Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / 048fe3c9c3456709581962bebe1f92ece0f79643 / . / docs / ReleaseNotes.html
blob: f3d87c6330c8fb5af1aab445e0ea6bb680c28a25 [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">
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]7 <title>LLVM 2.6 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
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]11<div class="doc_title">LLVM 2.6 Release Notes</div>
Mikhail Glushenkovea65d7d2008-10-13 02:08:34 +0000[diff] [blame]12
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]13<ol>
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]14 <li><a href="#intro">Introduction</a></li>
Chris Lattner96a445e2008-10-13 18:01:01 +0000[diff] [blame]15 <li><a href="#subproj">Sub-project Status Update</a></li>
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]16 <li><a href="#externalproj">External Projects Using LLVM 2.6</a></li>
17 <li><a href="#whatsnew">What's New in LLVM 2.6?</a></li>
Chris Lattner4b538b92004-04-30 22:17:12 +0000[diff] [blame]18 <li><a href="GettingStarted.html">Installation Instructions</a></li>
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]19 <li><a href="#portability">Portability and Supported Platforms</a></li>
Dan Gohman44aa9212008-10-14 16:23:02 +0000[diff] [blame]20 <li><a href="#knownproblems">Known Problems</a></li>
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]21 <li><a href="#additionalinfo">Additional Information</a></li>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]22</ol>
23
Chris Lattner7911ce22004-05-23 21:07:27 +0000[diff] [blame]24<div class="doc_author">
Dan Gohman44aa9212008-10-14 16:23:02 +0000[diff] [blame]25 <p>Written by the <a href="http://llvm.org">LLVM Team</a></p>
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]26</div>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]27
28<!-- *********************************************************************** -->
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]29<div class="doc_section">
30 <a name="intro">Introduction</a>
31</div>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]32<!-- *********************************************************************** -->
33
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]34<div class="doc_text">
35
Chris Lattnerf6662f92008-10-13 17:57:36 +0000[diff] [blame]36<p>This document contains the release notes for the LLVM Compiler
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]37Infrastructure, release 2.6. Here we describe the status of LLVM, including
Chris Lattnerf6662f92008-10-13 17:57:36 +0000[diff] [blame]38major improvements from the previous release and significant known problems.
Mikhail Glushenkovf795ef02009-03-01 18:09:47 +0000[diff] [blame]39All LLVM releases may be downloaded from the <a
Chris Lattnerf6662f92008-10-13 17:57:36 +0000[diff] [blame]40href="http://llvm.org/releases/">LLVM releases web site</a>.</p>
Chris Lattner19092612003-10-02 16:38:05 +0000[diff] [blame]41
Chris Lattner7506b1d2004-12-07 08:04:13 +0000[diff] [blame]42<p>For more information about LLVM, including information about the latest
Chris Lattnerc463b272005-10-29 07:07:09 +0000[diff] [blame]43release, please check out the <a href="http://llvm.org/">main LLVM
Chris Lattner47ad72c2003-10-07 21:38:31 +0000[diff] [blame]44web site</a>. If you have questions or comments, the <a
Chris Lattnerf6662f92008-10-13 17:57:36 +0000[diff] [blame]45href="http://mail.cs.uiuc.edu/mailman/listinfo/llvmdev">LLVM Developer's Mailing
46List</a> is a good place to send them.</p>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]47
Mikhail Glushenkovea65d7d2008-10-13 02:08:34 +0000[diff] [blame]48<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]49main LLVM web page, this document applies to the <i>next</i> release, not the
Gabor Greiffa933f82008-10-14 11:00:32 +0000[diff] [blame]50current one. To see the release notes for a specific release, please see the
Chris Lattner84977642007-09-21 03:54:09 +0000[diff] [blame]51<a href="http://llvm.org/releases/">releases page</a>.</p>
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]52
53</div>
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]54
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]55
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]56<!--
57Almost dead code.
58 include/llvm/Analysis/LiveValues.h => Dan
59 lib/Transforms/IPO/MergeFunctions.cpp => consider for 2.8.
60 llvm/Analysis/PointerTracking.h => Edwin wants this, consider for 2.8.
Chris Lattner76ef2982010-01-09 22:30:40 +0000[diff] [blame]61 ABCD, SCCVN, GEPSplitterPass
Chris Lattner048fe3c2010-01-16 21:25:13 +0000[diff] [blame^]62 MSIL backend?
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]63-->
64
65
66<!-- Unfinished features in 2.6:
67 gcc plugin.
68 strong phi elim
69 variable debug info for optimized code
Chris Lattnerab68e9e2009-02-26 22:33:38 +0000[diff] [blame]70 postalloc scheduler: anti dependence breaking, hazard recognizer?
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]71 metadata
72 loop dependence analysis
73 ELF Writer? How stable?
74 <li>PostRA scheduler improvements, ARM adoption (David Goodwin).</li>
75 2.7 supports the GDB 7.0 jit interfaces for debug info.
76 2.7 eliminates ADT/iterator.h
Chris Lattneracce85d2008-02-10 07:46:44 +0000[diff] [blame]77 -->
Mikhail Glushenkovea65d7d2008-10-13 02:08:34 +0000[diff] [blame]78
Chris Lattner547a3912008-10-12 19:47:48 +0000[diff] [blame]79 <!-- for announcement email:
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]80 Logo web page.
81 llvm devmtg
82 compiler_rt
83 KLEE web page at klee.llvm.org
84 Many new papers added to /pubs/
85 Mention gcc plugin.
86
Chris Lattner74c80df2009-02-25 06:34:50 +0000[diff] [blame]87 -->
Chris Lattneracce85d2008-02-10 07:46:44 +0000[diff] [blame]88
Chris Lattner96a445e2008-10-13 18:01:01 +0000[diff] [blame]89<!-- *********************************************************************** -->
90<div class="doc_section">
91 <a name="subproj">Sub-project Status Update</a>
Chris Lattnerea34f642008-06-08 21:34:41 +0000[diff] [blame]92</div>
Chris Lattner96a445e2008-10-13 18:01:01 +0000[diff] [blame]93<!-- *********************************************************************** -->
Chris Lattnerea34f642008-06-08 21:34:41 +0000[diff] [blame]94
95<div class="doc_text">
Chris Lattner8348b472008-06-08 21:58:17 +0000[diff] [blame]96<p>
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]97The LLVM 2.6 distribution currently consists of code from the core LLVM
98repository (which roughly includes the LLVM optimizers, code generators
99and supporting tools), the Clang repository and the llvm-gcc repository. In
100addition to this code, the LLVM Project includes other sub-projects that are in
101development. Here we include updates on these subprojects.
Bill Wendling63d8c552009-03-02 04:28:57 +0000[diff] [blame]102</p>
Chris Lattner96a445e2008-10-13 18:01:01 +0000[diff] [blame]103
Chris Lattner8348b472008-06-08 21:58:17 +0000[diff] [blame]104</div>
105
Chris Lattner96a445e2008-10-13 18:01:01 +0000[diff] [blame]106
Chris Lattner8348b472008-06-08 21:58:17 +0000[diff] [blame]107<!--=========================================================================-->
Chris Lattner96a445e2008-10-13 18:01:01 +0000[diff] [blame]108<div class="doc_subsection">
Chris Lattnerfb97b2d2008-10-13 18:11:54 +0000[diff] [blame]109<a name="clang">Clang: C/C++/Objective-C Frontend Toolkit</a>
Chris Lattner8348b472008-06-08 21:58:17 +0000[diff] [blame]110</div>
111
112<div class="doc_text">
113
114<p>The <a href="http://clang.llvm.org/">Clang project</a> is an effort to build
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]115a set of new 'LLVM native' front-end technologies for the C family of languages.
116LLVM 2.6 is the first release to officially include Clang, and it provides a
117production quality C and Objective-C compiler. If you are interested in <a
118href="http://clang.llvm.org/performance.html">fast compiles</a> and
119<a href="http://clang.llvm.org/diagnostics.html">good diagnostics</a>, we
120encourage you to try it out. Clang currently compiles typical Objective-C code
1213x faster than GCC and compiles C code about 30% faster than GCC at -O0 -g
122(which is when the most pressure is on the frontend).</p>
123
124<p>In addition to supporting these languages, C++ support is also <a
125href="http://clang.llvm.org/cxx_status.html">well under way</a>, and mainline
126Clang is able to parse the libstdc++ 4.2 headers and even codegen simple apps.
127If you are interested in Clang C++ support or any other Clang feature, we
128strongly encourage you to get involved on the <a
Chris Lattnerf40c40b2009-02-25 05:09:54 +0000[diff] [blame]129href="http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev">Clang front-end mailing
130list</a>.</p>
131
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]132<p>In the LLVM 2.6 time-frame, the Clang team has made many improvements:</p>
Bill Wendling741748a2008-10-27 09:27:33 +0000[diff] [blame]133
Daniel Dunbar13739432008-10-14 23:25:09 +0000[diff] [blame]134<ul>
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]135<li>C and Objective-C support are now considered production quality.</li>
136<li>AuroraUX, FreeBSD and OpenBSD are now supported.</li>
137<li>Most of Objective-C 2.0 is now supported with the GNU runtime.</li>
138<li>Many many bugs are fixed and lots of features have been added.</li>
Bill Wendling6bc15282009-03-02 04:28:18 +0000[diff] [blame]139</ul>
Chris Lattnerfb97b2d2008-10-13 18:11:54 +0000[diff] [blame]140</div>
141
142<!--=========================================================================-->
143<div class="doc_subsection">
144<a name="clangsa">Clang Static Analyzer</a>
145</div>
146
147<div class="doc_text">
148
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]149<p>Previously announced in the 2.4 and 2.5 LLVM releases, the Clang project also
Ted Kremenek8b3894c2009-02-27 07:01:32 +0000[diff] [blame]150includes an early stage static source code analysis tool for <a
151href="http://clang.llvm.org/StaticAnalysis.html">automatically finding bugs</a>
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]152in C and Objective-C programs. The tool performs checks to find
Ted Kremenek8b3894c2009-02-27 07:01:32 +0000[diff] [blame]153bugs that occur on a specific path within a program.</p>
Chris Lattnercc042612008-10-14 00:52:49 +0000[diff] [blame]154
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]155<p>In the LLVM 2.6 time-frame, the analyzer core has undergone several important
156improvements and cleanups and now includes a new <em>Checker</em> interface that
157is intended to eventually serve as a basis for domain-specific checks. Further,
158in addition to generating HTML files for reporting analysis results, the
159analyzer can now also emit bug reports in a structured XML format that is
160intended to be easily readable by other programs.</p>
Ted Kremenek3c3ec0c2008-10-14 05:14:21 +0000[diff] [blame]161
John Criswell524a5dd2009-03-02 15:28:15 +0000[diff] [blame]162<p>The set of checks performed by the static analyzer continues to expand, and
Ted Kremenek8b3894c2009-02-27 07:01:32 +0000[diff] [blame]163future plans for the tool include full source-level inter-procedural analysis
164and deeper checks such as buffer overrun detection. There are many opportunities
165to extend and enhance the static analyzer, and anyone interested in working on
166this project is encouraged to get involved!</p>
Chris Lattner8348b472008-06-08 21:58:17 +0000[diff] [blame]167
168</div>
169
Chris Lattner96a445e2008-10-13 18:01:01 +0000[diff] [blame]170<!--=========================================================================-->
171<div class="doc_subsection">
Nicolas Geoffray99a4d302008-10-14 19:23:04 +0000[diff] [blame]172<a name="vmkit">VMKit: JVM/CLI Virtual Machine Implementation</a>
Chris Lattner96a445e2008-10-13 18:01:01 +0000[diff] [blame]173</div>
174
175<div class="doc_text">
176<p>
Nicolas Geoffray99a4d302008-10-14 19:23:04 +0000[diff] [blame]177The <a href="http://vmkit.llvm.org/">VMKit project</a> is an implementation of
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]178a JVM and a CLI Virtual Machine (Microsoft .NET is an
179implementation of the CLI) using LLVM for static and just-in-time
180compilation.</p>
Chris Lattner96a445e2008-10-13 18:01:01 +0000[diff] [blame]181
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]182<p>
183VMKit version 0.26 builds with LLVM 2.6 and you can find it on its
184<a href="http://vmkit.llvm.org/releases/">web page</a>. The release includes
Nicolas Geoffray56d313d2008-10-15 20:25:04 +0000[diff] [blame]185bug fixes, cleanup and new features. The major changes are:</p>
Chris Lattner96a445e2008-10-13 18:01:01 +0000[diff] [blame]186
Nicolas Geoffray99a4d302008-10-14 19:23:04 +0000[diff] [blame]187<ul>
188
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]189<li>A new llcj tool to generate shared libraries or executables of Java
190 files.</li>
191<li>Cooperative garbage collection. </li>
192<li>Fast subtype checking (paper from Click et al [JGI'02]). </li>
193<li>Implementation of a two-word header for Java objects instead of the original
194 three-word header. </li>
195<li>Better Java specification-compliance: division by zero checks, stack
196 overflow checks, finalization and references support. </li>
Nicolas Geoffray99a4d302008-10-14 19:23:04 +0000[diff] [blame]197
198</ul>
Chris Lattner96a445e2008-10-13 18:01:01 +0000[diff] [blame]199</div>
200
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]201
202<!--=========================================================================-->
203<div class="doc_subsection">
204<a name="compiler-rt">compiler-rt: Compiler Runtime Library</a>
205</div>
206
207<div class="doc_text">
208<p>
209The new LLVM <a href="http://compiler-rt.llvm.org/">compiler-rt project</a>
210is a simple library that provides an implementation of the low-level
211target-specific hooks required by code generation and other runtime components.
212For example, when compiling for a 32-bit target, converting a double to a 64-bit
213unsigned integer is compiled into a runtime call to the "__fixunsdfdi"
214function. The compiler-rt library provides highly optimized implementations of
215this and other low-level routines (some are 3x faster than the equivalent
216libgcc routines).</p>
217
218<p>
219All of the code in the compiler-rt project is available under the standard LLVM
220License, a "BSD-style" license.</p>
221
222</div>
223
224<!--=========================================================================-->
225<div class="doc_subsection">
226<a name="klee">KLEE: Symbolic Execution and Automatic Test Case Generator</a>
227</div>
228
229<div class="doc_text">
230<p>
231The new LLVM <a href="http://klee.llvm.org/">KLEE project</a> is a symbolic
232execution framework for programs in LLVM bitcode form. KLEE tries to
233symbolically evaluate "all" paths through the application and records state
234transitions that lead to fault states. This allows it to construct testcases
235that lead to faults and can even be used to verify algorithms. For more
236details, please see the <a
237href="http://llvm.org/pubs/2008-12-OSDI-KLEE.html">OSDI 2008 paper</a> about
238KLEE.</p>
239
240</div>
241
242<!--=========================================================================-->
243<div class="doc_subsection">
244<a name="dragonegg">DragonEgg: GCC-4.5 as an LLVM frontend</a>
245</div>
246
247<div class="doc_text">
248<p>
249The goal of <a href="http://dragonegg.llvm.org/">DragonEgg</a> is to make
250gcc-4.5 act like llvm-gcc without requiring any gcc modifications whatsoever.
Duncan Sands78a1dcc2009-10-15 08:17:44 +0000[diff] [blame]251<a href="http://dragonegg.llvm.org/">DragonEgg</a> is a shared library (dragonegg.so)
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]252that is loaded by gcc at runtime. It uses the new gcc plugin architecture to
253disable the GCC optimizers and code generators, and schedule the LLVM optimizers
254and code generators (or direct output of LLVM IR) instead. Currently only Linux
255and Darwin are supported, and only on x86-32 and x86-64. It should be easy to
256add additional unix-like architectures and other processor families. In theory
257it should be possible to use <a href="http://dragonegg.llvm.org/">DragonEgg</a>
258with any language supported by gcc, however only C and Fortran work well for the
259moment. Ada and C++ work to some extent, while Java, Obj-C and Obj-C++ are so
260far entirely untested. Since gcc-4.5 has not yet been released, neither has
261<a href="http://dragonegg.llvm.org/">DragonEgg</a>. To build
262<a href="http://dragonegg.llvm.org/">DragonEgg</a> you will need to check out the
263development versions of <a href="http://gcc.gnu.org/svn.html/"> gcc</a>,
264<a href="http://llvm.org/docs/GettingStarted.html#checkout">llvm</a> and
265<a href="http://dragonegg.llvm.org/">DragonEgg</a> from their respective
266subversion repositories, and follow the instructions in the
267<a href="http://dragonegg.llvm.org/">DragonEgg</a> README.
268</p>
269
270</div>
271
272
273<!--=========================================================================-->
274<div class="doc_subsection">
275<a name="mc">llvm-mc: Machine Code Toolkit</a>
276</div>
277
278<div class="doc_text">
279<p>
280The LLVM Machine Code (MC) Toolkit project is a (very early) effort to build
281better tools for dealing with machine code, object file formats, etc. The idea
282is to be able to generate most of the target specific details of assemblers and
283disassemblers from existing LLVM target .td files (with suitable enhancements),
284and to build infrastructure for reading and writing common object file formats.
285One of the first deliverables is to build a full assembler and integrate it into
286the compiler, which is predicted to substantially reduce compile time in some
287scenarios.
288</p>
289
290<p>In the LLVM 2.6 timeframe, the MC framework has grown to the point where it
291can reliably parse and pretty print (with some encoding information) a
292darwin/x86 .s file successfully, and has the very early phases of a Mach-O
293assembler in progress. Beyond the MC framework itself, major refactoring of the
294LLVM code generator has started. The idea is to make the code generator reason
295about the code it is producing in a much more semantic way, rather than a
296textual way. For example, the code generator now uses MCSection objects to
297represent section assignments, instead of text strings that print to .section
298directives.</p>
299
300<p>MC is an early and ongoing project that will hopefully continue to lead to
301many improvements in the code generator and build infrastructure useful for many
302other situations.
303</p>
304
305</div>
306
307
Chris Lattnerab68e9e2009-02-26 22:33:38 +0000[diff] [blame]308<!-- *********************************************************************** -->
309<div class="doc_section">
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]310 <a name="externalproj">External Open Source Projects Using LLVM 2.6</a>
Chris Lattnerab68e9e2009-02-26 22:33:38 +0000[diff] [blame]311</div>
312<!-- *********************************************************************** -->
313
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]314<div class="doc_text">
315
316<p>An exciting aspect of LLVM is that it is used as an enabling technology for
317 a lot of other language and tools projects. This section lists some of the
318 projects that have already been updated to work with LLVM 2.6.</p>
319</div>
320
321
322<!--=========================================================================-->
323<div class="doc_subsection">
324<a name="Rubinius">Rubinius</a>
325</div>
326
327<div class="doc_text">
328<p><a href="http://github.com/evanphx/rubinius">Rubinius</a> is an environment
329for running Ruby code which strives to write as much of the core class
330implementation in Ruby as possible. Combined with a bytecode interpreting VM, it
331uses LLVM to optimize and compile ruby code down to machine code. Techniques
332such as type feedback, method inlining, and uncommon traps are all used to
333remove dynamism from ruby execution and increase performance.</p>
334
335<p>Since LLVM 2.5, Rubinius has made several major leaps forward, implementing
336a counter based JIT, type feedback and speculative method inlining.
337</p>
338
339</div>
340
341<!--=========================================================================-->
342<div class="doc_subsection">
343<a name="macruby">MacRuby</a>
344</div>
345
346<div class="doc_text">
347
348<p>
349<a href="http://macruby.org">MacRuby</a> is an implementation of Ruby on top of
350core Mac OS X technologies, such as the Objective-C common runtime and garbage
351collector and the CoreFoundation framework. It is principally developed by
352Apple and aims at enabling the creation of full-fledged Mac OS X applications.
353</p>
354
355<p>
356MacRuby uses LLVM for optimization passes, JIT and AOT compilation of Ruby
357expressions. It also uses zero-cost DWARF exceptions to implement Ruby exception
358handling.</p>
359
360</div>
361
362
Chris Lattnerab68e9e2009-02-26 22:33:38 +0000[diff] [blame]363<!--=========================================================================-->
364<div class="doc_subsection">
365<a name="pure">Pure</a>
366</div>
367
368<div class="doc_text">
369<p>
Chris Lattner917cc712009-03-02 02:37:32 +0000[diff] [blame]370<a href="http://pure-lang.googlecode.com/">Pure</a>
371is an algebraic/functional programming language based on term rewriting.
Chris Lattnerab68e9e2009-02-26 22:33:38 +0000[diff] [blame]372Programs are collections of equations which are used to evaluate expressions in
373a symbolic fashion. Pure offers dynamic typing, eager and lazy evaluation,
Mikhail Glushenkovf795ef02009-03-01 18:09:47 +0000[diff] [blame]374lexical closures, a hygienic macro system (also based on term rewriting),
Chris Lattnerab68e9e2009-02-26 22:33:38 +0000[diff] [blame]375built-in list and matrix support (including list and matrix comprehensions) and
376an easy-to-use C interface. The interpreter uses LLVM as a backend to
377 JIT-compile Pure programs to fast native code.</p>
378
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]379<p>Pure versions 0.31 and later have been tested and are known to work with
380LLVM 2.6 (and continue to work with older LLVM releases >= 2.3 as well).
Chris Lattnerab68e9e2009-02-26 22:33:38 +0000[diff] [blame]381</p>
382</div>
383
384
385<!--=========================================================================-->
386<div class="doc_subsection">
387<a name="ldc">LLVM D Compiler</a>
388</div>
389
390<div class="doc_text">
391<p>
Chris Lattner917cc712009-03-02 02:37:32 +0000[diff] [blame]392<a href="http://www.dsource.org/projects/ldc">LDC</a> is an implementation of
393the D Programming Language using the LLVM optimizer and code generator.
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]394The LDC project works great with the LLVM 2.6 release. General improvements in
John Criswell524a5dd2009-03-02 15:28:15 +0000[diff] [blame]395this
Chris Lattner917cc712009-03-02 02:37:32 +0000[diff] [blame]396cycle have included new inline asm constraint handling, better debug info
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]397support, general bug fixes and better x86-64 support. This has allowed
398some major improvements in LDC, getting it much closer to being as
Chris Lattnerab68e9e2009-02-26 22:33:38 +0000[diff] [blame]399fully featured as the original DMD compiler from DigitalMars.
400</p>
401</div>
402
Chris Lattnerbc31caf2009-02-28 18:58:01 +0000[diff] [blame]403<!--=========================================================================-->
404<div class="doc_subsection">
405<a name="RoadsendPHP">Roadsend PHP</a>
406</div>
407
408<div class="doc_text">
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]409<p>
410<a href="http://code.roadsend.com/rphp">Roadsend PHP</a> (rphp) is an open
Chris Lattnercade8222009-03-02 19:07:24 +0000[diff] [blame]411source implementation of the PHP programming
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]412language that uses LLVM for its optimizer, JIT and static compiler. This is a
Chris Lattnercade8222009-03-02 19:07:24 +0000[diff] [blame]413reimplementation of an earlier project that is now based on LLVM.</p>
Chris Lattnerbc31caf2009-02-28 18:58:01 +0000[diff] [blame]414</div>
Chris Lattnerab68e9e2009-02-26 22:33:38 +0000[diff] [blame]415
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]416<!--=========================================================================-->
417<div class="doc_subsection">
418<a name="UnladenSwallow">Unladen Swallow</a>
419</div>
420
421<div class="doc_text">
422<p>
423<a href="http://code.google.com/p/unladen-swallow/">Unladen Swallow</a> is a
424branch of <a href="http://python.org/">Python</a> intended to be fully
425compatible and significantly faster. It uses LLVM's optimization passes and JIT
426compiler.</p>
427</div>
428
429<!--=========================================================================-->
430<div class="doc_subsection">
431<a name="llvm-lua">llvm-lua</a>
432</div>
433
434<div class="doc_text">
435<p>
436<a href="http://code.google.com/p/llvm-lua/">LLVM-Lua</a> uses LLVM to add JIT
437and static compiling support to the Lua VM. Lua bytecode is analyzed to
438remove type checks, then LLVM is used to compile the bytecode down to machine
439code.</p>
440</div>
441
442<!--=========================================================================-->
443<div class="doc_subsection">
444<a name="icedtea">IcedTea Java Virtual Machine Implementation</a>
445</div>
446
447<div class="doc_text">
448<p>
449<a href="http://icedtea.classpath.org/wiki/Main_Page">IcedTea</a> provides a
450harness to build OpenJDK using only free software build tools and to provide
451replacements for the not-yet free parts of OpenJDK. One of the extensions that
452IcedTea provides is a new JIT compiler named <a
453href="http://icedtea.classpath.org/wiki/ZeroSharkFaq">Shark</a> which uses LLVM
454to provide native code generation without introducing processor-dependent
455code.
456</p>
457</div>
458
459
Chris Lattner8348b472008-06-08 21:58:17 +0000[diff] [blame]460
461<!-- *********************************************************************** -->
462<div class="doc_section">
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]463 <a name="whatsnew">What's New in LLVM 2.6?</a>
Chris Lattner8348b472008-06-08 21:58:17 +0000[diff] [blame]464</div>
465<!-- *********************************************************************** -->
466
Chris Lattnerf8e0b4e2008-06-08 22:59:35 +0000[diff] [blame]467<div class="doc_text">
468
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]469<p>This release includes a huge number of bug fixes, performance tweaks and
Chris Lattnerf6662f92008-10-13 17:57:36 +0000[diff] [blame]470minor improvements. Some of the major improvements and new features are listed
471in this section.
Chris Lattnerf8e0b4e2008-06-08 22:59:35 +0000[diff] [blame]472</p>
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]473
Chris Lattnerf8e0b4e2008-06-08 22:59:35 +0000[diff] [blame]474</div>
475
Chris Lattner8348b472008-06-08 21:58:17 +0000[diff] [blame]476<!--=========================================================================-->
Chris Lattnerea34f642008-06-08 21:34:41 +0000[diff] [blame]477<div class="doc_subsection">
Chris Lattner252b83d2008-02-06 18:00:06 +0000[diff] [blame]478<a name="majorfeatures">Major New Features</a>
479</div>
480
481<div class="doc_text">
482
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]483<p>LLVM 2.6 includes several major new capabilities:</p>
Chris Lattnerf3e5bc62007-05-14 06:56:09 +0000[diff] [blame]484
485<ul>
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]486<li>New <a href="#compiler-rt">compiler-rt</a>, <A href="#klee">KLEE</a>
487 and <a href="#mc">machine code toolkit</a> sub-projects.</li>
488<li>Debug information now includes line numbers when optimizations are enabled.
489 This allows statistical sampling tools like OProfile and Shark to map
490 samples back to source lines.</li>
491<li>LLVM now includes new experimental backends to support the MSP430, SystemZ
492 and BlackFin architectures.</li>
493<li>LLVM supports a new <a href="GoldPlugin.html">Gold Linker Plugin</a> which
494 enables support for <a href="LinkTimeOptimization.html">transparent
495 link-time optimization</a> on ELF targets when used with the Gold binutils
496 linker.</li>
497<li>LLVM now supports doing optimization and code generation on multiple
498 threads. Please see the <a href="ProgrammersManual.html#threading">LLVM
499 Programmer's Manual</a> for more information.</li>
500<li>LLVM now has experimental support for <a
501 href="http://nondot.org/~sabre/LLVMNotes/EmbeddedMetadata.txt">embedded
502 metadata</a> in LLVM IR, though the implementation is not guaranteed to be
503 final and the .bc file format may change in future releases. Debug info
504 does not yet use this format in LLVM 2.6.</li>
Chris Lattner8170c102008-02-10 08:18:42 +0000[diff] [blame]505</ul>
506
Chris Lattnerf3e5bc62007-05-14 06:56:09 +0000[diff] [blame]507</div>
508
Chris Lattner84977642007-09-21 03:54:09 +0000[diff] [blame]509<!--=========================================================================-->
510<div class="doc_subsection">
Chris Lattner917cc712009-03-02 02:37:32 +0000[diff] [blame]511<a name="coreimprovements">LLVM IR and Core Improvements</a>
Chris Lattnerf304ffc2008-02-10 08:17:19 +0000[diff] [blame]512</div>
513
514<div class="doc_text">
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]515<p>LLVM IR has several new features for better support of new targets and that
516expose new optimization opportunities:</p>
Chris Lattnerf304ffc2008-02-10 08:17:19 +0000[diff] [blame]517
Chris Lattner791f77b2008-06-05 06:25:56 +0000[diff] [blame]518<ul>
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]519<li>The <a href="LangRef.html#i_add">add</a>, <a
520 href="LangRef.html#i_sub">sub</a> and <a href="LangRef.html#i_mul">mul</a>
521 instructions have been split into integer and floating point versions (like
522 divide and remainder), introducing new <a
523 href="LangRef.html#i_fadd">fadd</a>, <a href="LangRef.html#i_fsub">fsub</a>,
524 and <a href="LangRef.html#i_fmul">fmul</a> instructions.</li>
525<li>The <a href="LangRef.html#i_add">add</a>, <a
526 href="LangRef.html#i_sub">sub</a> and <a href="LangRef.html#i_mul">mul</a>
527 instructions now support optional "nsw" and "nuw" bits which indicate that
528 the operation is guaranteed to not overflow (in the signed or
529 unsigned case, respectively). This gives the optimizer more information and
530 can be used for things like C signed integer values, which are undefined on
531 overflow.</li>
532<li>The <a href="LangRef.html#i_sdiv">sdiv</a> instruction now supports an
533 optional "exact" flag which indicates that the result of the division is
534 guaranteed to have a remainder of zero. This is useful for optimizing pointer
535 subtraction in C.</li>
536<li>The <a href="LangRef.html#i_getelementptr">getelementptr</a> instruction now
537 supports arbitrary integer index values for array/pointer indices. This
538 allows for better code generation on 16-bit pointer targets like PIC16.</li>
539<li>The <a href="LangRef.html#i_getelementptr">getelementptr</a> instruction now
540 supports an "inbounds" optimization hint that tells the optimizer that the
541 pointer is guaranteed to be within its allocated object.</li>
542<li>LLVM now support a series of new linkage types for global values which allow
543 for better optimization and new capabilities:
544 <ul>
545 <li><a href="LangRef.html#linkage_linkonce">linkonce_odr</a> and
546 <a href="LangRef.html#linkage_weak">weak_odr</a> have the same linkage
547 semantics as the non-"odr" linkage types. The difference is that these
548 linkage types indicate that all definitions of the specified function
549 are guaranteed to have the same semantics. This allows inlining
550 templates functions in C++ but not inlining weak functions in C,
551 which previously both got the same linkage type.</li>
552 <li><a href="LangRef.html#linkage_available_externally">available_externally
553 </a> is a new linkage type that gives the optimizer visibility into the
554 definition of a function (allowing inlining and side effect analysis)
555 but that does not cause code to be generated. This allows better
556 optimization of "GNU inline" functions, extern templates, etc.</li>
557 <li><a href="LangRef.html#linkage_linker_private">linker_private</a> is a
558 new linkage type (which is only useful on Mac OS X) that is used for
559 some metadata generation and other obscure things.</li>
560 </ul></li>
561<li>Finally, target-specific intrinsics can now return multiple values, which
562 is useful for modeling target operations with multiple results.</li>
Chris Lattnerf304ffc2008-02-10 08:17:19 +0000[diff] [blame]563</ul>
Mikhail Glushenkovea65d7d2008-10-13 02:08:34 +0000[diff] [blame]564
Chris Lattnerf304ffc2008-02-10 08:17:19 +0000[diff] [blame]565</div>
566
567<!--=========================================================================-->
568<div class="doc_subsection">
Chris Lattneracce85d2008-02-10 07:46:44 +0000[diff] [blame]569<a name="optimizer">Optimizer Improvements</a>
570</div>
571
572<div class="doc_text">
573
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]574<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]575release includes a few major enhancements and additions to the optimizers:</p>
Chris Lattneracce85d2008-02-10 07:46:44 +0000[diff] [blame]576
577<ul>
578
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]579<li>The <a href="Passes.html#scalarrepl">Scalar Replacement of Aggregates</a>
580 pass has many improvements that allow it to better promote vector unions,
581 variables which are memset, and much more strange code that can happen to
582 do bitfield accesses to register operations. An interesting change is that
583 it now produces "unusual" integer sizes (like i1704) in some cases and lets
584 other optimizers clean things up.</li>
585<li>The <a href="Passes.html#loop-reduce">Loop Strength Reduction</a> pass now
586 promotes small integer induction variables to 64-bit on 64-bit targets,
587 which provides a major performance boost for much numerical code. It also
588 promotes shorts to int on 32-bit hosts, etc. LSR now also analyzes pointer
589 expressions (e.g. getelementptrs), as well as integers.</li>
590<li>The <a href="Passes.html#gvn">GVN</a> pass now eliminates partial
591 redundancies of loads in simple cases.</li>
592<li>The <a href="Passes.html#inline">Inliner</a> now reuses stack space when
593 inlining similar arrays from multiple callees into one caller.</li>
594<li>LLVM includes a new experimental Static Single Information (SSI)
595 construction pass.</li>
Chris Lattner917cc712009-03-02 02:37:32 +0000[diff] [blame]596
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]597</ul>
Chris Lattner917cc712009-03-02 02:37:32 +0000[diff] [blame]598
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]599</div>
Chris Lattner917cc712009-03-02 02:37:32 +0000[diff] [blame]600
Chris Lattner917cc712009-03-02 02:37:32 +0000[diff] [blame]601
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]602<!--=========================================================================-->
603<div class="doc_subsection">
604<a name="executionengine">Interpreter and JIT Improvements</a>
605</div>
Chris Lattnerf3013872008-10-13 21:50:36 +0000[diff] [blame]606
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]607<div class="doc_text">
608
609<ul>
610<li>LLVM has a new "EngineBuilder" class which makes it more obvious how to
611 set up and configure an ExecutionEngine (a JIT or interpreter).</li>
612<li>The JIT now supports generating more than 16M of code.</li>
613<li>When configured with <tt>--with-oprofile</tt>, the JIT can now inform
614 OProfile about JIT'd code, allowing OProfile to get line number and function
615 name information for JIT'd functions.</li>
616<li>When "libffi" is available, the LLVM interpreter now uses it, which supports
617 calling almost arbitrary external (natively compiled) functions.</li>
618<li>Clients of the JIT can now register a 'JITEventListener' object to receive
619 callbacks when the JIT emits or frees machine code. The OProfile support
620 uses this mechanism.</li>
Chris Lattner0b832202008-06-08 02:45:07 +0000[diff] [blame]621</ul>
622
623</div>
624
625<!--=========================================================================-->
626<div class="doc_subsection">
Chris Lattner511433e2009-03-02 03:24:11 +0000[diff] [blame]627<a name="codegen">Target Independent Code Generator Improvements</a>
Chris Lattner0b832202008-06-08 02:45:07 +0000[diff] [blame]628</div>
629
630<div class="doc_text">
631
Mikhail Glushenkovf795ef02009-03-01 18:09:47 +0000[diff] [blame]632<p>We have put a significant amount of work into the code generator
633infrastructure, which allows us to implement more aggressive algorithms and make
634it run faster:</p>
Chris Lattner0b832202008-06-08 02:45:07 +0000[diff] [blame]635
636<ul>
Chris Lattner511433e2009-03-02 03:24:11 +0000[diff] [blame]637
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]638<li>The <tt>llc -asm-verbose</tt> option (exposed from llvm-gcc as <tt>-dA</tt>
639 and clang as <tt>-fverbose-asm</tt> or <tt>-dA</tt>) now adds a lot of
640 useful information in comments to
641 the generated .s file. This information includes location information (if
642 built with <tt>-g</tt>) and loop nest information.</li>
643<li>The code generator now supports a new MachineVerifier pass which is useful
644 for finding bugs in targets and codegen passes.</li>
645<li>The Machine LICM is now enabled by default. It hoists instructions out of
646 loops (such as constant pool loads, loads from read-only stubs, vector
647 constant synthesization code, etc.) and is currently configured to only do
648 so when the hoisted operation can be rematerialized.</li>
649<li>The Machine Sinking pass is now enabled by default. This pass moves
650 side-effect free operations down the CFG so that they are executed on fewer
651 paths through a function.</li>
652<li>The code generator now performs "stack slot coloring" of register spills,
653 which allows spill slots to be reused. This leads to smaller stack frames
654 in cases where there are lots of register spills.</li>
655<li>The register allocator has many improvements to take better advantage of
656 commutable operations, various spiller peephole optimizations, and can now
657 coalesce cross-register-class copies.</li>
658<li>Tblgen now supports multiclass inheritance and a number of new string and
659 list operations like <tt>!(subst)</tt>, <tt>!(foreach)</tt>, <tt>!car</tt>,
660 <tt>!cdr</tt>, <tt>!null</tt>, <tt>!if</tt>, <tt>!cast</tt>.
661 These make the .td files more expressive and allow more aggressive factoring
662 of duplication across instruction patterns.</li>
663<li>Target-specific intrinsics can now be added without having to hack VMCore to
664 add them. This makes it easier to maintain out-of-tree targets.</li>
665<li>The instruction selector is better at propagating information about values
666 (such as whether they are sign/zero extended etc.) across basic block
667 boundaries.</li>
668<li>The SelectionDAG datastructure has new nodes for representing buildvector
669 and <a href="http://llvm.org/PR2957">vector shuffle</a> operations. This
670 makes operations and pattern matching more efficient and easier to get
671 right.</li>
672<li>The Prolog/Epilog Insertion Pass now has experimental support for performing
673 the "shrink wrapping" optimization, which moves spills and reloads around in
674 the CFG to avoid doing saves on paths that don't need them.</li>
675<li>LLVM includes new experimental support for writing ELF .o files directly
676 from the compiler. It works well for many simple C testcases, but doesn't
677 support exception handling, debug info, inline assembly, etc.</li>
678<li>Targets can now specify register allocation hints through
679 <tt>MachineRegisterInfo::setRegAllocationHint</tt>. A regalloc hint consists
680 of hint type and physical register number. A hint type of zero specifies a
681 register allocation preference. Other hint type values are target specific
682 which are resolved by <tt>TargetRegisterInfo::ResolveRegAllocHint</tt>. An
683 example is the ARM target which uses register hints to request that the
684 register allocator provide an even / odd register pair to two virtual
685 registers.</li>
Chris Lattner84977642007-09-21 03:54:09 +0000[diff] [blame]686</ul>
Chris Lattner84977642007-09-21 03:54:09 +0000[diff] [blame]687</div>
688
Chris Lattner917cc712009-03-02 02:37:32 +0000[diff] [blame]689<!--=========================================================================-->
690<div class="doc_subsection">
691<a name="x86">X86-32 and X86-64 Target Improvements</a>
692</div>
693
694<div class="doc_text">
695<p>New features of the X86 target include:
696</p>
697
698<ul>
Chris Lattner511433e2009-03-02 03:24:11 +0000[diff] [blame]699
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]700<li>SSE 4.2 builtins are now supported.</li>
701<li>GCC-compatible soft float modes are now supported, which are typically used
702 by OS kernels.</li>
703<li>X86-64 now models implicit zero extensions better, which allows the code
704 generator to remove a lot of redundant zexts. It also models the 8-bit "H"
705 registers as subregs, which allows them to be used in some tricky
706 situations.</li>
707<li>X86-64 now supports the "local exec" and "initial exec" thread local storage
708 model.</li>
709<li>The vector forms of the <a href="LangRef.html#i_icmp">icmp</a> and <a
710 href="LangRef.html#i_fcmp">fcmp</a> instructions now select to efficient
711 SSE operations.</li>
712<li>Support for the win64 calling conventions have improved. The primary
713 missing feature is support for varargs function definitions. It seems to
714 work well for many win64 JIT purposes.</li>
715<li>The X86 backend has preliminary support for <a
716 href="CodeGenerator.html#x86_memory">mapping address spaces to segment
717 register references</a>. This allows you to write GS or FS relative memory
718 accesses directly in LLVM IR for cases where you know exactly what you're
719 doing (such as in an OS kernel). There are some known problems with this
720 support, but it works in simple cases.</li>
721<li>The X86 code generator has been refactored to move all global variable
722 reference logic to one place
723 (<tt>X86Subtarget::ClassifyGlobalReference</tt>) which
724 makes it easier to reason about.</li>
Chris Lattner511433e2009-03-02 03:24:11 +0000[diff] [blame]725
Chris Lattner917cc712009-03-02 02:37:32 +0000[diff] [blame]726</ul>
727
Chris Lattner917cc712009-03-02 02:37:32 +0000[diff] [blame]728</div>
Chris Lattner84977642007-09-21 03:54:09 +0000[diff] [blame]729
730<!--=========================================================================-->
731<div class="doc_subsection">
Chris Lattner11398992009-02-26 07:32:11 +0000[diff] [blame]732<a name="pic16">PIC16 Target Improvements</a>
733</div>
734
735<div class="doc_text">
736<p>New features of the PIC16 target include:
737</p>
738
739<ul>
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]740<li>Support for floating-point, indirect function calls, and
741 passing/returning aggregate types to functions.
742<li>The code generator is able to generate debug info into output COFF files.
743<li>Support for placing an object into a specific section or at a specific
744 address in memory.</li>
Chris Lattner11398992009-02-26 07:32:11 +0000[diff] [blame]745</ul>
746
747<p>Things not yet supported:</p>
748
749<ul>
Chris Lattner11398992009-02-26 07:32:11 +0000[diff] [blame]750<li>Variable arguments.</li>
Chris Lattner511433e2009-03-02 03:24:11 +0000[diff] [blame]751<li>Interrupts/programs.</li>
Chris Lattner11398992009-02-26 07:32:11 +0000[diff] [blame]752</ul>
753
754</div>
755
Chris Lattner11398992009-02-26 07:32:11 +0000[diff] [blame]756<!--=========================================================================-->
757<div class="doc_subsection">
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]758<a name="ARM">ARM Target Improvements</a>
Chris Lattnerc441fb82009-03-01 02:30:21 +0000[diff] [blame]759</div>
760
761<div class="doc_text">
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]762<p>New features of the ARM target include:
763</p>
Chris Lattnerc441fb82009-03-01 02:30:21 +0000[diff] [blame]764
765<ul>
Chris Lattnerc441fb82009-03-01 02:30:21 +0000[diff] [blame]766
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]767<li>Preliminary support for processors, such as the Cortex-A8 and Cortex-A9,
768that implement version v7-A of the ARM architecture. The ARM backend now
769supports both the Thumb2 and Advanced SIMD (Neon) instruction sets.</li>
Chris Lattnerc441fb82009-03-01 02:30:21 +0000[diff] [blame]770
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]771<li>The AAPCS-VFP "hard float" calling conventions are also supported with the
772<tt>-float-abi=hard</tt> flag.</li>
Chris Lattnerc441fb82009-03-01 02:30:21 +0000[diff] [blame]773
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]774<li>The ARM calling convention code is now tblgen generated instead of resorting
775 to C++ code.</li>
776</ul>
Chris Lattnerc441fb82009-03-01 02:30:21 +0000[diff] [blame]777
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]778<p>These features are still somewhat experimental
779and subject to change. The Neon intrinsics, in particular, may change in future
780releases of LLVM. ARMv7 support has progressed a lot on top of tree since 2.6
781branched.</p>
Chris Lattnerc441fb82009-03-01 02:30:21 +0000[diff] [blame]782
Chris Lattnerc441fb82009-03-01 02:30:21 +0000[diff] [blame]783
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]784</div>
Chris Lattnerc441fb82009-03-01 02:30:21 +0000[diff] [blame]785
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]786<!--=========================================================================-->
787<div class="doc_subsection">
788<a name="OtherTarget">Other Target Specific Improvements</a>
789</div>
Chris Lattnerc441fb82009-03-01 02:30:21 +0000[diff] [blame]790
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]791<div class="doc_text">
792<p>New features of other targets include:
793</p>
794
795<ul>
796<li>Mips now supports O32 Calling Convention.</li>
797<li>Many improvements to the 32-bit PowerPC SVR4 ABI (used on powerpc-linux)
798 support, lots of bugs fixed.</li>
799<li>Added support for the 64-bit PowerPC SVR4 ABI (used on powerpc64-linux).
800 Needs more testing.</li>
801</ul>
802
803</div>
804
805<!--=========================================================================-->
806<div class="doc_subsection">
807<a name="newapis">New Useful APIs</a>
808</div>
809
810<div class="doc_text">
811
812<p>This release includes a number of new APIs that are used internally, which
813 may also be useful for external clients.
814</p>
815
816<ul>
817<li>New <a href="http://llvm.org/doxygen/PrettyStackTrace_8h-source.html">
818 <tt>PrettyStackTrace</tt> class</a> allows crashes of llvm tools (and applications
819 that integrate them) to provide more detailed indication of what the
820 compiler was doing at the time of the crash (e.g. running a pass).
821 At the top level for each LLVM tool, it includes the command line arguments.
822 </li>
823<li>New <a href="http://llvm.org/doxygen/StringRef_8h-source.html">StringRef</a>
824 and <a href="http://llvm.org/doxygen/Twine_8h-source.html">Twine</a> classes
825 make operations on character ranges and
826 string concatenation to be more efficient. <tt>StringRef</tt> is just a <tt>const
827 char*</tt> with a length, <tt>Twine</tt> is a light-weight rope.</li>
828<li>LLVM has new <tt>WeakVH</tt>, <tt>AssertingVH</tt> and <tt>CallbackVH</tt>
829 classes, which make it easier to write LLVM IR transformations. <tt>WeakVH</tt>
830 is automatically drops to null when the referenced <tt>Value</tt> is deleted,
831 and is updated across a <tt>replaceAllUsesWith</tt> operation.
832 <tt>AssertingVH</tt> aborts the program if the
833 referenced value is destroyed while it is being referenced. <tt>CallbackVH</tt>
834 is a customizable class for handling value references. See <a
835 href="http://llvm.org/doxygen/ValueHandle_8h-source.html">ValueHandle.h</a>
836 for more information.</li>
837<li>The new '<a href="http://llvm.org/doxygen/Triple_8h-source.html">Triple
838 </a>' class centralizes a lot of logic that reasons about target
839 triples.</li>
840<li>The new '<a href="http://llvm.org/doxygen/ErrorHandling_8h-source.html">
841 llvm_report_error()</a>' set of APIs allows tools to embed the LLVM
842 optimizer and backend and recover from previously unrecoverable errors.</li>
843<li>LLVM has new abstractions for <a
844 href="http://llvm.org/doxygen/Atomic_8h-source.html">atomic operations</a>
845 and <a href="http://llvm.org/doxygen/RWMutex_8h-source.html">reader/writer
846 locks</a>.</li>
847<li>LLVM has new <a href="http://llvm.org/doxygen/SourceMgr_8h-source.html">
848 <tt>SourceMgr</tt> and <tt>SMLoc</tt> classes</a> which implement caret
849 diagnostics and basic include stack processing for simple parsers. It is
850 used by tablegen, llvm-mc, the .ll parser and FileCheck.</li>
851</ul>
852
853
854</div>
855
856<!--=========================================================================-->
857<div class="doc_subsection">
858<a name="otherimprovements">Other Improvements and New Features</a>
859</div>
860
861<div class="doc_text">
862<p>Other miscellaneous features include:</p>
863
864<ul>
865<li>LLVM now includes a new internal '<a
866 href="http://llvm.org/cmds/FileCheck.html">FileCheck</a>' tool which allows
867 writing much more accurate regression tests that run faster. Please see the
868 <a href="TestingGuide.html#FileCheck">FileCheck section of the Testing
869 Guide</a> for more information.</li>
870<li>LLVM profile information support has been significantly improved to produce
871correct use counts, and has support for edge profiling with reduced runtime
872overhead. Combined, the generated profile information is both more correct and
873imposes about half as much overhead (2.6. from 12% to 6% overhead on SPEC
874CPU2000).</li>
875<li>The C bindings (in the llvm/include/llvm-c directory) include many newly
876 supported APIs.</li>
877<li>LLVM 2.6 includes a brand new experimental LLVM bindings to the Ada2005
878 programming language.</li>
879
880<li>The LLVMC driver has several new features:
881 <ul>
882 <li>Dynamic plugins now work on Windows.</li>
883 <li>New option property: init. Makes possible to provide default values for
884 options defined in plugins (interface to <tt>cl::init</tt>).</li>
885 <li>New example: Skeleton, shows how to create a standalone LLVMC-based
886 driver.</li>
887 <li>New example: mcc16, a driver for the PIC16 toolchain.</li>
888 </ul>
889</li>
Mikhail Glushenkovf795ef02009-03-01 18:09:47 +0000[diff] [blame]890
Chris Lattnerc441fb82009-03-01 02:30:21 +0000[diff] [blame]891</ul>
892
893</div>
894
Chris Lattner77d29b12008-06-05 08:02:49 +0000[diff] [blame]895
896<!--=========================================================================-->
897<div class="doc_subsection">
Chris Lattnerf6662f92008-10-13 17:57:36 +0000[diff] [blame]898<a name="changes">Major Changes and Removed Features</a>
899</div>
900
901<div class="doc_text">
902
Chris Lattnereeb4da02008-10-13 22:06:31 +0000[diff] [blame]903<p>If you're already an LLVM user or developer with out-of-tree changes based
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]904on LLVM 2.5, this section lists some "gotchas" that you may run into upgrading
Chris Lattnereeb4da02008-10-13 22:06:31 +0000[diff] [blame]905from the previous release.</p>
Chris Lattnerf6662f92008-10-13 17:57:36 +0000[diff] [blame]906
907<ul>
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]908<li>The Itanium (IA64) backend has been removed. It was not actively supported
909 and had bitrotted.</li>
910<li>The BigBlock register allocator has been removed, it had also bitrotted.</li>
911<li>The C Backend (<tt>-march=c</tt>) is no longer considered part of the LLVM release
912criteria. We still want it to work, but no one is maintaining it and it lacks
913support for arbitrary precision integers and other important IR features.</li>
Chris Lattnerf6662f92008-10-13 17:57:36 +0000[diff] [blame]914
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]915<li>All LLVM tools now default to overwriting their output file, behaving more
916 like standard unix tools. Previously, this only happened with the '<tt>-f</tt>'
917 option.</li>
918<li>LLVM build now builds all libraries as .a files instead of some
919 libraries as relinked .o files. This requires some APIs like
920 InitializeAllTargets.h.
921 </li>
Chris Lattnereeb4da02008-10-13 22:06:31 +0000[diff] [blame]922</ul>
923
924
925<p>In addition, many APIs have changed in this release. Some of the major LLVM
926API changes are:</p>
927
928<ul>
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]929<li>All uses of <tt>hash_set</tt> and <tt>hash_map</tt> have been removed from
930 the LLVM tree and the wrapper headers have been removed.</li>
931<li>The llvm/Streams.h and <tt>DOUT</tt> member of Debug.h have been removed. The
932 <tt>llvm::Ostream</tt> class has been completely removed and replaced with
933 uses of <tt>raw_ostream</tt>.</li>
934<li>LLVM's global uniquing tables for <tt>Type</tt>s and <tt>Constant</tt>s have
935 been privatized into members of an <tt>LLVMContext</tt>. A number of APIs
936 now take an <tt>LLVMContext</tt> as a parameter. To smooth the transition
937 for clients that will only ever use a single context, the new
938 <tt>getGlobalContext()</tt> API can be used to access a default global
939 context which can be passed in any and all cases where a context is
940 required.
941<li>The <tt>getABITypeSize</tt> methods are now called <tt>getAllocSize</tt>.</li>
942<li>The <tt>Add</tt>, <tt>Sub</tt> and <tt>Mul</tt> operators are no longer
943 overloaded for floating-point types. Floating-point addition, subtraction
944 and multiplication are now represented with new operators <tt>FAdd</tt>,
945 <tt>FSub</tt> and <tt>FMul</tt>. In the <tt>IRBuilder</tt> API,
946 <tt>CreateAdd</tt>, <tt>CreateSub</tt>, <tt>CreateMul</tt> and
947 <tt>CreateNeg</tt> should only be used for integer arithmetic now;
948 <tt>CreateFAdd</tt>, <tt>CreateFSub</tt>, <tt>CreateFMul</tt> and
949 <tt>CreateFNeg</tt> should now be used for floating-point arithmetic.</li>
950<li>The <tt>DynamicLibrary</tt> class can no longer be constructed, its functionality has
951 moved to static member functions.</li>
952<li><tt>raw_fd_ostream</tt>'s constructor for opening a given filename now
953 takes an extra <tt>Force</tt> argument. If <tt>Force</tt> is set to
954 <tt>false</tt>, an error will be reported if a file with the given name
955 already exists. If <tt>Force</tt> is set to <tt>true</tt>, the file will
956 be silently truncated (which is the behavior before this flag was
957 added).</li>
958<li><tt>SCEVHandle</tt> no longer exists, because reference counting is no
959 longer done for <tt>SCEV*</tt> objects, instead <tt>const SCEV*</tt>
960 should be used.</li>
961
962<li>Many APIs, notably <tt>llvm::Value</tt>, now use the <tt>StringRef</tt>
963and <tt>Twine</tt> classes instead of passing <tt>const char*</tt>
964or <tt>std::string</tt>, as described in
965the <a href="ProgrammersManual.html#string_apis">Programmer's Manual</a>. Most
966clients should be unaffected by this transition, unless they are used to
967<tt>Value::getName()</tt> returning a string. Here are some tips on updating to
9682.6:
969 <ul>
970 <li><tt>getNameStr()</tt> is still available, and matches the old
971 behavior. Replacing <tt>getName()</tt> calls with this is an safe option,
972 although more efficient alternatives are now possible.</li>
973
974 <li>If you were just relying on <tt>getName()</tt> being able to be sent to
975 a <tt>std::ostream</tt>, consider migrating
976 to <tt>llvm::raw_ostream</tt>.</li>
977
978 <li>If you were using <tt>getName().c_str()</tt> to get a <tt>const
979 char*</tt> pointer to the name, you can use <tt>getName().data()</tt>.
980 Note that this string (as before), may not be the entire name if the
981 name contains embedded null characters.</li>
982
983 <li>If you were using <tt>operator +</tt> on the result of <tt>getName()</tt> and
984 treating the result as an <tt>std::string</tt>, you can either
985 use <tt>Twine::str</tt> to get the result as an <tt>std::string</tt>, or
986 could move to a <tt>Twine</tt> based design.</li>
987
988 <li><tt>isName()</tt> should be replaced with comparison
989 against <tt>getName()</tt> (this is now efficient).
990 </ul>
991</li>
992
993<li>The registration interfaces for backend Targets has changed (what was
994previously <tt>TargetMachineRegistry</tt>). For backend authors, see the <a
995href="WritingAnLLVMBackend.html#TargetRegistration">Writing An LLVM Backend</a>
996guide. For clients, the notable API changes are:
997 <ul>
998 <li><tt>TargetMachineRegistry</tt> has been renamed
999 to <tt>TargetRegistry</tt>.</li>
1000
1001 <li>Clients should move to using the <tt>TargetRegistry::lookupTarget()</tt>
1002 function to find targets.</li>
1003 </ul>
1004</li>
Devang Patelb34dd132008-10-14 20:03:43 +0000[diff] [blame]1005</ul>
Chris Lattnerf6662f92008-10-13 17:57:36 +0000[diff] [blame]1006
Chris Lattnerf6662f92008-10-13 17:57:36 +0000[diff] [blame]1007</div>
1008
1009
1010
Chris Lattner19092612003-10-02 16:38:05 +0000[diff] [blame]1011<!-- *********************************************************************** -->
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1012<div class="doc_section">
1013 <a name="portability">Portability and Supported Platforms</a>
1014</div>
Chris Lattner19092612003-10-02 16:38:05 +0000[diff] [blame]1015<!-- *********************************************************************** -->
1016
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1017<div class="doc_text">
1018
John Criswell0b5b5e92004-12-08 20:35:47 +0000[diff] [blame]1019<p>LLVM is known to work on the following platforms:</p>
Chris Lattner4654bdb2004-06-01 18:22:41 +0000[diff] [blame]1020
1021<ul>
Mikhail Glushenkovf795ef02009-03-01 18:09:47 +0000[diff] [blame]1022<li>Intel and AMD machines (IA32, X86-64, AMD64, EMT-64) running Red Hat
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]1023 Linux, Fedora Core, FreeBSD and AuroraUX (and probably other unix-like
1024 systems).</li>
Mikhail Glushenkovf795ef02009-03-01 18:09:47 +0000[diff] [blame]1025<li>PowerPC and X86-based Mac OS X systems, running 10.3 and above in 32-bit
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]1026 and 64-bit modes.</li>
Chris Lattner252b83d2008-02-06 18:00:06 +0000[diff] [blame]1027<li>Intel and AMD machines running on Win32 using MinGW libraries (native).</li>
Chris Lattner57a460e2007-05-23 04:39:32 +0000[diff] [blame]1028<li>Intel and AMD machines running on Win32 with the Cygwin libraries (limited
1029 support is available for native builds with Visual C++).</li>
Chris Lattner7e23d6e2009-10-16 16:30:58 +0000[diff] [blame]1030<li>Sun x86 and AMD64 machines running Solaris 10, OpenSolaris 0906.</li>
John Criswell9321fa82005-05-13 20:28:15 +0000[diff] [blame]1031<li>Alpha-based machines running Debian GNU/Linux.</li>
Chris Lattner4654bdb2004-06-01 18:22:41 +0000[diff] [blame]1032</ul>
1033
Chris Lattnerbc5786b2008-06-05 06:57:39 +0000[diff] [blame]1034<p>The core LLVM infrastructure uses GNU autoconf to adapt itself
Brian Gaekeb0fd7612004-05-09 05:28:35 +0000[diff] [blame]1035to the machine and operating system on which it is built. However, minor
1036porting may be required to get LLVM to work on new platforms. We welcome your
1037portability patches and reports of successful builds or error messages.</p>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]1038
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1039</div>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]1040
1041<!-- *********************************************************************** -->
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1042<div class="doc_section">
1043 <a name="knownproblems">Known Problems</a>
1044</div>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]1045<!-- *********************************************************************** -->
1046
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1047<div class="doc_text">
1048
Mikhail Glushenkovf795ef02009-03-01 18:09:47 +0000[diff] [blame]1049<p>This section contains significant known problems with the LLVM system,
Chris Lattnere18b32e2008-11-10 05:40:34 +0000[diff] [blame]1050listed by component. If you run into a problem, please check the <a
Chris Lattnerc463b272005-10-29 07:07:09 +0000[diff] [blame]1051href="http://llvm.org/bugs/">LLVM bug database</a> and submit a bug if
Chris Lattner5eccca42003-12-12 21:22:16 +0000[diff] [blame]1052there isn't already one.</p>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]1053
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]1054<ul>
1055<li>The llvm-gcc bootstrap will fail with some versions of binutils (e.g. 2.15)
1056 with a message of "<tt><a href="http://llvm.org/PR5004">Error: can not do 8
1057 byte pc-relative relocation</a></tt>" when building C++ code. We intend to
1058 fix this on mainline, but a workaround for 2.6 is to upgrade to binutils
1059 2.17 or later.</li>
1060
1061<li>LLVM will not correctly compile on Solaris and/or OpenSolaris
1062using the stock GCC 3.x.x series 'out the box',
Chris Lattner554ee4a2009-11-03 21:50:09 +0000[diff] [blame]1063See: <a href="GettingStarted.html#brokengcc">Broken versions of GCC and other tools</a>.
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]1064However, A <a href="http://pkg.auroraux.org/GCC">Modern GCC Build</a>
1065for x86/x86-64 has been made available from the third party AuroraUX Project
1066that has been meticulously tested for bootstrapping LLVM &amp; Clang.</li>
1067</ul>
1068
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1069</div>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]1070
Chris Lattnerf5ee1702004-03-14 02:03:02 +0000[diff] [blame]1071<!-- ======================================================================= -->
1072<div class="doc_subsection">
1073 <a name="experimental">Experimental features included with this release</a>
1074</div>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]1075
Chris Lattnerf5ee1702004-03-14 02:03:02 +0000[diff] [blame]1076<div class="doc_text">
1077
Misha Brukman6df9e2c2004-05-12 21:46:05 +0000[diff] [blame]1078<p>The following components of this LLVM release are either untested, known to
1079be broken or unreliable, or are in early development. These components should
1080not be relied on, and bugs should not be filed against them, but they may be
1081useful 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]1082components, please contact us on the <a
1083href="http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev">LLVMdev list</a>.</p>
Chris Lattnerf5ee1702004-03-14 02:03:02 +0000[diff] [blame]1084
1085<ul>
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]1086<li>The MSIL, Alpha, SPU, MIPS, PIC16, Blackfin, MSP430 and SystemZ backends are
1087 experimental.</li>
Bill Wendling99dac472009-03-02 07:54:14 +0000[diff] [blame]1088<li>The <tt>llc</tt> "<tt>-filetype=asm</tt>" (the default) is the only
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]1089 supported value for this option. The ELF writer is experimental.</li>
1090<li>The implementation of Andersen's Alias Analysis has many known bugs.</li>
Chris Lattnerf5ee1702004-03-14 02:03:02 +0000[diff] [blame]1091</ul>
1092
1093</div>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]1094
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1095<!-- ======================================================================= -->
1096<div class="doc_subsection">
Chris Lattner26299222006-11-18 07:51:14 +0000[diff] [blame]1097 <a name="x86-be">Known problems with the X86 back-end</a>
John Criswellc0c186d2005-11-08 21:11:33 +0000[diff] [blame]1098</div>
1099
1100<div class="doc_text">
1101
1102<ul>
Anton Korobeynikova6094be2008-06-08 10:24:13 +0000[diff] [blame]1103 <li>The X86 backend does not yet support
1104 all <a href="http://llvm.org/PR879">inline assembly that uses the X86
1105 floating point stack</a>. It supports the 'f' and 't' constraints, but not
1106 'u'.</li>
Chris Lattnere6e1b352008-06-08 21:19:07 +0000[diff] [blame]1107 <li>The X86 backend generates inefficient floating point code when configured
1108 to generate code for systems that don't have SSE2.</li>
Duncan Sands47eff2b2008-06-08 19:38:43 +0000[diff] [blame]1109 <li>Win64 code generation wasn't widely tested. Everything should work, but we
John Criswell524a5dd2009-03-02 15:28:15 +0000[diff] [blame]1110 expect small issues to happen. Also, llvm-gcc cannot build the mingw64
1111 runtime currently due
Anton Korobeynikova6094be2008-06-08 10:24:13 +0000[diff] [blame]1112 to <a href="http://llvm.org/PR2255">several</a>
John Criswell524a5dd2009-03-02 15:28:15 +0000[diff] [blame]1113 <a href="http://llvm.org/PR2257">bugs</a> and due to lack of support for
1114 the
1115 'u' inline assembly constraint and for X87 floating point inline assembly.</li>
Dan Gohman8207ba92008-06-08 23:05:11 +0000[diff] [blame]1116 <li>The X86-64 backend does not yet support the LLVM IR instruction
Chris Lattner917cc712009-03-02 02:37:32 +0000[diff] [blame]1117 <tt>va_arg</tt>. Currently, the llvm-gcc and front-ends support variadic
Dan Gohman8207ba92008-06-08 23:05:11 +0000[diff] [blame]1118 argument constructs on X86-64 by lowering them manually.</li>
Chris Lattner26299222006-11-18 07:51:14 +0000[diff] [blame]1119</ul>
1120
1121</div>
1122
1123<!-- ======================================================================= -->
1124<div class="doc_subsection">
1125 <a name="ppc-be">Known problems with the PowerPC back-end</a>
1126</div>
1127
1128<div class="doc_text">
1129
1130<ul>
Nicolas Geoffraye4285dc2007-05-15 09:21:28 +0000[diff] [blame]1131<li>The Linux PPC32/ABI support needs testing for the interpreter and static
Chris Lattner57a460e2007-05-23 04:39:32 +0000[diff] [blame]1132compilation, and lacks support for debug information.</li>
Chris Lattner26299222006-11-18 07:51:14 +0000[diff] [blame]1133</ul>
1134
1135</div>
1136
1137<!-- ======================================================================= -->
1138<div class="doc_subsection">
Chris Lattnerf3e5bc62007-05-14 06:56:09 +0000[diff] [blame]1139 <a name="arm-be">Known problems with the ARM back-end</a>
1140</div>
1141
1142<div class="doc_text">
1143
1144<ul>
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]1145<li>Support for the Advanced SIMD (Neon) instruction set is still incomplete
1146and not well tested. Some features may not work at all, and the code quality
1147may be poor in some cases.</li>
Chris Lattner57a460e2007-05-23 04:39:32 +0000[diff] [blame]1148<li>Thumb mode works only on ARMv6 or higher processors. On sub-ARMv6
Duncan Sandsc90d68b2007-09-26 15:59:54 +0000[diff] [blame]1149processors, thumb programs can crash or produce wrong
Chris Lattner57a460e2007-05-23 04:39:32 +0000[diff] [blame]1150results (<a href="http://llvm.org/PR1388">PR1388</a>).</li>
John Criswell524a5dd2009-03-02 15:28:15 +0000[diff] [blame]1151<li>Compilation for ARM Linux OABI (old ABI) is supported but not fully tested.
Chris Lattnerf3e5bc62007-05-14 06:56:09 +0000[diff] [blame]1152</li>
Chris Lattnerf3e5bc62007-05-14 06:56:09 +0000[diff] [blame]1153</ul>
1154
1155</div>
1156
1157<!-- ======================================================================= -->
1158<div class="doc_subsection">
Chris Lattner26299222006-11-18 07:51:14 +0000[diff] [blame]1159 <a name="sparc-be">Known problems with the SPARC back-end</a>
1160</div>
1161
1162<div class="doc_text">
1163
1164<ul>
John Criswell524a5dd2009-03-02 15:28:15 +0000[diff] [blame]1165<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]1166 support the 64-bit SPARC ABI (-m64).</li>
1167</ul>
1168
1169</div>
1170
1171<!-- ======================================================================= -->
1172<div class="doc_subsection">
Bruno Cardoso Lopesb7e1a4f2008-10-25 14:56:26 +0000[diff] [blame]1173 <a name="mips-be">Known problems with the MIPS back-end</a>
1174</div>
1175
1176<div class="doc_text">
1177
1178<ul>
Bruno Cardoso Lopesb7e1a4f2008-10-25 14:56:26 +0000[diff] [blame]1179<li>64-bit MIPS targets are not supported yet.</li>
1180</ul>
1181
1182</div>
1183
1184<!-- ======================================================================= -->
1185<div class="doc_subsection">
Chris Lattner26299222006-11-18 07:51:14 +0000[diff] [blame]1186 <a name="alpha-be">Known problems with the Alpha back-end</a>
1187</div>
1188
1189<div class="doc_text">
1190
1191<ul>
1192
1193<li>On 21164s, some rare FP arithmetic sequences which may trap do not have the
1194appropriate nops inserted to ensure restartability.</li>
1195
John Criswellc0c186d2005-11-08 21:11:33 +0000[diff] [blame]1196</ul>
1197</div>
1198
Chris Lattner26299222006-11-18 07:51:14 +0000[diff] [blame]1199<!-- ======================================================================= -->
1200<div class="doc_subsection">
Chris Lattnerf3e5bc62007-05-14 06:56:09 +0000[diff] [blame]1201 <a name="c-be">Known problems with the C back-end</a>
Chris Lattner26299222006-11-18 07:51:14 +0000[diff] [blame]1202</div>
1203
1204<div class="doc_text">
1205
1206<ul>
Chris Lattner5733b272008-06-05 06:35:40 +0000[diff] [blame]1207<li><a href="http://llvm.org/PR802">The C backend has only basic support for
1208 inline assembly code</a>.</li>
Chris Lattner725a0d82007-09-26 06:01:35 +0000[diff] [blame]1209<li><a href="http://llvm.org/PR1658">The C backend violates the ABI of common
1210 C++ programs</a>, preventing intermixing between C++ compiled by the CBE and
Gabor Greif4906abe2009-03-02 12:02:51 +0000[diff] [blame]1211 C++ code compiled with <tt>llc</tt> or native compilers.</li>
Duncan Sandsf74c0cc2008-02-10 13:40:55 +0000[diff] [blame]1212<li>The C backend does not support all exception handling constructs.</li>
Duncan Sands50723a92009-02-25 11:51:54 +0000[diff] [blame]1213<li>The C backend does not support arbitrary precision integers.</li>
Chris Lattner26299222006-11-18 07:51:14 +0000[diff] [blame]1214</ul>
1215
1216</div>
John Criswellc0c186d2005-11-08 21:11:33 +0000[diff] [blame]1217
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1218
1219<!-- ======================================================================= -->
1220<div class="doc_subsection">
Chris Lattner5733b272008-06-05 06:35:40 +0000[diff] [blame]1221 <a name="c-fe">Known problems with the llvm-gcc C front-end</a>
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1222</div>
Chris Lattner47588f92003-10-02 05:07:23 +0000[diff] [blame]1223
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1224<div class="doc_text">
Chris Lattnerc5d658a2006-03-03 00:34:26 +0000[diff] [blame]1225
Chris Lattner5733b272008-06-05 06:35:40 +0000[diff] [blame]1226<p>The only major language feature of GCC not supported by llvm-gcc is
1227 the <tt>__builtin_apply</tt> family of builtins. However, some extensions
1228 are only supported on some targets. For example, trampolines are only
Duncan Sands27aff872008-06-08 20:18:35 +0000[diff] [blame]1229 supported on some targets (these are used when you take the address of a
1230 nested function).</p>
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1231
Chris Lattner5733b272008-06-05 06:35:40 +0000[diff] [blame]1232<p>If you run into GCC extensions which are not supported, please let us know.
1233</p>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]1234
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1235</div>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]1236
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1237<!-- ======================================================================= -->
1238<div class="doc_subsection">
Chris Lattner5733b272008-06-05 06:35:40 +0000[diff] [blame]1239 <a name="c++-fe">Known problems with the llvm-gcc C++ front-end</a>
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1240</div>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]1241
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1242<div class="doc_text">
1243
Chris Lattnerf3e5bc62007-05-14 06:56:09 +0000[diff] [blame]1244<p>The C++ front-end is considered to be fully
Chris Lattner7506b1d2004-12-07 08:04:13 +0000[diff] [blame]1245tested and works for a number of non-trivial programs, including LLVM
Chris Lattnerf3e5bc62007-05-14 06:56:09 +0000[diff] [blame]1246itself, Qt, Mozilla, etc.</p>
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1247
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1248<ul>
Anton Korobeynikov0021fc12008-10-11 18:27:16 +0000[diff] [blame]1249<li>Exception handling works well on the X86 and PowerPC targets. Currently
Chris Lattnere18b32e2008-11-10 05:40:34 +0000[diff] [blame]1250 only Linux and Darwin targets are supported (both 32 and 64 bit).</li>
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1251</ul>
Chris Lattnerfcc54b32003-10-07 22:14:37 +0000[diff] [blame]1252
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1253</div>
1254
Chris Lattner1eb4df62008-10-30 03:58:13 +0000[diff] [blame]1255<!-- ======================================================================= -->
1256<div class="doc_subsection">
1257 <a name="fortran-fe">Known problems with the llvm-gcc Fortran front-end</a>
1258</div>
1259
1260<div class="doc_text">
Gabor Greifba10fe02008-11-04 21:50:59 +0000[diff] [blame]1261<ul>
Chris Lattner1eb4df62008-10-30 03:58:13 +0000[diff] [blame]1262<li>Fortran support generally works, but there are still several unresolved bugs
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]1263 in <a href="http://llvm.org/bugs/">Bugzilla</a>. Please see the
1264 tools/gfortran component for details.</li>
Gabor Greifba10fe02008-11-04 21:50:59 +0000[diff] [blame]1265</ul>
Chris Lattner1eb4df62008-10-30 03:58:13 +0000[diff] [blame]1266</div>
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1267
Chris Lattner2b659ef2008-02-12 06:29:45 +0000[diff] [blame]1268<!-- ======================================================================= -->
1269<div class="doc_subsection">
Chris Lattner5733b272008-06-05 06:35:40 +0000[diff] [blame]1270 <a name="ada-fe">Known problems with the llvm-gcc Ada front-end</a>
Chris Lattner2b659ef2008-02-12 06:29:45 +0000[diff] [blame]1271</div>
1272
1273<div class="doc_text">
John Criswell524a5dd2009-03-02 15:28:15 +0000[diff] [blame]1274The llvm-gcc 4.2 Ada compiler works fairly well; however, this is not a mature
1275technology, and problems should be expected.
Chris Lattner2b659ef2008-02-12 06:29:45 +0000[diff] [blame]1276<ul>
Duncan Sands27aff872008-06-08 20:18:35 +0000[diff] [blame]1277<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]1278to lack of trampoline support (pointers to nested functions) on other platforms.
1279However, 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]1280which does support trampolines.</li>
1281<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]1282This is due to lack of LLVM support for <tt>setjmp</tt>/<tt>longjmp</tt> style
1283exception handling, which is used internally by the compiler.
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]1284Workaround: configure with <tt>--disable-bootstrap</tt>.</li>
Duncan Sands978bcee2008-10-13 17:27:23 +0000[diff] [blame]1285<li>The c380004, <a href="http://llvm.org/PR2010">c393010</a>
1286and <a href="http://llvm.org/PR2421">cxg2021</a> ACATS tests fail
Duncan Sands326a4982009-02-25 11:59:06 +0000[diff] [blame]1287(c380004 also fails with gcc-4.2 mainline).
1288If the compiler is built with checks disabled then <a href="http://llvm.org/PR2010">c393010</a>
1289causes the compiler to go into an infinite loop, using up all system memory.</li>
Duncan Sandsdd3e6722009-03-02 16:35:57 +0000[diff] [blame]1290<li>Some GCC specific Ada tests continue to crash the compiler.</li>
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]1291<li>The <tt>-E</tt> binder option (exception backtraces)
Chris Lattner2b659ef2008-02-12 06:29:45 +0000[diff] [blame]1292<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]1293crashing if an exception is raised. Workaround: do not use <tt>-E</tt>.</li>
Chris Lattner2b659ef2008-02-12 06:29:45 +0000[diff] [blame]1294<li>Only discrete types <a href="http://llvm.org/PR1981">are allowed to start
1295or finish at a non-byte offset</a> in a record. Workaround: do not pack records
1296or use representation clauses that result in a field of a non-discrete type
1297starting or finishing in the middle of a byte.</li>
Chris Lattnere6e1b352008-06-08 21:19:07 +0000[diff] [blame]1298<li>The <tt>lli</tt> interpreter <a href="http://llvm.org/PR2009">considers
1299'main' as generated by the Ada binder to be invalid</a>.
1300Workaround: hand edit the file to use pointers for <tt>argv</tt> and
1301<tt>envp</tt> rather than integers.</li>
1302<li>The <tt>-fstack-check</tt> option <a href="http://llvm.org/PR2008">is
1303ignored</a>.</li>
Chris Lattner2b659ef2008-02-12 06:29:45 +0000[diff] [blame]1304</ul>
1305</div>
1306
Chris Lattner61358ab2009-10-13 17:48:04 +0000[diff] [blame]1307<!-- ======================================================================= -->
1308<div class="doc_subsection">
1309 <a name="ocaml-bindings">Known problems with the O'Caml bindings</a>
1310</div>
1311
1312<div class="doc_text">
1313
1314<p>The <tt>Llvm.Linkage</tt> module is broken, and has incorrect values. Only
1315<tt>Llvm.Linkage.External</tt>, <tt>Llvm.Linkage.Available_externally</tt>, and
1316<tt>Llvm.Linkage.Link_once</tt> will be correct. If you need any of the other linkage
1317modes, you'll have to write an external C library in order to expose the
1318functionality. This has been fixed in the trunk.</p>
1319</div>
1320
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]1321<!-- *********************************************************************** -->
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1322<div class="doc_section">
1323 <a name="additionalinfo">Additional Information</a>
1324</div>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]1325<!-- *********************************************************************** -->
1326
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1327<div class="doc_text">
1328
Chris Lattner416db102005-05-16 17:13:10 +0000[diff] [blame]1329<p>A wide variety of additional information is available on the <a
Chris Lattnerb4b0ce72007-05-18 00:44:29 +0000[diff] [blame]1330href="http://llvm.org">LLVM web page</a>, in particular in the <a
1331href="http://llvm.org/docs/">documentation</a> section. The web page also
Reid Spencer669ed452007-07-09 08:04:31 +0000[diff] [blame]1332contains versions of the API documentation which is up-to-date with the
1333Subversion version of the source code.
Misha Brukman109d9e82005-03-30 19:14:24 +0000[diff] [blame]1334You can access versions of these documents specific to this release by going
1335into the "<tt>llvm/doc/</tt>" directory in the LLVM tree.</p>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]1336
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1337<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]1338us via the <a href="http://llvm.org/docs/#maillist"> mailing
Chris Lattner5eccca42003-12-12 21:22:16 +0000[diff] [blame]1339lists</a>.</p>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]1340
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1341</div>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]1342
1343<!-- *********************************************************************** -->
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]1344
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1345<hr>
Misha Brukman2061e892003-11-22 01:23:39 +0000[diff] [blame]1346<address>
Misha Brukman38847d52003-12-21 22:53:21 +0000[diff] [blame]1347 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
Misha Brukman44408702008-12-11 17:34:48 +0000[diff] [blame]1348 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
Misha Brukman2061e892003-11-22 01:23:39 +0000[diff] [blame]1349 <a href="http://validator.w3.org/check/referer"><img
Misha Brukman44408702008-12-11 17:34:48 +0000[diff] [blame]1350 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]1351
Chris Lattnerb4b0ce72007-05-18 00:44:29 +0000[diff] [blame]1352 <a href="http://llvm.org/">LLVM Compiler Infrastructure</a><br>
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1353 Last modified: $Date$
Misha Brukman2061e892003-11-22 01:23:39 +0000[diff] [blame]1354</address>
Chris Lattner79c3fe12003-10-02 04:57:28 +0000[diff] [blame]1355
Misha Brukman500bc302003-11-22 00:38:41 +0000[diff] [blame]1356</body>
1357</html>