1 ======================== 2 LLVM Programmer's Manual 3 ======================== 4 5 .. contents:: 6 :local: 7 8 .. warning:: 9 This is always a work in progress. 10 11 .. _introduction: 12 13 Introduction 14 ============ 15 16 This document is meant to highlight some of the important classes and interfaces 17 available in the LLVM source-base. This manual is not intended to explain what 18 LLVM is, how it works, and what LLVM code looks like. It assumes that you know 19 the basics of LLVM and are interested in writing transformations or otherwise 20 analyzing or manipulating the code. 21 22 This document should get you oriented so that you can find your way in the 23 continuously growing source code that makes up the LLVM infrastructure. Note 24 that this manual is not intended to serve as a replacement for reading the 25 source code, so if you think there should be a method in one of these classes to 26 do something, but it's not listed, check the source. Links to the `doxygen 27 <http://llvm.org/doxygen/>`__ sources are provided to make this as easy as 28 possible. 29 30 The first section of this document describes general information that is useful 31 to know when working in the LLVM infrastructure, and the second describes the 32 Core LLVM classes. In the future this manual will be extended with information 33 describing how to use extension libraries, such as dominator information, CFG 34 traversal routines, and useful utilities like the ``InstVisitor`` (`doxygen 35 <http://llvm.org/doxygen/InstVisitor_8h-source.html>`__) template. 36 37 .. _general: 38 39 General Information 40 =================== 41 42 This section contains general information that is useful if you are working in 43 the LLVM source-base, but that isn't specific to any particular API. 44 45 .. _stl: 46 47 The C++ Standard Template Library 48 --------------------------------- 49 50 LLVM makes heavy use of the C++ Standard Template Library (STL), perhaps much 51 more than you are used to, or have seen before. Because of this, you might want 52 to do a little background reading in the techniques used and capabilities of the 53 library. There are many good pages that discuss the STL, and several books on 54 the subject that you can get, so it will not be discussed in this document. 55 56 Here are some useful links: 57 58 #. `cppreference.com 59 <http://en.cppreference.com/w/>`_ - an excellent 60 reference for the STL and other parts of the standard C++ library. 61 62 #. `C++ In a Nutshell <http://www.tempest-sw.com/cpp/>`_ - This is an O'Reilly 63 book in the making. It has a decent Standard Library Reference that rivals 64 Dinkumware's, and is unfortunately no longer free since the book has been 65 published. 66 67 #. `C++ Frequently Asked Questions <http://www.parashift.com/c++-faq-lite/>`_. 68 69 #. `SGI's STL Programmer's Guide <http://www.sgi.com/tech/stl/>`_ - Contains a 70 useful `Introduction to the STL 71 <http://www.sgi.com/tech/stl/stl_introduction.html>`_. 72 73 #. `Bjarne Stroustrup's C++ Page 74 <http://www.research.att.com/%7Ebs/C++.html>`_. 75 76 #. `Bruce Eckel's Thinking in C++, 2nd ed. Volume 2 Revision 4.0 77 (even better, get the book) 78 <http://www.mindview.net/Books/TICPP/ThinkingInCPP2e.html>`_. 79 80 You are also encouraged to take a look at the :doc:`LLVM Coding Standards 81 <CodingStandards>` guide which focuses on how to write maintainable code more 82 than where to put your curly braces. 83 84 .. _resources: 85 86 Other useful references 87 ----------------------- 88 89 #. `Using static and shared libraries across platforms 90 <http://www.fortran-2000.com/ArnaudRecipes/sharedlib.html>`_ 91 92 .. _apis: 93 94 Important and useful LLVM APIs 95 ============================== 96 97 Here we highlight some LLVM APIs that are generally useful and good to know 98 about when writing transformations. 99 100 .. _isa: 101 102 The ``isa<>``, ``cast<>`` and ``dyn_cast<>`` templates 103 ------------------------------------------------------ 104 105 The LLVM source-base makes extensive use of a custom form of RTTI. These 106 templates have many similarities to the C++ ``dynamic_cast<>`` operator, but 107 they don't have some drawbacks (primarily stemming from the fact that 108 ``dynamic_cast<>`` only works on classes that have a v-table). Because they are 109 used so often, you must know what they do and how they work. All of these 110 templates are defined in the ``llvm/Support/Casting.h`` (`doxygen 111 <http://llvm.org/doxygen/Casting_8h-source.html>`__) file (note that you very 112 rarely have to include this file directly). 113 114 ``isa<>``: 115 The ``isa<>`` operator works exactly like the Java "``instanceof``" operator. 116 It returns true or false depending on whether a reference or pointer points to 117 an instance of the specified class. This can be very useful for constraint 118 checking of various sorts (example below). 119 120 ``cast<>``: 121 The ``cast<>`` operator is a "checked cast" operation. It converts a pointer 122 or reference from a base class to a derived class, causing an assertion 123 failure if it is not really an instance of the right type. This should be 124 used in cases where you have some information that makes you believe that 125 something is of the right type. An example of the ``isa<>`` and ``cast<>`` 126 template is: 127 128 .. code-block:: c++ 129 130 static bool isLoopInvariant(const Value *V, const Loop *L) { 131 if (isa<Constant>(V) || isa<Argument>(V) || isa<GlobalValue>(V)) 132 return true; 133 134 // Otherwise, it must be an instruction... 135 return !L->contains(cast<Instruction>(V)->getParent()); 136 } 137 138 Note that you should **not** use an ``isa<>`` test followed by a ``cast<>``, 139 for that use the ``dyn_cast<>`` operator. 140 141 ``dyn_cast<>``: 142 The ``dyn_cast<>`` operator is a "checking cast" operation. It checks to see 143 if the operand is of the specified type, and if so, returns a pointer to it 144 (this operator does not work with references). If the operand is not of the 145 correct type, a null pointer is returned. Thus, this works very much like 146 the ``dynamic_cast<>`` operator in C++, and should be used in the same 147 circumstances. Typically, the ``dyn_cast<>`` operator is used in an ``if`` 148 statement or some other flow control statement like this: 149 150 .. code-block:: c++ 151 152 if (AllocationInst *AI = dyn_cast<AllocationInst>(Val)) { 153 // ... 154 } 155 156 This form of the ``if`` statement effectively combines together a call to 157 ``isa<>`` and a call to ``cast<>`` into one statement, which is very 158 convenient. 159 160 Note that the ``dyn_cast<>`` operator, like C++'s ``dynamic_cast<>`` or Java's 161 ``instanceof`` operator, can be abused. In particular, you should not use big 162 chained ``if/then/else`` blocks to check for lots of different variants of 163 classes. If you find yourself wanting to do this, it is much cleaner and more 164 efficient to use the ``InstVisitor`` class to dispatch over the instruction 165 type directly. 166 167 ``cast_or_null<>``: 168 The ``cast_or_null<>`` operator works just like the ``cast<>`` operator, 169 except that it allows for a null pointer as an argument (which it then 170 propagates). This can sometimes be useful, allowing you to combine several 171 null checks into one. 172 173 ``dyn_cast_or_null<>``: 174 The ``dyn_cast_or_null<>`` operator works just like the ``dyn_cast<>`` 175 operator, except that it allows for a null pointer as an argument (which it 176 then propagates). This can sometimes be useful, allowing you to combine 177 several null checks into one. 178 179 These five templates can be used with any classes, whether they have a v-table 180 or not. If you want to add support for these templates, see the document 181 :doc:`How to set up LLVM-style RTTI for your class hierarchy 182 <HowToSetUpLLVMStyleRTTI>` 183 184 .. _string_apis: 185 186 Passing strings (the ``StringRef`` and ``Twine`` classes) 187 --------------------------------------------------------- 188 189 Although LLVM generally does not do much string manipulation, we do have several 190 important APIs which take strings. Two important examples are the Value class 191 -- which has names for instructions, functions, etc. -- and the ``StringMap`` 192 class which is used extensively in LLVM and Clang. 193 194 These are generic classes, and they need to be able to accept strings which may 195 have embedded null characters. Therefore, they cannot simply take a ``const 196 char *``, and taking a ``const std::string&`` requires clients to perform a heap 197 allocation which is usually unnecessary. Instead, many LLVM APIs use a 198 ``StringRef`` or a ``const Twine&`` for passing strings efficiently. 199 200 .. _StringRef: 201 202 The ``StringRef`` class 203 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 204 205 The ``StringRef`` data type represents a reference to a constant string (a 206 character array and a length) and supports the common operations available on 207 ``std::string``, but does not require heap allocation. 208 209 It can be implicitly constructed using a C style null-terminated string, an 210 ``std::string``, or explicitly with a character pointer and length. For 211 example, the ``StringRef`` find function is declared as: 212 213 .. code-block:: c++ 214 215 iterator find(StringRef Key); 216 217 and clients can call it using any one of: 218 219 .. code-block:: c++ 220 221 Map.find("foo"); // Lookup "foo" 222 Map.find(std::string("bar")); // Lookup "bar" 223 Map.find(StringRef("\0baz", 4)); // Lookup "\0baz" 224 225 Similarly, APIs which need to return a string may return a ``StringRef`` 226 instance, which can be used directly or converted to an ``std::string`` using 227 the ``str`` member function. See ``llvm/ADT/StringRef.h`` (`doxygen 228 <http://llvm.org/doxygen/classllvm_1_1StringRef_8h-source.html>`__) for more 229 information. 230 231 You should rarely use the ``StringRef`` class directly, because it contains 232 pointers to external memory it is not generally safe to store an instance of the 233 class (unless you know that the external storage will not be freed). 234 ``StringRef`` is small and pervasive enough in LLVM that it should always be 235 passed by value. 236 237 The ``Twine`` class 238 ^^^^^^^^^^^^^^^^^^^ 239 240 The ``Twine`` (`doxygen <http://llvm.org/doxygen/classllvm_1_1Twine.html>`__) 241 class is an efficient way for APIs to accept concatenated strings. For example, 242 a common LLVM paradigm is to name one instruction based on the name of another 243 instruction with a suffix, for example: 244 245 .. code-block:: c++ 246 247 New = CmpInst::Create(..., SO->getName() + ".cmp"); 248 249 The ``Twine`` class is effectively a lightweight `rope 250 <http://en.wikipedia.org/wiki/Rope_(computer_science)>`_ which points to 251 temporary (stack allocated) objects. Twines can be implicitly constructed as 252 the result of the plus operator applied to strings (i.e., a C strings, an 253 ``std::string``, or a ``StringRef``). The twine delays the actual concatenation 254 of strings until it is actually required, at which point it can be efficiently 255 rendered directly into a character array. This avoids unnecessary heap 256 allocation involved in constructing the temporary results of string 257 concatenation. See ``llvm/ADT/Twine.h`` (`doxygen 258 <http://llvm.org/doxygen/Twine_8h_source.html>`__) and :ref:`here <dss_twine>` 259 for more information. 260 261 As with a ``StringRef``, ``Twine`` objects point to external memory and should 262 almost never be stored or mentioned directly. They are intended solely for use 263 when defining a function which should be able to efficiently accept concatenated 264 strings. 265 266 .. _function_apis: 267 268 Passing functions and other callable objects 269 -------------------------------------------- 270 271 Sometimes you may want a function to be passed a callback object. In order to 272 support lambda expressions and other function objects, you should not use the 273 traditional C approach of taking a function pointer and an opaque cookie: 274 275 .. code-block:: c++ 276 277 void takeCallback(bool (*Callback)(Function *, void *), void *Cookie); 278 279 Instead, use one of the following approaches: 280 281 Function template 282 ^^^^^^^^^^^^^^^^^ 283 284 If you don't mind putting the definition of your function into a header file, 285 make it a function template that is templated on the callable type. 286 287 .. code-block:: c++ 288 289 template<typename Callable> 290 void takeCallback(Callable Callback) { 291 Callback(1, 2, 3); 292 } 293 294 The ``function_ref`` class template 295 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 296 297 The ``function_ref`` 298 (`doxygen <http://llvm.org/doxygen/classllvm_1_1function_ref.html>`__) class 299 template represents a reference to a callable object, templated over the type 300 of the callable. This is a good choice for passing a callback to a function, 301 if you don't need to hold onto the callback after the function returns. 302 303 ``function_ref<Ret(Param1, Param2, ...)>`` can be implicitly constructed from 304 any callable object that can be called with arguments of type ``Param1``, 305 ``Param2``, ..., and returns a value that can be converted to type ``Ret``. 306 For example: 307 308 .. code-block:: c++ 309 310 void visitBasicBlocks(Function *F, function_ref<bool (BasicBlock*)> Callback) { 311 for (BasicBlock &BB : *F) 312 if (Callback(&BB)) 313 return; 314 } 315 316 can be called using: 317 318 .. code-block:: c++ 319 320 visitBasicBlocks(F, [&](BasicBlock *BB) { 321 if (process(BB)) 322 return isEmpty(BB); 323 return false; 324 }); 325 326 Note that a ``function_ref`` object contains pointers to external memory, so 327 it is not generally safe to store an instance of the class (unless you know 328 that the external storage will not be freed). 329 ``function_ref`` is small enough that it should always be passed by value. 330 331 ``std::function`` 332 ^^^^^^^^^^^^^^^^^ 333 334 You cannot use ``std::function`` within LLVM code, because it is not supported 335 by all our target toolchains. 336 337 338 .. _DEBUG: 339 340 The ``DEBUG()`` macro and ``-debug`` option 341 ------------------------------------------- 342 343 Often when working on your pass you will put a bunch of debugging printouts and 344 other code into your pass. After you get it working, you want to remove it, but 345 you may need it again in the future (to work out new bugs that you run across). 346 347 Naturally, because of this, you don't want to delete the debug printouts, but 348 you don't want them to always be noisy. A standard compromise is to comment 349 them out, allowing you to enable them if you need them in the future. 350 351 The ``llvm/Support/Debug.h`` (`doxygen 352 <http://llvm.org/doxygen/Debug_8h-source.html>`__) file provides a macro named 353 ``DEBUG()`` that is a much nicer solution to this problem. Basically, you can 354 put arbitrary code into the argument of the ``DEBUG`` macro, and it is only 355 executed if '``opt``' (or any other tool) is run with the '``-debug``' command 356 line argument: 357 358 .. code-block:: c++ 359 360 DEBUG(errs() << "I am here!\n"); 361 362 Then you can run your pass like this: 363 364 .. code-block:: none 365 366 $ opt < a.bc > /dev/null -mypass 367 <no output> 368 $ opt < a.bc > /dev/null -mypass -debug 369 I am here! 370 371 Using the ``DEBUG()`` macro instead of a home-brewed solution allows you to not 372 have to create "yet another" command line option for the debug output for your 373 pass. Note that ``DEBUG()`` macros are disabled for optimized builds, so they 374 do not cause a performance impact at all (for the same reason, they should also 375 not contain side-effects!). 376 377 One additional nice thing about the ``DEBUG()`` macro is that you can enable or 378 disable it directly in gdb. Just use "``set DebugFlag=0``" or "``set 379 DebugFlag=1``" from the gdb if the program is running. If the program hasn't 380 been started yet, you can always just run it with ``-debug``. 381 382 .. _DEBUG_TYPE: 383 384 Fine grained debug info with ``DEBUG_TYPE`` and the ``-debug-only`` option 385 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 386 387 Sometimes you may find yourself in a situation where enabling ``-debug`` just 388 turns on **too much** information (such as when working on the code generator). 389 If you want to enable debug information with more fine-grained control, you 390 can define the ``DEBUG_TYPE`` macro and use the ``-debug-only`` option as 391 follows: 392 393 .. code-block:: c++ 394 395 #undef DEBUG_TYPE 396 DEBUG(errs() << "No debug type\n"); 397 #define DEBUG_TYPE "foo" 398 DEBUG(errs() << "'foo' debug type\n"); 399 #undef DEBUG_TYPE 400 #define DEBUG_TYPE "bar" 401 DEBUG(errs() << "'bar' debug type\n")); 402 #undef DEBUG_TYPE 403 #define DEBUG_TYPE "" 404 DEBUG(errs() << "No debug type (2)\n"); 405 406 Then you can run your pass like this: 407 408 .. code-block:: none 409 410 $ opt < a.bc > /dev/null -mypass 411 <no output> 412 $ opt < a.bc > /dev/null -mypass -debug 413 No debug type 414 'foo' debug type 415 'bar' debug type 416 No debug type (2) 417 $ opt < a.bc > /dev/null -mypass -debug-only=foo 418 'foo' debug type 419 $ opt < a.bc > /dev/null -mypass -debug-only=bar 420 'bar' debug type 421 422 Of course, in practice, you should only set ``DEBUG_TYPE`` at the top of a file, 423 to specify the debug type for the entire module (if you do this before you 424 ``#include "llvm/Support/Debug.h"``, you don't have to insert the ugly 425 ``#undef``'s). Also, you should use names more meaningful than "foo" and "bar", 426 because there is no system in place to ensure that names do not conflict. If 427 two different modules use the same string, they will all be turned on when the 428 name is specified. This allows, for example, all debug information for 429 instruction scheduling to be enabled with ``-debug-type=InstrSched``, even if 430 the source lives in multiple files. 431 432 The ``DEBUG_WITH_TYPE`` macro is also available for situations where you would 433 like to set ``DEBUG_TYPE``, but only for one specific ``DEBUG`` statement. It 434 takes an additional first parameter, which is the type to use. For example, the 435 preceding example could be written as: 436 437 .. code-block:: c++ 438 439 DEBUG_WITH_TYPE("", errs() << "No debug type\n"); 440 DEBUG_WITH_TYPE("foo", errs() << "'foo' debug type\n"); 441 DEBUG_WITH_TYPE("bar", errs() << "'bar' debug type\n")); 442 DEBUG_WITH_TYPE("", errs() << "No debug type (2)\n"); 443 444 .. _Statistic: 445 446 The ``Statistic`` class & ``-stats`` option 447 ------------------------------------------- 448 449 The ``llvm/ADT/Statistic.h`` (`doxygen 450 <http://llvm.org/doxygen/Statistic_8h-source.html>`__) file provides a class 451 named ``Statistic`` that is used as a unified way to keep track of what the LLVM 452 compiler is doing and how effective various optimizations are. It is useful to 453 see what optimizations are contributing to making a particular program run 454 faster. 455 456 Often you may run your pass on some big program, and you're interested to see 457 how many times it makes a certain transformation. Although you can do this with 458 hand inspection, or some ad-hoc method, this is a real pain and not very useful 459 for big programs. Using the ``Statistic`` class makes it very easy to keep 460 track of this information, and the calculated information is presented in a 461 uniform manner with the rest of the passes being executed. 462 463 There are many examples of ``Statistic`` uses, but the basics of using it are as 464 follows: 465 466 #. Define your statistic like this: 467 468 .. code-block:: c++ 469 470 #define DEBUG_TYPE "mypassname" // This goes before any #includes. 471 STATISTIC(NumXForms, "The # of times I did stuff"); 472 473 The ``STATISTIC`` macro defines a static variable, whose name is specified by 474 the first argument. The pass name is taken from the ``DEBUG_TYPE`` macro, and 475 the description is taken from the second argument. The variable defined 476 ("NumXForms" in this case) acts like an unsigned integer. 477 478 #. Whenever you make a transformation, bump the counter: 479 480 .. code-block:: c++ 481 482 ++NumXForms; // I did stuff! 483 484 That's all you have to do. To get '``opt``' to print out the statistics 485 gathered, use the '``-stats``' option: 486 487 .. code-block:: none 488 489 $ opt -stats -mypassname < program.bc > /dev/null 490 ... statistics output ... 491 492 When running ``opt`` on a C file from the SPEC benchmark suite, it gives a 493 report that looks like this: 494 495 .. code-block:: none 496 497 7646 bitcodewriter - Number of normal instructions 498 725 bitcodewriter - Number of oversized instructions 499 129996 bitcodewriter - Number of bitcode bytes written 500 2817 raise - Number of insts DCEd or constprop'd 501 3213 raise - Number of cast-of-self removed 502 5046 raise - Number of expression trees converted 503 75 raise - Number of other getelementptr's formed 504 138 raise - Number of load/store peepholes 505 42 deadtypeelim - Number of unused typenames removed from symtab 506 392 funcresolve - Number of varargs functions resolved 507 27 globaldce - Number of global variables removed 508 2 adce - Number of basic blocks removed 509 134 cee - Number of branches revectored 510 49 cee - Number of setcc instruction eliminated 511 532 gcse - Number of loads removed 512 2919 gcse - Number of instructions removed 513 86 indvars - Number of canonical indvars added 514 87 indvars - Number of aux indvars removed 515 25 instcombine - Number of dead inst eliminate 516 434 instcombine - Number of insts combined 517 248 licm - Number of load insts hoisted 518 1298 licm - Number of insts hoisted to a loop pre-header 519 3 licm - Number of insts hoisted to multiple loop preds (bad, no loop pre-header) 520 75 mem2reg - Number of alloca's promoted 521 1444 cfgsimplify - Number of blocks simplified 522 523 Obviously, with so many optimizations, having a unified framework for this stuff 524 is very nice. Making your pass fit well into the framework makes it more 525 maintainable and useful. 526 527 .. _ViewGraph: 528 529 Viewing graphs while debugging code 530 ----------------------------------- 531 532 Several of the important data structures in LLVM are graphs: for example CFGs 533 made out of LLVM :ref:`BasicBlocks <BasicBlock>`, CFGs made out of LLVM 534 :ref:`MachineBasicBlocks <MachineBasicBlock>`, and :ref:`Instruction Selection 535 DAGs <SelectionDAG>`. In many cases, while debugging various parts of the 536 compiler, it is nice to instantly visualize these graphs. 537 538 LLVM provides several callbacks that are available in a debug build to do 539 exactly that. If you call the ``Function::viewCFG()`` method, for example, the 540 current LLVM tool will pop up a window containing the CFG for the function where 541 each basic block is a node in the graph, and each node contains the instructions 542 in the block. Similarly, there also exists ``Function::viewCFGOnly()`` (does 543 not include the instructions), the ``MachineFunction::viewCFG()`` and 544 ``MachineFunction::viewCFGOnly()``, and the ``SelectionDAG::viewGraph()`` 545 methods. Within GDB, for example, you can usually use something like ``call 546 DAG.viewGraph()`` to pop up a window. Alternatively, you can sprinkle calls to 547 these functions in your code in places you want to debug. 548 549 Getting this to work requires a small amount of setup. On Unix systems 550 with X11, install the `graphviz <http://www.graphviz.org>`_ toolkit, and make 551 sure 'dot' and 'gv' are in your path. If you are running on Mac OS X, download 552 and install the Mac OS X `Graphviz program 553 <http://www.pixelglow.com/graphviz/>`_ and add 554 ``/Applications/Graphviz.app/Contents/MacOS/`` (or wherever you install it) to 555 your path. The programs need not be present when configuring, building or 556 running LLVM and can simply be installed when needed during an active debug 557 session. 558 559 ``SelectionDAG`` has been extended to make it easier to locate *interesting* 560 nodes in large complex graphs. From gdb, if you ``call DAG.setGraphColor(node, 561 "color")``, then the next ``call DAG.viewGraph()`` would highlight the node in 562 the specified color (choices of colors can be found at `colors 563 <http://www.graphviz.org/doc/info/colors.html>`_.) More complex node attributes 564 can be provided with ``call DAG.setGraphAttrs(node, "attributes")`` (choices can 565 be found at `Graph attributes <http://www.graphviz.org/doc/info/attrs.html>`_.) 566 If you want to restart and clear all the current graph attributes, then you can 567 ``call DAG.clearGraphAttrs()``. 568 569 Note that graph visualization features are compiled out of Release builds to 570 reduce file size. This means that you need a Debug+Asserts or Release+Asserts 571 build to use these features. 572 573 .. _datastructure: 574 575 Picking the Right Data Structure for a Task 576 =========================================== 577 578 LLVM has a plethora of data structures in the ``llvm/ADT/`` directory, and we 579 commonly use STL data structures. This section describes the trade-offs you 580 should consider when you pick one. 581 582 The first step is a choose your own adventure: do you want a sequential 583 container, a set-like container, or a map-like container? The most important 584 thing when choosing a container is the algorithmic properties of how you plan to 585 access the container. Based on that, you should use: 586 587 588 * a :ref:`map-like <ds_map>` container if you need efficient look-up of a 589 value based on another value. Map-like containers also support efficient 590 queries for containment (whether a key is in the map). Map-like containers 591 generally do not support efficient reverse mapping (values to keys). If you 592 need that, use two maps. Some map-like containers also support efficient 593 iteration through the keys in sorted order. Map-like containers are the most 594 expensive sort, only use them if you need one of these capabilities. 595 596 * a :ref:`set-like <ds_set>` container if you need to put a bunch of stuff into 597 a container that automatically eliminates duplicates. Some set-like 598 containers support efficient iteration through the elements in sorted order. 599 Set-like containers are more expensive than sequential containers. 600 601 * a :ref:`sequential <ds_sequential>` container provides the most efficient way 602 to add elements and keeps track of the order they are added to the collection. 603 They permit duplicates and support efficient iteration, but do not support 604 efficient look-up based on a key. 605 606 * a :ref:`string <ds_string>` container is a specialized sequential container or 607 reference structure that is used for character or byte arrays. 608 609 * a :ref:`bit <ds_bit>` container provides an efficient way to store and 610 perform set operations on sets of numeric id's, while automatically 611 eliminating duplicates. Bit containers require a maximum of 1 bit for each 612 identifier you want to store. 613 614 Once the proper category of container is determined, you can fine tune the 615 memory use, constant factors, and cache behaviors of access by intelligently 616 picking a member of the category. Note that constant factors and cache behavior 617 can be a big deal. If you have a vector that usually only contains a few 618 elements (but could contain many), for example, it's much better to use 619 :ref:`SmallVector <dss_smallvector>` than :ref:`vector <dss_vector>`. Doing so 620 avoids (relatively) expensive malloc/free calls, which dwarf the cost of adding 621 the elements to the container. 622 623 .. _ds_sequential: 624 625 Sequential Containers (std::vector, std::list, etc) 626 --------------------------------------------------- 627 628 There are a variety of sequential containers available for you, based on your 629 needs. Pick the first in this section that will do what you want. 630 631 .. _dss_arrayref: 632 633 llvm/ADT/ArrayRef.h 634 ^^^^^^^^^^^^^^^^^^^ 635 636 The ``llvm::ArrayRef`` class is the preferred class to use in an interface that 637 accepts a sequential list of elements in memory and just reads from them. By 638 taking an ``ArrayRef``, the API can be passed a fixed size array, an 639 ``std::vector``, an ``llvm::SmallVector`` and anything else that is contiguous 640 in memory. 641 642 .. _dss_fixedarrays: 643 644 Fixed Size Arrays 645 ^^^^^^^^^^^^^^^^^ 646 647 Fixed size arrays are very simple and very fast. They are good if you know 648 exactly how many elements you have, or you have a (low) upper bound on how many 649 you have. 650 651 .. _dss_heaparrays: 652 653 Heap Allocated Arrays 654 ^^^^^^^^^^^^^^^^^^^^^ 655 656 Heap allocated arrays (``new[]`` + ``delete[]``) are also simple. They are good 657 if the number of elements is variable, if you know how many elements you will 658 need before the array is allocated, and if the array is usually large (if not, 659 consider a :ref:`SmallVector <dss_smallvector>`). The cost of a heap allocated 660 array is the cost of the new/delete (aka malloc/free). Also note that if you 661 are allocating an array of a type with a constructor, the constructor and 662 destructors will be run for every element in the array (re-sizable vectors only 663 construct those elements actually used). 664 665 .. _dss_tinyptrvector: 666 667 llvm/ADT/TinyPtrVector.h 668 ^^^^^^^^^^^^^^^^^^^^^^^^ 669 670 ``TinyPtrVector<Type>`` is a highly specialized collection class that is 671 optimized to avoid allocation in the case when a vector has zero or one 672 elements. It has two major restrictions: 1) it can only hold values of pointer 673 type, and 2) it cannot hold a null pointer. 674 675 Since this container is highly specialized, it is rarely used. 676 677 .. _dss_smallvector: 678 679 llvm/ADT/SmallVector.h 680 ^^^^^^^^^^^^^^^^^^^^^^ 681 682 ``SmallVector<Type, N>`` is a simple class that looks and smells just like 683 ``vector<Type>``: it supports efficient iteration, lays out elements in memory 684 order (so you can do pointer arithmetic between elements), supports efficient 685 push_back/pop_back operations, supports efficient random access to its elements, 686 etc. 687 688 The advantage of SmallVector is that it allocates space for some number of 689 elements (N) **in the object itself**. Because of this, if the SmallVector is 690 dynamically smaller than N, no malloc is performed. This can be a big win in 691 cases where the malloc/free call is far more expensive than the code that 692 fiddles around with the elements. 693 694 This is good for vectors that are "usually small" (e.g. the number of 695 predecessors/successors of a block is usually less than 8). On the other hand, 696 this makes the size of the SmallVector itself large, so you don't want to 697 allocate lots of them (doing so will waste a lot of space). As such, 698 SmallVectors are most useful when on the stack. 699 700 SmallVector also provides a nice portable and efficient replacement for 701 ``alloca``. 702 703 .. note:: 704 705 Prefer to use ``SmallVectorImpl<T>`` as a parameter type. 706 707 In APIs that don't care about the "small size" (most?), prefer to use 708 the ``SmallVectorImpl<T>`` class, which is basically just the "vector 709 header" (and methods) without the elements allocated after it. Note that 710 ``SmallVector<T, N>`` inherits from ``SmallVectorImpl<T>`` so the 711 conversion is implicit and costs nothing. E.g. 712 713 .. code-block:: c++ 714 715 // BAD: Clients cannot pass e.g. SmallVector<Foo, 4>. 716 hardcodedSmallSize(SmallVector<Foo, 2> &Out); 717 // GOOD: Clients can pass any SmallVector<Foo, N>. 718 allowsAnySmallSize(SmallVectorImpl<Foo> &Out); 719 720 void someFunc() { 721 SmallVector<Foo, 8> Vec; 722 hardcodedSmallSize(Vec); // Error. 723 allowsAnySmallSize(Vec); // Works. 724 } 725 726 Even though it has "``Impl``" in the name, this is so widely used that 727 it really isn't "private to the implementation" anymore. A name like 728 ``SmallVectorHeader`` would be more appropriate. 729 730 .. _dss_vector: 731 732 <vector> 733 ^^^^^^^^ 734 735 ``std::vector`` is well loved and respected. It is useful when SmallVector 736 isn't: when the size of the vector is often large (thus the small optimization 737 will rarely be a benefit) or if you will be allocating many instances of the 738 vector itself (which would waste space for elements that aren't in the 739 container). vector is also useful when interfacing with code that expects 740 vectors :). 741 742 One worthwhile note about std::vector: avoid code like this: 743 744 .. code-block:: c++ 745 746 for ( ... ) { 747 std::vector<foo> V; 748 // make use of V. 749 } 750 751 Instead, write this as: 752 753 .. code-block:: c++ 754 755 std::vector<foo> V; 756 for ( ... ) { 757 // make use of V. 758 V.clear(); 759 } 760 761 Doing so will save (at least) one heap allocation and free per iteration of the 762 loop. 763 764 .. _dss_deque: 765 766 <deque> 767 ^^^^^^^ 768 769 ``std::deque`` is, in some senses, a generalized version of ``std::vector``. 770 Like ``std::vector``, it provides constant time random access and other similar 771 properties, but it also provides efficient access to the front of the list. It 772 does not guarantee continuity of elements within memory. 773 774 In exchange for this extra flexibility, ``std::deque`` has significantly higher 775 constant factor costs than ``std::vector``. If possible, use ``std::vector`` or 776 something cheaper. 777 778 .. _dss_list: 779 780 <list> 781 ^^^^^^ 782 783 ``std::list`` is an extremely inefficient class that is rarely useful. It 784 performs a heap allocation for every element inserted into it, thus having an 785 extremely high constant factor, particularly for small data types. 786 ``std::list`` also only supports bidirectional iteration, not random access 787 iteration. 788 789 In exchange for this high cost, std::list supports efficient access to both ends 790 of the list (like ``std::deque``, but unlike ``std::vector`` or 791 ``SmallVector``). In addition, the iterator invalidation characteristics of 792 std::list are stronger than that of a vector class: inserting or removing an 793 element into the list does not invalidate iterator or pointers to other elements 794 in the list. 795 796 .. _dss_ilist: 797 798 llvm/ADT/ilist.h 799 ^^^^^^^^^^^^^^^^ 800 801 ``ilist<T>`` implements an 'intrusive' doubly-linked list. It is intrusive, 802 because it requires the element to store and provide access to the prev/next 803 pointers for the list. 804 805 ``ilist`` has the same drawbacks as ``std::list``, and additionally requires an 806 ``ilist_traits`` implementation for the element type, but it provides some novel 807 characteristics. In particular, it can efficiently store polymorphic objects, 808 the traits class is informed when an element is inserted or removed from the 809 list, and ``ilist``\ s are guaranteed to support a constant-time splice 810 operation. 811 812 These properties are exactly what we want for things like ``Instruction``\ s and 813 basic blocks, which is why these are implemented with ``ilist``\ s. 814 815 Related classes of interest are explained in the following subsections: 816 817 * :ref:`ilist_traits <dss_ilist_traits>` 818 819 * :ref:`iplist <dss_iplist>` 820 821 * :ref:`llvm/ADT/ilist_node.h <dss_ilist_node>` 822 823 * :ref:`Sentinels <dss_ilist_sentinel>` 824 825 .. _dss_packedvector: 826 827 llvm/ADT/PackedVector.h 828 ^^^^^^^^^^^^^^^^^^^^^^^ 829 830 Useful for storing a vector of values using only a few number of bits for each 831 value. Apart from the standard operations of a vector-like container, it can 832 also perform an 'or' set operation. 833 834 For example: 835 836 .. code-block:: c++ 837 838 enum State { 839 None = 0x0, 840 FirstCondition = 0x1, 841 SecondCondition = 0x2, 842 Both = 0x3 843 }; 844 845 State get() { 846 PackedVector<State, 2> Vec1; 847 Vec1.push_back(FirstCondition); 848 849 PackedVector<State, 2> Vec2; 850 Vec2.push_back(SecondCondition); 851 852 Vec1 |= Vec2; 853 return Vec1[0]; // returns 'Both'. 854 } 855 856 .. _dss_ilist_traits: 857 858 ilist_traits 859 ^^^^^^^^^^^^ 860 861 ``ilist_traits<T>`` is ``ilist<T>``'s customization mechanism. ``iplist<T>`` 862 (and consequently ``ilist<T>``) publicly derive from this traits class. 863 864 .. _dss_iplist: 865 866 iplist 867 ^^^^^^ 868 869 ``iplist<T>`` is ``ilist<T>``'s base and as such supports a slightly narrower 870 interface. Notably, inserters from ``T&`` are absent. 871 872 ``ilist_traits<T>`` is a public base of this class and can be used for a wide 873 variety of customizations. 874 875 .. _dss_ilist_node: 876 877 llvm/ADT/ilist_node.h 878 ^^^^^^^^^^^^^^^^^^^^^ 879 880 ``ilist_node<T>`` implements a the forward and backward links that are expected 881 by the ``ilist<T>`` (and analogous containers) in the default manner. 882 883 ``ilist_node<T>``\ s are meant to be embedded in the node type ``T``, usually 884 ``T`` publicly derives from ``ilist_node<T>``. 885 886 .. _dss_ilist_sentinel: 887 888 Sentinels 889 ^^^^^^^^^ 890 891 ``ilist``\ s have another specialty that must be considered. To be a good 892 citizen in the C++ ecosystem, it needs to support the standard container 893 operations, such as ``begin`` and ``end`` iterators, etc. Also, the 894 ``operator--`` must work correctly on the ``end`` iterator in the case of 895 non-empty ``ilist``\ s. 896 897 The only sensible solution to this problem is to allocate a so-called *sentinel* 898 along with the intrusive list, which serves as the ``end`` iterator, providing 899 the back-link to the last element. However conforming to the C++ convention it 900 is illegal to ``operator++`` beyond the sentinel and it also must not be 901 dereferenced. 902 903 These constraints allow for some implementation freedom to the ``ilist`` how to 904 allocate and store the sentinel. The corresponding policy is dictated by 905 ``ilist_traits<T>``. By default a ``T`` gets heap-allocated whenever the need 906 for a sentinel arises. 907 908 While the default policy is sufficient in most cases, it may break down when 909 ``T`` does not provide a default constructor. Also, in the case of many 910 instances of ``ilist``\ s, the memory overhead of the associated sentinels is 911 wasted. To alleviate the situation with numerous and voluminous 912 ``T``-sentinels, sometimes a trick is employed, leading to *ghostly sentinels*. 913 914 Ghostly sentinels are obtained by specially-crafted ``ilist_traits<T>`` which 915 superpose the sentinel with the ``ilist`` instance in memory. Pointer 916 arithmetic is used to obtain the sentinel, which is relative to the ``ilist``'s 917 ``this`` pointer. The ``ilist`` is augmented by an extra pointer, which serves 918 as the back-link of the sentinel. This is the only field in the ghostly 919 sentinel which can be legally accessed. 920 921 .. _dss_other: 922 923 Other Sequential Container options 924 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 925 926 Other STL containers are available, such as ``std::string``. 927 928 There are also various STL adapter classes such as ``std::queue``, 929 ``std::priority_queue``, ``std::stack``, etc. These provide simplified access 930 to an underlying container but don't affect the cost of the container itself. 931 932 .. _ds_string: 933 934 String-like containers 935 ---------------------- 936 937 There are a variety of ways to pass around and use strings in C and C++, and 938 LLVM adds a few new options to choose from. Pick the first option on this list 939 that will do what you need, they are ordered according to their relative cost. 940 941 Note that is is generally preferred to *not* pass strings around as ``const 942 char*``'s. These have a number of problems, including the fact that they 943 cannot represent embedded nul ("\0") characters, and do not have a length 944 available efficiently. The general replacement for '``const char*``' is 945 StringRef. 946 947 For more information on choosing string containers for APIs, please see 948 :ref:`Passing Strings <string_apis>`. 949 950 .. _dss_stringref: 951 952 llvm/ADT/StringRef.h 953 ^^^^^^^^^^^^^^^^^^^^ 954 955 The StringRef class is a simple value class that contains a pointer to a 956 character and a length, and is quite related to the :ref:`ArrayRef 957 <dss_arrayref>` class (but specialized for arrays of characters). Because 958 StringRef carries a length with it, it safely handles strings with embedded nul 959 characters in it, getting the length does not require a strlen call, and it even 960 has very convenient APIs for slicing and dicing the character range that it 961 represents. 962 963 StringRef is ideal for passing simple strings around that are known to be live, 964 either because they are C string literals, std::string, a C array, or a 965 SmallVector. Each of these cases has an efficient implicit conversion to 966 StringRef, which doesn't result in a dynamic strlen being executed. 967 968 StringRef has a few major limitations which make more powerful string containers 969 useful: 970 971 #. You cannot directly convert a StringRef to a 'const char*' because there is 972 no way to add a trailing nul (unlike the .c_str() method on various stronger 973 classes). 974 975 #. StringRef doesn't own or keep alive the underlying string bytes. 976 As such it can easily lead to dangling pointers, and is not suitable for 977 embedding in datastructures in most cases (instead, use an std::string or 978 something like that). 979 980 #. For the same reason, StringRef cannot be used as the return value of a 981 method if the method "computes" the result string. Instead, use std::string. 982 983 #. StringRef's do not allow you to mutate the pointed-to string bytes and it 984 doesn't allow you to insert or remove bytes from the range. For editing 985 operations like this, it interoperates with the :ref:`Twine <dss_twine>` 986 class. 987 988 Because of its strengths and limitations, it is very common for a function to 989 take a StringRef and for a method on an object to return a StringRef that points 990 into some string that it owns. 991 992 .. _dss_twine: 993 994 llvm/ADT/Twine.h 995 ^^^^^^^^^^^^^^^^ 996 997 The Twine class is used as an intermediary datatype for APIs that want to take a 998 string that can be constructed inline with a series of concatenations. Twine 999 works by forming recursive instances of the Twine datatype (a simple value 1000 object) on the stack as temporary objects, linking them together into a tree 1001 which is then linearized when the Twine is consumed. Twine is only safe to use 1002 as the argument to a function, and should always be a const reference, e.g.: 1003 1004 .. code-block:: c++ 1005 1006 void foo(const Twine &T); 1007 ... 1008 StringRef X = ... 1009 unsigned i = ... 1010 foo(X + "." + Twine(i)); 1011 1012 This example forms a string like "blarg.42" by concatenating the values 1013 together, and does not form intermediate strings containing "blarg" or "blarg.". 1014 1015 Because Twine is constructed with temporary objects on the stack, and because 1016 these instances are destroyed at the end of the current statement, it is an 1017 inherently dangerous API. For example, this simple variant contains undefined 1018 behavior and will probably crash: 1019 1020 .. code-block:: c++ 1021 1022 void foo(const Twine &T); 1023 ... 1024 StringRef X = ... 1025 unsigned i = ... 1026 const Twine &Tmp = X + "." + Twine(i); 1027 foo(Tmp); 1028 1029 ... because the temporaries are destroyed before the call. That said, Twine's 1030 are much more efficient than intermediate std::string temporaries, and they work 1031 really well with StringRef. Just be aware of their limitations. 1032 1033 .. _dss_smallstring: 1034 1035 llvm/ADT/SmallString.h 1036 ^^^^^^^^^^^^^^^^^^^^^^ 1037 1038 SmallString is a subclass of :ref:`SmallVector <dss_smallvector>` that adds some 1039 convenience APIs like += that takes StringRef's. SmallString avoids allocating 1040 memory in the case when the preallocated space is enough to hold its data, and 1041 it calls back to general heap allocation when required. Since it owns its data, 1042 it is very safe to use and supports full mutation of the string. 1043 1044 Like SmallVector's, the big downside to SmallString is their sizeof. While they 1045 are optimized for small strings, they themselves are not particularly small. 1046 This means that they work great for temporary scratch buffers on the stack, but 1047 should not generally be put into the heap: it is very rare to see a SmallString 1048 as the member of a frequently-allocated heap data structure or returned 1049 by-value. 1050 1051 .. _dss_stdstring: 1052 1053 std::string 1054 ^^^^^^^^^^^ 1055 1056 The standard C++ std::string class is a very general class that (like 1057 SmallString) owns its underlying data. sizeof(std::string) is very reasonable 1058 so it can be embedded into heap data structures and returned by-value. On the 1059 other hand, std::string is highly inefficient for inline editing (e.g. 1060 concatenating a bunch of stuff together) and because it is provided by the 1061 standard library, its performance characteristics depend a lot of the host 1062 standard library (e.g. libc++ and MSVC provide a highly optimized string class, 1063 GCC contains a really slow implementation). 1064 1065 The major disadvantage of std::string is that almost every operation that makes 1066 them larger can allocate memory, which is slow. As such, it is better to use 1067 SmallVector or Twine as a scratch buffer, but then use std::string to persist 1068 the result. 1069 1070 .. _ds_set: 1071 1072 Set-Like Containers (std::set, SmallSet, SetVector, etc) 1073 -------------------------------------------------------- 1074 1075 Set-like containers are useful when you need to canonicalize multiple values 1076 into a single representation. There are several different choices for how to do 1077 this, providing various trade-offs. 1078 1079 .. _dss_sortedvectorset: 1080 1081 A sorted 'vector' 1082 ^^^^^^^^^^^^^^^^^ 1083 1084 If you intend to insert a lot of elements, then do a lot of queries, a great 1085 approach is to use a vector (or other sequential container) with 1086 std::sort+std::unique to remove duplicates. This approach works really well if 1087 your usage pattern has these two distinct phases (insert then query), and can be 1088 coupled with a good choice of :ref:`sequential container <ds_sequential>`. 1089 1090 This combination provides the several nice properties: the result data is 1091 contiguous in memory (good for cache locality), has few allocations, is easy to 1092 address (iterators in the final vector are just indices or pointers), and can be 1093 efficiently queried with a standard binary search (e.g. 1094 ``std::lower_bound``; if you want the whole range of elements comparing 1095 equal, use ``std::equal_range``). 1096 1097 .. _dss_smallset: 1098 1099 llvm/ADT/SmallSet.h 1100 ^^^^^^^^^^^^^^^^^^^ 1101 1102 If you have a set-like data structure that is usually small and whose elements 1103 are reasonably small, a ``SmallSet<Type, N>`` is a good choice. This set has 1104 space for N elements in place (thus, if the set is dynamically smaller than N, 1105 no malloc traffic is required) and accesses them with a simple linear search. 1106 When the set grows beyond 'N' elements, it allocates a more expensive 1107 representation that guarantees efficient access (for most types, it falls back 1108 to std::set, but for pointers it uses something far better, :ref:`SmallPtrSet 1109 <dss_smallptrset>`. 1110 1111 The magic of this class is that it handles small sets extremely efficiently, but 1112 gracefully handles extremely large sets without loss of efficiency. The 1113 drawback is that the interface is quite small: it supports insertion, queries 1114 and erasing, but does not support iteration. 1115 1116 .. _dss_smallptrset: 1117 1118 llvm/ADT/SmallPtrSet.h 1119 ^^^^^^^^^^^^^^^^^^^^^^ 1120 1121 SmallPtrSet has all the advantages of ``SmallSet`` (and a ``SmallSet`` of 1122 pointers is transparently implemented with a ``SmallPtrSet``), but also supports 1123 iterators. If more than 'N' insertions are performed, a single quadratically 1124 probed hash table is allocated and grows as needed, providing extremely 1125 efficient access (constant time insertion/deleting/queries with low constant 1126 factors) and is very stingy with malloc traffic. 1127 1128 Note that, unlike ``std::set``, the iterators of ``SmallPtrSet`` are invalidated 1129 whenever an insertion occurs. Also, the values visited by the iterators are not 1130 visited in sorted order. 1131 1132 .. _dss_denseset: 1133 1134 llvm/ADT/DenseSet.h 1135 ^^^^^^^^^^^^^^^^^^^ 1136 1137 DenseSet is a simple quadratically probed hash table. It excels at supporting 1138 small values: it uses a single allocation to hold all of the pairs that are 1139 currently inserted in the set. DenseSet is a great way to unique small values 1140 that are not simple pointers (use :ref:`SmallPtrSet <dss_smallptrset>` for 1141 pointers). Note that DenseSet has the same requirements for the value type that 1142 :ref:`DenseMap <dss_densemap>` has. 1143 1144 .. _dss_sparseset: 1145 1146 llvm/ADT/SparseSet.h 1147 ^^^^^^^^^^^^^^^^^^^^ 1148 1149 SparseSet holds a small number of objects identified by unsigned keys of 1150 moderate size. It uses a lot of memory, but provides operations that are almost 1151 as fast as a vector. Typical keys are physical registers, virtual registers, or 1152 numbered basic blocks. 1153 1154 SparseSet is useful for algorithms that need very fast clear/find/insert/erase 1155 and fast iteration over small sets. It is not intended for building composite 1156 data structures. 1157 1158 .. _dss_sparsemultiset: 1159 1160 llvm/ADT/SparseMultiSet.h 1161 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 1162 1163 SparseMultiSet adds multiset behavior to SparseSet, while retaining SparseSet's 1164 desirable attributes. Like SparseSet, it typically uses a lot of memory, but 1165 provides operations that are almost as fast as a vector. Typical keys are 1166 physical registers, virtual registers, or numbered basic blocks. 1167 1168 SparseMultiSet is useful for algorithms that need very fast 1169 clear/find/insert/erase of the entire collection, and iteration over sets of 1170 elements sharing a key. It is often a more efficient choice than using composite 1171 data structures (e.g. vector-of-vectors, map-of-vectors). It is not intended for 1172 building composite data structures. 1173 1174 .. _dss_FoldingSet: 1175 1176 llvm/ADT/FoldingSet.h 1177 ^^^^^^^^^^^^^^^^^^^^^ 1178 1179 FoldingSet is an aggregate class that is really good at uniquing 1180 expensive-to-create or polymorphic objects. It is a combination of a chained 1181 hash table with intrusive links (uniqued objects are required to inherit from 1182 FoldingSetNode) that uses :ref:`SmallVector <dss_smallvector>` as part of its ID 1183 process. 1184 1185 Consider a case where you want to implement a "getOrCreateFoo" method for a 1186 complex object (for example, a node in the code generator). The client has a 1187 description of **what** it wants to generate (it knows the opcode and all the 1188 operands), but we don't want to 'new' a node, then try inserting it into a set 1189 only to find out it already exists, at which point we would have to delete it 1190 and return the node that already exists. 1191 1192 To support this style of client, FoldingSet perform a query with a 1193 FoldingSetNodeID (which wraps SmallVector) that can be used to describe the 1194 element that we want to query for. The query either returns the element 1195 matching the ID or it returns an opaque ID that indicates where insertion should 1196 take place. Construction of the ID usually does not require heap traffic. 1197 1198 Because FoldingSet uses intrusive links, it can support polymorphic objects in 1199 the set (for example, you can have SDNode instances mixed with LoadSDNodes). 1200 Because the elements are individually allocated, pointers to the elements are 1201 stable: inserting or removing elements does not invalidate any pointers to other 1202 elements. 1203 1204 .. _dss_set: 1205 1206 <set> 1207 ^^^^^ 1208 1209 ``std::set`` is a reasonable all-around set class, which is decent at many 1210 things but great at nothing. std::set allocates memory for each element 1211 inserted (thus it is very malloc intensive) and typically stores three pointers 1212 per element in the set (thus adding a large amount of per-element space 1213 overhead). It offers guaranteed log(n) performance, which is not particularly 1214 fast from a complexity standpoint (particularly if the elements of the set are 1215 expensive to compare, like strings), and has extremely high constant factors for 1216 lookup, insertion and removal. 1217 1218 The advantages of std::set are that its iterators are stable (deleting or 1219 inserting an element from the set does not affect iterators or pointers to other 1220 elements) and that iteration over the set is guaranteed to be in sorted order. 1221 If the elements in the set are large, then the relative overhead of the pointers 1222 and malloc traffic is not a big deal, but if the elements of the set are small, 1223 std::set is almost never a good choice. 1224 1225 .. _dss_setvector: 1226 1227 llvm/ADT/SetVector.h 1228 ^^^^^^^^^^^^^^^^^^^^ 1229 1230 LLVM's ``SetVector<Type>`` is an adapter class that combines your choice of a 1231 set-like container along with a :ref:`Sequential Container <ds_sequential>` The 1232 important property that this provides is efficient insertion with uniquing 1233 (duplicate elements are ignored) with iteration support. It implements this by 1234 inserting elements into both a set-like container and the sequential container, 1235 using the set-like container for uniquing and the sequential container for 1236 iteration. 1237 1238 The difference between SetVector and other sets is that the order of iteration 1239 is guaranteed to match the order of insertion into the SetVector. This property 1240 is really important for things like sets of pointers. Because pointer values 1241 are non-deterministic (e.g. vary across runs of the program on different 1242 machines), iterating over the pointers in the set will not be in a well-defined 1243 order. 1244 1245 The drawback of SetVector is that it requires twice as much space as a normal 1246 set and has the sum of constant factors from the set-like container and the 1247 sequential container that it uses. Use it **only** if you need to iterate over 1248 the elements in a deterministic order. SetVector is also expensive to delete 1249 elements out of (linear time), unless you use its "pop_back" method, which is 1250 faster. 1251 1252 ``SetVector`` is an adapter class that defaults to using ``std::vector`` and a 1253 size 16 ``SmallSet`` for the underlying containers, so it is quite expensive. 1254 However, ``"llvm/ADT/SetVector.h"`` also provides a ``SmallSetVector`` class, 1255 which defaults to using a ``SmallVector`` and ``SmallSet`` of a specified size. 1256 If you use this, and if your sets are dynamically smaller than ``N``, you will 1257 save a lot of heap traffic. 1258 1259 .. _dss_uniquevector: 1260 1261 llvm/ADT/UniqueVector.h 1262 ^^^^^^^^^^^^^^^^^^^^^^^ 1263 1264 UniqueVector is similar to :ref:`SetVector <dss_setvector>` but it retains a 1265 unique ID for each element inserted into the set. It internally contains a map 1266 and a vector, and it assigns a unique ID for each value inserted into the set. 1267 1268 UniqueVector is very expensive: its cost is the sum of the cost of maintaining 1269 both the map and vector, it has high complexity, high constant factors, and 1270 produces a lot of malloc traffic. It should be avoided. 1271 1272 .. _dss_immutableset: 1273 1274 llvm/ADT/ImmutableSet.h 1275 ^^^^^^^^^^^^^^^^^^^^^^^ 1276 1277 ImmutableSet is an immutable (functional) set implementation based on an AVL 1278 tree. Adding or removing elements is done through a Factory object and results 1279 in the creation of a new ImmutableSet object. If an ImmutableSet already exists 1280 with the given contents, then the existing one is returned; equality is compared 1281 with a FoldingSetNodeID. The time and space complexity of add or remove 1282 operations is logarithmic in the size of the original set. 1283 1284 There is no method for returning an element of the set, you can only check for 1285 membership. 1286 1287 .. _dss_otherset: 1288 1289 Other Set-Like Container Options 1290 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 1291 1292 The STL provides several other options, such as std::multiset and the various 1293 "hash_set" like containers (whether from C++ TR1 or from the SGI library). We 1294 never use hash_set and unordered_set because they are generally very expensive 1295 (each insertion requires a malloc) and very non-portable. 1296 1297 std::multiset is useful if you're not interested in elimination of duplicates, 1298 but has all the drawbacks of std::set. A sorted vector (where you don't delete 1299 duplicate entries) or some other approach is almost always better. 1300 1301 .. _ds_map: 1302 1303 Map-Like Containers (std::map, DenseMap, etc) 1304 --------------------------------------------- 1305 1306 Map-like containers are useful when you want to associate data to a key. As 1307 usual, there are a lot of different ways to do this. :) 1308 1309 .. _dss_sortedvectormap: 1310 1311 A sorted 'vector' 1312 ^^^^^^^^^^^^^^^^^ 1313 1314 If your usage pattern follows a strict insert-then-query approach, you can 1315 trivially use the same approach as :ref:`sorted vectors for set-like containers 1316 <dss_sortedvectorset>`. The only difference is that your query function (which 1317 uses std::lower_bound to get efficient log(n) lookup) should only compare the 1318 key, not both the key and value. This yields the same advantages as sorted 1319 vectors for sets. 1320 1321 .. _dss_stringmap: 1322 1323 llvm/ADT/StringMap.h 1324 ^^^^^^^^^^^^^^^^^^^^ 1325 1326 Strings are commonly used as keys in maps, and they are difficult to support 1327 efficiently: they are variable length, inefficient to hash and compare when 1328 long, expensive to copy, etc. StringMap is a specialized container designed to 1329 cope with these issues. It supports mapping an arbitrary range of bytes to an 1330 arbitrary other object. 1331 1332 The StringMap implementation uses a quadratically-probed hash table, where the 1333 buckets store a pointer to the heap allocated entries (and some other stuff). 1334 The entries in the map must be heap allocated because the strings are variable 1335 length. The string data (key) and the element object (value) are stored in the 1336 same allocation with the string data immediately after the element object. 1337 This container guarantees the "``(char*)(&Value+1)``" points to the key string 1338 for a value. 1339 1340 The StringMap is very fast for several reasons: quadratic probing is very cache 1341 efficient for lookups, the hash value of strings in buckets is not recomputed 1342 when looking up an element, StringMap rarely has to touch the memory for 1343 unrelated objects when looking up a value (even when hash collisions happen), 1344 hash table growth does not recompute the hash values for strings already in the 1345 table, and each pair in the map is store in a single allocation (the string data 1346 is stored in the same allocation as the Value of a pair). 1347 1348 StringMap also provides query methods that take byte ranges, so it only ever 1349 copies a string if a value is inserted into the table. 1350 1351 StringMap iteratation order, however, is not guaranteed to be deterministic, so 1352 any uses which require that should instead use a std::map. 1353 1354 .. _dss_indexmap: 1355 1356 llvm/ADT/IndexedMap.h 1357 ^^^^^^^^^^^^^^^^^^^^^ 1358 1359 IndexedMap is a specialized container for mapping small dense integers (or 1360 values that can be mapped to small dense integers) to some other type. It is 1361 internally implemented as a vector with a mapping function that maps the keys 1362 to the dense integer range. 1363 1364 This is useful for cases like virtual registers in the LLVM code generator: they 1365 have a dense mapping that is offset by a compile-time constant (the first 1366 virtual register ID). 1367 1368 .. _dss_densemap: 1369 1370 llvm/ADT/DenseMap.h 1371 ^^^^^^^^^^^^^^^^^^^ 1372 1373 DenseMap is a simple quadratically probed hash table. It excels at supporting 1374 small keys and values: it uses a single allocation to hold all of the pairs 1375 that are currently inserted in the map. DenseMap is a great way to map 1376 pointers to pointers, or map other small types to each other. 1377 1378 There are several aspects of DenseMap that you should be aware of, however. 1379 The iterators in a DenseMap are invalidated whenever an insertion occurs, 1380 unlike map. Also, because DenseMap allocates space for a large number of 1381 key/value pairs (it starts with 64 by default), it will waste a lot of space if 1382 your keys or values are large. Finally, you must implement a partial 1383 specialization of DenseMapInfo for the key that you want, if it isn't already 1384 supported. This is required to tell DenseMap about two special marker values 1385 (which can never be inserted into the map) that it needs internally. 1386 1387 DenseMap's find_as() method supports lookup operations using an alternate key 1388 type. This is useful in cases where the normal key type is expensive to 1389 construct, but cheap to compare against. The DenseMapInfo is responsible for 1390 defining the appropriate comparison and hashing methods for each alternate key 1391 type used. 1392 1393 .. _dss_valuemap: 1394 1395 llvm/IR/ValueMap.h 1396 ^^^^^^^^^^^^^^^^^^^ 1397 1398 ValueMap is a wrapper around a :ref:`DenseMap <dss_densemap>` mapping 1399 ``Value*``\ s (or subclasses) to another type. When a Value is deleted or 1400 RAUW'ed, ValueMap will update itself so the new version of the key is mapped to 1401 the same value, just as if the key were a WeakVH. You can configure exactly how 1402 this happens, and what else happens on these two events, by passing a ``Config`` 1403 parameter to the ValueMap template. 1404 1405 .. _dss_intervalmap: 1406 1407 llvm/ADT/IntervalMap.h 1408 ^^^^^^^^^^^^^^^^^^^^^^ 1409 1410 IntervalMap is a compact map for small keys and values. It maps key intervals 1411 instead of single keys, and it will automatically coalesce adjacent intervals. 1412 When then map only contains a few intervals, they are stored in the map object 1413 itself to avoid allocations. 1414 1415 The IntervalMap iterators are quite big, so they should not be passed around as 1416 STL iterators. The heavyweight iterators allow a smaller data structure. 1417 1418 .. _dss_map: 1419 1420 <map> 1421 ^^^^^ 1422 1423 std::map has similar characteristics to :ref:`std::set <dss_set>`: it uses a 1424 single allocation per pair inserted into the map, it offers log(n) lookup with 1425 an extremely large constant factor, imposes a space penalty of 3 pointers per 1426 pair in the map, etc. 1427 1428 std::map is most useful when your keys or values are very large, if you need to 1429 iterate over the collection in sorted order, or if you need stable iterators 1430 into the map (i.e. they don't get invalidated if an insertion or deletion of 1431 another element takes place). 1432 1433 .. _dss_mapvector: 1434 1435 llvm/ADT/MapVector.h 1436 ^^^^^^^^^^^^^^^^^^^^ 1437 1438 ``MapVector<KeyT,ValueT>`` provides a subset of the DenseMap interface. The 1439 main difference is that the iteration order is guaranteed to be the insertion 1440 order, making it an easy (but somewhat expensive) solution for non-deterministic 1441 iteration over maps of pointers. 1442 1443 It is implemented by mapping from key to an index in a vector of key,value 1444 pairs. This provides fast lookup and iteration, but has two main drawbacks: The 1445 key is stored twice and it doesn't support removing elements. 1446 1447 .. _dss_inteqclasses: 1448 1449 llvm/ADT/IntEqClasses.h 1450 ^^^^^^^^^^^^^^^^^^^^^^^ 1451 1452 IntEqClasses provides a compact representation of equivalence classes of small 1453 integers. Initially, each integer in the range 0..n-1 has its own equivalence 1454 class. Classes can be joined by passing two class representatives to the 1455 join(a, b) method. Two integers are in the same class when findLeader() returns 1456 the same representative. 1457 1458 Once all equivalence classes are formed, the map can be compressed so each 1459 integer 0..n-1 maps to an equivalence class number in the range 0..m-1, where m 1460 is the total number of equivalence classes. The map must be uncompressed before 1461 it can be edited again. 1462 1463 .. _dss_immutablemap: 1464 1465 llvm/ADT/ImmutableMap.h 1466 ^^^^^^^^^^^^^^^^^^^^^^^ 1467 1468 ImmutableMap is an immutable (functional) map implementation based on an AVL 1469 tree. Adding or removing elements is done through a Factory object and results 1470 in the creation of a new ImmutableMap object. If an ImmutableMap already exists 1471 with the given key set, then the existing one is returned; equality is compared 1472 with a FoldingSetNodeID. The time and space complexity of add or remove 1473 operations is logarithmic in the size of the original map. 1474 1475 .. _dss_othermap: 1476 1477 Other Map-Like Container Options 1478 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 1479 1480 The STL provides several other options, such as std::multimap and the various 1481 "hash_map" like containers (whether from C++ TR1 or from the SGI library). We 1482 never use hash_set and unordered_set because they are generally very expensive 1483 (each insertion requires a malloc) and very non-portable. 1484 1485 std::multimap is useful if you want to map a key to multiple values, but has all 1486 the drawbacks of std::map. A sorted vector or some other approach is almost 1487 always better. 1488 1489 .. _ds_bit: 1490 1491 Bit storage containers (BitVector, SparseBitVector) 1492 --------------------------------------------------- 1493 1494 Unlike the other containers, there are only two bit storage containers, and 1495 choosing when to use each is relatively straightforward. 1496 1497 One additional option is ``std::vector<bool>``: we discourage its use for two 1498 reasons 1) the implementation in many common compilers (e.g. commonly 1499 available versions of GCC) is extremely inefficient and 2) the C++ standards 1500 committee is likely to deprecate this container and/or change it significantly 1501 somehow. In any case, please don't use it. 1502 1503 .. _dss_bitvector: 1504 1505 BitVector 1506 ^^^^^^^^^ 1507 1508 The BitVector container provides a dynamic size set of bits for manipulation. 1509 It supports individual bit setting/testing, as well as set operations. The set 1510 operations take time O(size of bitvector), but operations are performed one word 1511 at a time, instead of one bit at a time. This makes the BitVector very fast for 1512 set operations compared to other containers. Use the BitVector when you expect 1513 the number of set bits to be high (i.e. a dense set). 1514 1515 .. _dss_smallbitvector: 1516 1517 SmallBitVector 1518 ^^^^^^^^^^^^^^ 1519 1520 The SmallBitVector container provides the same interface as BitVector, but it is 1521 optimized for the case where only a small number of bits, less than 25 or so, 1522 are needed. It also transparently supports larger bit counts, but slightly less 1523 efficiently than a plain BitVector, so SmallBitVector should only be used when 1524 larger counts are rare. 1525 1526 At this time, SmallBitVector does not support set operations (and, or, xor), and 1527 its operator[] does not provide an assignable lvalue. 1528 1529 .. _dss_sparsebitvector: 1530 1531 SparseBitVector 1532 ^^^^^^^^^^^^^^^ 1533 1534 The SparseBitVector container is much like BitVector, with one major difference: 1535 Only the bits that are set, are stored. This makes the SparseBitVector much 1536 more space efficient than BitVector when the set is sparse, as well as making 1537 set operations O(number of set bits) instead of O(size of universe). The 1538 downside to the SparseBitVector is that setting and testing of random bits is 1539 O(N), and on large SparseBitVectors, this can be slower than BitVector. In our 1540 implementation, setting or testing bits in sorted order (either forwards or 1541 reverse) is O(1) worst case. Testing and setting bits within 128 bits (depends 1542 on size) of the current bit is also O(1). As a general statement, 1543 testing/setting bits in a SparseBitVector is O(distance away from last set bit). 1544 1545 .. _common: 1546 1547 Helpful Hints for Common Operations 1548 =================================== 1549 1550 This section describes how to perform some very simple transformations of LLVM 1551 code. This is meant to give examples of common idioms used, showing the 1552 practical side of LLVM transformations. 1553 1554 Because this is a "how-to" section, you should also read about the main classes 1555 that you will be working with. The :ref:`Core LLVM Class Hierarchy Reference 1556 <coreclasses>` contains details and descriptions of the main classes that you 1557 should know about. 1558 1559 .. _inspection: 1560 1561 Basic Inspection and Traversal Routines 1562 --------------------------------------- 1563 1564 The LLVM compiler infrastructure have many different data structures that may be 1565 traversed. Following the example of the C++ standard template library, the 1566 techniques used to traverse these various data structures are all basically the 1567 same. For a enumerable sequence of values, the ``XXXbegin()`` function (or 1568 method) returns an iterator to the start of the sequence, the ``XXXend()`` 1569 function returns an iterator pointing to one past the last valid element of the 1570 sequence, and there is some ``XXXiterator`` data type that is common between the 1571 two operations. 1572 1573 Because the pattern for iteration is common across many different aspects of the 1574 program representation, the standard template library algorithms may be used on 1575 them, and it is easier to remember how to iterate. First we show a few common 1576 examples of the data structures that need to be traversed. Other data 1577 structures are traversed in very similar ways. 1578 1579 .. _iterate_function: 1580 1581 Iterating over the ``BasicBlock`` in a ``Function`` 1582 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 1583 1584 It's quite common to have a ``Function`` instance that you'd like to transform 1585 in some way; in particular, you'd like to manipulate its ``BasicBlock``\ s. To 1586 facilitate this, you'll need to iterate over all of the ``BasicBlock``\ s that 1587 constitute the ``Function``. The following is an example that prints the name 1588 of a ``BasicBlock`` and the number of ``Instruction``\ s it contains: 1589 1590 .. code-block:: c++ 1591 1592 // func is a pointer to a Function instance 1593 for (Function::iterator i = func->begin(), e = func->end(); i != e; ++i) 1594 // Print out the name of the basic block if it has one, and then the 1595 // number of instructions that it contains 1596 errs() << "Basic block (name=" << i->getName() << ") has " 1597 << i->size() << " instructions.\n"; 1598 1599 Note that i can be used as if it were a pointer for the purposes of invoking 1600 member functions of the ``Instruction`` class. This is because the indirection 1601 operator is overloaded for the iterator classes. In the above code, the 1602 expression ``i->size()`` is exactly equivalent to ``(*i).size()`` just like 1603 you'd expect. 1604 1605 .. _iterate_basicblock: 1606 1607 Iterating over the ``Instruction`` in a ``BasicBlock`` 1608 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 1609 1610 Just like when dealing with ``BasicBlock``\ s in ``Function``\ s, it's easy to 1611 iterate over the individual instructions that make up ``BasicBlock``\ s. Here's 1612 a code snippet that prints out each instruction in a ``BasicBlock``: 1613 1614 .. code-block:: c++ 1615 1616 // blk is a pointer to a BasicBlock instance 1617 for (BasicBlock::iterator i = blk->begin(), e = blk->end(); i != e; ++i) 1618 // The next statement works since operator<<(ostream&,...) 1619 // is overloaded for Instruction& 1620 errs() << *i << "\n"; 1621 1622 1623 However, this isn't really the best way to print out the contents of a 1624 ``BasicBlock``! Since the ostream operators are overloaded for virtually 1625 anything you'll care about, you could have just invoked the print routine on the 1626 basic block itself: ``errs() << *blk << "\n";``. 1627 1628 .. _iterate_insiter: 1629 1630 Iterating over the ``Instruction`` in a ``Function`` 1631 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 1632 1633 If you're finding that you commonly iterate over a ``Function``'s 1634 ``BasicBlock``\ s and then that ``BasicBlock``'s ``Instruction``\ s, 1635 ``InstIterator`` should be used instead. You'll need to include 1636 ``llvm/IR/InstIterator.h`` (`doxygen 1637 <http://llvm.org/doxygen/InstIterator_8h.html>`__) and then instantiate 1638 ``InstIterator``\ s explicitly in your code. Here's a small example that shows 1639 how to dump all instructions in a function to the standard error stream: 1640 1641 .. code-block:: c++ 1642 1643 #include "llvm/IR/InstIterator.h" 1644 1645 // F is a pointer to a Function instance 1646 for (inst_iterator I = inst_begin(F), E = inst_end(F); I != E; ++I) 1647 errs() << *I << "\n"; 1648 1649 Easy, isn't it? You can also use ``InstIterator``\ s to fill a work list with 1650 its initial contents. For example, if you wanted to initialize a work list to 1651 contain all instructions in a ``Function`` F, all you would need to do is 1652 something like: 1653 1654 .. code-block:: c++ 1655 1656 std::set<Instruction*> worklist; 1657 // or better yet, SmallPtrSet<Instruction*, 64> worklist; 1658 1659 for (inst_iterator I = inst_begin(F), E = inst_end(F); I != E; ++I) 1660 worklist.insert(&*I); 1661 1662 The STL set ``worklist`` would now contain all instructions in the ``Function`` 1663 pointed to by F. 1664 1665 .. _iterate_convert: 1666 1667 Turning an iterator into a class pointer (and vice-versa) 1668 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 1669 1670 Sometimes, it'll be useful to grab a reference (or pointer) to a class instance 1671 when all you've got at hand is an iterator. Well, extracting a reference or a 1672 pointer from an iterator is very straight-forward. Assuming that ``i`` is a 1673 ``BasicBlock::iterator`` and ``j`` is a ``BasicBlock::const_iterator``: 1674 1675 .. code-block:: c++ 1676 1677 Instruction& inst = *i; // Grab reference to instruction reference 1678 Instruction* pinst = &*i; // Grab pointer to instruction reference 1679 const Instruction& inst = *j; 1680 1681 However, the iterators you'll be working with in the LLVM framework are special: 1682 they will automatically convert to a ptr-to-instance type whenever they need to. 1683 Instead of derferencing the iterator and then taking the address of the result, 1684 you can simply assign the iterator to the proper pointer type and you get the 1685 dereference and address-of operation as a result of the assignment (behind the 1686 scenes, this is a result of overloading casting mechanisms). Thus the last line 1687 of the last example, 1688 1689 .. code-block:: c++ 1690 1691 Instruction *pinst = &*i; 1692 1693 is semantically equivalent to 1694 1695 .. code-block:: c++ 1696 1697 Instruction *pinst = i; 1698 1699 It's also possible to turn a class pointer into the corresponding iterator, and 1700 this is a constant time operation (very efficient). The following code snippet 1701 illustrates use of the conversion constructors provided by LLVM iterators. By 1702 using these, you can explicitly grab the iterator of something without actually 1703 obtaining it via iteration over some structure: 1704 1705 .. code-block:: c++ 1706 1707 void printNextInstruction(Instruction* inst) { 1708 BasicBlock::iterator it(inst); 1709 ++it; // After this line, it refers to the instruction after *inst 1710 if (it != inst->getParent()->end()) errs() << *it << "\n"; 1711 } 1712 1713 Unfortunately, these implicit conversions come at a cost; they prevent these 1714 iterators from conforming to standard iterator conventions, and thus from being 1715 usable with standard algorithms and containers. For example, they prevent the 1716 following code, where ``B`` is a ``BasicBlock``, from compiling: 1717 1718 .. code-block:: c++ 1719 1720 llvm::SmallVector<llvm::Instruction *, 16>(B->begin(), B->end()); 1721 1722 Because of this, these implicit conversions may be removed some day, and 1723 ``operator*`` changed to return a pointer instead of a reference. 1724 1725 .. _iterate_complex: 1726 1727 Finding call sites: a slightly more complex example 1728 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 1729 1730 Say that you're writing a FunctionPass and would like to count all the locations 1731 in the entire module (that is, across every ``Function``) where a certain 1732 function (i.e., some ``Function *``) is already in scope. As you'll learn 1733 later, you may want to use an ``InstVisitor`` to accomplish this in a much more 1734 straight-forward manner, but this example will allow us to explore how you'd do 1735 it if you didn't have ``InstVisitor`` around. In pseudo-code, this is what we 1736 want to do: 1737 1738 .. code-block:: none 1739 1740 initialize callCounter to zero 1741 for each Function f in the Module 1742 for each BasicBlock b in f 1743 for each Instruction i in b 1744 if (i is a CallInst and calls the given function) 1745 increment callCounter 1746 1747 And the actual code is (remember, because we're writing a ``FunctionPass``, our 1748 ``FunctionPass``-derived class simply has to override the ``runOnFunction`` 1749 method): 1750 1751 .. code-block:: c++ 1752 1753 Function* targetFunc = ...; 1754 1755 class OurFunctionPass : public FunctionPass { 1756 public: 1757 OurFunctionPass(): callCounter(0) { } 1758 1759 virtual runOnFunction(Function& F) { 1760 for (Function::iterator b = F.begin(), be = F.end(); b != be; ++b) { 1761 for (BasicBlock::iterator i = b->begin(), ie = b->end(); i != ie; ++i) { 1762 if (CallInst* callInst = dyn_cast<CallInst>(&*i)) { 1763 // We know we've encountered a call instruction, so we 1764 // need to determine if it's a call to the 1765 // function pointed to by m_func or not. 1766 if (callInst->getCalledFunction() == targetFunc) 1767 ++callCounter; 1768 } 1769 } 1770 } 1771 } 1772 1773 private: 1774 unsigned callCounter; 1775 }; 1776 1777 .. _calls_and_invokes: 1778 1779 Treating calls and invokes the same way 1780 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 1781 1782 You may have noticed that the previous example was a bit oversimplified in that 1783 it did not deal with call sites generated by 'invoke' instructions. In this, 1784 and in other situations, you may find that you want to treat ``CallInst``\ s and 1785 ``InvokeInst``\ s the same way, even though their most-specific common base 1786 class is ``Instruction``, which includes lots of less closely-related things. 1787 For these cases, LLVM provides a handy wrapper class called ``CallSite`` 1788 (`doxygen <http://llvm.org/doxygen/classllvm_1_1CallSite.html>`__) It is 1789 essentially a wrapper around an ``Instruction`` pointer, with some methods that 1790 provide functionality common to ``CallInst``\ s and ``InvokeInst``\ s. 1791 1792 This class has "value semantics": it should be passed by value, not by reference 1793 and it should not be dynamically allocated or deallocated using ``operator new`` 1794 or ``operator delete``. It is efficiently copyable, assignable and 1795 constructable, with costs equivalents to that of a bare pointer. If you look at 1796 its definition, it has only a single pointer member. 1797 1798 .. _iterate_chains: 1799 1800 Iterating over def-use & use-def chains 1801 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 1802 1803 Frequently, we might have an instance of the ``Value`` class (`doxygen 1804 <http://llvm.org/doxygen/classllvm_1_1Value.html>`__) and we want to determine 1805 which ``User`` s use the ``Value``. The list of all ``User``\ s of a particular 1806 ``Value`` is called a *def-use* chain. For example, let's say we have a 1807 ``Function*`` named ``F`` to a particular function ``foo``. Finding all of the 1808 instructions that *use* ``foo`` is as simple as iterating over the *def-use* 1809 chain of ``F``: 1810 1811 .. code-block:: c++ 1812 1813 Function *F = ...; 1814 1815 for (User *U : GV->users()) { 1816 if (Instruction *Inst = dyn_cast<Instruction>(U)) { 1817 errs() << "F is used in instruction:\n"; 1818 errs() << *Inst << "\n"; 1819 } 1820 1821 Alternatively, it's common to have an instance of the ``User`` Class (`doxygen 1822 <http://llvm.org/doxygen/classllvm_1_1User.html>`__) and need to know what 1823 ``Value``\ s are used by it. The list of all ``Value``\ s used by a ``User`` is 1824 known as a *use-def* chain. Instances of class ``Instruction`` are common 1825 ``User`` s, so we might want to iterate over all of the values that a particular 1826 instruction uses (that is, the operands of the particular ``Instruction``): 1827 1828 .. code-block:: c++ 1829 1830 Instruction *pi = ...; 1831 1832 for (Use &U : pi->operands()) { 1833 Value *v = U.get(); 1834 // ... 1835 } 1836 1837 Declaring objects as ``const`` is an important tool of enforcing mutation free 1838 algorithms (such as analyses, etc.). For this purpose above iterators come in 1839 constant flavors as ``Value::const_use_iterator`` and 1840 ``Value::const_op_iterator``. They automatically arise when calling 1841 ``use/op_begin()`` on ``const Value*``\ s or ``const User*``\ s respectively. 1842 Upon dereferencing, they return ``const Use*``\ s. Otherwise the above patterns 1843 remain unchanged. 1844 1845 .. _iterate_preds: 1846 1847 Iterating over predecessors & successors of blocks 1848 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 1849 1850 Iterating over the predecessors and successors of a block is quite easy with the 1851 routines defined in ``"llvm/Support/CFG.h"``. Just use code like this to 1852 iterate over all predecessors of BB: 1853 1854 .. code-block:: c++ 1855 1856 #include "llvm/Support/CFG.h" 1857 BasicBlock *BB = ...; 1858 1859 for (pred_iterator PI = pred_begin(BB), E = pred_end(BB); PI != E; ++PI) { 1860 BasicBlock *Pred = *PI; 1861 // ... 1862 } 1863 1864 Similarly, to iterate over successors use ``succ_iterator/succ_begin/succ_end``. 1865 1866 .. _simplechanges: 1867 1868 Making simple changes 1869 --------------------- 1870 1871 There are some primitive transformation operations present in the LLVM 1872 infrastructure that are worth knowing about. When performing transformations, 1873 it's fairly common to manipulate the contents of basic blocks. This section 1874 describes some of the common methods for doing so and gives example code. 1875 1876 .. _schanges_creating: 1877 1878 Creating and inserting new ``Instruction``\ s 1879 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 1880 1881 *Instantiating Instructions* 1882 1883 Creation of ``Instruction``\ s is straight-forward: simply call the constructor 1884 for the kind of instruction to instantiate and provide the necessary parameters. 1885 For example, an ``AllocaInst`` only *requires* a (const-ptr-to) ``Type``. Thus: 1886 1887 .. code-block:: c++ 1888 1889 AllocaInst* ai = new AllocaInst(Type::Int32Ty); 1890 1891 will create an ``AllocaInst`` instance that represents the allocation of one 1892 integer in the current stack frame, at run time. Each ``Instruction`` subclass 1893 is likely to have varying default parameters which change the semantics of the 1894 instruction, so refer to the `doxygen documentation for the subclass of 1895 Instruction <http://llvm.org/doxygen/classllvm_1_1Instruction.html>`_ that 1896 you're interested in instantiating. 1897 1898 *Naming values* 1899 1900 It is very useful to name the values of instructions when you're able to, as 1901 this facilitates the debugging of your transformations. If you end up looking 1902 at generated LLVM machine code, you definitely want to have logical names 1903 associated with the results of instructions! By supplying a value for the 1904 ``Name`` (default) parameter of the ``Instruction`` constructor, you associate a 1905 logical name with the result of the instruction's execution at run time. For 1906 example, say that I'm writing a transformation that dynamically allocates space 1907 for an integer on the stack, and that integer is going to be used as some kind 1908 of index by some other code. To accomplish this, I place an ``AllocaInst`` at 1909 the first point in the first ``BasicBlock`` of some ``Function``, and I'm 1910 intending to use it within the same ``Function``. I might do: 1911 1912 .. code-block:: c++ 1913 1914 AllocaInst* pa = new AllocaInst(Type::Int32Ty, 0, "indexLoc"); 1915 1916 where ``indexLoc`` is now the logical name of the instruction's execution value, 1917 which is a pointer to an integer on the run time stack. 1918 1919 *Inserting instructions* 1920 1921 There are essentially three ways to insert an ``Instruction`` into an existing 1922 sequence of instructions that form a ``BasicBlock``: 1923 1924 * Insertion into an explicit instruction list 1925 1926 Given a ``BasicBlock* pb``, an ``Instruction* pi`` within that ``BasicBlock``, 1927 and a newly-created instruction we wish to insert before ``*pi``, we do the 1928 following: 1929 1930 .. code-block:: c++ 1931 1932 BasicBlock *pb = ...; 1933 Instruction *pi = ...; 1934 Instruction *newInst = new Instruction(...); 1935 1936 pb->getInstList().insert(pi, newInst); // Inserts newInst before pi in pb 1937 1938 Appending to the end of a ``BasicBlock`` is so common that the ``Instruction`` 1939 class and ``Instruction``-derived classes provide constructors which take a 1940 pointer to a ``BasicBlock`` to be appended to. For example code that looked 1941 like: 1942 1943 .. code-block:: c++ 1944 1945 BasicBlock *pb = ...; 1946 Instruction *newInst = new Instruction(...); 1947 1948 pb->getInstList().push_back(newInst); // Appends newInst to pb 1949 1950 becomes: 1951 1952 .. code-block:: c++ 1953 1954 BasicBlock *pb = ...; 1955 Instruction *newInst = new Instruction(..., pb); 1956 1957 which is much cleaner, especially if you are creating long instruction 1958 streams. 1959 1960 * Insertion into an implicit instruction list 1961 1962 ``Instruction`` instances that are already in ``BasicBlock``\ s are implicitly 1963 associated with an existing instruction list: the instruction list of the 1964 enclosing basic block. Thus, we could have accomplished the same thing as the 1965 above code without being given a ``BasicBlock`` by doing: 1966 1967 .. code-block:: c++ 1968 1969 Instruction *pi = ...; 1970 Instruction *newInst = new Instruction(...); 1971 1972 pi->getParent()->getInstList().insert(pi, newInst); 1973 1974 In fact, this sequence of steps occurs so frequently that the ``Instruction`` 1975 class and ``Instruction``-derived classes provide constructors which take (as 1976 a default parameter) a pointer to an ``Instruction`` which the newly-created 1977 ``Instruction`` should precede. That is, ``Instruction`` constructors are 1978 capable of inserting the newly-created instance into the ``BasicBlock`` of a 1979 provided instruction, immediately before that instruction. Using an 1980 ``Instruction`` constructor with a ``insertBefore`` (default) parameter, the 1981 above code becomes: 1982 1983 .. code-block:: c++ 1984 1985 Instruction* pi = ...; 1986 Instruction* newInst = new Instruction(..., pi); 1987 1988 which is much cleaner, especially if you're creating a lot of instructions and 1989 adding them to ``BasicBlock``\ s. 1990 1991 * Insertion using an instance of ``IRBuilder`` 1992 1993 Inserting several ``Instruction``\ s can be quite laborious using the previous 1994 methods. The ``IRBuilder`` is a convenience class that can be used to add 1995 several instructions to the end of a ``BasicBlock`` or before a particular 1996 ``Instruction``. It also supports constant folding and renaming named 1997 registers (see ``IRBuilder``'s template arguments). 1998 1999 The example below demonstrates a very simple use of the ``IRBuilder`` where 2000 three instructions are inserted before the instruction ``pi``. The first two 2001 instructions are Call instructions and third instruction multiplies the return 2002 value of the two calls. 2003 2004 .. code-block:: c++ 2005 2006 Instruction *pi = ...; 2007 IRBuilder<> Builder(pi); 2008 CallInst* callOne = Builder.CreateCall(...); 2009 CallInst* callTwo = Builder.CreateCall(...); 2010 Value* result = Builder.CreateMul(callOne, callTwo); 2011 2012 The example below is similar to the above example except that the created 2013 ``IRBuilder`` inserts instructions at the end of the ``BasicBlock`` ``pb``. 2014 2015 .. code-block:: c++ 2016 2017 BasicBlock *pb = ...; 2018 IRBuilder<> Builder(pb); 2019 CallInst* callOne = Builder.CreateCall(...); 2020 CallInst* callTwo = Builder.CreateCall(...); 2021 Value* result = Builder.CreateMul(callOne, callTwo); 2022 2023 See :doc:`tutorial/LangImpl3` for a practical use of the ``IRBuilder``. 2024 2025 2026 .. _schanges_deleting: 2027 2028 Deleting Instructions 2029 ^^^^^^^^^^^^^^^^^^^^^ 2030 2031 Deleting an instruction from an existing sequence of instructions that form a 2032 BasicBlock_ is very straight-forward: just call the instruction's 2033 ``eraseFromParent()`` method. For example: 2034 2035 .. code-block:: c++ 2036 2037 Instruction *I = .. ; 2038 I->eraseFromParent(); 2039 2040 This unlinks the instruction from its containing basic block and deletes it. If 2041 you'd just like to unlink the instruction from its containing basic block but 2042 not delete it, you can use the ``removeFromParent()`` method. 2043 2044 .. _schanges_replacing: 2045 2046 Replacing an Instruction with another Value 2047 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 2048 2049 Replacing individual instructions 2050 """"""""""""""""""""""""""""""""" 2051 2052 Including "`llvm/Transforms/Utils/BasicBlockUtils.h 2053 <http://llvm.org/doxygen/BasicBlockUtils_8h-source.html>`_" permits use of two 2054 very useful replace functions: ``ReplaceInstWithValue`` and 2055 ``ReplaceInstWithInst``. 2056 2057 .. _schanges_deleting_sub: 2058 2059 Deleting Instructions 2060 """"""""""""""""""""" 2061 2062 * ``ReplaceInstWithValue`` 2063 2064 This function replaces all uses of a given instruction with a value, and then 2065 removes the original instruction. The following example illustrates the 2066 replacement of the result of a particular ``AllocaInst`` that allocates memory 2067 for a single integer with a null pointer to an integer. 2068 2069 .. code-block:: c++ 2070 2071 AllocaInst* instToReplace = ...; 2072 BasicBlock::iterator ii(instToReplace); 2073 2074 ReplaceInstWithValue(instToReplace->getParent()->getInstList(), ii, 2075 Constant::getNullValue(PointerType::getUnqual(Type::Int32Ty))); 2076 2077 * ``ReplaceInstWithInst`` 2078 2079 This function replaces a particular instruction with another instruction, 2080 inserting the new instruction into the basic block at the location where the 2081 old instruction was, and replacing any uses of the old instruction with the 2082 new instruction. The following example illustrates the replacement of one 2083 ``AllocaInst`` with another. 2084 2085 .. code-block:: c++ 2086 2087 AllocaInst* instToReplace = ...; 2088 BasicBlock::iterator ii(instToReplace); 2089 2090 ReplaceInstWithInst(instToReplace->getParent()->getInstList(), ii, 2091 new AllocaInst(Type::Int32Ty, 0, "ptrToReplacedInt")); 2092 2093 2094 Replacing multiple uses of Users and Values 2095 """"""""""""""""""""""""""""""""""""""""""" 2096 2097 You can use ``Value::replaceAllUsesWith`` and ``User::replaceUsesOfWith`` to 2098 change more than one use at a time. See the doxygen documentation for the 2099 `Value Class <http://llvm.org/doxygen/classllvm_1_1Value.html>`_ and `User Class 2100 <http://llvm.org/doxygen/classllvm_1_1User.html>`_, respectively, for more 2101 information. 2102 2103 .. _schanges_deletingGV: 2104 2105 Deleting GlobalVariables 2106 ^^^^^^^^^^^^^^^^^^^^^^^^ 2107 2108 Deleting a global variable from a module is just as easy as deleting an 2109 Instruction. First, you must have a pointer to the global variable that you 2110 wish to delete. You use this pointer to erase it from its parent, the module. 2111 For example: 2112 2113 .. code-block:: c++ 2114 2115 GlobalVariable *GV = .. ; 2116 2117 GV->eraseFromParent(); 2118 2119 2120 .. _create_types: 2121 2122 How to Create Types 2123 ------------------- 2124 2125 In generating IR, you may need some complex types. If you know these types 2126 statically, you can use ``TypeBuilder<...>::get()``, defined in 2127 ``llvm/Support/TypeBuilder.h``, to retrieve them. ``TypeBuilder`` has two forms 2128 depending on whether you're building types for cross-compilation or native 2129 library use. ``TypeBuilder<T, true>`` requires that ``T`` be independent of the 2130 host environment, meaning that it's built out of types from the ``llvm::types`` 2131 (`doxygen <http://llvm.org/doxygen/namespacellvm_1_1types.html>`__) namespace 2132 and pointers, functions, arrays, etc. built of those. ``TypeBuilder<T, false>`` 2133 additionally allows native C types whose size may depend on the host compiler. 2134 For example, 2135 2136 .. code-block:: c++ 2137 2138 FunctionType *ft = TypeBuilder<types::i<8>(types::i<32>*), true>::get(); 2139 2140 is easier to read and write than the equivalent 2141 2142 .. code-block:: c++ 2143 2144 std::vector<const Type*> params; 2145 params.push_back(PointerType::getUnqual(Type::Int32Ty)); 2146 FunctionType *ft = FunctionType::get(Type::Int8Ty, params, false); 2147 2148 See the `class comment 2149 <http://llvm.org/doxygen/TypeBuilder_8h-source.html#l00001>`_ for more details. 2150 2151 .. _threading: 2152 2153 Threads and LLVM 2154 ================ 2155 2156 This section describes the interaction of the LLVM APIs with multithreading, 2157 both on the part of client applications, and in the JIT, in the hosted 2158 application. 2159 2160 Note that LLVM's support for multithreading is still relatively young. Up 2161 through version 2.5, the execution of threaded hosted applications was 2162 supported, but not threaded client access to the APIs. While this use case is 2163 now supported, clients *must* adhere to the guidelines specified below to ensure 2164 proper operation in multithreaded mode. 2165 2166 Note that, on Unix-like platforms, LLVM requires the presence of GCC's atomic 2167 intrinsics in order to support threaded operation. If you need a 2168 multhreading-capable LLVM on a platform without a suitably modern system 2169 compiler, consider compiling LLVM and LLVM-GCC in single-threaded mode, and 2170 using the resultant compiler to build a copy of LLVM with multithreading 2171 support. 2172 2173 .. _shutdown: 2174 2175 Ending Execution with ``llvm_shutdown()`` 2176 ----------------------------------------- 2177 2178 When you are done using the LLVM APIs, you should call ``llvm_shutdown()`` to 2179 deallocate memory used for internal structures. 2180 2181 .. _managedstatic: 2182 2183 Lazy Initialization with ``ManagedStatic`` 2184 ------------------------------------------ 2185 2186 ``ManagedStatic`` is a utility class in LLVM used to implement static 2187 initialization of static resources, such as the global type tables. In a 2188 single-threaded environment, it implements a simple lazy initialization scheme. 2189 When LLVM is compiled with support for multi-threading, however, it uses 2190 double-checked locking to implement thread-safe lazy initialization. 2191 2192 .. _llvmcontext: 2193 2194 Achieving Isolation with ``LLVMContext`` 2195 ---------------------------------------- 2196 2197 ``LLVMContext`` is an opaque class in the LLVM API which clients can use to 2198 operate multiple, isolated instances of LLVM concurrently within the same 2199 address space. For instance, in a hypothetical compile-server, the compilation 2200 of an individual translation unit is conceptually independent from all the 2201 others, and it would be desirable to be able to compile incoming translation 2202 units concurrently on independent server threads. Fortunately, ``LLVMContext`` 2203 exists to enable just this kind of scenario! 2204 2205 Conceptually, ``LLVMContext`` provides isolation. Every LLVM entity 2206 (``Module``\ s, ``Value``\ s, ``Type``\ s, ``Constant``\ s, etc.) in LLVM's 2207 in-memory IR belongs to an ``LLVMContext``. Entities in different contexts 2208 *cannot* interact with each other: ``Module``\ s in different contexts cannot be 2209 linked together, ``Function``\ s cannot be added to ``Module``\ s in different 2210 contexts, etc. What this means is that is is safe to compile on multiple 2211 threads simultaneously, as long as no two threads operate on entities within the 2212 same context. 2213 2214 In practice, very few places in the API require the explicit specification of a 2215 ``LLVMContext``, other than the ``Type`` creation/lookup APIs. Because every 2216 ``Type`` carries a reference to its owning context, most other entities can 2217 determine what context they belong to by looking at their own ``Type``. If you 2218 are adding new entities to LLVM IR, please try to maintain this interface 2219 design. 2220 2221 For clients that do *not* require the benefits of isolation, LLVM provides a 2222 convenience API ``getGlobalContext()``. This returns a global, lazily 2223 initialized ``LLVMContext`` that may be used in situations where isolation is 2224 not a concern. 2225 2226 .. _jitthreading: 2227 2228 Threads and the JIT 2229 ------------------- 2230 2231 LLVM's "eager" JIT compiler is safe to use in threaded programs. Multiple 2232 threads can call ``ExecutionEngine::getPointerToFunction()`` or 2233 ``ExecutionEngine::runFunction()`` concurrently, and multiple threads can run 2234 code output by the JIT concurrently. The user must still ensure that only one 2235 thread accesses IR in a given ``LLVMContext`` while another thread might be 2236 modifying it. One way to do that is to always hold the JIT lock while accessing 2237 IR outside the JIT (the JIT *modifies* the IR by adding ``CallbackVH``\ s). 2238 Another way is to only call ``getPointerToFunction()`` from the 2239 ``LLVMContext``'s thread. 2240 2241 When the JIT is configured to compile lazily (using 2242 ``ExecutionEngine::DisableLazyCompilation(false)``), there is currently a `race 2243 condition <http://llvm.org/bugs/show_bug.cgi?id=5184>`_ in updating call sites 2244 after a function is lazily-jitted. It's still possible to use the lazy JIT in a 2245 threaded program if you ensure that only one thread at a time can call any 2246 particular lazy stub and that the JIT lock guards any IR access, but we suggest 2247 using only the eager JIT in threaded programs. 2248 2249 .. _advanced: 2250 2251 Advanced Topics 2252 =============== 2253 2254 This section describes some of the advanced or obscure API's that most clients 2255 do not need to be aware of. These API's tend manage the inner workings of the 2256 LLVM system, and only need to be accessed in unusual circumstances. 2257 2258 .. _SymbolTable: 2259 2260 The ``ValueSymbolTable`` class 2261 ------------------------------ 2262 2263 The ``ValueSymbolTable`` (`doxygen 2264 <http://llvm.org/doxygen/classllvm_1_1ValueSymbolTable.html>`__) class provides 2265 a symbol table that the :ref:`Function <c_Function>` and Module_ classes use for 2266 naming value definitions. The symbol table can provide a name for any Value_. 2267 2268 Note that the ``SymbolTable`` class should not be directly accessed by most 2269 clients. It should only be used when iteration over the symbol table names 2270 themselves are required, which is very special purpose. Note that not all LLVM 2271 Value_\ s have names, and those without names (i.e. they have an empty name) do 2272 not exist in the symbol table. 2273 2274 Symbol tables support iteration over the values in the symbol table with 2275 ``begin/end/iterator`` and supports querying to see if a specific name is in the 2276 symbol table (with ``lookup``). The ``ValueSymbolTable`` class exposes no 2277 public mutator methods, instead, simply call ``setName`` on a value, which will 2278 autoinsert it into the appropriate symbol table. 2279 2280 .. _UserLayout: 2281 2282 The ``User`` and owned ``Use`` classes' memory layout 2283 ----------------------------------------------------- 2284 2285 The ``User`` (`doxygen <http://llvm.org/doxygen/classllvm_1_1User.html>`__) 2286 class provides a basis for expressing the ownership of ``User`` towards other 2287 `Value instance <http://llvm.org/doxygen/classllvm_1_1Value.html>`_\ s. The 2288 ``Use`` (`doxygen <http://llvm.org/doxygen/classllvm_1_1Use.html>`__) helper 2289 class is employed to do the bookkeeping and to facilitate *O(1)* addition and 2290 removal. 2291 2292 .. _Use2User: 2293 2294 Interaction and relationship between ``User`` and ``Use`` objects 2295 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 2296 2297 A subclass of ``User`` can choose between incorporating its ``Use`` objects or 2298 refer to them out-of-line by means of a pointer. A mixed variant (some ``Use`` 2299 s inline others hung off) is impractical and breaks the invariant that the 2300 ``Use`` objects belonging to the same ``User`` form a contiguous array. 2301 2302 We have 2 different layouts in the ``User`` (sub)classes: 2303 2304 * Layout a) 2305 2306 The ``Use`` object(s) are inside (resp. at fixed offset) of the ``User`` 2307 object and there are a fixed number of them. 2308 2309 * Layout b) 2310 2311 The ``Use`` object(s) are referenced by a pointer to an array from the 2312 ``User`` object and there may be a variable number of them. 2313 2314 As of v2.4 each layout still possesses a direct pointer to the start of the 2315 array of ``Use``\ s. Though not mandatory for layout a), we stick to this 2316 redundancy for the sake of simplicity. The ``User`` object also stores the 2317 number of ``Use`` objects it has. (Theoretically this information can also be 2318 calculated given the scheme presented below.) 2319 2320 Special forms of allocation operators (``operator new``) enforce the following 2321 memory layouts: 2322 2323 * Layout a) is modelled by prepending the ``User`` object by the ``Use[]`` 2324 array. 2325 2326 .. code-block:: none 2327 2328 ...---.---.---.---.-------... 2329 | P | P | P | P | User 2330 '''---'---'---'---'-------''' 2331 2332 * Layout b) is modelled by pointing at the ``Use[]`` array. 2333 2334 .. code-block:: none 2335 2336 .-------... 2337 | User 2338 '-------''' 2339 | 2340 v 2341 .---.---.---.---... 2342 | P | P | P | P | 2343 '---'---'---'---''' 2344 2345 *(In the above figures* '``P``' *stands for the* ``Use**`` *that is stored in 2346 each* ``Use`` *object in the member* ``Use::Prev`` *)* 2347 2348 .. _Waymarking: 2349 2350 The waymarking algorithm 2351 ^^^^^^^^^^^^^^^^^^^^^^^^ 2352 2353 Since the ``Use`` objects are deprived of the direct (back)pointer to their 2354 ``User`` objects, there must be a fast and exact method to recover it. This is 2355 accomplished by the following scheme: 2356 2357 A bit-encoding in the 2 LSBits (least significant bits) of the ``Use::Prev`` 2358 allows to find the start of the ``User`` object: 2359 2360 * ``00`` --- binary digit 0 2361 2362 * ``01`` --- binary digit 1 2363 2364 * ``10`` --- stop and calculate (``s``) 2365 2366 * ``11`` --- full stop (``S``) 2367 2368 Given a ``Use*``, all we have to do is to walk till we get a stop and we either 2369 have a ``User`` immediately behind or we have to walk to the next stop picking 2370 up digits and calculating the offset: 2371 2372 .. code-block:: none 2373 2374 .---.---.---.---.---.---.---.---.---.---.---.---.---.---.---.---.---------------- 2375 | 1 | s | 1 | 0 | 1 | 0 | s | 1 | 1 | 0 | s | 1 | 1 | s | 1 | S | User (or User*) 2376 '---'---'---'---'---'---'---'---'---'---'---'---'---'---'---'---'---------------- 2377 |+15 |+10 |+6 |+3 |+1 2378 | | | | | __> 2379 | | | | __________> 2380 | | | ______________________> 2381 | | ______________________________________> 2382 | __________________________________________________________> 2383 2384 Only the significant number of bits need to be stored between the stops, so that 2385 the *worst case is 20 memory accesses* when there are 1000 ``Use`` objects 2386 associated with a ``User``. 2387 2388 .. _ReferenceImpl: 2389 2390 Reference implementation 2391 ^^^^^^^^^^^^^^^^^^^^^^^^ 2392 2393 The following literate Haskell fragment demonstrates the concept: 2394 2395 .. code-block:: haskell 2396 2397 > import Test.QuickCheck 2398 > 2399 > digits :: Int -> [Char] -> [Char] 2400 > digits 0 acc = '0' : acc 2401 > digits 1 acc = '1' : acc 2402 > digits n acc = digits (n `div` 2) $ digits (n `mod` 2) acc 2403 > 2404 > dist :: Int -> [Char] -> [Char] 2405 > dist 0 [] = ['S'] 2406 > dist 0 acc = acc 2407 > dist 1 acc = let r = dist 0 acc in 's' : digits (length r) r 2408 > dist n acc = dist (n - 1) $ dist 1 acc 2409 > 2410 > takeLast n ss = reverse $ take n $ reverse ss 2411 > 2412 > test = takeLast 40 $ dist 20 [] 2413 > 2414 2415 Printing <test> gives: ``"1s100000s11010s10100s1111s1010s110s11s1S"`` 2416 2417 The reverse algorithm computes the length of the string just by examining a 2418 certain prefix: 2419 2420 .. code-block:: haskell 2421 2422 > pref :: [Char] -> Int 2423 > pref "S" = 1 2424 > pref ('s':'1':rest) = decode 2 1 rest 2425 > pref (_:rest) = 1 + pref rest 2426 > 2427 > decode walk acc ('0':rest) = decode (walk + 1) (acc * 2) rest 2428 > decode walk acc ('1':rest) = decode (walk + 1) (acc * 2 + 1) rest 2429 > decode walk acc _ = walk + acc 2430 > 2431 2432 Now, as expected, printing <pref test> gives ``40``. 2433 2434 We can *quickCheck* this with following property: 2435 2436 .. code-block:: haskell 2437 2438 > testcase = dist 2000 [] 2439 > testcaseLength = length testcase 2440 > 2441 > identityProp n = n > 0 && n <= testcaseLength ==> length arr == pref arr 2442 > where arr = takeLast n testcase 2443 > 2444 2445 As expected <quickCheck identityProp> gives: 2446 2447 :: 2448 2449 *Main> quickCheck identityProp 2450 OK, passed 100 tests. 2451 2452 Let's be a bit more exhaustive: 2453 2454 .. code-block:: haskell 2455 2456 > 2457 > deepCheck p = check (defaultConfig { configMaxTest = 500 }) p 2458 > 2459 2460 And here is the result of <deepCheck identityProp>: 2461 2462 :: 2463 2464 *Main> deepCheck identityProp 2465 OK, passed 500 tests. 2466 2467 .. _Tagging: 2468 2469 Tagging considerations 2470 ^^^^^^^^^^^^^^^^^^^^^^ 2471 2472 To maintain the invariant that the 2 LSBits of each ``Use**`` in ``Use`` never 2473 change after being set up, setters of ``Use::Prev`` must re-tag the new 2474 ``Use**`` on every modification. Accordingly getters must strip the tag bits. 2475 2476 For layout b) instead of the ``User`` we find a pointer (``User*`` with LSBit 2477 set). Following this pointer brings us to the ``User``. A portable trick 2478 ensures that the first bytes of ``User`` (if interpreted as a pointer) never has 2479 the LSBit set. (Portability is relying on the fact that all known compilers 2480 place the ``vptr`` in the first word of the instances.) 2481 2482 .. _coreclasses: 2483 2484 The Core LLVM Class Hierarchy Reference 2485 ======================================= 2486 2487 ``#include "llvm/IR/Type.h"`` 2488 2489 header source: `Type.h <http://llvm.org/doxygen/Type_8h-source.html>`_ 2490 2491 doxygen info: `Type Clases <http://llvm.org/doxygen/classllvm_1_1Type.html>`_ 2492 2493 The Core LLVM classes are the primary means of representing the program being 2494 inspected or transformed. The core LLVM classes are defined in header files in 2495 the ``include/llvm/`` directory, and implemented in the ``lib/VMCore`` 2496 directory. 2497 2498 .. _Type: 2499 2500 The Type class and Derived Types 2501 -------------------------------- 2502 2503 ``Type`` is a superclass of all type classes. Every ``Value`` has a ``Type``. 2504 ``Type`` cannot be instantiated directly but only through its subclasses. 2505 Certain primitive types (``VoidType``, ``LabelType``, ``FloatType`` and 2506 ``DoubleType``) have hidden subclasses. They are hidden because they offer no 2507 useful functionality beyond what the ``Type`` class offers except to distinguish 2508 themselves from other subclasses of ``Type``. 2509 2510 All other types are subclasses of ``DerivedType``. Types can be named, but this 2511 is not a requirement. There exists exactly one instance of a given shape at any 2512 one time. This allows type equality to be performed with address equality of 2513 the Type Instance. That is, given two ``Type*`` values, the types are identical 2514 if the pointers are identical. 2515 2516 .. _m_Type: 2517 2518 Important Public Methods 2519 ^^^^^^^^^^^^^^^^^^^^^^^^ 2520 2521 * ``bool isIntegerTy() const``: Returns true for any integer type. 2522 2523 * ``bool isFloatingPointTy()``: Return true if this is one of the five 2524 floating point types. 2525 2526 * ``bool isSized()``: Return true if the type has known size. Things 2527 that don't have a size are abstract types, labels and void. 2528 2529 .. _derivedtypes: 2530 2531 Important Derived Types 2532 ^^^^^^^^^^^^^^^^^^^^^^^ 2533 2534 ``IntegerType`` 2535 Subclass of DerivedType that represents integer types of any bit width. Any 2536 bit width between ``IntegerType::MIN_INT_BITS`` (1) and 2537 ``IntegerType::MAX_INT_BITS`` (~8 million) can be represented. 2538 2539 * ``static const IntegerType* get(unsigned NumBits)``: get an integer 2540 type of a specific bit width. 2541 2542 * ``unsigned getBitWidth() const``: Get the bit width of an integer type. 2543 2544 ``SequentialType`` 2545 This is subclassed by ArrayType, PointerType and VectorType. 2546 2547 * ``const Type * getElementType() const``: Returns the type of each 2548 of the elements in the sequential type. 2549 2550 ``ArrayType`` 2551 This is a subclass of SequentialType and defines the interface for array 2552 types. 2553 2554 * ``unsigned getNumElements() const``: Returns the number of elements 2555 in the array. 2556 2557 ``PointerType`` 2558 Subclass of SequentialType for pointer types. 2559 2560 ``VectorType`` 2561 Subclass of SequentialType for vector types. A vector type is similar to an 2562 ArrayType but is distinguished because it is a first class type whereas 2563 ArrayType is not. Vector types are used for vector operations and are usually 2564 small vectors of of an integer or floating point type. 2565 2566 ``StructType`` 2567 Subclass of DerivedTypes for struct types. 2568 2569 .. _FunctionType: 2570 2571 ``FunctionType`` 2572 Subclass of DerivedTypes for function types. 2573 2574 * ``bool isVarArg() const``: Returns true if it's a vararg function. 2575 2576 * ``const Type * getReturnType() const``: Returns the return type of the 2577 function. 2578 2579 * ``const Type * getParamType (unsigned i)``: Returns the type of the ith 2580 parameter. 2581 2582 * ``const unsigned getNumParams() const``: Returns the number of formal 2583 parameters. 2584 2585 .. _Module: 2586 2587 The ``Module`` class 2588 -------------------- 2589 2590 ``#include "llvm/IR/Module.h"`` 2591 2592 header source: `Module.h <http://llvm.org/doxygen/Module_8h-source.html>`_ 2593 2594 doxygen info: `Module Class <http://llvm.org/doxygen/classllvm_1_1Module.html>`_ 2595 2596 The ``Module`` class represents the top level structure present in LLVM 2597 programs. An LLVM module is effectively either a translation unit of the 2598 original program or a combination of several translation units merged by the 2599 linker. The ``Module`` class keeps track of a list of :ref:`Function 2600 <c_Function>`\ s, a list of GlobalVariable_\ s, and a SymbolTable_. 2601 Additionally, it contains a few helpful member functions that try to make common 2602 operations easy. 2603 2604 .. _m_Module: 2605 2606 Important Public Members of the ``Module`` class 2607 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 2608 2609 * ``Module::Module(std::string name = "")`` 2610 2611 Constructing a Module_ is easy. You can optionally provide a name for it 2612 (probably based on the name of the translation unit). 2613 2614 * | ``Module::iterator`` - Typedef for function list iterator 2615 | ``Module::const_iterator`` - Typedef for const_iterator. 2616 | ``begin()``, ``end()``, ``size()``, ``empty()`` 2617 2618 These are forwarding methods that make it easy to access the contents of a 2619 ``Module`` object's :ref:`Function <c_Function>` list. 2620 2621 * ``Module::FunctionListType &getFunctionList()`` 2622 2623 Returns the list of :ref:`Function <c_Function>`\ s. This is necessary to use 2624 when you need to update the list or perform a complex action that doesn't have 2625 a forwarding method. 2626 2627 ---------------- 2628 2629 * | ``Module::global_iterator`` - Typedef for global variable list iterator 2630 | ``Module::const_global_iterator`` - Typedef for const_iterator. 2631 | ``global_begin()``, ``global_end()``, ``global_size()``, ``global_empty()`` 2632 2633 These are forwarding methods that make it easy to access the contents of a 2634 ``Module`` object's GlobalVariable_ list. 2635 2636 * ``Module::GlobalListType &getGlobalList()`` 2637 2638 Returns the list of GlobalVariable_\ s. This is necessary to use when you 2639 need to update the list or perform a complex action that doesn't have a 2640 forwarding method. 2641 2642 ---------------- 2643 2644 * ``SymbolTable *getSymbolTable()`` 2645 2646 Return a reference to the SymbolTable_ for this ``Module``. 2647 2648 ---------------- 2649 2650 * ``Function *getFunction(StringRef Name) const`` 2651 2652 Look up the specified function in the ``Module`` SymbolTable_. If it does not 2653 exist, return ``null``. 2654 2655 * ``Function *getOrInsertFunction(const std::string &Name, const FunctionType 2656 *T)`` 2657 2658 Look up the specified function in the ``Module`` SymbolTable_. If it does not 2659 exist, add an external declaration for the function and return it. 2660 2661 * ``std::string getTypeName(const Type *Ty)`` 2662 2663 If there is at least one entry in the SymbolTable_ for the specified Type_, 2664 return it. Otherwise return the empty string. 2665 2666 * ``bool addTypeName(const std::string &Name, const Type *Ty)`` 2667 2668 Insert an entry in the SymbolTable_ mapping ``Name`` to ``Ty``. If there is 2669 already an entry for this name, true is returned and the SymbolTable_ is not 2670 modified. 2671 2672 .. _Value: 2673 2674 The ``Value`` class 2675 ------------------- 2676 2677 ``#include "llvm/IR/Value.h"`` 2678 2679 header source: `Value.h <http://llvm.org/doxygen/Value_8h-source.html>`_ 2680 2681 doxygen info: `Value Class <http://llvm.org/doxygen/classllvm_1_1Value.html>`_ 2682 2683 The ``Value`` class is the most important class in the LLVM Source base. It 2684 represents a typed value that may be used (among other things) as an operand to 2685 an instruction. There are many different types of ``Value``\ s, such as 2686 Constant_\ s, Argument_\ s. Even Instruction_\ s and :ref:`Function 2687 <c_Function>`\ s are ``Value``\ s. 2688 2689 A particular ``Value`` may be used many times in the LLVM representation for a 2690 program. For example, an incoming argument to a function (represented with an 2691 instance of the Argument_ class) is "used" by every instruction in the function 2692 that references the argument. To keep track of this relationship, the ``Value`` 2693 class keeps a list of all of the ``User``\ s that is using it (the User_ class 2694 is a base class for all nodes in the LLVM graph that can refer to ``Value``\ s). 2695 This use list is how LLVM represents def-use information in the program, and is 2696 accessible through the ``use_*`` methods, shown below. 2697 2698 Because LLVM is a typed representation, every LLVM ``Value`` is typed, and this 2699 Type_ is available through the ``getType()`` method. In addition, all LLVM 2700 values can be named. The "name" of the ``Value`` is a symbolic string printed 2701 in the LLVM code: 2702 2703 .. code-block:: llvm 2704 2705 %foo = add i32 1, 2 2706 2707 .. _nameWarning: 2708 2709 The name of this instruction is "foo". **NOTE** that the name of any value may 2710 be missing (an empty string), so names should **ONLY** be used for debugging 2711 (making the source code easier to read, debugging printouts), they should not be 2712 used to keep track of values or map between them. For this purpose, use a 2713 ``std::map`` of pointers to the ``Value`` itself instead. 2714 2715 One important aspect of LLVM is that there is no distinction between an SSA 2716 variable and the operation that produces it. Because of this, any reference to 2717 the value produced by an instruction (or the value available as an incoming 2718 argument, for example) is represented as a direct pointer to the instance of the 2719 class that represents this value. Although this may take some getting used to, 2720 it simplifies the representation and makes it easier to manipulate. 2721 2722 .. _m_Value: 2723 2724 Important Public Members of the ``Value`` class 2725 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 2726 2727 * | ``Value::use_iterator`` - Typedef for iterator over the use-list 2728 | ``Value::const_use_iterator`` - Typedef for const_iterator over the 2729 use-list 2730 | ``unsigned use_size()`` - Returns the number of users of the value. 2731 | ``bool use_empty()`` - Returns true if there are no users. 2732 | ``use_iterator use_begin()`` - Get an iterator to the start of the 2733 use-list. 2734 | ``use_iterator use_end()`` - Get an iterator to the end of the use-list. 2735 | ``User *use_back()`` - Returns the last element in the list. 2736 2737 These methods are the interface to access the def-use information in LLVM. 2738 As with all other iterators in LLVM, the naming conventions follow the 2739 conventions defined by the STL_. 2740 2741 * ``Type *getType() const`` 2742 This method returns the Type of the Value. 2743 2744 * | ``bool hasName() const`` 2745 | ``std::string getName() const`` 2746 | ``void setName(const std::string &Name)`` 2747 2748 This family of methods is used to access and assign a name to a ``Value``, be 2749 aware of the :ref:`precaution above <nameWarning>`. 2750 2751 * ``void replaceAllUsesWith(Value *V)`` 2752 2753 This method traverses the use list of a ``Value`` changing all User_\ s of the 2754 current value to refer to "``V``" instead. For example, if you detect that an 2755 instruction always produces a constant value (for example through constant 2756 folding), you can replace all uses of the instruction with the constant like 2757 this: 2758 2759 .. code-block:: c++ 2760 2761 Inst->replaceAllUsesWith(ConstVal); 2762 2763 .. _User: 2764 2765 The ``User`` class 2766 ------------------ 2767 2768 ``#include "llvm/IR/User.h"`` 2769 2770 header source: `User.h <http://llvm.org/doxygen/User_8h-source.html>`_ 2771 2772 doxygen info: `User Class <http://llvm.org/doxygen/classllvm_1_1User.html>`_ 2773 2774 Superclass: Value_ 2775 2776 The ``User`` class is the common base class of all LLVM nodes that may refer to 2777 ``Value``\ s. It exposes a list of "Operands" that are all of the ``Value``\ s 2778 that the User is referring to. The ``User`` class itself is a subclass of 2779 ``Value``. 2780 2781 The operands of a ``User`` point directly to the LLVM ``Value`` that it refers 2782 to. Because LLVM uses Static Single Assignment (SSA) form, there can only be 2783 one definition referred to, allowing this direct connection. This connection 2784 provides the use-def information in LLVM. 2785 2786 .. _m_User: 2787 2788 Important Public Members of the ``User`` class 2789 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 2790 2791 The ``User`` class exposes the operand list in two ways: through an index access 2792 interface and through an iterator based interface. 2793 2794 * | ``Value *getOperand(unsigned i)`` 2795 | ``unsigned getNumOperands()`` 2796 2797 These two methods expose the operands of the ``User`` in a convenient form for 2798 direct access. 2799 2800 * | ``User::op_iterator`` - Typedef for iterator over the operand list 2801 | ``op_iterator op_begin()`` - Get an iterator to the start of the operand 2802 list. 2803 | ``op_iterator op_end()`` - Get an iterator to the end of the operand list. 2804 2805 Together, these methods make up the iterator based interface to the operands 2806 of a ``User``. 2807 2808 2809 .. _Instruction: 2810 2811 The ``Instruction`` class 2812 ------------------------- 2813 2814 ``#include "llvm/IR/Instruction.h"`` 2815 2816 header source: `Instruction.h 2817 <http://llvm.org/doxygen/Instruction_8h-source.html>`_ 2818 2819 doxygen info: `Instruction Class 2820 <http://llvm.org/doxygen/classllvm_1_1Instruction.html>`_ 2821 2822 Superclasses: User_, Value_ 2823 2824 The ``Instruction`` class is the common base class for all LLVM instructions. 2825 It provides only a few methods, but is a very commonly used class. The primary 2826 data tracked by the ``Instruction`` class itself is the opcode (instruction 2827 type) and the parent BasicBlock_ the ``Instruction`` is embedded into. To 2828 represent a specific type of instruction, one of many subclasses of 2829 ``Instruction`` are used. 2830 2831 Because the ``Instruction`` class subclasses the User_ class, its operands can 2832 be accessed in the same way as for other ``User``\ s (with the 2833 ``getOperand()``/``getNumOperands()`` and ``op_begin()``/``op_end()`` methods). 2834 An important file for the ``Instruction`` class is the ``llvm/Instruction.def`` 2835 file. This file contains some meta-data about the various different types of 2836 instructions in LLVM. It describes the enum values that are used as opcodes 2837 (for example ``Instruction::Add`` and ``Instruction::ICmp``), as well as the 2838 concrete sub-classes of ``Instruction`` that implement the instruction (for 2839 example BinaryOperator_ and CmpInst_). Unfortunately, the use of macros in this 2840 file confuses doxygen, so these enum values don't show up correctly in the 2841 `doxygen output <http://llvm.org/doxygen/classllvm_1_1Instruction.html>`_. 2842 2843 .. _s_Instruction: 2844 2845 Important Subclasses of the ``Instruction`` class 2846 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 2847 2848 .. _BinaryOperator: 2849 2850 * ``BinaryOperator`` 2851 2852 This subclasses represents all two operand instructions whose operands must be 2853 the same type, except for the comparison instructions. 2854 2855 .. _CastInst: 2856 2857 * ``CastInst`` 2858 This subclass is the parent of the 12 casting instructions. It provides 2859 common operations on cast instructions. 2860 2861 .. _CmpInst: 2862 2863 * ``CmpInst`` 2864 2865 This subclass respresents the two comparison instructions, 2866 `ICmpInst <LangRef.html#i_icmp>`_ (integer opreands), and 2867 `FCmpInst <LangRef.html#i_fcmp>`_ (floating point operands). 2868 2869 .. _TerminatorInst: 2870 2871 * ``TerminatorInst`` 2872 2873 This subclass is the parent of all terminator instructions (those which can 2874 terminate a block). 2875 2876 .. _m_Instruction: 2877 2878 Important Public Members of the ``Instruction`` class 2879 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 2880 2881 * ``BasicBlock *getParent()`` 2882 2883 Returns the BasicBlock_ that this 2884 ``Instruction`` is embedded into. 2885 2886 * ``bool mayWriteToMemory()`` 2887 2888 Returns true if the instruction writes to memory, i.e. it is a ``call``, 2889 ``free``, ``invoke``, or ``store``. 2890 2891 * ``unsigned getOpcode()`` 2892 2893 Returns the opcode for the ``Instruction``. 2894 2895 * ``Instruction *clone() const`` 2896 2897 Returns another instance of the specified instruction, identical in all ways 2898 to the original except that the instruction has no parent (i.e. it's not 2899 embedded into a BasicBlock_), and it has no name. 2900 2901 .. _Constant: 2902 2903 The ``Constant`` class and subclasses 2904 ------------------------------------- 2905 2906 Constant represents a base class for different types of constants. It is 2907 subclassed by ConstantInt, ConstantArray, etc. for representing the various 2908 types of Constants. GlobalValue_ is also a subclass, which represents the 2909 address of a global variable or function. 2910 2911 .. _s_Constant: 2912 2913 Important Subclasses of Constant 2914 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 2915 2916 * ConstantInt : This subclass of Constant represents an integer constant of 2917 any width. 2918 2919 * ``const APInt& getValue() const``: Returns the underlying 2920 value of this constant, an APInt value. 2921 2922 * ``int64_t getSExtValue() const``: Converts the underlying APInt value to an 2923 int64_t via sign extension. If the value (not the bit width) of the APInt 2924 is too large to fit in an int64_t, an assertion will result. For this 2925 reason, use of this method is discouraged. 2926 2927 * ``uint64_t getZExtValue() const``: Converts the underlying APInt value 2928 to a uint64_t via zero extension. IF the value (not the bit width) of the 2929 APInt is too large to fit in a uint64_t, an assertion will result. For this 2930 reason, use of this method is discouraged. 2931 2932 * ``static ConstantInt* get(const APInt& Val)``: Returns the ConstantInt 2933 object that represents the value provided by ``Val``. The type is implied 2934 as the IntegerType that corresponds to the bit width of ``Val``. 2935 2936 * ``static ConstantInt* get(const Type *Ty, uint64_t Val)``: Returns the 2937 ConstantInt object that represents the value provided by ``Val`` for integer 2938 type ``Ty``. 2939 2940 * ConstantFP : This class represents a floating point constant. 2941 2942 * ``double getValue() const``: Returns the underlying value of this constant. 2943 2944 * ConstantArray : This represents a constant array. 2945 2946 * ``const std::vector<Use> &getValues() const``: Returns a vector of 2947 component constants that makeup this array. 2948 2949 * ConstantStruct : This represents a constant struct. 2950 2951 * ``const std::vector<Use> &getValues() const``: Returns a vector of 2952 component constants that makeup this array. 2953 2954 * GlobalValue : This represents either a global variable or a function. In 2955 either case, the value is a constant fixed address (after linking). 2956 2957 .. _GlobalValue: 2958 2959 The ``GlobalValue`` class 2960 ------------------------- 2961 2962 ``#include "llvm/IR/GlobalValue.h"`` 2963 2964 header source: `GlobalValue.h 2965 <http://llvm.org/doxygen/GlobalValue_8h-source.html>`_ 2966 2967 doxygen info: `GlobalValue Class 2968 <http://llvm.org/doxygen/classllvm_1_1GlobalValue.html>`_ 2969 2970 Superclasses: Constant_, User_, Value_ 2971 2972 Global values ( GlobalVariable_\ s or :ref:`Function <c_Function>`\ s) are the 2973 only LLVM values that are visible in the bodies of all :ref:`Function 2974 <c_Function>`\ s. Because they are visible at global scope, they are also 2975 subject to linking with other globals defined in different translation units. 2976 To control the linking process, ``GlobalValue``\ s know their linkage rules. 2977 Specifically, ``GlobalValue``\ s know whether they have internal or external 2978 linkage, as defined by the ``LinkageTypes`` enumeration. 2979 2980 If a ``GlobalValue`` has internal linkage (equivalent to being ``static`` in C), 2981 it is not visible to code outside the current translation unit, and does not 2982 participate in linking. If it has external linkage, it is visible to external 2983 code, and does participate in linking. In addition to linkage information, 2984 ``GlobalValue``\ s keep track of which Module_ they are currently part of. 2985 2986 Because ``GlobalValue``\ s are memory objects, they are always referred to by 2987 their **address**. As such, the Type_ of a global is always a pointer to its 2988 contents. It is important to remember this when using the ``GetElementPtrInst`` 2989 instruction because this pointer must be dereferenced first. For example, if 2990 you have a ``GlobalVariable`` (a subclass of ``GlobalValue)`` that is an array 2991 of 24 ints, type ``[24 x i32]``, then the ``GlobalVariable`` is a pointer to 2992 that array. Although the address of the first element of this array and the 2993 value of the ``GlobalVariable`` are the same, they have different types. The 2994 ``GlobalVariable``'s type is ``[24 x i32]``. The first element's type is 2995 ``i32.`` Because of this, accessing a global value requires you to dereference 2996 the pointer with ``GetElementPtrInst`` first, then its elements can be accessed. 2997 This is explained in the `LLVM Language Reference Manual 2998 <LangRef.html#globalvars>`_. 2999 3000 .. _m_GlobalValue: 3001 3002 Important Public Members of the ``GlobalValue`` class 3003 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 3004 3005 * | ``bool hasInternalLinkage() const`` 3006 | ``bool hasExternalLinkage() const`` 3007 | ``void setInternalLinkage(bool HasInternalLinkage)`` 3008 3009 These methods manipulate the linkage characteristics of the ``GlobalValue``. 3010 3011 * ``Module *getParent()`` 3012 3013 This returns the Module_ that the 3014 GlobalValue is currently embedded into. 3015 3016 .. _c_Function: 3017 3018 The ``Function`` class 3019 ---------------------- 3020 3021 ``#include "llvm/IR/Function.h"`` 3022 3023 header source: `Function.h <http://llvm.org/doxygen/Function_8h-source.html>`_ 3024 3025 doxygen info: `Function Class 3026 <http://llvm.org/doxygen/classllvm_1_1Function.html>`_ 3027 3028 Superclasses: GlobalValue_, Constant_, User_, Value_ 3029 3030 The ``Function`` class represents a single procedure in LLVM. It is actually 3031 one of the more complex classes in the LLVM hierarchy because it must keep track 3032 of a large amount of data. The ``Function`` class keeps track of a list of 3033 BasicBlock_\ s, a list of formal Argument_\ s, and a SymbolTable_. 3034 3035 The list of BasicBlock_\ s is the most commonly used part of ``Function`` 3036 objects. The list imposes an implicit ordering of the blocks in the function, 3037 which indicate how the code will be laid out by the backend. Additionally, the 3038 first BasicBlock_ is the implicit entry node for the ``Function``. It is not 3039 legal in LLVM to explicitly branch to this initial block. There are no implicit 3040 exit nodes, and in fact there may be multiple exit nodes from a single 3041 ``Function``. If the BasicBlock_ list is empty, this indicates that the 3042 ``Function`` is actually a function declaration: the actual body of the function 3043 hasn't been linked in yet. 3044 3045 In addition to a list of BasicBlock_\ s, the ``Function`` class also keeps track 3046 of the list of formal Argument_\ s that the function receives. This container 3047 manages the lifetime of the Argument_ nodes, just like the BasicBlock_ list does 3048 for the BasicBlock_\ s. 3049 3050 The SymbolTable_ is a very rarely used LLVM feature that is only used when you 3051 have to look up a value by name. Aside from that, the SymbolTable_ is used 3052 internally to make sure that there are not conflicts between the names of 3053 Instruction_\ s, BasicBlock_\ s, or Argument_\ s in the function body. 3054 3055 Note that ``Function`` is a GlobalValue_ and therefore also a Constant_. The 3056 value of the function is its address (after linking) which is guaranteed to be 3057 constant. 3058 3059 .. _m_Function: 3060 3061 Important Public Members of the ``Function`` 3062 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 3063 3064 * ``Function(const FunctionType *Ty, LinkageTypes Linkage, 3065 const std::string &N = "", Module* Parent = 0)`` 3066 3067 Constructor used when you need to create new ``Function``\ s to add the 3068 program. The constructor must specify the type of the function to create and 3069 what type of linkage the function should have. The FunctionType_ argument 3070 specifies the formal arguments and return value for the function. The same 3071 FunctionType_ value can be used to create multiple functions. The ``Parent`` 3072 argument specifies the Module in which the function is defined. If this 3073 argument is provided, the function will automatically be inserted into that 3074 module's list of functions. 3075 3076 * ``bool isDeclaration()`` 3077 3078 Return whether or not the ``Function`` has a body defined. If the function is 3079 "external", it does not have a body, and thus must be resolved by linking with 3080 a function defined in a different translation unit. 3081 3082 * | ``Function::iterator`` - Typedef for basic block list iterator 3083 | ``Function::const_iterator`` - Typedef for const_iterator. 3084 | ``begin()``, ``end()``, ``size()``, ``empty()`` 3085 3086 These are forwarding methods that make it easy to access the contents of a 3087 ``Function`` object's BasicBlock_ list. 3088 3089 * ``Function::BasicBlockListType &getBasicBlockList()`` 3090 3091 Returns the list of BasicBlock_\ s. This is necessary to use when you need to 3092 update the list or perform a complex action that doesn't have a forwarding 3093 method. 3094 3095 * | ``Function::arg_iterator`` - Typedef for the argument list iterator 3096 | ``Function::const_arg_iterator`` - Typedef for const_iterator. 3097 | ``arg_begin()``, ``arg_end()``, ``arg_size()``, ``arg_empty()`` 3098 3099 These are forwarding methods that make it easy to access the contents of a 3100 ``Function`` object's Argument_ list. 3101 3102 * ``Function::ArgumentListType &getArgumentList()`` 3103 3104 Returns the list of Argument_. This is necessary to use when you need to 3105 update the list or perform a complex action that doesn't have a forwarding 3106 method. 3107 3108 * ``BasicBlock &getEntryBlock()`` 3109 3110 Returns the entry ``BasicBlock`` for the function. Because the entry block 3111 for the function is always the first block, this returns the first block of 3112 the ``Function``. 3113 3114 * | ``Type *getReturnType()`` 3115 | ``FunctionType *getFunctionType()`` 3116 3117 This traverses the Type_ of the ``Function`` and returns the return type of 3118 the function, or the FunctionType_ of the actual function. 3119 3120 * ``SymbolTable *getSymbolTable()`` 3121 3122 Return a pointer to the SymbolTable_ for this ``Function``. 3123 3124 .. _GlobalVariable: 3125 3126 The ``GlobalVariable`` class 3127 ---------------------------- 3128 3129 ``#include "llvm/IR/GlobalVariable.h"`` 3130 3131 header source: `GlobalVariable.h 3132 <http://llvm.org/doxygen/GlobalVariable_8h-source.html>`_ 3133 3134 doxygen info: `GlobalVariable Class 3135 <http://llvm.org/doxygen/classllvm_1_1GlobalVariable.html>`_ 3136 3137 Superclasses: GlobalValue_, Constant_, User_, Value_ 3138 3139 Global variables are represented with the (surprise surprise) ``GlobalVariable`` 3140 class. Like functions, ``GlobalVariable``\ s are also subclasses of 3141 GlobalValue_, and as such are always referenced by their address (global values 3142 must live in memory, so their "name" refers to their constant address). See 3143 GlobalValue_ for more on this. Global variables may have an initial value 3144 (which must be a Constant_), and if they have an initializer, they may be marked 3145 as "constant" themselves (indicating that their contents never change at 3146 runtime). 3147 3148 .. _m_GlobalVariable: 3149 3150 Important Public Members of the ``GlobalVariable`` class 3151 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 3152 3153 * ``GlobalVariable(const Type *Ty, bool isConstant, LinkageTypes &Linkage, 3154 Constant *Initializer = 0, const std::string &Name = "", Module* Parent = 0)`` 3155 3156 Create a new global variable of the specified type. If ``isConstant`` is true 3157 then the global variable will be marked as unchanging for the program. The 3158 Linkage parameter specifies the type of linkage (internal, external, weak, 3159 linkonce, appending) for the variable. If the linkage is InternalLinkage, 3160 WeakAnyLinkage, WeakODRLinkage, LinkOnceAnyLinkage or LinkOnceODRLinkage, then 3161 the resultant global variable will have internal linkage. AppendingLinkage 3162 concatenates together all instances (in different translation units) of the 3163 variable into a single variable but is only applicable to arrays. See the 3164 `LLVM Language Reference <LangRef.html#modulestructure>`_ for further details 3165 on linkage types. Optionally an initializer, a name, and the module to put 3166 the variable into may be specified for the global variable as well. 3167 3168 * ``bool isConstant() const`` 3169 3170 Returns true if this is a global variable that is known not to be modified at 3171 runtime. 3172 3173 * ``bool hasInitializer()`` 3174 3175 Returns true if this ``GlobalVariable`` has an intializer. 3176 3177 * ``Constant *getInitializer()`` 3178 3179 Returns the initial value for a ``GlobalVariable``. It is not legal to call 3180 this method if there is no initializer. 3181 3182 .. _BasicBlock: 3183 3184 The ``BasicBlock`` class 3185 ------------------------ 3186 3187 ``#include "llvm/IR/BasicBlock.h"`` 3188 3189 header source: `BasicBlock.h 3190 <http://llvm.org/doxygen/BasicBlock_8h-source.html>`_ 3191 3192 doxygen info: `BasicBlock Class 3193 <http://llvm.org/doxygen/classllvm_1_1BasicBlock.html>`_ 3194 3195 Superclass: Value_ 3196 3197 This class represents a single entry single exit section of the code, commonly 3198 known as a basic block by the compiler community. The ``BasicBlock`` class 3199 maintains a list of Instruction_\ s, which form the body of the block. Matching 3200 the language definition, the last element of this list of instructions is always 3201 a terminator instruction (a subclass of the TerminatorInst_ class). 3202 3203 In addition to tracking the list of instructions that make up the block, the 3204 ``BasicBlock`` class also keeps track of the :ref:`Function <c_Function>` that 3205 it is embedded into. 3206 3207 Note that ``BasicBlock``\ s themselves are Value_\ s, because they are 3208 referenced by instructions like branches and can go in the switch tables. 3209 ``BasicBlock``\ s have type ``label``. 3210 3211 .. _m_BasicBlock: 3212 3213 Important Public Members of the ``BasicBlock`` class 3214 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ 3215 3216 * ``BasicBlock(const std::string &Name = "", Function *Parent = 0)`` 3217 3218 The ``BasicBlock`` constructor is used to create new basic blocks for 3219 insertion into a function. The constructor optionally takes a name for the 3220 new block, and a :ref:`Function <c_Function>` to insert it into. If the 3221 ``Parent`` parameter is specified, the new ``BasicBlock`` is automatically 3222 inserted at the end of the specified :ref:`Function <c_Function>`, if not 3223 specified, the BasicBlock must be manually inserted into the :ref:`Function 3224 <c_Function>`. 3225 3226 * | ``BasicBlock::iterator`` - Typedef for instruction list iterator 3227 | ``BasicBlock::const_iterator`` - Typedef for const_iterator. 3228 | ``begin()``, ``end()``, ``front()``, ``back()``, 3229 ``size()``, ``empty()`` 3230 STL-style functions for accessing the instruction list. 3231 3232 These methods and typedefs are forwarding functions that have the same 3233 semantics as the standard library methods of the same names. These methods 3234 expose the underlying instruction list of a basic block in a way that is easy 3235 to manipulate. To get the full complement of container operations (including 3236 operations to update the list), you must use the ``getInstList()`` method. 3237 3238 * ``BasicBlock::InstListType &getInstList()`` 3239 3240 This method is used to get access to the underlying container that actually 3241 holds the Instructions. This method must be used when there isn't a 3242 forwarding function in the ``BasicBlock`` class for the operation that you 3243 would like to perform. Because there are no forwarding functions for 3244 "updating" operations, you need to use this if you want to update the contents 3245 of a ``BasicBlock``. 3246 3247 * ``Function *getParent()`` 3248 3249 Returns a pointer to :ref:`Function <c_Function>` the block is embedded into, 3250 or a null pointer if it is homeless. 3251 3252 * ``TerminatorInst *getTerminator()`` 3253 3254 Returns a pointer to the terminator instruction that appears at the end of the 3255 ``BasicBlock``. If there is no terminator instruction, or if the last 3256 instruction in the block is not a terminator, then a null pointer is returned. 3257 3258 .. _Argument: 3259 3260 The ``Argument`` class 3261 ---------------------- 3262 3263 This subclass of Value defines the interface for incoming formal arguments to a 3264 function. A Function maintains a list of its formal arguments. An argument has 3265 a pointer to the parent Function. 3266 3267 3268