1LLI(1) LLVM LLI(1)
2
6 lli - directly execute programs from LLVM bitcode
7
9 lli [options] [filename] [program args]
10
12 lli directly executes programs in LLVM bitcode format. It takes a pro‐
13 gram in LLVM bitcode format and executes it using a just-in-time com‐
14 piler or an interpreter.
15 lli is not an emulator. It will not execute IR of different architec‐
17 tures and it can only interpret (or JIT-compile) for the host architec‐
18 ture.
19 The JIT compiler takes the same arguments as other tools, like llc, but
21 they don't necessarily work for the interpreter.
22 If filename is not specified, then lli reads the LLVM bitcode for the
24 program from standard input.
25 The optional args specified on the command line are passed to the pro‐
27 gram as arguments.
28
30 -fake-argv0=executable
31 Override the argv[0] value passed into the executing program.
32 -force-interpreter={false,true}
34 If set to true, use the interpreter even if a just-in-time com‐
35 piler is available for this architecture. Defaults to false.
36 -help Print a summary of command line options.
38 -load=pluginfilename
40 Causes lli to load the plugin (shared object) named pluginfile‐
41 name and use it for optimization.
42 -stats Print statistics from the code-generation passes. This is only
44 meaningful for the just-in-time compiler, at present.
45 -time-passes
47 Record the amount of time needed for each code-generation pass
48 and print it to standard error.
49 -version
51 Print out the version of lli and exit without doing anything
52 else.
53
55 -mtriple=target triple
56 Override the target triple specified in the input bitcode file
57 with the specified string. This may result in a crash if you
58 pick an architecture which is not compatible with the current
59 system.
60 -march=arch
62 Specify the architecture for which to generate assembly, over‐
63 riding the target encoded in the bitcode file. See the output
64 of llc -help for a list of valid architectures. By default this
65 is inferred from the target triple or autodetected to the cur‐
66 rent architecture.
67 -mcpu=cpuname
69 Specify a specific chip in the current architecture to generate
70 code for. By default this is inferred from the target triple
71 and autodetected to the current architecture. For a list of
72 available CPUs, use: llvm-as < /dev/null | llc -march=xyz
73 -mcpu=help
74 -mattr=a1,+a2,-a3,...
76 Override or control specific attributes of the target, such as
77 whether SIMD operations are enabled or not. The default set of
78 attributes is set by the current CPU. For a list of available
79 attributes, use: llvm-as < /dev/null | llc -march=xyz
80 -mattr=help
81
83 -disable-excess-fp-precision
84 Disable optimizations that may increase floating point preci‐
85 sion.
86 -enable-no-infs-fp-math
88 Enable optimizations that assume no Inf values.
89 -enable-no-nans-fp-math
91 Enable optimizations that assume no NAN values.
92 -enable-unsafe-fp-math
94 Causes lli to enable optimizations that may decrease floating
95 point precision.
96 -soft-float
98 Causes lli to generate software floating point library calls
99 instead of equivalent hardware instructions.
100
102 -code-model=model
103 Choose the code model from:
104 default: Target default code model
106 small: Small code model
107 kernel: Kernel code model
108 medium: Medium code model
109 large: Large code model
110 -disable-post-RA-scheduler
112 Disable scheduling after register allocation.
113 -disable-spill-fusing
115 Disable fusing of spill code into instructions.
116 -jit-enable-eh
118 Exception handling should be enabled in the just-in-time com‐
119 piler.
120 -join-liveintervals
122 Coalesce copies (default=true).
123 -nozero-initialized-in-bss
125 Don't place zero-initialized symbols into the BSS section.
126 -pre-RA-sched=scheduler
128 Instruction schedulers available (before register allocation):
129 =default: Best scheduler for the target
131 =none: No scheduling: breadth first sequencing
132 =simple: Simple two pass scheduling: minimize critical path and maximize processor utilization
133 =simple-noitin: Simple two pass scheduling: Same as simple except using generic latency
134 =list-burr: Bottom-up register reduction list scheduling
135 =list-tdrr: Top-down register reduction list scheduling
136 =list-td: Top-down list scheduler -print-machineinstrs - Print generated machine code
137 -regalloc=allocator
139 Register allocator to use (default=linearscan)
140 =bigblock: Big-block register allocator
142 =linearscan: linear scan register allocator =local - local register allocator
143 =simple: simple register allocator
144 -relocation-model=model
146 Choose relocation model from:
147 =default: Target default relocation model
149 =static: Non-relocatable code =pic - Fully relocatable, position independent code
150 =dynamic-no-pic: Relocatable external references, non-relocatable code
151 -spiller
153 Spiller to use (default=local)
154 =simple: simple spiller
156 =local: local spiller
157 -x86-asm-syntax=syntax
159 Choose style of code to emit from X86 backend:
160 =att: Emit AT&T-style assembly
162 =intel: Emit Intel-style assembly
163
165 If lli fails to load the program, it will exit with an exit code of 1.
166 Otherwise, it will return the exit code of the program it executes.
167
169 llc
170
172 Maintained by The LLVM Team (http://llvm.org/).
173
175 2003-2019, LLVM Project
1766 2019-07-25 LLI(1)