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