Testing Guide¶

This comprehensive guide covers testing strategies, frameworks, and best practices for Nexus embedded applications.

Overview ¶

Testing is critical for embedded systems reliability. Nexus provides comprehensive testing infrastructure including unit tests, integration tests, and property-based tests.

Testing Pyramid:

┌─────────────────┐
│  System Tests   │  ← Few, slow, expensive
├─────────────────┤
│ Integration     │  ← Some, medium speed
│     Tests       │
├─────────────────┤
│   Unit Tests    │  ← Many, fast, cheap
└─────────────────┘

Key Concepts:

Unit testing with Google Test
Integration testing
Property-based testing with Hypothesis
Hardware-in-the-loop (HIL) testing
Code coverage analysis
Continuous integration

Test Infrastructure ¶

Test Framework ¶

Nexus uses Google Test (gtest) for C++ tests and provides C wrappers for testing C code.

Features:

Rich assertion macros
Test fixtures for setup/teardown
Parameterized tests
Death tests
Test discovery
XML/JSON output

Directory Structure:

tests/
├── hal/                    # HAL unit tests
│   ├── test_gpio.cpp
│   ├── test_uart.cpp
│   └── ...
├── osal/                   # OSAL unit tests
│   ├── test_task.cpp
│   ├── test_mutex.cpp
│   └── ...
├── framework/              # Framework tests
│   ├── log/
│   ├── shell/
│   └── config/
├── integration/            # Integration tests
│   ├── test_hal_osal.cpp
│   └── ...
└── property/               # Property-based tests
    ├── test_gpio_properties.cpp
    └── ...

Building Tests ¶

CMake Configuration:

# Enable tests (native platform only)
cmake -B build \
    -DNEXUS_PLATFORM=native \
    -DNEXUS_BUILD_TESTS=ON \
    -DCMAKE_BUILD_TYPE=Debug

# Build tests
cmake --build build

# Build specific test
cmake --build build --target test_gpio

Test Targets:

test_hal - All HAL tests
test_osal - All OSAL tests
test_log - Log framework tests
test_shell - Shell framework tests
test_config - Config framework tests
test_integration - Integration tests

Running Tests ¶

Run All Tests:

# Using Python script (recommended)
python scripts/test/test.py

# Using CTest
cd build
ctest -C Debug --output-on-failure

Run Specific Tests:

# Run specific test suite
python scripts/test/test.py -f "GPIO*"

# Run specific test case
./build/tests/hal/test_gpio --gtest_filter="GPIOTest.WriteRead"

# Run with verbose output
python scripts/test/test.py -v

Test Options:

# List all tests
./build/tests/hal/test_gpio --gtest_list_tests

# Run tests matching pattern
./build/tests/hal/test_gpio --gtest_filter="GPIO*"

# Repeat tests
./build/tests/hal/test_gpio --gtest_repeat=10

# Shuffle test order
./build/tests/hal/test_gpio --gtest_shuffle

# Generate XML output
./build/tests/hal/test_gpio --gtest_output=xml:results.xml

Unit Testing ¶

Writing Unit Tests ¶

Basic Test Structure:

#include <gtest/gtest.h>
extern "C" {
#include "hal/nx_gpio.h"
#include "hal/nx_factory.h"
}

// Test fixture
class GPIOTest : public ::testing::Test {
protected:
    void SetUp() override {
        // Setup before each test
        nx_hal_init();
    }

    void TearDown() override {
        // Cleanup after each test
        nx_hal_deinit();
    }
};

// Test case
TEST_F(GPIOTest, WriteRead) {
    // Arrange
    nx_gpio_write_t* gpio = nx_factory_gpio_write('A', 5);
    ASSERT_NE(gpio, nullptr);

    // Act
    gpio->write(gpio, 1);
    uint8_t value = gpio->read(gpio);

    // Assert
    EXPECT_EQ(value, 1);

    // Cleanup
    nx_factory_gpio_release((nx_gpio_t*)gpio);
}

Assertion Macros:

// Fatal assertions (stop test on failure)
ASSERT_TRUE(condition);
ASSERT_FALSE(condition);
ASSERT_EQ(expected, actual);
ASSERT_NE(val1, val2);
ASSERT_LT(val1, val2);
ASSERT_LE(val1, val2);
ASSERT_GT(val1, val2);
ASSERT_GE(val1, val2);
ASSERT_STREQ(str1, str2);
ASSERT_STRNE(str1, str2);
ASSERT_NEAR(val1, val2, abs_error);

// Non-fatal assertions (continue test on failure)
EXPECT_TRUE(condition);
EXPECT_FALSE(condition);
EXPECT_EQ(expected, actual);
// ... (same as ASSERT_* variants)

Testing C Functions:

extern "C" {
#include "my_module.h"
}

TEST(MyModuleTest, FunctionTest) {
    int result = my_c_function(42);
    EXPECT_EQ(result, 84);
}

Test Fixtures ¶

Shared Setup/Teardown:

class UARTTest : public ::testing::Test {
protected:
    nx_uart_t* uart;

    void SetUp() override {
        nx_hal_init();
        uart = nx_factory_uart(0);
        ASSERT_NE(uart, nullptr);
    }

    void TearDown() override {
        if (uart) {
            nx_factory_uart_release(uart);
        }
        nx_hal_deinit();
    }
};

TEST_F(UARTTest, SendReceive) {
    // uart is available here
    nx_tx_sync_t* tx = uart->get_tx_sync(uart);
    ASSERT_NE(tx, nullptr);
    // ... test code
}

TEST_F(UARTTest, Baudrate) {
    // uart is available here too
    // ... test code
}

Parameterized Tests:

class GPIOPinTest : public ::testing::TestWithParam<uint8_t> {
protected:
    void SetUp() override {
        nx_hal_init();
    }

    void TearDown() override {
        nx_hal_deinit();
    }
};

TEST_P(GPIOPinTest, AllPins) {
    uint8_t pin = GetParam();
    nx_gpio_write_t* gpio = nx_factory_gpio_write('A', pin);
    ASSERT_NE(gpio, nullptr);

    gpio->write(gpio, 1);
    EXPECT_EQ(gpio->read(gpio), 1);

    nx_factory_gpio_release((nx_gpio_t*)gpio);
}

// Test pins 0-15
INSTANTIATE_TEST_SUITE_P(
    AllPins,
    GPIOPinTest,
    ::testing::Range<uint8_t>(0, 16)
);

Mocking ¶

Manual Mocks:

// Mock HAL for testing upper layers
class MockGPIO : public nx_gpio_write_t {
public:
    uint8_t last_written_value = 0;
    uint8_t read_value = 0;

    static void mock_write(nx_gpio_write_t* self, uint8_t value) {
        MockGPIO* mock = static_cast<MockGPIO*>(self);
        mock->last_written_value = value;
    }

    static uint8_t mock_read(nx_gpio_write_t* self) {
        MockGPIO* mock = static_cast<MockGPIO*>(self);
        return mock->read_value;
    }

    MockGPIO() {
        this->write = mock_write;
        this->read = mock_read;
    }
};

TEST(ApplicationTest, UsesGPIO) {
    MockGPIO mock_gpio;
    mock_gpio.read_value = 1;

    // Test application code with mock
    application_function(&mock_gpio);

    EXPECT_EQ(mock_gpio.last_written_value, 1);
}

Google Mock (gmock):

#include <gmock/gmock.h>

class MockUART {
public:
    MOCK_METHOD(int, send, (const uint8_t* data, size_t len), ());
    MOCK_METHOD(int, receive, (uint8_t* data, size_t len), ());
};

TEST(ProtocolTest, SendsCorrectData) {
    MockUART mock_uart;

    // Expect send to be called with specific data
    EXPECT_CALL(mock_uart, send(_, 10))
        .Times(1)
        .WillOnce(::testing::Return(10));

    // Test code that uses mock_uart
    protocol_send(&mock_uart, data, 10);
}

Integration Testing ¶

HAL + OSAL Integration ¶

Test Multiple Layers:

TEST(IntegrationTest, GPIOWithTask) {
    // Initialize both layers
    nx_hal_init();
    osal_init();

    // Create GPIO
    nx_gpio_write_t* led = nx_factory_gpio_write('A', 5);
    ASSERT_NE(led, nullptr);

    // Create task that uses GPIO
    auto task_func = [](void* arg) {
        nx_gpio_write_t* gpio = static_cast<nx_gpio_write_t*>(arg);
        for (int i = 0; i < 10; i++) {
            gpio->toggle(gpio);
            osal_task_delay(100);
        }
    };

    osal_task_handle_t task;
    osal_task_create(task_func, "test", 512, led, 1, &task);

    // Wait for task to complete
    osal_task_delay(1500);

    // Cleanup
    osal_task_delete(task);
    nx_factory_gpio_release((nx_gpio_t*)led);
    osal_deinit();
    nx_hal_deinit();
}

Framework Integration ¶

Test Framework Components:

TEST(IntegrationTest, LogWithUART) {
    // Initialize HAL and Log framework
    nx_hal_init();

    nx_uart_t* uart = nx_factory_uart(0);
    ASSERT_NE(uart, nullptr);

    log_init(nullptr);
    log_backend_t* backend = log_backend_uart_create(uart);
    log_backend_register(backend);

    // Test logging
    LOG_INFO("Test message");

    // Verify message was sent (check UART mock)
    // ...

    // Cleanup
    log_backend_unregister("uart");
    log_backend_uart_destroy(backend);
    log_deinit();
    nx_factory_uart_release(uart);
    nx_hal_deinit();
}

Property-Based Testing ¶

Overview ¶

Property-based testing generates random inputs to verify properties that should always hold true.

Benefits:

Finds edge cases
Tests many scenarios automatically
Documents expected behavior
Complements example-based tests

Using Hypothesis ¶

Python Property Tests:

from hypothesis import given, strategies as st
import ctypes

# Load native library
lib = ctypes.CDLL('./build/libhal.so')

@given(st.integers(min_value=0, max_value=255))
def test_gpio_write_read(value):
    """Property: Written value should be readable"""
    gpio = lib.nx_factory_gpio_write(ord('A'), 5)
    assert gpio is not None

    lib.nx_gpio_write(gpio, value)
    read_value = lib.nx_gpio_read(gpio)

    assert read_value == value

    lib.nx_factory_gpio_release(gpio)

C++ Property Tests:

#include <gtest/gtest.h>
#include <random>

TEST(GPIOPropertyTest, WriteReadProperty) {
    std::random_device rd;
    std::mt19937 gen(rd());
    std::uniform_int_distribution<> dis(0, 1);

    nx_hal_init();
    nx_gpio_write_t* gpio = nx_factory_gpio_write('A', 5);
    ASSERT_NE(gpio, nullptr);

    // Test 1000 random values
    for (int i = 0; i < 1000; i++) {
        uint8_t value = dis(gen);
        gpio->write(gpio, value);
        uint8_t read = gpio->read(gpio);
        EXPECT_EQ(read, value) << "Failed on iteration " << i;
    }

    nx_factory_gpio_release((nx_gpio_t*)gpio);
    nx_hal_deinit();
}

Common Properties ¶

Idempotence:

TEST(PropertyTest, SetModeIdempotent) {
    // Setting mode twice should have same effect as once
    nx_gpio_t* gpio = nx_factory_gpio('A', 5);

    gpio->set_mode(gpio, NX_GPIO_MODE_OUTPUT_PP);
    gpio->set_mode(gpio, NX_GPIO_MODE_OUTPUT_PP);

    // Verify mode is set correctly
    // ...
}

Commutativity:

TEST(PropertyTest, ConfigOrderIndependent) {
    // Order of configuration shouldn't matter
    nx_gpio_t* gpio1 = nx_factory_gpio('A', 5);
    gpio1->set_mode(gpio1, NX_GPIO_MODE_OUTPUT_PP);
    gpio1->set_pull(gpio1, NX_GPIO_PULL_UP);

    nx_gpio_t* gpio2 = nx_factory_gpio('A', 6);
    gpio2->set_pull(gpio2, NX_GPIO_PULL_UP);
    gpio2->set_mode(gpio2, NX_GPIO_MODE_OUTPUT_PP);

    // Both should behave identically
    // ...
}

Inverse Operations:

TEST(PropertyTest, ToggleTwiceReturnsOriginal) {
    nx_gpio_write_t* gpio = nx_factory_gpio_write('A', 5);

    gpio->write(gpio, 0);
    gpio->toggle(gpio);
    gpio->toggle(gpio);

    EXPECT_EQ(gpio->read(gpio), 0);
}

Code Coverage ¶

Measuring Coverage ¶

Enable Coverage:

# Configure with coverage
cmake -B build \
    -DNEXUS_PLATFORM=native \
    -DNEXUS_BUILD_TESTS=ON \
    -DNEXUS_ENABLE_COVERAGE=ON \
    -DCMAKE_BUILD_TYPE=Debug

# Build and run tests
cmake --build build
cd build
ctest

Generate Coverage Report:

# Linux/macOS
cd scripts/coverage
./run_coverage_linux.sh

# Windows
cd scripts\coverage
.\run_coverage_windows.ps1

View Report:

# Linux
xdg-open ../../coverage_html/index.html

# macOS
open ../../coverage_html/index.html

# Windows
start ..\..\coverage_report\html\index.html

Coverage Targets ¶

Nexus Coverage Goals:

HAL: 100% line coverage for native platform
OSAL: 100% line coverage for all backends
Framework: >95% line coverage
Integration: >90% line coverage

Coverage Metrics:

Line Coverage: Percentage of lines executed
Function Coverage: Percentage of functions called
Branch Coverage: Percentage of branches taken

Example Report:

File                    Lines    Exec    Cover
------------------------------------------------
hal/src/nx_gpio.c        245     245    100.0%
hal/src/nx_uart.c        312     308     98.7%
hal/src/nx_spi.c         198     195     98.5%
osal/src/task.c          156     156    100.0%
framework/log/log.c      423     418     98.8%
------------------------------------------------
TOTAL                   1334    1322     99.1%

Improving Coverage ¶

Identify Uncovered Code:

# Find uncovered lines
lcov --list coverage.info | grep "0.0%"

Add Missing Tests:

// Cover error paths
TEST(GPIOTest, InvalidPin) {
    nx_gpio_t* gpio = nx_factory_gpio('A', 99);  // Invalid pin
    EXPECT_EQ(gpio, nullptr);
}

// Cover edge cases
TEST(GPIOTest, BoundaryValues) {
    // Test minimum pin
    nx_gpio_t* gpio0 = nx_factory_gpio('A', 0);
    EXPECT_NE(gpio0, nullptr);

    // Test maximum pin
    nx_gpio_t* gpio15 = nx_factory_gpio('A', 15);
    EXPECT_NE(gpio15, nullptr);
}

Hardware-in-the-Loop Testing ¶

HIL Test Setup ¶

Test Rig Components:

Target hardware (STM32F4 Discovery)
Debug probe (ST-Link, J-Link)
Test fixtures (buttons, LEDs, sensors)
Host computer running tests

Test Architecture:

┌──────────────┐
│ Host PC      │
│ (Test Runner)│
└──────┬───────┘
       │ USB
┌──────┴───────┐
│ Debug Probe  │
│ (ST-Link)    │
└──────┬───────┘
       │ SWD
┌──────┴───────┐
│ Target MCU   │
│ (STM32F4)    │
└──────────────┘

Automated HIL Tests ¶

Python Test Script:

import serial
import time
import subprocess

class HILTest:
    def __init__(self, serial_port, elf_file):
        self.serial = serial.Serial(serial_port, 115200, timeout=1)
        self.elf_file = elf_file

    def flash_firmware(self):
        """Flash firmware to target"""
        cmd = [
            'openocd',
            '-f', 'interface/stlink.cfg',
            '-f', 'target/stm32f4x.cfg',
            '-c', f'program {self.elf_file} verify reset exit'
        ]
        subprocess.run(cmd, check=True)
        time.sleep(1)  # Wait for reset

    def send_command(self, cmd):
        """Send command via UART"""
        self.serial.write(f"{cmd}\r\n".encode())
        time.sleep(0.1)

    def read_response(self):
        """Read response from UART"""
        return self.serial.read(self.serial.in_waiting).decode()

    def test_gpio_toggle(self):
        """Test GPIO toggle command"""
        self.send_command("led green on")
        response = self.read_response()
        assert "LED green ON" in response

        self.send_command("led green off")
        response = self.read_response()
        assert "LED green OFF" in response

# Run tests
if __name__ == '__main__':
    test = HILTest('/dev/ttyUSB0', 'build/app.elf')
    test.flash_firmware()
    test.test_gpio_toggle()
    print("HIL tests passed!")

Continuous Integration ¶

GitHub Actions ¶

Workflow Configuration:

name: Tests

on: [push, pull_request]

jobs:
  test:
    runs-on: ubuntu-latest

    steps:
    - uses: actions/checkout@v3
      with:
        submodules: recursive

    - name: Install Dependencies
      run: |
        sudo apt-get update
        sudo apt-get install -y cmake ninja-build lcov

    - name: Configure
      run: |
        cmake -B build -G Ninja \
          -DNEXUS_PLATFORM=native \
          -DNEXUS_BUILD_TESTS=ON \
          -DNEXUS_ENABLE_COVERAGE=ON \
          -DCMAKE_BUILD_TYPE=Debug

    - name: Build
      run: cmake --build build

    - name: Test
      run: |
        cd build
        ctest --output-on-failure

    - name: Coverage
      run: |
        cd build
        lcov --capture --directory . --output-file coverage.info
        lcov --remove coverage.info '/usr/*' --output-file coverage.info
        lcov --list coverage.info

    - name: Upload Coverage
      uses: codecov/codecov-action@v3
      with:
        files: ./build/coverage.info

Test Organization ¶

Test Naming ¶

Conventions:

Test file: test_<module>.cpp
Test suite: <Module>Test
Test case: <Action><Expected>

Examples:

// File: test_gpio.cpp
class GPIOTest : public ::testing::Test { };

TEST_F(GPIOTest, WriteHighReadsHigh) { }
TEST_F(GPIOTest, WriteLowReadsLow) { }
TEST_F(GPIOTest, ToggleChangesState) { }
TEST_F(GPIOTest, InvalidPinReturnsNull) { }

Test Documentation ¶

Document Test Intent:

/**
 * \brief           Test GPIO write and read operations
 * \details         Verifies that writing a value to a GPIO pin
 *                  and then reading it back returns the same value.
 *                  This tests the basic GPIO functionality.
 */
TEST_F(GPIOTest, WriteReadConsistency) {
    // Test implementation
}

Best Practices ¶

Write Tests First (TDD) * Define expected behavior * Write failing test * Implement feature * Verify test passes
Test One Thing * Each test should verify one behavior * Keep tests focused and simple * Use descriptive names
Arrange-Act-Assert Pattern * Arrange: Set up test conditions * Act: Execute code under test * Assert: Verify expected results
Independent Tests * Tests should not depend on each other * Use fixtures for shared setup * Clean up resources in teardown
Fast Tests * Unit tests should run in milliseconds * Use mocks to avoid slow operations * Parallelize test execution
Readable Tests * Use clear variable names * Add comments for complex logic * Keep tests simple
Maintainable Tests * Refactor test code like production code * Extract common test utilities * Keep tests up to date
Comprehensive Coverage * Test happy paths * Test error paths * Test edge cases * Test boundary conditions

Troubleshooting ¶

Common Issues ¶

Tests Fail Intermittently

Check for race conditions
Verify timing assumptions
Use proper synchronization
Increase timeouts if needed

Tests Pass Locally, Fail in CI

Check environment differences
Verify dependencies
Check file paths
Review CI logs

Low Code Coverage

Identify uncovered code
Add missing tests
Test error paths
Test edge cases

Slow Test Execution

Profile test execution
Use mocks for slow operations
Parallelize tests
Optimize test setup

Nexus 嵌入式平台

导航

Related Topics

Testing Guide¶

Overview ¶

Test Infrastructure ¶

Test Framework ¶

Building Tests ¶

Running Tests ¶

Unit Testing ¶

Writing Unit Tests ¶

Test Fixtures ¶

Mocking ¶

Integration Testing ¶

HAL + OSAL Integration ¶

Framework Integration ¶

Property-Based Testing ¶

Overview ¶

Using Hypothesis ¶

Common Properties ¶

Code Coverage ¶

Measuring Coverage ¶

Coverage Targets ¶

Improving Coverage ¶

Hardware-in-the-Loop Testing ¶

HIL Test Setup ¶

Automated HIL Tests ¶

Continuous Integration ¶

GitHub Actions ¶

Test Organization ¶

Test Naming ¶

Test Documentation ¶

Best Practices ¶

Troubleshooting ¶

Common Issues ¶

See Also ¶