blob: 75048cbe6c06a7123dd51298e2a424c7b129f865 [file] [log] [blame]
Guochun Shif4688a82002-07-17 23:05:56 +00001<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
2<html>
John Criswell85ed3612003-06-12 19:34:44 +00003 <head>
4 <title>Getting Started with LLVM System</title>
5 </head>
6
7 <body bgcolor=white>
8 <center><h1>Getting Started with the LLVM System<br><font size=3>By: <a
9 href="mailto:gshi1@uiuc.edu">Guochun Shi</a>,
John Criswellcb27f1e2003-07-01 21:55:57 +000010 <a href="mailto:sabre@nondot.org">Chris Lattner</a>,
11 <a href="mailto:criswell@uiuc.edu">John Criswell</a>, and
John Criswell85ed3612003-06-12 19:34:44 +000012 <a href="http://www.cs.uiuc.edu/~vadve">Vikram Adve</a>
13 </font></h1></center>
14
15 <!--=====================================================================-->
16 <h2><a name="Contents">Contents</a></h2>
17 <!--=====================================================================-->
18
19 <ul>
20 <li><a href="#overview">Overview</a>
John Criswellce760f62003-07-03 16:01:38 +000021 <ol>
John Criswell7a73b802003-06-30 21:59:07 +000022 <li><a href="#requirements">Requirements</a>
23 <ol>
24 <li><a href="#hardware">Hardware</a>
25 <li><a href="#software">Software</a>
26 </ol>
John Criswellce760f62003-07-03 16:01:38 +000027 </ol>
28 <li><a href="#starting">Getting started with LLVM</a>
29 <ol>
John Criswell85ed3612003-06-12 19:34:44 +000030 <li><a href="#quickstart">Getting started quickly (a summary)</a>
John Criswell85ed3612003-06-12 19:34:44 +000031 <li><a href="#terminology">Terminology and Notation</tt></a>
John Criswell85ed3612003-06-12 19:34:44 +000032 <li><a href="#environment">Setting up your environment</a>
John Criswell312a68c2003-07-07 19:27:35 +000033 <li><a href="#unpack">Unpacking the LLVM Archives</a>
John Criswellce760f62003-07-03 16:01:38 +000034 <li><a href="#checkout">Checkout LLVM from CVS</a>
35 <li><a href="#config">Local LLVM Configuration</tt></a>
36 <li><a href="#compile">Compiling the LLVM Suite Source Code</a>
37 <li><a href="#build">Building the LLVM C Front End</a>
38 <li><a href="#objfiles">The location for object files</tt></a>
John Criswell85ed3612003-06-12 19:34:44 +000039 </ol>
40 <li><a href="#layout">Program layout</a>
41 <ol>
42 <li><a href="#cvsdir">CVS directories</a>
John Criswell85ed3612003-06-12 19:34:44 +000043 <li><a href="#include"><tt>llvm/include</tt></a>
44 <li><a href="#lib"><tt>llvm/lib</tt></a>
45 <li><a href="#test"><tt>llvm/test</tt></a>
46 <li><a href="#tools"><tt>llvm/tools</tt></a>
47 </ol>
48 <li><a href="#tutorial">An example using the LLVM tool chain</a>
49 <li><a href="#links">Links</a>
50 </ul>
51
52
53 <!--=====================================================================-->
54 <center>
55 <h2><a name="overview"><b>Overview</b></a></h2>
56 </center>
57 <!--=====================================================================-->
58
John Criswellce760f62003-07-03 16:01:38 +000059 Welcome to LLVM! In order to get started, you first need to know some
60 basic information.
61
62 <p>
63 First, LLVM comes in two pieces. The first piece is the LLVM suite. This
64 contains all of the tools, libraries, and header files needed to use the
65 low level virtual machine. It also contains a test suite that can be used
66 to test the LLVM tools and the GCC front end.
67 <p>
68 The second piece is the GCC front end. This component provides a version
69 of GCC that compiles C code into LLVM bytecode. Currently, the C front end
70 is a modified version of GCC 3.4 (we track the GCC 3.4 development).
71 Once compiled into LLVM bytecode, a program can be manipulated with the
72 LLVM tools.
73
74 <!--=====================================================================-->
75 <h3><a name="requirements"><b>Requirements</b></a></h3>
76 <!--=====================================================================-->
77
78 Before you begin to use the LLVM system, review the requirements given
79 below. This may save you some trouble by knowing ahead of time what
80 hardware and software you will need.
81
82 <!--=====================================================================-->
83 <h4><a name="hardware"><b>Hardware</b></a></h4>
84 <!--=====================================================================-->
85 LLVM is known to work on the following platforms:
86 <ul>
87 <li> Linux on x86
88 <ul>
89 <li> Approximately 700 MB of Free Disk Space
90 <ul>
91 <li>Source code: 30 MB
92 <li>Object code: 670 MB
93 </ul>
94 </ul>
95 <li> Solaris on SparcV9 (Ultrasparc)
96 <ul>
97 <li> Approximately 1.03 GB of Free Disk Space
98 <ul>
99 <li>Source code: 30 MB
100 <li>Object code: 1000 MB
101 </ul>
102 </ul>
103 </ul>
104
105 LLVM <i>may</i> compile on other platforms. The LLVM utilities should work
106 on other platforms, so it should be possible to generate and produce LLVM
107 bytecode on unsupported platforms (although bytecode generated on one
108 platform may not work on another platform). However, the code generators
Misha Brukmanc56e5822003-07-03 16:29:36 +0000109 and Just-In-Time (JIT) compilers only generate SparcV9 or x86 machine code.
John Criswellce760f62003-07-03 16:01:38 +0000110
111 <!--=====================================================================-->
112 <h4><a name="software"><b>Software</b></a></h4>
113 <!--=====================================================================-->
114 <p>
John Criswell6dd90d22003-07-03 16:43:01 +0000115
116 Unpacking the distribution requires the following tools:
117 <dl compact>
118 <dt>GNU Zip (gzip)
119 <dt>GNU Tar
120 <dd>
121 These tools are needed to uncompress and unarchive the software.
122 Regular Solaris <tt>tar</tt> may work for unpacking the TAR archive but
123 is untested.
124 </dl>
125
John Criswellce760f62003-07-03 16:01:38 +0000126 Compiling LLVM requires that you have several different software packages
127 installed:
128
129 <dl compact>
130 <dt> GCC
131 <dd>
132 The GNU Compiler Collection must be installed with C and C++ language
133 support. GCC 3.2.x works, and GCC 3.x is generally supported.
134
135 <p>
136 Note that we currently do not support any other C++ compiler.
137 </p>
138
139 <dt> GNU Make
140 <dd>
141 The LLVM build system relies upon GNU Make extensions. Therefore, you
142 will need GNU Make (sometimes known as gmake) to build LLVM.
143 <p>
144
145 <dt> Flex and Bison
146 <dd>
147 The LLVM source code is built using flex and bison. You will not be
148 able to configure and compile LLVM without them.
149 <p>
150
151 <dt> GNU M4
152 <dd>
153 If you are installing Bison on your machine for the first time, you
154 will need GNU M4 (version 1.4 or higher).
155 </dl>
156
157 <p>
158 There are some additional tools that you may want to have when working with
159 LLVM:
160 </p>
161
162 <ul>
163 <li>GNU Autoconf
164 <li>GNU M4
165 <p>
166 If you want to make changes to the configure scripts, you will need
167 GNU autoconf (2.53 or higher), and consequently, GNU M4 (version 1.4
168 or higher).
169 </p>
170 </ul>
171
172
173 <p>The <a href="starting">next section</a> of this guide is meant to get
John Criswell85ed3612003-06-12 19:34:44 +0000174 you up and running with LLVM and to give you some basic information about
175 the LLVM environment. The <a href"#quickstart">first subsection</a> gives
176 a short summary for those who are already familiar with the system and
177 want to get started as quickly as possible.
178
179 <p>The later sections of this guide describe the <a
John Criswellcb27f1e2003-07-01 21:55:57 +0000180 href="#layout">general layout</a> of the the LLVM source-tree, a <a
John Criswell85ed3612003-06-12 19:34:44 +0000181 href="#tutorial">simple example</a> using the LLVM tool chain, and <a
182 href="#links">links</a> to find more information about LLVM or to get
183 help via e-mail.
184
185 <!--=====================================================================-->
186 <center>
187 <h2><a name="starting"><b>Getting Started</b></a></h2>
188 </center>
189 <!--=====================================================================-->
190
John Criswell7a73b802003-06-30 21:59:07 +0000191 <!--=====================================================================-->
John Criswell85ed3612003-06-12 19:34:44 +0000192 <h3><a name="quickstart"><b>Getting Started Quickly (A Summary)</b></a></h3>
193 <!--=====================================================================-->
194
195 Here's the short story for getting up and running quickly with LLVM:
Chris Lattner96768ea2003-02-14 04:22:13 +0000196 <ol>
John Criswellce760f62003-07-03 16:01:38 +0000197 <li>Build the LLVM suite
198 <ol>
199 <li>Find the path to the CVS repository containing LLVM (we'll call this <i>CVSROOTDIR</i>).
200 <li><tt>cd <i>where-you-want-llvm-to-live</i></tt>
201 <li><tt>cvs -d <i>CVSROOTDIR</i> checkout llvm</tt>
202 <li><tt>cd llvm</tt>
203 <li>Run <tt>configure</tt> to configure the Makefiles and header files.
204 Useful options include:
205 <ul>
206 <li><tt>--with-objroot=<i>directory</i></tt>
207 <br>
208 Specify where object files should be placed during the build.
John Criswell7a73b802003-06-30 21:59:07 +0000209
John Criswellce760f62003-07-03 16:01:38 +0000210 <li><tt>--with-llvmgccdir=<i>directory</i></tt>
211 <br>
212 Specify where the LLVM C frontend is going to be installed.
213 </ul>
214 <li>Set your LLVM_LIB_SEARCH_PATH environment variable.
215 <li><tt>gmake -k |& tee gnumake.out
216 &nbsp;&nbsp;&nbsp;# this is csh or tcsh syntax</tt>
217 </ol>
218 <p>
John Criswell6e0a6672003-07-03 16:03:39 +0000219 <li>Build the LLVM C Front End <b>(optional)</b>
John Criswellce760f62003-07-03 16:01:38 +0000220 <ol>
221 <li>Create a directory for the object files to live.
222 <li><tt>cd <i>object file directory</i></tt>
223 <li>Run <tt><i>Pathname-to-where-the-source-code-lives</i>/configure --prefix=<i>LLVMGCCDIR</i></tt> to configure GCC.
224 <li><tt>make bootstrap</tt>
225 <li><tt>make install</tt>
226 </ol>
Chris Lattner96768ea2003-02-14 04:22:13 +0000227 </ol>
John Criswell85ed3612003-06-12 19:34:44 +0000228
229 <p>See <a href="#environment">Setting up your environment</a> on tips to
230 simplify working with the LLVM front-end and compiled tools. See the
231 other sub-sections below for other useful details in working with LLVM,
232 or go straight to <a href="#layout">Program Layout</a> to learn about the
233 layout of the source code tree.
234
235 <!------------------------------------------------------------------------->
236 <h3><a name="terminology">Terminology and Notation</a></h3>
237 <!------------------------------------------------------------------------->
238
239 <p>Throughout this manual, the following names are used to denote paths
240 specific to the local system and working environment. <i>These are not
John Criswell7a73b802003-06-30 21:59:07 +0000241 environment variables you need to set but just strings used in the rest
242 of this document below</i>. In any of the examples below, simply replace
John Criswell85ed3612003-06-12 19:34:44 +0000243 each of these names with the appropriate pathname on your local system.
Chris Lattner1db872d2002-09-06 16:26:13 +0000244 All these paths are absolute:</p>
John Criswellce760f62003-07-03 16:01:38 +0000245 <dl compact>
246 <dt>CVSROOTDIR
247 <dd>
248 This is the path for the CVS repository containing the LLVM source
249 code. Ask the person responsible for your local LLVM installation to
250 give you this path.
251 <p>
252
253 <dt>OBJ_ROOT
254 <dd>
255 This is the top level directory for where the LLVM suite object files
256 will be placed during the build.
257 <p>
258
259 <dt>LLVMGCCDIR
260 <dd>
261 This is the pathname to the location where the LLVM C Front End will
John Criswell6dd90d22003-07-03 16:43:01 +0000262 be installed. Note that the C front end does not need to be installed
John Criswellce760f62003-07-03 16:01:38 +0000263 during the LLVM suite build; you will just need to know where it will
264 go for configuring the build system and running the test suite later.
265 <p>
John Criswella27028b2003-07-03 16:49:40 +0000266 For the pre-built C front end binaries, the LLVMGCCDIR is
267 <tt>cfrontend/<i>platform</i>/llvm-gcc</tt>.
John Criswellce760f62003-07-03 16:01:38 +0000268
269 <dt>GCCSRC
270 <dd>
271 This is the pathname of the directory where the LLVM C front end source
272 code can be found.
273 <p>
274
275 <dt>GCCOBJ
276 <dd>
277 This is the pathname of the directory where the LLVM C front end object
278 code will be placed during the build. It can be safely removed once
279 the build is complete.
280 </dl>
281
282 <!------------------------------------------------------------------------->
283 <h3><a name="environment">Setting up your environment</a></h3>
284 <!------------------------------------------------------------------------->
285
286 <p>
287 In order to compile and use LLVM, you will need to set some environment
288 variables. There are also some shell aliases which you may find useful.
289 You can set these on the command line, or better yet, set them in your
290 <tt>.cshrc</tt> or <tt>.profile</tt>.
291
292 <dl compact>
293 <dt><tt>LLVM_LIB_SEARCH_PATH</tt>=<tt><i>LLVMGCCDIR</i>/llvm-gcc/bytecode-libs</tt>
294 <dd>
295 This environment variable helps the LLVM C front end find bytecode
296 libraries that it will need for compilation.
297 <p>
298
299 <dt><tt>PATH</tt>=<tt>${PATH}:<i>OBJ_ROOT</i>/llvm/tools/Debug</tt>
300 <dd>
301 Adding this directory to the end of your path will allow the
302 compilation of the C front end to find the LLVM tools. The LLVM tools
303 are needed for the C front end compile.
304 <p>
305
306 <dt><tt>CC</tt>=<i>Pathname to your GCC compiler</i>
307 <dd>
308 The GCC compiler that you want to use must be the first C compiler in
309 your <tt>PATH</tt>. Otherwise, set this variable so that
310 <tt>configure</tt> will use the GCC compiler that you want to use.
311 <p>
312
313 <dt><tt>CXX</tt>=<i>Pathname to your GCC C++ compiler</i>
314 <dd>
315 The GCC compiler that you want to use must be the first C++ compiler in
316 your <tt>PATH</tt>. Otherwise, set this variable so that
317 <tt>configure</tt> will use the GCC compiler that you want to use.
318 <p>
319
320 <dt><tt>CVSROOT</tt>=<i>CVSROOT</i>
321 <dd>
322 This environment variable tells CVS where to find the CVS repository.
323 <p>
324
325 <dt>alias llvmgcc <i>LLVMGCCDIR</i><tt>/bin/llvm-gcc</tt>
326 <dd>
327 This alias allows you to use the LLVM C front end without putting it in
328 your <tt>PATH</tt> or typing in its complete pathname.
329 </dl>
John Criswell85ed3612003-06-12 19:34:44 +0000330
331 <!------------------------------------------------------------------------->
John Criswell312a68c2003-07-07 19:27:35 +0000332 <h3><a name="unpack">Unpacking the LLVM Archives</a></h3>
333 <!------------------------------------------------------------------------->
334
335 <p>
336 If you have the LLVM distribution, you will need to unpack it before you
337 can begin to compile it. LLVM is distributed as a set of four files. Each
338 file is a TAR archive that is compressed with the gzip program.
339 </p>
340
341 <p> The four files are the following:
342 <dl compact>
343 <dt>llvm.tar.gz
344 <dd>This is the source code to the LLVM suite.
345 <p>
346
347 <dt>cfrontend.sparc.tar.gz
348 <dd>This is the binary release of the C front end for Solaris/Sparc.
349 <p>
350
351 <dt>cfrontend.x86.tar.gz
352 <dd>This is the binary release of the C front end for Linux/x86.
353 <p>
354
355 <dt>cfrontend-src.tar.gz
356 <dd>This is the source code release of the C front end.
357 <p>
358 </dl>
359
360 <p>
361 To unpack the files, take each one, unzip it, and then untar it. A fast
362 way to do that is with the following:
363 </p>
364
365 <tt>gunzip --stdout <i>name of file</i> | tar -xvf -</tt>
366
367 <p>
368 For example, to extract the LLVM source code, use the following command:
369 </p>
370
371 <tt>gunzip --stdout llvm.tar.gz | tar -xvf -</tt>
372
373 <!------------------------------------------------------------------------->
John Criswell85ed3612003-06-12 19:34:44 +0000374 <h3><a name="checkout">Checkout LLVM from CVS</a></h3>
375 <!------------------------------------------------------------------------->
376
John Criswell312a68c2003-07-07 19:27:35 +0000377 <p>If you have access to our CVS repository, you can get a fresh copy of
378 the entire source code. All you need to do is check it out from CVS as
379 follows:
John Criswell85ed3612003-06-12 19:34:44 +0000380 <ul>
381 <li><tt>cd <i>where-you-want-llvm-to-live</i></tt>
382 <li><tt>cvs -d <i>CVSROOTDIR</i> checkout llvm</tt></p>
383 </ul>
384
385 <p>This will create an '<tt>llvm</tt>' directory in the current
386 directory and fully populate it with the LLVM source code, Makefiles,
387 test directories, and local copies of documentation files.</p>
388
John Criswellce760f62003-07-03 16:01:38 +0000389 <p>
390 Note that the C front end is not included in the CVS repository. You
391 should have either downloaded the source, or better yet, downloaded the
392 binary distribution for your platform.
393 </p>
394
John Criswell85ed3612003-06-12 19:34:44 +0000395 <!------------------------------------------------------------------------->
John Criswellce760f62003-07-03 16:01:38 +0000396 <h3><a name="config">Local LLVM Configuration</a></h3>
John Criswell85ed3612003-06-12 19:34:44 +0000397 <!------------------------------------------------------------------------->
398
John Criswellce760f62003-07-03 16:01:38 +0000399 <p>Once checked out from the CVS repository, the LLVM suite source code
400 must be configured via the <tt>configure</tt> script. This script sets
401 variables in <tt>llvm/Makefile.config</tt> and
John Criswell7a73b802003-06-30 21:59:07 +0000402 <tt>llvm/include/Config/config.h</tt>.
403
404 <p>
405 The following environment variables are used by <tt>configure</tt> to
406 configure Makefile.config:
407 </p>
John Criswell85ed3612003-06-12 19:34:44 +0000408
409 <ul>
John Criswellce760f62003-07-03 16:01:38 +0000410 <p><li><i>CXX</i> = Pathname of the C++ compiler to use.
411 <p><li><i>CC</i> = Pathname of the C compiler to use.
John Criswell85ed3612003-06-12 19:34:44 +0000412 </ul>
413
John Criswell7a73b802003-06-30 21:59:07 +0000414 The following options can be used to set or enable LLVM specific options:
415
John Criswellce760f62003-07-03 16:01:38 +0000416 <dl compact>
417 <dt><i>--with-objroot=OBJ_ROOT</i>
418 <dd>
John Criswell7a73b802003-06-30 21:59:07 +0000419 Path to the directory where
420 object files, libraries, and executables should be placed.
John Criswellce760f62003-07-03 16:01:38 +0000421 If this is set to <tt>.</tt>, then the object files will be placed
422 within the source code tree. If left unspecified, the default value is
423 <tt>.</tt>.
John Criswell7a73b802003-06-30 21:59:07 +0000424 (See the Section on <a href=#objfiles>
425 The location for LLVM object files</a>
426 for more information.)
John Criswellce760f62003-07-03 16:01:38 +0000427 <p>
428 <dt><i>--with-llvmgccdir=LLVMGCCDIR</i>
429 <dd>
430 Path to the location where the LLVM C front end binaries and
431 associated libraries will be installed.
432 <p>
433 <dt><i>--enable-optimized</i>
434 <dd>
John Criswell7a73b802003-06-30 21:59:07 +0000435 Enables optimized compilation (debugging symbols are removed and GCC
John Criswellce760f62003-07-03 16:01:38 +0000436 optimization flags are enabled). The default is to use an unoptimized
437 build (also known as a debug build).
438 <p>
439 <dt><i>--enable-jit</i>
440 <dd>
John Criswell7a73b802003-06-30 21:59:07 +0000441 Compile the Just In Time (JIT) functionality. This is not available
John Criswellce760f62003-07-03 16:01:38 +0000442 on all platforms. The default is dependent on platform, so it is best
443 to explicitly enable it if you want it.
444 </dl>
John Criswell7a73b802003-06-30 21:59:07 +0000445
446 In addition to running <tt>configure</tt>, you must set the
John Criswell85ed3612003-06-12 19:34:44 +0000447 <tt>LLVM_LIB_SEARCH_PATH</tt> environment variable in your startup scripts.
448 This environment variable is used to locate "system" libraries like
449 "<tt>-lc</tt>" and "<tt>-lm</tt>" when linking. This variable should be set
450 to the absolute path for the bytecode-libs subdirectory of the C front-end
John Criswellce760f62003-07-03 16:01:38 +0000451 install, or LLVMGCCDIR/llvm-gcc/bytecode-libs. For example, one might
452 set <tt>LLVM_LIB_SEARCH_PATH</tt> to
John Criswell85ed3612003-06-12 19:34:44 +0000453 <tt>/home/vadve/lattner/local/x86/llvm-gcc/bytecode-libs</tt> for the X86
454 version of the C front-end on our research machines.<p>
455
456 <!------------------------------------------------------------------------->
John Criswellce760f62003-07-03 16:01:38 +0000457 <h3><a name="compile">Compiling the LLVM Suite Source Code</a></h3>
458 <!------------------------------------------------------------------------->
459
460 Once you have configured LLVM, you can build it. There are three types of
461 builds:
462
463 <dl compact>
464 <dt>Debug Builds
465 <dd>
466 These builds are the default. They compile the tools and libraries
467 with debugging information.
468 <p>
469
470 <dt>Release (Optimized) Builds
471 <dd>
472 These builds are enabled with the <tt>--enable-optimized</tt> option to
473 <tt>configure</tt>. They compile the tools and libraries with GCC
474 optimizer flags on and strip debugging information from the libraries
475 and executables it generates.
476 <p>
477
478 <dt>Profile Builds
479 <dd>
480 These builds are for use with profiling. They compile profiling
481 information into the code for use with programs like <tt>gprof</tt>.
482 Profile builds must be started by setting variables on the
483 <tt>make</tt> command line.
484 </dl>
485
486 Once you have LLVM configured, you can build it by entering the top level
487 <tt>llvm</tt> directory and issuing the following command:
488 <p>
489 <tt>make</tt>
490
491 <p>
492 If you have multiple processors in your machine, you may wish to use some
493 of the parallel build options provided by GNU Make. For example, you could
494 use the command:
495 </p>
496
497 <p>
498 <tt>make -j2</tt>
499
500 <p>
501 There are several other targets which are useful when working with the LLVM
502 source code:
503
504 <dl compact>
505 <dt><tt>make clean</tt>
506 <dd>
507 Removes all files generated by the build. This includes object files,
508 generated C/C++ files, libraries, and executables.
509 <p>
510
511 <dt><tt>make distclean</tt>
512 <dd>
513 Removes everything that <tt>make clean</tt> does, but also removes
514 files generated by <tt>configure</tt>. It attempts to return the
515 source tree to the original state in which it was shipped.
516 <p>
517 </dl>
518
519 It is also possible to override default values from <tt>configure</tt> by
520 declaring variables on the command line. The following are some examples:
521
522 <dl compact>
523 <dt><tt>make ENABLE_OPTIMIZED=1</tt>
524 <dd>
525 Perform a Release (Optimized) build.
526 <p>
527
528 <dt><tt>make ENABLE_PROFILING=1</tt>
529 <dd>
530 Perform a Profiling build.
531 <p>
532
533 <dt><tt>make VERBOSE=1</tt>
534 <dd>
535 Print what <tt>make</tt> is doing on standard output.
536 <p>
537 </dl>
538
539 Every directory in the LLVM source tree includes a <tt>Makefile</tt> to
540 build it and any subdirectories that it contains. Entering any directory
541 inside the LLVM source tree and typing <tt>make</tt> should rebuild
542 anything in or below that directory that is out of date.
543
544 <!------------------------------------------------------------------------->
545 <h3><a name="build">Building the LLVM C Front End</a></h3>
546 <!------------------------------------------------------------------------->
547
548 <b>
549 <p>
550 This step is optional if you have the C front end binary distrubtion for
551 your platform.
552 </p>
553 </b>
554
555 Now that you have the LLVM Suite built, you can build the C front end. For
556 those of you that have built GCC before, the process is very similar.
557 <p>
558 Be forewarned, though: the build system for the C front end is not as
559 polished as the rest of the LLVM code, so there will be many warnings and
560 errors that you will need to ignore for now:
561
562 <ol>
563 <li>Ensure that <tt><i>OBJ_ROOT</i>/llvm/tools/Debug</tt> is at the
564 <i>end</i> of your <tt>PATH</tt> environment variable.
565
566 <li><tt>cd <i>GCCOBJ</i></tt>
567
568 <li>Configure the source code:
569 <ul>
570 <li>On Linux/x86, use
571 <ul>
572 <li><tt><i>GCCSRC</i>/configure --prefix=<i>LLVMGCCDIR</i>
573 --enable-languages=c</tt>
574 </ul>
575
576 <li>On Solaris/Sparc, use
577 <ul>
578 <li><tt><i>GCCSRC</i>/configure --prefix=<i>LLVMGCCDIR</i>
579 --enable-languages=c --target=sparcv9-sun-solaris2</tt>
580 </ul>
581 </ul>
582
583 <li><tt>make bootstrap</tt>
584
585 <li>The build will eventually fail. Don't worry; chances are good that
586 everything that needed to build is built.
587
588 <li><tt>make install</tt>
589 </ol>
590
591 At this point, you should have a working copy of the LLVM C front end
592 installed in <i>LLVMGCCDIR</i>.
593
594 <!------------------------------------------------------------------------->
John Criswell85ed3612003-06-12 19:34:44 +0000595 <h3><a name="objfiles">The location for LLVM object files</a></h3>
596 <!------------------------------------------------------------------------->
597
John Criswellce760f62003-07-03 16:01:38 +0000598 <p>The LLVM build system sends most output files generated during the build
599 into the directory defined by the variable <i>OBJ_ROOT</i> in
600 <tt>llvm/Makefile.config</tt>, which is set by the <i>--with-objroot</i>
601 option in <tt>configure</tt>. This can be either just your normal LLVM
John Criswell85ed3612003-06-12 19:34:44 +0000602 source tree or some other directory writable by you. You may wish to put
603 object files on a different filesystem either to keep them from being backed
604 up or to speed up local builds.
605
John Criswellce760f62003-07-03 16:01:38 +0000606 <p>
607 If <i>OBJ_ROOT</i> is specified, then the build system will create a
608 directory tree underneath it that resembles the source code's pathname
609 relative to your home directory.
610 </p>
John Criswell85ed3612003-06-12 19:34:44 +0000611
John Criswellce760f62003-07-03 16:01:38 +0000612 <p>
613 For example, suppose that <i>OBJ_ROOT</i> is set to <tt>/tmp</tt> and the
614 LLVM suite source code is located in <tt>/usr/home/joe/src/llvm</tt>, where
615 <tt>/usr/home/joe</tt> is the home directory of a user named Joe. Then,
616 the object files will be placed in <tt>/tmp/src/llvm</tt>.
617 </p>
John Criswell85ed3612003-06-12 19:34:44 +0000618
John Criswellce760f62003-07-03 16:01:38 +0000619 <p>
620 The LLVM build will place files underneath <i>OBJ_ROOT</i> in directories
621 named after the build type:
622 </p>
John Criswell85ed3612003-06-12 19:34:44 +0000623
John Criswellce760f62003-07-03 16:01:38 +0000624 <dl compact>
625 <dt>Debug Builds
626 <dd>
627 <dl compact>
628 <dt>Tools
629 <dd><tt><i>OBJ_ROOT</i>/llvm/tools/Debug</tt>
630 <dt>Libraries
631 <dd><tt><i>OBJ_ROOT</i>/llvm/lib/Debug</tt>
632 </dl>
633 <p>
John Criswell85ed3612003-06-12 19:34:44 +0000634
John Criswellce760f62003-07-03 16:01:38 +0000635 <dt>Release Builds
636 <dd>
637 <dl compact>
638 <dt>Tools
639 <dd><tt><i>OBJ_ROOT</i>/llvm/tools/Release</tt>
640 <dt>Libraries
641 <dd><tt><i>OBJ_ROOT</i>/llvm/lib/Release</tt>
642 </dl>
643 <p>
Chris Lattner7fe7f812002-07-24 19:51:14 +0000644
John Criswellce760f62003-07-03 16:01:38 +0000645 <dt>Profile Builds
646 <dd>
647 <dl compact>
648 <dt>Tools
649 <dd><tt><i>OBJ_ROOT</i>/llvm/tools/Profile</tt>
650 <dt>Libraries
651 <dd><tt><i>OBJ_ROOT</i>/llvm/lib/Profile</tt>
652 </dl>
653 </dl>
John Criswell85ed3612003-06-12 19:34:44 +0000654
655 <!--=====================================================================-->
656 <center>
657 <h2><a name="layout"><b>Program Layout</b></a></h2>
658 </center>
659 <!--=====================================================================-->
660
John Criswellce760f62003-07-03 16:01:38 +0000661 <p>One useful source of information about the LLVM source base is the LLVM <a
John Criswell85ed3612003-06-12 19:34:44 +0000662 href="http://www.doxygen.org">doxygen</a> documentation, available at <tt><a
663 href="http://llvm.cs.uiuc.edu/doxygen/">http://llvm.cs.uiuc.edu/doxygen/</a></tt>. The
664 following is a brief introduction to code layout:</p>
665
666
667 <!------------------------------------------------------------------------->
668 <h3><a name="cvsdir"><tt>CVS</tt> directories</a></h3>
669 <!------------------------------------------------------------------------->
670
671 Every directory checked out of CVS will contain a <tt>CVS</tt> directory;
672 for the most part these can just be ignored.
673
674
675 <!------------------------------------------------------------------------->
John Criswell85ed3612003-06-12 19:34:44 +0000676 <h3><a name="include"><tt>llvm/include</tt></a></h3>
677 <!------------------------------------------------------------------------->
678
679 This directory contains public header files exported from the LLVM
John Criswell7a73b802003-06-30 21:59:07 +0000680 library. The three main subdirectories of this directory are:<p>
John Criswell85ed3612003-06-12 19:34:44 +0000681
John Criswell8df90e02003-06-11 20:46:40 +0000682 <ol>
John Criswell85ed3612003-06-12 19:34:44 +0000683 <li><tt>llvm/include/llvm</tt> - This directory contains all of the LLVM
684 specific header files. This directory also has subdirectories for
685 different portions of LLVM: <tt>Analysis</tt>, <tt>CodeGen</tt>,
686 <tt>Reoptimizer</tt>, <tt>Target</tt>, <tt>Transforms</tt>, etc...
687
688 <li><tt>llvm/include/Support</tt> - This directory contains generic
John Criswellce760f62003-07-03 16:01:38 +0000689 support libraries that are independent of LLVM, but are used by LLVM.
John Criswell7a73b802003-06-30 21:59:07 +0000690 For example, some C++ STL utilities and a Command Line option processing
691 library.
692
693 <li><tt>llvm/include/Config</tt> - This directory contains header files
694 configured by the <tt>configure</tt> script. They wrap "standard" UNIX
695 and C header files. Source code can include these header files which
696 automatically take care of the conditional #includes that the configure
697 script generates.
Chris Lattner7fe7f812002-07-24 19:51:14 +0000698 </ol>
John Criswell85ed3612003-06-12 19:34:44 +0000699
700 <!------------------------------------------------------------------------->
701 <h3><a name="lib"><tt>llvm/lib</tt></a></h3>
702 <!------------------------------------------------------------------------->
703
John Criswell7a73b802003-06-30 21:59:07 +0000704 This directory contains most of the source files of the LLVM system. In
705 LLVM almost all
706 code exists in libraries, making it very easy to share code among the
John Criswell85ed3612003-06-12 19:34:44 +0000707 different <a href="#tools">tools</a>.<p>
708
709 <dl compact>
710 <dt><tt>llvm/lib/VMCore/</tt><dd> This directory holds the core LLVM
711 source files that implement core classes like Instruction and BasicBlock.
712
713 <dt><tt>llvm/lib/AsmParser/</tt><dd> This directory holds the source code
714 for the LLVM assembly language parser library.
715
716 <dt><tt>llvm/lib/ByteCode/</tt><dd> This directory holds code for reading
717 and write LLVM bytecode.
718
719 <dt><tt>llvm/lib/CWriter/</tt><dd> This directory implements the LLVM to C
720 converter.
721
722 <dt><tt>llvm/lib/Analysis/</tt><dd> This directory contains a variety of
723 different program analyses, such as Dominator Information, Call Graphs,
724 Induction Variables, Interval Identification, Natural Loop Identification,
725 etc...
726
727 <dt><tt>llvm/lib/Transforms/</tt><dd> This directory contains the source
728 code for the LLVM to LLVM program transformations, such as Aggressive Dead
729 Code Elimination, Sparse Conditional Constant Propagation, Inlining, Loop
730 Invarient Code Motion, Dead Global Elimination, Pool Allocation, and many
731 others...
732
733 <dt><tt>llvm/lib/Target/</tt><dd> This directory contains files that
734 describe various target architectures for code generation. For example,
735 the llvm/lib/Target/Sparc directory holds the Sparc machine
736 description.<br>
737
738 <dt><tt>llvm/lib/CodeGen/</tt><dd> This directory contains the major parts
739 of the code generator: Instruction Selector, Instruction Scheduling, and
740 Register Allocation.
741
742 <dt><tt>llvm/lib/Reoptimizer/</tt><dd> This directory holds code related
743 to the runtime reoptimizer framework that is currently under development.
744
745 <dt><tt>llvm/lib/Support/</tt><dd> This directory contains the source code
746 that corresponds to the header files located in
747 <tt>llvm/include/Support/</tt>.
748 </dl>
749
750 <!------------------------------------------------------------------------->
751 <h3><a name="test"><tt>llvm/test</tt></a></h3>
752 <!------------------------------------------------------------------------->
753
754 <p>This directory contains regression tests and source code that is used to
755 test the LLVM infrastructure...</p>
756
757 <!------------------------------------------------------------------------->
758 <h3><a name="tools"><tt>llvm/tools</tt></a></h3>
759 <!------------------------------------------------------------------------->
760
761 <p>The <b>tools</b> directory contains the executables built out of the
762 libraries above, which form the main part of the user interface. You can
763 always get help for a tool by typing <tt>tool_name --help</tt>. The
764 following is a brief introduction to the most important tools.</p>
765
766 <dl compact>
767 <dt><tt><b>as</b></tt><dd>The assembler transforms the human readable
768 LLVM assembly to LLVM bytecode.<p>
769
770 <dt><tt><b>dis</b></tt><dd>The disassembler transforms the LLVM bytecode
771 to human readable LLVM assembly. Additionally it can convert LLVM
772 bytecode to C, which is enabled with the <tt>-c</tt> option.<p>
773
774 <dt><tt><b>lli</b></tt><dd> <tt>lli</tt> is the LLVM interpreter, which
775 can directly execute LLVM bytecode (although very slowly...). In addition
John Criswellce760f62003-07-03 16:01:38 +0000776 to a simple interpreter, <tt>lli</tt> is also has debugger and tracing
John Criswell85ed3612003-06-12 19:34:44 +0000777 modes (entered by specifying <tt>-debug</tt> or <tt>-trace</tt> on the
Misha Brukmanc56e5822003-07-03 16:29:36 +0000778 command line, respectively). Finally, for architectures that support it
779 (currently only x86 and Sparc), by default, <tt>lli</tt> will function as
780 a Just-In-Time compiler (if the functionality was compiled in), and will
781 execute the code <i>much</i> faster than the interpreter.<p>
John Criswell85ed3612003-06-12 19:34:44 +0000782
783 <dt><tt><b>llc</b></tt><dd> <tt>llc</tt> is the LLVM backend compiler,
Misha Brukmanc56e5822003-07-03 16:29:36 +0000784 which translates LLVM bytecode to a SPARC or x86 assembly file.<p>
John Criswell85ed3612003-06-12 19:34:44 +0000785
786 <dt><tt><b>llvmgcc</b></tt><dd> <tt>llvmgcc</tt> is a GCC based C frontend
787 that has been retargeted to emit LLVM code as the machine code output. It
788 works just like any other GCC compiler, taking the typical <tt>-c, -S, -E,
789 -o</tt> options that are typically used. The source code for the
790 <tt>llvmgcc</tt> tool is currently not included in the LLVM cvs tree
791 because it is quite large and not very interesting.<p>
792
793 <ol>
794 <dt><tt><b>gccas</b></tt><dd> This tool is invoked by the
795 <tt>llvmgcc</tt> frontend as the "assembler" part of the compiler. This
796 tool actually assembles LLVM assembly to LLVM bytecode,
John Criswell7a73b802003-06-30 21:59:07 +0000797 performs a variety of optimizations,
John Criswell85ed3612003-06-12 19:34:44 +0000798 and outputs LLVM bytecode. Thus when you invoke <tt>llvmgcc -c x.c -o
799 x.o</tt>, you are causing <tt>gccas</tt> to be run, which writes the
800 <tt>x.o</tt> file (which is an LLVM bytecode file that can be
801 disassembled or manipulated just like any other bytecode file). The
802 command line interface to <tt>gccas</tt> is designed to be as close as
803 possible to the <b>system</b> '<tt>as</tt>' utility so that the gcc
John Criswellce760f62003-07-03 16:01:38 +0000804 frontend itself did not have to be modified to interface to a "weird"
John Criswell85ed3612003-06-12 19:34:44 +0000805 assembler.<p>
806
807 <dt><tt><b>gccld</b></tt><dd> <tt>gccld</tt> links together several LLVM
808 bytecode files into one bytecode file and does some optimization. It is
809 the linker invoked by the gcc frontend when multiple .o files need to be
810 linked together. Like <tt>gccas</tt> the command line interface of
811 <tt>gccld</tt> is designed to match the system linker, to aid
812 interfacing with the GCC frontend.<p>
813 </ol>
814
815 <dt><tt><b>opt</b></tt><dd> <tt>opt</tt> reads LLVM bytecode, applies a
816 series of LLVM to LLVM transformations (which are specified on the command
817 line), and then outputs the resultant bytecode. The '<tt>opt --help</tt>'
818 command is a good way to get a list of the program transformations
819 available in LLVM.<p>
820
821
822 <dt><tt><b>analyze</b></tt><dd> <tt>analyze</tt> is used to run a specific
823 analysis on an input LLVM bytecode file and print out the results. It is
824 primarily useful for debugging analyses, or familiarizing yourself with
825 what an analysis does.<p>
826
827 </dl>
828
829 <!--=====================================================================-->
830 <h2><a name="tutorial">An example using the LLVM tool chain</h2>
831 <!--=====================================================================-->
832
833 <ol>
834 <li>First, create a simple C file, name it 'hello.c':
835 <pre>
836 #include &lt;stdio.h&gt;
837 int main() {
838 printf("hello world\n");
839 return 0;
840 }
841 </pre>
842
843 <li>Next, compile the C file into a LLVM bytecode file:<p>
844
845 <tt>% llvmgcc hello.c -o hello</tt><p>
846
847 This will create two result files: <tt>hello</tt> and
848 <tt>hello.bc</tt>. The <tt>hello.bc</tt> is the LLVM bytecode that
849 corresponds the the compiled program and the library facilities that it
850 required. <tt>hello</tt> is a simple shell script that runs the bytecode
851 file with <tt>lli</tt>, making the result directly executable.<p>
852
853 <li>Run the program. To make sure the program ran, execute one of the
854 following commands:<p>
John Criswell8df90e02003-06-11 20:46:40 +0000855
John Criswell85ed3612003-06-12 19:34:44 +0000856 <tt>% ./hello</tt><p>
857
858 or<p>
859
860 <tt>% lli hello.bc</tt><p>
861
862 <li>Use the <tt>dis</tt> utility to take a look at the LLVM assembly
863 code:<p>
864
865 <tt>% dis < hello.bc | less</tt><p>
866
867 <li>Compile the program to native Sparc assembly using the code
Misha Brukmanc56e5822003-07-03 16:29:36 +0000868 generator (assuming you are currently on a Sparc system):<p>
John Criswell85ed3612003-06-12 19:34:44 +0000869
870 <tt>% llc hello.bc -o hello.s</tt><p>
871
872 <li>Assemble the native sparc assemble file into a program:<p>
873
874 <tt>% /opt/SUNWspro/bin/cc -xarch=v9 hello.s -o hello.sparc</tt><p>
875
876 <li>Execute the native sparc program:<p>
877
878 <tt>% ./hello.sparc</tt><p>
879
880 </ol>
881
882
883 <!--=====================================================================-->
John Criswellce760f62003-07-03 16:01:38 +0000884 <h2><a name="help">Common Problems</a></h2>
885 <!--=====================================================================-->
886
887 Below are common problems and their remedies:
888
889 <dl compact>
890 <dt><b>When I run configure, it finds the wrong C compiler.</b>
891 <dd>
892 The <tt>configure</tt> script attempts to locate first <tt>gcc</tt> and
893 then <tt>cc</tt>, unless it finds compiler paths set in <tt>CC</tt> and
894 <tt>CXX</tt> for the C and C++ compiler, respectively.
895
896 If <tt>configure</tt> finds the wrong compiler, either adjust your
897 <tt>PATH</tt> environment variable or set <tt>CC</tt> and <tt>CXX</tt>
898 explicitly.
899 <p>
900
901 <dt><b>I compile the code, and I get some error about /localhome</b>.
902 <dd>
903 There are several possible causes for this. The first is that you
904 didn't set a pathname properly when using <tt>configure</tt>, and it
905 defaulted to a pathname that we use on our research machines.
906 <p>
907 Another possibility is that we hardcoded a path in our Makefiles. If
908 you see this, please email the LLVM bug mailing list with the name of
909 the offending Makefile and a description of what is wrong with it.
910
911 <dt><b>The <tt>configure</tt> script finds the right C compiler, but it
912 uses the LLVM linker from a previous build. What do I do?</b>
913 <dd>
914 The <tt>configure</tt> script uses the <tt>PATH</tt> to find
915 executables, so if it's grabbing the wrong linker/assembler/etc, there
916 are two ways to fix it:
917 <ol>
918 <li>Adjust your <tt>PATH</tt> environment variable so that the
919 correct program appears first in the <tt>PATH</tt>. This may work,
920 but may not be convenient when you want them <i>first</i> in your
921 path for other work.
922 <p>
923
924 <li>Run <tt>configure</tt> with an alternative <tt>PATH</tt> that
925 is correct. In a Borne compatible shell, the syntax would be:
926 <p>
927 <tt>PATH=<the path without the bad program> ./configure ...</tt>
928 <p>
929 This is still somewhat inconvenient, but it allows
930 <tt>configure</tt> to do its work without having to adjust your
931 <tt>PATH</tt> permanently.
932 </ol>
933 </dl>
934
935 <!--=====================================================================-->
John Criswell85ed3612003-06-12 19:34:44 +0000936 <h2><a name="links">Links</a></h2>
937 <!--=====================================================================-->
938
939 <p>This document is just an <b>introduction</b> to how to use LLVM to do
940 some simple things... there are many more interesting and complicated things
941 that you can do that aren't documented here (but we'll gladly accept a patch
942 if you want to write something up!). For more information about LLVM, check
943 out:</p>
944
945 <ul>
946 <li><a href="http://llvm.cs.uiuc.edu/">LLVM homepage</a></li>
947 <li><a href="http://llvm.cs.uiuc.edu/doxygen/">LLVM doxygen tree</a></li>
John Criswellce760f62003-07-03 16:01:38 +0000948 <li><a href="http://llvm.cs.uiuc.edu/docs/Projects.html">Starting a Project that Uses LLVM</a></li>
John Criswell85ed3612003-06-12 19:34:44 +0000949 </ul>
950
951 <hr>
952
953 If you have any questions or run into any snags (or you have any
954 additions...), please send an email to
955 <a href="mailto:sabre@nondot.org">Chris Lattner</a>.</p>
956
957 <!-- Created: Mon Jul 1 02:29:02 CDT 2002 -->
958 <!-- hhmts start -->
959Last modified: Tue Jun 3 22:06:43 CDT 2003
960<!-- hhmts end -->
961 </body>
Guochun Shif4688a82002-07-17 23:05:56 +0000962</html>