<html xmlns="http://www.w3.org/1999/xhtml" xml:lang="en" lang="en">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
-<meta name="generator" content="Docutils 0.4: http://docutils.sourceforge.net/" />
+<meta name="generator" content="Docutils 0.6: http://docutils.sourceforge.net/" />
<title>Customizing LLVMC: Reference Manual</title>
<link rel="stylesheet" href="llvm.css" type="text/css" />
</head>
<body>
<div class="document" id="customizing-llvmc-reference-manual">
+<h1 class="title">Customizing LLVMC: Reference Manual</h1>
-<div class="doc_title">Customizing LLVMC: Reference Manual</div>
-
-<div class="doc_warning">
- <p>Note: This document is a work-in-progress. Additions and clarifications
- are welcome.</p>
+<!-- This file was automatically generated by rst2html.
+Please do not edit directly!
+The ReST source lives in the directory 'tools/llvmc/doc'. -->
+<div class="contents topic" id="contents">
+<p class="topic-title first">Contents</p>
+<ul class="simple">
+<li><a class="reference internal" href="#introduction" id="id7">Introduction</a></li>
+<li><a class="reference internal" href="#compiling-with-llvmc" id="id8">Compiling with <tt class="docutils literal">llvmc</tt></a></li>
+<li><a class="reference internal" href="#predefined-options" id="id9">Predefined options</a></li>
+<li><a class="reference internal" href="#compiling-llvmc-based-drivers" id="id10">Compiling LLVMC-based drivers</a></li>
+<li><a class="reference internal" href="#customizing-llvmc-the-compilation-graph" id="id11">Customizing LLVMC: the compilation graph</a></li>
+<li><a class="reference internal" href="#describing-options" id="id12">Describing options</a></li>
+<li><a class="reference internal" href="#conditional-evaluation" id="id13">Conditional evaluation</a></li>
+<li><a class="reference internal" href="#writing-a-tool-description" id="id14">Writing a tool description</a><ul>
+<li><a class="reference internal" href="#id4" id="id15">Actions</a></li>
+</ul>
+</li>
+<li><a class="reference internal" href="#language-map" id="id16">Language map</a></li>
+<li><a class="reference internal" href="#option-preprocessor" id="id17">Option preprocessor</a></li>
+<li><a class="reference internal" href="#more-advanced-topics" id="id18">More advanced topics</a><ul>
+<li><a class="reference internal" href="#hooks-and-environment-variables" id="id19">Hooks and environment variables</a></li>
+<li><a class="reference internal" href="#debugging" id="id20">Debugging</a></li>
+<li><a class="reference internal" href="#conditioning-on-the-executable-name" id="id21">Conditioning on the executable name</a></li>
+</ul>
+</li>
+</ul>
</div>
-
+<div class="doc_author">
+<p>Written by <a href="mailto:foldr@codedgers.com">Mikhail Glushenkov</a></p>
+</div><div class="section" id="introduction">
+<h1><a class="toc-backref" href="#id7">Introduction</a></h1>
<p>LLVMC is a generic compiler driver, designed to be customizable and
-extensible. It plays the same role for LLVM as the <tt class="docutils literal"><span class="pre">gcc</span></tt> program
-does for GCC - LLVMC's job is essentially to transform a set of input
-files into a set of targets depending on configuration rules and user
-options. What makes LLVMC different is that these transformation rules
-are completely customizable - in fact, LLVMC knows nothing about the
-specifics of transformation (even the command-line options are mostly
-not hard-coded) and regards the transformation structure as an
-abstract graph. This makes it possible to adapt LLVMC for other
+extensible. It plays the same role for LLVM as the <tt class="docutils literal">gcc</tt> program does for
+GCC - LLVMC's job is essentially to transform a set of input files into a set of
+targets depending on configuration rules and user options. What makes LLVMC
+different is that these transformation rules are completely customizable - in
+fact, LLVMC knows nothing about the specifics of transformation (even the
+command-line options are mostly not hard-coded) and regards the transformation
+structure as an abstract graph. The structure of this graph is described in
+high-level TableGen code, from which an efficient C++ representation is
+automatically derived. This makes it possible to adapt LLVMC for other
purposes - for example, as a build tool for game resources.</p>
-<p>Because LLVMC employs TableGen <a class="footnote-reference" href="#id2" id="id1" name="id1">[1]</a> as its configuration language, you
+<p>Because LLVMC employs <a class="reference external" href="http://llvm.org/docs/TableGenFundamentals.html">TableGen</a> as its configuration language, you
need to be familiar with it to customize LLVMC.</p>
-<div class="contents topic">
-<ul class="simple">
-<li><a class="reference" href="#compiling-with-llvmc" id="id3" name="id3">Compiling with LLVMC</a></li>
-<li><a class="reference" href="#predefined-options" id="id4" name="id4">Predefined options</a></li>
-<li><a class="reference" href="#customizing-llvmc-the-compilation-graph" id="id5" name="id5">Customizing LLVMC: the compilation graph</a></li>
-<li><a class="reference" href="#writing-a-tool-description" id="id6" name="id6">Writing a tool description</a></li>
-<li><a class="reference" href="#option-list-specifying-all-options-in-a-single-place" id="id7" name="id7">Option list - specifying all options in a single place</a></li>
-<li><a class="reference" href="#using-hooks-and-environment-variables-in-the-cmd-line-property" id="id8" name="id8">Using hooks and environment variables in the <tt class="docutils literal"><span class="pre">cmd_line</span></tt> property</a></li>
-<li><a class="reference" href="#conditional-evaluation-the-case-expression" id="id9" name="id9">Conditional evaluation: the <tt class="docutils literal"><span class="pre">case</span></tt> expression</a></li>
-<li><a class="reference" href="#language-map" id="id10" name="id10">Language map</a></li>
-<li><a class="reference" href="#references" id="id11" name="id11">References</a></li>
-</ul>
</div>
-
-<div class="doc_author">Written by Mikhail Glushenkov</div>
-
-<div class="doc_text">
-<div class="doc_section"><a class="toc-backref" href="#id3" id="compiling-with-llvmc" name="compiling-with-llvmc">Compiling with LLVMC</a></div>
-<p>LLVMC tries hard to be as compatible with <tt class="docutils literal"><span class="pre">gcc</span></tt> as possible,
+<div class="section" id="compiling-with-llvmc">
+<h1><a class="toc-backref" href="#id8">Compiling with <tt class="docutils literal">llvmc</tt></a></h1>
+<p>LLVMC tries hard to be as compatible with <tt class="docutils literal">gcc</tt> as possible,
although there are some small differences. Most of the time, however,
you shouldn't be able to notice them:</p>
<pre class="literal-block">
$ # This works as expected:
-$ llvmc2 -O3 -Wall hello.cpp
+$ llvmc -O3 -Wall hello.cpp
$ ./a.out
hello
</pre>
-<p>One nice feature of LLVMC is that one doesn't have to distinguish
-between different compilers for different languages (think <tt class="docutils literal"><span class="pre">g++</span></tt> and
-<tt class="docutils literal"><span class="pre">gcc</span></tt>) - the right toolchain is chosen automatically based on input
-language names (which are, in turn, determined from file
-extensions). If you want to force files ending with ".c" to compile as
-C++, use the <tt class="docutils literal"><span class="pre">-x</span></tt> option, just like you would do it with <tt class="docutils literal"><span class="pre">gcc</span></tt>:</p>
+<p>One nice feature of LLVMC is that one doesn't have to distinguish between
+different compilers for different languages (think <tt class="docutils literal">g++</tt> vs. <tt class="docutils literal">gcc</tt>) - the
+right toolchain is chosen automatically based on input language names (which
+are, in turn, determined from file extensions). If you want to force files
+ending with ".c" to compile as C++, use the <tt class="docutils literal"><span class="pre">-x</span></tt> option, just like you would
+do it with <tt class="docutils literal">gcc</tt>:</p>
<pre class="literal-block">
-$ llvmc2 -x c hello.cpp
-$ # hello.cpp is really a C file
+$ # hello.c is really a C++ file
+$ llvmc -x c++ hello.c
$ ./a.out
hello
</pre>
object files you should provide the <tt class="docutils literal"><span class="pre">--linker</span></tt> option since it's
impossible for LLVMC to choose the right linker in that case:</p>
<pre class="literal-block">
-$ llvmc2 -c hello.cpp
-$ llvmc2 hello.o
+$ llvmc -c hello.cpp
+$ llvmc hello.o
[A lot of link-time errors skipped]
-$ llvmc2 --linker=c++ hello.o
+$ llvmc --linker=c++ hello.o
$ ./a.out
hello
</pre>
+<p>By default, LLVMC uses <tt class="docutils literal"><span class="pre">llvm-gcc</span></tt> to compile the source code. It is also
+possible to choose the <tt class="docutils literal">clang</tt> compiler with the <tt class="docutils literal"><span class="pre">-clang</span></tt> option.</p>
</div>
-<div class="doc_text">
-<div class="doc_section"><a class="toc-backref" href="#id4" id="predefined-options" name="predefined-options">Predefined options</a></div>
-<p>LLVMC has some built-in options that can't be overridden in the
-configuration files:</p>
+<div class="section" id="predefined-options">
+<h1><a class="toc-backref" href="#id9">Predefined options</a></h1>
+<p>LLVMC has some built-in options that can't be overridden in the TableGen code:</p>
<ul class="simple">
-<li><tt class="docutils literal"><span class="pre">-o</span> <span class="pre">FILE</span></tt> - Output file name.</li>
-<li><tt class="docutils literal"><span class="pre">-x</span> <span class="pre">LANGUAGE</span></tt> - Specify the language of the following input files
+<li><tt class="docutils literal"><span class="pre">-o</span> FILE</tt> - Output file name.</li>
+<li><tt class="docutils literal"><span class="pre">-x</span> LANGUAGE</tt> - Specify the language of the following input files
until the next -x option.</li>
<li><tt class="docutils literal"><span class="pre">-v</span></tt> - Enable verbose mode, i.e. print out all executed commands.</li>
-<li><tt class="docutils literal"><span class="pre">--view-graph</span></tt> - Show a graphical representation of the compilation
-graph. Requires that you have <tt class="docutils literal"><span class="pre">dot</span></tt> and <tt class="docutils literal"><span class="pre">gv</span></tt> commands
-installed. Hidden option, useful for debugging.</li>
-<li><tt class="docutils literal"><span class="pre">--write-graph</span></tt> - Write a <tt class="docutils literal"><span class="pre">compilation-graph.dot</span></tt> file in the
-current directory with the compilation graph description in the
-Graphviz format. Hidden option, useful for debugging.</li>
-<li><tt class="docutils literal"><span class="pre">--save-temps</span></tt> - Write temporary files to the current directory
-and do not delete them on exit. Hidden option, useful for debugging.</li>
+<li><tt class="docutils literal"><span class="pre">--save-temps</span></tt> - Write temporary files to the current directory and do not
+delete them on exit. This option can also take an argument: the
+<tt class="docutils literal"><span class="pre">--save-temps=obj</span></tt> switch will write files into the directory specified with
+the <tt class="docutils literal"><span class="pre">-o</span></tt> option. The <tt class="docutils literal"><span class="pre">--save-temps=cwd</span></tt> and <tt class="docutils literal"><span class="pre">--save-temps</span></tt> switches are
+both synonyms for the default behaviour.</li>
+<li><tt class="docutils literal"><span class="pre">--temp-dir</span> DIRECTORY</tt> - Store temporary files in the given directory. This
+directory is deleted on exit unless <tt class="docutils literal"><span class="pre">--save-temps</span></tt> is specified. If
+<tt class="docutils literal"><span class="pre">--save-temps=obj</span></tt> is also specified, <tt class="docutils literal"><span class="pre">--temp-dir</span></tt> is given the
+precedence.</li>
+<li><tt class="docutils literal"><span class="pre">--check-graph</span></tt> - Check the compilation for common errors like mismatched
+output/input language names, multiple default edges and cycles. Exit with code
+zero if no errors were found, and return the number of found errors
+otherwise. Hidden option, useful for debugging.</li>
+<li><tt class="docutils literal"><span class="pre">--view-graph</span></tt> - Show a graphical representation of the compilation graph
+and exit. Requires that you have <tt class="docutils literal">dot</tt> and <tt class="docutils literal">gv</tt> programs installed. Hidden
+option, useful for debugging.</li>
+<li><tt class="docutils literal"><span class="pre">--write-graph</span></tt> - Write a <tt class="docutils literal"><span class="pre">compilation-graph.dot</span></tt> file in the current
+directory with the compilation graph description in Graphviz format (identical
+to the file used by the <tt class="docutils literal"><span class="pre">--view-graph</span></tt> option). The <tt class="docutils literal"><span class="pre">-o</span></tt> option can be
+used to set the output file name. Hidden option, useful for debugging.</li>
<li><tt class="docutils literal"><span class="pre">--help</span></tt>, <tt class="docutils literal"><span class="pre">--help-hidden</span></tt>, <tt class="docutils literal"><span class="pre">--version</span></tt> - These options have
their standard meaning.</li>
</ul>
</div>
-<div class="doc_text">
-<div class="doc_section"><a class="toc-backref" href="#id5" id="customizing-llvmc-the-compilation-graph" name="customizing-llvmc-the-compilation-graph">Customizing LLVMC: the compilation graph</a></div>
-<p>At the time of writing LLVMC does not support on-the-fly reloading of
-configuration, so to customize LLVMC you'll have to recompile the
-source code (which lives under <tt class="docutils literal"><span class="pre">$LLVM_DIR/tools/llvmc2</span></tt>). The
-default configuration files are <tt class="docutils literal"><span class="pre">Common.td</span></tt> (contains common
-definitions, don't forget to <tt class="docutils literal"><span class="pre">include</span></tt> it in your configuration
-files), <tt class="docutils literal"><span class="pre">Tools.td</span></tt> (tool descriptions) and <tt class="docutils literal"><span class="pre">Graph.td</span></tt> (compilation
-graph definition).</p>
-<p>To compile LLVMC with your own configuration file (say,``MyGraph.td``),
-run <tt class="docutils literal"><span class="pre">make</span></tt> like this:</p>
+<div class="section" id="compiling-llvmc-based-drivers">
+<h1><a class="toc-backref" href="#id10">Compiling LLVMC-based drivers</a></h1>
+<p>It's easiest to start working on your own LLVMC driver by copying the skeleton
+project which lives under <tt class="docutils literal">$LLVMC_DIR/examples/Skeleton</tt>:</p>
+<pre class="literal-block">
+$ cd $LLVMC_DIR/examples
+$ cp -r Skeleton MyDriver
+$ cd MyDriver
+$ ls
+AutoGenerated.td Hooks.cpp Main.cpp Makefile
+</pre>
+<p>As you can see, our basic driver consists of only three files (not counting the
+build script). <tt class="docutils literal">AutoGenerated.td</tt> contains TableGen description of the
+compilation graph; its format is documented in the following
+sections. <tt class="docutils literal">Hooks.cpp</tt> is an empty file that should be used for hook
+definitions (see <a class="reference internal" href="#hooks">below</a>). <tt class="docutils literal">Main.cpp</tt> is just a helper used to compile the
+auto-generated C++ code produced from TableGen source.</p>
+<p>The first thing that you should do is to change the <tt class="docutils literal">LLVMC_BASED_DRIVER</tt>
+variable in the <tt class="docutils literal">Makefile</tt>:</p>
<pre class="literal-block">
-$ cd $LLVM_DIR/tools/llvmc2
-$ make GRAPH=MyGraph.td TOOLNAME=my_llvmc
+LLVMC_BASED_DRIVER=MyDriver
</pre>
-<p>This will build an executable named <tt class="docutils literal"><span class="pre">my_llvmc</span></tt>. There are also
-several sample configuration files in the <tt class="docutils literal"><span class="pre">llvmc2/examples</span></tt>
-subdirectory that should help to get you started.</p>
-<p>Internally, LLVMC stores information about possible source
-transformations in form of a graph. Nodes in this graph represent
-tools, and edges between two nodes represent a transformation path. A
-special "root" node is used to mark entry points for the
-transformations. LLVMC also assigns a weight to each edge (more on
-this later) to choose between several alternative edges.</p>
-<p>The definition of the compilation graph (see file <tt class="docutils literal"><span class="pre">Graph.td</span></tt>) is
-just a list of edges:</p>
+<p>It can also be a good idea to put your TableGen code into a file with a less
+generic name:</p>
+<pre class="literal-block">
+$ touch MyDriver.td
+$ vim AutoGenerated.td
+[...]
+include "MyDriver.td"
+</pre>
+<p>If you have more than one TableGen source file, they all should be included from
+<tt class="docutils literal">AutoGenerated.td</tt>, since this file is used by the build system to generate
+C++ code.</p>
+<p>To build your driver, just <tt class="docutils literal">cd</tt> to its source directory and run <tt class="docutils literal">make</tt>. The
+resulting executable will be put into <tt class="docutils literal"><span class="pre">$LLVM_OBJ_DIR/$(BuildMode)/bin</span></tt>.</p>
+<p>If you're compiling LLVM with different source and object directories, then you
+must perform the following additional steps before running <tt class="docutils literal">make</tt>:</p>
+<pre class="literal-block">
+# LLVMC_SRC_DIR = $LLVM_SRC_DIR/tools/llvmc/
+# LLVMC_OBJ_DIR = $LLVM_OBJ_DIR/tools/llvmc/
+$ mkdir $LLVMC_OBJ_DIR/examples/MyDriver/
+$ cp $LLVMC_SRC_DIR/examples/MyDriver/Makefile \
+ $LLVMC_OBJ_DIR/examples/MyDriver/
+$ cd $LLVMC_OBJ_DIR/examples/MyDriver
+$ make
+</pre>
+</div>
+<div class="section" id="customizing-llvmc-the-compilation-graph">
+<h1><a class="toc-backref" href="#id11">Customizing LLVMC: the compilation graph</a></h1>
+<p>Each TableGen configuration file should include the common definitions:</p>
+<pre class="literal-block">
+include "llvm/CompilerDriver/Common.td"
+</pre>
+<p>Internally, LLVMC stores information about possible source transformations in
+form of a graph. Nodes in this graph represent tools, and edges between two
+nodes represent a transformation path. A special "root" node is used to mark
+entry points for the transformations. LLVMC also assigns a weight to each edge
+(more on this later) to choose between several alternative edges.</p>
+<p>The definition of the compilation graph (see file <tt class="docutils literal">llvmc/src/Base.td</tt> for an
+example) is just a list of edges:</p>
<pre class="literal-block">
def CompilationGraph : CompilationGraph<[
- Edge<root, llvm_gcc_c>,
- Edge<root, llvm_gcc_assembler>,
+ Edge<"root", "llvm_gcc_c">,
+ Edge<"root", "llvm_gcc_assembler">,
...
- Edge<llvm_gcc_c, llc>,
- Edge<llvm_gcc_cpp, llc>,
+ Edge<"llvm_gcc_c", "llc">,
+ Edge<"llvm_gcc_cpp", "llc">,
...
- OptionalEdge<llvm_gcc_c, opt, [(switch_on "opt")]>,
- OptionalEdge<llvm_gcc_cpp, opt, [(switch_on "opt")]>,
+ OptionalEdge<"llvm_gcc_c", "opt", (case (switch_on "opt"),
+ (inc_weight))>,
+ OptionalEdge<"llvm_gcc_cpp", "opt", (case (switch_on "opt"),
+ (inc_weight))>,
...
- OptionalEdge<llvm_gcc_assembler, llvm_gcc_cpp_linker,
+ OptionalEdge<"llvm_gcc_assembler", "llvm_gcc_cpp_linker",
(case (input_languages_contain "c++"), (inc_weight),
(or (parameter_equals "linker", "g++"),
(parameter_equals "linker", "c++")), (inc_weight))>,
]>;
</pre>
-<p>As you can see, the edges can be either default or optional, where
-optional edges are differentiated by sporting a <tt class="docutils literal"><span class="pre">case</span></tt> expression
-used to calculate the edge's weight.</p>
-<p>The default edges are assigned a weight of 1, and optional edges get a
-weight of 0 + 2*N where N is the number of tests that evaluated to
-true in the <tt class="docutils literal"><span class="pre">case</span></tt> expression. It is also possible to provide an
-integer parameter to <tt class="docutils literal"><span class="pre">inc_weight</span></tt> and <tt class="docutils literal"><span class="pre">dec_weight</span></tt> - in this case,
-the weight is increased (or decreased) by the provided value instead
-of the default 2.</p>
-<p>When passing an input file through the graph, LLVMC picks the edge
-with the maximum weight. To avoid ambiguity, there should be only one
-default edge between two nodes (with the exception of the root node,
-which gets a special treatment - there you are allowed to specify one
-default edge <em>per language</em>).</p>
-<p>To get a visual representation of the compilation graph (useful for
-debugging), run <tt class="docutils literal"><span class="pre">llvmc2</span> <span class="pre">--view-graph</span></tt>. You will need <tt class="docutils literal"><span class="pre">dot</span></tt> and
-<tt class="docutils literal"><span class="pre">gsview</span></tt> installed for this to work properly.</p>
+<p>As you can see, the edges can be either default or optional, where optional
+edges are differentiated by an additional <tt class="docutils literal">case</tt> expression used to calculate
+the weight of this edge. Notice also that we refer to tools via their names (as
+strings). This makes it possible to add edges to an existing compilation graph
+without having to know about all tool definitions used in the graph.</p>
+<p>The default edges are assigned a weight of 1, and optional edges get a weight of
+0 + 2*N where N is the number of tests that evaluated to true in the <tt class="docutils literal">case</tt>
+expression. It is also possible to provide an integer parameter to
+<tt class="docutils literal">inc_weight</tt> and <tt class="docutils literal">dec_weight</tt> - in this case, the weight is increased (or
+decreased) by the provided value instead of the default 2. Default weight of an
+optional edge can be changed by using the <tt class="docutils literal">default</tt> clause of the <tt class="docutils literal">case</tt>
+construct.</p>
+<p>When passing an input file through the graph, LLVMC picks the edge with the
+maximum weight. To avoid ambiguity, there should be only one default edge
+between two nodes (with the exception of the root node, which gets a special
+treatment - there you are allowed to specify one default edge <em>per language</em>).</p>
+<p>When multiple compilation graphs are defined, they are merged together. Multiple
+edges with the same end nodes are not allowed (i.e. the graph is not a
+multigraph), and will lead to a compile-time error.</p>
+<p>To get a visual representation of the compilation graph (useful for debugging),
+run <tt class="docutils literal">llvmc <span class="pre">--view-graph</span></tt>. You will need <tt class="docutils literal">dot</tt> and <tt class="docutils literal">gsview</tt> installed for
+this to work properly.</p>
</div>
-<div class="doc_text">
-<div class="doc_section"><a class="toc-backref" href="#id6" id="writing-a-tool-description" name="writing-a-tool-description">Writing a tool description</a></div>
-<p>As was said earlier, nodes in the compilation graph represent tools,
-which are described separately. A tool definition looks like this
-(taken from the <tt class="docutils literal"><span class="pre">Tools.td</span></tt> file):</p>
+<div class="section" id="describing-options">
+<h1><a class="toc-backref" href="#id12">Describing options</a></h1>
+<p>Command-line options supported by the driver are defined by using an
+<tt class="docutils literal">OptionList</tt>:</p>
<pre class="literal-block">
-def llvm_gcc_cpp : Tool<[
- (in_language "c++"),
- (out_language "llvm-assembler"),
- (output_suffix "bc"),
- (cmd_line "llvm-g++ -c $INFILE -o $OUTFILE -emit-llvm"),
- (sink)
- ]>;
-</pre>
-<p>This defines a new tool called <tt class="docutils literal"><span class="pre">llvm_gcc_cpp</span></tt>, which is an alias for
-<tt class="docutils literal"><span class="pre">llvm-g++</span></tt>. As you can see, a tool definition is just a list of
-properties; most of them should be self-explanatory. The <tt class="docutils literal"><span class="pre">sink</span></tt>
-property means that this tool should be passed all command-line
-options that lack explicit descriptions.</p>
-<p>The complete list of the currently implemented tool properties follows:</p>
-<ul class="simple">
-<li>Possible tool properties:<ul>
-<li><tt class="docutils literal"><span class="pre">in_language</span></tt> - input language name.</li>
-<li><tt class="docutils literal"><span class="pre">out_language</span></tt> - output language name.</li>
-<li><tt class="docutils literal"><span class="pre">output_suffix</span></tt> - output file suffix.</li>
-<li><tt class="docutils literal"><span class="pre">cmd_line</span></tt> - the actual command used to run the tool. You can
-use <tt class="docutils literal"><span class="pre">$INFILE</span></tt> and <tt class="docutils literal"><span class="pre">$OUTFILE</span></tt> variables, output redirection
-with <tt class="docutils literal"><span class="pre">></span></tt>, hook invocations (<tt class="docutils literal"><span class="pre">$CALL</span></tt>), environment variables
-(via <tt class="docutils literal"><span class="pre">$ENV</span></tt>) and the <tt class="docutils literal"><span class="pre">case</span></tt> construct (more on this below).</li>
-<li><tt class="docutils literal"><span class="pre">join</span></tt> - this tool is a "join node" in the graph, i.e. it gets a
-list of input files and joins them together. Used for linkers.</li>
-<li><tt class="docutils literal"><span class="pre">sink</span></tt> - all command-line options that are not handled by other
-tools are passed to this tool.</li>
-</ul>
-</li>
-</ul>
-<p>The next tool definition is slightly more complex:</p>
-<pre class="literal-block">
-def llvm_gcc_linker : Tool<[
- (in_language "object-code"),
- (out_language "executable"),
- (output_suffix "out"),
- (cmd_line "llvm-gcc $INFILE -o $OUTFILE"),
- (join),
- (prefix_list_option "L", (forward),
- (help "add a directory to link path")),
- (prefix_list_option "l", (forward),
- (help "search a library when linking")),
- (prefix_list_option "Wl", (unpack_values),
- (help "pass options to linker"))
- ]>;
+def Options : OptionList<[
+(switch_option "E", (help "Help string")),
+(alias_option "quiet", "q")
+...
+]>;
</pre>
-<p>This tool has a "join" property, which means that it behaves like a
-linker. This tool also defines several command-line options: <tt class="docutils literal"><span class="pre">-l</span></tt>,
-<tt class="docutils literal"><span class="pre">-L</span></tt> and <tt class="docutils literal"><span class="pre">-Wl</span></tt> which have their usual meaning. An option has two
-attributes: a name and a (possibly empty) list of properties. All
-currently implemented option types and properties are described below:</p>
+<p>As you can see, the option list is just a list of DAGs, where each DAG is an
+option description consisting of the option name and some properties. More than
+one option list can be defined (they are all merged together in the end), which
+can be handy if one wants to separate option groups syntactically.</p>
<ul>
<li><p class="first">Possible option types:</p>
<blockquote>
<ul class="simple">
-<li><tt class="docutils literal"><span class="pre">switch_option</span></tt> - a simple boolean switch, for example <tt class="docutils literal"><span class="pre">-time</span></tt>.</li>
-<li><tt class="docutils literal"><span class="pre">parameter_option</span></tt> - option that takes an argument, for example
-<tt class="docutils literal"><span class="pre">-std=c99</span></tt>;</li>
-<li><tt class="docutils literal"><span class="pre">parameter_list_option</span></tt> - same as the above, but more than one
-occurence of the option is allowed.</li>
-<li><tt class="docutils literal"><span class="pre">prefix_option</span></tt> - same as the parameter_option, but the option name
-and parameter value are not separated.</li>
-<li><tt class="docutils literal"><span class="pre">prefix_list_option</span></tt> - same as the above, but more than one
-occurence of the option is allowed; example: <tt class="docutils literal"><span class="pre">-lm</span> <span class="pre">-lpthread</span></tt>.</li>
-<li><tt class="docutils literal"><span class="pre">alias_option</span></tt> - a special option type for creating
-aliases. Unlike other option types, aliases are not allowed to
-have any properties besides the aliased option name. Usage
-example: <tt class="docutils literal"><span class="pre">(alias_option</span> <span class="pre">"preprocess",</span> <span class="pre">"E")</span></tt></li>
+<li><tt class="docutils literal">switch_option</tt> - a simple boolean switch without arguments, for example
+<tt class="docutils literal"><span class="pre">-O2</span></tt> or <tt class="docutils literal"><span class="pre">-time</span></tt>. At most one occurrence is allowed by default.</li>
+<li><tt class="docutils literal">parameter_option</tt> - option that takes one argument, for example
+<tt class="docutils literal"><span class="pre">-std=c99</span></tt>. It is also allowed to use spaces instead of the equality
+sign: <tt class="docutils literal"><span class="pre">-std</span> c99</tt>. At most one occurrence is allowed.</li>
+<li><tt class="docutils literal">parameter_list_option</tt> - same as the above, but more than one option
+occurrence is allowed.</li>
+<li><tt class="docutils literal">prefix_option</tt> - same as the parameter_option, but the option name and
+argument do not have to be separated. Example: <tt class="docutils literal"><span class="pre">-ofile</span></tt>. This can be also
+specified as <tt class="docutils literal"><span class="pre">-o</span> file</tt>; however, <tt class="docutils literal"><span class="pre">-o=file</span></tt> will be parsed incorrectly
+(<tt class="docutils literal">=file</tt> will be interpreted as option value). At most one occurrence is
+allowed.</li>
+<li><tt class="docutils literal">prefix_list_option</tt> - same as the above, but more than one occurrence of
+the option is allowed; example: <tt class="docutils literal"><span class="pre">-lm</span> <span class="pre">-lpthread</span></tt>.</li>
+<li><tt class="docutils literal">alias_option</tt> - a special option type for creating aliases. Unlike other
+option types, aliases are not allowed to have any properties besides the
+aliased option name.
+Usage example: <tt class="docutils literal">(alias_option "preprocess", "E")</tt></li>
+<li><tt class="docutils literal">switch_list_option</tt> - like <tt class="docutils literal">switch_option</tt> with the <tt class="docutils literal">zero_or_more</tt>
+property, but remembers how many times the switch was turned on. Useful
+mostly for forwarding. Example: when <tt class="docutils literal"><span class="pre">-foo</span></tt> is a switch option (with the
+<tt class="docutils literal">zero_or_more</tt> property), the command <tt class="docutils literal">driver <span class="pre">-foo</span> <span class="pre">-foo</span></tt> is forwarded
+as <tt class="docutils literal"><span class="pre">some-tool</span> <span class="pre">-foo</span></tt>, but when <tt class="docutils literal"><span class="pre">-foo</span></tt> is a switch list, the same command
+is forwarded as <tt class="docutils literal"><span class="pre">some-tool</span> <span class="pre">-foo</span> <span class="pre">-foo</span></tt>.</li>
</ul>
</blockquote>
</li>
<li><p class="first">Possible option properties:</p>
<blockquote>
<ul class="simple">
-<li><tt class="docutils literal"><span class="pre">append_cmd</span></tt> - append a string to the tool invocation command.</li>
-<li><tt class="docutils literal"><span class="pre">forward</span></tt> - forward this option unchanged.</li>
-<li><tt class="docutils literal"><span class="pre">output_suffix</span></tt> - modify the output suffix of this
-tool. Example : <tt class="docutils literal"><span class="pre">(switch</span> <span class="pre">"E",</span> <span class="pre">(output_suffix</span> <span class="pre">"i")</span></tt>.</li>
-<li><tt class="docutils literal"><span class="pre">stop_compilation</span></tt> - stop compilation after this phase.</li>
-<li><tt class="docutils literal"><span class="pre">unpack_values</span></tt> - used for for splitting and forwarding
-comma-separated lists of options, e.g. <tt class="docutils literal"><span class="pre">-Wa,-foo=bar,-baz</span></tt> is
-converted to <tt class="docutils literal"><span class="pre">-foo=bar</span> <span class="pre">-baz</span></tt> and appended to the tool invocation
-command.</li>
-<li><tt class="docutils literal"><span class="pre">help</span></tt> - help string associated with this option. Used for
-<tt class="docutils literal"><span class="pre">--help</span></tt> output.</li>
-<li><tt class="docutils literal"><span class="pre">required</span></tt> - this option is obligatory.</li>
+<li><tt class="docutils literal">help</tt> - help string associated with this option. Used for <tt class="docutils literal"><span class="pre">--help</span></tt>
+output.</li>
+<li><tt class="docutils literal">required</tt> - this option must be specified exactly once (or, in case of
+the list options without the <tt class="docutils literal">multi_val</tt> property, at least
+once). Incompatible with <tt class="docutils literal">optional</tt> and <tt class="docutils literal">one_or_more</tt>.</li>
+<li><tt class="docutils literal">optional</tt> - the option can be specified either zero times or exactly
+once. The default for switch options. Useful only for list options in
+conjunction with <tt class="docutils literal">multi_val</tt>. Incompatible with <tt class="docutils literal">required</tt>,
+<tt class="docutils literal">zero_or_more</tt> and <tt class="docutils literal">one_or_more</tt>.</li>
+<li><tt class="docutils literal">one_or_more</tt> - the option must be specified at least once. Can be useful
+to allow switch options be both obligatory and be specified multiple
+times. For list options is useful only in conjunction with <tt class="docutils literal">multi_val</tt>;
+for ordinary it is synonymous with <tt class="docutils literal">required</tt>. Incompatible with
+<tt class="docutils literal">required</tt>, <tt class="docutils literal">optional</tt> and <tt class="docutils literal">zero_or_more</tt>.</li>
+<li><tt class="docutils literal">zero_or_more</tt> - the option can be specified zero or more times. Useful
+to allow a single switch option to be specified more than
+once. Incompatible with <tt class="docutils literal">required</tt>, <tt class="docutils literal">optional</tt> and <tt class="docutils literal">one_or_more</tt>.</li>
+<li><tt class="docutils literal">hidden</tt> - the description of this option will not appear in
+the <tt class="docutils literal"><span class="pre">--help</span></tt> output (but will appear in the <tt class="docutils literal"><span class="pre">--help-hidden</span></tt>
+output).</li>
+<li><tt class="docutils literal">really_hidden</tt> - the option will not be mentioned in any help
+output.</li>
+<li><tt class="docutils literal">comma_separated</tt> - Indicates that any commas specified for an option's
+value should be used to split the value up into multiple values for the
+option. This property is valid only for list options. In conjunction with
+<tt class="docutils literal">forward_value</tt> can be used to implement option forwarding in style of
+gcc's <tt class="docutils literal"><span class="pre">-Wa,</span></tt>.</li>
+<li><tt class="docutils literal">multi_val n</tt> - this option takes <em>n</em> arguments (can be useful in some
+special cases). Usage example: <tt class="docutils literal">(parameter_list_option "foo", (multi_val
+3))</tt>; the command-line syntax is '-foo a b c'. Only list options can have
+this attribute; you can, however, use the <tt class="docutils literal">one_or_more</tt>, <tt class="docutils literal">optional</tt>
+and <tt class="docutils literal">required</tt> properties.</li>
+<li><tt class="docutils literal">init</tt> - this option has a default value, either a string (if it is a
+parameter), or a boolean (if it is a switch; as in C++, boolean constants
+are called <tt class="docutils literal">true</tt> and <tt class="docutils literal">false</tt>). List options can't have <tt class="docutils literal">init</tt>
+attribute.
+Usage examples: <tt class="docutils literal">(switch_option "foo", (init true))</tt>; <tt class="docutils literal">(prefix_option
+"bar", (init <span class="pre">"baz"))</span></tt>.</li>
</ul>
</blockquote>
</li>
</ul>
</div>
-<div class="doc_text">
-<div class="doc_section"><a class="toc-backref" href="#id7" id="option-list-specifying-all-options-in-a-single-place" name="option-list-specifying-all-options-in-a-single-place">Option list - specifying all options in a single place</a></div>
-<p>It can be handy to have all information about options gathered in a
-single place to provide an overview. This can be achieved by using a
-so-called <tt class="docutils literal"><span class="pre">OptionList</span></tt>:</p>
-<pre class="literal-block">
-def Options : OptionList<[
-(switch_option "E", (help "Help string")),
-(alias_option "quiet", "q")
-...
-]>;
-</pre>
-<p><tt class="docutils literal"><span class="pre">OptionList</span></tt> is also a good place to specify option aliases.</p>
-<p>Tool-specific option properties like <tt class="docutils literal"><span class="pre">append_cmd</span></tt> have (obviously)
-no meaning in the context of <tt class="docutils literal"><span class="pre">OptionList</span></tt>, so the only properties
-allowed there are <tt class="docutils literal"><span class="pre">help</span></tt> and <tt class="docutils literal"><span class="pre">required</span></tt>.</p>
-<p>Option lists are used at the file scope. See file
-<tt class="docutils literal"><span class="pre">examples/Clang.td</span></tt> for an example of <tt class="docutils literal"><span class="pre">OptionList</span></tt> usage.</p>
-</div>
-<div class="doc_text">
-<div class="doc_section"><a class="toc-backref" href="#id8" id="using-hooks-and-environment-variables-in-the-cmd-line-property" name="using-hooks-and-environment-variables-in-the-cmd-line-property">Using hooks and environment variables in the <tt class="docutils literal"><span class="pre">cmd_line</span></tt> property</a></div>
-<p>Normally, LLVMC executes programs from the system <tt class="docutils literal"><span class="pre">PATH</span></tt>. Sometimes,
-this is not sufficient: for example, we may want to specify tool names
-in the configuration file. This can be achieved via the mechanism of
-hooks - to compile LLVMC with your hooks, just drop a .cpp file into
-<tt class="docutils literal"><span class="pre">tools/llvmc2</span></tt> directory. Hooks should live in the <tt class="docutils literal"><span class="pre">hooks</span></tt>
-namespace and have the signature <tt class="docutils literal"><span class="pre">std::string</span> <span class="pre">hooks::MyHookName</span>
-<span class="pre">(void)</span></tt>. They can be used from the <tt class="docutils literal"><span class="pre">cmd_line</span></tt> tool property:</p>
-<pre class="literal-block">
-(cmd_line "$CALL(MyHook)/path/to/file -o $CALL(AnotherHook)")
-</pre>
-<p>It is also possible to use environment variables in the same manner:</p>
-<pre class="literal-block">
-(cmd_line "$ENV(VAR1)/path/to/file -o $ENV(VAR2)")
-</pre>
-<p>To change the command line string based on user-provided options use
-the <tt class="docutils literal"><span class="pre">case</span></tt> expression (documented below):</p>
-<pre class="literal-block">
-(cmd_line
- (case
- (switch_on "E"),
- "llvm-g++ -E -x c $INFILE -o $OUTFILE",
- (default),
- "llvm-g++ -c -x c $INFILE -o $OUTFILE -emit-llvm"))
-</pre>
-</div>
-<div class="doc_text">
-<div class="doc_section"><a class="toc-backref" href="#id9" id="conditional-evaluation-the-case-expression" name="conditional-evaluation-the-case-expression">Conditional evaluation: the <tt class="docutils literal"><span class="pre">case</span></tt> expression</a></div>
-<p>The 'case' construct can be used to calculate weights of the optional
-edges and to choose between several alternative command line strings
-in the <tt class="docutils literal"><span class="pre">cmd_line</span></tt> tool property. It is designed after the
-similarly-named construct in functional languages and takes the form
-<tt class="docutils literal"><span class="pre">(case</span> <span class="pre">(test_1),</span> <span class="pre">statement_1,</span> <span class="pre">(test_2),</span> <span class="pre">statement_2,</span> <span class="pre">...</span> <span class="pre">(test_N),</span>
-<span class="pre">statement_N)</span></tt>. The statements are evaluated only if the corresponding
-tests evaluate to true.</p>
+<div class="section" id="conditional-evaluation">
+<span id="case"></span><h1><a class="toc-backref" href="#id13">Conditional evaluation</a></h1>
+<p>The 'case' construct is the main means by which programmability is achieved in
+LLVMC. It can be used to calculate edge weights, program actions and modify the
+shell commands to be executed. The 'case' expression is designed after the
+similarly-named construct in functional languages and takes the form <tt class="docutils literal">(case
+(test_1), statement_1, (test_2), statement_2, ... (test_N), statement_N)</tt>. The
+statements are evaluated only if the corresponding tests evaluate to true.</p>
<p>Examples:</p>
<pre class="literal-block">
+// Edge weight calculation
+
// Increases edge weight by 5 if "-A" is provided on the
// command-line, and by 5 more if "-B" is also provided.
(case
(switch_on "A"), (inc_weight 5),
(switch_on "B"), (inc_weight 5))
-// Evaluates to "cmdline1" if option "-A" is provided on the
-// command line, otherwise to "cmdline2"
+
+// Tool command line specification
+
+// Evaluates to "cmdline1" if the option "-A" is provided on the
+// command line; to "cmdline2" if "-B" is provided;
+// otherwise to "cmdline3".
+
(case
(switch_on "A"), "cmdline1",
(switch_on "B"), "cmdline2",
(default), "cmdline3")
</pre>
-<p>Note the slight difference in 'case' expression handling in contexts
-of edge weights and command line specification - in the second example
-the value of the <tt class="docutils literal"><span class="pre">"B"</span></tt> switch is never checked when switch <tt class="docutils literal"><span class="pre">"A"</span></tt> is
-enabled, and the whole expression always evaluates to <tt class="docutils literal"><span class="pre">"cmdline1"</span></tt> in
-that case.</p>
+<p>Note the slight difference in 'case' expression handling in contexts of edge
+weights and command line specification - in the second example the value of the
+<tt class="docutils literal">"B"</tt> switch is never checked when switch <tt class="docutils literal">"A"</tt> is enabled, and the whole
+expression always evaluates to <tt class="docutils literal">"cmdline1"</tt> in that case.</p>
<p>Case expressions can also be nested, i.e. the following is legal:</p>
<pre class="literal-block">
(case (switch_on "E"), (case (switch_on "o"), ..., (default), ...)
(default), ...)
</pre>
-<p>You should, however, try to avoid doing that because it hurts
-readability. It is usually better to split tool descriptions and/or
-use TableGen inheritance instead.</p>
+<p>You should, however, try to avoid doing that because it hurts readability. It is
+usually better to split tool descriptions and/or use TableGen inheritance
+instead.</p>
<ul class="simple">
<li>Possible tests are:<ul>
-<li><tt class="docutils literal"><span class="pre">switch_on</span></tt> - Returns true if a given command-line option is
-provided by the user. Example: <tt class="docutils literal"><span class="pre">(switch_on</span> <span class="pre">"opt")</span></tt>. Note that
-you have to define all possible command-line options separately in
-the tool descriptions. See the next doc_text for the discussion of
-different kinds of command-line options.</li>
-<li><tt class="docutils literal"><span class="pre">parameter_equals</span></tt> - Returns true if a command-line parameter equals
-a given value. Example: <tt class="docutils literal"><span class="pre">(parameter_equals</span> <span class="pre">"W",</span> <span class="pre">"all")</span></tt>.</li>
-<li><tt class="docutils literal"><span class="pre">element_in_list</span></tt> - Returns true if a command-line parameter list
-includes a given value. Example: <tt class="docutils literal"><span class="pre">(parameter_in_list</span> <span class="pre">"l",</span> <span class="pre">"pthread")</span></tt>.</li>
-<li><tt class="docutils literal"><span class="pre">input_languages_contain</span></tt> - Returns true if a given language
-belongs to the current input language set. Example:
-<tt class="docutils literal"><span class="pre">`(input_languages_contain</span> <span class="pre">"c++")</span></tt>.</li>
-<li><tt class="docutils literal"><span class="pre">in_language</span></tt> - Evaluates to true if the language of the input
-file equals to the argument. Valid only when using <tt class="docutils literal"><span class="pre">case</span></tt>
-expression in a <tt class="docutils literal"><span class="pre">cmd_line</span></tt> tool property. Example:
-<tt class="docutils literal"><span class="pre">`(in_language</span> <span class="pre">"c++")</span></tt>.</li>
-<li><tt class="docutils literal"><span class="pre">not_empty</span></tt> - Returns true if a given option (which should be
-either a parameter or a parameter list) is set by the
-user. Example: <tt class="docutils literal"><span class="pre">`(not_empty</span> <span class="pre">"o")</span></tt>.</li>
-<li><tt class="docutils literal"><span class="pre">default</span></tt> - Always evaluates to true. Should always be the last
-test in the <tt class="docutils literal"><span class="pre">case</span></tt> expression.</li>
-<li><tt class="docutils literal"><span class="pre">and</span></tt> - A standard logical combinator that returns true iff all
-of its arguments return true. Used like this: <tt class="docutils literal"><span class="pre">(and</span> <span class="pre">(test1),</span>
-<span class="pre">(test2),</span> <span class="pre">...</span> <span class="pre">(testN))</span></tt>. Nesting of <tt class="docutils literal"><span class="pre">and</span></tt> and <tt class="docutils literal"><span class="pre">or</span></tt> is allowed,
-but not encouraged.</li>
-<li><tt class="docutils literal"><span class="pre">or</span></tt> - Another logical combinator that returns true only if any
-one of its arguments returns true. Example: <tt class="docutils literal"><span class="pre">(or</span> <span class="pre">(test1),</span>
-<span class="pre">(test2),</span> <span class="pre">...</span> <span class="pre">(testN))</span></tt>.</li>
+<li><tt class="docutils literal">switch_on</tt> - Returns true if a given command-line switch is provided by
+the user. Can be given multiple arguments, in that case <tt class="docutils literal">(switch_on "foo",
+"bar", "baz")</tt> is equivalent to <tt class="docutils literal">(and (switch_on <span class="pre">"foo"),</span> (switch_on
+<span class="pre">"bar"),</span> (switch_on <span class="pre">"baz"))</span></tt>.
+Example: <tt class="docutils literal">(switch_on "opt")</tt>.</li>
+<li><tt class="docutils literal">any_switch_on</tt> - Given a number of switch options, returns true if any of
+the switches is turned on.
+Example: <tt class="docutils literal">(any_switch_on "foo", "bar", "baz")</tt> is equivalent to <tt class="docutils literal">(or
+(switch_on <span class="pre">"foo"),</span> (switch_on <span class="pre">"bar"),</span> (switch_on <span class="pre">"baz"))</span></tt>.</li>
+<li><tt class="docutils literal">parameter_equals</tt> - Returns true if a command-line parameter (first
+argument) equals a given value (second argument).
+Example: <tt class="docutils literal">(parameter_equals "W", "all")</tt>.</li>
+<li><tt class="docutils literal">element_in_list</tt> - Returns true if a command-line parameter list (first
+argument) contains a given value (second argument).
+Example: <tt class="docutils literal">(element_in_list "l", "pthread")</tt>.</li>
+<li><tt class="docutils literal">input_languages_contain</tt> - Returns true if a given language
+belongs to the current input language set.
+Example: <tt class="docutils literal">(input_languages_contain <span class="pre">"c++")</span></tt>.</li>
+<li><tt class="docutils literal">in_language</tt> - Evaluates to true if the input file language is equal to
+the argument. At the moment works only with <tt class="docutils literal">command</tt> and <tt class="docutils literal">actions</tt> (on
+non-join nodes).
+Example: <tt class="docutils literal">(in_language <span class="pre">"c++")</span></tt>.</li>
+<li><tt class="docutils literal">not_empty</tt> - Returns true if a given option (which should be either a
+parameter or a parameter list) is set by the user. Like <tt class="docutils literal">switch_on</tt>, can
+be also given multiple arguments.
+Examples: <tt class="docutils literal">(not_empty "o")</tt>, <tt class="docutils literal">(not_empty "o", "l")</tt>.</li>
+<li><tt class="docutils literal">any_not_empty</tt> - Returns true if <tt class="docutils literal">not_empty</tt> returns true for any of
+the provided options.
+Example: <tt class="docutils literal">(any_not_empty "foo", "bar", "baz")</tt> is equivalent to <tt class="docutils literal">(or
+(not_empty <span class="pre">"foo"),</span> (not_empty <span class="pre">"bar"),</span> (not_empty <span class="pre">"baz"))</span></tt>.</li>
+<li><tt class="docutils literal">empty</tt> - The opposite of <tt class="docutils literal">not_empty</tt>. Equivalent to <tt class="docutils literal">(not (not_empty
+X))</tt>. Can be given multiple arguments.</li>
+<li><tt class="docutils literal">any_not_empty</tt> - Returns true if <tt class="docutils literal">not_empty</tt> returns true for any of
+the provided options.
+Example: <tt class="docutils literal">(any_empty "foo", "bar", "baz")</tt> is equivalent to <tt class="docutils literal">(or
+(not_empty <span class="pre">"foo"),</span> (not_empty <span class="pre">"bar"),</span> (not_empty <span class="pre">"baz"))</span></tt>.</li>
+<li><tt class="docutils literal">single_input_file</tt> - Returns true if there was only one input file
+provided on the command-line. Used without arguments:
+<tt class="docutils literal">(single_input_file)</tt>.</li>
+<li><tt class="docutils literal">multiple_input_files</tt> - Equivalent to <tt class="docutils literal">(not (single_input_file))</tt> (the
+case of zero input files is considered an error).</li>
+<li><tt class="docutils literal">default</tt> - Always evaluates to true. Should always be the last
+test in the <tt class="docutils literal">case</tt> expression.</li>
+<li><tt class="docutils literal">and</tt> - A standard logical combinator that returns true iff all of
+its arguments return true. Used like this: <tt class="docutils literal">(and (test1), (test2),
+... (testN))</tt>. Nesting of <tt class="docutils literal">and</tt> and <tt class="docutils literal">or</tt> is allowed, but not
+encouraged.</li>
+<li><tt class="docutils literal">or</tt> - A logical combinator that returns true iff any of its arguments
+return true.
+Example: <tt class="docutils literal">(or (test1), (test2), ... (testN))</tt>.</li>
+<li><tt class="docutils literal">not</tt> - Standard unary logical combinator that negates its
+argument.
+Example: <tt class="docutils literal">(not (or (test1), (test2), ... <span class="pre">(testN)))</span></tt>.</li>
</ul>
</li>
</ul>
</div>
-<div class="doc_text">
-<div class="doc_section"><a class="toc-backref" href="#id10" id="language-map" name="language-map">Language map</a></div>
-<p>One last thing that you will need to modify when adding support for a
-new language to LLVMC is the language map, which defines mappings from
-file extensions to language names. It is used to choose the proper
-toolchain(s) for a given input file set. Language map definition is
-located in the file <tt class="docutils literal"><span class="pre">Tools.td</span></tt> and looks like this:</p>
+<div class="section" id="writing-a-tool-description">
+<h1><a class="toc-backref" href="#id14">Writing a tool description</a></h1>
+<p>As was said earlier, nodes in the compilation graph represent tools, which are
+described separately. A tool definition looks like this (taken from the
+<tt class="docutils literal">llvmc/src/Base.td</tt> file):</p>
+<pre class="literal-block">
+def llvm_gcc_cpp : Tool<[
+ (in_language "c++"),
+ (out_language "llvm-assembler"),
+ (output_suffix "bc"),
+ (command "llvm-g++ -c -emit-llvm"),
+ (sink)
+ ]>;
+</pre>
+<p>This defines a new tool called <tt class="docutils literal">llvm_gcc_cpp</tt>, which is an alias for
+<tt class="docutils literal"><span class="pre">llvm-g++</span></tt>. As you can see, a tool definition is just a list of properties;
+most of them should be self-explanatory. The <tt class="docutils literal">sink</tt> property means that this
+tool should be passed all command-line options that aren't mentioned in the
+option list.</p>
+<p>The complete list of all currently implemented tool properties follows.</p>
+<ul class="simple">
+<li>Possible tool properties:<ul>
+<li><tt class="docutils literal">in_language</tt> - input language name. Can be given multiple arguments, in
+case the tool supports multiple input languages. Used for typechecking and
+mapping file extensions to tools.</li>
+<li><tt class="docutils literal">out_language</tt> - output language name. Multiple output languages are
+allowed. Used for typechecking the compilation graph.</li>
+<li><tt class="docutils literal">output_suffix</tt> - output file suffix. Can also be changed dynamically, see
+documentation on <a class="reference internal" href="#actions">actions</a>.</li>
+</ul>
+</li>
+</ul>
+<blockquote>
+<ul class="simple">
+<li><tt class="docutils literal">command</tt> - the actual command used to run the tool. You can use output
+redirection with <tt class="docutils literal">></tt>, hook invocations (<tt class="docutils literal">$CALL</tt>), environment variables
+(via <tt class="docutils literal">$ENV</tt>) and the <tt class="docutils literal">case</tt> construct.</li>
+<li><tt class="docutils literal">join</tt> - this tool is a "join node" in the graph, i.e. it gets a list of
+input files and joins them together. Used for linkers.</li>
+<li><tt class="docutils literal">sink</tt> - all command-line options that are not handled by other tools are
+passed to this tool.</li>
+<li><tt class="docutils literal">actions</tt> - A single big <tt class="docutils literal">case</tt> expression that specifies how this tool
+reacts on command-line options (described in more detail <a class="reference internal" href="#actions">below</a>).</li>
+</ul>
+</blockquote>
+<blockquote>
+<ul class="simple">
+<li><tt class="docutils literal">out_file_option</tt>, <tt class="docutils literal">in_file_option</tt> - Options appended to the
+<tt class="docutils literal">command</tt> string to designate output and input files. Default values are
+<tt class="docutils literal"><span class="pre">"-o"</span></tt> and <tt class="docutils literal">""</tt>, respectively.</li>
+</ul>
+</blockquote>
+<div class="section" id="id4">
+<span id="actions"></span><h2><a class="toc-backref" href="#id15">Actions</a></h2>
+<p>A tool often needs to react to command-line options, and this is precisely what
+the <tt class="docutils literal">actions</tt> property is for. The next example illustrates this feature:</p>
+<pre class="literal-block">
+def llvm_gcc_linker : Tool<[
+ (in_language "object-code"),
+ (out_language "executable"),
+ (output_suffix "out"),
+ (command "llvm-gcc"),
+ (join),
+ (actions (case (not_empty "L"), (forward "L"),
+ (not_empty "l"), (forward "l"),
+ (not_empty "dummy"),
+ [(append_cmd "-dummy1"), (append_cmd "-dummy2")])
+ ]>;
+</pre>
+<p>The <tt class="docutils literal">actions</tt> tool property is implemented on top of the omnipresent <tt class="docutils literal">case</tt>
+expression. It associates one or more different <em>actions</em> with given
+conditions - in the example, the actions are <tt class="docutils literal">forward</tt>, which forwards a given
+option unchanged, and <tt class="docutils literal">append_cmd</tt>, which appends a given string to the tool
+execution command. Multiple actions can be associated with a single condition by
+using a list of actions (used in the example to append some dummy options). The
+same <tt class="docutils literal">case</tt> construct can also be used in the <tt class="docutils literal">cmd_line</tt> property to modify
+the tool command line.</p>
+<p>The "join" property used in the example means that this tool behaves like a
+linker.</p>
+<p>The list of all possible actions follows.</p>
+<ul>
+<li><p class="first">Possible actions:</p>
+<blockquote>
+<ul class="simple">
+<li><tt class="docutils literal">append_cmd</tt> - Append a string to the tool invocation command.
+Example: <tt class="docutils literal">(case (switch_on <span class="pre">"pthread"),</span> (append_cmd <span class="pre">"-lpthread"))</span></tt>.</li>
+<li><tt class="docutils literal">error</tt> - Exit with error.
+Example: <tt class="docutils literal">(error "Mixing <span class="pre">-c</span> and <span class="pre">-S</span> is not <span class="pre">allowed!")</span></tt>.</li>
+<li><tt class="docutils literal">warning</tt> - Print a warning.
+Example: <tt class="docutils literal">(warning "Specifying both <span class="pre">-O1</span> and <span class="pre">-O2</span> is <span class="pre">meaningless!")</span></tt>.</li>
+<li><tt class="docutils literal">forward</tt> - Forward the option unchanged.
+Example: <tt class="docutils literal">(forward "Wall")</tt>.</li>
+<li><tt class="docutils literal">forward_as</tt> - Change the option's name, but forward the argument
+unchanged.
+Example: <tt class="docutils literal">(forward_as "O0", <span class="pre">"--disable-optimization")</span></tt>.</li>
+<li><tt class="docutils literal">forward_value</tt> - Forward only option's value. Cannot be used with switch
+options (since they don't have values), but works fine with lists.
+Example: <tt class="docutils literal">(forward_value <span class="pre">"Wa,")</span></tt>.</li>
+<li><tt class="docutils literal">forward_transformed_value</tt> - As above, but applies a hook to the
+option's value before forwarding (see <a class="reference internal" href="#hooks">below</a>). When
+<tt class="docutils literal">forward_transformed_value</tt> is applied to a list
+option, the hook must have signature
+<tt class="docutils literal"><span class="pre">std::string</span> <span class="pre">hooks::HookName</span> (const <span class="pre">std::vector<std::string>&)</span></tt>.
+Example: <tt class="docutils literal">(forward_transformed_value "m", "ConvertToMAttr")</tt>.</li>
+<li><tt class="docutils literal">output_suffix</tt> - Modify the output suffix of this tool.
+Example: <tt class="docutils literal">(output_suffix "i")</tt>.</li>
+<li><tt class="docutils literal">stop_compilation</tt> - Stop compilation after this tool processes its
+input. Used without arguments.
+Example: <tt class="docutils literal">(stop_compilation)</tt>.</li>
+</ul>
+</blockquote>
+</li>
+</ul>
+</div>
+</div>
+<div class="section" id="language-map">
+<h1><a class="toc-backref" href="#id16">Language map</a></h1>
+<p>If you are adding support for a new language to LLVMC, you'll need to modify the
+language map, which defines mappings from file extensions to language names. It
+is used to choose the proper toolchain(s) for a given input file set. Language
+map definition looks like this:</p>
<pre class="literal-block">
def LanguageMap : LanguageMap<
[LangToSuffixes<"c++", ["cc", "cp", "cxx", "cpp", "CPP", "c++", "C"]>,
...
]>;
</pre>
+<p>For example, without those definitions the following command wouldn't work:</p>
+<pre class="literal-block">
+$ llvmc hello.cpp
+llvmc: Unknown suffix: cpp
+</pre>
+<p>The language map entries are needed only for the tools that are linked from the
+root node. A tool can have multiple output languages.</p>
+</div>
+<div class="section" id="option-preprocessor">
+<h1><a class="toc-backref" href="#id17">Option preprocessor</a></h1>
+<p>It is sometimes useful to run error-checking code before processing the
+compilation graph. For example, if optimization options "-O1" and "-O2" are
+implemented as switches, we might want to output a warning if the user invokes
+the driver with both of these options enabled.</p>
+<p>The <tt class="docutils literal">OptionPreprocessor</tt> feature is reserved specially for these
+occasions. Example (adapted from <tt class="docutils literal">llvm/src/Base.td.in</tt>):</p>
+<pre class="literal-block">
+def Preprocess : OptionPreprocessor<
+(case (not (any_switch_on "O0", "O1", "O2", "O3")),
+ (set_option "O2"),
+ (and (switch_on "O3"), (any_switch_on "O0", "O1", "O2")),
+ (unset_option "O0", "O1", "O2"),
+ (and (switch_on "O2"), (any_switch_on "O0", "O1")),
+ (unset_option "O0", "O1"),
+ (and (switch_on "O1"), (switch_on "O0")),
+ (unset_option "O0"))
+>;
+</pre>
+<p>Here, <tt class="docutils literal">OptionPreprocessor</tt> is used to unset all spurious <tt class="docutils literal"><span class="pre">-O</span></tt> options so
+that they are not forwarded to the compiler. If no optimization options are
+specified, <tt class="docutils literal"><span class="pre">-O2</span></tt> is enabled.</p>
+<p><tt class="docutils literal">OptionPreprocessor</tt> is basically a single big <tt class="docutils literal">case</tt> expression, which is
+evaluated only once right after the driver is started. The only allowed actions
+in <tt class="docutils literal">OptionPreprocessor</tt> are <tt class="docutils literal">error</tt>, <tt class="docutils literal">warning</tt>, and two special actions:
+<tt class="docutils literal">unset_option</tt> and <tt class="docutils literal">set_option</tt>. As their names suggest, they can be used to
+set or unset a given option. To set an option with <tt class="docutils literal">set_option</tt>, use the
+two-argument form: <tt class="docutils literal">(set_option "parameter", VALUE)</tt>. Here, <tt class="docutils literal">VALUE</tt> can be
+either a string, a string list, or a boolean constant.</p>
+<p>For convenience, <tt class="docutils literal">set_option</tt> and <tt class="docutils literal">unset_option</tt> also work with multiple
+arguments. That is, instead of <tt class="docutils literal">[(unset_option <span class="pre">"A"),</span> (unset_option <span class="pre">"B")]</span></tt> you
+can use <tt class="docutils literal">(unset_option "A", "B")</tt>. Obviously, <tt class="docutils literal">(set_option "A", "B")</tt> is
+only valid if both <tt class="docutils literal">A</tt> and <tt class="docutils literal">B</tt> are switches.</p>
</div>
-<div class="doc_text">
-<div class="doc_section"><a class="toc-backref" href="#id11" id="references" name="references">References</a></div>
-<table class="docutils footnote" frame="void" id="id2" rules="none">
-<colgroup><col class="label" /><col /></colgroup>
-<tbody valign="top">
-<tr><td class="label"><a class="fn-backref" href="#id1" name="id2">[1]</a></td><td>TableGen Fundamentals
-<a class="reference" href="http://llvm.cs.uiuc.edu/docs/TableGenFundamentals.html">http://llvm.cs.uiuc.edu/docs/TableGenFundamentals.html</a></td></tr>
-</tbody>
-</table>
+<div class="section" id="more-advanced-topics">
+<h1><a class="toc-backref" href="#id18">More advanced topics</a></h1>
+<div class="section" id="hooks-and-environment-variables">
+<span id="hooks"></span><h2><a class="toc-backref" href="#id19">Hooks and environment variables</a></h2>
+<p>Normally, LLVMC searches for programs in the system <tt class="docutils literal">PATH</tt>. Sometimes, this is
+not sufficient: for example, we may want to specify tool paths or names in the
+configuration file. This can be achieved via the hooks mechanism. To write your
+own hooks, add their definitions to the <tt class="docutils literal">Hooks.cpp</tt> or drop a <tt class="docutils literal">.cpp</tt> file
+into your driver directory. Hooks should live in the <tt class="docutils literal">hooks</tt> namespace and
+have the signature <tt class="docutils literal"><span class="pre">std::string</span> <span class="pre">hooks::MyHookName</span> ([const char* Arg0 [ const
+char* Arg2 [, <span class="pre">...]]])</span></tt>. They can be used from the <tt class="docutils literal">command</tt> tool property:</p>
+<pre class="literal-block">
+(command "$CALL(MyHook)/path/to/file -o $CALL(AnotherHook)")
+</pre>
+<p>To pass arguments to hooks, use the following syntax:</p>
+<pre class="literal-block">
+(command "$CALL(MyHook, 'Arg1', 'Arg2', 'Arg # 3')/path/to/file -o1 -o2")
+</pre>
+<p>It is also possible to use environment variables in the same manner:</p>
+<pre class="literal-block">
+(command "$ENV(VAR1)/path/to/file -o $ENV(VAR2)")
+</pre>
+<p>To change the command line string based on user-provided options use
+the <tt class="docutils literal">case</tt> expression (documented <a class="reference internal" href="#case">above</a>):</p>
+<pre class="literal-block">
+(command
+ (case
+ (switch_on "E"),
+ "llvm-g++ -E -x c $INFILE -o $OUTFILE",
+ (default),
+ "llvm-g++ -c -x c $INFILE -o $OUTFILE -emit-llvm"))
+</pre>
</div>
+<div class="section" id="debugging">
+<h2><a class="toc-backref" href="#id20">Debugging</a></h2>
+<p>When writing LLVMC-based drivers, it can be useful to get a visual view of the
+resulting compilation graph. This can be achieved via the command line option
+<tt class="docutils literal"><span class="pre">--view-graph</span></tt> (which assumes that <a class="reference external" href="http://www.graphviz.org/">Graphviz</a> and <a class="reference external" href="http://pages.cs.wisc.edu/~ghost/">Ghostview</a> are
+installed). There is also a <tt class="docutils literal"><span class="pre">--write-graph</span></tt> option that creates a Graphviz
+source file (<tt class="docutils literal"><span class="pre">compilation-graph.dot</span></tt>) in the current directory.</p>
+<p>Another useful <tt class="docutils literal">llvmc</tt> option is <tt class="docutils literal"><span class="pre">--check-graph</span></tt>. It checks the compilation
+graph for common errors like mismatched output/input language names, multiple
+default edges and cycles. When invoked with <tt class="docutils literal"><span class="pre">--check-graph</span></tt>, <tt class="docutils literal">llvmc</tt> doesn't
+perform any compilation tasks and returns the number of encountered errors as
+its status code. In the future, these checks will be performed at compile-time
+and this option will disappear.</p>
</div>
+<div class="section" id="conditioning-on-the-executable-name">
+<h2><a class="toc-backref" href="#id21">Conditioning on the executable name</a></h2>
+<p>For now, the executable name (the value passed to the driver in <tt class="docutils literal">argv[0]</tt>) is
+accessible only in the C++ code (i.e. hooks). Use the following code:</p>
+<pre class="literal-block">
+namespace llvmc {
+extern const char* ProgramName;
+}
+
+namespace hooks {
+
+std::string MyHook() {
+//...
+if (strcmp(ProgramName, "mydriver") == 0) {
+ //...
+
+}
+
+} // end namespace hooks
+</pre>
+<p>In general, you're encouraged not to make the behaviour dependent on the
+executable file name, and use command-line switches instead. See for example how
+the <tt class="docutils literal">llvmc</tt> program behaves when it needs to choose the correct linker options
+(think <tt class="docutils literal">g++</tt> vs. <tt class="docutils literal">gcc</tt>).</p>
<hr />
<address>
- <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
- src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!" /></a>
- <a href="http://validator.w3.org/check/referer"><img
- src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!" /></a>
- <a href="http://llvm.org">The LLVM Compiler Infrastructure</a><br/>
- Last modified: $Date$
-</address>
+<a href="http://jigsaw.w3.org/css-validator/check/referer">
+<img src="http://jigsaw.w3.org/css-validator/images/vcss-blue"
+ alt="Valid CSS" /></a>
+<a href="http://validator.w3.org/check?uri=referer">
+<img src="http://www.w3.org/Icons/valid-xhtml10-blue"
+ alt="Valid XHTML 1.0 Transitional"/></a>
+
+<a href="mailto:foldr@codedgers.com">Mikhail Glushenkov</a><br />
+<a href="http://llvm.org">LLVM Compiler Infrastructure</a><br />
+
+Last modified: $Date$
+</address></div>
+</div>
+</div>
</body>
</html>
projects / oota-llvm.git / blobdiff