xref: /openbsd/gnu/llvm/llvm/docs/Passes.rst (revision 771fbea0)
1..
2    If Passes.html is up to date, the following "one-liner" should print
3    an empty diff.
4
5    egrep -e '^<tr><td><a href="#.*">-.*</a></td><td>.*</td></tr>$' \
6          -e '^  <a name=".*">.*</a>$' < Passes.html >html; \
7    perl >help <<'EOT' && diff -u help html; rm -f help html
8    open HTML, "<Passes.html" or die "open: Passes.html: $!\n";
9    while (<HTML>) {
10      m:^<tr><td><a href="#(.*)">-.*</a></td><td>.*</td></tr>$: or next;
11      $order{$1} = sprintf("%03d", 1 + int %order);
12    }
13    open HELP, "../Release/bin/opt -help|" or die "open: opt -help: $!\n";
14    while (<HELP>) {
15      m:^    -([^ ]+) +- (.*)$: or next;
16      my $o = $order{$1};
17      $o = "000" unless defined $o;
18      push @x, "$o<tr><td><a href=\"#$1\">-$1</a></td><td>$2</td></tr>\n";
19      push @y, "$o  <a name=\"$1\">-$1: $2</a>\n";
20    }
21    @x = map { s/^\d\d\d//; $_ } sort @x;
22    @y = map { s/^\d\d\d//; $_ } sort @y;
23    print @x, @y;
24    EOT
25
26    This (real) one-liner can also be helpful when converting comments to HTML:
27
28    perl -e '$/ = undef; for (split(/\n/, <>)) { s:^ *///? ?::; print "  <p>\n" if !$on && $_ =~ /\S/; print "  </p>\n" if $on && $_ =~ /^\s*$/; print "  $_\n"; $on = ($_ =~ /\S/); } print "  </p>\n" if $on'
29
30====================================
31LLVM's Analysis and Transform Passes
32====================================
33
34.. contents::
35    :local:
36
37Introduction
38============
39
40This document serves as a high level summary of the optimization features that
41LLVM provides.  Optimizations are implemented as Passes that traverse some
42portion of a program to either collect information or transform the program.
43The table below divides the passes that LLVM provides into three categories.
44Analysis passes compute information that other passes can use or for debugging
45or program visualization purposes.  Transform passes can use (or invalidate)
46the analysis passes.  Transform passes all mutate the program in some way.
47Utility passes provides some utility but don't otherwise fit categorization.
48For example passes to extract functions to bitcode or write a module to bitcode
49are neither analysis nor transform passes.  The table of contents above
50provides a quick summary of each pass and links to the more complete pass
51description later in the document.
52
53Analysis Passes
54===============
55
56This section describes the LLVM Analysis Passes.
57
58``-aa-eval``: Exhaustive Alias Analysis Precision Evaluator
59-----------------------------------------------------------
60
61This is a simple N^2 alias analysis accuracy evaluator.  Basically, for each
62function in the program, it simply queries to see how the alias analysis
63implementation answers alias queries between each pair of pointers in the
64function.
65
66This is inspired and adapted from code by: Naveen Neelakantam, Francesco
67Spadini, and Wojciech Stryjewski.
68
69``-basic-aa``: Basic Alias Analysis (stateless AA impl)
70-------------------------------------------------------
71
72A basic alias analysis pass that implements identities (two different globals
73cannot alias, etc), but does no stateful analysis.
74
75``-basiccg``: Basic CallGraph Construction
76------------------------------------------
77
78Yet to be written.
79
80``-count-aa``: Count Alias Analysis Query Responses
81---------------------------------------------------
82
83A pass which can be used to count how many alias queries are being made and how
84the alias analysis implementation being used responds.
85
86.. _passes-da:
87
88``-da``: Dependence Analysis
89----------------------------
90
91Dependence analysis framework, which is used to detect dependences in memory
92accesses.
93
94``-debug-aa``: AA use debugger
95------------------------------
96
97This simple pass checks alias analysis users to ensure that if they create a
98new value, they do not query AA without informing it of the value.  It acts as
99a shim over any other AA pass you want.
100
101Yes keeping track of every value in the program is expensive, but this is a
102debugging pass.
103
104``-domfrontier``: Dominance Frontier Construction
105-------------------------------------------------
106
107This pass is a simple dominator construction algorithm for finding forward
108dominator frontiers.
109
110``-domtree``: Dominator Tree Construction
111-----------------------------------------
112
113This pass is a simple dominator construction algorithm for finding forward
114dominators.
115
116
117``-dot-callgraph``: Print Call Graph to "dot" file
118--------------------------------------------------
119
120This pass, only available in ``opt``, prints the call graph into a ``.dot``
121graph.  This graph can then be processed with the "dot" tool to convert it to
122postscript or some other suitable format.
123
124``-dot-cfg``: Print CFG of function to "dot" file
125-------------------------------------------------
126
127This pass, only available in ``opt``, prints the control flow graph into a
128``.dot`` graph.  This graph can then be processed with the :program:`dot` tool
129to convert it to postscript or some other suitable format.
130
131``-dot-cfg-only``: Print CFG of function to "dot" file (with no function bodies)
132--------------------------------------------------------------------------------
133
134This pass, only available in ``opt``, prints the control flow graph into a
135``.dot`` graph, omitting the function bodies.  This graph can then be processed
136with the :program:`dot` tool to convert it to postscript or some other suitable
137format.
138
139``-dot-dom``: Print dominance tree of function to "dot" file
140------------------------------------------------------------
141
142This pass, only available in ``opt``, prints the dominator tree into a ``.dot``
143graph.  This graph can then be processed with the :program:`dot` tool to
144convert it to postscript or some other suitable format.
145
146``-dot-dom-only``: Print dominance tree of function to "dot" file (with no function bodies)
147-------------------------------------------------------------------------------------------
148
149This pass, only available in ``opt``, prints the dominator tree into a ``.dot``
150graph, omitting the function bodies.  This graph can then be processed with the
151:program:`dot` tool to convert it to postscript or some other suitable format.
152
153``-dot-postdom``: Print postdominance tree of function to "dot" file
154--------------------------------------------------------------------
155
156This pass, only available in ``opt``, prints the post dominator tree into a
157``.dot`` graph.  This graph can then be processed with the :program:`dot` tool
158to convert it to postscript or some other suitable format.
159
160``-dot-postdom-only``: Print postdominance tree of function to "dot" file (with no function bodies)
161---------------------------------------------------------------------------------------------------
162
163This pass, only available in ``opt``, prints the post dominator tree into a
164``.dot`` graph, omitting the function bodies.  This graph can then be processed
165with the :program:`dot` tool to convert it to postscript or some other suitable
166format.
167
168``-globalsmodref-aa``: Simple mod/ref analysis for globals
169----------------------------------------------------------
170
171This simple pass provides alias and mod/ref information for global values that
172do not have their address taken, and keeps track of whether functions read or
173write memory (are "pure").  For this simple (but very common) case, we can
174provide pretty accurate and useful information.
175
176``-instcount``: Counts the various types of ``Instruction``\ s
177--------------------------------------------------------------
178
179This pass collects the count of all instructions and reports them.
180
181``-intervals``: Interval Partition Construction
182-----------------------------------------------
183
184This analysis calculates and represents the interval partition of a function,
185or a preexisting interval partition.
186
187In this way, the interval partition may be used to reduce a flow graph down to
188its degenerate single node interval partition (unless it is irreducible).
189
190``-iv-users``: Induction Variable Users
191---------------------------------------
192
193Bookkeeping for "interesting" users of expressions computed from induction
194variables.
195
196``-lazy-value-info``: Lazy Value Information Analysis
197-----------------------------------------------------
198
199Interface for lazy computation of value constraint information.
200
201``-libcall-aa``: LibCall Alias Analysis
202---------------------------------------
203
204LibCall Alias Analysis.
205
206``-lint``: Statically lint-checks LLVM IR
207-----------------------------------------
208
209This pass statically checks for common and easily-identified constructs which
210produce undefined or likely unintended behavior in LLVM IR.
211
212It is not a guarantee of correctness, in two ways.  First, it isn't
213comprehensive.  There are checks which could be done statically which are not
214yet implemented.  Some of these are indicated by TODO comments, but those
215aren't comprehensive either.  Second, many conditions cannot be checked
216statically.  This pass does no dynamic instrumentation, so it can't check for
217all possible problems.
218
219Another limitation is that it assumes all code will be executed.  A store
220through a null pointer in a basic block which is never reached is harmless, but
221this pass will warn about it anyway.
222
223Optimization passes may make conditions that this pass checks for more or less
224obvious.  If an optimization pass appears to be introducing a warning, it may
225be that the optimization pass is merely exposing an existing condition in the
226code.
227
228This code may be run before :ref:`instcombine <passes-instcombine>`.  In many
229cases, instcombine checks for the same kinds of things and turns instructions
230with undefined behavior into unreachable (or equivalent).  Because of this,
231this pass makes some effort to look through bitcasts and so on.
232
233``-loops``: Natural Loop Information
234------------------------------------
235
236This analysis is used to identify natural loops and determine the loop depth of
237various nodes of the CFG.  Note that the loops identified may actually be
238several natural loops that share the same header node... not just a single
239natural loop.
240
241``-memdep``: Memory Dependence Analysis
242---------------------------------------
243
244An analysis that determines, for a given memory operation, what preceding
245memory operations it depends on.  It builds on alias analysis information, and
246tries to provide a lazy, caching interface to a common kind of alias
247information query.
248
249``-module-debuginfo``: Decodes module-level debug info
250------------------------------------------------------
251
252This pass decodes the debug info metadata in a module and prints in a
253(sufficiently-prepared-) human-readable form.
254
255For example, run this pass from ``opt`` along with the ``-analyze`` option, and
256it'll print to standard output.
257
258``-postdomfrontier``: Post-Dominance Frontier Construction
259----------------------------------------------------------
260
261This pass is a simple post-dominator construction algorithm for finding
262post-dominator frontiers.
263
264``-postdomtree``: Post-Dominator Tree Construction
265--------------------------------------------------
266
267This pass is a simple post-dominator construction algorithm for finding
268post-dominators.
269
270``-print-alias-sets``: Alias Set Printer
271----------------------------------------
272
273Yet to be written.
274
275``-print-callgraph``: Print a call graph
276----------------------------------------
277
278This pass, only available in ``opt``, prints the call graph to standard error
279in a human-readable form.
280
281``-print-callgraph-sccs``: Print SCCs of the Call Graph
282-------------------------------------------------------
283
284This pass, only available in ``opt``, prints the SCCs of the call graph to
285standard error in a human-readable form.
286
287``-print-cfg-sccs``: Print SCCs of each function CFG
288----------------------------------------------------
289
290This pass, only available in ``opt``, printsthe SCCs of each function CFG to
291standard error in a human-readable fom.
292
293``-print-dom-info``: Dominator Info Printer
294-------------------------------------------
295
296Dominator Info Printer.
297
298``-print-externalfnconstants``: Print external fn callsites passed constants
299----------------------------------------------------------------------------
300
301This pass, only available in ``opt``, prints out call sites to external
302functions that are called with constant arguments.  This can be useful when
303looking for standard library functions we should constant fold or handle in
304alias analyses.
305
306``-print-function``: Print function to stderr
307---------------------------------------------
308
309The ``PrintFunctionPass`` class is designed to be pipelined with other
310``FunctionPasses``, and prints out the functions of the module as they are
311processed.
312
313``-print-module``: Print module to stderr
314-----------------------------------------
315
316This pass simply prints out the entire module when it is executed.
317
318.. _passes-print-used-types:
319
320``-print-used-types``: Find Used Types
321--------------------------------------
322
323This pass is used to seek out all of the types in use by the program.  Note
324that this analysis explicitly does not include types only used by the symbol
325table.
326
327``-regions``: Detect single entry single exit regions
328-----------------------------------------------------
329
330The ``RegionInfo`` pass detects single entry single exit regions in a function,
331where a region is defined as any subgraph that is connected to the remaining
332graph at only two spots.  Furthermore, a hierarchical region tree is built.
333
334.. _passes-scalar-evolution:
335
336``-scalar-evolution``: Scalar Evolution Analysis
337------------------------------------------------
338
339The ``ScalarEvolution`` analysis can be used to analyze and catagorize scalar
340expressions in loops.  It specializes in recognizing general induction
341variables, representing them with the abstract and opaque ``SCEV`` class.
342Given this analysis, trip counts of loops and other important properties can be
343obtained.
344
345This analysis is primarily useful for induction variable substitution and
346strength reduction.
347
348``-scev-aa``: ScalarEvolution-based Alias Analysis
349--------------------------------------------------
350
351Simple alias analysis implemented in terms of ``ScalarEvolution`` queries.
352
353This differs from traditional loop dependence analysis in that it tests for
354dependencies within a single iteration of a loop, rather than dependencies
355between different iterations.
356
357``ScalarEvolution`` has a more complete understanding of pointer arithmetic
358than ``BasicAliasAnalysis``' collection of ad-hoc analyses.
359
360``-stack-safety``: Stack Safety Analysis
361------------------------------------------------
362
363The ``StackSafety`` analysis can be used to determine if stack allocated
364variables can be considered safe from memory access bugs.
365
366This analysis' primary purpose is to be used by sanitizers to avoid unnecessary
367instrumentation of safe variables.
368
369``-targetdata``: Target Data Layout
370-----------------------------------
371
372Provides other passes access to information on how the size and alignment
373required by the target ABI for various data types.
374
375Transform Passes
376================
377
378This section describes the LLVM Transform Passes.
379
380``-adce``: Aggressive Dead Code Elimination
381-------------------------------------------
382
383ADCE aggressively tries to eliminate code.  This pass is similar to :ref:`DCE
384<passes-dce>` but it assumes that values are dead until proven otherwise.  This
385is similar to :ref:`SCCP <passes-sccp>`, except applied to the liveness of
386values.
387
388``-always-inline``: Inliner for ``always_inline`` functions
389-----------------------------------------------------------
390
391A custom inliner that handles only functions that are marked as "always
392inline".
393
394``-argpromotion``: Promote 'by reference' arguments to scalars
395--------------------------------------------------------------
396
397This pass promotes "by reference" arguments to be "by value" arguments.  In
398practice, this means looking for internal functions that have pointer
399arguments.  If it can prove, through the use of alias analysis, that an
400argument is *only* loaded, then it can pass the value into the function instead
401of the address of the value.  This can cause recursive simplification of code
402and lead to the elimination of allocas (especially in C++ template code like
403the STL).
404
405This pass also handles aggregate arguments that are passed into a function,
406scalarizing them if the elements of the aggregate are only loaded.  Note that
407it refuses to scalarize aggregates which would require passing in more than
408three operands to the function, because passing thousands of operands for a
409large array or structure is unprofitable!
410
411Note that this transformation could also be done for arguments that are only
412stored to (returning the value instead), but does not currently.  This case
413would be best handled when and if LLVM starts supporting multiple return values
414from functions.
415
416``-bb-vectorize``: Basic-Block Vectorization
417--------------------------------------------
418
419This pass combines instructions inside basic blocks to form vector
420instructions.  It iterates over each basic block, attempting to pair compatible
421instructions, repeating this process until no additional pairs are selected for
422vectorization.  When the outputs of some pair of compatible instructions are
423used as inputs by some other pair of compatible instructions, those pairs are
424part of a potential vectorization chain.  Instruction pairs are only fused into
425vector instructions when they are part of a chain longer than some threshold
426length.  Moreover, the pass attempts to find the best possible chain for each
427pair of compatible instructions.  These heuristics are intended to prevent
428vectorization in cases where it would not yield a performance increase of the
429resulting code.
430
431``-block-placement``: Profile Guided Basic Block Placement
432----------------------------------------------------------
433
434This pass is a very simple profile guided basic block placement algorithm.  The
435idea is to put frequently executed blocks together at the start of the function
436and hopefully increase the number of fall-through conditional branches.  If
437there is no profile information for a particular function, this pass basically
438orders blocks in depth-first order.
439
440``-break-crit-edges``: Break critical edges in CFG
441--------------------------------------------------
442
443Break all of the critical edges in the CFG by inserting a dummy basic block.
444It may be "required" by passes that cannot deal with critical edges.  This
445transformation obviously invalidates the CFG, but can update forward dominator
446(set, immediate dominators, tree, and frontier) information.
447
448``-codegenprepare``: Optimize for code generation
449-------------------------------------------------
450
451This pass munges the code in the input function to better prepare it for
452SelectionDAG-based code generation.  This works around limitations in its
453basic-block-at-a-time approach.  It should eventually be removed.
454
455``-constmerge``: Merge Duplicate Global Constants
456-------------------------------------------------
457
458Merges duplicate global constants together into a single constant that is
459shared.  This is useful because some passes (i.e., TraceValues) insert a lot of
460string constants into the program, regardless of whether or not an existing
461string is available.
462
463``-constprop``: Simple constant propagation
464-------------------------------------------
465
466This pass implements constant propagation and merging.  It looks for
467instructions involving only constant operands and replaces them with a constant
468value instead of an instruction.  For example:
469
470.. code-block:: llvm
471
472  add i32 1, 2
473
474becomes
475
476.. code-block:: llvm
477
478  i32 3
479
480NOTE: this pass has a habit of making definitions be dead.  It is a good idea
481to run a :ref:`Dead Instruction Elimination <passes-die>` pass sometime after
482running this pass.
483
484.. _passes-dce:
485
486``-dce``: Dead Code Elimination
487-------------------------------
488
489Dead code elimination is similar to :ref:`dead instruction elimination
490<passes-die>`, but it rechecks instructions that were used by removed
491instructions to see if they are newly dead.
492
493``-deadargelim``: Dead Argument Elimination
494-------------------------------------------
495
496This pass deletes dead arguments from internal functions.  Dead argument
497elimination removes arguments which are directly dead, as well as arguments
498only passed into function calls as dead arguments of other functions.  This
499pass also deletes dead arguments in a similar way.
500
501This pass is often useful as a cleanup pass to run after aggressive
502interprocedural passes, which add possibly-dead arguments.
503
504``-deadtypeelim``: Dead Type Elimination
505----------------------------------------
506
507This pass is used to cleanup the output of GCC.  It eliminate names for types
508that are unused in the entire translation unit, using the :ref:`find used types
509<passes-print-used-types>` pass.
510
511.. _passes-die:
512
513``-die``: Dead Instruction Elimination
514--------------------------------------
515
516Dead instruction elimination performs a single pass over the function, removing
517instructions that are obviously dead.
518
519``-dse``: Dead Store Elimination
520--------------------------------
521
522A trivial dead store elimination that only considers basic-block local
523redundant stores.
524
525.. _passes-functionattrs:
526
527``-functionattrs``: Deduce function attributes
528----------------------------------------------
529
530A simple interprocedural pass which walks the call-graph, looking for functions
531which do not access or only read non-local memory, and marking them
532``readnone``/``readonly``.  In addition, it marks function arguments (of
533pointer type) "``nocapture``" if a call to the function does not create any
534copies of the pointer value that outlive the call.  This more or less means
535that the pointer is only dereferenced, and not returned from the function or
536stored in a global.  This pass is implemented as a bottom-up traversal of the
537call-graph.
538
539``-globaldce``: Dead Global Elimination
540---------------------------------------
541
542This transform is designed to eliminate unreachable internal globals from the
543program.  It uses an aggressive algorithm, searching out globals that are known
544to be alive.  After it finds all of the globals which are needed, it deletes
545whatever is left over.  This allows it to delete recursive chunks of the
546program which are unreachable.
547
548``-globalopt``: Global Variable Optimizer
549-----------------------------------------
550
551This pass transforms simple global variables that never have their address
552taken.  If obviously true, it marks read/write globals as constant, deletes
553variables only stored to, etc.
554
555``-gvn``: Global Value Numbering
556--------------------------------
557
558This pass performs global value numbering to eliminate fully and partially
559redundant instructions.  It also performs redundant load elimination.
560
561.. _passes-indvars:
562
563``-indvars``: Canonicalize Induction Variables
564----------------------------------------------
565
566This transformation analyzes and transforms the induction variables (and
567computations derived from them) into simpler forms suitable for subsequent
568analysis and transformation.
569
570This transformation makes the following changes to each loop with an
571identifiable induction variable:
572
573* All loops are transformed to have a *single* canonical induction variable
574  which starts at zero and steps by one.
575* The canonical induction variable is guaranteed to be the first PHI node in
576  the loop header block.
577* Any pointer arithmetic recurrences are raised to use array subscripts.
578
579If the trip count of a loop is computable, this pass also makes the following
580changes:
581
582* The exit condition for the loop is canonicalized to compare the induction
583  value against the exit value.  This turns loops like:
584
585  .. code-block:: c++
586
587    for (i = 7; i*i < 1000; ++i)
588
589    into
590
591  .. code-block:: c++
592
593    for (i = 0; i != 25; ++i)
594
595* Any use outside of the loop of an expression derived from the indvar is
596  changed to compute the derived value outside of the loop, eliminating the
597  dependence on the exit value of the induction variable.  If the only purpose
598  of the loop is to compute the exit value of some derived expression, this
599  transformation will make the loop dead.
600
601This transformation should be followed by strength reduction after all of the
602desired loop transformations have been performed.  Additionally, on targets
603where it is profitable, the loop could be transformed to count down to zero
604(the "do loop" optimization).
605
606``-inline``: Function Integration/Inlining
607------------------------------------------
608
609Bottom-up inlining of functions into callees.
610
611.. _passes-instcombine:
612
613``-instcombine``: Combine redundant instructions
614------------------------------------------------
615
616Combine instructions to form fewer, simple instructions.  This pass does not
617modify the CFG. This pass is where algebraic simplification happens.
618
619This pass combines things like:
620
621.. code-block:: llvm
622
623  %Y = add i32 %X, 1
624  %Z = add i32 %Y, 1
625
626into:
627
628.. code-block:: llvm
629
630  %Z = add i32 %X, 2
631
632This is a simple worklist driven algorithm.
633
634This pass guarantees that the following canonicalizations are performed on the
635program:
636
637#. If a binary operator has a constant operand, it is moved to the right-hand
638   side.
639#. Bitwise operators with constant operands are always grouped so that shifts
640   are performed first, then ``or``\ s, then ``and``\ s, then ``xor``\ s.
641#. Compare instructions are converted from ``<``, ``>``, ``≤``, or ``≥`` to
642   ``=`` or ``≠`` if possible.
643#. All ``cmp`` instructions on boolean values are replaced with logical
644   operations.
645#. ``add X, X`` is represented as ``mul X, 2`` ⇒ ``shl X, 1``
646#. Multiplies with a constant power-of-two argument are transformed into
647   shifts.
648#. … etc.
649
650This pass can also simplify calls to specific well-known function calls (e.g.
651runtime library functions).  For example, a call ``exit(3)`` that occurs within
652the ``main()`` function can be transformed into simply ``return 3``. Whether or
653not library calls are simplified is controlled by the
654:ref:`-functionattrs <passes-functionattrs>` pass and LLVM's knowledge of
655library calls on different targets.
656
657.. _passes-aggressive-instcombine:
658
659``-aggressive-instcombine``: Combine expression patterns
660--------------------------------------------------------
661
662Combine expression patterns to form expressions with fewer, simple instructions.
663This pass does not modify the CFG.
664
665For example, this pass reduce width of expressions post-dominated by TruncInst
666into smaller width when applicable.
667
668It differs from instcombine pass in that it contains pattern optimization that
669requires higher complexity than the O(1), thus, it should run fewer times than
670instcombine pass.
671
672``-internalize``: Internalize Global Symbols
673--------------------------------------------
674
675This pass loops over all of the functions in the input module, looking for a
676main function.  If a main function is found, all other functions and all global
677variables with initializers are marked as internal.
678
679``-ipconstprop``: Interprocedural constant propagation
680------------------------------------------------------
681
682This pass implements an *extremely* simple interprocedural constant propagation
683pass.  It could certainly be improved in many different ways, like using a
684worklist.  This pass makes arguments dead, but does not remove them.  The
685existing dead argument elimination pass should be run after this to clean up
686the mess.
687
688``-ipsccp``: Interprocedural Sparse Conditional Constant Propagation
689--------------------------------------------------------------------
690
691An interprocedural variant of :ref:`Sparse Conditional Constant Propagation
692<passes-sccp>`.
693
694``-jump-threading``: Jump Threading
695-----------------------------------
696
697Jump threading tries to find distinct threads of control flow running through a
698basic block.  This pass looks at blocks that have multiple predecessors and
699multiple successors.  If one or more of the predecessors of the block can be
700proven to always cause a jump to one of the successors, we forward the edge
701from the predecessor to the successor by duplicating the contents of this
702block.
703
704An example of when this can occur is code like this:
705
706.. code-block:: c++
707
708  if () { ...
709    X = 4;
710  }
711  if (X < 3) {
712
713In this case, the unconditional branch at the end of the first if can be
714revectored to the false side of the second if.
715
716.. _passes-lcssa:
717
718``-lcssa``: Loop-Closed SSA Form Pass
719-------------------------------------
720
721This pass transforms loops by placing phi nodes at the end of the loops for all
722values that are live across the loop boundary.  For example, it turns the left
723into the right code:
724
725.. code-block:: c++
726
727  for (...)                for (...)
728      if (c)                   if (c)
729          X1 = ...                 X1 = ...
730      else                     else
731          X2 = ...                 X2 = ...
732      X3 = phi(X1, X2)         X3 = phi(X1, X2)
733  ... = X3 + 4              X4 = phi(X3)
734                              ... = X4 + 4
735
736This is still valid LLVM; the extra phi nodes are purely redundant, and will be
737trivially eliminated by ``InstCombine``.  The major benefit of this
738transformation is that it makes many other loop optimizations, such as
739``LoopUnswitch``\ ing, simpler. You can read more in the
740:ref:`loop terminology section for the LCSSA form <loop-terminology-lcssa>`.
741
742.. _passes-licm:
743
744``-licm``: Loop Invariant Code Motion
745-------------------------------------
746
747This pass performs loop invariant code motion, attempting to remove as much
748code from the body of a loop as possible.  It does this by either hoisting code
749into the preheader block, or by sinking code to the exit blocks if it is safe.
750This pass also promotes must-aliased memory locations in the loop to live in
751registers, thus hoisting and sinking "invariant" loads and stores.
752
753This pass uses alias analysis for two purposes:
754
755#. Moving loop invariant loads and calls out of loops.  If we can determine
756   that a load or call inside of a loop never aliases anything stored to, we
757   can hoist it or sink it like any other instruction.
758
759#. Scalar Promotion of Memory.  If there is a store instruction inside of the
760   loop, we try to move the store to happen AFTER the loop instead of inside of
761   the loop.  This can only happen if a few conditions are true:
762
763   #. The pointer stored through is loop invariant.
764   #. There are no stores or loads in the loop which *may* alias the pointer.
765      There are no calls in the loop which mod/ref the pointer.
766
767   If these conditions are true, we can promote the loads and stores in the
768   loop of the pointer to use a temporary alloca'd variable.  We then use the
769   :ref:`mem2reg <passes-mem2reg>` functionality to construct the appropriate
770   SSA form for the variable.
771
772``-loop-deletion``: Delete dead loops
773-------------------------------------
774
775This file implements the Dead Loop Deletion Pass.  This pass is responsible for
776eliminating loops with non-infinite computable trip counts that have no side
777effects or volatile instructions, and do not contribute to the computation of
778the function's return value.
779
780.. _passes-loop-extract:
781
782``-loop-extract``: Extract loops into new functions
783---------------------------------------------------
784
785A pass wrapper around the ``ExtractLoop()`` scalar transformation to extract
786each top-level loop into its own new function.  If the loop is the *only* loop
787in a given function, it is not touched.  This is a pass most useful for
788debugging via bugpoint.
789
790``-loop-extract-single``: Extract at most one loop into a new function
791----------------------------------------------------------------------
792
793Similar to :ref:`Extract loops into new functions <passes-loop-extract>`, this
794pass extracts one natural loop from the program into a function if it can.
795This is used by :program:`bugpoint`.
796
797``-loop-reduce``: Loop Strength Reduction
798-----------------------------------------
799
800This pass performs a strength reduction on array references inside loops that
801have as one or more of their components the loop induction variable.  This is
802accomplished by creating a new value to hold the initial value of the array
803access for the first iteration, and then creating a new GEP instruction in the
804loop to increment the value by the appropriate amount.
805
806.. _passes-loop-rotate:
807
808``-loop-rotate``: Rotate Loops
809------------------------------
810
811A simple loop rotation transformation.  A summary of it can be found in
812:ref:`Loop Terminology for Rotated Loops <loop-terminology-loop-rotate>`.
813
814
815.. _passes-loop-simplify:
816
817``-loop-simplify``: Canonicalize natural loops
818----------------------------------------------
819
820This pass performs several transformations to transform natural loops into a
821simpler form, which makes subsequent analyses and transformations simpler and
822more effective. A summary of it can be found in
823:ref:`Loop Terminology, Loop Simplify Form <loop-terminology-loop-simplify>`.
824
825Loop pre-header insertion guarantees that there is a single, non-critical entry
826edge from outside of the loop to the loop header.  This simplifies a number of
827analyses and transformations, such as :ref:`LICM <passes-licm>`.
828
829Loop exit-block insertion guarantees that all exit blocks from the loop (blocks
830which are outside of the loop that have predecessors inside of the loop) only
831have predecessors from inside of the loop (and are thus dominated by the loop
832header).  This simplifies transformations such as store-sinking that are built
833into LICM.
834
835This pass also guarantees that loops will have exactly one backedge.
836
837Note that the :ref:`simplifycfg <passes-simplifycfg>` pass will clean up blocks
838which are split out but end up being unnecessary, so usage of this pass should
839not pessimize generated code.
840
841This pass obviously modifies the CFG, but updates loop information and
842dominator information.
843
844``-loop-unroll``: Unroll loops
845------------------------------
846
847This pass implements a simple loop unroller.  It works best when loops have
848been canonicalized by the :ref:`indvars <passes-indvars>` pass, allowing it to
849determine the trip counts of loops easily.
850
851``-loop-unroll-and-jam``: Unroll and Jam loops
852----------------------------------------------
853
854This pass implements a simple unroll and jam classical loop optimisation pass.
855It transforms loop from:
856
857.. code-block:: c++
858
859  for i.. i+= 1              for i.. i+= 4
860    for j..                    for j..
861      code(i, j)                 code(i, j)
862                                 code(i+1, j)
863                                 code(i+2, j)
864                                 code(i+3, j)
865                             remainder loop
866
867Which can be seen as unrolling the outer loop and "jamming" (fusing) the inner
868loops into one. When variables or loads can be shared in the new inner loop, this
869can lead to significant performance improvements. It uses
870:ref:`Dependence Analysis <passes-da>` for proving the transformations are safe.
871
872.. _passes-loop-unswitch:
873
874``-loop-unswitch``: Unswitch loops
875----------------------------------
876
877This pass transforms loops that contain branches on loop-invariant conditions
878to have multiple loops.  For example, it turns the left into the right code:
879
880.. code-block:: c++
881
882  for (...)                  if (lic)
883      A                          for (...)
884      if (lic)                       A; B; C
885          B                  else
886      C                          for (...)
887                                     A; C
888
889This can increase the size of the code exponentially (doubling it every time a
890loop is unswitched) so we only unswitch if the resultant code will be smaller
891than a threshold.
892
893This pass expects :ref:`LICM <passes-licm>` to be run before it to hoist
894invariant conditions out of the loop, to make the unswitching opportunity
895obvious.
896
897``-loweratomic``: Lower atomic intrinsics to non-atomic form
898------------------------------------------------------------
899
900This pass lowers atomic intrinsics to non-atomic form for use in a known
901non-preemptible environment.
902
903The pass does not verify that the environment is non-preemptible (in general
904this would require knowledge of the entire call graph of the program including
905any libraries which may not be available in bitcode form); it simply lowers
906every atomic intrinsic.
907
908``-lowerinvoke``: Lower invokes to calls, for unwindless code generators
909------------------------------------------------------------------------
910
911This transformation is designed for use by code generators which do not yet
912support stack unwinding.  This pass converts ``invoke`` instructions to
913``call`` instructions, so that any exception-handling ``landingpad`` blocks
914become dead code (which can be removed by running the ``-simplifycfg`` pass
915afterwards).
916
917``-lowerswitch``: Lower ``SwitchInst``\ s to branches
918-----------------------------------------------------
919
920Rewrites switch instructions with a sequence of branches, which allows targets
921to get away with not implementing the switch instruction until it is
922convenient.
923
924.. _passes-mem2reg:
925
926``-mem2reg``: Promote Memory to Register
927----------------------------------------
928
929This file promotes memory references to be register references.  It promotes
930alloca instructions which only have loads and stores as uses.  An ``alloca`` is
931transformed by using dominator frontiers to place phi nodes, then traversing
932the function in depth-first order to rewrite loads and stores as appropriate.
933This is just the standard SSA construction algorithm to construct "pruned" SSA
934form.
935
936``-memcpyopt``: MemCpy Optimization
937-----------------------------------
938
939This pass performs various transformations related to eliminating ``memcpy``
940calls, or transforming sets of stores into ``memset``\ s.
941
942``-mergefunc``: Merge Functions
943-------------------------------
944
945This pass looks for equivalent functions that are mergable and folds them.
946
947Total-ordering is introduced among the functions set: we define comparison
948that answers for every two functions which of them is greater. It allows to
949arrange functions into the binary tree.
950
951For every new function we check for equivalent in tree.
952
953If equivalent exists we fold such functions. If both functions are overridable,
954we move the functionality into a new internal function and leave two
955overridable thunks to it.
956
957If there is no equivalent, then we add this function to tree.
958
959Lookup routine has O(log(n)) complexity, while whole merging process has
960complexity of O(n*log(n)).
961
962Read
963:doc:`this <MergeFunctions>`
964article for more details.
965
966``-mergereturn``: Unify function exit nodes
967-------------------------------------------
968
969Ensure that functions have at most one ``ret`` instruction in them.
970Additionally, it keeps track of which node is the new exit node of the CFG.
971
972``-partial-inliner``: Partial Inliner
973-------------------------------------
974
975This pass performs partial inlining, typically by inlining an ``if`` statement
976that surrounds the body of the function.
977
978``-prune-eh``: Remove unused exception handling info
979----------------------------------------------------
980
981This file implements a simple interprocedural pass which walks the call-graph,
982turning invoke instructions into call instructions if and only if the callee
983cannot throw an exception.  It implements this as a bottom-up traversal of the
984call-graph.
985
986``-reassociate``: Reassociate expressions
987-----------------------------------------
988
989This pass reassociates commutative expressions in an order that is designed to
990promote better constant propagation, GCSE, :ref:`LICM <passes-licm>`, PRE, etc.
991
992For example: 4 + (x + 5) ⇒ x + (4 + 5)
993
994In the implementation of this algorithm, constants are assigned rank = 0,
995function arguments are rank = 1, and other values are assigned ranks
996corresponding to the reverse post order traversal of current function (starting
997at 2), which effectively gives values in deep loops higher rank than values not
998in loops.
999
1000``-reg2mem``: Demote all values to stack slots
1001----------------------------------------------
1002
1003This file demotes all registers to memory references.  It is intended to be the
1004inverse of :ref:`mem2reg <passes-mem2reg>`.  By converting to ``load``
1005instructions, the only values live across basic blocks are ``alloca``
1006instructions and ``load`` instructions before ``phi`` nodes.  It is intended
1007that this should make CFG hacking much easier.  To make later hacking easier,
1008the entry block is split into two, such that all introduced ``alloca``
1009instructions (and nothing else) are in the entry block.
1010
1011``-sroa``: Scalar Replacement of Aggregates
1012------------------------------------------------------
1013
1014The well-known scalar replacement of aggregates transformation.  This transform
1015breaks up ``alloca`` instructions of aggregate type (structure or array) into
1016individual ``alloca`` instructions for each member if possible.  Then, if
1017possible, it transforms the individual ``alloca`` instructions into nice clean
1018scalar SSA form.
1019
1020.. _passes-sccp:
1021
1022``-sccp``: Sparse Conditional Constant Propagation
1023--------------------------------------------------
1024
1025Sparse conditional constant propagation and merging, which can be summarized
1026as:
1027
1028* Assumes values are constant unless proven otherwise
1029* Assumes BasicBlocks are dead unless proven otherwise
1030* Proves values to be constant, and replaces them with constants
1031* Proves conditional branches to be unconditional
1032
1033Note that this pass has a habit of making definitions be dead.  It is a good
1034idea to run a :ref:`DCE <passes-dce>` pass sometime after running this pass.
1035
1036.. _passes-simplifycfg:
1037
1038``-simplifycfg``: Simplify the CFG
1039----------------------------------
1040
1041Performs dead code elimination and basic block merging.  Specifically:
1042
1043* Removes basic blocks with no predecessors.
1044* Merges a basic block into its predecessor if there is only one and the
1045  predecessor only has one successor.
1046* Eliminates PHI nodes for basic blocks with a single predecessor.
1047* Eliminates a basic block that only contains an unconditional branch.
1048
1049``-sink``: Code sinking
1050-----------------------
1051
1052This pass moves instructions into successor blocks, when possible, so that they
1053aren't executed on paths where their results aren't needed.
1054
1055``-strip``: Strip all symbols from a module
1056-------------------------------------------
1057
1058Performs code stripping.  This transformation can delete:
1059
1060* names for virtual registers
1061* symbols for internal globals and functions
1062* debug information
1063
1064Note that this transformation makes code much less readable, so it should only
1065be used in situations where the strip utility would be used, such as reducing
1066code size or making it harder to reverse engineer code.
1067
1068``-strip-dead-debug-info``: Strip debug info for unused symbols
1069---------------------------------------------------------------
1070
1071.. FIXME: this description is the same as for -strip
1072
1073performs code stripping. this transformation can delete:
1074
1075* names for virtual registers
1076* symbols for internal globals and functions
1077* debug information
1078
1079note that this transformation makes code much less readable, so it should only
1080be used in situations where the strip utility would be used, such as reducing
1081code size or making it harder to reverse engineer code.
1082
1083``-strip-dead-prototypes``: Strip Unused Function Prototypes
1084------------------------------------------------------------
1085
1086This pass loops over all of the functions in the input module, looking for dead
1087declarations and removes them.  Dead declarations are declarations of functions
1088for which no implementation is available (i.e., declarations for unused library
1089functions).
1090
1091``-strip-debug-declare``: Strip all ``llvm.dbg.declare`` intrinsics
1092-------------------------------------------------------------------
1093
1094.. FIXME: this description is the same as for -strip
1095
1096This pass implements code stripping.  Specifically, it can delete:
1097
1098#. names for virtual registers
1099#. symbols for internal globals and functions
1100#. debug information
1101
1102Note that this transformation makes code much less readable, so it should only
1103be used in situations where the 'strip' utility would be used, such as reducing
1104code size or making it harder to reverse engineer code.
1105
1106``-strip-nondebug``: Strip all symbols, except dbg symbols, from a module
1107-------------------------------------------------------------------------
1108
1109.. FIXME: this description is the same as for -strip
1110
1111This pass implements code stripping.  Specifically, it can delete:
1112
1113#. names for virtual registers
1114#. symbols for internal globals and functions
1115#. debug information
1116
1117Note that this transformation makes code much less readable, so it should only
1118be used in situations where the 'strip' utility would be used, such as reducing
1119code size or making it harder to reverse engineer code.
1120
1121``-tailcallelim``: Tail Call Elimination
1122----------------------------------------
1123
1124This file transforms calls of the current function (self recursion) followed by
1125a return instruction with a branch to the entry of the function, creating a
1126loop.  This pass also implements the following extensions to the basic
1127algorithm:
1128
1129#. Trivial instructions between the call and return do not prevent the
1130   transformation from taking place, though currently the analysis cannot
1131   support moving any really useful instructions (only dead ones).
1132#. This pass transforms functions that are prevented from being tail recursive
1133   by an associative expression to use an accumulator variable, thus compiling
1134   the typical naive factorial or fib implementation into efficient code.
1135#. TRE is performed if the function returns void, if the return returns the
1136   result returned by the call, or if the function returns a run-time constant
1137   on all exits from the function.  It is possible, though unlikely, that the
1138   return returns something else (like constant 0), and can still be TRE'd.  It
1139   can be TRE'd if *all other* return instructions in the function return the
1140   exact same value.
1141#. If it can prove that callees do not access theier caller stack frame, they
1142   are marked as eligible for tail call elimination (by the code generator).
1143
1144Utility Passes
1145==============
1146
1147This section describes the LLVM Utility Passes.
1148
1149``-deadarghaX0r``: Dead Argument Hacking (BUGPOINT USE ONLY; DO NOT USE)
1150------------------------------------------------------------------------
1151
1152Same as dead argument elimination, but deletes arguments to functions which are
1153external.  This is only for use by :doc:`bugpoint <Bugpoint>`.
1154
1155``-extract-blocks``: Extract Basic Blocks From Module (for bugpoint use)
1156------------------------------------------------------------------------
1157
1158This pass is used by bugpoint to extract all blocks from the module into their
1159own functions.
1160
1161``-instnamer``: Assign names to anonymous instructions
1162------------------------------------------------------
1163
1164This is a little utility pass that gives instructions names, this is mostly
1165useful when diffing the effect of an optimization because deleting an unnamed
1166instruction can change all other instruction numbering, making the diff very
1167noisy.
1168
1169.. _passes-verify:
1170
1171``-verify``: Module Verifier
1172----------------------------
1173
1174Verifies an LLVM IR code.  This is useful to run after an optimization which is
1175undergoing testing.  Note that llvm-as verifies its input before emitting
1176bitcode, and also that malformed bitcode is likely to make LLVM crash.  All
1177language front-ends are therefore encouraged to verify their output before
1178performing optimizing transformations.
1179
1180#. Both of a binary operator's parameters are of the same type.
1181#. Verify that the indices of mem access instructions match other operands.
1182#. Verify that arithmetic and other things are only performed on first-class
1183   types.  Verify that shifts and logicals only happen on integrals f.e.
1184#. All of the constants in a switch statement are of the correct type.
1185#. The code is in valid SSA form.
1186#. It is illegal to put a label into any other type (like a structure) or to
1187   return one.
1188#. Only phi nodes can be self referential: ``%x = add i32 %x``, ``%x`` is
1189   invalid.
1190#. PHI nodes must have an entry for each predecessor, with no extras.
1191#. PHI nodes must be the first thing in a basic block, all grouped together.
1192#. PHI nodes must have at least one entry.
1193#. All basic blocks should only end with terminator insts, not contain them.
1194#. The entry node to a function must not have predecessors.
1195#. All Instructions must be embedded into a basic block.
1196#. Functions cannot take a void-typed parameter.
1197#. Verify that a function's argument list agrees with its declared type.
1198#. It is illegal to specify a name for a void value.
1199#. It is illegal to have an internal global value with no initializer.
1200#. It is illegal to have a ``ret`` instruction that returns a value that does
1201   not agree with the function return value type.
1202#. Function call argument types match the function prototype.
1203#. All other things that are tested by asserts spread about the code.
1204
1205Note that this does not provide full security verification (like Java), but
1206instead just tries to ensure that code is well-formed.
1207
1208.. _passes-view-cfg:
1209
1210``-view-cfg``: View CFG of function
1211-----------------------------------
1212
1213Displays the control flow graph using the GraphViz tool.
1214
1215``-view-cfg-only``: View CFG of function (with no function bodies)
1216------------------------------------------------------------------
1217
1218Displays the control flow graph using the GraphViz tool, but omitting function
1219bodies.
1220
1221``-view-dom``: View dominance tree of function
1222----------------------------------------------
1223
1224Displays the dominator tree using the GraphViz tool.
1225
1226``-view-dom-only``: View dominance tree of function (with no function bodies)
1227-----------------------------------------------------------------------------
1228
1229Displays the dominator tree using the GraphViz tool, but omitting function
1230bodies.
1231
1232``-view-postdom``: View postdominance tree of function
1233------------------------------------------------------
1234
1235Displays the post dominator tree using the GraphViz tool.
1236
1237``-view-postdom-only``: View postdominance tree of function (with no function bodies)
1238-------------------------------------------------------------------------------------
1239
1240Displays the post dominator tree using the GraphViz tool, but omitting function
1241bodies.
1242
1243``-transform-warning``: Report missed forced transformations
1244------------------------------------------------------------
1245
1246Emits warnings about not yet applied forced transformations (e.g. from
1247``#pragma omp simd``).
1248