docs/ReleaseNotes.html

   1 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
   2                       "http://www.w3.org/TR/html4/strict.dtd">
   3 <html>
   4 <head>
   5   <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
   6   <link rel="stylesheet" href="llvm.css" type="text/css">
   7   <title>LLVM 2.0 Release Notes</title>
   8 </head>
   9 <body>
  10
  11 <div class="doc_title">LLVM 2.0 Release Notes</div>
  12
  13 <ol>
  14   <li><a href="#intro">Introduction</a></li>
  15   <li><a href="#whatsnew">What's New?</a></li>
  16   <li><a href="GettingStarted.html">Installation Instructions</a></li>
  17   <li><a href="#portability">Portability and Supported Platforms</a></li>
  18   <li><a href="#knownproblems">Known Problems</a>
  19   <li><a href="#additionalinfo">Additional Information</a></li>
  20 </ol>
  21
  22 <div class="doc_author">
  23   <p>Written by the <a href="http://llvm.org">LLVM Team</a><p>
  24 </div>
  25
  26 <!-- *********************************************************************** -->
  27 <div class="doc_section">
  28   <a name="intro">Introduction</a>
  29 </div>
  30 <!-- *********************************************************************** -->
  31
  32 <div class="doc_text">
  33
  34 <p>This document contains the release notes for the LLVM compiler
  35 infrastructure, release 2.0.  Here we describe the status of LLVM, including
  36 major improvements from the previous release and any known problems.  All LLVM
  37 releases may be downloaded from the <a href="http://llvm.org/releases/">LLVM
  38 releases web site</a>.</p>
  39
  40 <p>For more information about LLVM, including information about the latest
  41 release, please check out the <a href="http://llvm.org/">main LLVM
  42 web site</a>.  If you have questions or comments, the <a
  43 href="http://mail.cs.uiuc.edu/mailman/listinfo/llvmdev">LLVM developer's mailing
  44 list</a> is a good place to send them.</p>
  45
  46 <p>Note that if you are reading this file from CVS or the main LLVM web page,
  47 this document applies to the <i>next</i> release, not the current one.  To see
  48 the release notes for the current or previous releases, see the <a
  49 href="http://llvm.org/releases/">releases page</a>.</p>
  50
  51 </div>
  52
  53 <!-- *********************************************************************** -->
  54 <div class="doc_section">
  55   <a name="whatsnew">What's New?</a>
  56 </div>
  57 <!-- *********************************************************************** -->
  58
  59 <div class="doc_text">
  60
  61 <p>This is the eleventh public release of the LLVM Compiler Infrastructure.
  62 Being the first major release since 1.0, this release is different in several
  63 ways from our previous releases:</p>
  64
  65 <ol>
  66 <li>We took this as an opportunity to
  67 break backwards compatibility with the LLVM 1.x bytecode and .ll file format.
  68 If you have LLVM 1.9 .ll files that you would like to upgrade to LLVM 2.x, we
  69 recommend the use of the stand alone <a href="#llvm-upgrade">llvm-upgrade</a>
  70 tool (which is included with 2.0).  We intend to keep compatibility with .ll
  71 and .bc formats within the 2.x release series, like we did within the 1.x
  72 series.</li>
  73 <li>There are several significant change to the LLVM IR and internal APIs, such
  74     as a major overhaul of the type system, the completely new bitcode file
  75     format, etc (described below).</li>
  76 <li>We designed the release around a 6 month release cycle instead of the usual
  77     3-month cycle.  This gave us extra time to develop and test some of the
  78     more invasive features in this release.</li>
  79 <li>LLVM 2.0 no longer supports the llvm-gcc3 front-end.  Users are required to
  80     upgrade to llvm-gcc4.  llvm-gcc4 includes many features over
  81     llvm-gcc3, is faster, and is <a href="CFEBuildInstrs.html">much easier to
  82     build from source</a>.</li>
  83 </ol>
  84
  85 <p>Note that while this is a major version bump, this release has been
  86    extensively tested on a wide range of software.  It is easy to say that this
  87    is our best release yet, in terms of both features and correctness.  This is
  88    the first LLVM release to correctly compile and optimize major software like
  89    LLVM itself, Mozilla/Seamonkey, Qt 4.3rc1, kOffice, etc out of the box on
  90    linux/x86.
  91    </p>
  92
  93 </div>
  94
  95 <!--=========================================================================-->
  96 <div class="doc_subsection">
  97 <a name="newfeatures">New Features in LLVM 2.0</a>
  98 </div>
  99
 100 <!--_________________________________________________________________________-->
 101 <div class="doc_subsubsection"><a name="majorchanges">Major Changes</a></div>
 102 <div class="doc_text">
 103
 104 <p>Changes to the LLVM IR itself:</p>
 105
 106 <ul>
 107
 108 <li>Integer types are now completely signless. This means that we
 109     have types like i8/i16/i32 instead of ubyte/sbyte/short/ushort/int
 110     etc. LLVM operations that depend on sign have been split up into
 111     separate instructions (<a href="http://llvm.org/PR950">PR950</a>).  This
 112     eliminates cast instructions that just change the sign of the operands (e.g.
 113     int -> uint), which reduces the size of the IR and makes optimizers
 114     simpler to write.</li>
 115
 116 <li>Integer types with arbitrary bitwidths (e.g. i13, i36, i42, i1057, etc) are
 117     now supported in the LLVM IR and optimizations (<a
 118     href="http://llvm.org/PR1043">PR1043</a>).  However, neither llvm-gcc
 119     (<a href="http://llvm.org/PR1284">PR1284</a>) nor the native code generators
 120     (<a href="http://llvm.org/PR1270">PR1270</a>) support non-standard width
 121     integers yet.</li>
 122
 123 <li>'Type planes' have been removed (<a href="http://llvm.org/PR411">PR411</a>).
 124     It is no longer possible to have two values with the same name in the
 125     same symbol table.  This simplifies LLVM internals, allowing significant
 126     speedups.</li>
 127
 128 <li>Global variables and functions in .ll files are now prefixed with
 129     @ instead of % (<a href="http://llvm.org/PR645">PR645</a>).</li>
 130
 131 <li>The LLVM 1.x "bytecode" format has been replaced with a
 132     completely new binary representation, named 'bitcode'. The <a
 133     href="BitCodeFormat.html">Bitcode Format</a> brings a
 134     number of advantages to the LLVM over the old bytecode format: it is denser
 135     (files are smaller), more extensible, requires less memory to read,
 136     is easier to keep backwards compatible (so LLVM 2.5 will read 2.0 .bc
 137     files), and has many other nice features.</li>
 138
 139 <li>Load and store instructions now track the alignment of their pointer
 140     (<a href="http://www.llvm.org/PR400">PR400</a>).  This allows the IR to
 141     express loads that are not sufficiently aligned (e.g. due to '<tt>#pragma
 142     packed</tt>') or to capture extra alignment information.</li>
 143 </ul>
 144
 145 <p>Major new features:</p>
 146
 147 <ul>
 148
 149 <li>A number of ELF features are now supported by LLVM, including 'visibility',
 150     extern weak linkage, Thread Local Storage (TLS) with the <tt>__thread</tt>
 151     keyword, and symbol aliases.
 152     Among other things, this means that many of the special options needed to
 153     configure llvm-gcc on linux are no longer needed, and special hacks to build
 154     large C++ libraries like Qt are not needed.</li>
 155
 156 <li>LLVM now has a new MSIL backend. <tt>llc -march=msil</tt> will now turn LLVM
 157     into MSIL (".net") bytecode.  This is still fairly early development
 158     with a number of limitations.</li>
 159
 160 <li>A new <a href="CommandGuide/html/llvm-upgrade.html">llvm-upgrade</a> tool
 161     exists to migrates LLVM 1.9 .ll files to LLVM 2.0 syntax.</li>
 162 </ul>
 163
 164 </div>
 165
 166
 167 <!--_________________________________________________________________________-->
 168 <div class="doc_subsubsection"><a name="llvmgccfeatures">llvm-gcc
 169 Improvements</a></div>
 170 <div class="doc_text">
 171 <p>New features include:
 172 </p>
 173
 174 <ul>
 175 <li>Precompiled Headers (PCH) are now supported.</li>
 176
 177 <li>"<tt>#pragma packed</tt>" is now supported, as are the various features
 178     described above (visibility, extern weak linkage, __thread, aliases,
 179     etc).</li>
 180
 181 <li>Tracking function parameter/result attributes is now possible.</li>
 182
 183 <li>Many internal enhancements have been added, such as improvements to
 184     NON_LVALUE_EXPR, arrays with non-zero base, structs with variable sized
 185     fields, VIEW_CONVERT_EXPR, CEIL_DIV_EXPR, nested functions, and many other
 186     things.  This is primarily to supports non-C GCC front-ends, like Ada.</li>
 187
 188 <li>It is simpler to configure llvm-gcc for linux.</li>
 189
 190 </ul>
 191
 192 </div>
 193
 194 <!--_________________________________________________________________________-->
 195 <div class="doc_subsubsection"><a name="optimizer">Optimizer
 196 Improvements</a></div>
 197
 198 <div class="doc_text">
 199 <p>New features include:
 200 </p>
 201
 202 <ul>
 203 <li>The <a href="WritingAnLLVMPass.html">pass manager</a> has been entirely
 204     rewritten, making it significantly smaller, simpler, and more extensible.
 205     Support has been added to run <tt>FunctionPass</tt>es interlaced with
 206     <tt>CallGraphSCCPass</tt>es, we now support loop transformations
 207     explicitly with <tt>LoopPass</tt>, and <tt>ModulePass</tt>es may now use the
 208     result of <tt>FunctionPass</tt>es.</li>
 209
 210 <li>LLVM 2.0 includes a new loop rotation pass, which converts "for loops" into
 211     "do/while loops", where the condition is at the bottom of the loop.</li>
 212
 213 <li>The Loop Strength Reduction pass has been improved, and we now support
 214     sinking expressions across blocks to reduce register pressure.</li>
 215
 216 <li>The <tt>-scalarrepl</tt> pass can now promote unions containing FP values
 217     into a register, it can also handle unions of vectors of the same
 218     size.</li>
 219
 220 <li>The [Post]DominatorSet classes have been removed from LLVM and clients
 221     switched to use the more-efficient ETForest class instead.</li>
 222
 223 <li>The ImmediateDominator class has also been removed, and clients have been
 224     switched to use DominatorTree instead.</li>
 225
 226 <li>The predicate simplifier pass has been improved, making it able to do
 227     simple value range propagation and eliminate more conditionals.  However,
 228     note that predsimplify is not enabled by default in llvm-gcc.</li>
 229
 230 </ul>
 231
 232 </div>
 233
 234 <!--_________________________________________________________________________-->
 235 <div class="doc_subsubsection"><a name="codegen">Code
 236 Generator Enhancements</a></div>
 237
 238 <div class="doc_text">
 239 <p>
 240 New features include:
 241 </p>
 242
 243 <ul>
 244
 245 <li>LLVM now supports software floating point, which allows LLVM to target
 246     chips that don't have hardware FPUs (e.g. ARM thumb mode).</li>
 247
 248 <li>A new register scavenger has been implemented, which is useful for
 249     finding free registers after register allocation.  This is useful when
 250     rewriting frame references on RISC targets, for example.</li>
 251
 252 <li>Heuristics have been added to avoid coalescing vregs with very large live
 253     ranges to physregs.  This was bad because it effectively pinned the physical
 254     register for the entire lifetime of the virtual register (<a
 255     href="http://llvm.org/PR711">PR711</a>).</li>
 256
 257 <li>Support now exists for very simple (but still very useful)
 258     rematerialization the register allocator, enough to move
 259     instructions like "load immediate" and constant pool loads.</li>
 260
 261 <li>Switch statement lowering is significantly better, improving codegen for
 262     sparse switches that have dense subregions, and implemented support
 263     for the shift/and trick.</li>
 264
 265 <li>LLVM now supports tracking physreg sub-registers and super-registers
 266     in the code generator, and includes extensive register
 267     allocator changes to track them.</li>
 268
 269 <li>There is initial support for virtreg sub-registers
 270     (<a href="http://llvm.org/PR1350">PR1350</a>).</li>
 271
 272 </ul>
 273
 274 <p>
 275 Other improvements include:
 276 </p>
 277
 278 <ul>
 279
 280 <li>Inline assembly support is much more solid that before.
 281     The two primary features still missing are support for 80-bit floating point
 282     stack registers on X86 (<a href="http://llvm.org/PR879">PR879</a>), and
 283     support for inline asm in the C backend (<a
 284     href="http://llvm.org/PR802">PR802</a>).</li>
 285
 286 <li>DWARF debug information generation has been improved.  LLVM now passes
 287     most of the GDB testsuite on MacOS and debug info is more dense.</li>
 288
 289 <li>Codegen support for Zero-cost DWARF exception handling has been added (<a
 290     href="http://llvm.org/PR592">PR592</a>).  It is mostly
 291     complete and just in need of continued bug fixes and optimizations at
 292     this point.  However, support in llvm-g++ is disabled with an
 293     #ifdef for the 2.0 release  (<a
 294     href="http://llvm.org/PR870">PR870</a>).</li>
 295
 296 <li>The code generator now has more accurate and general hooks for
 297     describing addressing modes ("isLegalAddressingMode") to
 298     optimizations like loop strength reduction and code sinking.</li>
 299
 300 <li>Progress has been made on a direct Mach-o .o file writer. Many small
 301     apps work, but it is still not quite complete.</li>
 302
 303 </ul>
 304
 305 <p>In addition, the LLVM target description format has itself been extended in
 306  several ways:</p>
 307
 308 <ul>
 309 <li>TargetData now supports better target parameterization in
 310     the .ll/.bc files, eliminating the 'pointersize/endianness' attributes
 311     in the files (<a href="http://llvm.org/PR761">PR761</a>).</li>
 312
 313 <li>TargetData was generalized for finer grained alignment handling,
 314     handling of vector alignment, and handling of preferred alignment</li>
 315
 316 <li>LLVM now supports describing target calling conventions
 317     explicitly in .td files, reducing the amount of C++ code that needs
 318     to be written for a port.</li>
 319
 320 </ul>
 321
 322 </div>
 323
 324 <!--_________________________________________________________________________-->
 325 <div class="doc_subsubsection"><a name="specifictargets">Target-Specific
 326 Improvements</a></div>
 327
 328 <div class="doc_text">
 329
 330 <p>X86-specific Code Generator Enhancements:
 331 </p>
 332
 333 <ul>
 334 <li>The MMX instruction set is now supported through intrinsics.</li>
 335 <li>The scheduler was improved to better reduce register pressure on
 336     X86 and other targets that are register pressure sensitive.</li>
 337 <li>Linux/x86-64 support is much better.</li>
 338 <li>PIC support for linux/x86 has been added.</li>
 339 <li>The X86 backend now supports the GCC regparm attribute.</li>
 340 <li>LLVM now supports inline asm with multiple constraint letters per operand
 341     (like "mri") which is common in X86 inline asms.</li>
 342 </ul>
 343
 344 <p>ARM-specific Code Generator Enhancements:</p>
 345
 346 <ul>
 347 <li>The ARM code generator is now stable and fully supported.</li>
 348
 349 <li>There are major new features, including support for ARM
 350     v4-v6 chips, vfp support, soft float point support, pre/postinc support,
 351     load/store multiple generation, constant pool entry motion (to support
 352     large functions), inline asm support, weak linkage support, static
 353     ctor/dtor support and many bug fixes.</li>
 354
 355 <li>Added support for Thumb code generation (<tt>llc -march=thumb</tt>).</li>
 356
 357 <li>The ARM backend now supports the ARM AAPCS/EABI ABI and PIC codegen on
 358     arm/linux.</li>
 359
 360 <li>Several bugs were fixed for DWARF debug info generation on arm/linux.</li>
 361
 362 </ul>
 363
 364 <p>PowerPC-specific Code Generator Enhancements:</p>
 365
 366 <ul>
 367 <li>The PowerPC 64 JIT now supports addressing code loaded above the 2G
 368      boundary.</li>
 369
 370 <li>Improved support for the Linux/ppc ABI and the linux/ppc JIT is fully
 371     functional now.  llvm-gcc and static compilation are not fully supported
 372     yet though.</li>
 373
 374 <li>Many PowerPC 64 bug fixes.</li>
 375
 376 </ul>
 377
 378 </div>
 379
 380
 381 <!--_________________________________________________________________________-->
 382 <div class="doc_subsubsection"><a name="other">Other Improvements</a></div>
 383 <div class="doc_text">
 384
 385 <p>More specific changes include:</p>
 386
 387 <ul>
 388 <li>LLVM no longer relies on static destructors to shut itself down.  Instead,
 389     it lazily initializes itself and shuts down when <tt>llvm_shutdown()</tt> is
 390     explicitly called.</li>
 391
 392 <li>LLVM now has significantly fewer static constructors, reducing startup time.
 393     </li>
 394
 395 <li>Several classes have been refactored to reduce the amount of code that
 396     gets linked into apps that use the JIT.</li>
 397
 398 <li>Construction of intrinsic function declarations has been simplified.</li>
 399
 400 <li>The gccas/gccld tools have been replaced with small shell scripts.</li>
 401
 402 <li>Support has been added to llvm-test for running on low-memory
 403     or slow machines (make SMALL_PROBLEM_SIZE=1).</li>
 404
 405 </ul>
 406 </div>
 407
 408 <!--_________________________________________________________________________-->
 409 <div class="doc_subsubsection"><a name="apichanges">API Changes</a></div>
 410 <div class="doc_text">
 411
 412 <p>LLVM 2.0 contains a revamp of the type system and several other significant
 413 internal changes.  If you are programming to the C++ API, be aware of the
 414 following major changes:</p>
 415
 416 <ul>
 417 <li>Pass registration is slightly different in LLVM 2.0 (you now need an
 418    <tt>intptr_t</tt> in your constructor), as explained in the <a
 419    href="WritingAnLLVMPass.html#basiccode">Writing an LLVM Pass</a>
 420    document.</li>
 421
 422 <li><tt>ConstantBool</tt>, <tt>ConstantIntegral</tt> and <tt>ConstantInt</tt>
 423     classes have been merged together, we now just have
 424     <tt>ConstantInt</tt>.</li>
 425
 426 <li><tt>Type::IntTy</tt>, <tt>Type::UIntTy</tt>, <tt>Type::SByteTy</tt>, ... are
 427     replaced by <tt>Type::Int8Ty</tt>, <tt>Type::Int16Ty</tt>, etc.  LLVM types
 428     have always corresponded to fixed size types
 429     (e.g. long was always 64-bits), but the type system no longer includes
 430     information about the sign of the type.  Also, the
 431     <tt>Type::isPrimitiveType()</tt> method now returns false for integers.</li>
 432
 433 <li>Several classes (<tt>CallInst</tt>, <tt>GetElementPtrInst</tt>,
 434     <tt>ConstantArray</tt>, etc), that once took <tt>std::vector</tt> as
 435      arguments now take ranges instead.   For example, you can create a
 436     <tt>GetElementPtrInst</tt> with code like:
 437
 438     <pre>
 439       Value *Ops[] = { Op1, Op2, Op3 };
 440       GEP = new GetElementPtrInst(BasePtr, Ops, 3);
 441     </pre>
 442
 443     This avoids creation of a temporary vector (and a call to malloc/free).  If
 444     you have an <tt>std::vector</tt>, use code like this:
 445     <pre>
 446       std::vector&lt;Value*&gt; Ops = ...;
 447       GEP = new GetElementPtrInst(BasePtr, &amp;Ops[0], Ops.size());
 448     </pre>
 449
 450     </li>
 451
 452 <li><tt>CastInst</tt> is now abstract and its functionality is split into
 453     several parts, one for each of the <a href="LangRef.html#convertops">new
 454     cast instructions</a>.</li>
 455
 456 <li><tt>Instruction::getNext()/getPrev()</tt> are now private (along with
 457     <tt>BasicBlock::getNext</tt>, etc), for efficiency reasons (they are now no
 458     longer just simple pointers).  Please use <tt>BasicBlock::iterator</tt>, etc
 459     instead.
 460 </li>
 461
 462 <li><tt>Module::getNamedFunction()</tt> is now called
 463     <tt>Module::getFunction()</tt>.</li>
 464
 465 <li><tt>SymbolTable.h</tt> has been split into <tt>ValueSymbolTable.h</tt> and
 466 <tt>TypeSymbolTable.h</tt>.</li>
 467 </ul>
 468 </div>
 469
 470
 471 <!-- *********************************************************************** -->
 472 <div class="doc_section">
 473   <a name="portability">Portability and Supported Platforms</a>
 474 </div>
 475 <!-- *********************************************************************** -->
 476
 477 <div class="doc_text">
 478
 479 <p>LLVM is known to work on the following platforms:</p>
 480
 481 <ul>
 482 <li>Intel and AMD machines running Red Hat Linux, Fedora Core and FreeBSD
 483       (and probably other unix-like systems).</li>
 484 <li>PowerPC and X86-based Mac OS X systems, running 10.2 and above in 32-bit and
 485     64-bit modes.</li>
 486 <li>Intel and AMD machines running on Win32 using MinGW libraries (native)</li>
 487 <li>Intel and AMD machines running on Win32 with the Cygwin libraries (limited
 488     support is available for native builds with Visual C++).</li>
 489 <li>Sun UltraSPARC workstations running Solaris 8.</li>
 490 <li>Alpha-based machines running Debian GNU/Linux.</li>
 491 <li>Itanium-based machines running Linux and HP-UX.</li>
 492 </ul>
 493
 494 <p>The core LLVM infrastructure uses
 495 <a href="http://www.gnu.org/software/autoconf/">GNU autoconf</a> to adapt itself
 496 to the machine and operating system on which it is built.  However, minor
 497 porting may be required to get LLVM to work on new platforms.  We welcome your
 498 portability patches and reports of successful builds or error messages.</p>
 499
 500 </div>
 501
 502 <!-- *********************************************************************** -->
 503 <div class="doc_section">
 504   <a name="knownproblems">Known Problems</a>
 505 </div>
 506 <!-- *********************************************************************** -->
 507
 508 <div class="doc_text">
 509
 510 <p>This section contains all known problems with the LLVM system, listed by
 511 component.  As new problems are discovered, they will be added to these
 512 sections.  If you run into a problem, please check the <a
 513 href="http://llvm.org/bugs/">LLVM bug database</a> and submit a bug if
 514 there isn't already one.</p>
 515
 516 </div>
 517
 518 <!-- ======================================================================= -->
 519 <div class="doc_subsection">
 520   <a name="experimental">Experimental features included with this release</a>
 521 </div>
 522
 523 <div class="doc_text">
 524
 525 <p>The following components of this LLVM release are either untested, known to
 526 be broken or unreliable, or are in early development.  These components should
 527 not be relied on, and bugs should not be filed against them, but they may be
 528 useful to some people.  In particular, if you would like to work on one of these
 529 components, please contact us on the <a href="http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev">LLVMdev list</a>.</p>
 530
 531 <ul>
 532 <li>The <tt>-cee</tt> pass is known to be buggy, and may be removed in in a
 533     future release.</li>
 534 <li>C++ EH support is disabled for this release.</li>
 535 <li>The MSIL backend is experimental.</li>
 536 <li>The IA64 code generator is experimental.</li>
 537 <li>The Alpha JIT is experimental.</li>
 538 <li>"<tt>-filetype=asm</tt>" (the default) is the only supported value for the
 539     <tt>-filetype</tt> llc option.</li>
 540 </ul>
 541
 542 </div>
 543
 544 <!-- ======================================================================= -->
 545 <div class="doc_subsection">
 546   <a name="x86-be">Known problems with the X86 back-end</a>
 547 </div>
 548
 549 <div class="doc_text">
 550
 551 <ul>
 552 <li>The X86 backend does not yet support <a href="http://llvm.org/PR879">inline
 553     assembly that uses the X86 floating point stack</a>.</li>
 554 </ul>
 555
 556 </div>
 557
 558 <!-- ======================================================================= -->
 559 <div class="doc_subsection">
 560   <a name="ppc-be">Known problems with the PowerPC back-end</a>
 561 </div>
 562
 563 <div class="doc_text">
 564
 565 <ul>
 566 <li><a href="http://llvm.org/PR642">PowerPC backend does not correctly
 567 implement ordered FP comparisons</a>.</li>
 568 <li>The Linux PPC32/ABI support needs testing for the interpreter and static
 569 compilation, and lacks support for debug information.</li>
 570 </ul>
 571
 572 </div>
 573
 574 <!-- ======================================================================= -->
 575 <div class="doc_subsection">
 576   <a name="arm-be">Known problems with the ARM back-end</a>
 577 </div>
 578
 579 <div class="doc_text">
 580
 581 <ul>
 582 <li>Thumb mode works only on ARMv6 or higher processors. On sub-ARMv6
 583 processors, thumb program can crash or produces wrong
 584 results (<a href="http://llvm.org/PR1388">PR1388</a>).</li>
 585 <li>Compilation for ARM Linux OABI (old ABI) is supported, but not fully tested.
 586 </li>
 587 <li>There is a bug in QEMU-ARM (<= 0.9.0) which causes it to incorrectly execute
 588 programs compiled with LLVM.  Please use more recent versions of QEMU.</li>
 589 </ul>
 590
 591 </div>
 592
 593 <!-- ======================================================================= -->
 594 <div class="doc_subsection">
 595   <a name="sparc-be">Known problems with the SPARC back-end</a>
 596 </div>
 597
 598 <div class="doc_text">
 599
 600 <ul>
 601 <li>The SPARC backend only supports the 32-bit SPARC ABI (-m32), it does not
 602     support the 64-bit SPARC ABI (-m64).</li>
 603 </ul>
 604
 605 </div>
 606
 607 <!-- ======================================================================= -->
 608 <div class="doc_subsection">
 609   <a name="alpha-be">Known problems with the Alpha back-end</a>
 610 </div>
 611
 612 <div class="doc_text">
 613
 614 <ul>
 615
 616 <li>On 21164s, some rare FP arithmetic sequences which may trap do not have the
 617 appropriate nops inserted to ensure restartability.</li>
 618
 619 </ul>
 620 </div>
 621
 622 <!-- ======================================================================= -->
 623 <div class="doc_subsection">
 624   <a name="ia64-be">Known problems with the IA64 back-end</a>
 625 </div>
 626
 627 <div class="doc_text">
 628
 629 <ul>
 630
 631 <li>C++ programs are likely to fail on IA64, as calls to <tt>setjmp</tt> are
 632 made where the argument is not 16-byte aligned, as required on IA64. (Strictly
 633 speaking this is not a bug in the IA64 back-end; it will also be encountered
 634 when building C++ programs using the C back-end.)</li>
 635
 636 <li>The C++ front-end does not use <a href="http://llvm.org/PR406">IA64
 637 ABI compliant layout of v-tables</a>.  In particular, it just stores function
 638 pointers instead of function descriptors in the vtable.  This bug prevents
 639 mixing C++ code compiled with LLVM with C++ objects compiled by other C++
 640 compilers.</li>
 641
 642 <li>There are a few ABI violations which will lead to problems when mixing LLVM
 643 output with code built with other compilers, particularly for floating-point
 644 programs.</li>
 645
 646 <li>Defining vararg functions is not supported (but calling them is ok).</li>
 647
 648 <li>The Itanium backend has bitrotted somewhat.</li>
 649 </ul>
 650
 651 </div>
 652
 653 <!-- ======================================================================= -->
 654 <div class="doc_subsection">
 655   <a name="c-be">Known problems with the C back-end</a>
 656 </div>
 657
 658 <div class="doc_text">
 659
 660 <ul>
 661 <li><a href="http://llvm.org/PR802">The C backend does not support inline
 662     assembly code</a>.</li>
 663 </ul>
 664
 665 </div>
 666
 667
 668 <!-- ======================================================================= -->
 669 <div class="doc_subsection">
 670   <a name="c-fe">Known problems with the C front-end</a>
 671 </div>
 672
 673 <!-- _______________________________________________________________________ -->
 674 <div class="doc_subsubsection">Bugs</div>
 675
 676 <div class="doc_text">
 677
 678 <p>llvm-gcc4 does not currently support <a href="http://llvm.org/PR869">Link-Time
 679 Optimization</a> on most platforms "out-of-the-box".  Please inquire on the
 680 llvmdev mailing list if you are interested.</p>
 681
 682 </div>
 683
 684 <!-- _______________________________________________________________________ -->
 685 <div class="doc_subsubsection">
 686   Notes
 687 </div>
 688
 689 <div class="doc_text">
 690 <ul>
 691
 692 <li><p>"long double" is silently transformed by the front-end into "double".  There
 693 is no support for floating point data types of any size other than 32 and 64
 694 bits.</p></li>
 695
 696 <li><p>llvm-gcc does <b>not</b> support <tt>__builtin_apply</tt> yet.
 697   See <a href="http://gcc.gnu.org/onlinedocs/gcc/Constructing-Calls.html#Constructing%20Calls">Constructing Calls</a>: Dispatching a call to another function.</p>
 698 </li>
 699
 700 <li><p>llvm-gcc <b>partially</b> supports these GCC extensions:</p>
 701   <ol>
 702   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Nested-Functions.html#Nested%20Functions">Nested Functions</a>: As in Algol and Pascal, lexical scoping of functions.<br>
 703       Nested functions are supported, but llvm-gcc does not support non-local
 704       gotos or taking the address of a nested function.</li>
 705
 706   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Function-Attributes.html#Function%20Attributes">Function Attributes</a>:
 707
 708       Declaring that functions have no side effects or that they can never
 709       return.<br>
 710
 711       <b>Supported:</b> <tt>alias</tt>, <tt>always_inline</tt>, <tt>cdecl</tt>,
 712       <tt>constructor</tt>, <tt>destructor</tt>,
 713       <tt>deprecated</tt>, <tt>fastcall</tt>, <tt>format</tt>,
 714       <tt>format_arg</tt>, <tt>non_null</tt>, <tt>noreturn</tt>, <tt>regparm</tt>
 715       <tt>section</tt>, <tt>stdcall</tt>, <tt>unused</tt>, <tt>used</tt>,
 716       <tt>visibility</tt>, <tt>warn_unused_result</tt>, <tt>weak</tt><br>
 717
 718       <b>Ignored:</b> <tt>noinline</tt>, <tt>pure</tt>, <tt>const</tt>, <tt>nothrow</tt>,
 719       <tt>malloc</tt>, <tt>no_instrument_function</tt></li>
 720   </ol>
 721 </li>
 722
 723 <li><p>llvm-gcc supports the vast majority of GCC extensions, including:</p>
 724
 725   <ol>
 726   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Pragmas.html#Pragmas">Pragmas</a>: Pragmas accepted by GCC.</li>
 727   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Local-Labels.html#Local%20Labels">Local Labels</a>: Labels local to a block.</li>
 728   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Other-Builtins.html#Other%20Builtins">Other Builtins</a>:
 729       Other built-in functions.</li>
 730   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Variable-Attributes.html#Variable%20Attributes">Variable Attributes</a>:
 731       Specifying attributes of variables.</li>
 732   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Type-Attributes.html#Type%20Attributes">Type Attributes</a>:   Specifying attributes of types.</li>
 733   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Thread_002dLocal.html">Thread-Local</a>: Per-thread variables.</li>
 734   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Variable-Length.html#Variable%20Length">Variable Length</a>:
 735       Arrays whose length is computed at run time.</li>
 736   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Labels-as-Values.html#Labels%20as%20Values">Labels as Values</a>: Getting pointers to labels and computed gotos.</li>
 737   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Statement-Exprs.html#Statement%20Exprs">Statement Exprs</a>:   Putting statements and declarations inside expressions.</li>
 738   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Typeof.html#Typeof">Typeof</a>: <code>typeof</code>: referring to the type of an expression.</li>
 739   <li><a href="http://gcc.gnu.org/onlinedocs/gcc-3.4.0/gcc/Lvalues.html#Lvalues">Lvalues</a>: Using <code>?:</code>, "<code>,</code>" and casts in lvalues.</li>
 740   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Conditionals.html#Conditionals">Conditionals</a>: Omitting the middle operand of a <code>?:</code> expression.</li>
 741   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Long-Long.html#Long%20Long">Long Long</a>: Double-word integers.</li>
 742   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Complex.html#Complex">Complex</a>:   Data types for complex numbers.</li>
 743   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Hex-Floats.html#Hex%20Floats">Hex Floats</a>:Hexadecimal floating-point constants.</li>
 744   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html#Zero%20Length">Zero Length</a>: Zero-length arrays.</li>
 745   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Empty-Structures.html#Empty%20Structures">Empty Structures</a>: Structures with no members.</li>
 746   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Variadic-Macros.html#Variadic%20Macros">Variadic Macros</a>: Macros with a variable number of arguments.</li>
 747   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Escaped-Newlines.html#Escaped%20Newlines">Escaped Newlines</a>:  Slightly looser rules for escaped newlines.</li>
 748   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Extended-Asm.html#Extended%20Asm">Extended Asm</a>: Assembler instructions with C expressions as operands.</li>
 749   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Constraints.html#Constraints">Constraints</a>: Constraints for asm operands.</li>
 750   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Asm-Labels.html#Asm%20Labels">Asm Labels</a>: Specifying the assembler name to use for a C symbol.</li>
 751   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Explicit-Reg-Vars.html#Explicit%20Reg%20Vars">Explicit Reg Vars</a>: Defining variables residing in specified registers.</li>
 752   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Vector-Extensions.html#Vector%20Extensions">Vector Extensions</a>: Using vector instructions through built-in functions.</li>
 753   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Target-Builtins.html#Target%20Builtins">Target Builtins</a>:   Built-in functions specific to particular targets.</li>
 754   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Subscripting.html#Subscripting">Subscripting</a>: Any array can be subscripted, even if not an lvalue.</li>
 755   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Pointer-Arith.html#Pointer%20Arith">Pointer Arith</a>: Arithmetic on <code>void</code>-pointers and function pointers.</li>
 756   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Initializers.html#Initializers">Initializers</a>: Non-constant initializers.</li>
 757   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Compound-Literals.html#Compound%20Literals">Compound Literals</a>: Compound literals give structures, unions,
 758 or arrays as values.</li>
 759   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Designated-Inits.html#Designated%20Inits">Designated Inits</a>: Labeling elements of initializers.</li>
 760   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Cast-to-Union.html#Cast%20to%20Union">Cast to Union</a>: Casting to union type from any member of the union.</li>
 761   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Case-Ranges.html#Case%20Ranges">Case Ranges</a>: `case 1 ... 9' and such.</li>
 762   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Mixed-Declarations.html#Mixed%20Declarations">Mixed Declarations</a>: Mixing declarations and code.</li>
 763   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Function-Prototypes.html#Function%20Prototypes">Function Prototypes</a>: Prototype declarations and old-style definitions.</li>
 764   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/C_002b_002b-Comments.html#C_002b_002b-Comments">C++ Comments</a>: C++ comments are recognized.</li>
 765   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Dollar-Signs.html#Dollar%20Signs">Dollar Signs</a>: Dollar sign is allowed in identifiers.</li>
 766   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Character-Escapes.html#Character%20Escapes">Character Escapes</a>: <code>\e</code> stands for the character &lt;ESC&gt;.</li>
 767   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Alignment.html#Alignment">Alignment</a>: Inquiring about the alignment of a type or variable.</li>
 768   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Inline.html#Inline">Inline</a>: Defining inline functions (as fast as macros).</li>
 769   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Alternate-Keywords.html#Alternate%20Keywords">Alternate Keywords</a>:<code>__const__</code>, <code>__asm__</code>, etc., for header files.</li>
 770   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Incomplete-Enums.html#Incomplete%20Enums">Incomplete Enums</a>:  <code>enum foo;</code>, with details to follow.</li>
 771   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Function-Names.html#Function%20Names">Function Names</a>: Printable strings which are the name of the current function.</li>
 772   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Return-Address.html#Return%20Address">Return Address</a>: Getting the return or frame address of a function.</li>
 773   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Unnamed-Fields.html#Unnamed%20Fields">Unnamed Fields</a>: Unnamed struct/union fields within structs/unions.</li>
 774   <li><a href="http://gcc.gnu.org/onlinedocs/gcc/Attribute-Syntax.html#Attribute%20Syntax">Attribute Syntax</a>: Formal syntax for attributes.</li>
 775   </ol></li>
 776
 777 </ul>
 778
 779 <p>If you run into GCC extensions which have not been included in any of these
 780 lists, please let us know (also including whether or not they work).</p>
 781
 782 </div>
 783
 784 <!-- ======================================================================= -->
 785 <div class="doc_subsection">
 786   <a name="c++-fe">Known problems with the C++ front-end</a>
 787 </div>
 788
 789 <div class="doc_text">
 790
 791 <p>The C++ front-end is considered to be fully
 792 tested and works for a number of non-trivial programs, including LLVM
 793 itself, Qt, Mozilla, etc.</p>
 794
 795 <ul>
 796 <li>llvm-gcc4 only has partial support for <a href="http://llvm.org/PR870">C++
 797 Exception Handling</a>, and it is not enabled by default.</li>
 798
 799 <!-- NO EH Support!
 800
 801 <li>Destructors for local objects are not always run when a <tt>longjmp</tt> is
 802     performed. In particular, destructors for objects in the <tt>longjmp</tt>ing
 803     function and in the <tt>setjmp</tt> receiver function may not be run.
 804     Objects in intervening stack frames will be destroyed, however (which is
 805     better than most compilers).</li>
 806
 807 <li>The LLVM C++ front-end follows the <a
 808     href="http://www.codesourcery.com/cxx-abi">Itanium C++ ABI</a>.
 809     This document, which is not Itanium specific, specifies a standard for name
 810     mangling, class layout, v-table layout, RTTI formats, and other C++
 811     representation issues.  Because we use this API, code generated by the LLVM
 812     compilers should be binary compatible with machine code generated by other
 813     Itanium ABI C++ compilers (such as G++, the Intel and HP compilers, etc).
 814     <i>However</i>, the exception handling mechanism used by llvm-gcc3 is very
 815     different from the model used in the Itanium ABI, so <b>exceptions will not
 816     interact correctly</b>. </li>
 817 -->
 818 </ul>
 819
 820 </div>
 821
 822
 823
 824 <!-- *********************************************************************** -->
 825 <div class="doc_section">
 826   <a name="additionalinfo">Additional Information</a>
 827 </div>
 828 <!-- *********************************************************************** -->
 829
 830 <div class="doc_text">
 831
 832 <p>A wide variety of additional information is available on the <a
 833 href="http://llvm.org">LLVM web page</a>, in particular in the <a
 834 href="http://llvm.org/docs/">documentation</a> section.  The web page also
 835 contains versions of the API documentation which is up-to-date with the CVS
 836 version of the source code.
 837 You can access versions of these documents specific to this release by going
 838 into the "<tt>llvm/doc/</tt>" directory in the LLVM tree.</p>
 839
 840 <p>If you have any questions or comments about LLVM, please feel free to contact
 841 us via the <a href="http://llvm.org/docs/#maillist"> mailing
 842 lists</a>.</p>
 843
 844 </div>
 845
 846 <!-- *********************************************************************** -->
 847
 848 <hr>
 849 <address>
 850   <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
 851   src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
 852   <a href="http://validator.w3.org/check/referer"><img
 853   src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!" /></a>
 854
 855   <a href="http://llvm.org/">LLVM Compiler Infrastructure</a><br>
 856   Last modified: $Date$
 857 </address>
 858
 859 </body>
 860 </html>