8 Introduction --- What is a pass?
9 ================================
11 The LLVM Pass Framework is an important part of the LLVM system, because LLVM
12 passes are where most of the interesting parts of the compiler exist. Passes
13 perform the transformations and optimizations that make up the compiler, they
14 build the analysis results that are used by these transformations, and they
15 are, above all, a structuring technique for compiler code.
17 All LLVM passes are subclasses of the `Pass
18 <http://llvm.org/doxygen/classllvm_1_1Pass.html>`_ class, which implement
19 functionality by overriding virtual methods inherited from ``Pass``. Depending
20 on how your pass works, you should inherit from the :ref:`ModulePass
21 <writing-an-llvm-pass-ModulePass>` , :ref:`CallGraphSCCPass
22 <writing-an-llvm-pass-CallGraphSCCPass>`, :ref:`FunctionPass
23 <writing-an-llvm-pass-FunctionPass>` , or :ref:`LoopPass
24 <writing-an-llvm-pass-LoopPass>`, or :ref:`RegionPass
25 <writing-an-llvm-pass-RegionPass>`, or :ref:`BasicBlockPass
26 <writing-an-llvm-pass-BasicBlockPass>` classes, which gives the system more
27 information about what your pass does, and how it can be combined with other
28 passes. One of the main features of the LLVM Pass Framework is that it
29 schedules passes to run in an efficient way based on the constraints that your
30 pass meets (which are indicated by which class they derive from).
32 We start by showing you how to construct a pass, everything from setting up the
33 code, to compiling, loading, and executing it. After the basics are down, more
34 advanced features are discussed.
36 Quick Start --- Writing hello world
37 ===================================
39 Here we describe how to write the "hello world" of passes. The "Hello" pass is
40 designed to simply print out the name of non-external functions that exist in
41 the program being compiled. It does not modify the program at all, it just
42 inspects it. The source code and files for this pass are available in the LLVM
43 source tree in the ``lib/Transforms/Hello`` directory.
45 .. _writing-an-llvm-pass-makefile:
47 Setting up the build environment
48 --------------------------------
50 First, configure and build LLVM. Next, you need to create a new directory
51 somewhere in the LLVM source base. For this example, we'll assume that you
52 made ``lib/Transforms/Hello``. Finally, you must set up a build script
53 (``Makefile``) that will compile the source code for the new pass. To do this,
54 copy the following into ``Makefile``:
58 # Makefile for hello pass
60 # Path to top level of LLVM hierarchy
63 # Name of the library to build
66 # Make the shared library become a loadable module so the tools can
67 # dlopen/dlsym on the resulting library.
70 # Include the makefile implementation stuff
71 include $(LEVEL)/Makefile.common
73 This makefile specifies that all of the ``.cpp`` files in the current directory
74 are to be compiled and linked together into a shared object
75 ``$(LEVEL)/Debug+Asserts/lib/Hello.so`` that can be dynamically loaded by the
76 :program:`opt` or :program:`bugpoint` tools via their :option:`-load` options.
77 If your operating system uses a suffix other than ``.so`` (such as Windows or Mac
78 OS X), the appropriate extension will be used.
80 If you are used CMake to build LLVM, see :ref:`cmake-out-of-source-pass`.
82 Now that we have the build scripts set up, we just need to write the code for
85 .. _writing-an-llvm-pass-basiccode:
90 Now that we have a way to compile our new pass, we just have to write it.
95 #include "llvm/Pass.h"
96 #include "llvm/IR/Function.h"
97 #include "llvm/Support/raw_ostream.h"
99 Which are needed because we are writing a `Pass
100 <http://llvm.org/doxygen/classllvm_1_1Pass.html>`_, we are operating on
101 `Function <http://llvm.org/doxygen/classllvm_1_1Function.html>`_\ s, and we will
102 be doing some printing.
108 using namespace llvm;
110 ... which is required because the functions from the include files live in the
119 ... which starts out an anonymous namespace. Anonymous namespaces are to C++
120 what the "``static``" keyword is to C (at global scope). It makes the things
121 declared inside of the anonymous namespace visible only to the current file.
122 If you're not familiar with them, consult a decent C++ book for more
125 Next, we declare our pass itself:
129 struct Hello : public FunctionPass {
131 This declares a "``Hello``" class that is a subclass of :ref:`FunctionPass
132 <writing-an-llvm-pass-FunctionPass>`. The different builtin pass subclasses
133 are described in detail :ref:`later <writing-an-llvm-pass-pass-classes>`, but
134 for now, know that ``FunctionPass`` operates on a function at a time.
139 Hello() : FunctionPass(ID) {}
141 This declares pass identifier used by LLVM to identify pass. This allows LLVM
142 to avoid using expensive C++ runtime information.
146 bool runOnFunction(Function &F) override {
148 errs().write_escaped(F.getName()) << "\n";
151 }; // end of struct Hello
152 } // end of anonymous namespace
154 We declare a :ref:`runOnFunction <writing-an-llvm-pass-runOnFunction>` method,
155 which overrides an abstract virtual method inherited from :ref:`FunctionPass
156 <writing-an-llvm-pass-FunctionPass>`. This is where we are supposed to do our
157 thing, so we just print out our message with the name of each function.
163 We initialize pass ID here. LLVM uses ID's address to identify a pass, so
164 initialization value is not important.
168 static RegisterPass<Hello> X("hello", "Hello World Pass",
169 false /* Only looks at CFG */,
170 false /* Analysis Pass */);
172 Lastly, we :ref:`register our class <writing-an-llvm-pass-registration>`
173 ``Hello``, giving it a command line argument "``hello``", and a name "Hello
174 World Pass". The last two arguments describe its behavior: if a pass walks CFG
175 without modifying it then the third argument is set to ``true``; if a pass is
176 an analysis pass, for example dominator tree pass, then ``true`` is supplied as
179 As a whole, the ``.cpp`` file looks like:
183 #include "llvm/Pass.h"
184 #include "llvm/IR/Function.h"
185 #include "llvm/Support/raw_ostream.h"
187 using namespace llvm;
190 struct Hello : public FunctionPass {
192 Hello() : FunctionPass(ID) {}
194 bool runOnFunction(Function &F) override {
196 errs().write_escaped(F.getName()) << '\n';
203 static RegisterPass<Hello> X("hello", "Hello World Pass", false, false);
205 Now that it's all together, compile the file with a simple "``gmake``" command
206 from the top level of your build directory and you should get a new file
207 "``Debug+Asserts/lib/Hello.so``". Note that everything in this file is
208 contained in an anonymous namespace --- this reflects the fact that passes
209 are self contained units that do not need external interfaces (although they
210 can have them) to be useful.
212 Running a pass with ``opt``
213 ---------------------------
215 Now that you have a brand new shiny shared object file, we can use the
216 :program:`opt` command to run an LLVM program through your pass. Because you
217 registered your pass with ``RegisterPass``, you will be able to use the
218 :program:`opt` tool to access it, once loaded.
220 To test it, follow the example at the end of the :doc:`GettingStarted` to
221 compile "Hello World" to LLVM. We can now run the bitcode file (hello.bc) for
222 the program through our transformation like this (or course, any bitcode file
225 .. code-block:: console
227 $ opt -load ../../Debug+Asserts/lib/Hello.so -hello < hello.bc > /dev/null
232 The :option:`-load` option specifies that :program:`opt` should load your pass
233 as a shared object, which makes "``-hello``" a valid command line argument
234 (which is one reason you need to :ref:`register your pass
235 <writing-an-llvm-pass-registration>`). Because the Hello pass does not modify
236 the program in any interesting way, we just throw away the result of
237 :program:`opt` (sending it to ``/dev/null``).
239 To see what happened to the other string you registered, try running
240 :program:`opt` with the :option:`-help` option:
242 .. code-block:: console
244 $ opt -load ../../Debug+Asserts/lib/Hello.so -help
245 OVERVIEW: llvm .bc -> .bc modular optimizer
247 USAGE: opt [options] <input bitcode>
250 Optimizations available:
252 -globalopt - Global Variable Optimizer
253 -globalsmodref-aa - Simple mod/ref analysis for globals
254 -gvn - Global Value Numbering
255 -hello - Hello World Pass
256 -indvars - Induction Variable Simplification
257 -inline - Function Integration/Inlining
260 The pass name gets added as the information string for your pass, giving some
261 documentation to users of :program:`opt`. Now that you have a working pass,
262 you would go ahead and make it do the cool transformations you want. Once you
263 get it all working and tested, it may become useful to find out how fast your
264 pass is. The :ref:`PassManager <writing-an-llvm-pass-passmanager>` provides a
265 nice command line option (:option:`--time-passes`) that allows you to get
266 information about the execution time of your pass along with the other passes
267 you queue up. For example:
269 .. code-block:: console
271 $ opt -load ../../Debug+Asserts/lib/Hello.so -hello -time-passes < hello.bc > /dev/null
275 ===============================================================================
276 ... Pass execution timing report ...
277 ===============================================================================
278 Total Execution Time: 0.02 seconds (0.0479059 wall clock)
280 ---User Time--- --System Time-- --User+System-- ---Wall Time--- --- Pass Name ---
281 0.0100 (100.0%) 0.0000 ( 0.0%) 0.0100 ( 50.0%) 0.0402 ( 84.0%) Bitcode Writer
282 0.0000 ( 0.0%) 0.0100 (100.0%) 0.0100 ( 50.0%) 0.0031 ( 6.4%) Dominator Set Construction
283 0.0000 ( 0.0%) 0.0000 ( 0.0%) 0.0000 ( 0.0%) 0.0013 ( 2.7%) Module Verifier
284 0.0000 ( 0.0%) 0.0000 ( 0.0%) 0.0000 ( 0.0%) 0.0033 ( 6.9%) Hello World Pass
285 0.0100 (100.0%) 0.0100 (100.0%) 0.0200 (100.0%) 0.0479 (100.0%) TOTAL
287 As you can see, our implementation above is pretty fast. The additional
288 passes listed are automatically inserted by the :program:`opt` tool to verify
289 that the LLVM emitted by your pass is still valid and well formed LLVM, which
290 hasn't been broken somehow.
292 Now that you have seen the basics of the mechanics behind passes, we can talk
293 about some more details of how they work and how to use them.
295 .. _writing-an-llvm-pass-pass-classes:
297 Pass classes and requirements
298 =============================
300 One of the first things that you should do when designing a new pass is to
301 decide what class you should subclass for your pass. The :ref:`Hello World
302 <writing-an-llvm-pass-basiccode>` example uses the :ref:`FunctionPass
303 <writing-an-llvm-pass-FunctionPass>` class for its implementation, but we did
304 not discuss why or when this should occur. Here we talk about the classes
305 available, from the most general to the most specific.
307 When choosing a superclass for your ``Pass``, you should choose the **most
308 specific** class possible, while still being able to meet the requirements
309 listed. This gives the LLVM Pass Infrastructure information necessary to
310 optimize how passes are run, so that the resultant compiler isn't unnecessarily
313 The ``ImmutablePass`` class
314 ---------------------------
316 The most plain and boring type of pass is the "`ImmutablePass
317 <http://llvm.org/doxygen/classllvm_1_1ImmutablePass.html>`_" class. This pass
318 type is used for passes that do not have to be run, do not change state, and
319 never need to be updated. This is not a normal type of transformation or
320 analysis, but can provide information about the current compiler configuration.
322 Although this pass class is very infrequently used, it is important for
323 providing information about the current target machine being compiled for, and
324 other static information that can affect the various transformations.
326 ``ImmutablePass``\ es never invalidate other transformations, are never
327 invalidated, and are never "run".
329 .. _writing-an-llvm-pass-ModulePass:
331 The ``ModulePass`` class
332 ------------------------
334 The `ModulePass <http://llvm.org/doxygen/classllvm_1_1ModulePass.html>`_ class
335 is the most general of all superclasses that you can use. Deriving from
336 ``ModulePass`` indicates that your pass uses the entire program as a unit,
337 referring to function bodies in no predictable order, or adding and removing
338 functions. Because nothing is known about the behavior of ``ModulePass``
339 subclasses, no optimization can be done for their execution.
341 A module pass can use function level passes (e.g. dominators) using the
342 ``getAnalysis`` interface ``getAnalysis<DominatorTree>(llvm::Function *)`` to
343 provide the function to retrieve analysis result for, if the function pass does
344 not require any module or immutable passes. Note that this can only be done
345 for functions for which the analysis ran, e.g. in the case of dominators you
346 should only ask for the ``DominatorTree`` for function definitions, not
349 To write a correct ``ModulePass`` subclass, derive from ``ModulePass`` and
350 overload the ``runOnModule`` method with the following signature:
352 The ``runOnModule`` method
353 ^^^^^^^^^^^^^^^^^^^^^^^^^^
357 virtual bool runOnModule(Module &M) = 0;
359 The ``runOnModule`` method performs the interesting work of the pass. It
360 should return ``true`` if the module was modified by the transformation and
363 .. _writing-an-llvm-pass-CallGraphSCCPass:
365 The ``CallGraphSCCPass`` class
366 ------------------------------
368 The `CallGraphSCCPass
369 <http://llvm.org/doxygen/classllvm_1_1CallGraphSCCPass.html>`_ is used by
370 passes that need to traverse the program bottom-up on the call graph (callees
371 before callers). Deriving from ``CallGraphSCCPass`` provides some mechanics
372 for building and traversing the ``CallGraph``, but also allows the system to
373 optimize execution of ``CallGraphSCCPass``\ es. If your pass meets the
374 requirements outlined below, and doesn't meet the requirements of a
375 :ref:`FunctionPass <writing-an-llvm-pass-FunctionPass>` or :ref:`BasicBlockPass
376 <writing-an-llvm-pass-BasicBlockPass>`, you should derive from
377 ``CallGraphSCCPass``.
379 ``TODO``: explain briefly what SCC, Tarjan's algo, and B-U mean.
381 To be explicit, CallGraphSCCPass subclasses are:
383 #. ... *not allowed* to inspect or modify any ``Function``\ s other than those
384 in the current SCC and the direct callers and direct callees of the SCC.
385 #. ... *required* to preserve the current ``CallGraph`` object, updating it to
386 reflect any changes made to the program.
387 #. ... *not allowed* to add or remove SCC's from the current Module, though
388 they may change the contents of an SCC.
389 #. ... *allowed* to add or remove global variables from the current Module.
390 #. ... *allowed* to maintain state across invocations of :ref:`runOnSCC
391 <writing-an-llvm-pass-runOnSCC>` (including global data).
393 Implementing a ``CallGraphSCCPass`` is slightly tricky in some cases because it
394 has to handle SCCs with more than one node in it. All of the virtual methods
395 described below should return ``true`` if they modified the program, or
396 ``false`` if they didn't.
398 The ``doInitialization(CallGraph &)`` method
399 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
403 virtual bool doInitialization(CallGraph &CG);
405 The ``doInitialization`` method is allowed to do most of the things that
406 ``CallGraphSCCPass``\ es are not allowed to do. They can add and remove
407 functions, get pointers to functions, etc. The ``doInitialization`` method is
408 designed to do simple initialization type of stuff that does not depend on the
409 SCCs being processed. The ``doInitialization`` method call is not scheduled to
410 overlap with any other pass executions (thus it should be very fast).
412 .. _writing-an-llvm-pass-runOnSCC:
414 The ``runOnSCC`` method
415 ^^^^^^^^^^^^^^^^^^^^^^^
419 virtual bool runOnSCC(CallGraphSCC &SCC) = 0;
421 The ``runOnSCC`` method performs the interesting work of the pass, and should
422 return ``true`` if the module was modified by the transformation, ``false``
425 The ``doFinalization(CallGraph &)`` method
426 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
430 virtual bool doFinalization(CallGraph &CG);
432 The ``doFinalization`` method is an infrequently used method that is called
433 when the pass framework has finished calling :ref:`runOnSCC
434 <writing-an-llvm-pass-runOnSCC>` for every SCC in the program being compiled.
436 .. _writing-an-llvm-pass-FunctionPass:
438 The ``FunctionPass`` class
439 --------------------------
441 In contrast to ``ModulePass`` subclasses, `FunctionPass
442 <http://llvm.org/doxygen/classllvm_1_1Pass.html>`_ subclasses do have a
443 predictable, local behavior that can be expected by the system. All
444 ``FunctionPass`` execute on each function in the program independent of all of
445 the other functions in the program. ``FunctionPass``\ es do not require that
446 they are executed in a particular order, and ``FunctionPass``\ es do not modify
449 To be explicit, ``FunctionPass`` subclasses are not allowed to:
451 #. Inspect or modify a ``Function`` other than the one currently being processed.
452 #. Add or remove ``Function``\ s from the current ``Module``.
453 #. Add or remove global variables from the current ``Module``.
454 #. Maintain state across invocations of :ref:`runOnFunction
455 <writing-an-llvm-pass-runOnFunction>` (including global data).
457 Implementing a ``FunctionPass`` is usually straightforward (See the :ref:`Hello
458 World <writing-an-llvm-pass-basiccode>` pass for example).
459 ``FunctionPass``\ es may overload three virtual methods to do their work. All
460 of these methods should return ``true`` if they modified the program, or
461 ``false`` if they didn't.
463 .. _writing-an-llvm-pass-doInitialization-mod:
465 The ``doInitialization(Module &)`` method
466 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
470 virtual bool doInitialization(Module &M);
472 The ``doInitialization`` method is allowed to do most of the things that
473 ``FunctionPass``\ es are not allowed to do. They can add and remove functions,
474 get pointers to functions, etc. The ``doInitialization`` method is designed to
475 do simple initialization type of stuff that does not depend on the functions
476 being processed. The ``doInitialization`` method call is not scheduled to
477 overlap with any other pass executions (thus it should be very fast).
479 A good example of how this method should be used is the `LowerAllocations
480 <http://llvm.org/doxygen/LowerAllocations_8cpp-source.html>`_ pass. This pass
481 converts ``malloc`` and ``free`` instructions into platform dependent
482 ``malloc()`` and ``free()`` function calls. It uses the ``doInitialization``
483 method to get a reference to the ``malloc`` and ``free`` functions that it
484 needs, adding prototypes to the module if necessary.
486 .. _writing-an-llvm-pass-runOnFunction:
488 The ``runOnFunction`` method
489 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
493 virtual bool runOnFunction(Function &F) = 0;
495 The ``runOnFunction`` method must be implemented by your subclass to do the
496 transformation or analysis work of your pass. As usual, a ``true`` value
497 should be returned if the function is modified.
499 .. _writing-an-llvm-pass-doFinalization-mod:
501 The ``doFinalization(Module &)`` method
502 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
506 virtual bool doFinalization(Module &M);
508 The ``doFinalization`` method is an infrequently used method that is called
509 when the pass framework has finished calling :ref:`runOnFunction
510 <writing-an-llvm-pass-runOnFunction>` for every function in the program being
513 .. _writing-an-llvm-pass-LoopPass:
515 The ``LoopPass`` class
516 ----------------------
518 All ``LoopPass`` execute on each loop in the function independent of all of the
519 other loops in the function. ``LoopPass`` processes loops in loop nest order
520 such that outer most loop is processed last.
522 ``LoopPass`` subclasses are allowed to update loop nest using ``LPPassManager``
523 interface. Implementing a loop pass is usually straightforward.
524 ``LoopPass``\ es may overload three virtual methods to do their work. All
525 these methods should return ``true`` if they modified the program, or ``false``
528 The ``doInitialization(Loop *, LPPassManager &)`` method
529 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
533 virtual bool doInitialization(Loop *, LPPassManager &LPM);
535 The ``doInitialization`` method is designed to do simple initialization type of
536 stuff that does not depend on the functions being processed. The
537 ``doInitialization`` method call is not scheduled to overlap with any other
538 pass executions (thus it should be very fast). ``LPPassManager`` interface
539 should be used to access ``Function`` or ``Module`` level analysis information.
541 .. _writing-an-llvm-pass-runOnLoop:
543 The ``runOnLoop`` method
544 ^^^^^^^^^^^^^^^^^^^^^^^^
548 virtual bool runOnLoop(Loop *, LPPassManager &LPM) = 0;
550 The ``runOnLoop`` method must be implemented by your subclass to do the
551 transformation or analysis work of your pass. As usual, a ``true`` value
552 should be returned if the function is modified. ``LPPassManager`` interface
553 should be used to update loop nest.
555 The ``doFinalization()`` method
556 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
560 virtual bool doFinalization();
562 The ``doFinalization`` method is an infrequently used method that is called
563 when the pass framework has finished calling :ref:`runOnLoop
564 <writing-an-llvm-pass-runOnLoop>` for every loop in the program being compiled.
566 .. _writing-an-llvm-pass-RegionPass:
568 The ``RegionPass`` class
569 ------------------------
571 ``RegionPass`` is similar to :ref:`LoopPass <writing-an-llvm-pass-LoopPass>`,
572 but executes on each single entry single exit region in the function.
573 ``RegionPass`` processes regions in nested order such that the outer most
574 region is processed last.
576 ``RegionPass`` subclasses are allowed to update the region tree by using the
577 ``RGPassManager`` interface. You may overload three virtual methods of
578 ``RegionPass`` to implement your own region pass. All these methods should
579 return ``true`` if they modified the program, or ``false`` if they did not.
581 The ``doInitialization(Region *, RGPassManager &)`` method
582 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
586 virtual bool doInitialization(Region *, RGPassManager &RGM);
588 The ``doInitialization`` method is designed to do simple initialization type of
589 stuff that does not depend on the functions being processed. The
590 ``doInitialization`` method call is not scheduled to overlap with any other
591 pass executions (thus it should be very fast). ``RPPassManager`` interface
592 should be used to access ``Function`` or ``Module`` level analysis information.
594 .. _writing-an-llvm-pass-runOnRegion:
596 The ``runOnRegion`` method
597 ^^^^^^^^^^^^^^^^^^^^^^^^^^
601 virtual bool runOnRegion(Region *, RGPassManager &RGM) = 0;
603 The ``runOnRegion`` method must be implemented by your subclass to do the
604 transformation or analysis work of your pass. As usual, a true value should be
605 returned if the region is modified. ``RGPassManager`` interface should be used to
608 The ``doFinalization()`` method
609 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
613 virtual bool doFinalization();
615 The ``doFinalization`` method is an infrequently used method that is called
616 when the pass framework has finished calling :ref:`runOnRegion
617 <writing-an-llvm-pass-runOnRegion>` for every region in the program being
620 .. _writing-an-llvm-pass-BasicBlockPass:
622 The ``BasicBlockPass`` class
623 ----------------------------
625 ``BasicBlockPass``\ es are just like :ref:`FunctionPass's
626 <writing-an-llvm-pass-FunctionPass>` , except that they must limit their scope
627 of inspection and modification to a single basic block at a time. As such,
628 they are **not** allowed to do any of the following:
630 #. Modify or inspect any basic blocks outside of the current one.
631 #. Maintain state across invocations of :ref:`runOnBasicBlock
632 <writing-an-llvm-pass-runOnBasicBlock>`.
633 #. Modify the control flow graph (by altering terminator instructions)
634 #. Any of the things forbidden for :ref:`FunctionPasses
635 <writing-an-llvm-pass-FunctionPass>`.
637 ``BasicBlockPass``\ es are useful for traditional local and "peephole"
638 optimizations. They may override the same :ref:`doInitialization(Module &)
639 <writing-an-llvm-pass-doInitialization-mod>` and :ref:`doFinalization(Module &)
640 <writing-an-llvm-pass-doFinalization-mod>` methods that :ref:`FunctionPass's
641 <writing-an-llvm-pass-FunctionPass>` have, but also have the following virtual
642 methods that may also be implemented:
644 The ``doInitialization(Function &)`` method
645 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
649 virtual bool doInitialization(Function &F);
651 The ``doInitialization`` method is allowed to do most of the things that
652 ``BasicBlockPass``\ es are not allowed to do, but that ``FunctionPass``\ es
653 can. The ``doInitialization`` method is designed to do simple initialization
654 that does not depend on the ``BasicBlock``\ s being processed. The
655 ``doInitialization`` method call is not scheduled to overlap with any other
656 pass executions (thus it should be very fast).
658 .. _writing-an-llvm-pass-runOnBasicBlock:
660 The ``runOnBasicBlock`` method
661 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
665 virtual bool runOnBasicBlock(BasicBlock &BB) = 0;
667 Override this function to do the work of the ``BasicBlockPass``. This function
668 is not allowed to inspect or modify basic blocks other than the parameter, and
669 are not allowed to modify the CFG. A ``true`` value must be returned if the
670 basic block is modified.
672 The ``doFinalization(Function &)`` method
673 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
677 virtual bool doFinalization(Function &F);
679 The ``doFinalization`` method is an infrequently used method that is called
680 when the pass framework has finished calling :ref:`runOnBasicBlock
681 <writing-an-llvm-pass-runOnBasicBlock>` for every ``BasicBlock`` in the program
682 being compiled. This can be used to perform per-function finalization.
684 The ``MachineFunctionPass`` class
685 ---------------------------------
687 A ``MachineFunctionPass`` is a part of the LLVM code generator that executes on
688 the machine-dependent representation of each LLVM function in the program.
690 Code generator passes are registered and initialized specially by
691 ``TargetMachine::addPassesToEmitFile`` and similar routines, so they cannot
692 generally be run from the :program:`opt` or :program:`bugpoint` commands.
694 A ``MachineFunctionPass`` is also a ``FunctionPass``, so all the restrictions
695 that apply to a ``FunctionPass`` also apply to it. ``MachineFunctionPass``\ es
696 also have additional restrictions. In particular, ``MachineFunctionPass``\ es
697 are not allowed to do any of the following:
699 #. Modify or create any LLVM IR ``Instruction``\ s, ``BasicBlock``\ s,
700 ``Argument``\ s, ``Function``\ s, ``GlobalVariable``\ s,
701 ``GlobalAlias``\ es, or ``Module``\ s.
702 #. Modify a ``MachineFunction`` other than the one currently being processed.
703 #. Maintain state across invocations of :ref:`runOnMachineFunction
704 <writing-an-llvm-pass-runOnMachineFunction>` (including global data).
706 .. _writing-an-llvm-pass-runOnMachineFunction:
708 The ``runOnMachineFunction(MachineFunction &MF)`` method
709 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
713 virtual bool runOnMachineFunction(MachineFunction &MF) = 0;
715 ``runOnMachineFunction`` can be considered the main entry point of a
716 ``MachineFunctionPass``; that is, you should override this method to do the
717 work of your ``MachineFunctionPass``.
719 The ``runOnMachineFunction`` method is called on every ``MachineFunction`` in a
720 ``Module``, so that the ``MachineFunctionPass`` may perform optimizations on
721 the machine-dependent representation of the function. If you want to get at
722 the LLVM ``Function`` for the ``MachineFunction`` you're working on, use
723 ``MachineFunction``'s ``getFunction()`` accessor method --- but remember, you
724 may not modify the LLVM ``Function`` or its contents from a
725 ``MachineFunctionPass``.
727 .. _writing-an-llvm-pass-registration:
732 In the :ref:`Hello World <writing-an-llvm-pass-basiccode>` example pass we
733 illustrated how pass registration works, and discussed some of the reasons that
734 it is used and what it does. Here we discuss how and why passes are
737 As we saw above, passes are registered with the ``RegisterPass`` template. The
738 template parameter is the name of the pass that is to be used on the command
739 line to specify that the pass should be added to a program (for example, with
740 :program:`opt` or :program:`bugpoint`). The first argument is the name of the
741 pass, which is to be used for the :option:`-help` output of programs, as well
742 as for debug output generated by the :option:`--debug-pass` option.
744 If you want your pass to be easily dumpable, you should implement the virtual
752 virtual void print(llvm::raw_ostream &O, const Module *M) const;
754 The ``print`` method must be implemented by "analyses" in order to print a
755 human readable version of the analysis results. This is useful for debugging
756 an analysis itself, as well as for other people to figure out how an analysis
757 works. Use the opt ``-analyze`` argument to invoke this method.
759 The ``llvm::raw_ostream`` parameter specifies the stream to write the results
760 on, and the ``Module`` parameter gives a pointer to the top level module of the
761 program that has been analyzed. Note however that this pointer may be ``NULL``
762 in certain circumstances (such as calling the ``Pass::dump()`` from a
763 debugger), so it should only be used to enhance debug output, it should not be
766 .. _writing-an-llvm-pass-interaction:
768 Specifying interactions between passes
769 --------------------------------------
771 One of the main responsibilities of the ``PassManager`` is to make sure that
772 passes interact with each other correctly. Because ``PassManager`` tries to
773 :ref:`optimize the execution of passes <writing-an-llvm-pass-passmanager>` it
774 must know how the passes interact with each other and what dependencies exist
775 between the various passes. To track this, each pass can declare the set of
776 passes that are required to be executed before the current pass, and the passes
777 which are invalidated by the current pass.
779 Typically this functionality is used to require that analysis results are
780 computed before your pass is run. Running arbitrary transformation passes can
781 invalidate the computed analysis results, which is what the invalidation set
782 specifies. If a pass does not implement the :ref:`getAnalysisUsage
783 <writing-an-llvm-pass-getAnalysisUsage>` method, it defaults to not having any
784 prerequisite passes, and invalidating **all** other passes.
786 .. _writing-an-llvm-pass-getAnalysisUsage:
788 The ``getAnalysisUsage`` method
789 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
793 virtual void getAnalysisUsage(AnalysisUsage &Info) const;
795 By implementing the ``getAnalysisUsage`` method, the required and invalidated
796 sets may be specified for your transformation. The implementation should fill
797 in the `AnalysisUsage
798 <http://llvm.org/doxygen/classllvm_1_1AnalysisUsage.html>`_ object with
799 information about which passes are required and not invalidated. To do this, a
800 pass may call any of the following methods on the ``AnalysisUsage`` object:
802 The ``AnalysisUsage::addRequired<>`` and ``AnalysisUsage::addRequiredTransitive<>`` methods
803 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
805 If your pass requires a previous pass to be executed (an analysis for example),
806 it can use one of these methods to arrange for it to be run before your pass.
807 LLVM has many different types of analyses and passes that can be required,
808 spanning the range from ``DominatorSet`` to ``BreakCriticalEdges``. Requiring
809 ``BreakCriticalEdges``, for example, guarantees that there will be no critical
810 edges in the CFG when your pass has been run.
812 Some analyses chain to other analyses to do their job. For example, an
813 `AliasAnalysis <AliasAnalysis>` implementation is required to :ref:`chain
814 <aliasanalysis-chaining>` to other alias analysis passes. In cases where
815 analyses chain, the ``addRequiredTransitive`` method should be used instead of
816 the ``addRequired`` method. This informs the ``PassManager`` that the
817 transitively required pass should be alive as long as the requiring pass is.
819 The ``AnalysisUsage::addPreserved<>`` method
820 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
822 One of the jobs of the ``PassManager`` is to optimize how and when analyses are
823 run. In particular, it attempts to avoid recomputing data unless it needs to.
824 For this reason, passes are allowed to declare that they preserve (i.e., they
825 don't invalidate) an existing analysis if it's available. For example, a
826 simple constant folding pass would not modify the CFG, so it can't possibly
827 affect the results of dominator analysis. By default, all passes are assumed
828 to invalidate all others.
830 The ``AnalysisUsage`` class provides several methods which are useful in
831 certain circumstances that are related to ``addPreserved``. In particular, the
832 ``setPreservesAll`` method can be called to indicate that the pass does not
833 modify the LLVM program at all (which is true for analyses), and the
834 ``setPreservesCFG`` method can be used by transformations that change
835 instructions in the program but do not modify the CFG or terminator
836 instructions (note that this property is implicitly set for
837 :ref:`BasicBlockPass <writing-an-llvm-pass-BasicBlockPass>`\ es).
839 ``addPreserved`` is particularly useful for transformations like
840 ``BreakCriticalEdges``. This pass knows how to update a small set of loop and
841 dominator related analyses if they exist, so it can preserve them, despite the
842 fact that it hacks on the CFG.
844 Example implementations of ``getAnalysisUsage``
845 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
849 // This example modifies the program, but does not modify the CFG
850 void LICM::getAnalysisUsage(AnalysisUsage &AU) const {
851 AU.setPreservesCFG();
852 AU.addRequired<LoopInfoWrapperPass>();
855 .. _writing-an-llvm-pass-getAnalysis:
857 The ``getAnalysis<>`` and ``getAnalysisIfAvailable<>`` methods
858 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
860 The ``Pass::getAnalysis<>`` method is automatically inherited by your class,
861 providing you with access to the passes that you declared that you required
862 with the :ref:`getAnalysisUsage <writing-an-llvm-pass-getAnalysisUsage>`
863 method. It takes a single template argument that specifies which pass class
864 you want, and returns a reference to that pass. For example:
868 bool LICM::runOnFunction(Function &F) {
869 LoopInfo &LI = getAnalysis<LoopInfoWrapperPass>().getLoopInfo();
873 This method call returns a reference to the pass desired. You may get a
874 runtime assertion failure if you attempt to get an analysis that you did not
875 declare as required in your :ref:`getAnalysisUsage
876 <writing-an-llvm-pass-getAnalysisUsage>` implementation. This method can be
877 called by your ``run*`` method implementation, or by any other local method
878 invoked by your ``run*`` method.
880 A module level pass can use function level analysis info using this interface.
885 bool ModuleLevelPass::runOnModule(Module &M) {
887 DominatorTree &DT = getAnalysis<DominatorTree>(Func);
891 In above example, ``runOnFunction`` for ``DominatorTree`` is called by pass
892 manager before returning a reference to the desired pass.
894 If your pass is capable of updating analyses if they exist (e.g.,
895 ``BreakCriticalEdges``, as described above), you can use the
896 ``getAnalysisIfAvailable`` method, which returns a pointer to the analysis if
897 it is active. For example:
901 if (DominatorSet *DS = getAnalysisIfAvailable<DominatorSet>()) {
902 // A DominatorSet is active. This code will update it.
905 Implementing Analysis Groups
906 ----------------------------
908 Now that we understand the basics of how passes are defined, how they are used,
909 and how they are required from other passes, it's time to get a little bit
910 fancier. All of the pass relationships that we have seen so far are very
911 simple: one pass depends on one other specific pass to be run before it can
912 run. For many applications, this is great, for others, more flexibility is
915 In particular, some analyses are defined such that there is a single simple
916 interface to the analysis results, but multiple ways of calculating them.
917 Consider alias analysis for example. The most trivial alias analysis returns
918 "may alias" for any alias query. The most sophisticated analysis a
919 flow-sensitive, context-sensitive interprocedural analysis that can take a
920 significant amount of time to execute (and obviously, there is a lot of room
921 between these two extremes for other implementations). To cleanly support
922 situations like this, the LLVM Pass Infrastructure supports the notion of
925 Analysis Group Concepts
926 ^^^^^^^^^^^^^^^^^^^^^^^
928 An Analysis Group is a single simple interface that may be implemented by
929 multiple different passes. Analysis Groups can be given human readable names
930 just like passes, but unlike passes, they need not derive from the ``Pass``
931 class. An analysis group may have one or more implementations, one of which is
932 the "default" implementation.
934 Analysis groups are used by client passes just like other passes are: the
935 ``AnalysisUsage::addRequired()`` and ``Pass::getAnalysis()`` methods. In order
936 to resolve this requirement, the :ref:`PassManager
937 <writing-an-llvm-pass-passmanager>` scans the available passes to see if any
938 implementations of the analysis group are available. If none is available, the
939 default implementation is created for the pass to use. All standard rules for
940 :ref:`interaction between passes <writing-an-llvm-pass-interaction>` still
943 Although :ref:`Pass Registration <writing-an-llvm-pass-registration>` is
944 optional for normal passes, all analysis group implementations must be
945 registered, and must use the :ref:`INITIALIZE_AG_PASS
946 <writing-an-llvm-pass-RegisterAnalysisGroup>` template to join the
947 implementation pool. Also, a default implementation of the interface **must**
948 be registered with :ref:`RegisterAnalysisGroup
949 <writing-an-llvm-pass-RegisterAnalysisGroup>`.
951 As a concrete example of an Analysis Group in action, consider the
952 `AliasAnalysis <http://llvm.org/doxygen/classllvm_1_1AliasAnalysis.html>`_
953 analysis group. The default implementation of the alias analysis interface
954 (the `basicaa <http://llvm.org/doxygen/structBasicAliasAnalysis.html>`_ pass)
955 just does a few simple checks that don't require significant analysis to
956 compute (such as: two different globals can never alias each other, etc).
957 Passes that use the `AliasAnalysis
958 <http://llvm.org/doxygen/classllvm_1_1AliasAnalysis.html>`_ interface (for
959 example the `gcse <http://llvm.org/doxygen/structGCSE.html>`_ pass), do not
960 care which implementation of alias analysis is actually provided, they just use
961 the designated interface.
963 From the user's perspective, commands work just like normal. Issuing the
964 command ``opt -gcse ...`` will cause the ``basicaa`` class to be instantiated
965 and added to the pass sequence. Issuing the command ``opt -somefancyaa -gcse
966 ...`` will cause the ``gcse`` pass to use the ``somefancyaa`` alias analysis
967 (which doesn't actually exist, it's just a hypothetical example) instead.
969 .. _writing-an-llvm-pass-RegisterAnalysisGroup:
971 Using ``RegisterAnalysisGroup``
972 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
974 The ``RegisterAnalysisGroup`` template is used to register the analysis group
975 itself, while the ``INITIALIZE_AG_PASS`` is used to add pass implementations to
976 the analysis group. First, an analysis group should be registered, with a
977 human readable name provided for it. Unlike registration of passes, there is
978 no command line argument to be specified for the Analysis Group Interface
979 itself, because it is "abstract":
983 static RegisterAnalysisGroup<AliasAnalysis> A("Alias Analysis");
985 Once the analysis is registered, passes can declare that they are valid
986 implementations of the interface by using the following code:
991 // Declare that we implement the AliasAnalysis interface
992 INITIALIZE_AG_PASS(FancyAA, AliasAnalysis , "somefancyaa",
993 "A more complex alias analysis implementation",
994 false, // Is CFG Only?
995 true, // Is Analysis?
996 false); // Is default Analysis Group implementation?
999 This just shows a class ``FancyAA`` that uses the ``INITIALIZE_AG_PASS`` macro
1000 both to register and to "join" the `AliasAnalysis
1001 <http://llvm.org/doxygen/classllvm_1_1AliasAnalysis.html>`_ analysis group.
1002 Every implementation of an analysis group should join using this macro.
1007 // Declare that we implement the AliasAnalysis interface
1008 INITIALIZE_AG_PASS(BasicAA, AliasAnalysis, "basicaa",
1009 "Basic Alias Analysis (default AA impl)",
1010 false, // Is CFG Only?
1011 true, // Is Analysis?
1012 true); // Is default Analysis Group implementation?
1015 Here we show how the default implementation is specified (using the final
1016 argument to the ``INITIALIZE_AG_PASS`` template). There must be exactly one
1017 default implementation available at all times for an Analysis Group to be used.
1018 Only default implementation can derive from ``ImmutablePass``. Here we declare
1019 that the `BasicAliasAnalysis
1020 <http://llvm.org/doxygen/structBasicAliasAnalysis.html>`_ pass is the default
1021 implementation for the interface.
1026 The `Statistic <http://llvm.org/doxygen/Statistic_8h-source.html>`_ class is
1027 designed to be an easy way to expose various success metrics from passes.
1028 These statistics are printed at the end of a run, when the :option:`-stats`
1029 command line option is enabled on the command line. See the :ref:`Statistics
1030 section <Statistic>` in the Programmer's Manual for details.
1032 .. _writing-an-llvm-pass-passmanager:
1034 What PassManager does
1035 ---------------------
1037 The `PassManager <http://llvm.org/doxygen/PassManager_8h-source.html>`_ `class
1038 <http://llvm.org/doxygen/classllvm_1_1PassManager.html>`_ takes a list of
1039 passes, ensures their :ref:`prerequisites <writing-an-llvm-pass-interaction>`
1040 are set up correctly, and then schedules passes to run efficiently. All of the
1041 LLVM tools that run passes use the PassManager for execution of these passes.
1043 The PassManager does two main things to try to reduce the execution time of a
1046 #. **Share analysis results.** The ``PassManager`` attempts to avoid
1047 recomputing analysis results as much as possible. This means keeping track
1048 of which analyses are available already, which analyses get invalidated, and
1049 which analyses are needed to be run for a pass. An important part of work
1050 is that the ``PassManager`` tracks the exact lifetime of all analysis
1051 results, allowing it to :ref:`free memory
1052 <writing-an-llvm-pass-releaseMemory>` allocated to holding analysis results
1053 as soon as they are no longer needed.
1055 #. **Pipeline the execution of passes on the program.** The ``PassManager``
1056 attempts to get better cache and memory usage behavior out of a series of
1057 passes by pipelining the passes together. This means that, given a series
1058 of consecutive :ref:`FunctionPass <writing-an-llvm-pass-FunctionPass>`, it
1059 will execute all of the :ref:`FunctionPass
1060 <writing-an-llvm-pass-FunctionPass>` on the first function, then all of the
1061 :ref:`FunctionPasses <writing-an-llvm-pass-FunctionPass>` on the second
1062 function, etc... until the entire program has been run through the passes.
1064 This improves the cache behavior of the compiler, because it is only
1065 touching the LLVM program representation for a single function at a time,
1066 instead of traversing the entire program. It reduces the memory consumption
1067 of compiler, because, for example, only one `DominatorSet
1068 <http://llvm.org/doxygen/classllvm_1_1DominatorSet.html>`_ needs to be
1069 calculated at a time. This also makes it possible to implement some
1070 :ref:`interesting enhancements <writing-an-llvm-pass-SMP>` in the future.
1072 The effectiveness of the ``PassManager`` is influenced directly by how much
1073 information it has about the behaviors of the passes it is scheduling. For
1074 example, the "preserved" set is intentionally conservative in the face of an
1075 unimplemented :ref:`getAnalysisUsage <writing-an-llvm-pass-getAnalysisUsage>`
1076 method. Not implementing when it should be implemented will have the effect of
1077 not allowing any analysis results to live across the execution of your pass.
1079 The ``PassManager`` class exposes a ``--debug-pass`` command line options that
1080 is useful for debugging pass execution, seeing how things work, and diagnosing
1081 when you should be preserving more analyses than you currently are. (To get
1082 information about all of the variants of the ``--debug-pass`` option, just type
1083 "``opt -help-hidden``").
1085 By using the --debug-pass=Structure option, for example, we can see how our
1086 :ref:`Hello World <writing-an-llvm-pass-basiccode>` pass interacts with other
1087 passes. Lets try it out with the gcse and licm passes:
1089 .. code-block:: console
1091 $ opt -load ../../Debug+Asserts/lib/Hello.so -gcse -licm --debug-pass=Structure < hello.bc > /dev/null
1093 Function Pass Manager
1094 Dominator Set Construction
1095 Immediate Dominators Construction
1096 Global Common Subexpression Elimination
1097 -- Immediate Dominators Construction
1098 -- Global Common Subexpression Elimination
1099 Natural Loop Construction
1100 Loop Invariant Code Motion
1101 -- Natural Loop Construction
1102 -- Loop Invariant Code Motion
1104 -- Dominator Set Construction
1109 This output shows us when passes are constructed and when the analysis results
1110 are known to be dead (prefixed with "``--``"). Here we see that GCSE uses
1111 dominator and immediate dominator information to do its job. The LICM pass
1112 uses natural loop information, which uses dominator sets, but not immediate
1113 dominators. Because immediate dominators are no longer useful after the GCSE
1114 pass, it is immediately destroyed. The dominator sets are then reused to
1115 compute natural loop information, which is then used by the LICM pass.
1117 After the LICM pass, the module verifier runs (which is automatically added by
1118 the :program:`opt` tool), which uses the dominator set to check that the
1119 resultant LLVM code is well formed. After it finishes, the dominator set
1120 information is destroyed, after being computed once, and shared by three
1123 Lets see how this changes when we run the :ref:`Hello World
1124 <writing-an-llvm-pass-basiccode>` pass in between the two passes:
1126 .. code-block:: console
1128 $ opt -load ../../Debug+Asserts/lib/Hello.so -gcse -hello -licm --debug-pass=Structure < hello.bc > /dev/null
1130 Function Pass Manager
1131 Dominator Set Construction
1132 Immediate Dominators Construction
1133 Global Common Subexpression Elimination
1134 -- Dominator Set Construction
1135 -- Immediate Dominators Construction
1136 -- Global Common Subexpression Elimination
1139 Dominator Set Construction
1140 Natural Loop Construction
1141 Loop Invariant Code Motion
1142 -- Natural Loop Construction
1143 -- Loop Invariant Code Motion
1145 -- Dominator Set Construction
1153 Here we see that the :ref:`Hello World <writing-an-llvm-pass-basiccode>` pass
1154 has killed the Dominator Set pass, even though it doesn't modify the code at
1155 all! To fix this, we need to add the following :ref:`getAnalysisUsage
1156 <writing-an-llvm-pass-getAnalysisUsage>` method to our pass:
1160 // We don't modify the program, so we preserve all analyses
1161 void getAnalysisUsage(AnalysisUsage &AU) const override {
1162 AU.setPreservesAll();
1165 Now when we run our pass, we get this output:
1167 .. code-block:: console
1169 $ opt -load ../../Debug+Asserts/lib/Hello.so -gcse -hello -licm --debug-pass=Structure < hello.bc > /dev/null
1170 Pass Arguments: -gcse -hello -licm
1172 Function Pass Manager
1173 Dominator Set Construction
1174 Immediate Dominators Construction
1175 Global Common Subexpression Elimination
1176 -- Immediate Dominators Construction
1177 -- Global Common Subexpression Elimination
1180 Natural Loop Construction
1181 Loop Invariant Code Motion
1182 -- Loop Invariant Code Motion
1183 -- Natural Loop Construction
1185 -- Dominator Set Construction
1193 Which shows that we don't accidentally invalidate dominator information
1194 anymore, and therefore do not have to compute it twice.
1196 .. _writing-an-llvm-pass-releaseMemory:
1198 The ``releaseMemory`` method
1199 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1203 virtual void releaseMemory();
1205 The ``PassManager`` automatically determines when to compute analysis results,
1206 and how long to keep them around for. Because the lifetime of the pass object
1207 itself is effectively the entire duration of the compilation process, we need
1208 some way to free analysis results when they are no longer useful. The
1209 ``releaseMemory`` virtual method is the way to do this.
1211 If you are writing an analysis or any other pass that retains a significant
1212 amount of state (for use by another pass which "requires" your pass and uses
1213 the :ref:`getAnalysis <writing-an-llvm-pass-getAnalysis>` method) you should
1214 implement ``releaseMemory`` to, well, release the memory allocated to maintain
1215 this internal state. This method is called after the ``run*`` method for the
1216 class, before the next call of ``run*`` in your pass.
1218 Registering dynamically loaded passes
1219 =====================================
1221 *Size matters* when constructing production quality tools using LLVM, both for
1222 the purposes of distribution, and for regulating the resident code size when
1223 running on the target system. Therefore, it becomes desirable to selectively
1224 use some passes, while omitting others and maintain the flexibility to change
1225 configurations later on. You want to be able to do all this, and, provide
1226 feedback to the user. This is where pass registration comes into play.
1228 The fundamental mechanisms for pass registration are the
1229 ``MachinePassRegistry`` class and subclasses of ``MachinePassRegistryNode``.
1231 An instance of ``MachinePassRegistry`` is used to maintain a list of
1232 ``MachinePassRegistryNode`` objects. This instance maintains the list and
1233 communicates additions and deletions to the command line interface.
1235 An instance of ``MachinePassRegistryNode`` subclass is used to maintain
1236 information provided about a particular pass. This information includes the
1237 command line name, the command help string and the address of the function used
1238 to create an instance of the pass. A global static constructor of one of these
1239 instances *registers* with a corresponding ``MachinePassRegistry``, the static
1240 destructor *unregisters*. Thus a pass that is statically linked in the tool
1241 will be registered at start up. A dynamically loaded pass will register on
1242 load and unregister at unload.
1244 Using existing registries
1245 -------------------------
1247 There are predefined registries to track instruction scheduling
1248 (``RegisterScheduler``) and register allocation (``RegisterRegAlloc``) machine
1249 passes. Here we will describe how to *register* a register allocator machine
1252 Implement your register allocator machine pass. In your register allocator
1253 ``.cpp`` file add the following include:
1257 #include "llvm/CodeGen/RegAllocRegistry.h"
1259 Also in your register allocator ``.cpp`` file, define a creator function in the
1264 FunctionPass *createMyRegisterAllocator() {
1265 return new MyRegisterAllocator();
1268 Note that the signature of this function should match the type of
1269 ``RegisterRegAlloc::FunctionPassCtor``. In the same file add the "installing"
1270 declaration, in the form:
1274 static RegisterRegAlloc myRegAlloc("myregalloc",
1275 "my register allocator help string",
1276 createMyRegisterAllocator);
1278 Note the two spaces prior to the help string produces a tidy result on the
1279 :option:`-help` query.
1281 .. code-block:: console
1285 -regalloc - Register allocator to use (default=linearscan)
1286 =linearscan - linear scan register allocator
1287 =local - local register allocator
1288 =simple - simple register allocator
1289 =myregalloc - my register allocator help string
1292 And that's it. The user is now free to use ``-regalloc=myregalloc`` as an
1293 option. Registering instruction schedulers is similar except use the
1294 ``RegisterScheduler`` class. Note that the
1295 ``RegisterScheduler::FunctionPassCtor`` is significantly different from
1296 ``RegisterRegAlloc::FunctionPassCtor``.
1298 To force the load/linking of your register allocator into the
1299 :program:`llc`/:program:`lli` tools, add your creator function's global
1300 declaration to ``Passes.h`` and add a "pseudo" call line to
1301 ``llvm/Codegen/LinkAllCodegenComponents.h``.
1303 Creating new registries
1304 -----------------------
1306 The easiest way to get started is to clone one of the existing registries; we
1307 recommend ``llvm/CodeGen/RegAllocRegistry.h``. The key things to modify are
1308 the class name and the ``FunctionPassCtor`` type.
1310 Then you need to declare the registry. Example: if your pass registry is
1311 ``RegisterMyPasses`` then define:
1315 MachinePassRegistry RegisterMyPasses::Registry;
1317 And finally, declare the command line option for your passes. Example:
1321 cl::opt<RegisterMyPasses::FunctionPassCtor, false,
1322 RegisterPassParser<RegisterMyPasses> >
1324 cl::init(&createDefaultMyPass),
1325 cl::desc("my pass option help"));
1327 Here the command option is "``mypass``", with ``createDefaultMyPass`` as the
1330 Using GDB with dynamically loaded passes
1331 ----------------------------------------
1333 Unfortunately, using GDB with dynamically loaded passes is not as easy as it
1334 should be. First of all, you can't set a breakpoint in a shared object that
1335 has not been loaded yet, and second of all there are problems with inlined
1336 functions in shared objects. Here are some suggestions to debugging your pass
1339 For sake of discussion, I'm going to assume that you are debugging a
1340 transformation invoked by :program:`opt`, although nothing described here
1343 Setting a breakpoint in your pass
1344 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1346 First thing you do is start gdb on the opt process:
1348 .. code-block:: console
1352 Copyright 2000 Free Software Foundation, Inc.
1353 GDB is free software, covered by the GNU General Public License, and you are
1354 welcome to change it and/or distribute copies of it under certain conditions.
1355 Type "show copying" to see the conditions.
1356 There is absolutely no warranty for GDB. Type "show warranty" for details.
1357 This GDB was configured as "sparc-sun-solaris2.6"...
1360 Note that :program:`opt` has a lot of debugging information in it, so it takes
1361 time to load. Be patient. Since we cannot set a breakpoint in our pass yet
1362 (the shared object isn't loaded until runtime), we must execute the process,
1363 and have it stop before it invokes our pass, but after it has loaded the shared
1364 object. The most foolproof way of doing this is to set a breakpoint in
1365 ``PassManager::run`` and then run the process with the arguments you want:
1367 .. code-block:: console
1369 $ (gdb) break llvm::PassManager::run
1370 Breakpoint 1 at 0x2413bc: file Pass.cpp, line 70.
1371 (gdb) run test.bc -load $(LLVMTOP)/llvm/Debug+Asserts/lib/[libname].so -[passoption]
1372 Starting program: opt test.bc -load $(LLVMTOP)/llvm/Debug+Asserts/lib/[libname].so -[passoption]
1373 Breakpoint 1, PassManager::run (this=0xffbef174, M=@0x70b298) at Pass.cpp:70
1374 70 bool PassManager::run(Module &M) { return PM->run(M); }
1377 Once the :program:`opt` stops in the ``PassManager::run`` method you are now
1378 free to set breakpoints in your pass so that you can trace through execution or
1379 do other standard debugging stuff.
1381 Miscellaneous Problems
1382 ^^^^^^^^^^^^^^^^^^^^^^
1384 Once you have the basics down, there are a couple of problems that GDB has,
1385 some with solutions, some without.
1387 * Inline functions have bogus stack information. In general, GDB does a pretty
1388 good job getting stack traces and stepping through inline functions. When a
1389 pass is dynamically loaded however, it somehow completely loses this
1390 capability. The only solution I know of is to de-inline a function (move it
1391 from the body of a class to a ``.cpp`` file).
1393 * Restarting the program breaks breakpoints. After following the information
1394 above, you have succeeded in getting some breakpoints planted in your pass.
1395 Nex thing you know, you restart the program (i.e., you type "``run``" again),
1396 and you start getting errors about breakpoints being unsettable. The only
1397 way I have found to "fix" this problem is to delete the breakpoints that are
1398 already set in your pass, run the program, and re-set the breakpoints once
1399 execution stops in ``PassManager::run``.
1401 Hopefully these tips will help with common case debugging situations. If you'd
1402 like to contribute some tips of your own, just contact `Chris
1403 <mailto:sabre@nondot.org>`_.
1405 Future extensions planned
1406 -------------------------
1408 Although the LLVM Pass Infrastructure is very capable as it stands, and does
1409 some nifty stuff, there are things we'd like to add in the future. Here is
1412 .. _writing-an-llvm-pass-SMP:
1417 Multiple CPU machines are becoming more common and compilation can never be
1418 fast enough: obviously we should allow for a multithreaded compiler. Because
1419 of the semantics defined for passes above (specifically they cannot maintain
1420 state across invocations of their ``run*`` methods), a nice clean way to
1421 implement a multithreaded compiler would be for the ``PassManager`` class to
1422 create multiple instances of each pass object, and allow the separate instances
1423 to be hacking on different parts of the program at the same time.
1425 This implementation would prevent each of the passes from having to implement
1426 multithreaded constructs, requiring only the LLVM core to have locking in a few
1427 places (for global resources). Although this is a simple extension, we simply
1428 haven't had time (or multiprocessor machines, thus a reason) to implement this.
1429 Despite that, we have kept the LLVM passes SMP ready, and you should too.