Using -Wl,--wrap for mocking in C July 19, 2016 on Drew DeVault's blog

One of the comforts I’ve grown used to in higher level languages when testing my code is mocking. The idea is that in order to test some code in isolation, you should “mock” the behavior of things it depends on. Let’s see a (contrived) example:

int read_to_end(FILE *f, char *buf) {
    int r = 0, l;
    while (!feof(f)) {
        l = fread(buf, 1, 256, f);
        r += l;
        buf += l;
    }
    return r;
}

If we want to test this function without mocking, we would need to actually open a specially crafted file and provide a FILE* to the function. However, with the linker --wrap flag, we can define a wrapper function. Using -Wl,[flag] in your C compiler command line will pass [flag] to the linker. Gold (GNU) and lld (LLVM) both support the wrap flag, which specifies a function to be “wrapped”. If I use -Wl,--wrap=fread, then the code above will be compiled like so:

int read_to_end(FILE *f, char *buf) {
    int r = 0, l;
    while (!feof(f)) {
        l = __wrap_fread(buf, 1, 256, f);
        r += l;
        buf += l;
    }
    return r;
}

And if I add -Wl,--wrap=feof we’ll get this:

int read_to_end(FILE *f, char *buf) {
    int r = 0, l;
    while (!__wrap_feof(f)) {
        l = __wrap_fread(buf, 1, 256, f);
        r += l;
        buf += l;
    }
    return r;
}

Now, we can define some functions that do the behavior we need to test instead of invoking fread directly:

int feof_return_value = 0;

int __wrap_feof(FILE *f) {
    assert(f == (FILE *)0x1234);
    return feof_return_value;
}

void test_read_to_end_eof() {
    // ...
    feof_return_value = 1;
    read_to_end((FILE *)0x1234, buf);
    // ...
}

Using --wrap also conveniently defines __real_feof and __real_fread if we need them.

Unfortunately, you can’t have two different wrappers for the same function in an executable. This could lead to having to write several executables for each, or making your wrapper function smart enough to have several configurable outcomes.

Eventually I intend to write my own test framework for C, which will use wrappers to support mocking. I want wrappers to be done automatically and have it behave something like this:

static int fake_fread(void *ptr, size_t size,
        size_t nmemb, FILE *stream) {
    const char *hello = "Hello world!";
    memcpy(ptr, hello, strlen(hello) + 1);
    return strlen(hello) + 1;
}

void test_read_to_end() {
    FILE *test = (FILE *)0x1234;
    char *buffer = char[1024];

    mock_t *mock_feof = configure_mock(feof, "p");
    mock_feof->call(0)->returns(0);
    mock_feof->returns(1);

    // pzzp is pointer, size_t, size_t, pointer
    // Tells us what the fread arguments look like
    mock_t *mock_fread = configure_mock(fread, "pzzp");
    mock_fread->exec(fake_fread);

    read_to_end(test, buffer);

    assert(mock_feof->call_count == 2);
    assert((FILE*)mock_feof->call(0)->args(0) == test);

    assert(mock_fread->call_count == 1);
    assert((FILE*)mock_fread->call(0)->args(0) == buffer);
    assert((FILE*)mock_fread->call(0)->args(3) == test);

    assert(strcmp(buffer, "Hello world!") == 0);
}

Articles from blogs I read Generated by openring

What's cooking on SourceHut? Q2 2025

Hello everyone! We’re back with another quarterly update for you on how things are going at SourceHut. It’s been a busy few months, and we’re happy to share what we’re working on with you all. Drew’s updates As readers of my personal blog might have noticed w…

via Blogs on Sourcehut May 29, 2025

My perfect music synchronization solution

I'm buying DRM free music for a long time, and I synchronized those music files across my devices using Syncthing. But my library weight growth, now weighting 60G… Some of my devices can't afford this! I had to find an alternative. Years ago I swit…

via Willow's feed May 23, 2025

The Future of Customer Support is Lies, I Guess

Update, 2025-05-22: TrueNAS was kind enough to reach out and let me know that their support process does not normally incorporate LLMs. They’re talking about what happened internally, and intend to prevent it from happening again through improved document…

via Aphyr: Posts May 21, 2025