CMock: A Summary

ThrowTheSwitch.org

This documentation is released under a Creative Commons 3.0 Attribution Share-Alike License

What Exactly Are We Talking About Here?

CMock is a nice little tool which takes your header files and creates a Mock interface for it so that you can more easily unit test modules that touch other modules. For each function prototype in your header, like this one:

int DoesSomething(int a, int b);

…you get an automatically generated DoesSomething function that you can link to instead of your real DoesSomething function. By using this Mocked version, you can then verify that it receives the data you want, and make it return whatever data you desire, make it throw errors when you want, and more… Create these for everything your latest real module touches, and you’re suddenly in a position of power: You can control and verify every detail of your latest creation.

To make that easier, CMock also gives you a bunch of functions like the ones below, so you can tell that generated DoesSomething function how to behave for each test:

void DoesSomething_ExpectAndReturn(int a, int b, int toReturn);
void DoesSomething_ExpectAndThrow(int a, int b, EXCEPTION_T error);
void DoesSomething_StubWithCallback(CMOCK_DoesSomething_CALLBACK YourCallback);
void DoesSomething_IgnoreAndReturn(int toReturn);

You can pile a bunch of these back to back, and it remembers what you wanted to pass when, like so:

test_CallsDoesSomething_ShouldDoJustThat(void)
{
    DoesSomething_ExpectAndReturn(1,2,3);
    DoesSomething_ExpectAndReturn(4,5,6);
    DoesSomething_ExpectAndThrow(7,8, STATUS_ERROR_OOPS);

    CallsDoesSomething( );
}

This test will call CallsDoesSomething, which is the function we are testing. We are expecting that function to call DoesSomething three times. The first time, we check to make sure it’s called as DoesSomething(1, 2) and we’ll magically return a 3. The second time we check for DoesSomething(4, 5) and we’ll return a 6. The third time we verify DoesSomething(7, 8) and we’ll throw an error instead of returning anything. If CallsDoesSomething gets any of this wrong, it fails the test. It will fail if you didn’t call DoesSomething enough, or too much, or with the wrong arguments, or in the wrong order.

CMock is based on Unity, which it uses for all internal testing. It uses Ruby to do all the main work (versions 2.0.0 and above).

Installing

The first thing you need to do to install CMock is to get yourself a copy of Ruby. If you’re on linux or osx, you probably already have it. You can prove it by typing the following:

ruby --version

If it replied in a way that implies ignorance, then you’re going to need to install it. You can go to ruby-lang to get the latest version. You’re also going to need to do that if it replied with a version that is older than 2.0.0. Go ahead. We’ll wait.

Once you have Ruby, you have three options:

Generated Mock Module Summary

In addition to the mocks themselves, CMock will generate the following functions for use in your tests. The expect functions are always generated. The other functions are only generated if those plugins are enabled:

Expect:

Your basic staple Expects which will be used for most of your day to day CMock work. By calling this, you are telling CMock that you expect that function to be called during your test. It also specifies which arguments you expect it to be called with, and what return value you want returned when that happens. You can call this function multiple times back to back in order to queue up multiple calls.

  • void func(void) => void func_Expect(void)

  • void func(params) => void func_Expect(expected_params)

  • retval func(void) => void func_ExpectAndReturn(retval_to_return)

  • retval func(params) => void func_ExpectAndReturn(expected_params, retval_to_return)

ExpectAnyArgs:

This behaves just like the Expects calls, except that it doesn’t really care what the arguments are that the mock gets called with. It still counts the number of times the mock is called and it still handles return values if there are some. Note that an ExpectAnyArgs call is not generated for functions that have no arguments, because it would act exactly like the existing Expect and ExpectAndReturn calls.

  • void func(params) => void func_ExpectAnyArgs(void)

  • retval func(params) => void func_ExpectAnyArgsAndReturn(retval_to_return)

Array:

An ExpectWithArray is another variant of Expect. Like expect, it cares about the number of times a mock is called, the arguments it is called with, and the values it is to return. This variant has another feature, though. For anything that resembles a pointer or array, it breaks the argument into TWO arguments. The first is the original pointer. The second specify the number of elements it is to verify of that array. If you specify 1, it’ll check one object. If 2, it’ll assume your pointer is pointing at the first of two elements in an array. If you specify zero elements and UNITY_COMPARE_PTRS_ON_ZERO_ARRAY is defined, then this assertion can also be used to directly compare the pointers to verify that they are pointing to the same memory address.

  • void func(void) => (nothing. In fact, an additional function is only generated if the params list contains pointers)

  • void func(ptr * param, other) => void func_ExpectWithArray(ptr* param, int param_depth, other)

  • retval func(void) => (nothing. In fact, an additional function is only generated if the params list contains pointers)

  • retval func(other, ptr* param) => void func_ExpectWithArrayAndReturn(other, ptr* param, int param_depth, retval_to_return)

Ignore:

Maybe you don’t care about the number of times a particular function is called or the actual arguments it is called with. In that case, you want to use Ignore. Ignore only needs to be called once per test. It will then ignore any further calls to that particular mock. The IgnoreAndReturn works similarly, except that it has the added benefit of knowing what to return when that call happens. If the mock is called more times than IgnoreAndReturn was called, it will keep returning the last value without complaint. If it’s called fewer times, it will also ignore that. You SAID you didn’t care how many times it was called, right?

  • void func(void) => void func_Ignore(void)

  • void func(params) => void func_Ignore(void)

  • retval func(void) => void func_IgnoreAndReturn(retval_to_return)

  • retval func(params) => void func_IgnoreAndReturn(retval_to_return)

StopIgnore:

Maybe you want to ignore a particular function for part of a test but dont want to ignore it later on. In that case, you want to use StopIgnore which will cancel the previously called Ignore or IgnoreAndReturn requiring you to Expect or otherwise handle the call to a function.

  • void func(void) => void func_StopIgnore(void)

  • void func(params) => void func_StopIgnore(void)

  • retval func(void) => void func_StopIgnore(void)

  • retval func(params) => void func_StopIgnore(void)

IgnoreStateless:

This plugin is similar to the Ignore plugin, but the IgnoreAndReturn functions are stateless. So the Ignored function will always return the last specified return value and does not queue the return values as the IgnoreAndReturn of the default plugin will.

To stop ignoring a function you can call StopIgnore or simply overwrite the Ignore (resp. IgnoreAndReturn) with an Expect (resp. ExpectAndReturn). Note that calling Ignore (resp IgnoreAndReturn) will clear your previous called Expect (resp. ExpectAndReturn), so they are not restored after StopIgnore is called.

You can use this plugin by using :ignore_stateless instead of :ignore in your CMock configuration file.

The generated functions are the same as Ignore and StopIgnore above.

Ignore Arg:

Maybe you overall want to use Expect and its similar variations, but you don’t care what is passed to a particular argument. This is particularly useful when that argument is a pointer to a value that is supposed to be filled in by the function. You don’t want to use ExpectAnyArgs, because you still care about the other arguments. Instead, after an Expect call is made, you can call this function. It tells CMock to ignore a particular argument for the rest of this test, for this mock function. You may call multiple instances of this to ignore multiple arguments after each expectation if desired.

  • void func(params) => void func_IgnoreArg_paramName(void)

ReturnThruPtr:

Another option which operates on a particular argument of a function is the ReturnThruPtr plugin. For every argument that resembles a pointer or reference, CMock generates an instance of this function. Just as the AndReturn functions support injecting one or more return values into a queue, this function lets you specify one or more return values which are queued up and copied into the space being pointed at each time the mock is called.

  • void func(param1) => void func_ReturnThruPtr_paramName(val_to_return)

  • => void func_ReturnArrayThruPtr_paramName(val_to_return, len)

  • => void func_ReturnMemThruPtr_paramName(val_to_return, size)

For example, consider the following function:

BOOL divide(uint numerator, uint denominator, uint* result){ if (denominator == 0){ return FALSE; } *result = numerator/denominator; return TRUE; }

We might want to mock this function so that regardless of the inputs, it returns TRUE and a result of 23. We could do so like this:

uint result_1 = 23; divide_ExpectAnyArgsAndReturn(TRUE); divide_ReturnThruPtr_result(&result_1);

If we want to expect a numerator of 5 and a denominator of 2 and return a result of 42: uint result_1 = 42; divide_ExpectAndReturn(5,2,NULL,TRUE); divide_IgnoreArg_result(); divide_ReturnThruPtr_result(&result_1);

Callback:

If all those other options don’t work, and you really need to do something custom, you still have a choice. As soon as you stub a callback in a test, it will call the callback whenever the mock is encountered and return the retval returned from the callback (if any).

  • void func(void) => void func_[AddCallback,Stub](CMOCK_func_CALLBACK callback) where CMOCK_func_CALLBACK looks like: void func(int NumCalls)

  • void func(params) => void func_[AddCallback,Stub](CMOCK_func_CALLBACK callback) where CMOCK_func_CALLBACK looks like: void func(params, int NumCalls)

  • retval func(void) => void func_[AddCallback,Stub](CMOCK_func_CALLBACK callback) where CMOCK_func_CALLBACK looks like: retval func(int NumCalls)

  • retval func(params) => void func_[AddCallback,Stub](CMOCK_func_CALLBACK callback) where CMOCK_func_CALLBACK looks like: retval func(params, int NumCalls)

You can choose from two options:

  • func_AddCallback tells the mock to check its arguments and calling order (based on any Expects you’ve set up) before calling the callback.

  • func_Stub tells the mock to skip all the normal checks and jump directly to the callback instead. In this case, you are replacing the normal mock calls with your own custom stub function.

There is also an older name, func_StubWithCallback, which is just an alias for either func_AddCallback or func_Stub depending on setting of the :callback_after_arg_check toggle. This is deprecated and we recommend using the two options above.

Cexception:

Finally, if you are using Cexception for error handling, you can use this to throw errors from inside mocks. Like Expects, it remembers which call was supposed to throw the error, and it still checks parameters first.

  • void func(void) => void func_ExpectAndThrow(value_to_throw)

  • void func(params) => void func_ExpectAndThrow(expected_params, value_to_throw)

  • retval func(void) => void func_ExpectAndThrow(value_to_throw)

  • retval func(params) => void func_ExpectAndThrow(expected_params, value_to_throw)

Running CMock

CMock is a Ruby script and class. You can therefore use it directly from the command line, or include it in your own scripts or rakefiles.

Mocking from the Command Line

After unpacking CMock, you will find cmock.rb in the ‘lib’ directory. This is the file that you want to run. It takes a list of header files to be mocked, as well as an optional yaml file for a more detailed configuration (see config options below).

For example, this will create three mocks using the configuration specified in MyConfig.yml:

ruby cmock.rb -oMyConfig.yml super.h duper.h awesome.h

And this will create two mocks using the default configuration:

ruby cmock.rb ../mocking/stuff/is/fun.h ../try/it/yourself.h

Mocking From Scripts or Rake

CMock can be used directly from your own scripts or from a rakefile. Start by including cmock.rb, then create an instance of CMock. When you create your instance, you may initialize it in one of three ways.

You may specify nothing, allowing it to run with default settings:

require 'cmock.rb'
cmock = CMock.new

You may specify a YAML file containing the configuration options you desire:

cmock = CMock.new('../MyConfig.yml')

You may specify the options explicitly:

cmock = Cmock.new(:plugins => [:cexception, :ignore], :mock_path => 'my/mocks/')

Creating Skeletons:

Not only is CMock able to generate mock files from a header file, but it is also able to generate (and update) skeleton C files from headers. It does this by creating a (mostly) empty implementation for every function that is declared in the header. If you later add to that header list, just run this feature again and it will add prototypes for the missing functions!

Like the normal usecase for CMock, this feature can be used from the command line or from within its ruby API. For example, from the command line, add --skeleton to generate a skeleton instead:

ruby cmock.rb --skeleton ../create/c/for/this.h

Config Options:

The following configuration options can be specified in the yaml file or directly when instantiating.

Passed as Ruby, they look like this:

    { :attributes => [“__funky”, “__intrinsic”], :when_ptr => :compare }

Defined in the yaml file, they look more like this:

    :cmock:
      :attributes:
        - __funky
        - __intrinsic
      :when_ptr: :compare

In all cases, you can just include the things that you want to override from the defaults. We’ve tried to specify what the defaults are below.

  • :attributes: These are attributes that CMock should ignore for you for testing purposes. Custom compiler extensions and externs are handy things to put here. If your compiler is choking on some extended syntax, this is often a good place to look.

    • defaults: [‘__ramfunc’, ‘__irq’, ‘__fiq’, ‘register’, ‘extern’]

    • note: this option will reinsert these attributes onto the mock’s calls. If that isn’t what you are looking for, check out :strippables.

  • :c_calling_conventions: Similarly, CMock may need to understand which C calling conventions might show up in your codebase. If it encounters something it doesn’t recognize, it’s not going to mock it. We have the most common covered, but there are many compilers out there, and therefore many other options.

    • defaults: [‘__stdcall’, ‘__cdecl’, ‘__fastcall’]

    • note: this option will reinsert these attributes onto the mock’s calls. If that isn’t what you are looking for, check out :strippables.

  • :callback_after_arg_check: Tell :callback plugin to do the normal argument checking before it calls the callback function by setting this to true. When false, the callback function is called instead of the argument verification.

    • default: false

  • :callback_include_count: Tell :callback plugin to include an extra parameter to specify the number of times the callback has been called. If set to false, the callback has the same interface as the mocked function. This can be handy when you’re wanting to use callback as a stub.

    • default: true

  • :cexception_include: Tell :cexception plugin where to find CException.h… You only need to define this if it’s not in your build path already… which it usually will be for the purpose of your builds.

    • default: nil

  • :enforce_strict_ordering: CMock always enforces the order that you call a particular function, so if you expect GrabNabber(int size) to be called three times, it will verify that the sizes are in the order you specified. You might also want to make sure that all different functions are called in a particular order. If so, set this to true.

    • default: false

  • :framework: Currently the only option is :unity. Eventually if we support other unity test frameworks (or if you write one for us), they’ll get added here.

    : default: :unity

  • :includes: An array of additional include files which should be added to the mocks. Useful for global types and definitions used in your project. There are more specific versions if you care WHERE in the mock files the includes get placed. You can define any or all of these options.

    • :includes

    • :includes_h_pre_orig_header

    • :includes_h_post_orig_header

    • :includes_c_pre_header

    • :includes_c_post_header

    • default: nil #for all 5 options

  • :memcmp_if_unknown: C developers create a lot of types, either through typedef or preprocessor macros. CMock isn’t going to automatically know what you were thinking all the time (though it tries its best). If it comes across a type it doesn’t recognize, you have a choice on how you want it to handle it. It can either perform a raw memory comparison and report any differences, or it can fail with a meaningful message. Either way, this feature will only happen after all other mechanisms have failed (The thing encountered isn’t a standard type. It isn’t in the :treat_as list. It isn’t in a custom unity_helper).

    • default: true

  • :mock_path: The directory where you would like the mock files generated to be placed.

    • default: mocks

  • :mock_prefix: The prefix to prepend to your mock files. For example, if it’s Mock, a file “USART.h” will get a mock called “MockUSART.c”. This CAN be used with a suffix at the same time.

    • default: Mock

  • :mock_suffix: The suffix to append to your mock files. For example, it it’s _Mock, a file “USART.h” will get a mock called “USART_Mock.h”. This CAN be used with a prefix at the same time.

    • default: “”

  • :plugins: An array of which plugins to enable. ‘:expect’ is always active. Also available currently:

    • :ignore

    • :ignore_stateless

    • :ignore_arg

    • :expect_any_args

    • :array

    • :cexception

    • :callback

    • :return_thru_ptr

  • :strippables: An array containing a list of items to remove from the header before deciding what should be mocked. This can be something simple like a compiler extension CMock wouldn’t recognize, or could be a regex to reject certain function name patterns. This is a great way to get rid of compiler extensions when your test compiler doesn’t support them. For example, use :strippables: ['(?:functionName\s*\(+.*?\)+)'] to prevent a function functionName from being mocked. By default, it is ignoring all gcc attribute extensions.

    • default: ['(?:__attribute__\s*\(+.*?\)+)']

  • :exclude_setjmp_h: Some embedded systems don’t have <setjmp.h> available. Setting this to true removes references to this header file and the ability to use cexception.

    • default: false

  • :subdir: This is a relative subdirectory for your mocks. Set this to e.g. “sys” in order to create a mock for sys/types.h in (:mock_path)/sys/.

    • default: “”

  • :treat_as: The :treat_as list is a shortcut for when you have created typedefs of standard types. Why create a custom unity helper for UINT16 when the unity function TEST_ASSERT_EQUAL_HEX16 will work just perfectly? Just add ‘UINT16’ => ‘HEX16’ to your list (actually, don’t. We already did that one for you). Maybe you have a type that is a pointer to an array of unsigned characters? No problem, just add ‘UINT8_T*’ => ‘HEX8*’

    • NOTE: unlike the other options, your specifications MERGE with the default list. Therefore, if you want to override something, you must reassign it to something else (or to nil if you don’t want it)

    • default:

      • ‘int’: ‘INT’

      • ‘char’: ‘INT8’

      • ‘short’: ‘INT16’

      • ‘long’: ‘INT’

      • ‘int8’: ‘INT8’

      • ‘int16’: ‘INT16’

      • ‘int32’: ‘INT’

      • ‘int8_t’: ‘INT8’

      • ‘int16_t’: ‘INT16’

      • ‘int32_t’: ‘INT’

      • ‘INT8_T’: ‘INT8’

      • ‘INT16_T’: ‘INT16’

      • ‘INT32_T’: ‘INT’

      • ‘bool’: ‘INT’

      • ‘bool_t’: ‘INT’

      • ‘BOOL’: ‘INT’

      • ‘BOOL_T’: ‘INT’

      • ‘unsigned int’: ‘HEX32’

      • ‘unsigned long’: ‘HEX32’

      • ‘uint32’: ‘HEX32’

      • ‘uint32_t’: ‘HEX32’

      • ‘UINT32’: ‘HEX32’

      • ‘UINT32_T’: ‘HEX32’

      • ‘void*’: ‘HEX8_ARRAY’

      • ‘unsigned short’: ‘HEX16’

      • ‘uint16’: ‘HEX16’

      • ‘uint16_t’: ‘HEX16’

      • ‘UINT16’: ‘HEX16’

      • ‘UINT16_T’: ‘HEX16’

      • ‘unsigned char’: ‘HEX8’

      • ‘uint8’: ‘HEX8’

      • ‘uint8_t’: ‘HEX8’

      • ‘UINT8’: ‘HEX8’

      • ‘UINT8_T’: ‘HEX8’

      • ‘char*’: ‘STRING’

      • ‘pCHAR’: ‘STRING’

      • ‘cstring’: ‘STRING’

      • ‘CSTRING’: ‘STRING’

      • ‘float’: ‘FLOAT’

      • ‘double’: ‘FLOAT’

  • :treat_as_array: A specialized sort of :treat_as to be used when you’ve created a typedef of an array type, such as typedef int TenIntegers[10];. This is a hash of typedef name to element type. For example:

      { "TenIntegers" => "int",
        "ArrayOfFloat" => "float" }
    

    Telling CMock about these typedefs allows it to be more intelligent about parameters of such types, so that you can use features like ExpectWithArray and ReturnArrayThruPtr with them.

  • :treat_as_void: We’ve seen “fun” legacy systems typedef ‘void’ with a custom type, like MY_VOID. Add any instances of those to this list to help CMock understand how to deal with your code.

    • default: []

  • :treat_externs: This specifies how you want CMock to handle functions that have been marked as extern in the header file. Should it mock them?

    • :include will mock externed functions

    • :exclude will ignore externed functions (default).

  • :treat_inlines: This specifies how you want CMock to handle functions that have been marked as inline in the header file. Should it mock them?

    • :include will mock inlined functions

    • :exclude will ignore inlined functions (default).

    CMock will look for the following default patterns (simplified from the actual regex):

    • “static inline”

    • “inline static”

    • “inline”

    • “static” You can override these patterns, check out :inline_function_patterns.

    Enabling this feature does require a change in the build system that is using CMock. To understand why, we need to give some more info on how we are handling inline functions internally. Let’s say we want to mock a header called example.h. example.h contains inline functions, we cannot include this header in the mocks or test code if we want to mock the inline functions simply because the inline functions contain an implementation that we want to override in our mocks! So, to circumvent this, we generate a new header, also named example.h, in the same directory as mock_example.h/c . This newly generated header should/is exactly the same as the original header, only difference is the inline functions are transformed to ‘normal’ functions declarations. Placing the new header in the same directory as mock_example.h/c ensures that they will include the new header and not the old one. However, CMock has no control in how the build system is configured and which include paths the test code is compiled with. In order for the test code to also see the newly generated header ,and not the old header with inline functions, the build system has to add the mock folder to the include paths. Furthermore, we need to keep the order of include paths in mind. We have to set the mock folder before the other includes to avoid the test code including the original header instead of the newly generated header (without inline functions).

  • :unity_helper_path: If you have created a header with your own extensions to unity to handle your own types, you can set this argument to that path. CMock will then automagically pull in your helpers and use them. The only trick is that you make sure you follow the naming convention: UNITY_TEST_ASSERT_EQUAL_YourType. If it finds macros of the right shape that match that pattern, it’ll use them.

    • default: []

  • :verbosity: How loud should CMock be?

    • 0 for errors only

    • 1 for errors and warnings

    • 2 for normal (default)

    • 3 for verbose

  • :weak: When set this to some value, the generated mocks are defined as weak symbols using the configured format. This allows them to be overridden in particular tests.

    • Set to ‘__attribute ((weak))’ for weak mocks when using GCC.

    • Set to any non-empty string for weak mocks when using IAR.

    • default: “”

  • :when_no_prototypes: When you give CMock a header file and ask it to create a mock out of it, it usually contains function prototypes (otherwise what was the point?). You can control what happens when this isn’t true. You can set this to :warn, :ignore, or :error

    • default: :warn

  • :when_ptr: You can customize how CMock deals with pointers (c strings result in string comparisons… we’re talking about other pointers here). Your options are :compare_ptr to just verify the pointers are the same, :compare_data or :smart to verify that the data is the same. :compare_data and :smart behaviors will change slightly based on if you have the array plugin enabled. By default, they compare a single element of what is being pointed to. So if you have a pointer to a struct called ORGAN_T, it will compare one ORGAN_T (whatever that is).

    • default: :smart

  • :array_size_type:

  • :array_size_name: When the :array plugin is disabled, these options do nothing.

    When the :array plugin is enabled, these options allow CMock to recognize functions with parameters that might refer to an array, like the following, and treat them more intelligently:

    • void GoBananas(Banana * bananas, int num_bananas)

    • int write_data(int fd, const uint8_t * data, uint32_t size)

    To recognize functions like these, CMock looks for a parameter list containing a pointer (which could be an array) followed by something that could be an array size. “Something”, by default, means an int or size_t parameter with a name containing “size” or “len”.

    :array_size_type is a list of additional types (besides int and size_t) that could be used for an array size parameter. For example, to get CMock to recognize that uint32_t size is an array size, you’d need to say:

      cfg[:array_size_type] = ['uint32_t']
    

    :array_size_name is a regular expression used to match an array size parameter by name. By default, it’s ‘size|len’. To get CMock to recognize a name like num_bananas, you could tell it to also accept names containing ‘num_’ like this:

      cfg[:array_size_name] = 'size|len|num_'
    

    Parameters must match both :array_size_type and :array_size_name (and must come right after a pointer parameter) to be treated as an array size.

    Once you’ve told it how to recognize your arrays, CMock will give you _Expect calls that work more like _ExpectWithArray, and compare an array of objects rather than just a single object.

    For example, if you write the following, CMock will check that GoBananas is called and passed an array containing a green banana followed by a yellow banana:

      Banana b[2] = {GreenBanana, YellowBanana};
      GoBananas_Expect(b, 2);
    

    In other words, GoBananas_Expect(b, 2) now works just the same as:

      GoBananas_ExpectWithArray(b, 2, 2);
    
  • :fail_on_unexpected_calls: By default, CMock will fail a test if a mock is called without _Expect and _Ignore called first. While this forces test writers to be more explicit in their expectations, it can clutter tests with _Expect or _Ignore calls for functions which are not the focus of the test. While this is a good indicator that this module should be refactored, some users are not fans of the additional noise.

    Therefore, :fail_on_unexpected_calls can be set to false to force all mocks to start with the assumption that they are operating as _Ignore unless otherwise specified.

    • default: true

    • note: If this option is disabled, the mocked functions will return a default value (0) when called (and only if they have to return something of course).

  • :inline_function_patterns: An array containing a list of strings to detect inline functions. This option is only taken into account if you enable :treat_inlines. These strings are interpreted as regex patterns so be sure to escape certain characters. For example, use :inline_function_patterns: ['static inline __attribute__ \(\(always_inline\)\)'] to recognize static inline __attribute__ ((always_inline)) int my_func(void) as an inline function. The default patterns are are:

    • default: [‘(static\s+inline|inline\s+static)\s*’, ‘(\bstatic\b|\binline\b)\s*’]

    • note: The order of patterns is important here! We go from specific patterns (‘static inline’) to general patterns (‘inline’), otherwise we would miss functions that use ‘static inline’ iso ‘inline’.

Compiled Options:

A number of #defines also exist for customizing the cmock experience. Feel free to pass these into your compiler or whatever is most convenient. CMock will otherwise do its best to guess what you want based on other settings, particularly Unity’s settings.

  • CMOCK_MEM_STATIC or CMOCK_MEM_DYNAMIC Define one of these to determine if you want to dynamically add memory during tests as required from the heap. If static, you can control the total footprint of Cmock. If dynamic, you will need to make sure you make some heap space available for Cmock.

  • CMOCK_MEM_SIZE In static mode this is the total amount of memory you are allocating to Cmock. In Dynamic mode this is the size of each chunk allocated at once (larger numbers grab more memory but require fewer mallocs).

  • CMOCK_MEM_ALIGN The way to align your data to. Not everything is as flexible as a PC, as most embedded designers know. This defaults to 2, meaning align to the closest 2^2 -> 4 bytes (32 bits). You can turn off alignment by setting 0, force alignment to the closest uint16 with 1 or even to the closest uint64 with 3.

  • CMOCK_MEM_PTR_AS_INT This is used internally to hold pointers… it needs to be big enough. On most processors a pointer is the same as an unsigned long… but maybe that’s not true for yours?

  • CMOCK_MEM_INDEX_TYPE This needs to be something big enough to point anywhere in Cmock’s memory space… usually it’s a size_t.

Other Tips

resetTest

While this isn’t strictly a CMock feature, often users of CMock are using either the test runner generator scripts in Unity or using Ceedling. In either case, there is a handy function called resetTest which gets generated with your runner. You can then use this handy function in your tests themselves. Call it during a test to have CMock validate everything to this point and start over clean. This is really useful when wanting to test a function in an iterative manner with different arguments.

C++ Support

C++ unit test/mocking frameworks often use a completely different approach (vs. CMock) that relies on overloading virtual class members and does not support directly mocking static class member methods or free functions (i.e., functions in plain C). One workaround is to wrap the non-virtual functions in an object that exposes them as virtual methods and modify your code to inject mocks at run-time… but there is another way!

Simply use CMock to mock the static member methods and a C++ mocking framework to handle the virtual methods. (Yes, you can mix mocks from CMock and a C++ mocking framework together in the same test!)

Keep in mind that since C++ mocking frameworks often link the real object to the unit test too, we need to resolve multiple definition errors with something like the following in the source of the real implementation for any functions that CMock mocks:

#if defined(TEST)
    __attribute__((weak))
#endif

To address potential issues with re-using the same function name in different namespaces/classes, the generated function names include the namespace(s) and class. For example:

namespace MyNamespace {
    class MyClass {
        static int DoesSomething(int a, int b);
    };
}

Will generate functions like

void MyNamespace_MyClass_DoesSomething_ExpectAndReturn(int a, int b, int toReturn);

Examples

You can look in the examples directory for a couple of examples on how you might tool CMock into your build process. You may also want to consider using Ceedling. Please note that these examples are meant to show how the build process works. They have failing tests ON PURPOSE to show what that would look like. Don’t be alarmed. ;)