1LLC(1) LLVM LLC(1)
2
6 llc - LLVM static compiler
7
9 llc [options] [filename]
10
12 The llc command compiles LLVM source inputs into assembly language for
13 a specified architecture. The assembly language output can then be
14 passed through a native assembler and linker to generate a native exe‐
15 cutable.
16 The choice of architecture for the output assembly code is automati‐
18 cally determined from the input file, unless the -march option is used
19 to override the default.
20
22 If filename is "-" or omitted, llc reads from standard input. Other‐
23 wise, it will from filename. Inputs can be in either the LLVM assembly
24 language format (.ll) or the LLVM bitcode format (.bc).
25 If the -o option is omitted, then llc will send its output to standard
27 output if the input is from standard input. If the -o option specifies
28 "-", then the output will also be sent to standard output.
29 If no -o option is specified and an input file other than "-" is speci‐
31 fied, then llc creates the output filename by taking the input file‐
32 name, removing any existing .bc extension, and adding a .s suffix.
33 Other llc options are described below.
35 End-user Options
37 -help Print a summary of command line options.
38 -O=uint
40 Generate code at different optimization levels. These corre‐
41 spond to the -O0, -O1, -O2, and -O3 optimization levels used by
42 clang.
43 -mtriple=<target triple>
45 Override the target triple specified in the input file with the
46 specified string.
47 -march=<arch>
49 Specify the architecture for which to generate assembly, over‐
50 riding the target encoded in the input file. See the output of
51 llc -help for a list of valid architectures. By default this is
52 inferred from the target triple or autodetected to the current
53 architecture.
54 -mcpu=<cpuname>
56 Specify a specific chip in the current architecture to generate
57 code for. By default this is inferred from the target triple
58 and autodetected to the current architecture. For a list of
59 available CPUs, use:
60 llvm-as < /dev/null | llc -march=xyz -mcpu=help
62 -filetype=<output file type>
64 Specify what kind of output llc should generated. Options are:
65 asm for textual assembly ( '.s'), obj for native object files
66 ('.o') and null for not emitting anything (for performance test‐
67 ing).
68 Note that not all targets support all options.
70 -mattr=a1,+a2,-a3,...
72 Override or control specific attributes of the target, such as
73 whether SIMD operations are enabled or not. The default set of
74 attributes is set by the current CPU. For a list of available
75 attributes, use:
76 llvm-as < /dev/null | llc -march=xyz -mattr=help
78 --frame-pointer
80 Specify effect of frame pointer elimination optimization
81 (all,non-leaf,none).
82 --disable-excess-fp-precision
84 Disable optimizations that may produce excess precision for
85 floating point. Note that this option can dramatically slow
86 down code on some systems (e.g. X86).
87 --enable-no-infs-fp-math
89 Enable optimizations that assume no Inf values.
90 --enable-no-nans-fp-math
92 Enable optimizations that assume no NAN values.
93 --enable-unsafe-fp-math
95 Enable optimizations that make unsafe assumptions about IEEE
96 math (e.g. that addition is associative) or may not work for all
97 input ranges. These optimizations allow the code generator to
98 make use of some instructions which would otherwise not be
99 usable (such as fsin on X86).
100 --stats
102 Print statistics recorded by code-generation passes.
103 --time-passes
105 Record the amount of time needed for each pass and print a
106 report to standard error.
107 --load=<dso_path>
109 Dynamically load dso_path (a path to a dynamically shared
110 object) that implements an LLVM target. This will permit the
111 target name to be used with the -march option so that code can
112 be generated for that target.
113 -meabi=[default|gnu|4|5]
115 Specify which EABI version should conform to. Valid EABI ver‐
116 sions are gnu, 4 and 5. Default value (default) depends on the
117 triple.
118 -stack-size-section
120 Emit the .stack_sizes section which contains stack size meta‐
121 data. The section contains an array of pairs of function symbol
122 values (pointer size) and stack sizes (unsigned LEB128). The
123 stack size values only include the space allocated in the func‐
124 tion prologue. Functions with dynamic stack allocations are not
125 included.
126 Tuning/Configuration Options
128 --print-machineinstrs
129 Print generated machine code between compilation phases (useful
130 for debugging).
131 --regalloc=<allocator>
133 Specify the register allocator to use. Valid register alloca‐
134 tors are:
135 basic
137 Basic register allocator.
138 fast
140 Fast register allocator. It is the default for unoptimized
141 code.
142 greedy
144 Greedy register allocator. It is the default for optimized
145 code.
146 pbqp
148 Register allocator based on 'Partitioned Boolean Quadratic
149 Programming'.
150 --spiller=<spiller>
152 Specify the spiller to use for register allocators that support
153 it. Currently this option is used only by the linear scan reg‐
154 ister allocator. The default spiller is local. Valid spillers
155 are:
156 simple
158 Simple spiller
159 local
161 Local spiller
162 Intel IA-32-specific Options
164 --x86-asm-syntax=[att|intel]
165 Specify whether to emit assembly code in AT&T syntax (the
166 default) or Intel syntax.
167
169 If llc succeeds, it will exit with 0. Otherwise, if an error occurs,
170 it will exit with a non-zero value.
171
173 lli
174
176 Maintained by the LLVM Team (https://llvm.org/).
177
179 2003-2019, LLVM Project
1808 2019-11-13 LLC(1)