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 3.0 Release Notes</title> 8</head> 9<body> 10 11<h1>LLVM 3.0 Release Notes</h1> 12 13<img align=right src="http://llvm.org/img/DragonSmall.png" 14 width="136" height="136" alt="LLVM Dragon Logo"> 15 16<ol> 17 <li><a href="#intro">Introduction</a></li> 18 <li><a href="#subproj">Sub-project Status Update</a></li> 19 <li><a href="#externalproj">External Projects Using LLVM 3.0</a></li> 20 <li><a href="#whatsnew">What's New in LLVM 3.0?</a></li> 21 <li><a href="GettingStarted.html">Installation Instructions</a></li> 22 <li><a href="#knownproblems">Known Problems</a></li> 23 <li><a href="#additionalinfo">Additional Information</a></li> 24</ol> 25 26<div class="doc_author"> 27 <p>Written by the <a href="http://llvm.org/">LLVM Team</a></p> 28</div> 29 30<!-- 31<h1 style="color:red">These are in-progress notes for the upcoming LLVM 3.0 32release.<br> 33You may prefer the 34<a href="http://llvm.org/releases/2.9/docs/ReleaseNotes.html">LLVM 2.9 35Release Notes</a>.</h1> 36 --> 37 38<!-- *********************************************************************** --> 39<h2> 40 <a name="intro">Introduction</a> 41</h2> 42<!-- *********************************************************************** --> 43 44<div> 45 46<p>This document contains the release notes for the LLVM Compiler 47 Infrastructure, release 3.0. Here we describe the status of LLVM, including 48 major improvements from the previous release and significant known problems. 49 All LLVM releases may be downloaded from 50 the <a href="http://llvm.org/releases/">LLVM releases web site</a>.</p> 51 52<p>For more information about LLVM, including information about the latest 53 release, please check out the <a href="http://llvm.org/">main LLVM web 54 site</a>. If you have questions or comments, 55 the <a href="http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev">LLVM 56 Developer's Mailing List</a> is a good place to send them.</p> 57 58<p>Note that if you are reading this file from a Subversion checkout or the main 59 LLVM web page, this document applies to the <i>next</i> release, not the 60 current one. To see the release notes for a specific release, please see the 61 <a href="http://llvm.org/releases/">releases page</a>.</p> 62 63</div> 64 65<!-- Features that need text if they're finished for 3.1: 66 ARM EHABI 67 combiner-aa? 68 strong phi elim 69 loop dependence analysis 70 CorrelatedValuePropagation 71 lib/Transforms/IPO/MergeFunctions.cpp => consider for 3.1. 72 --> 73 74<!-- *********************************************************************** --> 75<h2> 76 <a name="subproj">Sub-project Status Update</a> 77</h2> 78<!-- *********************************************************************** --> 79 80<div> 81 82<p>The LLVM 3.0 distribution currently consists of code from the core LLVM 83 repository (which roughly includes the LLVM optimizers, code generators and 84 supporting tools), the Clang repository and the llvm-gcc repository. In 85 addition to this code, the LLVM Project includes other sub-projects that are 86 in development. Here we include updates on these subprojects.</p> 87 88<!--=========================================================================--> 89<h3> 90<a name="clang">Clang: C/C++/Objective-C Frontend Toolkit</a> 91</h3> 92 93<div> 94 95<p><a href="http://clang.llvm.org/">Clang</a> is an LLVM front end for the C, 96 C++, and Objective-C languages. Clang aims to provide a better user 97 experience through expressive diagnostics, a high level of conformance to 98 language standards, fast compilation, and low memory use. Like LLVM, Clang 99 provides a modular, library-based architecture that makes it suitable for 100 creating or integrating with other development tools. Clang is considered a 101 production-quality compiler for C, Objective-C, C++ and Objective-C++ on x86 102 (32- and 64-bit), and for darwin/arm targets.</p> 103 104<p>In the LLVM 3.0 time-frame, the Clang team has made many improvements:</p> 105 106<ul> 107 <li>Greatly improved support for building C++ applications, with greater 108 stability and better diagnostics.</li> 109 110 <li><a href="http://clang.llvm.org/cxx_status.html">Improved support</a> for 111 the <a href="http://www.iso.org/iso/iso_catalogue/catalogue_tc/catalogue_detail.htm?csnumber=50372">C++ 112 2011</a> standard, including implementations of non-static data member 113 initializers, alias templates, delegating constructors, the range-based 114 for loop, and implicitly-generated move constructors and move assignment 115 operators, among others.</li> 116 117 <li>Implemented support for some features of the upcoming C1x standard, 118 including static assertions and generic selections.</li> 119 120 <li>Better detection of include and linking paths for system headers and 121 libraries, especially for Linux distributions.</li> 122 123 <li>Implemented support 124 for <a href="http://clang.llvm.org/docs/AutomaticReferenceCounting.html">Automatic 125 Reference Counting</a> for Objective-C.</li> 126 127 <li>Implemented a number of optimizations in <tt>libclang</tt>, the Clang C 128 interface, to improve the performance of code completion and the mapping 129 from source locations to abstract syntax tree nodes.</li> 130</ul> 131 132 133<p>If Clang rejects your code but another compiler accepts it, please take a 134 look at the <a href="http://clang.llvm.org/compatibility.html">language 135 compatibility</a> guide to make sure this is not intentional or a known 136 issue.</p> 137 138</div> 139 140<!--=========================================================================--> 141<h3> 142<a name="dragonegg">DragonEgg: GCC front-ends, LLVM back-end</a> 143</h3> 144 145<div> 146<p><a href="http://dragonegg.llvm.org/">DragonEgg</a> is a 147 <a href="http://gcc.gnu.org/wiki/plugins">gcc plugin</a> that replaces GCC's 148 optimizers and code generators with LLVM's. Currently it requires a patched 149 version of gcc-4.5. The plugin can target the x86-32 and x86-64 processor 150 families and has been used successfully on the Darwin, FreeBSD and Linux 151 platforms. The Ada, C, C++ and Fortran languages work well. The plugin is 152 capable of compiling plenty of Obj-C, Obj-C++ and Java but it is not known 153 whether the compiled code actually works or not!</p> 154 155<p>The 3.0 release has the following notable changes:</p> 156 157<ul> 158<!-- 159<li></li> 160--> 161</ul> 162 163</div> 164 165<!--=========================================================================--> 166<h3> 167<a name="compiler-rt">compiler-rt: Compiler Runtime Library</a> 168</h3> 169 170<div> 171 172<p>The new LLVM <a href="http://compiler-rt.llvm.org/">compiler-rt project</a> 173 is a simple library that provides an implementation of the low-level 174 target-specific hooks required by code generation and other runtime 175 components. For example, when compiling for a 32-bit target, converting a 176 double to a 64-bit unsigned integer is compiled into a runtime call to the 177 "__fixunsdfdi" function. The compiler-rt library provides highly optimized 178 implementations of this and other low-level routines (some are 3x faster than 179 the equivalent libgcc routines).</p> 180 181<p>In the LLVM 3.0 timeframe,</p> 182 183</div> 184 185<!--=========================================================================--> 186<h3> 187<a name="lldb">LLDB: Low Level Debugger</a> 188</h3> 189 190<div> 191 192<p>LLDB has advanced by leaps and bounds in the 3.0 timeframe. It is 193 dramatically more stable and useful, and includes both a 194 new <a href="http://lldb.llvm.org/tutorial.html">tutorial</a> and 195 a <a href="http://lldb.llvm.org/lldb-gdb.html">side-by-side comparison with 196 GDB</a>.</p> 197 198</div> 199 200<!--=========================================================================--> 201<h3> 202<a name="libc++">libc++: C++ Standard Library</a> 203</h3> 204 205<div> 206 207<p>Like compiler_rt, libc++ is now <a href="DeveloperPolicy.html#license">dual 208 licensed</a> under the MIT and UIUC license, allowing it to be used more 209 permissively.</p> 210 211</div> 212 213 214<!--=========================================================================--> 215<h3> 216<a name="LLBrowse">LLBrowse: IR Browser</a> 217</h3> 218 219<div> 220 221<p><a href="http://llvm.org/svn/llvm-project/llbrowse/trunk/doc/LLBrowse.html"> 222 LLBrowse</a> is an interactive viewer for LLVM modules. It can load any LLVM 223 module and displays its contents as an expandable tree view, facilitating an 224 easy way to inspect types, functions, global variables, or metadata nodes. It 225 is fully cross-platform, being based on the popular wxWidgets GUI 226 toolkit.</p> 227 228</div> 229 230<!--=========================================================================--> 231<h3> 232<a name="vmkit">VMKit</a> 233</h3> 234 235<div> 236 237<p>The <a href="http://vmkit.llvm.org/">VMKit project</a> is an implementation 238 of a Java Virtual Machine (Java VM or JVM) that uses LLVM for static and 239 just-in-time compilation. As of LLVM 3.0, VMKit now supports generational 240 garbage collectors. The garbage collectors are provided by the MMTk 241 framework, and VMKit can be configured to use one of the numerous implemented 242 collectors of MMTk.</p> 243 244</div> 245 246 247<!--=========================================================================--> 248<!-- 249<h3> 250<a name="klee">KLEE: A Symbolic Execution Virtual Machine</a> 251</h3> 252 253<div> 254<p> 255<a href="http://klee.llvm.org/">KLEE</a> is a symbolic execution framework for 256programs in LLVM bitcode form. KLEE tries to symbolically evaluate "all" paths 257through the application and records state transitions that lead to fault 258states. This allows it to construct testcases that lead to faults and can even 259be used to verify some algorithms. 260</p> 261 262<p>UPDATE!</p> 263</div>--> 264 265</div> 266 267<!-- *********************************************************************** --> 268<h2> 269 <a name="externalproj">External Open Source Projects Using LLVM 3.0</a> 270</h2> 271<!-- *********************************************************************** --> 272 273<div> 274 275<p>An exciting aspect of LLVM is that it is used as an enabling technology for 276 a lot of other language and tools projects. This section lists some of the 277 projects that have already been updated to work with LLVM 3.0.</p> 278 279<!--=========================================================================--> 280<h3>AddressSanitizer</h3> 281 282<div> 283 284<p><a href="http://code.google.com/p/address-sanitizer/">AddressSanitizer</a> 285 uses compiler instrumentation and a specialized malloc library to find C/C++ 286 bugs such as use-after-free and out-of-bound accesses to heap, stack, and 287 globals. The key feature of the tool is speed: the average slowdown 288 introduced by AddressSanitizer is less than 2x.</p> 289 290</div> 291 292<!--=========================================================================--> 293<h3>ClamAV</h3> 294 295<div> 296 297<p><a href="http://www.clamav.net">Clam AntiVirus</a> is an open source (GPL) 298 anti-virus toolkit for UNIX, designed especially for e-mail scanning on mail 299 gateways.</p> 300 301<p>Since version 0.96 it 302 has <a href="http://vrt-sourcefire.blogspot.com/2010/09/introduction-to-clamavs-low-level.html">bytecode 303 signatures</a> that allow writing detections for complex malware.</p> 304 305<p>It uses LLVM's JIT to speed up the execution of bytecode on X86, X86-64, 306 PPC32/64, falling back to its own interpreter otherwise. The git version was 307 updated to work with LLVM 3.0.</p> 308 309</div> 310 311<!--=========================================================================--> 312<h3>clReflect</h3> 313 314<div> 315 316<p><a href="https://bitbucket.org/dwilliamson/clreflect">clReflect</a> is a C++ 317 parser that uses clang/LLVM to derive a light-weight reflection database 318 suitable for use in game development. It comes with a very simple runtime 319 library for loading and querying the database, requiring no external 320 dependencies (including CRT), and an additional utility library for object 321 management and serialisation.</p> 322 323</div> 324 325<!--=========================================================================--> 326<h3>Cling C++ Interpreter</h3> 327 328<div> 329 330<p><a href="http://cern.ch/cling">Cling</a> is an interactive compiler interface 331 (aka C++ interpreter). It uses LLVM's JIT and clang; it currently supports 332 C++ and C. It has a prompt interface, runs source files, calls into shared 333 libraries, prints the value of expressions, even does runtime lookup of 334 identifiers (dynamic scopes). And it just behaves like one would expect from 335 an interpreter.</p> 336 337</div> 338 339<!--=========================================================================--> 340<!-- FIXME: Comment out 341<h3>Crack Programming Language</h3> 342 343<div> 344<p> 345<a href="http://code.google.com/p/crack-language/">Crack</a> aims to provide the 346ease of development of a scripting language with the performance of a compiled 347language. The language derives concepts from C++, Java and Python, incorporating 348object-oriented programming, operator overloading and strong typing.</p> 349</div> 350--> 351 352<!--=========================================================================--> 353<h3>Glasgow Haskell Compiler (GHC)</h3> 354 355<div> 356 357<p>GHC is an open source, state-of-the-art programming suite for Haskell, a 358 standard lazy functional programming language. It includes an optimizing 359 static compiler generating good code for a variety of platforms, together 360 with an interactive system for convenient, quick development.</p> 361 362<p>GHC 7.0 and onwards include an LLVM code generator, supporting LLVM 2.8 and 363 later. Since LLVM 2.9, GHC now includes experimental support for the ARM 364 platform with LLVM 3.0.</p> 365 366</div> 367 368<!--=========================================================================--> 369<h3>gwXscript</h3> 370 371<div> 372 373<p><a href="http://botwars.tk/gwscript/">gwXscript</a> is an object oriented, 374 aspect oriented programming language which can create both executables (ELF, 375 EXE) and shared libraries (DLL, SO, DYNLIB). The compiler is implemented in 376 its own language and translates scripts into LLVM-IR which can be optimized 377 and translated into native code by the LLVM framework. Source code in 378 gwScript contains definitions that expand the namespaces. So you can build 379 your project and simply 'plug out' features by removing a file. The remaining 380 project does not leave scars since you directly separate concerns by the 381 'template' feature of gwX. It is also possible to add new features to a 382 project by just adding files and without editing the original project. This 383 language is used for example to create games or content management systems 384 that should be extendable.</p> 385 386<p>gwXscript is strongly typed and offers comfort with its native types string, 387 hash and array. You can easily write new libraries in gwXscript or native 388 code. gwXscript is type safe and users should not be able to crash your 389 program or execute malicious code except code that is eating CPU time.</p> 390 391</div> 392 393<!--=========================================================================--> 394<h3>include-what-you-use</h3> 395 396<div> 397 398<p><a href="http://code.google.com/p/include-what-you-use">include-what-you-use</a> 399 is a tool to ensure that a file directly <code>#include</code>s 400 all <code>.h</code> files that provide a symbol that the file uses. It also 401 removes superfluous <code>#include</code>s from source files.</p> 402 403</div> 404 405<!--=========================================================================--> 406<h3>LanguageKit and Pragmatic Smalltalk</h3> 407 408<div> 409 410<p><a href="http://etoileos.com/etoile/features/languagekit/">LanguageKit</a> is 411 a framework for implementing dynamic languages sharing an object model with 412 Objective-C. It provides static and JIT compilation using LLVM along with 413 its own interpreter. Pragmatic Smalltalk is a dialect of Smalltalk, built on 414 top of LanguageKit, that interfaces directly with Objective-C, sharing the 415 same object representation and message sending behaviour. These projects are 416 developed as part of the Étoié desktop environment.</p> 417 418</div> 419 420<!--=========================================================================--> 421<h3>LuaAV</h3> 422 423<div> 424 425<p><a href="http://lua-av.mat.ucsb.edu/blog/">LuaAV</a> is a real-time 426 audiovisual scripting environment based around the Lua language and a 427 collection of libraries for sound, graphics, and other media protocols. LuaAV 428 uses LLVM and Clang to JIT compile efficient user-defined audio synthesis 429 routines specified in a declarative syntax.</p> 430 431</div> 432 433<!--=========================================================================--> 434<h3>Mono</h3> 435 436<div> 437 438<p>An open source, cross-platform implementation of C# and the CLR that is 439 binary compatible with Microsoft.NET. Has an optional, dynamically-loaded 440 LLVM code generation backend in Mini, the JIT compiler.</p> 441 442<p>Note that we use a Git mirror of LLVM with some patches. See: 443 https://github.com/mono/llvm</p> 444 445</div> 446 447<!--=========================================================================--> 448<h3>Portable OpenCL (pocl)</h3> 449 450<div> 451 452<p>Portable OpenCL is an open source implementation of the OpenCL standard which 453 can be easily adapted for new targets. One of the goals of the project is 454 improving performance portability of OpenCL programs, avoiding the need for 455 target-dependent manual optimizations. A "native" target is included, which 456 allows running OpenCL kernels on the host (CPU).</p> 457 458</div> 459 460<!--=========================================================================--> 461<h3>Pure</h3> 462 463<div> 464<p><a href="http://pure-lang.googlecode.com/">Pure</a> is an 465 algebraic/functional programming language based on term rewriting. Programs 466 are collections of equations which are used to evaluate expressions in a 467 symbolic fashion. The interpreter uses LLVM as a backend to JIT-compile Pure 468 programs to fast native code. Pure offers dynamic typing, eager and lazy 469 evaluation, lexical closures, a hygienic macro system (also based on term 470 rewriting), built-in list and matrix support (including list and matrix 471 comprehensions) and an easy-to-use interface to C and other programming 472 languages (including the ability to load LLVM bitcode modules, and inline C, 473 C++, Fortran and Faust code in Pure programs if the corresponding LLVM-enabled 474 compilers are installed).</p> 475 476<p>Pure version 0.48 has been tested and is known to work with LLVM 3.0 477 (and continues to work with older LLVM releases >= 2.5).</p> 478 479</div> 480 481<!--=========================================================================--> 482<h3>Renderscript</h3> 483 484<div> 485 486<p><a href="http://developer.android.com/guide/topics/renderscript/index.html">Renderscript</a> 487 is Android's advanced 3D graphics rendering and compute API. It provides a 488 portable C99-based language with extensions to facilitate common use cases 489 for enhancing graphics and thread level parallelism. The Renderscript 490 compiler frontend is based on Clang/LLVM. It emits a portable bitcode format 491 for the actual compiled script code, as well as reflects a Java interface for 492 developers to control the execution of the compiled bitcode. Executable 493 machine code is then generated from this bitcode by an LLVM backend on the 494 device. Renderscript is thus able to provide a mechanism by which Android 495 developers can improve performance of their applications while retaining 496 portability.</p> 497 498</div> 499 500<!--=========================================================================--> 501<h3>SAFECode</h3> 502 503<div> 504 505<p><a href="http://safecode.cs.illinois.edu">SAFECode</a> is a memory safe C/C++ 506 compiler built using LLVM. It takes standard, unannotated C/C++ code, 507 analyzes the code to ensure that memory accesses and array indexing 508 operations are safe, and instruments the code with run-time checks when 509 safety cannot be proven statically. SAFECode can be used as a debugging aid 510 (like Valgrind) to find and repair memory safety bugs. It can also be used 511 to protect code from security attacks at run-time.</p> 512 513</div> 514 515<!--=========================================================================--> 516<h3>The Stupid D Compiler (SDC)</h3> 517 518<div> 519 520<p><a href="https://github.com/bhelyer/SDC">The Stupid D Compiler</a> is a 521 project seeking to write a self-hosting compiler for the D programming 522 language without using the frontend of the reference compiler (DMD).</p> 523 524</div> 525 526<!--=========================================================================--> 527<h3>TTA-based Co-design Environment (TCE)</h3> 528 529<div> 530 531<p>TCE is a toolset for designing application-specific processors (ASP) based on 532 the Transport triggered architecture (TTA). The toolset provides a complete 533 co-design flow from C/C++ programs down to synthesizable VHDL and parallel 534 program binaries. Processor customization points include the register files, 535 function units, supported operations, and the interconnection network.</p> 536 537<p>TCE uses Clang and LLVM for C/C++ language support, target independent 538 optimizations and also for parts of code generation. It generates new 539 LLVM-based code generators "on the fly" for the designed TTA processors and 540 loads them in to the compiler backend as runtime libraries to avoid 541 per-target recompilation of larger parts of the compiler chain.</p> 542 543</div> 544 545<!--=========================================================================--> 546<h3>Tart Programming Language</h3> 547 548<div> 549 550<p><a href="http://code.google.com/p/tart/">Tart</a> is a general-purpose, 551 strongly typed programming language designed for application 552 developers. Strongly inspired by Python and C#, Tart focuses on practical 553 solutions for the professional software developer, while avoiding the clutter 554 and boilerplate of legacy languages like Java and C++. Although Tart is still 555 in development, the current implementation supports many features expected of 556 a modern programming language, such as garbage collection, powerful 557 bidirectional type inference, a greatly simplified syntax for template 558 metaprogramming, closures and function literals, reflection, operator 559 overloading, explicit mutability and immutability, and much more. Tart is 560 flexible enough to accommodate a broad range of programming styles and 561 philosophies, while maintaining a strong commitment to simplicity, minimalism 562 and elegance in design.</p> 563 564</div> 565 566<!--=========================================================================--> 567<h3>ThreadSanitizer</h3> 568 569<div> 570 571<p><a href="http://code.google.com/p/data-race-test/">ThreadSanitizer</a> is a 572 data race detector for (mostly) C and C++ code, available for Linux, Mac OS 573 and Windows. On different systems, we use binary instrumentation frameworks 574 (Valgrind and Pin) as frontends that generate the program events for the race 575 detection algorithm. On Linux, there's an option of using LLVM-based 576 compile-time instrumentation.</p> 577 578</div> 579 580<!--=========================================================================--> 581<h3>The ZooLib C++ Cross-Platform Application Framework</h3> 582 583<div> 584 585<p><a href="http://www.zoolib.org/">ZooLib</a> is Open Source under the MIT 586 License. It provides GUI, filesystem access, TCP networking, thread-safe 587 memory management, threading and locking for Mac OS X, Classic Mac OS, 588 Microsoft Windows, POSIX operating systems with X11, BeOS, Haiku, Apple's iOS 589 and Research in Motion's BlackBerry.</p> 590 591<p>My current work is to use CLang's static analyzer to improve ZooLib's code 592 quality. I also plan to set up LLVM compiles of the demo programs and test 593 programs using CLang and LLVM on all the platforms that CLang, LLVM and 594 ZooLib all support.</p> 595 596</div> 597 598<!--=========================================================================--> 599<!-- 600<h3>PinaVM</h3> 601 602<div> 603<p><a href="http://gitorious.org/pinavm/pages/Home">PinaVM</a> is an open 604source, <a href="http://www.systemc.org/">SystemC</a> front-end. Unlike many 605other front-ends, PinaVM actually executes the elaboration of the 606program analyzed using LLVM's JIT infrastructure. It later enriches the 607bitcode with SystemC-specific information.</p> 608</div> 609--> 610 611 612<!--=========================================================================--> 613<!-- 614<h3 id="icedtea">IcedTea Java Virtual Machine Implementation</h3> 615 616<div> 617<p> 618<a href="http://icedtea.classpath.org/wiki/Main_Page">IcedTea</a> provides a 619harness to build OpenJDK using only free software build tools and to provide 620replacements for the not-yet free parts of OpenJDK. One of the extensions that 621IcedTea provides is a new JIT compiler named <a 622href="http://icedtea.classpath.org/wiki/ZeroSharkFaq">Shark</a> which uses LLVM 623to provide native code generation without introducing processor-dependent 624code. 625</p> 626 627<p> OpenJDK 7 b112, IcedTea6 1.9 and IcedTea7 1.13 and later have been tested 628and are known to work with LLVM 3.0 (and continue to work with older LLVM 629releases >= 2.6 as well).</p> 630</div> 631--> 632 633<!--=========================================================================--> 634<!-- 635<h3>Polly - Polyhedral optimizations for LLVM</h3> 636 637<div> 638<p>Polly is a project that aims to provide advanced memory access optimizations 639to better take advantage of SIMD units, cache hierarchies, multiple cores or 640even vector accelerators for LLVM. Built around an abstract mathematical 641description based on Z-polyhedra, it provides the infrastructure to develop 642advanced optimizations in LLVM and to connect complex external optimizers. In 643its first year of existence Polly already provides an exact value-based 644dependency analysis as well as basic SIMD and OpenMP code generation support. 645Furthermore, Polly can use PoCC(Pluto) an advanced optimizer for data-locality 646and parallelism.</p> 647</div> 648--> 649 650<!--=========================================================================--> 651<!-- 652<h3>Rubinius</h3> 653 654<div> 655 <p><a href="http://github.com/evanphx/rubinius">Rubinius</a> is an environment 656 for running Ruby code which strives to write as much of the implementation in 657 Ruby as possible. Combined with a bytecode interpreting VM, it uses LLVM to 658 optimize and compile ruby code down to machine code. Techniques such as type 659 feedback, method inlining, and deoptimization are all used to remove dynamism 660 from ruby execution and increase performance.</p> 661</div> 662--> 663 664<!--=========================================================================--> 665<!-- 666<h3> 667<a name="FAUST">FAUST Real-Time Audio Signal Processing Language</a> 668</h3> 669 670<div> 671<p> 672<a href="http://faust.grame.fr">FAUST</a> is a compiled language for real-time 673audio signal processing. The name FAUST stands for Functional AUdio STream. Its 674programming model combines two approaches: functional programming and block 675diagram composition. In addition with the C, C++, JAVA output formats, the 676Faust compiler can now generate LLVM bitcode, and works with LLVM 2.7-3.0.</p> 677 678</div> 679--> 680 681</div> 682 683<!-- *********************************************************************** --> 684<h2> 685 <a name="whatsnew">What's New in LLVM 3.0?</a> 686</h2> 687<!-- *********************************************************************** --> 688 689<div> 690 691<p>This release includes a huge number of bug fixes, performance tweaks and 692 minor improvements. Some of the major improvements and new features are 693 listed in this section.</p> 694 695<!--=========================================================================--> 696<h3> 697<a name="majorfeatures">Major New Features</a> 698</h3> 699 700<div> 701 702<p>LLVM 3.0 includes several major new capabilities:</p> 703 704<ul> 705 706<!-- 707<li></li> 708--> 709 710</ul> 711 712</div> 713 714<!--=========================================================================--> 715<h3> 716<a name="coreimprovements">LLVM IR and Core Improvements</a> 717</h3> 718 719<div> 720 721<p>LLVM IR has several new features for better support of new targets and that 722 expose new optimization opportunities:</p> 723 724<p>One of the biggest changes is that 3.0 has a new exception handling 725 system. The old system used LLVM intrinsics to convey the exception handling 726 information to the code generator. It worked in most cases, but not 727 all. Inlining was especially difficult to get right. Also, the intrinsics 728 could be moved away from the <code>invoke</code> instruction, making it hard 729 to recover that information.</p> 730 731<p>The new EH system makes exception handling a first-class member of the IR. It 732 adds two new instructions:</p> 733 734<ul> 735 <li><a href="LangRef.html#i_landingpad"><code>landingpad</code></a> — 736 this instruction defines a landing pad basic block. It contains all of the 737 information that's needed by the code generator. It's also required to be 738 the first non-PHI instruction in the landing pad. In addition, a landing 739 pad may be jumped to only by the unwind edge of an <code>invoke</code> 740 instruction.</li> 741 742 <li><a href="LangRef.html#i_resume"><code>resume</code></a> — this 743 instruction causes the current exception to resume traveling up the 744 stack. It replaces the <code>@llvm.eh.resume</code> intrinsic.</li> 745</ul> 746 747<p>Converting from the old EH API to the new EH API is rather simple, because a 748 lot of complexity has been removed. The two intrinsics, 749 <code>@llvm.eh.exception</code> and <code>@llvm.eh.selector</code> have been 750 superceded by the <code>landingpad</code> instruction. Instead of generating 751 a call to <code>@llvm.eh.exception</code> and <code>@llvm.eh.selector</code>: 752 753<div class="doc_code"> 754<pre> 755Function *ExcIntr = Intrinsic::getDeclaration(TheModule, 756 Intrinsic::eh_exception); 757Function *SlctrIntr = Intrinsic::getDeclaration(TheModule, 758 Intrinsic::eh_selector); 759 760// The exception pointer. 761Value *ExnPtr = Builder.CreateCall(ExcIntr, "exc_ptr"); 762 763std::vector<Value*> Args; 764Args.push_back(ExnPtr); 765Args.push_back(Builder.CreateBitCast(Personality, 766 Type::getInt8PtrTy(Context))); 767 768<i>// Add selector clauses to Args.</i> 769 770// The selector call. 771Builder.CreateCall(SlctrIntr, Args, "exc_sel"); 772</pre> 773</div> 774 775<p>You should instead generate a <code>landingpad</code> instruction, that 776 returns an exception object and selector value:</p> 777 778<div class="doc_code"> 779<pre> 780LandingPadInst *LPadInst = 781 Builder.CreateLandingPad(StructType::get(Int8PtrTy, Int32Ty, NULL), 782 Personality, 0); 783 784Value *LPadExn = Builder.CreateExtractValue(LPadInst, 0); 785Builder.CreateStore(LPadExn, getExceptionSlot()); 786 787Value *LPadSel = Builder.CreateExtractValue(LPadInst, 1); 788Builder.CreateStore(LPadSel, getEHSelectorSlot()); 789</pre> 790</div> 791 792<p>It's now trivial to add the individual clauses to the <code>landingpad</code> 793 instruction.</p> 794 795<div class="doc_code"> 796<pre> 797<i><b>// Adding a catch clause</b></i> 798Constant *TypeInfo = getTypeInfo(); 799LPadInst->addClause(TypeInfo); 800 801<i><b>// Adding a C++ catch-all</b></i> 802LPadInst->addClause(Constant::getNullValue(Builder.getInt8PtrTy())); 803 804<i><b>// Adding a cleanup</b></i> 805LPadInst->setCleanup(true); 806 807<i><b>// Adding a filter clause</b></i> 808std::vector<Constant*> TypeInfos; 809Constant *TypeInfo = getFilterTypeInfo(); 810TypeInfos.push_back(Builder.CreateBitCast(TypeInfo, Builder.getInt8PtrTy())); 811 812ArrayType *FilterTy = ArrayType::get(Int8PtrTy, TypeInfos.size()); 813LPadInst->addClause(ConstantArray::get(FilterTy, TypeInfos)); 814</pre> 815</div> 816 817<p>Converting from using the <code>@llvm.eh.resume</code> intrinsic to 818 the <code>resume</code> instruction is trivial. It takes the exception 819 pointer and exception selector values returned by 820 the <code>landingpad</code> instruction:</p> 821 822<div class="doc_code"> 823<pre> 824Type *UnwindDataTy = StructType::get(Builder.getInt8PtrTy(), 825 Builder.getInt32Ty(), NULL); 826Value *UnwindData = UndefValue::get(UnwindDataTy); 827Value *ExcPtr = Builder.CreateLoad(getExceptionObjSlot()); 828Value *ExcSel = Builder.CreateLoad(getExceptionSelSlot()); 829UnwindData = Builder.CreateInsertValue(UnwindData, ExcPtr, 0, "exc_ptr"); 830UnwindData = Builder.CreateInsertValue(UnwindData, ExcSel, 1, "exc_sel"); 831Builder.CreateResume(UnwindData); 832</pre> 833</div> 834 835</div> 836 837<!--=========================================================================--> 838<h3> 839<a name="optimizer">Optimizer Improvements</a> 840</h3> 841 842<div> 843 844<p>In addition to a large array of minor performance tweaks and bug fixes, this 845 release includes a few major enhancements and additions to the 846 optimizers:</p> 847 848<ul> 849<!-- 850<li></li> 851--> 852</li> 853 854</ul> 855 856</div> 857 858<!--=========================================================================--> 859<h3> 860<a name="mc">MC Level Improvements</a> 861</h3> 862 863<div> 864 865<p>The LLVM Machine Code (aka MC) subsystem was created to solve a number of 866 problems in the realm of assembly, disassembly, object file format handling, 867 and a number of other related areas that CPU instruction-set level tools work 868 in.</p> 869 870<ul> 871<!-- 872<li></li> 873--> 874</ul> 875 876<p>For more information, please see 877 the <a href="http://blog.llvm.org/2010/04/intro-to-llvm-mc-project.html">Intro 878 to the LLVM MC Project Blog Post</a>.</p> 879 880</div> 881 882<!--=========================================================================--> 883<h3> 884<a name="codegen">Target Independent Code Generator Improvements</a> 885</h3> 886 887<div> 888 889<p>We have put a significant amount of work into the code generator 890 infrastructure, which allows us to implement more aggressive algorithms and 891 make it run faster:</p> 892 893<ul> 894<!-- 895<li></li> 896--> 897</ul> 898</div> 899 900<!--=========================================================================--> 901<h3> 902<a name="x86">X86-32 and X86-64 Target Improvements</a> 903</h3> 904 905<div> 906 907<p>New features and major changes in the X86 target include:</p> 908 909<ul> 910 911 <li>The CRC32 intrinsics have been renamed. The intrinsics were previously 912 <code>@llvm.x86.sse42.crc32.[8|16|32]</code> 913 and <code>@llvm.x86.sse42.crc64.[8|64]</code>. They have been renamed to 914 <code>@llvm.x86.sse42.crc32.32.[8|16|32]</code> and 915 <code>@llvm.x86.sse42.crc32.64.[8|64]</code>.</li> 916 917</ul> 918 919</div> 920 921<!--=========================================================================--> 922<h3> 923<a name="ARM">ARM Target Improvements</a> 924</h3> 925 926<div> 927 928<p>New features of the ARM target include:</p> 929 930<ul> 931<!-- 932<li></li> 933--> 934</ul> 935</div> 936 937<!--=========================================================================--> 938<h3> 939<a name="OtherTS">Other Target Specific Improvements</a> 940</h3> 941 942<p>PPC32/ELF va_arg was implemented.</p> 943<p>PPC32 initial support for .o file writing was implemented.</p> 944 945<div> 946 947<ul> 948<!-- 949<li></li> 950--> 951</ul> 952 953</div> 954 955<!--=========================================================================--> 956<h3> 957<a name="changes">Major Changes and Removed Features</a> 958</h3> 959 960<div> 961 962<p>If you're already an LLVM user or developer with out-of-tree changes based on 963 LLVM 2.9, this section lists some "gotchas" that you may run into upgrading 964 from the previous release.</p> 965 966<ul> 967 <li>The <code>LLVMC</code> front end code was removed while separating 968 out language independence.</li> 969 <li>The <code>LowerSetJmp</code> pass wasn't used effectively by any 970 target and has been removed.</li> 971 <li>The old <code>TailDup</code> pass was not used in the standard pipeline 972 and was unable to update ssa form, so it has been removed. 973 <li>The syntax of volatile loads and stores in IR has been changed to 974 "<code>load volatile</code>"/"<code>store volatile</code>". The old 975 syntax ("<code>volatile load</code>"/"<code>volatile store</code>") 976 is still accepted, but is now considered deprecated.</li> 977 <li>The old atomic intrinscs (<code>llvm.memory.barrier</code> and 978 <code>llvm.atomic.*</code>) are now gone. Please use the new atomic 979 instructions, described in the <a href="Atomics.html">atomics guide</a>. 980</ul> 981 982<h4>Windows (32-bit)</h4> 983<div> 984 985<ul> 986 <li>On Win32(MinGW32 and MSVC), Windows 2000 will not be supported. 987 Windows XP or higher is required.</li> 988</ul> 989 990</div> 991 992</div> 993 994<!--=========================================================================--> 995<h3> 996<a name="api_changes">Internal API Changes</a> 997</h3> 998 999<div> 1000 1001<p>In addition, many APIs have changed in this release. Some of the major 1002 LLVM API changes are:</p> 1003 1004<ul> 1005 <li>The biggest and most pervasive change is that llvm::Type's are no longer 1006 returned or accepted as 'const' values. Instead, just pass around 1007 non-const Type's.</li> 1008 1009 <li><code>PHINode::reserveOperandSpace</code> has been removed. Instead, you 1010 must specify how many operands to reserve space for when you create the 1011 PHINode, by passing an extra argument 1012 into <code>PHINode::Create</code>.</li> 1013 1014 <li>PHINodes no longer store their incoming BasicBlocks as operands. Instead, 1015 the list of incoming BasicBlocks is stored separately, and can be accessed 1016 with new functions <code>PHINode::block_begin</code> 1017 and <code>PHINode::block_end</code>.</li> 1018 1019 <li>Various functions now take an <code>ArrayRef</code> instead of either a 1020 pair of pointers (or iterators) to the beginning and end of a range, or a 1021 pointer and a length. Others now return an <code>ArrayRef</code> instead 1022 of a reference to a <code>SmallVector</code> 1023 or <code>std::vector</code>. These include: 1024<ul> 1025<!-- Please keep this list sorted. --> 1026<li><code>CallInst::Create</code></li> 1027<li><code>ComputeLinearIndex</code> (in <code>llvm/CodeGen/Analysis.h</code>)</li> 1028<li><code>ConstantArray::get</code></li> 1029<li><code>ConstantExpr::getExtractElement</code></li> 1030<li><code>ConstantExpr::getGetElementPtr</code></li> 1031<li><code>ConstantExpr::getInBoundsGetElementPtr</code></li> 1032<li><code>ConstantExpr::getIndices</code></li> 1033<li><code>ConstantExpr::getInsertElement</code></li> 1034<li><code>ConstantExpr::getWithOperands</code></li> 1035<li><code>ConstantFoldCall</code> (in <code>llvm/Analysis/ConstantFolding.h</code>)</li> 1036<li><code>ConstantFoldInstOperands</code> (in <code>llvm/Analysis/ConstantFolding.h</code>)</li> 1037<li><code>ConstantVector::get</code></li> 1038<li><code>DIBuilder::createComplexVariable</code></li> 1039<li><code>DIBuilder::getOrCreateArray</code></li> 1040<li><code>ExtractValueInst::Create</code></li> 1041<li><code>ExtractValueInst::getIndexedType</code></li> 1042<li><code>ExtractValueInst::getIndices</code></li> 1043<li><code>FindInsertedValue</code> (in <code>llvm/Analysis/ValueTracking.h</code>)</li> 1044<li><code>gep_type_begin</code> (in <code>llvm/Support/GetElementPtrTypeIterator.h</code>)</li> 1045<li><code>gep_type_end</code> (in <code>llvm/Support/GetElementPtrTypeIterator.h</code>)</li> 1046<li><code>GetElementPtrInst::Create</code></li> 1047<li><code>GetElementPtrInst::CreateInBounds</code></li> 1048<li><code>GetElementPtrInst::getIndexedType</code></li> 1049<li><code>InsertValueInst::Create</code></li> 1050<li><code>InsertValueInst::getIndices</code></li> 1051<li><code>InvokeInst::Create</code></li> 1052<li><code>IRBuilder::CreateCall</code></li> 1053<li><code>IRBuilder::CreateExtractValue</code></li> 1054<li><code>IRBuilder::CreateGEP</code></li> 1055<li><code>IRBuilder::CreateInBoundsGEP</code></li> 1056<li><code>IRBuilder::CreateInsertValue</code></li> 1057<li><code>IRBuilder::CreateInvoke</code></li> 1058<li><code>MDNode::get</code></li> 1059<li><code>MDNode::getIfExists</code></li> 1060<li><code>MDNode::getTemporary</code></li> 1061<li><code>MDNode::getWhenValsUnresolved</code></li> 1062<li><code>SimplifyGEPInst</code> (in <code>llvm/Analysis/InstructionSimplify.h</code>)</li> 1063<li><code>TargetData::getIndexedOffset</code></li> 1064</ul></li> 1065 1066 <li>All forms of <code>StringMap::getOrCreateValue</code> have been remove 1067 except for the one which takes a <code>StringRef</code>.</li> 1068 1069 <li>The <code>LLVMBuildUnwind</code> function from the C API was removed. The 1070 LLVM <code>unwind</code> instruction has been deprecated for a long time 1071 and isn't used by the current front-ends. So this was removed during the 1072 exception handling rewrite.</li> 1073 1074 <li>The <code>LLVMAddLowerSetJmpPass</code> function from the C API was 1075 removed because the <code>LowerSetJmp</code> pass was removed.</li> 1076 1077 <li>The <code>DIBuilder</code> interface used by front ends to encode 1078 debugging information in the LLVM IR now expects clients to 1079 use <code>DIBuilder::finalize()</code> at the end of translation unit to 1080 complete debugging information encoding.</li> 1081 1082 <li>The way the type system works has been 1083 rewritten: <code>PATypeHolder</code> and <code>OpaqueType</code> are gone, 1084 and all APIs deal with <code>Type*</code> instead of <code>const 1085 Type*</code>. If you need to create recursive structures, then create a 1086 named structure, and use <code>setBody()</code> when all its elements are 1087 built. Type merging and refining is gone too: named structures are not 1088 merged with other structures, even if their layout is identical. (of 1089 course anonymous structures are still uniqued by layout).</li> 1090 1091 <li>TargetSelect.h moved to Support/ from Target/</li> 1092 1093 <li>UpgradeIntrinsicCall no longer upgrades pre-2.9 intrinsic calls (for 1094 example <code>llvm.memset.i32</code>).</li> 1095 1096 <li>It is mandatory to initialize all out-of-tree passes too and their dependencies now with 1097 <code>INITIALIZE_PASS{BEGIN,END,}</code> 1098 and <code>INITIALIZE_{PASS,AG}_DEPENDENCY</code>.</li> 1099 1100 <li>The interface for MemDepResult in MemoryDependenceAnalysis has been 1101 enhanced with new return types Unknown and NonFuncLocal, in addition to 1102 the existing types Clobber, Def, and NonLocal.</li> 1103</ul> 1104 1105</div> 1106 1107</div> 1108 1109<!-- *********************************************************************** --> 1110<h2> 1111 <a name="knownproblems">Known Problems</a> 1112</h2> 1113<!-- *********************************************************************** --> 1114 1115<div> 1116 1117<p>This section contains significant known problems with the LLVM system, listed 1118 by component. If you run into a problem, please check 1119 the <a href="http://llvm.org/bugs/">LLVM bug database</a> and submit a bug if 1120 there isn't already one.</p> 1121 1122<!-- ======================================================================= --> 1123<h3> 1124 <a name="experimental">Experimental features included with this release</a> 1125</h3> 1126 1127<div> 1128 1129<p>The following components of this LLVM release are either untested, known to 1130 be broken or unreliable, or are in early development. These components 1131 should not be relied on, and bugs should not be filed against them, but they 1132 may be useful to some people. In particular, if you would like to work on 1133 one of these components, please contact us on 1134 the <a href="http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev">LLVMdev 1135 list</a>.</p> 1136 1137<ul> 1138 <li>The Alpha, Blackfin, CellSPU, MicroBlaze, MSP430, MIPS, PTX, SystemZ and 1139 XCore backends are experimental.</li> 1140 1141 <li><tt>llc</tt> "<tt>-filetype=obj</tt>" is experimental on all targets other 1142 than darwin and ELF X86 systems.</li> 1143</ul> 1144 1145</div> 1146 1147<!-- ======================================================================= --> 1148<h3> 1149 <a name="x86-be">Known problems with the X86 back-end</a> 1150</h3> 1151 1152<div> 1153 1154<ul> 1155 <li>The X86 backend does not yet support 1156 all <a href="http://llvm.org/PR879">inline assembly that uses the X86 1157 floating point stack</a>. It supports the 'f' and 't' constraints, but 1158 not 'u'.</li> 1159 1160 <li>The X86-64 backend does not yet support the LLVM IR instruction 1161 <tt>va_arg</tt>. Currently, front-ends support variadic argument 1162 constructs on X86-64 by lowering them manually.</li> 1163 1164 <li>Windows x64 (aka Win64) code generator has a few issues. 1165 <ul> 1166 <li>llvm-gcc cannot build the mingw-w64 runtime currently due to lack of 1167 support for the 'u' inline assembly constraint and for X87 floating 1168 point inline assembly.</li> 1169 1170 <li>On mingw-w64, you will see unresolved symbol <tt>__chkstk</tt> due 1171 to <a href="http://llvm.org/bugs/show_bug.cgi?id=8919">Bug 8919</a>. 1172 It is fixed 1173 in <a href="http://lists.cs.uiuc.edu/pipermail/llvm-commits/Week-of-Mon-20110321/118499.html">r128206</a>.</li> 1174 1175 <li>Miss-aligned MOVDQA might crash your program. It is due to 1176 <a href="http://llvm.org/bugs/show_bug.cgi?id=9483">Bug 9483</a>, lack 1177 of handling aligned internal globals.</li> 1178 </ul> 1179 </li> 1180 1181</ul> 1182 1183</div> 1184 1185<!-- ======================================================================= --> 1186<h3> 1187 <a name="ppc-be">Known problems with the PowerPC back-end</a> 1188</h3> 1189 1190<div> 1191 1192<ul> 1193 <li>The PPC32/ELF support lacks PIC support.</li> 1194</ul> 1195 1196</div> 1197 1198<!-- ======================================================================= --> 1199<h3> 1200 <a name="arm-be">Known problems with the ARM back-end</a> 1201</h3> 1202 1203<div> 1204 1205<ul> 1206 <li>Thumb mode works only on ARMv6 or higher processors. On sub-ARMv6 1207 processors, thumb programs can crash or produce wrong results 1208 (<a href="http://llvm.org/PR1388">PR1388</a>).</li> 1209 1210 <li>Compilation for ARM Linux OABI (old ABI) is supported but not fully 1211 tested.</li> 1212</ul> 1213 1214</div> 1215 1216<!-- ======================================================================= --> 1217<h3> 1218 <a name="sparc-be">Known problems with the SPARC back-end</a> 1219</h3> 1220 1221<div> 1222 1223<ul> 1224 <li>The SPARC backend only supports the 32-bit SPARC ABI (-m32); it does not 1225 support the 64-bit SPARC ABI (-m64).</li> 1226</ul> 1227 1228</div> 1229 1230<!-- ======================================================================= --> 1231<h3> 1232 <a name="mips-be">Known problems with the MIPS back-end</a> 1233</h3> 1234 1235<div> 1236 1237<ul> 1238 <li>64-bit MIPS targets are not supported yet.</li> 1239</ul> 1240 1241</div> 1242 1243<!-- ======================================================================= --> 1244<h3> 1245 <a name="alpha-be">Known problems with the Alpha back-end</a> 1246</h3> 1247 1248<div> 1249 1250<ul> 1251 <li>On 21164s, some rare FP arithmetic sequences which may trap do not have 1252 the appropriate nops inserted to ensure restartability.</li> 1253</ul> 1254 1255</div> 1256 1257<!-- ======================================================================= --> 1258<h3> 1259 <a name="c-be">Known problems with the C back-end</a> 1260</h3> 1261 1262<div> 1263 1264<p>The C backend has numerous problems and is not being actively maintained. 1265 Depending on it for anything serious is not advised.</p> 1266 1267<ul> 1268 <li><a href="http://llvm.org/PR802">The C backend has only basic support for 1269 inline assembly code</a>.</li> 1270 1271 <li><a href="http://llvm.org/PR1658">The C backend violates the ABI of common 1272 C++ programs</a>, preventing intermixing between C++ compiled by the CBE 1273 and C++ code compiled with <tt>llc</tt> or native compilers.</li> 1274 1275 <li>The C backend does not support all exception handling constructs.</li> 1276 1277 <li>The C backend does not support arbitrary precision integers.</li> 1278</ul> 1279 1280</div> 1281 1282 1283<!-- ======================================================================= --> 1284<h3> 1285 <a name="llvm-gcc">Known problems with the llvm-gcc front-end</a> 1286</h3> 1287 1288<div> 1289 1290<p><b>LLVM 2.9 was the last release of llvm-gcc.</b></p> 1291 1292<p>llvm-gcc is generally very stable for the C family of languages. The only 1293 major language feature of GCC not supported by llvm-gcc is the 1294 <tt>__builtin_apply</tt> family of builtins. However, some extensions 1295 are only supported on some targets. For example, trampolines are only 1296 supported on some targets (these are used when you take the address of a 1297 nested function).</p> 1298 1299<p>Fortran support generally works, but there are still several unresolved bugs 1300 in <a href="http://llvm.org/bugs/">Bugzilla</a>. Please see the 1301 tools/gfortran component for details. Note that llvm-gcc is missing major 1302 Fortran performance work in the frontend and library that went into GCC after 1303 4.2. If you are interested in Fortran, we recommend that you consider using 1304 <a href="#dragonegg">dragonegg</a> instead.</p> 1305 1306<p>The llvm-gcc 4.2 Ada compiler has basic functionality, but is no longer being 1307 actively maintained. If you are interested in Ada, we recommend that you 1308 consider using <a href="#dragonegg">dragonegg</a> instead.</p> 1309 1310</div> 1311 1312</div> 1313 1314<!-- *********************************************************************** --> 1315<h2> 1316 <a name="additionalinfo">Additional Information</a> 1317</h2> 1318<!-- *********************************************************************** --> 1319 1320<div> 1321 1322<p>A wide variety of additional information is available on 1323 the <a href="http://llvm.org/">LLVM web page</a>, in particular in 1324 the <a href="http://llvm.org/docs/">documentation</a> section. The web page 1325 also contains versions of the API documentation which is up-to-date with the 1326 Subversion version of the source code. You can access versions of these 1327 documents specific to this release by going into the "<tt>llvm/doc/</tt>" 1328 directory in the LLVM tree.</p> 1329 1330<p>If you have any questions or comments about LLVM, please feel free to contact 1331 us via the <a href="http://llvm.org/docs/#maillist"> mailing lists</a>.</p> 1332 1333</div> 1334 1335<!-- *********************************************************************** --> 1336 1337<hr> 1338<address> 1339 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img 1340 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a> 1341 <a href="http://validator.w3.org/check/referer"><img 1342 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a> 1343 1344 <a href="http://llvm.org/">LLVM Compiler Infrastructure</a><br> 1345 Last modified: $Date: 2011-11-01 00:51:35 -0400 (Tue, 01 Nov 2011) $ 1346</address> 1347 1348</body> 1349</html> 1350