8. Software Test Framework#

8.1. The RTEMS Test Framework#

The RTEMS Test Framework helps you to write test suites. It has the following features:

Implemented in standard C11
Tests can be written in C or C++
Runs on at least FreeBSD, MSYS2, Linux and RTEMS
Test runner and test case code can be in separate translation units
Test cases are automatically registered at link-time
Test cases may have a test fixture
Test checks for various standard types
Supports test case planning
Test case scoped dynamic memory
Test case destructors
Test case resource accounting to show that no resources are leaked during the test case execution
Supports early test case exit, e.g. in case a malloc() fails
Individual test case and overall test suite duration is reported
Procedures for code runtime measurements in RTEMS
Easy to parse test report to generate for example human readable test reports
Low overhead time measurement of short time sequences (using cycle counter hardware if a available)
Configurable time service provider for a monotonic clock
Low global memory overhead for test cases and test checks
Supports multi-threaded execution and interrupts in test cases
A simple (polled) put character function is sufficient to produce the test report
Only text, global data and a stack pointer must be set up to run a test suite
No dynamic memory is used by the framework itself
No memory is aggregated throughout the test case execution

8.1.1. Nomenclature#

A test suite is a collection of test cases. A test case consists of individual test actions and checks. A test check determines if the outcome of a test action meets its expectation. A test action is a program sequence with an observable outcome, for example a function invocation with a return status. If a test action produces the expected outcome as determined by the corresponding test check, then this test check passes, otherwise this test check fails. The test check failures of a test case are summed up. A test case passes, if the failure count of this test case is zero, otherwise the test case fails. The test suite passes if all test cases pass, otherwise it fails.

8.1.2. Test Cases#

You can write a test case with the T_TEST_CASE() macro followed by a function body:

T_TEST_CASE(name)
{
   /* Your test case code */
}

The test case name must be a valid C designator. The test case names must be unique within the test suite. Just link modules with test cases to the test runner to form a test suite. The test cases are automatically registered via static C constructors.

Listing 8.1 Test Case Example#

#include <t.h>

static int add(int a, int b)
{
    return a + b;
}

T_TEST_CASE(a_test_case)
{
    int actual_value;

    actual_value = add(1, 1);
    T_eq_int(actual_value, 2);
    T_true(false, "a test failure message");
}

Listing 8.2 Test Case Report#

B:a_test_case
P:0:8:UI1:test-simple.c:13
F:1:8:UI1:test-simple.c:14:a test failure message
E:a_test_case:N:2:F:1:D:0.001657

The B line indicates the begin of test case a_test_case. The P line shows that the test check in file test-simple.c at line 13 executed by task UI1 on processor 0 as the test step 0 passed. The invocation of add() in line 12 is the test action of test step 0. The F lines shows that the test check in file test-simple.c at line 14 executed by task UI1 on processor 0 as the test step 1 failed with a message of “a test failure message”. The E line indicates the end of test case a_test_case resulting in a total of two test steps (N) and one test failure (F). The test case execution duration (D) was 0.001657 seconds. For test report details see: Test Reporting.

8.1.3. Test Fixture#

You can write a test case with a test fixture with the T_TEST_CASE_FIXTURE() macro followed by a function body:

T_TEST_CASE_FIXTURE(name, fixture)
{
   /* Your test case code */
}

The test case name must be a valid C designator. The test case names must be unique within the test suite. The fixture must point to a statically initialized read-only object of type T_fixture.

typedef struct T_fixture {
    void (*setup)(void *context);
    void (*stop)(void *context);
    void (*teardown)(void *context);
    void (*scope)(void *context, char *buffer, size_t size);
    void *initial_context;
} T_fixture;

The test fixture provides methods to setup, stop, and teardown a test case as well as the scope for log messages. A context is passed to each of the methods. The initial context is defined by the read-only fixture object. The context can be obtained by the T_fixture_context() function. The context can be changed within the scope of one test case by the T_set_fixture_context() function. The next test case execution using the same fixture will start again with the initial context defined by the read-only fixture object. Setting the context can be used for example to dynamically allocate a test environment in the setup method.

The test case fixtures of a test case are organized as a stack. Fixtures can be dynamically added to a test case and removed from a test case via the T_push_fixture() and T_pop_fixture() functions.

void *T_push_fixture(T_fixture_node *node, const T_fixture *fixture);

void T_pop_fixture(void);

The T_push_fixture() function needs an uninitialized fixture node which must exist until T_pop_fixture() is called. It returns the initial context of the fixture. At the end of a test case all pushed fixtures are popped automatically. A call of T_pop_fixture() invokes the teardown method of the fixture and must correspond to a previous call to T_push_fixture().

Listing 8.3 Test Fixture Example#

#include <t.h>

static int initial_value = 3;

static int counter;

static void
setup(void *ctx)
{
    int *c;

    T_log(T_QUIET, "setup begin");
    T_eq_ptr(ctx, &initial_value);
    T_eq_ptr(ctx, T_fixture_context());
    c = ctx;
    counter = *c;
    T_set_fixture_context(&counter);
    T_eq_ptr(&counter, T_fixture_context());
    T_log(T_QUIET, "setup end");
}

static void
stop(void *ctx)
{
    int *c;

    T_log(T_QUIET, "stop begin");
    T_eq_ptr(ctx, &counter);
    c = ctx;
    ++(*c);
    T_log(T_QUIET, "stop end");
}

static void
teardown(void *ctx)
{
    int *c;

    T_log(T_QUIET, "teardown begin");
    T_eq_ptr(ctx, &counter);
    c = ctx;
    T_eq_int(*c, 4);
    T_log(T_QUIET, "teardown end");
}

static const T_fixture fixture = {
    .setup = setup,
    .stop = stop,
    .teardown = teardown,
    .initial_context = &initial_value
};

T_TEST_CASE_FIXTURE(fixture, &fixture)
{
    T_assert_true(true, "all right");
    T_assert_true(false, "test fails and we stop the test case");
    T_log(T_QUIET, "not reached");
}

Listing 8.4 Test Fixture Report#

B:fixture
L:setup begin
P:0:0:UI1:test-fixture.c:13
P:1:0:UI1:test-fixture.c:14
P:2:0:UI1:test-fixture.c:18
L:setup end
P:3:0:UI1:test-fixture.c:55
F:4:0:UI1:test-fixture.c:56:test fails and we stop the test case
L:stop begin
P:5:0:UI1:test-fixture.c:28
L:stop end
L:teardown begin
P:6:0:UI1:test-fixture.c:40
P:7:0:UI1:test-fixture.c:42
L:teardown end
E:fixture:N:8:F:1

8.1.4. Test Case Planning#

A non-quiet test check fetches and increments the test step counter atomically. For each test case execution the planned steps can be specified with the T_plan() function.

void T_plan(unsigned int planned_steps);

This function must be invoked at most once in each test case execution. If the planned test steps are set with this function, then the final test steps after the test case execution must be equal to the planned steps, otherwise the test case fails.

Use the T_step_*(step, …) test check variants to ensure that the test case execution follows exactly the planned steps.

Listing 8.5 Test Planning Example#

#include <t.h>

T_TEST_CASE(wrong_step)
{
    T_plan(2);
    T_step_true(0, true, "all right");
    T_step_true(2, true, "wrong step");
}

T_TEST_CASE(plan_ok)
{
    T_plan(1);
    T_step_true(0, true, "all right");
}

T_TEST_CASE(plan_failed)
{
    T_plan(2);
    T_step_true(0, true, "not enough steps");
    T_quiet_true(true, "quiet test do not count");
}

T_TEST_CASE(double_plan)
{
    T_plan(99);
    T_plan(2);
}

T_TEST_CASE(steps)
{
    T_step(0, "a");
    T_plan(3);
    T_step(1, "b");
    T_step(2, "c");
}

Listing 8.6 Test Planning Report#

B:wrong_step
P:0:0:UI1:test-plan.c:6
F:1:0:UI1:test-plan.c:7:planned step (2)
E:wrong_step:N:2:F:1
B:plan_ok
P:0:0:UI1:test-plan.c:13
E:plan_ok:N:1:F:0
B:plan_failed
P:0:0:UI1:test-plan.c:19
F:*:0:UI1:*:*:actual steps (1), planned steps (2)
E:plan_failed:N:1:F:1
B:double_plan
F:*:0:UI1:*:*:planned steps (99) already set
E:double_plan:N:0:F:1
B:steps
P:0:0:UI1:test-plan.c:31
P:1:0:UI1:test-plan.c:33
P:2:0:UI1:test-plan.c:34
E:steps:N:3:F:0

8.1.5. Test Case Resource Accounting#

The framework can check if various resources have leaked during a test case execution. The resource checkers are specified by the test run configuration. On RTEMS, checks for the following resources are available

workspace and heap memory,
file descriptors,
POSIX keys and key value pairs,
RTEMS barriers,
RTEMS user extensions,
RTEMS message queues,
RTEMS partitions,
RTEMS periods,
RTEMS regions,
RTEMS semaphores,
RTEMS tasks, and
RTEMS timers.

Listing 8.7 Resource Accounting Example#

#include <t.h>

#include <stdlib.h>

#include <rtems.h>

T_TEST_CASE(missing_sema_delete)
{
    rtems_status_code sc;
    rtems_id id;

    sc = rtems_semaphore_create(rtems_build_name('S', 'E', 'M', 'A'), 0,
        RTEMS_COUNTING_SEMAPHORE, 0, &id);
    T_rsc_success(sc);
}

T_TEST_CASE(missing_free)
{
    void *p;

    p = malloc(1);
    T_not_null(p);
}

Listing 8.8 Resource Accounting Report#

B:missing_sema_delete
P:0:0:UI1:test-leak.c:14
F:*:0:UI1:*:*:RTEMS semaphore leak (1)
E:missing_sema_delete:N:1:F:1:D:0.004013
B:missing_free
P:0:0:UI1:test-leak.c:22
F:*:0:UI1:*:*:memory leak in workspace or heap
E:missing_free:N:1:F:1:D:0.003944

8.1.6. Test Case Scoped Dynamic Memory#

You can allocate dynamic memory which is automatically freed after the current test case execution. You can provide an optional destroy function to T_zalloc() which is called right before the memory is freed. The T_zalloc() function initializes the memory to zero.

void *T_malloc(size_t size);

void *T_calloc(size_t nelem, size_t elsize);

void *T_zalloc(size_t size, void (*destroy)(void *));

void T_free(void *ptr);

Listing 8.9 Test Case Scoped Dynamic Memory Example#

#include <t.h>

T_TEST_CASE(malloc_free)
{
    void *p;

    p = T_malloc(1);
    T_assert_not_null(p);
    T_free(p);
}

T_TEST_CASE(malloc_auto)
{
    void *p;

    p = T_malloc(1);
    T_assert_not_null(p);
}

static void
destroy(void *p)
{
    int *i;

    i = p;
    T_step_eq_int(2, *i, 1);
}

T_TEST_CASE(zalloc_auto)
{
    int *i;

    T_plan(3);
    i = T_zalloc(sizeof(*i), destroy);
    T_step_assert_not_null(0, i);
    T_step_eq_int(1, *i, 0);
    *i = 1;
}

Listing 8.10 Test Case Scoped Dynamic Memory Report#

B:malloc_free
P:0:0:UI1:test-malloc.c:8
E:malloc_free:N:1:F:0:D:0.005200
B:malloc_auto
P:0:0:UI1:test-malloc.c:17
E:malloc_auto:N:1:F:0:D:0.004790
B:zalloc_auto
P:0:0:UI1:test-malloc.c:35
P:1:0:UI1:test-malloc.c:36
P:2:0:UI1:test-malloc.c:26
E:zalloc_auto:N:3:F:0:D:0.006583

8.1.7. Test Case Destructors#

You can add test case destructors with T_add_destructor(). The destructors are called automatically at the test case end before the resource accounting takes place. Optionally, a registered destructor can be removed before the test case end with T_remove_destructor(). The T_destructor structure of a destructor must exist after the return from the test case body. It is recommended to use statically allocated memory. Do not use stack memory or dynamic memory obtained via T_malloc(), T_calloc() or T_zalloc() for the T_destructor structure.

void T_add_destructor(T_destructor *destructor,
   void (*destroy)(T_destructor *));

void T_remove_destructor(T_destructor *destructor);

Listing 8.11 Test Case Destructor Example#

#include <t.h>

static void
destroy(T_destructor *dtor)
{
    (void)dtor;
    T_step(0, "destroy");
}

T_TEST_CASE(destructor)
{
    static T_destructor dtor;

    T_plan(1);
    T_add_destructor(&dtor, destroy);
}

Listing 8.12 Test Case Destructor Report#

B:destructor
P:0:0:UI1:test-destructor.c:7
E:destructor:N:1:F:0:D:0.003714

8.1.8. Test Checks#

A test check determines if the actual value presented to the test check has the expected properties. The actual value should represent the outcome of a test action. If a test action produces the expected outcome as determined by the corresponding test check, then this test check passes, otherwise this test check fails. A failed test check does not stop the test case execution immediately unless the T_assert_*() test variant is used. Each test check increments the test step counter unless the T_quiet_*() test variant is used. The test step counter is initialized to zero before the test case begins to execute. The T_step_*(step, …) test check variants verify that the test step counter is equal to the planned test step value, otherwise the test check fails.

8.1.8.1. Test Check Variant Conventions#

The T_quiet_*() test check variants do not increment the test step counter and only print a message if the test check fails. This is helpful in case a test check appears in a tight loop.

The T_step_*(step, …) test check variants check in addition that the test step counter is equal to the specified test step value, otherwise the test check fails.

The T_assert_*() and T_step_assert_*(step, …) test check variants stop the current test case execution if the test check fails.

8.1.8.2. Test Check Parameter Conventions#

The following names for test check parameters are used throughout the test checks:

step: The planned test step for this test check.
a: The actual value to check against an expected value. It is usually the first parameter in all test checks, except in the T_step_*(step, …) test check variants, here it is the second parameter.
e: The expected value of a test check. This parameter is optional. Some test checks have an implicit expected value. If present, then this parameter is directly after the actual value parameter of the test check.
fmt: A printf()-like format string. Floating-point and exotic formats may be not supported.

8.1.8.3. Test Check Condition Conventions#

The following names for test check conditions are used:

eq: The actual value must equal the expected value.
ne: The actual value must not equal the value of the second parameter.
ge: The actual value must be greater than or equal to the expected value.
gt: The actual value must be greater than the expected value.
le: The actual value must be less than or equal to the expected value.
lt: The actual value must be less than the expected value.

If the actual value satisfies the test check condition, then the test check passes, otherwise it fails.

8.1.8.4. Test Check Type Conventions#

The following names for test check types are used:

ptr: The test value must be a pointer (void *).
mem: The test value must be a memory area with a specified length.
str: The test value must be a null byte terminated string.
nstr: The length of the test value string is limited to a specified maximum.
char: The test value must be a character (char).
schar: The test value must be a signed character (signed char).
uchar: The test value must be an unsigned character (unsigned char).
short: The test value must be a short integer (short).
ushort: The test value must be an unsigned short integer (unsigned short).
int: The test value must be an integer (int).
uint: The test value must be an unsigned integer (unsigned int).
long: The test value must be a long integer (long).
ulong: The test value must be an unsigned long integer (unsigned long).
ll: The test value must be a long long integer (long long).
ull: The test value must be an unsigned long long integer (unsigned long long).
i8: The test value must be a signed 8-bit integer (int8_t).
u8: The test value must be an unsigned 8-bit integer (uint8_t).
i16: The test value must be a signed 16-bit integer (int16_t).
u16: The test value must be an unsigned 16-bit integer (uint16_t).
i32: The test value must be a signed 32-bit integer (int32_t).
u32: The test value must be an unsigned 32-bit integer (uint32_t).
i64: The test value must be a signed 64-bit integer (int64_t).
u64: The test value must be an unsigned 64-bit integer (uint64_t).
iptr: The test value must be of type intptr_t.
uptr: The test value must be of type uintptr_t.
ssz: The test value must be of type ssize_t.
sz: The test value must be of type size_t.

8.1.8.5. Integers#

Let xyz be the type variant which shall be one of schar, uchar, short, ushort, int, uint, long, ulong, ll, ull, i8, u8, i16, u16, i32, u32, i64, u64, iptr, uptr, ssz, and sz.

Let I be the type name which shall be compatible to the type variant.

The following test checks for integers are available:

void T_eq_xyz(I a, I e);
void T_assert_eq_xyz(I a, I e);
void T_quiet_eq_xyz(I a, I e);
void T_step_eq_xyz(unsigned int step, I a, I e);
void T_step_assert_eq_xyz(unsigned int step, I a, I e);

void T_ne_xyz(I a, I e);
void T_assert_ne_xyz(I a, I e);
void T_quiet_ne_xyz(I a, I e);
void T_step_ne_xyz(unsigned int step, I a, I e);
void T_step_assert_ne_xyz(unsigned int step, I a, I e);

void T_ge_xyz(I a, I e);
void T_assert_ge_xyz(I a, I e);
void T_quiet_ge_xyz(I a, I e);
void T_step_ge_xyz(unsigned int step, I a, I e);
void T_step_assert_ge_xyz(unsigned int step, I a, I e);

void T_gt_xyz(I a, I e);
void T_assert_gt_xyz(I a, I e);
void T_quiet_gt_xyz(I a, I e);
void T_step_gt_xyz(unsigned int step, I a, I e);
void T_step_assert_gt_xyz(unsigned int step, I a, I e);

void T_le_xyz(I a, I e);
void T_assert_le_xyz(I a, I e);
void T_quiet_le_xyz(I a, I e);
void T_step_le_xyz(unsigned int step, I a, I e);
void T_step_assert_le_xyz(unsigned int step, I a, I e);

void T_lt_xyz(I a, I e);
void T_assert_lt_xyz(I a, I e);
void T_quiet_lt_xyz(I a, I e);
void T_step_lt_xyz(unsigned int step, I a, I e);
void T_step_assert_lt_xyz(unsigned int step, I a, I e);

An automatically generated message is printed in case the test check fails.

8.1.8.6. Boolean Expressions#

The following test checks for boolean expressions are available:

void T_true(bool a, const char *fmt, ...);
void T_assert_true(bool a, const char *fmt, ...);
void T_quiet_true(bool a, const char *fmt, ...);
void T_step_true(unsigned int step, bool a, const char *fmt, ...);
void T_step_assert_true(unsigned int step, bool a, const char *fmt, ...);

void T_false(bool a, const char *fmt, ...);
void T_assert_false(bool a, const char *fmt, ...);
void T_quiet_false(bool a, const char *fmt, ...);
void T_step_false(unsigned int step, bool a, const char *fmt, ...);
void T_step_assert_false(unsigned int step, bool a, const char *fmt, ...);

The message is only printed in case the test check fails. The format parameter is mandatory.

Listing 8.13 Boolean Test Checks Example#

#include <t.h>

T_TEST_CASE(example)
{
    T_true(true, "test passes, no message output");
    T_true(false, "test fails");
    T_quiet_true(true, "quiet test passes, no output at all");
    T_quiet_true(false, "quiet test fails");
    T_step_true(2, true, "step test passes, no message output");
    T_step_true(3, false, "step test fails");
    T_assert_false(true, "this is a format %s", "string");
}

Listing 8.14 Boolean Test Checks Report#

B:example
P:0:0:UI1:test-example.c:5
F:1:0:UI1:test-example.c:6:test fails
F:*:0:UI1:test-example.c:8:quiet test fails
P:2:0:UI1:test-example.c:9
F:3:0:UI1:test-example.c:10:step test fails
F:4:0:UI1:test-example.c:11:this is a format string
E:example:N:5:F:4

8.1.8.7. Generic Types#

The following test checks for data types with an equality (==) or inequality (!=) operator are available:

void T_eq(T a, T e, const char *fmt, ...);
void T_assert_eq(T a, T e, const char *fmt, ...);
void T_quiet_eq(T a, T e, const char *fmt, ...);
void T_step_eq(unsigned int step, T a, T e, const char *fmt, ...);
void T_step_assert_eq(unsigned int step, T a, T e, const char *fmt, ...);

void T_ne(T a, T e, const char *fmt, ...);
void T_assert_ne(T a, T e, const char *fmt, ...);
void T_quiet_ne(T a, T e, const char *fmt, ...);
void T_step_ne(unsigned int step, T a, T e, const char *fmt, ...);
void T_step_assert_ne(unsigned int step, T a, T e, const char *fmt, ...);

The type name T specifies an arbitrary type which must support the corresponding operator. The message is only printed in case the test check fails. The format parameter is mandatory.

8.1.8.8. Pointers#

The following test checks for pointers are available:

void T_eq_ptr(const void *a, const void *e);
void T_assert_eq_ptr(const void *a, const void *e);
void T_quiet_eq_ptr(const void *a, const void *e);
void T_step_eq_ptr(unsigned int step, const void *a, const void *e);
void T_step_assert_eq_ptr(unsigned int step, const void *a, const void *e);

void T_ne_ptr(const void *a, const void *e);
void T_assert_ne_ptr(const void *a, const void *e);
void T_quiet_ne_ptr(const void *a, const void *e);
void T_step_ne_ptr(unsigned int step, const void *a, const void *e);
void T_step_assert_ne_ptr(unsigned int step, const void *a, const void *e);

void T_null(const void *a);
void T_assert_null(const void *a);
void T_quiet_null(const void *a);
void T_step_null(unsigned int step, const void *a);
void T_step_assert_null(unsigned int step, const void *a);

void T_not_null(const void *a);
void T_assert_not_null(const void *a);
void T_quiet_not_null(const void *a);
void T_step_not_null(unsigned int step, const void *a);
void T_step_assert_not_null(unsigned int step, const void *a);

An automatically generated message is printed in case the test check fails.

8.1.8.9. Memory Areas#

The following test checks for memory areas are available:

void T_eq_mem(const void *a, const void *e, size_t n);
void T_assert_eq_mem(const void *a, const void *e, size_t n);
void T_quiet_eq_mem(const void *a, const void *e, size_t n);
void T_step_eq_mem(unsigned int step, const void *a, const void *e, size_t n);
void T_step_assert_eq_mem(unsigned int step, const void *a, const void *e, size_t n);

void T_ne_mem(const void *a, const void *e, size_t n);
void T_assert_ne_mem(const void *a, const void *e, size_t n);
void T_quiet_ne_mem(const void *a, const void *e, size_t n);
void T_step_ne_mem(unsigned int step, const void *a, const void *e, size_t n);
void T_step_assert_ne_mem(unsigned int step, const void *a, const void *e, size_t n);

The memcmp() function is used to compare the memory areas. An automatically generated message is printed in case the test check fails.

8.1.8.10. Strings#

The following test checks for strings are available:

void T_eq_str(const char *a, const char *e);
void T_assert_eq_str(const char *a, const char *e);
void T_quiet_eq_str(const char *a, const char *e);
void T_step_eq_str(unsigned int step, const char *a, const char *e);
void T_step_assert_eq_str(unsigned int step, const char *a, const char *e);

void T_ne_str(const char *a, const char *e);
void T_assert_ne_str(const char *a, const char *e);
void T_quiet_ne_str(const char *a, const char *e);
void T_step_ne_str(unsigned int step, const char *a, const char *e);
void T_step_assert_ne_str(unsigned int step, const char *a, const char *e);

void T_eq_nstr(const char *a, const char *e, size_t n);
void T_assert_eq_nstr(const char *a, const char *e, size_t n);
void T_quiet_eq_nstr(const char *a, const char *e, size_t n);
void T_step_eq_nstr(unsigned int step, const char *a, const char *e, size_t n);
void T_step_assert_eq_nstr(unsigned int step, const char *a, const char *e, size_t n);

void T_ne_nstr(const char *a, const char *e, size_t n);
void T_assert_ne_nstr(const char *a, const char *e, size_t n);
void T_quiet_ne_nstr(const char *a, const char *e, size_t n);
void T_step_ne_nstr(unsigned int step, const char *a, const char *e, size_t n);
void T_step_assert_ne_nstr(unsigned int step, const char *a, const char *e, size_t n);

The strcmp() and strncmp() functions are used to compare the strings. An automatically generated message is printed in case the test check fails.

8.1.8.11. Characters#

The following test checks for characters (char) are available:

void T_eq_char(char a, char e);
void T_assert_eq_char(char a, char e);
void T_quiet_eq_char(char a, char e);
void T_step_eq_char(unsigned int step, char a, char e);
void T_step_assert_eq_char(unsigned int step, char a, char e);

void T_ne_char(char a, char e);
void T_assert_ne_char(char a, char e);
void T_quiet_ne_char(char a, char e);
void T_step_ne_char(unsigned int step, char a, char e);
void T_step_assert_ne_char(unsigned int step, char a, char e);

An automatically generated message is printed in case the test check fails.

8.1.8.12. RTEMS Status Codes#

The following test checks for RTEMS status codes are available:

void T_rsc(rtems_status_code a, rtems_status_code e);
void T_assert_rsc(rtems_status_code a, rtems_status_code e);
void T_quiet_rsc(rtems_status_code a, rtems_status_code e);
void T_step_rsc(unsigned int step, rtems_status_code a, rtems_status_code e);
void T_step_assert_rsc(unsigned int step, rtems_status_code a, rtems_status_code e);

void T_rsc_success(rtems_status_code a);
void T_assert_rsc_success(rtems_status_code a);
void T_quiet_rsc_success(rtems_status_code a);
void T_step_rsc_success(unsigned int step, rtems_status_code a);
void T_step_assert_rsc_success(unsigned int step, rtems_status_code a);

An automatically generated message is printed in case the test check fails.

8.1.8.13. POSIX Error Numbers#

The following test checks for POSIX error numbers are available:

void T_eno(int a, int e);
void T_assert_eno(int a, int e);
void T_quiet_eno(int a, int e);
void T_step_eno(unsigned int step, int a, int e);
void T_step_assert_eno(unsigned int step, int a, int e);

void T_eno_success(int a);
void T_assert_eno_success(int a);
void T_quiet_eno_success(int a);
void T_step_eno_success(unsigned int step, int a);
void T_step_assert_eno_success(unsigned int step, int a);

The actual and expected value must be a POSIX error number, e.g. EINVAL, ENOMEM, etc. An automatically generated message is printed in case the test check fails.

8.1.8.14. POSIX Status Codes#

The following test checks for POSIX status codes are available:

void T_psx_error(int a, int eno);
void T_assert_psx_error(int a, int eno);
void T_quiet_psx_error(int a, int eno);
void T_step_psx_error(unsigned int step, int a, int eno);
void T_step_assert_psx_error(unsigned int step, int a, int eno);

void T_psx_success(int a);
void T_assert_psx_success(int a);
void T_quiet_psx_success(int a);
void T_step_psx_success(unsigned int step, int a);
void T_step_assert_psx_success(unsigned int step, int a);

The eno value must be a POSIX error number, e.g. EINVAL, ENOMEM, etc. An actual value of zero indicates success. An actual value of minus one indicates an error. An automatically generated message is printed in case the test check fails.

Listing 8.15 POSIX Status Code Example#

#include <t.h>

#include <sys/stat.h>
#include <errno.h>

T_TEST_CASE(stat)
{
    struct stat st;
    int status;

    errno = 0;
    status = stat("foobar", &st);
    T_psx_error(status, ENOENT);
}

Listing 8.16 POSIX Status Code Report#

B:stat
P:0:0:UI1:test-psx.c:13
E:stat:N:1:F:0

8.1.9. Log Messages and Formatted Output#

You can print log messages with the T_log() function:

void T_log(T_verbosity verbosity, char const *fmt, ...);

A newline is automatically added to terminate the log message line.

Listing 8.17 Log Message Example#

#include <t.h>

T_TEST_CASE(log)
{
    T_log(T_NORMAL, "a log message %i, %i, %i", 1, 2, 3);
    T_set_verbosity(T_QUIET);
    T_log(T_NORMAL, "not verbose enough");
}

Listing 8.18 Log Message Report#

B:log
L:a log message 1, 2, 3
E:log:N:0:F:0

You can use the following functions to print formatted output:

int T_printf(char const *, ...);

int T_vprintf(char const *, va_list);

int T_snprintf(char *, size_t, const char *, ...);

In contrast to the corresponding standard C library functions, floating-point and exotic formats may not be supported. On some architectures supported by RTEMS, floating-point operations are only supported in special tasks and may be forbidden in interrupt context. The formatted output functions provided by the test framework work in every context.

8.1.10. Utility#

You can stop a test case via the T_stop() function. This function does not return. You can indicate unreachable code paths with the T_unreachable() function. If this function is called, then the test case stops.

You can busy wait with the T_busy() function:

void T_busy(uint_fast32_t count);

It performs a busy loop with the specified iteration count. This function is optimized to not perform memory accesses and should have a small jitter. The loop iterations have a processor-specific duration.

You can get an iteration count for the T_busy() function which corresponds roughly to one clock tick interval with the T_get_one_clock_tick_busy() function:

uint_fast32_t T_get_one_clock_tick_busy(void);

This function requires a clock driver. It must be called from thread context with interrupts enabled. It may return a different value each time it is called.

8.1.11. Time Services#

The test framework provides two unsigned integer types for time values. The T_ticks unsigned integer type is used by the T_tick() function which measures time using the highest frequency counter available on the platform. It should only be used to measure small time intervals. The T_time unsigned integer type is used by the T_now() function which returns the current monotonic clock value of the platform, e.g. CLOCK_MONOTONIC.

T_ticks T_tick(void);

T_time T_now(void);

The reference time point for these two clocks is unspecified. You can obtain the test case begin time with the T_case_begin_time() function.

T_time T_case_begin_time(void);

You can convert time into ticks with the T_time_to_ticks() function and vice versa with the T_ticks_to_time() function.

T_time T_ticks_to_time(T_ticks ticks);

T_ticks T_time_to_ticks(T_time time);

You can convert seconds and nanoseconds values into a combined time value with the T_seconds_and_nanoseconds_to_time() function. You can convert a time value into separate seconds and nanoseconds values with the T_time_to_seconds_and_nanoseconds() function.

T_time T_seconds_and_nanoseconds_to_time(uint32_t s, uint32_t ns);

void T_time_to_seconds_and_nanoseconds(T_time time, uint32_t *s, uint32_t *ns);

You can convert a time value into a string represention. The time unit of the string representation is seconds. The precision of the string represention may be nanoseconds, microseconds, milliseconds, or seconds. You have to provide a buffer for the string (T_time_string).

const char *T_time_to_string_ns(T_time time, T_time_string buffer);

const char *T_time_to_string_us(T_time time, T_time_string buffer);

const char *T_time_to_string_ms(T_time time, T_time_string buffer);

const char *T_time_to_string_s(T_time time, T_time_string buffer);

Listing 8.19 Time String Example#

#include <t.h>

T_TEST_CASE(time_to_string)
{
    T_time_string ts;
    T_time t;
    uint32_t s;
    uint32_t ns;

    t = T_seconds_and_nanoseconds_to_time(0, 123456789);
    T_eq_str(T_time_to_string_ns(t, ts), "0.123456789");
    T_eq_str(T_time_to_string_us(t, ts), "0.123456");
    T_eq_str(T_time_to_string_ms(t, ts), "0.123");
    T_eq_str(T_time_to_string_s(t, ts), "0");

    T_time_to_seconds_and_nanoseconds(t, &s, &ns);
    T_eq_u32(s, 0);
    T_eq_u32(ns, 123456789);
}

Listing 8.20 Time String Report#

B:time_to_string
P:0:0:UI1:test-time.c:11
P:1:0:UI1:test-time.c:12
P:2:0:UI1:test-time.c:13
P:3:0:UI1:test-time.c:14
P:4:0:UI1:test-time.c:17
P:5:0:UI1:test-time.c:18
E:time_to_string:N:6:F:0:D:0.005250

You can convert a tick value into a string represention. The time unit of the string representation is seconds. The precision of the string represention may be nanoseconds, microseconds, milliseconds, or seconds. You have to provide a buffer for the string (T_time_string).

const char *T_ticks_to_string_ns(T_ticks ticks, T_time_string buffer);

const char *T_ticks_to_string_us(T_ticks ticks, T_time_string buffer);

const char *T_ticks_to_string_ms(T_ticks ticks, T_time_string buffer);

const char *T_ticks_to_string_s(T_ticks ticks, T_time_string buffer);

8.1.12. Code Runtime Measurements#

You can measure the runtime of code fragments in several execution environment variants with the T_measure_runtime() function. This function needs a context which must be created with the T_measure_runtime_create() function. The context is automatically destroyed after the test case execution.

typedef struct {
    size_t sample_count;
} T_measure_runtime_config;

typedef struct {
    const char *name;
    int flags;
    void (*setup)(void *arg);
    void (*body)(void *arg);
    bool (*teardown)(void *arg, T_ticks *delta, uint32_t tic, uint32_t toc,
        unsigned int retry);
    void *arg;
} T_measure_runtime_request;

T_measure_runtime_context *T_measure_runtime_create(
    const T_measure_runtime_config *config);

void T_measure_runtime(T_measure_runtime_context *ctx,
    const T_measure_runtime_request *request);

The runtime measurement is performed for the body request handler of the measurement request (T_measure_runtime_request). The optional setup request handler is called before each invocation of the body request handler. The optional teardown request handler is called after each invocation of the body request handler. It has several parameters and a return status. If it returns true, then this measurement sample value is recorded, otherwise the measurement is retried. The delta parameter is the current measurement sample value. It can be altered by the teardown request handler. The tic and toc parameters are the system tick values before and after the request body invocation. The retry parameter is the current retry counter. The runtime of the operational setup and teardown request handlers is not measured.

You can control some aspects of the measurement through the request flags (use zero for the default):

T_MEASURE_RUNTIME_ALLOW_CLOCK_ISR: Allow clock interrupts during the measurement. By default, measurements during which a clock interrupt happened are discarded unless it happens two times in a row.
T_MEASURE_RUNTIME_REPORT_SAMPLES: Report all measurement samples.
T_MEASURE_RUNTIME_DISABLE_FULL_CACHE: Disable the FullCache execution environment variant.
T_MEASURE_RUNTIME_DISABLE_HOT_CACHE: Disable the HotCache execution environment variant.
T_MEASURE_RUNTIME_DISABLE_DIRTY_CACHE: Disable the DirtyCache execution environment variant.
T_MEASURE_RUNTIME_DISABLE_MINOR_LOAD: Disable the Load execution environment variants with a load worker count less than the processor count.
T_MEASURE_RUNTIME_DISABLE_MAX_LOAD: Disable the Load execution environment variant with a load worker count equal to the processor count.

The execution environment variants (M:V) are:

FullCache

Before the body request handler is invoked a memory area with twice the size of the outer-most data cache is completely read. This fills the data cache with valid cache lines which are unrelated to the body request handler. The cache is full with valid data and loading memory used by the handler needs to evict cache lines.

You can disable this variant with the T_MEASURE_RUNTIME_DISABLE_FULL_CACHE request flag.

HotCache

Before the body request handler is invoked the body request handler is called without measuring the runtime. The aim is to load all data used by the body request handler to the cache.

You can disable this variant with the T_MEASURE_RUNTIME_DISABLE_HOT_CACHE request flag.

DirtyCache

Before the body request handler is invoked a memory area with twice the size of the outer-most data cache is completely written with new data. This should produce a data cache with dirty cache lines which are unrelated to the body request handler. In addition, the entire instruction cache is invalidated.

You can disable this variant with the T_MEASURE_RUNTIME_DISABLE_DIRTY_CACHE request flag.

Load/<WorkerCount>

This variant tries to get close to worst-case conditions. The cache is set up according to the DirtyCache variant. In addition, other processors try to fully load the memory system. The load is produced through writes to a memory area with twice the size of the outer-most data cache. The load variant is performed multiple times with a different set of active load worker threads. The <WorkerCount> value is the count of active workers which ranges from one to the processor count.

You can disable these variants with the T_MEASURE_RUNTIME_DISABLE_MINOR_LOAD and T_MEASURE_RUNTIME_DISABLE_MAX_LOAD request flags.

On SPARC, the body request handler is called with a register window setting so that window overflow traps will occur in the next level function call.

Each execution in an environment variant produces a sample set of body request handler runtime measurements. The minimum (M:MI), first quartile (M:Q1), median (M:Q2), third quartile (M:Q3), maximum (M:MX), median absolute deviation (M:MAD), and the sum of the sample values (M:D) is reported.

Listing 8.21 Code Runtime Measurement Example#

#include <t.h>

static void
empty(void *arg)
{
    (void)arg;
}

T_TEST_CASE(measure_empty)
{
    static const T_measure_runtime_config config = {
        .sample_count = 1024
    };
    T_measure_runtime_context *ctx;
    T_measure_runtime_request req;

    ctx = T_measure_runtime_create(&config);
    T_assert_not_null(ctx);

    memset(&req, 0, sizeof(req));
    req.name = "Empty";
    req.body = empty;
    T_measure_runtime(ctx, &req);
}

Listing 8.22 Code Runtime Measurement Report#

B:measure_empty
P:0:0:UI1:test-rtems-measure.c:18
M:B:Empty
M:V:FullCache
M:N:1024
M:MI:0.000000000
M:Q1:0.000000000
M:Q2:0.000000000
M:Q3:0.000000000
M:MX:0.000000009
M:MAD:0.000000000
M:D:0.000000485
M:E:Empty:D:0.208984183
M:B:Empty
M:V:HotCache
M:N:1024
M:MI:0.000000003
M:Q1:0.000000003
M:Q2:0.000000003
M:Q3:0.000000003
M:MX:0.000000006
M:MAD:0.000000000
M:D:0.000002626
M:E:Empty:D:0.000017046
M:B:Empty
M:V:DirtyCache
M:N:1024
M:MI:0.000000007
M:Q1:0.000000007
M:Q2:0.000000007
M:Q3:0.000000008
M:MX:0.000000559
M:MAD:0.000000000
M:D:0.000033244
M:E:Empty:D:1.887834875
M:B:Empty
M:V:Load/1
M:N:1024
M:MI:0.000000000
M:Q1:0.000000002
M:Q2:0.000000002
M:Q3:0.000000003
M:MX:0.000000288
M:MAD:0.000000000
M:D:0.000002421
M:E:Empty:D:0.001798809
[... 22 more load variants ...]
M:E:Empty:D:0.021252583
M:B:Empty
M:V:Load/24
M:N:1024
M:MI:0.000000001
M:Q1:0.000000002
M:Q2:0.000000002
M:Q3:0.000000003
M:MX:0.000001183
M:MAD:0.000000000
M:D:0.000003406
M:E:Empty:D:0.015188063
E:measure_empty:N:1:F:0:D:14.284869

8.1.13. Interrupt Tests#

In the operating system implementation you may have two kinds of critical sections. Firstly, there are low-level critical sections protected by interrupts disabled and maybe also some SMP spin lock. Secondly, there are high-level critical sections which are protected by disabled thread dispatching. The high-level critical sections may contain several low-level critical sections. Between these low-level critical sections interrupts may happen which could alter the code path taken in the high-level critical section.

The test framework provides support to write test cases for high-level critical sections though the T_interrupt_test() function:

typedef enum {
    T_INTERRUPT_TEST_INITIAL,
    T_INTERRUPT_TEST_ACTION,
    T_INTERRUPT_TEST_BLOCKED,
    T_INTERRUPT_TEST_CONTINUE,
    T_INTERRUPT_TEST_DONE,
    T_INTERRUPT_TEST_EARLY,
    T_INTERRUPT_TEST_INTERRUPT,
    T_INTERRUPT_TEST_LATE,
    T_INTERRUPT_TEST_TIMEOUT
} T_interrupt_test_state;

typedef struct {
    void                   (*prepare)(void *arg);
    void                   (*action)(void *arg);
    T_interrupt_test_state (*interrupt)(void *arg);
    void                   (*blocked)(void *arg);
    uint32_t                 max_iteration_count;
} T_interrupt_test_config;

T_interrupt_test_state T_interrupt_test(
    const T_interrupt_test_config *config,
    void                          *arg
);

This function returns T_INTERRUPT_TEST_DONE if the test condition was satisfied within the maximum iteration count, otherwise it returns T_INTERRUPT_TEST_TIMEOUT. The interrupt test run uses the specified configuration and passes the specified argument to all configured handlers. The function shall be called from thread context with interrupts enabled.

The interrupt test uses an adaptive bisection algorithm to try to hit the code section under test by an interrupt. In each test iteration, it waits for a time point one quarter of the clock tick interval after a clock tick using the monotonic clock. Then it performs a busy wait using T_busy() with a busy count controlled by the adaptive bisection algorithm. The test maintains a sample set of upper and lower bound busy wait count values. Initially, the lower bound values are zero and the upper bound values are set to a value returned by T_get_one_clock_tick_busy(). The busy wait count for an iteration is set to the middle point between the arithmetic mean of the lower and upper bound sample values. After the action handler returns, the set of lower and upper bound sample values is updated based on the test state. If the test state is T_INTERRUPT_TEST_EARLY, then the oldest upper bound sample value is replaced by the busy wait count used to delay the action and the latest lower bound sample value is slightly decreased. Reducing the lower bound helps to avoid a zero length interval between the upper and lower bounds. If the test state is T_INTERRUPT_TEST_LATE, then the oldest lower bound sample value is replaced by the busy wait count used to delay the action and the latest upper bound sample value is slightly increased. In all other test states the timing values remain as is. Using the arithmetic mean of a sample set dampens the effect of each test iteration and is an heuristic to mitigate the influence of jitters in the action code execution.

The optional prepare handler should prepare the system so that the action handler can be called. It is called in a tight loop, so all the time consuming setup should be done before T_interrupt_test() is called. During the preparation the test state is T_INTERRUPT_TEST_INITIAL. The preparation handler shall not change the test state.

The action handler should call the function which executes the code section under test. The execution path up to the code section under test should have a low jitter. Otherwise, the adaptive bisection algorithm may not find the right spot.

The interrupt handler should check if the test condition is satisfied or a new iteration is necessary. This handler is called in interrupt context. It shall return T_INTERRUPT_TEST_DONE if the test condition is satisfied and the test run is done. It shall return T_INTERRUPT_TEST_EARLY if the interrupt happened too early to satisfy the test condition. It shall return T_INTERRUPT_TEST_LATE if the interrupt happened too late to satisfy the test condition. It shall return T_INTERRUPT_TEST_CONTINUE if the test should continue with the current timing settings. Other states shall not be returned. It is critical to return the early and late states if the test condition was not satisfied, otherwise the adaptive bisection algorithm may not work. The returned state is used to try to change the test state from T_INTERRUPT_TEST_ACTION to the returned state.

The optional blocked handler is invoked if the executing thread blocks during the action processing. It should remove the blocking condition of the thread so that the next iteration can start. It can use T_interrupt_change_state() to change the interrupt test state.

The max iteration count configuration member defines the maximum iteration count of the test loop. If the maximum iteration count is reached before the test condition is satisfied, then T_interrupt_test() returns T_INTERRUPT_TEST_TIMEOUT.

The interrupt and blocked handlers may be called in arbitrary test states.

The action, interrupt, and blocked handlers can use T_interrupt_test_get_state() to get the current test state:

T_interrupt_test_state T_interrupt_test_get_state(void);

The action, interrupt, and blocked handlers can use T_interrupt_test_change_state() to try to change the test state from an expected state to a desired state:

T_interrupt_test_state T_interrupt_test_change_state(
    T_interrupt_test_state expected_state,
    T_interrupt_test_state desired_state
);

The function returns the previous state. If it differs from the expected state, then the requested state change to the desired state did not take place. In an SMP configuration, do not call this function in a tight loop. It could lock up the test run. To busy wait for a state change, use T_interrupt_test_get_state().

The action handler can use T_interrupt_test_busy_wait_for_interrupt() to busy wait for the interrupt:

void T_interrupt_test_busy_wait_for_interrupt(void);

This is useful if the action code does not block to wait for the interrupt. If the action handler just returns the test code immediately prepares the next iteration and may miss an interrupt which happens too late.

8.1.14. Test Runner#

You can call the T_main() function to run all registered test cases.

int T_main(const T_config *config);

The T_main() function returns 0 if all test cases passed, otherwise it returns 1. Concurrent execution of the T_main() function is undefined behaviour.

You can ask if you execute within the context of the test runner with the T_is_runner() function:

bool T_is_runner(void);

It returns true if you execute within the context of the test runner (the context which executes for example T_main()). Otherwise it returns false, for example if you execute in another task, in interrupt context, nobody executes T_main(), or during system initialization on another processor.

On RTEMS, you have to register the test cases with the T_register() function before you call T_main(). This makes it possible to run low level tests, for example without the operating system directly in boot_card() or during device driver initialization. On other platforms, the T_register() is a no operation.

void T_register(void);

You can run test cases also individually. Use T_run_initialize() to initialize the test runner. Call T_run_all() to run all or T_run_by_name() to run specific registered test cases. Call T_case_begin() to begin a freestanding test case and call T_case_end() to finish it. Finally, call T_run_finalize().

void T_run_initialize(const T_config *config);

void T_run_all(void);

void T_run_by_name(const char *name);

void T_case_begin(const char *name, const T_fixture *fixture);

void T_case_end(void);

bool T_run_finalize(void);

The T_run_finalize() function returns true if all test cases passed, otherwise it returns false. Concurrent execution of the runner functions (including T_main()) is undefined behaviour. The test suite configuration must be persistent throughout the test run.

typedef enum {
    T_EVENT_RUN_INITIALIZE,
    T_EVENT_CASE_EARLY,
    T_EVENT_CASE_BEGIN,
    T_EVENT_CASE_END,
    T_EVENT_CASE_LATE,
    T_EVENT_RUN_FINALIZE
} T_event;

typedef void (*T_action)(T_event, const char *);

typedef void (*T_putchar)(int, void *);

typedef struct {
    const char *name;
    char *buf;
    size_t buf_size;
    T_putchar putchar;
    void *putchar_arg;
    T_verbosity verbosity;
    T_time (*now)(void);
    size_t action_count;
    const T_action *actions;
} T_config;

With the test suite configuration you can specifiy the test suite name, the put character handler used the output the test report, the initial verbosity, the monotonic time provider and an optional set of test suite actions. Only the test runner calls the put character handler, other tasks or interrupt handlers write to a buffer which is emptied by the test runner on demand. You have to specifiy this buffer in the test configuration. The test suite actions are called with the test suite name for test suite run events (T_EVENT_RUN_INITIALIZE and T_EVENT_RUN_FINALIZE) and the test case name for the test case events (T_EVENT_CASE_EARLY, T_EVENT_CASE_BEGIN, T_EVENT_CASE_END and T_EVENT_CASE_LATE).

8.1.15. Test Verbosity#

Three test verbosity levels are defined:

T_QUIET: Only the test suite begin, system, test case end, and test suite end lines are printed.
T_NORMAL: Prints everything except passed test lines.
T_VERBOSE: Prints everything.

The test verbosity level can be set within the scope of one test case with the T_set_verbosity() function:

T_verbosity T_set_verbosity(T_verbosity new_verbosity);

The function returns the previous verbosity. After the test case, the configured verbosity is automatically restored.

An example with T_QUIET verbosity:

A:xyz
S:Platform:RTEMS
[...]
E:a:N:2:F:1
E:b:N:0:F:1
E:c:N:1:F:1
E:d:N:6:F:0
Z:xyz:C:4:N:9:F:3

The same example with T_NORMAL verbosity:

A:xyz
S:Platform:RTEMS
[...]
B:a
F:1:0:UI1:test-verbosity.c:6:test fails
E:a:N:2:F:1
B:b
F:*:0:UI1:test-verbosity.c:12:quiet test fails
E:b:N:0:F:1
B:c
F:0:0:UI1:test-verbosity.c:17:this is a format string
E:c:N:1:F:1
B:d
E:d:N:6:F:0
Z:xyz:C:4:N:9:F:3

The same example with T_VERBOSE verbosity:

A:xyz
S:Platform:RTEMS
[...]
B:a
P:0:0:UI1:test-verbosity.c:5
F:1:0:UI1:test-verbosity.c:6:test fails
E:a:N:2:F:1
B:b
F:*:0:UI1:test-verbosity.c:12:quiet test fails
E:b:N:0:F:1
B:c
F:0:0:UI1:test-verbosity.c:17:this is a format string
E:c:N:1:F:1
B:d
P:0:0:UI1:test-verbosity.c:22
P:1:0:UI1:test-verbosity.c:23
P:2:0:UI1:test-verbosity.c:24
P:3:0:UI1:test-verbosity.c:25
P:4:0:UI1:test-verbosity.c:26
P:5:0:UI1:test-verbosity.c:27
E:d:N:6:F:0
Z:xyz:C:4:N:9:F:3

8.1.16. Test Reporting#

The test reporting is line based which should be easy to parse with a simple state machine. Each line consists of a set of fields separated by colon characters (:). The first character of the line determines the line format:

A

A test suite begin line. It has the format:

A:<TestSuite>

A description of the field follows:

<TestSuite>: The test suite name. Must not contain colon characters (:).

S

A test suite system line. It has the format:

S:<Key>:<Value>

A description of the fields follows:

<Key>: A key string. Must not contain colon characters (:).
<Value>: An arbitrary key value string. May contain colon characters (:).

B

A test case begin line. It has the format:

B:<TestCase>

A description of the field follows:

<TestCase>: A test case name. Must not contain colon characters (:).

P

A test pass line. It has the format:

P:<Step>:<Processor>:<Task>:<File>:<Line>

A description of the fields follows:

<Step>: Each non-quiet test has a unique test step counter value in each test case execution. The test step counter is set to zero before the test case executes. For quiet test checks, there is no associated test step and the character * instead of an integer is used to indicate this.
<Processor>: The processor index of the processor which executed at least one instruction of the corresponding test.
<Task>: The name of the task which executed the corresponding test if the test executed in task context. The name ISR indicates that the test executed in interrupt context. The name ? indicates that the test executed in an arbitrary context with no valid executing task.
<File>: The name of the source file which contains the corresponding test. A source file of * indicates that no test source file is associated with the test, e.g. it was produced by the test framework itself.
<Line>: The line of the test statement in the source file which contains the corresponding test. A line number of * indicates that no test source file is associated with the test, e.g. it was produced by the test framework itself.

F

A test failure line. It has the format:

F:<Step>:<Processor>:<Task>:<File>:<Line>:<Message>

A description of the fields follows:

<Step> <Processor> <Task> <File> <Line>: See above P line.
<Message>: An arbitrary message string. May contain colon characters (:).

L

A log message line. It has the format:

L:<Message>

A description of the field follows:

<Message>: An arbitrary message string. May contain colon characters (:).

E

A test case end line. It has the format:

E:<TestCase>:N:<Steps>:F:<Failures>:D:<Duration>

A description of the fields follows:

<TestCase>: A test case name. Must not contain colon characters (:).
<Steps>: The final test step counter of a test case. Quiet test checks produce no test steps.
<Failures>: The count of failed test checks of a test case.
<Duration>: The test case duration in seconds.

Z

A test suite end line. It has the format:

Z:<TestSuite>:C:<TestCases>:N:<OverallSteps>:F:<OverallFailures>:D:<Duration>

A description of the fields follows:

<TestSuite>: The test suite name. Must not contain colon characters (:).
<TestCases>: The count of test cases in the test suite.
<OverallSteps>: The overall count of test steps in the test suite.
<OverallFailures>: The overall count of failed test cases in the test suite.
<Duration>: The test suite duration in seconds.

Y

Auxiliary information line. Issued after the test suite end. It has the format:

Y:ReportHash:SHA256:<Hash>

A description of the fields follows:

<Hash>: The SHA256 hash value of the test suite report from the begin to the end of the test suite.

M

A code runtime measurement line. It has the formats:

M:B:<Name>

M:V:<Variant>

M:N:<SampleCount>

M:S:<Count>:<Value>

M:MI:<Minimum>

M:Q1:<FirstQuartile>

M:Q2:<Median>

M:Q3:<ThirdQuartile>

M:MX:<Maximum>

M:MAD:<MedianAbsoluteDeviation>

M:D:<SumOfSampleValues>

M:E:<Name>:D:<Duration>

A description of the fields follows:

<Name>: A code runtime measurement name. Must not contain colon characters (:).
<Variant>: The execution variant which is one of FullCache, HotCache, DirtyCache, or Load/<WorkerCount>. The <WorkerCount> is the count of active workers which ranges from one to the processor count.
<SampleCount>: The sample count as defined by the runtime measurement configuration.
<Count>: The count of samples with the same value.
<Value>: A sample value in seconds.
<Minimum>: The minimum of the sample set in seconds.
<FirstQuartile>: The first quartile of the sample set in seconds.
<Median>: The median of the sample set in seconds.
<ThirdQuartile>: The third quartile of the sample set in seconds.
<Maximum>: The maximum of the sample set in seconds.
<MedianAbsoluteDeviation>: The median absolute deviation of the sample set in seconds.
<SumOfSampleValues>: The sum of all sample values of the sample set in seconds.
<Duration>: The runtime measurement duration in seconds. It includes time to set up the execution environment variant.

Listing 8.23 Example Test Report#

A:xyz
S:Platform:RTEMS
S:Compiler:7.4.0 20181206 (RTEMS 5, RSB e0aec65182449a4e22b820e773087636edaf5b32, Newlib 1d35a003f)
S:Version:5.0.0.820977c5af17c1ca2f79800d64bd87ce70a24c68
S:BSP:erc32
S:RTEMS_DEBUG:1
S:RTEMS_MULTIPROCESSING:0
S:RTEMS_POSIX_API:1
S:RTEMS_PROFILING:0
S:RTEMS_SMP:1
B:timer
P:0:0:UI1:test-rtems.c:26
P:1:0:UI1:test-rtems.c:29
P:2:0:UI1:test-rtems.c:33
P:3:0:ISR:test-rtems.c:14
P:4:0:ISR:test-rtems.c:15
P:5:0:UI1:test-rtems.c:38
P:6:0:UI1:test-rtems.c:39
P:7:0:UI1:test-rtems.c:42
E:timer:N:8:F:0:D:0.019373
B:rsc_success
P:0:0:UI1:test-rtems.c:59
F:1:0:UI1:test-rtems.c:60:RTEMS_INVALID_NUMBER == RTEMS_SUCCESSFUL
F:*:0:UI1:test-rtems.c:62:RTEMS_INVALID_NUMBER == RTEMS_SUCCESSFUL
P:2:0:UI1:test-rtems.c:63
F:3:0:UI1:test-rtems.c:64:RTEMS_INVALID_NUMBER == RTEMS_SUCCESSFUL
E:rsc_success:N:4:F:3:D:0.011128
B:rsc
P:0:0:UI1:test-rtems.c:48
F:1:0:UI1:test-rtems.c:49:RTEMS_INVALID_NUMBER == RTEMS_INVALID_ID
F:*:0:UI1:test-rtems.c:51:RTEMS_INVALID_NUMBER == RTEMS_INVALID_ID
P:2:0:UI1:test-rtems.c:52
F:3:0:UI1:test-rtems.c:53:RTEMS_INVALID_NUMBER == RTEMS_INVALID_ID
E:rsc:N:4:F:3:D:0.011083
Z:xyz:C:3:N:16:F:6:D:0.047201
Y:ReportHash:SHA256:e5857c520dd9c9b7c15d4a76d78c21ccc46619c30a869ecd11bbcd1885155e0b

8.1.17. Test Report Validation#

You can add the T_report_hash_sha256() test suite action to the test suite configuration to generate and report the SHA256 hash value of the test suite report. The hash value covers everything reported by the test suite run from the begin to the end. This can be used to check that the report generated on the target is identical to the report received on the report consumer side. The hash value is reported after the end of test suite line (Z) as auxiliary information in a Y line. Consumers may have to reverse a \n to \r\n conversion before the hash is calculated. Such a conversion could be performed by a particular put character handler provided by the test suite configuration.

8.1.18. Supported Platforms#

The framework runs on FreeBSD, MSYS2, Linux and RTEMS.

8.2. Test Framework Requirements for RTEMS#

The requirements on a test framework suitable for RTEMS are:

8.2.1. License Requirements#

TF.License.Permissive: The test framework shall have a permissive open source license such as BSD-2-Clause.

8.2.2. Portability Requirements#

TF.Portability

The test framework shall be portable.

TF.Portability.RTEMS

The test framework shall run on RTEMS.

TF.Portability.POSIX

The test framework shall be portable to POSIX compatible operating systems. This allows to run test cases of standard C/POSIX/etc. APIs on multiple platforms.

TF.Portability.POSIX.Linux: The test framework shall run on Linux.
TF.Portability.POSIX.FreeBSD: The test framework shall run on FreeBSD.

TF.Portability.C11

The test framework shall be written in C11.

TF.Portability.Static

Test framework shall not use dynamic memory for basic services.

TF.Portability.Small

The test framework shall be small enough to support low-end platforms (e.g. 64KiB of RAM/ROM should be sufficient to test the architecture port, e.g. no complex stuff such as file systems, etc.).

TF.Portability.Small.LinkTimeConfiguration

The test framework shall be configured at link-time.

TF.Portability.Small.Modular

The test framework shall be modular so that only necessary parts end up in the final executable.

TF.Portability.Small.Memory

The test framework shall not aggregate data during test case executions.

8.2.3. Reporting Requirements#

TF.Reporting

Test results shall be reported.

TF.Reporting.Verbosity

The test report verbosity shall be configurable. This allows different test run scenarios, e.g. regression test runs, full test runs with test report verification against the planned test output.

TF.Reporting.Verification

It shall be possible to use regular expressions to verify test reports line by line.

TF.Reporting.Compact

Test output shall be compact to avoid long test runs on platforms with a slow output device, e.g. 9600 Baud UART.

TF.Reporting.PutChar

A simple output one character function provided by the platform shall be sufficient to report the test results.

TF.Reporting.NonBlocking

The ouptut functions shall be non-blocking.

TF.Reporting.Printf

The test framework shall provide printf()-like output functions.

TF.Reporting.Printf.WithFP: There shall be a printf()-like output function with floating point support.
TF.Reporting.Printf.WithoutFP: There shall be a printf()-like output function without floating point support on RTEMS.

TF.Reporting.Platform

The test platform shall be reported.

TF.Reporting.Platform.RTEMS.Git: The RTEMS source Git commit shall be reported.
TF.Reporting.Platform.RTEMS.Arch: The RTEMS architecture name shall be reported.
TF.Reporting.Platform.RTEMS.BSP: The RTEMS BSP name shall be reported.
TF.Reporting.Platform.RTEMS.Tools: The RTEMS tool chain version shall be reported.
TF.Reporting.Platform.RTEMS.Config.Debug: The shall be reported if RTEMS_DEBUG is defined.
TF.Reporting.Platform.RTEMS.Config.Multiprocessing: The shall be reported if RTEMS_MULTIPROCESSING is defined.
TF.Reporting.Platform.RTEMS.Config.POSIX: The shall be reported if RTEMS_POSIX_API is defined.
TF.Reporting.Platform.RTEMS.Config.Profiling: The shall be reported if RTEMS_PROFILING is defined.
TF.Reporting.Platform.RTEMS.Config.SMP: The shall be reported if RTEMS_SMP is defined.

TF.Reporting.TestCase

The test cases shall be reported.

TF.Reporting.TestCase.Begin: The test case begin shall be reported.
TF.Reporting.TestCase.End: The test case end shall be reported.
TF.Reporting.TestCase.Tests: The count of test checks of the test case shall be reported.
TF.Reporting.TestCase.Failures: The count of failed test checks of the test case shall be reported.
TF.Reporting.TestCase.Timing: Test case timing shall be reported.
TF.Reporting.TestCase.Tracing: Automatic tracing and reporting of thread context switches and interrupt service routines shall be optionally performed.

8.2.4. Environment Requirements#

TF.Environment

The test framework shall support all environment conditions of the platform.

TF.Environment.SystemStart: The test framework shall run during early stages of the system start, e.g. valid stack pointer, initialized data and cleared BSS, nothing more.
TF.Environment.BeforeDeviceDrivers: The test framework shall run before device drivers are initialized.
TF.Environment.InterruptContext: The test framework shall support test case code in interrupt context.

8.2.5. Usability Requirements#

TF.Usability

The test framework shall be easy to use.

TF.Usability.TestCase

It shall be possible to write test cases.

TF.Usability.TestCase.Independence

It shall be possible to write test cases in modules independent of the test runner.

TF.Usability.TestCase.AutomaticRegistration

Test cases shall be registered automatically, e.g. via constructors or linker sets.

TF.Usability.TestCase.Order

It shall be possible to sort the registered test cases (e.g. random, by name) before they are executed.

TF.Usability.TestCase.Resources

It shall be possible to use resources with a life time restricted to the test case.

TF.Usability.TestCase.Resources.Memory: It shall be possible to dynamically allocate memory which is automatically freed once the test case completed.
TF.Usability.TestCase.Resources.File: It shall be possible to create a file which is automatically unlinked once the test case completed.
TF.Usability.TestCase.Resources.Directory: It shall be possible to create a directory which is automatically removed once the test case completed.
TF.Usability.TestCase.Resources.FileDescriptor: It shall be possible to open a file descriptor which is automatically closed once the test case completed.

TF.Usability.TestCase.Fixture

It shall be possible to use a text fixture for test cases.

TF.Usability.TestCase.Fixture.SetUp: It shall be possible to provide a set up handler for each test case.
TF.Usability.TestCase.Fixture.TearDown: It shall be possible to provide a tear down handler for each test case.

TF.Usability.TestCase.Context

The test case context shall be verified a certain points.

TF.Usability.TestCase.Context.VerifyAtEnd

After a test case exection it shall be verified that the context is equal to the context at the test case begin. This helps to ensure that test cases are independent of each other.

TF.Usability.TestCase.Context.VerifyThread

The test framework shall provide a function to ensure that the test case code executes in normal thread context. This helps to ensure that operating system service calls return to a sane context.

TF.Usability.TestCase.Context.Configurable

The context verified in test case shall be configurable at link-time.

TF.Usability.TestCase.Context.ThreadDispatchDisableLevel

It shall be possible to verify the thread dispatch disable level.

TF.Usability.TestCase.Context.ISRNestLevel

It shall be possible to verify the ISR nest level.

TF.Usability.TestCase.Context.InterruptLevel

It shall be possible to verify the interrupt level (interrupts enabled/disabled).

TF.Usability.TestCase.Context.Workspace

It shall be possible to verify the workspace.

TF.Usability.TestCase.Context.Heap

It shall be possible to verify the heap.

TF.Usability.TestCase.Context.OpenFileDescriptors

It shall be possible to verify the open file descriptors.

TF.Usability.TestCase.Context.Classic

It shall be possible to verify Classic API objects.

TF.Usability.TestCase.Context.Classic.Barrier: It shall be possible to verify Classic API Barrier objects.
TF.Usability.TestCase.Context.Classic.Extensions: It shall be possible to verify Classic API User Extensions objects.
TF.Usability.TestCase.Context.Classic.MessageQueues: It shall be possible to verify Classic API Message Queue objects.
TF.Usability.TestCase.Context.Classic.Partitions: It shall be possible to verify Classic API Partition objects.
TF.Usability.TestCase.Context.Classic.Periods: It shall be possible to verify Classic API Rate Monotonic Period objects.
TF.Usability.TestCase.Context.Classic.Regions: It shall be possible to verify Classic API Region objects.
TF.Usability.TestCase.Context.Classic.Semaphores: It shall be possible to verify Classic API Semaphore objects.
TF.Usability.TestCase.Context.Classic.Tasks: It shall be possible to verify Classic API Task objects.
TF.Usability.TestCase.Context.Classic.Timers: It shall be possible to verify Classic API Timer objects.

TF.Usability.TestCase.Context.POSIX

It shall be possible to verify POSIX API objects.

TF.Usability.TestCase.Context.POSIX.Keys: It shall be possible to verify POSIX API Key objects.
TF.Usability.TestCase.Context.POSIX.KeyValuePairs: It shall be possible to verify POSIX API Key Value Pair objects.
TF.Usability.TestCase.Context.POSIX.MessageQueues: It shall be possible to verify POSIX API Message Queue objects.
TF.Usability.TestCase.Context.POSIX.Semaphores: It shall be possible to verify POSIX API Named Semaphores objects.
TF.Usability.TestCase.Context.POSIX.Shms: It shall be possible to verify POSIX API Shared Memory objects.
TF.Usability.TestCase.Context.POSIX.Threads: It shall be possible to verify POSIX API Thread objects.
TF.Usability.TestCase.Context.POSIX.Timers: It shall be possible to verify POSIX API Timer objects.

TF.Usability.Assert

There shall be functions to assert test objectives.

TF.Usability.Assert.Safe: Test assert functions shall be safe to use, e.g. assert(a == b) vs. assert(a = b) vs. assert_eq(a, b).
TF.Usability.Assert.Continue: There shall be assert functions which allow the test case to continue in case of an assertion failure.
TF.Usability.Assert.Abort: There shall be assert functions which abourt the test case in case of an assertion failure.

TF.Usability.EasyToWrite

It shall be easy to write test code, e.g. avoid long namespace prefix rtems_test_*.

TF.Usability.Threads

The test framework shall support multi-threading.

TF.Usability.Pattern

The test framework shall support test patterns.

TF.Usability.Pattern.Interrupts: The test framework shall support test cases which use interrupts, e.g. spintrcritical*.
TF.Usability.Pattern.Parallel: The test framework shall support test cases which want to run code in parallel on SMP machines.
TF.Usability.Pattern.Timing: The test framework shall support test cases which want to measure the timing of code sections under various platform conditions, e.g. dirty cache, empty cache, hot cache, with load from other processors, etc..

TF.Usability.Configuration

The test framework shall be configurable.

TF.Usability.Configuration.Time: The timestamp function shall be configurable, e.g. to allow test runs without a clock driver.

8.2.6. Performance Requirements#

TF.Performance.RTEMS.No64BitDivision: The test framework shall not use 64-bit divisions on RTEMS.

8.3. Off-the-shelf Test Frameworks#

There are several off-the-shelf test frameworks for C/C++. The first obstacle for test frameworks is the license requirement (TF.License.Permissive).

8.3.1. bdd-for-c#

In the bdd-for-c framework the complete test suite must be contained in one file and the main function is generated. This violates TF.Usability.TestCase.Independence.

8.3.2. CBDD#

The CBDD framework uses the C blocks extension from clang. This violates TF.Portability.C11.

8.3.3. Google Test#

Google Test 1.8.1 was supported by RTEMS. Unfortunately, it is written in C++ and is too heavy weight for low-end platforms. Otherwise it is a nice framework. We have archived it in case someone wants to try to bring it back.

8.3.4. Unity#

The Unity Test API does not meet our requirements. There was a discussion on the mailing list in 2013.

8.4. Standard Test Report Formats#

8.4.1. JUnit XML#

A common test report format is JUnit XML.

<?xml version="1.0" encoding="UTF-8" ?>
<testsuites id="xyz" name="abc" tests="225" failures="1262" time="0.001">
  <testsuite id="def" name="ghi" tests="45" failures="17" time="0.001">
    <testcase id="jkl" name="mno" time="0.001">
      <failure message="pqr" type="stu"></failure>
      <system-out>stdout</system-out>
      <system-err>stderr</system-err>
    </testcase>
  </testsuite>
</testsuites>

The major problem with this format is that you have to output the failure count of all test suites and the individual test suite before the test case output. You know the failure count only after a complete test run. This runs contrary to requirement TF.Portability.Small.Memory. It is also a bit verbose (TF.Reporting.Compact).

It is easy to convert a full test report generated by The RTEMS Test Framework to the JUnit XML format.

8.4.2. Test Anything Protocol#

The Test Anything Protocol (TAP) is easy to consume and produce.

1..4
ok 1 - Input file opened
not ok 2 - First line of the input valid
ok 3 - Read the rest of the file
not ok 4 - Summarized correctly # TODO Not written yet

You have to know in advance how many test statements you want to execute in a test case. The problem with this format is that there is no standard way to provide auxiliary data such as test timing or a tracing report.

It is easy to convert a full test report generated by The RTEMS Test Framework to the TAP format.

Software Test Framework

Contents

8. Software Test Framework#

8.1. The RTEMS Test Framework#

8.1.1. Nomenclature#

8.1.2. Test Cases#

8.1.3. Test Fixture#

8.1.4. Test Case Planning#

8.1.5. Test Case Resource Accounting#

8.1.6. Test Case Scoped Dynamic Memory#

8.1.7. Test Case Destructors#

8.1.8. Test Checks#

8.1.8.1. Test Check Variant Conventions#

8.1.8.2. Test Check Parameter Conventions#

8.1.8.3. Test Check Condition Conventions#

8.1.8.4. Test Check Type Conventions#

8.1.8.5. Integers#

8.1.8.6. Boolean Expressions#

8.1.8.7. Generic Types#

8.1.8.8. Pointers#

8.1.8.9. Memory Areas#

8.1.8.10. Strings#

8.1.8.11. Characters#

8.1.8.12. RTEMS Status Codes#

8.1.8.13. POSIX Error Numbers#

8.1.8.14. POSIX Status Codes#

8.1.9. Log Messages and Formatted Output#

8.1.10. Utility#

8.1.11. Time Services#

8.1.12. Code Runtime Measurements#

8.1.13. Interrupt Tests#

8.1.14. Test Runner#

8.1.15. Test Verbosity#

8.1.16. Test Reporting#

8.1.17. Test Report Validation#

8.1.18. Supported Platforms#

8.2. Test Framework Requirements for RTEMS#

8.2.1. License Requirements#

8.2.2. Portability Requirements#

8.2.3. Reporting Requirements#

8.2.4. Environment Requirements#

8.2.5. Usability Requirements#

8.2.6. Performance Requirements#

8.3. Off-the-shelf Test Frameworks#

8.3.1. bdd-for-c#

8.3.2. CBDD#

8.3.3. Google Test#

8.3.4. Unity#

8.4. Standard Test Report Formats#

8.4.1. JUnit XML#

8.4.2. Test Anything Protocol#