1======================================================
2LLVM Link Time Optimization: Design and Implementation
3======================================================
4
5.. contents::
6   :local:
7
8Description
9===========
10
11LLVM features powerful intermodular optimizations which can be used at link
12time.  Link Time Optimization (LTO) is another name for intermodular
13optimization when performed during the link stage. This document describes the
14interface and design between the LTO optimizer and the linker.
15
16Design Philosophy
17=================
18
19The LLVM Link Time Optimizer provides complete transparency, while doing
20intermodular optimization, in the compiler tool chain. Its main goal is to let
21the developer take advantage of intermodular optimizations without making any
22significant changes to the developer's makefiles or build system. This is
23achieved through tight integration with the linker. In this model, the linker
24treates LLVM bitcode files like native object files and allows mixing and
25matching among them. The linker uses `libLTO`_, a shared object, to handle LLVM
26bitcode files. This tight integration between the linker and LLVM optimizer
27helps to do optimizations that are not possible in other models. The linker
28input allows the optimizer to avoid relying on conservative escape analysis.
29
30.. _libLTO-example:
31
32Example of link time optimization
33---------------------------------
34
35The following example illustrates the advantages of LTO's integrated approach
36and clean interface. This example requires a system linker which supports LTO
37through the interface described in this document.  Here, clang transparently
38invokes system linker.
39
40* Input source file ``a.c`` is compiled into LLVM bitcode form.
41* Input source file ``main.c`` is compiled into native object code.
42
43.. code-block:: c++
44
45  --- a.h ---
46  extern int foo1(void);
47  extern void foo2(void);
48  extern void foo4(void);
49
50  --- a.c ---
51  #include "a.h"
52
53  static signed int i = 0;
54
55  void foo2(void) {
56    i = -1;
57  }
58
59  static int foo3() {
60    foo4();
61    return 10;
62  }
63
64  int foo1(void) {
65    int data = 0;
66
67    if (i < 0)
68      data = foo3();
69
70    data = data + 42;
71    return data;
72  }
73
74  --- main.c ---
75  #include <stdio.h>
76  #include "a.h"
77
78  void foo4(void) {
79    printf("Hi\n");
80  }
81
82  int main() {
83    return foo1();
84  }
85
86To compile, run:
87
88.. code-block:: console
89
90  % clang -emit-llvm -c a.c -o a.o   # <-- a.o is LLVM bitcode file
91  % clang -c main.c -o main.o        # <-- main.o is native object file
92  % clang a.o main.o -o main         # <-- standard link command without modifications
93
94* In this example, the linker recognizes that ``foo2()`` is an externally
95  visible symbol defined in LLVM bitcode file. The linker completes its usual
96  symbol resolution pass and finds that ``foo2()`` is not used
97  anywhere. This information is used by the LLVM optimizer and it
98  removes ``foo2()``.
99
100* As soon as ``foo2()`` is removed, the optimizer recognizes that condition ``i
101  < 0`` is always false, which means ``foo3()`` is never used. Hence, the
102  optimizer also removes ``foo3()``.
103
104* And this in turn, enables linker to remove ``foo4()``.
105
106This example illustrates the advantage of tight integration with the
107linker. Here, the optimizer can not remove ``foo3()`` without the linker's
108input.
109
110Alternative Approaches
111----------------------
112
113**Compiler driver invokes link time optimizer separately.**
114    In this model the link time optimizer is not able to take advantage of
115    information collected during the linker's normal symbol resolution phase.
116    In the above example, the optimizer can not remove ``foo2()`` without the
117    linker's input because it is externally visible. This in turn prohibits the
118    optimizer from removing ``foo3()``.
119
120**Use separate tool to collect symbol information from all object files.**
121    In this model, a new, separate, tool or library replicates the linker's
122    capability to collect information for link time optimization. Not only is
123    this code duplication difficult to justify, but it also has several other
124    disadvantages.  For example, the linking semantics and the features provided
125    by the linker on various platform are not unique. This means, this new tool
126    needs to support all such features and platforms in one super tool or a
127    separate tool per platform is required. This increases maintenance cost for
128    link time optimizer significantly, which is not necessary. This approach
129    also requires staying synchronized with linker developements on various
130    platforms, which is not the main focus of the link time optimizer. Finally,
131    this approach increases end user's build time due to the duplication of work
132    done by this separate tool and the linker itself.
133
134Multi-phase communication between ``libLTO`` and linker
135=======================================================
136
137The linker collects information about symbol definitions and uses in various
138link objects which is more accurate than any information collected by other
139tools during typical build cycles.  The linker collects this information by
140looking at the definitions and uses of symbols in native .o files and using
141symbol visibility information. The linker also uses user-supplied information,
142such as a list of exported symbols. LLVM optimizer collects control flow
143information, data flow information and knows much more about program structure
144from the optimizer's point of view.  Our goal is to take advantage of tight
145integration between the linker and the optimizer by sharing this information
146during various linking phases.
147
148Phase 1 : Read LLVM Bitcode Files
149---------------------------------
150
151The linker first reads all object files in natural order and collects symbol
152information. This includes native object files as well as LLVM bitcode files.
153To minimize the cost to the linker in the case that all .o files are native
154object files, the linker only calls ``lto_module_create()`` when a supplied
155object file is found to not be a native object file.  If ``lto_module_create()``
156returns that the file is an LLVM bitcode file, the linker then iterates over the
157module using ``lto_module_get_symbol_name()`` and
158``lto_module_get_symbol_attribute()`` to get all symbols defined and referenced.
159This information is added to the linker's global symbol table.
160
161
162The lto* functions are all implemented in a shared object libLTO.  This allows
163the LLVM LTO code to be updated independently of the linker tool.  On platforms
164that support it, the shared object is lazily loaded.
165
166Phase 2 : Symbol Resolution
167---------------------------
168
169In this stage, the linker resolves symbols using global symbol table.  It may
170report undefined symbol errors, read archive members, replace weak symbols, etc.
171The linker is able to do this seamlessly even though it does not know the exact
172content of input LLVM bitcode files.  If dead code stripping is enabled then the
173linker collects the list of live symbols.
174
175Phase 3 : Optimize Bitcode Files
176--------------------------------
177
178After symbol resolution, the linker tells the LTO shared object which symbols
179are needed by native object files.  In the example above, the linker reports
180that only ``foo1()`` is used by native object files using
181``lto_codegen_add_must_preserve_symbol()``.  Next the linker invokes the LLVM
182optimizer and code generators using ``lto_codegen_compile()`` which returns a
183native object file creating by merging the LLVM bitcode files and applying
184various optimization passes.
185
186Phase 4 : Symbol Resolution after optimization
187----------------------------------------------
188
189In this phase, the linker reads optimized a native object file and updates the
190internal global symbol table to reflect any changes. The linker also collects
191information about any changes in use of external symbols by LLVM bitcode
192files. In the example above, the linker notes that ``foo4()`` is not used any
193more. If dead code stripping is enabled then the linker refreshes the live
194symbol information appropriately and performs dead code stripping.
195
196After this phase, the linker continues linking as if it never saw LLVM bitcode
197files.
198
199.. _libLTO:
200
201``libLTO``
202==========
203
204``libLTO`` is a shared object that is part of the LLVM tools, and is intended
205for use by a linker. ``libLTO`` provides an abstract C interface to use the LLVM
206interprocedural optimizer without exposing details of LLVM's internals. The
207intention is to keep the interface as stable as possible even when the LLVM
208optimizer continues to evolve. It should even be possible for a completely
209different compilation technology to provide a different libLTO that works with
210their object files and the standard linker tool.
211
212``lto_module_t``
213----------------
214
215A non-native object file is handled via an ``lto_module_t``.  The following
216functions allow the linker to check if a file (on disk or in a memory buffer) is
217a file which libLTO can process:
218
219.. code-block:: c
220
221  lto_module_is_object_file(const char*)
222  lto_module_is_object_file_for_target(const char*, const char*)
223  lto_module_is_object_file_in_memory(const void*, size_t)
224  lto_module_is_object_file_in_memory_for_target(const void*, size_t, const char*)
225
226If the object file can be processed by ``libLTO``, the linker creates a
227``lto_module_t`` by using one of:
228
229.. code-block:: c
230
231  lto_module_create(const char*)
232  lto_module_create_from_memory(const void*, size_t)
233
234and when done, the handle is released via
235
236.. code-block:: c
237
238  lto_module_dispose(lto_module_t)
239
240
241The linker can introspect the non-native object file by getting the number of
242symbols and getting the name and attributes of each symbol via:
243
244.. code-block:: c
245
246  lto_module_get_num_symbols(lto_module_t)
247  lto_module_get_symbol_name(lto_module_t, unsigned int)
248  lto_module_get_symbol_attribute(lto_module_t, unsigned int)
249
250The attributes of a symbol include the alignment, visibility, and kind.
251
252``lto_code_gen_t``
253------------------
254
255Once the linker has loaded each non-native object files into an
256``lto_module_t``, it can request ``libLTO`` to process them all and generate a
257native object file.  This is done in a couple of steps.  First, a code generator
258is created with:
259
260.. code-block:: c
261
262  lto_codegen_create()
263
264Then, each non-native object file is added to the code generator with:
265
266.. code-block:: c
267
268  lto_codegen_add_module(lto_code_gen_t, lto_module_t)
269
270The linker then has the option of setting some codegen options.  Whether or not
271to generate DWARF debug info is set with:
272
273.. code-block:: c
274
275  lto_codegen_set_debug_model(lto_code_gen_t)
276
277Which kind of position independence is set with:
278
279.. code-block:: c
280
281  lto_codegen_set_pic_model(lto_code_gen_t)
282
283And each symbol that is referenced by a native object file or otherwise must not
284be optimized away is set with:
285
286.. code-block:: c
287
288  lto_codegen_add_must_preserve_symbol(lto_code_gen_t, const char*)
289
290After all these settings are done, the linker requests that a native object file
291be created from the modules with the settings using:
292
293.. code-block:: c
294
295  lto_codegen_compile(lto_code_gen_t, size*)
296
297which returns a pointer to a buffer containing the generated native object file.
298The linker then parses that and links it with the rest of the native object
299files.
300