Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / 7ae3ac831e3968c0e4d98332503cee772d9e0681 / . / docs / ReleaseNotes.html
blob: 94e5c7b398a460c00bb237d53301cd7852a4235f [file] [log] [blame]
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1<!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 +0000[diff] [blame]5 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]6 <link rel="stylesheet" href="llvm.css" type="text/css">
Dan Gohmanb44f6c62010-05-03 23:51:05 +0000[diff] [blame]7 <title>LLVM 2.8 Release Notes</title>
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]8</head>
9<body>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]10
Dan Gohmanb44f6c62010-05-03 23:51:05 +0000[diff] [blame]11<div class="doc_title">LLVM 2.8 Release Notes</div>
Mikhail Glushenkovea65d7d2008-10-13 02:08:34 +0000[diff] [blame]12
Chris Lattner0e464a92010-03-17 04:02:39 +0000[diff] [blame]13<img align=right src="http://llvm.org/img/DragonSmall.png"
Gabor Greifee2187a2010-04-22 10:21:43 +0000[diff] [blame]14 width="136" height="136" alt="LLVM Dragon Logo">
Chris Lattner0e464a92010-03-17 04:02:39 +0000[diff] [blame]15
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]16<ol>
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]17 <li><a href="#intro">Introduction</a></li>
Chris Lattner96a445e2008-10-13 18:01:01 +0000[diff] [blame]18 <li><a href="#subproj">Sub-project Status Update</a></li>
Dan Gohmanb44f6c62010-05-03 23:51:05 +0000[diff] [blame]19 <li><a href="#externalproj">External Projects Using LLVM 2.8</a></li>
20 <li><a href="#whatsnew">What's New in LLVM 2.8?</a></li>
Chris Lattner4b538b92004-04-30 22:17:12 +0000[diff] [blame]21 <li><a href="GettingStarted.html">Installation Instructions</a></li>
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]22 <li><a href="#portability">Portability and Supported Platforms</a></li>
Dan Gohman44aa9212008-10-14 16:23:02 +0000[diff] [blame]23 <li><a href="#knownproblems">Known Problems</a></li>
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]24 <li><a href="#additionalinfo">Additional Information</a></li>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]25</ol>
26
Chris Lattner7911ce22004-05-23 21:07:27 +0000[diff] [blame]27<div class="doc_author">
Dan Gohman44aa9212008-10-14 16:23:02 +0000[diff] [blame]28 <p>Written by the <a href="http://llvm.org">LLVM Team</a></p>
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]29</div>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]30
Chris Lattner7b91eda2010-04-22 05:41:35 +0000[diff] [blame]31<h1 style="color:red">These are in-progress notes for the upcoming LLVM 2.8
Jeffrey Yasskinbec48772010-01-28 01:14:43 +0000[diff] [blame]32release.<br>
33You may prefer the
Dan Gohmanb44f6c62010-05-03 23:51:05 +0000[diff] [blame]34<a href="http://llvm.org/releases/2.7/docs/ReleaseNotes.html">LLVM 2.7
35Release Notes</a>.</h1>
Jeffrey Yasskinbec48772010-01-28 01:14:43 +0000[diff] [blame]36
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]37<!-- *********************************************************************** -->
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]38<div class="doc_section">
39 <a name="intro">Introduction</a>
40</div>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]41<!-- *********************************************************************** -->
42
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]43<div class="doc_text">
44
Chris Lattnerf6662f92008-10-13 17:57:36 +0000[diff] [blame]45<p>This document contains the release notes for the LLVM Compiler
Dan Gohman7ae3ac82010-05-03 23:52:21 +0000[diff] [blame^]46Infrastructure, release 2.8. Here we describe the status of LLVM, including
Chris Lattnerf6662f92008-10-13 17:57:36 +0000[diff] [blame]47major improvements from the previous release and significant known problems.
Mikhail Glushenkovf795ef02009-03-01 18:09:47 +0000[diff] [blame]48All LLVM releases may be downloaded from the <a
Chris Lattnerf6662f92008-10-13 17:57:36 +0000[diff] [blame]49href="http://llvm.org/releases/">LLVM releases web site</a>.</p>
Chris Lattner19092612003-10-02 16:38:05 +0000[diff] [blame]50
Chris Lattner7506b1d2004-12-07 08:04:13 +0000[diff] [blame]51<p>For more information about LLVM, including information about the latest
Chris Lattnerc463b272005-10-29 07:07:09 +0000[diff] [blame]52release, please check out the <a href="http://llvm.org/">main LLVM
Chris Lattner47ad72c2003-10-07 21:38:31 +0000[diff] [blame]53web site</a>. If you have questions or comments, the <a
Chris Lattnerc66bfef2010-03-17 04:41:49 +0000[diff] [blame]54href="http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev">LLVM Developer's
55Mailing List</a> is a good place to send them.</p>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]56
Mikhail Glushenkovea65d7d2008-10-13 02:08:34 +0000[diff] [blame]57<p>Note that if you are reading this file from a Subversion checkout or the
Chris Lattner84977642007-09-21 03:54:09 +0000[diff] [blame]58main LLVM web page, this document applies to the <i>next</i> release, not the
Gabor Greiffa933f82008-10-14 11:00:32 +0000[diff] [blame]59current one. To see the release notes for a specific release, please see the
Chris Lattner84977642007-09-21 03:54:09 +0000[diff] [blame]60<a href="http://llvm.org/releases/">releases page</a>.</p>
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]61
62</div>
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]63
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]64
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]65<!--
66Almost dead code.
67 include/llvm/Analysis/LiveValues.h => Dan
68 lib/Transforms/IPO/MergeFunctions.cpp => consider for 2.8.
69 llvm/Analysis/PointerTracking.h => Edwin wants this, consider for 2.8.
Chris Lattner00736fc2010-04-13 06:37:00 +0000[diff] [blame]70 ABCD, GEPSplitterPass
Chris Lattner048fe3c2010-01-16 21:25:13 +0000[diff] [blame]71 MSIL backend?
Chris Lattnerdc910082010-03-17 06:41:58 +0000[diff] [blame]72 lib/Transforms/Utils/SSI.cpp -> ABCD depends on it.
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]73-->
74
75
Dan Gohmanb44f6c62010-05-03 23:51:05 +0000[diff] [blame]76<!-- Features that need text if they're finished for 2.8:
Chris Lattner914ce462010-04-22 06:28:20 +0000[diff] [blame]77 combiner-aa?
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]78 strong phi elim
Chris Lattnerb7c85b42010-04-21 05:17:40 +0000[diff] [blame]79 llvm.dbg.value: variable debug info for optimized code
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]80 loop dependence analysis
Chris Lattneracce85d2008-02-10 07:46:44 +0000[diff] [blame]81 -->
Mikhail Glushenkovea65d7d2008-10-13 02:08:34 +0000[diff] [blame]82
Chris Lattner547a3912008-10-12 19:47:48 +0000[diff] [blame]83 <!-- for announcement email:
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]84 Logo web page.
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]85 Many new papers added to /pubs/
Chris Lattner74c80df2009-02-25 06:34:50 +0000[diff] [blame]86 -->
Chris Lattneracce85d2008-02-10 07:46:44 +0000[diff] [blame]87
Chris Lattner96a445e2008-10-13 18:01:01 +0000[diff] [blame]88<!-- *********************************************************************** -->
89<div class="doc_section">
90 <a name="subproj">Sub-project Status Update</a>
Chris Lattnerea34f642008-06-08 21:34:41 +0000[diff] [blame]91</div>
Chris Lattner96a445e2008-10-13 18:01:01 +0000[diff] [blame]92<!-- *********************************************************************** -->
Chris Lattnerea34f642008-06-08 21:34:41 +0000[diff] [blame]93
94<div class="doc_text">
Chris Lattner8348b472008-06-08 21:58:17 +0000[diff] [blame]95<p>
Dan Gohmanb44f6c62010-05-03 23:51:05 +0000[diff] [blame]96The LLVM 2.8 distribution currently consists of code from the core LLVM
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]97repository (which roughly includes the LLVM optimizers, code generators
98and supporting tools), the Clang repository and the llvm-gcc repository. In
99addition to this code, the LLVM Project includes other sub-projects that are in
100development. Here we include updates on these subprojects.
Bill Wendling63d8c552009-03-02 04:28:57 +0000[diff] [blame]101</p>
Chris Lattner96a445e2008-10-13 18:01:01 +0000[diff] [blame]102
Chris Lattner8348b472008-06-08 21:58:17 +0000[diff] [blame]103</div>
104
Chris Lattner96a445e2008-10-13 18:01:01 +0000[diff] [blame]105
Chris Lattner8348b472008-06-08 21:58:17 +0000[diff] [blame]106<!--=========================================================================-->
Chris Lattner96a445e2008-10-13 18:01:01 +0000[diff] [blame]107<div class="doc_subsection">
Chris Lattnerfb97b2d2008-10-13 18:11:54 +0000[diff] [blame]108<a name="clang">Clang: C/C++/Objective-C Frontend Toolkit</a>
Chris Lattner8348b472008-06-08 21:58:17 +0000[diff] [blame]109</div>
110
111<div class="doc_text">
112
Chris Lattner095539f2010-04-26 17:42:18 +0000[diff] [blame]113<p><a href="http://clang.llvm.org/">Clang</a> is an LLVM front end for the C,
114C++, and Objective-C languages. Clang aims to provide a better user experience
115through expressive diagnostics, a high level of conformance to language
116standards, fast compilation, and low memory use. Like LLVM, Clang provides a
117modular, library-based architecture that makes it suitable for creating or
118integrating with other development tools. Clang is considered a
119production-quality compiler for C and Objective-C on x86 (32- and 64-bit).</p>
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]120
Jeffrey Yasskinbec48772010-01-28 01:14:43 +0000[diff] [blame]121<p>In the LLVM 2.7 time-frame, the Clang team has made many improvements:</p>
Bill Wendling741748a2008-10-27 09:27:33 +0000[diff] [blame]122
Daniel Dunbar13739432008-10-14 23:25:09 +0000[diff] [blame]123<ul>
Douglas Gregorf720a4e2010-04-22 20:42:40 +0000[diff] [blame]124
Chris Lattner095539f2010-04-26 17:42:18 +0000[diff] [blame]125<li>C++ Support: Clang is now capable of self-hosting! While still
126alpha-quality, Clang's C++ support has matured enough to build LLVM and Clang,
127and C++ is now enabled by default. See the <a
128href="http://clang.llvm.org/cxx_compatibility.html">Clang C++ compatibility
129page</a> for common C++ migration issues.</li>
Douglas Gregorf720a4e2010-04-22 20:42:40 +0000[diff] [blame]130
Chris Lattner095539f2010-04-26 17:42:18 +0000[diff] [blame]131<li>Objective-C: Clang now includes experimental support for an updated
132Objective-C ABI on non-Darwin platforms. This includes support for non-fragile
133instance variables and accelerated proxies, as well as greater potential for
134future optimisations. The new ABI is used when compiling with the
135-fobjc-nonfragile-abi and -fgnu-runtime options. Code compiled with these
136options may be mixed with code compiled with GCC or clang using the old GNU ABI,
137but requires the libobjc2 runtime from the GNUstep project.</li>
David Chisnall8e0bd802010-04-25 19:13:33 +0000[diff] [blame]138
Chris Lattner095539f2010-04-26 17:42:18 +0000[diff] [blame]139<li>New warnings: Clang contains a number of new warnings, including
140control-flow warnings (unreachable code, missing return statements in a
141non-<code>void</code> function, etc.), sign-comparison warnings, and improved
142format-string warnings.</li>
Daniel Dunbar60dadb62010-03-25 16:09:18 +0000[diff] [blame]143
Daniel Dunbar60dadb62010-03-25 16:09:18 +0000[diff] [blame]144<li>CIndex API and Python bindings: Clang now includes a C API as part of the
Gabor Greif1440dbe2010-04-25 21:27:54 +0000[diff] [blame]145CIndex library. Although we may make some changes to the API in the future, it
Daniel Dunbar60dadb62010-03-25 16:09:18 +0000[diff] [blame]146is intended to be stable and has been designed for use by external projects. See
147the Clang
148doxygen <a href="http://clang.llvm.org/doxygen/group__CINDEX.html">CIndex</a>
Wesley Peck1f433612010-04-22 14:19:00 +0000[diff] [blame]149documentation for more details. The CIndex API also includes a preliminary
Daniel Dunbar60dadb62010-03-25 16:09:18 +0000[diff] [blame]150set of Python bindings.</li>
151
152<li>ARM Support: Clang now has ABI support for both the Darwin and Linux ARM
153ABIs. Coupled with many improvements to the LLVM ARM backend, Clang is now
Gabor Greif1440dbe2010-04-25 21:27:54 +0000[diff] [blame]154suitable for use as a beta quality ARM compiler.</li>
David Chisnall8e0bd802010-04-25 19:13:33 +0000[diff] [blame]155
Bill Wendling6bc15282009-03-02 04:28:18 +0000[diff] [blame]156</ul>
Chris Lattnerfb97b2d2008-10-13 18:11:54 +0000[diff] [blame]157</div>
158
159<!--=========================================================================-->
160<div class="doc_subsection">
161<a name="clangsa">Clang Static Analyzer</a>
162</div>
163
164<div class="doc_text">
165
Chris Lattner8cdd7932010-04-22 06:38:11 +0000[diff] [blame]166<p>The <a href="http://clang-analyzer.llvm.org/">Clang Static Analyzer</a>
167 project is an effort to use static source code analysis techniques to
168 automatically find bugs in C and Objective-C programs (and hopefully <a
169 href="http://clang-analyzer.llvm.org/dev_cxx.html">C++ in the
170 future</a>!). The tool is very good at finding bugs that occur on specific
171 paths through code, such as on error conditions.</p>
Chris Lattnercc042612008-10-14 00:52:49 +0000[diff] [blame]172
Chris Lattner8cdd7932010-04-22 06:38:11 +0000[diff] [blame]173<p>In the LLVM 2.7 time-frame, the analyzer core has made several major and
174 minor improvements, including better support for tracking the fields of
175 structures, initial support (not enabled by default yet) for doing
176 interprocedural (cross-function) analysis, and new checks have been added.
177</p>
Chris Lattner8348b472008-06-08 21:58:17 +0000[diff] [blame]178
179</div>
180
Chris Lattner96a445e2008-10-13 18:01:01 +0000[diff] [blame]181<!--=========================================================================-->
182<div class="doc_subsection">
Nicolas Geoffray99a4d302008-10-14 19:23:04 +0000[diff] [blame]183<a name="vmkit">VMKit: JVM/CLI Virtual Machine Implementation</a>
Chris Lattner96a445e2008-10-13 18:01:01 +0000[diff] [blame]184</div>
185
186<div class="doc_text">
187<p>
Nicolas Geoffray99a4d302008-10-14 19:23:04 +0000[diff] [blame]188The <a href="http://vmkit.llvm.org/">VMKit project</a> is an implementation of
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]189a JVM and a CLI Virtual Machine (Microsoft .NET is an
190implementation of the CLI) using LLVM for static and just-in-time
191compilation.</p>
Chris Lattner96a445e2008-10-13 18:01:01 +0000[diff] [blame]192
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]193<p>
Chris Lattnerbf8e5162010-03-29 18:34:13 +0000[diff] [blame]194With the release of LLVM 2.7, VMKit has shifted to a great framework for writing
195virtual machines. VMKit now offers precise and efficient garbage collection with
196multi-threading support, thanks to the MMTk memory management toolkit, as well
197as just in time and ahead of time compilation with LLVM. The major changes in
198VMKit 0.27 are:</p>
Chris Lattner96a445e2008-10-13 18:01:01 +0000[diff] [blame]199
Nicolas Geoffray99a4d302008-10-14 19:23:04 +0000[diff] [blame]200<ul>
201
Chris Lattnerbf8e5162010-03-29 18:34:13 +0000[diff] [blame]202<li>Garbage collection: VMKit now uses the MMTk toolkit for garbage collectors.
203 The first collector to be ported is the MarkSweep collector, which is precise,
204 and drastically improves the performance of VMKit.</li>
205<li>Line number information in the JVM: by using the debug metadata of LLVM, the
206 JVM now supports precise line number information, useful when printing a stack
207 trace.</li>
208<li>Interface calls in the JVM: we implemented a variant of the Interface Method
209 Table technique for interface calls in the JVM.
210</li>
Nicolas Geoffray99a4d302008-10-14 19:23:04 +0000[diff] [blame]211
212</ul>
Chris Lattner96a445e2008-10-13 18:01:01 +0000[diff] [blame]213</div>
214
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]215
216<!--=========================================================================-->
217<div class="doc_subsection">
218<a name="compiler-rt">compiler-rt: Compiler Runtime Library</a>
219</div>
220
221<div class="doc_text">
222<p>
223The new LLVM <a href="http://compiler-rt.llvm.org/">compiler-rt project</a>
224is a simple library that provides an implementation of the low-level
225target-specific hooks required by code generation and other runtime components.
226For example, when compiling for a 32-bit target, converting a double to a 64-bit
227unsigned integer is compiled into a runtime call to the "__fixunsdfdi"
228function. The compiler-rt library provides highly optimized implementations of
229this and other low-level routines (some are 3x faster than the equivalent
230libgcc routines).</p>
231
232<p>
233All of the code in the compiler-rt project is available under the standard LLVM
Chris Lattnerb7c85b42010-04-21 05:17:40 +0000[diff] [blame]234License, a "BSD-style" license. New in LLVM 2.7: compiler_rt now
235supports ARM targets.</p>
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]236
237</div>
238
239<!--=========================================================================-->
240<div class="doc_subsection">
Duncan Sands749fd832010-04-02 09:23:15 +0000[diff] [blame]241<a name="dragonegg">DragonEgg: llvm-gcc ported to gcc-4.5</a>
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]242</div>
243
244<div class="doc_text">
245<p>
Duncan Sands749fd832010-04-02 09:23:15 +0000[diff] [blame]246<a href="http://dragonegg.llvm.org/">DragonEgg</a> is a port of llvm-gcc to
247gcc-4.5. Unlike llvm-gcc, which makes many intrusive changes to the underlying
248gcc-4.2 code, dragonegg in theory does not require any gcc-4.5 modifications
249whatsoever (currently one small patch is needed). This is thanks to the new
250<a href="http://gcc.gnu.org/wiki/plugins">gcc plugin architecture</a>, which
251makes it possible to modify the behaviour of gcc at runtime by loading a plugin,
252which is nothing more than a dynamic library which conforms to the gcc plugin
253interface. DragonEgg is a gcc plugin that causes the LLVM optimizers to be run
254instead of the gcc optimizers, and the LLVM code generators instead of the gcc
255code generators, just like llvm-gcc. To use it, you add
256"-fplugin=path/dragonegg.so" to the gcc-4.5 command line, and gcc-4.5 magically
257becomes llvm-gcc-4.5!
258</p>
259
260<p>
261DragonEgg is still a work in progress. Currently C works very well, while C++,
262Ada and Fortran work fairly well. All other languages either don't work at all,
263or only work poorly. For the moment only the x86-32 and x86-64 targets are
Duncan Sands51a51742010-04-20 19:40:58 +0000[diff] [blame]264supported, and only on linux and darwin (darwin needs an additional gcc patch).
Duncan Sands749fd832010-04-02 09:23:15 +0000[diff] [blame]265</p>
266
267<p>
Duncan Sandscb9dda62010-04-21 13:51:48 +0000[diff] [blame]268DragonEgg is a new project which is seeing its first release with llvm-2.7.
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]269</p>
270
271</div>
272
273
274<!--=========================================================================-->
275<div class="doc_subsection">
276<a name="mc">llvm-mc: Machine Code Toolkit</a>
277</div>
278
279<div class="doc_text">
280<p>
Chris Lattner8cdd7932010-04-22 06:38:11 +0000[diff] [blame]281The LLVM Machine Code (aka MC) sub-project of LLVM was created to solve a number
282of problems in the realm of assembly, disassembly, object file format handling,
283and a number of other related areas that CPU instruction-set level tools work
284in. It is a sub-project of LLVM which provides it with a number of advantages
285over other compilers that do not have tightly integrated assembly-level tools.
286For a gentle introduction, please see the <a
287href="http://blog.llvm.org/2010/04/intro-to-llvm-mc-project.html">Intro to the
288LLVM MC Project Blog Post</a>.
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]289</p>
290
Chris Lattner8cdd7932010-04-22 06:38:11 +0000[diff] [blame]291<p>2.7 includes major parts of the work required by the new MC Project. A few
292 targets have been refactored to support it, and work is underway to support a
Gabor Greife32adf52010-04-25 21:30:22 +0000[diff] [blame]293 native assembler in LLVM. This work is not complete in LLVM 2.7, but it has
Chris Lattner8cdd7932010-04-22 06:38:11 +0000[diff] [blame]294 made substantially more progress on LLVM mainline.</p>
295
296<p>One minor example of what MC can do is to transcode an AT&amp;T syntax
Gabor Greifed387452010-04-22 10:25:23 +0000[diff] [blame]297 X86 .s file into intel syntax. You can do this with something like:</p>
Chris Lattner8cdd7932010-04-22 06:38:11 +0000[diff] [blame]298<pre>
299 llvm-mc foo.s -output-asm-variant=1 -o foo-intel.s
300</pre>
Chris Lattner450a31e2010-04-21 06:23:40 +0000[diff] [blame]301
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]302</div>
303
304
Chris Lattnerab68e9e2009-02-26 22:33:38 +0000[diff] [blame]305<!-- *********************************************************************** -->
306<div class="doc_section">
Jeffrey Yasskinbec48772010-01-28 01:14:43 +0000[diff] [blame]307 <a name="externalproj">External Open Source Projects Using LLVM 2.7</a>
Chris Lattnerab68e9e2009-02-26 22:33:38 +0000[diff] [blame]308</div>
309<!-- *********************************************************************** -->
310
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]311<div class="doc_text">
312
313<p>An exciting aspect of LLVM is that it is used as an enabling technology for
314 a lot of other language and tools projects. This section lists some of the
Jeffrey Yasskinbec48772010-01-28 01:14:43 +0000[diff] [blame]315 projects that have already been updated to work with LLVM 2.7.</p>
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]316</div>
317
Chris Lattnerab68e9e2009-02-26 22:33:38 +0000[diff] [blame]318<!--=========================================================================-->
319<div class="doc_subsection">
320<a name="pure">Pure</a>
321</div>
322
323<div class="doc_text">
324<p>
Chris Lattner917cc712009-03-02 02:37:32 +0000[diff] [blame]325<a href="http://pure-lang.googlecode.com/">Pure</a>
326is an algebraic/functional programming language based on term rewriting.
Chris Lattnerab68e9e2009-02-26 22:33:38 +0000[diff] [blame]327Programs are collections of equations which are used to evaluate expressions in
328a symbolic fashion. Pure offers dynamic typing, eager and lazy evaluation,
Mikhail Glushenkovf795ef02009-03-01 18:09:47 +0000[diff] [blame]329lexical closures, a hygienic macro system (also based on term rewriting),
Chris Lattnerab68e9e2009-02-26 22:33:38 +0000[diff] [blame]330built-in list and matrix support (including list and matrix comprehensions) and
331an easy-to-use C interface. The interpreter uses LLVM as a backend to
332 JIT-compile Pure programs to fast native code.</p>
333
Chris Lattner477a1fd2010-03-17 17:25:49 +0000[diff] [blame]334<p>Pure versions 0.43 and later have been tested and are known to work with
335LLVM 2.7 (and continue to work with older LLVM releases >= 2.5).</p>
336
Chris Lattnerab68e9e2009-02-26 22:33:38 +0000[diff] [blame]337</div>
338
Chris Lattnerbc31caf2009-02-28 18:58:01 +0000[diff] [blame]339<!--=========================================================================-->
340<div class="doc_subsection">
341<a name="RoadsendPHP">Roadsend PHP</a>
342</div>
343
344<div class="doc_text">
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]345<p>
346<a href="http://code.roadsend.com/rphp">Roadsend PHP</a> (rphp) is an open
Chris Lattnercade8222009-03-02 19:07:24 +0000[diff] [blame]347source implementation of the PHP programming
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]348language that uses LLVM for its optimizer, JIT and static compiler. This is a
Chris Lattner0e464a92010-03-17 04:02:39 +0000[diff] [blame]349reimplementation of an earlier project that is now based on LLVM.
Chris Lattner477a1fd2010-03-17 17:25:49 +0000[diff] [blame]350</p>
Chris Lattnerbc31caf2009-02-28 18:58:01 +0000[diff] [blame]351</div>
Chris Lattnerab68e9e2009-02-26 22:33:38 +0000[diff] [blame]352
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]353<!--=========================================================================-->
354<div class="doc_subsection">
355<a name="UnladenSwallow">Unladen Swallow</a>
356</div>
357
358<div class="doc_text">
359<p>
360<a href="http://code.google.com/p/unladen-swallow/">Unladen Swallow</a> is a
361branch of <a href="http://python.org/">Python</a> intended to be fully
362compatible and significantly faster. It uses LLVM's optimization passes and JIT
Chris Lattnerb5f6feb2010-03-18 06:52:15 +0000[diff] [blame]363compiler.
Chris Lattner0e464a92010-03-17 04:02:39 +0000[diff] [blame]364</p>
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]365</div>
366
367<!--=========================================================================-->
368<div class="doc_subsection">
Chris Lattner8d4bbbe2010-03-29 17:50:39 +0000[diff] [blame]369<a name="tce">TTA-based Codesign Environment (TCE)</a>
370</div>
371
372<div class="doc_text">
373<p>
374<a href="http://tce.cs.tut.fi/">TCE</a> is a toolset for designing
375application-specific processors (ASP) based on the Transport triggered
376architecture (TTA). The toolset provides a complete co-design flow from C/C++
377programs down to synthesizable VHDL and parallel program binaries. Processor
378customization points include the register files, function units, supported
379operations, and the interconnection network.</p>
380
381<p>TCE uses llvm-gcc/Clang and LLVM for C/C++ language support, target
382independent optimizations and also for parts of code generation. It generates
383new LLVM-based code generators "on the fly" for the designed TTA processors and
384loads them in to the compiler backend as runtime libraries to avoid per-target
385recompilation of larger parts of the compiler chain.</p>
386
387</div>
Chris Lattner8348b472008-06-08 21:58:17 +0000[diff] [blame]388
John Criswellf77cdab2010-04-06 14:52:14 +0000[diff] [blame]389<!--=========================================================================-->
390<div class="doc_subsection">
391<a name="safecode">SAFECode Compiler</a>
392</div>
393
394<div class="doc_text">
395<p>
396<a href="http://safecode.cs.illinois.edu">SAFECode</a> is a memory safe C
397compiler built using LLVM. It takes standard, unannotated C code, analyzes the
398code to ensure that memory accesses and array indexing operations are safe, and
399instruments the code with run-time checks when safety cannot be proven
400statically.
401</p>
402</div>
403
Chris Lattnerfb1a7392010-04-22 17:28:36 +0000[diff] [blame]404<!--=========================================================================-->
405<div class="doc_subsection">
406<a name="icedtea">IcedTea Java Virtual Machine Implementation</a>
407</div>
John Criswellf77cdab2010-04-06 14:52:14 +0000[diff] [blame]408
Chris Lattnerfb1a7392010-04-22 17:28:36 +0000[diff] [blame]409<div class="doc_text">
410<p>
411<a href="http://icedtea.classpath.org/wiki/Main_Page">IcedTea</a> provides a
412harness to build OpenJDK using only free software build tools and to provide
413replacements for the not-yet free parts of OpenJDK. One of the extensions that
414IcedTea provides is a new JIT compiler named <a
415href="http://icedtea.classpath.org/wiki/ZeroSharkFaq">Shark</a> which uses LLVM
416to provide native code generation without introducing processor-dependent
417code.
418</p>
419<p>Icedtea6 1.8 and later have been tested and are known to work with
420LLVM 2.7 (and continue to work with older LLVM releases >= 2.6 as well).
421</p>
422</div>
423
424<!--=========================================================================-->
425<div class="doc_subsection">
426<a name="llvm-lua">LLVM-Lua</a>
427</div>
428
429<div class="doc_text">
430<p>
431<a href="http://code.google.com/p/llvm-lua/">LLVM-Lua</a> uses LLVM
432 to add JIT and static compiling support to the Lua VM. Lua
433bytecode is analyzed to remove type checks, then LLVM is used to compile the
434bytecode down to machine code.
435</p>
436<p>LLVM-Lua 1.2.0 have been tested and is known to work with LLVM 2.7.
437</p>
438</div>
Chris Lattner3a1d4cf2010-04-22 21:34:16 +0000[diff] [blame]439
440<!--=========================================================================-->
441<div class="doc_subsection">
442<a name="MacRuby">MacRuby</a>
443</div>
444
445<div class="doc_text">
446<p>
447<a href="http://macruby.org">MacRuby</a> is an implementation of Ruby based on
448core Mac OS technologies, sponsored by Apple Inc. It uses LLVM at runtime for
449optimization passes, JIT compilation and exception handling. It also allows
450static (ahead-of-time) compilation of Ruby code straight to machine code.
451</p>
452<p>The upcoming MacRuby 0.6 release works with LLVM 2.7.
453</p>
454</div>
455
Chris Lattner7c8e7962010-04-26 17:38:10 +0000[diff] [blame]456<!--=========================================================================-->
457<div class="doc_subsection">
458<a name="GHC">Glasgow Haskell Compiler (GHC)</a>
459</div>
460
461<div class="doc_text">
462<p>
463<a href="http://www.haskell.org/ghc/">GHC</a> is an open source,
464state-of-the-art programming suite for Haskell, a standard lazy
465functional programming language. It includes an optimizing static
466compiler generating good code for a variety of platforms, together
467with an interactive system for convenient, quick development.</p>
468
469<p>In addition to the existing C and native code generators, GHC now
470supports an <a
471href="http://hackage.haskell.org/trac/ghc/wiki/Commentary/Compiler/Backends/LLVM">LLVM
472code generator</a>. GHC supports LLVM 2.7.</p>
473
474</div>
475
Chris Lattner3a1d4cf2010-04-22 21:34:16 +0000[diff] [blame]476
Chris Lattner8348b472008-06-08 21:58:17 +0000[diff] [blame]477<!-- *********************************************************************** -->
478<div class="doc_section">
Jeffrey Yasskinbec48772010-01-28 01:14:43 +0000[diff] [blame]479 <a name="whatsnew">What's New in LLVM 2.7?</a>
Chris Lattner8348b472008-06-08 21:58:17 +0000[diff] [blame]480</div>
481<!-- *********************************************************************** -->
482
Chris Lattnerf8e0b4e2008-06-08 22:59:35 +0000[diff] [blame]483<div class="doc_text">
484
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]485<p>This release includes a huge number of bug fixes, performance tweaks and
Chris Lattnerf6662f92008-10-13 17:57:36 +0000[diff] [blame]486minor improvements. Some of the major improvements and new features are listed
487in this section.
Chris Lattnerf8e0b4e2008-06-08 22:59:35 +0000[diff] [blame]488</p>
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]489
Chris Lattner914ce462010-04-22 06:28:20 +0000[diff] [blame]490</div>
491
492<!--=========================================================================-->
493<div class="doc_subsection">
494<a name="orgchanges">LLVM Community Changes</a>
495</div>
496
497<div class="doc_text">
498
Chris Lattnerb7c85b42010-04-21 05:17:40 +0000[diff] [blame]499<p>In addition to changes to the code, between LLVM 2.6 and 2.7, a number of
500organization changes have happened:
501</p>
502
503<ul>
Chris Lattner7b91eda2010-04-22 05:41:35 +0000[diff] [blame]504<li>LLVM has a new <a href="http://llvm.org/Logo.html">official logo</a>!</li>
Chris Lattnerb7c85b42010-04-21 05:17:40 +0000[diff] [blame]505
Chris Lattnerb7c85b42010-04-21 05:17:40 +0000[diff] [blame]506<li>Ted Kremenek and Doug Gregor have stepped forward as <a
507 href="http://llvm.org/docs/DeveloperPolicy.html#owners">Code Owners</a> of the
Chris Lattner914ce462010-04-22 06:28:20 +0000[diff] [blame]508 Clang static analyzer and the Clang frontend, respectively.</li>
Chris Lattner450a31e2010-04-21 06:23:40 +0000[diff] [blame]509
510<li>LLVM now has an <a href="http://blog.llvm.org">official Blog</a> at
511 <a href="http://blog.llvm.org">http://blog.llvm.org</a>. This is a great way
512 to learn about new LLVM-related features as they are implemented. Several
513 features in this release are already explained on the blog.</li>
514
Chris Lattnerb7c85b42010-04-21 05:17:40 +0000[diff] [blame]515<li>The LLVM web pages are now checked into the SVN server, in the "www",
516 "www-pubs" and "www-releases" SVN modules. Previously they were hidden in a
Chris Lattner7b91eda2010-04-22 05:41:35 +0000[diff] [blame]517 largely inaccessible old CVS server.</li>
Chris Lattner450a31e2010-04-21 06:23:40 +0000[diff] [blame]518
519<li><a href="http://llvm.org">llvm.org</a> is now hosted on a new (and much
520 faster) server. It is still graciously hosted at the University of Illinois
521 of Urbana Champaign.</li>
Chris Lattnerb7c85b42010-04-21 05:17:40 +0000[diff] [blame]522</ul>
Chris Lattnerf8e0b4e2008-06-08 22:59:35 +0000[diff] [blame]523</div>
524
Chris Lattner8348b472008-06-08 21:58:17 +0000[diff] [blame]525<!--=========================================================================-->
Chris Lattnerea34f642008-06-08 21:34:41 +0000[diff] [blame]526<div class="doc_subsection">
Chris Lattner252b83d2008-02-06 18:00:06 +0000[diff] [blame]527<a name="majorfeatures">Major New Features</a>
528</div>
529
530<div class="doc_text">
531
Jeffrey Yasskinbec48772010-01-28 01:14:43 +0000[diff] [blame]532<p>LLVM 2.7 includes several major new capabilities:</p>
Chris Lattnerf3e5bc62007-05-14 06:56:09 +0000[diff] [blame]533
534<ul>
Chris Lattner7b91eda2010-04-22 05:41:35 +0000[diff] [blame]535<li>2.7 includes initial support for the <a
Chris Lattner450a31e2010-04-21 06:23:40 +0000[diff] [blame]536 href="http://en.wikipedia.org/wiki/MicroBlaze">MicroBlaze</a> target.
537 MicroBlaze is a soft processor core designed for Xilinx FPGAs.</li>
Chris Lattnerb7c85b42010-04-21 05:17:40 +0000[diff] [blame]538
Chris Lattner7b91eda2010-04-22 05:41:35 +0000[diff] [blame]539<li>2.7 includes a new LLVM IR "extensible metadata" feature. This feature
Chris Lattner450a31e2010-04-21 06:23:40 +0000[diff] [blame]540 supports many different use cases, including allowing front-end authors to
541 encode source level information into LLVM IR, which is consumed by later
542 language-specific passes. This is a great way to do high-level optimizations
543 like devirtualization, type-based alias analysis, etc. See the <a
544 href="http://blog.llvm.org/2010/04/extensible-metadata-in-llvm-ir.html">
545 Extensible Metadata Blog Post</a> for more information.</li>
546
Chris Lattner7b91eda2010-04-22 05:41:35 +0000[diff] [blame]547<li>2.7 encodes <a href="SourceLevelDebugging.html">debug information</a>
Chris Lattner450a31e2010-04-21 06:23:40 +0000[diff] [blame]548in a completely new way, built on extensible metadata. The new implementation
549is much more memory efficient and paves the way for improvements to optimized
550code debugging experience.</li>
Chris Lattnerb7c85b42010-04-21 05:17:40 +0000[diff] [blame]551
Chris Lattner7b91eda2010-04-22 05:41:35 +0000[diff] [blame]552<li>2.7 now directly supports taking the address of a label and doing an
Chris Lattner450a31e2010-04-21 06:23:40 +0000[diff] [blame]553 indirect branch through a pointer. This is particularly useful for
554 interpreter loops, and is used to implement the GCC "address of label"
555 extension. For more information, see the <a
556href="http://blog.llvm.org/2010/01/address-of-label-and-indirect-branches.html">
557Address of Label and Indirect Branches in LLVM IR Blog Post</a>.
Chris Lattnerb7c85b42010-04-21 05:17:40 +0000[diff] [blame]558
Chris Lattner7b91eda2010-04-22 05:41:35 +0000[diff] [blame]559<li>2.7 is the first release to start supporting APIs for assembling and
Chris Lattner450a31e2010-04-21 06:23:40 +0000[diff] [blame]560 disassembling target machine code. These APIs are useful for a variety of
561 low level clients, and are surfaced in the new "enhanced disassembly" API.
562 For more information see the <a
563 href="http://blog.llvm.org/2010/01/x86-disassembler.html">The X86
564 Disassembler Blog Post</a> for more information.</li>
Chris Lattnerb7c85b42010-04-21 05:17:40 +0000[diff] [blame]565
Chris Lattner8cdd7932010-04-22 06:38:11 +0000[diff] [blame]566<li>2.7 includes major parts of the work required by the new MC Project,
567 see the <a href="#mc">MC update above</a> for more information.</li>
Chris Lattnerb7c85b42010-04-21 05:17:40 +0000[diff] [blame]568
Chris Lattner8170c102008-02-10 08:18:42 +0000[diff] [blame]569</ul>
Chris Lattnerdc910082010-03-17 06:41:58 +0000[diff] [blame]570
Chris Lattnerf3e5bc62007-05-14 06:56:09 +0000[diff] [blame]571</div>
572
Chris Lattner84977642007-09-21 03:54:09 +0000[diff] [blame]573<!--=========================================================================-->
574<div class="doc_subsection">
Chris Lattner917cc712009-03-02 02:37:32 +0000[diff] [blame]575<a name="coreimprovements">LLVM IR and Core Improvements</a>
Chris Lattnerf304ffc2008-02-10 08:17:19 +0000[diff] [blame]576</div>
577
578<div class="doc_text">
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]579<p>LLVM IR has several new features for better support of new targets and that
580expose new optimization opportunities:</p>
Chris Lattnerf304ffc2008-02-10 08:17:19 +0000[diff] [blame]581
Chris Lattner791f77b2008-06-05 06:25:56 +0000[diff] [blame]582<ul>
Chris Lattner7b91eda2010-04-22 05:41:35 +0000[diff] [blame]583<li>LLVM IR now supports a 16-bit "half float" data type through <a
584 href="LangRef.html#int_fp16">two new intrinsics</a> and APFloat support.</li>
585<li>LLVM IR supports two new <a href="LangRef.html#fnattrs">function
Chris Lattner450a31e2010-04-21 06:23:40 +0000[diff] [blame]586 attributes</a>: inlinehint and alignstack(n). The former is a hint to the
Chris Lattner7b91eda2010-04-22 05:41:35 +0000[diff] [blame]587 optimizer that a function was declared 'inline' and thus the inliner should
588 weight it higher when considering inlining it. The later
Chris Lattner450a31e2010-04-21 06:23:40 +0000[diff] [blame]589 indicates to the code generator that the function diverges from the platform
590 ABI on stack alignment.</li>
591<li>The new <a href="LangRef.html#int_objectsize">llvm.objectsize</a> intrinsic
592 allows the optimizer to infer the sizes of memory objects in some cases.
Chris Lattner7b91eda2010-04-22 05:41:35 +0000[diff] [blame]593 This intrinsic is used to implement the GCC <tt>__builtin_object_size</tt>
Chris Lattner450a31e2010-04-21 06:23:40 +0000[diff] [blame]594 extension.</li>
Chris Lattner914ce462010-04-22 06:28:20 +0000[diff] [blame]595<li>LLVM IR now supports marking load and store instructions with <a
596 href="LangRef.html#i_load">"non-temporal" hints</a> (building on the new
597 metadata feature). This hint encourages the code
598 generator to generate non-temporal accesses when possible, which are useful
599 for code that is carefully managing cache behavior. Currently, only the
600 X86 backend provides target support for this feature.</li>
601
Chris Lattner450a31e2010-04-21 06:23:40 +0000[diff] [blame]602<li>LLVM 2.7 has pre-alpha support for <a
Gabor Greifc8b3af92010-04-22 10:11:24 +0000[diff] [blame]603 href="LangRef.html#t_union">unions in LLVM IR</a>.
Chris Lattner914ce462010-04-22 06:28:20 +0000[diff] [blame]604 Unfortunately, this support is not really usable in 2.7, so if you're
Chris Lattner450a31e2010-04-21 06:23:40 +0000[diff] [blame]605 interested in pushing it forward, please help contribute to LLVM mainline.</li>
Gabor Greifee2187a2010-04-22 10:21:43 +0000[diff] [blame]606
Chris Lattnerf304ffc2008-02-10 08:17:19 +0000[diff] [blame]607</ul>
Mikhail Glushenkovea65d7d2008-10-13 02:08:34 +0000[diff] [blame]608
Chris Lattnerf304ffc2008-02-10 08:17:19 +0000[diff] [blame]609</div>
610
611<!--=========================================================================-->
612<div class="doc_subsection">
Chris Lattneracce85d2008-02-10 07:46:44 +0000[diff] [blame]613<a name="optimizer">Optimizer Improvements</a>
614</div>
615
616<div class="doc_text">
617
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]618<p>In addition to a large array of minor performance tweaks and bug fixes, this
Chris Lattnerf3013872008-10-13 21:50:36 +0000[diff] [blame]619release includes a few major enhancements and additions to the optimizers:</p>
Chris Lattneracce85d2008-02-10 07:46:44 +0000[diff] [blame]620
621<ul>
622
Chris Lattner77640fe2010-04-27 07:28:11 +0000[diff] [blame]623<li>The inliner now merges arrays stack objects in different callees when
Chris Lattner7b91eda2010-04-22 05:41:35 +0000[diff] [blame]624 inlining multiple call sites into one function. This reduces the stack size
625 of the resultant function.</li>
626<li>The -basicaa alias analysis pass (which is the default) has been improved to
627 be less dependent on "type safe" pointers. It can now look through bitcasts
628 and other constructs more aggressively, allowing better load/store
629 optimization.</li>
630<li>The load elimination optimization in the GVN Pass [<a
631href="http://blog.llvm.org/2009/12/introduction-to-load-elimination-in-gvn.html">intro
632 blog post</a>] has been substantially improved to be more aggressive about
633 partial redundancy elimination and do more aggressive phi translation. Please
634 see the <a
635 href="http://blog.llvm.org/2009/12/advanced-topics-in-redundant-load.html">
636 Advanced Topics in Redundant Load Elimination with a Focus on PHI Translation
637 Blog Post</a> for more details.</li>
638<li>The module <a href="LangRef.html#datalayout">target data string</a> now
Wesley Peck68d1fe92010-04-22 13:28:34 +0000[diff] [blame]639 includes a notion of 'native' integer data types for the target. This
640 helps mid-level optimizations avoid promoting complex sequences of
641 operations to data types that are not natively supported (e.g. converting
642 i32 operations to i64 on 32-bit chips).</li>
Chris Lattner7b91eda2010-04-22 05:41:35 +0000[diff] [blame]643<li>The mid-level optimizer is now conservative when operating on a module with
644 no target data. Previously, it would default to SparcV9 settings, which is
645 not what most people expected.</li>
Chris Lattnera54c1f72010-04-21 06:42:24 +0000[diff] [blame]646<li>Jump threading is now much more aggressive at simplifying correlated
Chris Lattner7b91eda2010-04-22 05:41:35 +0000[diff] [blame]647 conditionals and threading blocks with otherwise complex logic. It has
648 subsumed the old "Conditional Propagation" pass, and -condprop has been
649 removed from LLVM 2.7.</li>
650<li>The -instcombine pass has been refactored from being one huge file to being
651 a library of its own. Internally, it uses a customized IRBuilder to clean
652 it up and simplify it.</li>
Chris Lattnerb7c85b42010-04-21 05:17:40 +0000[diff] [blame]653
Chris Lattner7b91eda2010-04-22 05:41:35 +0000[diff] [blame]654<li>The optimal edge profiling pass is reliable and much more complete than in
655 2.6. It can be used with the llvm-prof tool but isn't wired up to the
656 llvm-gcc and clang command line options yet.</li>
Chris Lattner917cc712009-03-02 02:37:32 +0000[diff] [blame]657
Chris Lattner7b91eda2010-04-22 05:41:35 +0000[diff] [blame]658<li>A new experimental alias analysis implementation, -scev-aa, has been added.
659 It uses LLVM's Scalar Evolution implementation to do symbolic analysis of
660 pointer offset expressions to disambiguate pointers. It can catch a few
661 cases that basicaa cannot, particularly in complex loop nests.</li>
Chris Lattnera54c1f72010-04-21 06:42:24 +0000[diff] [blame]662
Chris Lattner914ce462010-04-22 06:28:20 +0000[diff] [blame]663<li>The default pass ordering has been tweaked for improved optimization
Chris Lattner7b91eda2010-04-22 05:41:35 +0000[diff] [blame]664 effectiveness.</li>
Chris Lattnera54c1f72010-04-21 06:42:24 +0000[diff] [blame]665
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]666</ul>
Chris Lattner917cc712009-03-02 02:37:32 +0000[diff] [blame]667
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]668</div>
Chris Lattner917cc712009-03-02 02:37:32 +0000[diff] [blame]669
Chris Lattner917cc712009-03-02 02:37:32 +0000[diff] [blame]670
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]671<!--=========================================================================-->
672<div class="doc_subsection">
673<a name="executionengine">Interpreter and JIT Improvements</a>
674</div>
Chris Lattnerf3013872008-10-13 21:50:36 +0000[diff] [blame]675
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]676<div class="doc_text">
677
678<ul>
Wesley Peck640604a2010-04-22 13:36:27 +0000[diff] [blame]679<li>The JIT now supports generating debug information and is compatible with
680the new GDB 7.0 (and later) interfaces for registering dynamically generated
681debug info.</li>
Chris Lattnerb7c85b42010-04-21 05:17:40 +0000[diff] [blame]682
Chris Lattner914ce462010-04-22 06:28:20 +0000[diff] [blame]683<li>The JIT now <a href="http://llvm.org/PR5184">defaults
Jeffrey Yasskin01eba392010-01-29 19:10:38 +0000[diff] [blame]684to compiling eagerly</a> to avoid a race condition in the lazy JIT.
685Clients that still want the lazy JIT can switch it on by calling
686<tt>ExecutionEngine::DisableLazyCompilation(false)</tt>.</li>
Chris Lattnerb7c85b42010-04-21 05:17:40 +0000[diff] [blame]687
Jeffrey Yasskin40966a72010-02-11 01:07:39 +0000[diff] [blame]688<li>It is now possible to create more than one JIT instance in the same process.
689These JITs can generate machine code in parallel,
690although <a href="http://llvm.org/docs/ProgrammersManual.html#jitthreading">you
691still have to obey the other threading restrictions</a>.</li>
Chris Lattnerb7c85b42010-04-21 05:17:40 +0000[diff] [blame]692
Chris Lattner0b832202008-06-08 02:45:07 +0000[diff] [blame]693</ul>
694
695</div>
696
697<!--=========================================================================-->
698<div class="doc_subsection">
Chris Lattner511433e2009-03-02 03:24:11 +0000[diff] [blame]699<a name="codegen">Target Independent Code Generator Improvements</a>
Chris Lattner0b832202008-06-08 02:45:07 +0000[diff] [blame]700</div>
701
702<div class="doc_text">
703
Mikhail Glushenkovf795ef02009-03-01 18:09:47 +0000[diff] [blame]704<p>We have put a significant amount of work into the code generator
705infrastructure, which allows us to implement more aggressive algorithms and make
706it run faster:</p>
Chris Lattner0b832202008-06-08 02:45:07 +0000[diff] [blame]707
708<ul>
Chris Lattner914ce462010-04-22 06:28:20 +0000[diff] [blame]709<li>The 'llc -asm-verbose' option (which is now the default) has been enhanced
710 to emit many useful comments to .s files indicating information about spill
711 slots and loop nest structure. This should make it much easier to read and
712 understand assembly files. This is wired up in llvm-gcc and clang to
713 the <tt>-fverbose-asm</tt> option.</li>
714
Dan Gohman9c675f12010-04-22 20:50:43 +0000[diff] [blame]715<li>New LSR with "full strength reduction" mode, which can reduce address
716 register pressure in loops where address generation is important.</li>
Chris Lattner914ce462010-04-22 06:28:20 +0000[diff] [blame]717
718<li>A new codegen level Common Subexpression Elimination pass (MachineCSE)
719 is available and enabled by default. It catches redundancies exposed by
720 lowering.</li>
721<li>A new pre-register-allocation tail duplication pass is available and enabled
722 by default, it can substantially improve branch prediction quality in some
723 cases.</li>
724<li>A new sign and zero extension optimization pass (OptimizeExtsPass)
725 is available and enabled by default. This pass can takes advantage
726 architecture features like x86-64 implicit zero extension behavior and
727 sub-registers.</li>
728<li>The code generator now supports a mode where it attempts to preserve the
729 order of instructions in the input code. This is important for source that
730 is hand scheduled and extremely sensitive to scheduling. It is compatible
731 with the GCC <tt>-fno-schedule-insns</tt> option.</li>
732<li>The target-independent code generator now supports generating code with
733 arbitrary numbers of result values. Returning more values than was
734 previously supported is handled by returning through a hidden pointer. In
735 2.7, only the X86 and XCore targets have adopted support for this
736 though.</li>
737<li>The code generator now supports generating code that follows the
738 <a href="LangRef.html#callingconv">Glasgow Haskell Compiler Calling
739 Convention</a> and ABI.</li>
740<li>The "<a href="CodeGenerator.html#selectiondag_select">DAG instruction
741 selection</a>" phase of the code generator has been largely rewritten for
742 2.7. Previously, tblgen spit out tons of C++ code which was compiled and
743 linked into the target to do the pattern matching, now it emits a much
744 smaller table which is read by the target-independent code. The primary
745 advantages of this approach is that the size and compile time of various
746 targets is much improved. The X86 code generator shrunk by 1.5MB of code,
747 for example.</li>
748<li>Almost the entire code generator has switched to emitting code through the
749 MC interfaces instead of printing textually to the .s file. This led to a
750 number of cleanups and speedups. In 2.7, debug an exception handling
751 information does not go through MC yet.</li>
Chris Lattner84977642007-09-21 03:54:09 +0000[diff] [blame]752</ul>
Chris Lattner84977642007-09-21 03:54:09 +0000[diff] [blame]753</div>
754
Chris Lattner917cc712009-03-02 02:37:32 +0000[diff] [blame]755<!--=========================================================================-->
756<div class="doc_subsection">
757<a name="x86">X86-32 and X86-64 Target Improvements</a>
758</div>
759
760<div class="doc_text">
761<p>New features of the X86 target include:
762</p>
763
764<ul>
Chris Lattnera54c1f72010-04-21 06:42:24 +0000[diff] [blame]765<li>The X86 backend now optimizes tails calls much more aggressively for
766 functions that use the standard C calling convention.</li>
767<li>The X86 backend now models scalar SSE registers as subregs of the SSE vector
768 registers, making the code generator more aggressive in cases where scalars
769 and vector types are mixed.</li>
Chris Lattner511433e2009-03-02 03:24:11 +0000[diff] [blame]770
Chris Lattner917cc712009-03-02 02:37:32 +0000[diff] [blame]771</ul>
772
Chris Lattner917cc712009-03-02 02:37:32 +0000[diff] [blame]773</div>
Chris Lattner84977642007-09-21 03:54:09 +0000[diff] [blame]774
775<!--=========================================================================-->
776<div class="doc_subsection">
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]777<a name="ARM">ARM Target Improvements</a>
Chris Lattnerc441fb82009-03-01 02:30:21 +0000[diff] [blame]778</div>
779
780<div class="doc_text">
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]781<p>New features of the ARM target include:
782</p>
Chris Lattnerc441fb82009-03-01 02:30:21 +0000[diff] [blame]783
784<ul>
Chris Lattnerc441fb82009-03-01 02:30:21 +0000[diff] [blame]785
Chris Lattner450a31e2010-04-21 06:23:40 +0000[diff] [blame]786<li>The ARM backend now generates instructions in unified assembly syntax.</li>
787
788<li>llvm-gcc now has complete support for the ARM v7 NEON instruction set. This
789 support differs slightly from the GCC implementation. Please see the
790 <a
791href="http://blog.llvm.org/2010/04/arm-advanced-simd-neon-intrinsics-and.html">
792 ARM Advanced SIMD (NEON) Intrinsics and Types in LLVM Blog Post</a> for
793 helpful information if migrating code from GCC to LLVM-GCC.</li>
794
Wesley Peck9e385b12010-04-22 13:43:14 +0000[diff] [blame]795<li>The ARM and Thumb code generators now use register scavenging for stack
Jim Grosbach269e0fe2010-04-22 18:28:43 +0000[diff] [blame]796 object address materialization. This allows the use of R3 as a general
797 purpose register in Thumb1 code, as it was previous reserved for use in
798 stack address materialization. Secondly, sequential uses of the same
799 value will now re-use the materialized constant.</li>
800
Wesley Peck9e385b12010-04-22 13:43:14 +0000[diff] [blame]801<li>The ARM backend now has good support for ARMv4 targets and has been tested
Chris Lattner450a31e2010-04-21 06:23:40 +0000[diff] [blame]802 on StrongARM hardware. Previously, LLVM only supported ARMv4T and
803 newer chips.</li>
Jim Grosbach41a77662010-04-22 18:33:31 +0000[diff] [blame]804
805<li>Atomic builtins are now supported for ARMv6 and ARMv7 (__sync_synchronize,
806 __sync_fetch_and_add, etc.).</li>
807
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]808</ul>
Chris Lattnerc441fb82009-03-01 02:30:21 +0000[diff] [blame]809
Chris Lattnerc441fb82009-03-01 02:30:21 +0000[diff] [blame]810
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]811</div>
Chris Lattnerc441fb82009-03-01 02:30:21 +0000[diff] [blame]812
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]813<!--=========================================================================-->
814<div class="doc_subsection">
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]815<a name="newapis">New Useful APIs</a>
816</div>
817
818<div class="doc_text">
819
820<p>This release includes a number of new APIs that are used internally, which
821 may also be useful for external clients.
822</p>
823
824<ul>
Chris Lattner7b91eda2010-04-22 05:41:35 +0000[diff] [blame]825<li>The optimizer uses the new CodeMetrics class to measure the size of code.
Chris Lattner83cd35a2010-04-22 17:39:38 +0000[diff] [blame]826 Various passes (like the inliner, loop unswitcher, etc) all use this to make
827 more accurate estimates of the code size impact of various
Chris Lattner7b91eda2010-04-22 05:41:35 +0000[diff] [blame]828 optimizations.</li>
829<li>A new <a href="http://llvm.org/doxygen/InstructionSimplify_8h-source.html">
Wesley Peckc0287ba2010-04-22 13:50:46 +0000[diff] [blame]830 llvm/Analysis/InstructionSimplify.h</a> interface is available for doing
Chris Lattner7b91eda2010-04-22 05:41:35 +0000[diff] [blame]831 symbolic simplification of instructions (e.g. <tt>a+0</tt> -&gt; <tt>a</tt>)
832 without requiring the instruction to exist. This centralizes a lot of
833 ad-hoc symbolic manipulation code scattered in various passes.</li>
834<li>The optimizer now uses a new <a
835 href="http://llvm.org/doxygen/SSAUpdater_8h-source.html">SSAUpdater</a>
836 class which efficiently supports
837 doing unstructured SSA update operations. This centralized a bunch of code
Wesley Peckc0287ba2010-04-22 13:50:46 +0000[diff] [blame]838 scattered throughout various passes (e.g. jump threading, lcssa,
839 loop rotate, etc) for doing this sort of thing. The code generator has a
840 similar <a href="http://llvm.org/doxygen/MachineSSAUpdater_8h-source.html">
Chris Lattner7b91eda2010-04-22 05:41:35 +0000[diff] [blame]841 MachineSSAUpdater</a> class.</li>
842<li>The <a href="http://llvm.org/doxygen/Regex_8h-source.html">
843 llvm/Support/Regex.h</a> header exposes a platform independent regular
Chris Lattner450a31e2010-04-21 06:23:40 +0000[diff] [blame]844 expression API. Building on this, the <a
845 href="TestingGuide.html#FileCheck">FileCheck</a> utility now supports
846 regular exressions.</li>
847<li>raw_ostream now supports a circular "debug stream" accessed with "dbgs()".
848 By default, this stream works the same way as "errs()", but if you pass
849 <tt>-debug-buffer-size=1000</tt> to opt, the debug stream is capped to a
850 fixed sized circular buffer and the output is printed at the end of the
851 program's execution. This is helpful if you have a long lived compiler
852 process and you're interested in seeing snapshots in time.</li>
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]853</ul>
854
855
856</div>
857
858<!--=========================================================================-->
859<div class="doc_subsection">
860<a name="otherimprovements">Other Improvements and New Features</a>
861</div>
862
863<div class="doc_text">
864<p>Other miscellaneous features include:</p>
865
866<ul>
Chris Lattner450a31e2010-04-21 06:23:40 +0000[diff] [blame]867<li>You can now build LLVM as a big dynamic library (e.g. "libllvm2.7.so"). To
868 get this, configure LLVM with the --enable-shared option.</li>
869
Wesley Pecka0c44842010-04-22 13:53:54 +0000[diff] [blame]870<li>LLVM command line tools now overwrite their output by default. Previously,
871 they would only do this with -f. This makes them more convenient to use, and
Chris Lattnerb7c85b42010-04-21 05:17:40 +0000[diff] [blame]872 behave more like standard unix tools.</li>
873
874<li>The opt and llc tools now autodetect whether their input is a .ll or .bc
875 file, and automatically do the right thing. This means you don't need to
876 explicitly use the llvm-as tool for most things.</li>
Chris Lattnerc441fb82009-03-01 02:30:21 +0000[diff] [blame]877</ul>
878
879</div>
880
Chris Lattner77d29b12008-06-05 08:02:49 +0000[diff] [blame]881
882<!--=========================================================================-->
883<div class="doc_subsection">
Chris Lattnerf6662f92008-10-13 17:57:36 +0000[diff] [blame]884<a name="changes">Major Changes and Removed Features</a>
885</div>
886
887<div class="doc_text">
888
Chris Lattnereeb4da02008-10-13 22:06:31 +0000[diff] [blame]889<p>If you're already an LLVM user or developer with out-of-tree changes based
Jeffrey Yasskinbec48772010-01-28 01:14:43 +0000[diff] [blame]890on LLVM 2.6, this section lists some "gotchas" that you may run into upgrading
Chris Lattnereeb4da02008-10-13 22:06:31 +0000[diff] [blame]891from the previous release.</p>
Chris Lattnerf6662f92008-10-13 17:57:36 +0000[diff] [blame]892
893<ul>
Chris Lattnerb7c85b42010-04-21 05:17:40 +0000[diff] [blame]894
895<li>
896The Andersen's alias analysis ("anders-aa") pass, the Predicate Simplifier
897("predsimplify") pass, the LoopVR pass, the GVNPRE pass, and the random sampling
898profiling ("rsprofiling") passes have all been removed. They were not being
899actively maintained and had substantial problems. If you are interested in
900these components, you are welcome to ressurect them from SVN, fix the
901correctness problems, and resubmit them to mainline.</li>
902
903<li>LLVM now defaults to building most libraries with RTTI turned off, providing
904a code size reduction. Packagers who are interested in building LLVM to support
905plugins that require RTTI information should build with "make REQUIRE_RTTI=1"
906and should read the new <a href="Packaging.html">Advice on Packaging LLVM</a>
907document.</li>
908
Jeffrey Yasskinbc83d062010-02-09 23:03:44 +0000[diff] [blame]909<li>The LLVM interpreter now defaults to <em>not</em> using <tt>libffi</tt> even
910if you have it installed. This makes it more likely that an LLVM built on one
911system will work when copied to a similar system. To use <tt>libffi</tt>,
Chris Lattner914ce462010-04-22 06:28:20 +0000[diff] [blame]912configure with <tt>--enable-libffi</tt>.</li>
Chris Lattnereeb4da02008-10-13 22:06:31 +0000[diff] [blame]913
Chris Lattnerb7c85b42010-04-21 05:17:40 +0000[diff] [blame]914<li>Debug information uses a completely different representation, an LLVM 2.6
915.bc file should work with LLVM 2.7, but debug info won't come forward.</li>
916
917<li>The LLVM 2.6 (and earlier) "malloc" and "free" instructions got removed,
918 along with LowerAllocations pass. Now you should just use a call to the
919 malloc and free functions in libc. These calls are optimized as well as
920 the old instructions were.</li>
921</ul>
Chris Lattnereeb4da02008-10-13 22:06:31 +0000[diff] [blame]922
923<p>In addition, many APIs have changed in this release. Some of the major LLVM
924API changes are:</p>
925
926<ul>
Jeffrey Yasskin4fcd6072010-01-28 01:41:20 +0000[diff] [blame]927
Dan Gohmanb44f6c62010-05-03 23:51:05 +0000[diff] [blame]928<li>The <tt>add</tt>, <tt>sub</tt>, and <tt>mul</tt> instructions no longer
929support floating-point operands. The <tt>fadd</tt>, <tt>fsub</tt>, and
930<tt>fmul</tt> instructions should be used for this purpose instead.</li>
Daniel Dunbar4acdede2010-02-10 04:09:52 +0000[diff] [blame]931
Devang Patelb34dd132008-10-14 20:03:43 +0000[diff] [blame]932</ul>
Chris Lattnerf6662f92008-10-13 17:57:36 +0000[diff] [blame]933
Chris Lattnerf6662f92008-10-13 17:57:36 +0000[diff] [blame]934</div>
935
936
937
Chris Lattner19092612003-10-02 16:38:05 +0000[diff] [blame]938<!-- *********************************************************************** -->
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]939<div class="doc_section">
940 <a name="portability">Portability and Supported Platforms</a>
941</div>
Chris Lattner19092612003-10-02 16:38:05 +0000[diff] [blame]942<!-- *********************************************************************** -->
943
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]944<div class="doc_text">
945
John Criswell0b5b5e92004-12-08 20:35:47 +0000[diff] [blame]946<p>LLVM is known to work on the following platforms:</p>
Chris Lattner4654bdb2004-06-01 18:22:41 +0000[diff] [blame]947
948<ul>
Mikhail Glushenkovf795ef02009-03-01 18:09:47 +0000[diff] [blame]949<li>Intel and AMD machines (IA32, X86-64, AMD64, EMT-64) running Red Hat
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]950 Linux, Fedora Core, FreeBSD and AuroraUX (and probably other unix-like
951 systems).</li>
Chris Lattner914ce462010-04-22 06:28:20 +0000[diff] [blame]952<li>PowerPC and X86-based Mac OS X systems, running 10.4 and above in 32-bit
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]953 and 64-bit modes.</li>
Chris Lattner252b83d2008-02-06 18:00:06 +0000[diff] [blame]954<li>Intel and AMD machines running on Win32 using MinGW libraries (native).</li>
Chris Lattner57a460e2007-05-23 04:39:32 +0000[diff] [blame]955<li>Intel and AMD machines running on Win32 with the Cygwin libraries (limited
956 support is available for native builds with Visual C++).</li>
Chris Lattner7e23d6e2009-10-16 16:30:58 +0000[diff] [blame]957<li>Sun x86 and AMD64 machines running Solaris 10, OpenSolaris 0906.</li>
John Criswell9321fa82005-05-13 20:28:15 +0000[diff] [blame]958<li>Alpha-based machines running Debian GNU/Linux.</li>
Chris Lattner4654bdb2004-06-01 18:22:41 +0000[diff] [blame]959</ul>
960
Chris Lattnerbc5786b2008-06-05 06:57:39 +0000[diff] [blame]961<p>The core LLVM infrastructure uses GNU autoconf to adapt itself
Brian Gaekeb0fd7612004-05-09 05:28:35 +0000[diff] [blame]962to the machine and operating system on which it is built. However, minor
963porting may be required to get LLVM to work on new platforms. We welcome your
964portability patches and reports of successful builds or error messages.</p>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]965
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]966</div>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]967
968<!-- *********************************************************************** -->
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]969<div class="doc_section">
970 <a name="knownproblems">Known Problems</a>
971</div>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]972<!-- *********************************************************************** -->
973
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]974<div class="doc_text">
975
Mikhail Glushenkovf795ef02009-03-01 18:09:47 +0000[diff] [blame]976<p>This section contains significant known problems with the LLVM system,
Chris Lattnere18b32e2008-11-10 05:40:34 +0000[diff] [blame]977listed by component. If you run into a problem, please check the <a
Chris Lattnerc463b272005-10-29 07:07:09 +0000[diff] [blame]978href="http://llvm.org/bugs/">LLVM bug database</a> and submit a bug if
Chris Lattner5eccca42003-12-12 21:22:16 +0000[diff] [blame]979there isn't already one.</p>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]980
Chris Lattner477a1fd2010-03-17 17:25:49 +0000[diff] [blame]981<ul>
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]982<li>LLVM will not correctly compile on Solaris and/or OpenSolaris
983using the stock GCC 3.x.x series 'out the box',
Chris Lattner554ee4a2009-11-03 21:50:09 +0000[diff] [blame]984See: <a href="GettingStarted.html#brokengcc">Broken versions of GCC and other tools</a>.
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]985However, A <a href="http://pkg.auroraux.org/GCC">Modern GCC Build</a>
986for x86/x86-64 has been made available from the third party AuroraUX Project
987that has been meticulously tested for bootstrapping LLVM &amp; Clang.</li>
988</ul>
989
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]990</div>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]991
Chris Lattnerf5ee1702004-03-14 02:03:02 +0000[diff] [blame]992<!-- ======================================================================= -->
993<div class="doc_subsection">
994 <a name="experimental">Experimental features included with this release</a>
995</div>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]996
Chris Lattnerf5ee1702004-03-14 02:03:02 +0000[diff] [blame]997<div class="doc_text">
998
Misha Brukman6df9e2c2004-05-12 21:46:05 +0000[diff] [blame]999<p>The following components of this LLVM release are either untested, known to
1000be broken or unreliable, or are in early development. These components should
1001not be relied on, and bugs should not be filed against them, but they may be
1002useful to some people. In particular, if you would like to work on one of these
Chris Lattner2b659ef2008-02-12 06:29:45 +0000[diff] [blame]1003components, please contact us on the <a
1004href="http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev">LLVMdev list</a>.</p>
Chris Lattnerf5ee1702004-03-14 02:03:02 +0000[diff] [blame]1005
1006<ul>
Wesley Peck7c4a1212010-03-18 14:31:30 +0000[diff] [blame]1007<li>The MSIL, Alpha, SPU, MIPS, PIC16, Blackfin, MSP430, SystemZ and MicroBlaze
1008 backends are experimental.</li>
Chris Lattner914ce462010-04-22 06:28:20 +0000[diff] [blame]1009<li><tt>llc</tt> "<tt>-filetype=asm</tt>" (the default) is the only
Chris Lattnerc66bfef2010-03-17 04:41:49 +0000[diff] [blame]1010 supported value for this option. The MachO writer is experimental, and
1011 works much better in mainline SVN.</li>
Chris Lattnerf5ee1702004-03-14 02:03:02 +0000[diff] [blame]1012</ul>
1013
1014</div>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]1015
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1016<!-- ======================================================================= -->
1017<div class="doc_subsection">
Chris Lattner26299222006-11-18 07:51:14 +0000[diff] [blame]1018 <a name="x86-be">Known problems with the X86 back-end</a>
John Criswellc0c186d2005-11-08 21:11:33 +0000[diff] [blame]1019</div>
1020
1021<div class="doc_text">
1022
1023<ul>
Anton Korobeynikova6094be2008-06-08 10:24:13 +0000[diff] [blame]1024 <li>The X86 backend does not yet support
1025 all <a href="http://llvm.org/PR879">inline assembly that uses the X86
1026 floating point stack</a>. It supports the 'f' and 't' constraints, but not
1027 'u'.</li>
Chris Lattnere6e1b352008-06-08 21:19:07 +0000[diff] [blame]1028 <li>The X86 backend generates inefficient floating point code when configured
1029 to generate code for systems that don't have SSE2.</li>
Duncan Sands47eff2b2008-06-08 19:38:43 +0000[diff] [blame]1030 <li>Win64 code generation wasn't widely tested. Everything should work, but we
John Criswell524a5dd2009-03-02 15:28:15 +0000[diff] [blame]1031 expect small issues to happen. Also, llvm-gcc cannot build the mingw64
Chris Lattner914ce462010-04-22 06:28:20 +0000[diff] [blame]1032 runtime currently due to lack of support for the 'u' inline assembly
1033 constraint and for X87 floating point inline assembly.</li>
Dan Gohman8207ba92008-06-08 23:05:11 +0000[diff] [blame]1034 <li>The X86-64 backend does not yet support the LLVM IR instruction
Chris Lattner914ce462010-04-22 06:28:20 +0000[diff] [blame]1035 <tt>va_arg</tt>. Currently, front-ends support variadic
Dan Gohman8207ba92008-06-08 23:05:11 +0000[diff] [blame]1036 argument constructs on X86-64 by lowering them manually.</li>
Chris Lattner26299222006-11-18 07:51:14 +0000[diff] [blame]1037</ul>
1038
1039</div>
1040
1041<!-- ======================================================================= -->
1042<div class="doc_subsection">
1043 <a name="ppc-be">Known problems with the PowerPC back-end</a>
1044</div>
1045
1046<div class="doc_text">
1047
1048<ul>
Nicolas Geoffraye4285dc2007-05-15 09:21:28 +0000[diff] [blame]1049<li>The Linux PPC32/ABI support needs testing for the interpreter and static
Chris Lattner57a460e2007-05-23 04:39:32 +0000[diff] [blame]1050compilation, and lacks support for debug information.</li>
Chris Lattner26299222006-11-18 07:51:14 +0000[diff] [blame]1051</ul>
1052
1053</div>
1054
1055<!-- ======================================================================= -->
1056<div class="doc_subsection">
Chris Lattnerf3e5bc62007-05-14 06:56:09 +0000[diff] [blame]1057 <a name="arm-be">Known problems with the ARM back-end</a>
1058</div>
1059
1060<div class="doc_text">
1061
1062<ul>
Chris Lattner57a460e2007-05-23 04:39:32 +0000[diff] [blame]1063<li>Thumb mode works only on ARMv6 or higher processors. On sub-ARMv6
Duncan Sandsc90d68b2007-09-26 15:59:54 +0000[diff] [blame]1064processors, thumb programs can crash or produce wrong
Chris Lattner57a460e2007-05-23 04:39:32 +0000[diff] [blame]1065results (<a href="http://llvm.org/PR1388">PR1388</a>).</li>
John Criswell524a5dd2009-03-02 15:28:15 +0000[diff] [blame]1066<li>Compilation for ARM Linux OABI (old ABI) is supported but not fully tested.
Chris Lattnerf3e5bc62007-05-14 06:56:09 +0000[diff] [blame]1067</li>
Chris Lattnerf3e5bc62007-05-14 06:56:09 +0000[diff] [blame]1068</ul>
1069
1070</div>
1071
1072<!-- ======================================================================= -->
1073<div class="doc_subsection">
Chris Lattner26299222006-11-18 07:51:14 +0000[diff] [blame]1074 <a name="sparc-be">Known problems with the SPARC back-end</a>
1075</div>
1076
1077<div class="doc_text">
1078
1079<ul>
John Criswell524a5dd2009-03-02 15:28:15 +0000[diff] [blame]1080<li>The SPARC backend only supports the 32-bit SPARC ABI (-m32); it does not
Chris Lattner26299222006-11-18 07:51:14 +0000[diff] [blame]1081 support the 64-bit SPARC ABI (-m64).</li>
1082</ul>
1083
1084</div>
1085
1086<!-- ======================================================================= -->
1087<div class="doc_subsection">
Bruno Cardoso Lopesb7e1a4f2008-10-25 14:56:26 +0000[diff] [blame]1088 <a name="mips-be">Known problems with the MIPS back-end</a>
1089</div>
1090
1091<div class="doc_text">
1092
1093<ul>
Bruno Cardoso Lopesb7e1a4f2008-10-25 14:56:26 +0000[diff] [blame]1094<li>64-bit MIPS targets are not supported yet.</li>
1095</ul>
1096
1097</div>
1098
1099<!-- ======================================================================= -->
1100<div class="doc_subsection">
Chris Lattner26299222006-11-18 07:51:14 +0000[diff] [blame]1101 <a name="alpha-be">Known problems with the Alpha back-end</a>
1102</div>
1103
1104<div class="doc_text">
1105
1106<ul>
1107
1108<li>On 21164s, some rare FP arithmetic sequences which may trap do not have the
1109appropriate nops inserted to ensure restartability.</li>
1110
John Criswellc0c186d2005-11-08 21:11:33 +0000[diff] [blame]1111</ul>
1112</div>
1113
Chris Lattner26299222006-11-18 07:51:14 +0000[diff] [blame]1114<!-- ======================================================================= -->
1115<div class="doc_subsection">
Chris Lattnerf3e5bc62007-05-14 06:56:09 +0000[diff] [blame]1116 <a name="c-be">Known problems with the C back-end</a>
Chris Lattner26299222006-11-18 07:51:14 +0000[diff] [blame]1117</div>
1118
1119<div class="doc_text">
1120
1121<ul>
Chris Lattner5733b272008-06-05 06:35:40 +0000[diff] [blame]1122<li><a href="http://llvm.org/PR802">The C backend has only basic support for
1123 inline assembly code</a>.</li>
Chris Lattner725a0d82007-09-26 06:01:35 +0000[diff] [blame]1124<li><a href="http://llvm.org/PR1658">The C backend violates the ABI of common
1125 C++ programs</a>, preventing intermixing between C++ compiled by the CBE and
Gabor Greif4906abe2009-03-02 12:02:51 +0000[diff] [blame]1126 C++ code compiled with <tt>llc</tt> or native compilers.</li>
Duncan Sandsf74c0cc2008-02-10 13:40:55 +0000[diff] [blame]1127<li>The C backend does not support all exception handling constructs.</li>
Duncan Sands50723a92009-02-25 11:51:54 +0000[diff] [blame]1128<li>The C backend does not support arbitrary precision integers.</li>
Chris Lattner26299222006-11-18 07:51:14 +0000[diff] [blame]1129</ul>
1130
1131</div>
John Criswellc0c186d2005-11-08 21:11:33 +0000[diff] [blame]1132
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1133
1134<!-- ======================================================================= -->
1135<div class="doc_subsection">
Chris Lattnerc66bfef2010-03-17 04:41:49 +0000[diff] [blame]1136 <a name="c-fe">Known problems with the llvm-gcc C and C++ front-end</a>
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1137</div>
Chris Lattner47588f92003-10-02 05:07:23 +0000[diff] [blame]1138
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1139<div class="doc_text">
Chris Lattnerc5d658a2006-03-03 00:34:26 +0000[diff] [blame]1140
Chris Lattner5733b272008-06-05 06:35:40 +0000[diff] [blame]1141<p>The only major language feature of GCC not supported by llvm-gcc is
1142 the <tt>__builtin_apply</tt> family of builtins. However, some extensions
1143 are only supported on some targets. For example, trampolines are only
Duncan Sands27aff872008-06-08 20:18:35 +0000[diff] [blame]1144 supported on some targets (these are used when you take the address of a
1145 nested function).</p>
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1146
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1147</div>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]1148
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1149<!-- ======================================================================= -->
1150<div class="doc_subsection">
Chris Lattner1eb4df62008-10-30 03:58:13 +0000[diff] [blame]1151 <a name="fortran-fe">Known problems with the llvm-gcc Fortran front-end</a>
1152</div>
1153
1154<div class="doc_text">
Gabor Greifba10fe02008-11-04 21:50:59 +0000[diff] [blame]1155<ul>
Chris Lattner1eb4df62008-10-30 03:58:13 +0000[diff] [blame]1156<li>Fortran support generally works, but there are still several unresolved bugs
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]1157 in <a href="http://llvm.org/bugs/">Bugzilla</a>. Please see the
1158 tools/gfortran component for details.</li>
Gabor Greifba10fe02008-11-04 21:50:59 +0000[diff] [blame]1159</ul>
Chris Lattner1eb4df62008-10-30 03:58:13 +0000[diff] [blame]1160</div>
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1161
Chris Lattner2b659ef2008-02-12 06:29:45 +0000[diff] [blame]1162<!-- ======================================================================= -->
1163<div class="doc_subsection">
Chris Lattner5733b272008-06-05 06:35:40 +0000[diff] [blame]1164 <a name="ada-fe">Known problems with the llvm-gcc Ada front-end</a>
Chris Lattner2b659ef2008-02-12 06:29:45 +0000[diff] [blame]1165</div>
1166
1167<div class="doc_text">
John Criswell524a5dd2009-03-02 15:28:15 +0000[diff] [blame]1168The llvm-gcc 4.2 Ada compiler works fairly well; however, this is not a mature
1169technology, and problems should be expected.
Chris Lattner2b659ef2008-02-12 06:29:45 +0000[diff] [blame]1170<ul>
Duncan Sands27aff872008-06-08 20:18:35 +0000[diff] [blame]1171<li>The Ada front-end currently only builds on X86-32. This is mainly due
John Criswell524a5dd2009-03-02 15:28:15 +0000[diff] [blame]1172to lack of trampoline support (pointers to nested functions) on other platforms.
1173However, it <a href="http://llvm.org/PR2006">also fails to build on X86-64</a>
Chris Lattner2b659ef2008-02-12 06:29:45 +0000[diff] [blame]1174which does support trampolines.</li>
1175<li>The Ada front-end <a href="http://llvm.org/PR2007">fails to bootstrap</a>.
Duncan Sands326a4982009-02-25 11:59:06 +0000[diff] [blame]1176This is due to lack of LLVM support for <tt>setjmp</tt>/<tt>longjmp</tt> style
1177exception handling, which is used internally by the compiler.
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]1178Workaround: configure with <tt>--disable-bootstrap</tt>.</li>
Duncan Sands978bcee2008-10-13 17:27:23 +0000[diff] [blame]1179<li>The c380004, <a href="http://llvm.org/PR2010">c393010</a>
1180and <a href="http://llvm.org/PR2421">cxg2021</a> ACATS tests fail
Duncan Sands326a4982009-02-25 11:59:06 +0000[diff] [blame]1181(c380004 also fails with gcc-4.2 mainline).
1182If the compiler is built with checks disabled then <a href="http://llvm.org/PR2010">c393010</a>
1183causes the compiler to go into an infinite loop, using up all system memory.</li>
Duncan Sandsdd3e6722009-03-02 16:35:57 +0000[diff] [blame]1184<li>Some GCC specific Ada tests continue to crash the compiler.</li>
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]1185<li>The <tt>-E</tt> binder option (exception backtraces)
Chris Lattner2b659ef2008-02-12 06:29:45 +0000[diff] [blame]1186<a href="http://llvm.org/PR1982">does not work</a> and will result in programs
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]1187crashing if an exception is raised. Workaround: do not use <tt>-E</tt>.</li>
Chris Lattner2b659ef2008-02-12 06:29:45 +0000[diff] [blame]1188<li>Only discrete types <a href="http://llvm.org/PR1981">are allowed to start
1189or finish at a non-byte offset</a> in a record. Workaround: do not pack records
1190or use representation clauses that result in a field of a non-discrete type
1191starting or finishing in the middle of a byte.</li>
Chris Lattnere6e1b352008-06-08 21:19:07 +0000[diff] [blame]1192<li>The <tt>lli</tt> interpreter <a href="http://llvm.org/PR2009">considers
1193'main' as generated by the Ada binder to be invalid</a>.
1194Workaround: hand edit the file to use pointers for <tt>argv</tt> and
1195<tt>envp</tt> rather than integers.</li>
1196<li>The <tt>-fstack-check</tt> option <a href="http://llvm.org/PR2008">is
1197ignored</a>.</li>
Chris Lattner2b659ef2008-02-12 06:29:45 +0000[diff] [blame]1198</ul>
1199</div>
1200
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]1201<!-- *********************************************************************** -->
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1202<div class="doc_section">
1203 <a name="additionalinfo">Additional Information</a>
1204</div>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]1205<!-- *********************************************************************** -->
1206
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1207<div class="doc_text">
1208
Chris Lattner416db102005-05-16 17:13:10 +0000[diff] [blame]1209<p>A wide variety of additional information is available on the <a
Chris Lattnerb4b0ce72007-05-18 00:44:29 +0000[diff] [blame]1210href="http://llvm.org">LLVM web page</a>, in particular in the <a
1211href="http://llvm.org/docs/">documentation</a> section. The web page also
Reid Spencer669ed452007-07-09 08:04:31 +0000[diff] [blame]1212contains versions of the API documentation which is up-to-date with the
1213Subversion version of the source code.
Misha Brukman109d9e82005-03-30 19:14:24 +0000[diff] [blame]1214You can access versions of these documents specific to this release by going
1215into the "<tt>llvm/doc/</tt>" directory in the LLVM tree.</p>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]1216
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1217<p>If you have any questions or comments about LLVM, please feel free to contact
Chris Lattnerc463b272005-10-29 07:07:09 +0000[diff] [blame]1218us via the <a href="http://llvm.org/docs/#maillist"> mailing
Chris Lattner5eccca42003-12-12 21:22:16 +0000[diff] [blame]1219lists</a>.</p>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]1220
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1221</div>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]1222
1223<!-- *********************************************************************** -->
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]1224
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1225<hr>
Misha Brukman2061e892003-11-22 01:23:39 +0000[diff] [blame]1226<address>
Misha Brukman38847d52003-12-21 22:53:21 +0000[diff] [blame]1227 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
Misha Brukman44408702008-12-11 17:34:48 +0000[diff] [blame]1228 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
Misha Brukman2061e892003-11-22 01:23:39 +0000[diff] [blame]1229 <a href="http://validator.w3.org/check/referer"><img
Misha Brukman44408702008-12-11 17:34:48 +0000[diff] [blame]1230 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
Misha Brukman2061e892003-11-22 01:23:39 +0000[diff] [blame]1231
Chris Lattnerb4b0ce72007-05-18 00:44:29 +0000[diff] [blame]1232 <a href="http://llvm.org/">LLVM Compiler Infrastructure</a><br>
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1233 Last modified: $Date$
Misha Brukman2061e892003-11-22 01:23:39 +0000[diff] [blame]1234</address>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]1235
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1236</body>
1237</html>