blob: 113ce930ef02d6f2e47ec154bb3a20227071329b [file] [log] [blame]
Misha Brukman500bc302003-11-22 00:38:41 +00001<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
3<html>
4<head>
Reid Spencer6454ed32004-11-18 18:38:58 +00005 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
Daniel Dunbaradea4972012-04-19 20:20:34 +00006 <link rel="stylesheet" href="_static/llvm.css" type="text/css">
Chris Lattner21e1b7a2011-12-13 17:55:30 +00007 <title>LLVM 3.1 Release Notes</title>
Misha Brukman500bc302003-11-22 00:38:41 +00008</head>
9<body>
Chris Lattner79c3fe12003-10-02 04:57:28 +000010
Chris Lattner21e1b7a2011-12-13 17:55:30 +000011<h1>LLVM 3.1 Release Notes</h1>
Mikhail Glushenkovea65d7d2008-10-13 02:08:34 +000012
Jakub Staszakf8de54f2011-12-06 23:33:07 +000013<div>
14<img style="float:right" src="http://llvm.org/img/DragonSmall.png"
15 width="136" height="136" alt="LLVM Dragon Logo">
16</div>
Chris Lattner0e464a92010-03-17 04:02:39 +000017
Chris Lattner79c3fe12003-10-02 04:57:28 +000018<ol>
Misha Brukman500bc302003-11-22 00:38:41 +000019 <li><a href="#intro">Introduction</a></li>
Chris Lattner96a445e2008-10-13 18:01:01 +000020 <li><a href="#subproj">Sub-project Status Update</a></li>
Chris Lattner21e1b7a2011-12-13 17:55:30 +000021 <li><a href="#externalproj">External Projects Using LLVM 3.1</a></li>
22 <li><a href="#whatsnew">What's New in LLVM?</a></li>
Chris Lattner4b538b92004-04-30 22:17:12 +000023 <li><a href="GettingStarted.html">Installation Instructions</a></li>
Dan Gohman44aa9212008-10-14 16:23:02 +000024 <li><a href="#knownproblems">Known Problems</a></li>
Misha Brukman500bc302003-11-22 00:38:41 +000025 <li><a href="#additionalinfo">Additional Information</a></li>
Chris Lattner79c3fe12003-10-02 04:57:28 +000026</ol>
27
Chris Lattner7911ce22004-05-23 21:07:27 +000028<div class="doc_author">
NAKAMURA Takumib9a33632011-04-09 02:13:37 +000029 <p>Written by the <a href="http://llvm.org/">LLVM Team</a></p>
Misha Brukman500bc302003-11-22 00:38:41 +000030</div>
Chris Lattner79c3fe12003-10-02 04:57:28 +000031
Chris Lattner21e1b7a2011-12-13 17:55:30 +000032<h1 style="color:red">These are in-progress notes for the upcoming LLVM 3.1
Jeffrey Yasskinbec48772010-01-28 01:14:43 +000033release.<br>
34You may prefer the
Chris Lattner21e1b7a2011-12-13 17:55:30 +000035<a href="http://llvm.org/releases/3.0/docs/ReleaseNotes.html">LLVM 3.0
Dan Gohmanb44f6c62010-05-03 23:51:05 +000036Release Notes</a>.</h1>
Jeffrey Yasskinbec48772010-01-28 01:14:43 +000037
Chris Lattner79c3fe12003-10-02 04:57:28 +000038<!-- *********************************************************************** -->
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +000039<h2>
Misha Brukman500bc302003-11-22 00:38:41 +000040 <a name="intro">Introduction</a>
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +000041</h2>
Chris Lattner79c3fe12003-10-02 04:57:28 +000042<!-- *********************************************************************** -->
43
NAKAMURA Takumi074eeaa2011-04-21 01:52:00 +000044<div>
Misha Brukman500bc302003-11-22 00:38:41 +000045
Chris Lattnerf6662f92008-10-13 17:57:36 +000046<p>This document contains the release notes for the LLVM Compiler
Chris Lattner21e1b7a2011-12-13 17:55:30 +000047 Infrastructure, release 3.1. Here we describe the status of LLVM, including
Chris Lattner1ab8ce92011-11-27 18:47:37 +000048 major improvements from the previous release, improvements in various
49 subprojects of LLVM, and some of the current users of the code.
Bill Wendling7b7fa742011-10-26 18:46:16 +000050 All LLVM releases may be downloaded from
51 the <a href="http://llvm.org/releases/">LLVM releases web site</a>.</p>
Chris Lattner19092612003-10-02 16:38:05 +000052
Chris Lattner7506b1d2004-12-07 08:04:13 +000053<p>For more information about LLVM, including information about the latest
Bill Wendling7b7fa742011-10-26 18:46:16 +000054 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>
Chris Lattner79c3fe12003-10-02 04:57:28 +000058
Bill Wendling7b7fa742011-10-26 18:46:16 +000059<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>
Misha Brukman500bc302003-11-22 00:38:41 +000063
64</div>
Michael J. Spencer60f790c2011-11-28 18:20:09 +000065
66
Chris Lattner96a445e2008-10-13 18:01:01 +000067<!-- *********************************************************************** -->
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +000068<h2>
Chris Lattner96a445e2008-10-13 18:01:01 +000069 <a name="subproj">Sub-project Status Update</a>
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +000070</h2>
Chris Lattner96a445e2008-10-13 18:01:01 +000071<!-- *********************************************************************** -->
Chris Lattnerea34f642008-06-08 21:34:41 +000072
NAKAMURA Takumi074eeaa2011-04-21 01:52:00 +000073<div>
Bill Wendling7b7fa742011-10-26 18:46:16 +000074
Chris Lattner21e1b7a2011-12-13 17:55:30 +000075<p>The LLVM 3.1 distribution currently consists of code from the core LLVM
Bill Wendling7b7fa742011-10-26 18:46:16 +000076 repository (which roughly includes the LLVM optimizers, code generators and
Bill Wendling1fefd092012-05-14 08:11:53 +000077 supporting tools), and the Clang repository. In addition to this code, the
78 LLVM Project includes other sub-projects that are in development. Here we
79 include updates on these subprojects.</p>
Chris Lattner96a445e2008-10-13 18:01:01 +000080
Chris Lattner8348b472008-06-08 21:58:17 +000081<!--=========================================================================-->
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +000082<h3>
Chris Lattnerfb97b2d2008-10-13 18:11:54 +000083<a name="clang">Clang: C/C++/Objective-C Frontend Toolkit</a>
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +000084</h3>
Chris Lattner8348b472008-06-08 21:58:17 +000085
NAKAMURA Takumi074eeaa2011-04-21 01:52:00 +000086<div>
Chris Lattner8348b472008-06-08 21:58:17 +000087
Chris Lattner095539f2010-04-26 17:42:18 +000088<p><a href="http://clang.llvm.org/">Clang</a> is an LLVM front end for the C,
Bill Wendling7b7fa742011-10-26 18:46:16 +000089 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
Chris Lattner1ab8ce92011-11-27 18:47:37 +000095 (32- and 64-bit), and for Darwin/ARM targets.</p>
Chris Lattner61358ab2009-10-13 17:48:04 +000096
Benjamin Kramer88c8e692012-05-13 10:40:08 +000097<p>In the LLVM 3.1 time-frame, the Clang team has made many improvements.
98 Highlights include:</p>
Douglas Gregorba087df2011-10-15 00:48:01 +000099<ul>
Benjamin Kramer88c8e692012-05-13 10:40:08 +0000100 <li>Greatly expanded <a href="http://clang.llvm.org/cxx_status.html">C++11
101 support</a> including lambdas, initializer lists, constexpr, user-defined
102 literals, and atomics.</li>
103 <li>A new <a href="http://clang.llvm.org/docs/Tooling.html">tooling</a>
104 library to ease building of clang-based standalone tools.</li>
105 <li>Extended support for
106 <a href="http://clang.llvm.org/docs/ObjectiveCLiterals.html">literals in
107 Objective C</a>.</li>
Douglas Gregorba087df2011-10-15 00:48:01 +0000108</ul>
Chris Lattner21e1b7a2011-12-13 17:55:30 +0000109
Bill Wendling1fefd092012-05-14 08:11:53 +0000110<p>For more details about the changes to Clang since the 3.0 release, see the
111 <a href="http://clang.llvm.org/docs/ReleaseNotes.html">Clang release
112 notes.</a></p>
Michael J. Spencer60f790c2011-11-28 18:20:09 +0000113
Duncan Sandsf3ba7af2011-04-06 08:07:40 +0000114<p>If Clang rejects your code but another compiler accepts it, please take a
Bill Wendling7b7fa742011-10-26 18:46:16 +0000115 look at the <a href="http://clang.llvm.org/compatibility.html">language
116 compatibility</a> guide to make sure this is not intentional or a known
117 issue.</p>
Bill Wendling741748a2008-10-27 09:27:33 +0000118
Chris Lattnerfb97b2d2008-10-13 18:11:54 +0000119</div>
120
121<!--=========================================================================-->
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000122<h3>
Duncan Sands528a5102011-04-04 11:09:08 +0000123<a name="dragonegg">DragonEgg: GCC front-ends, LLVM back-end</a>
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000124</h3>
Chris Lattner61358ab2009-10-13 17:48:04 +0000125
NAKAMURA Takumi074eeaa2011-04-21 01:52:00 +0000126<div>
Bill Wendling1fefd092012-05-14 08:11:53 +0000127
Bill Wendling7b7fa742011-10-26 18:46:16 +0000128<p><a href="http://dragonegg.llvm.org/">DragonEgg</a> is a
129 <a href="http://gcc.gnu.org/wiki/plugins">gcc plugin</a> that replaces GCC's
Duncan Sands5abd10a2012-05-11 19:59:43 +0000130 optimizers and code generators with LLVM's. It works with gcc-4.5 and gcc-4.6
131 (and partially with gcc-4.7), can target the x86-32/x86-64 and ARM processor
132 families, and has been successfully used on the Darwin, FreeBSD, KFreeBSD,
133 Linux and OpenBSD platforms. It fully supports Ada, C, C++ and Fortran. It
134 has partial support for Go, Java, Obj-C and Obj-C++.</p>
Duncan Sands749fd832010-04-02 09:23:15 +0000135
Chris Lattner21e1b7a2011-12-13 17:55:30 +0000136<p>The 3.1 release has the following notable changes:</p>
Bill Wendling7b7fa742011-10-26 18:46:16 +0000137
Bill Wendling1fefd092012-05-14 08:11:53 +0000138<ul>
Duncan Sands5abd10a2012-05-11 19:59:43 +0000139 <li>Partial support for gcc-4.7. Ada support is poor, but other languages work
140 fairly well.</li>
141
142 <li>Support for ARM processors. Some essential gcc headers that are needed to
143 build DragonEgg for ARM are not installed by gcc. To work around this,
144 copy the missing headers from the gcc source tree.</li>
145
146 <li>Better optimization for Fortran by exploiting the fact that Fortran scalar
147 arguments have 'restrict' semantics.</li>
148
149 <li>Better optimization for all languages by passing information about type
150 aliasing and type ranges to the LLVM optimizers.</li>
151
152 <li>A regression test-suite was added.</li>
Duncan Sands4b1da2b2010-09-30 17:37:34 +0000153</ul>
Chris Lattner61358ab2009-10-13 17:48:04 +0000154
155</div>
156
Chris Lattner75547712010-10-03 23:49:06 +0000157<!--=========================================================================-->
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000158<h3>
Chris Lattner75547712010-10-03 23:49:06 +0000159<a name="compiler-rt">compiler-rt: Compiler Runtime Library</a>
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000160</h3>
Chris Lattner75547712010-10-03 23:49:06 +0000161
NAKAMURA Takumi074eeaa2011-04-21 01:52:00 +0000162<div>
Bill Wendling7b7fa742011-10-26 18:46:16 +0000163
164<p>The new LLVM <a href="http://compiler-rt.llvm.org/">compiler-rt project</a>
165 is a simple library that provides an implementation of the low-level
166 target-specific hooks required by code generation and other runtime
167 components. For example, when compiling for a 32-bit target, converting a
168 double to a 64-bit unsigned integer is compiled into a runtime call to the
169 "__fixunsdfdi" function. The compiler-rt library provides highly optimized
170 implementations of this and other low-level routines (some are 3x faster than
171 the equivalent libgcc routines).</p>
Chris Lattner75547712010-10-03 23:49:06 +0000172
David Chisnall6dd3bdb2012-05-15 13:06:46 +0000173<p>As of 3.1, compiler-rt includes the helper functions for atomic operations,
174 allowing atomic operations on arbitrary-sized quantities to work. These
175 functions follow the specification defined by gcc and are used by clang.</p>
176
Chris Lattner75547712010-10-03 23:49:06 +0000177</div>
Chris Lattner61358ab2009-10-13 17:48:04 +0000178
179<!--=========================================================================-->
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000180<h3>
Chris Lattnere07043c2010-09-29 05:30:03 +0000181<a name="lldb">LLDB: Low Level Debugger</a>
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000182</h3>
Chris Lattnere07043c2010-09-29 05:30:03 +0000183
NAKAMURA Takumi074eeaa2011-04-21 01:52:00 +0000184<div>
Michael J. Spencer60f790c2011-11-28 18:20:09 +0000185
Benjamin Kramerabe446c2012-05-13 10:21:51 +0000186<p><a href="http://lldb.llvm.org">LLDB</a> is a ground-up implementation of a
187 command line debugger, as well as a debugger API that can be used from other
188 applications. LLDB makes use of the Clang parser to provide high-fidelity
189 expression parsing (particularly for C++) and uses the LLVM JIT for target
190 support.</p>
Chris Lattnere07043c2010-09-29 05:30:03 +0000191
Chris Lattnere07043c2010-09-29 05:30:03 +0000192</div>
193
194<!--=========================================================================-->
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000195<h3>
Chris Lattnere07043c2010-09-29 05:30:03 +0000196<a name="libc++">libc++: C++ Standard Library</a>
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000197</h3>
Chris Lattnere07043c2010-09-29 05:30:03 +0000198
NAKAMURA Takumi074eeaa2011-04-21 01:52:00 +0000199<div>
Chris Lattnere07043c2010-09-29 05:30:03 +0000200
Bill Wendling7b7fa742011-10-26 18:46:16 +0000201<p>Like compiler_rt, libc++ is now <a href="DeveloperPolicy.html#license">dual
202 licensed</a> under the MIT and UIUC license, allowing it to be used more
203 permissively.</p>
Chris Lattnere07043c2010-09-29 05:30:03 +0000204
David Chisnall6dd3bdb2012-05-15 13:06:46 +0000205<p>Within the LLVM 3.1 time-frame there were the following highlights:</p>
206
207<ul>
208 <li>The <code>&lt;atomic&gt;</code> header is now passing all tests, when
209 compiling with clang and linking against the support code from
210 compiler-rt.</li>
211 <li>FreeBSD now includes libc++ as part of the base system.</li>
212 <li>libc++ has been ported to Solaris and, in combination with libcxxrt and
213 clang, is working with a large body of existing code.</li>
214</ul>
215
Chris Lattnere07043c2010-09-29 05:30:03 +0000216</div>
217
Chris Lattner3d6a80a2011-04-07 03:08:22 +0000218<!--=========================================================================-->
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000219<h3>
Chris Lattner3d6a80a2011-04-07 03:08:22 +0000220<a name="vmkit">VMKit</a>
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000221</h3>
Chris Lattner3d6a80a2011-04-07 03:08:22 +0000222
NAKAMURA Takumi074eeaa2011-04-21 01:52:00 +0000223<div>
Bill Wendling7b7fa742011-10-26 18:46:16 +0000224
Bill Wendling1fefd092012-05-14 08:11:53 +0000225<p>The <a href="http://vmkit.llvm.org/">VMKit project</a> is an implementation
226 of a Java Virtual Machine (Java VM or JVM) that uses LLVM for static and
227 just-in-time compilation.</p>
Nicolas Geoffray54d5df92011-11-10 23:37:56 +0000228
Bill Wendling1fefd092012-05-14 08:11:53 +0000229<p>In the LLVM 3.1 time-frame, VMKit has had significant improvements on both
230 runtime and startup performance.</p>
Bill Wendling7b7fa742011-10-26 18:46:16 +0000231
Chris Lattner3d6a80a2011-04-07 03:08:22 +0000232</div>
Michael J. Spencer60f790c2011-11-28 18:20:09 +0000233
Tobias Grosser76213b82012-04-16 17:17:00 +0000234
235<!--=========================================================================-->
236<h3>
Tobias Grosser99a0b282012-04-16 17:18:49 +0000237<a name="Polly">Polly: Polyhedral Optimizer</a>
Tobias Grosser76213b82012-04-16 17:17:00 +0000238</h3>
239
240<div>
241
Bill Wendling1fefd092012-05-14 08:11:53 +0000242<p><a href="http://polly.llvm.org/">Polly</a> is an <em>experimental</em>
Tobias Grosser76213b82012-04-16 17:17:00 +0000243 optimizer for data locality and parallelism. It currently provides high-level
244 loop optimizations and automatic parallelisation (using the OpenMP run time).
245 Work in the area of automatic SIMD and accelerator code generation was
Bill Wendling1fefd092012-05-14 08:11:53 +0000246 started.</p>
Tobias Grosser76213b82012-04-16 17:17:00 +0000247
Bill Wendling1fefd092012-05-14 08:11:53 +0000248<p>Within the LLVM 3.1 time-frame there were the following highlights:</p>
Tobias Grosser76213b82012-04-16 17:17:00 +0000249
Bill Wendling1fefd092012-05-14 08:11:53 +0000250<ul>
Tobias Grosser76213b82012-04-16 17:17:00 +0000251 <li>Polly became an official LLVM project</li>
Bill Wendling1fefd092012-05-14 08:11:53 +0000252 <li>Polly can be loaded directly into clang (enabled by '-O3 -mllvm -polly')</li>
253 <li>An automatic scheduling optimizer (derived
254 from <a href="http://pluto-compiler.sourceforge.net/">Pluto</a>) was
255 integrated. It performs loop transformations to optimize for data-locality
256 and parallelism. The transformations include, but are not limited to
257 interchange, fusion, fission, skewing and tiling.</li>
258</ul>
Tobias Grosser76213b82012-04-16 17:17:00 +0000259
260</div>
261
NAKAMURA Takumi074eeaa2011-04-21 01:52:00 +0000262</div>
Daniel Dunbar97b01a82010-10-04 17:39:47 +0000263
Chris Lattnerab68e9e2009-02-26 22:33:38 +0000264<!-- *********************************************************************** -->
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000265<h2>
Chris Lattner21e1b7a2011-12-13 17:55:30 +0000266 <a name="externalproj">External Open Source Projects Using LLVM 3.1</a>
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000267</h2>
Chris Lattnerab68e9e2009-02-26 22:33:38 +0000268<!-- *********************************************************************** -->
269
NAKAMURA Takumi074eeaa2011-04-21 01:52:00 +0000270<div>
Chris Lattner61358ab2009-10-13 17:48:04 +0000271
272<p>An exciting aspect of LLVM is that it is used as an enabling technology for
273 a lot of other language and tools projects. This section lists some of the
Chris Lattner21e1b7a2011-12-13 17:55:30 +0000274 projects that have already been updated to work with LLVM 3.1.</p>
Chris Lattner75547712010-10-03 23:49:06 +0000275
Bill Wendlingf8cfe2e2012-05-13 09:52:48 +0000276<h3>FAUST</h3>
277
278<div>
279
Benjamin Kramerabe446c2012-05-13 10:21:51 +0000280<p><a href="http://faust.grame.fr/">FAUST</a> is a compiled language for
281 real-time audio signal processing. The name FAUST stands for Functional
282 AUdio STream. Its programming model combines two approaches: functional
283 programming and block diagram composition. In addition with the C, C++, Java,
284 JavaScript output formats, the Faust compiler can generate LLVM bitcode, and
285 works with LLVM 2.7-3.1.</p>
Bill Wendlingf8cfe2e2012-05-13 09:52:48 +0000286
287</div>
288
Bill Wendlingde86bea2012-05-11 21:42:37 +0000289<h3>Glasgow Haskell Compiler (GHC)</h3>
290
291<div>
292
Benjamin Kramerabe446c2012-05-13 10:21:51 +0000293<p><a href="http://www.haskell.org/ghc/">GHC</a> is an open source compiler and
294 programming suite for Haskell, a lazy functional programming language. It
295 includes an optimizing static compiler generating good code for a variety of
296 platforms, together with an interactive system for convenient, quick
297 development.</p>
Bill Wendlingde86bea2012-05-11 21:42:37 +0000298
299<p>GHC 7.0 and onwards include an LLVM code generator, supporting LLVM 2.8 and
300 later.</p>
301
302</div>
303
Bill Wendlinge9a73bb2012-05-14 06:23:51 +0000304<h3>Julia</h3>
305
306<div>
307
308<p><a href="https://github.com/JuliaLang/julia">Julia</a> is a high-level,
309 high-performance dynamic language for technical computing. It provides a
310 sophisticated compiler, distributed parallel execution, numerical accuracy,
311 and an extensive mathematical function library. The compiler uses type
312 inference to generate fast code without any type declarations, and uses
313 LLVM's optimization passes and JIT compiler. The
314 <a href="http://julialang.org/"> Julia Language</a> is designed
315 around multiple dispatch, giving programs a large degree of flexibility. It
316 is ready for use on many kinds of problems.</p>
317
318</div>
319
Bill Wendlingf2f92ba2012-05-13 10:00:58 +0000320<h3>LLVM D Compiler</h3>
321
322<div>
323
324<p><a href="https://github.com/ldc-developers/ldc">LLVM D Compiler</a> (LDC) is
325 a compiler for the D programming Language. It is based on the DMD frontend
326 and uses LLVM as backend.</p>
327
328</div>
329
Bill Wendling91071052012-05-13 09:55:24 +0000330<h3>Open Shading Language</h3>
331
332<div>
333
Benjamin Kramerabe446c2012-05-13 10:21:51 +0000334<p><a href="https://github.com/imageworks/OpenShadingLanguage/">Open Shading
335 Language (OSL)</a> is a small but rich language for programmable shading in
336 advanced global illumination renderers and other applications, ideal for
337 describing materials, lights, displacement, and pattern generation. It uses
338 LLVM to JIT complex shader networks to x86 code at runtime.</p>
Bill Wendling91071052012-05-13 09:55:24 +0000339
340<p>OSL was developed by Sony Pictures Imageworks for use in its in-house
341 renderer used for feature film animation and visual effects, and is
Benjamin Kramerabe446c2012-05-13 10:21:51 +0000342 distributed as open source software with the "New BSD" license.</p>
Bill Wendling91071052012-05-13 09:55:24 +0000343
344</div>
345
Bill Wendling59a23e32012-05-13 09:59:27 +0000346<h3>Portable OpenCL (pocl)</h3>
347
348<div>
349
350<p>In addition to producing an easily portable open source OpenCL
Benjamin Kramerabe446c2012-05-13 10:21:51 +0000351 implementation, another major goal of <a href="http://pocl.sourceforge.net/">
352 pocl</a> is improving performance portability of OpenCL programs with
353 compiler optimizations, reducing the need for target-dependent manual
354 optimizations. An important part of pocl is a set of LLVM passes used to
355 statically parallelize multiple work-items with the kernel compiler, even in
356 the presence of work-group barriers. This enables static parallelization of
357 the fine-grained static concurrency in the work groups in multiple ways
358 (SIMD, VLIW, superscalar,...).</p>
Bill Wendling59a23e32012-05-13 09:59:27 +0000359
360</div>
361
Bill Wendlingde86bea2012-05-11 21:42:37 +0000362<h3>Pure</h3>
363
364<div>
365
Bill Wendling015d2ad2012-05-11 22:38:33 +0000366<p><a href="http://pure-lang.googlecode.com/">Pure</a> is an
367 algebraic/functional programming language based on term rewriting. Programs
368 are collections of equations which are used to evaluate expressions in a
369 symbolic fashion. The interpreter uses LLVM as a backend to JIT-compile Pure
370 programs to fast native code. Pure offers dynamic typing, eager and lazy
371 evaluation, lexical closures, a hygienic macro system (also based on term
372 rewriting), built-in list and matrix support (including list and matrix
373 comprehensions) and an easy-to-use interface to C and other programming
374 languages (including the ability to load LLVM bitcode modules, and inline C,
375 C++, Fortran and Faust code in Pure programs if the corresponding
376 LLVM-enabled compilers are installed).</p>
Bill Wendlingde86bea2012-05-11 21:42:37 +0000377
378<p>Pure version 0.54 has been tested and is known to work with LLVM 3.1 (and
379 continues to work with older LLVM releases >= 2.5).</p>
380
381</div>
Bill Wendling644ce532011-10-26 09:25:01 +0000382
Bill Wendling59a23e32012-05-13 09:59:27 +0000383<h3>TTA-based Co-design Environment (TCE)</h3>
384
385<div>
386
Benjamin Kramerabe446c2012-05-13 10:21:51 +0000387<p><a href="http://tce.cs.tut.fi/">TCE</a> is a toolset for designing
388 application-specific processors (ASP) based on the Transport triggered
389 architecture (TTA). The toolset provides a complete co-design flow from C/C++
390 programs down to synthesizable VHDL/Verilog and parallel program binaries.
391 Processor customization points include the register files, function units,
392 supported operations, and the interconnection network.</p>
Bill Wendling59a23e32012-05-13 09:59:27 +0000393
394<p>TCE uses Clang and LLVM for C/C++ language support, target independent
395 optimizations and also for parts of code generation. It generates new
396 LLVM-based code generators "on the fly" for the designed TTA processors and
397 loads them in to the compiler backend as runtime libraries to avoid
398 per-target recompilation of larger parts of the compiler chain.</p>
399
400</div>
401
NAKAMURA Takumi074eeaa2011-04-21 01:52:00 +0000402</div>
403
Chris Lattner8348b472008-06-08 21:58:17 +0000404<!-- *********************************************************************** -->
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000405<h2>
Chris Lattner21e1b7a2011-12-13 17:55:30 +0000406 <a name="whatsnew">What's New in LLVM 3.1?</a>
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000407</h2>
Chris Lattner8348b472008-06-08 21:58:17 +0000408<!-- *********************************************************************** -->
409
NAKAMURA Takumi074eeaa2011-04-21 01:52:00 +0000410<div>
Chris Lattnerf8e0b4e2008-06-08 22:59:35 +0000411
Chris Lattner61358ab2009-10-13 17:48:04 +0000412<p>This release includes a huge number of bug fixes, performance tweaks and
Bill Wendling7b7fa742011-10-26 18:46:16 +0000413 minor improvements. Some of the major improvements and new features are
414 listed in this section.</p>
Chris Lattner61358ab2009-10-13 17:48:04 +0000415
Chris Lattner914ce462010-04-22 06:28:20 +0000416<!--=========================================================================-->
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000417<h3>
Chris Lattner252b83d2008-02-06 18:00:06 +0000418<a name="majorfeatures">Major New Features</a>
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000419</h3>
Chris Lattner252b83d2008-02-06 18:00:06 +0000420
NAKAMURA Takumi074eeaa2011-04-21 01:52:00 +0000421<div>
Chris Lattner252b83d2008-02-06 18:00:06 +0000422
Chris Lattner1ab8ce92011-11-27 18:47:37 +0000423 <!-- Features that need text if they're finished for 3.1:
424 ARM EHABI
425 combiner-aa?
426 strong phi elim
427 loop dependence analysis
428 CorrelatedValuePropagation
429 lib/Transforms/IPO/MergeFunctions.cpp => consider for 3.1.
Chris Lattner1c80fbf2011-11-27 20:51:47 +0000430 Integrated assembler on by default for arm/thumb?
431
Chris Lattner1ab8ce92011-11-27 18:47:37 +0000432 -->
433
Chris Lattner1c80fbf2011-11-27 20:51:47 +0000434 <!-- Near dead:
Chris Lattnerdec23b62011-11-15 22:13:27 +0000435 Analysis/RegionInfo.h + Dom Frontiers
436 SparseBitVector: used in LiveVar.
Chris Lattner5a1731d2011-11-27 08:32:32 +0000437 llvm/lib/Archive - replace with lib object?
Chris Lattner1c80fbf2011-11-27 20:51:47 +0000438 -->
Chris Lattner6a007d12011-11-25 20:33:27 +0000439
Chris Lattner21e1b7a2011-12-13 17:55:30 +0000440<p>LLVM 3.1 includes several major changes and big features:</p>
Bill Wendlingbc5f6dd2011-10-26 18:33:01 +0000441
Chris Lattner791f77b2008-06-05 06:25:56 +0000442<ul>
Evan Cheng28681862011-12-14 22:57:45 +0000443 <li><a href="../tools/clang/docs/AddressSanitizer.html">AddressSanitizer</a>,
444 a fast memory error detector.</li>
445 <li><a href="CodeGenerator.html#machineinstrbundle">MachineInstr Bundles</a>,
446 Support to model instruction bundling / packing.</li>
Jim Grosbach6e6b8222012-02-23 23:52:06 +0000447 <li><a href="#armintegratedassembler">ARM Integrated Assembler</a>,
448 A full featured assembler and direct-to-object support for ARM.</li>
Chandler Carruth268fde42012-04-17 01:10:35 +0000449 <li><a href="#blockplacement">Basic Block Placement</a>
450 Probability driven basic block placement.</li>
Chris Lattnerf304ffc2008-02-10 08:17:19 +0000451</ul>
Mikhail Glushenkovea65d7d2008-10-13 02:08:34 +0000452
Bill Wendlingbc5f6dd2011-10-26 18:33:01 +0000453</div>
454
Michael J. Spencer60f790c2011-11-28 18:20:09 +0000455
Chris Lattner4f0fe432011-11-27 19:26:30 +0000456<!--=========================================================================-->
457<h3>
458<a name="coreimprovements">LLVM IR and Core Improvements</a>
459</h3>
460
461<div>
462
463<p>LLVM IR has several new features for better support of new targets and that
464 expose new optimization opportunities:</p>
465
Bill Wendling1fefd092012-05-14 08:11:53 +0000466<ul>
467 <li>A new type representing 16 bit <i>half</i> floating point values has
468 been added.</li>
469 <li>IR now supports vectors of pointers, including vector GEPs.</li>
470 <li>Module flags have been introduced. They convey information about the
471 module as a whole to LLVM subsystems. This is currently used to encode
472 Objective C ABI information.</li>
473 <li>Loads can now have range metadata attached to them to describe the
474 possible values being loaded.</li>
475 <li>The <tt>llvm.ctlz</tt> and <tt>llvm.cttz</tt> intrinsics now have an
476 additional argument which indicates whether the behavior of the intrinsic
477 is undefined on a zero input. This can be used to generate more efficient
478 code on platforms that only have instructions which don't return the type
479 size when counting bits in 0.</li>
480</ul>
481
Andrew Trick5aab6382011-11-06 17:59:24 +0000482</div>
483
484<!--=========================================================================-->
485<h3>
Chris Lattneracce85d2008-02-10 07:46:44 +0000486<a name="optimizer">Optimizer Improvements</a>
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000487</h3>
Chris Lattneracce85d2008-02-10 07:46:44 +0000488
NAKAMURA Takumi074eeaa2011-04-21 01:52:00 +0000489<div>
Chris Lattneracce85d2008-02-10 07:46:44 +0000490
Chris Lattner064caf92011-11-27 21:30:28 +0000491<p>In addition to many minor performance tweaks and bug fixes, this
Bill Wendling7b7fa742011-10-26 18:46:16 +0000492 release includes a few major enhancements and additions to the
493 optimizers:</p>
Chris Lattneracce85d2008-02-10 07:46:44 +0000494
495<ul>
Brendon Cahoonff730392012-04-14 16:54:12 +0000496 <li>The loop unroll pass now is able to unroll loops with run-time trip counts.
497 This feature is turned off by default, and is enabled with the
498 <code>-unroll-runtime</code> flag.</li>
Hal Finkel9068bf52012-04-16 03:49:43 +0000499 <li>A new basic-block autovectorization pass is available. Pass
500 <code>-vectorize</code> to run this pass along with some associated
Hal Finkel12c10b32012-04-16 17:06:49 +0000501 post-vectorization cleanup passes. For more information, see the EuroLLVM
502 2012 slides: <a href="http://llvm.org/devmtg/2012-04-12/Slides/Hal_Finkel.pdf">
503 Autovectorization with LLVM</a>.</li>
Benjamin Kramerf26b4152012-05-13 12:01:16 +0000504 <li>Inline cost heuristics have been completely overhauled and now closely
505 model constant propagation through call sites, disregard trivially dead
506 code costs, and can model C++ STL iterator patterns.</li>
Chris Lattner11b66112010-10-04 02:42:39 +0000507</ul>
508
Chris Lattner61358ab2009-10-13 17:48:04 +0000509</div>
Chris Lattner917cc712009-03-02 02:37:32 +0000510
Chris Lattner0b832202008-06-08 02:45:07 +0000511<!--=========================================================================-->
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000512<h3>
Chris Lattner4ba2b652010-09-30 16:31:33 +0000513<a name="mc">MC Level Improvements</a>
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000514</h3>
Chris Lattner4ba2b652010-09-30 16:31:33 +0000515
NAKAMURA Takumi074eeaa2011-04-21 01:52:00 +0000516<div>
Bill Wendling7b7fa742011-10-26 18:46:16 +0000517
518<p>The LLVM Machine Code (aka MC) subsystem was created to solve a number of
519 problems in the realm of assembly, disassembly, object file format handling,
520 and a number of other related areas that CPU instruction-set level tools work
Chris Lattner2f206022011-11-27 22:03:34 +0000521 in. For more information, please see
522 the <a href="http://blog.llvm.org/2010/04/intro-to-llvm-mc-project.html">Intro
523 to the LLVM MC Project Blog Post</a>.</p>
Chris Lattner7d9b6b42010-10-02 21:59:30 +0000524
Chris Lattner7d9b6b42010-10-02 21:59:30 +0000525<ul>
Benjamin Kramerf26b4152012-05-13 12:01:16 +0000526 <li>The integrated assembler can optionally emit debug information when
527 assembling a </tt>.s</tt> file. It can be enabled by passing the
528 <tt>-g</tt> option to <tt>llvm-mc</tt>.</li>
Chris Lattner7d9b6b42010-10-02 21:59:30 +0000529</ul>
530
NAKAMURA Takumi45c435a2011-04-05 08:24:22 +0000531</div>
Chris Lattner4ba2b652010-09-30 16:31:33 +0000532
Chris Lattner4ba2b652010-09-30 16:31:33 +0000533<!--=========================================================================-->
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000534<h3>
Chris Lattner511433e2009-03-02 03:24:11 +0000535<a name="codegen">Target Independent Code Generator Improvements</a>
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000536</h3>
Chris Lattner0b832202008-06-08 02:45:07 +0000537
NAKAMURA Takumi074eeaa2011-04-21 01:52:00 +0000538<div>
Chris Lattner0b832202008-06-08 02:45:07 +0000539
Bill Wendling190ec9b2012-02-16 10:23:43 +0000540<p>We have changed the way that the Type Legalizer legalizes vectors. The type
541 legalizer now attempts to promote integer elements. This enabled the
542 implementation of vector-select. Additionally, we see a performance boost on
543 workloads which use vectors of chars and shorts, since they are now promoted
544 to 32-bit types, which are better supported by the SIMD instruction set.
545 Floating point types are still widened as before.</p>
Nadav Rotem75597662011-12-20 08:02:50 +0000546
547
Mikhail Glushenkovf795ef02009-03-01 18:09:47 +0000548<p>We have put a significant amount of work into the code generator
Bill Wendling7b7fa742011-10-26 18:46:16 +0000549 infrastructure, which allows us to implement more aggressive algorithms and
550 make it run faster:</p>
Chris Lattner0b832202008-06-08 02:45:07 +0000551
552<ul>
Jakob Stoklund Olesen9897c622011-12-19 16:53:40 +0000553 <li>TableGen can now synthesize register classes that are only needed to
Bill Wendling190ec9b2012-02-16 10:23:43 +0000554 represent combinations of constraints from instructions and sub-registers.
555 The synthetic register classes inherit most of their properties form their
556 closest user-defined super-class.</li>
Jakob Stoklund Olesend9e5c762012-01-05 00:26:49 +0000557 <li><code>MachineRegisterInfo</code> now allows the reserved registers to be
Bill Wendling190ec9b2012-02-16 10:23:43 +0000558 frozen when register allocation starts. Target hooks should use the
Nadav Rotemfdc309c2012-02-26 08:35:53 +0000559 <code>MRI-&gt;canReserveReg(FramePtr)</code> method to avoid accidentally
Bill Wendling190ec9b2012-02-16 10:23:43 +0000560 disabling frame pointer elimination during register allocation.</li>
Jakob Stoklund Olesen7739cad2012-01-16 19:22:00 +0000561 <li>A new kind of <code>MachineOperand</code> provides a compact
Bill Wendling190ec9b2012-02-16 10:23:43 +0000562 representation of large clobber lists on call instructions. The register
563 mask operand references a bit mask of preserved registers. Everything else
564 is clobbered.</li>
Benjamin Kramer3a2d5652012-05-13 13:10:35 +0000565 <li>The DWARF debug info writer gained support for emitting data for the
566 <a href="SourceLevelDebugging.html#acceltable">name accelerator tables
567 DWARF extension</a>. It is used by LLDB to speed up name lookup.</li>
Chris Lattner84977642007-09-21 03:54:09 +0000568</ul>
Bill Wendling190ec9b2012-02-16 10:23:43 +0000569
Anshuman Dasgupta4b479552012-04-12 15:17:35 +0000570<p> We added new TableGen infrastructure to support bundling for
571 Very Long Instruction Word (VLIW) architectures. TableGen can now
572 automatically generate a deterministic finite automaton from a VLIW
573 target's schedule description which can be queried to determine
574 legal groupings of instructions in a bundle.</p>
575
Anshuman Dasgupta3c1ded22012-04-14 20:59:13 +0000576<p> We have added a new target independent VLIW packetizer based on the
Anshuman Dasguptacff391a2012-04-14 20:57:13 +0000577 DFA infrastructure to group machine instructions into bundles.</p>
578
Chris Lattner84977642007-09-21 03:54:09 +0000579</div>
580
Chandler Carruth268fde42012-04-17 01:10:35 +0000581<h4>
582<a name="blockplacement">Basic Block Placement</a>
583</h4>
584<div>
585<p>A probability based block placement and code layout algorithm was added to
586LLVM's code generator. This layout pass supports probabilities derived from
587static heuristics as well as source code annotations such as
588<code>__builtin_expect</code>.</p>
589</div>
590
Chris Lattner917cc712009-03-02 02:37:32 +0000591<!--=========================================================================-->
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000592<h3>
Chris Lattner917cc712009-03-02 02:37:32 +0000593<a name="x86">X86-32 and X86-64 Target Improvements</a>
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000594</h3>
Chris Lattner917cc712009-03-02 02:37:32 +0000595
NAKAMURA Takumi074eeaa2011-04-21 01:52:00 +0000596<div>
Bill Wendling7b7fa742011-10-26 18:46:16 +0000597
598<p>New features and major changes in the X86 target include:</p>
Chris Lattner917cc712009-03-02 02:37:32 +0000599
600<ul>
Benjamin Kramer3a2d5652012-05-13 13:10:35 +0000601 <li>Greatly improved support for AVX2.</li>
602 <li>Lots of bug fixes and improvements for AVX1.</li>
603 <li>Support for the FMA4 and XOP instruction set extensions.</li>
Jakob Stoklund Olesen26246612012-02-16 18:22:39 +0000604 <li>Call instructions use the new register mask operands for faster compile
605 times and better support for different calling conventions. The old WINCALL
606 instructions are no longer needed.</li>
NAKAMURA Takumi847307a2012-04-07 02:24:20 +0000607 <li>DW2 Exception Handling is enabled on Cygwin and MinGW.</li>
Benjamin Kramer3a2d5652012-05-13 13:10:35 +0000608 <li>Support for implicit TLS model used with MSVC runtime.</li>
Chris Lattner917cc712009-03-02 02:37:32 +0000609</ul>
610
Chris Lattner917cc712009-03-02 02:37:32 +0000611</div>
Chris Lattner84977642007-09-21 03:54:09 +0000612
613<!--=========================================================================-->
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000614<h3>
Chris Lattner61358ab2009-10-13 17:48:04 +0000615<a name="ARM">ARM Target Improvements</a>
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000616</h3>
Chris Lattnerc441fb82009-03-01 02:30:21 +0000617
NAKAMURA Takumi074eeaa2011-04-21 01:52:00 +0000618<div>
Bill Wendling7b7fa742011-10-26 18:46:16 +0000619
620<p>New features of the ARM target include:</p>
Chris Lattnerc441fb82009-03-01 02:30:21 +0000621
622<ul>
Jakob Stoklund Olesenf5bb45f2011-12-16 16:07:41 +0000623 <li>The constant island pass now supports basic block and constant pool entry
Jim Grosbach6e6b8222012-02-23 23:52:06 +0000624 alignments greater than 4 bytes.</li>
625 <li>On Darwin, the ARM target now has a full-featured integrated assembler.
626 </li>
627</ul>
Chris Lattner7b95c382011-11-15 22:23:46 +0000628
Jim Grosbach6e6b8222012-02-23 23:52:06 +0000629<h4>
630<a name="armintegratedassembler">ARM Integrated Assembler</a>
631</h4>
632<div>
633<p>The ARM target now includes a full featured macro assembler, including
634direct-to-object module support for clang. The assembler is currently enabled
635by default for Darwin only pending testing and any additional necessary
636platform specific support for Linux.</p>
Michael J. Spencer60f790c2011-11-28 18:20:09 +0000637
Jim Grosbach6e6b8222012-02-23 23:52:06 +0000638<p>Full support is included for Thumb1, Thumb2 and ARM modes, along with
639subtarget and CPU specific extensions for VFP2, VFP3 and NEON.</p>
640
641<p>The assembler is Unified Syntax only (see ARM Architecural Reference Manual
642for details). While there is some, and growing, support for pre-unfied (divided)
643syntax, there are still significant gaps in that support.</p>
644</div>
NAKAMURA Takumi9c55f592012-03-27 11:25:16 +0000645
646</div>
Chris Lattner1efe27e2011-04-06 00:45:11 +0000647<!--=========================================================================-->
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000648<h3>
Akira Hatanaka5381cbf2011-11-15 21:33:05 +0000649<a name="MIPS">MIPS Target Improvements</a>
650</h3>
651
652<div>
Akira Hatanaka19e74212012-05-14 18:40:07 +0000653New features and major changes in the MIPS target include:</p>
Akira Hatanaka5381cbf2011-11-15 21:33:05 +0000654
655<ul>
Akira Hatanaka19e74212012-05-14 18:40:07 +0000656 <li>MIPS32 little-endian direct object code emission is functional.</li>
657 <li>MIPS64 little-endian code generation is largely functional for N64 ABI in assembly printing mode with the exception of handling of long double (f128) type.</li>
658 <li>Support for new instructions has been added, which includes swap-bytes
659 instructions (WSBH and DSBH), floating point multiply-add/subtract and
660 negative multiply-add/subtract instructions, and floating
661 point load/store instructions with reg+reg addressing (LWXC1, etc.)</li>
662 <li>Various fixes to improve performance have been implemented.</li>
663 <li>Post-RA scheduling is now enabled at -O3.</li>
664 <li>Support for soft-float code generation has been added.</li>
Akira Hatanaka6ed743e2012-05-15 20:06:41 +0000665 <li>clang driver's support for MIPS 64-bits targets.</li>
666 <li>Support for MIPS floating point ABI option in clang driver.</li>
Akira Hatanaka5381cbf2011-11-15 21:33:05 +0000667</ul>
668</div>
Chris Lattner7b95c382011-11-15 22:23:46 +0000669
670<!--=========================================================================-->
671<h3>
Justin Holewinskidc23c142012-05-13 17:32:35 +0000672<a name="PTX">PTX Target Improvements</a>
673</h3>
674
675<div>
676
677<p>An outstanding conditional inversion bug was fixed in this release.</p>
678
679<p><b>NOTE</b>: LLVM 3.1 marks the last release of the PTX back-end, in its
680 current form. The back-end is currently being replaced by the NVPTX
681 back-end, currently in SVN ToT.</p>
682
683</div>
684
685<!--=========================================================================-->
686<h3>
Chris Lattner1efe27e2011-04-06 00:45:11 +0000687<a name="OtherTS">Other Target Specific Improvements</a>
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000688</h3>
Chris Lattnerc441fb82009-03-01 02:30:21 +0000689
NAKAMURA Takumi074eeaa2011-04-21 01:52:00 +0000690<div>
Bill Wendling7b7fa742011-10-26 18:46:16 +0000691
Chris Lattner1efe27e2011-04-06 00:45:11 +0000692<ul>
Bill Wendling1fefd092012-05-14 08:11:53 +0000693 <li>Support for Qualcomm's Hexagon VLIW processor has been added.</li>
Chris Lattner1efe27e2011-04-06 00:45:11 +0000694</ul>
Bill Wendling7b7fa742011-10-26 18:46:16 +0000695
Chris Lattner1efe27e2011-04-06 00:45:11 +0000696</div>
Chris Lattner77d29b12008-06-05 08:02:49 +0000697
698<!--=========================================================================-->
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000699<h3>
Chris Lattnerf6662f92008-10-13 17:57:36 +0000700<a name="changes">Major Changes and Removed Features</a>
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000701</h3>
Chris Lattnerf6662f92008-10-13 17:57:36 +0000702
NAKAMURA Takumi074eeaa2011-04-21 01:52:00 +0000703<div>
Chris Lattnerf6662f92008-10-13 17:57:36 +0000704
Bill Wendling2626dba2011-08-03 22:18:20 +0000705<p>If you're already an LLVM user or developer with out-of-tree changes based on
Chris Lattner21e1b7a2011-12-13 17:55:30 +0000706 LLVM 3.1, this section lists some "gotchas" that you may run into upgrading
Bill Wendling2626dba2011-08-03 22:18:20 +0000707 from the previous release.</p>
Chris Lattnerf6662f92008-10-13 17:57:36 +0000708
709<ul>
Benjamin Kramer2fb68012012-05-13 11:46:05 +0000710 <li>LLVM's build system now requires a python 2 interpreter to be present at
711 build time. A perl interpreter is no longer required.</li>
Benjamin Kramer5fdfaf52012-05-13 11:28:46 +0000712 <li>The C backend has been removed. It had numerous problems, to the point of
713 not being able to compile any nontrivial program.</li>
714 <li>The Alpha, Blackfin and SystemZ targets have been removed due to lack of
715 maintenance.</li>
Bill Wendling190ec9b2012-02-16 10:23:43 +0000716 <li>LLVM 3.1 removes support for reading LLVM 2.9 bitcode files. Going
717 forward, we aim for all future versions of LLVM to read bitcode files and
718 <tt>.ll</tt> files produced by LLVM 3.0 and later.</li>
719 <li>The <tt>unwind</tt> instruction is now gone. With the introduction of the
720 new exception handling system in LLVM 3.0, the <tt>unwind</tt> instruction
721 became obsolete.</li>
Joerg Sonnenbergerdba86d82012-04-26 20:10:07 +0000722 <li>LLVM 3.0 and earlier automatically added the returns_twice fo functions
723 like setjmp based on the name. This functionality was removed in 3.1.
724 This affects Clang users, if -ffreestanding is used.</li>
Devang Patelb34dd132008-10-14 20:03:43 +0000725</ul>
Chris Lattnerf6662f92008-10-13 17:57:36 +0000726
Chris Lattnerf6662f92008-10-13 17:57:36 +0000727</div>
728
Daniel Dunbarf0233c62010-10-04 20:11:41 +0000729<!--=========================================================================-->
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000730<h3>
Chris Lattner1efe27e2011-04-06 00:45:11 +0000731<a name="api_changes">Internal API Changes</a>
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000732</h3>
Daniel Dunbarf0233c62010-10-04 20:11:41 +0000733
NAKAMURA Takumi074eeaa2011-04-21 01:52:00 +0000734<div>
Daniel Dunbarf0233c62010-10-04 20:11:41 +0000735
Chris Lattner1efe27e2011-04-06 00:45:11 +0000736<p>In addition, many APIs have changed in this release. Some of the major
Bill Wendling16005252011-08-02 06:20:17 +0000737 LLVM API changes are:</p>
Daniel Dunbarf0233c62010-10-04 20:11:41 +0000738
739<ul>
Nick Lewycky98a92d12012-03-21 22:58:28 +0000740 <li>Target specific options have been moved from global variables to members
741 on the new <code>TargetOptions</code> class, which is local to each
742 <code>TargetMachine</code>. As a consequence, the associated flags will
743 no longer be accepted by <tt>clang -mllvm</tt>. This includes:
744<ul>
Nick Lewycky8a618a22012-03-23 00:56:26 +0000745<li><code>llvm::PrintMachineCode</code></li>
746<li><code>llvm::NoFramePointerElim</code></li>
747<li><code>llvm::NoFramePointerElimNonLeaf</code></li>
748<li><code>llvm::DisableFramePointerElim(const MachineFunction &)</code></li>
749<li><code>llvm::LessPreciseFPMADOption</code></li>
750<li><code>llvm::LessPrecideFPMAD()</code></li>
751<li><code>llvm::NoExcessFPPrecision</code></li>
752<li><code>llvm::UnsafeFPMath</code></li>
753<li><code>llvm::NoInfsFPMath</code></li>
754<li><code>llvm::NoNaNsFPMath</code></li>
755<li><code>llvm::HonorSignDependentRoundingFPMathOption</code></li>
756<li><code>llvm::HonorSignDependentRoundingFPMath()</code></li>
757<li><code>llvm::UseSoftFloat</code></li>
758<li><code>llvm::FloatABIType</code></li>
759<li><code>llvm::NoZerosInBSS</code></li>
760<li><code>llvm::JITExceptionHandling</code></li>
761<li><code>llvm::JITEmitDebugInfo</code></li>
762<li><code>llvm::JITEmitDebugInfoToDisk</code></li>
763<li><code>llvm::GuaranteedTailCallOpt</code></li>
764<li><code>llvm::StackAlignmentOverride</code></li>
765<li><code>llvm::RealignStack</code></li>
766<li><code>llvm::DisableJumpTables</code></li>
767<li><code>llvm::EnableFastISel</code></li>
768<li><code>llvm::getTrapFunctionName()</code></li>
769<li><code>llvm::EnableSegmentedStacks</code></li>
Nick Lewycky98a92d12012-03-21 22:58:28 +0000770</ul></li>
Bill Wendling1fefd092012-05-14 08:11:53 +0000771
772 <li>The <code>MDBuilder</code> class has been added to simplify the creation
773 of metadata.</li>
Daniel Dunbarf0233c62010-10-04 20:11:41 +0000774</ul>
Bill Wendling7b7fa742011-10-26 18:46:16 +0000775
Daniel Dunbarf0233c62010-10-04 20:11:41 +0000776</div>
Chris Lattnerf6662f92008-10-13 17:57:36 +0000777
Nadav Rotemfdc309c2012-02-26 08:35:53 +0000778<!--=========================================================================-->
779<h3>
780<a name="tools_changes">Tools Changes</a>
781</h3>
782
783<div>
784
785<p>In addition, some tools have changed in this release. Some of the changes
786 are:</p>
787
788
789<ul>
Benjamin Kramer5fdfaf52012-05-13 11:28:46 +0000790 <li><tt>llvm-stress</tt> is a command line tool for generating random
791 <tt>.ll</tt> files to fuzz different LLVM components. </li>
792 <li>The <tt>llvm-ld</tt> tool has been removed. The clang driver provides a
793 more reliable solution for turning a set of bitcode files into a binary.
794 To merge bitcode files <tt>llvm-link</tt> can be used instead.</li>
Nadav Rotemfdc309c2012-02-26 08:35:53 +0000795</ul>
796
797</div>
798
Gregory Szorcca347492012-05-12 21:12:22 +0000799
800<!--=========================================================================-->
801<h3>
802<a name="python">Python Bindings</a>
803</h3>
804
805<div>
806
807<p>Officially supported Python bindings have been added! Feature support is far
808from complete. The current bindings support interfaces to:</p>
809<ul>
810 <li>Object File Interface</li>
811 <li>Disassembler</li>
812</ul>
813
814<p>Using the Object File Interface, it is possible to inspect binary object files.
815Think of it as a Python version of readelf or llvm-objdump.</p>
816
817<p>Support for additional features is currently being developed by community
818contributors. If you are interested in shaping the direction of the Python
819bindings, please express your intent on IRC or the developers list.</p>
820
821</div>
822
Nadav Rotemfdc309c2012-02-26 08:35:53 +0000823</div>
824
Chris Lattner79c3fe12003-10-02 04:57:28 +0000825<!-- *********************************************************************** -->
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000826<h2>
Misha Brukman500bc302003-11-22 00:38:41 +0000827 <a name="knownproblems">Known Problems</a>
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000828</h2>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000829<!-- *********************************************************************** -->
830
NAKAMURA Takumi074eeaa2011-04-21 01:52:00 +0000831<div>
Misha Brukman500bc302003-11-22 00:38:41 +0000832
Chris Lattner70e22012011-11-27 19:38:20 +0000833<p>LLVM is generally a production quality compiler, and is used by a broad range
834 of applications and shipping in many products. That said, not every
835 subsystem is as mature as the aggregate, particularly the more obscure
836 targets. If you run into a problem, please check the <a
837 href="http://llvm.org/bugs/">LLVM bug database</a> and submit a bug if
838 there isn't already one or ask on the <a
839 href="http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev">LLVMdev
840 list</a>.</p>
Michael J. Spencer60f790c2011-11-28 18:20:09 +0000841
Chris Lattner70e22012011-11-27 19:38:20 +0000842 <p>Known problem areas include:</p>
Michael J. Spencer60f790c2011-11-28 18:20:09 +0000843
Chris Lattnerf5ee1702004-03-14 02:03:02 +0000844<ul>
Benjamin Kramer5fdfaf52012-05-13 11:28:46 +0000845 <li>The CellSPU, MSP430, PTX and XCore backends are experimental.</li>
Bill Wendling7b7fa742011-10-26 18:46:16 +0000846
Chris Lattner70e22012011-11-27 19:38:20 +0000847 <li>The integrated assembler, disassembler, and JIT is not supported by
848 several targets. If an integrated assembler is not supported, then a
849 system assembler is required. For more details, see the <a
850 href="CodeGenerator.html#targetfeatures">Target Features Matrix</a>.
851 </li>
Chris Lattner26299222006-11-18 07:51:14 +0000852</ul>
853
854</div>
John Criswellc0c186d2005-11-08 21:11:33 +0000855
Chris Lattner79c3fe12003-10-02 04:57:28 +0000856<!-- *********************************************************************** -->
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000857<h2>
Misha Brukman500bc302003-11-22 00:38:41 +0000858 <a name="additionalinfo">Additional Information</a>
NAKAMURA Takumi06c6d9a2011-04-18 01:17:51 +0000859</h2>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000860<!-- *********************************************************************** -->
861
NAKAMURA Takumi074eeaa2011-04-21 01:52:00 +0000862<div>
Misha Brukman500bc302003-11-22 00:38:41 +0000863
Bill Wendling7b7fa742011-10-26 18:46:16 +0000864<p>A wide variety of additional information is available on
865 the <a href="http://llvm.org/">LLVM web page</a>, in particular in
866 the <a href="http://llvm.org/docs/">documentation</a> section. The web page
867 also contains versions of the API documentation which is up-to-date with the
868 Subversion version of the source code. You can access versions of these
869 documents specific to this release by going into the "<tt>llvm/doc/</tt>"
870 directory in the LLVM tree.</p>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000871
Misha Brukman500bc302003-11-22 00:38:41 +0000872<p>If you have any questions or comments about LLVM, please feel free to contact
Bill Wendling7b7fa742011-10-26 18:46:16 +0000873 us via the <a href="http://llvm.org/docs/#maillist"> mailing lists</a>.</p>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000874
Misha Brukman500bc302003-11-22 00:38:41 +0000875</div>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000876
877<!-- *********************************************************************** -->
Chris Lattner79c3fe12003-10-02 04:57:28 +0000878
Misha Brukman500bc302003-11-22 00:38:41 +0000879<hr>
Misha Brukman2061e892003-11-22 01:23:39 +0000880<address>
Misha Brukman38847d52003-12-21 22:53:21 +0000881 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
Misha Brukman44408702008-12-11 17:34:48 +0000882 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
Misha Brukman2061e892003-11-22 01:23:39 +0000883 <a href="http://validator.w3.org/check/referer"><img
Misha Brukman44408702008-12-11 17:34:48 +0000884 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
Misha Brukman2061e892003-11-22 01:23:39 +0000885
Chris Lattnerb4b0ce72007-05-18 00:44:29 +0000886 <a href="http://llvm.org/">LLVM Compiler Infrastructure</a><br>
Misha Brukman500bc302003-11-22 00:38:41 +0000887 Last modified: $Date$
Misha Brukman2061e892003-11-22 01:23:39 +0000888</address>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000889
Misha Brukman500bc302003-11-22 00:38:41 +0000890</body>
891</html>