1 ========================
2 LLVM Programmer's Manual
3 ========================
9 This is always a work in progress.
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.
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
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.
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.
47 The C++ Standard Template Library
48 ---------------------------------
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.
56 Here are some useful links:
59 <http://en.cppreference.com/w/>`_ - an excellent
60 reference for the STL and other parts of the standard C++ library.
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
67 #. `C++ Frequently Asked Questions <http://www.parashift.com/c++-faq-lite/>`_.
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>`_.
73 #. `Bjarne Stroustrup's C++ Page
74 <http://www.research.att.com/%7Ebs/C++.html>`_.
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>`_.
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.
86 Other useful references
87 -----------------------
89 #. `Using static and shared libraries across platforms
90 <http://www.fortran-2000.com/ArnaudRecipes/sharedlib.html>`_
94 Important and useful LLVM APIs
95 ==============================
97 Here we highlight some LLVM APIs that are generally useful and good to know
98 about when writing transformations.
102 The ``isa<>``, ``cast<>`` and ``dyn_cast<>`` templates
103 ------------------------------------------------------
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).
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).
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<>``
130 static bool isLoopInvariant(const Value *V, const Loop *L) {
131 if (isa<Constant>(V) || isa<Argument>(V) || isa<GlobalValue>(V))
134 // Otherwise, it must be an instruction...
135 return !L->contains(cast<Instruction>(V)->getParent());
138 Note that you should **not** use an ``isa<>`` test followed by a ``cast<>``,
139 for that use the ``dyn_cast<>`` operator.
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:
152 if (AllocationInst *AI = dyn_cast<AllocationInst>(Val)) {
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
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
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.
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.
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>`
186 Passing strings (the ``StringRef`` and ``Twine`` classes)
187 ---------------------------------------------------------
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.
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.
202 The ``StringRef`` class
203 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
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.
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:
215 iterator find(StringRef Key);
217 and clients can call it using any one of:
221 Map.find("foo"); // Lookup "foo"
222 Map.find(std::string("bar")); // Lookup "bar"
223 Map.find(StringRef("\0baz", 4)); // Lookup "\0baz"
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
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
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:
247 New = CmpInst::Create(..., SO->getName() + ".cmp");
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.
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
268 The ``DEBUG()`` macro and ``-debug`` option
269 -------------------------------------------
271 Often when working on your pass you will put a bunch of debugging printouts and
272 other code into your pass. After you get it working, you want to remove it, but
273 you may need it again in the future (to work out new bugs that you run across).
275 Naturally, because of this, you don't want to delete the debug printouts, but
276 you don't want them to always be noisy. A standard compromise is to comment
277 them out, allowing you to enable them if you need them in the future.
279 The ``llvm/Support/Debug.h`` (`doxygen
280 <http://llvm.org/doxygen/Debug_8h-source.html>`__) file provides a macro named
281 ``DEBUG()`` that is a much nicer solution to this problem. Basically, you can
282 put arbitrary code into the argument of the ``DEBUG`` macro, and it is only
283 executed if '``opt``' (or any other tool) is run with the '``-debug``' command
288 DEBUG(errs() << "I am here!\n");
290 Then you can run your pass like this:
294 $ opt < a.bc > /dev/null -mypass
296 $ opt < a.bc > /dev/null -mypass -debug
299 Using the ``DEBUG()`` macro instead of a home-brewed solution allows you to not
300 have to create "yet another" command line option for the debug output for your
301 pass. Note that ``DEBUG()`` macros are disabled for optimized builds, so they
302 do not cause a performance impact at all (for the same reason, they should also
303 not contain side-effects!).
305 One additional nice thing about the ``DEBUG()`` macro is that you can enable or
306 disable it directly in gdb. Just use "``set DebugFlag=0``" or "``set
307 DebugFlag=1``" from the gdb if the program is running. If the program hasn't
308 been started yet, you can always just run it with ``-debug``.
312 Fine grained debug info with ``DEBUG_TYPE`` and the ``-debug-only`` option
313 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
315 Sometimes you may find yourself in a situation where enabling ``-debug`` just
316 turns on **too much** information (such as when working on the code generator).
317 If you want to enable debug information with more fine-grained control, you
318 define the ``DEBUG_TYPE`` macro and the ``-debug`` only option as follows:
323 DEBUG(errs() << "No debug type\n");
324 #define DEBUG_TYPE "foo"
325 DEBUG(errs() << "'foo' debug type\n");
327 #define DEBUG_TYPE "bar"
328 DEBUG(errs() << "'bar' debug type\n"));
330 #define DEBUG_TYPE ""
331 DEBUG(errs() << "No debug type (2)\n");
333 Then you can run your pass like this:
337 $ opt < a.bc > /dev/null -mypass
339 $ opt < a.bc > /dev/null -mypass -debug
344 $ opt < a.bc > /dev/null -mypass -debug-only=foo
346 $ opt < a.bc > /dev/null -mypass -debug-only=bar
349 Of course, in practice, you should only set ``DEBUG_TYPE`` at the top of a file,
350 to specify the debug type for the entire module (if you do this before you
351 ``#include "llvm/Support/Debug.h"``, you don't have to insert the ugly
352 ``#undef``'s). Also, you should use names more meaningful than "foo" and "bar",
353 because there is no system in place to ensure that names do not conflict. If
354 two different modules use the same string, they will all be turned on when the
355 name is specified. This allows, for example, all debug information for
356 instruction scheduling to be enabled with ``-debug-type=InstrSched``, even if
357 the source lives in multiple files.
359 The ``DEBUG_WITH_TYPE`` macro is also available for situations where you would
360 like to set ``DEBUG_TYPE``, but only for one specific ``DEBUG`` statement. It
361 takes an additional first parameter, which is the type to use. For example, the
362 preceding example could be written as:
366 DEBUG_WITH_TYPE("", errs() << "No debug type\n");
367 DEBUG_WITH_TYPE("foo", errs() << "'foo' debug type\n");
368 DEBUG_WITH_TYPE("bar", errs() << "'bar' debug type\n"));
369 DEBUG_WITH_TYPE("", errs() << "No debug type (2)\n");
373 The ``Statistic`` class & ``-stats`` option
374 -------------------------------------------
376 The ``llvm/ADT/Statistic.h`` (`doxygen
377 <http://llvm.org/doxygen/Statistic_8h-source.html>`__) file provides a class
378 named ``Statistic`` that is used as a unified way to keep track of what the LLVM
379 compiler is doing and how effective various optimizations are. It is useful to
380 see what optimizations are contributing to making a particular program run
383 Often you may run your pass on some big program, and you're interested to see
384 how many times it makes a certain transformation. Although you can do this with
385 hand inspection, or some ad-hoc method, this is a real pain and not very useful
386 for big programs. Using the ``Statistic`` class makes it very easy to keep
387 track of this information, and the calculated information is presented in a
388 uniform manner with the rest of the passes being executed.
390 There are many examples of ``Statistic`` uses, but the basics of using it are as
393 #. Define your statistic like this:
397 #define DEBUG_TYPE "mypassname" // This goes before any #includes.
398 STATISTIC(NumXForms, "The # of times I did stuff");
400 The ``STATISTIC`` macro defines a static variable, whose name is specified by
401 the first argument. The pass name is taken from the ``DEBUG_TYPE`` macro, and
402 the description is taken from the second argument. The variable defined
403 ("NumXForms" in this case) acts like an unsigned integer.
405 #. Whenever you make a transformation, bump the counter:
409 ++NumXForms; // I did stuff!
411 That's all you have to do. To get '``opt``' to print out the statistics
412 gathered, use the '``-stats``' option:
416 $ opt -stats -mypassname < program.bc > /dev/null
417 ... statistics output ...
419 When running ``opt`` on a C file from the SPEC benchmark suite, it gives a
420 report that looks like this:
424 7646 bitcodewriter - Number of normal instructions
425 725 bitcodewriter - Number of oversized instructions
426 129996 bitcodewriter - Number of bitcode bytes written
427 2817 raise - Number of insts DCEd or constprop'd
428 3213 raise - Number of cast-of-self removed
429 5046 raise - Number of expression trees converted
430 75 raise - Number of other getelementptr's formed
431 138 raise - Number of load/store peepholes
432 42 deadtypeelim - Number of unused typenames removed from symtab
433 392 funcresolve - Number of varargs functions resolved
434 27 globaldce - Number of global variables removed
435 2 adce - Number of basic blocks removed
436 134 cee - Number of branches revectored
437 49 cee - Number of setcc instruction eliminated
438 532 gcse - Number of loads removed
439 2919 gcse - Number of instructions removed
440 86 indvars - Number of canonical indvars added
441 87 indvars - Number of aux indvars removed
442 25 instcombine - Number of dead inst eliminate
443 434 instcombine - Number of insts combined
444 248 licm - Number of load insts hoisted
445 1298 licm - Number of insts hoisted to a loop pre-header
446 3 licm - Number of insts hoisted to multiple loop preds (bad, no loop pre-header)
447 75 mem2reg - Number of alloca's promoted
448 1444 cfgsimplify - Number of blocks simplified
450 Obviously, with so many optimizations, having a unified framework for this stuff
451 is very nice. Making your pass fit well into the framework makes it more
452 maintainable and useful.
456 Viewing graphs while debugging code
457 -----------------------------------
459 Several of the important data structures in LLVM are graphs: for example CFGs
460 made out of LLVM :ref:`BasicBlocks <BasicBlock>`, CFGs made out of LLVM
461 :ref:`MachineBasicBlocks <MachineBasicBlock>`, and :ref:`Instruction Selection
462 DAGs <SelectionDAG>`. In many cases, while debugging various parts of the
463 compiler, it is nice to instantly visualize these graphs.
465 LLVM provides several callbacks that are available in a debug build to do
466 exactly that. If you call the ``Function::viewCFG()`` method, for example, the
467 current LLVM tool will pop up a window containing the CFG for the function where
468 each basic block is a node in the graph, and each node contains the instructions
469 in the block. Similarly, there also exists ``Function::viewCFGOnly()`` (does
470 not include the instructions), the ``MachineFunction::viewCFG()`` and
471 ``MachineFunction::viewCFGOnly()``, and the ``SelectionDAG::viewGraph()``
472 methods. Within GDB, for example, you can usually use something like ``call
473 DAG.viewGraph()`` to pop up a window. Alternatively, you can sprinkle calls to
474 these functions in your code in places you want to debug.
476 Getting this to work requires a small amount of configuration. On Unix systems
477 with X11, install the `graphviz <http://www.graphviz.org>`_ toolkit, and make
478 sure 'dot' and 'gv' are in your path. If you are running on Mac OS/X, download
479 and install the Mac OS/X `Graphviz program
480 <http://www.pixelglow.com/graphviz/>`_ and add
481 ``/Applications/Graphviz.app/Contents/MacOS/`` (or wherever you install it) to
482 your path. Once in your system and path are set up, rerun the LLVM configure
483 script and rebuild LLVM to enable this functionality.
485 ``SelectionDAG`` has been extended to make it easier to locate *interesting*
486 nodes in large complex graphs. From gdb, if you ``call DAG.setGraphColor(node,
487 "color")``, then the next ``call DAG.viewGraph()`` would highlight the node in
488 the specified color (choices of colors can be found at `colors
489 <http://www.graphviz.org/doc/info/colors.html>`_.) More complex node attributes
490 can be provided with ``call DAG.setGraphAttrs(node, "attributes")`` (choices can
491 be found at `Graph attributes <http://www.graphviz.org/doc/info/attrs.html>`_.)
492 If you want to restart and clear all the current graph attributes, then you can
493 ``call DAG.clearGraphAttrs()``.
495 Note that graph visualization features are compiled out of Release builds to
496 reduce file size. This means that you need a Debug+Asserts or Release+Asserts
497 build to use these features.
501 Picking the Right Data Structure for a Task
502 ===========================================
504 LLVM has a plethora of data structures in the ``llvm/ADT/`` directory, and we
505 commonly use STL data structures. This section describes the trade-offs you
506 should consider when you pick one.
508 The first step is a choose your own adventure: do you want a sequential
509 container, a set-like container, or a map-like container? The most important
510 thing when choosing a container is the algorithmic properties of how you plan to
511 access the container. Based on that, you should use:
514 * a :ref:`map-like <ds_map>` container if you need efficient look-up of a
515 value based on another value. Map-like containers also support efficient
516 queries for containment (whether a key is in the map). Map-like containers
517 generally do not support efficient reverse mapping (values to keys). If you
518 need that, use two maps. Some map-like containers also support efficient
519 iteration through the keys in sorted order. Map-like containers are the most
520 expensive sort, only use them if you need one of these capabilities.
522 * a :ref:`set-like <ds_set>` container if you need to put a bunch of stuff into
523 a container that automatically eliminates duplicates. Some set-like
524 containers support efficient iteration through the elements in sorted order.
525 Set-like containers are more expensive than sequential containers.
527 * a :ref:`sequential <ds_sequential>` container provides the most efficient way
528 to add elements and keeps track of the order they are added to the collection.
529 They permit duplicates and support efficient iteration, but do not support
530 efficient look-up based on a key.
532 * a :ref:`string <ds_string>` container is a specialized sequential container or
533 reference structure that is used for character or byte arrays.
535 * a :ref:`bit <ds_bit>` container provides an efficient way to store and
536 perform set operations on sets of numeric id's, while automatically
537 eliminating duplicates. Bit containers require a maximum of 1 bit for each
538 identifier you want to store.
540 Once the proper category of container is determined, you can fine tune the
541 memory use, constant factors, and cache behaviors of access by intelligently
542 picking a member of the category. Note that constant factors and cache behavior
543 can be a big deal. If you have a vector that usually only contains a few
544 elements (but could contain many), for example, it's much better to use
545 :ref:`SmallVector <dss_smallvector>` than :ref:`vector <dss_vector>`. Doing so
546 avoids (relatively) expensive malloc/free calls, which dwarf the cost of adding
547 the elements to the container.
551 Sequential Containers (std::vector, std::list, etc)
552 ---------------------------------------------------
554 There are a variety of sequential containers available for you, based on your
555 needs. Pick the first in this section that will do what you want.
562 The ``llvm::ArrayRef`` class is the preferred class to use in an interface that
563 accepts a sequential list of elements in memory and just reads from them. By
564 taking an ``ArrayRef``, the API can be passed a fixed size array, an
565 ``std::vector``, an ``llvm::SmallVector`` and anything else that is contiguous
573 Fixed size arrays are very simple and very fast. They are good if you know
574 exactly how many elements you have, or you have a (low) upper bound on how many
579 Heap Allocated Arrays
580 ^^^^^^^^^^^^^^^^^^^^^
582 Heap allocated arrays (``new[]`` + ``delete[]``) are also simple. They are good
583 if the number of elements is variable, if you know how many elements you will
584 need before the array is allocated, and if the array is usually large (if not,
585 consider a :ref:`SmallVector <dss_smallvector>`). The cost of a heap allocated
586 array is the cost of the new/delete (aka malloc/free). Also note that if you
587 are allocating an array of a type with a constructor, the constructor and
588 destructors will be run for every element in the array (re-sizable vectors only
589 construct those elements actually used).
591 .. _dss_tinyptrvector:
593 llvm/ADT/TinyPtrVector.h
594 ^^^^^^^^^^^^^^^^^^^^^^^^
596 ``TinyPtrVector<Type>`` is a highly specialized collection class that is
597 optimized to avoid allocation in the case when a vector has zero or one
598 elements. It has two major restrictions: 1) it can only hold values of pointer
599 type, and 2) it cannot hold a null pointer.
601 Since this container is highly specialized, it is rarely used.
605 llvm/ADT/SmallVector.h
606 ^^^^^^^^^^^^^^^^^^^^^^
608 ``SmallVector<Type, N>`` is a simple class that looks and smells just like
609 ``vector<Type>``: it supports efficient iteration, lays out elements in memory
610 order (so you can do pointer arithmetic between elements), supports efficient
611 push_back/pop_back operations, supports efficient random access to its elements,
614 The advantage of SmallVector is that it allocates space for some number of
615 elements (N) **in the object itself**. Because of this, if the SmallVector is
616 dynamically smaller than N, no malloc is performed. This can be a big win in
617 cases where the malloc/free call is far more expensive than the code that
618 fiddles around with the elements.
620 This is good for vectors that are "usually small" (e.g. the number of
621 predecessors/successors of a block is usually less than 8). On the other hand,
622 this makes the size of the SmallVector itself large, so you don't want to
623 allocate lots of them (doing so will waste a lot of space). As such,
624 SmallVectors are most useful when on the stack.
626 SmallVector also provides a nice portable and efficient replacement for
631 Prefer to use ``SmallVectorImpl<T>`` as a parameter type.
633 In APIs that don't care about the "small size" (most?), prefer to use
634 the ``SmallVectorImpl<T>`` class, which is basically just the "vector
635 header" (and methods) without the elements allocated after it. Note that
636 ``SmallVector<T, N>`` inherits from ``SmallVectorImpl<T>`` so the
637 conversion is implicit and costs nothing. E.g.
641 // BAD: Clients cannot pass e.g. SmallVector<Foo, 4>.
642 hardcodedSmallSize(SmallVector<Foo, 2> &Out);
643 // GOOD: Clients can pass any SmallVector<Foo, N>.
644 allowsAnySmallSize(SmallVectorImpl<Foo> &Out);
647 SmallVector<Foo, 8> Vec;
648 hardcodedSmallSize(Vec); // Error.
649 allowsAnySmallSize(Vec); // Works.
652 Even though it has "``Impl``" in the name, this is so widely used that
653 it really isn't "private to the implementation" anymore. A name like
654 ``SmallVectorHeader`` would be more appropriate.
661 ``std::vector`` is well loved and respected. It is useful when SmallVector
662 isn't: when the size of the vector is often large (thus the small optimization
663 will rarely be a benefit) or if you will be allocating many instances of the
664 vector itself (which would waste space for elements that aren't in the
665 container). vector is also useful when interfacing with code that expects
668 One worthwhile note about std::vector: avoid code like this:
677 Instead, write this as:
687 Doing so will save (at least) one heap allocation and free per iteration of the
695 ``std::deque`` is, in some senses, a generalized version of ``std::vector``.
696 Like ``std::vector``, it provides constant time random access and other similar
697 properties, but it also provides efficient access to the front of the list. It
698 does not guarantee continuity of elements within memory.
700 In exchange for this extra flexibility, ``std::deque`` has significantly higher
701 constant factor costs than ``std::vector``. If possible, use ``std::vector`` or
709 ``std::list`` is an extremely inefficient class that is rarely useful. It
710 performs a heap allocation for every element inserted into it, thus having an
711 extremely high constant factor, particularly for small data types.
712 ``std::list`` also only supports bidirectional iteration, not random access
715 In exchange for this high cost, std::list supports efficient access to both ends
716 of the list (like ``std::deque``, but unlike ``std::vector`` or
717 ``SmallVector``). In addition, the iterator invalidation characteristics of
718 std::list are stronger than that of a vector class: inserting or removing an
719 element into the list does not invalidate iterator or pointers to other elements
727 ``ilist<T>`` implements an 'intrusive' doubly-linked list. It is intrusive,
728 because it requires the element to store and provide access to the prev/next
729 pointers for the list.
731 ``ilist`` has the same drawbacks as ``std::list``, and additionally requires an
732 ``ilist_traits`` implementation for the element type, but it provides some novel
733 characteristics. In particular, it can efficiently store polymorphic objects,
734 the traits class is informed when an element is inserted or removed from the
735 list, and ``ilist``\ s are guaranteed to support a constant-time splice
738 These properties are exactly what we want for things like ``Instruction``\ s and
739 basic blocks, which is why these are implemented with ``ilist``\ s.
741 Related classes of interest are explained in the following subsections:
743 * :ref:`ilist_traits <dss_ilist_traits>`
745 * :ref:`iplist <dss_iplist>`
747 * :ref:`llvm/ADT/ilist_node.h <dss_ilist_node>`
749 * :ref:`Sentinels <dss_ilist_sentinel>`
751 .. _dss_packedvector:
753 llvm/ADT/PackedVector.h
754 ^^^^^^^^^^^^^^^^^^^^^^^
756 Useful for storing a vector of values using only a few number of bits for each
757 value. Apart from the standard operations of a vector-like container, it can
758 also perform an 'or' set operation.
766 FirstCondition = 0x1,
767 SecondCondition = 0x2,
772 PackedVector<State, 2> Vec1;
773 Vec1.push_back(FirstCondition);
775 PackedVector<State, 2> Vec2;
776 Vec2.push_back(SecondCondition);
779 return Vec1[0]; // returns 'Both'.
782 .. _dss_ilist_traits:
787 ``ilist_traits<T>`` is ``ilist<T>``'s customization mechanism. ``iplist<T>``
788 (and consequently ``ilist<T>``) publicly derive from this traits class.
795 ``iplist<T>`` is ``ilist<T>``'s base and as such supports a slightly narrower
796 interface. Notably, inserters from ``T&`` are absent.
798 ``ilist_traits<T>`` is a public base of this class and can be used for a wide
799 variety of customizations.
803 llvm/ADT/ilist_node.h
804 ^^^^^^^^^^^^^^^^^^^^^
806 ``ilist_node<T>`` implements a the forward and backward links that are expected
807 by the ``ilist<T>`` (and analogous containers) in the default manner.
809 ``ilist_node<T>``\ s are meant to be embedded in the node type ``T``, usually
810 ``T`` publicly derives from ``ilist_node<T>``.
812 .. _dss_ilist_sentinel:
817 ``ilist``\ s have another specialty that must be considered. To be a good
818 citizen in the C++ ecosystem, it needs to support the standard container
819 operations, such as ``begin`` and ``end`` iterators, etc. Also, the
820 ``operator--`` must work correctly on the ``end`` iterator in the case of
821 non-empty ``ilist``\ s.
823 The only sensible solution to this problem is to allocate a so-called *sentinel*
824 along with the intrusive list, which serves as the ``end`` iterator, providing
825 the back-link to the last element. However conforming to the C++ convention it
826 is illegal to ``operator++`` beyond the sentinel and it also must not be
829 These constraints allow for some implementation freedom to the ``ilist`` how to
830 allocate and store the sentinel. The corresponding policy is dictated by
831 ``ilist_traits<T>``. By default a ``T`` gets heap-allocated whenever the need
832 for a sentinel arises.
834 While the default policy is sufficient in most cases, it may break down when
835 ``T`` does not provide a default constructor. Also, in the case of many
836 instances of ``ilist``\ s, the memory overhead of the associated sentinels is
837 wasted. To alleviate the situation with numerous and voluminous
838 ``T``-sentinels, sometimes a trick is employed, leading to *ghostly sentinels*.
840 Ghostly sentinels are obtained by specially-crafted ``ilist_traits<T>`` which
841 superpose the sentinel with the ``ilist`` instance in memory. Pointer
842 arithmetic is used to obtain the sentinel, which is relative to the ``ilist``'s
843 ``this`` pointer. The ``ilist`` is augmented by an extra pointer, which serves
844 as the back-link of the sentinel. This is the only field in the ghostly
845 sentinel which can be legally accessed.
849 Other Sequential Container options
850 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
852 Other STL containers are available, such as ``std::string``.
854 There are also various STL adapter classes such as ``std::queue``,
855 ``std::priority_queue``, ``std::stack``, etc. These provide simplified access
856 to an underlying container but don't affect the cost of the container itself.
860 String-like containers
861 ----------------------
863 There are a variety of ways to pass around and use strings in C and C++, and
864 LLVM adds a few new options to choose from. Pick the first option on this list
865 that will do what you need, they are ordered according to their relative cost.
867 Note that is is generally preferred to *not* pass strings around as ``const
868 char*``'s. These have a number of problems, including the fact that they
869 cannot represent embedded nul ("\0") characters, and do not have a length
870 available efficiently. The general replacement for '``const char*``' is
873 For more information on choosing string containers for APIs, please see
874 :ref:`Passing Strings <string_apis>`.
881 The StringRef class is a simple value class that contains a pointer to a
882 character and a length, and is quite related to the :ref:`ArrayRef
883 <dss_arrayref>` class (but specialized for arrays of characters). Because
884 StringRef carries a length with it, it safely handles strings with embedded nul
885 characters in it, getting the length does not require a strlen call, and it even
886 has very convenient APIs for slicing and dicing the character range that it
889 StringRef is ideal for passing simple strings around that are known to be live,
890 either because they are C string literals, std::string, a C array, or a
891 SmallVector. Each of these cases has an efficient implicit conversion to
892 StringRef, which doesn't result in a dynamic strlen being executed.
894 StringRef has a few major limitations which make more powerful string containers
897 #. You cannot directly convert a StringRef to a 'const char*' because there is
898 no way to add a trailing nul (unlike the .c_str() method on various stronger
901 #. StringRef doesn't own or keep alive the underlying string bytes.
902 As such it can easily lead to dangling pointers, and is not suitable for
903 embedding in datastructures in most cases (instead, use an std::string or
904 something like that).
906 #. For the same reason, StringRef cannot be used as the return value of a
907 method if the method "computes" the result string. Instead, use std::string.
909 #. StringRef's do not allow you to mutate the pointed-to string bytes and it
910 doesn't allow you to insert or remove bytes from the range. For editing
911 operations like this, it interoperates with the :ref:`Twine <dss_twine>`
914 Because of its strengths and limitations, it is very common for a function to
915 take a StringRef and for a method on an object to return a StringRef that points
916 into some string that it owns.
923 The Twine class is used as an intermediary datatype for APIs that want to take a
924 string that can be constructed inline with a series of concatenations. Twine
925 works by forming recursive instances of the Twine datatype (a simple value
926 object) on the stack as temporary objects, linking them together into a tree
927 which is then linearized when the Twine is consumed. Twine is only safe to use
928 as the argument to a function, and should always be a const reference, e.g.:
932 void foo(const Twine &T);
936 foo(X + "." + Twine(i));
938 This example forms a string like "blarg.42" by concatenating the values
939 together, and does not form intermediate strings containing "blarg" or "blarg.".
941 Because Twine is constructed with temporary objects on the stack, and because
942 these instances are destroyed at the end of the current statement, it is an
943 inherently dangerous API. For example, this simple variant contains undefined
944 behavior and will probably crash:
948 void foo(const Twine &T);
952 const Twine &Tmp = X + "." + Twine(i);
955 ... because the temporaries are destroyed before the call. That said, Twine's
956 are much more efficient than intermediate std::string temporaries, and they work
957 really well with StringRef. Just be aware of their limitations.
961 llvm/ADT/SmallString.h
962 ^^^^^^^^^^^^^^^^^^^^^^
964 SmallString is a subclass of :ref:`SmallVector <dss_smallvector>` that adds some
965 convenience APIs like += that takes StringRef's. SmallString avoids allocating
966 memory in the case when the preallocated space is enough to hold its data, and
967 it calls back to general heap allocation when required. Since it owns its data,
968 it is very safe to use and supports full mutation of the string.
970 Like SmallVector's, the big downside to SmallString is their sizeof. While they
971 are optimized for small strings, they themselves are not particularly small.
972 This means that they work great for temporary scratch buffers on the stack, but
973 should not generally be put into the heap: it is very rare to see a SmallString
974 as the member of a frequently-allocated heap data structure or returned
982 The standard C++ std::string class is a very general class that (like
983 SmallString) owns its underlying data. sizeof(std::string) is very reasonable
984 so it can be embedded into heap data structures and returned by-value. On the
985 other hand, std::string is highly inefficient for inline editing (e.g.
986 concatenating a bunch of stuff together) and because it is provided by the
987 standard library, its performance characteristics depend a lot of the host
988 standard library (e.g. libc++ and MSVC provide a highly optimized string class,
989 GCC contains a really slow implementation).
991 The major disadvantage of std::string is that almost every operation that makes
992 them larger can allocate memory, which is slow. As such, it is better to use
993 SmallVector or Twine as a scratch buffer, but then use std::string to persist
998 Set-Like Containers (std::set, SmallSet, SetVector, etc)
999 --------------------------------------------------------
1001 Set-like containers are useful when you need to canonicalize multiple values
1002 into a single representation. There are several different choices for how to do
1003 this, providing various trade-offs.
1005 .. _dss_sortedvectorset:
1010 If you intend to insert a lot of elements, then do a lot of queries, a great
1011 approach is to use a vector (or other sequential container) with
1012 std::sort+std::unique to remove duplicates. This approach works really well if
1013 your usage pattern has these two distinct phases (insert then query), and can be
1014 coupled with a good choice of :ref:`sequential container <ds_sequential>`.
1016 This combination provides the several nice properties: the result data is
1017 contiguous in memory (good for cache locality), has few allocations, is easy to
1018 address (iterators in the final vector are just indices or pointers), and can be
1019 efficiently queried with a standard binary or radix search.
1026 If you have a set-like data structure that is usually small and whose elements
1027 are reasonably small, a ``SmallSet<Type, N>`` is a good choice. This set has
1028 space for N elements in place (thus, if the set is dynamically smaller than N,
1029 no malloc traffic is required) and accesses them with a simple linear search.
1030 When the set grows beyond 'N' elements, it allocates a more expensive
1031 representation that guarantees efficient access (for most types, it falls back
1032 to std::set, but for pointers it uses something far better, :ref:`SmallPtrSet
1035 The magic of this class is that it handles small sets extremely efficiently, but
1036 gracefully handles extremely large sets without loss of efficiency. The
1037 drawback is that the interface is quite small: it supports insertion, queries
1038 and erasing, but does not support iteration.
1040 .. _dss_smallptrset:
1042 llvm/ADT/SmallPtrSet.h
1043 ^^^^^^^^^^^^^^^^^^^^^^
1045 SmallPtrSet has all the advantages of ``SmallSet`` (and a ``SmallSet`` of
1046 pointers is transparently implemented with a ``SmallPtrSet``), but also supports
1047 iterators. If more than 'N' insertions are performed, a single quadratically
1048 probed hash table is allocated and grows as needed, providing extremely
1049 efficient access (constant time insertion/deleting/queries with low constant
1050 factors) and is very stingy with malloc traffic.
1052 Note that, unlike ``std::set``, the iterators of ``SmallPtrSet`` are invalidated
1053 whenever an insertion occurs. Also, the values visited by the iterators are not
1054 visited in sorted order.
1061 DenseSet is a simple quadratically probed hash table. It excels at supporting
1062 small values: it uses a single allocation to hold all of the pairs that are
1063 currently inserted in the set. DenseSet is a great way to unique small values
1064 that are not simple pointers (use :ref:`SmallPtrSet <dss_smallptrset>` for
1065 pointers). Note that DenseSet has the same requirements for the value type that
1066 :ref:`DenseMap <dss_densemap>` has.
1070 llvm/ADT/SparseSet.h
1071 ^^^^^^^^^^^^^^^^^^^^
1073 SparseSet holds a small number of objects identified by unsigned keys of
1074 moderate size. It uses a lot of memory, but provides operations that are almost
1075 as fast as a vector. Typical keys are physical registers, virtual registers, or
1076 numbered basic blocks.
1078 SparseSet is useful for algorithms that need very fast clear/find/insert/erase
1079 and fast iteration over small sets. It is not intended for building composite
1082 .. _dss_sparsemultiset:
1084 llvm/ADT/SparseMultiSet.h
1085 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1087 SparseMultiSet adds multiset behavior to SparseSet, while retaining SparseSet's
1088 desirable attributes. Like SparseSet, it typically uses a lot of memory, but
1089 provides operations that are almost as fast as a vector. Typical keys are
1090 physical registers, virtual registers, or numbered basic blocks.
1092 SparseMultiSet is useful for algorithms that need very fast
1093 clear/find/insert/erase of the entire collection, and iteration over sets of
1094 elements sharing a key. It is often a more efficient choice than using composite
1095 data structures (e.g. vector-of-vectors, map-of-vectors). It is not intended for
1096 building composite data structures.
1100 llvm/ADT/FoldingSet.h
1101 ^^^^^^^^^^^^^^^^^^^^^
1103 FoldingSet is an aggregate class that is really good at uniquing
1104 expensive-to-create or polymorphic objects. It is a combination of a chained
1105 hash table with intrusive links (uniqued objects are required to inherit from
1106 FoldingSetNode) that uses :ref:`SmallVector <dss_smallvector>` as part of its ID
1109 Consider a case where you want to implement a "getOrCreateFoo" method for a
1110 complex object (for example, a node in the code generator). The client has a
1111 description of **what** it wants to generate (it knows the opcode and all the
1112 operands), but we don't want to 'new' a node, then try inserting it into a set
1113 only to find out it already exists, at which point we would have to delete it
1114 and return the node that already exists.
1116 To support this style of client, FoldingSet perform a query with a
1117 FoldingSetNodeID (which wraps SmallVector) that can be used to describe the
1118 element that we want to query for. The query either returns the element
1119 matching the ID or it returns an opaque ID that indicates where insertion should
1120 take place. Construction of the ID usually does not require heap traffic.
1122 Because FoldingSet uses intrusive links, it can support polymorphic objects in
1123 the set (for example, you can have SDNode instances mixed with LoadSDNodes).
1124 Because the elements are individually allocated, pointers to the elements are
1125 stable: inserting or removing elements does not invalidate any pointers to other
1133 ``std::set`` is a reasonable all-around set class, which is decent at many
1134 things but great at nothing. std::set allocates memory for each element
1135 inserted (thus it is very malloc intensive) and typically stores three pointers
1136 per element in the set (thus adding a large amount of per-element space
1137 overhead). It offers guaranteed log(n) performance, which is not particularly
1138 fast from a complexity standpoint (particularly if the elements of the set are
1139 expensive to compare, like strings), and has extremely high constant factors for
1140 lookup, insertion and removal.
1142 The advantages of std::set are that its iterators are stable (deleting or
1143 inserting an element from the set does not affect iterators or pointers to other
1144 elements) and that iteration over the set is guaranteed to be in sorted order.
1145 If the elements in the set are large, then the relative overhead of the pointers
1146 and malloc traffic is not a big deal, but if the elements of the set are small,
1147 std::set is almost never a good choice.
1151 llvm/ADT/SetVector.h
1152 ^^^^^^^^^^^^^^^^^^^^
1154 LLVM's ``SetVector<Type>`` is an adapter class that combines your choice of a
1155 set-like container along with a :ref:`Sequential Container <ds_sequential>` The
1156 important property that this provides is efficient insertion with uniquing
1157 (duplicate elements are ignored) with iteration support. It implements this by
1158 inserting elements into both a set-like container and the sequential container,
1159 using the set-like container for uniquing and the sequential container for
1162 The difference between SetVector and other sets is that the order of iteration
1163 is guaranteed to match the order of insertion into the SetVector. This property
1164 is really important for things like sets of pointers. Because pointer values
1165 are non-deterministic (e.g. vary across runs of the program on different
1166 machines), iterating over the pointers in the set will not be in a well-defined
1169 The drawback of SetVector is that it requires twice as much space as a normal
1170 set and has the sum of constant factors from the set-like container and the
1171 sequential container that it uses. Use it **only** if you need to iterate over
1172 the elements in a deterministic order. SetVector is also expensive to delete
1173 elements out of (linear time), unless you use it's "pop_back" method, which is
1176 ``SetVector`` is an adapter class that defaults to using ``std::vector`` and a
1177 size 16 ``SmallSet`` for the underlying containers, so it is quite expensive.
1178 However, ``"llvm/ADT/SetVector.h"`` also provides a ``SmallSetVector`` class,
1179 which defaults to using a ``SmallVector`` and ``SmallSet`` of a specified size.
1180 If you use this, and if your sets are dynamically smaller than ``N``, you will
1181 save a lot of heap traffic.
1183 .. _dss_uniquevector:
1185 llvm/ADT/UniqueVector.h
1186 ^^^^^^^^^^^^^^^^^^^^^^^
1188 UniqueVector is similar to :ref:`SetVector <dss_setvector>` but it retains a
1189 unique ID for each element inserted into the set. It internally contains a map
1190 and a vector, and it assigns a unique ID for each value inserted into the set.
1192 UniqueVector is very expensive: its cost is the sum of the cost of maintaining
1193 both the map and vector, it has high complexity, high constant factors, and
1194 produces a lot of malloc traffic. It should be avoided.
1196 .. _dss_immutableset:
1198 llvm/ADT/ImmutableSet.h
1199 ^^^^^^^^^^^^^^^^^^^^^^^
1201 ImmutableSet is an immutable (functional) set implementation based on an AVL
1202 tree. Adding or removing elements is done through a Factory object and results
1203 in the creation of a new ImmutableSet object. If an ImmutableSet already exists
1204 with the given contents, then the existing one is returned; equality is compared
1205 with a FoldingSetNodeID. The time and space complexity of add or remove
1206 operations is logarithmic in the size of the original set.
1208 There is no method for returning an element of the set, you can only check for
1213 Other Set-Like Container Options
1214 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1216 The STL provides several other options, such as std::multiset and the various
1217 "hash_set" like containers (whether from C++ TR1 or from the SGI library). We
1218 never use hash_set and unordered_set because they are generally very expensive
1219 (each insertion requires a malloc) and very non-portable.
1221 std::multiset is useful if you're not interested in elimination of duplicates,
1222 but has all the drawbacks of std::set. A sorted vector (where you don't delete
1223 duplicate entries) or some other approach is almost always better.
1227 Map-Like Containers (std::map, DenseMap, etc)
1228 ---------------------------------------------
1230 Map-like containers are useful when you want to associate data to a key. As
1231 usual, there are a lot of different ways to do this. :)
1233 .. _dss_sortedvectormap:
1238 If your usage pattern follows a strict insert-then-query approach, you can
1239 trivially use the same approach as :ref:`sorted vectors for set-like containers
1240 <dss_sortedvectorset>`. The only difference is that your query function (which
1241 uses std::lower_bound to get efficient log(n) lookup) should only compare the
1242 key, not both the key and value. This yields the same advantages as sorted
1247 llvm/ADT/StringMap.h
1248 ^^^^^^^^^^^^^^^^^^^^
1250 Strings are commonly used as keys in maps, and they are difficult to support
1251 efficiently: they are variable length, inefficient to hash and compare when
1252 long, expensive to copy, etc. StringMap is a specialized container designed to
1253 cope with these issues. It supports mapping an arbitrary range of bytes to an
1254 arbitrary other object.
1256 The StringMap implementation uses a quadratically-probed hash table, where the
1257 buckets store a pointer to the heap allocated entries (and some other stuff).
1258 The entries in the map must be heap allocated because the strings are variable
1259 length. The string data (key) and the element object (value) are stored in the
1260 same allocation with the string data immediately after the element object.
1261 This container guarantees the "``(char*)(&Value+1)``" points to the key string
1264 The StringMap is very fast for several reasons: quadratic probing is very cache
1265 efficient for lookups, the hash value of strings in buckets is not recomputed
1266 when looking up an element, StringMap rarely has to touch the memory for
1267 unrelated objects when looking up a value (even when hash collisions happen),
1268 hash table growth does not recompute the hash values for strings already in the
1269 table, and each pair in the map is store in a single allocation (the string data
1270 is stored in the same allocation as the Value of a pair).
1272 StringMap also provides query methods that take byte ranges, so it only ever
1273 copies a string if a value is inserted into the table.
1275 StringMap iteratation order, however, is not guaranteed to be deterministic, so
1276 any uses which require that should instead use a std::map.
1280 llvm/ADT/IndexedMap.h
1281 ^^^^^^^^^^^^^^^^^^^^^
1283 IndexedMap is a specialized container for mapping small dense integers (or
1284 values that can be mapped to small dense integers) to some other type. It is
1285 internally implemented as a vector with a mapping function that maps the keys
1286 to the dense integer range.
1288 This is useful for cases like virtual registers in the LLVM code generator: they
1289 have a dense mapping that is offset by a compile-time constant (the first
1290 virtual register ID).
1297 DenseMap is a simple quadratically probed hash table. It excels at supporting
1298 small keys and values: it uses a single allocation to hold all of the pairs
1299 that are currently inserted in the map. DenseMap is a great way to map
1300 pointers to pointers, or map other small types to each other.
1302 There are several aspects of DenseMap that you should be aware of, however.
1303 The iterators in a DenseMap are invalidated whenever an insertion occurs,
1304 unlike map. Also, because DenseMap allocates space for a large number of
1305 key/value pairs (it starts with 64 by default), it will waste a lot of space if
1306 your keys or values are large. Finally, you must implement a partial
1307 specialization of DenseMapInfo for the key that you want, if it isn't already
1308 supported. This is required to tell DenseMap about two special marker values
1309 (which can never be inserted into the map) that it needs internally.
1311 DenseMap's find_as() method supports lookup operations using an alternate key
1312 type. This is useful in cases where the normal key type is expensive to
1313 construct, but cheap to compare against. The DenseMapInfo is responsible for
1314 defining the appropriate comparison and hashing methods for each alternate key
1322 ValueMap is a wrapper around a :ref:`DenseMap <dss_densemap>` mapping
1323 ``Value*``\ s (or subclasses) to another type. When a Value is deleted or
1324 RAUW'ed, ValueMap will update itself so the new version of the key is mapped to
1325 the same value, just as if the key were a WeakVH. You can configure exactly how
1326 this happens, and what else happens on these two events, by passing a ``Config``
1327 parameter to the ValueMap template.
1329 .. _dss_intervalmap:
1331 llvm/ADT/IntervalMap.h
1332 ^^^^^^^^^^^^^^^^^^^^^^
1334 IntervalMap is a compact map for small keys and values. It maps key intervals
1335 instead of single keys, and it will automatically coalesce adjacent intervals.
1336 When then map only contains a few intervals, they are stored in the map object
1337 itself to avoid allocations.
1339 The IntervalMap iterators are quite big, so they should not be passed around as
1340 STL iterators. The heavyweight iterators allow a smaller data structure.
1347 std::map has similar characteristics to :ref:`std::set <dss_set>`: it uses a
1348 single allocation per pair inserted into the map, it offers log(n) lookup with
1349 an extremely large constant factor, imposes a space penalty of 3 pointers per
1350 pair in the map, etc.
1352 std::map is most useful when your keys or values are very large, if you need to
1353 iterate over the collection in sorted order, or if you need stable iterators
1354 into the map (i.e. they don't get invalidated if an insertion or deletion of
1355 another element takes place).
1359 llvm/ADT/MapVector.h
1360 ^^^^^^^^^^^^^^^^^^^^
1362 ``MapVector<KeyT,ValueT>`` provides a subset of the DenseMap interface. The
1363 main difference is that the iteration order is guaranteed to be the insertion
1364 order, making it an easy (but somewhat expensive) solution for non-deterministic
1365 iteration over maps of pointers.
1367 It is implemented by mapping from key to an index in a vector of key,value
1368 pairs. This provides fast lookup and iteration, but has two main drawbacks: The
1369 key is stored twice and it doesn't support removing elements.
1371 .. _dss_inteqclasses:
1373 llvm/ADT/IntEqClasses.h
1374 ^^^^^^^^^^^^^^^^^^^^^^^
1376 IntEqClasses provides a compact representation of equivalence classes of small
1377 integers. Initially, each integer in the range 0..n-1 has its own equivalence
1378 class. Classes can be joined by passing two class representatives to the
1379 join(a, b) method. Two integers are in the same class when findLeader() returns
1380 the same representative.
1382 Once all equivalence classes are formed, the map can be compressed so each
1383 integer 0..n-1 maps to an equivalence class number in the range 0..m-1, where m
1384 is the total number of equivalence classes. The map must be uncompressed before
1385 it can be edited again.
1387 .. _dss_immutablemap:
1389 llvm/ADT/ImmutableMap.h
1390 ^^^^^^^^^^^^^^^^^^^^^^^
1392 ImmutableMap is an immutable (functional) map implementation based on an AVL
1393 tree. Adding or removing elements is done through a Factory object and results
1394 in the creation of a new ImmutableMap object. If an ImmutableMap already exists
1395 with the given key set, then the existing one is returned; equality is compared
1396 with a FoldingSetNodeID. The time and space complexity of add or remove
1397 operations is logarithmic in the size of the original map.
1401 Other Map-Like Container Options
1402 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1404 The STL provides several other options, such as std::multimap and the various
1405 "hash_map" like containers (whether from C++ TR1 or from the SGI library). We
1406 never use hash_set and unordered_set because they are generally very expensive
1407 (each insertion requires a malloc) and very non-portable.
1409 std::multimap is useful if you want to map a key to multiple values, but has all
1410 the drawbacks of std::map. A sorted vector or some other approach is almost
1415 Bit storage containers (BitVector, SparseBitVector)
1416 ---------------------------------------------------
1418 Unlike the other containers, there are only two bit storage containers, and
1419 choosing when to use each is relatively straightforward.
1421 One additional option is ``std::vector<bool>``: we discourage its use for two
1422 reasons 1) the implementation in many common compilers (e.g. commonly
1423 available versions of GCC) is extremely inefficient and 2) the C++ standards
1424 committee is likely to deprecate this container and/or change it significantly
1425 somehow. In any case, please don't use it.
1432 The BitVector container provides a dynamic size set of bits for manipulation.
1433 It supports individual bit setting/testing, as well as set operations. The set
1434 operations take time O(size of bitvector), but operations are performed one word
1435 at a time, instead of one bit at a time. This makes the BitVector very fast for
1436 set operations compared to other containers. Use the BitVector when you expect
1437 the number of set bits to be high (i.e. a dense set).
1439 .. _dss_smallbitvector:
1444 The SmallBitVector container provides the same interface as BitVector, but it is
1445 optimized for the case where only a small number of bits, less than 25 or so,
1446 are needed. It also transparently supports larger bit counts, but slightly less
1447 efficiently than a plain BitVector, so SmallBitVector should only be used when
1448 larger counts are rare.
1450 At this time, SmallBitVector does not support set operations (and, or, xor), and
1451 its operator[] does not provide an assignable lvalue.
1453 .. _dss_sparsebitvector:
1458 The SparseBitVector container is much like BitVector, with one major difference:
1459 Only the bits that are set, are stored. This makes the SparseBitVector much
1460 more space efficient than BitVector when the set is sparse, as well as making
1461 set operations O(number of set bits) instead of O(size of universe). The
1462 downside to the SparseBitVector is that setting and testing of random bits is
1463 O(N), and on large SparseBitVectors, this can be slower than BitVector. In our
1464 implementation, setting or testing bits in sorted order (either forwards or
1465 reverse) is O(1) worst case. Testing and setting bits within 128 bits (depends
1466 on size) of the current bit is also O(1). As a general statement,
1467 testing/setting bits in a SparseBitVector is O(distance away from last set bit).
1471 Helpful Hints for Common Operations
1472 ===================================
1474 This section describes how to perform some very simple transformations of LLVM
1475 code. This is meant to give examples of common idioms used, showing the
1476 practical side of LLVM transformations.
1478 Because this is a "how-to" section, you should also read about the main classes
1479 that you will be working with. The :ref:`Core LLVM Class Hierarchy Reference
1480 <coreclasses>` contains details and descriptions of the main classes that you
1485 Basic Inspection and Traversal Routines
1486 ---------------------------------------
1488 The LLVM compiler infrastructure have many different data structures that may be
1489 traversed. Following the example of the C++ standard template library, the
1490 techniques used to traverse these various data structures are all basically the
1491 same. For a enumerable sequence of values, the ``XXXbegin()`` function (or
1492 method) returns an iterator to the start of the sequence, the ``XXXend()``
1493 function returns an iterator pointing to one past the last valid element of the
1494 sequence, and there is some ``XXXiterator`` data type that is common between the
1497 Because the pattern for iteration is common across many different aspects of the
1498 program representation, the standard template library algorithms may be used on
1499 them, and it is easier to remember how to iterate. First we show a few common
1500 examples of the data structures that need to be traversed. Other data
1501 structures are traversed in very similar ways.
1503 .. _iterate_function:
1505 Iterating over the ``BasicBlock`` in a ``Function``
1506 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1508 It's quite common to have a ``Function`` instance that you'd like to transform
1509 in some way; in particular, you'd like to manipulate its ``BasicBlock``\ s. To
1510 facilitate this, you'll need to iterate over all of the ``BasicBlock``\ s that
1511 constitute the ``Function``. The following is an example that prints the name
1512 of a ``BasicBlock`` and the number of ``Instruction``\ s it contains:
1516 // func is a pointer to a Function instance
1517 for (Function::iterator i = func->begin(), e = func->end(); i != e; ++i)
1518 // Print out the name of the basic block if it has one, and then the
1519 // number of instructions that it contains
1520 errs() << "Basic block (name=" << i->getName() << ") has "
1521 << i->size() << " instructions.\n";
1523 Note that i can be used as if it were a pointer for the purposes of invoking
1524 member functions of the ``Instruction`` class. This is because the indirection
1525 operator is overloaded for the iterator classes. In the above code, the
1526 expression ``i->size()`` is exactly equivalent to ``(*i).size()`` just like
1529 .. _iterate_basicblock:
1531 Iterating over the ``Instruction`` in a ``BasicBlock``
1532 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1534 Just like when dealing with ``BasicBlock``\ s in ``Function``\ s, it's easy to
1535 iterate over the individual instructions that make up ``BasicBlock``\ s. Here's
1536 a code snippet that prints out each instruction in a ``BasicBlock``:
1540 // blk is a pointer to a BasicBlock instance
1541 for (BasicBlock::iterator i = blk->begin(), e = blk->end(); i != e; ++i)
1542 // The next statement works since operator<<(ostream&,...)
1543 // is overloaded for Instruction&
1544 errs() << *i << "\n";
1547 However, this isn't really the best way to print out the contents of a
1548 ``BasicBlock``! Since the ostream operators are overloaded for virtually
1549 anything you'll care about, you could have just invoked the print routine on the
1550 basic block itself: ``errs() << *blk << "\n";``.
1552 .. _iterate_insiter:
1554 Iterating over the ``Instruction`` in a ``Function``
1555 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1557 If you're finding that you commonly iterate over a ``Function``'s
1558 ``BasicBlock``\ s and then that ``BasicBlock``'s ``Instruction``\ s,
1559 ``InstIterator`` should be used instead. You'll need to include
1560 ``llvm/Support/InstIterator.h`` (`doxygen
1561 <http://llvm.org/doxygen/InstIterator_8h-source.html>`__) and then instantiate
1562 ``InstIterator``\ s explicitly in your code. Here's a small example that shows
1563 how to dump all instructions in a function to the standard error stream:
1567 #include "llvm/Support/InstIterator.h"
1569 // F is a pointer to a Function instance
1570 for (inst_iterator I = inst_begin(F), E = inst_end(F); I != E; ++I)
1571 errs() << *I << "\n";
1573 Easy, isn't it? You can also use ``InstIterator``\ s to fill a work list with
1574 its initial contents. For example, if you wanted to initialize a work list to
1575 contain all instructions in a ``Function`` F, all you would need to do is
1580 std::set<Instruction*> worklist;
1581 // or better yet, SmallPtrSet<Instruction*, 64> worklist;
1583 for (inst_iterator I = inst_begin(F), E = inst_end(F); I != E; ++I)
1584 worklist.insert(&*I);
1586 The STL set ``worklist`` would now contain all instructions in the ``Function``
1589 .. _iterate_convert:
1591 Turning an iterator into a class pointer (and vice-versa)
1592 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1594 Sometimes, it'll be useful to grab a reference (or pointer) to a class instance
1595 when all you've got at hand is an iterator. Well, extracting a reference or a
1596 pointer from an iterator is very straight-forward. Assuming that ``i`` is a
1597 ``BasicBlock::iterator`` and ``j`` is a ``BasicBlock::const_iterator``:
1601 Instruction& inst = *i; // Grab reference to instruction reference
1602 Instruction* pinst = &*i; // Grab pointer to instruction reference
1603 const Instruction& inst = *j;
1605 However, the iterators you'll be working with in the LLVM framework are special:
1606 they will automatically convert to a ptr-to-instance type whenever they need to.
1607 Instead of derferencing the iterator and then taking the address of the result,
1608 you can simply assign the iterator to the proper pointer type and you get the
1609 dereference and address-of operation as a result of the assignment (behind the
1610 scenes, this is a result of overloading casting mechanisms). Thus the last line
1611 of the last example,
1615 Instruction *pinst = &*i;
1617 is semantically equivalent to
1621 Instruction *pinst = i;
1623 It's also possible to turn a class pointer into the corresponding iterator, and
1624 this is a constant time operation (very efficient). The following code snippet
1625 illustrates use of the conversion constructors provided by LLVM iterators. By
1626 using these, you can explicitly grab the iterator of something without actually
1627 obtaining it via iteration over some structure:
1631 void printNextInstruction(Instruction* inst) {
1632 BasicBlock::iterator it(inst);
1633 ++it; // After this line, it refers to the instruction after *inst
1634 if (it != inst->getParent()->end()) errs() << *it << "\n";
1637 Unfortunately, these implicit conversions come at a cost; they prevent these
1638 iterators from conforming to standard iterator conventions, and thus from being
1639 usable with standard algorithms and containers. For example, they prevent the
1640 following code, where ``B`` is a ``BasicBlock``, from compiling:
1644 llvm::SmallVector<llvm::Instruction *, 16>(B->begin(), B->end());
1646 Because of this, these implicit conversions may be removed some day, and
1647 ``operator*`` changed to return a pointer instead of a reference.
1649 .. _iterate_complex:
1651 Finding call sites: a slightly more complex example
1652 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1654 Say that you're writing a FunctionPass and would like to count all the locations
1655 in the entire module (that is, across every ``Function``) where a certain
1656 function (i.e., some ``Function *``) is already in scope. As you'll learn
1657 later, you may want to use an ``InstVisitor`` to accomplish this in a much more
1658 straight-forward manner, but this example will allow us to explore how you'd do
1659 it if you didn't have ``InstVisitor`` around. In pseudo-code, this is what we
1662 .. code-block:: none
1664 initialize callCounter to zero
1665 for each Function f in the Module
1666 for each BasicBlock b in f
1667 for each Instruction i in b
1668 if (i is a CallInst and calls the given function)
1669 increment callCounter
1671 And the actual code is (remember, because we're writing a ``FunctionPass``, our
1672 ``FunctionPass``-derived class simply has to override the ``runOnFunction``
1677 Function* targetFunc = ...;
1679 class OurFunctionPass : public FunctionPass {
1681 OurFunctionPass(): callCounter(0) { }
1683 virtual runOnFunction(Function& F) {
1684 for (Function::iterator b = F.begin(), be = F.end(); b != be; ++b) {
1685 for (BasicBlock::iterator i = b->begin(), ie = b->end(); i != ie; ++i) {
1686 if (CallInst* callInst = dyn_cast<CallInst>(&*i)) {
1687 // We know we've encountered a call instruction, so we
1688 // need to determine if it's a call to the
1689 // function pointed to by m_func or not.
1690 if (callInst->getCalledFunction() == targetFunc)
1698 unsigned callCounter;
1701 .. _calls_and_invokes:
1703 Treating calls and invokes the same way
1704 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1706 You may have noticed that the previous example was a bit oversimplified in that
1707 it did not deal with call sites generated by 'invoke' instructions. In this,
1708 and in other situations, you may find that you want to treat ``CallInst``\ s and
1709 ``InvokeInst``\ s the same way, even though their most-specific common base
1710 class is ``Instruction``, which includes lots of less closely-related things.
1711 For these cases, LLVM provides a handy wrapper class called ``CallSite``
1712 (`doxygen <http://llvm.org/doxygen/classllvm_1_1CallSite.html>`__) It is
1713 essentially a wrapper around an ``Instruction`` pointer, with some methods that
1714 provide functionality common to ``CallInst``\ s and ``InvokeInst``\ s.
1716 This class has "value semantics": it should be passed by value, not by reference
1717 and it should not be dynamically allocated or deallocated using ``operator new``
1718 or ``operator delete``. It is efficiently copyable, assignable and
1719 constructable, with costs equivalents to that of a bare pointer. If you look at
1720 its definition, it has only a single pointer member.
1724 Iterating over def-use & use-def chains
1725 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1727 Frequently, we might have an instance of the ``Value`` class (`doxygen
1728 <http://llvm.org/doxygen/classllvm_1_1Value.html>`__) and we want to determine
1729 which ``User`` s use the ``Value``. The list of all ``User``\ s of a particular
1730 ``Value`` is called a *def-use* chain. For example, let's say we have a
1731 ``Function*`` named ``F`` to a particular function ``foo``. Finding all of the
1732 instructions that *use* ``foo`` is as simple as iterating over the *def-use*
1739 for (Value::use_iterator i = F->use_begin(), e = F->use_end(); i != e; ++i)
1740 if (Instruction *Inst = dyn_cast<Instruction>(*i)) {
1741 errs() << "F is used in instruction:\n";
1742 errs() << *Inst << "\n";
1745 Note that dereferencing a ``Value::use_iterator`` is not a very cheap operation.
1746 Instead of performing ``*i`` above several times, consider doing it only once in
1747 the loop body and reusing its result.
1749 Alternatively, it's common to have an instance of the ``User`` Class (`doxygen
1750 <http://llvm.org/doxygen/classllvm_1_1User.html>`__) and need to know what
1751 ``Value``\ s are used by it. The list of all ``Value``\ s used by a ``User`` is
1752 known as a *use-def* chain. Instances of class ``Instruction`` are common
1753 ``User`` s, so we might want to iterate over all of the values that a particular
1754 instruction uses (that is, the operands of the particular ``Instruction``):
1758 Instruction *pi = ...;
1760 for (User::op_iterator i = pi->op_begin(), e = pi->op_end(); i != e; ++i) {
1765 Declaring objects as ``const`` is an important tool of enforcing mutation free
1766 algorithms (such as analyses, etc.). For this purpose above iterators come in
1767 constant flavors as ``Value::const_use_iterator`` and
1768 ``Value::const_op_iterator``. They automatically arise when calling
1769 ``use/op_begin()`` on ``const Value*``\ s or ``const User*``\ s respectively.
1770 Upon dereferencing, they return ``const Use*``\ s. Otherwise the above patterns
1775 Iterating over predecessors & successors of blocks
1776 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1778 Iterating over the predecessors and successors of a block is quite easy with the
1779 routines defined in ``"llvm/Support/CFG.h"``. Just use code like this to
1780 iterate over all predecessors of BB:
1784 #include "llvm/Support/CFG.h"
1785 BasicBlock *BB = ...;
1787 for (pred_iterator PI = pred_begin(BB), E = pred_end(BB); PI != E; ++PI) {
1788 BasicBlock *Pred = *PI;
1792 Similarly, to iterate over successors use ``succ_iterator/succ_begin/succ_end``.
1796 Making simple changes
1797 ---------------------
1799 There are some primitive transformation operations present in the LLVM
1800 infrastructure that are worth knowing about. When performing transformations,
1801 it's fairly common to manipulate the contents of basic blocks. This section
1802 describes some of the common methods for doing so and gives example code.
1804 .. _schanges_creating:
1806 Creating and inserting new ``Instruction``\ s
1807 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1809 *Instantiating Instructions*
1811 Creation of ``Instruction``\ s is straight-forward: simply call the constructor
1812 for the kind of instruction to instantiate and provide the necessary parameters.
1813 For example, an ``AllocaInst`` only *requires* a (const-ptr-to) ``Type``. Thus:
1817 AllocaInst* ai = new AllocaInst(Type::Int32Ty);
1819 will create an ``AllocaInst`` instance that represents the allocation of one
1820 integer in the current stack frame, at run time. Each ``Instruction`` subclass
1821 is likely to have varying default parameters which change the semantics of the
1822 instruction, so refer to the `doxygen documentation for the subclass of
1823 Instruction <http://llvm.org/doxygen/classllvm_1_1Instruction.html>`_ that
1824 you're interested in instantiating.
1828 It is very useful to name the values of instructions when you're able to, as
1829 this facilitates the debugging of your transformations. If you end up looking
1830 at generated LLVM machine code, you definitely want to have logical names
1831 associated with the results of instructions! By supplying a value for the
1832 ``Name`` (default) parameter of the ``Instruction`` constructor, you associate a
1833 logical name with the result of the instruction's execution at run time. For
1834 example, say that I'm writing a transformation that dynamically allocates space
1835 for an integer on the stack, and that integer is going to be used as some kind
1836 of index by some other code. To accomplish this, I place an ``AllocaInst`` at
1837 the first point in the first ``BasicBlock`` of some ``Function``, and I'm
1838 intending to use it within the same ``Function``. I might do:
1842 AllocaInst* pa = new AllocaInst(Type::Int32Ty, 0, "indexLoc");
1844 where ``indexLoc`` is now the logical name of the instruction's execution value,
1845 which is a pointer to an integer on the run time stack.
1847 *Inserting instructions*
1849 There are essentially two ways to insert an ``Instruction`` into an existing
1850 sequence of instructions that form a ``BasicBlock``:
1852 * Insertion into an explicit instruction list
1854 Given a ``BasicBlock* pb``, an ``Instruction* pi`` within that ``BasicBlock``,
1855 and a newly-created instruction we wish to insert before ``*pi``, we do the
1860 BasicBlock *pb = ...;
1861 Instruction *pi = ...;
1862 Instruction *newInst = new Instruction(...);
1864 pb->getInstList().insert(pi, newInst); // Inserts newInst before pi in pb
1866 Appending to the end of a ``BasicBlock`` is so common that the ``Instruction``
1867 class and ``Instruction``-derived classes provide constructors which take a
1868 pointer to a ``BasicBlock`` to be appended to. For example code that looked
1873 BasicBlock *pb = ...;
1874 Instruction *newInst = new Instruction(...);
1876 pb->getInstList().push_back(newInst); // Appends newInst to pb
1882 BasicBlock *pb = ...;
1883 Instruction *newInst = new Instruction(..., pb);
1885 which is much cleaner, especially if you are creating long instruction
1888 * Insertion into an implicit instruction list
1890 ``Instruction`` instances that are already in ``BasicBlock``\ s are implicitly
1891 associated with an existing instruction list: the instruction list of the
1892 enclosing basic block. Thus, we could have accomplished the same thing as the
1893 above code without being given a ``BasicBlock`` by doing:
1897 Instruction *pi = ...;
1898 Instruction *newInst = new Instruction(...);
1900 pi->getParent()->getInstList().insert(pi, newInst);
1902 In fact, this sequence of steps occurs so frequently that the ``Instruction``
1903 class and ``Instruction``-derived classes provide constructors which take (as
1904 a default parameter) a pointer to an ``Instruction`` which the newly-created
1905 ``Instruction`` should precede. That is, ``Instruction`` constructors are
1906 capable of inserting the newly-created instance into the ``BasicBlock`` of a
1907 provided instruction, immediately before that instruction. Using an
1908 ``Instruction`` constructor with a ``insertBefore`` (default) parameter, the
1913 Instruction* pi = ...;
1914 Instruction* newInst = new Instruction(..., pi);
1916 which is much cleaner, especially if you're creating a lot of instructions and
1917 adding them to ``BasicBlock``\ s.
1919 .. _schanges_deleting:
1921 Deleting Instructions
1922 ^^^^^^^^^^^^^^^^^^^^^
1924 Deleting an instruction from an existing sequence of instructions that form a
1925 BasicBlock_ is very straight-forward: just call the instruction's
1926 ``eraseFromParent()`` method. For example:
1930 Instruction *I = .. ;
1931 I->eraseFromParent();
1933 This unlinks the instruction from its containing basic block and deletes it. If
1934 you'd just like to unlink the instruction from its containing basic block but
1935 not delete it, you can use the ``removeFromParent()`` method.
1937 .. _schanges_replacing:
1939 Replacing an Instruction with another Value
1940 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1942 Replacing individual instructions
1943 """""""""""""""""""""""""""""""""
1945 Including "`llvm/Transforms/Utils/BasicBlockUtils.h
1946 <http://llvm.org/doxygen/BasicBlockUtils_8h-source.html>`_" permits use of two
1947 very useful replace functions: ``ReplaceInstWithValue`` and
1948 ``ReplaceInstWithInst``.
1950 .. _schanges_deleting_sub:
1952 Deleting Instructions
1953 """""""""""""""""""""
1955 * ``ReplaceInstWithValue``
1957 This function replaces all uses of a given instruction with a value, and then
1958 removes the original instruction. The following example illustrates the
1959 replacement of the result of a particular ``AllocaInst`` that allocates memory
1960 for a single integer with a null pointer to an integer.
1964 AllocaInst* instToReplace = ...;
1965 BasicBlock::iterator ii(instToReplace);
1967 ReplaceInstWithValue(instToReplace->getParent()->getInstList(), ii,
1968 Constant::getNullValue(PointerType::getUnqual(Type::Int32Ty)));
1970 * ``ReplaceInstWithInst``
1972 This function replaces a particular instruction with another instruction,
1973 inserting the new instruction into the basic block at the location where the
1974 old instruction was, and replacing any uses of the old instruction with the
1975 new instruction. The following example illustrates the replacement of one
1976 ``AllocaInst`` with another.
1980 AllocaInst* instToReplace = ...;
1981 BasicBlock::iterator ii(instToReplace);
1983 ReplaceInstWithInst(instToReplace->getParent()->getInstList(), ii,
1984 new AllocaInst(Type::Int32Ty, 0, "ptrToReplacedInt"));
1987 Replacing multiple uses of Users and Values
1988 """""""""""""""""""""""""""""""""""""""""""
1990 You can use ``Value::replaceAllUsesWith`` and ``User::replaceUsesOfWith`` to
1991 change more than one use at a time. See the doxygen documentation for the
1992 `Value Class <http://llvm.org/doxygen/classllvm_1_1Value.html>`_ and `User Class
1993 <http://llvm.org/doxygen/classllvm_1_1User.html>`_, respectively, for more
1996 .. _schanges_deletingGV:
1998 Deleting GlobalVariables
1999 ^^^^^^^^^^^^^^^^^^^^^^^^
2001 Deleting a global variable from a module is just as easy as deleting an
2002 Instruction. First, you must have a pointer to the global variable that you
2003 wish to delete. You use this pointer to erase it from its parent, the module.
2008 GlobalVariable *GV = .. ;
2010 GV->eraseFromParent();
2018 In generating IR, you may need some complex types. If you know these types
2019 statically, you can use ``TypeBuilder<...>::get()``, defined in
2020 ``llvm/Support/TypeBuilder.h``, to retrieve them. ``TypeBuilder`` has two forms
2021 depending on whether you're building types for cross-compilation or native
2022 library use. ``TypeBuilder<T, true>`` requires that ``T`` be independent of the
2023 host environment, meaning that it's built out of types from the ``llvm::types``
2024 (`doxygen <http://llvm.org/doxygen/namespacellvm_1_1types.html>`__) namespace
2025 and pointers, functions, arrays, etc. built of those. ``TypeBuilder<T, false>``
2026 additionally allows native C types whose size may depend on the host compiler.
2031 FunctionType *ft = TypeBuilder<types::i<8>(types::i<32>*), true>::get();
2033 is easier to read and write than the equivalent
2037 std::vector<const Type*> params;
2038 params.push_back(PointerType::getUnqual(Type::Int32Ty));
2039 FunctionType *ft = FunctionType::get(Type::Int8Ty, params, false);
2041 See the `class comment
2042 <http://llvm.org/doxygen/TypeBuilder_8h-source.html#l00001>`_ for more details.
2049 This section describes the interaction of the LLVM APIs with multithreading,
2050 both on the part of client applications, and in the JIT, in the hosted
2053 Note that LLVM's support for multithreading is still relatively young. Up
2054 through version 2.5, the execution of threaded hosted applications was
2055 supported, but not threaded client access to the APIs. While this use case is
2056 now supported, clients *must* adhere to the guidelines specified below to ensure
2057 proper operation in multithreaded mode.
2059 Note that, on Unix-like platforms, LLVM requires the presence of GCC's atomic
2060 intrinsics in order to support threaded operation. If you need a
2061 multhreading-capable LLVM on a platform without a suitably modern system
2062 compiler, consider compiling LLVM and LLVM-GCC in single-threaded mode, and
2063 using the resultant compiler to build a copy of LLVM with multithreading
2066 .. _startmultithreaded:
2068 Entering and Exiting Multithreaded Mode
2069 ---------------------------------------
2071 In order to properly protect its internal data structures while avoiding
2072 excessive locking overhead in the single-threaded case, the LLVM must intialize
2073 certain data structures necessary to provide guards around its internals. To do
2074 so, the client program must invoke ``llvm_start_multithreaded()`` before making
2075 any concurrent LLVM API calls. To subsequently tear down these structures, use
2076 the ``llvm_stop_multithreaded()`` call. You can also use the
2077 ``llvm_is_multithreaded()`` call to check the status of multithreaded mode.
2079 Note that both of these calls must be made *in isolation*. That is to say that
2080 no other LLVM API calls may be executing at any time during the execution of
2081 ``llvm_start_multithreaded()`` or ``llvm_stop_multithreaded``. It's is the
2082 client's responsibility to enforce this isolation.
2084 The return value of ``llvm_start_multithreaded()`` indicates the success or
2085 failure of the initialization. Failure typically indicates that your copy of
2086 LLVM was built without multithreading support, typically because GCC atomic
2087 intrinsics were not found in your system compiler. In this case, the LLVM API
2088 will not be safe for concurrent calls. However, it *will* be safe for hosting
2089 threaded applications in the JIT, though :ref:`care must be taken
2090 <jitthreading>` to ensure that side exits and the like do not accidentally
2091 result in concurrent LLVM API calls.
2095 Ending Execution with ``llvm_shutdown()``
2096 -----------------------------------------
2098 When you are done using the LLVM APIs, you should call ``llvm_shutdown()`` to
2099 deallocate memory used for internal structures. This will also invoke
2100 ``llvm_stop_multithreaded()`` if LLVM is operating in multithreaded mode. As
2101 such, ``llvm_shutdown()`` requires the same isolation guarantees as
2102 ``llvm_stop_multithreaded()``.
2104 Note that, if you use scope-based shutdown, you can use the
2105 ``llvm_shutdown_obj`` class, which calls ``llvm_shutdown()`` in its destructor.
2109 Lazy Initialization with ``ManagedStatic``
2110 ------------------------------------------
2112 ``ManagedStatic`` is a utility class in LLVM used to implement static
2113 initialization of static resources, such as the global type tables. Before the
2114 invocation of ``llvm_shutdown()``, it implements a simple lazy initialization
2115 scheme. Once ``llvm_start_multithreaded()`` returns, however, it uses
2116 double-checked locking to implement thread-safe lazy initialization.
2118 Note that, because no other threads are allowed to issue LLVM API calls before
2119 ``llvm_start_multithreaded()`` returns, it is possible to have
2120 ``ManagedStatic``\ s of ``llvm::sys::Mutex``\ s.
2122 The ``llvm_acquire_global_lock()`` and ``llvm_release_global_lock`` APIs provide
2123 access to the global lock used to implement the double-checked locking for lazy
2124 initialization. These should only be used internally to LLVM, and only if you
2125 know what you're doing!
2129 Achieving Isolation with ``LLVMContext``
2130 ----------------------------------------
2132 ``LLVMContext`` is an opaque class in the LLVM API which clients can use to
2133 operate multiple, isolated instances of LLVM concurrently within the same
2134 address space. For instance, in a hypothetical compile-server, the compilation
2135 of an individual translation unit is conceptually independent from all the
2136 others, and it would be desirable to be able to compile incoming translation
2137 units concurrently on independent server threads. Fortunately, ``LLVMContext``
2138 exists to enable just this kind of scenario!
2140 Conceptually, ``LLVMContext`` provides isolation. Every LLVM entity
2141 (``Module``\ s, ``Value``\ s, ``Type``\ s, ``Constant``\ s, etc.) in LLVM's
2142 in-memory IR belongs to an ``LLVMContext``. Entities in different contexts
2143 *cannot* interact with each other: ``Module``\ s in different contexts cannot be
2144 linked together, ``Function``\ s cannot be added to ``Module``\ s in different
2145 contexts, etc. What this means is that is is safe to compile on multiple
2146 threads simultaneously, as long as no two threads operate on entities within the
2149 In practice, very few places in the API require the explicit specification of a
2150 ``LLVMContext``, other than the ``Type`` creation/lookup APIs. Because every
2151 ``Type`` carries a reference to its owning context, most other entities can
2152 determine what context they belong to by looking at their own ``Type``. If you
2153 are adding new entities to LLVM IR, please try to maintain this interface
2156 For clients that do *not* require the benefits of isolation, LLVM provides a
2157 convenience API ``getGlobalContext()``. This returns a global, lazily
2158 initialized ``LLVMContext`` that may be used in situations where isolation is
2166 LLVM's "eager" JIT compiler is safe to use in threaded programs. Multiple
2167 threads can call ``ExecutionEngine::getPointerToFunction()`` or
2168 ``ExecutionEngine::runFunction()`` concurrently, and multiple threads can run
2169 code output by the JIT concurrently. The user must still ensure that only one
2170 thread accesses IR in a given ``LLVMContext`` while another thread might be
2171 modifying it. One way to do that is to always hold the JIT lock while accessing
2172 IR outside the JIT (the JIT *modifies* the IR by adding ``CallbackVH``\ s).
2173 Another way is to only call ``getPointerToFunction()`` from the
2174 ``LLVMContext``'s thread.
2176 When the JIT is configured to compile lazily (using
2177 ``ExecutionEngine::DisableLazyCompilation(false)``), there is currently a `race
2178 condition <http://llvm.org/bugs/show_bug.cgi?id=5184>`_ in updating call sites
2179 after a function is lazily-jitted. It's still possible to use the lazy JIT in a
2180 threaded program if you ensure that only one thread at a time can call any
2181 particular lazy stub and that the JIT lock guards any IR access, but we suggest
2182 using only the eager JIT in threaded programs.
2189 This section describes some of the advanced or obscure API's that most clients
2190 do not need to be aware of. These API's tend manage the inner workings of the
2191 LLVM system, and only need to be accessed in unusual circumstances.
2195 The ``ValueSymbolTable`` class
2196 ------------------------------
2198 The ``ValueSymbolTable`` (`doxygen
2199 <http://llvm.org/doxygen/classllvm_1_1ValueSymbolTable.html>`__) class provides
2200 a symbol table that the :ref:`Function <c_Function>` and Module_ classes use for
2201 naming value definitions. The symbol table can provide a name for any Value_.
2203 Note that the ``SymbolTable`` class should not be directly accessed by most
2204 clients. It should only be used when iteration over the symbol table names
2205 themselves are required, which is very special purpose. Note that not all LLVM
2206 Value_\ s have names, and those without names (i.e. they have an empty name) do
2207 not exist in the symbol table.
2209 Symbol tables support iteration over the values in the symbol table with
2210 ``begin/end/iterator`` and supports querying to see if a specific name is in the
2211 symbol table (with ``lookup``). The ``ValueSymbolTable`` class exposes no
2212 public mutator methods, instead, simply call ``setName`` on a value, which will
2213 autoinsert it into the appropriate symbol table.
2217 The ``User`` and owned ``Use`` classes' memory layout
2218 -----------------------------------------------------
2220 The ``User`` (`doxygen <http://llvm.org/doxygen/classllvm_1_1User.html>`__)
2221 class provides a basis for expressing the ownership of ``User`` towards other
2222 `Value instance <http://llvm.org/doxygen/classllvm_1_1Value.html>`_\ s. The
2223 ``Use`` (`doxygen <http://llvm.org/doxygen/classllvm_1_1Use.html>`__) helper
2224 class is employed to do the bookkeeping and to facilitate *O(1)* addition and
2229 Interaction and relationship between ``User`` and ``Use`` objects
2230 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2232 A subclass of ``User`` can choose between incorporating its ``Use`` objects or
2233 refer to them out-of-line by means of a pointer. A mixed variant (some ``Use``
2234 s inline others hung off) is impractical and breaks the invariant that the
2235 ``Use`` objects belonging to the same ``User`` form a contiguous array.
2237 We have 2 different layouts in the ``User`` (sub)classes:
2241 The ``Use`` object(s) are inside (resp. at fixed offset) of the ``User``
2242 object and there are a fixed number of them.
2246 The ``Use`` object(s) are referenced by a pointer to an array from the
2247 ``User`` object and there may be a variable number of them.
2249 As of v2.4 each layout still possesses a direct pointer to the start of the
2250 array of ``Use``\ s. Though not mandatory for layout a), we stick to this
2251 redundancy for the sake of simplicity. The ``User`` object also stores the
2252 number of ``Use`` objects it has. (Theoretically this information can also be
2253 calculated given the scheme presented below.)
2255 Special forms of allocation operators (``operator new``) enforce the following
2258 * Layout a) is modelled by prepending the ``User`` object by the ``Use[]``
2261 .. code-block:: none
2263 ...---.---.---.---.-------...
2264 | P | P | P | P | User
2265 '''---'---'---'---'-------'''
2267 * Layout b) is modelled by pointing at the ``Use[]`` array.
2269 .. code-block:: none
2280 *(In the above figures* '``P``' *stands for the* ``Use**`` *that is stored in
2281 each* ``Use`` *object in the member* ``Use::Prev`` *)*
2285 The waymarking algorithm
2286 ^^^^^^^^^^^^^^^^^^^^^^^^
2288 Since the ``Use`` objects are deprived of the direct (back)pointer to their
2289 ``User`` objects, there must be a fast and exact method to recover it. This is
2290 accomplished by the following scheme:
2292 A bit-encoding in the 2 LSBits (least significant bits) of the ``Use::Prev``
2293 allows to find the start of the ``User`` object:
2295 * ``00`` --- binary digit 0
2297 * ``01`` --- binary digit 1
2299 * ``10`` --- stop and calculate (``s``)
2301 * ``11`` --- full stop (``S``)
2303 Given a ``Use*``, all we have to do is to walk till we get a stop and we either
2304 have a ``User`` immediately behind or we have to walk to the next stop picking
2305 up digits and calculating the offset:
2307 .. code-block:: none
2309 .---.---.---.---.---.---.---.---.---.---.---.---.---.---.---.---.----------------
2310 | 1 | s | 1 | 0 | 1 | 0 | s | 1 | 1 | 0 | s | 1 | 1 | s | 1 | S | User (or User*)
2311 '---'---'---'---'---'---'---'---'---'---'---'---'---'---'---'---'----------------
2312 |+15 |+10 |+6 |+3 |+1
2315 | | | ______________________>
2316 | | ______________________________________>
2317 | __________________________________________________________>
2319 Only the significant number of bits need to be stored between the stops, so that
2320 the *worst case is 20 memory accesses* when there are 1000 ``Use`` objects
2321 associated with a ``User``.
2325 Reference implementation
2326 ^^^^^^^^^^^^^^^^^^^^^^^^
2328 The following literate Haskell fragment demonstrates the concept:
2330 .. code-block:: haskell
2332 > import Test.QuickCheck
2334 > digits :: Int -> [Char] -> [Char]
2335 > digits 0 acc = '0' : acc
2336 > digits 1 acc = '1' : acc
2337 > digits n acc = digits (n `div` 2) $ digits (n `mod` 2) acc
2339 > dist :: Int -> [Char] -> [Char]
2342 > dist 1 acc = let r = dist 0 acc in 's' : digits (length r) r
2343 > dist n acc = dist (n - 1) $ dist 1 acc
2345 > takeLast n ss = reverse $ take n $ reverse ss
2347 > test = takeLast 40 $ dist 20 []
2350 Printing <test> gives: ``"1s100000s11010s10100s1111s1010s110s11s1S"``
2352 The reverse algorithm computes the length of the string just by examining a
2355 .. code-block:: haskell
2357 > pref :: [Char] -> Int
2359 > pref ('s':'1':rest) = decode 2 1 rest
2360 > pref (_:rest) = 1 + pref rest
2362 > decode walk acc ('0':rest) = decode (walk + 1) (acc * 2) rest
2363 > decode walk acc ('1':rest) = decode (walk + 1) (acc * 2 + 1) rest
2364 > decode walk acc _ = walk + acc
2367 Now, as expected, printing <pref test> gives ``40``.
2369 We can *quickCheck* this with following property:
2371 .. code-block:: haskell
2373 > testcase = dist 2000 []
2374 > testcaseLength = length testcase
2376 > identityProp n = n > 0 && n <= testcaseLength ==> length arr == pref arr
2377 > where arr = takeLast n testcase
2380 As expected <quickCheck identityProp> gives:
2384 *Main> quickCheck identityProp
2385 OK, passed 100 tests.
2387 Let's be a bit more exhaustive:
2389 .. code-block:: haskell
2392 > deepCheck p = check (defaultConfig { configMaxTest = 500 }) p
2395 And here is the result of <deepCheck identityProp>:
2399 *Main> deepCheck identityProp
2400 OK, passed 500 tests.
2404 Tagging considerations
2405 ^^^^^^^^^^^^^^^^^^^^^^
2407 To maintain the invariant that the 2 LSBits of each ``Use**`` in ``Use`` never
2408 change after being set up, setters of ``Use::Prev`` must re-tag the new
2409 ``Use**`` on every modification. Accordingly getters must strip the tag bits.
2411 For layout b) instead of the ``User`` we find a pointer (``User*`` with LSBit
2412 set). Following this pointer brings us to the ``User``. A portable trick
2413 ensures that the first bytes of ``User`` (if interpreted as a pointer) never has
2414 the LSBit set. (Portability is relying on the fact that all known compilers
2415 place the ``vptr`` in the first word of the instances.)
2419 The Core LLVM Class Hierarchy Reference
2420 =======================================
2422 ``#include "llvm/Type.h"``
2424 header source: `Type.h <http://llvm.org/doxygen/Type_8h-source.html>`_
2426 doxygen info: `Type Clases <http://llvm.org/doxygen/classllvm_1_1Type.html>`_
2428 The Core LLVM classes are the primary means of representing the program being
2429 inspected or transformed. The core LLVM classes are defined in header files in
2430 the ``include/llvm/`` directory, and implemented in the ``lib/VMCore``
2435 The Type class and Derived Types
2436 --------------------------------
2438 ``Type`` is a superclass of all type classes. Every ``Value`` has a ``Type``.
2439 ``Type`` cannot be instantiated directly but only through its subclasses.
2440 Certain primitive types (``VoidType``, ``LabelType``, ``FloatType`` and
2441 ``DoubleType``) have hidden subclasses. They are hidden because they offer no
2442 useful functionality beyond what the ``Type`` class offers except to distinguish
2443 themselves from other subclasses of ``Type``.
2445 All other types are subclasses of ``DerivedType``. Types can be named, but this
2446 is not a requirement. There exists exactly one instance of a given shape at any
2447 one time. This allows type equality to be performed with address equality of
2448 the Type Instance. That is, given two ``Type*`` values, the types are identical
2449 if the pointers are identical.
2453 Important Public Methods
2454 ^^^^^^^^^^^^^^^^^^^^^^^^
2456 * ``bool isIntegerTy() const``: Returns true for any integer type.
2458 * ``bool isFloatingPointTy()``: Return true if this is one of the five
2459 floating point types.
2461 * ``bool isSized()``: Return true if the type has known size. Things
2462 that don't have a size are abstract types, labels and void.
2466 Important Derived Types
2467 ^^^^^^^^^^^^^^^^^^^^^^^
2470 Subclass of DerivedType that represents integer types of any bit width. Any
2471 bit width between ``IntegerType::MIN_INT_BITS`` (1) and
2472 ``IntegerType::MAX_INT_BITS`` (~8 million) can be represented.
2474 * ``static const IntegerType* get(unsigned NumBits)``: get an integer
2475 type of a specific bit width.
2477 * ``unsigned getBitWidth() const``: Get the bit width of an integer type.
2480 This is subclassed by ArrayType, PointerType and VectorType.
2482 * ``const Type * getElementType() const``: Returns the type of each
2483 of the elements in the sequential type.
2486 This is a subclass of SequentialType and defines the interface for array
2489 * ``unsigned getNumElements() const``: Returns the number of elements
2493 Subclass of SequentialType for pointer types.
2496 Subclass of SequentialType for vector types. A vector type is similar to an
2497 ArrayType but is distinguished because it is a first class type whereas
2498 ArrayType is not. Vector types are used for vector operations and are usually
2499 small vectors of of an integer or floating point type.
2502 Subclass of DerivedTypes for struct types.
2507 Subclass of DerivedTypes for function types.
2509 * ``bool isVarArg() const``: Returns true if it's a vararg function.
2511 * ``const Type * getReturnType() const``: Returns the return type of the
2514 * ``const Type * getParamType (unsigned i)``: Returns the type of the ith
2517 * ``const unsigned getNumParams() const``: Returns the number of formal
2522 The ``Module`` class
2523 --------------------
2525 ``#include "llvm/Module.h"``
2527 header source: `Module.h <http://llvm.org/doxygen/Module_8h-source.html>`_
2529 doxygen info: `Module Class <http://llvm.org/doxygen/classllvm_1_1Module.html>`_
2531 The ``Module`` class represents the top level structure present in LLVM
2532 programs. An LLVM module is effectively either a translation unit of the
2533 original program or a combination of several translation units merged by the
2534 linker. The ``Module`` class keeps track of a list of :ref:`Function
2535 <c_Function>`\ s, a list of GlobalVariable_\ s, and a SymbolTable_.
2536 Additionally, it contains a few helpful member functions that try to make common
2541 Important Public Members of the ``Module`` class
2542 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2544 * ``Module::Module(std::string name = "")``
2546 Constructing a Module_ is easy. You can optionally provide a name for it
2547 (probably based on the name of the translation unit).
2549 * | ``Module::iterator`` - Typedef for function list iterator
2550 | ``Module::const_iterator`` - Typedef for const_iterator.
2551 | ``begin()``, ``end()``, ``size()``, ``empty()``
2553 These are forwarding methods that make it easy to access the contents of a
2554 ``Module`` object's :ref:`Function <c_Function>` list.
2556 * ``Module::FunctionListType &getFunctionList()``
2558 Returns the list of :ref:`Function <c_Function>`\ s. This is necessary to use
2559 when you need to update the list or perform a complex action that doesn't have
2560 a forwarding method.
2564 * | ``Module::global_iterator`` - Typedef for global variable list iterator
2565 | ``Module::const_global_iterator`` - Typedef for const_iterator.
2566 | ``global_begin()``, ``global_end()``, ``global_size()``, ``global_empty()``
2568 These are forwarding methods that make it easy to access the contents of a
2569 ``Module`` object's GlobalVariable_ list.
2571 * ``Module::GlobalListType &getGlobalList()``
2573 Returns the list of GlobalVariable_\ s. This is necessary to use when you
2574 need to update the list or perform a complex action that doesn't have a
2579 * ``SymbolTable *getSymbolTable()``
2581 Return a reference to the SymbolTable_ for this ``Module``.
2585 * ``Function *getFunction(StringRef Name) const``
2587 Look up the specified function in the ``Module`` SymbolTable_. If it does not
2588 exist, return ``null``.
2590 * ``Function *getOrInsertFunction(const std::string &Name, const FunctionType
2593 Look up the specified function in the ``Module`` SymbolTable_. If it does not
2594 exist, add an external declaration for the function and return it.
2596 * ``std::string getTypeName(const Type *Ty)``
2598 If there is at least one entry in the SymbolTable_ for the specified Type_,
2599 return it. Otherwise return the empty string.
2601 * ``bool addTypeName(const std::string &Name, const Type *Ty)``
2603 Insert an entry in the SymbolTable_ mapping ``Name`` to ``Ty``. If there is
2604 already an entry for this name, true is returned and the SymbolTable_ is not
2612 ``#include "llvm/Value.h"``
2614 header source: `Value.h <http://llvm.org/doxygen/Value_8h-source.html>`_
2616 doxygen info: `Value Class <http://llvm.org/doxygen/classllvm_1_1Value.html>`_
2618 The ``Value`` class is the most important class in the LLVM Source base. It
2619 represents a typed value that may be used (among other things) as an operand to
2620 an instruction. There are many different types of ``Value``\ s, such as
2621 Constant_\ s, Argument_\ s. Even Instruction_\ s and :ref:`Function
2622 <c_Function>`\ s are ``Value``\ s.
2624 A particular ``Value`` may be used many times in the LLVM representation for a
2625 program. For example, an incoming argument to a function (represented with an
2626 instance of the Argument_ class) is "used" by every instruction in the function
2627 that references the argument. To keep track of this relationship, the ``Value``
2628 class keeps a list of all of the ``User``\ s that is using it (the User_ class
2629 is a base class for all nodes in the LLVM graph that can refer to ``Value``\ s).
2630 This use list is how LLVM represents def-use information in the program, and is
2631 accessible through the ``use_*`` methods, shown below.
2633 Because LLVM is a typed representation, every LLVM ``Value`` is typed, and this
2634 Type_ is available through the ``getType()`` method. In addition, all LLVM
2635 values can be named. The "name" of the ``Value`` is a symbolic string printed
2638 .. code-block:: llvm
2644 The name of this instruction is "foo". **NOTE** that the name of any value may
2645 be missing (an empty string), so names should **ONLY** be used for debugging
2646 (making the source code easier to read, debugging printouts), they should not be
2647 used to keep track of values or map between them. For this purpose, use a
2648 ``std::map`` of pointers to the ``Value`` itself instead.
2650 One important aspect of LLVM is that there is no distinction between an SSA
2651 variable and the operation that produces it. Because of this, any reference to
2652 the value produced by an instruction (or the value available as an incoming
2653 argument, for example) is represented as a direct pointer to the instance of the
2654 class that represents this value. Although this may take some getting used to,
2655 it simplifies the representation and makes it easier to manipulate.
2659 Important Public Members of the ``Value`` class
2660 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2662 * | ``Value::use_iterator`` - Typedef for iterator over the use-list
2663 | ``Value::const_use_iterator`` - Typedef for const_iterator over the
2665 | ``unsigned use_size()`` - Returns the number of users of the value.
2666 | ``bool use_empty()`` - Returns true if there are no users.
2667 | ``use_iterator use_begin()`` - Get an iterator to the start of the
2669 | ``use_iterator use_end()`` - Get an iterator to the end of the use-list.
2670 | ``User *use_back()`` - Returns the last element in the list.
2672 These methods are the interface to access the def-use information in LLVM.
2673 As with all other iterators in LLVM, the naming conventions follow the
2674 conventions defined by the STL_.
2676 * ``Type *getType() const``
2677 This method returns the Type of the Value.
2679 * | ``bool hasName() const``
2680 | ``std::string getName() const``
2681 | ``void setName(const std::string &Name)``
2683 This family of methods is used to access and assign a name to a ``Value``, be
2684 aware of the :ref:`precaution above <nameWarning>`.
2686 * ``void replaceAllUsesWith(Value *V)``
2688 This method traverses the use list of a ``Value`` changing all User_\ s of the
2689 current value to refer to "``V``" instead. For example, if you detect that an
2690 instruction always produces a constant value (for example through constant
2691 folding), you can replace all uses of the instruction with the constant like
2696 Inst->replaceAllUsesWith(ConstVal);
2703 ``#include "llvm/User.h"``
2705 header source: `User.h <http://llvm.org/doxygen/User_8h-source.html>`_
2707 doxygen info: `User Class <http://llvm.org/doxygen/classllvm_1_1User.html>`_
2711 The ``User`` class is the common base class of all LLVM nodes that may refer to
2712 ``Value``\ s. It exposes a list of "Operands" that are all of the ``Value``\ s
2713 that the User is referring to. The ``User`` class itself is a subclass of
2716 The operands of a ``User`` point directly to the LLVM ``Value`` that it refers
2717 to. Because LLVM uses Static Single Assignment (SSA) form, there can only be
2718 one definition referred to, allowing this direct connection. This connection
2719 provides the use-def information in LLVM.
2723 Important Public Members of the ``User`` class
2724 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2726 The ``User`` class exposes the operand list in two ways: through an index access
2727 interface and through an iterator based interface.
2729 * | ``Value *getOperand(unsigned i)``
2730 | ``unsigned getNumOperands()``
2732 These two methods expose the operands of the ``User`` in a convenient form for
2735 * | ``User::op_iterator`` - Typedef for iterator over the operand list
2736 | ``op_iterator op_begin()`` - Get an iterator to the start of the operand
2738 | ``op_iterator op_end()`` - Get an iterator to the end of the operand list.
2740 Together, these methods make up the iterator based interface to the operands
2746 The ``Instruction`` class
2747 -------------------------
2749 ``#include "llvm/Instruction.h"``
2751 header source: `Instruction.h
2752 <http://llvm.org/doxygen/Instruction_8h-source.html>`_
2754 doxygen info: `Instruction Class
2755 <http://llvm.org/doxygen/classllvm_1_1Instruction.html>`_
2757 Superclasses: User_, Value_
2759 The ``Instruction`` class is the common base class for all LLVM instructions.
2760 It provides only a few methods, but is a very commonly used class. The primary
2761 data tracked by the ``Instruction`` class itself is the opcode (instruction
2762 type) and the parent BasicBlock_ the ``Instruction`` is embedded into. To
2763 represent a specific type of instruction, one of many subclasses of
2764 ``Instruction`` are used.
2766 Because the ``Instruction`` class subclasses the User_ class, its operands can
2767 be accessed in the same way as for other ``User``\ s (with the
2768 ``getOperand()``/``getNumOperands()`` and ``op_begin()``/``op_end()`` methods).
2769 An important file for the ``Instruction`` class is the ``llvm/Instruction.def``
2770 file. This file contains some meta-data about the various different types of
2771 instructions in LLVM. It describes the enum values that are used as opcodes
2772 (for example ``Instruction::Add`` and ``Instruction::ICmp``), as well as the
2773 concrete sub-classes of ``Instruction`` that implement the instruction (for
2774 example BinaryOperator_ and CmpInst_). Unfortunately, the use of macros in this
2775 file confuses doxygen, so these enum values don't show up correctly in the
2776 `doxygen output <http://llvm.org/doxygen/classllvm_1_1Instruction.html>`_.
2780 Important Subclasses of the ``Instruction`` class
2781 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2785 * ``BinaryOperator``
2787 This subclasses represents all two operand instructions whose operands must be
2788 the same type, except for the comparison instructions.
2793 This subclass is the parent of the 12 casting instructions. It provides
2794 common operations on cast instructions.
2800 This subclass respresents the two comparison instructions,
2801 `ICmpInst <LangRef.html#i_icmp>`_ (integer opreands), and
2802 `FCmpInst <LangRef.html#i_fcmp>`_ (floating point operands).
2806 * ``TerminatorInst``
2808 This subclass is the parent of all terminator instructions (those which can
2813 Important Public Members of the ``Instruction`` class
2814 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2816 * ``BasicBlock *getParent()``
2818 Returns the BasicBlock_ that this
2819 ``Instruction`` is embedded into.
2821 * ``bool mayWriteToMemory()``
2823 Returns true if the instruction writes to memory, i.e. it is a ``call``,
2824 ``free``, ``invoke``, or ``store``.
2826 * ``unsigned getOpcode()``
2828 Returns the opcode for the ``Instruction``.
2830 * ``Instruction *clone() const``
2832 Returns another instance of the specified instruction, identical in all ways
2833 to the original except that the instruction has no parent (i.e. it's not
2834 embedded into a BasicBlock_), and it has no name.
2838 The ``Constant`` class and subclasses
2839 -------------------------------------
2841 Constant represents a base class for different types of constants. It is
2842 subclassed by ConstantInt, ConstantArray, etc. for representing the various
2843 types of Constants. GlobalValue_ is also a subclass, which represents the
2844 address of a global variable or function.
2848 Important Subclasses of Constant
2849 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2851 * ConstantInt : This subclass of Constant represents an integer constant of
2854 * ``const APInt& getValue() const``: Returns the underlying
2855 value of this constant, an APInt value.
2857 * ``int64_t getSExtValue() const``: Converts the underlying APInt value to an
2858 int64_t via sign extension. If the value (not the bit width) of the APInt
2859 is too large to fit in an int64_t, an assertion will result. For this
2860 reason, use of this method is discouraged.
2862 * ``uint64_t getZExtValue() const``: Converts the underlying APInt value
2863 to a uint64_t via zero extension. IF the value (not the bit width) of the
2864 APInt is too large to fit in a uint64_t, an assertion will result. For this
2865 reason, use of this method is discouraged.
2867 * ``static ConstantInt* get(const APInt& Val)``: Returns the ConstantInt
2868 object that represents the value provided by ``Val``. The type is implied
2869 as the IntegerType that corresponds to the bit width of ``Val``.
2871 * ``static ConstantInt* get(const Type *Ty, uint64_t Val)``: Returns the
2872 ConstantInt object that represents the value provided by ``Val`` for integer
2875 * ConstantFP : This class represents a floating point constant.
2877 * ``double getValue() const``: Returns the underlying value of this constant.
2879 * ConstantArray : This represents a constant array.
2881 * ``const std::vector<Use> &getValues() const``: Returns a vector of
2882 component constants that makeup this array.
2884 * ConstantStruct : This represents a constant struct.
2886 * ``const std::vector<Use> &getValues() const``: Returns a vector of
2887 component constants that makeup this array.
2889 * GlobalValue : This represents either a global variable or a function. In
2890 either case, the value is a constant fixed address (after linking).
2894 The ``GlobalValue`` class
2895 -------------------------
2897 ``#include "llvm/GlobalValue.h"``
2899 header source: `GlobalValue.h
2900 <http://llvm.org/doxygen/GlobalValue_8h-source.html>`_
2902 doxygen info: `GlobalValue Class
2903 <http://llvm.org/doxygen/classllvm_1_1GlobalValue.html>`_
2905 Superclasses: Constant_, User_, Value_
2907 Global values ( GlobalVariable_\ s or :ref:`Function <c_Function>`\ s) are the
2908 only LLVM values that are visible in the bodies of all :ref:`Function
2909 <c_Function>`\ s. Because they are visible at global scope, they are also
2910 subject to linking with other globals defined in different translation units.
2911 To control the linking process, ``GlobalValue``\ s know their linkage rules.
2912 Specifically, ``GlobalValue``\ s know whether they have internal or external
2913 linkage, as defined by the ``LinkageTypes`` enumeration.
2915 If a ``GlobalValue`` has internal linkage (equivalent to being ``static`` in C),
2916 it is not visible to code outside the current translation unit, and does not
2917 participate in linking. If it has external linkage, it is visible to external
2918 code, and does participate in linking. In addition to linkage information,
2919 ``GlobalValue``\ s keep track of which Module_ they are currently part of.
2921 Because ``GlobalValue``\ s are memory objects, they are always referred to by
2922 their **address**. As such, the Type_ of a global is always a pointer to its
2923 contents. It is important to remember this when using the ``GetElementPtrInst``
2924 instruction because this pointer must be dereferenced first. For example, if
2925 you have a ``GlobalVariable`` (a subclass of ``GlobalValue)`` that is an array
2926 of 24 ints, type ``[24 x i32]``, then the ``GlobalVariable`` is a pointer to
2927 that array. Although the address of the first element of this array and the
2928 value of the ``GlobalVariable`` are the same, they have different types. The
2929 ``GlobalVariable``'s type is ``[24 x i32]``. The first element's type is
2930 ``i32.`` Because of this, accessing a global value requires you to dereference
2931 the pointer with ``GetElementPtrInst`` first, then its elements can be accessed.
2932 This is explained in the `LLVM Language Reference Manual
2933 <LangRef.html#globalvars>`_.
2937 Important Public Members of the ``GlobalValue`` class
2938 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2940 * | ``bool hasInternalLinkage() const``
2941 | ``bool hasExternalLinkage() const``
2942 | ``void setInternalLinkage(bool HasInternalLinkage)``
2944 These methods manipulate the linkage characteristics of the ``GlobalValue``.
2946 * ``Module *getParent()``
2948 This returns the Module_ that the
2949 GlobalValue is currently embedded into.
2953 The ``Function`` class
2954 ----------------------
2956 ``#include "llvm/Function.h"``
2958 header source: `Function.h <http://llvm.org/doxygen/Function_8h-source.html>`_
2960 doxygen info: `Function Class
2961 <http://llvm.org/doxygen/classllvm_1_1Function.html>`_
2963 Superclasses: GlobalValue_, Constant_, User_, Value_
2965 The ``Function`` class represents a single procedure in LLVM. It is actually
2966 one of the more complex classes in the LLVM hierarchy because it must keep track
2967 of a large amount of data. The ``Function`` class keeps track of a list of
2968 BasicBlock_\ s, a list of formal Argument_\ s, and a SymbolTable_.
2970 The list of BasicBlock_\ s is the most commonly used part of ``Function``
2971 objects. The list imposes an implicit ordering of the blocks in the function,
2972 which indicate how the code will be laid out by the backend. Additionally, the
2973 first BasicBlock_ is the implicit entry node for the ``Function``. It is not
2974 legal in LLVM to explicitly branch to this initial block. There are no implicit
2975 exit nodes, and in fact there may be multiple exit nodes from a single
2976 ``Function``. If the BasicBlock_ list is empty, this indicates that the
2977 ``Function`` is actually a function declaration: the actual body of the function
2978 hasn't been linked in yet.
2980 In addition to a list of BasicBlock_\ s, the ``Function`` class also keeps track
2981 of the list of formal Argument_\ s that the function receives. This container
2982 manages the lifetime of the Argument_ nodes, just like the BasicBlock_ list does
2983 for the BasicBlock_\ s.
2985 The SymbolTable_ is a very rarely used LLVM feature that is only used when you
2986 have to look up a value by name. Aside from that, the SymbolTable_ is used
2987 internally to make sure that there are not conflicts between the names of
2988 Instruction_\ s, BasicBlock_\ s, or Argument_\ s in the function body.
2990 Note that ``Function`` is a GlobalValue_ and therefore also a Constant_. The
2991 value of the function is its address (after linking) which is guaranteed to be
2996 Important Public Members of the ``Function``
2997 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2999 * ``Function(const FunctionType *Ty, LinkageTypes Linkage,
3000 const std::string &N = "", Module* Parent = 0)``
3002 Constructor used when you need to create new ``Function``\ s to add the
3003 program. The constructor must specify the type of the function to create and
3004 what type of linkage the function should have. The FunctionType_ argument
3005 specifies the formal arguments and return value for the function. The same
3006 FunctionType_ value can be used to create multiple functions. The ``Parent``
3007 argument specifies the Module in which the function is defined. If this
3008 argument is provided, the function will automatically be inserted into that
3009 module's list of functions.
3011 * ``bool isDeclaration()``
3013 Return whether or not the ``Function`` has a body defined. If the function is
3014 "external", it does not have a body, and thus must be resolved by linking with
3015 a function defined in a different translation unit.
3017 * | ``Function::iterator`` - Typedef for basic block list iterator
3018 | ``Function::const_iterator`` - Typedef for const_iterator.
3019 | ``begin()``, ``end()``, ``size()``, ``empty()``
3021 These are forwarding methods that make it easy to access the contents of a
3022 ``Function`` object's BasicBlock_ list.
3024 * ``Function::BasicBlockListType &getBasicBlockList()``
3026 Returns the list of BasicBlock_\ s. This is necessary to use when you need to
3027 update the list or perform a complex action that doesn't have a forwarding
3030 * | ``Function::arg_iterator`` - Typedef for the argument list iterator
3031 | ``Function::const_arg_iterator`` - Typedef for const_iterator.
3032 | ``arg_begin()``, ``arg_end()``, ``arg_size()``, ``arg_empty()``
3034 These are forwarding methods that make it easy to access the contents of a
3035 ``Function`` object's Argument_ list.
3037 * ``Function::ArgumentListType &getArgumentList()``
3039 Returns the list of Argument_. This is necessary to use when you need to
3040 update the list or perform a complex action that doesn't have a forwarding
3043 * ``BasicBlock &getEntryBlock()``
3045 Returns the entry ``BasicBlock`` for the function. Because the entry block
3046 for the function is always the first block, this returns the first block of
3049 * | ``Type *getReturnType()``
3050 | ``FunctionType *getFunctionType()``
3052 This traverses the Type_ of the ``Function`` and returns the return type of
3053 the function, or the FunctionType_ of the actual function.
3055 * ``SymbolTable *getSymbolTable()``
3057 Return a pointer to the SymbolTable_ for this ``Function``.
3061 The ``GlobalVariable`` class
3062 ----------------------------
3064 ``#include "llvm/GlobalVariable.h"``
3066 header source: `GlobalVariable.h
3067 <http://llvm.org/doxygen/GlobalVariable_8h-source.html>`_
3069 doxygen info: `GlobalVariable Class
3070 <http://llvm.org/doxygen/classllvm_1_1GlobalVariable.html>`_
3072 Superclasses: GlobalValue_, Constant_, User_, Value_
3074 Global variables are represented with the (surprise surprise) ``GlobalVariable``
3075 class. Like functions, ``GlobalVariable``\ s are also subclasses of
3076 GlobalValue_, and as such are always referenced by their address (global values
3077 must live in memory, so their "name" refers to their constant address). See
3078 GlobalValue_ for more on this. Global variables may have an initial value
3079 (which must be a Constant_), and if they have an initializer, they may be marked
3080 as "constant" themselves (indicating that their contents never change at
3083 .. _m_GlobalVariable:
3085 Important Public Members of the ``GlobalVariable`` class
3086 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
3088 * ``GlobalVariable(const Type *Ty, bool isConstant, LinkageTypes &Linkage,
3089 Constant *Initializer = 0, const std::string &Name = "", Module* Parent = 0)``
3091 Create a new global variable of the specified type. If ``isConstant`` is true
3092 then the global variable will be marked as unchanging for the program. The
3093 Linkage parameter specifies the type of linkage (internal, external, weak,
3094 linkonce, appending) for the variable. If the linkage is InternalLinkage,
3095 WeakAnyLinkage, WeakODRLinkage, LinkOnceAnyLinkage or LinkOnceODRLinkage, then
3096 the resultant global variable will have internal linkage. AppendingLinkage
3097 concatenates together all instances (in different translation units) of the
3098 variable into a single variable but is only applicable to arrays. See the
3099 `LLVM Language Reference <LangRef.html#modulestructure>`_ for further details
3100 on linkage types. Optionally an initializer, a name, and the module to put
3101 the variable into may be specified for the global variable as well.
3103 * ``bool isConstant() const``
3105 Returns true if this is a global variable that is known not to be modified at
3108 * ``bool hasInitializer()``
3110 Returns true if this ``GlobalVariable`` has an intializer.
3112 * ``Constant *getInitializer()``
3114 Returns the initial value for a ``GlobalVariable``. It is not legal to call
3115 this method if there is no initializer.
3119 The ``BasicBlock`` class
3120 ------------------------
3122 ``#include "llvm/BasicBlock.h"``
3124 header source: `BasicBlock.h
3125 <http://llvm.org/doxygen/BasicBlock_8h-source.html>`_
3127 doxygen info: `BasicBlock Class
3128 <http://llvm.org/doxygen/classllvm_1_1BasicBlock.html>`_
3132 This class represents a single entry single exit section of the code, commonly
3133 known as a basic block by the compiler community. The ``BasicBlock`` class
3134 maintains a list of Instruction_\ s, which form the body of the block. Matching
3135 the language definition, the last element of this list of instructions is always
3136 a terminator instruction (a subclass of the TerminatorInst_ class).
3138 In addition to tracking the list of instructions that make up the block, the
3139 ``BasicBlock`` class also keeps track of the :ref:`Function <c_Function>` that
3140 it is embedded into.
3142 Note that ``BasicBlock``\ s themselves are Value_\ s, because they are
3143 referenced by instructions like branches and can go in the switch tables.
3144 ``BasicBlock``\ s have type ``label``.
3148 Important Public Members of the ``BasicBlock`` class
3149 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
3151 * ``BasicBlock(const std::string &Name = "", Function *Parent = 0)``
3153 The ``BasicBlock`` constructor is used to create new basic blocks for
3154 insertion into a function. The constructor optionally takes a name for the
3155 new block, and a :ref:`Function <c_Function>` to insert it into. If the
3156 ``Parent`` parameter is specified, the new ``BasicBlock`` is automatically
3157 inserted at the end of the specified :ref:`Function <c_Function>`, if not
3158 specified, the BasicBlock must be manually inserted into the :ref:`Function
3161 * | ``BasicBlock::iterator`` - Typedef for instruction list iterator
3162 | ``BasicBlock::const_iterator`` - Typedef for const_iterator.
3163 | ``begin()``, ``end()``, ``front()``, ``back()``,
3164 ``size()``, ``empty()``
3165 STL-style functions for accessing the instruction list.
3167 These methods and typedefs are forwarding functions that have the same
3168 semantics as the standard library methods of the same names. These methods
3169 expose the underlying instruction list of a basic block in a way that is easy
3170 to manipulate. To get the full complement of container operations (including
3171 operations to update the list), you must use the ``getInstList()`` method.
3173 * ``BasicBlock::InstListType &getInstList()``
3175 This method is used to get access to the underlying container that actually
3176 holds the Instructions. This method must be used when there isn't a
3177 forwarding function in the ``BasicBlock`` class for the operation that you
3178 would like to perform. Because there are no forwarding functions for
3179 "updating" operations, you need to use this if you want to update the contents
3180 of a ``BasicBlock``.
3182 * ``Function *getParent()``
3184 Returns a pointer to :ref:`Function <c_Function>` the block is embedded into,
3185 or a null pointer if it is homeless.
3187 * ``TerminatorInst *getTerminator()``
3189 Returns a pointer to the terminator instruction that appears at the end of the
3190 ``BasicBlock``. If there is no terminator instruction, or if the last
3191 instruction in the block is not a terminator, then a null pointer is returned.
3195 The ``Argument`` class
3196 ----------------------
3198 This subclass of Value defines the interface for incoming formal arguments to a
3199 function. A Function maintains a list of its formal arguments. An argument has
3200 a pointer to the parent Function.