replica
/
hanchenye-llvm-project
2
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Added how to add machine passes to command line options. 

llvm-svn: 29528
This commit is contained in:
Jim Laskey 2006-08-04 18:10:12 +00:00
parent cd4ff2b36f
commit 575df8b844
1 changed files with 216 additions and 77 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

293
llvm/docs/WritingAnLLVMPass.html
View File

@ -83,6 +83,11 @@
<ul> <ul>
<li><a href="#releaseMemory">The <tt>releaseMemory</tt> method</a></li> <li><a href="#releaseMemory">The <tt>releaseMemory</tt> method</a></li>
</ul></li> </ul></li>
<li><a href="#registering">Registering dynamically loaded passes</a>
<ul>
<li><a href="#registering_existing">Using existing registries</a></li>
<li><a href="#registering_new">Creating new registries</a></li>
</ul></li>
<li><a href="#debughints">Using GDB with dynamically loaded passes</a> <li><a href="#debughints">Using GDB with dynamically loaded passes</a>
<ul> <ul>
<li><a href="#breakpoint">Setting a breakpoint in your pass</a></li> <li><a href="#breakpoint">Setting a breakpoint in your pass</a></li>
@ -97,7 +102,8 @@
</ol> </ol>
<div class="doc_author"> <div class="doc_author">
<p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a></p> <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a>,
<a href="mailto:jlaskey@apple.com">Jim Laskey</a></p>
</div> </div>
<!-- *********************************************************************** --> <!-- *********************************************************************** -->
@ -163,7 +169,7 @@ source tree in the <tt>lib/Transforms/Hello</tt> directory.</p>
copy the following into <tt>Makefile</tt>:</p> copy the following into <tt>Makefile</tt>:</p>
<hr/> <hr/>
<pre> <div class="doc_code"><pre>
# Makefile for hello pass # Makefile for hello pass
# Path to top level of LLVM heirarchy # Path to top level of LLVM heirarchy
@ -181,7 +187,7 @@ LOADABLE_MODULE = 1
# Include the makefile implementation stuff # Include the makefile implementation stuff
include $(LEVEL)/Makefile.common include $(LEVEL)/Makefile.common
</pre> </pre></div>
<p>This makefile specifies that all of the <tt>.cpp</tt> files in the current <p>This makefile specifies that all of the <tt>.cpp</tt> files in the current
directory are to be compiled and linked together into a directory are to be compiled and linked together into a
@ -205,10 +211,10 @@ the pass itself.</p>
<p>Now that we have a way to compile our new pass, we just have to write it. <p>Now that we have a way to compile our new pass, we just have to write it.
Start out with:</p> Start out with:</p>
<pre> <div class="doc_code"><pre>
<b>#include</b> "<a href="http://llvm.org/doxygen/Pass_8h-source.html">llvm/Pass.h</a>" <b>#include</b> "<a href="http://llvm.org/doxygen/Pass_8h-source.html">llvm/Pass.h</a>"
<b>#include</b> "<a href="http://llvm.org/doxygen/Function_8h-source.html">llvm/Function.h</a>" <b>#include</b> "<a href="http://llvm.org/doxygen/Function_8h-source.html">llvm/Function.h</a>"
</pre> </pre></div>
<p>Which are needed because we are writing a <tt><a <p>Which are needed because we are writing a <tt><a
href="http://llvm.org/doxygen/classllvm_1_1Pass.html">Pass</a></tt>, and href="http://llvm.org/doxygen/classllvm_1_1Pass.html">Pass</a></tt>, and
@ -216,18 +222,18 @@ we are operating on <tt><a
href="http://llvm.org/doxygen/classllvm_1_1Function.html">Function</a></tt>'s.</p> href="http://llvm.org/doxygen/classllvm_1_1Function.html">Function</a></tt>'s.</p>
<p>Next we have:</p> <p>Next we have:</p>
<pre> <div class="doc_code"><pre>
<b>using namespace llvm;</b> <b>using namespace llvm;</b>
</pre> </pre></div>
<p>... which is required because the functions from the include files <p>... which is required because the functions from the include files
live in the llvm namespace. live in the llvm namespace.
</p> </p>
<p>Next we have:</p> <p>Next we have:</p>
<pre> <div class="doc_code"><pre>
<b>namespace</b> { <b>namespace</b> {
</pre> </pre></div>
<p>... which starts out an anonymous namespace. Anonymous namespaces are to C++ <p>... which starts out an anonymous namespace. Anonymous namespaces are to C++
what the "<tt>static</tt>" keyword is to C (at global scope). It makes the what the "<tt>static</tt>" keyword is to C (at global scope). It makes the
@ -237,9 +243,9 @@ information.</p>
<p>Next, we declare our pass itself:</p> <p>Next, we declare our pass itself:</p>
<pre> <div class="doc_code"><pre>
<b>struct</b> Hello : <b>public</b> <a href="#FunctionPass">FunctionPass</a> { <b>struct</b> Hello : <b>public</b> <a href="#FunctionPass">FunctionPass</a> {
</pre><p> </pre></div><p>
<p>This declares a "<tt>Hello</tt>" class that is a subclass of <tt><a <p>This declares a "<tt>Hello</tt>" class that is a subclass of <tt><a
href="http://llvm.org/doxygen/classllvm_1_1FunctionPass.html">FunctionPass</a></tt>. href="http://llvm.org/doxygen/classllvm_1_1FunctionPass.html">FunctionPass</a></tt>.
@ -248,13 +254,13 @@ href="#passtype">later</a>, but for now, know that <a
href="#FunctionPass"><tt>FunctionPass</tt></a>'s operate a function at a href="#FunctionPass"><tt>FunctionPass</tt></a>'s operate a function at a
time.</p> time.</p>
<pre> <div class="doc_code"><pre>
<b>virtual bool</b> <a href="#runOnFunction">runOnFunction</a>(Function &amp;F) { <b>virtual bool</b> <a href="#runOnFunction">runOnFunction</a>(Function &amp;F) {
std::cerr &lt;&lt; "<i>Hello: </i>" &lt;&lt; F.getName() &lt;&lt; "\n"; std::cerr &lt;&lt; "<i>Hello: </i>" &lt;&lt; F.getName() &lt;&lt; "\n";
<b>return false</b>; <b>return false</b>;
} }
}; <i>// end of struct Hello</i> }; <i>// end of struct Hello</i>
</pre> </pre></div>
<p>We declare a "<a href="#runOnFunction"><tt>runOnFunction</tt></a>" method, <p>We declare a "<a href="#runOnFunction"><tt>runOnFunction</tt></a>" method,
which overloads an abstract virtual method inherited from <a which overloads an abstract virtual method inherited from <a
@ -262,10 +268,10 @@ href="#FunctionPass"><tt>FunctionPass</tt></a>. This is where we are supposed
to do our thing, so we just print out our message with the name of each to do our thing, so we just print out our message with the name of each
function.</p> function.</p>
<pre> <div class="doc_code"><pre>
RegisterOpt&lt;Hello&gt; X("<i>hello</i>", "<i>Hello World Pass</i>"); RegisterOpt&lt;Hello&gt; X("<i>hello</i>", "<i>Hello World Pass</i>");
} <i>// end of anonymous namespace</i> } <i>// end of anonymous namespace</i>
</pre> </pre></div>
<p>Lastly, we register our class <tt>Hello</tt>, giving it a command line <p>Lastly, we register our class <tt>Hello</tt>, giving it a command line
argument "<tt>hello</tt>", and a name "<tt>Hello World Pass</tt>". There are argument "<tt>hello</tt>", and a name "<tt>Hello World Pass</tt>". There are
@ -275,7 +281,7 @@ depending on what it is to be used for. For "optimizations" we use the
<p>As a whole, the <tt>.cpp</tt> file looks like:</p> <p>As a whole, the <tt>.cpp</tt> file looks like:</p>
<pre> <div class="doc_code"><pre>
<b>#include</b> "<a href="http://llvm.org/doxygen/Pass_8h-source.html">llvm/Pass.h</a>" <b>#include</b> "<a href="http://llvm.org/doxygen/Pass_8h-source.html">llvm/Pass.h</a>"
<b>#include</b> "<a href="http://llvm.org/doxygen/Function_8h-source.html">llvm/Function.h</a>" <b>#include</b> "<a href="http://llvm.org/doxygen/Function_8h-source.html">llvm/Function.h</a>"
@ -291,7 +297,7 @@ depending on what it is to be used for. For "optimizations" we use the
RegisterOpt&lt;Hello&gt; X("<i>hello</i>", "<i>Hello World Pass</i>"); RegisterOpt&lt;Hello&gt; X("<i>hello</i>", "<i>Hello World Pass</i>");
} }
</pre> </pre></div>
<p>Now that it's all together, compile the file with a simple "<tt>gmake</tt>" <p>Now that it's all together, compile the file with a simple "<tt>gmake</tt>"
command in the local directory and you should get a new command in the local directory and you should get a new
@ -320,12 +326,12 @@ LLVM. We can now run the bytecode file (<tt>hello.bc</tt>) for the program
through our transformation like this (or course, any bytecode file will through our transformation like this (or course, any bytecode file will
work):</p> work):</p>
<pre> <div class="doc_code"><pre>
$ opt -load ../../../Debug/lib/Hello.so -hello &lt; hello.bc &gt; /dev/null $ opt -load ../../../Debug/lib/Hello.so -hello &lt; hello.bc &gt; /dev/null
Hello: __main Hello: __main
Hello: puts Hello: puts
Hello: main Hello: main
</pre> </pre></div>
<p>The '<tt>-load</tt>' option specifies that '<tt>opt</tt>' should load your <p>The '<tt>-load</tt>' option specifies that '<tt>opt</tt>' should load your
pass as a shared object, which makes '<tt>-hello</tt>' a valid command line pass as a shared object, which makes '<tt>-hello</tt>' a valid command line
@ -337,7 +343,7 @@ interesting way, we just throw away the result of <tt>opt</tt> (sending it to
<p>To see what happened to the other string you registered, try running <p>To see what happened to the other string you registered, try running
<tt>opt</tt> with the <tt>--help</tt> option:</p> <tt>opt</tt> with the <tt>--help</tt> option:</p>
<pre> <div class="doc_code"><pre>
$ opt -load ../../../Debug/lib/Hello.so --help $ opt -load ../../../Debug/lib/Hello.so --help
OVERVIEW: llvm .bc -&gt; .bc modular optimizer OVERVIEW: llvm .bc -&gt; .bc modular optimizer
@ -354,7 +360,7 @@ OPTIONS:
-inline - Function Integration/Inlining -inline - Function Integration/Inlining
-instcombine - Combine redundant instructions -instcombine - Combine redundant instructions
... ...
</pre> </pre></div>
<p>The pass name get added as the information string for your pass, giving some <p>The pass name get added as the information string for your pass, giving some
documentation to users of <tt>opt</tt>. Now that you have a working pass, you documentation to users of <tt>opt</tt>. Now that you have a working pass, you
@ -365,7 +371,7 @@ line option (<tt>--time-passes</tt>) that allows you to get information about
the execution time of your pass along with the other passes you queue up. For the execution time of your pass along with the other passes you queue up. For
example:</p> example:</p>
<pre> <div class="doc_code"><pre>
$ opt -load ../../../Debug/lib/Hello.so -hello -time-passes &lt; hello.bc &gt; /dev/null $ opt -load ../../../Debug/lib/Hello.so -hello -time-passes &lt; hello.bc &gt; /dev/null
Hello: __main Hello: __main
Hello: puts Hello: puts
@ -381,7 +387,7 @@ Hello: main
0.0000 ( 0.0%) 0.0000 ( 0.0%) 0.0000 ( 0.0%) 0.0013 ( 2.7%) Module Verifier 0.0000 ( 0.0%) 0.0000 ( 0.0%) 0.0000 ( 0.0%) 0.0013 ( 2.7%) Module Verifier
<b> 0.0000 ( 0.0%) 0.0000 ( 0.0%) 0.0000 ( 0.0%) 0.0033 ( 6.9%) Hello World Pass</b> <b> 0.0000 ( 0.0%) 0.0000 ( 0.0%) 0.0000 ( 0.0%) 0.0033 ( 6.9%) Hello World Pass</b>
0.0100 (100.0%) 0.0100 (100.0%) 0.0200 (100.0%) 0.0479 (100.0%) TOTAL 0.0100 (100.0%) 0.0100 (100.0%) 0.0200 (100.0%) 0.0479 (100.0%) TOTAL
</pre> </pre></div>
<p>As you can see, our implementation above is pretty fast :). The additional <p>As you can see, our implementation above is pretty fast :). The additional
passes listed are automatically inserted by the '<tt>opt</tt>' tool to verify passes listed are automatically inserted by the '<tt>opt</tt>' tool to verify
@ -467,9 +473,9 @@ following signature:</p>
<div class="doc_text"> <div class="doc_text">
<pre> <div class="doc_code"><pre>
<b>virtual bool</b> runOnModule(Module &amp;M) = 0; <b>virtual bool</b> runOnModule(Module &amp;M) = 0;
</pre> </pre></div>
<p>The <tt>runOnModule</tt> method performs the interesting work of the pass. <p>The <tt>runOnModule</tt> method performs the interesting work of the pass.
It should return true if the module was modified by the transformation and It should return true if the module was modified by the transformation and
@ -535,9 +541,9 @@ false if they didn't.</p>
<div class="doc_text"> <div class="doc_text">
<pre> <div class="doc_code"><pre>
<b>virtual bool</b> doInitialization(Module &amp;M); <b>virtual bool</b> doInitialization(Module &amp;M);
</pre> </pre></div>
<p>The <tt>doIninitialize</tt> method is allowed to do most of the things that <p>The <tt>doIninitialize</tt> method is allowed to do most of the things that
<tt>CallGraphSCCPass</tt>'s are not allowed to do. They can add and remove <tt>CallGraphSCCPass</tt>'s are not allowed to do. They can add and remove
@ -556,9 +562,9 @@ fast).</p>
<div class="doc_text"> <div class="doc_text">
<pre> <div class="doc_code"><pre>
<b>virtual bool</b> runOnSCC(const std::vector&lt;CallGraphNode *&gt; &amp;SCCM) = 0; <b>virtual bool</b> runOnSCC(const std::vector&lt;CallGraphNode *&gt; &amp;SCCM) = 0;
</pre> </pre></div>
<p>The <tt>runOnSCC</tt> method performs the interesting work of the pass, and <p>The <tt>runOnSCC</tt> method performs the interesting work of the pass, and
should return true if the module was modified by the transformation, false should return true if the module was modified by the transformation, false
@ -574,9 +580,9 @@ otherwise.</p>
<div class="doc_text"> <div class="doc_text">
<pre> <div class="doc_code"><pre>
<b>virtual bool</b> doFinalization(Module &amp;M); <b>virtual bool</b> doFinalization(Module &amp;M);
</pre> </pre></div>
<p>The <tt>doFinalization</tt> method is an infrequently used method that is <p>The <tt>doFinalization</tt> method is an infrequently used method that is
called when the pass framework has finished calling <a called when the pass framework has finished calling <a
@ -625,9 +631,9 @@ should return true if they modified the program, or false if they didn't.</p>
<div class="doc_text"> <div class="doc_text">
<pre> <div class="doc_code"><pre>
<b>virtual bool</b> doInitialization(Module &amp;M); <b>virtual bool</b> doInitialization(Module &amp;M);
</pre> </pre></div>
<p>The <tt>doIninitialize</tt> method is allowed to do most of the things that <p>The <tt>doIninitialize</tt> method is allowed to do most of the things that
<tt>FunctionPass</tt>'s are not allowed to do. They can add and remove <tt>FunctionPass</tt>'s are not allowed to do. They can add and remove
@ -653,9 +659,9 @@ free functions that it needs, adding prototypes to the module if necessary.</p>
<div class="doc_text"> <div class="doc_text">
<pre> <div class="doc_code"><pre>
<b>virtual bool</b> runOnFunction(Function &amp;F) = 0; <b>virtual bool</b> runOnFunction(Function &amp;F) = 0;
</pre><p> </pre></div><p>
<p>The <tt>runOnFunction</tt> method must be implemented by your subclass to do <p>The <tt>runOnFunction</tt> method must be implemented by your subclass to do
the transformation or analysis work of your pass. As usual, a true value should the transformation or analysis work of your pass. As usual, a true value should
@ -671,9 +677,9 @@ be returned if the function is modified.</p>
<div class="doc_text"> <div class="doc_text">
<pre> <div class="doc_code"><pre>
<b>virtual bool</b> doFinalization(Module &amp;M); <b>virtual bool</b> doFinalization(Module &amp;M);
</pre> </pre></div>
<p>The <tt>doFinalization</tt> method is an infrequently used method that is <p>The <tt>doFinalization</tt> method is an infrequently used method that is
called when the pass framework has finished calling <a called when the pass framework has finished calling <a
@ -719,9 +725,9 @@ href="#FunctionPass"><tt>FunctionPass</tt></a>'s have, but also have the followi
<div class="doc_text"> <div class="doc_text">
<pre> <div class="doc_code"><pre>
<b>virtual bool</b> doInitialization(Function &amp;F); <b>virtual bool</b> doInitialization(Function &amp;F);
</pre> </pre></div>
<p>The <tt>doIninitialize</tt> method is allowed to do most of the things that <p>The <tt>doIninitialize</tt> method is allowed to do most of the things that
<tt>BasicBlockPass</tt>'s are not allowed to do, but that <tt>BasicBlockPass</tt>'s are not allowed to do, but that
@ -740,9 +746,9 @@ fast).</p>
<div class="doc_text"> <div class="doc_text">
<pre> <div class="doc_code"><pre>
<b>virtual bool</b> runOnBasicBlock(BasicBlock &amp;BB) = 0; <b>virtual bool</b> runOnBasicBlock(BasicBlock &amp;BB) = 0;
</pre> </pre></div>
<p>Override this function to do the work of the <tt>BasicBlockPass</tt>. This <p>Override this function to do the work of the <tt>BasicBlockPass</tt>. This
function is not allowed to inspect or modify basic blocks other than the function is not allowed to inspect or modify basic blocks other than the
@ -759,9 +765,9 @@ if the basic block is modified.</p>
<div class="doc_text"> <div class="doc_text">
<pre> <div class="doc_code"><pre>
<b>virtual bool</b> doFinalization(Function &amp;F); <b>virtual bool</b> doFinalization(Function &amp;F);
</pre> </pre></div>
<p>The <tt>doFinalization</tt> method is an infrequently used method that is <p>The <tt>doFinalization</tt> method is an infrequently used method that is
called when the pass framework has finished calling <a called when the pass framework has finished calling <a
@ -805,9 +811,9 @@ data)</li>
<div class="doc_text"> <div class="doc_text">
<pre> <div class="doc_code"><pre>
<b>virtual bool</b> runOnMachineFunction(MachineFunction &amp;MF) = 0; <b>virtual bool</b> runOnMachineFunction(MachineFunction &amp;MF) = 0;
</pre> </pre></div>
<p><tt>runOnMachineFunction</tt> can be considered the main entry point of a <p><tt>runOnMachineFunction</tt> can be considered the main entry point of a
<tt>MachineFunctionPass</tt>; that is, you should override this method to do the <tt>MachineFunctionPass</tt>; that is, you should override this method to do the
@ -877,9 +883,9 @@ should implement the virtual <tt>print</tt> method:</p>
<div class="doc_text"> <div class="doc_text">
<pre> <div class="doc_code"><pre>
<b>virtual void</b> print(std::ostream &amp;O, <b>const</b> Module *M) <b>const</b>; <b>virtual void</b> print(std::ostream &amp;O, <b>const</b> Module *M) <b>const</b>;
</pre> </pre></div>
<p>The <tt>print</tt> method must be implemented by "analyses" in order to print <p>The <tt>print</tt> method must be implemented by "analyses" in order to print
a human readable version of the analysis results. This is useful for debugging a human readable version of the analysis results. This is useful for debugging
@ -927,9 +933,9 @@ having any prerequisite passes, and invalidating <b>all</b> other passes.</p>
<div class="doc_text"> <div class="doc_text">
<pre> <div class="doc_code"><pre>
<b>virtual void</b> getAnalysisUsage(AnalysisUsage &amp;Info) <b>const</b>; <b>virtual void</b> getAnalysisUsage(AnalysisUsage &amp;Info) <b>const</b>;
</pre> </pre></div>
<p>By implementing the <tt>getAnalysisUsage</tt> method, the required and <p>By implementing the <tt>getAnalysisUsage</tt> method, the required and
invalidated sets may be specified for your transformation. The implementation invalidated sets may be specified for your transformation. The implementation
@ -1008,24 +1014,24 @@ the fact that it hacks on the CFG.
<div class="doc_text"> <div class="doc_text">
<pre> <div class="doc_code"><pre>
<i>// This is an example implementation from an analysis, which does not modify <i>// This is an example implementation from an analysis, which does not modify
// the program at all, yet has a prerequisite.</i> // the program at all, yet has a prerequisite.</i>
<b>void</b> <a href="http://llvm.org/doxygen/classllvm_1_1PostDominanceFrontier.html">PostDominanceFrontier</a>::getAnalysisUsage(AnalysisUsage &amp;AU) <b>const</b> { <b>void</b> <a href="http://llvm.org/doxygen/classllvm_1_1PostDominanceFrontier.html">PostDominanceFrontier</a>::getAnalysisUsage(AnalysisUsage &amp;AU) <b>const</b> {
AU.setPreservesAll(); AU.setPreservesAll();
AU.addRequired&lt;<a href="http://llvm.org/doxygen/classllvm_1_1PostDominatorTree.html">PostDominatorTree</a>&gt;(); AU.addRequired&lt;<a href="http://llvm.org/doxygen/classllvm_1_1PostDominatorTree.html">PostDominatorTree</a>&gt;();
} }
</pre> </pre></div>
<p>and:</p> <p>and:</p>
<pre> <div class="doc_code"><pre>
<i>// This example modifies the program, but does not modify the CFG</i> <i>// This example modifies the program, but does not modify the CFG</i>
<b>void</b> <a href="http://llvm.org/doxygen/structLICM.html">LICM</a>::getAnalysisUsage(AnalysisUsage &amp;AU) <b>const</b> { <b>void</b> <a href="http://llvm.org/doxygen/structLICM.html">LICM</a>::getAnalysisUsage(AnalysisUsage &amp;AU) <b>const</b> {
AU.setPreservesCFG(); AU.setPreservesCFG();
AU.addRequired&lt;<a href="http://llvm.org/doxygen/classllvm_1_1LoopInfo.html">LoopInfo</a>&gt;(); AU.addRequired&lt;<a href="http://llvm.org/doxygen/classllvm_1_1LoopInfo.html">LoopInfo</a>&gt;();
} }
</pre> </pre></div>
</div> </div>
@ -1042,12 +1048,12 @@ required with the <a href="#getAnalysisUsage"><tt>getAnalysisUsage</tt></a>
method. It takes a single template argument that specifies which pass class you method. It takes a single template argument that specifies which pass class you
want, and returns a reference to that pass. For example:</p> want, and returns a reference to that pass. For example:</p>
<pre> <div class="doc_code"><pre>
bool LICM::runOnFunction(Function &amp;F) { bool LICM::runOnFunction(Function &amp;F) {
LoopInfo &amp;LI = getAnalysis&lt;LoopInfo&gt;(); LoopInfo &amp;LI = getAnalysis&lt;LoopInfo&gt;();
... ...
} }
</pre> </pre></div>
<p>This method call returns a reference to the pass desired. You may get a <p>This method call returns a reference to the pass desired. You may get a
runtime assertion failure if you attempt to get an analysis that you did not runtime assertion failure if you attempt to get an analysis that you did not
@ -1062,13 +1068,13 @@ If your pass is capable of updating analyses if they exist (e.g.,
<tt>getAnalysisToUpdate</tt> method, which returns a pointer to the analysis if <tt>getAnalysisToUpdate</tt> method, which returns a pointer to the analysis if
it is active. For example:</p> it is active. For example:</p>
<pre> <div class="doc_code"><pre>
... ...
if (DominatorSet *DS = getAnalysisToUpdate&lt;DominatorSet&gt;()) { if (DominatorSet *DS = getAnalysisToUpdate&lt;DominatorSet&gt;()) {
<i>// A DominatorSet is active. This code will update it.</i> <i>// A DominatorSet is active. This code will update it.</i>
} }
... ...
</pre> </pre></div>
</div> </div>
@ -1163,14 +1169,14 @@ an analysis should be registered, with a human readable name provided for it.
Unlike registration of passes, there is no command line argument to be specified Unlike registration of passes, there is no command line argument to be specified
for the Analysis Group Interface itself, because it is "abstract":</p> for the Analysis Group Interface itself, because it is "abstract":</p>
<pre> <div class="doc_code"><pre>
<b>static</b> RegisterAnalysisGroup&lt;<a href="http://llvm.org/doxygen/classllvm_1_1AliasAnalysis.html">AliasAnalysis</a>&gt; A("<i>Alias Analysis</i>"); <b>static</b> RegisterAnalysisGroup&lt;<a href="http://llvm.org/doxygen/classllvm_1_1AliasAnalysis.html">AliasAnalysis</a>&gt; A("<i>Alias Analysis</i>");
</pre> </pre></div>
<p>Once the analysis is registered, passes can declare that they are valid <p>Once the analysis is registered, passes can declare that they are valid
implementations of the interface by using the following code:</p> implementations of the interface by using the following code:</p>
<pre> <div class="doc_code"><pre>
<b>namespace</b> { <b>namespace</b> {
//<i> Analysis Group implementations <b>must</b> be registered normally...</i> //<i> Analysis Group implementations <b>must</b> be registered normally...</i>
RegisterOpt&lt;FancyAA&gt; RegisterOpt&lt;FancyAA&gt;
@ -1179,7 +1185,7 @@ implementations of the interface by using the following code:</p>
//<i> Declare that we implement the AliasAnalysis interface</i> //<i> Declare that we implement the AliasAnalysis interface</i>
RegisterAnalysisGroup&lt;<a href="http://llvm.org/doxygen/classllvm_1_1AliasAnalysis.html">AliasAnalysis</a>, FancyAA&gt; C; RegisterAnalysisGroup&lt;<a href="http://llvm.org/doxygen/classllvm_1_1AliasAnalysis.html">AliasAnalysis</a>, FancyAA&gt; C;
} }
</pre> </pre></div>
<p>This just shows a class <tt>FancyAA</tt> that is registered normally, then <p>This just shows a class <tt>FancyAA</tt> that is registered normally, then
uses the <tt>RegisterAnalysisGroup</tt> template to "join" the <tt><a uses the <tt>RegisterAnalysisGroup</tt> template to "join" the <tt><a
@ -1188,7 +1194,7 @@ analysis group. Every implementation of an analysis group should join using
this template. A single pass may join multiple different analysis groups with this template. A single pass may join multiple different analysis groups with
no problem.</p> no problem.</p>
<pre> <div class="doc_code"><pre>
<b>namespace</b> { <b>namespace</b> {
//<i> Analysis Group implementations <b>must</b> be registered normally...</i> //<i> Analysis Group implementations <b>must</b> be registered normally...</i>
RegisterOpt&lt;<a href="http://llvm.org/doxygen/structBasicAliasAnalysis.html">BasicAliasAnalysis</a>&gt; RegisterOpt&lt;<a href="http://llvm.org/doxygen/structBasicAliasAnalysis.html">BasicAliasAnalysis</a>&gt;
@ -1197,7 +1203,7 @@ no problem.</p>
//<i> Declare that we implement the AliasAnalysis interface</i> //<i> Declare that we implement the AliasAnalysis interface</i>
RegisterAnalysisGroup&lt;<a href="http://llvm.org/doxygen/classllvm_1_1AliasAnalysis.html">AliasAnalysis</a>, <a href="http://llvm.org/doxygen/structBasicAliasAnalysis.html">BasicAliasAnalysis</a>, <b>true</b>&gt; E; RegisterAnalysisGroup&lt;<a href="http://llvm.org/doxygen/classllvm_1_1AliasAnalysis.html">AliasAnalysis</a>, <a href="http://llvm.org/doxygen/structBasicAliasAnalysis.html">BasicAliasAnalysis</a>, <b>true</b>&gt; E;
} }
</pre> </pre></div>
<p>Here we show how the default implementation is specified (using the extra <p>Here we show how the default implementation is specified (using the extra
argument to the <tt>RegisterAnalysisGroup</tt> template). There must be exactly argument to the <tt>RegisterAnalysisGroup</tt> template). There must be exactly
@ -1290,7 +1296,7 @@ option, just type '<tt>opt --help-hidden</tt>').</p>
how our <a href="#basiccode">Hello World</a> pass interacts with other passes. how our <a href="#basiccode">Hello World</a> pass interacts with other passes.
Lets try it out with the <tt>gcse</tt> and <tt>licm</tt> passes:</p> Lets try it out with the <tt>gcse</tt> and <tt>licm</tt> passes:</p>
<pre> <div class="doc_code"><pre>
$ opt -load ../../../Debug/lib/Hello.so -gcse -licm --debug-pass=Structure &lt; hello.bc &gt; /dev/null $ opt -load ../../../Debug/lib/Hello.so -gcse -licm --debug-pass=Structure &lt; hello.bc &gt; /dev/null
Module Pass Manager Module Pass Manager
Function Pass Manager Function Pass Manager
@ -1308,7 +1314,7 @@ Module Pass Manager
-- Module Verifier -- Module Verifier
Bytecode Writer Bytecode Writer
--Bytecode Writer --Bytecode Writer
</pre> </pre></div>
<p>This output shows us when passes are constructed and when the analysis <p>This output shows us when passes are constructed and when the analysis
results are known to be dead (prefixed with '<tt>--</tt>'). Here we see that results are known to be dead (prefixed with '<tt>--</tt>'). Here we see that
@ -1327,7 +1333,7 @@ passes.</p>
<p>Lets see how this changes when we run the <a href="#basiccode">Hello <p>Lets see how this changes when we run the <a href="#basiccode">Hello
World</a> pass in between the two passes:</p> World</a> pass in between the two passes:</p>
<pre> <div class="doc_code"><pre>
$ opt -load ../../../Debug/lib/Hello.so -gcse -hello -licm --debug-pass=Structure &lt; hello.bc &gt; /dev/null $ opt -load ../../../Debug/lib/Hello.so -gcse -hello -licm --debug-pass=Structure &lt; hello.bc &gt; /dev/null
Module Pass Manager Module Pass Manager
Function Pass Manager Function Pass Manager
@ -1352,23 +1358,23 @@ Module Pass Manager
Hello: __main Hello: __main
Hello: puts Hello: puts
Hello: main Hello: main
</pre> </pre></div>
<p>Here we see that the <a href="#basiccode">Hello World</a> pass has killed the <p>Here we see that the <a href="#basiccode">Hello World</a> pass has killed the
Dominator Set pass, even though it doesn't modify the code at all! To fix this, Dominator Set pass, even though it doesn't modify the code at all! To fix this,
we need to add the following <a we need to add the following <a
href="#getAnalysisUsage"><tt>getAnalysisUsage</tt></a> method to our pass:</p> href="#getAnalysisUsage"><tt>getAnalysisUsage</tt></a> method to our pass:</p>
<pre> <div class="doc_code"><pre>
<i>// We don't modify the program, so we preserve all analyses</i> <i>// We don't modify the program, so we preserve all analyses</i>
<b>virtual void</b> getAnalysisUsage(AnalysisUsage &amp;AU) <b>const</b> { <b>virtual void</b> getAnalysisUsage(AnalysisUsage &amp;AU) <b>const</b> {
AU.setPreservesAll(); AU.setPreservesAll();
} }
</pre> </pre></div>
<p>Now when we run our pass, we get this output:</p> <p>Now when we run our pass, we get this output:</p>
<pre> <div class="doc_code"><pre>
$ opt -load ../../../Debug/lib/Hello.so -gcse -hello -licm --debug-pass=Structure &lt; hello.bc &gt; /dev/null $ opt -load ../../../Debug/lib/Hello.so -gcse -hello -licm --debug-pass=Structure &lt; hello.bc &gt; /dev/null
Pass Arguments: -gcse -hello -licm Pass Arguments: -gcse -hello -licm
Module Pass Manager Module Pass Manager
@ -1392,7 +1398,7 @@ Module Pass Manager
Hello: __main Hello: __main
Hello: puts Hello: puts
Hello: main Hello: main
</pre> </pre></div>
<p>Which shows that we don't accidentally invalidate dominator information <p>Which shows that we don't accidentally invalidate dominator information
anymore, and therefore do not have to compute it twice.</p> anymore, and therefore do not have to compute it twice.</p>
@ -1406,9 +1412,9 @@ anymore, and therefore do not have to compute it twice.</p>
<div class="doc_text"> <div class="doc_text">
<pre> <div class="doc_code"><pre>
<b>virtual void</b> releaseMemory(); <b>virtual void</b> releaseMemory();
</pre> </pre></div>
<p>The <tt>PassManager</tt> automatically determines when to compute analysis <p>The <tt>PassManager</tt> automatically determines when to compute analysis
results, and how long to keep them around for. Because the lifetime of the pass results, and how long to keep them around for. Because the lifetime of the pass
@ -1425,6 +1431,139 @@ class, before the next call of <tt>run*</tt> in your pass.</p>
</div> </div>
<!-- *********************************************************************** -->
<div class="doc_section">
<a name="registering">Registering dynamically loaded passes</a>
</div>
<!-- *********************************************************************** -->
<div class="doc_text">
<p><i>Size matters</i> when constructing production quality tools using llvm,
both for the purposes of distribution, and for regulating the resident code size
when running on the target system. Therefore, it becomes desirable to
selectively use some passes, while omitting others and maintain the flexibility
to change configurations later on. You want to be able to do all this, and,
provide feedback to the user. This is where pass registration comes into
play.</p>
<p>The fundamental mechanisms for pass registration are the
<tt>MachinePassRegistry</tt> class and subclasses of
<tt>MachinePassRegistryNode</tt>.</p>
<p>An instance of <tt>MachinePassRegistry</tt> is used to maintain a list of
<tt>MachinePassRegistryNode</tt> objects. This instance maintains the list and
communicates additions and deletions to the command line interface.</p>
<p>An instance of <tt>MachinePassRegistryNode</tt> subclass is used to maintain
information provided about a particular pass. This information includes the
command line name, the command help string and the address of the function used
to create an instance of the pass. A global static constructor of one of these
instances <i>registers</i> with a corresponding <tt>MachinePassRegistry</tt>,
the static destructor <i>unregisters</i>. Thus a pass that is statically linked
in the tool will be registered at start up. A dynamically loaded pass will
register on load and unregister at unload.</p>
</div>
<!-- _______________________________________________________________________ -->
<div class="doc_subsection">
<a name="registering_existing">Using existing registries</a>
</div>
<div class="doc_text">
<p>There are predefined registries to track instruction scheduling
(<tt>RegisterScheduler</tt>) and register allocation (<tt>RegisterRegAlloc</tt>)
machine passes. Here we will describe how to <i>register</i> a register
allocator machine pass.</p>
<p>Implement your register allocator machine pass. In your register allocator
.cpp file add the following include;</p>
<div class="doc_code"><pre>
#include ""llvm/CodeGen/RegAllocRegistry.h""
</pre></div>
<p>Also in your register allocator .cpp file, define a creator function in the
form; </p>
<div class="doc_code"><pre>
FunctionPass *createMyRegisterAllocator() {
return new MyRegisterAllocator();
}
</pre></div>
<p>Note that the signature of this function should match the type of
<tt>RegisterRegAlloc::FunctionPassCtor</tt>. In the same file add the
"installing" declaration, in the form;</p>
<div class="doc_code"><pre>
static RegisterRegAlloc myRegAlloc("myregalloc",
" my register allocator help string",
createMyRegisterAllocator);
</pre></div>
<p>Note the two spaces prior to the help string produces a tidy result on the
--help query.</p>
<div class="doc_code"><pre>
$ llc --help
...
-regalloc - Register allocator to use: (default = linearscan)
=linearscan - linear scan register allocator
=local - local register allocator
=simple - simple register allocator
=myregalloc - my register allocator help string
...
</pre></div>
<p>And that's it. The user is now free to use <tt>-regalloc=myregalloc</tt> as
an option. Registering instruction schedulers is similar except use the
<tt>RegisterRegAlloc</tt> class. Note that the
<tt>RegisterRegAlloc::FunctionPassCtor</tt> is significantly different from
<tt>RegisterRegAlloc::FunctionPassCtor</tt>.</p>
<p>To force the load/linking of your register allocator into the llc/lli tools,
add your creator function's global declaration to "Passes.h" and add a "pseudo"
call line to <tt>llvm/Codegen/LinkAllCodegenComponents.h</tt>.</p>
</div>
<!-- _______________________________________________________________________ -->
<div class="doc_subsection">
<a name="registering_new">Creating new registries</a>
</div>
<div class="doc_text">
<p>The easiest way to get started is to clone one of the existing registries; we
recommend <tt>llvm/CodeGen/RegAllocRegistry.h</tt>. The key things to modify
are the class name and the <tt>FunctionPassCtor</tt> type.</p>
<p>Then you need to declare the registry. Example: if your pass registry is
<tt>RegisterMyPasses</tt> then define;</p>
<div class="doc_code"><pre>
MachinePassRegistry RegisterMyPasses::Registry;
</pre></div>
<p>And finally, declare the command line option for your passes. Example:</p>
<div class="doc_code"><pre>
cl::opt&lt;RegisterMyPasses::FunctionPassCtor, false,
RegisterPassParser&lt;RegisterMyPasses&gt &gt
MyPassOpt("mypass",
cl::init(&amp;createDefaultMyPass),
cl::desc("my pass option help"));
</pre></div>
<p>Here the command option is "mypass", with createDefaultMyPass as the default
creator.</p>
</div>
<!-- *********************************************************************** --> <!-- *********************************************************************** -->
<div class="doc_section"> <div class="doc_section">
<a name="debughints">Using GDB with dynamically loaded passes</a> <a name="debughints">Using GDB with dynamically loaded passes</a>
@ -1454,7 +1593,7 @@ on that.</p>
<p>First thing you do is start <tt>gdb</tt> on the <tt>opt</tt> process:</p> <p>First thing you do is start <tt>gdb</tt> on the <tt>opt</tt> process:</p>
<pre> <div class="doc_code"><pre>
$ <b>gdb opt</b> $ <b>gdb opt</b>
GNU gdb 5.0 GNU gdb 5.0
Copyright 2000 Free Software Foundation, Inc. Copyright 2000 Free Software Foundation, Inc.
@ -1464,7 +1603,7 @@ Type "show copying" to see the conditions.
There is absolutely no warranty for GDB. Type "show warranty" for details. There is absolutely no warranty for GDB. Type "show warranty" for details.
This GDB was configured as "sparc-sun-solaris2.6"... This GDB was configured as "sparc-sun-solaris2.6"...
(gdb) (gdb)
</pre> </pre></div>
<p>Note that <tt>opt</tt> has a lot of debugging information in it, so it takes <p>Note that <tt>opt</tt> has a lot of debugging information in it, so it takes
time to load. Be patient. Since we cannot set a breakpoint in our pass yet time to load. Be patient. Since we cannot set a breakpoint in our pass yet
@ -1474,7 +1613,7 @@ object. The most foolproof way of doing this is to set a breakpoint in
<tt>PassManager::run</tt> and then run the process with the arguments you <tt>PassManager::run</tt> and then run the process with the arguments you
want:</p> want:</p>
<pre> <div class="doc_code"><pre>
(gdb) <b>break PassManager::run</b> (gdb) <b>break PassManager::run</b>
Breakpoint 1 at 0x2413bc: file Pass.cpp, line 70. Breakpoint 1 at 0x2413bc: file Pass.cpp, line 70.
(gdb) <b>run test.bc -load $(LLVMTOP)/llvm/Debug/lib/[libname].so -[passoption]</b> (gdb) <b>run test.bc -load $(LLVMTOP)/llvm/Debug/lib/[libname].so -[passoption]</b>
@ -1482,7 +1621,7 @@ Starting program: opt test.bc -load $(LLVMTOP)/llvm/Debug/lib/[libname].so -[pas
Breakpoint 1, PassManager::run (this=0xffbef174, M=@0x70b298) at Pass.cpp:70 Breakpoint 1, PassManager::run (this=0xffbef174, M=@0x70b298) at Pass.cpp:70
70 bool PassManager::run(Module &amp;M) { return PM-&gt;run(M); } 70 bool PassManager::run(Module &amp;M) { return PM-&gt;run(M); }
(gdb) (gdb)
</pre> </pre></div>
<p>Once the <tt>opt</tt> stops in the <tt>PassManager::run</tt> method you are <p>Once the <tt>opt</tt> stops in the <tt>PassManager::run</tt> method you are
now free to set breakpoints in your pass so that you can trace through execution now free to set breakpoints in your pass so that you can trace through execution