Gitiles
Code Review Sign In
gerrit-public.fairphone.software / fp2-dev / platform / external / llvm / f65224fa5f1ff6daabeef2f1e14cbec1f8d70c61 / . / docs / ReleaseNotes.html
blob: afa0619f31569b0a463af53abf57387c9059a195 [file] [log] [blame]
Anton Korobeynikov877b73b2008-06-08 10:23:46 +0000[diff] [blame]1
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]2<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
3 "http://www.w3.org/TR/html4/strict.dtd">
4<html>
5<head>
6 <meta http-equiv="Content-Type" content="text/html; charset=utf-8">
7 <link rel="stylesheet" href="llvm.css" type="text/css">
Chris Lattnerdaeb3232008-02-12 06:29:45 +0000[diff] [blame]8 <title>LLVM 2.3 Release Notes</title>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]9</head>
10<body>
11
Chris Lattnerdaeb3232008-02-12 06:29:45 +0000[diff] [blame]12<div class="doc_title">LLVM 2.3 Release Notes</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]13
14<ol>
15 <li><a href="#intro">Introduction</a></li>
16 <li><a href="#whatsnew">What's New?</a></li>
17 <li><a href="GettingStarted.html">Installation Instructions</a></li>
18 <li><a href="#portability">Portability and Supported Platforms</a></li>
19 <li><a href="#knownproblems">Known Problems</a>
20 <li><a href="#additionalinfo">Additional Information</a></li>
21</ol>
22
23<div class="doc_author">
24 <p>Written by the <a href="http://llvm.org">LLVM Team</a><p>
25</div>
26
Chris Lattnere45f18b2008-06-05 08:02:49 +0000[diff] [blame]27<!-- Done through Week-of-Mon-20080324.txt -->
28
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]29<!-- *********************************************************************** -->
30<div class="doc_section">
31 <a name="intro">Introduction</a>
32</div>
33<!-- *********************************************************************** -->
34
35<div class="doc_text">
36
37<p>This document contains the release notes for the LLVM compiler
Chris Lattnerdaeb3232008-02-12 06:29:45 +0000[diff] [blame]38infrastructure, release 2.3. Here we describe the status of LLVM, including
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]39major improvements from the previous release and any known problems. All LLVM
40releases may be downloaded from the <a href="http://llvm.org/releases/">LLVM
41releases web site</a>.</p>
42
43<p>For more information about LLVM, including information about the latest
44release, please check out the <a href="http://llvm.org/">main LLVM
45web site</a>. If you have questions or comments, the <a
46href="http://mail.cs.uiuc.edu/mailman/listinfo/llvmdev">LLVM developer's mailing
47list</a> is a good place to send them.</p>
48
49<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]50main LLVM web page, this document applies to the <i>next</i> release, not the
51current one. To see the release notes for a specific releases, please see the
52<a href="http://llvm.org/releases/">releases page</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]53
54</div>
55
56<!-- *********************************************************************** -->
57<div class="doc_section">
58 <a name="whatsnew">What's New?</a>
59</div>
60<!-- *********************************************************************** -->
61
62<div class="doc_text">
63
Chris Lattnerdaeb3232008-02-12 06:29:45 +0000[diff] [blame]64<p>This is the fourteenth public release of the LLVM Compiler Infrastructure.
Chris Lattnerc5a4e312008-06-05 06:25:56 +0000[diff] [blame]65It includes a large number of features and refinements from LLVM 2.2.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]66
67</div>
68
Chris Lattnerdaeb3232008-02-12 06:29:45 +0000[diff] [blame]69<!-- Unfinished features in 2.3:
Chris Lattner8fda4952008-02-10 07:46:44 +0000[diff] [blame]70 Machine LICM
71 Machine Sinking
72 LegalizeDAGTypes
73 -->
74
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]75<!--=========================================================================-->
76<div class="doc_subsection">
Chris Lattnere45f18b2008-06-05 08:02:49 +0000[diff] [blame]77<a name="changes">Major Changes in LLVM 2.3</a>
Chris Lattner41bf8e92008-02-10 07:04:35 +0000[diff] [blame]78</div>
79
80<div class="doc_text">
81
Chris Lattnerf65224f2008-06-08 21:34:41 +0000[diff] [blame^]82<p>LLVM 2.3 no longer supports llvm-gcc 4.0, it has been replaced with
83 llvm-gcc 4.2.</p>
Chris Lattnere0798082008-02-06 18:00:06 +0000[diff] [blame]84
Chris Lattnerf65224f2008-06-08 21:34:41 +0000[diff] [blame^]85<p>LLVM 2.3 no longer includes the <tt>llvm-upgrade</tt> tool. It was useful
86 for upgrading LLVM 1.9 files to LLVM 2.x syntax, but you can always use a
87 previous LLVM release to do this. One nice impact of this is that the LLVM
88 regression test suite no longer depends on llvm-upgrade, which makes it run
89 faster.</p>
90
91<p>The <tt>llvm2cpp</tt> tool has been folded into llc, use
Chris Lattnerd95b5962008-06-08 21:19:07 +0000[diff] [blame]92 <tt>llc -march=cpp</tt> instead of <tt>llvm2cpp</tt>.</p>
93
Chris Lattner0eef6e42008-06-08 02:45:07 +0000[diff] [blame]94<p>LLVM API Changes:</p>
Chris Lattnere45f18b2008-06-05 08:02:49 +0000[diff] [blame]95
Chris Lattner0eef6e42008-06-08 02:45:07 +0000[diff] [blame]96<ul>
Chris Lattnerf65224f2008-06-08 21:34:41 +0000[diff] [blame^]97<li>Several core LLVM IR classes have migrated to use the
98 '<tt>FOOCLASS::Create(...)</tt>' pattern instead of '<tt>new
99 FOOCLASS(...)</tt>' (e.g. where FOOCLASS=<tt>BasicBlock</tt>). We hope to
100 standardize on <tt>FOOCLASS::Create</tt> for all IR classes in the future,
101 but not all of them have been moved over yet.</li>
Chris Lattner0eef6e42008-06-08 02:45:07 +0000[diff] [blame]102<li>LLVM 2.3 renames the LLVMBuilder and LLVMFoldingBuilder classes to
103 IRBuilder.</li>
104<li>MRegisterInfo was renamed to TargetRegisterInfo.</li>
105<li>The MappedFile class is gone, please use MemoryBuffer instead.</li>
Chris Lattnerf65224f2008-06-08 21:34:41 +0000[diff] [blame^]106<li>The '<tt>-enable-eh</tt>' flag to llc has been removed. Now code should
107 encode whether it is safe to omit unwind information for a function by
108 tagging the Function object with the '<tt>nounwind</tt>' attribute.</li>
Chris Lattner0eef6e42008-06-08 02:45:07 +0000[diff] [blame]109
110</ul>
Chris Lattnere0798082008-02-06 18:00:06 +0000[diff] [blame]111</div>
112
113<!--=========================================================================-->
114<div class="doc_subsection">
Chris Lattnerf65224f2008-06-08 21:34:41 +0000[diff] [blame^]115<a name="otherprojects">Other LLVM Sub-Projects</a>
116</div>
117
118<div class="doc_text">
119
120<p>
121<p>The <a href="http://clang.llvm.org/">clang project</a> is an effort to build
122a set of new 'llvm native' front-end technologies for the LLVM optimizer
123and code generator. Currently, its C and Objective-C support is maturing
124nicely, and it has advanced source-to-source analysis and transformation
125capabilities. If you are interested in building source-level tools for C and
126Objective-C (and eventually C++), you should take a look. However, note that
127clang is not an official part of the LLVM 2.3 release. If you are interested in
128this project, please see its <a href="http://clang.llvm.org/">web site</a>.</p>
129
130
131</p>
132
133</div>
134
135<!--=========================================================================-->
136<div class="doc_subsection">
Chris Lattnere0798082008-02-06 18:00:06 +0000[diff] [blame]137<a name="majorfeatures">Major New Features</a>
138</div>
139
140<div class="doc_text">
141
Chris Lattnerdaeb3232008-02-12 06:29:45 +0000[diff] [blame]142<p>LLVM 2.3 includes several major new capabilities:</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]143
144<ul>
Chris Lattnerf65224f2008-06-08 21:34:41 +0000[diff] [blame^]145<li><p>The biggest change in LLVM 2.3 is Multiple Return Value (MRV) support.
146 MRVs allow LLVM IR to directly represent functions that return multiple
147 values without having to pass them "by reference" in the LLVM IR. This
148 allows a front-end to generate more efficient code, as MRVs are generally
149 returned in registers if a target supports them. See the <a
150 href="LangRef.html#i_getresult">LLVM IR Reference</a> for more details.</p>
151
152 <p>MRVs are fully supported in the LLVM IR, but are not yet fully supported in
153 on all targets. However, it is generally safe to return up to 2 values from
154 a function: most targets should be able to handle at least that. MRV
155 support is a critical requirement for X86-64 ABI support, as X86-64 requires
156 the ability to return multiple registers from functions, and we use MRVs to
157 accomplish this in a direct way.</p></li>
Chris Lattnerba948522008-06-05 06:57:39 +0000[diff] [blame]158
159
Chris Lattnerf65224f2008-06-08 21:34:41 +0000[diff] [blame^]160<li><p>LLVM 2.3 includes a complete reimplementation of the "<tt>llvmc</tt>"
161 tool. It is designed to overcome several problems with the original
162 <tt>llvmc</tt> and to provide a superset of the features of the
163 '<tt>gcc</tt>' driver.</p>
Chris Lattnerba948522008-06-05 06:57:39 +0000[diff] [blame]164
Chris Lattnerf65224f2008-06-08 21:34:41 +0000[diff] [blame^]165<p>The main features of <tt>llvmc2</tt> are:
166 <ul>
167 <li>Extended handling of command line options and smart rules for
168 dispatching them to different tools.</li>
169 <li>Flexible (and extensible) rules for defining different tools.</li>
170 <li>The different intermediate steps performed by tools are represented
171 as edges in the abstract graph.</li>
172 <li>The 'language' for driver behavior definition is tablegen and thus
173 it's relatively easy to add new features.</li>
174 <li>The definition of driver is transformed into set of C++ classes, thus
175 no runtime interpretation is needed.</li>
176 </ul></p>
177 </li>
Chris Lattnerba948522008-06-05 06:57:39 +0000[diff] [blame]178
Gabor Greif5e75e652008-06-05 18:39:01 +0000[diff] [blame]179
Chris Lattnerf65224f2008-06-08 21:34:41 +0000[diff] [blame^]180<li><p>LLVM 2.3 includes a completely rewritten interface for <a
181 href="LinkTimeOptimization.html">Link Time Optimization</a>. This interface
182 is written in C, which allows for easier integration with C code bases, and
183 incorporates improvements we learned about from the first incarnation of the
184 interface.</p></li>
Chris Lattnerba948522008-06-05 06:57:39 +0000[diff] [blame]185
Chris Lattnerf65224f2008-06-08 21:34:41 +0000[diff] [blame^]186<li><p>The <a href="tutorial/LangImpl1.html">Kaleidoscope tutorial</a> now
187 includes a "port" of the tutorial that <a
188 href="tutorial/OCamlLangImpl1.html">uses the Ocaml bindings</a> to implement
189 the Kaleidoscope language.</p></li>
Chris Lattnerba948522008-06-05 06:57:39 +0000[diff] [blame]190
Chris Lattner5ba16862008-02-10 08:18:42 +0000[diff] [blame]191</ul>
192
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]193</div>
194
Chris Lattnerc5a4e312008-06-05 06:25:56 +0000[diff] [blame]195
196<!--=========================================================================-->
197<div class="doc_subsection">
Chris Lattnerf65224f2008-06-08 21:34:41 +0000[diff] [blame^]198<a name="llvm-gcc">llvm-gcc 4.2 Improvements</a>
Chris Lattnerc5a4e312008-06-05 06:25:56 +0000[diff] [blame]199</div>
200
201<div class="doc_text">
202
Duncan Sandsa73561c2008-06-08 20:18:35 +0000[diff] [blame]203<p>LLVM 2.3 fully supports the llvm-gcc 4.2 front-end.</p>
Chris Lattnerc5a4e312008-06-05 06:25:56 +0000[diff] [blame]204
205<p>llvm-gcc 4.2 includes numerous fixes to better support the Objective-C
206front-end. Objective-C now works very well on Mac OS/X.</p>
207
208<p>llvm-gcc 4.2 includes many other fixes which improve conformance with the
209relevant parts of the GCC testsuite.</p>
210
Chris Lattnerc5a4e312008-06-05 06:25:56 +0000[diff] [blame]211</div>
212
213
Chris Lattnerb0d6c502007-09-21 03:54:09 +0000[diff] [blame]214<!--=========================================================================-->
215<div class="doc_subsection">
Chris Lattnerc3649452008-02-10 08:17:19 +0000[diff] [blame]216<a name="coreimprovements">LLVM Core Improvements</a>
217</div>
218
219<div class="doc_text">
220<p>New features include:
221</p>
222
Chris Lattnerc3649452008-02-10 08:17:19 +0000[diff] [blame]223
Chris Lattnerc5a4e312008-06-05 06:25:56 +0000[diff] [blame]224Common linkage?
225
Chris Lattner0eef6e42008-06-08 02:45:07 +0000[diff] [blame]226Atomic operation support, Alpha, X86, X86-64, PowerPC. "__sync_synchronize",
Anton Korobeynikov877b73b2008-06-08 10:23:46 +0000[diff] [blame]227"__sync_val_compare_and_swap", etc
Chris Lattnerc5a4e312008-06-05 06:25:56 +0000[diff] [blame]228
229<ul>
Gordon Henriksenf9b229b2008-06-05 12:51:50 +0000[diff] [blame]230<li>The C and Ocaml bindings have received additional improvements. The
231bindings now cover pass managers, several transformation passes, iteration
232over the LLVM IR, target data, and parameter attribute lists.</li>
Chris Lattnerc3649452008-02-10 08:17:19 +0000[diff] [blame]233</ul>
234
235</div>
236
237<!--=========================================================================-->
238<div class="doc_subsection">
Chris Lattner8fda4952008-02-10 07:46:44 +0000[diff] [blame]239<a name="optimizer">Optimizer Improvements</a>
240</div>
241
242<div class="doc_text">
243
Chris Lattnerc3649452008-02-10 08:17:19 +0000[diff] [blame]244<p>In addition to a huge array of bug fixes and minor performance tweaks, the
Chris Lattnerdaeb3232008-02-12 06:29:45 +0000[diff] [blame]245LLVM 2.3 optimizers support a few major enhancements:</p>
Chris Lattner8fda4952008-02-10 07:46:44 +0000[diff] [blame]246
247<ul>
248
Devang Patel67e46712008-06-05 21:44:00 +0000[diff] [blame]249<li>Loop index set splitting on by default.<p>
250This transformation hoists conditions from loop bodies and reduces loop's
251iteration space to improve performance. For example, <p>
252<pre>
Chris Lattnerabc3b352008-06-08 20:25:30 +0000[diff] [blame]253for (i = LB; i &lt; UB; ++i)
254 if (i &lt;= NV)
Devang Patel67e46712008-06-05 21:44:00 +0000[diff] [blame]255 LOOP_BODY
256</pre>
257is transformed into
258<pre>
259NUB = min(NV+1, UB)
Chris Lattnerabc3b352008-06-08 20:25:30 +0000[diff] [blame]260for (i = LB; i &lt; NUB; ++i)
Devang Patel67e46712008-06-05 21:44:00 +0000[diff] [blame]261 LOOP_BODY
262</pre>
263</li>
Chris Lattner0eef6e42008-06-08 02:45:07 +0000[diff] [blame]264
Duncan Sandsa73561c2008-06-08 20:18:35 +0000[diff] [blame]265<li>LLVM includes a new <tt>memcpy</tt> optimization pass removes
266dead <tt>memcpy</tt> calls, unneeded copies of aggregates, and performs
Chris Lattner0eef6e42008-06-08 02:45:07 +0000[diff] [blame]267return slot optimization. The LLVM optimizer now notices long sequences of
Duncan Sandsa73561c2008-06-08 20:18:35 +0000[diff] [blame]268consecutive stores and merges them into <tt>memcpy</tt>'s where profitable.</li>
Chris Lattner0eef6e42008-06-08 02:45:07 +0000[diff] [blame]269
Gabor Greif5e75e652008-06-05 18:39:01 +0000[diff] [blame]270<li>Alignment detection for vector memory references and for <tt>memcpy</tt> and
271<tt>memset</tt> is now more aggressive.</li>
Owen Anderson7f8ea062008-06-06 16:23:15 +0000[diff] [blame]272
Chris Lattner0eef6e42008-06-08 02:45:07 +0000[diff] [blame]273<li>The aggressive dead code elimination (ADCE) optimization has been rewritten
274to make it both faster and safer in the presence of code containing infinite
275loops. Some of its prior functionality has been factored out into the loop
276deletion pass, which <em>is</em> safe for infinite loops.</li>
277
278<li>Several optimizations have been sped up, leading to faster code generation
279 with the same code quality.</li>
280
281<li>The 'SimplifyLibCalls' pass, which optimizes calls to libc and libm
282 functions for C-based languages, has been rewritten to be a FunctionPass
283 instead a ModulePass. This allows it to be run more often and to be
284 included at -O1 in llvm-gcc. It was also extended to include more
Duncan Sandsa73561c2008-06-08 20:18:35 +0000[diff] [blame]285 optimizations and several corner case bugs were fixed.</li>
Chris Lattner0eef6e42008-06-08 02:45:07 +0000[diff] [blame]286
Duncan Sands8b4f3772008-06-08 19:38:43 +0000[diff] [blame]287<li>LLVM now includes a simple 'Jump Threading' pass, which attempts to simplify
Duncan Sandsa73561c2008-06-08 20:18:35 +0000[diff] [blame]288 conditional branches using information about predecessor blocks, simplifying
Chris Lattnerd95b5962008-06-08 21:19:07 +0000[diff] [blame]289 the control flow graph. This pass is pretty basic at this point, but
290 catches some important cases and provides a foundation to build on.</li>
291
292<li>Several corner case bugs which could lead to deleting volatile memory
293 accesses have been fixed.</li>
Chris Lattner0eef6e42008-06-08 02:45:07 +0000[diff] [blame]294</ul>
295
296</div>
297
298<!--=========================================================================-->
299<div class="doc_subsection">
300<a name="codegen">Code Generator Improvements</a>
301</div>
302
303<div class="doc_text">
304
305<p>We put a significant amount of work into the code generator infrastructure,
306which allows us to implement more aggressive algorithms and make it run
307faster:</p>
308
309<ul>
310<li>MemOperand in the code generator: describe me!.</li>
311
312<li>The target-independent code generator infrastructure now uses LLVM's APInt
313 class to handle integer values, which allows it to support integer types
314 larger than 64 bits. Note that support for such types is also dependent on
315 target-specific support. Use of APInt is also a step toward support for
316 non-power-of-2 integer sizes.</li>
317
Chris Lattnerd95b5962008-06-08 21:19:07 +0000[diff] [blame]318<li>LLVM 2.3 includes several compile time speedups for code with large basic
319 blocks, particular in the instruction selection phase, register allocation,
320 scheduling, and tail merging/jump threading.</li>
Chris Lattner0eef6e42008-06-08 02:45:07 +0000[diff] [blame]321
322<li>Several improvements which make llc's <tt>--view-sunit-dags</tt>
323 visualization of scheduling dependency graphs easier to understand.</li>
324
325<li>The code generator allows targets to write patterns that generate subreg
326 references directly in .td files now.</li>
327
328<li><tt>memcpy</tt> lowering in the backend is more aggressive, particularly for
329 <tt>memcpy</tt> calls introduced by the code generator when handling
330 pass-by-value structure argument copies.</li>
Chris Lattnerd95b5962008-06-08 21:19:07 +0000[diff] [blame]331
332<li>Inline assembly with multiple register results now returns those results
333 directly in the appropriate registers, rather than going through memory.
334 Inline assembly that uses constraints like "ir" with immediates now use the
335 'i' form when possible instead of always loading the value in a register.
336 This saves an instruction and reduces register use.</li>
337
338<li>Added support for PIC/GOT style tail calls on x86/32 and initial support
339 for tail calls on PowerPC 32 (it may also work on ppc64 but not
340 thoroughly tested).</li>
Chris Lattnerb0d6c502007-09-21 03:54:09 +0000[diff] [blame]341</ul>
342
343</div>
344
345
346<!--=========================================================================-->
347<div class="doc_subsection">
Chris Lattnere45f18b2008-06-05 08:02:49 +0000[diff] [blame]348<a name="x86specific">X86/X86-64 Specific Improvements</a>
Chris Lattnerb0d6c502007-09-21 03:54:09 +0000[diff] [blame]349</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]350
351<div class="doc_text">
Chris Lattnerc3649452008-02-10 08:17:19 +0000[diff] [blame]352<p>New target-specific features include:
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]353</p>
354
355<ul>
Chris Lattnerc5a4e312008-06-05 06:25:56 +0000[diff] [blame]356<li>llvm-gcc's X86-64 ABI conformance is far improved, particularly in the
Gabor Greif5e75e652008-06-05 18:39:01 +0000[diff] [blame]357 area of passing and returning structures by value. llvm-gcc compiled code
Chris Lattnerc5a4e312008-06-05 06:25:56 +0000[diff] [blame]358 now interoperates very well on X86-64 systems with other compilers.</li>
359
Duncan Sands8b4f3772008-06-08 19:38:43 +0000[diff] [blame]360<li>Support for Win64 was added. This includes code generation itself, JIT
Anton Korobeynikov77d15e22008-06-08 10:24:13 +0000[diff] [blame]361 support and necessary changes to llvm-gcc.</li>
362
Chris Lattnerc5a4e312008-06-05 06:25:56 +0000[diff] [blame]363<li>The LLVM X86 backend now supports the support SSE 4.1 instruction set, and
Chris Lattner0eef6e42008-06-08 02:45:07 +0000[diff] [blame]364 the llvm-gcc 4.2 front-end supports the SSE 4.1 compiler builtins. Various
365 generic vector operations (insert/extract/shuffle) are much more efficient
366 when SSE 4.1 is enabled. The JIT automatically takes advantage of these
367 instructions, but llvm-gcc must be explicitly told to use them, e.g. with
368 <tt>-march=penryn</tt>.</li>
Anton Korobeynikov77d15e22008-06-08 10:24:13 +0000[diff] [blame]369
Chris Lattnerc5a4e312008-06-05 06:25:56 +0000[diff] [blame]370<li>The X86 backend now does a number of optimizations that aim to avoid
371 converting numbers back and forth from SSE registers to the X87 floating
372 point stack.</li>
Anton Korobeynikov77d15e22008-06-08 10:24:13 +0000[diff] [blame]373
Chris Lattnerba948522008-06-05 06:57:39 +0000[diff] [blame]374<li>The X86 backend supports stack realignment, which is particularly useful for
375 vector code on OS's without 16-byte aligned stacks.</li>
Chris Lattnerc5a4e312008-06-05 06:25:56 +0000[diff] [blame]376
377<li>The X86 backend now supports the "sseregparm" options in GCC, which allow
378 functions to be tagged as passing floating point values in SSE
379 registers.</li>
380
381<li>Trampolines (taking the address of a nested function) now work on
Duncan Sandsa73561c2008-06-08 20:18:35 +0000[diff] [blame]382 Linux/X86-64.</li>
Anton Korobeynikov77d15e22008-06-08 10:24:13 +0000[diff] [blame]383
Chris Lattnere45f18b2008-06-05 08:02:49 +0000[diff] [blame]384<li><tt>__builtin_prefetch</tt> is now compiled into the appropriate prefetch
385 instructions instead of being ignored.</li>
Chris Lattnerc5a4e312008-06-05 06:25:56 +0000[diff] [blame]386
Duncan Sandsa73561c2008-06-08 20:18:35 +0000[diff] [blame]387<li>128-bit integers are now supported on X86-64 targets.</li>
Dan Gohman74e1ff52008-06-05 16:15:39 +0000[diff] [blame]388
Chris Lattner0eef6e42008-06-08 02:45:07 +0000[diff] [blame]389<li>The register allocator can now rematerialize PIC-base computations.</li>
390
Chris Lattnerd95b5962008-06-08 21:19:07 +0000[diff] [blame]391<li>The "t" and "f" inline assembly constraints for the X87 floating point stack
392 now work. However, the "u" constraint is still not fully supported.</li>
393
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]394</ul>
Chris Lattnerb0d6c502007-09-21 03:54:09 +0000[diff] [blame]395
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]396</div>
397
Chris Lattnerb0d6c502007-09-21 03:54:09 +0000[diff] [blame]398<!--=========================================================================-->
399<div class="doc_subsection">
Chris Lattnere45f18b2008-06-05 08:02:49 +0000[diff] [blame]400<a name="targetspecific">Other Target Specific Improvements</a>
401</div>
402
403<div class="doc_text">
404<p>New target-specific features include:
405</p>
406
407<ul>
Gabor Greif5e75e652008-06-05 18:39:01 +0000[diff] [blame]408<li>The LLVM C backend now supports vector code.</li>
Chris Lattnere45f18b2008-06-05 08:02:49 +0000[diff] [blame]409</ul>
410
411</div>
412
413
414
415<!--=========================================================================-->
416<div class="doc_subsection">
Chris Lattnerb0d6c502007-09-21 03:54:09 +0000[diff] [blame]417<a name="otherimprovements">Other Improvements</a>
418</div>
419
420<div class="doc_text">
421<p>New features include:
422</p>
423
424<ul>
Chris Lattnerba948522008-06-05 06:57:39 +0000[diff] [blame]425<li>LLVM now builds with GCC 4.3.</li>
Chris Lattnerd95b5962008-06-08 21:19:07 +0000[diff] [blame]426<li>Bugpoint now supports running custom scripts (with the <tt>-run-custom</tt>
427 option) to determine how to execute the command and whether it is making
428 forward process.</li>
Chris Lattnerb0d6c502007-09-21 03:54:09 +0000[diff] [blame]429</ul>
430
431</div>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]432
433<!-- *********************************************************************** -->
434<div class="doc_section">
435 <a name="portability">Portability and Supported Platforms</a>
436</div>
437<!-- *********************************************************************** -->
438
439<div class="doc_text">
440
441<p>LLVM is known to work on the following platforms:</p>
442
443<ul>
Gabor Greif5e75e652008-06-05 18:39:01 +0000[diff] [blame]444<li>Intel and AMD machines (IA32) running Red Hat Linux, Fedora Core and FreeBSD
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]445 (and probably other unix-like systems).</li>
Chris Lattnerbd633de92008-02-06 06:30:34 +0000[diff] [blame]446<li>PowerPC and X86-based Mac OS X systems, running 10.3 and above in 32-bit and
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]447 64-bit modes.</li>
Chris Lattnere0798082008-02-06 18:00:06 +0000[diff] [blame]448<li>Intel and AMD machines running on Win32 using MinGW libraries (native).</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]449<li>Intel and AMD machines running on Win32 with the Cygwin libraries (limited
450 support is available for native builds with Visual C++).</li>
Gabor Greif5e75e652008-06-05 18:39:01 +0000[diff] [blame]451<li>Sun UltraSPARC workstations running Solaris 10.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]452<li>Alpha-based machines running Debian GNU/Linux.</li>
Gabor Greif5e75e652008-06-05 18:39:01 +0000[diff] [blame]453<li>Itanium-based (IA64) machines running Linux and HP-UX.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]454</ul>
455
Chris Lattnerba948522008-06-05 06:57:39 +0000[diff] [blame]456<p>The core LLVM infrastructure uses GNU autoconf to adapt itself
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]457to the machine and operating system on which it is built. However, minor
458porting may be required to get LLVM to work on new platforms. We welcome your
459portability patches and reports of successful builds or error messages.</p>
460
461</div>
462
463<!-- *********************************************************************** -->
464<div class="doc_section">
465 <a name="knownproblems">Known Problems</a>
466</div>
467<!-- *********************************************************************** -->
468
469<div class="doc_text">
470
471<p>This section contains all known problems with the LLVM system, listed by
472component. As new problems are discovered, they will be added to these
473sections. If you run into a problem, please check the <a
474href="http://llvm.org/bugs/">LLVM bug database</a> and submit a bug if
475there isn't already one.</p>
476
477</div>
478
479<!-- ======================================================================= -->
480<div class="doc_subsection">
481 <a name="experimental">Experimental features included with this release</a>
482</div>
483
484<div class="doc_text">
485
486<p>The following components of this LLVM release are either untested, known to
487be broken or unreliable, or are in early development. These components should
488not be relied on, and bugs should not be filed against them, but they may be
489useful 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]490components, please contact us on the <a
491href="http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev">LLVMdev list</a>.</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]492
493<ul>
Chris Lattnerdaeb3232008-02-12 06:29:45 +0000[diff] [blame]494<li>The MSIL, IA64, Alpha, SPU, and MIPS backends are experimental.</li>
Duncan Sandsa73561c2008-06-08 20:18:35 +0000[diff] [blame]495<li>The llc "<tt>-filetype=asm</tt>" (the default) is the only supported
Chris Lattnerbd633de92008-02-06 06:30:34 +0000[diff] [blame]496 value for this option.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]497</ul>
498
499</div>
500
501<!-- ======================================================================= -->
502<div class="doc_subsection">
503 <a name="x86-be">Known problems with the X86 back-end</a>
504</div>
505
506<div class="doc_text">
507
508<ul>
Anton Korobeynikov77d15e22008-06-08 10:24:13 +0000[diff] [blame]509 <li>The X86 backend does not yet support
510 all <a href="http://llvm.org/PR879">inline assembly that uses the X86
511 floating point stack</a>. It supports the 'f' and 't' constraints, but not
512 'u'.</li>
Chris Lattnerd95b5962008-06-08 21:19:07 +0000[diff] [blame]513 <li>The X86 backend generates inefficient floating point code when configured
514 to generate code for systems that don't have SSE2.</li>
Duncan Sands8b4f3772008-06-08 19:38:43 +0000[diff] [blame]515 <li>Win64 code generation wasn't widely tested. Everything should work, but we
Anton Korobeynikov77d15e22008-06-08 10:24:13 +0000[diff] [blame]516 expect small issues to happen. Also, llvm-gcc cannot build mingw64 runtime
517 currently due
518 to <a href="http://llvm.org/PR2255">several</a>
519 <a href="http://llvm.org/PR2257">bugs</a> in FP stackifier
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]520</ul>
521
522</div>
523
524<!-- ======================================================================= -->
525<div class="doc_subsection">
526 <a name="ppc-be">Known problems with the PowerPC back-end</a>
527</div>
528
529<div class="doc_text">
530
531<ul>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]532<li>The Linux PPC32/ABI support needs testing for the interpreter and static
533compilation, and lacks support for debug information.</li>
534</ul>
535
536</div>
537
538<!-- ======================================================================= -->
539<div class="doc_subsection">
540 <a name="arm-be">Known problems with the ARM back-end</a>
541</div>
542
543<div class="doc_text">
544
545<ul>
546<li>Thumb mode works only on ARMv6 or higher processors. On sub-ARMv6
Duncan Sands1827fa62007-09-26 15:59:54 +0000[diff] [blame]547processors, thumb programs can crash or produce wrong
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]548results (<a href="http://llvm.org/PR1388">PR1388</a>).</li>
549<li>Compilation for ARM Linux OABI (old ABI) is supported, but not fully tested.
550</li>
Chris Lattnerd95b5962008-06-08 21:19:07 +0000[diff] [blame]551<li>There is a bug in QEMU-ARM (&lt;= 0.9.0) which causes it to incorrectly
552 execute
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]553programs compiled with LLVM. Please use more recent versions of QEMU.</li>
554</ul>
555
556</div>
557
558<!-- ======================================================================= -->
559<div class="doc_subsection">
560 <a name="sparc-be">Known problems with the SPARC back-end</a>
561</div>
562
563<div class="doc_text">
564
565<ul>
566<li>The SPARC backend only supports the 32-bit SPARC ABI (-m32), it does not
567 support the 64-bit SPARC ABI (-m64).</li>
568</ul>
569
570</div>
571
572<!-- ======================================================================= -->
573<div class="doc_subsection">
574 <a name="alpha-be">Known problems with the Alpha back-end</a>
575</div>
576
577<div class="doc_text">
578
579<ul>
580
581<li>On 21164s, some rare FP arithmetic sequences which may trap do not have the
582appropriate nops inserted to ensure restartability.</li>
583
584</ul>
585</div>
586
587<!-- ======================================================================= -->
588<div class="doc_subsection">
589 <a name="ia64-be">Known problems with the IA64 back-end</a>
590</div>
591
592<div class="doc_text">
593
594<ul>
595
596<li>C++ programs are likely to fail on IA64, as calls to <tt>setjmp</tt> are
597made where the argument is not 16-byte aligned, as required on IA64. (Strictly
598speaking this is not a bug in the IA64 back-end; it will also be encountered
599when building C++ programs using the C back-end.)</li>
600
601<li>The C++ front-end does not use <a href="http://llvm.org/PR406">IA64
602ABI compliant layout of v-tables</a>. In particular, it just stores function
603pointers instead of function descriptors in the vtable. This bug prevents
604mixing C++ code compiled with LLVM with C++ objects compiled by other C++
605compilers.</li>
606
607<li>There are a few ABI violations which will lead to problems when mixing LLVM
608output with code built with other compilers, particularly for floating-point
609programs.</li>
610
Duncan Sands8b4f3772008-06-08 19:38:43 +0000[diff] [blame]611<li>Defining vararg functions is not supported (but calling them is OK).</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]612
613<li>The Itanium backend has bitrotted somewhat.</li>
614</ul>
615
616</div>
617
618<!-- ======================================================================= -->
619<div class="doc_subsection">
620 <a name="c-be">Known problems with the C back-end</a>
621</div>
622
623<div class="doc_text">
624
625<ul>
Chris Lattner499c07c2008-06-05 06:35:40 +0000[diff] [blame]626<li><a href="http://llvm.org/PR802">The C backend has only basic support for
627 inline assembly code</a>.</li>
Chris Lattnere0194462007-09-26 06:01:35 +0000[diff] [blame]628<li><a href="http://llvm.org/PR1658">The C backend violates the ABI of common
629 C++ programs</a>, preventing intermixing between C++ compiled by the CBE and
Duncan Sandsa73561c2008-06-08 20:18:35 +0000[diff] [blame]630 C++ code compiled with llc or native compilers.</li>
Duncan Sandsbd0ad6e2008-02-10 13:40:55 +0000[diff] [blame]631<li>The C backend does not support all exception handling constructs.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]632</ul>
633
634</div>
635
636
637<!-- ======================================================================= -->
638<div class="doc_subsection">
Chris Lattner499c07c2008-06-05 06:35:40 +0000[diff] [blame]639 <a name="c-fe">Known problems with the llvm-gcc C front-end</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]640</div>
641
642<!-- _______________________________________________________________________ -->
643<div class="doc_subsubsection">Bugs</div>
644
645<div class="doc_text">
646
Chris Lattnere0798082008-02-06 18:00:06 +0000[diff] [blame]647<p>llvm-gcc does not currently support <a href="http://llvm.org/PR869">Link-Time
648Optimization</a> on most platforms "out-of-the-box". Please inquire on the
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]649llvmdev mailing list if you are interested.</p>
650
Chris Lattner499c07c2008-06-05 06:35:40 +0000[diff] [blame]651<p>The only major language feature of GCC not supported by llvm-gcc is
652 the <tt>__builtin_apply</tt> family of builtins. However, some extensions
653 are only supported on some targets. For example, trampolines are only
Duncan Sandsa73561c2008-06-08 20:18:35 +0000[diff] [blame]654 supported on some targets (these are used when you take the address of a
655 nested function).</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]656
Chris Lattner499c07c2008-06-05 06:35:40 +0000[diff] [blame]657<p>If you run into GCC extensions which are not supported, please let us know.
658</p>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]659
660</div>
661
662<!-- ======================================================================= -->
663<div class="doc_subsection">
Chris Lattner499c07c2008-06-05 06:35:40 +0000[diff] [blame]664 <a name="c++-fe">Known problems with the llvm-gcc C++ front-end</a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]665</div>
666
667<div class="doc_text">
668
669<p>The C++ front-end is considered to be fully
670tested and works for a number of non-trivial programs, including LLVM
671itself, Qt, Mozilla, etc.</p>
672
673<ul>
Chris Lattner499c07c2008-06-05 06:35:40 +0000[diff] [blame]674<li>Exception handling works well on the X86 and PowerPC targets, including
Duncan Sandsa73561c2008-06-08 20:18:35 +0000[diff] [blame]675X86-64 darwin. This works when linking to a libstdc++ compiled by GCC. It is
676supported on X86-64 linux, but that is disabled by default in this release.</li>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]677</ul>
678
679</div>
680
681
Chris Lattnerdaeb3232008-02-12 06:29:45 +0000[diff] [blame]682<!-- ======================================================================= -->
683<div class="doc_subsection">
Chris Lattner499c07c2008-06-05 06:35:40 +0000[diff] [blame]684 <a name="ada-fe">Known problems with the llvm-gcc Ada front-end</a>
Chris Lattnerdaeb3232008-02-12 06:29:45 +0000[diff] [blame]685</div>
686
687<div class="doc_text">
688The llvm-gcc 4.2 Ada compiler works fairly well, however this is not a mature
689technology and problems should be expected.
690<ul>
Duncan Sandsa73561c2008-06-08 20:18:35 +0000[diff] [blame]691<li>The Ada front-end currently only builds on X86-32. This is mainly due
Chris Lattnerdaeb3232008-02-12 06:29:45 +0000[diff] [blame]692to lack of trampoline support (pointers to nested functions) on other platforms,
Duncan Sandsa73561c2008-06-08 20:18:35 +0000[diff] [blame]693however 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]694which does support trampolines.</li>
695<li>The Ada front-end <a href="http://llvm.org/PR2007">fails to bootstrap</a>.
696Workaround: configure with --disable-bootstrap.</li>
697<li>The c380004 and <a href="http://llvm.org/PR2010">c393010</a> ACATS tests
Duncan Sandsa73561c2008-06-08 20:18:35 +0000[diff] [blame]698fail (c380004 also fails with gcc-4.2 mainline). When built at -O3, the
699<a href="http://llvm.org/PR2421">cxg2021</a> ACATS test also fails.</li>
700<li>Some gcc specific Ada tests continue to crash the compiler. The testsuite
701reports most tests as having failed even though they pass.</li>
Chris Lattnerdaeb3232008-02-12 06:29:45 +0000[diff] [blame]702<li>The -E binder option (exception backtraces)
703<a href="http://llvm.org/PR1982">does not work</a> and will result in programs
704crashing if an exception is raised. Workaround: do not use -E.</li>
705<li>Only discrete types <a href="http://llvm.org/PR1981">are allowed to start
706or finish at a non-byte offset</a> in a record. Workaround: do not pack records
707or use representation clauses that result in a field of a non-discrete type
708starting or finishing in the middle of a byte.</li>
Chris Lattnerd95b5962008-06-08 21:19:07 +0000[diff] [blame]709<li>The <tt>lli</tt> interpreter <a href="http://llvm.org/PR2009">considers
710'main' as generated by the Ada binder to be invalid</a>.
711Workaround: hand edit the file to use pointers for <tt>argv</tt> and
712<tt>envp</tt> rather than integers.</li>
713<li>The <tt>-fstack-check</tt> option <a href="http://llvm.org/PR2008">is
714ignored</a>.</li>
Chris Lattnerdaeb3232008-02-12 06:29:45 +0000[diff] [blame]715</ul>
716</div>
717
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]718<!-- *********************************************************************** -->
719<div class="doc_section">
720 <a name="additionalinfo">Additional Information</a>
721</div>
722<!-- *********************************************************************** -->
723
724<div class="doc_text">
725
726<p>A wide variety of additional information is available on the <a
727href="http://llvm.org">LLVM web page</a>, in particular in the <a
728href="http://llvm.org/docs/">documentation</a> section. The web page also
729contains versions of the API documentation which is up-to-date with the
730Subversion version of the source code.
731You can access versions of these documents specific to this release by going
732into the "<tt>llvm/doc/</tt>" directory in the LLVM tree.</p>
733
734<p>If you have any questions or comments about LLVM, please feel free to contact
735us via the <a href="http://llvm.org/docs/#maillist"> mailing
736lists</a>.</p>
737
738</div>
739
740<!-- *********************************************************************** -->
741
742<hr>
743<address>
744 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
745 src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a>
746 <a href="http://validator.w3.org/check/referer"><img
Chris Lattner5ba16862008-02-10 08:18:42 +0000[diff] [blame]747 src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!"></a>
Dan Gohmanf17a25c2007-07-18 16:29:46 +0000[diff] [blame]748
749 <a href="http://llvm.org/">LLVM Compiler Infrastructure</a><br>
750 Last modified: $Date$
751</address>
752
753</body>
754</html>