lli

lli is a command-line tool that executes LLVM bitcode by interpreting it and, in some configurations, by just-in-time compiling it to native code. It is part of the LLVM Project and is often described as the LLVM interpreter. Programs can be compiled to LLVM’s intermediate representation (IR) by front ends such as clang and then run directly with lli without generating machine code for a specific platform.

When invoked, lli loads a bitcode file, initializes the LLVM runtime, and executes the program’s entry point. It can work with additional bitcode libraries to extend functionality and may rely on the host system libraries for input, output, and other runtime services. The tool is designed to provide a convenient way to test and prototype LLVM IR, verify IR transformations, and debug low-level code without a separate native compilation step.

Implementation-wise, lli leverages the LLVM execution infrastructure, which in modern configurations includes both interpretation and just-in-time compilation capabilities. This allows hot spots in the code to be compiled on the fly for better performance on long-running tasks, while still supporting straightforward interpretation for shorter runs or debugging. The exact behavior can depend on the LLVM version and the available execution engine backends.

See also: LLVM Project, LLVM IR, Clang, ExecutionEngine, MCJIT, ORC JIT.