1 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN"
2 "http://www.w3.org/TR/html4/strict.dtd">
5 <title>LLVM's Analysis and Transform Passes</title>
6 <link rel="stylesheet" href="llvm.css" type="text/css">
7 <meta http-equiv="Content-Type" content="text/html; charset=UTF-8">
13 If Passes.html is up to date, the following "one-liner" should print
16 egrep -e '^<tr><td><a href="#.*">-.*</a></td><td>.*</td></tr>$' \
17 -e '^ <a name=".*">.*</a>$' < Passes.html >html; \
18 perl >help <<'EOT' && diff -u help html; rm -f help html
19 open HTML, "<Passes.html" or die "open: Passes.html: $!\n";
21 m:^<tr><td><a href="#(.*)">-.*</a></td><td>.*</td></tr>$: or next;
22 $order{$1} = sprintf("%03d", 1 + int %order);
24 open HELP, "../Release/bin/opt -help|" or die "open: opt -help: $!\n";
26 m:^ -([^ ]+) +- (.*)$: or next;
28 $o = "000" unless defined $o;
29 push @x, "$o<tr><td><a href=\"#$1\">-$1</a></td><td>$2</td></tr>\n";
30 push @y, "$o <a name=\"$1\">-$1: $2</a>\n";
32 @x = map { s/^\d\d\d//; $_ } sort @x;
33 @y = map { s/^\d\d\d//; $_ } sort @y;
37 This (real) one-liner can also be helpful when converting comments to HTML:
39 perl -e '$/ = 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'
43 <h1>LLVM's Analysis and Transform Passes</h1>
46 <li><a href="#intro">Introduction</a></li>
47 <li><a href="#analyses">Analysis Passes</a>
48 <li><a href="#transforms">Transform Passes</a></li>
49 <li><a href="#utilities">Utility Passes</a></li>
52 <div class="doc_author">
53 <p>Written by <a href="mailto:rspencer@x10sys.com">Reid Spencer</a>
54 and Gordon Henriksen</p>
57 <!-- ======================================================================= -->
58 <h2><a name="intro">Introduction</a></h2>
60 <p>This document serves as a high level summary of the optimization features
61 that LLVM provides. Optimizations are implemented as Passes that traverse some
62 portion of a program to either collect information or transform the program.
63 The table below divides the passes that LLVM provides into three categories.
64 Analysis passes compute information that other passes can use or for debugging
65 or program visualization purposes. Transform passes can use (or invalidate)
66 the analysis passes. Transform passes all mutate the program in some way.
67 Utility passes provides some utility but don't otherwise fit categorization.
68 For example passes to extract functions to bitcode or write a module to
69 bitcode are neither analysis nor transform passes.
70 <p>The table below provides a quick summary of each pass and links to the more
71 complete pass description later in the document.</p>
74 <tr><th colspan="2"><b>ANALYSIS PASSES</b></th></tr>
75 <tr><th>Option</th><th>Name</th></tr>
76 <tr><td><a href="#aa-eval">-aa-eval</a></td><td>Exhaustive Alias Analysis Precision Evaluator</td></tr>
77 <tr><td><a href="#basicaa">-basicaa</a></td><td>Basic Alias Analysis (stateless AA impl)</td></tr>
78 <tr><td><a href="#basiccg">-basiccg</a></td><td>Basic CallGraph Construction</td></tr>
79 <tr><td><a href="#count-aa">-count-aa</a></td><td>Count Alias Analysis Query Responses</td></tr>
80 <tr><td><a href="#debug-aa">-debug-aa</a></td><td>AA use debugger</td></tr>
81 <tr><td><a href="#domfrontier">-domfrontier</a></td><td>Dominance Frontier Construction</td></tr>
82 <tr><td><a href="#domtree">-domtree</a></td><td>Dominator Tree Construction</td></tr>
83 <tr><td><a href="#dot-callgraph">-dot-callgraph</a></td><td>Print Call Graph to 'dot' file</td></tr>
84 <tr><td><a href="#dot-cfg">-dot-cfg</a></td><td>Print CFG of function to 'dot' file</td></tr>
85 <tr><td><a href="#dot-cfg-only">-dot-cfg-only</a></td><td>Print CFG of function to 'dot' file (with no function bodies)</td></tr>
86 <tr><td><a href="#dot-dom">-dot-dom</a></td><td>Print dominance tree of function to 'dot' file</td></tr>
87 <tr><td><a href="#dot-dom-only">-dot-dom-only</a></td><td>Print dominance tree of function to 'dot' file (with no function bodies)</td></tr>
88 <tr><td><a href="#dot-postdom">-dot-postdom</a></td><td>Print postdominance tree of function to 'dot' file</td></tr>
89 <tr><td><a href="#dot-postdom-only">-dot-postdom-only</a></td><td>Print postdominance tree of function to 'dot' file (with no function bodies)</td></tr>
90 <tr><td><a href="#globalsmodref-aa">-globalsmodref-aa</a></td><td>Simple mod/ref analysis for globals</td></tr>
91 <tr><td><a href="#instcount">-instcount</a></td><td>Counts the various types of Instructions</td></tr>
92 <tr><td><a href="#intervals">-intervals</a></td><td>Interval Partition Construction</td></tr>
93 <tr><td><a href="#iv-users">-iv-users</a></td><td>Induction Variable Users</td></tr>
94 <tr><td><a href="#lazy-value-info">-lazy-value-info</a></td><td>Lazy Value Information Analysis</td></tr>
95 <tr><td><a href="#lda">-lda</a></td><td>Loop Dependence Analysis</td></tr>
96 <tr><td><a href="#libcall-aa">-libcall-aa</a></td><td>LibCall Alias Analysis</td></tr>
97 <tr><td><a href="#lint">-lint</a></td><td>Statically lint-checks LLVM IR</td></tr>
98 <tr><td><a href="#loops">-loops</a></td><td>Natural Loop Information</td></tr>
99 <tr><td><a href="#memdep">-memdep</a></td><td>Memory Dependence Analysis</td></tr>
100 <tr><td><a href="#module-debuginfo">-module-debuginfo</a></td><td>Decodes module-level debug info</td></tr>
101 <tr><td><a href="#no-aa">-no-aa</a></td><td>No Alias Analysis (always returns 'may' alias)</td></tr>
102 <tr><td><a href="#no-profile">-no-profile</a></td><td>No Profile Information</td></tr>
103 <tr><td><a href="#postdomfrontier">-postdomfrontier</a></td><td>Post-Dominance Frontier Construction</td></tr>
104 <tr><td><a href="#postdomtree">-postdomtree</a></td><td>Post-Dominator Tree Construction</td></tr>
105 <tr><td><a href="#print-alias-sets">-print-alias-sets</a></td><td>Alias Set Printer</td></tr>
106 <tr><td><a href="#print-callgraph">-print-callgraph</a></td><td>Print a call graph</td></tr>
107 <tr><td><a href="#print-callgraph-sccs">-print-callgraph-sccs</a></td><td>Print SCCs of the Call Graph</td></tr>
108 <tr><td><a href="#print-cfg-sccs">-print-cfg-sccs</a></td><td>Print SCCs of each function CFG</td></tr>
109 <tr><td><a href="#print-dbginfo">-print-dbginfo</a></td><td>Print debug info in human readable form</td></tr>
110 <tr><td><a href="#print-dom-info">-print-dom-info</a></td><td>Dominator Info Printer</td></tr>
111 <tr><td><a href="#print-externalfnconstants">-print-externalfnconstants</a></td><td>Print external fn callsites passed constants</td></tr>
112 <tr><td><a href="#print-function">-print-function</a></td><td>Print function to stderr</td></tr>
113 <tr><td><a href="#print-module">-print-module</a></td><td>Print module to stderr</td></tr>
114 <tr><td><a href="#print-used-types">-print-used-types</a></td><td>Find Used Types</td></tr>
115 <tr><td><a href="#profile-estimator">-profile-estimator</a></td><td>Estimate profiling information</td></tr>
116 <tr><td><a href="#profile-loader">-profile-loader</a></td><td>Load profile information from llvmprof.out</td></tr>
117 <tr><td><a href="#profile-verifier">-profile-verifier</a></td><td>Verify profiling information</td></tr>
118 <tr><td><a href="#regions">-regions</a></td><td>Detect single entry single exit regions</td></tr>
119 <tr><td><a href="#scalar-evolution">-scalar-evolution</a></td><td>Scalar Evolution Analysis</td></tr>
120 <tr><td><a href="#scev-aa">-scev-aa</a></td><td>ScalarEvolution-based Alias Analysis</td></tr>
121 <tr><td><a href="#targetdata">-targetdata</a></td><td>Target Data Layout</td></tr>
124 <tr><th colspan="2"><b>TRANSFORM PASSES</b></th></tr>
125 <tr><th>Option</th><th>Name</th></tr>
126 <tr><td><a href="#adce">-adce</a></td><td>Aggressive Dead Code Elimination</td></tr>
127 <tr><td><a href="#always-inline">-always-inline</a></td><td>Inliner for always_inline functions</td></tr>
128 <tr><td><a href="#argpromotion">-argpromotion</a></td><td>Promote 'by reference' arguments to scalars</td></tr>
129 <tr><td><a href="#block-placement">-block-placement</a></td><td>Profile Guided Basic Block Placement</td></tr>
130 <tr><td><a href="#break-crit-edges">-break-crit-edges</a></td><td>Break critical edges in CFG</td></tr>
131 <tr><td><a href="#codegenprepare">-codegenprepare</a></td><td>Optimize for code generation</td></tr>
132 <tr><td><a href="#constmerge">-constmerge</a></td><td>Merge Duplicate Global Constants</td></tr>
133 <tr><td><a href="#constprop">-constprop</a></td><td>Simple constant propagation</td></tr>
134 <tr><td><a href="#dce">-dce</a></td><td>Dead Code Elimination</td></tr>
135 <tr><td><a href="#deadargelim">-deadargelim</a></td><td>Dead Argument Elimination</td></tr>
136 <tr><td><a href="#deadtypeelim">-deadtypeelim</a></td><td>Dead Type Elimination</td></tr>
137 <tr><td><a href="#die">-die</a></td><td>Dead Instruction Elimination</td></tr>
138 <tr><td><a href="#dse">-dse</a></td><td>Dead Store Elimination</td></tr>
139 <tr><td><a href="#functionattrs">-functionattrs</a></td><td>Deduce function attributes</td></tr>
140 <tr><td><a href="#globaldce">-globaldce</a></td><td>Dead Global Elimination</td></tr>
141 <tr><td><a href="#globalopt">-globalopt</a></td><td>Global Variable Optimizer</td></tr>
142 <tr><td><a href="#gvn">-gvn</a></td><td>Global Value Numbering</td></tr>
143 <tr><td><a href="#indvars">-indvars</a></td><td>Canonicalize Induction Variables</td></tr>
144 <tr><td><a href="#inline">-inline</a></td><td>Function Integration/Inlining</td></tr>
145 <tr><td><a href="#insert-edge-profiling">-insert-edge-profiling</a></td><td>Insert instrumentation for edge profiling</td></tr>
146 <tr><td><a href="#insert-optimal-edge-profiling">-insert-optimal-edge-profiling</a></td><td>Insert optimal instrumentation for edge profiling</td></tr>
147 <tr><td><a href="#instcombine">-instcombine</a></td><td>Combine redundant instructions</td></tr>
148 <tr><td><a href="#internalize">-internalize</a></td><td>Internalize Global Symbols</td></tr>
149 <tr><td><a href="#ipconstprop">-ipconstprop</a></td><td>Interprocedural constant propagation</td></tr>
150 <tr><td><a href="#ipsccp">-ipsccp</a></td><td>Interprocedural Sparse Conditional Constant Propagation</td></tr>
151 <tr><td><a href="#jump-threading">-jump-threading</a></td><td>Jump Threading</td></tr>
152 <tr><td><a href="#lcssa">-lcssa</a></td><td>Loop-Closed SSA Form Pass</td></tr>
153 <tr><td><a href="#licm">-licm</a></td><td>Loop Invariant Code Motion</td></tr>
154 <tr><td><a href="#loop-deletion">-loop-deletion</a></td><td>Delete dead loops</td></tr>
155 <tr><td><a href="#loop-extract">-loop-extract</a></td><td>Extract loops into new functions</td></tr>
156 <tr><td><a href="#loop-extract-single">-loop-extract-single</a></td><td>Extract at most one loop into a new function</td></tr>
157 <tr><td><a href="#loop-reduce">-loop-reduce</a></td><td>Loop Strength Reduction</td></tr>
158 <tr><td><a href="#loop-rotate">-loop-rotate</a></td><td>Rotate Loops</td></tr>
159 <tr><td><a href="#loop-simplify">-loop-simplify</a></td><td>Canonicalize natural loops</td></tr>
160 <tr><td><a href="#loop-unroll">-loop-unroll</a></td><td>Unroll loops</td></tr>
161 <tr><td><a href="#loop-unswitch">-loop-unswitch</a></td><td>Unswitch loops</td></tr>
162 <tr><td><a href="#loweratomic">-loweratomic</a></td><td>Lower atomic intrinsics to non-atomic form</td></tr>
163 <tr><td><a href="#lowerinvoke">-lowerinvoke</a></td><td>Lower invoke and unwind, for unwindless code generators</td></tr>
164 <tr><td><a href="#lowerswitch">-lowerswitch</a></td><td>Lower SwitchInst's to branches</td></tr>
165 <tr><td><a href="#mem2reg">-mem2reg</a></td><td>Promote Memory to Register</td></tr>
166 <tr><td><a href="#memcpyopt">-memcpyopt</a></td><td>MemCpy Optimization</td></tr>
167 <tr><td><a href="#mergefunc">-mergefunc</a></td><td>Merge Functions</td></tr>
168 <tr><td><a href="#mergereturn">-mergereturn</a></td><td>Unify function exit nodes</td></tr>
169 <tr><td><a href="#partial-inliner">-partial-inliner</a></td><td>Partial Inliner</td></tr>
170 <tr><td><a href="#prune-eh">-prune-eh</a></td><td>Remove unused exception handling info</td></tr>
171 <tr><td><a href="#reassociate">-reassociate</a></td><td>Reassociate expressions</td></tr>
172 <tr><td><a href="#reg2mem">-reg2mem</a></td><td>Demote all values to stack slots</td></tr>
173 <tr><td><a href="#scalarrepl">-scalarrepl</a></td><td>Scalar Replacement of Aggregates (DT)</td></tr>
174 <tr><td><a href="#sccp">-sccp</a></td><td>Sparse Conditional Constant Propagation</td></tr>
175 <tr><td><a href="#simplify-libcalls">-simplify-libcalls</a></td><td>Simplify well-known library calls</td></tr>
176 <tr><td><a href="#simplifycfg">-simplifycfg</a></td><td>Simplify the CFG</td></tr>
177 <tr><td><a href="#sink">-sink</a></td><td>Code sinking</td></tr>
178 <tr><td><a href="#sretpromotion">-sretpromotion</a></td><td>Promote sret arguments to multiple ret values</td></tr>
179 <tr><td><a href="#strip">-strip</a></td><td>Strip all symbols from a module</td></tr>
180 <tr><td><a href="#strip-dead-debug-info">-strip-dead-debug-info</a></td><td>Strip debug info for unused symbols</td></tr>
181 <tr><td><a href="#strip-dead-prototypes">-strip-dead-prototypes</a></td><td>Strip Unused Function Prototypes</td></tr>
182 <tr><td><a href="#strip-debug-declare">-strip-debug-declare</a></td><td>Strip all llvm.dbg.declare intrinsics</td></tr>
183 <tr><td><a href="#strip-nondebug">-strip-nondebug</a></td><td>Strip all symbols, except dbg symbols, from a module</td></tr>
184 <tr><td><a href="#tailcallelim">-tailcallelim</a></td><td>Tail Call Elimination</td></tr>
185 <tr><td><a href="#tailduplicate">-tailduplicate</a></td><td>Tail Duplication</td></tr>
188 <tr><th colspan="2"><b>UTILITY PASSES</b></th></tr>
189 <tr><th>Option</th><th>Name</th></tr>
190 <tr><td><a href="#deadarghaX0r">-deadarghaX0r</a></td><td>Dead Argument Hacking (BUGPOINT USE ONLY; DO NOT USE)</td></tr>
191 <tr><td><a href="#extract-blocks">-extract-blocks</a></td><td>Extract Basic Blocks From Module (for bugpoint use)</td></tr>
192 <tr><td><a href="#instnamer">-instnamer</a></td><td>Assign names to anonymous instructions</td></tr>
193 <tr><td><a href="#preverify">-preverify</a></td><td>Preliminary module verification</td></tr>
194 <tr><td><a href="#verify">-verify</a></td><td>Module Verifier</td></tr>
195 <tr><td><a href="#view-cfg">-view-cfg</a></td><td>View CFG of function</td></tr>
196 <tr><td><a href="#view-cfg-only">-view-cfg-only</a></td><td>View CFG of function (with no function bodies)</td></tr>
197 <tr><td><a href="#view-dom">-view-dom</a></td><td>View dominance tree of function</td></tr>
198 <tr><td><a href="#view-dom-only">-view-dom-only</a></td><td>View dominance tree of function (with no function bodies)</td></tr>
199 <tr><td><a href="#view-postdom">-view-postdom</a></td><td>View postdominance tree of function</td></tr>
200 <tr><td><a href="#view-postdom-only">-view-postdom-only</a></td><td>View postdominance tree of function (with no function bodies)</td></tr>
205 <!-- ======================================================================= -->
206 <h2><a name="analyses">Analysis Passes</a></h2>
208 <p>This section describes the LLVM Analysis Passes.</p>
210 <!-------------------------------------------------------------------------- -->
212 <a name="aa-eval">-aa-eval: Exhaustive Alias Analysis Precision Evaluator</a>
215 <p>This is a simple N^2 alias analysis accuracy evaluator.
216 Basically, for each function in the program, it simply queries to see how the
217 alias analysis implementation answers alias queries between each pair of
218 pointers in the function.</p>
220 <p>This is inspired and adapted from code by: Naveen Neelakantam, Francesco
221 Spadini, and Wojciech Stryjewski.</p>
224 <!-------------------------------------------------------------------------- -->
226 <a name="basicaa">-basicaa: Basic Alias Analysis (stateless AA impl)</a>
230 This is the default implementation of the Alias Analysis interface
231 that simply implements a few identities (two different globals cannot alias,
232 etc), but otherwise does no analysis.
236 <!-------------------------------------------------------------------------- -->
238 <a name="basiccg">-basiccg: Basic CallGraph Construction</a>
241 <p>Yet to be written.</p>
244 <!-------------------------------------------------------------------------- -->
246 <a name="count-aa">-count-aa: Count Alias Analysis Query Responses</a>
250 A pass which can be used to count how many alias queries
251 are being made and how the alias analysis implementation being used responds.
255 <!-------------------------------------------------------------------------- -->
257 <a name="debug-aa">-debug-aa: AA use debugger</a>
261 This simple pass checks alias analysis users to ensure that if they
262 create a new value, they do not query AA without informing it of the value.
263 It acts as a shim over any other AA pass you want.
267 Yes keeping track of every value in the program is expensive, but this is
272 <!-------------------------------------------------------------------------- -->
274 <a name="domfrontier">-domfrontier: Dominance Frontier Construction</a>
278 This pass is a simple dominator construction algorithm for finding forward
283 <!-------------------------------------------------------------------------- -->
285 <a name="domtree">-domtree: Dominator Tree Construction</a>
289 This pass is a simple dominator construction algorithm for finding forward
294 <!-------------------------------------------------------------------------- -->
296 <a name="dot-callgraph">-dot-callgraph: Print Call Graph to 'dot' file</a>
300 This pass, only available in <code>opt</code>, prints the call graph into a
301 <code>.dot</code> graph. This graph can then be processed with the "dot" tool
302 to convert it to postscript or some other suitable format.
306 <!-------------------------------------------------------------------------- -->
308 <a name="dot-cfg">-dot-cfg: Print CFG of function to 'dot' file</a>
312 This pass, only available in <code>opt</code>, prints the control flow graph
313 into a <code>.dot</code> graph. This graph can then be processed with the
314 "dot" tool to convert it to postscript or some other suitable format.
318 <!-------------------------------------------------------------------------- -->
320 <a name="dot-cfg-only">-dot-cfg-only: Print CFG of function to 'dot' file (with no function bodies)</a>
324 This pass, only available in <code>opt</code>, prints the control flow graph
325 into a <code>.dot</code> graph, omitting the function bodies. This graph can
326 then be processed with the "dot" tool to convert it to postscript or some
327 other suitable format.
331 <!-------------------------------------------------------------------------- -->
333 <a name="dot-dom">-dot-dom: Print dominance tree of function to 'dot' file</a>
337 This pass, only available in <code>opt</code>, prints the dominator tree
338 into a <code>.dot</code> graph. This graph can then be processed with the
339 "dot" tool to convert it to postscript or some other suitable format.
343 <!-------------------------------------------------------------------------- -->
345 <a name="dot-dom-only">-dot-dom-only: Print dominance tree of function to 'dot' file (with no function bodies)</a>
349 This pass, only available in <code>opt</code>, prints the dominator tree
350 into a <code>.dot</code> graph, omitting the function bodies. This graph can
351 then be processed with the "dot" tool to convert it to postscript or some
352 other suitable format.
356 <!-------------------------------------------------------------------------- -->
358 <a name="dot-postdom">-dot-postdom: Print postdominance tree of function to 'dot' file</a>
362 This pass, only available in <code>opt</code>, prints the post dominator tree
363 into a <code>.dot</code> graph. This graph can then be processed with the
364 "dot" tool to convert it to postscript or some other suitable format.
368 <!-------------------------------------------------------------------------- -->
370 <a name="dot-postdom-only">-dot-postdom-only: Print postdominance tree of function to 'dot' file (with no function bodies)</a>
374 This pass, only available in <code>opt</code>, prints the post dominator tree
375 into a <code>.dot</code> graph, omitting the function bodies. This graph can
376 then be processed with the "dot" tool to convert it to postscript or some
377 other suitable format.
381 <!-------------------------------------------------------------------------- -->
383 <a name="globalsmodref-aa">-globalsmodref-aa: Simple mod/ref analysis for globals</a>
387 This simple pass provides alias and mod/ref information for global values
388 that do not have their address taken, and keeps track of whether functions
389 read or write memory (are "pure"). For this simple (but very common) case,
390 we can provide pretty accurate and useful information.
394 <!-------------------------------------------------------------------------- -->
396 <a name="instcount">-instcount: Counts the various types of Instructions</a>
400 This pass collects the count of all instructions and reports them
404 <!-------------------------------------------------------------------------- -->
406 <a name="intervals">-intervals: Interval Partition Construction</a>
410 This analysis calculates and represents the interval partition of a function,
411 or a preexisting interval partition.
415 In this way, the interval partition may be used to reduce a flow graph down
416 to its degenerate single node interval partition (unless it is irreducible).
420 <!-------------------------------------------------------------------------- -->
422 <a name="iv-users">-iv-users: Induction Variable Users</a>
425 <p>Bookkeeping for "interesting" users of expressions computed from
426 induction variables.</p>
429 <!-------------------------------------------------------------------------- -->
431 <a name="lazy-value-info">-lazy-value-info: Lazy Value Information Analysis</a>
434 <p>Interface for lazy computation of value constraint information.</p>
437 <!-------------------------------------------------------------------------- -->
439 <a name="lda">-lda: Loop Dependence Analysis</a>
442 <p>Loop dependence analysis framework, which is used to detect dependences in
443 memory accesses in loops.</p>
446 <!-------------------------------------------------------------------------- -->
448 <a name="libcall-aa">-libcall-aa: LibCall Alias Analysis</a>
451 <p>LibCall Alias Analysis.</p>
454 <!-------------------------------------------------------------------------- -->
456 <a name="lint">-lint: Statically lint-checks LLVM IR</a>
459 <p>This pass statically checks for common and easily-identified constructs
460 which produce undefined or likely unintended behavior in LLVM IR.</p>
462 <p>It is not a guarantee of correctness, in two ways. First, it isn't
463 comprehensive. There are checks which could be done statically which are
464 not yet implemented. Some of these are indicated by TODO comments, but
465 those aren't comprehensive either. Second, many conditions cannot be
466 checked statically. This pass does no dynamic instrumentation, so it
467 can't check for all possible problems.</p>
469 <p>Another limitation is that it assumes all code will be executed. A store
470 through a null pointer in a basic block which is never reached is harmless,
471 but this pass will warn about it anyway.</p>
473 <p>Optimization passes may make conditions that this pass checks for more or
474 less obvious. If an optimization pass appears to be introducing a warning,
475 it may be that the optimization pass is merely exposing an existing
476 condition in the code.</p>
478 <p>This code may be run before instcombine. In many cases, instcombine checks
479 for the same kinds of things and turns instructions with undefined behavior
480 into unreachable (or equivalent). Because of this, this pass makes some
481 effort to look through bitcasts and so on.
485 <!-------------------------------------------------------------------------- -->
487 <a name="loops">-loops: Natural Loop Information</a>
491 This analysis is used to identify natural loops and determine the loop depth
492 of various nodes of the CFG. Note that the loops identified may actually be
493 several natural loops that share the same header node... not just a single
498 <!-------------------------------------------------------------------------- -->
500 <a name="memdep">-memdep: Memory Dependence Analysis</a>
504 An analysis that determines, for a given memory operation, what preceding
505 memory operations it depends on. It builds on alias analysis information, and
506 tries to provide a lazy, caching interface to a common kind of alias
511 <!-------------------------------------------------------------------------- -->
513 <a name="module-debuginfo">-module-debuginfo: Decodes module-level debug info</a>
516 <p>This pass decodes the debug info metadata in a module and prints in a
517 (sufficiently-prepared-) human-readable form.
519 For example, run this pass from opt along with the -analyze option, and
520 it'll print to standard output.
524 <!-------------------------------------------------------------------------- -->
526 <a name="no-aa">-no-aa: No Alias Analysis (always returns 'may' alias)</a>
530 Always returns "I don't know" for alias queries. NoAA is unlike other alias
531 analysis implementations, in that it does not chain to a previous analysis. As
532 such it doesn't follow many of the rules that other alias analyses must.
536 <!-------------------------------------------------------------------------- -->
538 <a name="no-profile">-no-profile: No Profile Information</a>
542 The default "no profile" implementation of the abstract
543 <code>ProfileInfo</code> interface.
547 <!-------------------------------------------------------------------------- -->
549 <a name="postdomfrontier">-postdomfrontier: Post-Dominance Frontier Construction</a>
553 This pass is a simple post-dominator construction algorithm for finding
554 post-dominator frontiers.
558 <!-------------------------------------------------------------------------- -->
560 <a name="postdomtree">-postdomtree: Post-Dominator Tree Construction</a>
564 This pass is a simple post-dominator construction algorithm for finding
569 <!-------------------------------------------------------------------------- -->
571 <a name="print-alias-sets">-print-alias-sets: Alias Set Printer</a>
574 <p>Yet to be written.</p>
577 <!-------------------------------------------------------------------------- -->
579 <a name="print-callgraph">-print-callgraph: Print a call graph</a>
583 This pass, only available in <code>opt</code>, prints the call graph to
584 standard error in a human-readable form.
588 <!-------------------------------------------------------------------------- -->
590 <a name="print-callgraph-sccs">-print-callgraph-sccs: Print SCCs of the Call Graph</a>
594 This pass, only available in <code>opt</code>, prints the SCCs of the call
595 graph to standard error in a human-readable form.
599 <!-------------------------------------------------------------------------- -->
601 <a name="print-cfg-sccs">-print-cfg-sccs: Print SCCs of each function CFG</a>
605 This pass, only available in <code>opt</code>, prints the SCCs of each
606 function CFG to standard error in a human-readable form.
610 <!-------------------------------------------------------------------------- -->
612 <a name="print-dbginfo">-print-dbginfo: Print debug info in human readable form</a>
615 <p>Pass that prints instructions, and associated debug info:</p>
618 <li>source/line/col information</li>
619 <li>original variable name</li>
620 <li>original type name</li>
624 <!-------------------------------------------------------------------------- -->
626 <a name="print-dom-info">-print-dom-info: Dominator Info Printer</a>
629 <p>Dominator Info Printer.</p>
632 <!-------------------------------------------------------------------------- -->
634 <a name="print-externalfnconstants">-print-externalfnconstants: Print external fn callsites passed constants</a>
638 This pass, only available in <code>opt</code>, prints out call sites to
639 external functions that are called with constant arguments. This can be
640 useful when looking for standard library functions we should constant fold
641 or handle in alias analyses.
645 <!-------------------------------------------------------------------------- -->
647 <a name="print-function">-print-function: Print function to stderr</a>
651 The <code>PrintFunctionPass</code> class is designed to be pipelined with
652 other <code>FunctionPass</code>es, and prints out the functions of the module
653 as they are processed.
657 <!-------------------------------------------------------------------------- -->
659 <a name="print-module">-print-module: Print module to stderr</a>
663 This pass simply prints out the entire module when it is executed.
667 <!-------------------------------------------------------------------------- -->
669 <a name="print-used-types">-print-used-types: Find Used Types</a>
673 This pass is used to seek out all of the types in use by the program. Note
674 that this analysis explicitly does not include types only used by the symbol
678 <!-------------------------------------------------------------------------- -->
680 <a name="profile-estimator">-profile-estimator: Estimate profiling information</a>
683 <p>Profiling information that estimates the profiling information
684 in a very crude and unimaginative way.
688 <!-------------------------------------------------------------------------- -->
690 <a name="profile-loader">-profile-loader: Load profile information from llvmprof.out</a>
694 A concrete implementation of profiling information that loads the information
695 from a profile dump file.
699 <!-------------------------------------------------------------------------- -->
701 <a name="profile-verifier">-profile-verifier: Verify profiling information</a>
704 <p>Pass that checks profiling information for plausibility.</p>
707 <a name="regions">-regions: Detect single entry single exit regions</a>
711 The <code>RegionInfo</code> pass detects single entry single exit regions in a
712 function, where a region is defined as any subgraph that is connected to the
713 remaining graph at only two spots. Furthermore, an hierarchical region tree is
718 <!-------------------------------------------------------------------------- -->
720 <a name="scalar-evolution">-scalar-evolution: Scalar Evolution Analysis</a>
724 The <code>ScalarEvolution</code> analysis can be used to analyze and
725 catagorize scalar expressions in loops. It specializes in recognizing general
726 induction variables, representing them with the abstract and opaque
727 <code>SCEV</code> class. Given this analysis, trip counts of loops and other
728 important properties can be obtained.
732 This analysis is primarily useful for induction variable substitution and
737 <!-------------------------------------------------------------------------- -->
739 <a name="scev-aa">-scev-aa: ScalarEvolution-based Alias Analysis</a>
742 <p>Simple alias analysis implemented in terms of ScalarEvolution queries.
744 This differs from traditional loop dependence analysis in that it tests
745 for dependencies within a single iteration of a loop, rather than
746 dependencies between different iterations.
748 ScalarEvolution has a more complete understanding of pointer arithmetic
749 than BasicAliasAnalysis' collection of ad-hoc analyses.
753 <!-------------------------------------------------------------------------- -->
755 <a name="targetdata">-targetdata: Target Data Layout</a>
758 <p>Provides other passes access to information on how the size and alignment
759 required by the the target ABI for various data types.</p>
764 <!-- ======================================================================= -->
765 <h2><a name="transforms">Transform Passes</a></h2>
767 <p>This section describes the LLVM Transform Passes.</p>
769 <!-------------------------------------------------------------------------- -->
771 <a name="adce">-adce: Aggressive Dead Code Elimination</a>
774 <p>ADCE aggressively tries to eliminate code. This pass is similar to
775 <a href="#dce">DCE</a> but it assumes that values are dead until proven
776 otherwise. This is similar to <a href="#sccp">SCCP</a>, except applied to
777 the liveness of values.</p>
780 <!-------------------------------------------------------------------------- -->
782 <a name="always-inline">-always-inline: Inliner for always_inline functions</a>
785 <p>A custom inliner that handles only functions that are marked as
789 <!-------------------------------------------------------------------------- -->
791 <a name="argpromotion">-argpromotion: Promote 'by reference' arguments to scalars</a>
795 This pass promotes "by reference" arguments to be "by value" arguments. In
796 practice, this means looking for internal functions that have pointer
797 arguments. If it can prove, through the use of alias analysis, that an
798 argument is *only* loaded, then it can pass the value into the function
799 instead of the address of the value. This can cause recursive simplification
800 of code and lead to the elimination of allocas (especially in C++ template
805 This pass also handles aggregate arguments that are passed into a function,
806 scalarizing them if the elements of the aggregate are only loaded. Note that
807 it refuses to scalarize aggregates which would require passing in more than
808 three operands to the function, because passing thousands of operands for a
809 large array or structure is unprofitable!
813 Note that this transformation could also be done for arguments that are only
814 stored to (returning the value instead), but does not currently. This case
815 would be best handled when and if LLVM starts supporting multiple return
816 values from functions.
820 <!-------------------------------------------------------------------------- -->
822 <a name="block-placement">-block-placement: Profile Guided Basic Block Placement</a>
825 <p>This pass is a very simple profile guided basic block placement algorithm.
826 The idea is to put frequently executed blocks together at the start of the
827 function and hopefully increase the number of fall-through conditional
828 branches. If there is no profile information for a particular function, this
829 pass basically orders blocks in depth-first order.</p>
832 <!-------------------------------------------------------------------------- -->
834 <a name="break-crit-edges">-break-crit-edges: Break critical edges in CFG</a>
838 Break all of the critical edges in the CFG by inserting a dummy basic block.
839 It may be "required" by passes that cannot deal with critical edges. This
840 transformation obviously invalidates the CFG, but can update forward dominator
841 (set, immediate dominators, tree, and frontier) information.
845 <!-------------------------------------------------------------------------- -->
847 <a name="codegenprepare">-codegenprepare: Optimize for code generation</a>
850 This pass munges the code in the input function to better prepare it for
851 SelectionDAG-based code generation. This works around limitations in it's
852 basic-block-at-a-time approach. It should eventually be removed.
855 <!-------------------------------------------------------------------------- -->
857 <a name="constmerge">-constmerge: Merge Duplicate Global Constants</a>
861 Merges duplicate global constants together into a single constant that is
862 shared. This is useful because some passes (ie TraceValues) insert a lot of
863 string constants into the program, regardless of whether or not an existing
868 <!-------------------------------------------------------------------------- -->
870 <a name="constprop">-constprop: Simple constant propagation</a>
873 <p>This file implements constant propagation and merging. It looks for
874 instructions involving only constant operands and replaces them with a
875 constant value instead of an instruction. For example:</p>
876 <blockquote><pre>add i32 1, 2</pre></blockquote>
878 <blockquote><pre>i32 3</pre></blockquote>
879 <p>NOTE: this pass has a habit of making definitions be dead. It is a good
880 idea to to run a <a href="#die">DIE</a> (Dead Instruction Elimination) pass
881 sometime after running this pass.</p>
884 <!-------------------------------------------------------------------------- -->
886 <a name="dce">-dce: Dead Code Elimination</a>
890 Dead code elimination is similar to <a href="#die">dead instruction
891 elimination</a>, but it rechecks instructions that were used by removed
892 instructions to see if they are newly dead.
896 <!-------------------------------------------------------------------------- -->
898 <a name="deadargelim">-deadargelim: Dead Argument Elimination</a>
902 This pass deletes dead arguments from internal functions. Dead argument
903 elimination removes arguments which are directly dead, as well as arguments
904 only passed into function calls as dead arguments of other functions. This
905 pass also deletes dead arguments in a similar way.
909 This pass is often useful as a cleanup pass to run after aggressive
910 interprocedural passes, which add possibly-dead arguments.
914 <!-------------------------------------------------------------------------- -->
916 <a name="deadtypeelim">-deadtypeelim: Dead Type Elimination</a>
920 This pass is used to cleanup the output of GCC. It eliminate names for types
921 that are unused in the entire translation unit, using the <a
922 href="#findusedtypes">find used types</a> pass.
926 <!-------------------------------------------------------------------------- -->
928 <a name="die">-die: Dead Instruction Elimination</a>
932 Dead instruction elimination performs a single pass over the function,
933 removing instructions that are obviously dead.
937 <!-------------------------------------------------------------------------- -->
939 <a name="dse">-dse: Dead Store Elimination</a>
943 A trivial dead store elimination that only considers basic-block local
948 <!-------------------------------------------------------------------------- -->
950 <a name="functionattrs">-functionattrs: Deduce function attributes</a>
953 <p>A simple interprocedural pass which walks the call-graph, looking for
954 functions which do not access or only read non-local memory, and marking them
955 readnone/readonly. In addition, it marks function arguments (of pointer type)
956 'nocapture' if a call to the function does not create any copies of the pointer
957 value that outlive the call. This more or less means that the pointer is only
958 dereferenced, and not returned from the function or stored in a global.
959 This pass is implemented as a bottom-up traversal of the call-graph.
963 <!-------------------------------------------------------------------------- -->
965 <a name="globaldce">-globaldce: Dead Global Elimination</a>
969 This transform is designed to eliminate unreachable internal globals from the
970 program. It uses an aggressive algorithm, searching out globals that are
971 known to be alive. After it finds all of the globals which are needed, it
972 deletes whatever is left over. This allows it to delete recursive chunks of
973 the program which are unreachable.
977 <!-------------------------------------------------------------------------- -->
979 <a name="globalopt">-globalopt: Global Variable Optimizer</a>
983 This pass transforms simple global variables that never have their address
984 taken. If obviously true, it marks read/write globals as constant, deletes
985 variables only stored to, etc.
989 <!-------------------------------------------------------------------------- -->
991 <a name="gvn">-gvn: Global Value Numbering</a>
995 This pass performs global value numbering to eliminate fully and partially
996 redundant instructions. It also performs redundant load elimination.
1000 <!-------------------------------------------------------------------------- -->
1002 <a name="indvars">-indvars: Canonicalize Induction Variables</a>
1006 This transformation analyzes and transforms the induction variables (and
1007 computations derived from them) into simpler forms suitable for subsequent
1008 analysis and transformation.
1012 This transformation makes the following changes to each loop with an
1013 identifiable induction variable:
1017 <li>All loops are transformed to have a <em>single</em> canonical
1018 induction variable which starts at zero and steps by one.</li>
1019 <li>The canonical induction variable is guaranteed to be the first PHI node
1020 in the loop header block.</li>
1021 <li>Any pointer arithmetic recurrences are raised to use array
1026 If the trip count of a loop is computable, this pass also makes the following
1031 <li>The exit condition for the loop is canonicalized to compare the
1032 induction value against the exit value. This turns loops like:
1033 <blockquote><pre>for (i = 7; i*i < 1000; ++i)</pre></blockquote>
1035 <blockquote><pre>for (i = 0; i != 25; ++i)</pre></blockquote></li>
1036 <li>Any use outside of the loop of an expression derived from the indvar
1037 is changed to compute the derived value outside of the loop, eliminating
1038 the dependence on the exit value of the induction variable. If the only
1039 purpose of the loop is to compute the exit value of some derived
1040 expression, this transformation will make the loop dead.</li>
1044 This transformation should be followed by strength reduction after all of the
1045 desired loop transformations have been performed. Additionally, on targets
1046 where it is profitable, the loop could be transformed to count down to zero
1047 (the "do loop" optimization).
1051 <!-------------------------------------------------------------------------- -->
1053 <a name="inline">-inline: Function Integration/Inlining</a>
1057 Bottom-up inlining of functions into callees.
1061 <!-------------------------------------------------------------------------- -->
1063 <a name="insert-edge-profiling">-insert-edge-profiling: Insert instrumentation for edge profiling</a>
1067 This pass instruments the specified program with counters for edge profiling.
1068 Edge profiling can give a reasonable approximation of the hot paths through a
1069 program, and is used for a wide variety of program transformations.
1073 Note that this implementation is very naïve. It inserts a counter for
1074 <em>every</em> edge in the program, instead of using control flow information
1075 to prune the number of counters inserted.
1079 <!-------------------------------------------------------------------------- -->
1081 <a name="insert-optimal-edge-profiling">-insert-optimal-edge-profiling: Insert optimal instrumentation for edge profiling</a>
1084 <p>This pass instruments the specified program with counters for edge profiling.
1085 Edge profiling can give a reasonable approximation of the hot paths through a
1086 program, and is used for a wide variety of program transformations.
1090 <!-------------------------------------------------------------------------- -->
1092 <a name="instcombine">-instcombine: Combine redundant instructions</a>
1096 Combine instructions to form fewer, simple
1097 instructions. This pass does not modify the CFG This pass is where algebraic
1098 simplification happens.
1102 This pass combines things like:
1107 %Z = add i32 %Y, 1</pre></blockquote>
1114 >%Z = add i32 %X, 2</pre></blockquote>
1117 This is a simple worklist driven algorithm.
1121 This pass guarantees that the following canonicalizations are performed on
1126 <li>If a binary operator has a constant operand, it is moved to the right-
1128 <li>Bitwise operators with constant operands are always grouped so that
1129 shifts are performed first, then <code>or</code>s, then
1130 <code>and</code>s, then <code>xor</code>s.</li>
1131 <li>Compare instructions are converted from <code><</code>,
1132 <code>></code>, <code>≤</code>, or <code>≥</code> to
1133 <code>=</code> or <code>≠</code> if possible.</li>
1134 <li>All <code>cmp</code> instructions on boolean values are replaced with
1135 logical operations.</li>
1136 <li><code>add <var>X</var>, <var>X</var></code> is represented as
1137 <code>mul <var>X</var>, 2</code> ⇒ <code>shl <var>X</var>, 1</code></li>
1138 <li>Multiplies with a constant power-of-two argument are transformed into
1144 <!-------------------------------------------------------------------------- -->
1146 <a name="internalize">-internalize: Internalize Global Symbols</a>
1150 This pass loops over all of the functions in the input module, looking for a
1151 main function. If a main function is found, all other functions and all
1152 global variables with initializers are marked as internal.
1156 <!-------------------------------------------------------------------------- -->
1158 <a name="ipconstprop">-ipconstprop: Interprocedural constant propagation</a>
1162 This pass implements an <em>extremely</em> simple interprocedural constant
1163 propagation pass. It could certainly be improved in many different ways,
1164 like using a worklist. This pass makes arguments dead, but does not remove
1165 them. The existing dead argument elimination pass should be run after this
1166 to clean up the mess.
1170 <!-------------------------------------------------------------------------- -->
1172 <a name="ipsccp">-ipsccp: Interprocedural Sparse Conditional Constant Propagation</a>
1176 An interprocedural variant of <a href="#sccp">Sparse Conditional Constant
1181 <!-------------------------------------------------------------------------- -->
1183 <a name="jump-threading">-jump-threading: Jump Threading</a>
1187 Jump threading tries to find distinct threads of control flow running through
1188 a basic block. This pass looks at blocks that have multiple predecessors and
1189 multiple successors. If one or more of the predecessors of the block can be
1190 proven to always cause a jump to one of the successors, we forward the edge
1191 from the predecessor to the successor by duplicating the contents of this
1195 An example of when this can occur is code like this:
1202 if (X < 3) {</pre>
1205 In this case, the unconditional branch at the end of the first if can be
1206 revectored to the false side of the second if.
1210 <!-------------------------------------------------------------------------- -->
1212 <a name="lcssa">-lcssa: Loop-Closed SSA Form Pass</a>
1216 This pass transforms loops by placing phi nodes at the end of the loops for
1217 all values that are live across the loop boundary. For example, it turns
1218 the left into the right code:
1222 >for (...) for (...)
1227 X3 = phi(X1, X2) X3 = phi(X1, X2)
1228 ... = X3 + 4 X4 = phi(X3)
1232 This is still valid LLVM; the extra phi nodes are purely redundant, and will
1233 be trivially eliminated by <code>InstCombine</code>. The major benefit of
1234 this transformation is that it makes many other loop optimizations, such as
1235 LoopUnswitching, simpler.
1239 <!-------------------------------------------------------------------------- -->
1241 <a name="licm">-licm: Loop Invariant Code Motion</a>
1245 This pass performs loop invariant code motion, attempting to remove as much
1246 code from the body of a loop as possible. It does this by either hoisting
1247 code into the preheader block, or by sinking code to the exit blocks if it is
1248 safe. This pass also promotes must-aliased memory locations in the loop to
1249 live in registers, thus hoisting and sinking "invariant" loads and stores.
1253 This pass uses alias analysis for two purposes:
1257 <li>Moving loop invariant loads and calls out of loops. If we can determine
1258 that a load or call inside of a loop never aliases anything stored to,
1259 we can hoist it or sink it like any other instruction.</li>
1260 <li>Scalar Promotion of Memory - If there is a store instruction inside of
1261 the loop, we try to move the store to happen AFTER the loop instead of
1262 inside of the loop. This can only happen if a few conditions are true:
1264 <li>The pointer stored through is loop invariant.</li>
1265 <li>There are no stores or loads in the loop which <em>may</em> alias
1266 the pointer. There are no calls in the loop which mod/ref the
1269 If these conditions are true, we can promote the loads and stores in the
1270 loop of the pointer to use a temporary alloca'd variable. We then use
1271 the mem2reg functionality to construct the appropriate SSA form for the
1276 <!-------------------------------------------------------------------------- -->
1278 <a name="loop-deletion">-loop-deletion: Delete dead loops</a>
1282 This file implements the Dead Loop Deletion Pass. This pass is responsible
1283 for eliminating loops with non-infinite computable trip counts that have no
1284 side effects or volatile instructions, and do not contribute to the
1285 computation of the function's return value.
1289 <!-------------------------------------------------------------------------- -->
1291 <a name="loop-extract">-loop-extract: Extract loops into new functions</a>
1295 A pass wrapper around the <code>ExtractLoop()</code> scalar transformation to
1296 extract each top-level loop into its own new function. If the loop is the
1297 <em>only</em> loop in a given function, it is not touched. This is a pass most
1298 useful for debugging via bugpoint.
1302 <!-------------------------------------------------------------------------- -->
1304 <a name="loop-extract-single">-loop-extract-single: Extract at most one loop into a new function</a>
1308 Similar to <a href="#loop-extract">Extract loops into new functions</a>,
1309 this pass extracts one natural loop from the program into a function if it
1310 can. This is used by bugpoint.
1314 <!-------------------------------------------------------------------------- -->
1316 <a name="loop-reduce">-loop-reduce: Loop Strength Reduction</a>
1320 This pass performs a strength reduction on array references inside loops that
1321 have as one or more of their components the loop induction variable. This is
1322 accomplished by creating a new value to hold the initial value of the array
1323 access for the first iteration, and then creating a new GEP instruction in
1324 the loop to increment the value by the appropriate amount.
1328 <!-------------------------------------------------------------------------- -->
1330 <a name="loop-rotate">-loop-rotate: Rotate Loops</a>
1333 <p>A simple loop rotation transformation.</p>
1336 <!-------------------------------------------------------------------------- -->
1338 <a name="loop-simplify">-loop-simplify: Canonicalize natural loops</a>
1342 This pass performs several transformations to transform natural loops into a
1343 simpler form, which makes subsequent analyses and transformations simpler and
1348 Loop pre-header insertion guarantees that there is a single, non-critical
1349 entry edge from outside of the loop to the loop header. This simplifies a
1350 number of analyses and transformations, such as LICM.
1354 Loop exit-block insertion guarantees that all exit blocks from the loop
1355 (blocks which are outside of the loop that have predecessors inside of the
1356 loop) only have predecessors from inside of the loop (and are thus dominated
1357 by the loop header). This simplifies transformations such as store-sinking
1358 that are built into LICM.
1362 This pass also guarantees that loops will have exactly one backedge.
1366 Note that the simplifycfg pass will clean up blocks which are split out but
1367 end up being unnecessary, so usage of this pass should not pessimize
1372 This pass obviously modifies the CFG, but updates loop information and
1373 dominator information.
1377 <!-------------------------------------------------------------------------- -->
1379 <a name="loop-unroll">-loop-unroll: Unroll loops</a>
1383 This pass implements a simple loop unroller. It works best when loops have
1384 been canonicalized by the <a href="#indvars"><tt>-indvars</tt></a> pass,
1385 allowing it to determine the trip counts of loops easily.
1389 <!-------------------------------------------------------------------------- -->
1391 <a name="loop-unswitch">-loop-unswitch: Unswitch loops</a>
1395 This pass transforms loops that contain branches on loop-invariant conditions
1396 to have multiple loops. For example, it turns the left into the right code:
1408 This can increase the size of the code exponentially (doubling it every time
1409 a loop is unswitched) so we only unswitch if the resultant code will be
1410 smaller than a threshold.
1414 This pass expects LICM to be run before it to hoist invariant conditions out
1415 of the loop, to make the unswitching opportunity obvious.
1419 <!-------------------------------------------------------------------------- -->
1421 <a name="loweratomic">-loweratomic: Lower atomic intrinsics to non-atomic form</a>
1425 This pass lowers atomic intrinsics to non-atomic form for use in a known
1426 non-preemptible environment.
1430 The pass does not verify that the environment is non-preemptible (in
1431 general this would require knowledge of the entire call graph of the
1432 program including any libraries which may not be available in bitcode form);
1433 it simply lowers every atomic intrinsic.
1437 <!-------------------------------------------------------------------------- -->
1439 <a name="lowerinvoke">-lowerinvoke: Lower invoke and unwind, for unwindless code generators</a>
1443 This transformation is designed for use by code generators which do not yet
1444 support stack unwinding. This pass supports two models of exception handling
1445 lowering, the 'cheap' support and the 'expensive' support.
1449 'Cheap' exception handling support gives the program the ability to execute
1450 any program which does not "throw an exception", by turning 'invoke'
1451 instructions into calls and by turning 'unwind' instructions into calls to
1452 abort(). If the program does dynamically use the unwind instruction, the
1453 program will print a message then abort.
1457 'Expensive' exception handling support gives the full exception handling
1458 support to the program at the cost of making the 'invoke' instruction
1459 really expensive. It basically inserts setjmp/longjmp calls to emulate the
1460 exception handling as necessary.
1464 Because the 'expensive' support slows down programs a lot, and EH is only
1465 used for a subset of the programs, it must be specifically enabled by the
1466 <tt>-enable-correct-eh-support</tt> option.
1470 Note that after this pass runs the CFG is not entirely accurate (exceptional
1471 control flow edges are not correct anymore) so only very simple things should
1472 be done after the lowerinvoke pass has run (like generation of native code).
1473 This should not be used as a general purpose "my LLVM-to-LLVM pass doesn't
1474 support the invoke instruction yet" lowering pass.
1478 <!-------------------------------------------------------------------------- -->
1480 <a name="lowerswitch">-lowerswitch: Lower SwitchInst's to branches</a>
1484 Rewrites <tt>switch</tt> instructions with a sequence of branches, which
1485 allows targets to get away with not implementing the switch instruction until
1490 <!-------------------------------------------------------------------------- -->
1492 <a name="mem2reg">-mem2reg: Promote Memory to Register</a>
1496 This file promotes memory references to be register references. It promotes
1497 <tt>alloca</tt> instructions which only have <tt>load</tt>s and
1498 <tt>store</tt>s as uses. An <tt>alloca</tt> is transformed by using dominator
1499 frontiers to place <tt>phi</tt> nodes, then traversing the function in
1500 depth-first order to rewrite <tt>load</tt>s and <tt>store</tt>s as
1501 appropriate. This is just the standard SSA construction algorithm to construct
1506 <!-------------------------------------------------------------------------- -->
1508 <a name="memcpyopt">-memcpyopt: MemCpy Optimization</a>
1512 This pass performs various transformations related to eliminating memcpy
1513 calls, or transforming sets of stores into memset's.
1517 <!-------------------------------------------------------------------------- -->
1519 <a name="mergefunc">-mergefunc: Merge Functions</a>
1522 <p>This pass looks for equivalent functions that are mergable and folds them.
1524 A hash is computed from the function, based on its type and number of
1527 Once all hashes are computed, we perform an expensive equality comparison
1528 on each function pair. This takes n^2/2 comparisons per bucket, so it's
1529 important that the hash function be high quality. The equality comparison
1530 iterates through each instruction in each basic block.
1532 When a match is found the functions are folded. If both functions are
1533 overridable, we move the functionality into a new internal function and
1534 leave two overridable thunks to it.
1538 <!-------------------------------------------------------------------------- -->
1540 <a name="mergereturn">-mergereturn: Unify function exit nodes</a>
1544 Ensure that functions have at most one <tt>ret</tt> instruction in them.
1545 Additionally, it keeps track of which node is the new exit node of the CFG.
1549 <!-------------------------------------------------------------------------- -->
1551 <a name="partial-inliner">-partial-inliner: Partial Inliner</a>
1554 <p>This pass performs partial inlining, typically by inlining an if
1555 statement that surrounds the body of the function.
1559 <!-------------------------------------------------------------------------- -->
1561 <a name="prune-eh">-prune-eh: Remove unused exception handling info</a>
1565 This file implements a simple interprocedural pass which walks the call-graph,
1566 turning <tt>invoke</tt> instructions into <tt>call</tt> instructions if and
1567 only if the callee cannot throw an exception. It implements this as a
1568 bottom-up traversal of the call-graph.
1572 <!-------------------------------------------------------------------------- -->
1574 <a name="reassociate">-reassociate: Reassociate expressions</a>
1578 This pass reassociates commutative expressions in an order that is designed
1579 to promote better constant propagation, GCSE, LICM, PRE, etc.
1583 For example: 4 + (<var>x</var> + 5) ⇒ <var>x</var> + (4 + 5)
1587 In the implementation of this algorithm, constants are assigned rank = 0,
1588 function arguments are rank = 1, and other values are assigned ranks
1589 corresponding to the reverse post order traversal of current function
1590 (starting at 2), which effectively gives values in deep loops higher rank
1591 than values not in loops.
1595 <!-------------------------------------------------------------------------- -->
1597 <a name="reg2mem">-reg2mem: Demote all values to stack slots</a>
1601 This file demotes all registers to memory references. It is intented to be
1602 the inverse of <a href="#mem2reg"><tt>-mem2reg</tt></a>. By converting to
1603 <tt>load</tt> instructions, the only values live across basic blocks are
1604 <tt>alloca</tt> instructions and <tt>load</tt> instructions before
1605 <tt>phi</tt> nodes. It is intended that this should make CFG hacking much
1606 easier. To make later hacking easier, the entry block is split into two, such
1607 that all introduced <tt>alloca</tt> instructions (and nothing else) are in the
1612 <!-------------------------------------------------------------------------- -->
1614 <a name="scalarrepl">-scalarrepl: Scalar Replacement of Aggregates (DT)</a>
1618 The well-known scalar replacement of aggregates transformation. This
1619 transform breaks up <tt>alloca</tt> instructions of aggregate type (structure
1620 or array) into individual <tt>alloca</tt> instructions for each member if
1621 possible. Then, if possible, it transforms the individual <tt>alloca</tt>
1622 instructions into nice clean scalar SSA form.
1626 This combines a simple scalar replacement of aggregates algorithm with the <a
1627 href="#mem2reg"><tt>mem2reg</tt></a> algorithm because often interact,
1628 especially for C++ programs. As such, iterating between <tt>scalarrepl</tt>,
1629 then <a href="#mem2reg"><tt>mem2reg</tt></a> until we run out of things to
1634 <!-------------------------------------------------------------------------- -->
1636 <a name="sccp">-sccp: Sparse Conditional Constant Propagation</a>
1640 Sparse conditional constant propagation and merging, which can be summarized
1645 <li>Assumes values are constant unless proven otherwise</li>
1646 <li>Assumes BasicBlocks are dead unless proven otherwise</li>
1647 <li>Proves values to be constant, and replaces them with constants</li>
1648 <li>Proves conditional branches to be unconditional</li>
1652 Note that this pass has a habit of making definitions be dead. It is a good
1653 idea to to run a DCE pass sometime after running this pass.
1657 <!-------------------------------------------------------------------------- -->
1659 <a name="simplify-libcalls">-simplify-libcalls: Simplify well-known library calls</a>
1663 Applies a variety of small optimizations for calls to specific well-known
1664 function calls (e.g. runtime library functions). For example, a call
1665 <tt>exit(3)</tt> that occurs within the <tt>main()</tt> function can be
1666 transformed into simply <tt>return 3</tt>.
1670 <!-------------------------------------------------------------------------- -->
1672 <a name="simplifycfg">-simplifycfg: Simplify the CFG</a>
1676 Performs dead code elimination and basic block merging. Specifically:
1680 <li>Removes basic blocks with no predecessors.</li>
1681 <li>Merges a basic block into its predecessor if there is only one and the
1682 predecessor only has one successor.</li>
1683 <li>Eliminates PHI nodes for basic blocks with a single predecessor.</li>
1684 <li>Eliminates a basic block that only contains an unconditional
1689 <!-------------------------------------------------------------------------- -->
1691 <a name="sink">-sink: Code sinking</a>
1694 <p>This pass moves instructions into successor blocks, when possible, so that
1695 they aren't executed on paths where their results aren't needed.
1699 <!-------------------------------------------------------------------------- -->
1701 <a name="sretpromotion">-sretpromotion: Promote sret arguments to multiple ret values</a>
1705 This pass finds functions that return a struct (using a pointer to the struct
1706 as the first argument of the function, marked with the '<tt>sret</tt>' attribute) and
1707 replaces them with a new function that simply returns each of the elements of
1708 that struct (using multiple return values).
1712 This pass works under a number of conditions:
1716 <li>The returned struct must not contain other structs</li>
1717 <li>The returned struct must only be used to load values from</li>
1718 <li>The placeholder struct passed in is the result of an <tt>alloca</tt></li>
1722 <!-------------------------------------------------------------------------- -->
1724 <a name="strip">-strip: Strip all symbols from a module</a>
1728 performs code stripping. this transformation can delete:
1732 <li>names for virtual registers</li>
1733 <li>symbols for internal globals and functions</li>
1734 <li>debug information</li>
1738 note that this transformation makes code much less readable, so it should
1739 only be used in situations where the <tt>strip</tt> utility would be used,
1740 such as reducing code size or making it harder to reverse engineer code.
1744 <!-------------------------------------------------------------------------- -->
1746 <a name="strip-dead-debug-info">-strip-dead-debug-info: Strip debug info for unused symbols</a>
1750 performs code stripping. this transformation can delete:
1754 <li>names for virtual registers</li>
1755 <li>symbols for internal globals and functions</li>
1756 <li>debug information</li>
1760 note that this transformation makes code much less readable, so it should
1761 only be used in situations where the <tt>strip</tt> utility would be used,
1762 such as reducing code size or making it harder to reverse engineer code.
1766 <!-------------------------------------------------------------------------- -->
1768 <a name="strip-dead-prototypes">-strip-dead-prototypes: Strip Unused Function Prototypes</a>
1772 This pass loops over all of the functions in the input module, looking for
1773 dead declarations and removes them. Dead declarations are declarations of
1774 functions for which no implementation is available (i.e., declarations for
1775 unused library functions).
1779 <!-------------------------------------------------------------------------- -->
1781 <a name="strip-debug-declare">-strip-debug-declare: Strip all llvm.dbg.declare intrinsics</a>
1784 <p>This pass implements code stripping. Specifically, it can delete:</p>
1786 <li>names for virtual registers</li>
1787 <li>symbols for internal globals and functions</li>
1788 <li>debug information</li>
1791 Note that this transformation makes code much less readable, so it should
1792 only be used in situations where the 'strip' utility would be used, such as
1793 reducing code size or making it harder to reverse engineer code.
1797 <!-------------------------------------------------------------------------- -->
1799 <a name="strip-nondebug">-strip-nondebug: Strip all symbols, except dbg symbols, from a module</a>
1802 <p>This pass implements code stripping. Specifically, it can delete:</p>
1804 <li>names for virtual registers</li>
1805 <li>symbols for internal globals and functions</li>
1806 <li>debug information</li>
1809 Note that this transformation makes code much less readable, so it should
1810 only be used in situations where the 'strip' utility would be used, such as
1811 reducing code size or making it harder to reverse engineer code.
1815 <!-------------------------------------------------------------------------- -->
1817 <a name="tailcallelim">-tailcallelim: Tail Call Elimination</a>
1821 This file transforms calls of the current function (self recursion) followed
1822 by a return instruction with a branch to the entry of the function, creating
1823 a loop. This pass also implements the following extensions to the basic
1828 <li>Trivial instructions between the call and return do not prevent the
1829 transformation from taking place, though currently the analysis cannot
1830 support moving any really useful instructions (only dead ones).
1831 <li>This pass transforms functions that are prevented from being tail
1832 recursive by an associative expression to use an accumulator variable,
1833 thus compiling the typical naive factorial or <tt>fib</tt> implementation
1834 into efficient code.
1835 <li>TRE is performed if the function returns void, if the return
1836 returns the result returned by the call, or if the function returns a
1837 run-time constant on all exits from the function. It is possible, though
1838 unlikely, that the return returns something else (like constant 0), and
1839 can still be TRE'd. It can be TRE'd if <em>all other</em> return
1840 instructions in the function return the exact same value.
1841 <li>If it can prove that callees do not access theier caller stack frame,
1842 they are marked as eligible for tail call elimination (by the code
1847 <!-------------------------------------------------------------------------- -->
1849 <a name="tailduplicate">-tailduplicate: Tail Duplication</a>
1853 This pass performs a limited form of tail duplication, intended to simplify
1854 CFGs by removing some unconditional branches. This pass is necessary to
1855 straighten out loops created by the C front-end, but also is capable of
1856 making other code nicer. After this pass is run, the CFG simplify pass
1857 should be run to clean up the mess.
1863 <!-- ======================================================================= -->
1864 <h2><a name="utilities">Utility Passes</a></h2>
1866 <p>This section describes the LLVM Utility Passes.</p>
1868 <!-------------------------------------------------------------------------- -->
1870 <a name="deadarghaX0r">-deadarghaX0r: Dead Argument Hacking (BUGPOINT USE ONLY; DO NOT USE)</a>
1874 Same as dead argument elimination, but deletes arguments to functions which
1875 are external. This is only for use by <a
1876 href="Bugpoint.html">bugpoint</a>.</p>
1879 <!-------------------------------------------------------------------------- -->
1881 <a name="extract-blocks">-extract-blocks: Extract Basic Blocks From Module (for bugpoint use)</a>
1885 This pass is used by bugpoint to extract all blocks from the module into their
1889 <!-------------------------------------------------------------------------- -->
1891 <a name="instnamer">-instnamer: Assign names to anonymous instructions</a>
1894 <p>This is a little utility pass that gives instructions names, this is mostly
1895 useful when diffing the effect of an optimization because deleting an
1896 unnamed instruction can change all other instruction numbering, making the
1901 <!-------------------------------------------------------------------------- -->
1903 <a name="preverify">-preverify: Preliminary module verification</a>
1907 Ensures that the module is in the form required by the <a
1908 href="#verifier">Module Verifier</a> pass.
1912 Running the verifier runs this pass automatically, so there should be no need
1917 <!-------------------------------------------------------------------------- -->
1919 <a name="verify">-verify: Module Verifier</a>
1923 Verifies an LLVM IR code. This is useful to run after an optimization which is
1924 undergoing testing. Note that <tt>llvm-as</tt> verifies its input before
1925 emitting bitcode, and also that malformed bitcode is likely to make LLVM
1926 crash. All language front-ends are therefore encouraged to verify their output
1927 before performing optimizing transformations.
1931 <li>Both of a binary operator's parameters are of the same type.</li>
1932 <li>Verify that the indices of mem access instructions match other
1934 <li>Verify that arithmetic and other things are only performed on
1935 first-class types. Verify that shifts and logicals only happen on
1937 <li>All of the constants in a switch statement are of the correct type.</li>
1938 <li>The code is in valid SSA form.</li>
1939 <li>It is illegal to put a label into any other type (like a structure) or
1941 <li>Only phi nodes can be self referential: <tt>%x = add i32 %x, %x</tt> is
1943 <li>PHI nodes must have an entry for each predecessor, with no extras.</li>
1944 <li>PHI nodes must be the first thing in a basic block, all grouped
1946 <li>PHI nodes must have at least one entry.</li>
1947 <li>All basic blocks should only end with terminator insts, not contain
1949 <li>The entry node to a function must not have predecessors.</li>
1950 <li>All Instructions must be embedded into a basic block.</li>
1951 <li>Functions cannot take a void-typed parameter.</li>
1952 <li>Verify that a function's argument list agrees with its declared
1954 <li>It is illegal to specify a name for a void value.</li>
1955 <li>It is illegal to have a internal global value with no initializer.</li>
1956 <li>It is illegal to have a ret instruction that returns a value that does
1957 not agree with the function return value type.</li>
1958 <li>Function call argument types match the function prototype.</li>
1959 <li>All other things that are tested by asserts spread about the code.</li>
1963 Note that this does not provide full security verification (like Java), but
1964 instead just tries to ensure that code is well-formed.
1968 <!-------------------------------------------------------------------------- -->
1970 <a name="view-cfg">-view-cfg: View CFG of function</a>
1974 Displays the control flow graph using the GraphViz tool.
1978 <!-------------------------------------------------------------------------- -->
1980 <a name="view-cfg-only">-view-cfg-only: View CFG of function (with no function bodies)</a>
1984 Displays the control flow graph using the GraphViz tool, but omitting function
1989 <!-------------------------------------------------------------------------- -->
1991 <a name="view-dom">-view-dom: View dominance tree of function</a>
1995 Displays the dominator tree using the GraphViz tool.
1999 <!-------------------------------------------------------------------------- -->
2001 <a name="view-dom-only">-view-dom-only: View dominance tree of function (with no function bodies)</a>
2005 Displays the dominator tree using the GraphViz tool, but omitting function
2010 <!-------------------------------------------------------------------------- -->
2012 <a name="view-postdom">-view-postdom: View postdominance tree of function</a>
2016 Displays the post dominator tree using the GraphViz tool.
2020 <!-------------------------------------------------------------------------- -->
2022 <a name="view-postdom-only">-view-postdom-only: View postdominance tree of function (with no function bodies)</a>
2026 Displays the post dominator tree using the GraphViz tool, but omitting
2033 <!-- *********************************************************************** -->
2037 <a href="http://jigsaw.w3.org/css-validator/check/referer"><img
2038 src="http://jigsaw.w3.org/css-validator/images/vcss-blue" alt="Valid CSS"></a>
2039 <a href="http://validator.w3.org/check/referer"><img
2040 src="http://www.w3.org/Icons/valid-html401-blue" alt="Valid HTML 4.01"></a>
2042 <a href="mailto:rspencer@x10sys.com">Reid Spencer</a><br>
2043 <a href="http://llvm.org/">LLVM Compiler Infrastructure</a><br>
2044 Last modified: $Date$