<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" 
                      "http://www.w3.org/TR/html4/strict.dtd">
<html>
<head>
 <title>LLVM Link Time Optimization: Design and Implementation</title>
  <link rel="stylesheet" href="llvm.css" type="text/css">
</head>

<div class="doc_title">
  LLVM Link Time Optimization: Design and Implementation
</div>

<ul>
  <li><a href="#desc">Description</a></li>
  <li><a href="#design">Design Philosophy</a>
  <ul>
    <li><a href="#example1">Example of link time optimization</a></li>
    <li><a href="#alternative_approaches">Alternative Approaches</a></li>
  </ul></li>
  <li><a href="#multiphase">Multi-phase communication between LLVM and linker</a>
  <ul>
    <li><a href="#phase1">Phase 1 : Read LLVM Bytecode Files</a></li>
    <li><a href="#phase2">Phase 2 : Symbol Resolution</a></li>
    <li><a href="#phase3">Phase 3 : Optimize Bitcode Files</a></li>
    <li><a href="#phase4">Phase 4 : Symbol Resolution after optimization</a></li>
  </ul></li>
  <li><a href="#lto">libLTO</a>
  <ul>
    <li><a href="#lto_module_t">lto_module_t</a></li>
    <li><a href="#lto_code_gen_t">lto_code_gen_t</a></li>
  </ul>
</ul>

<div class="doc_author">
<p>Written by Devang Patel and Nick Kledzik</p>
</div>

<!-- *********************************************************************** -->
<div class="doc_section">
<a name="desc">Description</a>
</div>
<!-- *********************************************************************** -->

<div class="doc_text">
<p>
LLVM features powerful intermodular optimizations which can be used at link 
time.  Link Time Optimization (LTO) is another name for intermodular optimization 
when performed during the link stage. This document describes the interface 
and design between the LTO optimizer and the linker.</p>
</div>

<!-- *********************************************************************** -->
<div class="doc_section">
<a name="design">Design Philosophy</a>
</div>
<!-- *********************************************************************** -->

<div class="doc_text">
<p>
The LLVM Link Time Optimizer provides complete transparency, while doing 
intermodular optimization, in the compiler tool chain. Its main goal is to let 
the developer take advantage of intermodular optimizations without making any 
significant changes to the developer's makefiles or build system. This is 
achieved through tight integration with the linker. In this model, the linker 
treates LLVM bitcode files like native object files and allows mixing and 
matching among them. The linker uses <a href="#lto">libLTO</a>, a shared
object, to handle LLVM bitcode files. This tight integration between 
the linker and LLVM optimizer helps to do optimizations that are not possible 
in other models. The linker input allows the optimizer to avoid relying on 
conservative escape analysis.
</p>
</div>

<!-- ======================================================================= -->
<div class="doc_subsection">
  <a name="example1">Example of link time optimization</a>
</div>

<div class="doc_text">
  <p>The following example illustrates the advantages of LTO's integrated
  approach and clean interface. This example requires a system linker which
  supports LTO through the interface described in this document.  Here,
  llvm-gcc transparently invokes system linker. </p>
  <ul>
    <li> Input source file <tt>a.c</tt> is compiled into LLVM bitcode form.
    <li> Input source file <tt>main.c</tt> is compiled into native object code.
  </ul>
<div class="doc_code"><pre>
--- a.h ---
extern int foo1(void);
extern void foo2(void);
extern void foo4(void);
--- a.c ---
#include "a.h"

static signed int i = 0;

void foo2(void) {
 i = -1;
}

static int foo3() {
foo4();
return 10;
}

int foo1(void) {
int data = 0;

if (i &lt; 0) { data = foo3(); }

data = data + 42;
return data;
}

--- main.c ---
#include &lt;stdio.h&gt;
#include "a.h"

void foo4(void) {
 printf ("Hi\n");
}

int main() {
 return foo1();
}

--- command lines ---
$ llvm-gcc --emit-llvm -c a.c -o a.o  # &lt;-- a.o is LLVM bitcode file
$ llvm-gcc -c main.c -o main.o # &lt;-- main.o is native object file
$ llvm-gcc a.o main.o -o main # &lt;-- standard link command without any modifications
</pre></div>
  <p>In this example, the linker recognizes that <tt>foo2()</tt> is an 
  externally visible symbol defined in LLVM bitcode file. The linker completes 
  its usual symbol resolution 
  pass and finds that <tt>foo2()</tt> is not used anywhere. This information 
  is used by the LLVM optimizer and it removes <tt>foo2()</tt>. As soon as 
  <tt>foo2()</tt> is removed, the optimizer recognizes that condition 
  <tt>i &lt; 0</tt> is always false, which means <tt>foo3()</tt> is never 
  used. Hence, the optimizer removes <tt>foo3()</tt>, also.  And this in turn, 
  enables linker to remove <tt>foo4()</tt>.  This example illustrates the 
  advantage of tight integration with the linker. Here, the optimizer can not 
  remove <tt>foo3()</tt> without the linker's input.
  </p>
</div>

<!-- ======================================================================= -->
<div class="doc_subsection">
  <a name="alternative_approaches">Alternative Approaches</a>
</div>

<div class="doc_text">
  <dl>
    <dt><b>Compiler driver invokes link time optimizer separately.</b></dt>
    <dd>In this model the link time optimizer is not able to take advantage of 
    information collected during the linker's normal symbol resolution phase. 
    In the above example, the optimizer can not remove <tt>foo2()</tt> without 
    the linker's input because it is externally visible. This in turn prohibits
    the optimizer from removing <tt>foo3()</tt>.</dd>
    <dt><b>Use separate tool to collect symbol information from all object
    files.</b></dt>
    <dd>In this model, a new, separate, tool or library replicates the linker's
    capability to collect information for link time optimization. Not only is
    this code duplication difficult to justify, but it also has several other 
    disadvantages.  For example, the linking semantics and the features 
    provided by the linker on various platform are not unique. This means, 
    this new tool needs to support all such features and platforms in one 
    super tool or a separate tool per platform is required. This increases 
    maintance cost for link time optimizer significantly, which is not 
    necessary. This approach also requires staying synchronized with linker 
    developements on various platforms, which is not the main focus of the link 
    time optimizer. Finally, this approach increases end user's build time due 
    to the duplication of work done by this separate tool and the linker itself.
    </dd>
  </dl>
</div>

<!-- *********************************************************************** -->
<div class="doc_section">
  <a name="multiphase">Multi-phase communication between libLTO and linker</a>
</div>

<div class="doc_text">
  <p>The linker collects information about symbol defininitions and uses in 
  various link objects which is more accurate than any information collected 
  by other tools during typical build cycles.  The linker collects this 
  information by looking at the definitions and uses of symbols in native .o 
  files and using symbol visibility information. The linker also uses 
  user-supplied information, such as a list of exported symbols. LLVM 
  optimizer collects control flow information, data flow information and knows 
  much more about program structure from the optimizer's point of view. 
  Our goal is to take advantage of tight intergration between the linker and 
  the optimizer by sharing this information during various linking phases.
</p>
</div>

<!-- ======================================================================= -->
<div class="doc_subsection">
  <a name="phase1">Phase 1 : Read LLVM Bitcode Files</a>
</div>

<div class="doc_text">
  <p>The linker first reads all object files in natural order and collects 
  symbol information. This includes native object files as well as LLVM bitcode 
  files.  To minimize the cost to the linker in the case that all .o files
  are native object files, the linker only calls <tt>lto_module_create()</tt> 
  when a supplied object file is found to not be a native object file.  If
  <tt>lto_module_create()</tt> returns that the file is an LLVM bitcode file, 
  the linker
  then iterates over the module using <tt>lto_module_get_symbol_name()</tt> and
  <tt>lto_module_get_symbol_attribute()</tt> to get all symbols defined and 
  referenced.
  This information is added to the linker's global symbol table.
</p>
  <p>The lto* functions are all implemented in a shared object libLTO.  This
  allows the LLVM LTO code to be updated independently of the linker tool.
  On platforms that support it, the shared object is lazily loaded. 
</p>
</div>

<!-- ======================================================================= -->
<div class="doc_subsection">
  <a name="phase2">Phase 2 : Symbol Resolution</a>
</div>

<div class="doc_text">
  <p>In this stage, the linker resolves symbols using global symbol table. 
  It may report undefined symbol errors, read archive members, replace 
  weak symbols, etc.  The linker is able to do this seamlessly even though it 
  does not know the exact content of input LLVM bitcode files.  If dead code 
  stripping is enabled then the linker collects the list of live symbols.
  </p>
</div>

<!-- ======================================================================= -->
<div class="doc_subsection">
  <a name="phase3">Phase 3 : Optimize Bitcode Files</a>
</div>
<div class="doc_text">
  <p>After symbol resolution, the linker tells the LTO shared object which
  symbols are needed by native object files.  In the example above, the linker 
  reports that only <tt>foo1()</tt> is used by native object files using 
  <tt>lto_codegen_add_must_preserve_symbol()</tt>.  Next the linker invokes
  the LLVM optimizer and code generators using <tt>lto_codegen_compile()</tt>
  which returns a native object file creating by merging the LLVM bitcode files 
  and applying various optimization passes.  
</p>
</div>

<!-- ======================================================================= -->
<div class="doc_subsection">
  <a name="phase4">Phase 4 : Symbol Resolution after optimization</a>
</div>

<div class="doc_text">
  <p>In this phase, the linker reads optimized a native object file and 
  updates the internal global symbol table to reflect any changes. The linker 
  also collects information about any changes in use of external symbols by 
  LLVM bitcode files. In the examle above, the linker notes that 
  <tt>foo4()</tt> is not used any more. If dead code stripping is enabled then 
  the linker refreshes the live symbol information appropriately and performs 
  dead code stripping.</p>
  <p>After this phase, the linker continues linking as if it never saw LLVM 
  bitcode files.</p>
</div>

<!-- *********************************************************************** -->
<div class="doc_section">
<a name="lto">libLTO</a>
</div>

<div class="doc_text">
  <p><tt>libLTO</tt> is a shared object that is part of the LLVM tools, and 
  is intended for use by a linker. <tt>libLTO</tt> provides an abstract C 
  interface to use the LLVM interprocedural optimizer without exposing details 
  of LLVM's internals. The intention is to keep the interface as stable as 
  possible even when the LLVM optimizer continues to evolve. It should even
  be possible for a completely different compilation technology to provide
  a different libLTO that works with their object files and the standard
  linker tool.</p>
</div>

<!-- ======================================================================= -->
<div class="doc_subsection">
  <a name="lto_module_t">lto_module_t</a>
</div>

<div class="doc_text">
   <p>A non-native object file is handled via an <tt>lto_module_t</tt>.  
   The following functions allow the linker to check if a file (on disk
   or in a memory buffer) is a file which libLTO can process: <pre>
   lto_module_is_object_file(const char*)
   lto_module_is_object_file_for_target(const char*, const char*)
   lto_module_is_object_file_in_memory(const void*, size_t)
   lto_module_is_object_file_in_memory_for_target(const void*, size_t, const char*)</pre>
  If the object file can be processed by libLTO, the linker creates a
  <tt>lto_module_t</tt> by using one of <pre>
   lto_module_create(const char*)
   lto_module_create_from_memory(const void*, size_t)</pre>
  and when done, the handle is released via<pre>
   lto_module_dispose(lto_module_t)</pre>
  The linker can introspect the non-native object file by getting the number
  of symbols and getting the name and attributes of each symbol via: <pre>
   lto_module_get_num_symbols(lto_module_t)
   lto_module_get_symbol_name(lto_module_t, unsigned int)
   lto_module_get_symbol_attribute(lto_module_t, unsigned int)</pre>
  The attributes of a symbol include the alignment, visibility, and kind.
</p>
</div>

<!-- ======================================================================= -->
<div class="doc_subsection">
  <a name="lto_code_gen_t">lto_code_gen_t</a>
</div>

<div class="doc_text">
  <p>Once the linker has loaded each non-native object files into an
  <tt>lto_module_t</tt>, it can request libLTO to process them all and
  generate a native object file.  This is done in a couple of steps.
  First a code generator is created with:<pre>
    lto_codegen_create() </pre>
  then each non-native object file is added to the code generator with:<pre>
    lto_codegen_add_module(lto_code_gen_t, lto_module_t)</pre>
  The linker then has the option of setting some codegen options.  Whether
  or not to generate DWARF debug info is set with: <pre>
    lto_codegen_set_debug_model(lto_code_gen_t) </pre>
  Which kind of position independence is set with: <pre>
    lto_codegen_set_pic_model(lto_code_gen_t) </pre>
  And each symbol that is referenced by a native object file or otherwise
  must not be optimized away is set with: <pre>
    lto_codegen_add_must_preserve_symbol(lto_code_gen_t, const char*)</pre>
  After all these settings are done, the linker requests that a native 
  object file be created from the modules with the settings using:
    lto_codegen_compile(lto_code_gen_t, size*)</pre>
  which returns a pointer to a buffer containing the generated native
  object file.  The linker then parses that and links it with the rest 
  of the native object files.  
</div>

<!-- *********************************************************************** -->

<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>

  Devang Patel and Nick Kledzik<br>
  <a href="http://llvm.org">LLVM Compiler Infrastructure</a><br>
  Last modified: $Date$
</address>

</body>
</html>