1 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
5 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
6 <link rel="stylesheet" href="_static/llvm.css" type="text/css">
7 <title>LLVM 3.1 Release Notes</title>
11 <h1>LLVM 3.1 Release Notes</h1>
14 <img style="float:right" src="http://llvm.org/img/DragonSmall.png"
15 width="136" height="136" alt="LLVM Dragon Logo">
19 <li><a href="#intro">Introduction</a></li>
20 <li><a href="#subproj">Sub-project Status Update</a></li>
21 <li><a href="#externalproj">External Projects Using LLVM 3.1</a></li>
22 <li><a href="#whatsnew">What's New in LLVM?</a></li>
23 <li><a href="GettingStarted.html">Installation Instructions</a></li>
24 <li><a href="#knownproblems">Known Problems</a></li>
25 <li><a href="#additionalinfo">Additional Information</a></li>
28 <div class="doc_author">
29 <p>Written by the <a href="http://llvm.org/">LLVM Team</a></p>
32 <h1 style="color:red">These are in-progress notes for the upcoming LLVM 3.1
35 <a href="http://llvm.org/releases/3.0/docs/ReleaseNotes.html">LLVM 3.0
36 Release Notes</a>.</h1>
38 <!-- *********************************************************************** -->
40 <a name="intro">Introduction</a>
42 <!-- *********************************************************************** -->
46 <p>This document contains the release notes for the LLVM Compiler
47 Infrastructure, release 3.1. Here we describe the status of LLVM, including
48 major improvements from the previous release, improvements in various
49 subprojects of LLVM, and some of the current users of the code.
50 All LLVM releases may be downloaded from
51 the <a href="http://llvm.org/releases/">LLVM releases web site</a>.</p>
53 <p>For more information about LLVM, including information about the latest
54 release, please check out the <a href="http://llvm.org/">main LLVM web
55 site</a>. If you have questions or comments,
56 the <a href="http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev">LLVM
57 Developer's Mailing List</a> is a good place to send them.</p>
59 <p>Note that if you are reading this file from a Subversion checkout or the main
60 LLVM web page, this document applies to the <i>next</i> release, not the
61 current one. To see the release notes for a specific release, please see the
62 <a href="http://llvm.org/releases/">releases page</a>.</p>
67 <!-- *********************************************************************** -->
69 <a name="subproj">Sub-project Status Update</a>
71 <!-- *********************************************************************** -->
75 <p>The LLVM 3.1 distribution currently consists of code from the core LLVM
76 repository (which roughly includes the LLVM optimizers, code generators and
77 supporting tools), and the Clang repository. In
78 addition to this code, the LLVM Project includes other sub-projects that are
79 in development. Here we include updates on these subprojects.</p>
81 <!--=========================================================================-->
83 <a name="clang">Clang: C/C++/Objective-C Frontend Toolkit</a>
88 <p><a href="http://clang.llvm.org/">Clang</a> is an LLVM front end for the C,
89 C++, and Objective-C languages. Clang aims to provide a better user
90 experience through expressive diagnostics, a high level of conformance to
91 language standards, fast compilation, and low memory use. Like LLVM, Clang
92 provides a modular, library-based architecture that makes it suitable for
93 creating or integrating with other development tools. Clang is considered a
94 production-quality compiler for C, Objective-C, C++ and Objective-C++ on x86
95 (32- and 64-bit), and for Darwin/ARM targets.</p>
97 <p>In the LLVM 3.1 time-frame, the Clang team has made many improvements:</p>
102 <p>For more details about the changes to Clang since the 2.9 release, see the
103 <a href="http://clang.llvm.org/docs/ReleaseNotes.html">Clang release notes</a>
107 <p>If Clang rejects your code but another compiler accepts it, please take a
108 look at the <a href="http://clang.llvm.org/compatibility.html">language
109 compatibility</a> guide to make sure this is not intentional or a known
114 <!--=========================================================================-->
116 <a name="dragonegg">DragonEgg: GCC front-ends, LLVM back-end</a>
120 <p><a href="http://dragonegg.llvm.org/">DragonEgg</a> is a
121 <a href="http://gcc.gnu.org/wiki/plugins">gcc plugin</a> that replaces GCC's
122 optimizers and code generators with LLVM's. It works with gcc-4.5 or gcc-4.6,
123 targets the x86-32 and x86-64 processor families, and has been successfully
124 used on the Darwin, FreeBSD, KFreeBSD, Linux and OpenBSD platforms. It fully
125 supports Ada, C, C++ and Fortran. It has partial support for Go, Java, Obj-C
128 <p>The 3.1 release has the following notable changes:</p>
138 <!--=========================================================================-->
140 <a name="compiler-rt">compiler-rt: Compiler Runtime Library</a>
145 <p>The new LLVM <a href="http://compiler-rt.llvm.org/">compiler-rt project</a>
146 is a simple library that provides an implementation of the low-level
147 target-specific hooks required by code generation and other runtime
148 components. For example, when compiling for a 32-bit target, converting a
149 double to a 64-bit unsigned integer is compiled into a runtime call to the
150 "__fixunsdfdi" function. The compiler-rt library provides highly optimized
151 implementations of this and other low-level routines (some are 3x faster than
152 the equivalent libgcc routines).</p>
158 <!--=========================================================================-->
160 <a name="lldb">LLDB: Low Level Debugger</a>
165 <p>LLDB is a ground-up implementation of a command line debugger, as well as a
166 debugger API that can be used from other applications. LLDB makes use of the
167 Clang parser to provide high-fidelity expression parsing (particularly for
168 C++) and uses the LLVM JIT for target support.</p>
174 <!--=========================================================================-->
176 <a name="libc++">libc++: C++ Standard Library</a>
181 <p>Like compiler_rt, libc++ is now <a href="DeveloperPolicy.html#license">dual
182 licensed</a> under the MIT and UIUC license, allowing it to be used more
189 <!--=========================================================================-->
191 <a name="vmkit">VMKit</a>
196 <p>The <a href="http://vmkit.llvm.org/">VMKit project</a> is an
197 implementation of a Java Virtual Machine (Java VM or JVM) that uses LLVM for
198 static and just-in-time compilation.
200 <p>In the LLVM 3.1 time-frame, VMKit has had significant improvements on both
201 runtime and startup performance:</p>
210 <!--=========================================================================-->
212 <a name="Polly">Polly: Polyhedral Optimizer</a>
217 <p><a href="http://polly.llvm.org/">Polly</a> is an <em>experimental</em>
218 optimizer for data locality and parallelism. It currently provides high-level
219 loop optimizations and automatic parallelisation (using the OpenMP run time).
220 Work in the area of automatic SIMD and accelerator code generation was
223 <p>Within the LLVM 3.1 time-frame there were the following highlights:</p>
226 <li>Polly became an official LLVM project</li>
227 <li>Polly can be loaded directly into clang (Enabled by '-O3 -mllvm -polly'
229 <li>An automatic scheduling optimizer (derived from <a
230 href="http://pluto-compiler.sourceforge.net/">Pluto</a>) was integrated. It
231 performs loop transformations to optimize for data-locality and parallelism.
232 The transformations include, but are not limited to interchange, fusion,
233 fission, skewing and tiling.
241 <!-- *********************************************************************** -->
243 <a name="externalproj">External Open Source Projects Using LLVM 3.1</a>
245 <!-- *********************************************************************** -->
249 <p>An exciting aspect of LLVM is that it is used as an enabling technology for
250 a lot of other language and tools projects. This section lists some of the
251 projects that have already been updated to work with LLVM 3.1.</p>
253 ... to be filled in right before the release ...
257 <!-- *********************************************************************** -->
259 <a name="whatsnew">What's New in LLVM 3.1?</a>
261 <!-- *********************************************************************** -->
265 <p>This release includes a huge number of bug fixes, performance tweaks and
266 minor improvements. Some of the major improvements and new features are
267 listed in this section.</p>
269 <!--=========================================================================-->
271 <a name="majorfeatures">Major New Features</a>
276 <!-- Features that need text if they're finished for 3.1:
280 loop dependence analysis
281 CorrelatedValuePropagation
282 lib/Transforms/IPO/MergeFunctions.cpp => consider for 3.1.
283 Integrated assembler on by default for arm/thumb?
288 Analysis/RegionInfo.h + Dom Frontiers
289 SparseBitVector: used in LiveVar.
290 llvm/lib/Archive - replace with lib object?
293 <p>LLVM 3.1 includes several major changes and big features:</p>
296 <li><a href="../tools/clang/docs/AddressSanitizer.html">AddressSanitizer</a>,
297 a fast memory error detector.</li>
298 <li><a href="CodeGenerator.html#machineinstrbundle">MachineInstr Bundles</a>,
299 Support to model instruction bundling / packing.</li>
300 <li><a href="#armintegratedassembler">ARM Integrated Assembler</a>,
301 A full featured assembler and direct-to-object support for ARM.</li>
302 <li><a href="#blockplacement">Basic Block Placement</a>
303 Probability driven basic block placement.</li>
310 <!--=========================================================================-->
312 <a name="coreimprovements">LLVM IR and Core Improvements</a>
317 <p>LLVM IR has several new features for better support of new targets and that
318 expose new optimization opportunities:</p>
321 <li>IR support for half float</li>
322 <li>IR support for vectors of pointers, including vector GEPs.</li>
323 <li>Module flags have been introduced. They convey information about the
324 module as a whole to LLVM subsystems.</li>
325 <li>Loads can now have range metadata attached to them to describe the
326 possible values being loaded.</li>
327 <li>Inline cost heuristics have been completely overhauled and now closely
328 model constant propagation through call sites, disregard trivially dead
329 code costs, and can model C++ STL iterator patterns.</li>
334 <!--=========================================================================-->
336 <a name="optimizer">Optimizer Improvements</a>
341 <p>In addition to many minor performance tweaks and bug fixes, this
342 release includes a few major enhancements and additions to the
346 <li>The loop unroll pass now is able to unroll loops with run-time trip counts.
347 This feature is turned off by default, and is enabled with the
348 <code>-unroll-runtime</code> flag.</li>
349 <li>A new basic-block autovectorization pass is available. Pass
350 <code>-vectorize</code> to run this pass along with some associated
351 post-vectorization cleanup passes. For more information, see the EuroLLVM
352 2012 slides: <a href="http://llvm.org/devmtg/2012-04-12/Slides/Hal_Finkel.pdf">
353 Autovectorization with LLVM</a>.</li>
359 <!--=========================================================================-->
361 <a name="mc">MC Level Improvements</a>
366 <p>The LLVM Machine Code (aka MC) subsystem was created to solve a number of
367 problems in the realm of assembly, disassembly, object file format handling,
368 and a number of other related areas that CPU instruction-set level tools work
369 in. For more information, please see
370 the <a href="http://blog.llvm.org/2010/04/intro-to-llvm-mc-project.html">Intro
371 to the LLVM MC Project Blog Post</a>.</p>
379 <!--=========================================================================-->
381 <a name="codegen">Target Independent Code Generator Improvements</a>
386 <p>We have changed the way that the Type Legalizer legalizes vectors. The type
387 legalizer now attempts to promote integer elements. This enabled the
388 implementation of vector-select. Additionally, we see a performance boost on
389 workloads which use vectors of chars and shorts, since they are now promoted
390 to 32-bit types, which are better supported by the SIMD instruction set.
391 Floating point types are still widened as before.</p>
394 <p>We have put a significant amount of work into the code generator
395 infrastructure, which allows us to implement more aggressive algorithms and
396 make it run faster:</p>
399 <li>TableGen can now synthesize register classes that are only needed to
400 represent combinations of constraints from instructions and sub-registers.
401 The synthetic register classes inherit most of their properties form their
402 closest user-defined super-class.</li>
403 <li><code>MachineRegisterInfo</code> now allows the reserved registers to be
404 frozen when register allocation starts. Target hooks should use the
405 <code>MRI->canReserveReg(FramePtr)</code> method to avoid accidentally
406 disabling frame pointer elimination during register allocation.</li>
407 <li>A new kind of <code>MachineOperand</code> provides a compact
408 representation of large clobber lists on call instructions. The register
409 mask operand references a bit mask of preserved registers. Everything else
413 <p> We added new TableGen infrastructure to support bundling for
414 Very Long Instruction Word (VLIW) architectures. TableGen can now
415 automatically generate a deterministic finite automaton from a VLIW
416 target's schedule description which can be queried to determine
417 legal groupings of instructions in a bundle.</p>
419 <p> We have added a new target independent VLIW packetizer based on the
420 DFA infrastructure to group machine instructions into bundles.</p>
425 <a name="blockplacement">Basic Block Placement</a>
428 <p>A probability based block placement and code layout algorithm was added to
429 LLVM's code generator. This layout pass supports probabilities derived from
430 static heuristics as well as source code annotations such as
431 <code>__builtin_expect</code>.</p>
434 <!--=========================================================================-->
436 <a name="x86">X86-32 and X86-64 Target Improvements</a>
441 <p>New features and major changes in the X86 target include:</p>
444 <li>Bug fixes and improved support for AVX1</li>
445 <li>Support for AVX2 (still incomplete at this point)</li>
446 <li>Call instructions use the new register mask operands for faster compile
447 times and better support for different calling conventions. The old WINCALL
448 instructions are no longer needed.</li>
449 <li>DW2 Exception Handling is enabled on Cygwin and MinGW.</li>
450 <li>Support for implicit TLS model used with MS VC runtime</li>
455 <!--=========================================================================-->
457 <a name="ARM">ARM Target Improvements</a>
462 <p>New features of the ARM target include:</p>
465 <li>The constant island pass now supports basic block and constant pool entry
466 alignments greater than 4 bytes.</li>
467 <li>On Darwin, the ARM target now has a full-featured integrated assembler.
472 <a name="armintegratedassembler">ARM Integrated Assembler</a>
475 <p>The ARM target now includes a full featured macro assembler, including
476 direct-to-object module support for clang. The assembler is currently enabled
477 by default for Darwin only pending testing and any additional necessary
478 platform specific support for Linux.</p>
480 <p>Full support is included for Thumb1, Thumb2 and ARM modes, along with
481 subtarget and CPU specific extensions for VFP2, VFP3 and NEON.</p>
483 <p>The assembler is Unified Syntax only (see ARM Architecural Reference Manual
484 for details). While there is some, and growing, support for pre-unfied (divided)
485 syntax, there are still significant gaps in that support.</p>
489 <!--=========================================================================-->
491 <a name="MIPS">MIPS Target Improvements</a>
496 <p>This release has seen major new work on just about every aspect of the MIPS
497 backend. Some of the major new features include:</p>
504 <!--=========================================================================-->
506 <a name="OtherTS">Other Target Specific Improvements</a>
511 <p>Support for Qualcomm's Hexagon VLIW processor has been added.</p>
521 <!--=========================================================================-->
523 <a name="changes">Major Changes and Removed Features</a>
528 <p>If you're already an LLVM user or developer with out-of-tree changes based on
529 LLVM 3.1, this section lists some "gotchas" that you may run into upgrading
530 from the previous release.</p>
533 <li>LLVM 3.1 removes support for reading LLVM 2.9 bitcode files. Going
534 forward, we aim for all future versions of LLVM to read bitcode files and
535 <tt>.ll</tt> files produced by LLVM 3.0 and later.</li>
536 <li>The <tt>unwind</tt> instruction is now gone. With the introduction of the
537 new exception handling system in LLVM 3.0, the <tt>unwind</tt> instruction
538 became obsolete.</li>
544 <!--=========================================================================-->
546 <a name="api_changes">Internal API Changes</a>
551 <p>In addition, many APIs have changed in this release. Some of the major
552 LLVM API changes are:</p>
555 <li>Target specific options have been moved from global variables to members
556 on the new <code>TargetOptions</code> class, which is local to each
557 <code>TargetMachine</code>. As a consequence, the associated flags will
558 no longer be accepted by <tt>clang -mllvm</tt>. This includes:
560 <li><code>llvm::PrintMachineCode</code></li>
561 <li><code>llvm::NoFramePointerElim</code></li>
562 <li><code>llvm::NoFramePointerElimNonLeaf</code></li>
563 <li><code>llvm::DisableFramePointerElim(const MachineFunction &)</code></li>
564 <li><code>llvm::LessPreciseFPMADOption</code></li>
565 <li><code>llvm::LessPrecideFPMAD()</code></li>
566 <li><code>llvm::NoExcessFPPrecision</code></li>
567 <li><code>llvm::UnsafeFPMath</code></li>
568 <li><code>llvm::NoInfsFPMath</code></li>
569 <li><code>llvm::NoNaNsFPMath</code></li>
570 <li><code>llvm::HonorSignDependentRoundingFPMathOption</code></li>
571 <li><code>llvm::HonorSignDependentRoundingFPMath()</code></li>
572 <li><code>llvm::UseSoftFloat</code></li>
573 <li><code>llvm::FloatABIType</code></li>
574 <li><code>llvm::NoZerosInBSS</code></li>
575 <li><code>llvm::JITExceptionHandling</code></li>
576 <li><code>llvm::JITEmitDebugInfo</code></li>
577 <li><code>llvm::JITEmitDebugInfoToDisk</code></li>
578 <li><code>llvm::GuaranteedTailCallOpt</code></li>
579 <li><code>llvm::StackAlignmentOverride</code></li>
580 <li><code>llvm::RealignStack</code></li>
581 <li><code>llvm::DisableJumpTables</code></li>
582 <li><code>llvm::EnableFastISel</code></li>
583 <li><code>llvm::getTrapFunctionName()</code></li>
584 <li><code>llvm::EnableSegmentedStacks</code></li>
586 <li>The MDBuilder class has been added to simplify the creation of
593 <!--=========================================================================-->
595 <a name="tools_changes">Tools Changes</a>
600 <p>In addition, some tools have changed in this release. Some of the changes
605 <li>llvm-stress is a command line tool for generating random .ll files to fuzz
606 different LLVM components. </li>
607 <li>llvm-ld has been removed. Use llvm-link or Clang instead.</li>
619 <!-- *********************************************************************** -->
621 <a name="knownproblems">Known Problems</a>
623 <!-- *********************************************************************** -->
627 <p>LLVM is generally a production quality compiler, and is used by a broad range
628 of applications and shipping in many products. That said, not every
629 subsystem is as mature as the aggregate, particularly the more obscure
630 targets. If you run into a problem, please check the <a
631 href="http://llvm.org/bugs/">LLVM bug database</a> and submit a bug if
632 there isn't already one or ask on the <a
633 href="http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev">LLVMdev
636 <p>Known problem areas include:</p>
639 <li>The Alpha, Blackfin, CellSPU, MSP430, PTX, SystemZ and
640 XCore backends are experimental, and the Alpha, Blackfin and SystemZ
641 targets have already been removed from mainline.</li>
643 <li>The integrated assembler, disassembler, and JIT is not supported by
644 several targets. If an integrated assembler is not supported, then a
645 system assembler is required. For more details, see the <a
646 href="CodeGenerator.html#targetfeatures">Target Features Matrix</a>.
649 <li>The C backend has numerous problems and is not being actively maintained.
650 Depending on it for anything serious is not advised.</li>
655 <!-- *********************************************************************** -->
657 <a name="additionalinfo">Additional Information</a>
659 <!-- *********************************************************************** -->
663 <p>A wide variety of additional information is available on
664 the <a href="http://llvm.org/">LLVM web page</a>, in particular in
665 the <a href="http://llvm.org/docs/">documentation</a> section. The web page
666 also contains versions of the API documentation which is up-to-date with the
667 Subversion version of the source code. You can access versions of these
668 documents specific to this release by going into the "<tt>llvm/doc/</tt>"
669 directory in the LLVM tree.</p>
671 <p>If you have any questions or comments about LLVM, please feel free to contact
672 us via the <a href="http://llvm.org/docs/#maillist"> mailing lists</a>.</p>
676 <!-- *********************************************************************** -->
680 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
681 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
682 <a href="http://validator.w3.org/check/referer"><img
683 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
685 <a href="http://llvm.org/">LLVM Compiler Infrastructure</a><br>
686 Last modified: $Date$