How to Create a Custom Fault Injector

1. Under < LLFI_SRC_ROOT >/runtime_lib, create a class that is the child of class HardwareFaultInjector or class SoftwareFaultInjector. You have to implement the injectFault(long llfi_index, unsigned size, unsigned fi_bit,char *buf) in your class.

2. Register the class with RegisterFaultInjector(). In the registration, you also provide the name to find the fault injector.

3. Build the < LLFI_SRC_ROOT >/runtime_lib by adding the created file(s) to CMakeLists.txt in that directory, and go to the < LLFI_DST_ROOT > and run make

4. run the fault injection executable with the custom fault injector. You need to use the option fi_type=<name given in step 2> in llfi.config.runtime.txt to invoke the custom fault injector.

5. There are some sample fault injectors in < LLFI_SRC_ROOT >/runtime_lib/CommonFaultInjectors.cpp. Play it first to get an idea.

6. [GUI] To get the custom fault injector to show up in the LLFI GUI, add the name of the fault injector, exactly as registered, to the file LLFI_SRC_ROOT/gui/config/fault_type.txt (hardware injection only).

[FIDL] Alternatively, use the FIDL tool to generate a Custom Software Fault Injector and a Custom Instruction Selector instead.