5 llc - LLVM static compiler
9 B<llc> [I<options>] [I<filename>]
13 The B<llc> command compiles LLVM source inputs into assembly language for a
14 specified architecture. The assembly language output can then be passed through
15 a native assembler and linker to generate a native executable.
17 The choice of architecture for the output assembly code is automatically
18 determined from the input file, unless the B<-march> option is used to override
23 If I<filename> is - or omitted, B<llc> reads from standard input. Otherwise, it
24 will from I<filename>. Inputs can be in either the LLVM assembly language
25 format (.ll) or the LLVM bitcode format (.bc).
27 If the B<-o> option is omitted, then B<llc> will send its output to standard
28 output if the input is from standard input. If the B<-o> option specifies -,
29 then the output will also be sent to standard output.
31 If no B<-o> option is specified and an input file other than - is specified,
32 then B<llc> creates the output filename by taking the input filename,
33 removing any existing F<.bc> extension, and adding a F<.s> suffix.
35 Other B<llc> options are as follows:
37 =head2 End-user Options
43 Print a summary of command line options.
47 Generate code at different optimization levels. These correspond to the I<-O0>,
48 I<-O1>, I<-O2>, I<-O3>, and I<-O4> optimization levels used by B<llvm-gcc> and
51 =item B<-mtriple>=I<target triple>
53 Override the target triple specified in the input file with the specified
56 =item B<-march>=I<arch>
58 Specify the architecture for which to generate assembly, overriding the target
59 encoded in the input file. See the output of B<llc -help> for a list of
60 valid architectures. By default this is inferred from the target triple or
61 autodetected to the current architecture.
63 =item B<-mcpu>=I<cpuname>
65 Specify a specific chip in the current architecture to generate code for.
66 By default this is inferred from the target triple and autodetected to
67 the current architecture. For a list of available CPUs, use:
68 B<llvm-as E<lt> /dev/null | llc -march=xyz -mcpu=help>
70 =item B<-mattr>=I<a1,+a2,-a3,...>
72 Override or control specific attributes of the target, such as whether SIMD
73 operations are enabled or not. The default set of attributes is set by the
74 current CPU. For a list of available attributes, use:
75 B<llvm-as E<lt> /dev/null | llc -march=xyz -mattr=help>
77 =item B<--disable-fp-elim>
79 Disable frame pointer elimination optimization.
81 =item B<--disable-excess-fp-precision>
83 Disable optimizations that may produce excess precision for floating point.
84 Note that this option can dramatically slow down code on some systems
87 =item B<--enable-no-infs-fp-math>
89 Enable optimizations that assume no Inf values.
91 =item B<--enable-no-nans-fp-math>
93 Enable optimizations that assume no NAN values.
95 =item B<--enable-unsafe-fp-math>
97 Enable optimizations that make unsafe assumptions about IEEE math (e.g. that
98 addition is associative) or may not work for all input ranges. These
99 optimizations allow the code generator to make use of some instructions which
100 would otherwise not be usable (such as fsin on X86).
102 =item B<--enable-correct-eh-support>
104 Instruct the B<lowerinvoke> pass to insert code for correct exception handling
105 support. This is expensive and is by default omitted for efficiency.
109 Print statistics recorded by code-generation passes.
111 =item B<--time-passes>
113 Record the amount of time needed for each pass and print a report to standard
116 =item B<--load>=F<dso_path>
118 Dynamically load F<dso_path> (a path to a dynamically shared object) that
119 implements an LLVM target. This will permit the target name to be used with the
120 B<-march> option so that code can be generated for that target.
124 =head2 Tuning/Configuration Options
128 =item B<--print-machineinstrs>
130 Print generated machine code between compilation phases (useful for debugging).
132 =item B<--regalloc>=I<allocator>
134 Specify the register allocator to use. The default I<allocator> is I<local>.
135 Valid register allocators are:
141 Very simple "always spill" register allocator
145 Local register allocator
149 Linear scan global register allocator
151 =item I<iterativescan>
153 Iterative scan global register allocator
157 =item B<--spiller>=I<spiller>
159 Specify the spiller to use for register allocators that support it. Currently
160 this option is used only by the linear scan register allocator. The default
161 I<spiller> is I<local>. Valid spillers are:
177 =head2 Intel IA-32-specific Options
181 =item B<--x86-asm-syntax=att|intel>
183 Specify whether to emit assembly code in AT&T syntax (the default) or intel
190 If B<llc> succeeds, it will exit with 0. Otherwise, if an error occurs,
191 it will exit with a non-zero value.
199 Maintained by the LLVM Team (L<http://llvm.org>).