1 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
4 <title>Getting Started with LLVM System</title>
8 <center><h1>Getting Started with the LLVM System<br><font size=3>By: <a
9 href="mailto:gshi1@uiuc.edu">Guochun Shi</a>,
10 <a href="mailto:sabre@nondot.org">Chris Lattner</a>,
11 <a href="mailto:criswell@uiuc.edu">John Criswell</a>,
12 <a href="http://misha.brukman.net">Misha Brukman</a>, and
13 <a href="http://www.cs.uiuc.edu/~vadve">Vikram Adve</a>
16 <!--=====================================================================-->
17 <h2><a name="Contents">Contents</a></h2>
18 <!--=====================================================================-->
21 <li><a href="#overview">Overview</a>
23 <li><a href="#requirements">Requirements</a>
25 <li><a href="#hardware">Hardware</a>
26 <li><a href="#software">Software</a>
29 <li><a href="#starting">Getting started with LLVM</a>
31 <li><a href="#quickstart">Getting Started Quickly (A Summary)</a>
32 <li><a href="#terminology">Terminology and Notation</tt></a>
33 <li><a href="#environment">Setting Up Your Environment</a>
34 <li><a href="#unpack">Unpacking the LLVM Archives</a>
35 <li><a href="#checkout">Checkout LLVM from CVS</a>
36 <li><a href="#installcf">Install the C Front End</a>
37 <li><a href="#config">Local LLVM Configuration</tt></a>
38 <li><a href="#compile">Compiling the LLVM Suite Source Code</a>
39 <li><a href="#objfiles">The Location of LLVM Object Files</tt></a>
41 <li><a href="#layout">Program layout</a>
43 <li><a href="#cvsdir"><tt>CVS</tt> directories</a>
44 <li><a href="#include"><tt>llvm/include</tt></a>
45 <li><a href="#lib"><tt>llvm/lib</tt></a>
46 <li><a href="#test"><tt>llvm/test</tt></a>
47 <li><a href="#tools"><tt>llvm/tools</tt></a>
48 <li><a href="#utils"><tt>llvm/utils</tt></a>
50 <li><a href="#cfront">Compiling the LLVM C Front End</a>
51 <li><a href="#tutorial">An Example Using the LLVM Tool Chain</a>
52 <li><a href="#problems">Common Problems</a>
53 <li><a href="#links">Links</a>
57 <!--=====================================================================-->
59 <h2><a name="overview"><b>Overview</b></a></h2>
61 <!--=====================================================================-->
63 Welcome to LLVM! In order to get started, you first need to know some
67 First, LLVM comes in two pieces. The first piece is the LLVM suite. This
68 contains all of the tools, libraries, and header files needed to use the
69 low level virtual machine. It also contains a test suite that can be used
70 to test the LLVM tools and the C front end.
72 The second piece is the C front end. This component provides a version
73 of GCC that compiles C code into LLVM bytecode. Currently, the C front end
74 is a modified version of GCC 3.4 (we track the GCC 3.4 development).
75 Once compiled into LLVM bytecode, a program can be manipulated with the
76 LLVM tools from the LLVM suite.
78 <!--=====================================================================-->
79 <h3><a name="requirements"><b>Requirements</b></a></h3>
80 <!--=====================================================================-->
82 Before you begin to use the LLVM system, review the requirements given
83 below. This may save you some trouble by knowing ahead of time what
84 hardware and software you will need.
86 <!--=====================================================================-->
87 <h4><a name="hardware"><b>Hardware</b></a></h4>
88 <!--=====================================================================-->
89 LLVM is known to work on the following platforms:
93 <li> Approximately 760 MB of Free Disk Space
95 <li>Source code: 30 MB
96 <li>Object code: 670 MB
97 <li>C front end: 60 MB
103 <li> Solaris on SparcV9 (Ultrasparc)
105 <li> Approximately 1.24 GB of Free Disk Space
107 <li>Source code: 30 MB
108 <li>Object code: 1000 MB
109 <li>C front end: 210 MB
115 If you want to compile your own version of the C front end, you will need
116 additional disk space:
122 <li> Approximately 249 MB of Free Disk Space
124 <li>Source code: 146 MB
125 <li>Object code: 82 MB
126 <li>Installed binaries: 21 MB
134 <li> Approximately 264 MB of Free Disk Space
136 <li>Source code: 146 MB
137 <li>Object code: 93 MB
138 <li>Installed binaries: 25 MB
144 LLVM <i>may</i> compile on other platforms. The LLVM utilities should work
145 on other platforms, so it should be possible to generate and produce LLVM
146 bytecode on unsupported platforms (although bytecode generated on one
147 platform may not work on another platform). However, the code generators
148 and Just-In-Time (JIT) compilers only generate SparcV9 or x86 machine code.
151 <!--=====================================================================-->
152 <h4><a name="software"><b>Software</b></a></h4>
153 <!--=====================================================================-->
156 Unpacking the distribution requires the following tools:
161 These tools are needed to uncompress and unarchive the software.
162 Regular Solaris <tt>tar</tt> may work for unpacking the TAR archive but
166 Compiling LLVM requires that you have several different software packages
172 The GNU Compiler Collection must be installed with C and C++ language
173 support. GCC 3.2.x works, and GCC 3.x is generally supported.
176 Note that we currently do not support any other C++ compiler.
181 The LLVM build system relies upon GNU Make extensions. Therefore, you
182 will need GNU Make (sometimes known as gmake) to build LLVM.
187 The LLVM source code is built using flex and bison. You will not be
188 able to configure and compile LLVM without them.
193 If you are installing Bison on your machine for the first time, you
194 will need GNU M4 (version 1.4 or higher).
198 There are some additional tools that you may want to have when working with
206 If you want to make changes to the configure scripts, you will need
207 GNU autoconf (2.53 or higher), and consequently, GNU M4 (version 1.4
213 <p>The <a href="starting">next section</a> of this guide is meant to get
214 you up and running with LLVM and to give you some basic information about
215 the LLVM environment. The <a href"#quickstart">first subsection</a> gives
216 a short summary for those who are already familiar with the system and
217 want to get started as quickly as possible.
219 <p>The later sections of this guide describe the <a
220 href="#layout">general layout</a> of the the LLVM source-tree, a <a
221 href="#tutorial">simple example</a> using the LLVM tool chain, and <a
222 href="#links">links</a> to find more information about LLVM or to get
225 <!--=====================================================================-->
227 <h2><a name="starting"><b>Getting Started with LLVM</b></a></h2>
229 <!--=====================================================================-->
231 <!--=====================================================================-->
232 <h3><a name="quickstart"><b>Getting Started Quickly (A Summary)</b></a></h3>
233 <!--=====================================================================-->
235 Here's the short story for getting up and running quickly with LLVM:
237 <li>Install the C front end:
239 <li><tt>cd <i>where-you-want-the-C-front-end-to-live</i></tt>
240 <li><tt>gunzip --stdout cfrontend.<i>platform</i>.tar.gz | tar -xvf
246 <li>Get the Source Code
248 <li>With the distributed files:
250 <li><tt>cd <i>where-you-want-llvm-to-live</i></tt>
251 <li><tt>gunzip --stdout llvm.tar.gz | tar -xvf -</tt>
252 <li><tt>gunzip --stdout cfrontend.<i>platform</i>.tar.gz | tar
259 <li>With anonymous CVS access:
261 <li>Find the path to the CVS repository containing LLVM (we'll
262 call this <i>CVSROOTDIR</i>).
263 <li><tt>cd <i>where-you-want-llvm-to-live</i></tt>
264 <li><tt>cvs -d <i>CVSROOTDIR</i> checkout llvm</tt>
272 <li>Configure the LLVM Build Environment
274 <li>Run <tt>configure</tt> to configure the Makefiles and header
275 files for the default platform.
276 Useful options include:
278 <li><tt>--with-objroot=<i>directory</i></tt>
280 Specify where object files should be placed during the
283 <li><tt>--with-llvmgccdir=<i>directory</i></tt>
285 Specify where the LLVM C frontend is going to be installed.
291 <li>Build the LLVM Suite
293 <li>Set your LLVM_LIB_SEARCH_PATH environment variable.
294 <li><tt>gmake -k |& tee gnumake.out
295 # this is csh or tcsh syntax</tt>
302 <p>See <a href="#environment">Setting Up Your Environment</a> on tips to
303 simplify working with the LLVM front-end and compiled tools. See the
304 other sub-sections below for other useful details in working with LLVM,
305 or go straight to <a href="#layout">Program Layout</a> to learn about the
306 layout of the source code tree. For information on building the C front
307 end yourself, see <a href="#cfront">Compiling the LLVM C Front End</a> for
310 <!------------------------------------------------------------------------->
311 <h3><a name="terminology">Terminology and Notation</a></h3>
312 <!------------------------------------------------------------------------->
314 <p>Throughout this manual, the following names are used to denote paths
315 specific to the local system and working environment. <i>These are not
316 environment variables you need to set but just strings used in the rest
317 of this document below</i>. In any of the examples below, simply replace
318 each of these names with the appropriate pathname on your local system.
319 All these paths are absolute:</p>
323 This is the path for the CVS repository containing the LLVM source
324 code. Ask the person responsible for your local LLVM installation to
330 This is the top level directory for where the LLVM suite object files
331 will be placed during the build.
336 This is the pathname to the location where the LLVM C Front End will
337 be installed. Note that the C front end does not need to be installed
338 during the LLVM suite build; you will just need to know where it will
339 go for configuring the build system and running the test suite later.
341 For the pre-built C front end binaries, the LLVMGCCDIR is
342 <tt>cfrontend/<i>platform</i>/llvm-gcc</tt>.
346 This is the pathname of the directory where the LLVM C front end source
352 This is the pathname of the directory where the LLVM C front end object
353 code will be placed during the build. It can be safely removed once
354 the build is complete.
357 <!------------------------------------------------------------------------->
358 <h3><a name="environment">Setting Up Your Environment</a></h3>
359 <!------------------------------------------------------------------------->
362 In order to compile and use LLVM, you will need to set some environment
363 variables. There are also some shell aliases which you may find useful.
364 You can set these on the command line, or better yet, set them in your
365 <tt>.cshrc</tt> or <tt>.profile</tt>.
368 <dt><tt>LLVM_LIB_SEARCH_PATH</tt>=<tt><i>LLVMGCCDIR</i>/llvm-gcc/bytecode-libs</tt>
370 This environment variable helps the LLVM C front end find bytecode
371 libraries that it will need for compilation.
374 <dt>alias llvmgcc <i>LLVMGCCDIR</i><tt>/bin/llvm-gcc</tt>
376 This alias allows you to use the LLVM C front end without putting it in
377 your <tt>PATH</tt> or typing in its complete pathname.
380 <!------------------------------------------------------------------------->
381 <h3><a name="unpack">Unpacking the LLVM Archives</a></h3>
382 <!------------------------------------------------------------------------->
385 If you have the LLVM distribution, you will need to unpack it before you
386 can begin to compile it. LLVM is distributed as a set of four files. Each
387 file is a TAR archive that is compressed with the gzip program.
390 <p> The four files are as follows:
393 <dd>This is the source code to the LLVM suite.
396 <dt>cfrontend.sparc.tar.gz
397 <dd>This is the binary release of the C front end for Solaris/Sparc.
400 <dt>cfrontend.x86.tar.gz
401 <dd>This is the binary release of the C front end for Linux/x86.
404 <dt>cfrontend-src.tar.gz
405 <dd>This is the source code release of the C front end.
409 <!------------------------------------------------------------------------->
410 <h3><a name="checkout">Checkout LLVM from CVS</a></h3>
411 <!------------------------------------------------------------------------->
413 <p>If you have access to our CVS repository, you can get a fresh copy of
414 the entire source code. All you need to do is check it out from CVS as
417 <li><tt>cd <i>where-you-want-llvm-to-live</i></tt>
418 <li><tt>cvs -d <i>CVSROOTDIR</i> checkout llvm</tt></p>
421 <p>This will create an '<tt>llvm</tt>' directory in the current
422 directory and fully populate it with the LLVM source code, Makefiles,
423 test directories, and local copies of documentation files.</p>
426 Note that the C front end is not included in the CVS repository. You
427 should have either downloaded the source, or better yet, downloaded the
428 binary distribution for your platform.
431 <!------------------------------------------------------------------------->
432 <h3><a name="installcf">Install the C Front End</a></h3>
433 <!------------------------------------------------------------------------->
436 Before configuring and compiling the LLVM suite, it is best to extract the
437 LLVM C front end. While not used in building, the C front end is used by
438 the LLVM test suite, and its location must be given to the
439 <tt>configure</tt> script before the LLVM suite can be built.
443 To install the C front end, do the following:
445 <li><tt>cd <i>where-you-want-the-front-end-to-live</i></tt>
446 <li><tt>gunzip --stdout cfrontend.<i>platform</i>.tar.gz | tar -xvf
450 <!------------------------------------------------------------------------->
451 <h3><a name="config">Local LLVM Configuration</a></h3>
452 <!------------------------------------------------------------------------->
454 <p>Once checked out from the CVS repository, the LLVM suite source code
455 must be configured via the <tt>configure</tt> script. This script sets
456 variables in <tt>llvm/Makefile.config</tt> and
457 <tt>llvm/include/Config/config.h</tt>.
460 The following environment variables are used by the <tt>configure</tt>
461 script to configure the build system:
475 Tells <tt>configure</tt> which C compiler to use. By default,
476 <tt>configure</tt> will look for the first GCC compiler in
477 <tt>PATH</tt>. Use this variable to override
478 <tt>configure</tt>'s default behavior.
485 Tells <tt>configure</tt> which C++ compiler to use. By default,
486 <tt>configure</tt> will look for the first GCC compiler in
487 <tt>PATH</tt>. Use this variable to override
488 <tt>configure</tt>'s default behavior.
494 The following options can be used to set or enable LLVM specific options:
498 <dt><i>--with-objroot=OBJ_ROOT</i>
500 Path to the directory where
501 object files, libraries, and executables should be placed.
502 If this is set to <tt>.</tt>, then the object files will be placed
503 within the source code tree. If left unspecified, the default value is
507 If the USER environment variable is specified and the directory
508 <tt>/localhome/$USER</tt> exists, then the default value is
509 <tt>/localhome/$USER</tt>.
512 Otherwise, the default value is <tt>.</tt>.
514 (See the Section on <a href=#objfiles>
515 The Location of LLVM Object Files</a>
516 for more information.)
518 <dt><i>--with-llvmgccdir=LLVMGCCDIR</i>
520 Path to the location where the LLVM C front end binaries and
521 associated libraries will be installed.
523 <dt><i>--enable-optimized</i>
525 Enables optimized compilation (debugging symbols are removed and GCC
526 optimization flags are enabled). The default is to use an unoptimized
527 build (also known as a debug build).
529 <dt><i>--enable-jit</i>
531 Compile the Just In Time (JIT) functionality. This is not available
532 on all platforms. The default is dependent on platform, so it is best
533 to explicitly enable it if you want it.
536 In addition to running <tt>configure</tt>, you must set the
537 <tt>LLVM_LIB_SEARCH_PATH</tt> environment variable in your startup scripts.
538 This environment variable is used to locate "system" libraries like
539 "<tt>-lc</tt>" and "<tt>-lm</tt>" when linking. This variable should be set
540 to the absolute path for the bytecode-libs subdirectory of the C front-end
541 install, or LLVMGCCDIR/llvm-gcc/bytecode-libs. For example, one might
542 set <tt>LLVM_LIB_SEARCH_PATH</tt> to
543 <tt>/home/vadve/lattner/local/x86/llvm-gcc/bytecode-libs</tt> for the X86
544 version of the C front-end on our research machines.<p>
546 <!------------------------------------------------------------------------->
547 <h3><a name="compile">Compiling the LLVM Suite Source Code</a></h3>
548 <!------------------------------------------------------------------------->
550 Once you have configured LLVM, you can build it. There are three types of
556 These builds are the default. They compile the tools and libraries
557 with debugging information.
560 <dt>Release (Optimized) Builds
562 These builds are enabled with the <tt>--enable-optimized</tt> option to
563 <tt>configure</tt>. They compile the tools and libraries with GCC
564 optimizer flags on and strip debugging information from the libraries
565 and executables it generates.
570 These builds are for use with profiling. They compile profiling
571 information into the code for use with programs like <tt>gprof</tt>.
572 Profile builds must be started by setting variables on the
573 <tt>gmake</tt> command line.
576 Once you have LLVM configured, you can build it by entering the top level
577 <tt>llvm</tt> directory and issuing the following command:
582 If you have multiple processors in your machine, you may wish to use some
583 of the parallel build options provided by GNU Make. For example, you could
591 There are several other targets which are useful when working with the LLVM
595 <dt><tt>gmake clean</tt>
597 Removes all files generated by the build. This includes object files,
598 generated C/C++ files, libraries, and executables.
601 <dt><tt>gmake distclean</tt>
603 Removes everything that <tt>gmake clean</tt> does, but also removes
604 files generated by <tt>configure</tt>. It attempts to return the
605 source tree to the original state in which it was shipped.
609 It is also possible to override default values from <tt>configure</tt> by
610 declaring variables on the command line. The following are some examples:
613 <dt><tt>gmake ENABLE_OPTIMIZED=1</tt>
615 Perform a Release (Optimized) build.
618 <dt><tt>gmake ENABLE_PROFILING=1</tt>
620 Perform a Profiling build.
623 <dt><tt>gmake VERBOSE=1</tt>
625 Print what <tt>gmake</tt> is doing on standard output.
629 Every directory in the LLVM source tree includes a <tt>Makefile</tt> to
630 build it and any subdirectories that it contains. Entering any directory
631 inside the LLVM source tree and typing <tt>gmake</tt> should rebuild
632 anything in or below that directory that is out of date.
634 <!------------------------------------------------------------------------->
635 <h3><a name="objfiles">The Location of LLVM Object Files</a></h3>
636 <!------------------------------------------------------------------------->
638 <p>The LLVM build system sends most output files generated during the build
639 into the directory defined by the variable <i>OBJ_ROOT</i> in
640 <tt>llvm/Makefile.config</tt>, which is set by the <i>--with-objroot</i>
641 option in <tt>configure</tt>. This can be either just your normal LLVM
642 source tree or some other directory writable by you. You may wish to put
643 object files on a different filesystem either to keep them from being backed
644 up or to speed up local builds.
647 If <i>OBJ_ROOT</i> is specified, then the build system will create a
648 directory tree underneath it that resembles the source code's pathname
649 relative to your home directory (unless <i>OBJ_ROOT</i> is set to
650 <tt>.</tt>, in which case object files are placed within the LLVM source
656 <i>--with-objroot</i>=<tt>.</tt>
658 <i>--with-objroot</i>=<tt>`pwd`</tt>
659 are not the same thing. The former will simply place object files within
660 the source tree, while the latter will set the location of object files
661 using the source tree's relative path from the home directory.
665 For example, suppose that <i>OBJ_ROOT</i> is set to <tt>/tmp</tt> and the
666 LLVM suite source code is located in <tt>/usr/home/joe/src/llvm</tt>, where
667 <tt>/usr/home/joe</tt> is the home directory of a user named Joe. Then,
668 the object files will be placed in <tt>/tmp/src/llvm</tt>.
672 The LLVM build will place files underneath <i>OBJ_ROOT</i> in directories
673 named after the build type:
681 <dd><tt><i>OBJ_ROOT</i>/llvm/tools/Debug</tt>
683 <dd><tt><i>OBJ_ROOT</i>/llvm/lib/Debug</tt>
691 <dd><tt><i>OBJ_ROOT</i>/llvm/tools/Release</tt>
693 <dd><tt><i>OBJ_ROOT</i>/llvm/lib/Release</tt>
701 <dd><tt><i>OBJ_ROOT</i>/llvm/tools/Profile</tt>
703 <dd><tt><i>OBJ_ROOT</i>/llvm/lib/Profile</tt>
707 <!--=====================================================================-->
709 <h2><a name="layout"><b>Program Layout</b></a></h2>
711 <!--=====================================================================-->
713 <p>One useful source of information about the LLVM source base is the LLVM <a
714 href="http://www.doxygen.org">doxygen</a> documentation, available at <tt><a
715 href="http://llvm.cs.uiuc.edu/doxygen/">http://llvm.cs.uiuc.edu/doxygen/</a></tt>. The
716 following is a brief introduction to code layout:</p>
719 <!------------------------------------------------------------------------->
720 <h3><a name="cvsdir"><tt>CVS</tt> directories</a></h3>
721 <!------------------------------------------------------------------------->
723 Every directory checked out of CVS will contain a <tt>CVS</tt> directory;
724 for the most part these can just be ignored.
727 <!------------------------------------------------------------------------->
728 <h3><a name="include"><tt>llvm/include</tt></a></h3>
729 <!------------------------------------------------------------------------->
731 This directory contains public header files exported from the LLVM
732 library. The three main subdirectories of this directory are:<p>
735 <li><tt>llvm/include/llvm</tt> - This directory contains all of the LLVM
736 specific header files. This directory also has subdirectories for
737 different portions of LLVM: <tt>Analysis</tt>, <tt>CodeGen</tt>,
738 <tt>Reoptimizer</tt>, <tt>Target</tt>, <tt>Transforms</tt>, etc...
740 <li><tt>llvm/include/Support</tt> - This directory contains generic
741 support libraries that are independent of LLVM, but are used by LLVM.
742 For example, some C++ STL utilities and a Command Line option processing
745 <li><tt>llvm/include/Config</tt> - This directory contains header files
746 configured by the <tt>configure</tt> script. They wrap "standard" UNIX
747 and C header files. Source code can include these header files which
748 automatically take care of the conditional #includes that the configure
752 <!------------------------------------------------------------------------->
753 <h3><a name="lib"><tt>llvm/lib</tt></a></h3>
754 <!------------------------------------------------------------------------->
756 This directory contains most of the source files of the LLVM system. In
758 code exists in libraries, making it very easy to share code among the
759 different <a href="#tools">tools</a>.<p>
762 <dt><tt>llvm/lib/VMCore/</tt><dd> This directory holds the core LLVM
763 source files that implement core classes like Instruction and BasicBlock.
765 <dt><tt>llvm/lib/AsmParser/</tt><dd> This directory holds the source code
766 for the LLVM assembly language parser library.
768 <dt><tt>llvm/lib/ByteCode/</tt><dd> This directory holds code for reading
769 and write LLVM bytecode.
771 <dt><tt>llvm/lib/CWriter/</tt><dd> This directory implements the LLVM to C
774 <dt><tt>llvm/lib/Analysis/</tt><dd> This directory contains a variety of
775 different program analyses, such as Dominator Information, Call Graphs,
776 Induction Variables, Interval Identification, Natural Loop Identification,
779 <dt><tt>llvm/lib/Transforms/</tt><dd> This directory contains the source
780 code for the LLVM to LLVM program transformations, such as Aggressive Dead
781 Code Elimination, Sparse Conditional Constant Propagation, Inlining, Loop
782 Invarient Code Motion, Dead Global Elimination, and many others...
784 <dt><tt>llvm/lib/Target/</tt><dd> This directory contains files that
785 describe various target architectures for code generation. For example,
786 the llvm/lib/Target/Sparc directory holds the Sparc machine
789 <dt><tt>llvm/lib/CodeGen/</tt><dd> This directory contains the major parts
790 of the code generator: Instruction Selector, Instruction Scheduling, and
793 <dt><tt>llvm/lib/Reoptimizer/</tt><dd> This directory holds code related
794 to the runtime reoptimizer framework that is currently under development.
796 <dt><tt>llvm/lib/Support/</tt><dd> This directory contains the source code
797 that corresponds to the header files located in
798 <tt>llvm/include/Support/</tt>.
801 <!------------------------------------------------------------------------->
802 <h3><a name="test"><tt>llvm/test</tt></a></h3>
803 <!------------------------------------------------------------------------->
805 <p>This directory contains regression tests and source code that is used to
806 test the LLVM infrastructure...</p>
808 <!------------------------------------------------------------------------->
809 <h3><a name="tools"><tt>llvm/tools</tt></a></h3>
810 <!------------------------------------------------------------------------->
812 <p>The <b>tools</b> directory contains the executables built out of the
813 libraries above, which form the main part of the user interface. You can
814 always get help for a tool by typing <tt>tool_name --help</tt>. The
815 following is a brief introduction to the most important tools.</p>
818 <dt><tt><b>as</b></tt><dd>The assembler transforms the human readable
819 LLVM assembly to LLVM bytecode.<p>
821 <dt><tt><b>dis</b></tt><dd>The disassembler transforms the LLVM bytecode
822 to human readable LLVM assembly. Additionally it can convert LLVM
823 bytecode to C, which is enabled with the <tt>-c</tt> option.<p>
825 <dt><tt><b>lli</b></tt><dd> <tt>lli</tt> is the LLVM interpreter, which
826 can directly execute LLVM bytecode (although very slowly...). In addition
827 to a simple interpreter, <tt>lli</tt> is also has debugger and tracing
828 modes (entered by specifying <tt>-debug</tt> or <tt>-trace</tt> on the
829 command line, respectively). Finally, for architectures that support it
830 (currently only x86 and Sparc), by default, <tt>lli</tt> will function as
831 a Just-In-Time compiler (if the functionality was compiled in), and will
832 execute the code <i>much</i> faster than the interpreter.<p>
834 <dt><tt><b>llc</b></tt><dd> <tt>llc</tt> is the LLVM backend compiler,
835 which translates LLVM bytecode to a SPARC or x86 assembly file.<p>
837 <dt><tt><b>llvmgcc</b></tt><dd> <tt>llvmgcc</tt> is a GCC based C frontend
838 that has been retargeted to emit LLVM code as the machine code output. It
839 works just like any other GCC compiler, taking the typical <tt>-c, -S, -E,
840 -o</tt> options that are typically used. The source code for the
841 <tt>llvmgcc</tt> tool is currently not included in the LLVM cvs tree
842 because it is quite large and not very interesting.<p>
845 <dt><tt><b>gccas</b></tt><dd> This tool is invoked by the
846 <tt>llvmgcc</tt> frontend as the "assembler" part of the compiler. This
847 tool actually assembles LLVM assembly to LLVM bytecode,
848 performs a variety of optimizations,
849 and outputs LLVM bytecode. Thus when you invoke <tt>llvmgcc -c x.c -o
850 x.o</tt>, you are causing <tt>gccas</tt> to be run, which writes the
851 <tt>x.o</tt> file (which is an LLVM bytecode file that can be
852 disassembled or manipulated just like any other bytecode file). The
853 command line interface to <tt>gccas</tt> is designed to be as close as
854 possible to the <b>system</b> '<tt>as</tt>' utility so that the gcc
855 frontend itself did not have to be modified to interface to a "weird"
858 <dt><tt><b>gccld</b></tt><dd> <tt>gccld</tt> links together several LLVM
859 bytecode files into one bytecode file and does some optimization. It is
860 the linker invoked by the gcc frontend when multiple .o files need to be
861 linked together. Like <tt>gccas</tt> the command line interface of
862 <tt>gccld</tt> is designed to match the system linker, to aid
863 interfacing with the GCC frontend.<p>
866 <dt><tt><b>opt</b></tt><dd> <tt>opt</tt> reads LLVM bytecode, applies a
867 series of LLVM to LLVM transformations (which are specified on the command
868 line), and then outputs the resultant bytecode. The '<tt>opt --help</tt>'
869 command is a good way to get a list of the program transformations
870 available in LLVM.<p>
873 <dt><tt><b>analyze</b></tt><dd> <tt>analyze</tt> is used to run a specific
874 analysis on an input LLVM bytecode file and print out the results. It is
875 primarily useful for debugging analyses, or familiarizing yourself with
876 what an analysis does.<p>
880 <!------------------------------------------------------------------------->
881 <h3><a name="utils"><tt>llvm/utils</tt></a></h3>
882 <!------------------------------------------------------------------------->
884 This directory contains utilities for working with LLVM sourcecode, and some
885 of the utilities are actually required as part of the build process because
886 they are code generators for parts of LLVM infrastructure.
889 <td><tt><b>Burg/</b></tt><dd> <tt>Burg</tt> is an instruction selector
890 generator -- it builds trees on which it then performs pattern-matching to
891 select instructions according to the patterns the user has specified. Burg
892 is currently used in the Sparc V9 backend.<p>
894 <dt><tt><b>codegen-diff</b></tt><dd> <tt>codegen-diff</tt> is a script
895 that finds differences between code that LLC generates and code that LLI
896 generates. This is a useful tool if you are debugging one of them,
897 assuming that the other generates correct output. For the full user
898 manual, run <tt>`perldoc codegen-diff'</tt>.<p>
900 <dt><tt><b>cvsupdate</b></tt><dd> <tt>cvsupdate</tt> is a script that will
901 update your CVS tree, but produce a much cleaner and more organized output
902 than simply running <tt>`cvs up -dP'</tt> will. For example, it will group
903 together all the new and updated files and modified files in separate
904 sections, so you can see at a glance what has changed. If you are at the
905 top of your LLVM CVS tree, running <tt>utils/cvsupdate</tt> is the
906 preferred way of updating the tree.<p>
908 <dt><tt><b>emacs/</b></tt><dd> The <tt>emacs</tt> directory contains
909 syntax-highlighting files which will work with Emacs and XEmacs editors,
910 providing syntax highlighting support for LLVM assembly files and TableGen
911 description files. For information on how to use the syntax files, consult
912 the <tt>README</tt> file in that directory.<p>
914 <dt><tt><b>getsrcs.sh</b></tt><dd> The <tt>getsrcs.sh</tt> script finds
915 and outputs all non-generated source files, which is useful if one wishes
916 to do a lot of development across directories and does not want to
917 individually find each file. One way to use it is to run, for example:
918 <tt>xemacs `utils/getsources.sh`</tt> from the top of your LLVM source
921 <dt><tt><b>makellvm</b></tt><dd> The <tt>makellvm</tt> script compiles all
922 files in the current directory and then compiles and links the tool that
923 is the first argument. For example, assuming you are in the directory
924 <tt>llvm/lib/Target/Sparc</tt>, if <tt>makellvm</tt> is in your path,
925 simply running <tt>makellvm llc</tt> will make a build of the current
926 directory, switch to directory <tt>llvm/tools/llc</tt> and build it,
927 causing a re-linking of LLC.<p>
929 <dt><tt><b>NightlyTest.pl</b></tt> and
930 <tt><b>NightlyTestTemplate.html</b></tt><dd> These files are used in a
931 cron script to generate nightly status reports of the functionality of
932 tools, and the results can be seen by following the appropriate link on
933 the <a href="http://llvm.cs.uiuc.edu/">LLVM homepage</a>.<p>
935 <dt><tt><b>TableGen/</b></tt><dd> The <tt>TableGen</tt> directory contains
936 the tool used to generate register descriptions, instruction set
937 descriptions, and even assemblers from common TableGen description
940 <dt><tt><b>vim/</b></tt><dd> The <tt>vim</tt> directory contains
941 syntax-highlighting files which will work with the VIM editor, providing
942 syntax highlighting support for LLVM assembly files and TableGen
943 description files. For information on how to use the syntax files, consult
944 the <tt>README</tt> file in that directory.<p>
948 <!--=====================================================================-->
949 <h2><center><a name="cfront">Compiling the LLVM C Front End</center></h2>
950 <!--=====================================================================-->
954 This step is optional if you have the C front end binary distribution for
959 Now that you have the LLVM suite built, you can build the C front end. For
960 those of you that have built GCC before, the process is very similar.
962 Be forewarned, though: the build system for the C front end is not as
963 polished as the rest of the LLVM code, so there will be many warnings and
964 errors that you will need to ignore for now:
967 <li>Ensure that <tt><i>OBJ_ROOT</i>/llvm/tools/Debug</tt> is at the
968 <i>end</i> of your <tt>PATH</tt> environment variable. The front end
969 build needs to know where to find the LLVM tools, but you want to
970 ensure that these tools are not found before the system assembler and
971 linker that you normally use for compilation.
973 <li><tt>cd <i>GCCOBJ</i></tt>
975 <li>Configure the source code:
977 <li>On Linux/x86, use
979 <li><tt><i>GCCSRC</i>/configure --prefix=<i>LLVMGCCDIR</i>
980 --enable-languages=c</tt>
983 <li>On Solaris/Sparc, use
985 <li><tt><i>GCCSRC</i>/configure --prefix=<i>LLVMGCCDIR</i>
986 --enable-languages=c --target=sparcv9-sun-solaris2</tt>
992 <li>The build will eventually fail. Don't worry; chances are good that
993 everything that needed to build is built.
995 <li><tt>gmake -k install</tt>
999 Once this is done, you should have a built front end compiler in
1000 <tt><i>LLVMGCCDIR</i></tt>.
1003 <!--=====================================================================-->
1005 <center><a name="tutorial">An Example Using the LLVM Tool Chain</center>
1007 <!--=====================================================================-->
1010 <li>First, create a simple C file, name it 'hello.c':
1012 #include <stdio.h>
1014 printf("hello world\n");
1019 <li>Next, compile the C file into a LLVM bytecode file:<p>
1021 <tt>% llvmgcc hello.c -o hello</tt><p>
1023 This will create two result files: <tt>hello</tt> and
1024 <tt>hello.bc</tt>. The <tt>hello.bc</tt> is the LLVM bytecode that
1025 corresponds the the compiled program and the library facilities that it
1026 required. <tt>hello</tt> is a simple shell script that runs the bytecode
1027 file with <tt>lli</tt>, making the result directly executable.<p>
1029 <li>Run the program. To make sure the program ran, execute one of the
1030 following commands:<p>
1032 <tt>% ./hello</tt><p>
1036 <tt>% lli hello.bc</tt><p>
1038 <li>Use the <tt>dis</tt> utility to take a look at the LLVM assembly
1041 <tt>% dis < hello.bc | less</tt><p>
1043 <li>Compile the program to native Sparc assembly using the code
1044 generator (assuming you are currently on a Sparc system):<p>
1046 <tt>% llc hello.bc -o hello.s</tt><p>
1048 <li>Assemble the native sparc assemble file into a program:<p>
1050 <tt>% /opt/SUNWspro/bin/cc -xarch=v9 hello.s -o hello.sparc</tt><p>
1052 <li>Execute the native sparc program:<p>
1054 <tt>% ./hello.sparc</tt><p>
1059 <!--=====================================================================-->
1061 <center><a name="problems">Common Problems</a></center>
1063 <!--=====================================================================-->
1065 Below are common problems and their remedies:
1068 <dt><b>When I run configure, it finds the wrong C compiler.</b>
1070 The <tt>configure</tt> script attempts to locate first <tt>gcc</tt> and
1071 then <tt>cc</tt>, unless it finds compiler paths set in <tt>CC</tt> and
1072 <tt>CXX</tt> for the C and C++ compiler, respectively.
1074 If <tt>configure</tt> finds the wrong compiler, either adjust your
1075 <tt>PATH</tt> environment variable or set <tt>CC</tt> and <tt>CXX</tt>
1079 <dt><b>I compile the code, and I get some error about /localhome</b>.
1081 There are several possible causes for this. The first is that you
1082 didn't set a pathname properly when using <tt>configure</tt>, and it
1083 defaulted to a pathname that we use on our research machines.
1085 Another possibility is that we hardcoded a path in our Makefiles. If
1086 you see this, please email the LLVM bug mailing list with the name of
1087 the offending Makefile and a description of what is wrong with it.
1089 <dt><b>The <tt>configure</tt> script finds the right C compiler, but it
1090 uses the LLVM linker from a previous build. What do I do?</b>
1092 The <tt>configure</tt> script uses the <tt>PATH</tt> to find
1093 executables, so if it's grabbing the wrong linker/assembler/etc, there
1094 are two ways to fix it:
1096 <li>Adjust your <tt>PATH</tt> environment variable so that the
1097 correct program appears first in the <tt>PATH</tt>. This may work,
1098 but may not be convenient when you want them <i>first</i> in your
1099 path for other work.
1102 <li>Run <tt>configure</tt> with an alternative <tt>PATH</tt> that
1103 is correct. In a Borne compatible shell, the syntax would be:
1105 <tt>PATH=<the path without the bad program> ./configure ...</tt>
1107 This is still somewhat inconvenient, but it allows
1108 <tt>configure</tt> to do its work without having to adjust your
1109 <tt>PATH</tt> permanently.
1112 <dt><b>I've upgraded to a new version of LLVM, and I get strange build
1115 Sometimes changes to the LLVM source code alters how the build system
1116 works. Changes in libtool, autoconf, or header file dependencies are
1117 especially prone to this sort of problem.
1119 The best thing to try is to remove the old files and re-build. In most
1120 cases, this takes care of the problem. To do this, just type <tt>make
1121 clean</tt> and then <tt>make</tt> in the directory that fails to build.
1126 <!--=====================================================================-->
1127 <h2><center><a name="links">Links</a></center></h2>
1128 <!--=====================================================================-->
1130 <p>This document is just an <b>introduction</b> to how to use LLVM to do
1131 some simple things... there are many more interesting and complicated things
1132 that you can do that aren't documented here (but we'll gladly accept a patch
1133 if you want to write something up!). For more information about LLVM, check
1137 <li><a href="http://llvm.cs.uiuc.edu/">LLVM homepage</a></li>
1138 <li><a href="http://llvm.cs.uiuc.edu/doxygen/">LLVM doxygen tree</a></li>
1139 <li><a href="http://llvm.cs.uiuc.edu/docs/Projects.html">Starting a Project that Uses LLVM</a></li>
1144 If you have any questions or run into any snags (or you have any
1145 additions...), please send an email to
1146 <a href="mailto:sabre@nondot.org">Chris Lattner</a>.</p>
1148 <!-- Created: Mon Jul 1 02:29:02 CDT 2002 -->
1149 <!-- hhmts start -->
1150 Last modified: Mon Aug 11 13:52:22 CDT 2003