Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 1 | <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" |
| 2 | "http://www.w3.org/TR/html4/strict.dtd"> |
| 3 | <html> |
| 4 | <head> |
| 5 | <title>The LLVM Target-Independent Code Generator</title> |
| 6 | <link rel="stylesheet" href="llvm.css" type="text/css"> |
| 7 | </head> |
| 8 | <body> |
| 9 | |
| 10 | <div class="doc_title"> |
| 11 | The LLVM Target-Independent Code Generator |
| 12 | </div> |
| 13 | |
| 14 | <ol> |
| 15 | <li><a href="#introduction">Introduction</a> |
| 16 | <ul> |
| 17 | <li><a href="#required">Required components in the code generator</a></li> |
Chris Lattner | e35d3bb | 2005-10-16 00:36:38 +0000 | [diff] [blame] | 18 | <li><a href="#high-level-design">The high-level design of the code |
| 19 | generator</a></li> |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 20 | <li><a href="#tablegen">Using TableGen for target description</a></li> |
| 21 | </ul> |
| 22 | </li> |
| 23 | <li><a href="#targetdesc">Target description classes</a> |
| 24 | <ul> |
| 25 | <li><a href="#targetmachine">The <tt>TargetMachine</tt> class</a></li> |
| 26 | <li><a href="#targetdata">The <tt>TargetData</tt> class</a></li> |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 27 | <li><a href="#targetlowering">The <tt>TargetLowering</tt> class</a></li> |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 28 | <li><a href="#mregisterinfo">The <tt>MRegisterInfo</tt> class</a></li> |
| 29 | <li><a href="#targetinstrinfo">The <tt>TargetInstrInfo</tt> class</a></li> |
| 30 | <li><a href="#targetframeinfo">The <tt>TargetFrameInfo</tt> class</a></li> |
Chris Lattner | 47adebb | 2005-10-16 17:06:07 +0000 | [diff] [blame^] | 31 | <li><a href="#targetsubtarget">The <tt>TargetSubtarget</tt> class</a></li> |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 32 | <li><a href="#targetjitinfo">The <tt>TargetJITInfo</tt> class</a></li> |
| 33 | </ul> |
| 34 | </li> |
| 35 | <li><a href="#codegendesc">Machine code description classes</a> |
Chris Lattner | ec94f80 | 2004-06-04 00:16:02 +0000 | [diff] [blame] | 36 | <ul> |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 37 | <li><a href="#machineinstr">The <tt>MachineInstr</tt> class</a></li> |
Chris Lattner | ec94f80 | 2004-06-04 00:16:02 +0000 | [diff] [blame] | 38 | </ul> |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 39 | </li> |
| 40 | <li><a href="#codegenalgs">Target-independent code generation algorithms</a> |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 41 | <ul> |
| 42 | <li><a href="#instselect">Instruction Selection</a> |
| 43 | <ul> |
| 44 | <li><a href="#selectiondag_intro">Introduction to SelectionDAGs</a></li> |
| 45 | <li><a href="#selectiondag_process">SelectionDAG Code Generation |
| 46 | Process</a></li> |
| 47 | <li><a href="#selectiondag_build">Initial SelectionDAG |
| 48 | Construction</a></li> |
| 49 | <li><a href="#selectiondag_legalize">SelectionDAG Legalize Phase</a></li> |
| 50 | <li><a href="#selectiondag_optimize">SelectionDAG Optimization |
Chris Lattner | e35d3bb | 2005-10-16 00:36:38 +0000 | [diff] [blame] | 51 | Phase: the DAG Combiner</a></li> |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 52 | <li><a href="#selectiondag_select">SelectionDAG Select Phase</a></li> |
Chris Lattner | e35d3bb | 2005-10-16 00:36:38 +0000 | [diff] [blame] | 53 | <li><a href="#selectiondag_sched">SelectionDAG Scheduling and Emission |
| 54 | Phase</a></li> |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 55 | <li><a href="#selectiondag_future">Future directions for the |
| 56 | SelectionDAG</a></li> |
| 57 | </ul></li> |
| 58 | </ul> |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 59 | </li> |
| 60 | <li><a href="#targetimpls">Target description implementations</a> |
| 61 | <ul> |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 62 | <li><a href="#x86">The X86 backend</a></li> |
Chris Lattner | 10d6800 | 2004-06-01 17:18:11 +0000 | [diff] [blame] | 63 | </ul> |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 64 | </li> |
| 65 | |
| 66 | </ol> |
| 67 | |
| 68 | <div class="doc_author"> |
| 69 | <p>Written by <a href="mailto:sabre@nondot.org">Chris Lattner</a></p> |
| 70 | </div> |
| 71 | |
Chris Lattner | 10d6800 | 2004-06-01 17:18:11 +0000 | [diff] [blame] | 72 | <div class="doc_warning"> |
| 73 | <p>Warning: This is a work in progress.</p> |
| 74 | </div> |
| 75 | |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 76 | <!-- *********************************************************************** --> |
| 77 | <div class="doc_section"> |
| 78 | <a name="introduction">Introduction</a> |
| 79 | </div> |
| 80 | <!-- *********************************************************************** --> |
| 81 | |
| 82 | <div class="doc_text"> |
| 83 | |
| 84 | <p>The LLVM target-independent code generator is a framework that provides a |
| 85 | suite of reusable components for translating the LLVM internal representation to |
| 86 | the machine code for a specified target -- either in assembly form (suitable for |
| 87 | a static compiler) or in binary machine code format (usable for a JIT compiler). |
Chris Lattner | ec94f80 | 2004-06-04 00:16:02 +0000 | [diff] [blame] | 88 | The LLVM target-independent code generator consists of five main components:</p> |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 89 | |
| 90 | <ol> |
| 91 | <li><a href="#targetdesc">Abstract target description</a> interfaces which |
Reid Spencer | bdbcb8a | 2004-06-05 14:39:24 +0000 | [diff] [blame] | 92 | capture important properties about various aspects of the machine, independently |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 93 | of how they will be used. These interfaces are defined in |
| 94 | <tt>include/llvm/Target/</tt>.</li> |
| 95 | |
| 96 | <li>Classes used to represent the <a href="#codegendesc">machine code</a> being |
Reid Spencer | bdbcb8a | 2004-06-05 14:39:24 +0000 | [diff] [blame] | 97 | generated for a target. These classes are intended to be abstract enough to |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 98 | represent the machine code for <i>any</i> target machine. These classes are |
| 99 | defined in <tt>include/llvm/CodeGen/</tt>.</li> |
| 100 | |
| 101 | <li><a href="#codegenalgs">Target-independent algorithms</a> used to implement |
| 102 | various phases of native code generation (register allocation, scheduling, stack |
| 103 | frame representation, etc). This code lives in <tt>lib/CodeGen/</tt>.</li> |
| 104 | |
| 105 | <li><a href="#targetimpls">Implementations of the abstract target description |
| 106 | interfaces</a> for particular targets. These machine descriptions make use of |
| 107 | the components provided by LLVM, and can optionally provide custom |
| 108 | target-specific passes, to build complete code generators for a specific target. |
| 109 | Target descriptions live in <tt>lib/Target/</tt>.</li> |
| 110 | |
Chris Lattner | ec94f80 | 2004-06-04 00:16:02 +0000 | [diff] [blame] | 111 | <li><a href="#jit">The target-independent JIT components</a>. The LLVM JIT is |
| 112 | completely target independent (it uses the <tt>TargetJITInfo</tt> structure to |
| 113 | interface for target-specific issues. The code for the target-independent |
| 114 | JIT lives in <tt>lib/ExecutionEngine/JIT</tt>.</li> |
| 115 | |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 116 | </ol> |
| 117 | |
| 118 | <p> |
| 119 | Depending on which part of the code generator you are interested in working on, |
| 120 | different pieces of this will be useful to you. In any case, you should be |
| 121 | familiar with the <a href="#targetdesc">target description</a> and <a |
| 122 | href="#codegendesc">machine code representation</a> classes. If you want to add |
Reid Spencer | bdbcb8a | 2004-06-05 14:39:24 +0000 | [diff] [blame] | 123 | a backend for a new target, you will need to <a href="#targetimpls">implement the |
| 124 | target description</a> classes for your new target and understand the <a |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 125 | href="LangRef.html">LLVM code representation</a>. If you are interested in |
| 126 | implementing a new <a href="#codegenalgs">code generation algorithm</a>, it |
| 127 | should only depend on the target-description and machine code representation |
| 128 | classes, ensuring that it is portable. |
| 129 | </p> |
| 130 | |
| 131 | </div> |
| 132 | |
| 133 | <!-- ======================================================================= --> |
| 134 | <div class="doc_subsection"> |
| 135 | <a name="required">Required components in the code generator</a> |
| 136 | </div> |
| 137 | |
| 138 | <div class="doc_text"> |
| 139 | |
| 140 | <p>The two pieces of the LLVM code generator are the high-level interface to the |
| 141 | code generator and the set of reusable components that can be used to build |
| 142 | target-specific backends. The two most important interfaces (<a |
| 143 | href="#targetmachine"><tt>TargetMachine</tt></a> and <a |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 144 | href="#targetdata"><tt>TargetData</tt></a>) are the only ones that are |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 145 | required to be defined for a backend to fit into the LLVM system, but the others |
| 146 | must be defined if the reusable code generator components are going to be |
| 147 | used.</p> |
| 148 | |
| 149 | <p>This design has two important implications. The first is that LLVM can |
| 150 | support completely non-traditional code generation targets. For example, the C |
| 151 | backend does not require register allocation, instruction selection, or any of |
| 152 | the other standard components provided by the system. As such, it only |
| 153 | implements these two interfaces, and does its own thing. Another example of a |
| 154 | code generator like this is a (purely hypothetical) backend that converts LLVM |
| 155 | to the GCC RTL form and uses GCC to emit machine code for a target.</p> |
| 156 | |
Reid Spencer | bdbcb8a | 2004-06-05 14:39:24 +0000 | [diff] [blame] | 157 | <p>This design also implies that it is possible to design and |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 158 | implement radically different code generators in the LLVM system that do not |
| 159 | make use of any of the built-in components. Doing so is not recommended at all, |
| 160 | but could be required for radically different targets that do not fit into the |
| 161 | LLVM machine description model: programmable FPGAs for example.</p> |
Chris Lattner | 900bf8c | 2004-06-02 07:06:06 +0000 | [diff] [blame] | 162 | |
| 163 | <p><b>Important Note:</b> For historical reasons, the LLVM SparcV9 code |
| 164 | generator uses almost entirely different code paths than described in this |
| 165 | document. For this reason, there are some deprecated interfaces (such as |
| 166 | <tt>TargetRegInfo</tt> and <tt>TargetSchedInfo</tt>), which are only used by the |
| 167 | V9 backend and should not be used by any other targets. Also, all code in the |
| 168 | <tt>lib/Target/SparcV9</tt> directory and subdirectories should be considered |
| 169 | deprecated, and should not be used as the basis for future code generator work. |
Misha Brukman | f3709d6 | 2004-06-03 16:55:57 +0000 | [diff] [blame] | 170 | The SparcV9 backend is slowly being merged into the rest of the |
| 171 | target-independent code generators, but this is a low-priority process with no |
Chris Lattner | 900bf8c | 2004-06-02 07:06:06 +0000 | [diff] [blame] | 172 | predictable completion date.</p> |
| 173 | |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 174 | </div> |
| 175 | |
| 176 | <!-- ======================================================================= --> |
| 177 | <div class="doc_subsection"> |
Chris Lattner | 10d6800 | 2004-06-01 17:18:11 +0000 | [diff] [blame] | 178 | <a name="high-level-design">The high-level design of the code generator</a> |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 179 | </div> |
| 180 | |
| 181 | <div class="doc_text"> |
| 182 | |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 183 | <p>The LLVM target-independent code generator is designed to support efficient and |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 184 | quality code generation for standard register-based microprocessors. Code |
| 185 | generation in this model is divided into the following stages:</p> |
| 186 | |
| 187 | <ol> |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 188 | <li><b><a href="#instselect">Instruction Selection</a></b> - Determining an |
| 189 | efficient implementation of the input LLVM code in the target instruction set. |
| 190 | This stage produces the initial code for the program in the target instruction |
| 191 | set, then makes use of virtual registers in SSA form and physical registers that |
| 192 | represent any required register assignments due to target constraints or calling |
| 193 | conventions.</li> |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 194 | |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 195 | <li><b><a href="#ssamco">SSA-based Machine Code Optimizations</a></b> - This |
| 196 | optional stage consists of a series of machine-code optimizations that |
| 197 | operate on the SSA-form produced by the instruction selector. Optimizations |
| 198 | like modulo-scheduling, normal scheduling, or peephole optimization work here. |
| 199 | </li> |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 200 | |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 201 | <li><b><a name="#regalloc">Register Allocation</a></b> - The |
| 202 | target code is transformed from an infinite virtual register file in SSA form |
| 203 | to the concrete register file used by the target. This phase introduces spill |
| 204 | code and eliminates all virtual register references from the program.</li> |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 205 | |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 206 | <li><b><a name="#proepicode">Prolog/Epilog Code Insertion</a></b> - Once the |
| 207 | machine code has been generated for the function and the amount of stack space |
| 208 | required is known (used for LLVM alloca's and spill slots), the prolog and |
| 209 | epilog code for the function can be inserted and "abstract stack location |
| 210 | references" can be eliminated. This stage is responsible for implementing |
| 211 | optimizations like frame-pointer elimination and stack packing.</li> |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 212 | |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 213 | <li><b><a name="latemco">Late Machine Code Optimizations</a></b> - Optimizations |
| 214 | that operate on "final" machine code can go here, such as spill code scheduling |
| 215 | and peephole optimizations.</li> |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 216 | |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 217 | <li><b><a name="codemission">Code Emission</a></b> - The final stage actually |
| 218 | puts out the code for the current function, either in the target assembler |
| 219 | format or in machine code.</li> |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 220 | |
| 221 | </ol> |
| 222 | |
| 223 | <p> |
| 224 | The code generator is based on the assumption that the instruction selector will |
| 225 | use an optimal pattern matching selector to create high-quality sequences of |
Reid Spencer | bdbcb8a | 2004-06-05 14:39:24 +0000 | [diff] [blame] | 226 | native instructions. Alternative code generator designs based on pattern |
| 227 | expansion and |
| 228 | aggressive iterative peephole optimization are much slower. This design |
| 229 | permits efficient compilation (important for JIT environments) and |
| 230 | aggressive optimization (used when generating code offline) by allowing |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 231 | components of varying levels of sophistication to be used for any step of |
Reid Spencer | bdbcb8a | 2004-06-05 14:39:24 +0000 | [diff] [blame] | 232 | compilation.</p> |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 233 | |
| 234 | <p> |
| 235 | In addition to these stages, target implementations can insert arbitrary |
| 236 | target-specific passes into the flow. For example, the X86 target uses a |
| 237 | special pass to handle the 80x87 floating point stack architecture. Other |
| 238 | targets with unusual requirements can be supported with custom passes as needed. |
| 239 | </p> |
| 240 | |
| 241 | </div> |
| 242 | |
| 243 | |
| 244 | <!-- ======================================================================= --> |
| 245 | <div class="doc_subsection"> |
Chris Lattner | 10d6800 | 2004-06-01 17:18:11 +0000 | [diff] [blame] | 246 | <a name="tablegen">Using TableGen for target description</a> |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 247 | </div> |
| 248 | |
| 249 | <div class="doc_text"> |
| 250 | |
Chris Lattner | 5489e93 | 2004-06-01 18:35:00 +0000 | [diff] [blame] | 251 | <p>The target description classes require a detailed description of the target |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 252 | architecture. These target descriptions often have a large amount of common |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 253 | information (e.g., an <tt>add</tt> instruction is almost identical to a |
| 254 | <tt>sub</tt> instruction). |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 255 | In order to allow the maximum amount of commonality to be factored out, the LLVM |
| 256 | code generator uses the <a href="TableGenFundamentals.html">TableGen</a> tool to |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 257 | describe big chunks of the target machine, which allows the use of |
| 258 | domain-specific and target-specific abstractions to reduce the amount of |
| 259 | repetition. |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 260 | </p> |
| 261 | |
| 262 | </div> |
| 263 | |
| 264 | <!-- *********************************************************************** --> |
| 265 | <div class="doc_section"> |
| 266 | <a name="targetdesc">Target description classes</a> |
| 267 | </div> |
| 268 | <!-- *********************************************************************** --> |
| 269 | |
| 270 | <div class="doc_text"> |
| 271 | |
| 272 | <p>The LLVM target description classes (which are located in the |
| 273 | <tt>include/llvm/Target</tt> directory) provide an abstract description of the |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 274 | target machine; independent of any particular client. These classes are |
| 275 | designed to capture the <i>abstract</i> properties of the target (such as the |
| 276 | instructions and registers it has), and do not incorporate any particular pieces |
| 277 | of code generation algorithms. These interfaces do not take interference graphs |
| 278 | as inputs or other algorithm-specific data structures.</p> |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 279 | |
| 280 | <p>All of the target description classes (except the <tt><a |
| 281 | href="#targetdata">TargetData</a></tt> class) are designed to be subclassed by |
| 282 | the concrete target implementation, and have virtual methods implemented. To |
Reid Spencer | bdbcb8a | 2004-06-05 14:39:24 +0000 | [diff] [blame] | 283 | get to these implementations, the <tt><a |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 284 | href="#targetmachine">TargetMachine</a></tt> class provides accessors that |
| 285 | should be implemented by the target.</p> |
| 286 | |
| 287 | </div> |
| 288 | |
| 289 | <!-- ======================================================================= --> |
| 290 | <div class="doc_subsection"> |
| 291 | <a name="targetmachine">The <tt>TargetMachine</tt> class</a> |
| 292 | </div> |
| 293 | |
| 294 | <div class="doc_text"> |
| 295 | |
| 296 | <p>The <tt>TargetMachine</tt> class provides virtual methods that are used to |
| 297 | access the target-specific implementations of the various target description |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 298 | classes via the <tt>get*Info</tt> methods (<tt>getInstrInfo</tt>, |
| 299 | <tt>getRegisterInfo</tt>, <tt>getFrameInfo</tt>, etc.). This class is |
| 300 | designed to be specialized by |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 301 | a concrete target implementation (e.g., <tt>X86TargetMachine</tt>) which |
| 302 | implements the various virtual methods. The only required target description |
| 303 | class is the <a href="#targetdata"><tt>TargetData</tt></a> class, but if the |
| 304 | code generator components are to be used, the other interfaces should be |
| 305 | implemented as well.</p> |
| 306 | |
| 307 | </div> |
| 308 | |
| 309 | |
| 310 | <!-- ======================================================================= --> |
| 311 | <div class="doc_subsection"> |
| 312 | <a name="targetdata">The <tt>TargetData</tt> class</a> |
| 313 | </div> |
| 314 | |
| 315 | <div class="doc_text"> |
| 316 | |
| 317 | <p>The <tt>TargetData</tt> class is the only required target description class, |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 318 | and it is the only class that is not extensible. You cannot derived a new |
| 319 | class from it. <tt>TargetData</tt> specifies information about how the target |
| 320 | lays out memory for structures, the alignment requirements for various data |
| 321 | types, the size of pointers in the target, and whether the target is |
| 322 | little-endian or big-endian.</p> |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 323 | |
| 324 | </div> |
| 325 | |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 326 | <!-- ======================================================================= --> |
| 327 | <div class="doc_subsection"> |
| 328 | <a name="targetlowering">The <tt>TargetLowering</tt> class</a> |
| 329 | </div> |
| 330 | |
| 331 | <div class="doc_text"> |
| 332 | |
| 333 | <p>The <tt>TargetLowering</tt> class is used by SelectionDAG based instruction |
| 334 | selectors primarily to describe how LLVM code should be lowered to SelectionDAG |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 335 | operations. Among other things, this class indicates: |
| 336 | <ul><li>an initial register class to use for various ValueTypes,</li> |
| 337 | <li>which operations are natively supported by the target machine,</li> |
| 338 | <li>the return type of setcc operations, and</li> |
| 339 | <li>the type to use for shift amounts, etc</li>. |
| 340 | </ol></p> |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 341 | |
| 342 | </div> |
| 343 | |
| 344 | |
| 345 | |
| 346 | |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 347 | |
| 348 | <!-- ======================================================================= --> |
| 349 | <div class="doc_subsection"> |
Chris Lattner | 10d6800 | 2004-06-01 17:18:11 +0000 | [diff] [blame] | 350 | <a name="mregisterinfo">The <tt>MRegisterInfo</tt> class</a> |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 351 | </div> |
| 352 | |
| 353 | <div class="doc_text"> |
| 354 | |
| 355 | <p>The <tt>MRegisterInfo</tt> class (which will eventually be renamed to |
| 356 | <tt>TargetRegisterInfo</tt>) is used to describe the register file of the |
| 357 | target and any interactions between the registers.</p> |
| 358 | |
| 359 | <p>Registers in the code generator are represented in the code generator by |
| 360 | unsigned numbers. Physical registers (those that actually exist in the target |
| 361 | description) are unique small numbers, and virtual registers are generally |
| 362 | large.</p> |
| 363 | |
| 364 | <p>Each register in the processor description has an associated |
Chris Lattner | 88a06d2 | 2005-09-30 17:46:55 +0000 | [diff] [blame] | 365 | <tt>TargetRegisterDesc</tt> entry, which provides a textual name for the register |
| 366 | (used for assembly output and debugging dumps) and a set of aliases (used to |
| 367 | indicate that one register overlaps with another). |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 368 | </p> |
| 369 | |
| 370 | <p>In addition to the per-register description, the <tt>MRegisterInfo</tt> class |
| 371 | exposes a set of processor specific register classes (instances of the |
| 372 | <tt>TargetRegisterClass</tt> class). Each register class contains sets of |
| 373 | registers that have the same properties (for example, they are all 32-bit |
| 374 | integer registers). Each SSA virtual register created by the instruction |
| 375 | selector has an associated register class. When the register allocator runs, it |
| 376 | replaces virtual registers with a physical register in the set.</p> |
| 377 | |
| 378 | <p> |
| 379 | The target-specific implementations of these classes is auto-generated from a <a |
| 380 | href="TableGenFundamentals.html">TableGen</a> description of the register file. |
| 381 | </p> |
| 382 | |
| 383 | </div> |
| 384 | |
| 385 | <!-- ======================================================================= --> |
| 386 | <div class="doc_subsection"> |
Chris Lattner | 10d6800 | 2004-06-01 17:18:11 +0000 | [diff] [blame] | 387 | <a name="targetinstrinfo">The <tt>TargetInstrInfo</tt> class</a> |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 388 | </div> |
| 389 | |
Reid Spencer | 627cd00 | 2005-07-19 01:36:35 +0000 | [diff] [blame] | 390 | <div class="doc_text"> |
| 391 | <p>The <tt>TargetInstrInfo</tt> class is used to describe the machine |
| 392 | instructions supported by the target. It is essentially an array of |
| 393 | <tt>TargetInstrDescriptor</tt> objects, each of which describes one |
| 394 | instruction the target supports. Descriptors define things like the mnemonic |
Chris Lattner | a307978 | 2005-07-19 03:37:48 +0000 | [diff] [blame] | 395 | for the opcode, the number of operands, the list of implicit register uses |
| 396 | and defs, whether the instruction has certain target-independent properties |
| 397 | (accesses memory, is commutable, etc), and holds any target-specific flags.</p> |
Reid Spencer | 627cd00 | 2005-07-19 01:36:35 +0000 | [diff] [blame] | 398 | </div> |
| 399 | |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 400 | <!-- ======================================================================= --> |
| 401 | <div class="doc_subsection"> |
Chris Lattner | 10d6800 | 2004-06-01 17:18:11 +0000 | [diff] [blame] | 402 | <a name="targetframeinfo">The <tt>TargetFrameInfo</tt> class</a> |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 403 | </div> |
| 404 | |
Reid Spencer | 627cd00 | 2005-07-19 01:36:35 +0000 | [diff] [blame] | 405 | <div class="doc_text"> |
| 406 | <p>The <tt>TargetFrameInfo</tt> class is used to provide information about the |
| 407 | stack frame layout of the target. It holds the direction of stack growth, |
| 408 | the known stack alignment on entry to each function, and the offset to the |
| 409 | locals area. The offset to the local area is the offset from the stack |
| 410 | pointer on function entry to the first location where function data (local |
| 411 | variables, spill locations) can be stored.</p> |
Reid Spencer | 627cd00 | 2005-07-19 01:36:35 +0000 | [diff] [blame] | 412 | </div> |
Chris Lattner | 47adebb | 2005-10-16 17:06:07 +0000 | [diff] [blame^] | 413 | |
| 414 | <!-- ======================================================================= --> |
| 415 | <div class="doc_subsection"> |
| 416 | <a name="targetsubtarget">The <tt>TargetSubtarget</tt> class</a> |
| 417 | </div> |
| 418 | |
| 419 | <div class="doc_text"> |
| 420 | <p> |
| 421 | TODO |
| 422 | </p> |
| 423 | </div> |
| 424 | |
| 425 | |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 426 | <!-- ======================================================================= --> |
| 427 | <div class="doc_subsection"> |
Chris Lattner | 10d6800 | 2004-06-01 17:18:11 +0000 | [diff] [blame] | 428 | <a name="targetjitinfo">The <tt>TargetJITInfo</tt> class</a> |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 429 | </div> |
| 430 | |
| 431 | <!-- *********************************************************************** --> |
| 432 | <div class="doc_section"> |
| 433 | <a name="codegendesc">Machine code description classes</a> |
| 434 | </div> |
| 435 | <!-- *********************************************************************** --> |
| 436 | |
Chris Lattner | ec94f80 | 2004-06-04 00:16:02 +0000 | [diff] [blame] | 437 | <div class="doc_text"> |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 438 | |
Chris Lattner | ec94f80 | 2004-06-04 00:16:02 +0000 | [diff] [blame] | 439 | <p> |
| 440 | At the high-level, LLVM code is translated to a machine specific representation |
| 441 | formed out of MachineFunction, MachineBasicBlock, and <a |
| 442 | href="#machineinstr"><tt>MachineInstr</tt></a> instances |
| 443 | (defined in include/llvm/CodeGen). This representation is completely target |
| 444 | agnostic, representing instructions in their most abstract form: an opcode and a |
| 445 | series of operands. This representation is designed to support both SSA |
| 446 | representation for machine code, as well as a register allocated, non-SSA form. |
| 447 | </p> |
| 448 | |
| 449 | </div> |
| 450 | |
| 451 | <!-- ======================================================================= --> |
| 452 | <div class="doc_subsection"> |
| 453 | <a name="machineinstr">The <tt>MachineInstr</tt> class</a> |
| 454 | </div> |
| 455 | |
| 456 | <div class="doc_text"> |
| 457 | |
| 458 | <p>Target machine instructions are represented as instances of the |
| 459 | <tt>MachineInstr</tt> class. This class is an extremely abstract way of |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 460 | representing machine instructions. In particular, it only keeps track of |
| 461 | an opcode number and a set of operands.</p> |
Chris Lattner | ec94f80 | 2004-06-04 00:16:02 +0000 | [diff] [blame] | 462 | |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 463 | <p>The opcode number is a simple unsigned number that only has meaning to a |
Chris Lattner | ec94f80 | 2004-06-04 00:16:02 +0000 | [diff] [blame] | 464 | specific backend. All of the instructions for a target should be defined in |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 465 | the <tt>*InstrInfo.td</tt> file for the target. The opcode enum values |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 466 | are auto-generated from this description. The <tt>MachineInstr</tt> class does |
| 467 | not have any information about how to interpret the instruction (i.e., what the |
Chris Lattner | ec94f80 | 2004-06-04 00:16:02 +0000 | [diff] [blame] | 468 | semantics of the instruction are): for that you must refer to the |
| 469 | <tt><a href="#targetinstrinfo">TargetInstrInfo</a></tt> class.</p> |
| 470 | |
| 471 | <p>The operands of a machine instruction can be of several different types: |
| 472 | they can be a register reference, constant integer, basic block reference, etc. |
| 473 | In addition, a machine operand should be marked as a def or a use of the value |
| 474 | (though only registers are allowed to be defs).</p> |
| 475 | |
| 476 | <p>By convention, the LLVM code generator orders instruction operands so that |
| 477 | all register definitions come before the register uses, even on architectures |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 478 | that are normally printed in other orders. For example, the SPARC add |
Chris Lattner | ec94f80 | 2004-06-04 00:16:02 +0000 | [diff] [blame] | 479 | instruction: "<tt>add %i1, %i2, %i3</tt>" adds the "%i1", and "%i2" registers |
| 480 | and stores the result into the "%i3" register. In the LLVM code generator, |
| 481 | the operands should be stored as "<tt>%i3, %i1, %i2</tt>": with the destination |
| 482 | first.</p> |
| 483 | |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 484 | <p>Keeping destination (definition) operands at the beginning of the operand |
| 485 | list has several advantages. In particular, the debugging printer will print |
| 486 | the instruction like this:</p> |
Chris Lattner | ec94f80 | 2004-06-04 00:16:02 +0000 | [diff] [blame] | 487 | |
| 488 | <pre> |
| 489 | %r3 = add %i1, %i2 |
| 490 | </pre> |
| 491 | |
| 492 | <p>If the first operand is a def, and it is also easier to <a |
| 493 | href="#buildmi">create instructions</a> whose only def is the first |
| 494 | operand.</p> |
| 495 | |
| 496 | </div> |
| 497 | |
| 498 | <!-- _______________________________________________________________________ --> |
| 499 | <div class="doc_subsubsection"> |
| 500 | <a name="buildmi">Using the <tt>MachineInstrBuilder.h</tt> functions</a> |
| 501 | </div> |
| 502 | |
| 503 | <div class="doc_text"> |
| 504 | |
| 505 | <p>Machine instructions are created by using the <tt>BuildMI</tt> functions, |
| 506 | located in the <tt>include/llvm/CodeGen/MachineInstrBuilder.h</tt> file. The |
| 507 | <tt>BuildMI</tt> functions make it easy to build arbitrary machine |
| 508 | instructions. Usage of the <tt>BuildMI</tt> functions look like this: |
| 509 | </p> |
| 510 | |
| 511 | <pre> |
| 512 | // Create a 'DestReg = mov 42' (rendered in X86 assembly as 'mov DestReg, 42') |
| 513 | // instruction. The '1' specifies how many operands will be added. |
| 514 | MachineInstr *MI = BuildMI(X86::MOV32ri, 1, DestReg).addImm(42); |
| 515 | |
| 516 | // Create the same instr, but insert it at the end of a basic block. |
| 517 | MachineBasicBlock &MBB = ... |
| 518 | BuildMI(MBB, X86::MOV32ri, 1, DestReg).addImm(42); |
| 519 | |
| 520 | // Create the same instr, but insert it before a specified iterator point. |
| 521 | MachineBasicBlock::iterator MBBI = ... |
| 522 | BuildMI(MBB, MBBI, X86::MOV32ri, 1, DestReg).addImm(42); |
| 523 | |
| 524 | // Create a 'cmp Reg, 0' instruction, no destination reg. |
| 525 | MI = BuildMI(X86::CMP32ri, 2).addReg(Reg).addImm(0); |
| 526 | // Create an 'sahf' instruction which takes no operands and stores nothing. |
| 527 | MI = BuildMI(X86::SAHF, 0); |
| 528 | |
| 529 | // Create a self looping branch instruction. |
| 530 | BuildMI(MBB, X86::JNE, 1).addMBB(&MBB); |
| 531 | </pre> |
| 532 | |
| 533 | <p> |
| 534 | The key thing to remember with the <tt>BuildMI</tt> functions is that you have |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 535 | to specify the number of operands that the machine instruction will take. This |
| 536 | allows for efficient memory allocation. You also need to specify if operands |
| 537 | default to be uses of values, not definitions. If you need to add a definition |
| 538 | operand (other than the optional destination register), you must explicitly |
| 539 | mark it as such. |
Chris Lattner | ec94f80 | 2004-06-04 00:16:02 +0000 | [diff] [blame] | 540 | </p> |
| 541 | |
| 542 | </div> |
| 543 | |
| 544 | <!-- _______________________________________________________________________ --> |
| 545 | <div class="doc_subsubsection"> |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 546 | <a name="fixedregs">Fixed (preassigned) registers</a> |
Chris Lattner | ec94f80 | 2004-06-04 00:16:02 +0000 | [diff] [blame] | 547 | </div> |
| 548 | |
| 549 | <div class="doc_text"> |
| 550 | |
| 551 | <p>One important issue that the code generator needs to be aware of is the |
| 552 | presence of fixed registers. In particular, there are often places in the |
| 553 | instruction stream where the register allocator <em>must</em> arrange for a |
| 554 | particular value to be in a particular register. This can occur due to |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 555 | limitations of the instruction set (e.g., the X86 can only do a 32-bit divide |
Chris Lattner | ec94f80 | 2004-06-04 00:16:02 +0000 | [diff] [blame] | 556 | with the <tt>EAX</tt>/<tt>EDX</tt> registers), or external factors like calling |
| 557 | conventions. In any case, the instruction selector should emit code that |
| 558 | copies a virtual register into or out of a physical register when needed.</p> |
| 559 | |
| 560 | <p>For example, consider this simple LLVM example:</p> |
| 561 | |
| 562 | <pre> |
| 563 | int %test(int %X, int %Y) { |
| 564 | %Z = div int %X, %Y |
| 565 | ret int %Z |
| 566 | } |
| 567 | </pre> |
| 568 | |
| 569 | <p>The X86 instruction selector produces this machine code for the div |
| 570 | and ret (use |
| 571 | "<tt>llc X.bc -march=x86 -print-machineinstrs</tt>" to get this):</p> |
| 572 | |
| 573 | <pre> |
| 574 | ;; Start of div |
| 575 | %EAX = mov %reg1024 ;; Copy X (in reg1024) into EAX |
| 576 | %reg1027 = sar %reg1024, 31 |
| 577 | %EDX = mov %reg1027 ;; Sign extend X into EDX |
| 578 | idiv %reg1025 ;; Divide by Y (in reg1025) |
| 579 | %reg1026 = mov %EAX ;; Read the result (Z) out of EAX |
| 580 | |
| 581 | ;; Start of ret |
| 582 | %EAX = mov %reg1026 ;; 32-bit return value goes in EAX |
| 583 | ret |
| 584 | </pre> |
| 585 | |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 586 | <p>By the end of code generation, the register allocator has coalesced |
Chris Lattner | ec94f80 | 2004-06-04 00:16:02 +0000 | [diff] [blame] | 587 | the registers and deleted the resultant identity moves, producing the |
| 588 | following code:</p> |
| 589 | |
| 590 | <pre> |
| 591 | ;; X is in EAX, Y is in ECX |
| 592 | mov %EAX, %EDX |
| 593 | sar %EDX, 31 |
| 594 | idiv %ECX |
| 595 | ret |
| 596 | </pre> |
| 597 | |
| 598 | <p>This approach is extremely general (if it can handle the X86 architecture, |
| 599 | it can handle anything!) and allows all of the target specific |
| 600 | knowledge about the instruction stream to be isolated in the instruction |
| 601 | selector. Note that physical registers should have a short lifetime for good |
| 602 | code generation, and all physical registers are assumed dead on entry and |
| 603 | exit of basic blocks (before register allocation). Thus if you need a value |
| 604 | to be live across basic block boundaries, it <em>must</em> live in a virtual |
| 605 | register.</p> |
| 606 | |
| 607 | </div> |
| 608 | |
| 609 | <!-- _______________________________________________________________________ --> |
| 610 | <div class="doc_subsubsection"> |
| 611 | <a name="ssa">Machine code SSA form</a> |
| 612 | </div> |
| 613 | |
| 614 | <div class="doc_text"> |
| 615 | |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 616 | <p><tt>MachineInstr</tt>'s are initially selected in SSA-form, and |
Chris Lattner | ec94f80 | 2004-06-04 00:16:02 +0000 | [diff] [blame] | 617 | are maintained in SSA-form until register allocation happens. For the most |
| 618 | part, this is trivially simple since LLVM is already in SSA form: LLVM PHI nodes |
| 619 | become machine code PHI nodes, and virtual registers are only allowed to have a |
| 620 | single definition.</p> |
| 621 | |
| 622 | <p>After register allocation, machine code is no longer in SSA-form, as there |
| 623 | are no virtual registers left in the code.</p> |
| 624 | |
| 625 | </div> |
| 626 | |
| 627 | <!-- *********************************************************************** --> |
| 628 | <div class="doc_section"> |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 629 | <a name="codegenalgs">Target-independent code generation algorithms</a> |
| 630 | </div> |
| 631 | <!-- *********************************************************************** --> |
| 632 | |
| 633 | <div class="doc_text"> |
| 634 | |
| 635 | <p>This section documents the phases described in the <a |
| 636 | href="high-level-design">high-level design of the code generator</a>. It |
| 637 | explains how they work and some of the rationale behind their design.</p> |
| 638 | |
| 639 | </div> |
| 640 | |
| 641 | <!-- ======================================================================= --> |
| 642 | <div class="doc_subsection"> |
| 643 | <a name="instselect">Instruction Selection</a> |
| 644 | </div> |
| 645 | |
| 646 | <div class="doc_text"> |
| 647 | <p> |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 648 | Instruction Selection is the process of translating LLVM code presented to the |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 649 | code generator into target-specific machine instructions. There are several |
| 650 | well-known ways to do this in the literature. In LLVM there are two main forms: |
Chris Lattner | e35d3bb | 2005-10-16 00:36:38 +0000 | [diff] [blame] | 651 | the SelectionDAG based instruction selector framework and an old-style 'simple' |
| 652 | instruction selector (which effectively peephole selects each LLVM instruction |
| 653 | into a series of machine instructions). We recommend that all targets use the |
| 654 | SelectionDAG infrastructure. |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 655 | </p> |
| 656 | |
Chris Lattner | e35d3bb | 2005-10-16 00:36:38 +0000 | [diff] [blame] | 657 | <p>Portions of the DAG instruction selector are generated from the target |
| 658 | description files (<tt>*.td</tt>) files. Eventually, we aim for the entire |
| 659 | instruction selector to be generated from these <tt>.td</tt> files.</p> |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 660 | </div> |
| 661 | |
| 662 | <!-- _______________________________________________________________________ --> |
| 663 | <div class="doc_subsubsection"> |
| 664 | <a name="selectiondag_intro">Introduction to SelectionDAGs</a> |
| 665 | </div> |
| 666 | |
| 667 | <div class="doc_text"> |
| 668 | |
| 669 | <p> |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 670 | The SelectionDAG provides an abstraction for code representation in a way that |
| 671 | is amenable to instruction selection using automatic techniques |
| 672 | (e.g. dynamic-programming based optimal pattern matching selectors), It is also |
| 673 | well suited to other phases of code generation; in particular, instruction scheduling. Additionally, the SelectionDAG provides a host representation where a |
| 674 | large variety of very-low-level (but target-independent) |
| 675 | <a href="#selectiondag_optimize">optimizations</a> may be |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 676 | performed: ones which require extensive information about the instructions |
| 677 | efficiently supported by the target. |
| 678 | </p> |
| 679 | |
| 680 | <p> |
| 681 | The SelectionDAG is a Directed-Acyclic-Graph whose nodes are instances of the |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 682 | <tt>SDNode</tt> class. The primary payload of the <tt>SDNode</tt> is its |
| 683 | operation code (Opcode) that indicates what operation the node performs. |
| 684 | The various operation node types are described at the top of the |
| 685 | <tt>include/llvm/CodeGen/SelectionDAGNodes.h</tt> file. Depending on the |
| 686 | operation, nodes may contain additional information (e.g. the condition code |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 687 | for a SETCC node) contained in a derived class.</p> |
| 688 | |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 689 | <p>Although most operations define a single value, each node in the graph may |
| 690 | define multiple values. For example, a combined div/rem operation will define |
| 691 | both the dividend and the remainder. Many other situations require multiple |
| 692 | values as well. Each node also has some number of operands, which are edges |
| 693 | to the node defining the used value. Because nodes may define multiple values, |
| 694 | edges are represented by instances of the <tt>SDOperand</tt> class, which is |
| 695 | a <SDNode, unsigned> pair, indicating the node and result |
| 696 | value being used, respectively. Each value produced by an SDNode has an |
| 697 | associated MVT::ValueType, indicating what type the value is. |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 698 | </p> |
| 699 | |
| 700 | <p> |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 701 | SelectionDAGs contain two different kinds of values: those that represent data |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 702 | flow and those that represent control flow dependencies. Data values are simple |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 703 | edges with an integer or floating point value type. Control edges are |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 704 | represented as "chain" edges which are of type MVT::Other. These edges provide |
| 705 | an ordering between nodes that have side effects (such as |
| 706 | loads/stores/calls/return/etc). All nodes that have side effects should take a |
| 707 | token chain as input and produce a new one as output. By convention, token |
| 708 | chain inputs are always operand #0, and chain results are always the last |
| 709 | value produced by an operation.</p> |
| 710 | |
| 711 | <p> |
| 712 | A SelectionDAG has designated "Entry" and "Root" nodes. The Entry node is |
Chris Lattner | e0c1317 | 2005-05-09 15:41:03 +0000 | [diff] [blame] | 713 | always a marker node with an Opcode of ISD::EntryToken. The Root node is the |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 714 | final side-effecting node in the token chain. For example, in a single basic |
| 715 | block function, this would be the return node. |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 716 | </p> |
| 717 | |
| 718 | <p> |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 719 | One important concept for SelectionDAGs is the notion of a "legal" vs. "illegal" |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 720 | DAG. A legal DAG for a target is one that only uses supported operations and |
| 721 | supported types. On PowerPC, for example, a DAG with any values of i1, i8, i16, |
| 722 | or i64 type would be illegal. The <a href="#selectiondag_legalize">legalize</a> |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 723 | phase is responsible for turning an illegal DAG into a legal DAG. |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 724 | </p> |
| 725 | </div> |
| 726 | |
| 727 | <!-- _______________________________________________________________________ --> |
| 728 | <div class="doc_subsubsection"> |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 729 | <a name="selectiondag_process">SelectionDAG Instruction Selection Process</a> |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 730 | </div> |
| 731 | |
| 732 | <div class="doc_text"> |
| 733 | |
| 734 | <p> |
| 735 | SelectionDAG-based instruction selection consists of the following steps: |
| 736 | </p> |
| 737 | |
| 738 | <ol> |
| 739 | <li><a href="#selectiondag_build">Build initial DAG</a> - This stage performs |
| 740 | a simple translation from the input LLVM code to an illegal SelectionDAG. |
| 741 | </li> |
| 742 | <li><a href="#selectiondag_optimize">Optimize SelectionDAG</a> - This stage |
| 743 | performs simple optimizations on the SelectionDAG to simplify it and |
| 744 | recognize meta instructions (like rotates and div/rem pairs) for |
| 745 | targets that support these meta operations. This makes the resultant code |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 746 | more efficient and the 'select instructions from DAG' phase (below) simpler. |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 747 | </li> |
| 748 | <li><a href="#selectiondag_legalize">Legalize SelectionDAG</a> - This stage |
| 749 | converts the illegal SelectionDAG to a legal SelectionDAG, by eliminating |
| 750 | unsupported operations and data types.</li> |
| 751 | <li><a href="#selectiondag_optimize">Optimize SelectionDAG (#2)</a> - This |
| 752 | second run of the SelectionDAG optimized the newly legalized DAG, to |
| 753 | eliminate inefficiencies introduced by legalization.</li> |
| 754 | <li><a href="#selectiondag_select">Select instructions from DAG</a> - Finally, |
| 755 | the target instruction selector matches the DAG operations to target |
Chris Lattner | e35d3bb | 2005-10-16 00:36:38 +0000 | [diff] [blame] | 756 | instructions. This process translates the target-independent input DAG into |
| 757 | another DAG of target instructions.</li> |
| 758 | <li><a href="#selectiondag_sched">SelectionDAG Scheduling and Emission</a> |
| 759 | - The last phase assigns a linear order to the instructions in the |
| 760 | target-instruction DAG and emits them into the MachineFunction being |
| 761 | compiled. This step uses traditional prepass scheduling techniques.</li> |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 762 | </ol> |
| 763 | |
| 764 | <p>After all of these steps are complete, the SelectionDAG is destroyed and the |
| 765 | rest of the code generation passes are run.</p> |
| 766 | |
| 767 | </div> |
| 768 | |
| 769 | <!-- _______________________________________________________________________ --> |
| 770 | <div class="doc_subsubsection"> |
| 771 | <a name="selectiondag_build">Initial SelectionDAG Construction</a> |
| 772 | </div> |
| 773 | |
| 774 | <div class="doc_text"> |
| 775 | |
| 776 | <p> |
| 777 | The initial SelectionDAG is naively peephole expanded from the LLVM input by |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 778 | the <tt>SelectionDAGLowering</tt> class in the SelectionDAGISel.cpp file. The |
| 779 | intent of this pass is to expose as much low-level, target-specific details |
| 780 | to the SelectionDAG as possible. This pass is mostly hard-coded (e.g. an LLVM |
| 781 | add turns into an SDNode add while a geteelementptr is expanded into the obvious |
| 782 | arithmetic). This pass requires target-specific hooks to lower calls and |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 783 | returns, varargs, etc. For these features, the TargetLowering interface is |
| 784 | used. |
| 785 | </p> |
| 786 | |
| 787 | </div> |
| 788 | |
| 789 | <!-- _______________________________________________________________________ --> |
| 790 | <div class="doc_subsubsection"> |
| 791 | <a name="selectiondag_legalize">SelectionDAG Legalize Phase</a> |
| 792 | </div> |
| 793 | |
| 794 | <div class="doc_text"> |
| 795 | |
| 796 | <p>The Legalize phase is in charge of converting a DAG to only use the types and |
| 797 | operations that are natively supported by the target. This involves two major |
| 798 | tasks:</p> |
| 799 | |
| 800 | <ol> |
| 801 | <li><p>Convert values of unsupported types to values of supported types.</p> |
| 802 | <p>There are two main ways of doing this: promoting a small type to a larger |
Chris Lattner | fd84c2d | 2005-04-25 00:38:52 +0000 | [diff] [blame] | 803 | type (e.g. f32 -> f64, or i16 -> i32), and breaking up large |
| 804 | integer types |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 805 | to smaller ones (e.g. implementing i64 with i32 operations where |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 806 | possible). Type conversions can insert sign and zero extensions as |
| 807 | needed to make sure that the final code has the same behavior as the |
| 808 | input.</p> |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 809 | </li> |
| 810 | |
| 811 | <li><p>Eliminate operations that are not supported by the target in a supported |
| 812 | type.</p> |
| 813 | <p>Targets often have wierd constraints, such as not supporting every |
| 814 | operation on every supported datatype (e.g. X86 does not support byte |
| 815 | conditional moves). Legalize takes care of either open-coding another |
| 816 | sequence of operations to emulate the operation (this is known as |
| 817 | expansion), promoting to a larger type that supports the operation |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 818 | (promotion), or using a target-specific hook to implement the |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 819 | legalization.</p> |
| 820 | </li> |
| 821 | </ol> |
| 822 | |
| 823 | <p> |
| 824 | Instead of using a Legalize pass, we could require that every target-specific |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 825 | <a href="#selectiondag_optimize">selector</a> supports and expands every |
| 826 | operator and type even if they are not supported and may require many |
| 827 | instructions to implement (in fact, this is the approach taken by the |
| 828 | "simple" selectors). However, using a Legalize pass allows all of the |
| 829 | cannonicalization patterns to be shared across targets which makes it very |
| 830 | easy to optimize the cannonicalized code because it is still in the form of |
| 831 | a DAG. |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 832 | </p> |
| 833 | |
| 834 | </div> |
| 835 | |
| 836 | <!-- _______________________________________________________________________ --> |
| 837 | <div class="doc_subsubsection"> |
Chris Lattner | e35d3bb | 2005-10-16 00:36:38 +0000 | [diff] [blame] | 838 | <a name="selectiondag_optimize">SelectionDAG Optimization Phase: the DAG |
| 839 | Combiner</a> |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 840 | </div> |
| 841 | |
| 842 | <div class="doc_text"> |
| 843 | |
| 844 | <p> |
| 845 | The SelectionDAG optimization phase is run twice for code generation: once |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 846 | immediately after the DAG is built and once after legalization. The first run |
| 847 | of the pass allows the initial code to be cleaned up (e.g. performing |
| 848 | optimizations that depend on knowing that the operators have restricted type |
| 849 | inputs). The second run of the pass cleans up the messy code generated by the |
| 850 | Legalize pass, allowing Legalize to be very simple since it can ignore many |
| 851 | special cases. |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 852 | </p> |
| 853 | |
| 854 | <p> |
Chris Lattner | e35d3bb | 2005-10-16 00:36:38 +0000 | [diff] [blame] | 855 | One important class of optimizations performed is optimizing inserted sign and |
| 856 | zero extension instructions. We currently use ad-hoc techniques, but could move |
| 857 | to more rigorous techniques in the future. Here are some good |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 858 | papers on the subject:</p> |
| 859 | |
| 860 | <p> |
| 861 | "<a href="http://www.eecs.harvard.edu/~nr/pubs/widen-abstract.html">Widening |
| 862 | integer arithmetic</a>"<br> |
| 863 | Kevin Redwine and Norman Ramsey<br> |
| 864 | International Conference on Compiler Construction (CC) 2004 |
| 865 | </p> |
| 866 | |
| 867 | |
| 868 | <p> |
| 869 | "<a href="http://portal.acm.org/citation.cfm?doid=512529.512552">Effective |
| 870 | sign extension elimination</a>"<br> |
| 871 | Motohiro Kawahito, Hideaki Komatsu, and Toshio Nakatani<br> |
| 872 | Proceedings of the ACM SIGPLAN 2002 Conference on Programming Language Design |
| 873 | and Implementation. |
| 874 | </p> |
| 875 | |
| 876 | </div> |
| 877 | |
| 878 | <!-- _______________________________________________________________________ --> |
| 879 | <div class="doc_subsubsection"> |
| 880 | <a name="selectiondag_select">SelectionDAG Select Phase</a> |
| 881 | </div> |
| 882 | |
| 883 | <div class="doc_text"> |
| 884 | |
| 885 | <p>The Select phase is the bulk of the target-specific code for instruction |
| 886 | selection. This phase takes a legal SelectionDAG as input, and does simple |
| 887 | pattern matching on the DAG to generate code. In time, the Select phase will |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 888 | be automatically generated from the target's InstrInfo.td file, which is why we |
| 889 | want to make the Select phase as simple and mechanical as possible.</p> |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 890 | |
| 891 | </div> |
| 892 | |
| 893 | <!-- _______________________________________________________________________ --> |
| 894 | <div class="doc_subsubsection"> |
Chris Lattner | e35d3bb | 2005-10-16 00:36:38 +0000 | [diff] [blame] | 895 | <a name="selectiondag_sched">SelectionDAG Scheduling and Emission Phase</a> |
| 896 | </div> |
| 897 | |
| 898 | <div class="doc_text"> |
| 899 | |
| 900 | <p>The scheduling phase takes the DAG of target instructions from the selection |
| 901 | phase and assigns an order. The scheduler can pick an order depending on |
| 902 | various constraints of the machines (i.e. order for minimal register pressure or |
| 903 | try to cover instruction latencies). Once an order is established, the DAG is |
| 904 | converted to a list of <a href="#machineinstr">MachineInstr</a>s and the |
| 905 | Selection DAG is destroyed. |
| 906 | </p> |
| 907 | |
| 908 | </div> |
| 909 | |
| 910 | <!-- _______________________________________________________________________ --> |
| 911 | <div class="doc_subsubsection"> |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 912 | <a name="selectiondag_future">Future directions for the SelectionDAG</a> |
| 913 | </div> |
| 914 | |
| 915 | <div class="doc_text"> |
| 916 | |
| 917 | <ol> |
Chris Lattner | e35d3bb | 2005-10-16 00:36:38 +0000 | [diff] [blame] | 918 | <li>Optional function-at-a-time selection.</li> |
| 919 | <li>Auto-generate entire selector from .td file.</li> |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 920 | </li> |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 921 | </ol> |
| 922 | |
| 923 | </div> |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 924 | |
| 925 | <!-- ======================================================================= --> |
| 926 | <div class="doc_subsection"> |
| 927 | <a name="ssamco">SSA-based Machine Code Optimizations</a> |
| 928 | </div> |
| 929 | <div class="doc_text"><p>To Be Written</p></div> |
| 930 | <!-- ======================================================================= --> |
| 931 | <div class="doc_subsection"> |
| 932 | <a name="regalloc">Register Allocation</a> |
| 933 | </div> |
| 934 | <div class="doc_text"><p>To Be Written</p></div> |
| 935 | <!-- ======================================================================= --> |
| 936 | <div class="doc_subsection"> |
| 937 | <a name="proepicode">Prolog/Epilog Code Insertion</a> |
| 938 | </div> |
| 939 | <div class="doc_text"><p>To Be Written</p></div> |
| 940 | <!-- ======================================================================= --> |
| 941 | <div class="doc_subsection"> |
| 942 | <a name="latemco">Late Machine Code Optimizations</a> |
| 943 | </div> |
| 944 | <div class="doc_text"><p>To Be Written</p></div> |
| 945 | <!-- ======================================================================= --> |
| 946 | <div class="doc_subsection"> |
| 947 | <a name="codemission">Code Emission</a> |
| 948 | </div> |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 949 | |
| 950 | <!-- *********************************************************************** --> |
| 951 | <div class="doc_section"> |
Chris Lattner | ec94f80 | 2004-06-04 00:16:02 +0000 | [diff] [blame] | 952 | <a name="targetimpls">Target description implementations</a> |
| 953 | </div> |
| 954 | <!-- *********************************************************************** --> |
| 955 | |
| 956 | <div class="doc_text"> |
| 957 | |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 958 | <p>This section of the document explains features or design decisions that |
Chris Lattner | ec94f80 | 2004-06-04 00:16:02 +0000 | [diff] [blame] | 959 | are specific to the code generator for a particular target.</p> |
| 960 | |
| 961 | </div> |
| 962 | |
| 963 | |
| 964 | <!-- ======================================================================= --> |
| 965 | <div class="doc_subsection"> |
| 966 | <a name="x86">The X86 backend</a> |
| 967 | </div> |
| 968 | |
| 969 | <div class="doc_text"> |
| 970 | |
| 971 | <p> |
| 972 | The X86 code generator lives in the <tt>lib/Target/X86</tt> directory. This |
| 973 | code generator currently targets a generic P6-like processor. As such, it |
| 974 | produces a few P6-and-above instructions (like conditional moves), but it does |
| 975 | not make use of newer features like MMX or SSE. In the future, the X86 backend |
Chris Lattner | aa5bcb5 | 2005-01-28 17:22:53 +0000 | [diff] [blame] | 976 | will have sub-target support added for specific processor families and |
Chris Lattner | ec94f80 | 2004-06-04 00:16:02 +0000 | [diff] [blame] | 977 | implementations.</p> |
| 978 | |
| 979 | </div> |
| 980 | |
| 981 | <!-- _______________________________________________________________________ --> |
| 982 | <div class="doc_subsubsection"> |
Chris Lattner | 9b988be | 2005-07-12 00:20:49 +0000 | [diff] [blame] | 983 | <a name="x86_tt">X86 Target Triples Supported</a> |
| 984 | </div> |
| 985 | |
| 986 | <div class="doc_text"> |
| 987 | <p> |
| 988 | The following are the known target triples that are supported by the X86 |
| 989 | backend. This is not an exhaustive list, but it would be useful to add those |
| 990 | that people test. |
| 991 | </p> |
| 992 | |
| 993 | <ul> |
| 994 | <li><b>i686-pc-linux-gnu</b> - Linux</li> |
| 995 | <li><b>i386-unknown-freebsd5.3</b> - FreeBSD 5.3</li> |
| 996 | <li><b>i686-pc-cygwin</b> - Cygwin on Win32</li> |
| 997 | <li><b>i686-pc-mingw32</b> - MingW on Win32</li> |
| 998 | <li><b>i686-apple-darwin*</b> - Apple Darwin</li> |
| 999 | </ul> |
| 1000 | |
| 1001 | </div> |
| 1002 | |
| 1003 | <!-- _______________________________________________________________________ --> |
| 1004 | <div class="doc_subsubsection"> |
Chris Lattner | ec94f80 | 2004-06-04 00:16:02 +0000 | [diff] [blame] | 1005 | <a name="x86_memory">Representing X86 addressing modes in MachineInstrs</a> |
| 1006 | </div> |
| 1007 | |
| 1008 | <div class="doc_text"> |
| 1009 | |
Misha Brukman | 600df45 | 2005-02-17 22:22:24 +0000 | [diff] [blame] | 1010 | <p>The x86 has a very flexible way of accessing memory. It is capable of |
Chris Lattner | ec94f80 | 2004-06-04 00:16:02 +0000 | [diff] [blame] | 1011 | forming memory addresses of the following expression directly in integer |
| 1012 | instructions (which use ModR/M addressing):</p> |
| 1013 | |
| 1014 | <pre> |
| 1015 | Base+[1,2,4,8]*IndexReg+Disp32 |
| 1016 | </pre> |
| 1017 | |
Misha Brukman | 600df45 | 2005-02-17 22:22:24 +0000 | [diff] [blame] | 1018 | <p>In order to represent this, LLVM tracks no less than 4 operands for each |
| 1019 | memory operand of this form. This means that the "load" form of 'mov' has the |
| 1020 | following <tt>MachineOperand</tt>s in this order:</p> |
Chris Lattner | ec94f80 | 2004-06-04 00:16:02 +0000 | [diff] [blame] | 1021 | |
| 1022 | <pre> |
| 1023 | Index: 0 | 1 2 3 4 |
| 1024 | Meaning: DestReg, | BaseReg, Scale, IndexReg, Displacement |
| 1025 | OperandTy: VirtReg, | VirtReg, UnsImm, VirtReg, SignExtImm |
| 1026 | </pre> |
| 1027 | |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 1028 | <p>Stores, and all other instructions, treat the four memory operands in the |
| 1029 | same way, in the same order.</p> |
Chris Lattner | ec94f80 | 2004-06-04 00:16:02 +0000 | [diff] [blame] | 1030 | |
| 1031 | </div> |
| 1032 | |
| 1033 | <!-- _______________________________________________________________________ --> |
| 1034 | <div class="doc_subsubsection"> |
| 1035 | <a name="x86_names">Instruction naming</a> |
| 1036 | </div> |
| 1037 | |
| 1038 | <div class="doc_text"> |
| 1039 | |
| 1040 | <p> |
Reid Spencer | ad1f0cd | 2005-04-24 20:56:18 +0000 | [diff] [blame] | 1041 | An instruction name consists of the base name, a default operand size, and a |
| 1042 | a character per operand with an optional special size. For example:</p> |
Chris Lattner | ec94f80 | 2004-06-04 00:16:02 +0000 | [diff] [blame] | 1043 | |
| 1044 | <p> |
| 1045 | <tt>ADD8rr</tt> -> add, 8-bit register, 8-bit register<br> |
| 1046 | <tt>IMUL16rmi</tt> -> imul, 16-bit register, 16-bit memory, 16-bit immediate<br> |
| 1047 | <tt>IMUL16rmi8</tt> -> imul, 16-bit register, 16-bit memory, 8-bit immediate<br> |
| 1048 | <tt>MOVSX32rm16</tt> -> movsx, 32-bit register, 16-bit memory |
| 1049 | </p> |
| 1050 | |
| 1051 | </div> |
Chris Lattner | ce52b7e | 2004-06-01 06:48:00 +0000 | [diff] [blame] | 1052 | |
| 1053 | <!-- *********************************************************************** --> |
| 1054 | <hr> |
| 1055 | <address> |
| 1056 | <a href="http://jigsaw.w3.org/css-validator/check/referer"><img |
| 1057 | src="http://jigsaw.w3.org/css-validator/images/vcss" alt="Valid CSS!"></a> |
| 1058 | <a href="http://validator.w3.org/check/referer"><img |
| 1059 | src="http://www.w3.org/Icons/valid-html401" alt="Valid HTML 4.01!" /></a> |
| 1060 | |
| 1061 | <a href="mailto:sabre@nondot.org">Chris Lattner</a><br> |
| 1062 | <a href="http://llvm.cs.uiuc.edu">The LLVM Compiler Infrastructure</a><br> |
| 1063 | Last modified: $Date$ |
| 1064 | </address> |
| 1065 | |
| 1066 | </body> |
| 1067 | </html> |