Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / dc87c86efbf6b7473c2f1771383dba2011bb33eb / . / docs / ReleaseNotes.html
blob: bdaba71b7430d341e663a0b1df71528fd743ff45 [file] [log] [blame]
Dan Gohmanf17a25c2007-07-18 16:29:46 +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>
5 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
6 <link rel="stylesheet" href="llvm.css" type="text/css">
Chris Lattner95d25072009-02-25 04:41:31 +0000[diff] [blame]7 <title>LLVM 2.5 Release Notes</title>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]8</head>
9<body>
10
Chris Lattner95d25072009-02-25 04:41:31 +0000[diff] [blame]11<div class="doc_title">LLVM 2.5 Release Notes</div>
Mikhail Glushenkov808cd3e2008-10-13 02:08:34 +0000[diff] [blame]12
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]13<ol>
14 <li><a href="#intro">Introduction</a></li>
Chris Lattner78a130b2008-10-13 18:01:01 +0000[diff] [blame]15 <li><a href="#subproj">Sub-project Status Update</a></li>
Chris Lattnerb2cc5022009-02-26 22:33:38 +0000[diff] [blame]16 <li><a href="#externalproj">External Projects Using LLVM 2.5</a></li>
Chris Lattner47202c32009-03-02 02:37:32 +0000[diff] [blame]17 <li><a href="#whatsnew">What's New in LLVM 2.5?</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]18 <li><a href="GettingStarted.html">Installation Instructions</a></li>
19 <li><a href="#portability">Portability and Supported Platforms</a></li>
Dan Gohman0de549c2008-10-14 16:23:02 +0000[diff] [blame]20 <li><a href="#knownproblems">Known Problems</a></li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]21 <li><a href="#additionalinfo">Additional Information</a></li>
22</ol>
23
24<div class="doc_author">
Dan Gohman0de549c2008-10-14 16:23:02 +0000[diff] [blame]25 <p>Written by the <a href="http://llvm.org">LLVM Team</a></p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]26</div>
27
28<!-- *********************************************************************** -->
29<div class="doc_section">
30 <a name="intro">Introduction</a>
31</div>
32<!-- *********************************************************************** -->
33
34<div class="doc_text">
35
Chris Lattner92eacf82008-10-13 17:57:36 +0000[diff] [blame]36<p>This document contains the release notes for the LLVM Compiler
Chris Lattner95d25072009-02-25 04:41:31 +0000[diff] [blame]37Infrastructure, release 2.5. Here we describe the status of LLVM, including
Chris Lattner92eacf82008-10-13 17:57:36 +0000[diff] [blame]38major improvements from the previous release and significant known problems.
Mikhail Glushenkovf61066a2009-03-01 18:09:47 +0000[diff] [blame]39All LLVM releases may be downloaded from the <a
Chris Lattner92eacf82008-10-13 17:57:36 +0000[diff] [blame]40href="http://llvm.org/releases/">LLVM releases web site</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]41
42<p>For more information about LLVM, including information about the latest
43release, please check out the <a href="http://llvm.org/">main LLVM
44web site</a>. If you have questions or comments, the <a
Chris Lattner92eacf82008-10-13 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>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]47
Mikhail Glushenkov808cd3e2008-10-13 02:08:34 +0000[diff] [blame]48<p>Note that if you are reading this file from a Subversion checkout or the
Chris Lattnerb0d6c502007-09-21 03:54:09 +0000[diff] [blame]49main LLVM web page, this document applies to the <i>next</i> release, not the
Gabor Greif7c65c4f2008-10-14 11:00:32 +0000[diff] [blame]50current one. To see the release notes for a specific release, please see the
Chris Lattnerb0d6c502007-09-21 03:54:09 +0000[diff] [blame]51<a href="http://llvm.org/releases/">releases page</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]52
53</div>
54
Chris Lattner95d25072009-02-25 04:41:31 +0000[diff] [blame]55<!-- Unfinished features in 2.5:
Chris Lattner8fda4952008-02-10 07:46:44 +0000[diff] [blame]56 Machine LICM
57 Machine Sinking
Chris Lattnera53f4972009-02-25 06:34:50 +0000[diff] [blame]58 target-specific intrinsics
59 gold lto plugin
Chris Lattner56b64702009-02-25 07:32:53 +0000[diff] [blame]60 pre-alloc splitter, strong phi elim
Gabor Greife33c5322009-03-02 12:02:51 +0000[diff] [blame]61 <tt>llc -enable-value-prop</tt>, propagation of value info
62 (sign/zero ext info) from one MBB to another
Chris Lattnera53f4972009-02-25 06:34:50 +0000[diff] [blame]63 debug info for optimized code
Chris Lattner5c109932009-02-26 07:29:17 +0000[diff] [blame]64 interpreter + libffi
Chris Lattnerb2cc5022009-02-26 22:33:38 +0000[diff] [blame]65 postalloc scheduler: anti dependence breaking, hazard recognizer?
66
Chris Lattner47202c32009-03-02 02:37:32 +0000[diff] [blame]67initial support for debug line numbers when optimization enabled, not useful in
68 2.5 but will be for 2.6.
69
Chris Lattner8fda4952008-02-10 07:46:44 +0000[diff] [blame]70 -->
Mikhail Glushenkov808cd3e2008-10-13 02:08:34 +0000[diff] [blame]71
Chris Lattner4f076402008-10-12 19:47:48 +0000[diff] [blame]72 <!-- for announcement email:
Chris Lattnera53f4972009-02-25 06:34:50 +0000[diff] [blame]73 -->
Chris Lattner8fda4952008-02-10 07:46:44 +0000[diff] [blame]74
Chris Lattner78a130b2008-10-13 18:01:01 +0000[diff] [blame]75<!-- *********************************************************************** -->
76<div class="doc_section">
77 <a name="subproj">Sub-project Status Update</a>
Chris Lattnerf65224f2008-06-08 21:34:41 +0000[diff] [blame]78</div>
Chris Lattner78a130b2008-10-13 18:01:01 +0000[diff] [blame]79<!-- *********************************************************************** -->
Chris Lattnerf65224f2008-06-08 21:34:41 +0000[diff] [blame]80
81<div class="doc_text">
Chris Lattnerf6a0d382008-06-08 21:58:17 +0000[diff] [blame]82<p>
Chris Lattner95d25072009-02-25 04:41:31 +0000[diff] [blame]83The LLVM 2.5 distribution currently consists of code from the core LLVM
Bill Wendlinga8fb81d2009-03-02 04:28:18 +0000[diff] [blame]84repository &mdash;which roughly includes the LLVM optimizers, code generators
85and supporting tools &mdash; and the llvm-gcc repository. In addition to this
86code, the LLVM Project includes other sub-projects that are in development. The
87two which are the most actively developed are the <a href="#clang">Clang
88Project</a> and the <a href="#vmkit">VMKit Project</a>.
Bill Wendlingdde41b82009-03-02 04:28:57 +0000[diff] [blame]89</p>
Chris Lattner78a130b2008-10-13 18:01:01 +0000[diff] [blame]90
Chris Lattnerf6a0d382008-06-08 21:58:17 +0000[diff] [blame]91</div>
92
Chris Lattner78a130b2008-10-13 18:01:01 +0000[diff] [blame]93
Chris Lattnerf6a0d382008-06-08 21:58:17 +0000[diff] [blame]94<!--=========================================================================-->
Chris Lattner78a130b2008-10-13 18:01:01 +0000[diff] [blame]95<div class="doc_subsection">
Chris Lattnera3c683b2008-10-13 18:11:54 +0000[diff] [blame]96<a name="clang">Clang: C/C++/Objective-C Frontend Toolkit</a>
Chris Lattnerf6a0d382008-06-08 21:58:17 +0000[diff] [blame]97</div>
98
99<div class="doc_text">
100
101<p>The <a href="http://clang.llvm.org/">Clang project</a> is an effort to build
Bill Wendlinga8fb81d2009-03-02 04:28:18 +0000[diff] [blame]102a set of new 'LLVM native' front-end technologies for the LLVM optimizer and
103code generator. While Clang is not included in the LLVM 2.5 release, it is
104continuing to make major strides forward in all areas. Its C and Objective-C
Chris Lattnerc97507b2009-02-25 05:09:54 +0000[diff] [blame]105parsing and code generation support is now very solid. For example, it is
Bill Wendlinga8fb81d2009-03-02 04:28:18 +0000[diff] [blame]106capable of successfully building many real-world applications for X86-32
Daniel Dunbar568c9792009-03-02 20:08:12 +0000[diff] [blame]107and X86-64,
108including the <a href="http://wiki.freebsd.org/BuildingFreeBSDWithClang">FreeBSD
109kernel</a> and <a href="http://gcc.gnu.org/gcc-4.2/">gcc 4.2</a>. C++ is also
Bill Wendlinga8fb81d2009-03-02 04:28:18 +0000[diff] [blame]110making <a href="http://clang.llvm.org/cxx_status.html">incredible progress</a>,
Daniel Dunbar568c9792009-03-02 20:08:12 +0000[diff] [blame]111and work on templates has recently started. If you are
Chris Lattnerc97507b2009-02-25 05:09:54 +0000[diff] [blame]112interested in fast compiles and good diagnostics, we encourage you to try it out
Chris Lattnera53f4972009-02-25 06:34:50 +0000[diff] [blame]113by <a href="http://clang.llvm.org/get_started.html">building from mainline</a>
Mikhail Glushenkovf61066a2009-03-01 18:09:47 +0000[diff] [blame]114and reporting any issues you hit to the <a
Chris Lattnerc97507b2009-02-25 05:09:54 +0000[diff] [blame]115href="http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev">Clang front-end mailing
116list</a>.</p>
117
118<p>In the LLVM 2.5 time-frame, the Clang team has made many improvements:</p>
Bill Wendling385b0d32008-10-27 09:27:33 +0000[diff] [blame]119
Daniel Dunbarf3e35782008-10-14 23:25:09 +0000[diff] [blame]120<ul>
Chris Lattnerc97507b2009-02-25 05:09:54 +0000[diff] [blame]121<li>Clang now has a new driver, which is focused on providing a GCC-compatible
Mikhail Glushenkovf61066a2009-03-01 18:09:47 +0000[diff] [blame]122 interface.</li>
Daniel Dunbar568c9792009-03-02 20:08:12 +0000[diff] [blame]123<li>The X86-64 ABI is now supported, including support for the Apple
124 64-bit Objective-C runtime and zero cost exception handling.</li>
Mikhail Glushenkovf61066a2009-03-01 18:09:47 +0000[diff] [blame]125<li>Precompiled header support is now implemented.</li>
Chris Lattnerc97507b2009-02-25 05:09:54 +0000[diff] [blame]126<li>Objective-C support is significantly improved beyond LLVM 2.4, supporting
127 many features, such as Objective-C Garbage Collection.</li>
Daniel Dunbar568c9792009-03-02 20:08:12 +0000[diff] [blame]128<li>Variable length arrays are now fully supported.</li>
129<li>C99 designated initializers are now fully supported.</li>
130<li>Clang now includes all major compiler headers, including a
131 redesigned <i>tgmath.h</i> and several more intrinsic headers.</li>
Chris Lattner47202c32009-03-02 02:37:32 +0000[diff] [blame]132<li>Many many bugs are fixed and many features have been added.</li>
Bill Wendlinga8fb81d2009-03-02 04:28:18 +0000[diff] [blame]133</ul>
Chris Lattnera3c683b2008-10-13 18:11:54 +0000[diff] [blame]134</div>
135
136<!--=========================================================================-->
137<div class="doc_subsection">
138<a name="clangsa">Clang Static Analyzer</a>
139</div>
140
141<div class="doc_text">
142
Ted Kremeneka0d77d72009-02-27 07:01:32 +0000[diff] [blame]143<p>Previously announced in the last LLVM release, the Clang project also
144includes an early stage static source code analysis tool for <a
145href="http://clang.llvm.org/StaticAnalysis.html">automatically finding bugs</a>
146in C and Objective-C programs. The tool performs a growing set of checks to find
147bugs that occur on a specific path within a program.</p>
Chris Lattnerc2d84672008-10-14 00:52:49 +0000[diff] [blame]148
Ted Kremeneka0d77d72009-02-27 07:01:32 +0000[diff] [blame]149<p>In the LLVM 2.5 time-frame there have been many significant improvements to
150the analyzer's core path simulation engine and machinery for generating
151path-based bug reports to end-users. Particularly noteworthy improvements
152include experimental support for full field-sensitivity and reasoning about heap
153objects as well as an improved value-constraints subengine that does a much
154better job of reasoning about inequality relationships (e.g., <tt>x &gt; 2</tt>)
155between variables and constants.
Ted Kremenek0437b2c2008-10-14 05:14:21 +0000[diff] [blame]156
John Criswellaea8c752009-03-02 15:28:15 +0000[diff] [blame]157<p>The set of checks performed by the static analyzer continues to expand, and
Ted Kremeneka0d77d72009-02-27 07:01:32 +0000[diff] [blame]158future plans for the tool include full source-level inter-procedural analysis
159and deeper checks such as buffer overrun detection. There are many opportunities
160to extend and enhance the static analyzer, and anyone interested in working on
161this project is encouraged to get involved!</p>
Chris Lattnerf6a0d382008-06-08 21:58:17 +0000[diff] [blame]162
163</div>
164
Chris Lattner78a130b2008-10-13 18:01:01 +0000[diff] [blame]165<!--=========================================================================-->
166<div class="doc_subsection">
Nicolas Geoffray2b139382008-10-14 19:23:04 +0000[diff] [blame]167<a name="vmkit">VMKit: JVM/CLI Virtual Machine Implementation</a>
Chris Lattner78a130b2008-10-13 18:01:01 +0000[diff] [blame]168</div>
169
170<div class="doc_text">
171<p>
Nicolas Geoffray2b139382008-10-14 19:23:04 +0000[diff] [blame]172The <a href="http://vmkit.llvm.org/">VMKit project</a> is an implementation of
Chris Lattner78a130b2008-10-13 18:01:01 +0000[diff] [blame]173a JVM and a CLI Virtual Machines (Microsoft .NET is an
174implementation of the CLI) using the Just-In-Time compiler of LLVM.</p>
175
Nicolas Geoffray8c3565e2009-03-02 09:25:38 +0000[diff] [blame]176<p>Following LLVM 2.5, VMKit has its second release that you can find on its
Nicolas Geoffray20cfdd42008-10-15 20:25:04 +0000[diff] [blame]177<a href="http://vmkit.llvm.org/releases/">webpage</a>. The release includes
178bug fixes, cleanup and new features. The major changes are:</p>
Chris Lattner78a130b2008-10-13 18:01:01 +0000[diff] [blame]179
Nicolas Geoffray2b139382008-10-14 19:23:04 +0000[diff] [blame]180<ul>
181
Chris Lattner47bcf612009-02-26 05:20:36 +0000[diff] [blame]182<li>Ahead of Time compiler: compiles .class files to llvm .bc. VMKit uses this
Gabor Greifed585c62009-03-02 11:34:51 +0000[diff] [blame]183functionality to native compile the standard classes (e.g. java.lang.String).
Gabor Greife33c5322009-03-02 12:02:51 +0000[diff] [blame]184Users can compile AoT .class files into dynamic libraries and run them with the
Chris Lattner47bcf612009-02-26 05:20:36 +0000[diff] [blame]185help of VMKit.</li>
186
187<li>New exception model: the dwarf exception model is very slow for
188exception-intensive applications, so the JVM has had a new implementation of
189exceptions which check at each function call if an exception happened. There is
190a low performance penalty on applications without exceptions, but it is a big
191gain for exception-intensive applications. For example the jack benchmark in
192Spec JVM98 is 6x faster (performance gain of 83%).</li>
193
Nicolas Geoffray8c3565e2009-03-02 09:25:38 +0000[diff] [blame]194<li>User-level management of thread stacks, so that thread local data access
195at runtime is fast and portable. </li>
196
197<li>Implementation of biased locking for faster object synchronizations at
198runtime.</li>
199
200<li>New support for OSX/X64, Linux/X64 (with the Boehm GC) and Linux/ppc32.</li>
Nicolas Geoffray2b139382008-10-14 19:23:04 +0000[diff] [blame]201
202</ul>
Chris Lattner78a130b2008-10-13 18:01:01 +0000[diff] [blame]203</div>
204
Chris Lattnerb2cc5022009-02-26 22:33:38 +0000[diff] [blame]205<!-- *********************************************************************** -->
206<div class="doc_section">
207 <a name="externalproj">External Projects Using LLVM 2.5</a>
208</div>
209<!-- *********************************************************************** -->
210
211<!--=========================================================================-->
212<div class="doc_subsection">
213<a name="pure">Pure</a>
214</div>
215
216<div class="doc_text">
217<p>
Chris Lattner47202c32009-03-02 02:37:32 +0000[diff] [blame]218<a href="http://pure-lang.googlecode.com/">Pure</a>
219is an algebraic/functional programming language based on term rewriting.
Chris Lattnerb2cc5022009-02-26 22:33:38 +0000[diff] [blame]220Programs are collections of equations which are used to evaluate expressions in
221a symbolic fashion. Pure offers dynamic typing, eager and lazy evaluation,
Mikhail Glushenkovf61066a2009-03-01 18:09:47 +0000[diff] [blame]222lexical closures, a hygienic macro system (also based on term rewriting),
Chris Lattnerb2cc5022009-02-26 22:33:38 +0000[diff] [blame]223built-in list and matrix support (including list and matrix comprehensions) and
224an easy-to-use C interface. The interpreter uses LLVM as a backend to
225 JIT-compile Pure programs to fast native code.</p>
226
227<p>In addition to the usual algebraic data structures, Pure also has
228MATLAB-style matrices in order to support numeric computations and signal
229processing in an efficient way. Pure is mainly aimed at mathematical
230applications right now, but it has been designed as a general purpose language.
231The dynamic interpreter environment and the C interface make it possible to use
232it as a kind of functional scripting language for many application areas.
233</p>
234</div>
235
236
237<!--=========================================================================-->
238<div class="doc_subsection">
239<a name="ldc">LLVM D Compiler</a>
240</div>
241
242<div class="doc_text">
243<p>
Chris Lattner47202c32009-03-02 02:37:32 +0000[diff] [blame]244<a href="http://www.dsource.org/projects/ldc">LDC</a> is an implementation of
245the D Programming Language using the LLVM optimizer and code generator.
John Criswellaea8c752009-03-02 15:28:15 +0000[diff] [blame]246The LDC project works great with the LLVM 2.5 release. General improvements in
247this
Chris Lattner47202c32009-03-02 02:37:32 +0000[diff] [blame]248cycle have included new inline asm constraint handling, better debug info
249support, general bugfixes, and better x86-64 support. This has allowed
Chris Lattnerb2cc5022009-02-26 22:33:38 +0000[diff] [blame]250some major improvements in LDC, getting us much closer to being as
251fully featured as the original DMD compiler from DigitalMars.
252</p>
253</div>
254
Chris Lattner1ad599a2009-02-28 18:58:01 +0000[diff] [blame]255<!--=========================================================================-->
256<div class="doc_subsection">
257<a name="RoadsendPHP">Roadsend PHP</a>
258</div>
259
260<div class="doc_text">
Chris Lattner47202c32009-03-02 02:37:32 +0000[diff] [blame]261<p><a href="http://code.roadsend.com/rphp">Roadsend PHP</a> (rphp) is an open
Chris Lattner9e08de12009-03-02 19:07:24 +0000[diff] [blame]262source implementation of the PHP programming
263language that uses LLVM for its optimizer, JIT, and static compiler. This is a
264reimplementation of an earlier project that is now based on LLVM.</p>
Chris Lattner1ad599a2009-02-28 18:58:01 +0000[diff] [blame]265</div>
Chris Lattnerb2cc5022009-02-26 22:33:38 +0000[diff] [blame]266
Chris Lattnerf6a0d382008-06-08 21:58:17 +0000[diff] [blame]267
268<!-- *********************************************************************** -->
269<div class="doc_section">
Chris Lattner47202c32009-03-02 02:37:32 +0000[diff] [blame]270 <a name="whatsnew">What's New in LLVM 2.5?</a>
Chris Lattnerf6a0d382008-06-08 21:58:17 +0000[diff] [blame]271</div>
272<!-- *********************************************************************** -->
273
Chris Lattner5cce7a52008-06-08 22:59:35 +0000[diff] [blame]274<div class="doc_text">
275
Bill Wendling385b0d32008-10-27 09:27:33 +0000[diff] [blame]276<p>This release includes a huge number of bug fixes, performance tweaks, and
Chris Lattner92eacf82008-10-13 17:57:36 +0000[diff] [blame]277minor improvements. Some of the major improvements and new features are listed
278in this section.
Chris Lattner5cce7a52008-06-08 22:59:35 +0000[diff] [blame]279</p>
280</div>
281
Chris Lattnerf6a0d382008-06-08 21:58:17 +0000[diff] [blame]282<!--=========================================================================-->
Chris Lattnerf65224f2008-06-08 21:34:41 +0000[diff] [blame]283<div class="doc_subsection">
Chris Lattnere0798082008-02-06 18:00:06 +0000[diff] [blame]284<a name="majorfeatures">Major New Features</a>
285</div>
286
287<div class="doc_text">
288
Chris Lattner95d25072009-02-25 04:41:31 +0000[diff] [blame]289<p>LLVM 2.5 includes several major new capabilities:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]290
291<ul>
Chris Lattner47202c32009-03-02 02:37:32 +0000[diff] [blame]292<li>LLVM 2.5 includes a brand new <a
293href="http://en.wikipedia.org/wiki/XCore">XCore</a> backend.</li>
Chris Lattnera53f4972009-02-25 06:34:50 +0000[diff] [blame]294
Bill Wendling69afba82009-03-02 07:42:15 +0000[diff] [blame]295<li>llvm-gcc now generally supports the GFortran front-end, and the precompiled
Chris Lattner47202c32009-03-02 02:37:32 +0000[diff] [blame]296release binaries now support Fortran, even on Mac OS/X.</li>
Chris Lattnera53f4972009-02-25 06:34:50 +0000[diff] [blame]297
Chris Lattner47202c32009-03-02 02:37:32 +0000[diff] [blame]298<li>CMake is now used by the <a href="GettingStartedVS.html">LLVM build process
299on Windows</a>. It automatically generates Visual Studio project files (and
300more) from a set of simple text files. This makes it much easier to
301maintain. In time, we'd like to standardize on CMake for everything.</li>
Chris Lattner5c109932009-02-26 07:29:17 +0000[diff] [blame]302
Chris Lattner47202c32009-03-02 02:37:32 +0000[diff] [blame]303<li>LLVM 2.5 now uses (and includes) Google Test for unit testing.</li>
Chris Lattner5c109932009-02-26 07:29:17 +0000[diff] [blame]304
Chris Lattner47202c32009-03-02 02:37:32 +0000[diff] [blame]305<li>The LLVM native code generator now supports arbitrary precision integers.
306Types like <tt>i33</tt> have long been valid in the LLVM IR, but were previously
307only supported by the interpreter. Note that the C backend still does not
308support these.</li>
Chris Lattner56b64702009-02-25 07:32:53 +0000[diff] [blame]309
John Criswellaea8c752009-03-02 15:28:15 +0000[diff] [blame]310<li>LLVM 2.5 no longer uses 'bison,' so it is easier to build on Windows.</li>
Chris Lattner5ba16862008-02-10 08:18:42 +0000[diff] [blame]311</ul>
312
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]313</div>
314
Chris Lattnerc5a4e312008-06-05 06:25:56 +0000[diff] [blame]315
316<!--=========================================================================-->
317<div class="doc_subsection">
Chris Lattnerf65224f2008-06-08 21:34:41 +0000[diff] [blame]318<a name="llvm-gcc">llvm-gcc 4.2 Improvements</a>
Chris Lattnerc5a4e312008-06-05 06:25:56 +0000[diff] [blame]319</div>
320
321<div class="doc_text">
322
Chris Lattner4352cc82008-10-13 20:47:20 +0000[diff] [blame]323<p>LLVM fully supports the llvm-gcc 4.2 front-end, which marries the GCC
324front-ends and driver with the LLVM optimizer and code generator. It currently
325includes support for the C, C++, Objective-C, Ada, and Fortran front-ends.</p>
Chris Lattnerc5a4e312008-06-05 06:25:56 +0000[diff] [blame]326
Chris Lattner5cce7a52008-06-08 22:59:35 +0000[diff] [blame]327<ul>
Chris Lattner47202c32009-03-02 02:37:32 +0000[diff] [blame]328<li>In this release, the GCC inliner is completely disabled. Previously the GCC
329inliner was used to handle always-inline functions and other cases. This caused
330problems with code size growth, and it is completely disabled in this
331release.</li>
332
333<li>llvm-gcc (and LLVM in general) now support code generation for stack
334canaries, which is an effective form of <a
335href="http://en.wikipedia.org/wiki/Stack-smashing_protection">buffer overflow
336protection</a>. llvm-gcc supports this with the <tt>-fstack-protector</tt>
337command line option (just like GCC). In LLVM IR, you can request code
338generation for stack canaries with function attributes.
339</li>
Gabor Greif0e49cc82008-06-09 06:06:18 +0000[diff] [blame]340</ul>
Chris Lattnerc5a4e312008-06-05 06:25:56 +0000[diff] [blame]341
Chris Lattnerc5a4e312008-06-05 06:25:56 +0000[diff] [blame]342</div>
343
344
Chris Lattnerb0d6c502007-09-21 03:54:09 +0000[diff] [blame]345<!--=========================================================================-->
346<div class="doc_subsection">
Chris Lattner47202c32009-03-02 02:37:32 +0000[diff] [blame]347<a name="coreimprovements">LLVM IR and Core Improvements</a>
Chris Lattnerc3649452008-02-10 08:17:19 +0000[diff] [blame]348</div>
349
350<div class="doc_text">
Chris Lattner47202c32009-03-02 02:37:32 +0000[diff] [blame]351<p>LLVM IR has several new features that are used by our existing front-ends and
352can be useful if you are writing a front-end for LLVM:</p>
Chris Lattnerc3649452008-02-10 08:17:19 +0000[diff] [blame]353
Chris Lattnerc5a4e312008-06-05 06:25:56 +0000[diff] [blame]354<ul>
Chris Lattner47202c32009-03-02 02:37:32 +0000[diff] [blame]355<li>The <a href="LangRef.html#i_shufflevector">shufflevector</a> instruction
356has been generalized to allow different shuffle mask width than its input
357vectors. This allows you to use shufflevector to combine two
358"&lt;4 x float&gt;" vectors into a "&lt;8 x float&gt;" for example.</li>
359
360<li>LLVM IR now supports new intrinsics for computing and acting on <a
361href="LangRef.html#int_overflow">overflow of integer operations</a>. This allows
362efficient code generation for languages that must trap or throw an exception on
363overflow. While these intrinsics work on all targets, they only generate
364efficient code on X86 so far.</li>
365
366<li>LLVM IR now supports a new <a href="LangRef.html#linkage">private
367linkage</a> type to produce labels that are stripped by the assembler before it
368produces a .o file (thus they are invisible to the linker).</li>
369
370<li>LLVM IR supports two new attributes for better alias analysis. The <a
371href="LangRef.html#paramattrs">noalias</a> attribute can now be used on the
372return value of a function to indicate that it returns new memory (e.g.
Duncan Sandsdd8d2672009-03-02 16:58:00 +0000[diff] [blame]373'malloc', 'calloc', etc).
374The new <a href="LangRef.html#paramattrs">nocapture</a> attribute can be used
375on pointer arguments to indicate that the function does not return the pointer,
376store it in an object that outlives the call, or let the value of the pointer
377escape from the function in any other way.
378Note that it is the pointer itself that must not escape, not the value it
379points to: loading a value out of the pointer is perfectly fine.
380Many standard library functions (e.g. 'strlen', 'memcpy') have this property.
381<!-- The simplifylibcalls pass applies these attributes to standard libc functions. -->
382</li>
Chris Lattner47202c32009-03-02 02:37:32 +0000[diff] [blame]383
384<li>The parser for ".ll" files in lib/AsmParser is now completely rewritten as a
385recursive descent parser. This parser produces better error messages (including
John Criswellaea8c752009-03-02 15:28:15 +0000[diff] [blame]386caret diagnostics), is less fragile (less likely to crash on strange things),
387does not leak memory, is more efficient, and eliminates LLVM's last use of the
Chris Lattner47202c32009-03-02 02:37:32 +0000[diff] [blame]388'bison' tool.</li>
389
390<li>Debug information representation and manipulation internals have been
391 consolidated to use a new set of classes in
John Criswellaea8c752009-03-02 15:28:15 +0000[diff] [blame]392 <tt>llvm/Analysis/DebugInfo.h</tt>. These routines are more
Chris Lattner47202c32009-03-02 02:37:32 +0000[diff] [blame]393 efficient, robust, and extensible and replace the older mechanisms.
394 llvm-gcc, clang, and the code generator now use them to create and process
395 debug information.</li>
Mikhail Glushenkovf61066a2009-03-01 18:09:47 +0000[diff] [blame]396
Chris Lattnerc3649452008-02-10 08:17:19 +0000[diff] [blame]397</ul>
Mikhail Glushenkov808cd3e2008-10-13 02:08:34 +0000[diff] [blame]398
Chris Lattnerc3649452008-02-10 08:17:19 +0000[diff] [blame]399</div>
400
401<!--=========================================================================-->
402<div class="doc_subsection">
Chris Lattner8fda4952008-02-10 07:46:44 +0000[diff] [blame]403<a name="optimizer">Optimizer Improvements</a>
404</div>
405
406<div class="doc_text">
407
John Criswellaea8c752009-03-02 15:28:15 +0000[diff] [blame]408<p>In addition to a large array of bug fixes and minor performance tweaks, this
Chris Lattnerea9c2242008-10-13 21:50:36 +0000[diff] [blame]409release includes a few major enhancements and additions to the optimizers:</p>
Chris Lattner8fda4952008-02-10 07:46:44 +0000[diff] [blame]410
411<ul>
412
Chris Lattner47202c32009-03-02 02:37:32 +0000[diff] [blame]413<li>The loop optimizer now improves floating point induction variables in
414several ways, including adding shadow induction variables to avoid
415"integer &lt;-&gt; floating point" conversions in loops when safe.</li>
416
John Criswellaea8c752009-03-02 15:28:15 +0000[diff] [blame]417<li>The "-mem2reg" pass is now much faster on code with large basic blocks.</li>
Chris Lattner47202c32009-03-02 02:37:32 +0000[diff] [blame]418
419<li>The "-jump-threading" pass is more powerful: it is iterative
420 and handles threading based on values with fully and partially redundant
421 loads.</li>
422
423<li>The "-memdep" memory dependence analysis pass (used by GVN and memcpyopt) is
424 both faster and more aggressive.</li>
425
426<li>The "-scalarrepl" scalar replacement of aggregates pass is more aggressive
427 about promoting unions to registers.</li>
Chris Lattnerea9c2242008-10-13 21:50:36 +0000[diff] [blame]428
Chris Lattner0eef6e42008-06-08 02:45:07 +0000[diff] [blame]429</ul>
430
431</div>
432
433<!--=========================================================================-->
434<div class="doc_subsection">
Chris Lattner5f393542009-03-02 03:24:11 +0000[diff] [blame]435<a name="codegen">Target Independent Code Generator Improvements</a>
Chris Lattner0eef6e42008-06-08 02:45:07 +0000[diff] [blame]436</div>
437
438<div class="doc_text">
439
Mikhail Glushenkovf61066a2009-03-01 18:09:47 +0000[diff] [blame]440<p>We have put a significant amount of work into the code generator
441infrastructure, which allows us to implement more aggressive algorithms and make
442it run faster:</p>
Chris Lattner0eef6e42008-06-08 02:45:07 +0000[diff] [blame]443
444<ul>
Chris Lattner5f393542009-03-02 03:24:11 +0000[diff] [blame]445<li>The <a href="WritingAnLLVMBackend.html">Writing an LLVM Compiler
446Backend</a> document has been greatly expanded and is substantially more
447complete.</li>
448
449<li>The SelectionDAG type legalization logic has been completely rewritten, is
450now more powerful (it supports arbitrary precision integer types for example),
John Criswellaea8c752009-03-02 15:28:15 +0000[diff] [blame]451and is more correct in several corner cases. The type legalizer converts
Chris Lattner5f393542009-03-02 03:24:11 +0000[diff] [blame]452operations on types that are not natively supported by the target machine into
453equivalent code sequences that only use natively supported types. The old type
454legalizer is still available (for now) and will be used if
455<tt>-disable-legalize-types</tt> is passed to the code generator.
Duncan Sands335d5e42009-02-25 11:50:08 +0000[diff] [blame]456</li>
Chris Lattner0eef6e42008-06-08 02:45:07 +0000[diff] [blame]457
Chris Lattner5f393542009-03-02 03:24:11 +0000[diff] [blame]458<li>The code generator now supports widening illegal vectors to larger legal
459ones (for example, converting operations on &lt;3 x float&gt; to work on
460&lt;4 x float&gt;) which is very important for common graphics
461applications.</li>
Chris Lattner47202c32009-03-02 02:37:32 +0000[diff] [blame]462
Chris Lattner5f393542009-03-02 03:24:11 +0000[diff] [blame]463<li>The assembly printers for each target are now split out into their own
464libraries that are separate from the main code generation logic. This reduces
John Criswellaea8c752009-03-02 15:28:15 +0000[diff] [blame]465the code size of JIT compilers by not requiring them to be linked in.</li>
Chris Lattner47202c32009-03-02 02:37:32 +0000[diff] [blame]466
Chris Lattner5f393542009-03-02 03:24:11 +0000[diff] [blame]467<li>The 'fast' instruction selection path (used at -O0 and for fast JIT
468 compilers) now supports accelerating codegen for code that uses exception
469 handling constructs.</li>
470
471<li>The optional PBQP register allocator now supports register coalescing.</li>
Chris Lattnerb0d6c502007-09-21 03:54:09 +0000[diff] [blame]472</ul>
Chris Lattnerb0d6c502007-09-21 03:54:09 +0000[diff] [blame]473</div>
474
Chris Lattner47202c32009-03-02 02:37:32 +0000[diff] [blame]475<!--=========================================================================-->
476<div class="doc_subsection">
477<a name="x86">X86-32 and X86-64 Target Improvements</a>
478</div>
479
480<div class="doc_text">
481<p>New features of the X86 target include:
482</p>
483
484<ul>
Bill Wendling9c7c4382009-03-02 07:54:14 +0000[diff] [blame]485<li>The <tt><a href="LangRef.html#int_returnaddress">llvm.returnaddress</a></tt>
486intrinsic (which is used to implement <tt>__builtin_return_address</tt>) now
487supports non-zero stack depths on X86.</li>
Chris Lattner5f393542009-03-02 03:24:11 +0000[diff] [blame]488
489<li>The X86 backend now supports code generation of vector shift operations
490using SSE instructions.</li>
491
Bill Wendling9c7c4382009-03-02 07:54:14 +0000[diff] [blame]492<li>X86-64 code generation now takes advantage of red zone, unless the
493<tt>-mno-red-zone</tt> option is specified.</li>
Chris Lattner5f393542009-03-02 03:24:11 +0000[diff] [blame]494
495<li>The X86 backend now supports using address space #256 in LLVM IR as a way of
496performing memory references off the GS segment register. This allows a
497front-end to take advantage of very low-level programming techniques when
Gabor Greifed585c62009-03-02 11:34:51 +0000[diff] [blame]498targeting X86 CPUs. See <tt>test/CodeGen/X86/movgs.ll</tt> for a simple
Bill Wendling9c7c4382009-03-02 07:54:14 +0000[diff] [blame]499example.</li>
Chris Lattner5f393542009-03-02 03:24:11 +0000[diff] [blame]500
501<li>The X86 backend now supports a <tt>-disable-mmx</tt> command line option to
502 prevent use of MMX even on chips that support it. This is important for cases
Bill Wendling9c7c4382009-03-02 07:54:14 +0000[diff] [blame]503 where code does not contain the proper <tt>llvm.x86.mmx.emms</tt>
504 intrinsics.</li>
Chris Lattner5f393542009-03-02 03:24:11 +0000[diff] [blame]505
Bill Wendling9c7c4382009-03-02 07:54:14 +0000[diff] [blame]506<li>The X86 JIT now detects the new Intel <a
507 href="http://en.wikipedia.org/wiki/Intel_Core_i7">Core i7</a> and <a
John Criswellaea8c752009-03-02 15:28:15 +0000[diff] [blame]508 href="http://en.wikipedia.org/wiki/Intel_Atom">Atom</a> chips and
509 auto-configures itself appropriately for the features of these chips.</li>
Chris Lattner5f393542009-03-02 03:24:11 +0000[diff] [blame]510
511<li>The JIT now supports exception handling constructs on Linux/X86-64 and
512 Darwin/x86-64.</li>
Bill Wendling9c7c4382009-03-02 07:54:14 +0000[diff] [blame]513
Chris Lattner5f393542009-03-02 03:24:11 +0000[diff] [blame]514<li>The JIT supports Thread Local Storage (TLS) on Linux/X86-32 but not yet on
515 X86-64.</li>
Chris Lattner47202c32009-03-02 02:37:32 +0000[diff] [blame]516</ul>
517
Chris Lattner47202c32009-03-02 02:37:32 +0000[diff] [blame]518</div>
Chris Lattnerb0d6c502007-09-21 03:54:09 +0000[diff] [blame]519
520<!--=========================================================================-->
521<div class="doc_subsection">
Chris Lattnere1fbd752009-02-26 07:32:11 +0000[diff] [blame]522<a name="pic16">PIC16 Target Improvements</a>
523</div>
524
525<div class="doc_text">
526<p>New features of the PIC16 target include:
527</p>
528
529<ul>
530<li>Both direct and indirect load/stores work now.</li>
531<li>Logical, bitwise and conditional operations now work for integer data
532types.</li>
533<li>Function calls involving basic types work now.</li>
534<li>Support for integer arrays.</li>
John Criswellaea8c752009-03-02 15:28:15 +0000[diff] [blame]535<li>The compiler can now emit libcalls for operations not supported by m/c
Bill Wendling9c7c4382009-03-02 07:54:14 +0000[diff] [blame]536instructions.</li>
537<li>Support for both data and ROM address spaces.</li>
Chris Lattnere1fbd752009-02-26 07:32:11 +0000[diff] [blame]538</ul>
539
540<p>Things not yet supported:</p>
541
542<ul>
543<li>Floating point.</li>
Bill Wendling9c7c4382009-03-02 07:54:14 +0000[diff] [blame]544<li>Passing/returning aggregate types to and from functions.</li>
Chris Lattnere1fbd752009-02-26 07:32:11 +0000[diff] [blame]545<li>Variable arguments.</li>
546<li>Indirect function calls.</li>
Chris Lattner5f393542009-03-02 03:24:11 +0000[diff] [blame]547<li>Interrupts/programs.</li>
Chris Lattnere1fbd752009-02-26 07:32:11 +0000[diff] [blame]548<li>Debug info.</li>
Chris Lattnere1fbd752009-02-26 07:32:11 +0000[diff] [blame]549</ul>
550
551</div>
552
553
554<!--=========================================================================-->
555<div class="doc_subsection">
Chris Lattner7ccc40d2009-03-01 02:30:21 +0000[diff] [blame]556<a name="llvmc">Improvements in LLVMC</a>
557</div>
558
559<div class="doc_text">
560<p>New features include:</p>
561
562<ul>
Mikhail Glushenkovf61066a2009-03-01 18:09:47 +0000[diff] [blame]563<li>Beginning with LLVM 2.5, <tt>llvmc2</tt> is known as
564 just <tt>llvmc</tt>. The old <tt>llvmc</tt> driver was removed.</li>
Chris Lattner7ccc40d2009-03-01 02:30:21 +0000[diff] [blame]565
566<li>The Clang plugin was substantially improved and is now enabled
Mikhail Glushenkovf61066a2009-03-01 18:09:47 +0000[diff] [blame]567 by default. The command <tt>llvmc --clang</tt> can be now used as a
568 synonym to <tt>ccc</tt>.</li>
Chris Lattner7ccc40d2009-03-01 02:30:21 +0000[diff] [blame]569
Bill Wendling9c7c4382009-03-02 07:54:14 +0000[diff] [blame]570<li>There is now a <tt>--check-graph</tt> option, which is supposed to catch
Mikhail Glushenkovf61066a2009-03-01 18:09:47 +0000[diff] [blame]571 common errors like multiple default edges, mismatched output/input language
572 names and cycles. In general, these checks can't be done at compile-time
573 because of the need to support plugins.</li>
Chris Lattner7ccc40d2009-03-01 02:30:21 +0000[diff] [blame]574
Mikhail Glushenkovf61066a2009-03-01 18:09:47 +0000[diff] [blame]575<li>Plugins are now more flexible and can refer to compilation graph nodes and
576 options defined in other plugins. To manage dependencies, a priority-sorting
Bill Wendling9c7c4382009-03-02 07:54:14 +0000[diff] [blame]577 mechanism was introduced. This change affects the TableGen file syntax. See the
Mikhail Glushenkovf61066a2009-03-01 18:09:47 +0000[diff] [blame]578 documentation for details.</li>
Chris Lattner7ccc40d2009-03-01 02:30:21 +0000[diff] [blame]579
Mikhail Glushenkovf61066a2009-03-01 18:09:47 +0000[diff] [blame]580<li>Hooks can now be provided with arguments. The syntax is "<tt>$CALL(MyHook,
Bill Wendling9c7c4382009-03-02 07:54:14 +0000[diff] [blame]581 'Arg1', 'Arg2', 'Arg3')</tt>".</li>
Chris Lattner7ccc40d2009-03-01 02:30:21 +0000[diff] [blame]582
Mikhail Glushenkovf61066a2009-03-01 18:09:47 +0000[diff] [blame]583<li>A new option type: multi-valued option, for options that take more than one
584 argument (for example, "<tt>-foo a b c</tt>").</li>
Chris Lattner7ccc40d2009-03-01 02:30:21 +0000[diff] [blame]585
Mikhail Glushenkovf61066a2009-03-01 18:09:47 +0000[diff] [blame]586<li>New option properties: '<tt>one_or_more</tt>', '<tt>zero_or_more</tt>',
587'<tt>hidden</tt>' and '<tt>really_hidden</tt>'.</li>
Chris Lattner7ccc40d2009-03-01 02:30:21 +0000[diff] [blame]588
Mikhail Glushenkovf61066a2009-03-01 18:09:47 +0000[diff] [blame]589<li>The '<tt>case</tt>' expression gained an '<tt>error</tt>' action and
590 an '<tt>empty</tt>' test (equivalent to "<tt>(not (not_empty ...))</tt>").</li>
Chris Lattner7ccc40d2009-03-01 02:30:21 +0000[diff] [blame]591
592<li>Documentation now looks more consistent to the rest of the LLVM
593 docs. There is also a man page now.</li>
Mikhail Glushenkovf61066a2009-03-01 18:09:47 +0000[diff] [blame]594
Chris Lattner7ccc40d2009-03-01 02:30:21 +0000[diff] [blame]595</ul>
596
597</div>
598
Chris Lattnere45f18b2008-06-05 08:02:49 +0000[diff] [blame]599
600<!--=========================================================================-->
601<div class="doc_subsection">
Chris Lattner92eacf82008-10-13 17:57:36 +0000[diff] [blame]602<a name="changes">Major Changes and Removed Features</a>
603</div>
604
605<div class="doc_text">
606
Chris Lattnercc64a322008-10-13 22:06:31 +0000[diff] [blame]607<p>If you're already an LLVM user or developer with out-of-tree changes based
Chris Lattner95d25072009-02-25 04:41:31 +0000[diff] [blame]608on LLVM 2.4, this section lists some "gotchas" that you may run into upgrading
Chris Lattnercc64a322008-10-13 22:06:31 +0000[diff] [blame]609from the previous release.</p>
Chris Lattner92eacf82008-10-13 17:57:36 +0000[diff] [blame]610
611<ul>
612
Bill Wendling9c7c4382009-03-02 07:54:14 +0000[diff] [blame]613<li>llvm-gcc defaults to <tt>-fno-math-errno</tt> on all X86 targets.</li>
614
Chris Lattnercc64a322008-10-13 22:06:31 +0000[diff] [blame]615</ul>
616
617
618<p>In addition, many APIs have changed in this release. Some of the major LLVM
619API changes are:</p>
620
621<ul>
Gabor Greiff7ea27c2009-03-02 12:23:55 +0000[diff] [blame]622<li>Some deprecated interfaces to create <tt>Instruction</tt> subclasses, that
John Criswellaea8c752009-03-02 15:28:15 +0000[diff] [blame]623 were spelled with lower case "create," have been removed.</li>
Devang Patel0e3da1a2008-10-14 20:03:43 +0000[diff] [blame]624</ul>
Chris Lattner92eacf82008-10-13 17:57:36 +0000[diff] [blame]625
Chris Lattner92eacf82008-10-13 17:57:36 +0000[diff] [blame]626</div>
627
628
629
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]630<!-- *********************************************************************** -->
631<div class="doc_section">
632 <a name="portability">Portability and Supported Platforms</a>
633</div>
634<!-- *********************************************************************** -->
635
636<div class="doc_text">
637
638<p>LLVM is known to work on the following platforms:</p>
639
640<ul>
Mikhail Glushenkovf61066a2009-03-01 18:09:47 +0000[diff] [blame]641<li>Intel and AMD machines (IA32, X86-64, AMD64, EMT-64) running Red Hat
Chris Lattnerb1ad47c2008-11-10 05:40:34 +0000[diff] [blame]642Linux, Fedora Core and FreeBSD (and probably other unix-like systems).</li>
Mikhail Glushenkovf61066a2009-03-01 18:09:47 +0000[diff] [blame]643<li>PowerPC and X86-based Mac OS X systems, running 10.3 and above in 32-bit
Chris Lattnerb1ad47c2008-11-10 05:40:34 +0000[diff] [blame]644and 64-bit modes.</li>
Chris Lattnere0798082008-02-06 18:00:06 +0000[diff] [blame]645<li>Intel and AMD machines running on Win32 using MinGW libraries (native).</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]646<li>Intel and AMD machines running on Win32 with the Cygwin libraries (limited
647 support is available for native builds with Visual C++).</li>
Gabor Greif5e75e652008-06-05 18:39:01 +0000[diff] [blame]648<li>Sun UltraSPARC workstations running Solaris 10.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]649<li>Alpha-based machines running Debian GNU/Linux.</li>
Gabor Greif5e75e652008-06-05 18:39:01 +0000[diff] [blame]650<li>Itanium-based (IA64) machines running Linux and HP-UX.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]651</ul>
652
Chris Lattnerba948522008-06-05 06:57:39 +0000[diff] [blame]653<p>The core LLVM infrastructure uses GNU autoconf to adapt itself
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]654to the machine and operating system on which it is built. However, minor
655porting may be required to get LLVM to work on new platforms. We welcome your
656portability patches and reports of successful builds or error messages.</p>
657
658</div>
659
660<!-- *********************************************************************** -->
661<div class="doc_section">
662 <a name="knownproblems">Known Problems</a>
663</div>
664<!-- *********************************************************************** -->
665
666<div class="doc_text">
667
Mikhail Glushenkovf61066a2009-03-01 18:09:47 +0000[diff] [blame]668<p>This section contains significant known problems with the LLVM system,
Chris Lattnerb1ad47c2008-11-10 05:40:34 +0000[diff] [blame]669listed by component. If you run into a problem, please check the <a
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]670href="http://llvm.org/bugs/">LLVM bug database</a> and submit a bug if
671there isn't already one.</p>
672
673</div>
674
675<!-- ======================================================================= -->
676<div class="doc_subsection">
677 <a name="experimental">Experimental features included with this release</a>
678</div>
679
680<div class="doc_text">
681
682<p>The following components of this LLVM release are either untested, known to
683be broken or unreliable, or are in early development. These components should
684not be relied on, and bugs should not be filed against them, but they may be
685useful to some people. In particular, if you would like to work on one of these
Chris Lattnerdaeb3232008-02-12 06:29:45 +0000[diff] [blame]686components, please contact us on the <a
687href="http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev">LLVMdev list</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]688
689<ul>
Gabor Greifaa53e562008-10-14 11:31:14 +0000[diff] [blame]690<li>The MSIL, IA64, Alpha, SPU, MIPS, and PIC16 backends are experimental.</li>
Bill Wendling9c7c4382009-03-02 07:54:14 +0000[diff] [blame]691<li>The <tt>llc</tt> "<tt>-filetype=asm</tt>" (the default) is the only
692 supported value for this option.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]693</ul>
694
695</div>
696
697<!-- ======================================================================= -->
698<div class="doc_subsection">
699 <a name="x86-be">Known problems with the X86 back-end</a>
700</div>
701
702<div class="doc_text">
703
704<ul>
Anton Korobeynikov77d15e22008-06-08 10:24:13 +0000[diff] [blame]705 <li>The X86 backend does not yet support
706 all <a href="http://llvm.org/PR879">inline assembly that uses the X86
707 floating point stack</a>. It supports the 'f' and 't' constraints, but not
708 'u'.</li>
Chris Lattnerd95b5962008-06-08 21:19:07 +0000[diff] [blame]709 <li>The X86 backend generates inefficient floating point code when configured
710 to generate code for systems that don't have SSE2.</li>
Duncan Sands8b4f3772008-06-08 19:38:43 +0000[diff] [blame]711 <li>Win64 code generation wasn't widely tested. Everything should work, but we
John Criswellaea8c752009-03-02 15:28:15 +0000[diff] [blame]712 expect small issues to happen. Also, llvm-gcc cannot build the mingw64
713 runtime currently due
Anton Korobeynikov77d15e22008-06-08 10:24:13 +0000[diff] [blame]714 to <a href="http://llvm.org/PR2255">several</a>
John Criswellaea8c752009-03-02 15:28:15 +0000[diff] [blame]715 <a href="http://llvm.org/PR2257">bugs</a> and due to lack of support for
716 the
717 'u' inline assembly constraint and for X87 floating point inline assembly.</li>
Dan Gohman3882ecb2008-06-08 23:05:11 +0000[diff] [blame]718 <li>The X86-64 backend does not yet support the LLVM IR instruction
Chris Lattner47202c32009-03-02 02:37:32 +0000[diff] [blame]719 <tt>va_arg</tt>. Currently, the llvm-gcc and front-ends support variadic
Dan Gohman3882ecb2008-06-08 23:05:11 +0000[diff] [blame]720 argument constructs on X86-64 by lowering them manually.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]721</ul>
722
723</div>
724
725<!-- ======================================================================= -->
726<div class="doc_subsection">
727 <a name="ppc-be">Known problems with the PowerPC back-end</a>
728</div>
729
730<div class="doc_text">
731
732<ul>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]733<li>The Linux PPC32/ABI support needs testing for the interpreter and static
734compilation, and lacks support for debug information.</li>
735</ul>
736
737</div>
738
739<!-- ======================================================================= -->
740<div class="doc_subsection">
741 <a name="arm-be">Known problems with the ARM back-end</a>
742</div>
743
744<div class="doc_text">
745
746<ul>
747<li>Thumb mode works only on ARMv6 or higher processors. On sub-ARMv6
Duncan Sands1827fa62007-09-26 15:59:54 +0000[diff] [blame]748processors, thumb programs can crash or produce wrong
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]749results (<a href="http://llvm.org/PR1388">PR1388</a>).</li>
John Criswellaea8c752009-03-02 15:28:15 +0000[diff] [blame]750<li>Compilation for ARM Linux OABI (old ABI) is supported but not fully tested.
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]751</li>
Chris Lattnerd95b5962008-06-08 21:19:07 +0000[diff] [blame]752<li>There is a bug in QEMU-ARM (&lt;= 0.9.0) which causes it to incorrectly
753 execute
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]754programs compiled with LLVM. Please use more recent versions of QEMU.</li>
755</ul>
756
757</div>
758
759<!-- ======================================================================= -->
760<div class="doc_subsection">
761 <a name="sparc-be">Known problems with the SPARC back-end</a>
762</div>
763
764<div class="doc_text">
765
766<ul>
John Criswellaea8c752009-03-02 15:28:15 +0000[diff] [blame]767<li>The SPARC backend only supports the 32-bit SPARC ABI (-m32); it does not
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]768 support the 64-bit SPARC ABI (-m64).</li>
769</ul>
770
771</div>
772
773<!-- ======================================================================= -->
774<div class="doc_subsection">
Bruno Cardoso Lopes62d45d72008-10-25 14:56:26 +0000[diff] [blame]775 <a name="mips-be">Known problems with the MIPS back-end</a>
776</div>
777
778<div class="doc_text">
779
780<ul>
781<li>The O32 ABI is not fully supported.</li>
782<li>64-bit MIPS targets are not supported yet.</li>
783</ul>
784
785</div>
786
787<!-- ======================================================================= -->
788<div class="doc_subsection">
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]789 <a name="alpha-be">Known problems with the Alpha back-end</a>
790</div>
791
792<div class="doc_text">
793
794<ul>
795
796<li>On 21164s, some rare FP arithmetic sequences which may trap do not have the
797appropriate nops inserted to ensure restartability.</li>
798
799</ul>
800</div>
801
802<!-- ======================================================================= -->
803<div class="doc_subsection">
804 <a name="ia64-be">Known problems with the IA64 back-end</a>
805</div>
806
807<div class="doc_text">
808
809<ul>
John Criswellaea8c752009-03-02 15:28:15 +0000[diff] [blame]810<li>The Itanium backend is highly experimental and has a number of known
Chris Lattnerf06141a2008-06-08 23:12:47 +0000[diff] [blame]811 issues. We are looking for a maintainer for the Itanium backend. If you
Gabor Greif45fc8b12008-10-15 10:47:24 +0000[diff] [blame]812 are interested, please contact the LLVMdev mailing list.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]813</ul>
814
815</div>
816
817<!-- ======================================================================= -->
818<div class="doc_subsection">
819 <a name="c-be">Known problems with the C back-end</a>
820</div>
821
822<div class="doc_text">
823
824<ul>
Chris Lattner499c07c2008-06-05 06:35:40 +0000[diff] [blame]825<li><a href="http://llvm.org/PR802">The C backend has only basic support for
826 inline assembly code</a>.</li>
Chris Lattnere0194462007-09-26 06:01:35 +0000[diff] [blame]827<li><a href="http://llvm.org/PR1658">The C backend violates the ABI of common
828 C++ programs</a>, preventing intermixing between C++ compiled by the CBE and
Gabor Greife33c5322009-03-02 12:02:51 +0000[diff] [blame]829 C++ code compiled with <tt>llc</tt> or native compilers.</li>
Duncan Sandsbd0ad6e2008-02-10 13:40:55 +0000[diff] [blame]830<li>The C backend does not support all exception handling constructs.</li>
Duncan Sandse6e264c2009-02-25 11:51:54 +0000[diff] [blame]831<li>The C backend does not support arbitrary precision integers.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]832</ul>
833
834</div>
835
836
837<!-- ======================================================================= -->
838<div class="doc_subsection">
Chris Lattner499c07c2008-06-05 06:35:40 +0000[diff] [blame]839 <a name="c-fe">Known problems with the llvm-gcc C front-end</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]840</div>
841
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]842<div class="doc_text">
843
Chris Lattnere0798082008-02-06 18:00:06 +0000[diff] [blame]844<p>llvm-gcc does not currently support <a href="http://llvm.org/PR869">Link-Time
845Optimization</a> on most platforms "out-of-the-box". Please inquire on the
Gabor Greif45fc8b12008-10-15 10:47:24 +0000[diff] [blame]846LLVMdev mailing list if you are interested.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]847
Chris Lattner499c07c2008-06-05 06:35:40 +0000[diff] [blame]848<p>The only major language feature of GCC not supported by llvm-gcc is
849 the <tt>__builtin_apply</tt> family of builtins. However, some extensions
850 are only supported on some targets. For example, trampolines are only
Duncan Sandsa73561c2008-06-08 20:18:35 +0000[diff] [blame]851 supported on some targets (these are used when you take the address of a
852 nested function).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]853
Chris Lattner499c07c2008-06-05 06:35:40 +0000[diff] [blame]854<p>If you run into GCC extensions which are not supported, please let us know.
855</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]856
857</div>
858
859<!-- ======================================================================= -->
860<div class="doc_subsection">
Chris Lattner499c07c2008-06-05 06:35:40 +0000[diff] [blame]861 <a name="c++-fe">Known problems with the llvm-gcc C++ front-end</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]862</div>
863
864<div class="doc_text">
865
866<p>The C++ front-end is considered to be fully
867tested and works for a number of non-trivial programs, including LLVM
868itself, Qt, Mozilla, etc.</p>
869
870<ul>
Anton Korobeynikovd9a17932008-10-11 18:27:16 +0000[diff] [blame]871<li>Exception handling works well on the X86 and PowerPC targets. Currently
Chris Lattnerb1ad47c2008-11-10 05:40:34 +0000[diff] [blame]872 only Linux and Darwin targets are supported (both 32 and 64 bit).</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]873</ul>
874
875</div>
876
Chris Lattner6c8a1b82008-10-30 03:58:13 +0000[diff] [blame]877<!-- ======================================================================= -->
878<div class="doc_subsection">
879 <a name="fortran-fe">Known problems with the llvm-gcc Fortran front-end</a>
880</div>
881
882<div class="doc_text">
Gabor Greif1fc8fbc2008-11-04 21:50:59 +0000[diff] [blame]883<ul>
Chris Lattner6c8a1b82008-10-30 03:58:13 +0000[diff] [blame]884<li>Fortran support generally works, but there are still several unresolved bugs
885 in Bugzilla. Please see the tools/gfortran component for details.</li>
Gabor Greif1fc8fbc2008-11-04 21:50:59 +0000[diff] [blame]886</ul>
Chris Lattner6c8a1b82008-10-30 03:58:13 +0000[diff] [blame]887</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]888
Chris Lattnerdaeb3232008-02-12 06:29:45 +0000[diff] [blame]889<!-- ======================================================================= -->
890<div class="doc_subsection">
Chris Lattner499c07c2008-06-05 06:35:40 +0000[diff] [blame]891 <a name="ada-fe">Known problems with the llvm-gcc Ada front-end</a>
Chris Lattnerdaeb3232008-02-12 06:29:45 +0000[diff] [blame]892</div>
893
894<div class="doc_text">
John Criswellaea8c752009-03-02 15:28:15 +0000[diff] [blame]895The llvm-gcc 4.2 Ada compiler works fairly well; however, this is not a mature
896technology, and problems should be expected.
Chris Lattnerdaeb3232008-02-12 06:29:45 +0000[diff] [blame]897<ul>
Duncan Sandsa73561c2008-06-08 20:18:35 +0000[diff] [blame]898<li>The Ada front-end currently only builds on X86-32. This is mainly due
John Criswellaea8c752009-03-02 15:28:15 +0000[diff] [blame]899to lack of trampoline support (pointers to nested functions) on other platforms.
900However, it <a href="http://llvm.org/PR2006">also fails to build on X86-64</a>
Chris Lattnerdaeb3232008-02-12 06:29:45 +0000[diff] [blame]901which does support trampolines.</li>
902<li>The Ada front-end <a href="http://llvm.org/PR2007">fails to bootstrap</a>.
Duncan Sandsf73473e2009-02-25 11:59:06 +0000[diff] [blame]903This is due to lack of LLVM support for <tt>setjmp</tt>/<tt>longjmp</tt> style
904exception handling, which is used internally by the compiler.
Chris Lattnerdaeb3232008-02-12 06:29:45 +0000[diff] [blame]905Workaround: configure with --disable-bootstrap.</li>
Duncan Sands2dbc2e82008-10-13 17:27:23 +0000[diff] [blame]906<li>The c380004, <a href="http://llvm.org/PR2010">c393010</a>
907and <a href="http://llvm.org/PR2421">cxg2021</a> ACATS tests fail
Duncan Sandsf73473e2009-02-25 11:59:06 +0000[diff] [blame]908(c380004 also fails with gcc-4.2 mainline).
909If the compiler is built with checks disabled then <a href="http://llvm.org/PR2010">c393010</a>
910causes the compiler to go into an infinite loop, using up all system memory.</li>
Duncan Sandsc2a29a02009-03-02 16:35:57 +0000[diff] [blame]911<li>Some GCC specific Ada tests continue to crash the compiler.</li>
Chris Lattnerdaeb3232008-02-12 06:29:45 +0000[diff] [blame]912<li>The -E binder option (exception backtraces)
913<a href="http://llvm.org/PR1982">does not work</a> and will result in programs
914crashing if an exception is raised. Workaround: do not use -E.</li>
915<li>Only discrete types <a href="http://llvm.org/PR1981">are allowed to start
916or finish at a non-byte offset</a> in a record. Workaround: do not pack records
917or use representation clauses that result in a field of a non-discrete type
918starting or finishing in the middle of a byte.</li>
Chris Lattnerd95b5962008-06-08 21:19:07 +0000[diff] [blame]919<li>The <tt>lli</tt> interpreter <a href="http://llvm.org/PR2009">considers
920'main' as generated by the Ada binder to be invalid</a>.
921Workaround: hand edit the file to use pointers for <tt>argv</tt> and
922<tt>envp</tt> rather than integers.</li>
923<li>The <tt>-fstack-check</tt> option <a href="http://llvm.org/PR2008">is
924ignored</a>.</li>
Chris Lattnerdaeb3232008-02-12 06:29:45 +0000[diff] [blame]925</ul>
926</div>
927
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]928<!-- *********************************************************************** -->
929<div class="doc_section">
930 <a name="additionalinfo">Additional Information</a>
931</div>
932<!-- *********************************************************************** -->
933
934<div class="doc_text">
935
936<p>A wide variety of additional information is available on the <a
937href="http://llvm.org">LLVM web page</a>, in particular in the <a
938href="http://llvm.org/docs/">documentation</a> section. The web page also
939contains versions of the API documentation which is up-to-date with the
940Subversion version of the source code.
941You can access versions of these documents specific to this release by going
942into the "<tt>llvm/doc/</tt>" directory in the LLVM tree.</p>
943
944<p>If you have any questions or comments about LLVM, please feel free to contact
945us via the <a href="http://llvm.org/docs/#maillist"> mailing
946lists</a>.</p>
947
948</div>
949
950<!-- *********************************************************************** -->
951
952<hr>
953<address>
954 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
Misha Brukman947321d2008-12-11 17:34:48 +0000[diff] [blame]955 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]956 <a href="http://validator.w3.org/check/referer"><img
Misha Brukman947321d2008-12-11 17:34:48 +0000[diff] [blame]957 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]958
959 <a href="http://llvm.org/">LLVM Compiler Infrastructure</a><br>
960 Last modified: $Date$
961</address>
962
963</body>
964</html>