LLVM: bugpoint tool


NAME

bugpoint

SYNOPSIS

bugpoint [options] [input llvm ll/bc files] [LLVM passes] --args <program arguments>...

DESCRIPTION

The bugpoint tool is a generally useful tool for narrowing down problems in LLVM tools and passes. It can be used to debug three types of failures: optimizer crashes, miscompilations by optimizers, or invalid native code generation. It aims to reduce testcases to something useful. For example, if gccas crashes while optimizing a file, it will identify the optimization (or combination of optimizations) that causes the crash, and reduce the file down to a small example which triggers the crash.

bugpoint has been designed to be a useful tool without requiring any hooks into the LLVM intrastructure at all. It works with any and all LLVM passes and code generators, and does not need to "know" how they work. Because of this, it may appear to do a lot of stupid things or miss obvious simplifications. Remember, however, that computer time is much cheaper than programmer time, so if it takes a long time to reduce a testcase it is still worth it. :)

Automatic Mode Selection

bugpoint reads the specified list of .bc or .ll files specified on the command-line and links them together. If any LLVM passes are specified on the command line, it runs these passes on the resultant module. If any of the passes crash, or if they produce a malformed LLVM module, bugpoint enters crash debugging mode.

Otherwise, if the -output option was not specified, bugpoint runs the initial program with the C backend (which is assumed to generate good code) to generate a reference output. Once bugpoint has a reference output to match, it tries executing the original program with the selected code generator. If the resultant output is different than the reference output, it exters code generator debugging mode.

Otherwise, bugpoint runs the LLVM program after all of the LLVM passes have been applied to it. If the executed program matches the reference output, there is no problem bugpoint can debug. Otherwise, it enters miscompilation debugging mode.

Crash debugging mode

If an optimizer crashes, bugpoint will try a variety of techniques to narrow down the list of passes and the code to a more manageable amount. First, bugpoint figures out which combination of passes trigger the bug. This is useful when debugging a problem exposed by gccas for example, because it has over 30 optimization it runs.

Next, bugpoint tries removing functions from the module, to reduce the size of the testcase to a reasonable amount. Usually it is able to get it down to a single function for intraprocedural optimizations. Once the number of functions has been reduced, it attempts to delete various edges in the control flow graph, to reduce the size of the function as much as possible. Finally, bugpoint deletes any individual LLVM instructions whose absense does not eliminate the failure. At the end, bugpoint should tell you what passes crash, give you a bytecode file, and give you instructions on how to reproduce the failure with opt or analyze.

Code generator debugging mode

The code generator debugger attempts to narrow down the amount of code that is being miscompiled by the selected code generator. To do this, it takes the LLVM program and partitions it into two pieces: one piece which it compiles with the C backend (into a shared object), and one piece which it runs with either the JIT or the static LLC compiler. It uses several techniques to reduce the amount of code pushed through the LLVM code generator, to reduce the potential scope of the problem. After it is finished, it emits two bytecode files (the "test" [to be compiled with the code generator] and "safe" [to be compiled with the C backend] modules), and instructions for reproducing the problem. This module assume the C backend produces good code.

If you are using this mode and get an error message that says "Non-instruction is using an external function!", try using the -run-llc option instead of the -run-jit option. This is due to an unimplemented feature in the code generator debugging mode.

Miscompilation debugging mode

The miscompilation debugging mode works similarly to the code generator debugging mode. It works by splitting the program into two pieces, running the optimizations specified on one piece, relinking the program, then executing it. It attempts to narrow down the list of passes to the one (or few) which are causing the miscompilation, then reduce the portion of the program which is being miscompiled. This module assumes that the selected code generator is working properly.

Advice for using bugpoint

bugpoint can be a remarkably useful tool, but it sometimes works in non-obvious ways. Here are some hints and tips:

  1. In code generator and miscompilation debugging modes, bugpoint only works with programs that have deterministic output. Thus, if the program outputs the date, time, or any other "random" data, it should be masked out.
  2. In code generator and miscompilation debugging modes, debugging will go faster if you manually modify the program or its inputs to reduce the runtime, but still exhibit the problem.
  3. bugpoint is extremely useful when working on a new optimization: it helps track down regressions quickly. To avoid having to relink bugpoint every time you change your optization however, have bugpoint dynamically load your optimization with the -load option.
  4. bugpoint can generate a lot of output and run for a long period of time. It is often useful to capture the output of the program to file. For example:
    bugpoint ..... |& tee bugpoint.log

OPTIONS

EXIT STATUS

If bugpoint succeeds in finding a problem, it will exit with 0. Otherwise, if an error occurs, it will exit with a non-zero value.

SEE ALSO

opt, analyze
Maintained by the LLVM Team.