Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / 0c5465e27e51577bdee6797f31a9bd5c7545d5c7 / . / docs / ReleaseNotes-2.6.html
blob: 8b8b7d69bb2e09ac9577b12f95c019a195db34ba [file] [log] [blame]
Duncan Sands54fbb412009-06-24 08:38:48 +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">
7 <title>LLVM 2.6 Release Notes</title>
8</head>
9<body>
10
11<div class="doc_title">LLVM 2.6 Release Notes</div>
12
13<ol>
14 <li><a href="#intro">Introduction</a></li>
15 <li><a href="#subproj">Sub-project Status Update</a></li>
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>
18 <li><a href="GettingStarted.html">Installation Instructions</a></li>
19 <li><a href="#portability">Portability and Supported Platforms</a></li>
20 <li><a href="#knownproblems">Known Problems</a></li>
21 <li><a href="#additionalinfo">Additional Information</a></li>
22</ol>
23
24<div class="doc_author">
25 <p>Written by the <a href="http://llvm.org">LLVM Team</a></p>
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
36<p>This document contains the release notes for the LLVM Compiler
37Infrastructure, release 2.6. Here we describe the status of LLVM, including
38major improvements from the previous release and significant known problems.
39All LLVM releases may be downloaded from the <a
40href="http://llvm.org/releases/">LLVM releases web site</a>.</p>
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
45href="http://mail.cs.uiuc.edu/mailman/listinfo/llvmdev">LLVM Developer's Mailing
46List</a> is a good place to send them.</p>
47
48<p>Note that if you are reading this file from a Subversion checkout or the
49main LLVM web page, this document applies to the <i>next</i> release, not the
50current one. To see the release notes for a specific release, please see the
51<a href="http://llvm.org/releases/">releases page</a>.</p>
52
53</div>
Chris Lattner169c7012009-10-08 07:01:46 +0000[diff] [blame]54
Chris Lattnerb2d43472009-10-05 02:12:39 +0000[diff] [blame]55
Chris Lattner6aaf6902009-10-08 06:27:53 +0000[diff] [blame]56<!--
57Almost dead code.
Chris Lattner169c7012009-10-08 07:01:46 +0000[diff] [blame]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 Lattner6aaf6902009-10-08 06:27:53 +0000[diff] [blame]61-->
Chris Lattnerb2d43472009-10-05 02:12:39 +0000[diff] [blame]62
Chris Lattnerb2d43472009-10-05 02:12:39 +0000[diff] [blame]63
64<!-- Unfinished features in 2.6:
Chris Lattner169c7012009-10-08 07:01:46 +0000[diff] [blame]65 gcc plugin.
Chris Lattnerb2d43472009-10-05 02:12:39 +0000[diff] [blame]66 strong phi elim
67 variable debug info for optimized code
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]68 postalloc scheduler: anti dependence breaking, hazard recognizer?
Chris Lattnerb2d43472009-10-05 02:12:39 +0000[diff] [blame]69 metadata
70 loop dependence analysis
Chris Lattner169c7012009-10-08 07:01:46 +0000[diff] [blame]71 ELF Writer? How stable?
72 <li>PostRA scheduler improvements, ARM adoption (David Goodwin).</li>
Chris Lattner0d6011062009-10-09 06:24:25 +0000[diff] [blame]73 2.7 supports the GDB 7.0 jit interfaces for debug info.
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]74 -->
75
76 <!-- for announcement email:
Chris Lattnerb2d43472009-10-05 02:12:39 +0000[diff] [blame]77 Logo web page.
78 llvm devmtg
79 compiler_rt
Chris Lattner51ead6d2009-10-10 19:00:55 +0000[diff] [blame]80 KLEE web page at klee.llvm.org
Chris Lattner6aaf6902009-10-08 06:27:53 +0000[diff] [blame]81 Many new papers added to /pubs/
82 Mention gcc plugin.
83
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]84 -->
85
86<!-- *********************************************************************** -->
87<div class="doc_section">
88 <a name="subproj">Sub-project Status Update</a>
89</div>
90<!-- *********************************************************************** -->
91
92<div class="doc_text">
93<p>
94The LLVM 2.6 distribution currently consists of code from the core LLVM
Chris Lattner4d91d4d2009-10-10 18:33:13 +0000[diff] [blame]95repository (which roughly includes the LLVM optimizers, code generators
96and supporting tools), the Clang repository and the llvm-gcc repository. In
97addition to this code, the LLVM Project includes other sub-projects that are in
98development. Here we include updates on these subprojects.
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]99</p>
100
101</div>
102
103
104<!--=========================================================================-->
105<div class="doc_subsection">
106<a name="clang">Clang: C/C++/Objective-C Frontend Toolkit</a>
107</div>
108
109<div class="doc_text">
110
Chris Lattner401ec6d2009-10-09 05:01:15 +0000[diff] [blame]111<p>The <a href="http://clang.llvm.org/">Clang project</a> is an effort to build
112a set of new 'LLVM native' front-end technologies for the C family of languages.
113LLVM 2.6 is the first release to officially include Clang, and it provides a
Chris Lattner4d91d4d2009-10-10 18:33:13 +0000[diff] [blame]114production quality C and Objective-C compiler. If you are interested in <a
115href="http://clang.llvm.org/performance.html">fast compiles</a> and
116<a href="http://clang.llvm.org/diagnostics.html">good diagnostics</a>, we
117encourage you to try it out. Clang currently compiles typical Objective-C code
1183x faster than GCC and compiles C code about 30% faster than GCC at -O0 -g
119(which is when the most pressure is on the frontend).</p>
Chris Lattner401ec6d2009-10-09 05:01:15 +0000[diff] [blame]120
121<p>In addition to supporting these languages, C++ support is also <a
122href="http://clang.llvm.org/cxx_status.html">well under way</a>, and mainline
123Clang is able to parse the libstdc++ 4.2 headers and even codegen simple apps.
124If you are interested in Clang C++ support or any other Clang feature, we
125strongly encourage you to get involved on the <a
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]126href="http://lists.cs.uiuc.edu/mailman/listinfo/cfe-dev">Clang front-end mailing
127list</a>.</p>
128
129<p>In the LLVM 2.6 time-frame, the Clang team has made many improvements:</p>
130
131<ul>
Chris Lattner401ec6d2009-10-09 05:01:15 +0000[diff] [blame]132<li>C and Objective-C support are now considered production quality.</li>
Chris Lattner4d91d4d2009-10-10 18:33:13 +0000[diff] [blame]133<li>AuroraUX, FreeBSD, and OpenBSD are now supported.</li>
Chris Lattner401ec6d2009-10-09 05:01:15 +0000[diff] [blame]134<li>Most of Objective-C 2.0 is now supported with the GNU runtime.</li>
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]135<li>Many many bugs are fixed and many features have been added.</li>
136</ul>
137</div>
138
139<!--=========================================================================-->
140<div class="doc_subsection">
141<a name="clangsa">Clang Static Analyzer</a>
142</div>
143
144<div class="doc_text">
145
Ted Kremenek0c5465e2009-10-11 03:10:25 +0000[diff] [blame^]146<p>Previously announced in the 2.4 and 2.5 LLVM releases, the Clang project also
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]147includes an early stage static source code analysis tool for <a
148href="http://clang.llvm.org/StaticAnalysis.html">automatically finding bugs</a>
Duncan Sandsccedda42009-10-10 19:16:25 +0000[diff] [blame]149in C and Objective-C programs. The tool performs checks to find
Ted Kremenek0c5465e2009-10-11 03:10:25 +0000[diff] [blame^]150bugs that occur on a specific path within a program.</p>
151
152<p>In the LLVM 2.6 time-frame, the analyzer core has undergone several important
153improvements and cleanups and now includes a new <em>Checker</em> interface that
154is intended to eventually serve as a basis for domain-specific checks. Further,
155in addition to generating HTML files for reporting analysis results, the
156analyzer can now also emit bug reports in a structured XML format that is
157intended to be easily readable by other programs.</p>
158
159<p>The set of checks performed by the static analyzer continues to expand, and
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]160future plans for the tool include full source-level inter-procedural analysis
161and deeper checks such as buffer overrun detection. There are many opportunities
162to extend and enhance the static analyzer, and anyone interested in working on
163this project is encouraged to get involved!</p>
164
165</div>
166
167<!--=========================================================================-->
168<div class="doc_subsection">
169<a name="vmkit">VMKit: JVM/CLI Virtual Machine Implementation</a>
170</div>
171
172<div class="doc_text">
173<p>
174The <a href="http://vmkit.llvm.org/">VMKit project</a> is an implementation of
Nicolas Geoffray12c0e562009-10-09 10:13:08 +0000[diff] [blame]175a JVM and a CLI Virtual Machine (Microsoft .NET is an
Nicolas Geoffray0dcee362009-10-09 10:17:14 +0000[diff] [blame]176implementation of the CLI) using LLVM for static and just-in-time
177compilation.</p>
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]178
Nicolas Geoffray0dcee362009-10-09 10:17:14 +0000[diff] [blame]179<p>
180VMKit version 0.26 builds with LLVM 2.6 and you can find it on its
Duncan Sandsdc5bd2b2009-10-10 19:30:16 +0000[diff] [blame]181<a href="http://vmkit.llvm.org/releases/">web page</a>. The release includes
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]182bug fixes, cleanup and new features. The major changes are:</p>
183
184<ul>
185
Nicolas Geoffray0dcee362009-10-09 10:17:14 +0000[diff] [blame]186<li>A new llcj tool to generate shared libraries or executables of Java
187 files.</li>
Nicolas Geoffray13eff6a2009-10-09 13:17:57 +0000[diff] [blame]188<li>Cooperative garbage collection. </li>
Nicolas Geoffray12c0e562009-10-09 10:13:08 +0000[diff] [blame]189<li>Fast subtype checking (paper from Click et al [JGI'02]). </li>
Duncan Sandsdc5bd2b2009-10-10 19:30:16 +0000[diff] [blame]190<li>Implementation of a two-word header for Java objects instead of the original
Nicolas Geoffray0dcee362009-10-09 10:17:14 +0000[diff] [blame]191 three-word header. </li>
192<li>Better Java specification-compliance: division by zero checks, stack
193 overflow checks, finalization and references support. </li>
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]194
195</ul>
196</div>
197
Chris Lattner6e6d33c2009-10-09 05:55:04 +0000[diff] [blame]198
199<!--=========================================================================-->
200<div class="doc_subsection">
201<a name="compiler-rt">compiler-rt: Compiler Runtime Library</a>
202</div>
203
204<div class="doc_text">
205<p>
206The new LLVM <a href="http://compiler-rt.llvm.org/">compiler-rt project</a>
207is a simple library that provides an implementation of the low-level
208target-specific hooks required by code generation and other runtime components.
209For example, when compiling for a 32-bit target, converting a double to a 64-bit
Duncan Sandsd4f53352009-10-10 20:06:04 +0000[diff] [blame]210unsigned integer is compiled into a runtime call to the "__fixunsdfdi"
Chris Lattner4d91d4d2009-10-10 18:33:13 +0000[diff] [blame]211function. The compiler-rt library provides highly optimized implementations of
212this and other low-level routines (some are 3x faster than the equivalent
213libgcc routines).</p>
Chris Lattner6e6d33c2009-10-09 05:55:04 +0000[diff] [blame]214
215<p>
216All of the code in the compiler-rt project is available under the standard LLVM
217License, a "BSD-style" license.</p>
218
219</div>
220
221<!--=========================================================================-->
222<div class="doc_subsection">
Chris Lattner51ead6d2009-10-10 19:00:55 +0000[diff] [blame]223<a name="klee">KLEE: Symbolic Execution and Automatic Test Case Generator</a>
Chris Lattner6e6d33c2009-10-09 05:55:04 +0000[diff] [blame]224</div>
225
226<div class="doc_text">
227<p>
Chris Lattner51ead6d2009-10-10 19:00:55 +0000[diff] [blame]228The new LLVM <a href="http://klee.llvm.org/">KLEE project</a> is a symbolic
229execution framework for programs in LLVM bitcode form. KLEE tries to
Chris Lattner6e6d33c2009-10-09 05:55:04 +0000[diff] [blame]230symbolically evaluate "all" paths through the application and records state
231transitions that lead to fault states. This allows it to construct testcases
232that lead to faults and can even be used to verify algorithms. For more
233details, please see the <a
234href="http://llvm.org/pubs/2008-12-OSDI-KLEE.html">OSDI 2008 paper</a> about
Chris Lattner51ead6d2009-10-10 19:00:55 +0000[diff] [blame]235KLEE.</p>
Chris Lattner6e6d33c2009-10-09 05:55:04 +0000[diff] [blame]236
237</div>
238
239<!--=========================================================================-->
240<div class="doc_subsection">
241<a name="dragonegg">Dragon Egg: An LLVM backend plugin for GCC</a>
242</div>
243
244<div class="doc_text">
245<p>
246<b>Duncan needs to write me</b>.
247</p>
248
249</div>
250
251
Chris Lattner169c7012009-10-08 07:01:46 +0000[diff] [blame]252<!--=========================================================================-->
253<div class="doc_subsection">
254<a name="mc">llvm-mc: Machine Code Toolkit</a>
255</div>
256
257<div class="doc_text">
258<p>
Chris Lattner0d6011062009-10-09 06:24:25 +0000[diff] [blame]259The LLVM Machine Code (MC) Toolkit project is a (very early) effort to build
260better tools for dealing with machine code, object file formats, etc. The idea
261is to be able to generate most of the target specific details of assemblers and
262disassemblers from existing LLVM target .td files (with suitable enhancements),
263and to build infrastructure for reading and writing common object file formats.
264One of the first deliverables is to build a full assembler and integrate it into
265the compiler, which is predicted to substantially reduce compile time in some
266scenarios.
267</p>
Chris Lattner169c7012009-10-08 07:01:46 +0000[diff] [blame]268
Chris Lattner0d6011062009-10-09 06:24:25 +0000[diff] [blame]269<p>In the LLVM 2.6 timeframe, the MC framework has grown to the point where it
270can reliably parse and pretty print (with some encoding information) a
271darwin/x86 .s file successfully, and has the very early phases of a Mach-O
272assembler in progress. Beyond the MC framework itself, major refactoring of the
273LLVM code generator has started. The idea is to make the code generator reason
274about the code it is producing in a much more semantic way, rather than a
275textual way. For example, the code generator now uses MCSection objects to
276represent section assignments, instead of text strings that print to .section
277directives.</p>
278
279<p>MC is an early and ongoing project that will hopefully continue to lead to
280many improvements in the code generator and build infrastructure useful for many
281other situations.
Chris Lattner169c7012009-10-08 07:01:46 +0000[diff] [blame]282</p>
283
284</div>
285
286
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]287<!-- *********************************************************************** -->
288<div class="doc_section">
289 <a name="externalproj">External Projects Using LLVM 2.6</a>
290</div>
291<!-- *********************************************************************** -->
292
Chris Lattner4d91d4d2009-10-10 18:33:13 +0000[diff] [blame]293<div class="doc_text">
294
295<p>An exciting aspect of LLVM is that it is used as an enabling technology for
296 a lot of other language and tools projects. This section lists some of the
297 projects that have already been updated to work with LLVM 2.6.</p>
298</div>
299
300
Chris Lattner169c7012009-10-08 07:01:46 +0000[diff] [blame]301<!--=========================================================================-->
302<div class="doc_subsection">
303<a name="Rubinius">Rubinius</a>
304</div>
305
306<div class="doc_text">
307<p><a href="http://github.com/evanphx/rubinius">Rubinius</a> is an environment
308for running Ruby code which strives to write as much of the core class
309implementation in Ruby as possible. Combined with a bytecode interpreting VM, it
310uses LLVM to optimize and compile ruby code down to machine code. Techniques
311such as type feedback, method inlining, and uncommon traps are all used to
312remove dynamism from ruby execution and increase performance.</p>
Chris Lattner59a30272009-10-08 16:01:33 +0000[diff] [blame]313
314<p>Since LLVM 2.5, Rubinius has made several major leaps forward, implementing
315a counter based JIT, type feedback, and speculative method inlining.
316</p>
317
Chris Lattner169c7012009-10-08 07:01:46 +0000[diff] [blame]318</div>
Chris Lattner2375deb2009-09-30 06:27:22 +0000[diff] [blame]319
320<!--=========================================================================-->
321<div class="doc_subsection">
322<a name="macruby">MacRuby</a>
323</div>
324
325<div class="doc_text">
326
327<p>
328<a href="http://macruby.org">MacRuby</a> is an implementation of Ruby on top of
329core Mac OS X technologies, such as the Objective-C common runtime and garbage
330collector, and the CoreFoundation framework. It is principally developed by
331Apple and aims at enabling the creation of full-fledged Mac OS X applications.
332</p>
333
334<p>
335MacRuby uses LLVM for optimization passes, JIT and AOT compilation of Ruby
336expressions. It also uses zero-cost DWARF exceptions to implement Ruby exception
337handling.</p>
338
339</div>
340
341
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]342<!--=========================================================================-->
343<div class="doc_subsection">
344<a name="pure">Pure</a>
345</div>
346
347<div class="doc_text">
348<p>
349<a href="http://pure-lang.googlecode.com/">Pure</a>
350is an algebraic/functional programming language based on term rewriting.
351Programs are collections of equations which are used to evaluate expressions in
352a symbolic fashion. Pure offers dynamic typing, eager and lazy evaluation,
353lexical closures, a hygienic macro system (also based on term rewriting),
354built-in list and matrix support (including list and matrix comprehensions) and
355an easy-to-use C interface. The interpreter uses LLVM as a backend to
356 JIT-compile Pure programs to fast native code.</p>
357
Chris Lattner59a30272009-10-08 16:01:33 +0000[diff] [blame]358<p>Pure versions 0.31 and later have been tested and are known to work with
359LLVM 2.6 (and continue to work with older LLVM releases >= 2.3 as well).
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]360</p>
361</div>
362
363
364<!--=========================================================================-->
365<div class="doc_subsection">
366<a name="ldc">LLVM D Compiler</a>
367</div>
368
369<div class="doc_text">
370<p>
371<a href="http://www.dsource.org/projects/ldc">LDC</a> is an implementation of
372the D Programming Language using the LLVM optimizer and code generator.
373The LDC project works great with the LLVM 2.6 release. General improvements in
374this
375cycle have included new inline asm constraint handling, better debug info
Duncan Sandsdc5bd2b2009-10-10 19:30:16 +0000[diff] [blame]376support, general bug fixes, and better x86-64 support. This has allowed
Chris Lattner8b04a442009-10-10 23:05:42 +0000[diff] [blame]377some major improvements in LDC, getting it much closer to being as
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]378fully featured as the original DMD compiler from DigitalMars.
379</p>
380</div>
381
382<!--=========================================================================-->
383<div class="doc_subsection">
384<a name="RoadsendPHP">Roadsend PHP</a>
385</div>
386
387<div class="doc_text">
Chris Lattner169c7012009-10-08 07:01:46 +0000[diff] [blame]388<p>
Chris Lattner169c7012009-10-08 07:01:46 +0000[diff] [blame]389<a href="http://code.roadsend.com/rphp">Roadsend PHP</a> (rphp) is an open
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]390source implementation of the PHP programming
391language that uses LLVM for its optimizer, JIT, and static compiler. This is a
392reimplementation of an earlier project that is now based on LLVM.</p>
393</div>
394
Jeffrey Yasskin3a07d2f2009-06-24 21:09:13 +0000[diff] [blame]395<!--=========================================================================-->
396<div class="doc_subsection">
Chris Lattner6e6d33c2009-10-09 05:55:04 +0000[diff] [blame]397<a name="UnladenSwallow">Unladen Swallow</a>
Jeffrey Yasskin3a07d2f2009-06-24 21:09:13 +0000[diff] [blame]398</div>
399
400<div class="doc_text">
Chris Lattner169c7012009-10-08 07:01:46 +0000[diff] [blame]401<p>
Chris Lattner169c7012009-10-08 07:01:46 +0000[diff] [blame]402<a href="http://code.google.com/p/unladen-swallow/">Unladen Swallow</a> is a
Jeffrey Yasskin3a07d2f2009-06-24 21:09:13 +0000[diff] [blame]403branch of <a href="http://python.org/">Python</a> intended to be fully
404compatible and significantly faster. It uses LLVM's optimization passes and JIT
405compiler.</p>
406</div>
407
Chris Lattner6e6d33c2009-10-09 05:55:04 +0000[diff] [blame]408<!--=========================================================================-->
409<div class="doc_subsection">
410<a name="llvm-lua">llvm-lua</a>
411</div>
412
413<div class="doc_text">
414<p>
415<a href="http://code.google.com/p/llvm-lua/">LLVM-Lua</a> uses LLVM to add JIT
Duncan Sandsd4f53352009-10-10 20:06:04 +0000[diff] [blame]416and static compiling support to the Lua VM. Lua bytecode is analyzed to
417remove type checks, then LLVM is used to compile the bytecode down to machine
Chris Lattner6e6d33c2009-10-09 05:55:04 +0000[diff] [blame]418code.</p>
419</div>
420
Jeffrey Yasskin123b3922009-06-24 21:26:42 +0000[diff] [blame]421
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]422
423<!-- *********************************************************************** -->
424<div class="doc_section">
425 <a name="whatsnew">What's New in LLVM 2.6?</a>
426</div>
427<!-- *********************************************************************** -->
428
429<div class="doc_text">
430
431<p>This release includes a huge number of bug fixes, performance tweaks, and
432minor improvements. Some of the major improvements and new features are listed
433in this section.
434</p>
Chris Lattner6aaf6902009-10-08 06:27:53 +0000[diff] [blame]435
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]436</div>
437
438<!--=========================================================================-->
439<div class="doc_subsection">
440<a name="majorfeatures">Major New Features</a>
441</div>
442
443<div class="doc_text">
444
445<p>LLVM 2.6 includes several major new capabilities:</p>
446
447<ul>
Chris Lattner51ead6d2009-10-10 19:00:55 +0000[diff] [blame]448<li>New <a href="#compiler-rt">compiler-rt</a>, <A href="#klee">KLEE</a>,
Chris Lattner0d6011062009-10-09 06:24:25 +0000[diff] [blame]449 and <a href="#mc">machine code toolkit</a> sub-projects.</li>
450<li>Debug information now includes line numbers when optimizations are enabled.
451 This allows statistical sampling tools like oprofile and Shark to map
452 samples back to source lines.</li>
453<li>LLVM now includes new experimental backends to support the MSP430, SystemZ,
Chris Lattner6e6d33c2009-10-09 05:55:04 +0000[diff] [blame]454 and BlackFin architectures.</li>
455<li>LLVM supports a new <a href="GoldPlugin.html">Gold Linker Plugin</a> which
456 enables support for <a href="LinkTimeOptimization.html">transparent
457 link-time optimization</a> on ELF targets when used with the Gold binutils
458 linker.</li>
Chris Lattner4d91d4d2009-10-10 18:33:13 +0000[diff] [blame]459<li>LLVM now supports doing optimization and code generation on multiple
460 threads. Please see the <a href="ProgrammersManual.html#threading">LLVM
461 Programmer's Manual</a> for more information.</li>
Chris Lattner896d2582009-10-09 06:36:25 +0000[diff] [blame]462<li>LLVM now has experimental support for <a
463 href="http://nondot.org/~sabre/LLVMNotes/EmbeddedMetadata.txt">embedded
464 metadata</a> in LLVM IR, though the implementation is not guaranteed to be
465 final and the .bc file format may change in future releases. Debug info
Chris Lattner4d91d4d2009-10-10 18:33:13 +0000[diff] [blame]466 does not yet use this format in LLVM 2.6.</li>
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]467</ul>
468
469</div>
470
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]471<!--=========================================================================-->
472<div class="doc_subsection">
473<a name="coreimprovements">LLVM IR and Core Improvements</a>
474</div>
475
476<div class="doc_text">
Chris Lattner4d91d4d2009-10-10 18:33:13 +0000[diff] [blame]477<p>LLVM IR has several new features for better support of new targets and that
478expose new optimization opportunities:</p>
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]479
480<ul>
Chris Lattner4d91d4d2009-10-10 18:33:13 +0000[diff] [blame]481<li>The <a href="LangRef.html#i_add">add</a>, <a
482 href="LangRef.html#i_sub">sub</a>, and <a href="LangRef.html#i_mul">mul</a>
Duncan Sandsd4f53352009-10-10 20:06:04 +0000[diff] [blame]483 instructions have been split into integer and floating point versions (like
Chris Lattner4d91d4d2009-10-10 18:33:13 +0000[diff] [blame]484 divide and remainder), introducing new <a
485 href="LangRef.html#i_fadd">fadd</a>, <a href="LangRef.html#i_fsub">fsub</a>,
486 and <a href="LangRef.html#i_fmul">fmul</a> instructions.</li>
487<li>The <a href="LangRef.html#i_add">add</a>, <a
488 href="LangRef.html#i_sub">sub</a>, and <a href="LangRef.html#i_mul">mul</a>
489 instructions now support optional "nsw" and "nuw" bits which indicate that
490 the operation is guaranteed to not overflow (in the signed or
491 unsigned case, respectively). This gives the optimizer more information and
Duncan Sandsd4f53352009-10-10 20:06:04 +0000[diff] [blame]492 can be used for things like C signed integer values, which are undefined on
Chris Lattner4d91d4d2009-10-10 18:33:13 +0000[diff] [blame]493 overflow.</li>
494<li>The <a href="LangRef.html#i_sdiv">sdiv</a> instruction now supports an
495 optional "exact" flag which indicates that the result of the division is
Duncan Sandsd4f53352009-10-10 20:06:04 +0000[diff] [blame]496 guaranteed to have a remainder of zero. This is useful for optimizing pointer
Chris Lattner4d91d4d2009-10-10 18:33:13 +0000[diff] [blame]497 subtraction in C.</li>
498<li>The <a href="LangRef.html#i_getelementptr">getelementptr</a> instruction now
499 supports arbitrary integer index values for array/pointer indices. This
Benjamin Kramer08f0add2009-10-10 19:45:13 +0000[diff] [blame]500 allows for better code generation on 16-bit targets like PIC16.</li>
Chris Lattner4d91d4d2009-10-10 18:33:13 +0000[diff] [blame]501<li>The <a href="LangRef.html#i_getelementptr">getelementptr</a> instruction now
502 supports an "inbounds" optimization hint that tells the optimizer that the
503 pointer is guaranteed to be within its allocated object.</li>
504<li>LLVM now support a series of new linkage types for global values which allow
505 for better optimization and new capabilities:
506 <ul>
507 <li><a href="LangRef.html#linkage_linkonce">linkonce_odr</a> and
508 <a href="LangRef.html#linkage_weak">weak_odr</a> have the same linkage
509 semantics as the non-"odr" linkage types. The difference is that these
510 linkage types indicate that all definitions of the specified function
511 are guaranteed to have the same semantics. This allows inlining
512 templates functions in C++ but not inlining weak functions in C,
513 which previously both got the same linkage type.</li>
514 <li><a href="LangRef.html#linkage_available_externally">available_externally
515 </a> is a new linkage type that gives the optimizer visibility into the
516 definition of a function (allowing inlining and side effect analysis)
517 but that does not cause code to be generated. This allows better
518 optimization of "GNU inline" functions, extern templates, etc.</li>
519 <li><a href="LangRef.html#linkage_linker_private">linker_private</a> is a
520 new linkage type (which is only useful on Mac OS X) that is used for
521 some metadata generation and other obscure things.</li>
522 </ul></li>
523<li>Finally, target-specific intrinsics can now return multiple values, which
524 is useful for modeling target operations with multiple results.</li>
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]525</ul>
526
527</div>
528
529<!--=========================================================================-->
530<div class="doc_subsection">
531<a name="optimizer">Optimizer Improvements</a>
532</div>
533
534<div class="doc_text">
535
Chris Lattner51ead6d2009-10-10 19:00:55 +0000[diff] [blame]536<p>In addition to a large array of minor performance tweaks and bug fixes, this
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]537release includes a few major enhancements and additions to the optimizers:</p>
538
539<ul>
540
Chris Lattner51ead6d2009-10-10 19:00:55 +0000[diff] [blame]541<li>The <a href="Passes.html#scalarrepl">Scalar Replacement of Aggregates</a>
542 pass has many improvements that allow it to better promote vector unions,
Duncan Sandsd4f53352009-10-10 20:06:04 +0000[diff] [blame]543 variables which are memset, and much more strange code that can happen to
544 do bitfield accesses to register operations. An interesting change is that
Chris Lattner51ead6d2009-10-10 19:00:55 +0000[diff] [blame]545 it now produces "unusual" integer sizes (like i1704) in some cases and lets
546 other optimizers clean things up.</li>
547<li>The <a href="Passes.html#loop-reduce">Loop Strength Reduction</a> pass now
548 promotes small integer induction variables to 64-bit on 64-bit targets,
Duncan Sandsd4f53352009-10-10 20:06:04 +0000[diff] [blame]549 which provides a major performance boost for much numerical code. It also
Chris Lattner51ead6d2009-10-10 19:00:55 +0000[diff] [blame]550 promotes shorts to int on 32-bit hosts, etc. LSR now also analyzes pointer
551 expressions (e.g. getelementptrs), as well as integers.</li>
552<li>The <a href="Passes.html#gvn">GVN</a> pass now eliminates partial
553 redundancies of loads in simple cases.</li>
554<li>The <a href="Passes.html#inline">Inliner</a> now reuses stack space when
Duncan Sandsdc5bd2b2009-10-10 19:30:16 +0000[diff] [blame]555 inlining similar arrays from multiple callees into one caller.</li>
Chris Lattner51ead6d2009-10-10 19:00:55 +0000[diff] [blame]556<li>LLVM includes a new experimental Static Single Information (SSI)
557 construction pass.</li>
Chris Lattner6aaf6902009-10-08 06:27:53 +0000[diff] [blame]558</li>
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]559
560</ul>
561
562</div>
563
Chris Lattnerfdf24e72009-10-10 21:40:13 +0000[diff] [blame]564
565<!--=========================================================================-->
566<div class="doc_subsection">
567<a name="executionengine">Interpreter and JIT Improvements</a>
568</div>
569
570<div class="doc_text">
571
572<ul>
573<li>LLVM has a new "EngineBuilder" class which makes it more obvious how to
574 set up and configure an ExecutionEngine (a JIT or interpreter).</li>
575<li>The JIT now supports generating more than 16M of code.</li>
576<li>When configured with --with-oprofile, the JIT can now inform oprofile about
577 JIT'd code, allowing oprofile to get line number and function name
578 information for JIT'd functions.</li>
579<li>When "libffi" is available, the LLVM interpreter now uses it, which supports
580 calling almost arbitrary external (natively compiled) functions.</li>
581<li>Clients of the JIT can now register a 'JITEventListener' object to receive
582 callbacks when the JIT emits or frees machine code. The OProfile support
583 uses this mechanism.</li>
584</ul>
585
586</div>
587
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]588<!--=========================================================================-->
589<div class="doc_subsection">
590<a name="codegen">Target Independent Code Generator Improvements</a>
591</div>
592
593<div class="doc_text">
594
595<p>We have put a significant amount of work into the code generator
596infrastructure, which allows us to implement more aggressive algorithms and make
597it run faster:</p>
598
599<ul>
600
Chris Lattner51ead6d2009-10-10 19:00:55 +0000[diff] [blame]601<li>The <tt>llc -asm-verbose</tt> option (exposed from llvm-gcc and clang as
602 <tt>-fverbose-asm</tt>) now adds a lot of useful information in comments to
603 the generated .s file. This information includes location information (if
604 built with <tt>-g</tt>) and loop nest information.</li>
605<li>The code generator now supports a new MachineVerifier pass which is useful
Duncan Sandsdc5bd2b2009-10-10 19:30:16 +0000[diff] [blame]606 for finding bugs in targets and codegen passes.</li>
Chris Lattner51ead6d2009-10-10 19:00:55 +0000[diff] [blame]607<li>The Machine LICM is now enabled by default. It hoists instructions out of
Duncan Sandsdc5bd2b2009-10-10 19:30:16 +0000[diff] [blame]608 loops (such as constant pool loads, loads from read-only stubs, vector
Chris Lattner51ead6d2009-10-10 19:00:55 +0000[diff] [blame]609 constant synthesization code, etc) and is currently configured to only do so
610 when the hoisted operation can be rematerialized.</li>
611<li>The Machine Sinking pass is now enabled by default. This pass moves
612 side-effect free operations down the CFG so that they are executed on fewer
613 paths through a function.</li>
Chris Lattnerfdf24e72009-10-10 21:40:13 +0000[diff] [blame]614<li>The code generator now performs "stack slot coloring" of register spills,
Chris Lattner10abb752009-10-10 19:26:21 +0000[diff] [blame]615 which allows spill slots to be reused. This leads to smaller stack frames
616 in cases where there are lots of register spills.</li>
617<li>The register allocator has many improvements to take better advantage of
618 commutable operations, various spiller peephole optimizations, and can now
619 coalesce cross-register-class copies.</li>
Chris Lattner6aaf6902009-10-08 06:27:53 +0000[diff] [blame]620<li>Tblgen now supports multiclass inheritance and a number of new string and
621 list operations like !(subst), !(foreach), !car, !cdr, !null, !if, !cast.
622 These make the .td files more expressive and allow more aggressive factoring
623 of duplication across instruction patterns.</li>
Chris Lattner51ead6d2009-10-10 19:00:55 +0000[diff] [blame]624<li>Target-specific intrinsics can now be added without having to hack VMCore to
625 add them. This makes it easier to maintain out-of-tree targets.</li>
Chris Lattner10abb752009-10-10 19:26:21 +0000[diff] [blame]626<li>The instruction selector is better at propagating information about values
627 (such as whether they are sign/zero extended etc) across basic block
628 boundaries.</li>
Chris Lattnerfdf24e72009-10-10 21:40:13 +0000[diff] [blame]629<li>The SelectionDAG datastructure has new nodes for representing buildvector
630 and <a href="http://llvm.org/PR2957">vector shuffle</a> operations. This
631 makes operations and pattern matching more efficient and easier to get
632 right.</li>
Chris Lattner51ead6d2009-10-10 19:00:55 +0000[diff] [blame]633<li>The Prolog/Epilog Insertion Pass now has experimental support for performing
634 the "shrink wrapping" optimization, which moves spills and reloads around in
635 the CFG to avoid doing saves on paths that don't need them.</li>
636<li>LLVM includes new experimental support for writing ELF .o files directly
637 from the compiler. It works well for many simple C testcases, but doesn't
638 support exception handling, debug info, inline assembly, etc.</li>
Chris Lattner6e6d33c2009-10-09 05:55:04 +0000[diff] [blame]639<li>Targets can now specify register allocation hints through
Chris Lattner51ead6d2009-10-10 19:00:55 +0000[diff] [blame]640 MachineRegisterInfo:: setRegAllocationHint. A regalloc hint consists of hint
641 type and physical register number. A hint type of zero specifies a register
642 allocation preference. Other hint type values are target specific which are
Duncan Sandsd4f53352009-10-10 20:06:04 +0000[diff] [blame]643 resolved by TargetRegisterInfo::ResolveRegAllocHint. An example is the ARM
644 target which uses register hints to request that the register allocator
Chris Lattner51ead6d2009-10-10 19:00:55 +0000[diff] [blame]645 provide an even / odd register pair to two virtual registers.</li>
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]646</ul>
647</div>
648
649<!--=========================================================================-->
650<div class="doc_subsection">
651<a name="x86">X86-32 and X86-64 Target Improvements</a>
652</div>
653
654<div class="doc_text">
655<p>New features of the X86 target include:
656</p>
657
658<ul>
659
Chris Lattnerfdf24e72009-10-10 21:40:13 +0000[diff] [blame]660<li>SSE 4.2 builtins are now supported.</li>
661<li>GCC-compatible soft float modes are now supported, which are typically used
662 by OS kernels.</li>
663<li>X86-64 now models implicit zero extensions better, which allows the code
664 generator to remove a lot of redundant zexts. It also models the 8-bit "H"
665 registers as sugregs, which allows they to be used in some tricky
666 situations.</li>
667<li>X86-64 now supports the "local exec" and "initial exec" thread local storage
668 model.</li>
669<li>The vector forms of the <a href="LangRef.html#i_icmp">icmp</a> and <a
670 href="LangRef.html#i_fcmp">fcmp</a> instructions now select to efficient
671 SSE operations.</li>
672<li>The X86 backend has preliminary support for <a
673 href="CodeGenerator.html#x86_memory">mapping address spaces to segment
674 register references</a>. This allows you to write GS or FS relative memory
675 accesses directly in LLVM IR for cases where you know exactly what you're
676 doing (such as in an OS kernel). There are some known problems with this
677 support, but it works in simple cases.</li>
678<li>The X86 code generator has been refactored to move all global variable
679 reference logic to one place
680 (<tt>X86Subtarget::ClassifyGlobalReference</tt>) which
681 makes it easier to reason about.</li>
Chris Lattner6aaf6902009-10-08 06:27:53 +0000[diff] [blame]682</li>
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]683</ul>
684
685</div>
686
687<!--=========================================================================-->
688<div class="doc_subsection">
689<a name="pic16">PIC16 Target Improvements</a>
690</div>
691
692<div class="doc_text">
693<p>New features of the PIC16 target include:
694</p>
695
696<ul>
Chris Lattner6aaf6902009-10-08 06:27:53 +0000[diff] [blame]697<li>Support for floating-point, indirect function calls, and
698 passing/returning aggregate types to functions.
699<li>The code generator is able to generate debug info into output COFF files.
700<li>Support for placing an object into a specific section or at a specific
701 address in memory.</li>
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]702</ul>
703
704<p>Things not yet supported:</p>
705
706<ul>
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]707<li>Variable arguments.</li>
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]708<li>Interrupts/programs.</li>
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]709</ul>
710
711</div>
712
Bob Wilson80333842009-08-12 21:19:49 +0000[diff] [blame]713<!--=========================================================================-->
714<div class="doc_subsection">
715<a name="ARM">ARM Target Improvements</a>
716</div>
717
718<div class="doc_text">
719<p>New features of the ARM target include:
720</p>
721
722<ul>
723
724<li>Preliminary support for processors, such as the Cortex-A8 and Cortex-A9,
Sandeep Patel16eac042009-08-20 15:01:16 +0000[diff] [blame]725that implement version v7-A of the ARM architecture. The ARM backend now
Chris Lattnerfdf24e72009-10-10 21:40:13 +0000[diff] [blame]726supports both the Thumb2 and Advanced SIMD (Neon) instruction sets.</li>
727
728<li>The AAPCS-VFP "hard float" calling conventions are also supported with the
729<tt>-float-abi=hard</tt> flag.</li>
730
731<li>The ARM calling convention code is now tblgen generated instead of C++
732 code.</li>
Bob Wilson80333842009-08-12 21:19:49 +0000[diff] [blame]733</li>
Chris Lattner6aaf6902009-10-08 06:27:53 +0000[diff] [blame]734
Chris Lattnerfdf24e72009-10-10 21:40:13 +0000[diff] [blame]735
736<p>These features are still somewhat experimental
737and subject to change. The Neon intrinsics, in particular, may change in future
738releases of LLVM.</p>
Chris Lattner6aaf6902009-10-08 06:27:53 +0000[diff] [blame]739
Bob Wilson80333842009-08-12 21:19:49 +0000[diff] [blame]740</ul>
741
742</div>
743
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]744<!--=========================================================================-->
745<div class="doc_subsection">
Chris Lattner6aaf6902009-10-08 06:27:53 +0000[diff] [blame]746<a name="OtherTarget">Other Target Specific Improvements</a>
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]747</div>
748
749<div class="doc_text">
Chris Lattner6aaf6902009-10-08 06:27:53 +0000[diff] [blame]750<p>New features of other targets include:
751</p>
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]752
753<ul>
Chris Lattner6aaf6902009-10-08 06:27:53 +0000[diff] [blame]754<li>Mips now supports O32 Calling Convention.</li>
Chris Lattner896d2582009-10-09 06:36:25 +0000[diff] [blame]755<li>Many improvements to the 32-bit PowerPC SVR4 ABI (used on powerpc-linux)
756 support, lots of bugs fixed.</li>
757<li>Added support for the 64-bit PowerPC SVR4 ABI (used on powerpc64-linux).
758 Needs more testing.</li>
Chris Lattner6aaf6902009-10-08 06:27:53 +0000[diff] [blame]759</ul>
760
761</div>
762
763<!--=========================================================================-->
764<div class="doc_subsection">
765<a name="newapis">New Useful APIs</a>
766</div>
767
768<div class="doc_text">
769
Chris Lattner835f2532009-10-10 22:02:58 +0000[diff] [blame]770<p>This release includes a number of new APIs that are used internally, which
771 may also be useful for external clients.
772</p>
773
Chris Lattner6aaf6902009-10-08 06:27:53 +0000[diff] [blame]774<ul>
Chris Lattner835f2532009-10-10 22:02:58 +0000[diff] [blame]775<li>New <a href="http://llvm.org/doxygen/PrettyStackTrace_8h-source.html">
776 PrettyStackTrace classes</a> allows crashes of llvm tools (and applications
777 that integrate them) to provide more detailed indication of what the
778 compiler was doing at the time of the crash (e.g. running a pass).
779 At the top level for each LLVM tool, it includes the command line arguments.
780 </li>
781<li>New <a href="http://llvm.org/doxygen/StringRef_8h-source.html">StringRef</a>
782 and <a href="http://llvm.org/doxygen/Twine_8h-source.html">Twine</a> classes
783 make operations on character ranges and
784 string concatenation to be more efficient. StringRef is just a <tt>const
785 char*</tt> with a length, Twine is a light-weight rope.</li>
786<li>LLVM has new WeakVH, AssertingVH, and CallbackVH classes, which make it
787 easier to write LLVM IR transformations. WeakVH is automatically drops to
788 null when the referenced Value is deleted, and is updated across a
789 replaceAllUsesWith operation. AssertingVH aborts the program if the
790 referenced value is destroyed while it is being referenced. CallbackVH is
791 a customizable class for handling value references. See <a
792 href="http://llvm.org/doxygen/ValueHandle_8h-source.html">ValueHandle.h</a>
793 for more information.</li>
794<li>The new '<a href="http://llvm.org/doxygen/Triple_8h-source.html">Triple
795 </a>' class centralizes a lot of logic that reasons about target
796 triples.</li>
Chris Lattnerb9a51552009-10-10 22:15:25 +0000[diff] [blame]797<li>The new '<a href="http://llvm.org/doxygen/ErrorHandling_8h-source.html">
Chris Lattner835f2532009-10-10 22:02:58 +0000[diff] [blame]798 llvm_report_error()</a>' set of APIs allows tools to embed the LLVM
799 optimizer and backend and recover from previously unrecoverable errors.</li>
800<li>LLVM has new abstractions for <a
801 href="http://llvm.org/doxygen/Atomic_8h-source.html">atomic operations</a>
802 and <a href="http://llvm.org/doxygen/RWMutex_8h-source.html">reader/writer
803 locks</a>.</li>
804<li>LLVM has new <a href="http://llvm.org/doxygen/SourceMgr_8h-source.html">
805 SourceMgr and SMLoc classes</a> which implement caret diagnostics and basic
806 include stack processing for simple parsers. It is used by tablegen,
807 llvm-mc, the .ll parser and FileCheck.</li>
Chris Lattner0d6011062009-10-09 06:24:25 +0000[diff] [blame]808</ul>
Chris Lattner6aaf6902009-10-08 06:27:53 +0000[diff] [blame]809
810
811</div>
812
813<!--=========================================================================-->
814<div class="doc_subsection">
815<a name="otherimprovements">Other Improvements and New Features</a>
816</div>
817
818<div class="doc_text">
819<p>Other miscellaneous features include:</p>
820
821<ul>
Chris Lattner0d6011062009-10-09 06:24:25 +0000[diff] [blame]822<li>LLVM now includes a new internal '<a
823 href="http://llvm.org/cmds/FileCheck.html">FileCheck</a>' tool which allows
824 writing much more accurate regression tests that run faster. Please see the
825 <a href="TestingGuide.html#FileCheck">FileCheck section of the Testing
826 Guide</a> for more information.</li>
Chris Lattner6e6d33c2009-10-09 05:55:04 +0000[diff] [blame]827<li>LLVM profile information support has been significantly improved to produce
828correct use counts, and has support for edge profiling with reduced runtime
829overhead. Combined, the generated profile information is both more correct and
830imposes about half as much overhead (2.6. from 12% to 6% overhead on SPEC
831CPU2000).</li>
Chris Lattner835f2532009-10-10 22:02:58 +0000[diff] [blame]832<li>The C bindings (in the llvm/include/llvm-c directory) include many newly
833 supported APIs.</li>
Chris Lattner0d6011062009-10-09 06:24:25 +0000[diff] [blame]834<li>LLVM 2.6 includes a brand new experimental LLVM bindings to the Ada2005
Chris Lattner835f2532009-10-10 22:02:58 +0000[diff] [blame]835 programming language.</li>
Chris Lattner169c7012009-10-08 07:01:46 +0000[diff] [blame]836
Chris Lattner835f2532009-10-10 22:02:58 +0000[diff] [blame]837<li>The LLVMC driver has several new features:
838 <ul>
839 <li>Dynamic plugins now work on Windows.</li>
840 <li>New option property: init. Makes possible to provide default values for
841 options defined in plugins (interface to cl::init).</li>
842 <li>New example: Skeleton, shows how to create a standalone LLVMC-based
843 driver.</li>
844 <li>New example: mcc16, a driver for the PIC16 toolchain.</li>
845 </ul>
846</li>
Chris Lattner169c7012009-10-08 07:01:46 +0000[diff] [blame]847
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]848</ul>
849
850</div>
851
852
853<!--=========================================================================-->
854<div class="doc_subsection">
855<a name="changes">Major Changes and Removed Features</a>
856</div>
857
858<div class="doc_text">
859
860<p>If you're already an LLVM user or developer with out-of-tree changes based
Owen Andersonbe3fe4e2009-07-02 16:48:38 +0000[diff] [blame]861on LLVM 2.5, this section lists some "gotchas" that you may run into upgrading
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]862from the previous release.</p>
863
864<ul>
Chris Lattner8767bdc2009-10-10 22:13:38 +0000[diff] [blame]865<li>The Itanium (IA64) backend has been removed. It was not actively supported
866 and had bitrotted.</li>
Duncan Sandsd4f53352009-10-10 20:06:04 +0000[diff] [blame]867<li>The BigBlock register allocator has been removed, it had also bitrotted.</li>
Chris Lattner6aaf6902009-10-08 06:27:53 +0000[diff] [blame]868<li>The C Backend (-march=c) is no longer considered part of the LLVM release
869criteria. We still want it to work, but no one is maintaining it and it lacks
870support for arbitrary precision integers and other important IR features.</li>
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]871
Chris Lattner8767bdc2009-10-10 22:13:38 +0000[diff] [blame]872<li>All LLVM tools now default to overwriting their output file, behaving more
873 like standard unix tools. Previously, this only happened with the '-f'
874 option.</li>
875<li>LLVM build now builds all libraries as .a files instead of some
Chris Lattner169c7012009-10-08 07:01:46 +0000[diff] [blame]876 libraries as relinked .o files. This requires some APIs like
Chris Lattner8767bdc2009-10-10 22:13:38 +0000[diff] [blame]877 InitializeAllTargets.h.
878 </li>
879</ul>
Chris Lattner169c7012009-10-08 07:01:46 +0000[diff] [blame]880
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]881
882<p>In addition, many APIs have changed in this release. Some of the major LLVM
883API changes are:</p>
884
885<ul>
Chris Lattner8767bdc2009-10-10 22:13:38 +0000[diff] [blame]886<li>All uses of hash_set and hash_map have been removed from the LLVM tree and
887 the wrapper headers have been removed.</li>
888<li>The llvm/Streams.h and DOUT member of Debug.h have been removed. The
889 llvm::Ostream class has been completely removed and replaced with uses of
890 raw_ostream.</li>
Owen Andersonbe3fe4e2009-07-02 16:48:38 +0000[diff] [blame]891<li>LLVM's global uniquing tables for <tt>Type</tt>s and <tt>Constant</tt>s have
892 been privatized into members of an <tt>LLVMContext</tt>. A number of APIs
893 now take an <tt>LLVMContext</tt> as a parameter. To smooth the transition
894 for clients that will only ever use a single context, the new
895 <tt>getGlobalContext()</tt> API can be used to access a default global
896 context which can be passed in any and all cases where a context is
897 required.
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]898<li>The <tt>getABITypeSize</tt> methods are now called <tt>getAllocSize</tt>.</li>
Dan Gohman79537c92009-07-07 20:05:15 +0000[diff] [blame]899<li>The <tt>Add</tt>, <tt>Sub</tt>, and <tt>Mul</tt> operators are no longer
900 overloaded for floating-point types. Floating-point addition, subtraction,
901 and multiplication are now represented with new operators <tt>FAdd</tt>,
902 <tt>FSub</tt>, and <tt>FMul</tt>. In the <tt>IRBuilder</tt> API,
903 <tt>CreateAdd</tt>, <tt>CreateSub</tt>, <tt>CreateMul</tt>, and
904 <tt>CreateNeg</tt> should only be used for integer arithmetic now;
905 <tt>CreateFAdd</tt>, <tt>CreateFSub</tt>, <tt>CreateFMul</tt>, and
906 <tt>CreateFNeg</tt> should now be used for floating-point arithmetic.</li>
Daniel Dunbara7d9e052009-07-12 20:41:27 +0000[diff] [blame]907<li>The DynamicLibrary class can no longer be constructed, its functionality has
908 moved to static member functions.</li>
Dan Gohmanb6b11702009-07-15 19:59:19 +0000[diff] [blame]909<li><tt>raw_fd_ostream</tt>'s constructor for opening a given filename now
910 takes an extra <tt>Force</tt> argument. If <tt>Force</tt> is set to
911 <tt>false</tt>, an error will be reported if a file with the given name
912 already exists. If <tt>Force</tt> is set to <tt>true</tt>, the file will
913 be silently truncated (which is the behavior before this flag was
914 added).</li>
Edwin Török07768e22009-07-21 20:27:10 +0000[diff] [blame]915<li><tt>SCEVHandle</tt> no longer exists, because reference counting is no
916longer done for <tt>SCEV*</tt> objects, instead <tt>const SCEV*</tt> should be
917used.</li>
Daniel Dunbarbbfb87a2009-07-25 05:26:53 +0000[diff] [blame]918
Daniel Dunbare3572ba2009-07-25 04:41:11 +0000[diff] [blame]919<li>Many APIs, notably <tt>llvm::Value</tt>, now use the <tt>StringRef</tt>
920and <tt>Twine</tt> classes instead of passing <tt>const char*</tt>
921or <tt>std::string</tt>, as described in
922the <a href="ProgrammersManual.html#string_apis">Programmer's Manual</a>. Most
Chris Lattner8767bdc2009-10-10 22:13:38 +0000[diff] [blame]923clients should be unaffected by this transition, unless they are used to
924<tt>Value::getName()</tt> returning a string. Here are some tips on updating to
9252.6:
Daniel Dunbarbbfb87a2009-07-25 05:26:53 +0000[diff] [blame]926 <ul>
927 <li><tt>getNameStr()</tt> is still available, and matches the old
928 behavior. Replacing <tt>getName()</tt> calls with this is an safe option,
929 although more efficient alternatives are now possible.</li>
930
931 <li>If you were just relying on <tt>getName()</tt> being able to be sent to
932 a <tt>std::ostream</tt>, consider migrating
933 to <tt>llvm::raw_ostream</tt>.</li>
934
935 <li>If you were using <tt>getName().c_str()</tt> to get a <tt>const
936 char*</tt> pointer to the name, you can use <tt>getName().data()</tt>.
937 Note that this string (as before), may not be the entire name if the
Benjamin Kramer08f0add2009-10-10 19:45:13 +0000[diff] [blame]938 name contains embedded null characters.</li>
Daniel Dunbarbbfb87a2009-07-25 05:26:53 +0000[diff] [blame]939
940 <li>If you were using operator plus on the result of <tt>getName()</tt> and
941 treating the result as an <tt>std::string</tt>, you can either
Duncan Sandsd4f53352009-10-10 20:06:04 +0000[diff] [blame]942 use <tt>Twine::str</tt> to get the result as an <tt>std::string</tt>, or
Daniel Dunbarbbfb87a2009-07-25 05:26:53 +0000[diff] [blame]943 could move to a <tt>Twine</tt> based design.</li>
Daniel Dunbare03513b2009-07-25 23:55:21 +0000[diff] [blame]944
945 <li><tt>isName()</tt> should be replaced with comparison
Benjamin Kramer7b2136d2009-08-05 15:42:44 +0000[diff] [blame]946 against <tt>getName()</tt> (this is now efficient).
Daniel Dunbarbbfb87a2009-07-25 05:26:53 +0000[diff] [blame]947 </ul>
948</li>
949
Daniel Dunbar48224ee2009-07-26 02:12:58 +0000[diff] [blame]950<li>The registration interfaces for backend Targets has changed (what was
Chris Lattner8767bdc2009-10-10 22:13:38 +0000[diff] [blame]951previously TargetMachineRegistry). For backend authors, see the <a
952href="WritingAnLLVMBackend.html#TargetRegistration">Writing An LLVM Backend</a>
953guide. For clients, the notable API changes are:
Daniel Dunbarc9a70092009-07-26 05:41:39 +0000[diff] [blame]954 <ul>
955 <li><tt>TargetMachineRegistry</tt> has been renamed
956 to <tt>TargetRegistry</tt>.</li>
957
958 <li>Clients should move to using the <tt>TargetRegistry::lookupTarget()</tt>
959 function to find targets.</li>
960 </ul>
961</li>
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]962</ul>
963
964</div>
965
966
967
968<!-- *********************************************************************** -->
969<div class="doc_section">
970 <a name="portability">Portability and Supported Platforms</a>
971</div>
972<!-- *********************************************************************** -->
973
974<div class="doc_text">
975
976<p>LLVM is known to work on the following platforms:</p>
977
978<ul>
979<li>Intel and AMD machines (IA32, X86-64, AMD64, EMT-64) running Red Hat
Chris Lattner8767bdc2009-10-10 22:13:38 +0000[diff] [blame]980 Linux, Fedora Core, FreeBSD and AuroraUX (and probably other unix-like
981 systems).</li>
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]982<li>PowerPC and X86-based Mac OS X systems, running 10.3 and above in 32-bit
Chris Lattner8767bdc2009-10-10 22:13:38 +0000[diff] [blame]983 and 64-bit modes.</li>
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]984<li>Intel and AMD machines running on Win32 using MinGW libraries (native).</li>
985<li>Intel and AMD machines running on Win32 with the Cygwin libraries (limited
986 support is available for native builds with Visual C++).</li>
987<li>Sun UltraSPARC workstations running Solaris 10.</li>
988<li>Alpha-based machines running Debian GNU/Linux.</li>
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]989</ul>
990
991<p>The core LLVM infrastructure uses GNU autoconf to adapt itself
992to the machine and operating system on which it is built. However, minor
993porting may be required to get LLVM to work on new platforms. We welcome your
994portability patches and reports of successful builds or error messages.</p>
995
996</div>
997
998<!-- *********************************************************************** -->
999<div class="doc_section">
1000 <a name="knownproblems">Known Problems</a>
1001</div>
1002<!-- *********************************************************************** -->
1003
1004<div class="doc_text">
1005
1006<p>This section contains significant known problems with the LLVM system,
1007listed by component. If you run into a problem, please check the <a
1008href="http://llvm.org/bugs/">LLVM bug database</a> and submit a bug if
1009there isn't already one.</p>
1010
Chris Lattnera8f23072009-07-21 22:47:03 +0000[diff] [blame]1011<ul>
Chris Lattner4050d252009-07-21 23:17:26 +0000[diff] [blame]1012<li>LLVM will not correctly compile on Solaris and/or OpenSolaris
Chris Lattnera8f23072009-07-21 22:47:03 +0000[diff] [blame]1013using the stock GCC 3.x.x series 'out the box',
1014See: <a href="#brokengcc">Broken versions of GCC and other tools</a>.
1015However, A <a href="http://pkg.auroraux.org/GCC">Modern GCC Build</a>
1016for x86/x64 has been made available from the third party AuroraUX Project
Chris Lattner6aaf6902009-10-08 06:27:53 +0000[diff] [blame]1017that has been meticulously tested for bootstrapping LLVM &amp; Clang.</li>
Chris Lattnera8f23072009-07-21 22:47:03 +0000[diff] [blame]1018</ul>
1019
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]1020</div>
1021
1022<!-- ======================================================================= -->
1023<div class="doc_subsection">
1024 <a name="experimental">Experimental features included with this release</a>
1025</div>
1026
1027<div class="doc_text">
1028
1029<p>The following components of this LLVM release are either untested, known to
1030be broken or unreliable, or are in early development. These components should
1031not be relied on, and bugs should not be filed against them, but they may be
1032useful to some people. In particular, if you would like to work on one of these
1033components, please contact us on the <a
1034href="http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev">LLVMdev list</a>.</p>
1035
1036<ul>
Chris Lattner8767bdc2009-10-10 22:13:38 +0000[diff] [blame]1037<li>The MSIL, Alpha, SPU, MIPS, PIC16, Blackfin, MSP430 and SystemZ backends are
1038 experimental.</li>
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]1039<li>The <tt>llc</tt> "<tt>-filetype=asm</tt>" (the default) is the only
Chris Lattner8767bdc2009-10-10 22:13:38 +0000[diff] [blame]1040 supported value for this option. The ELF writer is experimental.</li>
1041<li>The implementation of Andersen's Alias Analysis has many known bugs.</li>
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]1042</ul>
1043
1044</div>
1045
1046<!-- ======================================================================= -->
1047<div class="doc_subsection">
1048 <a name="x86-be">Known problems with the X86 back-end</a>
1049</div>
1050
1051<div class="doc_text">
1052
1053<ul>
1054 <li>The X86 backend does not yet support
1055 all <a href="http://llvm.org/PR879">inline assembly that uses the X86
1056 floating point stack</a>. It supports the 'f' and 't' constraints, but not
1057 'u'.</li>
1058 <li>The X86 backend generates inefficient floating point code when configured
1059 to generate code for systems that don't have SSE2.</li>
1060 <li>Win64 code generation wasn't widely tested. Everything should work, but we
1061 expect small issues to happen. Also, llvm-gcc cannot build the mingw64
1062 runtime currently due
1063 to <a href="http://llvm.org/PR2255">several</a>
1064 <a href="http://llvm.org/PR2257">bugs</a> and due to lack of support for
1065 the
1066 'u' inline assembly constraint and for X87 floating point inline assembly.</li>
1067 <li>The X86-64 backend does not yet support the LLVM IR instruction
1068 <tt>va_arg</tt>. Currently, the llvm-gcc and front-ends support variadic
1069 argument constructs on X86-64 by lowering them manually.</li>
1070</ul>
1071
1072</div>
1073
1074<!-- ======================================================================= -->
1075<div class="doc_subsection">
1076 <a name="ppc-be">Known problems with the PowerPC back-end</a>
1077</div>
1078
1079<div class="doc_text">
1080
1081<ul>
1082<li>The Linux PPC32/ABI support needs testing for the interpreter and static
1083compilation, and lacks support for debug information.</li>
1084</ul>
1085
1086</div>
1087
1088<!-- ======================================================================= -->
1089<div class="doc_subsection">
1090 <a name="arm-be">Known problems with the ARM back-end</a>
1091</div>
1092
1093<div class="doc_text">
1094
1095<ul>
Bob Wilson80333842009-08-12 21:19:49 +0000[diff] [blame]1096<li>Support for the Advanced SIMD (Neon) instruction set is still incomplete
1097and not well tested. Some features may not work at all, and the code quality
1098may be poor in some cases.</li>
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]1099<li>Thumb mode works only on ARMv6 or higher processors. On sub-ARMv6
1100processors, thumb programs can crash or produce wrong
1101results (<a href="http://llvm.org/PR1388">PR1388</a>).</li>
1102<li>Compilation for ARM Linux OABI (old ABI) is supported but not fully tested.
1103</li>
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]1104</ul>
1105
1106</div>
1107
1108<!-- ======================================================================= -->
1109<div class="doc_subsection">
1110 <a name="sparc-be">Known problems with the SPARC back-end</a>
1111</div>
1112
1113<div class="doc_text">
1114
1115<ul>
1116<li>The SPARC backend only supports the 32-bit SPARC ABI (-m32); it does not
1117 support the 64-bit SPARC ABI (-m64).</li>
1118</ul>
1119
1120</div>
1121
1122<!-- ======================================================================= -->
1123<div class="doc_subsection">
1124 <a name="mips-be">Known problems with the MIPS back-end</a>
1125</div>
1126
1127<div class="doc_text">
1128
1129<ul>
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]1130<li>64-bit MIPS targets are not supported yet.</li>
1131</ul>
1132
1133</div>
1134
1135<!-- ======================================================================= -->
1136<div class="doc_subsection">
1137 <a name="alpha-be">Known problems with the Alpha back-end</a>
1138</div>
1139
1140<div class="doc_text">
1141
1142<ul>
1143
1144<li>On 21164s, some rare FP arithmetic sequences which may trap do not have the
1145appropriate nops inserted to ensure restartability.</li>
1146
1147</ul>
1148</div>
1149
1150<!-- ======================================================================= -->
1151<div class="doc_subsection">
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]1152 <a name="c-be">Known problems with the C back-end</a>
1153</div>
1154
1155<div class="doc_text">
1156
1157<ul>
1158<li><a href="http://llvm.org/PR802">The C backend has only basic support for
1159 inline assembly code</a>.</li>
1160<li><a href="http://llvm.org/PR1658">The C backend violates the ABI of common
1161 C++ programs</a>, preventing intermixing between C++ compiled by the CBE and
1162 C++ code compiled with <tt>llc</tt> or native compilers.</li>
1163<li>The C backend does not support all exception handling constructs.</li>
1164<li>The C backend does not support arbitrary precision integers.</li>
1165</ul>
1166
1167</div>
1168
1169
1170<!-- ======================================================================= -->
1171<div class="doc_subsection">
1172 <a name="c-fe">Known problems with the llvm-gcc C front-end</a>
1173</div>
1174
1175<div class="doc_text">
1176
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]1177<p>The only major language feature of GCC not supported by llvm-gcc is
1178 the <tt>__builtin_apply</tt> family of builtins. However, some extensions
1179 are only supported on some targets. For example, trampolines are only
1180 supported on some targets (these are used when you take the address of a
1181 nested function).</p>
1182
1183<p>If you run into GCC extensions which are not supported, please let us know.
1184</p>
1185
1186</div>
1187
1188<!-- ======================================================================= -->
1189<div class="doc_subsection">
1190 <a name="c++-fe">Known problems with the llvm-gcc C++ front-end</a>
1191</div>
1192
1193<div class="doc_text">
1194
1195<p>The C++ front-end is considered to be fully
1196tested and works for a number of non-trivial programs, including LLVM
1197itself, Qt, Mozilla, etc.</p>
1198
1199<ul>
1200<li>Exception handling works well on the X86 and PowerPC targets. Currently
1201 only Linux and Darwin targets are supported (both 32 and 64 bit).</li>
1202</ul>
1203
1204</div>
1205
1206<!-- ======================================================================= -->
1207<div class="doc_subsection">
1208 <a name="fortran-fe">Known problems with the llvm-gcc Fortran front-end</a>
1209</div>
1210
1211<div class="doc_text">
1212<ul>
1213<li>Fortran support generally works, but there are still several unresolved bugs
1214 in Bugzilla. Please see the tools/gfortran component for details.</li>
1215</ul>
1216</div>
1217
1218<!-- ======================================================================= -->
1219<div class="doc_subsection">
1220 <a name="ada-fe">Known problems with the llvm-gcc Ada front-end</a>
1221</div>
1222
1223<div class="doc_text">
1224The llvm-gcc 4.2 Ada compiler works fairly well; however, this is not a mature
1225technology, and problems should be expected.
1226<ul>
1227<li>The Ada front-end currently only builds on X86-32. This is mainly due
1228to lack of trampoline support (pointers to nested functions) on other platforms.
1229However, it <a href="http://llvm.org/PR2006">also fails to build on X86-64</a>
1230which does support trampolines.</li>
1231<li>The Ada front-end <a href="http://llvm.org/PR2007">fails to bootstrap</a>.
1232This is due to lack of LLVM support for <tt>setjmp</tt>/<tt>longjmp</tt> style
1233exception handling, which is used internally by the compiler.
1234Workaround: configure with --disable-bootstrap.</li>
1235<li>The c380004, <a href="http://llvm.org/PR2010">c393010</a>
1236and <a href="http://llvm.org/PR2421">cxg2021</a> ACATS tests fail
1237(c380004 also fails with gcc-4.2 mainline).
1238If the compiler is built with checks disabled then <a href="http://llvm.org/PR2010">c393010</a>
1239causes the compiler to go into an infinite loop, using up all system memory.</li>
1240<li>Some GCC specific Ada tests continue to crash the compiler.</li>
1241<li>The -E binder option (exception backtraces)
1242<a href="http://llvm.org/PR1982">does not work</a> and will result in programs
1243crashing if an exception is raised. Workaround: do not use -E.</li>
1244<li>Only discrete types <a href="http://llvm.org/PR1981">are allowed to start
1245or finish at a non-byte offset</a> in a record. Workaround: do not pack records
1246or use representation clauses that result in a field of a non-discrete type
1247starting or finishing in the middle of a byte.</li>
1248<li>The <tt>lli</tt> interpreter <a href="http://llvm.org/PR2009">considers
1249'main' as generated by the Ada binder to be invalid</a>.
1250Workaround: hand edit the file to use pointers for <tt>argv</tt> and
1251<tt>envp</tt> rather than integers.</li>
1252<li>The <tt>-fstack-check</tt> option <a href="http://llvm.org/PR2008">is
1253ignored</a>.</li>
1254</ul>
1255</div>
1256
Erick Tryzelaarb4cf9752009-09-28 04:42:55 +0000[diff] [blame]1257<!-- ======================================================================= -->
1258<div class="doc_subsection">
1259 <a name="ocaml-bindingse">Known problems with the O'Caml bindings</a>
1260</div>
1261
1262<div class="doc_text">
1263
1264<p>The Llvm.Linkage module is broken, and has incorrect values. Only
1265Llvm.Linkage.External, Llvm.Linkage.Available_externally, and
1266Llvm.Linkage.Link_once will be correct. If you need any of the other linkage
1267modes, you'll have to write an external C library in order to expose the
1268functionality. This has been fixed in the trunk.</p>
1269</div>
1270
Duncan Sands54fbb412009-06-24 08:38:48 +0000[diff] [blame]1271<!-- *********************************************************************** -->
1272<div class="doc_section">
1273 <a name="additionalinfo">Additional Information</a>
1274</div>
1275<!-- *********************************************************************** -->
1276
1277<div class="doc_text">
1278
1279<p>A wide variety of additional information is available on the <a
1280href="http://llvm.org">LLVM web page</a>, in particular in the <a
1281href="http://llvm.org/docs/">documentation</a> section. The web page also
1282contains versions of the API documentation which is up-to-date with the
1283Subversion version of the source code.
1284You can access versions of these documents specific to this release by going
1285into the "<tt>llvm/doc/</tt>" directory in the LLVM tree.</p>
1286
1287<p>If you have any questions or comments about LLVM, please feel free to contact
1288us via the <a href="http://llvm.org/docs/#maillist"> mailing
1289lists</a>.</p>
1290
1291</div>
1292
1293<!-- *********************************************************************** -->
1294
1295<hr>
1296<address>
1297 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
1298 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
1299 <a href="http://validator.w3.org/check/referer"><img
1300 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
1301
1302 <a href="http://llvm.org/">LLVM Compiler Infrastructure</a><br>
1303 Last modified: $Date$
1304</address>
1305
1306</body>
1307</html>