Run a fuzzer

Once you have built your fuzzer, there are a few ways to execute it.

Run a fuzzer on a Fuchsia device

You can use the fx fuzz tool to run a fuzzer on your local device. This is the recommended way to run your fuzzer while you are developing it. The tool is useful in that it knows how to find fuzzing-related files and knows various common options.

  • To see available commands and options:
  • To see available fuzzers:
  • To start a fuzzer:

package and fuzzer match those reported by fx fuzz list, and may be abbreviated. For commands that accept a single fuzzer, e.g. check, the abbreviated name must uniquely identify exactly one fuzzer.

When starting a fuzzer, the tool will echo the command it is invoking, prefixed by +. This can be useful if you want to manually invoke the fuzzer.

If the fuzzer finds any results, you can use fx fuzz to investigate them further. See Handling results found through fuzzing.

Run a fuzzer on a host platform

You can run host fuzzers built by the Fuchsia build system, although the extra tooling of fx fuzz is not supported. This means you will need to manually run them and reproduce the results they find. To build host fuzzers, set fuzz_host=true in the fuzzers_package[gn fuzzers package].

For example:

fuzzers_package("overnet_fuzzers") {
  cpp_fuzzers = [ "packet_protocol:packet_protocol_fuzzer" ]
  fuzz_host = true

Upon building, you can find the host fuzzers in the host variant output directory. For example, the fuzzer above would be produced at //out/default/host_x64-asan-fuzzer.

Run a fuzzer on ClusterFuzz

This is the easiest and most recommended way to run a fuzzer after initial development. To run your fuzzer on ClusterFuzz, you simply need to ensure it is a GN dependency of //bundles/buildbot:core. Practically, this means including it in your code's “tests” GN target.

For example:

   group("tests") {
     deps = [

If you are unsure if your fuzzer is included in the dependency graph, you can check using gn path.

For example:

For //examples/fuzzers, this yields:

All fuzzers in that dependency graph will be made available to ClusterFuzz to select and run. If ClusterFuzz opens bugs, you can use its outputs to reproduce its findings. See Handling results found through fuzzing.