/usr/share/doc/gcc-4.5-base/NEWS.html is in gcc-4.5 4.5.3-12ubuntu2.
This file is owned by root:root, with mode 0o644.
The actual contents of the file can be viewed below.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 | <?xml version="1.0" encoding="ISO-8859-1"?>
<!DOCTYPE html
PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"
"http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en" lang="en">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1" />
<link rev="made" href="mailto:gcc@gcc.gnu.org" />
<link rel="shortcut icon" href="http://gcc.gnu.org/favicon.ico" />
<link rel="stylesheet" type="text/css" href="/gnu.css" />
<link rel="stylesheet" type="text/css" href="http://gcc.gnu.org/gcc.css" />
<title>
GCC 4.5 Release Series — Changes, New Features, and Fixes
- GNU Project - Free Software Foundation (FSF)</title>
</head>
<!-- GCC maintainers, please do not hesitate to update/contribute entries
concerning those part of GCC you maintain! 2002-03-23, Gerald.
-->
<body bgcolor="#FFFFFF" text="#000000" link="#1F00FF" alink="#FF0000" vlink="#9900DD">
<h1 align="center">
GCC 4.5 Release Series<br />Changes, New Features, and Fixes
</h1>
<h2>Caveats</h2>
<ul>
<li id="mpccaveat">GCC now requires the <a
href="http://www.multiprecision.org/">MPC</a> library in order to
build. See the <a
href="http://gcc.gnu.org/install/prerequisites.html">prerequisites
page</a> for version requirements.</li>
<li><p>Support for a number of older systems and recently
unmaintained or untested target ports of GCC has been declared
obsolete in GCC 4.5. Unless there is activity to revive them, the
next release of GCC will have their sources permanently
<strong>removed</strong>.</p>
<p id="obsoleted">The following ports for individual systems on
particular architectures have been obsoleted:</p>
<ul>
<li>IRIX releases before 6.5 (mips-sgi-irix5*,
mips-sgi-irix6.[0-4])</li>
<li>Solaris 7 (*-*-solaris2.7)</li>
<li>Tru64 UNIX releases before V5.1 (alpha*-dec-osf4*,
alpha-dec-osf5.0*)</li>
<li>Details for the IRIX, Solaris 7, and Tru64 UNIX obsoletions can
be found in the <a
href="http://gcc.gnu.org/ml/gcc/2010-01/msg00510.html">announcement</a>.</li>
</ul>
</li>
<li>Support has been removed for all the
<a href="../gcc-4.4/changes.html#obsoleted">configurations obsoleted
in GCC 4.4</a>.</li>
<li>Support has been removed for the <code>protoize</code>
and <code>unprotoize</code> utilities, obsoleted in GCC 4.4.</li>
<li>Support has been removed for tuning for Itanium1 (Merced) variants.
Note that code tuned for Itanium2 should also run correctly on Itanium1.</li>
<li>GCC now generates unwind info also for epilogues. DWARF debuginfo
generated by GCC now uses more features of DWARF3 than it used to
do and also some DWARF4 features. GDB older than 7.0 is not able to
handle either of these, so to debug GCC 4.5 generated binaries or
libraries GDB 7.0 or later is needed. You can disable use of DWARF4
features with <code>-gdwarf-3 -gstrict-dwarf</code> options, or with
<code>-gdwarf-2 -gstrict-dwarf</code> restrict GCC to just DWARF2
standard, but epilogue unwind info is emitted unconditionally whenever
unwind info is emitted. </li>
<li>On x86 targets, code containing floating-point calculations may
run significantly slower when compiled with GCC 4.5 in strict C99
conformance mode than they did with earlier GCC versions. This is due
to stricter standard conformance of the compiler and can be avoided by
using the option <code>-fexcess-precision=fast</code>; also see
<a href="#x86">below</a>.</li>
<li>The function attribute <code>noinline</code> no longer prevents GCC
from cloning the function. A new attribute <code>noclone</code>
has been introduced for this purpose. Cloning a function means
that it is duplicated and the new copy is specialized for certain
contexts (for example when a parameter is a known constant).</li>
</ul>
<h2>General Optimizer Improvements</h2>
<ul>
<li>The <code>-save-temps</code> now takes an optional argument. The
<code>-save-temps</code> and <code>-save-temps=cwd</code> switches write
the temporary files in the current working directory based on the original
source file. The <code>-save-temps=obj</code> switch will write files into
the directory specified with the <code>-o</code> option, and the
intermediate filenames are based on the output file. This will allow the
user to get the compiler intermediate files when doing parallel builds
without two builds of the same filename located in different directories
from interfering with each other.</li>
<li>Debugging dumps are now created in the same directory as the
object file rather than in the current working directory. This
allows the user to get debugging dumps when doing parallel builds
without two builds of the same filename interfering with each other.</li>
<li id="mpcopts">GCC has been integrated with the <a
href="http://www.multiprecision.org/">MPC</a> library. This
allows GCC to evaluate complex arithmetic at compile time <a
href="http://gcc.gnu.org/bugzilla/show_bug.cgi?id=30789">more
accurately</a>. It also allows GCC to evaluate calls to complex
built-in math functions having constant arguments and replace them
at compile time with their mathematically equivalent results. In
doing so, GCC can generate correct results regardless of the math
library implementation or floating point precision of the host
platform. This also allows GCC to generate identical results
regardless of whether one compiles in native or cross-compile
configurations to a particular target. The following built-in
functions take advantage of this new capability:
<code>cacos</code>, <code>cacosh</code>, <code>casin</code>,
<code>casinh</code>, <code>catan</code>, <code>catanh</code>,
<code>ccos</code>, <code>ccosh</code>, <code>cexp</code>,
<code>clog</code>, <code>cpow</code>, <code>csin</code>,
<code>csinh</code>, <code>csqrt</code>, <code>ctan</code>, and
<code>ctanh</code>. The <code>float</code> and <code>long
double</code> variants of these functions (e.g. <code>csinf</code>
and <code>csinl</code>) are also handled.</li>
<li>A new link-time optimizer has been added (<code><a
href="http://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html#index-flto-801"
>-flto</a></code>).
When this option is used, GCC generates a bytecode representation of
each input file and writes it to special ELF sections in each
object file. When the object files are linked together, all the
function bodies are read from these ELF sections and instantiated
as if they had been part of the same translation unit. This
enables interprocedural optimizations to work across different
files (and even different languages), potentially improving the
performance of the generated code. To use the link-timer optimizer,
<code>-flto</code> needs to be specified at compile time and during
the final link. If the program does not require any symbols to be
exported, it is possible to combine <code>-flto</code> and the
experimental <code><a
href="http://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html#index-fwhopr-802"
>-fwhopr</a></code> with <code><a
href="http://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html#index-fwhole-program-800"
>-fwhole-program</a></code> to allow the interprocedural optimizers
to use more aggressive assumptions.</li>
<li>The automatic parallelization pass was enhanced to support
parallelization of outer loops.</li>
<li>Automatic parallelization can be enabled as part of Graphite.
In addition to <code>-ftree-parallelize-loops=</code>, specify
<code>-floop-parallelize-all</code> to enable the Graphite-based
optimization.</li>
<li>The infrastructure for optimizing based on
<a href="http://gcc.gnu.org/onlinedocs/gcc/Restricted-Pointers.html">restrict qualified pointers</a>
has been rewritten and should result in code generation improvements.
Optimizations based on restrict qualified pointers are now also available
when using <code>-fno-strict-aliasing</code>.</li>
<li>There is a new optimization pass that attempts to change prototype
of functions to avoid unused parameters, pass only relevant parts of
structures and turn arguments passed by reference to arguments passed
by value when possible. It is enabled by <code>-O2</code> and above
as well as <code>-Os</code> and can be manually invoked using the new
command-line switch <code>-fipa-sra</code>. </li>
<li>GCC now optimize exception handling code. In particular cleanup
regions that are proved to not have any effect are optimized out.</li>
</ul>
<h2>New Languages and Language specific improvements</h2>
<h3>All languages</h3>
<ul>
<li>The <code>-fshow-column</code> option is now on by default. This
means error messages now have a column associated with them.</li>
</ul>
<h3>Ada</h3>
<ul>
<li>Compilation of programs heavily using discriminated record types
with variant parts has been sped up and generates more compact code.</li>
<li>Stack checking now works reasonably well on most plaforms. In
some specific cases, stack overflows may still fail to be detected,
but a compile-time warning will be issued for these cases.</li>
</ul>
<h3>C family</h3>
<ul>
<li>If a header named in a <code>#include</code> directive is not
found, the compiler exits immediately. This avoids a cascade of
errors arising from declarations expected to be found in that
header being missing.</li>
<li>A new built-in function <code>__builtin_unreachable()</code>
has been added that tells the compiler that control will never
reach that point. It may be used after <code>asm</code>
statements that terminate by transferring control elsewhere, and
in other places that are known to be unreachable.</li>
<li>The <code>-Wlogical-op</code> option now warns for logical
expressions such as <code>(c == 1 && c == 2)</code> and <code>(c
!= 1 || c != 2)</code>, which are likely to be mistakes. This
option is disabled by default.</li>
<li>An <code>asm goto</code> feature has been added to
allow <code>asm</code> statements that jump to C labels.</li>
<li>C++0x raw strings are supported for C++ and for C
with <code>-std=gnu99</code>.</li>
<li>The <code>deprecated</code> attribute now takes an optional
string argument, for example,
<code>__attribute__((deprecated("text string")))</code>, that will
be printed together with the deprecation warning.</li>
</ul>
<h3>C</h3>
<ul>
<li>The <code>-Wenum-compare</code> option, which warns when
comparing values of different enum types, now works for C. It
formerly only worked for C++. This warning is enabled
by <code>-Wall</code>. It may be avoided by using a
type cast.</li>
<li>The <code>-Wcast-qual</code> option now warns about casts
which are unsafe in that they permit const-correctness to be
violated without further warnings. Specifically, it warns about
cases where a qualifier is added when all the lower types are
not <code>const</code>. For example, it warns about a cast
from <code>char **</code> to <code>const char **</code>.</li>
<li>The <code>-Wc++-compat</code> option is significantly
improved. It issues new warnings for:
<ul>
<li>Using C++ reserved operator names as identifiers.</li>
<li>Conversions to enum types without explicit casts.</li>
<li>Using va_arg with an enum type.</li>
<li>Using different enum types in the two branches
of <code>?:</code>.</li>
<li>Using <code>++</code> or <code>--</code> on a variable of
enum type.</li>
<li>Using the same name as both a struct, union or enum tag
and a typedef, unless the typedef refers to the tagged type
itself.</li>
<li>Using a struct, union, or enum which is defined within
another struct or union.</li>
<li>A struct field defined using a typedef if there is a field
in the struct, or an enclosing struct, whose name is the
typedef name.</li>
<li>Duplicate definitions at file scope.</li>
<li>Uninitialized const variables.</li>
<li>A global variable with an anonymous struct, union, or enum
type.</li>
<li>Using a string constant to initialize a char array whose
size is the length of the string.</li>
</ul></li>
<li>The new <code>-Wjump-misses-init</code> option warns about
cases where a <code>goto</code> or <code>switch</code> skips the
initialization of a variable. This sort of branch is an error in
C++ but not in C. This warning is enabled
by <code>-Wc++-compat</code>.</li>
<li>GCC now ensures that a
C99-conforming <code><stdint.h></code> is present on most
targets, and uses information about the types in this header to
implement the Fortran bindings to those types. GCC does not
ensure the presence of such a header, and does not implement the
Fortran bindings, on the following targets: NetBSD, VxWorks, VMS,
SymbianOS, WinCE, LynxOS, Netware, QNX, Interix, TPF.</li>
<li>GCC now implements C90- and C99-conforming rules for constant
expressions. This may cause warnings or errors for some code
using expressions that can be folded to a constant but are not
constant expressions as defined by ISO C.</li>
<li>All known target-independent C90 and C90 Amendment 1
conformance bugs, and all known target-independent C99 conformance
bugs not related to floating point or extended identifiers, have
been fixed.</li>
<li>The C decimal floating point support now includes support for
the <code>FLOAT_CONST_DECIMAL64</code> pragma.</li>
<li>The named address space feature from ISO/IEC TR 18037 is now
supported. This is currently only implemented for the SPU
processor.</li>
</ul>
<h3><a name="C++">C++</a></h3>
<ul>
<li>Improved <a href="cxx0x_status.html">experimental support for the
upcoming C++0x</a> ISO C++ standard, including support for raw strings,
lambda expressions and <code>explicit</code> type conversion
operators.</li>
<li>When printing the name of a class template specialization, G++ will
now omit any template arguments which come from default template
arguments. This behavior (and the pretty-printing of function template
specializations as template signature and arguments) can be disabled
with the <code>-fno-pretty-templates</code> option.</li>
<li>Access control is now applied to <code>typedef</code> names used in
a template, which may cause G++ to reject some ill-formed code that was
accepted by earlier releases. The <code>-fno-access-control</code>
option can be used as a temporary workaround until the code is
corrected.</li>
<li>Compilation time for code that uses templates should now scale
linearly with the number of instantiations rather than quadratically,
as template instantiations are now looked up using hash tables.</li>
<li>Declarations of functions that look like builtin declarations of
library functions are only considered to be redeclarations if they
are declared with extern "C". This may cause problems with
code that omits extern "C" on hand-written declarations of
C library functions such as <code>abort</code>
or <code>memcpy</code>. Such code is ill-formed, but was accepted by
earlier releases.</li>
<li>Diagnostics that used to complain about passing non-POD types to
<code>...</code> or jumping past the declaration of a non-POD
variable now check for triviality rather than PODness, as per
C++0x.</li>
<li>In C++0x mode local and anonymous classes are now allowed as
template arguments, and in declarations of variables and functions
with linkage, so long as any such declaration that is used is also
defined (<a
href="http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#757"
>DR 757</a>).</li>
<li>Labels may now have attributes, as has been permitted for a
while in C. This is only permitted when the label definition and
the attribute specifier is followed by a semicolon—i.e., the
label applies to an empty statement. The only useful attribute
for a label is <code>unused</code>.</li>
<li>
G++ now implements
<a href="http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#176">DR
176</a>. Previously G++ did not support using the
injected-class-name of a template base class as a type name, and
lookup of the name found the declaration of the template in the
enclosing scope. Now lookup of the name finds the
injected-class-name, which can be used either as a type or as a
template, depending on whether or not the name is followed by a
template argument list. As a result of this change, some code that
was previously accepted may be ill-formed because
<ol>
<li>The injected-class-name is not accessible because it's from a
private base, or</li>
<li>The injected-class-name cannot be used as an argument for a
template template parameter.</li>
</ol>
In either of these cases, the code can be fixed by adding a
nested-name-specifier to explicitly name the template. The first can
be worked around with <code>-fno-access-control</code>; the second is
only rejected with <code>-pedantic</code>.
</li>
<li>A new standard mangling for SIMD vector types has been added, to
avoid name clashes on systems with vectors of varying length. By
default the compiler still uses the old mangling, but emits aliases
with the new mangling on targets that support strong aliases. Users
can switch over entirely to the new mangling
with <code>-fabi-version=4</code> or <code>-fabi-version=0</code>.
<code>-Wabi</code> will now warn about code that uses the old mangling.</li>
</ul>
<h4>Runtime Library (libstdc++)</h4>
<ul>
<li><a href="http://gcc.gnu.org/onlinedocs/libstdc++/manual/status.html#status.iso.200x">
Improved experimental support for the upcoming ISO C++ standard,
C++0x</a>, including:
<ul>
<li>Support for <future>, <functional>,
and <random>.</li>
<li>Existing facilities now exploit explicit operators and the
newly implemented core C++0x features.</li>
</ul>
</li>
<li>
<p>An experimental <a href="http://gcc.gnu.org/onlinedocs/libstdc++/manual/profile_mode.html">
profile mode </a> has been added. This is an implementation of
many C++ standard library constructs with an additional analysis
layer that gives performance improvement advice based on
recognition of suboptimal usage patterns. For example,
</p>
<pre>
#include <vector>
int main()
{
std::vector<int> v;
for (int k = 0; k < 1024; ++k)
v.insert(v.begin(), k);
}
</pre>
<p>
When instrumented via the profile mode, can return suggestions about
the initial size and choice of the container used as follows:
</p>
<pre>
vector-to-list: improvement = 5: call stack = 0x804842c ...
: advice = change std::vector to std::list
vector-size: improvement = 3: call stack = 0x804842c ...
: advice = change initial container size from 0 to 1024
</pre>
<p>
These constructs can be substituted for the normal libstdc++
constructs on a piecemeal basis, or all existing components can be
transformed via the <code>-D_GLIBCXX_PROFILE</code> macro.
</p>
</li>
<li><a href="http://gcc.gnu.org/onlinedocs/libstdc++/manual/status.html#status.iso.tr24733">Support for decimal floating-point arithmetic</a>
(aka ISO C++ TR 24733) has been added. This support is in header file
<code><decimal/decimal></code>, uses namespace
<code>std::decimal</code>, and includes classes <code>decimal32</code>,
<code>decimal64</code>, and <code>decimal128</code>.</li>
<li>Sources have been audited for application of function attributes
<code>nothrow</code>, <code>const</code>, <code>pure</code>, and
<code>noreturn</code>.</li>
<li>Python pretty-printers have been added for many standard
library components that simplify the internal representation and
present a more intuitive view of components when used with
appropriately-advanced versions of GDB. For more information,
please consult the
more <a href="http://sourceware.org/gdb/wiki/STLSupport">detailed
description</a>.</li>
<li>The default behavior for comparing typeinfo names has changed,
so in <code><typeinfo></code>,
<code>__GXX_MERGED_TYPEINFO_NAMES</code> now defaults to zero.</li>
<li>The new <code>-static-libstdc++</code> option
directs <code>g++</code> to link the C++ library statically, even
if the default would normally be to link it dynamically.</li>
</ul>
<h3><a name="Fortran">Fortran</a></h3>
<ul>
<li>The <code>COMMON</code> default padding has been changed –
instead of adding the padding before a variable it is now added
afterwards, which increases the compatibility with other vendors
and helps to obtain the correct output in some cases. Cf. also the
<code>-falign-commons</code> option (<a href="../gcc-4.4/changes.html">added
in 4.4</a>).</li>
<li>The <code>-finit-real=</code> option now also supports the value
<code>snan</code> for signalling not-a-number; to be effective,
one additionally needs to enable trapping (e.g. via
<code>-ffpe-trap=</code>). Note: Compile-time optimizations can
turn a signalling NaN into a quiet one.</li>
<li>The new option <code>-fcheck=</code> has been added with the
options <code>bounds</code>, <code>array-temps</code>,
<code>do</code>, <code>pointer</code>, and <code>recursive</code>. The
<code>bounds</code> and <code>array-temps</code> options are
equivalent to <code>-fbounds-check</code> and
<code>-fcheck-array-temporaries</code>. The <code>do</code>
option checks for invalid modification of loop iteration variables,
and the <code>recursive</code> option tests for recursive calls
to subroutines/functions which are not marked as recursive. With
<code>pointer</code> pointer association checks in calls are performed;
however, neither undefined pointers nor pointers in expressions are
handled. Using <code>-fcheck=all</code> enables all these run-time
checks.</li>
<li>The run-time checking <code>-fcheck=bounds</code> now warns
about invalid string lengths of character dummy arguments. Additionally,
more compile-time checks have been added.</li>
<li>The new option <a
href="http://gcc.gnu.org/onlinedocs/gfortran/Code-Gen-Options.html"
><code>-fno-protect-parens</code></a> has been added; if set, the
compiler may reorder REAL and COMPLEX expressions without regard
to parentheses.</li>
<li>GNU Fortran no longer links against <code>libgfortranbegin</code>.
As before, <code>MAIN__</code> (assembler symbol name) is the actual
Fortran main program, which is invoked by the <code>main</code> function.
However, <code>main</code> is now generated and put in the same object
file as <code>MAIN__</code>. For the time being,
<code>libgfortranbegin</code> still exists for backward
compatibility. For details see the new <a
href="http://gcc.gnu.org/onlinedocs/gfortran/Mixed-Language-Programming.html">Mixed-Language
Programming</a> chapter in the manual.</li>
<li>The I/O library was restructured for performance and cleaner
code.</li>
<li>Array assignments and <code>WHERE</code> are now run in parallel when
OpenMP's <code>WORKSHARE</code> is used.</li>
<li>The experimental option <code>-fwhole-file</code> was added. The option
allows whole-file checking of procedure arguments and allows for better
optimizations. It can also be used with <code>-fwhole-program</code>,
which is now also supported in gfortran.</li>
<li>More Fortran 2003 and Fortran 2008 mathematical functions can
now be used as initialization expressions.</li>
<li>Some extended attributes such as <code>STDCALL</code> are now
supported via the <a
href="http://gcc.gnu.org/onlinedocs/gfortran/GNU-Fortran-Compiler-Directives.html">
<code>GCC$</code> compiler directive</a>.</li>
<li>For Fortran 77 compatibility: If <code>-fno-sign-zero</code> is
used, the <code>SIGN</code> intrinsic behaves now as if zero were always
positive.</li>
<li>For legacy compatibiliy: On Cygwin and MinGW, the special files
<code>CONOUT$</code> and <code>CONIN$</code> (and <code>CONERR$</code>
which maps to <code>CONOUT$</code>) are now supported.</li>
<li>Fortran 2003 support has been extended:
<ul>
<li>Procedure-pointer function results and procedure-pointer
components (including PASS),</li>
<li>allocatable scalars (experimental),</li>
<li><code>DEFERRED</code> type-bound procedures,</li>
<li>the <code>ERRMSG=</code> argument of the <code>ALLOCATE</code>
and <code>DEALLOCATE</code> statements have been implemented.</li>
<li>The <code>ALLOCATE</code> statement supports type-specs and
the <code>SOURCE=</code> argument.</li>
<li><code>OPERATOR(*)</code> and <code>ASSIGNMENT(=)</code> are now
allowed as <code>GENERIC</code> type-bound procedure (i.e. as
type-bound operators).</li>
<li>Rounding (<code>ROUND=</code>, <code>RZ</code>, ...) for output
is now supported.</li>
<li>The <code>INT_FAST{8,16,32,64,128}_T</code> kind type parameters
of the intrinsic module <code>ISO_C_BINDING</code> are now
supported, except for the targets listed above as ones where
GCC does not have <code><stdint.h></code> type information.
</li>
<li>Extensible derived types with type-bound procedure or procedure
pointer with <code>PASS</code> attribute now have to use
<code>CLASS</code> in line with the Fortran 2003 standard; the
workaround to use <code>TYPE</code> is no longer supported.</li>
<li><a href="http://gcc.gnu.org/wiki/OOP">Experimental, incomplete
support for polymorphism</a>, including <code>CLASS</code>,
<code>SELECT TYPE</code> and dynamic dispatch of type-bound
procedure calls. Some features do not work yet such as
unlimited polymorphism (<code>CLASS(*)</code>).</li>
</ul>
</li>
<li>Fortran 2008 support has been extended:
<ul>
<li>The <code>OPEN</code> statement now supports the
<code>NEWUNIT=</code> option, which returns a unique file unit,
thus preventing inadvertent use of the same unit in different parts
of the program.</li>
<li>Support for unlimited format items has been added.</li>
<li>The <code>INT{8,16,32}</code> and <code>REAL{32,64,128}</code>
kind type parameters of the intrinsic module
<code>ISO_FORTRAN_ENV</code> are now supported.</li>
<li>Using complex arguments with <code>TAN</code>, <code>SINH</code>,
<code>COSH</code>, <code>TANH</code>, <code>ASIN</code>,
<code>ACOS</code>, and <code>ATAN</code> is now possible; the
functions <code>ASINH</code>, <code>ACOSH</code>, and
<code>ATANH</code> have been added (for real and complex arguments)
and <code>ATAN(Y,X)</code> is now an alias for <code>ATAN2(Y,X).</code>
</li>
<li>The <code>BLOCK</code> construct has been implemented.</li>
</ul>
</li>
</ul>
<h3>Java (GCJ)</h3>
<h2 id="targets">New Targets and Target Specific Improvements</h2>
<h3>AIX</h3>
<ul>
<li>Full cross-toolchain support now available with GNU Binutils</li>
</ul>
<h3>ARM</h3>
<ul>
<li>GCC now supports the Cortex-M0 and Cortex-A5 processors.</li>
<li>GCC now supports the ARM v7E-M architecture.</li>
<li>GCC now supports VFPv4-based FPUs and FPUs with
single-precision-only VFP.</li>
<li>GCC has many improvements to optimization for other ARM
processors, including scheduling support for the integer pipeline
on Cortex-A9.</li>
<li>GCC now supports the IEEE 754-2008 half-precision
floating-point type, and a variant ARM-specific half-precision
type. This type is specified using <code>__fp16</code>, with the
layout determined by <code>-mfp16-format</code>. With
appropriate <code>-mfpu</code> options, the Cortex-A9 and VFPv4
half-precision instructions will be used.</li>
<li>GCC now supports the variant of AAPCS that uses VFP registers
for parameter passing and return values.</li>
</ul>
<h3>AVR</h3>
<ul>
<li>The <code>-mno-tablejump</code> option has been removed because it
has the same effect as the <code>-fno-jump-tables</code> option.</li>
<li>Added support for these new AVR devices:
<ul>
<li>ATmega8U2</li>
<li>ATmega16U2</li>
<li>ATmega32U2</li>
</ul>
</li>
</ul>
<h3 id="x86">IA-32/x86-64</h3>
<ul>
<li>GCC now will set the default for <code>-march=</code> based on
the configure target.</li>
<li>GCC now supports handling floating-point excess precision
arising from use of the x87 floating-point unit in a way that
conforms to ISO C99. This is enabled
with <code>-fexcess-precision=standard</code> and with standards
conformance options such as <code>-std=c99</code>, and may be
disabled using <code>-fexcess-precision=fast</code>.</li>
<li>Support for the Intel Atom processor is now available through the
<code>-march=atom</code> and <code>-mtune=atom</code> options.</li>
<li>A new <code>-mcrc32</code> option is now available to enable
<code>crc32</code> intrinsics.</li>
<li>A new <code>-mmovbe</code> option is now available to enable GCC
to use the <code>movbe</code> instruction to implement
<code>__builtin_bswap32</code> and <code>__builtin_bswap64</code>.
</li>
<li>SSE math now can be enabled by default at configure time with the
new <code>--with-fpmath=sse</code> option.</li>
<li>There is a new intrinsic header file, <x86intrin.h>. It
should be included before using any IA-32/x86-64 intrinsics.</li>
<li>Support for the XOP, FMA4, and LWP instruction sets for the AMD
Orochi processors are now available with
the <code>-mxop</code>, <code>-mfma4</code>,
and <code>-mlwp</code> options.</li>
<li> The <code>-mabm</code> option enables GCC to use
the <code>popcnt</code> and <code>lzcnt</code> instructions on AMD
processors.</li>
<li> The <code>-mpopcnt</code> option enables GCC to use
the <code>popcnt</code> instructions on both AMD and Intel
processors.</li>
</ul>
<h3>M68K/ColdFire</h3>
<ul>
<li>GCC now supports ColdFire 51xx, 5221x, 5225x, 52274, 52277,
5301x and 5441x devices.</li>
<li>GCC now supports thread-local storage (TLS) on M68K and
ColdFire processors.</li>
</ul>
<h3 id="mep">MeP</h3>
<p>Support has been added for the Toshiba Media embedded Processor (MeP, or
mep-elf) embedded target.</p>
<h3>MIPS</h3>
<ul>
<li>GCC now supports MIPS 1004K processors.</li>
<li>GCC can now be configured with
options <code>--with-arch-32</code>, <code>--with-arch-64</code>,
<code>--with-tune-32</code> and <code>--with-tune-64</code> to
control the default optimization separately for 32-bit and 64-bit
modes.</li>
<li>MIPS targets now support an alternative <code>_mcount</code> interface,
in which register <code>$12</code> points to the function's save slot
for register <code>$31</code>. This interface is selected by the
<code>-mcount-ra-address</code> option; see the documentation for
more details.</li>
<li>GNU/Linux targets can now generate read-only <code>.eh_frame</code>
sections. This optimization requires GNU binutils 2.20 or above, and
is only available if GCC is configured with a suitable version of
binutils.</li>
<li>GNU/Linux targets can now attach special relocations to indirect
calls, so that the linker can turn them into direct jumps or branches.
This optimization requires GNU binutils 2.20 or later, and is
automatically selected if GCC is configured with an appropriate
version of binutils. It can be explicitly enabled or disabled
using the <code>-mrelax-pic-calls</code> command-line option.</li>
<li>GCC now generates more heavily-optimized atomic operations on
Octeon processors.</li>
<li>MIPS targets now support the <code>-fstack-protector</code> option.</li>
<li>GCC now supports an <code>-msynci</code> option, which specifies
that <code>synci</code> is enough to flush the instruction cache,
without help from the operating system. GCC uses this information
to optimize automatically-generated cache flush operations, such as
those used for nested functions in C. There is also a
<code>--with-synci</code> configure-time option, which makes
<code>-msynci</code> the default.</li>
<li>GCC supports four new function attributes for interrupt
handlers: <code>interrupt</code>, <code>use_shadow_register_set</code>,
<code>keep_interrupts_masked</code> and
<code>use_debug_exception_return</code>. See the documentation
for more details about these attributes.</li>
</ul>
<h3 id="picochip">picochip</h3>
<h3 id="rs6000">RS/6000 (POWER/PowerPC)</h3>
<ul>
<li>GCC now supports the Power ISA 2.06, which includes the VSX
instructions that add vector 64-bit floating point support, new population
count instructions, and conversions between floating point and unsigned types.</li>
<li>Support for the power7 processor is now available through the
<code>-mcpu=power7</code> and <code>-mtune=power7</code>.</li>
<li>GCC will now vectorize loops that contain simple math functions like copysign
when generating code for altivec or VSX targets.</li>
<li>Support for the A2 processor is now available through the
<code>-mcpu=a2</code> and <code>-mtune=a2</code> options.</li>
<li>Support for the 476 processor is now available through the
<code>-mcpu={476,476fp}</code> and <code>-mtune={476,476fp}</code>
options.</li>
<li>Support for the e500mc64 processor is now available through
the <code>-mcpu=e500mc64</code>
and <code>-mtune=e500mc64</code> options.</li>
<li>GCC can now be configured with
options <code>--with-cpu-32</code>, <code>--with-cpu-64</code>,
<code>--with-tune-32</code> and <code>--with-tune-64</code> to
control the default optimization separately for 32-bit and 64-bit
modes.</li>
</ul>
<h3 id="rx">RX</h3>
<p>Support has been added for the Renesas RX Processor (rx-elf) target.</p>
<h2 id="os">Operating Systems</h2>
<h3 id="windows">Windows (Cygwin and MinGW)</h3>
<ul>
<li>GCC now installs all the major language runtime libraries as DLLs
when configured with the <code>--enable-shared</code> option.</li>
<li>GCC now makes use of the new support for aligned common variables in
versions of binutils >= 2.20 to fix bugs in the support for SSE data
types.</li>
<li>Improvements to the libffi support library increase the reliability
of code generated by GCJ on all Windows platforms. Libgcj is enabled
by default for the first time.</li>
<li>Libtool improvements simplify installation by placing the generated
DLLs in the correct binaries directory.</li>
<li>Numerous other minor bugfixes and improvements, and substantial
enhancements to the Fortran language support library.</li>
</ul>
<h2>Documentation improvements</h2>
<h2>Other significant improvements</h2>
<h3>Plugins</h3>
<ul>
<li>It is now possible to extend the compiler without having to
modify its source code. A new option <code>-fplugin=file.so</code>
tells GCC to load the shared object <code>file.so</code> and execute
it as part of the compiler. The internal documentation describes
the details on how plugins can interact with the compiler.</li>
</ul>
<h3>Installation changes</h3>
<ul>
<li>
The move to newer autotools changed default installation directories
and switches to control them:
The <code>--with-datarootdir</code>, <code>--with-docdir</code>,
<code>--with-pdfdir</code>, and <code>--with-htmldir</code> switches are
not used any more. Instead, you can now use <code>--datarootdir</code>,
<code>--docdir</code>, <code>--htmldir</code>, and <code>--pdfdir</code>.
The default installation directories have changed as follows according to
the GNU Coding Standards:
<table>
<tbody>
<tr>
<td>datarootdir</td>
<td>read-only architecture-independent data root [PREFIX/share]</td>
</tr>
<tr>
<td>localedir</td>
<td>locale-specific message catalogs [DATAROOTDIR/locale]</td>
</tr>
<tr>
<td>docdir</td>
<td>documentation root [DATAROOTDIR/doc/PACKAGE]</td>
</tr>
<tr>
<td>htmldir</td>
<td>html documentation [DOCDIR]</td>
</tr>
<tr>
<td>dvidir</td>
<td>dvi documentation [DOCDIR]</td>
</tr>
<tr>
<td>pdfdir</td>
<td>pdf documentation [DOCDIR]</td>
</tr>
<tr>
<td>psdir</td>
<td>ps documentation [DOCDIR]</td>
</tr>
</tbody>
</table>
The following variables have new default values:
<table>
<tbody>
<tr>
<td>datadir</td>
<td>read-only architecture-independent data [DATAROOTDIR]</td>
</tr>
<tr>
<td>infodir</td>
<td>info documentation [DATAROOTDIR/info]</td>
</tr>
<tr>
<td>mandir</td>
<td>man documentation [DATAROOTDIR/man]</td>
</tr>
</tbody>
</table>
</li>
</ul>
<!-- ==================================================================== -->
<div class="copyright">
<p>Please send FSF & GNU inquiries & questions to
<a href="mailto:gnu@gnu.org">gnu@gnu.org</a>.
There are also <a href="http://www.gnu.org/home.html#ContactInfo">other ways
to contact</a> the FSF.</p>
<p>These pages are maintained by
<a href="http://gcc.gnu.org/about.html">the GCC team</a>.</p>
<address>For questions related to the use of GCC, please consult these web
pages and the <a href="http://gcc.gnu.org/onlinedocs/">GCC manuals</a>. If
that fails, the <a href="mailto:gcc-help@gcc.gnu.org">gcc-help@gcc.gnu.org</a>
mailing list might help.<br />
Please send comments on these web pages and the development of GCC to our
developer mailing list at <a href="mailto:gcc@gnu.org">gcc@gnu.org</a>
or <a href="mailto:gcc@gcc.gnu.org">gcc@gcc.gnu.org</a>. All of our lists
have <a href="http://gcc.gnu.org/lists.html">public archives</a>.
</address>
<p>Copyright (C) Free Software Foundation, Inc.,
51 Franklin St, Fifth Floor, Boston, MA 02110, USA.</p>
<p>Verbatim copying and distribution of this entire article is
permitted in any medium, provided this notice is preserved.</p>
<table width="100%" border="0"><tr><td>
<!-- IGNORE DIFF -->Last modified 2010-04-17
</td><td align="right" valign="bottom">
<a href="http://validator.w3.org/check/referer">
<img src="http://www.w3.org/Icons/valid-xhtml10"
alt="Valid XHTML 1.0" border="0" width="88" height="31" />
</a>
</td></tr></table>
</div>
<!-- ==================================================================== -->
</body>
</html>
|