [92]8117 member function pointers and multiple inheritance
4172 * [93]8205 using declaration and multiple inheritance
4173 * [94]8645 unnecessary non-zero checks in stl_tree.h
4174 * [95]8724 explicit destructor call for incomplete class allowed
4175 * [96]8805 compile time regression with many member variables
4176 * [97]8691 -O3 and -fno-implicit-templates are incompatible
4177 * [98]8700 unhelpful error message for binding temp to reference
4178 * [99]8724 explicit destructor call for incomplete class allowed
4179 * [100]8949 numeric_limits<>::denorm_min() and is_iec559 problems
4180 * [101]9016 Failure to consistently constant fold "constant" C++
4181 objects
4182 * [102]9053 g++ confused about ambiguity of overloaded function
4183 templates
4184 * [103]9152 undefined virtual thunks
4185 * [104]9182 basic_filebuf<> does not report errors in codecvt<>::out
4186 * [105]9297 data corruption due to codegen bug (when copying.)
4187 * [106]9318 i/ostream::operator>>/<<(streambuf*) broken
4188 * [107]9320 Incorrect usage of traits_type::int_type in
4189 stdio_filebuf
4190 * [108]9400 bogus -Wshadow warning: shadowed declaration of this in
4191 local classes
4192 * [109]9424 i/ostream::operator>>/<<(streambuf*) drops characters
4193 * [110]9425 filebuf::pbackfail broken (DUP: [111]9439)
4194 * [112]9474 GCC freezes in compiling a weird code mixing <iostream>
4195 and <iostream.h>
4196 * [113]9548 Incorrect results from setf(ios::fixed) and
4197 precision(-1) [114][DR 231]
4198 * [115]9555 ostream inserters fail to set badbit on exception
4199 * [116]9561 ostream inserters rethrow exception of wrong type
4200 * [117]9563 ostream::sentry returns true after a failed preparation
4201 * [118]9582 one-definition rule violation in std::allocator
4202 * [119]9622 __PRETTY_FUNCTION__ incorrect in template destructors
4203 * [120]9683 bug in initialization chains for static const variables
4204 from template classes
4205 * [121]9791 -Woverloaded-virtual reports hiding of destructor
4206 * [122]9817 collate::compare doesn't handle nul characters
4207 * [123]9825 filebuf::sputbackc breaks sbumpc
4208 * [124]9826 operator>>(basic_istream, basic_string) fails to compile
4209 with custom traits
4210 * [125]9924 Multiple using statements for builtin functions not
4211 allowed
4212 * [126]9946 destructor is not called for temporary object
4213 * [127]9964 filebuf::close() sometimes fails to close file
4214 * [128]9988 filebuf::overflow writes EOF to file
4215 * [129]10033 optimization breaks polymorphic references w/ typeid
4216 operator
4217 * [130]10097 filebuf::underflow drops characters
4218 * [131]10132 filebuf destructor can throw exceptions
4219 * [132]10180 gcc fails to warn about non-inlined function
4220 * [133]10199 method parametrized by template does not work
4221 everywhere
4222 * [134]10300 use of array-new (nothrow) in segfaults on NULL return
4223 * [135]10427 Stack corruption with variable-length automatic arrays
4224 and virtual destructors
4225 * [136]10503 Compilation never stops in fixed_type_or_null
4227 Objective-C
4229 * [137]5956 selectors aren't matched properly when added to the
4230 selector table
4232 Fortran compiler and library
4234 * [138]1832 list directed i/o overflow hangs, -fbounds-check doesn't
4235 detect
4236 * [139]3924 g77 generates code that is rejected by GAS if COFF debug
4237 info requested
4238 * [140]5634 doc: explain that configure --prefix=~/... does not work
4239 * [141]6367 multiple repeat counts confuse namelist read into array
4240 * [142]6491 Logical operations error on logicals when using
4241 -fugly-logint
4242 * [143]6742 Generation of C++ Prototype for FORTRAN and extern "C"
4243 * [144]7113 Failure of g77.f-torture/execute/f90-intrinsic-bit.f -Os
4244 on irix6.5
4245 * [145]7236 OPEN(...,RECL=nnn,...) without ACCESS='DIRECT' should
4246 assume a direct access file
4247 * [146]7278 g77 "bug"; the executable misbehaves (with -O2
4248 -fno-automatic)
4249 * [147]7384 DATE_AND_TIME milliseconds field inactive on Windows
4250 * [148]7388 Incorrect output with 0-based array of characters
4251 * [149]8587 Double complex zero ** double precision number -> NaN
4252 instead of zero
4253 * [150]9038 -ffixed-line-length-none -x f77-cpp-input gives:
4254 Warning: unknown register name line-length-none
4255 * [151]10197 Direct access files not unformatted by default
4257 Java compiler and library
4259 * [152]6005 gcj fails to build rhug on alpha
4260 * [153]6389 System.getProperty("") should always throw an
4261 IllegalArgumentException
4262 * [154]6576 java.util.ResourceBundle.getResource ignores locale
4263 * [155]6652 new java.io.File("").getCanonicalFile() throws exception
4264 * [156]7060 getMethod() doesn't search super interface
4265 * [157]7073 bytecode interpreter gives wrong answer for interface
4266 getSuperclass()
4267 * [158]7180 possible bug in
4268 javax.naming.spi.NamingManager.getPlusPath()
4269 * [159]7416 java.security startup refs "GNU libgcj.security"
4270 * [160]7570 Runtime.exec with null envp: child doesn't inherit
4271 parent env (DUP: [161]7578)
4272 * [162]7611 Internal error while compiling libjava with -O
4273 * [163]7709 NullPointerException in _Jv_ResolvePoolEntry
4274 * [164]7766 ZipInputStream.available returns 0 immediately after
4275 construction
4276 * [165]7785 Calendar.getTimeInMillis/setTimeInMillis should be
4277 public
4278 * [166]7786 TimeZone.getDSTSavings() from JDK1.4 not implemented
4279 * [167]8142 '$' in class names vs. dlopen 'dynamic string tokens'
4280 * [168]8234 ZipInputStream chokes when InputStream.read() returns
4281 small chunks
4282 * [169]8415 reflection bug: exception info for Method
4283 * [170]8481 java.Random.nextInt(int) may return negative
4284 * [171]8593 Error reading GZIPped files with BufferedReader
4285 * [172]8759 java.beans.Introspector has no flushCaches() or
4286 flushFromCaches() methods
4287 * [173]8997 spin() calls Thread.sleep
4288 * [174]9253 on win32, java.io.File.listFiles("C:\\") returns pwd
4289 instead of the root content of C:
4290 * [175]9254 java::lang::Object::wait(), threads-win32.cc returns
4291 wrong return codes
4292 * [176]9271 Severe bias in java.security.SecureRandom
4294 Ada compiler and library
4296 * [177]6767 make gnatlib-shared fails on -laddr2line
4297 * [178]9911 gnatmake fails to link when GCC configured with
4298 --with-sjlj-exceptions=yes
4299 * [179]10020 Can't bootstrap gcc on AIX with Ada enabled
4300 * [180]10546 Ada tasking not working on Red Hat 9
4302 preprocessor
4304 * [181]7029 preprocessor should ignore #warning with -M
4306 ARM-specific
4308 * [182]2903 [arm] Optimization bug with long long arithmetic
4309 * [183]7873 arm-linux-gcc fails when assigning address to a bit
4310 field
4312 FreeBSD-specific
4314 * [184]7680 float functions undefined in math.h/cmath with #define
4315 _XOPEN_SOURCE
4317 HP-UX or HP-PA-specific
4319 * [185]8705 [HP-PA] ICE in emit_move_insn_1, in expr.c
4320 * [186]9986 [HP-UX] Incorrect transformation of fputs_unlocked to
4321 fputc_unlocked
4322 * [187]10056 [HP-PA] ICE at -O2 when building c++ code from doxygen
4324 m68hc11-specific
4326 * [188]6744 Bad assembler code generated: reference to pseudo
4327 register z
4328 * [189]7361 Internal compiler error in reload_cse_simplify_operands,
4329 in reload1.c
4331 MIPS-specific
4333 * [190]9496 [mips-linux] bug in optimizer?
4335 PowerPC-specific
4337 * [191]7067 -Os with -mcpu=powerpc optimizes for speed (?) instead
4338 of space
4339 * [192]8480 reload ICEs for LAPACK code on powerpc64-linux
4340 * [193]8784 [AIX] Internal compiler error in simplify_gen_subreg
4341 * [194]10315 [powerpc] ICE: in extract_insn, in recog.c
4343 SPARC-specific
4345 * [195]10267 (documentation) Wrong build instructions for
4346 *-*-solaris2*
4348 x86-specific (Intel/AMD)
4350 * [196]7916 ICE in instantiate_virtual_register_1
4351 * [197]7926 (c++) i486 instructions in header files make c++
4352 programs crash on i386
4353 * [198]8555 ICE in gen_split_1231
4354 * [199]8994 ICE with -O -march=pentium4
4355 * [200]9426 ICE with -fssa -funroll-loops -fprofile-arcs
4356 * [201]9806 ICE in inline assembly with -fPIC flag
4357 * [202]10077 gcc -msse2 generates movd to move dwords between xmm
4358 regs
4359 * [203]10233 64-bit comparison only comparing bottom 32-bits
4360 * [204]10286 type-punning doesn't work with __m64 and -O
4361 * [205]10308 [x86] ICE with -O -fgcse or -O2
4362 _________________________________________________________________
4364 GCC 3.3.1
4366 Bug Fixes
4368 This section lists the problem reports (PRs) from GCC's bug tracking
4369 system that are known to be fixed in the 3.3.1 release. This list
4370 might not be complete (that is, it is possible that some PRs that have
4371 been fixed are not listed here).
4373 Bootstrap failures
4375 * [206]11272 [Solaris] make bootstrap fails while building libstdc++
4377 Internal compiler errors (multi-platform)
4379 * [207]5754 ICE on invalid nested template class
4380 * [208]6597 ICE in set_mem_alias_set compiling Qt with -O2 on ia64
4381 and --enable-checking
4382 * [209]6949 (c++) ICE in tsubst_decl, in cp/pt.c
4383 * [210]7053 (c++) ICE when declaring a function already defined as a
4384 friend method of a template class
4385 * [211]8164 (c++) ICE when using different const expressions as
4386 template parameter
4387 * [212]8384 (c++) ICE in is_base_type, in dwarf2out.c
4388 * [213]9559 (c++) ICE with invalid initialization of a static const
4389 * [214]9649 (c++) ICE in finish_member_declaration, in
4390 cp/semantics.c when redeclaring a static member variable
4391 * [215]9864 (fortran) ICE in add_abstract_origin_attribute, in
4392 dwarfout.c with -g -O -finline-functions
4393 * [216]10432 (c++) ICE in poplevel, in cp/decl.c
4394 * [217]10475 ICE in subreg_highpart_offset for code with long long
4395 * [218]10635 (c++) ICE when dereferencing an incomplete type casted
4396 from a void pointer
4397 * [219]10661 (c++) ICE in instantiate_decl, in cp/pt.c while
4398 instantiating static member variables
4399 * [220]10700 ICE in copy_to_mode_reg on 64-bit targets
4400 * [221]10712 (c++) ICE in constructor_name_full, in cp/decl2.c
4401 * [222]10796 (c++) ICE when defining an enum with two values: -1 and
4402 MAX_INT_64BIT
4403 * [223]10890 ICE in merge_assigned_reloads building Linux 2.4.2x
4404 sched.c
4405 * [224]10939 (c++) ICE with template code
4406 * [225]10956 (c++) ICE when specializing a template member function
4407 of a template class, in tsubst, in cp/pt.c
4408 * [226]11041 (c++) ICE: const myclass &x = *x; (when operator*()
4409 defined)
4410 * [227]11059 (c++) ICE with empty union
4411 * [228]11083 (c++) ICE in commit_one_edge_insertion, in cfgrtl.c
4412 with -O2 -fnon-call-exceptions
4413 * [229]11105 (c++) ICE in mangle_conv_op_name_for_type
4414 * [230]11149 (c++) ICE on error when instantiation with call
4415 function of a base type
4416 * [231]11228 (c++) ICE on new-expression using array operator new
4417 and default-initialization
4418 * [232]11282 (c++) Infinite memory usage after syntax error
4419 * [233]11301 (fortran) ICE with -fno-globals
4420 * [234]11308 (c++) ICE when using an enum type name as if it were a
4421 class or namespace
4422 * [235]11473 (c++) ICE with -gstabs when empty struct inherits from
4423 an empty struct
4424 * [236]11503 (c++) ICE when instantiating template with ADDR_EXPR
4425 * [237]11513 (c++) ICE in push_template_decl_real, in cp/pt.c:
4426 template member functions
4428 Optimization bugs
4430 * [238]11198 -O2 -frename-registers generates wrong code (aliasing
4431 problem)
4432 * [239]11304 Wrong code production with -fomit-frame-pointer
4433 * [240]11381 volatile memory access optimized away
4434 * [241]11536 [strength-reduce] -O2 optimization produces wrong code
4435 * [242]11557 constant folding bug generates wrong code
4437 C front end
4439 * [243]5897 No warning for statement after return
4440 * [244]11279 DWARF-2 output mishandles large enums
4442 Preprocessor bugs
4444 * [245]11022 no warning for non-compatible macro redefinition
4446 C++ compiler and library
4448 * [246]2330 static_cast<>() to a private base is allowed
4449 * [247]5388 Incorrect message "operands to ?: have different types"
4450 * [248]5390 Libiberty fails to demangle multi-digit template
4451 parameters
4452 * [249]7877 Incorrect parameter passing to specializations of member
4453 function templates
4454 * [250]9393 Anonymous namespaces and compiling the same file twice
4455 * [251]10032 -pedantic converts some errors to warnings
4456 * [252]10468 const typeof(x) is non-const, but only in templates
4457 * [253]10527 confused error message with "new int()" parameter
4458 initializer
4459 * [254]10679 parameter MIN_INLINE_INSNS is not honored
4460 * [255]10682 gcc chokes on a typedef for an enum inside a class
4461 template
4462 * [256]10689 pow(std::complex(0),1/3) returns (nan, nan) instead of
4463 0.
4464 * [257]10845 template member function (with nested template as
4465 parameter) cannot be called anymore if another unrelated template
4466 member function is defined
4467 * [258]10849 Cannot define an out-of-class specialization of a
4468 private nested template class
4469 * [259]10888 Suppress -Winline warnings for system headers
4470 * [260]10929 -Winline warns about functions for which no definition
4471 is visible
4472 * [261]10931 valid conversion static_cast<const unsigned
4473 int&>(lvalue-of-type-int) is rejected
4474 * [262]10940 Bad code with explicit specialization
4475 * [263]10968 If member function implicitly instantiated, explicit
4476 instantiation of class fails to instantiate it
4477 * [264]10990 Cannot convert with dynamic_cast<> to a private base
4478 class from within a member function
4479 * [265]11039 Bad interaction between implicit typename deprecation
4480 and friendship
4481 * [266]11062 (libstdc++) avoid __attribute__ ((unused)); say
4482 "__unused__" instead
4483 * [267]11095 C++ iostream manipulator causes segfault when called
4484 with negative argument
4485 * [268]11098 g++ doesn't emit complete debugging information for
4486 local variables in destructors
4487 * [269]11137 Linux shared library constructors not called unless
4488 there's one global object
4489 * [270]11154 spurious ambiguity report for template class
4490 specialization
4491 * [271]11329 Compiler cannot find user defined implicit typecast
4492 * [272]11332 Spurious error with casts in ?: expression
4493 * [273]11431 static_cast behavior with subclasses when default
4494 constructor available
4495 * [274]11528 money_get facet does not accept "$.00" as valid
4496 * [275]11546 Type lookup problems in out-of-line definition of a
4497 class doubly nested from a template class
4498 * [276]11567 C++ code containing templated member function with same
4499 name as pure virtual member function results in linking failure
4500 * [277]11645 Failure to deal with using and private inheritance
4502 Java compiler and library
4504 * [278]5179 Qualified static field access doesn't initialize its
4505 class
4506 * [279]8204 gcj -O2 to native reorders certain instructions
4507 improperly
4508 * [280]10838 java.io.ObjectInputStream syntax error
4509 * [281]10886 The RMI registry that comes with GCJ does not work
4510 correctly
4511 * [282]11349 JNDI URL context factories not located correctly
4513 x86-specific (Intel/AMD)
4515 * [283]4823 ICE on inline assembly code
4516 * [284]8878 miscompilation with -O and SSE
4517 * [285]9815 (c++ library) atomicity.h - fails to compile with -O3
4518 -masm=intel
4519 * [286]10402 (inline assembly) [x86] ICE in merge_assigned_reloads,
4520 in reload1.c
4521 * [287]10504 ICE with SSE2 code and -O3 -mcpu=pentium4 -msse2
4522 * [288]10673 ICE for x86-64 on freebsd libc vfprintf.c source
4523 * [289]11044 [x86] out of range loop instructions for FP code on K6
4524 * [290]11089 ICE: instantiate_virtual_regs_lossage while using SSE
4525 built-ins
4526 * [291]11420 [x86_64] gcc generates invalid asm code when "-O -fPIC"
4527 is used
4529 SPARC- or Solaris- specific
4531 * [292]9362 solaris 'as' dies when fed .s and "-gstabs"
4532 * [293]10142 [SPARC64] gcc produces wrong code when passing
4533 structures by value
4534 * [294]10663 New configure check aborts with Sun tools.
4535 * [295]10835 combinatorial explosion in scheduler on HyperSPARC
4536 * [296]10876 ICE in calculate_giv_inc when building KDE
4537 * [297]10955 wrong code at -O3 for structure argument in context of
4538 structure return
4539 * [298]11018 -mcpu=ultrasparc busts tar-1.13.25
4540 * [299]11556 [sparc64] ICE in gen_reg_rtx() while compiling 2.6.x
4541 Linux kernel
4543 ia64 specific
4545 * [300]10907 gcc violates the ia64 ABI (GP must be preserved)
4546 * [301]11320 scheduler bug (in machine depended reorganization pass)
4547 * [302]11599 bug with conditional and __builtin_prefetch
4549 PowerPC specific
4551 * [303]9745 [powerpc] gcc mis-compiles libmcrypt (alias problem
4552 during loop)
4553 * [304]10871 error in rs6000_stack_info save_size computation
4554 * [305]11440 gcc mis-compiles c++ code (libkhtml) with -O2,
4555 -fno-gcse cures it
4557 m68k-specific
4559 * [306]7594 [m68k] ICE on legal code associated with simplify-rtx
4560 * [307]10557 [m68k] ICE in subreg_offset_representable_p
4561 * [308]11054 [m68k] ICE in reg_overlap_mentioned_p
4563 ARM-specific
4565 * [309]10834 [arm] GCC 3.3 still generates incorrect instructions
4566 for functions with __attribute__ ((interrupt ("IRQ")))
4567 * [310]10842 [arm] Clobbered link register is copied to pc under
4568 certain circumstances
4569 * [311]11052 [arm] noce_process_if_block() can lose REG_INC notes
4570 * [312]11183 [arm] ICE in change_address_1 (3.3) / subreg_hard_regno
4571 (3.4)
4573 MIPS-specific
4575 * [313]11084 ICE in propagate_one_insn, in flow.c
4577 SH-specific
4579 * [314]10331 can't compile c++ part of gcc cross compiler for sh-elf
4580 * [315]10413 [SH] ICE in reload_cse_simplify_operands, in reload1.c
4581 * [316]11096 i686-linux to sh-linux cross compiler fails to compile
4582 C++ files
4584 GNU/Linux (or Hurd?) specific
4586 * [317]2873 Bogus fixinclude of stdio.h from glibc 2.2.3
4588 UnixWare specific
4590 * [318]3163 configure bug: gcc/aclocal.m4 mmap test fails on
4591 UnixWare 7.1.1
4593 Cygwin (or mingw) specific
4595 * [319]5287 ICE with dllimport attribute
4596 * [320]10148 [MingW/CygWin] Compiler dumps core
4598 DJGPP specific
4600 * [321]8787 GCC fails to emit .intel_syntax when invoked with
4601 -masm=intel on DJGPP
4603 Darwin (and MacOS X) specific
4605 * [322]10900 trampolines crash
4607 Documentation
4609 * [323]1607 (c++) Format attributes on methods undocumented
4610 * [324]4252 Invalid option `-fdump-translation-unit'
4611 * [325]4490 Clarify restrictions on -m96bit-long-double,
4612 -m128bit-long-double
4613 * [326]10355 document an issue with regparm attribute on some
4614 systems (e.g. Solaris)
4615 * [327]10726 (fortran) Documentation for function "IDate Intrinsic
4616 (Unix)" is wrong
4617 * [328]10805 document bug in old version of Sun assembler
4618 * [329]10815 warn against GNU binutils on AIX
4619 * [330]10877 document need for newer binutils on i?86-*-linux-gnu
4620 * [331]11280 Manual incorrect with respect to -freorder-blocks
4621 * [332]11466 Document -mlittle-endian and its restrictions for the
4622 sparc64 port
4624 Testsuite bugs (compiler itself is not affected)
4626 * [333]10737 newer bison causes g++.dg/parse/crash2.C to incorrectly
4627 report failure
4628 * [334]10810 gcc-3.3 fails make check: buffer overrun in
4629 test_demangle.c
4630 _________________________________________________________________
4632 GCC 3.3.2
4634 Bug Fixes
4636 This section lists the problem reports (PRs) from [335]GCC's bug
4637 tracking system that are known to be fixed in the 3.3.2 release. This
4638 list might not be complete (that is, it is possible that some PRs that
4639 have been fixed are not listed here).
4641 Bootstrap failures and problems
4643 * [336]8336 [SCO5] bootstrap config still tries to use COFF options
4644 * [337]9330 [alpha-osf] Bootstrap failure on Compaq Tru64 with
4645 --enable-threads=posix
4646 * [338]9631 [hppa64-linux] gcc-3.3 fails to bootstrap
4647 * [339]9877 fixincludes makes a bad sys/byteorder.h on svr5
4648 (UnixWare 7.1.1)
4649 * [340]11687 xstormy16-elf build fails in libf2c
4650 * [341]12263 [SGI IRIX] bootstrap fails during compile of
4651 libf2c/libI77/backspace.c
4652 * [342]12490 buffer overflow in scan-decls.c (during Solaris 9
4653 fix-header processing)
4655 Internal compiler errors (multi-platform)
4657 * [343]7277 Casting integers to vector types causes ICE
4658 * [344]7939 (c++) ICE on invalid function template specialization
4659 * [345]11063 (c++) ICE on parsing initialization list of const array
4660 member
4661 * [346]11207 ICE with negative index in array element designator
4662 * [347]11522 (fortran) g77 dwarf-2 ICE in
4663 add_abstract_origin_attribute
4664 * [348]11595 (c++) ICE on duplicate label definition
4665 * [349]11646 (c++) ICE in commit_one_edge_insertion with
4666 -fnon-call-exceptions -fgcse -O
4667 * [350]11665 ICE in struct initializer when taking address
4668 * [351]11852 (c++) ICE with bad struct initializer.
4669 * [352]11878 (c++) ICE in cp_expr_size
4670 * [353]11883 ICE with any -O on mercury-generated C code
4671 * [354]11991 (c++) ICE in cxx_incomplete_type_diagnostic, in
4672 cp/typeck2.c when applying typeid operator to template template
4673 parameter
4674 * [355]12146 ICE in lookup_template_function, in cp/pt.c
4675 * [356]12215 ICE in make_label_edge with -fnon-call-exceptions
4676 -fno-gcse -O2
4677 * [357]12369 (c++) ICE with templates and friends
4678 * [358]12446 ICE in emit_move_insn on complicated array reference
4679 * [359]12510 ICE in final_scan_insn
4680 * [360]12544 ICE with large parameters used in nested functions
4682 C and optimization bugs
4684 * [361]9862 spurious warnings with -W -finline-functions
4685 * [362]10962 lookup_field is a linear search on a linked list (can
4686 be slow if large struct)
4687 * [363]11370 -Wunreachable-code gives false complaints
4688 * [364]11637 invalid assembly with -fnon-call-exceptions
4689 * [365]11885 Problem with bitfields in packed structs
4690 * [366]12082 Inappropriate unreachable code warnings
4691 * [367]12180 Inline optimization fails for variadic function
4692 * [368]12340 loop unroller + gcse produces wrong code
4694 C++ compiler and library
4696 * [369]3907 nested template parameter collides with member name
4697 * [370]5293 confusing message when binding a temporary to a
4698 reference
4699 * [371]5296 [DR115] Pointers to functions and to template functions
4700 behave differently in deduction
4701 * [372]7939 ICE on function template specialization
4702 * [373]8656 Unable to assign function with __attribute__ and pointer
4703 return type to an appropriate variable
4704 * [374]10147 Confusing error message for invalid template function
4705 argument
4706 * [375]11400 std::search_n() makes assumptions about Size parameter
4707 * [376]11409 issues with using declarations, overloading, and
4708 built-in functions
4709 * [377]11740 ctype<wchar_t>::do_is(mask, wchar_t) doesn't handle
4710 multiple bits in mask
4711 * [378]11786 operator() call on variable in other namespace not
4712 recognized
4713 * [379]11867 static_cast ignores ambiguity
4714 * [380]11928 bug with conversion operators that are typedefs
4715 * [381]12114 Uninitialized memory accessed in dtor
4716 * [382]12163 static_cast + explicit constructor regression
4717 * [383]12181 Wrong code with comma operator and c++
4718 * [384]12236 regparm and fastcall messes up parameters
4719 * [385]12266 incorrect instantiation of unneeded template during
4720 overload resolution
4721 * [386]12296 istream::peek() doesn't set eofbit
4722 * [387]12298 [sjlj exceptions] Stack unwind destroys
4723 not-yet-constructed object
4724 * [388]12369 ICE with templates and friends
4725 * [389]12337 apparently infinite loop in g++
4726 * [390]12344 stdcall attribute ignored if function returns a pointer
4727 * [391]12451 missing(late) class forward declaration in cxxabi.h
4728 * [392]12486 g++ accepts invalid use of a qualified name
4730 x86 specific (Intel/AMD)
4732 * [393]8869 [x86 MMX] ICE with const variable optimization and MMX
4733 builtins
4734 * [394]9786 ICE in fixup_abnormal_edges with -fnon-call-exceptions
4735 -O2
4736 * [395]11689 g++3.3 emits un-assembleable code for k6 architecture
4737 * [396]12116 [k6] Invalid assembly output values with X-MAME code
4738 * [397]12070 ICE converting between double and long double with
4739 -msoft-float
4741 ia64-specific
4743 * [398]11184 [ia64 hpux] ICE on __builtin_apply building libobjc
4744 * [399]11535 __builtin_return_address may not work on ia64
4745 * [400]11693 [ia64] ICE in gen_nop_type
4746 * [401]12224 [ia64] Thread-local storage doesn't work
4748 PowerPC-specific
4750 * [402]11087 [powerpc64-linux] GCC miscompiles raid1.c from linux
4751 kernel
4752 * [403]11319 loop miscompiled on ppc32
4753 * [404]11949 ICE Compiler segfault with ffmpeg -maltivec code
4755 SPARC-specific
4757 * [405]11662 wrong code for expr. with cast to long long and
4758 exclusive or
4759 * [406]11965 invalid assembler code for a shift < 32 operation
4760 * [407]12301 (c++) stack corruption when a returned expression
4761 throws an exception
4763 Alpha-specific
4765 * [408]11717 [alpha-linux] unrecognizable insn compiling for.c of
4766 kernel 2.4.22-pre8
4768 HPUX-specific
4770 * [409]11313 problem with #pragma weak and static inline functions
4771 * [410]11712 __STDC_EXT__ not defined for C++ by default anymore?
4773 Solaris specific
4775 * [411]12166 Profiled programs crash if PROFDIR is set
4777 Solaris-x86 specific
4779 * [412]12101 i386 Solaris no longer works with GNU as?
4781 Miscellaneous embedded target-specific bugs
4783 * [413]10988 [m32r-elf] wrong blockmove code with -O3
4784 * [414]11805 [h8300-unknown-coff] [H8300] ICE for simple code with
4785 -O2
4786 * [415]11902 [sh4] spec file improperly inserts rpath even when none
4787 needed
4788 * [416]11903 [sh4] -pthread fails to link due to error in spec file
4789 on sh4
4790 _________________________________________________________________
4792 GCC 3.3.3
4794 Minor features
4796 In addition to the bug fixes documented below, this release contains
4797 few minor features such as:
4798 * Support for --with-sysroot
4799 * Support for automatic detection of executable stacks
4800 * Support for SSE3 instructions
4801 * Support for thread local storage debugging under GDB on S390
4803 Bug Fixes
4805 This section lists the problem reports (PRs) from [417]GCC's bug
4806 tracking system that are known to be fixed in the 3.3.3 release. This
4807 list might not be complete (that is, it is possible that some PRs that
4808 have been fixed are not listed here).
4810 Bootstrap failures and issues
4812 * [418]11890 Building cross gcc-3.3.1 for sparc-sun-solaris2.6 fails
4813 * [419]12399 boehm-gc fails (when building a cross compiler):
4814 libtool unable to infer tagged configuration
4815 * [420]13068 mklibgcc.in doesn't handle multi-level multilib
4816 subdirectories properly
4818 Internal compiler errors (multi-platform)
4820 * [421]10060 ICE (stack overflow) on huge file (300k lines) due to
4821 recursive behaviour of copy_rtx_if_shared, in emit_rtl.c
4822 * [422]10555 (c++) ICE on undefined template argument
4823 * [423]10706 (c++) ICE in mangle_class_name_for_template
4824 * [424]11496 (fortran) error in flow_loops_find when -funroll-loops
4825 active
4826 * [425]11741 ICE in pre_insert_copy_insn, in gcse.c
4827 * [426]12440 GCC crashes during compilation of quicktime4linux 2.0.0
4828 * [427]12632 (fortran) -fbounds-check ICE
4829 * [428]12712 (c++) ICE on short legit C++ code fragment with gcc
4830 3.3.2
4831 * [429]12726 (c++) ICE (segfault) on trivial code
4832 * [430]12890 (c++) ICE on compilation of class with throwing method
4833 * [431]12900 (c++) ICE in rtl_verify_flow_info_1
4834 * [432]13060 (fortran) ICE in fixup_var_refs_1, in function.c on
4835 correct code with -O2 -fno-force-mem
4836 * [433]13289 (c++) ICE in regenerate_decl_from_template on recursive
4837 template
4838 * [434]13318 ICE: floating point exception in the loop optimizer
4839 * [435]13392 (c++) ICE in convert_from_eh_region_ranges_1, in
4840 except.c
4841 * [436]13574 (c++) invalid array default initializer in class lets
4842 gcc consume all memory and die
4843 * [437]13475 ICE on SIMD variables with partial value initialization
4844 * [438]13797 (c++) ICE on invalid template parameter
4845 * [439]13824 (java) gcj SEGV with simple .java program
4847 C and optimization bugs
4849 * [440]8776 loop invariants are not removed (most likely)
4850 * [441]10339 [sparc,ppc,ppc64] Invalid optimization: replacing
4851 strncmp by memcmp
4852 * [442]11350 undefined labels with -Os -fPIC
4853 * [443]12826 Optimizer removes reference through volatile pointer
4854 * [444]12500 stabs debug info: void no longer a predefined / builtin
4855 type
4856 * [445]12941 builtin-bitops-1.c miscompilation (latent bug)
4857 * [446]12953 tree inliner bug (in inline_forbidden_p) and fix
4858 * [447]13041 linux-2.6/sound/core/oss/rate.c miscompiled
4859 * [448]13507 spurious printf format warning
4860 * [449]13382 Type information for const pointer disappears during
4861 optimization.
4862 * [450]13394 noreturn attribute ignored on recursive invokation
4863 * [451]13400 Compiled code crashes storing to read-only location
4864 * [452]13521 Endless loop in calculate_global_regs_live
4866 C++ compiler and library
4868 Some of the bug fixes in this list were made to implement decisions
4869 that the ISO C++ standards committee has made concerning several
4870 defect reports (DRs). Links in the list below point to detailed
4871 discussion of the relevant defect report.
4872 * [453]2094 unimplemented: use of `ptrmem_cst' in template type
4873 unification
4874 * [454]2294 using declaration confusion
4875 * [455]5050 template instantiation depth exceeds limit: recursion
4876 problem?
4877 * [456]9371 Bad exception handling in
4878 i/ostream::operator>>/<<(streambuf*)
4879 * [457]9546 bad exception handling in ostream members
4880 * [458]10081 basic_ios::_M_cache_locale leaves NULL members in the
4881 face of unknown locales
4882 * [459]10093 [460][DR 61] Setting failbit in exceptions doesn't work
4883 * [461]10095 istream::operator>>(int&) sets ios::badbit when
4884 ios::failbit is set.
4885 * [462]11554 Warning about reordering of initializers doesn't
4886 mention location of constructor
4887 * [463]12297 istream::sentry::sentry() handles eof() incorrectly.
4888 * [464]12352 Exception safety problems in src/localename.cc
4889 * [465]12438 Memory leak in locale::combine()
4890 * [466]12540 Memory leak in locale::locale(const char*)
4891 * [467]12594 DRs [468]60 [TC] and [469]63 [TC] not implemented
4892 * [470]12657 Resolution of [471]DR 292 (WP) still unimplemented
4893 * [472]12696 memory eating infinite loop in diagnostics (error
4894 recovery problem)
4895 * [473]12815 Code compiled with optimization behaves unexpectedly
4896 * [474]12862 Conflicts between typedefs/enums and namespace member
4897 declarations
4898 * [475]12926 Wrong value after assignment in initialize list using
4899 bit-fields
4900 * [476]12967 Resolution of [477]DR 300 [WP] still unimplemented
4901 * [478]12971 Resolution of [479]DR 328 [WP] still unimplemented
4902 * [480]13007 basic_streambuf::pubimbue, imbue wrong
4903 * [481]13009 Implicitly-defined assignment operator writes to wrong
4904 memory
4905 * [482]13057 regparm attribute not applied to destructor
4906 * [483]13070 -Wformat option ignored in g++
4907 * [484]13081 forward template declarations in <complex> let inlining
4908 fail
4909 * [485]13239 Assertion does not seem to work correctly anymore
4910 * [486]13262 "xxx is private within this context" when initializing
4911 a self-contained template class
4912 * [487]13290 simple typo in concept checking for std::generate_n
4913 * [488]13323 Template code does not compile in presence of typedef
4914 * [489]13369 __verify_grouping (and __add_grouping?) not correct
4915 * [490]13371 infinite loop with packed struct and inlining
4916 * [491]13445 Template argument replacement "dereferences" a typedef
4917 * [492]13461 Fails to access protected-ctor from public constant
4918 * [493]13462 Non-standard-conforming type set::pointer
4919 * [494]13478 gcc uses wrong constructor to initialize a const
4920 reference
4921 * [495]13544 "conflicting types" for enums in different scopes
4922 * [496]13650 string::compare should not (always) use
4923 traits_type::length()
4924 * [497]13683 bogus warning about passing non-PODs through ellipsis
4925 * [498]13688 Derived class is denied access to protected base class
4926 member class
4927 * [499]13774 Member variable cleared in virtual multiple inheritance
4928 class
4929 * [500]13884 Protect sstream.tcc from extern template use
4931 Java compiler and library
4933 * [501]10746 [win32] garbage collection crash in GCJ
4935 Objective-C compiler and library
4937 * [502]11433 Crash due to dereferencing null pointer when querying
4938 protocol
4940 Fortran compiler and library
4942 * [503]12633 logical expression gives incorrect result with
4943 -fugly-logint option
4944 * [504]13037 [gcse-lm] g77 generates incorrect code
4945 * [505]13213 Hex constant problem when compiling with -fugly-logint
4946 and -ftypeless-boz
4948 x86-specific (Intel/AMD)
4950 * [506]4490 ICE with -m128bit-long-double
4951 * [507]12292 [x86_64] ICE: RTL check: expected code `const_int',
4952 have `reg' in make_field_assignment, in combine.c
4953 * [508]12441 ICE: can't find a register to spill
4954 * [509]12943 array static-init failure under -fpic, -fPIC
4955 * [510]13608 Incorrect code with -O3 -ffast-math
4957 PowerPC-specific
4959 * [511]11598 testcase gcc.dg/20020118-1.c fails runtime check of
4960 __attribute__((aligned(16)))
4961 * [512]11793 ICE in extract_insn, in recog.c (const_vector's)
4962 * [513]12467 vmsumubm emitted when vmsummbm appropriate (typo in
4963 altivec.md)
4964 * [514]12537 g++ generates writeable text sections
4966 SPARC-specific
4968 * [515]12496 wrong result for __atomic_add(&value, -1) when using
4969 -O0 -m64
4970 * [516]12865 mprotect call to make trampoline executable may fail
4971 * [517]13354 ICE in sparc_emit_set_const32
4973 ARM-specific
4975 * [518]10467 [arm] ICE in pre_insert_copy_insn,
4977 ia64-specific
4979 * [519]11226 ICE passing struct arg with two floats
4980 * [520]11227 ICE for _Complex float, _Complex long double args
4981 * [521]12644 GCC 3.3.2 fails to compile glibc on ia64
4982 * [522]13149 build gcc-3.3.2 1305 error:unrecognizable insn
4983 * Various fixes for libunwind
4985 Alpha-specific
4987 * [523]12654 Incorrect comparison code generated for Alpha
4988 * [524]12965 SEGV+ICE in cc1plus on alpha-linux with -O2
4989 * [525]13031 ICE (unrecognizable insn) when building
4990 gnome-libs-1.4.2
4992 HPPA-specific
4994 * [526]11634 [hppa] ICE in verify_local_live_at_start, in flow.c
4995 * [527]12158 [hppa] compilation does not terminate at -O1
4997 S390-specific
4999 * [528]11992 Wrong built-in code for memcmp with length 1<<24: only
5000 (1<<24)-1 possible for CLCL-Instruction
5002 SH-specific
5004 * [529]9365 segfault in gen_far_branch (config/sh/sh.c)
5005 * [530]10392 optimizer generates faulty array indexing
5006 * [531]11322 SH profiler outputs multiple definitions of symbol
5007 * [532]13069 gcc/config/sh/rtems.h broken
5008 * [533]13302 Putting a va_list in a struct causes seg fault
5009 * [534]13585 Incorrect optimization of call to sfunc
5010 * Fix inappropriately exported libgcc functions from the shared
5011 library
5013 Other embedded target specific
5015 * [535]8916 [mcore] unsigned char assign gets hosed.
5016 * [536]11576 [h8300] ICE in change_address_1, in emit-rtl.c
5017 * [537]13122 [h8300] local variable gets corrupted by function call
5018 when -fomit-frame-pointer is given
5019 * [538]13256 [cris] strict_low_part mistreated in delay slots
5020 * [539]13373 [mcore] optimization with -frerun-cse-after-loop
5021 -fexpensive-optimizations produces wrong code on mcore
5023 GNU HURD-specific
5025 * [540]12561 gcc/config/t-gnu needs updating to work with
5026 --with-sysroot
5028 Tru64 Unix specific
5030 * [541]6243 testsuite fails almost all tests due to no libintl in
5031 LD_LIBRARY_PATH during test.
5032 * [542]11397 weak aliases broken on Tru64 UNIX
5034 AIX-specific
5036 * [543]12505 build failure due to defines of uchar in cpphash.h and
5037 sys/types.h
5038 * [544]13150 WEAK symbols not exported by collect2
5040 IRIX-specific
5042 * [545]12666 fixincludes problem on IRIX 6.5.19m
5044 Solaris-specific
5046 * [546]12969 Including sys/byteorder.h breaks configure checks
5048 Testsuite problems (compiler is not affected)
5050 * [547]10819 testsuite creates CR+LF on compiler version lines in
5051 test summary files
5052 * [548]11612 abi_check not finding correct libgcc_s.so.1
5054 Miscellaneous
5056 * [549]13211 using -###, incorrect warnings about unused linker file
5057 are produced
5058 _________________________________________________________________
5060 GCC 3.3.4
5062 This is the [550]list of problem reports (PRs) from GCC's bug tracking
5063 system that are known to be fixed in the 3.3.4 release. This list
5064 might not be complete (that is, it is possible that some PRs that have
5065 been fixed are not listed here).
5066 _________________________________________________________________
5068 GCC 3.3.5
5070 This is the [551]list of problem reports (PRs) from GCC's bug tracking
5071 system that are known to be fixed in the 3.3.5 release. This list
5072 might not be complete (that is, it is possible that some PRs that have
5073 been fixed are not listed here).
5074 _________________________________________________________________
5076 GCC 3.3.6
5078 This is the [552]list of problem reports (PRs) from GCC's bug tracking
5079 system that are known to be fixed in the 3.3.6 release. This list
5080 might not be complete (that is, it is possible that some PRs that have
5081 been fixed are not listed here).
5083 Please send FSF & GNU inquiries & questions to [553]gnu@gnu.org. There
5084 are also [554]other ways to contact the FSF.
5086 These pages are maintained by [555]the GCC team.
5089 For questions related to the use of GCC, please consult these web
5090 pages and the [556]GCC manuals. If that fails, the
5091 [557]gcc-help@gcc.gnu.org mailing list might help.
5092 Please send comments on these web pages and the development of GCC
5093 to our developer mailing list at [558]gcc@gnu.org or
5094 [559]gcc@gcc.gnu.org. All of our lists have [560]public archives.
5096 Copyright (C) Free Software Foundation, Inc., 51 Franklin St, Fifth
5097 Floor, Boston, MA 02110, USA.
5099 Verbatim copying and distribution of this entire article is permitted
5100 in any medium, provided this notice is preserved.
5101 Last modified 2006-11-12 [561]Valid XHTML 1.0
5103 References
5105 1. http://gcc.gnu.org/gcc-3.3/changes.html#3.3.6
5106 2. http://gcc.gnu.org/gcc-3.1/changes.html#obsolete_systems
5107 3. http://gcc.gnu.org/gcc-3.3/changes.html#obsolete_systems
5108 4. http://gcc.gnu.org/gcc-3.3/changes.html#nonnull_attribute
5109 5. http://gcc.gnu.org/news/dfa.html
5110 6. http://gcc.gnu.org/gcc-3.3/c99status.html
5111 7. http://gcc.gnu.org/onlinedocs/gcc-3.3.6/g77/News.html
5112 8. http://gcc.gnu.org/PR10140
5113 9. http://gcc.gnu.org/PR10198
5114 10. http://gcc.gnu.org/PR10338
5115 11. http://gcc.gnu.org/PR3581
5116 12. http://gcc.gnu.org/PR4382
5117 13. http://gcc.gnu.org/PR5533
5118 14. http://gcc.gnu.org/PR6387
5119 15. http://gcc.gnu.org/PR6412
5120 16. http://gcc.gnu.org/PR6620
5121 17. http://gcc.gnu.org/PR6663
5122 18. http://gcc.gnu.org/PR7068
5123 19. http://gcc.gnu.org/PR7083
5124 20. http://gcc.gnu.org/PR7647
5125 21. http://gcc.gnu.org/PR7675
5126 22. http://gcc.gnu.org/PR7718
5127 23. http://gcc.gnu.org/PR8116
5128 24. http://gcc.gnu.org/PR8358
5129 25. http://gcc.gnu.org/PR8511
5130 26. http://gcc.gnu.org/PR8564
5131 27. http://gcc.gnu.org/PR8660
5132 28. http://gcc.gnu.org/PR8766
5133 29. http://gcc.gnu.org/PR8803
5134 30. http://gcc.gnu.org/PR8846
5135 31. http://gcc.gnu.org/PR8906
5136 32. http://gcc.gnu.org/PR9216
5137 33. http://gcc.gnu.org/PR9261
5138 34. http://gcc.gnu.org/PR9263
5139 35. http://gcc.gnu.org/PR9429
5140 36. http://gcc.gnu.org/PR9516
5141 37. http://gcc.gnu.org/PR9600
5142 38. http://gcc.gnu.org/PR9629
5143 39. http://gcc.gnu.org/PR9672
5144 40. http://gcc.gnu.org/PR9749
5145 41. http://gcc.gnu.org/PR9794
5146 42. http://gcc.gnu.org/PR9829
5147 43. http://gcc.gnu.org/PR9916
5148 44. http://gcc.gnu.org/PR9936
5149 45. http://gcc.gnu.org/PR10262
5150 46. http://gcc.gnu.org/PR10278
5151 47. http://gcc.gnu.org/PR10446
5152 48. http://gcc.gnu.org/PR10451
5153 49. http://gcc.gnu.org/PR10506
5154 50. http://gcc.gnu.org/PR10549
5155 51. http://gcc.gnu.org/PR2001
5156 52. http://gcc.gnu.org/PR2391
5157 53. http://gcc.gnu.org/PR2960
5158 54. http://gcc.gnu.org/PR4046
5159 55. http://gcc.gnu.org/PR6405
5160 56. http://gcc.gnu.org/PR6798
5161 57. http://gcc.gnu.org/PR6871
5162 58. http://gcc.gnu.org/PR6909
5163 59. http://gcc.gnu.org/PR7189
5164 60. http://gcc.gnu.org/PR7642
5165 61. http://gcc.gnu.org/PR8634
5166 62. http://gcc.gnu.org/PR8750
5167 63. http://gcc.gnu.org/PR2161
5168 64. http://gcc.gnu.org/PR4319
5169 65. http://gcc.gnu.org/PR8602
5170 66. http://gcc.gnu.org/PR9177
5171 67. http://gcc.gnu.org/PR9853
5172 68. http://gcc.gnu.org/PR45
5173 69. http://gcc.gnu.org/PR3784
5174 70. http://gcc.gnu.org/PR764
5175 71. http://gcc.gnu.org/PR5116
5176 72. http://gcc.gnu.org/PR2862
5177 73. http://gcc.gnu.org/PR3663
5178 74. http://gcc.gnu.org/PR3797
5179 75. http://gcc.gnu.org/PR3948
5180 76. http://gcc.gnu.org/PR4137
5181 77. http://gcc.gnu.org/PR4361
5182 78. http://gcc.gnu.org/PR4802
5183 79. http://gcc.gnu.org/PR5837
5184 80. http://gcc.gnu.org/PR4803
5185 81. http://gcc.gnu.org/PR5094
5186 82. http://gcc.gnu.org/PR5730
5187 83. http://gcc.gnu.org/PR6713
5188 84. http://gcc.gnu.org/PR7015
5189 85. http://gcc.gnu.org/PR7086
5190 86. http://gcc.gnu.org/PR7099
5191 87. http://gcc.gnu.org/PR7247
5192 88. http://gcc.gnu.org/PR7441
5193 89. http://gcc.gnu.org/PR7768
5194 90. http://gcc.gnu.org/PR7804
5195 91. http://gcc.gnu.org/PR8099
5196 92. http://gcc.gnu.org/PR8117
5197 93. http://gcc.gnu.org/PR8205
5198 94. http://gcc.gnu.org/PR8645
5199 95. http://gcc.gnu.org/PR8724
5200 96. http://gcc.gnu.org/PR8805
5201 97. http://gcc.gnu.org/PR8691
5202 98. http://gcc.gnu.org/PR8700
5203 99. http://gcc.gnu.org/PR8724
5204 100. http://gcc.gnu.org/PR8949
5205 101. http://gcc.gnu.org/PR9016
5206 102. http://gcc.gnu.org/PR9053
5207 103. http://gcc.gnu.org/PR9152
5208 104. http://gcc.gnu.org/PR9182
5209 105. http://gcc.gnu.org/PR9297
5210 106. http://gcc.gnu.org/PR9318
5211 107. http://gcc.gnu.org/PR9320
5212 108. http://gcc.gnu.org/PR9400
5213 109. http://gcc.gnu.org/PR9424
5214 110. http://gcc.gnu.org/PR9425
5215 111. http://gcc.gnu.org/PR9439
5216 112. http://gcc.gnu.org/PR9474
5217 113. http://gcc.gnu.org/PR9548
5218 114. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#231
5219 115. http://gcc.gnu.org/PR9555
5220 116. http://gcc.gnu.org/PR9561
5221 117. http://gcc.gnu.org/PR9563
5222 118. http://gcc.gnu.org/PR9582
5223 119. http://gcc.gnu.org/PR9622
5224 120. http://gcc.gnu.org/PR9683
5225 121. http://gcc.gnu.org/PR9791
5226 122. http://gcc.gnu.org/PR9817
5227 123. http://gcc.gnu.org/PR9825
5228 124. http://gcc.gnu.org/PR9826
5229 125. http://gcc.gnu.org/PR9924
5230 126. http://gcc.gnu.org/PR9946
5231 127. http://gcc.gnu.org/PR9964
5232 128. http://gcc.gnu.org/PR9988
5233 129. http://gcc.gnu.org/PR10033
5234 130. http://gcc.gnu.org/PR10097
5235 131. http://gcc.gnu.org/PR10132
5236 132. http://gcc.gnu.org/PR10180
5237 133. http://gcc.gnu.org/PR10199
5238 134. http://gcc.gnu.org/PR10300
5239 135. http://gcc.gnu.org/PR10427
5240 136. http://gcc.gnu.org/PR10503
5241 137. http://gcc.gnu.org/PR5956
5242 138. http://gcc.gnu.org/PR1832
5243 139. http://gcc.gnu.org/PR3924
5244 140. http://gcc.gnu.org/PR5634
5245 141. http://gcc.gnu.org/PR6367
5246 142. http://gcc.gnu.org/PR6491
5247 143. http://gcc.gnu.org/PR6742
5248 144. http://gcc.gnu.org/PR7113
5249 145. http://gcc.gnu.org/PR7236
5250 146. http://gcc.gnu.org/PR7278
5251 147. http://gcc.gnu.org/PR7384
5252 148. http://gcc.gnu.org/PR7388
5253 149. http://gcc.gnu.org/PR8587
5254 150. http://gcc.gnu.org/PR9038
5255 151. http://gcc.gnu.org/PR10197
5256 152. http://gcc.gnu.org/PR6005
5257 153. http://gcc.gnu.org/PR6389
5258 154. http://gcc.gnu.org/PR6576
5259 155. http://gcc.gnu.org/PR6652
5260 156. http://gcc.gnu.org/PR7060
5261 157. http://gcc.gnu.org/PR7073
5262 158. http://gcc.gnu.org/PR7180
5263 159. http://gcc.gnu.org/PR7416
5264 160. http://gcc.gnu.org/PR7570
5265 161. http://gcc.gnu.org/PR7578
5266 162. http://gcc.gnu.org/PR7611
5267 163. http://gcc.gnu.org/PR7709
5268 164. http://gcc.gnu.org/PR7766
5269 165. http://gcc.gnu.org/PR7785
5270 166. http://gcc.gnu.org/PR7786
5271 167. http://gcc.gnu.org/PR8142
5272 168. http://gcc.gnu.org/PR8234
5273 169. http://gcc.gnu.org/PR8415
5274 170. http://gcc.gnu.org/PR8481
5275 171. http://gcc.gnu.org/PR8593
5276 172. http://gcc.gnu.org/PR8759
5277 173. http://gcc.gnu.org/PR8997
5278 174. http://gcc.gnu.org/PR9253
5279 175. http://gcc.gnu.org/PR9254
5280 176. http://gcc.gnu.org/PR9271
5281 177. http://gcc.gnu.org/PR6767
5282 178. http://gcc.gnu.org/PR9911
5283 179. http://gcc.gnu.org/PR10020
5284 180. http://gcc.gnu.org/PR10546
5285 181. http://gcc.gnu.org/PR7029
5286 182. http://gcc.gnu.org/PR2903
5287 183. http://gcc.gnu.org/PR7873
5288 184. http://gcc.gnu.org/PR7680
5289 185. http://gcc.gnu.org/PR8705
5290 186. http://gcc.gnu.org/PR9986
5291 187. http://gcc.gnu.org/PR10056
5292 188. http://gcc.gnu.org/PR6744
5293 189. http://gcc.gnu.org/PR7361
5294 190. http://gcc.gnu.org/PR9496
5295 191. http://gcc.gnu.org/PR7067
5296 192. http://gcc.gnu.org/PR8480
5297 193. http://gcc.gnu.org/PR8784
5298 194. http://gcc.gnu.org/PR10315
5299 195. http://gcc.gnu.org/PR10267
5300 196. http://gcc.gnu.org/PR7916
5301 197. http://gcc.gnu.org/PR7926
5302 198. http://gcc.gnu.org/PR8555
5303 199. http://gcc.gnu.org/PR8994
5304 200. http://gcc.gnu.org/PR9426
5305 201. http://gcc.gnu.org/PR9806
5306 202. http://gcc.gnu.org/PR10077
5307 203. http://gcc.gnu.org/PR10233
5308 204. http://gcc.gnu.org/PR10286
5309 205. http://gcc.gnu.org/PR10308
5310 206. http://gcc.gnu.org/PR11272
5311 207. http://gcc.gnu.org/PR5754
5312 208. http://gcc.gnu.org/PR6597
5313 209. http://gcc.gnu.org/PR6949
5314 210. http://gcc.gnu.org/PR7053
5315 211. http://gcc.gnu.org/PR8164
5316 212. http://gcc.gnu.org/PR8384
5317 213. http://gcc.gnu.org/PR9559
5318 214. http://gcc.gnu.org/PR9649
5319 215. http://gcc.gnu.org/PR9864
5320 216. http://gcc.gnu.org/PR10432
5321 217. http://gcc.gnu.org/PR10475
5322 218. http://gcc.gnu.org/PR10635
5323 219. http://gcc.gnu.org/PR10661
5324 220. http://gcc.gnu.org/PR10700
5325 221. http://gcc.gnu.org/PR10712
5326 222. http://gcc.gnu.org/PR10796
5327 223. http://gcc.gnu.org/PR10890
5328 224. http://gcc.gnu.org/PR10939
5329 225. http://gcc.gnu.org/PR10956
5330 226. http://gcc.gnu.org/PR11041
5331 227. http://gcc.gnu.org/PR11059
5332 228. http://gcc.gnu.org/PR11083
5333 229. http://gcc.gnu.org/PR11105
5334 230. http://gcc.gnu.org/PR11149
5335 231. http://gcc.gnu.org/PR11228
5336 232. http://gcc.gnu.org/PR11282
5337 233. http://gcc.gnu.org/PR11301
5338 234. http://gcc.gnu.org/PR11308
5339 235. http://gcc.gnu.org/PR11473
5340 236. http://gcc.gnu.org/PR11503
5341 237. http://gcc.gnu.org/PR11513
5342 238. http://gcc.gnu.org/PR11198
5343 239. http://gcc.gnu.org/PR11304
5344 240. http://gcc.gnu.org/PR11381
5345 241. http://gcc.gnu.org/PR11536
5346 242. http://gcc.gnu.org/PR11557
5347 243. http://gcc.gnu.org/PR5897
5348 244. http://gcc.gnu.org/PR11279
5349 245. http://gcc.gnu.org/PR11022
5350 246. http://gcc.gnu.org/PR2330
5351 247. http://gcc.gnu.org/PR5388
5352 248. http://gcc.gnu.org/PR5390
5353 249. http://gcc.gnu.org/PR7877
5354 250. http://gcc.gnu.org/PR9393
5355 251. http://gcc.gnu.org/PR10032
5356 252. http://gcc.gnu.org/PR10468
5357 253. http://gcc.gnu.org/PR10527
5358 254. http://gcc.gnu.org/PR10679
5359 255. http://gcc.gnu.org/PR10682
5360 256. http://gcc.gnu.org/PR10689
5361 257. http://gcc.gnu.org/PR10845
5362 258. http://gcc.gnu.org/PR10849
5363 259. http://gcc.gnu.org/PR10888
5364 260. http://gcc.gnu.org/PR10929
5365 261. http://gcc.gnu.org/PR10931
5366 262. http://gcc.gnu.org/PR10940
5367 263. http://gcc.gnu.org/PR10968
5368 264. http://gcc.gnu.org/PR10990
5369 265. http://gcc.gnu.org/PR11039
5370 266. http://gcc.gnu.org/PR11062
5371 267. http://gcc.gnu.org/PR11095
5372 268. http://gcc.gnu.org/PR11098
5373 269. http://gcc.gnu.org/PR11137
5374 270. http://gcc.gnu.org/PR11154
5375 271. http://gcc.gnu.org/PR11329
5376 272. http://gcc.gnu.org/PR11332
5377 273. http://gcc.gnu.org/PR11431
5378 274. http://gcc.gnu.org/PR11528
5379 275. http://gcc.gnu.org/PR11546
5380 276. http://gcc.gnu.org/PR11567
5381 277. http://gcc.gnu.org/PR11645
5382 278. http://gcc.gnu.org/PR5179
5383 279. http://gcc.gnu.org/PR8204
5384 280. http://gcc.gnu.org/PR10838
5385 281. http://gcc.gnu.org/PR10886
5386 282. http://gcc.gnu.org/PR11349
5387 283. http://gcc.gnu.org/PR4823
5388 284. http://gcc.gnu.org/PR8878
5389 285. http://gcc.gnu.org/PR9815
5390 286. http://gcc.gnu.org/PR10402
5391 287. http://gcc.gnu.org/PR10504
5392 288. http://gcc.gnu.org/PR10673
5393 289. http://gcc.gnu.org/PR11044
5394 290. http://gcc.gnu.org/PR11089
5395 291. http://gcc.gnu.org/PR11420
5396 292. http://gcc.gnu.org/PR9362
5397 293. http://gcc.gnu.org/PR10142
5398 294. http://gcc.gnu.org/PR10663
5399 295. http://gcc.gnu.org/PR10835
5400 296. http://gcc.gnu.org/PR10876
5401 297. http://gcc.gnu.org/PR10955
5402 298. http://gcc.gnu.org/PR11018
5403 299. http://gcc.gnu.org/PR11556
5404 300. http://gcc.gnu.org/PR10907
5405 301. http://gcc.gnu.org/PR11320
5406 302. http://gcc.gnu.org/PR11599
5407 303. http://gcc.gnu.org/PR9745
5408 304. http://gcc.gnu.org/PR10871
5409 305. http://gcc.gnu.org/PR11440
5410 306. http://gcc.gnu.org/PR7594
5411 307. http://gcc.gnu.org/PR10557
5412 308. http://gcc.gnu.org/PR11054
5413 309. http://gcc.gnu.org/PR10834
5414 310. http://gcc.gnu.org/PR10842
5415 311. http://gcc.gnu.org/PR11052
5416 312. http://gcc.gnu.org/PR11183
5417 313. http://gcc.gnu.org/PR11084
5418 314. http://gcc.gnu.org/PR10331
5419 315. http://gcc.gnu.org/PR10413
5420 316. http://gcc.gnu.org/PR11096
5421 317. http://gcc.gnu.org/PR2873
5422 318. http://gcc.gnu.org/PR3163
5423 319. http://gcc.gnu.org/PR5287
5424 320. http://gcc.gnu.org/PR10148
5425 321. http://gcc.gnu.org/PR8787
5426 322. http://gcc.gnu.org/PR10900
5427 323. http://gcc.gnu.org/PR1607
5428 324. http://gcc.gnu.org/PR4252
5429 325. http://gcc.gnu.org/PR4490
5430 326. http://gcc.gnu.org/PR10355
5431 327. http://gcc.gnu.org/PR10726
5432 328. http://gcc.gnu.org/PR10805
5433 329. http://gcc.gnu.org/PR10815
5434 330. http://gcc.gnu.org/PR10877
5435 331. http://gcc.gnu.org/PR11280
5436 332. http://gcc.gnu.org/PR11466
5437 333. http://gcc.gnu.org/PR10737
5438 334. http://gcc.gnu.org/PR10810
5439 335. http://gcc.gnu.org/bugzilla/
5440 336. http://gcc.gnu.org/PR8336
5441 337. http://gcc.gnu.org/PR9330
5442 338. http://gcc.gnu.org/PR9631
5443 339. http://gcc.gnu.org/PR9877
5444 340. http://gcc.gnu.org/PR11687
5445 341. http://gcc.gnu.org/PR12263
5446 342. http://gcc.gnu.org/PR12490
5447 343. http://gcc.gnu.org/PR7277
5448 344. http://gcc.gnu.org/PR7939
5449 345. http://gcc.gnu.org/PR11063
5450 346. http://gcc.gnu.org/PR11207
5451 347. http://gcc.gnu.org/PR11522
5452 348. http://gcc.gnu.org/PR11595
5453 349. http://gcc.gnu.org/PR11646
5454 350. http://gcc.gnu.org/PR11665
5455 351. http://gcc.gnu.org/PR11852
5456 352. http://gcc.gnu.org/PR11878
5457 353. http://gcc.gnu.org/PR11883
5458 354. http://gcc.gnu.org/PR11991
5459 355. http://gcc.gnu.org/PR12146
5460 356. http://gcc.gnu.org/PR12215
5461 357. http://gcc.gnu.org/PR12369
5462 358. http://gcc.gnu.org/PR12446
5463 359. http://gcc.gnu.org/PR12510
5464 360. http://gcc.gnu.org/PR12544
5465 361. http://gcc.gnu.org/PR9862
5466 362. http://gcc.gnu.org/PR10962
5467 363. http://gcc.gnu.org/PR11370
5468 364. http://gcc.gnu.org/PR11637
5469 365. http://gcc.gnu.org/PR11885
5470 366. http://gcc.gnu.org/PR12082
5471 367. http://gcc.gnu.org/PR12180
5472 368. http://gcc.gnu.org/PR12340
5473 369. http://gcc.gnu.org/PR3907
5474 370. http://gcc.gnu.org/PR5293
5475 371. http://gcc.gnu.org/PR5296
5476 372. http://gcc.gnu.org/PR7939
5477 373. http://gcc.gnu.org/PR8656
5478 374. http://gcc.gnu.org/PR10147
5479 375. http://gcc.gnu.org/PR11400
5480 376. http://gcc.gnu.org/PR11409
5481 377. http://gcc.gnu.org/PR11740
5482 378. http://gcc.gnu.org/PR11786
5483 379. http://gcc.gnu.org/PR11867
5484 380. http://gcc.gnu.org/PR11928
5485 381. http://gcc.gnu.org/PR12114
5486 382. http://gcc.gnu.org/PR12163
5487 383. http://gcc.gnu.org/PR12181
5488 384. http://gcc.gnu.org/PR12236
5489 385. http://gcc.gnu.org/PR12266
5490 386. http://gcc.gnu.org/PR12296
5491 387. http://gcc.gnu.org/PR12298
5492 388. http://gcc.gnu.org/PR12369
5493 389. http://gcc.gnu.org/PR12337
5494 390. http://gcc.gnu.org/PR12344
5495 391. http://gcc.gnu.org/PR12451
5496 392. http://gcc.gnu.org/PR12486
5497 393. http://gcc.gnu.org/PR8869
5498 394. http://gcc.gnu.org/PR9786
5499 395. http://gcc.gnu.org/PR11689
5500 396. http://gcc.gnu.org/PR12116
5501 397. http://gcc.gnu.org/PR12070
5502 398. http://gcc.gnu.org/PR11184
5503 399. http://gcc.gnu.org/PR11535
5504 400. http://gcc.gnu.org/PR11693
5505 401. http://gcc.gnu.org/PR12224
5506 402. http://gcc.gnu.org/PR11087
5507 403. http://gcc.gnu.org/PR11319
5508 404. http://gcc.gnu.org/PR11949
5509 405. http://gcc.gnu.org/PR11662
5510 406. http://gcc.gnu.org/PR11965
5511 407. http://gcc.gnu.org/PR12301
5512 408. http://gcc.gnu.org/PR11717
5513 409. http://gcc.gnu.org/PR11313
5514 410. http://gcc.gnu.org/PR11712
5515 411. http://gcc.gnu.org/PR12166
5516 412. http://gcc.gnu.org/PR12101
5517 413. http://gcc.gnu.org/PR10988
5518 414. http://gcc.gnu.org/PR11805
5519 415. http://gcc.gnu.org/PR11902
5520 416. http://gcc.gnu.org/PR11903
5521 417. http://gcc.gnu.org/bugzilla/
5522 418. http://gcc.gnu.org/PR11890
5523 419. http://gcc.gnu.org/PR12399
5524 420. http://gcc.gnu.org/PR13068
5525 421. http://gcc.gnu.org/PR10060
5526 422. http://gcc.gnu.org/PR10555
5527 423. http://gcc.gnu.org/PR10706
5528 424. http://gcc.gnu.org/PR11496
5529 425. http://gcc.gnu.org/PR11741
5530 426. http://gcc.gnu.org/PR12440
5531 427. http://gcc.gnu.org/PR12632
5532 428. http://gcc.gnu.org/PR12712
5533 429. http://gcc.gnu.org/PR12726
5534 430. http://gcc.gnu.org/PR12890
5535 431. http://gcc.gnu.org/PR12900
5536 432. http://gcc.gnu.org/PR13060
5537 433. http://gcc.gnu.org/PR13289
5538 434. http://gcc.gnu.org/PR13318
5539 435. http://gcc.gnu.org/PR13392
5540 436. http://gcc.gnu.org/PR13574
5541 437. http://gcc.gnu.org/PR13475
5542 438. http://gcc.gnu.org/PR13797
5543 439. http://gcc.gnu.org/PR13824
5544 440. http://gcc.gnu.org/PR8776
5545 441. http://gcc.gnu.org/PR10339
5546 442. http://gcc.gnu.org/PR11350
5547 443. http://gcc.gnu.org/PR12826
5548 444. http://gcc.gnu.org/PR12500
5549 445. http://gcc.gnu.org/PR12941
5550 446. http://gcc.gnu.org/PR12953
5551 447. http://gcc.gnu.org/PR13041
5552 448. http://gcc.gnu.org/PR13507
5553 449. http://gcc.gnu.org/PR13382
5554 450. http://gcc.gnu.org/PR13394
5555 451. http://gcc.gnu.org/PR13400
5556 452. http://gcc.gnu.org/PR13521
5557 453. http://gcc.gnu.org/PR2094
5558 454. http://gcc.gnu.org/PR2294
5559 455. http://gcc.gnu.org/PR5050
5560 456. http://gcc.gnu.org/PR9371
5561 457. http://gcc.gnu.org/PR9546
5562 458. http://gcc.gnu.org/PR10081
5563 459. http://gcc.gnu.org/PR10093
5564 460. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#61
5565 461. http://gcc.gnu.org/PR10095
5566 462. http://gcc.gnu.org/PR11554
5567 463. http://gcc.gnu.org/PR12297
5568 464. http://gcc.gnu.org/PR12352
5569 465. http://gcc.gnu.org/PR12438
5570 466. http://gcc.gnu.org/PR12540
5571 467. http://gcc.gnu.org/PR12594
5572 468. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#60
5573 469. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#63
5574 470. http://gcc.gnu.org/PR12657
5575 471. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#292
5576 472. http://gcc.gnu.org/PR12696
5577 473. http://gcc.gnu.org/PR12815
5578 474. http://gcc.gnu.org/PR12862
5579 475. http://gcc.gnu.org/PR12926
5580 476. http://gcc.gnu.org/PR12967
5581 477. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html
5582 478. http://gcc.gnu.org/PR12971
5583 479. http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#328
5584 480. http://gcc.gnu.org/PR13007
5585 481. http://gcc.gnu.org/PR13009
5586 482. http://gcc.gnu.org/PR13057
5587 483. http://gcc.gnu.org/PR13070
5588 484. http://gcc.gnu.org/PR13081
5589 485. http://gcc.gnu.org/PR13239
5590 486. http://gcc.gnu.org/PR13262
5591 487. http://gcc.gnu.org/PR13290
5592 488. http://gcc.gnu.org/PR13323
5593 489. http://gcc.gnu.org/PR13369
5594 490. http://gcc.gnu.org/PR13371
5595 491. http://gcc.gnu.org/PR13445
5596 492. http://gcc.gnu.org/PR13461
5597 493. http://gcc.gnu.org/PR13462
5598 494. http://gcc.gnu.org/PR13478
5599 495. http://gcc.gnu.org/PR13544
5600 496. http://gcc.gnu.org/PR13650
5601 497. http://gcc.gnu.org/PR13683
5602 498. http://gcc.gnu.org/PR13688
5603 499. http://gcc.gnu.org/PR13774
5604 500. http://gcc.gnu.org/PR13884
5605 501. http://gcc.gnu.org/PR10746
5606 502. http://gcc.gnu.org/PR11433
5607 503. http://gcc.gnu.org/PR12633
5608 504. http://gcc.gnu.org/PR13037
5609 505. http://gcc.gnu.org/PR13213
5610 506. http://gcc.gnu.org/PR4490
5611 507. http://gcc.gnu.org/PR12292
5612 508. http://gcc.gnu.org/PR12441
5613 509. http://gcc.gnu.org/PR12943
5614 510. http://gcc.gnu.org/PR13608
5615 511. http://gcc.gnu.org/PR11598
5616 512. http://gcc.gnu.org/PR11793
5617 513. http://gcc.gnu.org/PR12467
5618 514. http://gcc.gnu.org/PR12537
5619 515. http://gcc.gnu.org/PR12496
5620 516. http://gcc.gnu.org/PR12865
5621 517. http://gcc.gnu.org/PR13354
5622 518. http://gcc.gnu.org/PR10467
5623 519. http://gcc.gnu.org/PR11226
5624 520. http://gcc.gnu.org/PR11227
5625 521. http://gcc.gnu.org/PR12644
5626 522. http://gcc.gnu.org/PR13149
5627 523. http://gcc.gnu.org/PR12654
5628 524. http://gcc.gnu.org/PR12965
5629 525. http://gcc.gnu.org/PR13031
5630 526. http://gcc.gnu.org/PR11634
5631 527. http://gcc.gnu.org/PR12158
5632 528. http://gcc.gnu.org/PR11992
5633 529. http://gcc.gnu.org/PR9365
5634 530. http://gcc.gnu.org/PR10392
5635 531. http://gcc.gnu.org/PR11322
5636 532. http://gcc.gnu.org/PR13069
5637 533. http://gcc.gnu.org/PR13302
5638 534. http://gcc.gnu.org/PR13585
5639 535. http://gcc.gnu.org/PR8916
5640 536. http://gcc.gnu.org/PR11576
5641 537. http://gcc.gnu.org/PR13122
5642 538. http://gcc.gnu.org/PR13256
5643 539. http://gcc.gnu.org/PR13373
5644 540. http://gcc.gnu.org/PR12561
5645 541. http://gcc.gnu.org/PR6243
5646 542. http://gcc.gnu.org/PR11397
5647 543. http://gcc.gnu.org/PR12505
5648 544. http://gcc.gnu.org/PR13150
5649 545. http://gcc.gnu.org/PR12666
5650 546. http://gcc.gnu.org/PR12969
5651 547. http://gcc.gnu.org/PR10819
5652 548. http://gcc.gnu.org/PR11612
5653 549. http://gcc.gnu.org/PR13211
5654 550. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.3.4
5655 551. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.3.5
5656 552. http://gcc.gnu.org/bugzilla/buglist.cgi?bug_status=RESOLVED&resolution=FIXED&target_milestone=3.3.6
5657 553. mailto:gnu@gnu.org
5658 554. http://www.gnu.org/home.html#ContactInfo
5659 555. http://gcc.gnu.org/about.html
5660 556. http://gcc.gnu.org/onlinedocs/
5661 557. mailto:gcc-help@gcc.gnu.org
5662 558. mailto:gcc@gnu.org
5663 559. mailto:gcc@gcc.gnu.org
5664 560. http://gcc.gnu.org/lists.html
5665 561. http://validator.w3.org/check/referer
5666 ======================================================================
5667 http://gcc.gnu.org/gcc-3.2/index.html
5669 GCC 3.2 Release Series
5671 April 25, 2003
5673 The [1]GNU project and the GCC developers are pleased to announce the
5674 release of GCC 3.2.3.
5676 The purpose of the GCC 3.2 release series is to provide a stable
5677 platform for OS distributors to use building their next releases. A
5678 primary objective was to stabilize the C++ ABI; we believe that the
5679 interface to the compiler and the C++ standard library are now
5680 relatively stable.
5682 Be aware that C++ code compiled by GCC 3.2.x will (in general) not
5683 interoperate with code compiled by GCC 3.1.1 or earlier.
5685 Please refer to our [2]detailed list of news, caveats, and bug-fixes
5686 for further information.
5688 Release History
5690 GCC 3.2.3
5691 April 25, 2003 ([3]changes)
5693 GCC 3.2.2
5694 February 5, 2003 ([4]changes)
5696 GCC 3.2.1
5697 November 19, 2002 ([5]changes)
5699 GCC 3.2
5700 August 14, 2002 ([6]changes)
5702 References and Acknowledgements
5704 GCC used to stand for the GNU C Compiler, but since the compiler
5705 supports several other languages aside from C, it now stands for the
5706 GNU Compiler Collection.
5708 A list of [7]successful builds is updated as new information becomes
5709 available.
5711 The GCC developers would like to thank the numerous people that have
5712 contributed new features, improvements, bug fixes, and other changes
5713 as well as test results to GCC. This [8]amazing group of volunteers is
5714 what makes GCC successful.
5716 For additional information about GCC please refer to the [9]GCC
5717 project web site or contact the [10]GCC development mailing list.
5719 To obtain GCC please use [11]our mirror sites, one of the [12]GNU
5720 mirror sites, or our CVS server.
5722 Please send FSF & GNU inquiries & questions to [13]gnu@gnu.org. There
5723 are also [14]other ways to contact the FSF.
5725 These pages are maintained by [15]the GCC team.
5728 For questions related to the use of GCC, please consult these web
5729 pages and the [16]GCC manuals. If that fails, the
5730 [17]gcc-help@gcc.gnu.org mailing list might help.
5731 Please send comments on these web pages and the development of GCC
5732 to our developer mailing list at [18]gcc@gnu.org or
5733 [19]gcc@gcc.gnu.org. All of our lists have [20]public archives.
5735 Copyright (C) Free Software Foundation, Inc., 51 Franklin St, Fifth
5736 Floor, Boston, MA 02110, USA.
5738 Verbatim copying and distribution of this entire article is permitted
5739 in any medium, provided this notice is preserved.
5740 Last modified 2006-06-21 [21]Valid XHTML 1.0
5742 References
5744 1. http://www.gnu.org/
5745 2. http://gcc.gnu.org/gcc-3.2/changes.html
5746 3. http://gcc.gnu.org/gcc-3.2/changes.html#3.2.3
5747 4. http://gcc.gnu.org/gcc-3.2/changes.html#3.2.2
5748 5. http://gcc.gnu.org/gcc-3.2/changes.html#3.2.1
5749 6. http://gcc.gnu.org/gcc-3.2/changes.html#3.2
5750 7. http://gcc.gnu.org/gcc-3.2/buildstat.html
5751 8. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
5752 9. http://gcc.gnu.org/index.html
5753 10. mailto:gcc@gcc.gnu.org
5754 11. http://gcc.gnu.org/mirrors.html
5755 12. http://www.gnu.org/order/ftp.html
5756 13. mailto:gnu@gnu.org
5757 14. http://www.gnu.org/home.html#ContactInfo
5758 15. http://gcc.gnu.org/about.html
5759 16. http://gcc.gnu.org/onlinedocs/
5760 17. mailto:gcc-help@gcc.gnu.org
5761 18. mailto:gcc@gnu.org
5762 19. mailto:gcc@gcc.gnu.org
5763 20. http://gcc.gnu.org/lists.html
5764 21. http://validator.w3.org/check/referer
5765 ======================================================================
5766 http://gcc.gnu.org/gcc-3.2/changes.html
5768 GCC 3.2 Release Series
5769 Changes, New Features, and Fixes
5771 The latest release in the 3.2 release series is [1]GCC 3.2.3.
5773 Caveats and New Features
5775 Caveats
5777 * The C++ compiler does not correctly zero-initialize
5778 pointers-to-data members. You must explicitly initialize them. For
5779 example: int S::*m(0); will work, but depending on
5780 default-initialization to zero will not work. This bug cannot be
5781 fixed in GCC 3.2 without inducing unacceptable risks. It will be
5782 fixed in GCC 3.3.
5783 * This GCC release is based on the GCC 3.1 sourcebase, and thus has
5784 all the [2]changes in the GCC 3.1 series. In addition, GCC 3.2 has
5785 a number of C++ ABI fixes which make its C++ compiler generate
5786 binary code which is incompatible with the C++ compilers found in
5787 earlier GCC releases, including GCC 3.1 and GCC 3.1.1.
5789 Frontend Enhancements
5791 C/C++/Objective-C
5793 * The method of constructing the list of directories to be searched
5794 for header files has been revised. If a directory named by a -I
5795 option is a standard system include directory, the option is
5796 ignored to ensure that the default search order for system
5797 directories and the special treatment of system header files are
5798 not defeated.
5799 * The C and Objective-C compilers no longer accept the "Naming
5800 Types" extension (typedef foo = bar); it was already unavailable
5801 in C++. Code which uses it will need to be changed to use the
5802 "typeof" extension instead: typedef typeof(bar) foo. (We have
5803 removed this extension without a period of deprecation because it
5804 has caused the compiler to crash since version 3.0 and no one
5805 noticed until very recently. Thus we conclude it is not in
5806 widespread use.)
5808 C++
5810 * GCC 3.2 fixed serveral differences between the C++ ABI implemented
5811 in GCC and the multi-vendor standard, but more have been found
5812 since the release. 3.2.1 adds a new warning, -Wabi, to warn about
5813 code which is affected by these bugs. We will fix these bugs in
5814 some future release, once we are confident that all have been
5815 found; until then, it is our intention to make changes to the ABI
5816 only if they are necessary for correct compilation of C++, as
5817 opposed to conformance to the ABI documents.
5818 * For details on how to build an ABI compliant compiler for
5819 GNU/Linux systems, check the [3]common C++ ABI page.
5821 New Targets and Target Specific Improvements
5823 IA-32
5825 * Fixed a number of bugs in SSE and MMX intrinsics.
5826 * Fixed common compiler crashes with SSE instruction set enabled
5827 (implied by -march=pentium3, pentium4, athlon-xp)
5828 * __m128 and __m128i is not 128bit aligned when used in structures.
5830 x86-64
5832 * A bug whereby the compiler could generate bad code for bzero has
5833 been fixed.
5834 * ABI fixes (implying ABI incompatibilities with previous version in
5835 some corner cases)
5836 * Fixed prefetch code generation
5837 _________________________________________________________________
5839 GCC 3.2.3
5841 3.2.3 is a bug fix release only; there are no new features that were
5842 not present in GCC 3.2.2.
5844 Bug Fixes
5846 This section lists the problem reports (PRs) from GCC's bug tracking
5847 system that are known to be fixed in the 3.2.3 release. This list
5848 might not be complete (that is, it is possible that some PRs that have
5849 been fixed are not listed here), and some of the titles have been
5850 changed to make them more clear.
5852 Internal Compiler Errors (multi-platform)
5854 * [4]3782: (c++) -quiet -fstats produces a segmentation fault in
5855 cc1plus
5856 * [5]6440: (c++) template specializations cause ICE
5857 * [6]7050: (c++) ICE on: (i ? get_string() : throw)
5858 * [7]7741: ICE on conflicting types (make_decl_rtl in varasm.c)
5859 * [8]7982: (c++) ICE due to infinite recursion (using STL set)
5860 * [9]8068: exceedingly high (infinite) memory usage
5861 * [10]8178: ICE with __builtin_ffs
5862 * [11]8396: ICE in copy_to_mode_reg, in explow.c
5863 * [12]8674: (c++) ICE in cp_expr_size, in cp/cp-lang.c
5864 * [13]9768: ICE when optimizing inline code at -O2
5865 * [14]9798: (c++) Infinite recursion (segfault) in
5866 cp/decl.c:push_using_directive with recursive using directives
5867 * [15]9799: mismatching structure initializer with nested flexible
5868 array member: ICE
5869 * [16]9928: ICE on duplicate enum declaration
5870 * [17]10114: ICE in mem_loc_descriptor, in dwarf2out.c (affects
5871 sparc, alpha)
5872 * [18]10352: ICE in find_reloads_toplev
5873 * [19]10336: ICE with -Wunreachable-code
5875 C/optimizer bugs:
5877 * [20]8224: Incorrect joining of signed and unsigned division
5878 * [21]8613: -O2 produces wrong code with builtin strlen and
5879 postincrements
5880 * [22]8828: gcc reports some code is unreachable when it is not
5881 * [23]9226: GCSE breaking argument passing
5882 * [24]9853: miscompilation of non-constant structure initializer
5883 * [25]9797: C99-style struct initializers are miscompiled
5884 * [26]9967: Some standard C function calls should not be replaced
5885 when optimizing for size
5886 * [27]10116: ce2: invalid merge of join_bb in the context of switch
5887 statements
5888 * [28]10171: wrong code for inlined function
5889 * [29]10175: -Wunreachable-code doesn't work for single lines
5891 C++ compiler and library:
5893 * [30]8316: Confusing diagnostic for code that misuses conversion
5894 operators
5895 * [31]9169: filebuf output fails if codecvt<>::out returns noconv
5896 * [32]9420: incomplete type incorrectly reported
5897 * [33]9459: typeof in return type specification of template not
5898 supported
5899 * [34]9507: filebuf::open handles ios_base::ate incorrectly
5900 * [35]9538: Out-of-bounds memory access in streambuf::sputbackc
5901 * [36]9602: Total confusion about template/friend/virtual/abstract
5902 * [37]9993: destructor not called for local object created within
5903 and returned from infinite loop
5904 * [38]10167: ieee_1003.1-2001 locale specialisations on a
5905 glibc-2.3.2 system
5907 Java compiler and library:
5909 * [39]9652: libgcj build fails on irix6.5.1[78]
5910 * [40]10144: gas on solaris complains about bad .stabs lines for
5911 java, native as unaffected
5913 x86-specific (Intel/AMD):
5915 * [41]8746: gcc miscompiles Linux kernel ppa driver on x86
5916 * [42]9888: -mcpu=k6 -Os produces out of range loop instructions
5917 * [43]9638: Cross-build for target i386-elf and i586-pc-linux-gnu
5918 failed
5919 * [44]9954: Cross-build for target i586-pc-linux-gnu (--with-newlib)
5920 failed
5922 SPARC-specific:
5924 * [45]7784: [Sparc] ICE in extract_insn, in recog.c
5925 * [46]7796: sparc extra failure with -m64 on execute/930921-1.c in
5926 unroll.c
5927 * [47]8281: ICE when compiling with -O2 -fPIC for Ultrasparc
5928 * [48]8366: [Sparc] C testsuite failure with -m64 -fpic -O in
5929 execute/loop-2d.c
5930 * [49]8726: gcc -O2 miscompiles Samba 2.2.7 on 32-bit sparc
5931 * [50]9414: Scheduling bug on Ultrasparc
5932 * [51]10067: GCC-3.2.2 outputs invalid asm on sparc64
5934 m68k-specific:
5936 * [52]7248: broken "inclusive or" code
5937 * [53]8343: m68k-elf/rtems ICE at instantiate_virtual_regs_1
5939 PowerPC-specific:
5941 * [54]9732: Wrong code with -O2 -fPIC
5942 * [55]10073: ICE: powerpc cannot split insn
5944 Alpha-specific:
5946 * [56]7702: optimization problem on a DEC alpha under OSF1
5947 * [57]9671: gcc.3.2.2 does not build on a HP Tru64 Unix v5.1B system
5949 HP-specific:
5951 * [58]8694: <string> breaks <ctype.h> on HP-UX 10.20 (DUP: 9275)
5952 * [59]9953: (ada) gcc 3.2.x can't build 3.3-branch ada on HP-UX 10
5953 (missing symbol)
5954 * [60]10271: Floating point args don't get reloaded across function
5955 calls with -O2
5957 MIPS specific:
5959 * [61]6362: mips-irix6 gcc-3.1 C testsuite failure with -mips4 in
5960 compile/920501-4.c
5962 CRIS specific:
5964 * [62]10377: gcc-3.2.2 creates bad assembler code for cris
5966 Miscellaneous and minor bugs:
5968 * [63]6955: collect2 says "core dumped" when there is no core
5969 _________________________________________________________________
5971 GCC 3.2.2
5973 Beginning with 3.2.2, GCC's Makefile suite supports redirection of
5974 make install by means of the DESTDIR variable. Parts of the GCC tree
5975 have featured that support long before, but now it is available even
5976 from the top level.
5978 Other than that, GCC 3.2.2 is a bug fix release only; there are no new
5979 features that were not present in GCC 3.2.1.
5981 Bug Fixes
5983 On the following i386-based systems GCC 3.2.1 broke the C ABI wrt.
5984 functions returning structures: Cygwin, FreeBSD (GCC 3.2.1 as shipped
5985 with FreeBSD 5.0 does not have this problem), Interix, a.out-based
5986 Linux and NetBSD, OpenBSD, and Darwin. GCC 3.2.2 reverts this ABI
5987 change, and thus restores ABI-compatibility with previous releases
5988 (except GCC 3.2.1) on these platforms.
5990 This section lists the problem reports (PRs) from GCC's bug tracking
5991 system that are known to be fixed in the 3.2.2 release. This list
5992 might not be complete (that is, it is possible that some PRs that have
5993 been fixed are not listed here) and some of the titles have been
5994 changed to make them more clear.
5996 Internal Compiler Errors (multi-platform)
5998 * [64]5919: (c++) ICE when passing variable array to template
5999 function
6000 * [65]7129: (c++) ICE with min/max assignment operators (<?= and
6001 >?=)
6002 * [66]7507: ICE with -O2 when address of called function is a
6003 complicated expression
6004 * [67]7622: ICE with nested inline functions if function's address
6005 is taken
6006 * [68]7681: (fortran) ICE in compensate_edge, in reg-stack.c (also
6007 PR [69]9258)
6008 * [70]8031: (c++) ICE in code comparing typeids and casting from
6009 virtual base
6010 * [71]8275: ICE in simplify_subreg
6011 * [72]8332: (c++) builtin strlen/template interaction causes ICE
6012 * [73]8372: (c++) ICE on explicit call of destructor
6013 * [74]8439: (c, not c++) empty struct causes ICE
6014 * [75]8442: (c++) ICE with nested template classes
6015 * [76]8518: ICE when compiling mplayer ("extern inline" issue)
6016 * [77]8615: (c++) ICE with out-of-range character constant template
6017 argument
6018 * [78]8663: (c++) ICE in cp_expr_size, at cp-lang.c:307
6019 * [79]8799: (c++) ICE: error reporting routines re-entered
6020 * [80]9328: (c++) ICE with typeof(X) for overloaded X
6021 * [81]9465: (preprocessor) cpp -traditional ICE on null bytes
6023 C++ (compiler and library) bugs
6025 * [82]47: scoping in nested classes is broken
6026 * [83]6745: problems with iostream rdbuf() member function
6027 * [84]8214: conversion from const char* const to char* sometimes
6028 accepted illegally
6029 * [85]8493: builtin strlen and overload resolution (same bug as
6030 [86]8332)
6031 * [87]8503: strange behaviour of function types
6032 * [88]8727: compiler confused by inheritance from an anonymous
6033 struct
6034 * [89]7445: poor performance of std::locale::classic() in
6035 multi-threaded applications
6036 * [90]8230: mishandling of overflow in vector<T>::resize
6037 * [91]8399: sync_with_stdio(false) breaks unformatted input
6038 * [92]8662: illegal access of private member of unnamed class is
6039 accepted
6040 * [93]8707: "make distclean" fails in libstdc++-v3 directory
6041 * [94]8708: __USE_MALLOC doesn't work
6042 * [95]8790: Use of non-thread-safe strtok in src/localename.cc
6043 * [96]8887: Bug in date formats with --enable-clocale=generic
6044 * [97]9076: Call Frame Instructions are not handled correctly during
6045 unwind operation
6046 * [98]9151: std::setprecision limited to 16 digits when outputting a
6047 double to a stream
6048 * [99]9168: codecvt<char, char, mbstate_t> overwrites output buffers
6049 * [100]9269: libstdc++ headers: explicit specialization of function
6050 must precede its first use
6051 * [101]9322: return value of basic_streambuf<>::getloc affected by
6052 locale::global
6053 * [102]9433: segfault in runtime support for dynamic_cast
6055 C and optimizer bugs
6057 * [103]8032: GCC incorrectly initializes static structs that have
6058 flexible arrays
6059 * [104]8639: simple arithmetic expression broken
6060 * [105]8794: optimization improperly eliminates certain expressions
6061 * [106]8832: traditional "asm volatile" code is illegally optimized
6062 * [107]8988: loop optimizer bug: with -O2, code is generated that
6063 segfaults (found on i386, bug present for all platforms)
6064 * [108]9492: structure copy clobbers subsequent stores to structure
6066 Objective-C bugs
6068 * [109]9267: Objective-C parser won't build with newer bison
6069 versions (e.g. 1.875)
6071 Ada bugs
6073 * [110]8344: Ada build problem due to conflict between gcc/final.o,
6074 gcc/ada/final.o
6076 Preprocessor bugs
6078 * [111]8524: _Pragma within macros is improperly expanded
6079 * [112]8880: __WCHAR_TYPE__ macro incorrectly set to "long int" with
6080 -fshort-wchar
6082 ARM-specific
6084 * [113]9090: arm ICE with >= -O2; regression from gcc-2.95
6086 x86-specific (Intel/AMD)
6088 * [114]8588: ICE in extract_insn, at recog.c:NNNN (shift
6089 instruction)
6090 * [115]8599: loop unroll bug with -march=k6-3
6091 * [116]9506: ABI breakage in structure return (affects BSD and
6092 Cygwin, but not GNU/Linux)
6094 FreeBSD 5.0 specific
6096 * [117]9484: GCC 3.2.1 Bootstrap failure on FreeBSD 5.0
6098 RTEMS-specific
6100 * [118]9292: hppa1.1-rtems configurery problems
6101 * [119]9293: [m68k-elf/rtems] config/m68k/t-crtstuff bug
6102 * [120]9295: [mips-rtems] config/mips/rtems.h init/fini issue
6103 * [121]9296: gthr-rtems regression
6104 * [122]9316: powerpc-rtems: extending multilibs
6106 HP-PA specific
6108 * [123]9493: ICE with -O2 when building a simple function
6110 Documentation
6112 * [124]7341: hyperlink to gcov in GCC documentation doesn't work
6113 * [125]8947: Please add a warning about "-malign-double" in docs
6114 * [126]7448, [127]8882: typo cleanups
6115 _________________________________________________________________
6117 GCC 3.2.1
6119 3.2.1 adds a new warning, -Wabi. This option warns when GNU C++
6120 generates code that is known not to be binary-compatible with the
6121 vendor-neutral ia32/ia64 ABI. Please consult the GCC manual, included
6122 in the distribution, for details.
6124 This release also removes an old GCC extension, "naming types", and
6125 the documentation now directs users to use a different GCC extension,
6126 __typeof__, instead. The feature had evidently been broken for a
6127 while.
6129 Otherwise, 3.2.1 is a bug fix release only; other than bug fixes and
6130 the new warning there are no new features that were not present in GCC
6131 3.2.
6133 In addition, the previous fix for [128]PR 7445 (poor performance of
6134 std::locale::classic() in multi-threaded applications) was reverted
6135 ("unfixed"), because the "fix" was not thread-safe.
6137 Bug Fixes
6139 This section lists the problem reports (PRs) from GCC's bug tracking
6140 system that are known to be fixed in the 3.2.1 release. This list
6141 might not be complete (that is, it is possible that some PRs that have
6142 been fixed are not listed here). As you can see, the number of bug
6143 fixes is quite large, so it is strongly recommended that users of
6144 earlier GCC 3.x releases upgrade to GCC 3.2.1.
6146 Internal Compiler Errors (multi-platform)
6148 * [129]2521: (c++) ICE in build_ptrmemfunc, in cp/typeck.c
6149 * [130]5661: (c++) ICE instantiating template on array of unknown
6150 size (bad code)
6151 * [131]6419: (c++) ICE in make_decl_rtl for "longest" attribute on
6152 64-bit platforms
6153 * [132]6994: (c++) ICE in find_function_data
6154 * [133]7150: preprocessor: GCC -dM -E gives an ICE
6155 * [134]7160: ICE when optimizing branches without a return value
6156 * [135]7228: (c++) ICE when using member template and template
6157 function
6158 * [136]7266: (c++) ICE with -pedantic on missing typename
6159 * [137]7353: ICE from use of "Naming Types" extension, see above
6160 * [138]7411: ICE in instantiate_virtual_regs_1, in function.c
6161 * [139]7478: (c++) ICE on static_cast inside template
6162 * [140]7526: preprocessor core dump when _Pragma implies #pragma
6163 dependency
6164 * [141]7721: (c++) ICE on simple (but incorrect) template ([142]7803
6165 is a duplicate)
6166 * [143]7754: (c++) ICE on union with template parameter
6167 * [144]7788: (c++) redeclaring a definition as an incomplete class
6168 causes ICE
6169 * [145]8031: (c++) ICE in comptypes, in cp/typeck.c
6170 * [146]8055: preprocessor dies with SIG11 when building FreeBSD
6171 kernel
6172 * [147]8067: (c++) ICE due to mishandling of __FUNCTION__ and
6173 related variables
6174 * [148]8134: (c++) ICE in force_store_init_value on legal code
6175 * [149]8149: (c++) ICE on incomplete type
6176 * [150]8160: (c++) ICE in build_modify_expr, in cp/typeck.c: array
6177 initialization
6179 C++ (compiler and library) bugs
6181 * [151]5607: No pointer adjustment in covariant return types
6182 * [152]6579: Infinite loop with statement expressions in member
6183 initialization
6184 * [153]6803: Default copy constructor bug in GCC 3.1
6185 * [154]7176: g++ confused by friend and static member with same name
6186 * [155]7188: Segfault with template class and recursive (incorrect)
6187 initializer list
6188 * [156]7306: Regression: GCC 3.x fails to compile code with virtual
6189 inheritance if a method has a variable number of arguments
6190 * [157]7461: ctype<char>::classic_table() returns offset array on
6191 Cygwin
6192 * [158]7524: f(const float arg[3]) fails
6193 * [159]7584: Erroneous ambiguous base error on using declaration
6194 * [160]7676: Member template overloading problem
6195 * [161]7679: infinite loop when a right parenthesis is missing
6196 * [162]7811: default locale not taken from environment
6197 * [163]7961: compare( char *) implemented incorrectly in
6198 basic_string<>
6199 * [164]8071: basic_ostream::operator<<(streambuf*) loops forever if
6200 streambuf::underflow() leaves gptr() NULL (dups: [165]8127,
6201 [166]6745)
6202 * [167]8096: deque::at() throws std::range_error instead of
6203 std::out_of_range
6204 * [168]8127: cout << cin.rdbuf() infinite loop
6205 * [169]8218: Excessively large memory consumed for classes with
6206 large array members
6207 * [170]8287: GCC 3.2: Destructor called for non-constructed local
6208 object
6209 * [171]8347: empty vector range used in string construction causes
6210 core dump
6211 * [172]8348: fail() flag is set in istringstream when eof() flag is
6212 set
6213 * [173]8391: regression: infinite loop in cp/decl2.c(finish_file)
6215 C and optimizer bugs
6217 * [174]6627: -fno-align-functions doesn't seem to disable function
6218 alignment
6219 * [175]6631: life_analysis misoptimizes code to initialize fields of
6220 a structure
6221 * [176]7102: unsigned char division results in floating exception
6222 * [177]7120: Run once loop should *always* be unrolled
6223 (pessimization)
6224 * [178]7209: Bug involving array referencing and ?: operator
6225 * [179]7515: invalid inlining of global function with -O3
6226 * [180]7814: incorrect scheduling for glibc-2.2.92 strcpy test
6227 * [181]8467: bug in sibling call optimization
6229 Preprocessor bugs
6231 * [182]4890: incorrect line markers from the traditional
6232 preprocessor
6233 * [183]7357: -M option omits system headers files (making it the
6234 same as -MM)
6235 * [184]7358: Changes to Sun's make Dependencies
6236 * [185]7602: C++ header files found in CPLUS_INCLUDE_PATH treated as
6237 C headers
6238 * [186]7862: Interrupting GCC -MD removes .d file but not .o
6239 * [187]8190: Failed compilation deletes -MD dependency file
6240 * [188]8524: _Pragma within macro is improperly expanded
6242 x86 specific (Intel/AMD)
6244 * [189]5351: (i686-only) function pass-by-value structure copy
6245 corrupts stack ([190]7591 is a duplicate)
6246 * [191]6845, [192]7034, [193]7124, [194]7174: ICE's with
6247 -march=pentium3/pentium2/athlon (these are all the same underlying
6248 bug, in MMX register use)
6249 * [195]7134, [196]7375, [197]7390: ICE with -march=athlon (maybe
6250 same as above?)
6251 * [198]6890: xmmintrin.h, _MM_TRANSPOSE4_PS is broken
6252 * [199]6981: wrong code in 64-bit manipulation on x86
6253 * [200]7242: GCC -mcpu=pentium[23] doesn't define
6254 __tune_pentiumpro__ macro
6255 * [201]7396: ix86: cmpgt_ss, cmpge_ss, cmpngt_ss, and cmpnge_ss SSE
6256 intrinsics are broken
6257 * [202]7630: GCC 3.2 breaks on Mozilla 1.0's JS sources with
6258 -march=pentium4
6259 * [203]7693: Typo in i386 mmintrin.h header
6260 * [204]7723: ICE - Pentium3 sse - GCC 3.2
6261 * [205]7951: ICE on -march=pentium4 -O2 -mfpmath=sse
6262 * [206]8146: (i686 only) gcc 3.2 miscompiles gcc 2.95.3
6264 PowerPC specific
6266 * [207]5967: GCC bug when profiling nested functions on powerpc
6267 * [208]6984: wrong code generated with -O2, -O3, -Os for do-while
6268 loop on PowerPC
6269 * [209]7114: PowerPC: ICE building strcoll.op from glibc-2.2.5
6270 * [210]7130: miscompiled code for GCC-3.1 in powerpc linux with
6271 -funroll-all-loops
6272 * [211]7133: PowerPC ICE: unrecognizable insn
6273 * [212]7380: ICE in extract_insn, at recog.c:2148
6274 * [213]8252: ICE on Altivec code with optimization turned on
6275 * [214]8451: Altivec ICE in GCC 3.2
6277 HP/PA specific
6279 * [215]7250: __ashrdi3 returns wrong value on 32 bit hppa
6281 SPARC specific
6283 * [216]6668: when using --disable-multilib, libgcc_s.so is installed
6284 in the wrong place on sparc-solaris
6285 * [217]7151: ICE when compiling for UltraSPARC
6286 * [218]7335: SPARC: ICE in verify_wide_reg (flow.c:557) with long
6287 double and -O1
6288 * [219]7842: [REGRESSION] SPARC code gen bug
6290 ARM specific
6292 * [220]7856: [arm] invalid offset in constant pool reference
6293 * [221]7967: optimization produces wrong code (ARM)
6295 Alpha specific
6297 * [222]7374: __builtin_fabsl broken on alpha
6299 IBM s390 specific
6301 * [223]7370: ICE in fixup_var_refs_1 on s390x
6302 * [224]7409: loop optimization bug on s390x-linux-gnu
6303 * [225]8232: s390x: ICE when using bcmp with int length argument
6305 SCO specific
6307 * [226]7623: SCO OpenServer build fails with machmode.def: undefined
6308 symbol: BITS_PER_UNIT
6310 m68k/Coldfire specific
6312 * [227]8314: crtbegin, crtend need to be multilib'ed for this
6313 platform
6315 Documentation
6317 * [228]761: Document some undocumented options
6318 * [229]5610: Fix documentation about invoking SSE instructions
6319 (-mfpmath=sse)
6320 * [230]7484: List -Wmissing-declarations as C-only option
6321 * [231]7531: -mcmodel not documented for x86-64
6322 * [232]8120: Update documentation of bad use of ##
6323 _________________________________________________________________
6325 GCC 3.2
6327 3.2 is a small bug fix release, but there is a change to the
6328 application binary interface (ABI), hence the change to the second
6329 part of the version number.
6331 The main purpose of the 3.2 release is to correct a couple of problems
6332 in the C++ ABI, with the intention of providing a stable interface
6333 going forward. Accordingly, 3.2 is only a small change to 3.1.1.
6335 Bug Fixes
6337 C++
6339 * [233]7320: g++ 3.2 relocation problem
6340 * [234]7470: vtable: virtual function pointers not in declaration
6341 order
6343 libstdc++
6345 * [235]6410: Trouble with non-ASCII monetary symbols and wchar_t
6346 * [236]6503, [237]6642, [238]7186: Problems with comparing or
6347 subtracting various types of const and non-const iterators
6348 * [239]7216: ambiguity with basic_iostream::traits_type
6349 * [240]7220: problem with basic_istream::ignore(0,delimiter)
6350 * [241]7222: locale::operator==() doesn't work on std::locale("")
6351 * [242]7286: placement operator delete issue
6352 * [243]7442: cxxabi.h does not match the C++ ABI
6353 * [244]7445: poor performance of std::locale::classic() in
6354 multi-threaded applications
6356 x86-64 specific
6358 * [245]7291: off-by-one in generated inline bzero code for x86-64
6360 Please send FSF & GNU inquiries & questions to [246]gnu@gnu.org. There
6361 are also [247]other ways to contact the FSF.
6363 These pages are maintained by [248]the GCC team.
6366 For questions related to the use of GCC, please consult these web
6367 pages and the [249]GCC manuals. If that fails, the
6368 [250]gcc-help@gcc.gnu.org mailing list might help.
6369 Please send comments on these web pages and the development of GCC
6370 to our developer mailing list at [251]gcc@gnu.org or
6371 [252]gcc@gcc.gnu.org. All of our lists have [253]public archives.
6373 Copyright (C) Free Software Foundation, Inc., 51 Franklin St, Fifth
6374 Floor, Boston, MA 02110, USA.
6376 Verbatim copying and distribution of this entire article is permitted
6377 in any medium, provided this notice is preserved.
6378 Last modified 2006-06-21 [254]Valid XHTML 1.0
6380 References
6382 1. http://gcc.gnu.org/gcc-3.2/changes.html#3.2.3
6383 2. http://gcc.gnu.org/gcc-3.1/changes.html
6384 3. http://gcc.gnu.org/gcc-3.2/c++-abi.html
6385 4. http://gcc.gnu.org/PR3782
6386 5. http://gcc.gnu.org/PR6440
6387 6. http://gcc.gnu.org/PR7050
6388 7. http://gcc.gnu.org/PR7741
6389 8. http://gcc.gnu.org/PR7982
6390 9. http://gcc.gnu.org/PR8068
6391 10. http://gcc.gnu.org/PR8178
6392 11. http://gcc.gnu.org/PR8396
6393 12. http://gcc.gnu.org/PR8674
6394 13. http://gcc.gnu.org/PR9768
6395 14. http://gcc.gnu.org/PR9798
6396 15. http://gcc.gnu.org/PR9799
6397 16. http://gcc.gnu.org/PR9928
6398 17. http://gcc.gnu.org/PR10114
6399 18. http://gcc.gnu.org/PR10352
6400 19. http://gcc.gnu.org/PR10336
6401 20. http://gcc.gnu.org/PR8224
6402 21. http://gcc.gnu.org/PR8613
6403 22. http://gcc.gnu.org/PR8828
6404 23. http://gcc.gnu.org/PR9226
6405 24. http://gcc.gnu.org/PR9853
6406 25. http://gcc.gnu.org/PR9797
6407 26. http://gcc.gnu.org/PR9967
6408 27. http://gcc.gnu.org/PR10116
6409 28. http://gcc.gnu.org/PR10171
6410 29. http://gcc.gnu.org/PR10175
6411 30. http://gcc.gnu.org/PR8316
6412 31. http://gcc.gnu.org/PR9169
6413 32. http://gcc.gnu.org/PR9420
6414 33. http://gcc.gnu.org/PR9459
6415 34. http://gcc.gnu.org/PR9507
6416 35. http://gcc.gnu.org/PR9538
6417 36. http://gcc.gnu.org/PR9602
6418 37. http://gcc.gnu.org/PR9993
6419 38. http://gcc.gnu.org/PR10167
6420 39. http://gcc.gnu.org/PR9652
6421 40. http://gcc.gnu.org/PR10144
6422 41. http://gcc.gnu.org/PR8746
6423 42. http://gcc.gnu.org/PR9888
6424 43. http://gcc.gnu.org/PR9638
6425 44. http://gcc.gnu.org/PR9954
6426 45. http://gcc.gnu.org/PR7784
6427 46. http://gcc.gnu.org/PR7796
6428 47. http://gcc.gnu.org/PR8281
6429 48. http://gcc.gnu.org/PR8366
6430 49. http://gcc.gnu.org/PR8726
6431 50. http://gcc.gnu.org/PR9414
6432 51. http://gcc.gnu.org/PR10067
6433 52. http://gcc.gnu.org/PR7248
6434 53. http://gcc.gnu.org/PR8343
6435 54. http://gcc.gnu.org/PR9732
6436 55. http://gcc.gnu.org/PR10073
6437 56. http://gcc.gnu.org/PR7702
6438 57. http://gcc.gnu.org/PR9671
6439 58. http://gcc.gnu.org/PR8694
6440 59. http://gcc.gnu.org/PR9953
6441 60. http://gcc.gnu.org/PR10271
6442 61. http://gcc.gnu.org/PR6362
6443 62. http://gcc.gnu.org/PR10377
6444 63. http://gcc.gnu.org/PR6955
6445 64. http://gcc.gnu.org/PR5919
6446 65. http://gcc.gnu.org/PR7129
6447 66. http://gcc.gnu.org/PR7507
6448 67. http://gcc.gnu.org/PR7622
6449 68. http://gcc.gnu.org/PR7681
6450 69. http://gcc.gnu.org/PR9528
6451 70. http://gcc.gnu.org/PR8031
6452 71. http://gcc.gnu.org/PR8275
6453 72. http://gcc.gnu.org/PR8332
6454 73. http://gcc.gnu.org/PR8372
6455 74. http://gcc.gnu.org/PR8439
6456 75. http://gcc.gnu.org/PR8442
6457 76. http://gcc.gnu.org/PR8518
6458 77. http://gcc.gnu.org/PR8615
6459 78. http://gcc.gnu.org/PR8663
6460 79. http://gcc.gnu.org/PR8799
6461 80. http://gcc.gnu.org/PR9328
6462 81. http://gcc.gnu.org/PR9465
6463 82. http://gcc.gnu.org/PR47
6464 83. http://gcc.gnu.org/PR6745
6465 84. http://gcc.gnu.org/PR8214
6466 85. http://gcc.gnu.org/PR8493
6467 86. http://gcc.gnu.org/PR8332
6468 87. http://gcc.gnu.org/PR8503
6469 88. http://gcc.gnu.org/PR8727
6470 89. http://gcc.gnu.org/PR7445
6471 90. http://gcc.gnu.org/PR8230
6472 91. http://gcc.gnu.org/PR8399
6473 92. http://gcc.gnu.org/PR8662
6474 93. http://gcc.gnu.org/PR8707
6475 94. http://gcc.gnu.org/PR8708
6476 95. http://gcc.gnu.org/PR8790
6477 96. http://gcc.gnu.org/PR8887
6478 97. http://gcc.gnu.org/PR9076
6479 98. http://gcc.gnu.org/PR9151
6480 99. http://gcc.gnu.org/PR9168
6481 100. http://gcc.gnu.org/PR9269
6482 101. http://gcc.gnu.org/PR9322
6483 102. http://gcc.gnu.org/PR9433
6484 103. http://gcc.gnu.org/PR8032
6485 104. http://gcc.gnu.org/PR8639
6486 105. http://gcc.gnu.org/PR8794
6487 106. http://gcc.gnu.org/PR8832
6488 107. http://gcc.gnu.org/PR8988
6489 108. http://gcc.gnu.org/PR9492
6490 109. http://gcc.gnu.org/PR9267
6491 110. http://gcc.gnu.org/PR8344
6492 111. http://gcc.gnu.org/PR8524
6493 112. http://gcc.gnu.org/PR8880
6494 113. http://gcc.gnu.org/PR9090
6495 114. http://gcc.gnu.org/PR8588
6496 115. http://gcc.gnu.org/PR8599
6497 116. http://gcc.gnu.org/PR9506
6498 117. http://gcc.gnu.org/PR9484
6499 118. http://gcc.gnu.org/PR9292
6500 119. http://gcc.gnu.org/PR9293
6501 120. http://gcc.gnu.org/PR9295
6502 121. http://gcc.gnu.org/PR9296
6503 122. http://gcc.gnu.org/PR9316
6504 123. http://gcc.gnu.org/PR9493
6505 124. http://gcc.gnu.org/PR7341
6506 125. http://gcc.gnu.org/PR8947
6507 126. http://gcc.gnu.org/PR7448
6508 127. http://gcc.gnu.org/PR8882
6509 128. http://gcc.gnu.org/PR7445
6510 129. http://gcc.gnu.org/PR2521
6511 130. http://gcc.gnu.org/PR5661
6512 131. http://gcc.gnu.org/PR6419
6513 132. http://gcc.gnu.org/PR6994
6514 133. http://gcc.gnu.org/PR7150
6515 134. http://gcc.gnu.org/PR7160
6516 135. http://gcc.gnu.org/PR7228
6517 136. http://gcc.gnu.org/PR7266
6518 137. http://gcc.gnu.org/PR7353
6519 138. http://gcc.gnu.org/PR7411
6520 139. http://gcc.gnu.org/PR7478
6521 140. http://gcc.gnu.org/PR7526
6522 141. http://gcc.gnu.org/PR7721
6523 142. http://gcc.gnu.org/PR7803
6524 143. http://gcc.gnu.org/PR7754
6525 144. http://gcc.gnu.org/PR7788
6526 145. http://gcc.gnu.org/PR8031
6527 146. http://gcc.gnu.org/PR8055
6528 147. http://gcc.gnu.org/PR8067
6529 148. http://gcc.gnu.org/PR8134
6530 149. http://gcc.gnu.org/PR8149
6531 150. http://gcc.gnu.org/PR8160
6532 151. http://gcc.gnu.org/PR5607
6533 152. http://gcc.gnu.org/PR6579
6534 153. http://gcc.gnu.org/PR6803
6535 154. http://gcc.gnu.org/PR7176
6536 155. http://gcc.gnu.org/PR7188
6537 156. http://gcc.gnu.org/PR7306
6538 157. http://gcc.gnu.org/PR7461
6539 158. http://gcc.gnu.org/PR7524
6540 159. http://gcc.gnu.org/PR7584
6541 160. http://gcc.gnu.org/PR7676
6542 161. http://gcc.gnu.org/PR7679
6543 162. http://gcc.gnu.org/PR7811
6544 163. http://gcc.gnu.org/PR7961
6545 164. http://gcc.gnu.org/PR8071
6546 165. http://gcc.gnu.org/PR8127
6547 166. http://gcc.gnu.org/PR6745
6548 167. http://gcc.gnu.org/PR8096
6549 168. http://gcc.gnu.org/PR8127
6550 169. http://gcc.gnu.org/PR8218
6551 170. http://gcc.gnu.org/PR8287
6552 171. http://gcc.gnu.org/PR8347
6553 172. http://gcc.gnu.org/PR8348
6554 173. http://gcc.gnu.org/PR8391
6555 174. http://gcc.gnu.org/PR6627
6556 175. http://gcc.gnu.org/PR6631
6557 176. http://gcc.gnu.org/PR7102
6558 177. http://gcc.gnu.org/PR7120
6559 178. http://gcc.gnu.org/PR7209
6560 179. http://gcc.gnu.org/PR7515
6561 180. http://gcc.gnu.org/PR7814
6562 181. http://gcc.gnu.org/PR8467
6563 182. http://gcc.gnu.org/PR4890
6564 183. http://gcc.gnu.org/PR7357
6565 184. http://gcc.gnu.org/PR7358
6566 185. http://gcc.gnu.org/PR7602
6567 186. http://gcc.gnu.org/PR7862
6568 187. http://gcc.gnu.org/PR8190
6569 188. http://gcc.gnu.org/PR8524
6570 189. http://gcc.gnu.org/PR5351
6571 190. http://gcc.gnu.org/PR7591
6572 191. http://gcc.gnu.org/PR6845
6573 192. http://gcc.gnu.org/PR7034
6574 193. http://gcc.gnu.org/PR7124
6575 194. http://gcc.gnu.org/PR7174
6576 195. http://gcc.gnu.org/PR7134
6577 196. http://gcc.gnu.org/PR7375
6578 197. http://gcc.gnu.org/PR7390
6579 198. http://gcc.gnu.org/PR6890
6580 199. http://gcc.gnu.org/PR6981
6581 200. http://gcc.gnu.org/PR7242
6582 201. http://gcc.gnu.org/PR7396
6583 202. http://gcc.gnu.org/PR7630
6584 203. http://gcc.gnu.org/PR7693
6585 204. http://gcc.gnu.org/PR7723
6586 205. http://gcc.gnu.org/PR7951
6587 206. http://gcc.gnu.org/PR8146
6588 207. http://gcc.gnu.org/PR5967
6589 208. http://gcc.gnu.org/PR6984
6590 209. http://gcc.gnu.org/PR7114
6591 210. http://gcc.gnu.org/PR7130
6592 211. http://gcc.gnu.org/PR7133
6593 212. http://gcc.gnu.org/PR7380
6594 213. http://gcc.gnu.org/PR8252
6595 214. http://gcc.gnu.org/PR8451
6596 215. http://gcc.gnu.org/PR7250
6597 216. http://gcc.gnu.org/PR6668
6598 217. http://gcc.gnu.org/PR7151
6599 218. http://gcc.gnu.org/PR7335
6600 219. http://gcc.gnu.org/PR7842
6601 220. http://gcc.gnu.org/PR7856
6602 221. http://gcc.gnu.org/PR7967
6603 222. http://gcc.gnu.org/PR7374
6604 223. http://gcc.gnu.org/PR7370
6605 224. http://gcc.gnu.org/PR7409
6606 225. http://gcc.gnu.org/PR8232
6607 226. http://gcc.gnu.org/PR7623
6608 227. http://gcc.gnu.org/PR8314
6609 228. http://gcc.gnu.org/PR761
6610 229. http://gcc.gnu.org/PR5610
6611 230. http://gcc.gnu.org/PR7484
6612 231. http://gcc.gnu.org/PR7531
6613 232. http://gcc.gnu.org/PR8120
6614 233. http://gcc.gnu.org/PR7320
6615 234. http://gcc.gnu.org/PR7470
6616 235. http://gcc.gnu.org/PR6410
6617 236. http://gcc.gnu.org/PR6503
6618 237. http://gcc.gnu.org/PR6642
6619 238. http://gcc.gnu.org/PR7186
6620 239. http://gcc.gnu.org/PR7216
6621 240. http://gcc.gnu.org/PR7220
6622 241. http://gcc.gnu.org/PR7222
6623 242. http://gcc.gnu.org/PR7286
6624 243. http://gcc.gnu.org/PR7442
6625 244. http://gcc.gnu.org/PR7445
6626 245. http://gcc.gnu.org/PR7291
6627 246. mailto:gnu@gnu.org
6628 247. http://www.gnu.org/home.html#ContactInfo
6629 248. http://gcc.gnu.org/about.html
6630 249. http://gcc.gnu.org/onlinedocs/
6631 250. mailto:gcc-help@gcc.gnu.org
6632 251. mailto:gcc@gnu.org
6633 252. mailto:gcc@gcc.gnu.org
6634 253. http://gcc.gnu.org/lists.html
6635 254. http://validator.w3.org/check/referer
6636 ======================================================================
6637 http://gcc.gnu.org/gcc-3.1/index.html
6639 GCC 3.1
6641 July 27, 2002
6643 The [1]GNU project and the GCC developers are pleased to announce the
6644 release of GCC 3.1.1.
6646 The links below still apply to GCC 3.1.1.
6648 May 15, 2002
6650 The [2]GNU project and the GCC developers are pleased to announce the
6651 release of GCC 3.1.
6653 GCC used to stand for the GNU C Compiler, but since the compiler
6654 supports several other languages aside from C, it now stands for the
6655 GNU Compiler Collection.
6657 A list of [3]successful builds is updated as new information becomes
6658 available.
6660 The GCC developers would like to thank the numerous people that have
6661 contributed [4]new features, improvements, bug fixes, and other
6662 changes as well as test results to GCC. This [5]amazing group of
6663 volunteers is what makes GCC successful.
6665 For additional information about GCC please refer to the [6]GCC
6666 project web site or contact the [7]GCC development mailing list.
6668 To obtain GCC please use [8]our mirror sites, one of the [9]GNU mirror
6669 sites, or our CVS server.
6670 _________________________________________________________________
6672 Please send FSF & GNU inquiries & questions to [10]gnu@gnu.org. There
6673 are also [11]other ways to contact the FSF.
6675 These pages are maintained by [12]the GCC team.
6678 For questions related to the use of GCC, please consult these web
6679 pages and the [13]GCC manuals. If that fails, the
6680 [14]gcc-help@gcc.gnu.org mailing list might help.
6681 Please send comments on these web pages and the development of GCC
6682 to our developer mailing list at [15]gcc@gnu.org or
6683 [16]gcc@gcc.gnu.org. All of our lists have [17]public archives.
6685 Copyright (C) Free Software Foundation, Inc., 51 Franklin St, Fifth
6686 Floor, Boston, MA 02110, USA.
6688 Verbatim copying and distribution of this entire article is permitted
6689 in any medium, provided this notice is preserved.
6690 Last modified 2006-06-21 [18]Valid XHTML 1.0
6692 References
6694 1. http://www.gnu.org/
6695 2. http://www.gnu.org/
6696 3. http://gcc.gnu.org/gcc-3.1/buildstat.html
6697 4. http://gcc.gnu.org/gcc-3.1/changes.html
6698 5. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
6699 6. http://gcc.gnu.org/index.html
6700 7. mailto:gcc@gcc.gnu.org
6701 8. http://gcc.gnu.org/mirrors.html
6702 9. http://www.gnu.org/order/ftp.html
6703 10. mailto:gnu@gnu.org
6704 11. http://www.gnu.org/home.html#ContactInfo
6705 12. http://gcc.gnu.org/about.html
6706 13. http://gcc.gnu.org/onlinedocs/
6707 14. mailto:gcc-help@gcc.gnu.org
6708 15. mailto:gcc@gnu.org
6709 16. mailto:gcc@gcc.gnu.org
6710 17. http://gcc.gnu.org/lists.html
6711 18. http://validator.w3.org/check/referer
6712 ======================================================================
6713 http://gcc.gnu.org/gcc-3.1/changes.html
6715 GCC 3.1 Release Series
6716 Changes, New Features, and Fixes
6718 Additional changes in GCC 3.1.1
6720 * A bug related to how structures and unions are returned has been
6721 fixed for powerpc-*-netbsd*.
6722 * An important bug in the implementation of -fprefetch-loop-arrays
6723 has been fixed. Previously the optimization prefetched random
6724 blocks of memory for most targets except for i386.
6725 * The Java compiler now compiles Java programs much faster and also
6726 works with parallel make.
6727 * Nested functions have been fixed for mips*-*-netbsd*.
6728 * Some missing floating point support routines have beed added for
6729 mips*-*-netbsd*.
6730 * This [1]message gives additional information about the bugs fixed
6731 in this release.
6733 Caveats
6735 * The -traditional C compiler option has been deprecated and will be
6736 removed in GCC 3.3. (It remains possible to preprocess non-C code
6737 with the traditional preprocessor.)
6738 * The default debugging format for most ELF platforms (including
6739 GNU/Linux and FreeBSD; notable exception is Solaris) has changed
6740 from stabs to DWARF2. This requires GDB 5.1.1 or later.
6742 General Optimizer Improvements
6744 * Jan Hubicka, SuSE Labs, together with Richard Henderson, Red Hat,
6745 and Andreas Jaeger, SuSE Labs, has contributed [2]infrastructure
6746 for profile driven optimizations.
6747 Options -fprofile-arcs and -fbranch-probabilities can now be used
6748 to improve speed of the generated code by profiling the actual
6749 program behaviour on typical runs. In the absence of profile info
6750 the compiler attempts to guess the profile statically.
6751 * [3]SPEC2000 and SPEC95 benchmark suites are now used daily to
6752 monitor performance of the generated code.
6753 According to the SPECInt2000 results on an AMD Athlon CPU, the
6754 code generated by GCC 3.1 is 6% faster on the average (8.2% faster
6755 with profile feedback) compared to GCC 3.0. The code produced by
6756 GCC 3.0 is about 2.1% faster compared to 2.95.3. Tests were done
6757 using the -O2 -march=athlon command-line options.
6758 * Alexandre Oliva, of Red Hat, has generalized the tree inlining
6759 infrastructure developed by CodeSourcery, LLC for the C++ front
6760 end, so that it is now used in the C front end too. Inlining
6761 functions as trees exposes them earlier to the compiler, giving it
6762 more opportunities for optimization.
6763 * Support for data prefetching instructions has been added to the
6764 GCC back end and several targets. A new __builtin_prefetch
6765 intrinsic is available to explicitly insert prefetch instructions
6766 and experimental support for loop array prefetching has been added
6767 (see -fprefetch-loop-array documentation).
6768 * Support for emitting debugging information for macros has been
6769 added for DWARF2. It is activated using -g3.
6771 New Languages and Language specific improvements
6773 C/C++
6775 * A few more [4]ISO C99 features.
6776 * The preprocessor is 10-50% faster than the preprocessor in GCC
6777 3.0.
6778 * The preprocessor's symbol table has been merged with the symbol
6779 table of the C, C++ and Objective-C front ends.
6780 * The preprocessor consumes less memory than the preprocessor in GCC
6781 3.0, often significantly so. On normal input files, it typically
6782 consumes less memory than pre-3.0 cccp-based GCC, too.
6784 C++
6786 * -fhonor-std and -fno-honor-std have been removed. -fno-honor-std
6787 was a workaround to allow std compliant code to work with the
6788 non-std compliant libstdc++-v2. libstdc++-v3 is std compliant.
6789 * The C++ ABI has been fixed so that void (A::*)() const is mangled
6790 as "M1AKFvvE", rather than "MK1AFvvE" as before. This change only
6791 affects pointer to cv-qualified member function types.
6792 * The C++ ABI has been changed to correctly handle this code:
6793 struct A {
6794 void operator delete[] (void *, size_t);
6795 };
6797 struct B : public A {
6798 };
6800 new B[10];
6802 The amount of storage allocated for the array will be greater than
6803 it was in 3.0, in order to store the number of elements in the
6804 array, so that the correct size can be passed to operator delete[]
6805 when the array is deleted. Previously, the value passed to
6806 operator delete[] was unpredictable.
6807 This change will only affect code that declares a two-argument
6808 operator delete[] with a second parameter of type size_t in a base
6809 class, and does not override that definition in a derived class.
6810 * The C++ ABI has been changed so that:
6811 struct A {
6812 void operator delete[] (void *, size_t);
6813 void operator delete[] (void *);
6814 };
6816 does not cause unnecessary storage to be allocated when an array
6817 of A objects is allocated.
6818 This change will only affect code that declares both of these
6819 forms of operator delete[], and declared the two-argument form
6820 before the one-argument form.
6821 * The C++ ABI has been changed so that when a parameter is passed by
6822 value, any cleanup for that parameter is performed in the caller,
6823 as specified by the ia64 C++ ABI, rather than the called function
6824 as before. As a result, classes with a non-trivial destructor but
6825 a trivial copy constructor will be passed and returned by
6826 invisible reference, rather than by bitwise copy as before.
6827 * G++ now supports the "named return value optimization": for code
6828 like
6829 A f () {
6830 A a;
6831 ...
6832 return a;
6833 }
6835 G++ will allocate a in the return value slot, so that the return
6836 becomes a no-op. For this to work, all return statements in the
6837 function must return the same variable.
6838 * Improvements to the C++ library are listed in [5]the libstdc++-v3
6839 FAQ.
6841 Objective-C
6843 * Annoying linker warnings (due to incorrect code being generated)
6844 have been fixed.
6845 * If a class method cannot be found, the compiler no longer issues a
6846 warning if a corresponding instance method exists in the root
6847 class.
6848 * Forward @protocol declarations have been fixed.
6849 * Loading of categories has been fixed in certain situations (GNU
6850 run time only).
6851 * The class lookup in the run-time library has been rewritten so
6852 that class method dispatch is more than twice as fast as it used
6853 to be (GNU run time only).
6855 Java
6857 * libgcj now includes RMI, java.lang.ref.*, javax.naming, and
6858 javax.transaction.
6859 * Property files and other system resources can be compiled into
6860 executables which use libgcj using the new gcj --resource feature.
6861 * libgcj has been ported to more platforms. In particular there is
6862 now a mostly-functional mingw32 (Windows) target port.
6863 * JNI and CNI invocation interfaces were implemented, so
6864 gcj-compiled Java code can now be called from a C/C++ application.
6865 * gcj can now use builtin functions for certain known methods, for
6866 instance Math.cos.
6867 * gcj can now automatically remove redundant array-store checks in
6868 some common cases.
6869 * The --no-store-checks optimization option was added. This can be
6870 used to omit runtime store checks for code which is known not to
6871 throw ArrayStoreException
6872 * The following third party interface standards were added to
6873 libgcj: org.w3c.dom and org.xml.sax.
6874 * java.security has been merged with GNU Classpath. The new package
6875 is now JDK 1.2 compliant, and much more complete.
6876 * A bytecode verifier was added to the libgcj interpreter.
6877 * java.lang.Character was rewritten to comply with the Unicode 3.0
6878 standard, and improve performance.
6879 * Partial support for many more locales was added to libgcj.
6880 * Socket timeouts have been implemented.
6881 * libgcj has been merged into a single shared library. There are no
6882 longer separate shared libraries for the garbage collector and
6883 zlib.
6884 * Several performance improvements were made to gcj and libgcj:
6885 + Hash synchronization (thin locks)
6886 + A special allocation path for finalizer-free objects
6887 + Thread-local allocation
6888 + Parallel GC, and other GC tweaks
6890 Fortran
6892 Fortran improvements are listed in [6]the Fortran documentation.
6894 Ada
6896 [7]Ada Core Technologies, Inc, has contributed its GNAT Ada 95 front
6897 end and associated tools. The GNAT compiler fully implements the Ada
6898 language as defined by the ISO/IEC 8652 standard.
6900 Please note that the integration of the Ada front end is still work in
6901 progress.
6903 New Targets and Target Specific Improvements
6905 * Hans-Peter Nilsson has contributed a port to [8]MMIX, the CPU
6906 architecture used in new editions of Donald E. Knuth's The Art of
6907 Computer Programming.
6908 * [9]Axis Communications has contributed its port to the CRIS CPU
6909 architecture, used in the ETRAX system-on-a-chip series. See
6910 [10]Axis' developer site for technical information.
6911 * Alexandre Oliva, of Red Hat, has contributed a port to the
6912 [11]SuperH SH5 64-bit RISC microprocessor architecture, extending
6913 the existing SH port.
6914 * UltraSPARC is fully supported in 64-bit mode. The option -m64
6915 enables it.
6916 * For compatibility with the Sun compiler #pragma redefine_extname
6917 has been implemented on Solaris.
6918 * The x86 back end has had some noticeable work done to it.
6919 + SuSE Labs developers Jan Hubicka, Bo Thorsen and Andreas
6920 Jaeger have contributed a port to the AMD x86-64
6921 architecture. For more information on x86-64 see
6922 [12]http://www.x86-64.org.
6923 + The compiler now supports MMX, 3DNow!, SSE, and SSE2
6924 instructions. Options -mmmx, -m3dnow, -msse, and -msse2 will
6925 enable the respective instruction sets. Intel C++ compatible
6926 MMX/3DNow!/SSE intrinsics are implemented. SSE2 intrinsics
6927 will be added in next major release.
6928 + Following those improvements, targets for Pentium MMX, K6-2,
6929 K6-3, Pentium III, Pentium 4, and Athlon 4 Mobile/XP/MP were
6930 added. Refer to the documentation on -march= and -mcpu=
6931 options for details.
6932 + For those targets that support it, -mfpmath=sse will cause
6933 the compiler to generate SSE/SSE2 instructions for floating
6934 point math instead of x87 instructions. Usually, this will
6935 lead to quicker code -- especially on the Pentium 4. Note
6936 that only scalar floating point instructions are used and GCC
6937 does not exploit SIMD features yet.
6938 + Prefetch support has been added to the Pentium III, Pentium
6939 4, K6-2, K6-3, and Athlon series.
6940 + Code generated for floating point to integer conversions has
6941 been improved leading to better performance of many 3D
6942 applications.
6943 * The PowerPC back end has added 64-bit PowerPC GNU/Linux support.
6944 * C++ support for AIX has been improved.
6945 * Aldy Hernandez, of Red Hat, Inc, has contributed extensions to the
6946 PowerPC port supporting the AltiVec programming model (SIMD). The
6947 support, though presently useful, is experimental and is expected
6948 to stabilize for 3.2. The support is written to conform to
6949 Motorola's AltiVec specs. See -maltivec.
6951 Obsolete Systems
6953 Support for a number of older systems has been declared obsolete in
6954 GCC 3.1. Unless there is activity to revive them, the next release of
6955 GCC will have their sources permanently removed.
6957 All configurations of the following processor architectures have been
6958 declared obsolete:
6959 * MIL-STD-1750A, 1750a-*-*
6960 * AMD A29k, a29k-*-*
6961 * Convex, c*-convex-*
6962 * Clipper, clipper-*-*
6963 * Elxsi, elxsi-*-*
6964 * Intel i860, i860-*-*
6965 * Sun picoJava, pj-*-* and pjl-*-*
6966 * Western Electric 32000, we32k-*-*
6968 Most configurations of the following processor architectures have been
6969 declared obsolete, but we are preserving a few systems which may have
6970 active developers. It is unlikely that the remaining systems will
6971 survive much longer unless we see definite signs of port activity.
6972 * Motorola 88000 except
6973 + Generic a.out, m88k-*-aout*
6974 + Generic SVR4, m88k-*-sysv4
6975 + OpenBSD, m88k-*-openbsd*
6976 * NS32k except
6977 + NetBSD, ns32k-*-netbsd*
6978 + OpenBSD, ns32k-*-openbsd*.
6979 * ROMP except
6980 + OpenBSD, romp-*-openbsd*.
6982 Finally, only some configurations of these processor architectures are
6983 being obsoleted.
6984 * Alpha:
6985 + OSF/1, alpha*-*-osf[123]*. (Digital Unix and Tru64 Unix, aka
6986 alpha*-*-osf[45], are still supported.)
6987 * ARM:
6988 + RISCiX, arm-*-riscix*.
6989 * i386:
6990 + 386BSD, i?86-*-bsd*
6991 + Chorus, i?86-*-chorusos*
6992 + DG/UX, i?86-*-dgux*
6993 + FreeBSD 1.x, i?86-*-freebsd1.*
6994 + IBM AIX, i?86-*-aix*
6995 + ISC UNIX, i?86-*-isc*
6996 + Linux with pre-BFD linker, i?86-*-linux*oldld*
6997 + NEXTstep, i?86-next-*
6998 + OSF UNIX, i?86-*-osf1* and i?86-*-osfrose*
6999 + RTEMS/coff, i?86-*-rtemscoff*
7000 + RTEMS/go32, i?86-go32-rtems*
7001 + Sequent/BSD, i?86-sequent-bsd*
7002 + Sequent/ptx before version 3, i?86-sequent-ptx[12]* and
7003 i?86-sequent-sysv3*
7004 + SunOS, i?86-*-sunos*
7005 * Motorola 68000:
7006 + Altos, m68[k0]*-altos-*
7007 + Apollo, m68[k0]*-apollo-*
7008 + Apple A/UX, m68[k0]*-apple-*
7009 + Bull, m68[k0]*-bull-*
7010 + Convergent, m68[k0]*-convergent-*
7011 + Generic SVR3, m68[k0]*-*-sysv3*
7012 + ISI, m68[k0]*-isi-*
7013 + LynxOS, m68[k0]*-*-lynxos*
7014 + NEXT, m68[k0]*-next-*
7015 + RTEMS/coff, m68[k0]*-*-rtemscoff*
7016 + Sony, m68[k0]*-sony-*
7017 * MIPS:
7018 + DEC Ultrix, mips-*-ultrix* and mips-dec-*
7019 + Generic BSD, mips-*-bsd*
7020 + Generic System V, mips-*-sysv*
7021 + IRIX before version 5, mips-sgi-irix[1234]*
7022 + RiscOS, mips-*-riscos*
7023 + Sony, mips-sony-*
7024 + Tandem, mips-tandem-*
7025 * SPARC:
7026 + RTEMS/a.out, sparc-*-rtemsaout*.
7028 Documentation improvements
7030 * The old manual ("Using and Porting the GNU Compiler Collection")
7031 has been replaced by a users manual ("Using the GNU Compiler
7032 Collection") and a separate internals reference manual ("GNU
7033 Compiler Collection Internals").
7034 * More complete and much improved documentation about GCC's internal
7035 representation used by the C and C++ front ends.
7036 * Many cleanups and improvements in general.
7038 Please send FSF & GNU inquiries & questions to [13]gnu@gnu.org. There
7039 are also [14]other ways to contact the FSF.
7041 These pages are maintained by [15]the GCC team.
7044 For questions related to the use of GCC, please consult these web
7045 pages and the [16]GCC manuals. If that fails, the
7046 [17]gcc-help@gcc.gnu.org mailing list might help.
7047 Please send comments on these web pages and the development of GCC
7048 to our developer mailing list at [18]gcc@gnu.org or
7049 [19]gcc@gcc.gnu.org. All of our lists have [20]public archives.
7051 Copyright (C) Free Software Foundation, Inc., 51 Franklin St, Fifth
7052 Floor, Boston, MA 02110, USA.
7054 Verbatim copying and distribution of this entire article is permitted
7055 in any medium, provided this notice is preserved.
7056 Last modified 2006-10-22 [21]Valid XHTML 1.0
7058 References
7060 1. http://gcc.gnu.org/ml/gcc/2002-07/msg01208.html
7061 2. http://gcc.gnu.org/news/profiledriven.html
7062 3. http://gcc.gnu.org/benchmarks/
7063 4. http://gcc.gnu.org/gcc-3.1/c99status.html
7064 5. http://gcc.gnu.org/onlinedocs/libstdc++/faq/index.html#4_1
7065 6. http://gcc.gnu.org/onlinedocs/gcc-3.1.1/g77/News.html
7066 7. http://www.gnat.com/
7067 8. http://www-cs-faculty.stanford.edu/~knuth/mmix.html
7068 9. http://www.axis.com/
7069 10. http://developer.axis.com/
7070 11. http://www.superh.com/
7071 12. http://www.x86-64.org/
7072 13. mailto:gnu@gnu.org
7073 14. http://www.gnu.org/home.html#ContactInfo
7074 15. http://gcc.gnu.org/about.html
7075 16. http://gcc.gnu.org/onlinedocs/
7076 17. mailto:gcc-help@gcc.gnu.org
7077 18. mailto:gcc@gnu.org
7078 19. mailto:gcc@gcc.gnu.org
7079 20. http://gcc.gnu.org/lists.html
7080 21. http://validator.w3.org/check/referer
7081 ======================================================================
7082 http://gcc.gnu.org/gcc-3.0/gcc-3.0.html
7084 GCC 3.0.4
7086 February 20, 2002
7088 The [1]GNU project and the GCC developers are pleased to announce the
7089 release of GCC 3.0.4, which is a bug-fix release for the GCC 3.0
7090 series.
7092 GCC used to stand for the GNU C Compiler, but since the compiler
7093 supports several other languages aside from C, it now stands for the
7094 GNU Compiler Collection.
7096 GCC 3.0.x has several new optimizations, new targets, new languages
7097 and many other new features, relative to GCC 2.95.x. See the [2]new
7098 features page for a more complete list.
7100 A list of [3]successful builds is updated as new information becomes
7101 available.
7103 The GCC developers would like to thank the numerous people that have
7104 contributed new features, test results, bug fixes, etc to GCC. This
7105 [4]amazing group of volunteers is what makes GCC successful.
7107 And finally, we can't in good conscience fail to mention some
7108 [5]caveats to using GCC 3.0.x.
7110 For additional information about GCC please refer to the [6]GCC
7111 project web site or contact the [7]GCC development mailing list.
7113 To obtain GCC please use [8]our mirror sites, one of the [9]GNU mirror
7114 sites, or our CVS server.
7115 _________________________________________________________________
7117 Previous 3.0.x Releases
7119 December 20, 2001: GCC 3.0.3 has been released.
7120 October 25, 2001: GCC 3.0.2 has been released.
7121 August 20, 2001: GCC 3.0.1 has been released.
7122 June 18, 2001: GCC 3.0 has been released.
7124 Please send FSF & GNU inquiries & questions to [10]gnu@gnu.org. There
7125 are also [11]other ways to contact the FSF.
7127 These pages are maintained by [12]the GCC team.
7130 For questions related to the use of GCC, please consult these web
7131 pages and the [13]GCC manuals. If that fails, the
7132 [14]gcc-help@gcc.gnu.org mailing list might help.
7133 Please send comments on these web pages and the development of GCC
7134 to our developer mailing list at [15]gcc@gnu.org or
7135 [16]gcc@gcc.gnu.org. All of our lists have [17]public archives.
7137 Copyright (C) Free Software Foundation, Inc., 51 Franklin St, Fifth
7138 Floor, Boston, MA 02110, USA.
7140 Verbatim copying and distribution of this entire article is permitted
7141 in any medium, provided this notice is preserved.
7142 Last modified 2007-02-01 [18]Valid XHTML 1.0
7144 References
7146 1. http://www.gnu.org/
7147 2. http://gcc.gnu.org/gcc-3.0/features.html
7148 3. http://gcc.gnu.org/gcc-3.0/buildstat.html
7149 4. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
7150 5. http://gcc.gnu.org/gcc-3.0/caveats.html
7151 6. http://gcc.gnu.org/index.html
7152 7. mailto:gcc@gcc.gnu.org
7153 8. http://gcc.gnu.org/mirrors.html
7154 9. http://www.gnu.org/order/ftp.html
7155 10. mailto:gnu@gnu.org
7156 11. http://www.gnu.org/home.html#ContactInfo
7157 12. http://gcc.gnu.org/about.html
7158 13. http://gcc.gnu.org/onlinedocs/
7159 14. mailto:gcc-help@gcc.gnu.org
7160 15. mailto:gcc@gnu.org
7161 16. mailto:gcc@gcc.gnu.org
7162 17. http://gcc.gnu.org/lists.html
7163 18. http://validator.w3.org/check/referer
7164 ======================================================================
7165 http://gcc.gnu.org/gcc-3.0/features.html
7167 GCC 3.0 New Features
7169 Additional changes in GCC 3.0.4
7171 * GCC 3.0 now supports newer versions of the [1]NetBSD operating
7172 system, which use the ELF object file format, on x86 processors.
7173 * Correct debugging information is generated from functions that
7174 have lines from multiple files (e.g. yacc output).
7175 * A fix for whitespace handling in the -traditional preprocessor,
7176 which can affect Fortran.
7177 * Fixes to the exception handling runtime.
7178 * More fixes for bad code generation in C++.
7179 * A fix for shared library generation under AIX 4.3.
7180 * Documentation updates.
7181 * Port of GCC to Tensilica's Xtensa processor contributed.
7182 * A fix for compiling the PPC Linux kernel (FAT fs wouldn't link).
7184 Additional changes in GCC 3.0.3
7186 * A fix to correct an accidental change to the PowerPC ABI.
7187 * Fixes for bad code generation on a variety of architectures.
7188 * Improvements to the debugging information generated for C++
7189 classes.
7190 * Fixes for bad code generation in C++.
7191 * A fix to avoid crashes in the C++ demangler.
7192 * A fix to the C++ standard library to avoid buffer overflows.
7193 * Miscellaneous improvements for a variety of architectures.
7195 Additional changes in GCC 3.0.2
7197 * Fixes for bad code generation during loop unrolling.
7198 * Fixes for bad code generation by the sibling call optimization.
7199 * Minor improvements to x86 code generation.
7200 * Implementation of function descriptors in C++ vtables for IA64.
7201 * Numerous minor bug-fixes.
7203 Additional changes in GCC 3.0.1
7205 * C++ fixes for incorrect code-generation.
7206 * Improved cross-compiling support for the C++ standard library.
7207 * Fixes for some embedded targets that worked in GCC 2.95.3, but not
7208 in GCC 3.0.
7209 * Fixes for various exception-handling bugs.
7210 * A port to the S/390 architecture.
7212 General Optimizer Improvements
7214 * [2]Basic block reordering pass.
7215 * New if-conversion pass with support for conditional (predicated)
7216 execution.
7217 * New tail call and sibling call elimination optimizations.
7218 * New register renaming pass.
7219 * New (experimental) [3]static single assignment (SSA)
7220 representation support.
7221 * New dead-code elimination pass implemented using the SSA
7222 representation.
7223 * [4]Global null pointer test elimination.
7224 * [5]Global code hoisting/unification.
7225 * More builtins and optimizations for stdio.h, string.h and old BSD
7226 functions, as well as for ISO C99 functions.
7227 * New builtin __builtin_expect for giving hints to the branch
7228 predictor.
7230 New Languages and Language specific improvements
7232 * The GNU Compiler for the Java(TM) language (GCJ) is now integrated
7233 and supported, including the run-time library containing most
7234 common non-GUI Java classes, a bytecode interpreter, and the Boehm
7235 conservative garbage collector. Many bugs have been fixed. GCJ can
7236 compile Java source or Java bytecodes to either native code or
7237 Java class files, and supports native methods written in either
7238 the standard JNI or the more efficient and convenient CNI.
7239 * Here is a [6]partial list of C++ improvements, both new features
7240 and those no longer supported.
7241 * New C++ ABI. On the IA-64 platform GCC is capable of
7242 inter-operating with other IA-64 compilers.
7243 * The new ABI also significantly reduces the size of symbol and
7244 debug information.
7245 * New [7]C++ support library and many C++ bug fixes, vastly
7246 improving our conformance to the ISO C++ standard.
7247 * New [8]inliner for C++.
7248 * Rewritten C preprocessor, integrated into the C, C++ and Objective
7249 C compilers, with very many improvements including ISO C99 support
7250 and [9]improvements to dependency generation.
7251 * Support for more [10]ISO C99 features.
7252 * Many improvements to support for checking calls to format
7253 functions such as printf and scanf, including support for ISO C99
7254 format features, extensions from the Single Unix Specification and
7255 GNU libc 2.2, checking of strfmon formats and features to assist
7256 in auditing for format string security bugs.
7257 * New warnings for C code that may have undefined semantics because
7258 of violations of sequence point rules in the C standard (such as a
7259 = a++;, a[n] = b[n++]; and a[i++] = i;), included in -Wall.
7260 * Additional warning option -Wfloat-equal.
7261 * Improvements to -Wtraditional.
7262 * Fortran improvements are listed in [11]the Fortran documentation.
7264 New Targets and Target Specific Improvements
7266 * New x86 back-end, generating much improved code.
7267 * Support for a generic i386-elf target contributed.
7268 * New option to emit x86 assembly code using Intel style syntax
7269 (-mintel-syntax).
7270 * HPUX 11 support contributed.
7271 * Improved PowerPC code generation, including scheduled prologue and
7272 epilogue.
7273 * Port of GCC to Intel's IA-64 processor contributed.
7274 * Port of GCC to Motorola's MCore 210 and 340 contributed.
7275 * New unified back-end for Arm, Thumb and StrongArm contributed.
7276 * Port of GCC to Intel's XScale processor contributed.
7277 * Port of GCC to Atmel's AVR microcontrollers contributed.
7278 * Port of GCC to Mitsubishi's D30V processor contributed.
7279 * Port of GCC to Matsushita's AM33 processor (a member of the
7280 MN10300 processor family) contributed.
7281 * Port of GCC to Fujitsu's FR30 processor contributed.
7282 * Port of GCC to Motorola's 68HC11 and 68HC12 processors
7283 contributed.
7284 * Port of GCC to Sun's picoJava processor core contributed.
7286 Documentation improvements
7288 * Substantially rewritten and improved C preprocessor manual.
7289 * Many improvements to other documentation.
7290 * Manpages for gcc, cpp and gcov are now generated automatically
7291 from the master Texinfo manual, eliminating the problem of
7292 manpages being out of date. (The generated manpages are only
7293 extracts from the full manual, which is provided in Texinfo form,
7294 from which info, HTML, other formats and a printed manual can be
7295 generated.)
7296 * Generated info files are included in the release tarballs
7297 alongside their Texinfo sources, avoiding problems on some
7298 platforms with building makeinfo as part of the GCC distribution.
7300 Other significant improvements
7302 * Garbage collection used internally by the compiler for most memory
7303 allocation instead of obstacks.
7304 * Lengauer and Tarjan algorithm used for computing dominators in the
7305 CFG. This algorithm can be significantly faster and more space
7306 efficient than our older algorithm.
7307 * gccbug script provided to assist in submitting bug reports to our
7308 bug tracking system. (Bug reports previously submitted directly to
7309 our mailing lists, for which you received no bug tracking number,
7310 should be submitted again using gccbug if you can reproduce the
7311 problem with GCC 3.0.)
7312 * The internal libgcc library is [12]built as a shared library on
7313 systems that support it.
7314 * Extensive testsuite included with GCC, with many new tests. In
7315 addition to tests for GCC bugs that have been fixed, many tests
7316 have been added for language features, compiler warnings and
7317 builtin functions.
7318 * Additional language-independent warning options -Wpacked,
7319 -Wpadded, -Wunreachable-code and -Wdisabled-optimization.
7320 * Target-independent options -falign-functions, -falign-loops and
7321 -falign-jumps.
7323 Plus a great many bug fixes and almost all the [13]features found in
7324 GCC 2.95.
7326 Please send FSF & GNU inquiries & questions to [14]gnu@gnu.org. There
7327 are also [15]other ways to contact the FSF.
7329 These pages are maintained by [16]the GCC team.
7332 For questions related to the use of GCC, please consult these web
7333 pages and the [17]GCC manuals. If that fails, the
7334 [18]gcc-help@gcc.gnu.org mailing list might help.
7335 Please send comments on these web pages and the development of GCC
7336 to our developer mailing list at [19]gcc@gnu.org or
7337 [20]gcc@gcc.gnu.org. All of our lists have [21]public archives.
7339 Copyright (C) Free Software Foundation, Inc., 51 Franklin St, Fifth
7340 Floor, Boston, MA 02110, USA.
7342 Verbatim copying and distribution of this entire article is permitted
7343 in any medium, provided this notice is preserved.
7344 Last modified 2007-02-01 [22]Valid XHTML 1.0
7346 References
7348 1. http://www.netbsd.org/
7349 2. http://gcc.gnu.org/news/reorder.html
7350 3. http://gcc.gnu.org/news/ssa.html
7351 4. http://gcc.gnu.org/news/null.html
7352 5. http://gcc.gnu.org/news/unify.html
7353 6. http://gcc.gnu.org/gcc-3.0/c++features.html
7354 7. http://gcc.gnu.org/libstdc++/
7355 8. http://gcc.gnu.org/news/inlining.html
7356 9. http://gcc.gnu.org/news/dependencies.html
7357 10. http://gcc.gnu.org/gcc-3.0/c99status.html
7358 11. http://gcc.gnu.org/onlinedocs/g77/News.html
7359 12. http://gcc.gnu.org/gcc-3.0/libgcc.html
7360 13. http://gcc.gnu.org/gcc-2.95/features.html
7361 14. mailto:gnu@gnu.org
7362 15. http://www.gnu.org/home.html#ContactInfo
7363 16. http://gcc.gnu.org/about.html
7364 17. http://gcc.gnu.org/onlinedocs/
7365 18. mailto:gcc-help@gcc.gnu.org
7366 19. mailto:gcc@gnu.org
7367 20. mailto:gcc@gcc.gnu.org
7368 21. http://gcc.gnu.org/lists.html
7369 22. http://validator.w3.org/check/referer
7370 ======================================================================
7371 http://gcc.gnu.org/gcc-3.0/caveats.html
7373 GCC 3.0 Caveats
7375 * -fstrict-aliasing is now part of -O2 and higher optimization
7376 levels. This allows the compiler to assume the strictest aliasing
7377 rules applicable to the language being compiled. For C and C++,
7378 this activates optimizations based on the type of expressions.
7379 This optimization may thus break old, non-compliant code.
7380 * Enumerations are now properly promoted to int in function
7381 parameters and function returns. Normally this change is not
7382 visible, but when using -fshort-enums this is an ABI change.
7383 * The undocumented extension that allowed C programs to have a label
7384 at the end of a compound statement has been deprecated and may be
7385 removed in a future version. Programs that now generate a warning
7386 about this may be fixed by adding a null statement (a single
7387 semicolon) after the label.
7388 * The poorly documented extension that allowed string constants in
7389 C, C++ and Objective C to contain unescaped newlines has been
7390 deprecated and may be removed in a future version. Programs using
7391 this extension may be fixed in several ways: the bare newline may
7392 be replaced by \n, or preceded by \n\, or string concatenation may
7393 be used with the bare newline preceded by \n" and " placed at the
7394 start of the next line.
7395 * The Chill compiler is not included in GCC 3.0, because of the lack
7396 of a volunteer to convert it to use garbage collection.
7397 * Certain non-standard iostream methods from earlier versions of
7398 libstdc++ are not included in libstdc++ v3, i.e. filebuf::attach,
7399 ostream::form, and istream::gets. Here are workaround hints for:
7400 [1]ostream::form, [2]filebuf::attach.
7401 * The new C++ ABI is not yet fully supported by current (as of
7402 2001-07-01) releases and development versions of GDB, or any
7403 earlier versions. There is a problem setting breakpoints by line
7404 number, and other related issues that have been fixed in GCC 3.0
7405 but not yet handled in GDB:
7406 [3]http://gcc.gnu.org/ml/gcc-bugs/2001-06/msg00421.html
7408 Please send FSF & GNU inquiries & questions to [4]gnu@gnu.org. There
7409 are also [5]other ways to contact the FSF.
7411 These pages are maintained by [6]the GCC team.
7414 For questions related to the use of GCC, please consult these web
7415 pages and the [7]GCC manuals. If that fails, the
7416 [8]gcc-help@gcc.gnu.org mailing list might help.
7417 Please send comments on these web pages and the development of GCC
7418 to our developer mailing list at [9]gcc@gnu.org or
7419 [10]gcc@gcc.gnu.org. All of our lists have [11]public archives.
7421 Copyright (C) Free Software Foundation, Inc., 51 Franklin St, Fifth
7422 Floor, Boston, MA 02110, USA.
7424 Verbatim copying and distribution of this entire article is permitted
7425 in any medium, provided this notice is preserved.
7426 Last modified 2006-06-21 [12]Valid XHTML 1.0
7428 References
7430 1. http://gcc.gnu.org/onlinedocs/libstdc++/21_strings/howto.html
7431 2. http://gcc.gnu.org/onlinedocs/libstdc++/ext/howto.html
7432 3. http://gcc.gnu.org/ml/gcc-bugs/2001-06/msg00421.html
7433 4. mailto:gnu@gnu.org
7434 5. http://www.gnu.org/home.html#ContactInfo
7435 6. http://gcc.gnu.org/about.html
7436 7. http://gcc.gnu.org/onlinedocs/
7437 8. mailto:gcc-help@gcc.gnu.org
7438 9. mailto:gcc@gnu.org
7439 10. mailto:gcc@gcc.gnu.org
7440 11. http://gcc.gnu.org/lists.html
7441 12. http://validator.w3.org/check/referer
7442 ======================================================================
7443 http://gcc.gnu.org/gcc-2.95/index.html
7445 GCC 2.95
7447 March 16, 2001: The GNU project and the GCC developers are pleased to
7448 announce the release of GCC version 2.95.3.
7450 Release History
7452 GCC 2.95.3
7453 March 16, 2001
7455 GCC 2.95.2
7456 October 27, 1999
7458 GCC 2.95.1
7459 August 19, 1999
7461 GCC 2.95
7462 July 31, 1999. This is the first release of GCC since the April
7463 1999 GCC/EGCS reunification and includes nearly a year's worth
7464 of new development and bugfixes.
7466 References and Acknowledgements
7468 GCC used to stand for the GNU C Compiler, but since the compiler
7469 supports several other languages aside from C, it now stands for the
7470 GNU Compiler Collection.
7472 The whole suite has been extensively [1]regression tested and
7473 [2]package tested. It should be reliable and suitable for widespread
7474 use.
7476 The compiler has several new optimizations, new targets, new languages
7477 and other new features. See the [3]new features page for a more
7478 complete list of new features found in the GCC 2.95 releases.
7480 The sources include installation instructions in both HTML and
7481 plaintext forms in the install directory in the distribution. However,
7482 the most up to date [4]installation instructions and [5]build/test
7483 status are on the web pages. We will update those pages as new
7484 information becomes available.
7486 The GCC developers would like to thank the numerous people that have
7487 contributed new features, test results, bugfixes, etc to GCC. This
7488 [6]amazing group of volunteers is what makes GCC successful.
7490 And finally, we can't in good conscience fail to mention some
7491 [7]caveats to using GCC 2.95.
7493 Download GCC 2.95 from the [8]GNU FTP server (ftp://ftp.gnu.org)
7494 [9]Find a GNU mirror site
7495 [10]Find a GCC mirror site
7497 For additional information about GCC please see the [11]GCC project
7498 web server or contact the [12]GCC development mailing list.
7500 Please send FSF & GNU inquiries & questions to [13]gnu@gnu.org. There
7501 are also [14]other ways to contact the FSF.
7503 These pages are maintained by [15]the GCC team.
7506 For questions related to the use of GCC, please consult these web
7507 pages and the [16]GCC manuals. If that fails, the
7508 [17]gcc-help@gcc.gnu.org mailing list might help.
7509 Please send comments on these web pages and the development of GCC
7510 to our developer mailing list at [18]gcc@gnu.org or
7511 [19]gcc@gcc.gnu.org. All of our lists have [20]public archives.
7513 Copyright (C) Free Software Foundation, Inc., 51 Franklin St, Fifth
7514 Floor, Boston, MA 02110, USA.
7516 Verbatim copying and distribution of this entire article is permitted
7517 in any medium, provided this notice is preserved.
7518 Last modified 2006-06-21 [21]Valid XHTML 1.0
7520 References
7522 1. http://gcc.gnu.org/gcc-2.95/regress.html
7523 2. http://gcc.gnu.org/gcc-2.95/othertest.html
7524 3. http://gcc.gnu.org/gcc-2.95/features.html
7525 4. http://gcc.gnu.org/install/
7526 5. http://gcc.gnu.org/gcc-2.95/buildstat.html
7527 6. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
7528 7. http://gcc.gnu.org/gcc-2.95/caveats.html
7529 8. ftp://ftp.gnu.org/pub/gnu/gcc/
7530 9. http://www.gnu.org/order/ftp.html
7531 10. http://gcc.gnu.org/mirrors.html
7532 11. http://gcc.gnu.org/index.html
7533 12. mailto:gcc@gcc.gnu.org
7534 13. mailto:gnu@gnu.org
7535 14. http://www.gnu.org/home.html#ContactInfo
7536 15. http://gcc.gnu.org/about.html
7537 16. http://gcc.gnu.org/onlinedocs/
7538 17. mailto:gcc-help@gcc.gnu.org
7539 18. mailto:gcc@gnu.org
7540 19. mailto:gcc@gcc.gnu.org
7541 20. http://gcc.gnu.org/lists.html
7542 21. http://validator.w3.org/check/referer
7543 ======================================================================
7544 http://gcc.gnu.org/gcc-2.95/features.html
7546 GCC 2.95 New Features
7548 * General Optimizer Improvements:
7549 + [1]Localized register spilling to improve speed and code
7550 density especially on small register class machines.
7551 + [2]Global CSE using lazy code motion algorithms.
7552 + [3]Improved global constant/copy propagation.
7553 + [4]Improved control flow graph analysis and manipulation.
7554 + [5]Local dead store elimination.
7555 + [6]Memory Load hoisting/store sinking in loops.
7556 + [7]Type based alias analysis is enabled by default. Note this
7557 feature will expose bugs in the Linux kernel. Please refer to
7558 the FAQ (as shipped with GCC 2.95) for additional information
7559 on this issue.
7560 + Major revamp of GIV detection, combination and simplification
7561 to improve loop performance.
7562 + Major improvements to register allocation and reloading.
7563 * New Languages and Language specific improvements
7564 + [8]Many C++ improvements.
7565 + [9]Many Fortran improvements.
7566 + [10]Java front-end has been integrated. [11]runtime library
7567 is available separately.
7568 + [12]ISO C99 support
7569 + [13]Chill front-end and runtime has been integrated.
7570 + Boehm garbage collector support in libobjc.
7571 + More support for various pragmas which appear in vendor
7572 include files
7573 * New Targets and Target Specific Improvements
7574 + [14]SPARC backend rewrite.
7575 + -mschedule=8000 will optimize code for PA8000 class
7576 processors; -mpa-risc-2-0 will generate code for PA2.0
7577 processors
7578 + Various micro-optimizations for the ia32 port. K6
7579 optimizations
7580 + Compiler will attempt to align doubles in the stack on the
7581 ia32 port
7582 + Alpha EV6 support
7583 + PowerPC 750
7584 + RS6000/PowerPC: -mcpu=401 was added as an alias for
7585 -mcpu=403. -mcpu=e603e was added to do -mcpu=603e and
7586 -msoft-float.
7587 + c3x, c4x
7588 + HyperSPARC
7589 + SparcLite86x
7590 + sh4
7591 + Support for new systems (OpenBSD, FreeBSD, UWIN, Interix,
7592 arm-linux)
7593 + vxWorks targets include support for vxWorks threads
7594 + StrongARM 110 and ARM9 support added. ARM Scheduling
7595 parameters rewritten.
7596 + Various changes to the MIPS port to avoid assembler macros,
7597 which in turn improves performance
7598 + Various performance improvements to the i960 port.
7599 + Major rewrite of ns32k port
7600 * Other significant improvements
7601 + [15]Ability to dump cfg information and display it using vcg.
7602 + The new faster scheme for fixing vendor header files is
7603 enabled by default.
7604 + Experimental internationalization support.
7605 + multibyte character support
7606 + Some compile-time speedups for pathological problems
7607 + Better support for complex types
7608 * Plus the usual mountain of bugfixes
7609 * Core compiler is based on the gcc2 development tree from Sept 30,
7610 1998, so we have all of the [16]features found in GCC 2.8.
7612 Additional Changes in GCC 2.95.1
7614 * Generic bugfixes and improvements
7615 + Various documentation fixes related to the GCC/EGCS merger.
7616 + Fix memory management bug which could lead to spurious
7617 aborts, core dumps or random parsing errors in the compiler.
7618 + Fix a couple bugs in the dwarf1 and dwarf2 debug record
7619 support.
7620 + Fix infinite loop in the CSE optimizer.
7621 + Avoid undefined behavior in compiler FP emulation code
7622 + Fix install problem when prefix is overridden on the make
7623 install command.
7624 + Fix problem with unwanted installation of assert.h on some
7625 systems.
7626 + Fix problem with finding the wrong assembler in a single tree
7627 build.
7628 + Avoid increasing the known alignment of a register that is
7629 already known to be a pointer.
7630 * Platform specific bugfixes and improvements
7631 + Codegen bugfix for prologue/epilogue for cpu32 target.
7632 + Fix long long code generation bug for the Coldfire target.
7633 + Fix various aborts in the SH compiler.
7634 + Fix bugs in libgcc support library for the SH.
7635 + Fix alpha ev6 code generation bug.
7636 + Fix problems with EXIT_SUCCESS/EXIT_FAILURE redefinitions on
7637 AIX platforms.
7638 + Fix -fpic code generation bug for rs6000/ppc svr4 targets.
7639 + Fix varargs/stdarg code generation bug for rs6000/ppc svr4
7640 targets.
7641 + Fix weak symbol handling for rs6000/ppc svr4 targets.
7642 + Fix various problems with 64bit code generation for the
7643 rs6000/ppc port.
7644 + Fix codegen bug which caused tetex to be mis-compiled on the
7645 x86.
7646 + Fix compiler abort in new cfg code exposed by x86 port.
7647 + Fix out of range array reference in code convert flat
7648 registers to the x87 stacked FP register file.
7649 + Fix minor vxworks configuration bug.
7650 + Fix return type of bsearch for SunOS 4.x.
7651 * Language & Runtime specific fixes.
7652 + The G++ signature extension has been deprecated. It will be
7653 removed in the next major release of G++. Use of signatures
7654 will result in a warning from the compiler.
7655 + Several bugs relating to templates and namespaces were fixed.
7656 + A bug that caused crashes when combining templates with -g on
7657 DWARF1 platforms was fixed.
7658 + Pointers-to-members, virtual functions, and multiple
7659 inheritance should now work together correctly.
7660 + Some code-generation bugs relating to function try blocks
7661 were fixed.
7662 + G++ is a little bit more lenient with certain archaic
7663 constructs than in GCC 2.95.
7664 + Fix to prevent shared library version #s from bring truncated
7665 to 1 digit
7666 + Fix missing std:: in the libstdc++ library.
7667 + Fix stream locking problems in libio.
7668 + Fix problem in java compiler driver.
7670 Additional Changes in GCC 2.95.2
7672 The -fstrict-aliasing is not enabled by default for GCC 2.95.2. While
7673 the optimizations performed by -fstrict-aliasing are valid according
7674 to the C and C++ standards, the optimization have caused some
7675 problems, particularly with old non-conforming code.
7677 The GCC developers are experimenting with ways to warn users about
7678 code which violates the C/C++ standards, but those warnings are not
7679 ready for widespread use at this time. Rather than wait for those
7680 warnings the GCC developers have chosen to disable -fstrict-aliasing
7681 by default for the GCC 2.95.2 release.
7683 We strongly encourage developers to find and fix code which violates
7684 the C/C++ standards as -fstrict-aliasing may be enabled by default in
7685 future releases. Use the option -fstrict-aliasing to re-enable these
7686 optimizations.
7687 * Generic bugfixes and improvements
7688 + Fix incorrectly optimized memory reference in global common
7689 subexpression elimination (GCSE) optimization pass.
7690 + Fix code generation bug in regmove.c in which it could
7691 incorrectly change a "const" value.
7692 + Fix bug in optimization of conditionals involving volatile
7693 memory references.
7694 + Avoid over-allocation of stack space for some procedures.
7695 + Fixed bug in the compiler which caused incorrect optimization
7696 of an obscure series of bit manipulations, shifts and
7697 arithmetic.
7698 + Fixed register allocator bug which caused teTeX to be
7699 mis-compiled on SPARC targets.
7700 + Avoid incorrect optimization of degenerate case statements
7701 for certain targets such as the ARM.
7702 + Fix out of range memory reference in the jump optimizer.
7703 + Avoid dereferencing null pointer in fix-header.
7704 + Fix test for GCC specific features so that it is possible to
7705 bootstrap with gcc-2.6.2 and older versions of GCC.
7706 + Fix typo in scheduler which could potentially cause out of
7707 range memory accesses.
7708 + Avoid incorrect loop reversal which caused incorrect code for
7709 certain loops on PowerPC targets.
7710 + Avoid incorrect optimization of switch statements on certain
7711 targets (for example the ARM).
7712 * Platform specific bugfixes and improvements
7713 + Work around bug in Sun V5.0 compilers which caused bootstrap
7714 comparison failures on SPARC targets.
7715 + Fix SPARC backend bug which caused aborts in final.c.
7716 + Fix sparc-hal-solaris2* configuration fragments.
7717 + Fix bug in sparc block profiling.
7718 + Fix obscure code generation bug for the PARISC targets.
7719 + Define __STDC_EXT__ for HPUX configurations.
7720 + Various POWERPC64 code generation bugfixes.
7721 + Fix abort for PPC targets using ELF (ex GNU/Linux).
7722 + Fix collect2 problems for AIX targets.
7723 + Correct handling of .file directive for PPC targets.
7724 + Fix bug in fix_trunc x86 patterns.
7725 + Fix x86 port to correctly pop the FP stack for functions that
7726 return structures in memory.
7727 + Fix minor bug in strlen x86 pattern.
7728 + Use stabs debugging instead of dwarf1 for x86-solaris
7729 targets.
7730 + Fix template repository code to handle leading underscore in
7731 mangled names.
7732 + Fix weak/weak alias support for OpenBSD.
7733 + GNU/Linux for the ARM has C++ compatible include files.
7734 * Language & Runtime specific fixes.
7735 + Fix handling of constructor attribute in the C front-end
7736 which caused problems building the Chill runtime library on
7737 some targets.
7738 + Fix minor problem merging type qualifiers in the C front-end.
7739 + Fix aliasing bug for pointers and references (C/C++).
7740 + Fix incorrect "non-constant initializer bug" when
7741 -traditional or -fwritable-strings is enabled.
7742 + Fix build error for Chill front-end on SunOS.
7743 + Do not complain about duplicate instantiations when using
7744 -frepo (C++).
7745 + Fix array bounds handling in C++ front-end which caused
7746 problems with dwarf debugging information in some
7747 circumstances.
7748 + Fix minor namespace problem.
7749 + Fix problem linking java programs.
7751 Additional Changes in GCC 2.95.3
7753 * Generic bugfixes and improvements
7754 + Fix numerous problems that caused incorrect optimization in
7755 the register reloading code.
7756 + Fix numerous problems that caused incorrect optimization in
7757 the loop optimizer.
7758 + Fix aborts in the functions build_insn_chain and scan_loops
7759 under some circumstances.
7760 + Fix an alias analysis bug.
7761 + Fix an infinite compilation bug in the combiner.
7762 + A few problems with complex number support have been fixed.
7763 + It is no longer possible for gcc to act as a fork bomb when
7764 installed incorrectly.
7765 + The -fpack-struct option should be recognized now.
7766 + Fixed a bug that caused incorrect code to be generated due to
7767 a lost stack adjustment.
7768 * Platform specific bugfixes and improvements
7769 + Support building ARM toolchains hosted on Windows.
7770 + Fix attribute calculations in ARM toolchains.
7771 + arm-linux support has been improved.
7772 + Fix a PIC failure on sparc targets.
7773 + On ix86 targets, the regparm attribute should now work
7774 reliably.
7775 + Several updates for the h8300 port.
7776 + Fix problem building libio with glibc 2.2.
7778 Please send FSF & GNU inquiries & questions to [17]gnu@gnu.org. There
7779 are also [18]other ways to contact the FSF.
7781 These pages are maintained by [19]the GCC team.
7784 For questions related to the use of GCC, please consult these web
7785 pages and the [20]GCC manuals. If that fails, the
7786 [21]gcc-help@gcc.gnu.org mailing list might help.
7787 Please send comments on these web pages and the development of GCC
7788 to our developer mailing list at [22]gcc@gnu.org or
7789 [23]gcc@gcc.gnu.org. All of our lists have [24]public archives.
7791 Copyright (C) Free Software Foundation, Inc., 51 Franklin St, Fifth
7792 Floor, Boston, MA 02110, USA.
7794 Verbatim copying and distribution of this entire article is permitted
7795 in any medium, provided this notice is preserved.
7796 Last modified 2006-06-21 [25]Valid XHTML 1.0
7798 References
7800 1. http://gcc.gnu.org/news/spill.html
7801 2. http://gcc.gnu.org/news/lcm.html
7802 3. http://gcc.gnu.org/news/cprop.html
7803 4. http://gcc.gnu.org/news/cfg.html
7804 5. http://gcc.gnu.org/news/dse.html
7805 6. http://gcc.gnu.org/news/hoist.html
7806 7. http://gcc.gnu.org/news/alias.html
7807 8. http://gcc.gnu.org/gcc-2.95/c++features.html
7808 9. http://gcc.gnu.org/onlinedocs/g77/News.html
7809 10. http://gcc.gnu.org/java/gcj-announce.txt
7810 11. http://gcc.gnu.org/news/javaannounce.html
7811 12. http://gcc.gnu.org/c99status.html
7812 13. http://gcc.gnu.org/news/chill.html
7813 14. http://gcc.gnu.org/news/sparc.html
7814 15. http://gcc.gnu.org/news/egcs-vcg.html
7815 16. http://gcc.gnu.org/egcs-1.0/features-2.8.html
7816 17. mailto:gnu@gnu.org
7817 18. http://www.gnu.org/home.html#ContactInfo
7818 19. http://gcc.gnu.org/about.html
7819 20. http://gcc.gnu.org/onlinedocs/
7820 21. mailto:gcc-help@gcc.gnu.org
7821 22. mailto:gcc@gnu.org
7822 23. mailto:gcc@gcc.gnu.org
7823 24. http://gcc.gnu.org/lists.html
7824 25. http://validator.w3.org/check/referer
7825 ======================================================================
7826 http://gcc.gnu.org/gcc-2.95/caveats.html
7828 GCC 2.95 Caveats
7830 * GCC 2.95 will issue an error for invalid asm statements that had
7831 been silently accepted by earlier versions of the compiler. This
7832 is particularly noticeable when compiling older versions of the
7833 Linux kernel (2.0.xx). Please refer to the FAQ (as shipped with
7834 GCC 2.95) for more information on this issue.
7835 * GCC 2.95 implements type based alias analysis to disambiguate
7836 memory references. Some programs, particularly the Linux kernel
7837 violate ANSI/ISO aliasing rules and therefore may not operate
7838 correctly when compiled with GCC 2.95. Please refer to the FAQ (as
7839 shipped with GCC 2.95) for more information on this issue.
7840 * GCC 2.95 has a known bug in its handling of complex variables for
7841 64bit targets. Instead of silently generating incorrect code, GCC
7842 2.95 will issue a fatal error for situations it can not handle.
7843 This primarily affects the Fortran community as Fortran makes more
7844 use of complex variables than C or C++.
7845 * GCC 2.95 has an integrated libstdc++, but does not have an
7846 integrated libg++. Furthermore old libg++ releases will not work
7847 with GCC 2.95. You can retrieve a recent copy of libg++ from the
7848 [1]GCC ftp server.
7849 Note most C++ programs only need libstdc++.
7850 * Exception handling may not work with shared libraries,
7851 particularly on alphas, hppas, rs6000/powerpc and mips based
7852 platforms. Exception handling is known to work on x86 GNU/Linux
7853 platforms with shared libraries.
7854 * In general, GCC 2.95 is more rigorous about rejecting invalid C++
7855 code or deprecated C++ constructs than G++ 2.7, G++ 2.8, EGCS 1.0,
7856 or EGCS 1.1. As a result it may be necessary to fix C++ code
7857 before it will compile with GCC 2.95.
7858 * G++ is also converting toward the ISO C++ standard; as a result
7859 code which was previously valid (and thus accepted by other
7860 compilers and older versions of g++) may no longer be accepted.
7861 The flag -fpermissive may allow some non-conforming code to
7862 compile with GCC 2.95.
7863 * GCC 2.95 compiled C++ code is not binary compatible with EGCS
7864 1.1.x, EGCS 1.0.x or GCC 2.8.x.
7865 * GCC 2.95 does not have changes from the GCC 2.8 tree that were
7866 made between Sept 30, 1998 and April 30, 1999 (the official end of
7867 the GCC 2.8 project). Future GCC releases will include all the
7868 changes from the defunct GCC 2.8 sources.
7870 Please send FSF & GNU inquiries & questions to [2]gnu@gnu.org. There
7871 are also [3]other ways to contact the FSF.
7873 These pages are maintained by [4]the GCC team.
7876 For questions related to the use of GCC, please consult these web
7877 pages and the [5]GCC manuals. If that fails, the
7878 [6]gcc-help@gcc.gnu.org mailing list might help.
7879 Please send comments on these web pages and the development of GCC
7880 to our developer mailing list at [7]gcc@gnu.org or
7881 [8]gcc@gcc.gnu.org. All of our lists have [9]public archives.
7883 Copyright (C) Free Software Foundation, Inc., 51 Franklin St, Fifth
7884 Floor, Boston, MA 02110, USA.
7886 Verbatim copying and distribution of this entire article is permitted
7887 in any medium, provided this notice is preserved.
7888 Last modified 2006-06-21 [10]Valid XHTML 1.0
7890 References
7892 1. ftp://gcc.gnu.org/pub/gcc/infrastructure/libg++-2.8.1.3.tar.gz
7893 2. mailto:gnu@gnu.org
7894 3. http://www.gnu.org/home.html#ContactInfo
7895 4. http://gcc.gnu.org/about.html
7896 5. http://gcc.gnu.org/onlinedocs/
7897 6. mailto:gcc-help@gcc.gnu.org
7898 7. mailto:gcc@gnu.org
7899 8. mailto:gcc@gcc.gnu.org
7900 9. http://gcc.gnu.org/lists.html
7901 10. http://validator.w3.org/check/referer
7902 ======================================================================
7903 http://gcc.gnu.org/egcs-1.1/index.html
7905 EGCS 1.1
7907 September 3, 1998: We are pleased to announce the release of EGCS 1.1.
7908 December 1, 1998: We are pleased to announce the release of EGCS
7909 1.1.1.
7910 March 15, 1999: We are pleased to announce the release of EGCS 1.1.2.
7912 EGCS is a free software project to further the development of the GNU
7913 compilers using an open development environment.
7915 EGCS 1.1 is a major new release of the EGCS compiler system. It has
7916 been [1]extensively tested and is believed to be stable and suitable
7917 for widespread use.
7919 EGCS 1.1 is based on an June 6, 1998 snapshot of the GCC 2.8
7920 development sources; it contains all of the new features found in GCC
7921 2.8.1 as well as all new development from GCC up to June 6, 1998.
7923 EGCS 1.1 also contains many improvements and features not found in GCC
7924 or in older versions of EGCS:
7925 * Global common subexpression elimination and global constant/copy
7926 propagation (aka [2]gcse)
7927 * Ongoing improvements to the [3]alias analysis support to allow for
7928 better optimizations throughout the compiler.
7929 * Vastly improved [4]C++ compiler and integrated C++ runtime
7930 libraries.
7931 * Fixes for the /tmp symlink race security problems.
7932 * New targets including mips16, arm-thumb and 64 bit PowerPC.
7933 * Improvements to GNU Fortran (g77) compiler and runtime library
7934 made since g77 version 0.5.23.
7936 See the [5]new features page for a more complete list of new features
7937 found in EGCS 1.1 releases.
7939 EGCS 1.1.1 is a minor update to fix several serious problems in EGCS
7940 1.1:
7941 * General improvements and fixes
7942 + Avoid some stack overflows when compiling large functions.
7943 + Avoid incorrect loop invariant code motions.
7944 + Fix some core dumps on Linux kernel code.
7945 + Bring back the imake -Di386 and friends fix from EGCS 1.0.2.
7946 + Fix code generation problem in gcse.
7947 + Various documentation related fixes.
7948 * g++/libstdc++ improvements and fixes
7949 + MT safe EH fix for setjmp/longjmp based exception handling.
7950 + Fix a few bad interactions between optimization and exception
7951 handling.
7952 + Fixes for demangling of template names starting with "__".
7953 + Fix a bug that would fail to run destructors in some cases
7954 with -O2.
7955 + Fix 'new' of classes with virtual bases.
7956 + Fix crash building Qt on the Alpha.
7957 + Fix failure compiling WIFEXITED macro on GNU/Linux.
7958 + Fix some -frepo failures.
7959 * g77 and libf2c improvements and fixes
7960 + Various documentation fixes.
7961 + Avoid compiler crash on RAND intrinsic.
7962 + Fix minor bugs in makefiles exposed by BSD make programs.
7963 + Define _XOPEN_SOURCE for libI77 build to avoid potential
7964 problems on some 64-bit systems.
7965 + Fix problem with implicit endfile on rewind.
7966 + Fix spurious recursive I/O errors.
7967 * platform specific improvements and fixes
7968 + Match all versions of UnixWare7.
7969 + Do not assume x86 SVR4 or UnixWare targets can handle stabs.
7970 + Fix PPC/RS6000 LEGITIMIZE_ADDRESS macro and bug in conversion
7971 from unsigned ints to double precision floats.
7972 + Fix ARM ABI issue with NetBSD.
7973 + Fix a few arm code generation bugs.
7974 + Fixincludes will fix additional broken SCO OpenServer header
7975 files.
7976 + Fix a m68k backend bug which caused invalid offsets in reg+d
7977 addresses.
7978 + Fix problems with 64bit AIX 4.3 support.
7979 + Fix handling of long longs for varargs/stdarg functions on
7980 the ppc.
7981 + Minor fixes to CPP predefines for Windows.
7982 + Fix code generation problems with gpr<->fpr copies for 64bit
7983 ppc.
7984 + Fix a few coldfire code generation bugs.
7985 + Fix some more header file problems on SunOS 4.x.
7986 + Fix assert.h handling for RTEMS.
7987 + Fix Windows handling of TREE_SYMBOL_REFERENCED.
7988 + Fix x86 compiler abort in reg-stack pass.
7989 + Fix cygwin/windows problem with section attributes.
7990 + Fix Alpha code generation problem exposed by SMP Linux
7991 kernels.
7992 + Fix typo in m68k 32->64bit integer conversion.
7993 + Make sure target libraries build with -fPIC for PPC & Alpha
7994 targets.
7996 EGCS 1.1.2 is a minor update to fix several serious problems in EGCS
7997 1.1.1:
7998 * General improvements and fixes
7999 + Fix bug in loop optimizer which caused the SPARC (and
8000 potentially other) ports to segfault.
8001 + Fix infinite recursion in alias analysis and combiner code.
8002 + Fix bug in regclass preferencing.
8003 + Fix incorrect loop reversal which caused incorrect code to be
8004 generated for several targets.
8005 + Fix return value for builtin memcpy.
8006 + Reduce compile time for certain loops which exposed quadratic
8007 behavior in the loop optimizer.
8008 + Fix bug which caused volatile memory to be written multiple
8009 times when only one write was needed/desired.
8010 + Fix compiler abort in caller-save.c
8011 + Fix combiner bug which caused incorrect code generation for
8012 certain division by constant operations.
8013 + Fix incorrect code generation due to a bug in range check
8014 optimizations.
8015 + Fix incorrect code generation due to mis-handling of
8016 clobbered values in CSE.
8017 + Fix compiler abort/segfault due to incorrect register
8018 splitting when unrolling loops.
8019 + Fix code generation involving autoincremented addresses with
8020 ternary operators.
8021 + Work around bug in the scheduler which caused qt to be
8022 mis-compiled on some platforms.
8023 + Fix code generation problems with -fshort-enums.
8024 + Tighten security for temporary files.
8025 + Improve compile time for codes which make heavy use of
8026 overloaded functions.
8027 + Fix multiply defined constructor/destructor symbol problems.
8028 + Avoid setting bogus RPATH environment variable during
8029 bootstrap.
8030 + Avoid GNU-make dependencies in the texinfo subdir.
8031 + Install CPP wrapper script in $(prefix)/bin if --enable-cpp.
8032 --enable-cpp=<dirname> can be used to specify an additional
8033 install directory for the cpp wrapper script.
8034 + Fix CSE bug which caused incorrect label-label refs to appear
8035 on some platforms.
8036 + Avoid linking in EH routines from libgcc if they are not
8037 needed.
8038 + Avoid obscure bug in aliasing code.
8039 + Fix bug in weak symbol handling.
8040 * Platform-specific improvements and fixes
8041 + Fix detection of PPro/PII on Unixware 7.
8042 + Fix compiler segfault when building spec99 and other programs
8043 for SPARC targets.
8044 + Fix code-generation bugs for integer and floating point
8045 conditional move instructions on the PPro/PII.
8046 + Use fixincludes to fix byteorder problems on i?86-*-sysv.
8047 + Fix build failure for the arc port.
8048 + Fix floating point format configuration for i?86-gnu port.
8049 + Fix problems with hppa1.0-hp-hpux10.20 configuration when
8050 threads are enabled.
8051 + Fix coldfire code generation bugs.
8052 + Fix "unrecognized insn" problems for Alpha and PPC ports.
8053 + Fix h8/300 code generation problem with floating point values
8054 in memory.
8055 + Fix unrecognized insn problems for the m68k port.
8056 + Fix namespace-pollution problem for the x86 port.
8057 + Fix problems with old assembler on x86 NeXT systems.
8058 + Fix PIC code-generation problems for the SPARC port.
8059 + Fix minor bug with LONG_CALLS in PowerPC SVR4 support.
8060 + Fix minor ISO namespace violation in Alpha varargs/stdarg
8061 support.
8062 + Fix incorrect "braf" instruction usage for the SH port.
8063 + Fix minor bug in va-sh which prevented its use with -ansi.
8064 + Fix problems recognizing and supporting FreeBSD.
8065 + Handle OpenBSD systems correctly.
8066 + Minor fixincludes fix for Digital UNIX 4.0B.
8067 + Fix problems with ctors/dtors in SCO shared libraries.
8068 + Abort instead of generating incorrect code for PPro/PII
8069 floating point conditional moves.
8070 + Avoid multiply defined symbols on Linux/GNU systems using
8071 libc-5.4.xx.
8072 + Fix abort in alpha compiler.
8073 * Fortran-specific fixes
8074 + Fix the IDate intrinsic (VXT) (in libg2c) so the returned
8075 year is in the documented, non-Y2K-compliant range of 0-99,
8076 instead of being returned as 100 in the year 2000.
8077 + Fix the `Date_and_Time' intrinsic (in libg2c) to return the
8078 milliseconds value properly in Values(8).
8079 + Fix the `LStat' intrinsic (in libg2c) to return device-ID
8080 information properly in SArray(7).
8082 Each release includes installation instructions in both HTML and
8083 plaintext forms (see the INSTALL directory in the toplevel directory
8084 of the distribution). However, we also keep the most up to date
8085 [6]installation instructions and [7]build/test status on our web page.
8086 We will update those pages as new information becomes available.
8088 The EGCS project would like to thank the numerous people that have
8089 contributed new features, test results, bugfixes, etc. This [8]amazing
8090 group of volunteers is what makes EGCS successful.
8092 And finally, we can't in good conscience fail to mention some
8093 [9]caveats to using EGCS 1.1.
8095 Download EGCS from egcs.cygnus.com (USA California).
8097 The EGCS 1.1 release is also available on many mirror sites.
8098 [10]Goto mirror list to find a closer site.
8100 Please send FSF & GNU inquiries & questions to [11]gnu@gnu.org. There
8101 are also [12]other ways to contact the FSF.
8103 These pages are maintained by [13]the GCC team.
8106 For questions related to the use of GCC, please consult these web
8107 pages and the [14]GCC manuals. If that fails, the
8108 [15]gcc-help@gcc.gnu.org mailing list might help.
8109 Please send comments on these web pages and the development of GCC
8110 to our developer mailing list at [16]gcc@gnu.org or
8111 [17]gcc@gcc.gnu.org. All of our lists have [18]public archives.
8113 Copyright (C) Free Software Foundation, Inc., 51 Franklin St, Fifth
8114 Floor, Boston, MA 02110, USA.
8116 Verbatim copying and distribution of this entire article is permitted
8117 in any medium, provided this notice is preserved.
8118 Last modified 2006-06-21 [19]Valid XHTML 1.0
8120 References
8122 1. http://gcc.gnu.org/egcs-1.1/egcs-1.1-test.html
8123 2. http://gcc.gnu.org/news/gcse.html
8124 3. http://gcc.gnu.org/news/alias.html
8125 4. http://gcc.gnu.org/egcs-1.1/c++features.html
8126 5. http://gcc.gnu.org/egcs-1.1/features.html
8127 6. http://gcc.gnu.org/install/
8128 7. http://gcc.gnu.org/egcs-1.1/buildstat.html
8129 8. http://gcc.gnu.org/onlinedocs/gcc/Contributors.html
8130 9. http://gcc.gnu.org/egcs-1.1/caveats.html
8131 10. http://gcc.gnu.org/mirrors.html
8132 11. mailto:gnu@gnu.org
8133 12. http://www.gnu.org/home.html#ContactInfo
8134 13. http://gcc.gnu.org/about.html
8135 14. http://gcc.gnu.org/onlinedocs/
8136 15. mailto:gcc-help@gcc.gnu.org
8137 16. mailto:gcc@gnu.org
8138 17. mailto:gcc@gcc.gnu.org
8139 18. http://gcc.gnu.org/lists.html
8140 19. http://validator.w3.org/check/referer
8141 ======================================================================
8142 http://gcc.gnu.org/egcs-1.1/features.html
8144 EGCS 1.1 new features
8146 * Integrated GNU Fortran (g77) compiler and runtime library with
8147 improvements, based on [1]g77 version 0.5.23.
8148 * Vast improvements in the C++ compiler; so many they have [2]page
8149 of their own!
8150 * Compiler implements [3]global common subexpression elimination and
8151 global copy/constant propagation.
8152 * More major improvements in the [4]alias analysis code.
8153 * More major improvements in the exception handling code to improve
8154 performance, lower static overhead and provide the infrastructure
8155 for future improvements.
8156 * The infamous /tmp symlink race security problems have been fixed.
8157 * The regmove optimization pass has been nearly completely rewritten
8158 to improve performance of generated code.
8159 * The compiler now recomputes register usage information before
8160 local register allocation. By providing more accurate information
8161 to the priority based allocator, we get better register
8162 allocation.
8163 * The register reloading phase of the compiler optimizes spill code
8164 much better than in previous releases.
8165 * Some bad interactions between the register allocator and
8166 instruction scheduler have been fixed, resulting in much better
8167 code for certain programs. Additionally, we have tuned the
8168 scheduler in various ways to improve performance of generated code
8169 for some architectures.
8170 * The compiler's branch shortening algorithms have been
8171 significantly improved to work better on targets which align jump
8172 targets.
8173 * The compiler now supports -Os to prefer optimizing for code space
8174 over optimizing for code speed.
8175 * The compiler will now totally eliminate library calls which
8176 compute constant values. This primarily helps targets with no
8177 integer div/mul support and targets without floating point
8178 support.
8179 * The compiler now supports an extensive "--help" option.
8180 * cpplib has been greatly improved and may be suitable for limited
8181 use.
8182 * Memory footprint for the compiler has been significantly reduced
8183 for some pathological cases.
8184 * The time to build EGCS has been improved for certain targets
8185 (particularly the alpha and mips platforms).
8186 * Many infrastructure improvements throughout the compiler, plus the
8187 usual mountain of bugfixes and minor improvements.
8188 * Target dependent improvements:
8189 + SPARC port now includes V8 plus and V9 support as well as
8190 performance tuning for Ultra class machines. The SPARC port
8191 now uses the Haifa scheduler.
8192 + Alpha port has been tuned for the EV6 processor and has an
8193 optimized expansion of memcpy/bzero. The Alpha port now uses
8194 the Haifa scheduler.
8195 + RS6000/PowerPC: support for the Power64 architecture and AIX
8196 4.3. The RS6000/PowerPC port now uses the Haifa scheduler.
8197 + x86: Alignment of static store data and jump targets is per
8198 Intel recommendations now. Various improvements throughout
8199 the x86 port to improve performance on Pentium processors
8200 (including improved epilogue sequences for Pentium chips and
8201 backend improvements which should help register allocation on
8202 all x86 variants. Conditional move support has been fixed and
8203 enabled for PPro processors. The x86 port also better
8204 supports 64bit operations now. Unixware 7, a System V Release
8205 5 target, is now supported and SCO OpenServer targets can
8206 support GAS.
8207 + MIPS has improved multiply/multiply-add support and now
8208 includes mips16 ISA support.
8209 + M68k has many micro-optimizations and Coldfire fixes.
8210 * Core compiler is based on the GCC development tree from June 9,
8211 1998, so we have all of the [5]features found in GCC 2.8.
8213 Please send FSF & GNU inquiries & questions to [6]gnu@gnu.org. There
8214 are also [7]other ways to contact the FSF.
8216 These pages are maintained by [8]the GCC team.
8219 For questions related to the use of GCC, please consult these web
8220 pages and the [9]GCC manuals. If that fails, the
8221 [10]gcc-help@gcc.gnu.org mailing list might help.
8222 Please send comments on these web pages and the development of GCC
8223 to our developer mailing list at [11]gcc@gnu.org or
8224 [12]gcc@gcc.gnu.org. All of our lists have [13]public archives.
8226 Copyright (C) Free Software Foundation, Inc., 51 Franklin St, Fifth
8227 Floor, Boston, MA 02110, USA.
8229 Verbatim copying and distribution of this entire article is permitted
8230 in any medium, provided this notice is preserved.
8231 Last modified 2006-06-21 [14]Valid XHTML 1.0
8233 References
8235 1. http://gcc.gnu.org/onlinedocs/g77/News.html
8236 2. http://gcc.gnu.org/egcs-1.1/c++features.html
8237 3. http://gcc.gnu.org/news/gcse.html
8238 4. http://gcc.gnu.org/news/alias.html
8239 5. http://gcc.gnu.org/egcs-1.0/features-2.8.html
8240 6. mailto:gnu@gnu.org
8241 7. http://www.gnu.org/home.html#ContactInfo
8242 8. http://gcc.gnu.org/about.html
8243 9. http://gcc.gnu.org/onlinedocs/
8244 10. mailto:gcc-help@gcc.gnu.org
8245 11. mailto:gcc@gnu.org
8246 12. mailto:gcc@gcc.gnu.org
8247 13. http://gcc.gnu.org/lists.html
8248 14. http://validator.w3.org/check/referer
8249 ======================================================================
8250 http://gcc.gnu.org/egcs-1.1/caveats.html
8252 EGCS 1.1 Caveats
8254 * EGCS has an integrated libstdc++, but does not have an integrated
8255 libg++. Furthermore old libg++ releases will not work with EGCS;
8256 HJ Lu has made a libg++-2.8.1.2 snapshot available which may work
8257 with EGCS.
8258 Note most C++ programs only need libstdc++.
8259 * Exception handling may not work with shared libraries,
8260 particularly on alphas, hppas, rs6000/powerpc and mips based
8261 platforms. Exception handling is known to work on x86-linux
8262 platforms with shared libraries.
8263 * Some versions of the Linux kernel have bugs which prevent them
8264 from being compiled or from running when compiled by EGCS. See the
8265 FAQ (as shipped with EGCS 1.1) for additional information.
8266 * In general, EGCS is more rigorous about rejecting invalid C++ code
8267 or deprecated C++ constructs than g++-2.7, g++-2.8 or EGCS 1.0. As
8268 a result it may be necessary to fix C++ code before it will
8269 compile with EGCS.
8270 * G++ is also converting toward the ISO C++ standard; as a result
8271 code which was previously valid (and thus accepted by other
8272 compilers and older versions of g++) may no longer be accepted.
8273 * EGCS 1.1 compiled C++ code is not binary compatible with EGCS
8274 1.0.x or GCC 2.8.x due to changes necessary to support thread safe
8275 exception handling.
8277 Please send FSF & GNU inquiries & questions to [1]gnu@gnu.org. There
8278 are also [2]other ways to contact the FSF.
8280 These pages are maintained by [3]the GCC team.
8283 For questions related to the use of GCC, please consult these web
8284 pages and the [4]GCC manuals. If that fails, the
8285 [5]gcc-help@gcc.gnu.org mailing list might help.
8286 Please send comments on these web pages and the development of GCC
8287 to our developer mailing list at [6]gcc@gnu.org or
8288 [7]gcc@gcc.gnu.org. All of our lists have [8]public archives.
8290 Copyright (C) Free Software Foundation, Inc., 51 Franklin St, Fifth
8291 Floor, Boston, MA 02110, USA.
8293 Verbatim copying and distribution of this entire article is permitted
8294 in any medium, provided this notice is preserved.
8295 Last modified 2006-06-21 [9]Valid XHTML 1.0
8297 References
8299 1. mailto:gnu@gnu.org
8300 2. http://www.gnu.org/home.html#ContactInfo
8301 3. http://gcc.gnu.org/about.html
8302 4. http://gcc.gnu.org/onlinedocs/
8303 5. mailto:gcc-help@gcc.gnu.org
8304 6. mailto:gcc@gnu.org
8305 7. mailto:gcc@gcc.gnu.org
8306 8. http://gcc.gnu.org/lists.html
8307 9. http://validator.w3.org/check/referer
8308 ======================================================================
8309 http://gcc.gnu.org/egcs-1.0/index.html
8311 EGCS 1.0
8313 December 3, 1997: We are pleased to announce the release of EGCS 1.0.
8314 January 6, 1998: We are pleased to announce the release of EGCS 1.0.1.
8315 March 16, 1998: We are pleased to announce the release of EGCS 1.0.2.
8316 May 15, 1998 We are pleased to announce the release of EGCS 1.0.3.
8318 EGCS is a collaborative effort involving several groups of hackers
8319 using an open development model to accelerate development and testing
8320 of GNU compilers and runtime libraries.
8322 An important goal of EGCS is to allow wide scale testing of
8323 experimental features and optimizations; therefore, EGCS contains some
8324 features and optimizations which are still under development. However,
8325 EGCS has been carefully tested and should be comparable in quality to
8326 most GCC releases.
8328 EGCS 1.0 is based on an August 2, 1997 snapshot of the GCC 2.8
8329 development sources; it contains nearly all of the new features found
8330 in GCC 2.8.
8332 EGCS 1.0 also contains many improvements and features not found in GCC
8333 2.7 and even the GCC 2.8 series (which was released after the original
8334 EGCS 1.0 release).
8335 * Integrated C++ runtime libraries, including support for most major
8336 GNU/Linux systems!
8337 * The integrated libstdc++ library includes a verbatim copy of SGI's
8338 STL release.
8339 * Integrated GNU Fortran compiler.
8340 * New instruction scheduler.
8341 * New alias analysis code.
8343 See the [1]new features page for a more complete list of new features.
8345 EGCS 1.0.1 is a minor update to the EGCS 1.0 compiler to fix a few
8346 critical bugs and add support for Red Hat 5.0 Linux. Changes since the
8347 EGCS 1.0 release:
8348 * Add support for Red Hat 5.0 Linux and better support for Linux
8349 systems using glibc2.
8350 Many programs failed to link when compiled with EGCS 1.0 on Red
8351 Hat 5.0 or on systems with newer versions of glibc2. EGCS 1.0.1
8352 should fix these problems.
8353 * Compatibility with both EGCS 1.0 and GCC 2.8 libgcc exception
8354 handling interfaces.
8355 To avoid future compatibility problems, we strongly urge anyone
8356 who is planning on distributing shared libraries that contain C++
8357 code to upgrade to EGCS 1.0.1 first.
8358 Soon after EGCS 1.0 was released, the GCC developers made some
8359 incompatible changes in libgcc's exception handling interfaces.
8360 These changes were needed to solve problems on some platforms.
8361 This means that GCC 2.8.0, when released, will not be seamlessly
8362 compatible with shared libraries built by EGCS 1.0. The reason is
8363 that the libgcc.a in GCC 2.8.0 will not contain a function needed
8364 by the old interface.
8365 The result of this is that there may be compatibility problems
8366 with shared libraries built by EGCS 1.0 when used with GCC 2.8.0.
8367 With EGCS 1.0.1, generated code uses the new (GCC 2.8.0)
8368 interface, and libgcc.a has the support routines for both the old
8369 and the new interfaces (so EGCS 1.0.1 and EGCS 1.0 code can be
8370 freely mixed, and EGCS 1.0.1 and GCC 2.8.0 code can be freely
8371 mixed).
8372 The maintainers of GCC 2.x have decided against including seamless
8373 support for the old interface in 2.8.0, since it was never
8374 "official", so to avoid future compatibility problems we recommend
8375 against distributing any shared libraries built by EGCS 1.0 that
8376 contain C++ code (upgrade to 1.0.1 and use that).
8377 * Various bugfixes in the x86, hppa, mips, and rs6000/ppc backends.
8378 The x86 changes fix code generation errors exposed when building
8379 glibc2 and the Linux dynamic linker (ld.so).
8380 The hppa change fixes a compiler abort when configured for use
8381 with RTEMS.
8382 The MIPS changes fix problems with the definition of LONG_MAX on
8383 newer systems, allow for command line selection of the target ABI,
8384 and fix one code generation problem.
8385 The rs6000/ppc change fixes some problems with passing structures
8386 to varargs/stdarg functions.
8387 * A few machine independent bugfixes, mostly to fix code generation
8388 errors when building Linux kernels or glibc.
8389 * Fix a few critical exception handling and template bugs in the C++
8390 compiler.
8391 * Fix Fortran namelist bug on alphas.
8392 * Fix build problems on x86-solaris systems.
8394 EGCS 1.0.2 is a minor update to the EGCS 1.0.1 compiler to fix several
8395 serious problems in EGCS 1.0.1.
8396 * General improvements and fixes
8397 + Memory consumption significantly reduced, especially for
8398 templates and inline functions.
8399 + Fix various problems with glibc2.1.
8400 + Fix loop optimization bug exposed by rs6000/ppc port.
8401 + Fix to avoid potential code generation problems in jump.c.
8402 + Fix some undefined symbol problems in dwarf1 debug support.
8403 * g++/libstdc++ improvements and fixes
8404 + libstdc++ in the EGCS release has been updated and should be
8405 link compatible with libstdc++-2.8.
8406 + Various fixes in libio/libstdc++ to work better on Linux
8407 systems.
8408 + Fix problems with duplicate symbols on systems that do not
8409 support weak symbols.
8410 + Memory corruption bug and undefined symbols in bastring have
8411 been fixed.
8412 + Various exception handling fixes.
8413 + Fix compiler abort for very long thunk names.
8414 * g77 improvements and fixes
8415 + Fix compiler crash for omitted bound in Fortran CASE
8416 statement.
8417 + Add missing entries to g77 lang-options.
8418 + Fix problem with -fpedantic in the g77 compiler.
8419 + Fix "backspace" problem with g77 on alphas.
8420 + Fix x86 backend problem with Fortran literals and -fpic.
8421 + Fix some of the problems with negative subscripts for g77 on
8422 alphas.
8423 + Fixes for Fortran builds on cygwin32/mingw32.
8424 * platform specific improvements and fixes
8425 + Fix long double problems on x86 (exposed by glibc).
8426 + x86 ports define i386 again to keep imake happy.
8427 + Fix exception handling support on NetBSD ports.
8428 + Several changes to collect2 to fix many problems with AIX.
8429 + Define __ELF__ for rs6000/linux.
8430 + Fix -mcall-linux problem on rs6000/linux.
8431 + Fix stdarg/vararg problem for rs6000/linux.
8432 + Allow autoconf to select a proper install problem on AIX 3.1.
8433 + m68k port support includes -mcpu32 option as well as cpu32
8434 multilibs.
8435 + Fix stdarg bug for irix6.
8436 + Allow EGCS to build on irix5 without the gnu assembler.
8437 + Fix problem with static linking on sco5.
8438 + Fix bootstrap on sco5 with native compiler.
8439 + Fix for abort building newlib on H8 target.
8440 + Fix fixincludes handling of math.h on SunOS.
8441 + Minor fix for Motorola 3300 m68k systems.
8443 EGCS 1.0.3 is a minor update to the EGCS 1.0.2 compiler to fix a few
8444 problems reported by Red Hat for builds of Red Hat 5.1.
8445 * Generic bugfixes:
8446 + Fix a typo in the libio library which resulted in incorrect
8447 behavior of istream::get.
8448 + Fix the Fortran negative array index problem.
8449 + Fix a major problem with the ObjC runtime thread support
8450 exposed by glibc2.
8451 + Reduce memory consumption of the Haifa scheduler.
8452 * Target specific bugfixes:
8453 + Fix one x86 floating point code generation bug exposed by
8454 glibc2 builds.
8455 + Fix one x86 internal compiler error exposed by glibc2 builds.
8456 + Fix profiling bugs on the Alpha.
8457 + Fix ImageMagick & emacs 20.2 build problems on the Alpha.
8458 + Fix rs6000/ppc bug when converting values from integer types
8459 to floating point types.
8461 The EGCS 1.0 releases include installation instructions in both HTML
8462 and plaintext forms (see the INSTALL directory in the toplevel
8463 directory of the distribution). However, we also keep the most up to
8464 date [2]installation instructions and [3]build/test status on our web
8465 page. We will update those pages as new information becomes available.
8467 And, we can't in good conscience fail to mention some [4]caveats to
8468 using EGCS.
8470 Update: Big thanks to Stanford for providing a high speed link for
8471 downloading EGCS (go.cygnus.com)!
8473 Download EGCS from ftp.cygnus.com (USA California) or go.cygnus.com
8474 (USA California -- High speed link provided by Stanford).
8476 The EGCS 1.0 release is also available many mirror sites.
8477 [5]Goto mirror list to find a closer site
8479 We'd like to thank the numerous people that have contributed new
8480 features, test results, bugfixes, etc. Unfortunately, they're far too
8481 numerous to mention by name.
8483 Please send FSF & GNU inquiries & questions to [6]gnu@gnu.org. There
8484 are also [7]other ways to contact the FSF.
8486 These pages are maintained by [8]the GCC team.
8489 For questions related to the use of GCC, please consult these web
8490 pages and the [9]GCC manuals. If that fails, the
8491 [10]gcc-help@gcc.gnu.org mailing list might help.
8492 Please send comments on these web pages and the development of GCC
8493 to our developer mailing list at [11]gcc@gnu.org or
8494 [12]gcc@gcc.gnu.org. All of our lists have [13]public archives.
8496 Copyright (C) Free Software Foundation, Inc., 51 Franklin St, Fifth
8497 Floor, Boston, MA 02110, USA.
8499 Verbatim copying and distribution of this entire article is permitted
8500 in any medium, provided this notice is preserved.
8501 Last modified 2006-06-21 [14]Valid XHTML 1.0
8503 References
8505 1. http://gcc.gnu.org/egcs-1.0/features.html
8506 2. http://gcc.gnu.org/install/
8507 3. http://gcc.gnu.org/egcs-1.0/buildstat.html
8508 4. http://gcc.gnu.org/egcs-1.0/caveats.html
8509 5. http://gcc.gnu.org/mirrors.html
8510 6. mailto:gnu@gnu.org
8511 7. http://www.gnu.org/home.html#ContactInfo
8512 8. http://gcc.gnu.org/about.html
8513 9. http://gcc.gnu.org/onlinedocs/
8514 10. mailto:gcc-help@gcc.gnu.org
8515 11. mailto:gcc@gnu.org
8516 12. mailto:gcc@gcc.gnu.org
8517 13. http://gcc.gnu.org/lists.html
8518 14. http://validator.w3.org/check/referer
8519 ======================================================================
8520 http://gcc.gnu.org/egcs-1.0/features.html
8522 EGCS 1.0 features
8524 * Core compiler is based on the gcc2 development tree from Aug 2,
8525 1997, so we have most of the [1]features found in GCC 2.8.
8526 * Integrated GNU Fortran compiler based on g77-0.5.22-19970929.
8527 * Vast improvements in the C++ compiler; so many they have [2]page
8528 of their own!
8529 * Integrated C++ runtime libraries, including support for most major
8530 linux systems!
8531 * New instruction scheduler from IBM Haifa which includes support
8532 for function wide instruction scheduling as well as superscalar
8533 scheduling.
8534 * Significantly improved alias analysis code.
8535 * Improved register allocation for two address machines.
8536 * Significant code generation improvements for Fortran code on
8537 Alphas.
8538 * Various optimizations from the g77 project as well as improved
8539 loop optimizations.
8540 * Dwarf2 debug format support for some targets.
8541 * egcs libstdc++ includes the SGI STL implementation without
8542 changes.
8543 * As a result of these and other changes, egcs libstc++ is not
8544 binary compatible with previous releases of libstdc++.
8545 * Various new ports -- UltraSPARC, Irix6.2 & Irix6.3 support, The
8546 SCO Openserver 5 family (5.0.{0,2,4} and Internet FastStart 1.0
8547 and 1.1), Support for RTEMS on several embedded targets, Support
8548 for arm-linux, Mitsubishi M32R, Hitachi H8/S, Matsushita MN102 and
8549 MN103, NEC V850, Sparclet, Solaris & Linux on PowerPCs, etc.
8550 * Integrated testsuites for gcc, g++, g77, libstdc++ and libio.
8551 * RS6000/PowerPC ports generate code which can run on all
8552 RS6000/PowerPC variants by default.
8553 * -mcpu= and -march= switches for the x86 port to allow better
8554 control over how the x86 port generates code.
8555 * Includes the template repository patch (aka repo patch); note the
8556 new template code makes repo obsolete for ELF systems using gnu-ld
8557 such as Linux.
8558 * Plus the usual assortment of bugfixes and improvements.
8560 Please send FSF & GNU inquiries & questions to [3]gnu@gnu.org. There
8561 are also [4]other ways to contact the FSF.
8563 These pages are maintained by [5]the GCC team.
8566 For questions related to the use of GCC, please consult these web
8567 pages and the [6]GCC manuals. If that fails, the
8568 [7]gcc-help@gcc.gnu.org mailing list might help.
8569 Please send comments on these web pages and the development of GCC
8570 to our developer mailing list at [8]gcc@gnu.org or
8571 [9]gcc@gcc.gnu.org. All of our lists have [10]public archives.
8573 Copyright (C) Free Software Foundation, Inc., 51 Franklin St, Fifth
8574 Floor, Boston, MA 02110, USA.
8576 Verbatim copying and distribution of this entire article is permitted
8577 in any medium, provided this notice is preserved.
8578 Last modified 2006-06-21 [11]Valid XHTML 1.0
8580 References
8582 1. http://gcc.gnu.org/egcs-1.0/features-2.8.html
8583 2. http://gcc.gnu.org/egcs-1.0/c++features.html
8584 3. mailto:gnu@gnu.org
8585 4. http://www.gnu.org/home.html#ContactInfo
8586 5. http://gcc.gnu.org/about.html
8587 6. http://gcc.gnu.org/onlinedocs/
8588 7. mailto:gcc-help@gcc.gnu.org
8589 8. mailto:gcc@gnu.org
8590 9. mailto:gcc@gcc.gnu.org
8591 10. http://gcc.gnu.org/lists.html
8592 11. http://validator.w3.org/check/referer
8593 ======================================================================
8594 http://gcc.gnu.org/egcs-1.0/caveats.html
8596 EGCS 1.0 Caveats
8598 * EGCS has an integrated libstdc++, but does not have an integrated
8599 libg++. Furthermore old libg++ releases will not work with egc; HJ
8600 Lu has made a libg++-2.8.1.2 available which may work with EGCS.
8601 Note most C++ programs only need libstdc++.
8602 * Note that using -pedantic or -Wreturn-type can cause an explosion
8603 in the amount of memory needed for template-heavy C++ code, such
8604 as code that uses STL. Also note that -Wall includes
8605 -Wreturn-type, so if you use -Wall you will need to specify
8606 -Wno-return-type to turn it off.
8607 * Exception handling may not work with shared libraries,
8608 particularly on alphas, hppas, and mips based platforms. Exception
8609 handling is known to work on x86-linux platforms with shared
8610 libraries.
8611 * Some versions of the Linux kernel have bugs which prevent them
8612 from being compiled or from running when compiled by EGCS. See the
8613 FAQ (as shipped with EGCS 1.0) for additional information.
8614 * In general, EGCS is more rigorous about rejecting invalid C++ code
8615 or deprecated C++ constructs than G++ 2.7. As a result it may be
8616 necessary to fix C++ code before it will compile with EGCS.
8617 * G++ is also aggressively tracking the C++ standard; as a result
8618 code which was previously valid (and thus accepted by other
8619 compilers and older versions of G++) may no longer be accepted.
8620 * EGCS 1.0 may not work with Red Hat Linux 5.0 on all targets. EGCS
8621 1.0.x and later releases should work with Red Hat Linux 5.0.
8623 Please send FSF & GNU inquiries & questions to [1]gnu@gnu.org. There
8624 are also [2]other ways to contact the FSF.
8626 These pages are maintained by [3]the GCC team.
8629 For questions related to the use of GCC, please consult these web
8630 pages and the [4]GCC manuals. If that fails, the
8631 [5]gcc-help@gcc.gnu.org mailing list might help.
8632 Please send comments on these web pages and the development of GCC
8633 to our developer mailing list at [6]gcc@gnu.org or
8634 [7]gcc@gcc.gnu.org. All of our lists have [8]public archives.
8636 Copyright (C) Free Software Foundation, Inc., 51 Franklin St, Fifth
8637 Floor, Boston, MA 02110, USA.
8639 Verbatim copying and distribution of this entire article is permitted
8640 in any medium, provided this notice is preserved.
8641 Last modified 2006-06-21 [9]Valid XHTML 1.0
8643 References
8645 1. mailto:gnu@gnu.org
8646 2. http://www.gnu.org/home.html#ContactInfo
8647 3. http://gcc.gnu.org/about.html
8648 4. http://gcc.gnu.org/onlinedocs/
8649 5. mailto:gcc-help@gcc.gnu.org
8650 6. mailto:gcc@gnu.org
8651 7. mailto:gcc@gcc.gnu.org
8652 8. http://gcc.gnu.org/lists.html
8653 9. http://validator.w3.org/check/referer
8654 ======================================================================