mirror of https://github.com/ARMmbed/mbed-os.git
345 lines
13 KiB
C++
345 lines
13 KiB
C++
/*
|
|
* Copyright (c) 2018-2019 Arm Limited and affiliates.
|
|
* SPDX-License-Identifier: Apache-2.0
|
|
*
|
|
* Licensed under the Apache License, Version 2.0 (the "License");
|
|
* you may not use this file except in compliance with the License.
|
|
* You may obtain a copy of the License at
|
|
*
|
|
* http://www.apache.org/licenses/LICENSE-2.0
|
|
*
|
|
* Unless required by applicable law or agreed to in writing, software
|
|
* distributed under the License is distributed on an "AS IS" BASIS,
|
|
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
* See the License for the specific language governing permissions and
|
|
* limitations under the License.
|
|
*/
|
|
#if !DEVICE_WATCHDOG
|
|
#error [NOT_SUPPORTED] Watchdog not supported for this target
|
|
#else
|
|
|
|
#include "greentea-client/test_env.h"
|
|
#include "utest/utest.h"
|
|
#include "unity/unity.h"
|
|
#include "drivers/Watchdog.h"
|
|
#include "Watchdog_reset_tests.h"
|
|
#include "mbed.h"
|
|
|
|
#define TIMEOUT_MS 100UL
|
|
|
|
/* This value is used to calculate the time to kick the watchdog.
|
|
* Given the watchdog timeout is set to TIMEOUT_MS, the kick will be performed
|
|
* with a delay of (TIMEOUT_MS - KICK_ADVANCE_MS), after the init.
|
|
*
|
|
* It is common for the watchdog peripheral to use a low precision clock source,
|
|
* e.g. the LSI RC acts as a clock source for the IWDG on ST targets.
|
|
* According to the ST spec, the 37 kHz LSI is guaranteed to have a frequency
|
|
* around 37-38 kHz, but the actual frequency range guaranteed by the production
|
|
* tests is 26 kHz up to 56 kHz.
|
|
* Bearing that in mind, a 100 ms timeout value may actually last as long as 142 ms
|
|
* and as short as 66 ms.
|
|
* The value of 35 ms is used to cover the worst case scenario (66 ms).
|
|
*/
|
|
#define KICK_ADVANCE_MS 35UL
|
|
|
|
#define MSG_VALUE_DUMMY "0"
|
|
#define CASE_DATA_INVALID 0xffffffffUL
|
|
#define CASE_DATA_PHASE2_OK 0xfffffffeUL
|
|
|
|
#define MSG_VALUE_LEN 24
|
|
#define MSG_KEY_LEN 24
|
|
|
|
#define MSG_KEY_DEVICE_READY "ready"
|
|
#define MSG_KEY_START_CASE "start_case"
|
|
#define MSG_KEY_DEVICE_RESET "dev_reset"
|
|
|
|
/* To prevent a loss of Greentea data, the serial buffers have to be flushed
|
|
* before the UART peripheral shutdown. The UART shutdown happens when the
|
|
* device is entering the deepsleep mode or performing a reset.
|
|
*
|
|
* With the current API, it is not possible to check if the hardware buffers
|
|
* are empty. However, it is possible to determine the time required for the
|
|
* buffers to flush.
|
|
*
|
|
* Assuming the biggest Tx FIFO of 128 bytes (as for CY8CPROTO_062_4343W)
|
|
* and a default UART config (9600, 8N1), flushing the Tx FIFO wold take:
|
|
* (1 start_bit + 8 data_bits + 1 stop_bit) * 128 * 1000 / 9600 = 133.3 ms.
|
|
* To be on the safe side, set the wait time to 150 ms.
|
|
*/
|
|
#define SERIAL_FLUSH_TIME_MS 150
|
|
|
|
#define TIMEOUT_US (1000 * (TIMEOUT_MS))
|
|
#define KICK_ADVANCE_US (1000 * (KICK_ADVANCE_MS))
|
|
#define SERIAL_FLUSH_TIME_US (1000 * (SERIAL_FLUSH_TIME_MS))
|
|
|
|
using utest::v1::Case;
|
|
using utest::v1::Specification;
|
|
using utest::v1::Harness;
|
|
|
|
using namespace mbed;
|
|
|
|
struct testcase_data {
|
|
int index;
|
|
int start_index;
|
|
uint32_t received_data;
|
|
};
|
|
|
|
testcase_data current_case;
|
|
|
|
Ticker wdg_kicking_ticker;
|
|
|
|
bool send_reset_notification(testcase_data *tcdata, uint32_t delay_ms)
|
|
{
|
|
char msg_value[12];
|
|
int str_len = snprintf(msg_value, sizeof msg_value, "%02x,%08lx", tcdata->start_index + tcdata->index, delay_ms);
|
|
if (str_len < 0) {
|
|
utest_printf("Failed to compose a value string to be sent to host.");
|
|
return false;
|
|
}
|
|
greentea_send_kv(MSG_KEY_DEVICE_RESET, msg_value);
|
|
return true;
|
|
}
|
|
|
|
void test_simple_reset()
|
|
{
|
|
// Phase 2. -- verify the test results.
|
|
// Verify if this test case passed based on data received from host.
|
|
if (current_case.received_data != CASE_DATA_INVALID) {
|
|
TEST_ASSERT_EQUAL(CASE_DATA_PHASE2_OK, current_case.received_data);
|
|
current_case.received_data = CASE_DATA_INVALID;
|
|
return;
|
|
}
|
|
|
|
// Phase 1. -- run the test code.
|
|
// Init the watchdog and wait for a device reset.
|
|
if (send_reset_notification(¤t_case, 2 * TIMEOUT_MS + SERIAL_FLUSH_TIME_MS) == false) {
|
|
TEST_ASSERT_MESSAGE(0, "Dev-host communication error.");
|
|
return;
|
|
}
|
|
wait_us(SERIAL_FLUSH_TIME_US); // Wait for the serial buffers to flush.
|
|
Watchdog &watchdog = Watchdog::get_instance();
|
|
TEST_ASSERT_FALSE(watchdog.is_running());
|
|
TEST_ASSERT_TRUE(watchdog.start(TIMEOUT_MS));
|
|
TEST_ASSERT_TRUE(watchdog.is_running());
|
|
// Watchdog should fire before twice the timeout value.
|
|
wait_us(2 * TIMEOUT_US); // Device reset expected.
|
|
|
|
// Watchdog reset should have occurred during a wait above.
|
|
|
|
hal_watchdog_kick();
|
|
wdg_kicking_ticker.attach_us(mbed::callback(hal_watchdog_kick), 20000); // For testsuite failure handling.
|
|
TEST_ASSERT_MESSAGE(0, "Watchdog did not reset the device as expected.");
|
|
}
|
|
|
|
#if DEVICE_SLEEP
|
|
void test_sleep_reset()
|
|
{
|
|
// Phase 2. -- verify the test results.
|
|
if (current_case.received_data != CASE_DATA_INVALID) {
|
|
TEST_ASSERT_EQUAL(CASE_DATA_PHASE2_OK, current_case.received_data);
|
|
current_case.received_data = CASE_DATA_INVALID;
|
|
return;
|
|
}
|
|
|
|
// Phase 1. -- run the test code.
|
|
if (send_reset_notification(¤t_case, 2 * TIMEOUT_MS + SERIAL_FLUSH_TIME_MS) == false) {
|
|
TEST_ASSERT_MESSAGE(0, "Dev-host communication error.");
|
|
return;
|
|
}
|
|
wait_us(SERIAL_FLUSH_TIME_US); // Wait for the serial buffers to flush.
|
|
Watchdog &watchdog = Watchdog::get_instance();
|
|
TEST_ASSERT_FALSE(watchdog.is_running());
|
|
TEST_ASSERT_TRUE(watchdog.start(TIMEOUT_MS));
|
|
TEST_ASSERT_TRUE(watchdog.is_running());
|
|
sleep_manager_lock_deep_sleep();
|
|
if (sleep_manager_can_deep_sleep()) {
|
|
TEST_ASSERT_MESSAGE(0, "Deepsleep should be disallowed.");
|
|
return;
|
|
}
|
|
// Watchdog should fire before twice the timeout value.
|
|
ThisThread::sleep_for(2 * TIMEOUT_MS); // Device reset expected.
|
|
sleep_manager_unlock_deep_sleep();
|
|
|
|
// Watchdog reset should have occurred during the sleep above.
|
|
|
|
hal_watchdog_kick();
|
|
wdg_kicking_ticker.attach_us(mbed::callback(hal_watchdog_kick), 20000); // For testsuite failure handling.
|
|
TEST_ASSERT_MESSAGE(0, "Watchdog did not reset the device as expected.");
|
|
}
|
|
|
|
#if DEVICE_LPTICKER
|
|
void test_deepsleep_reset()
|
|
{
|
|
// Phase 2. -- verify the test results.
|
|
if (current_case.received_data != CASE_DATA_INVALID) {
|
|
TEST_ASSERT_EQUAL(CASE_DATA_PHASE2_OK, current_case.received_data);
|
|
current_case.received_data = CASE_DATA_INVALID;
|
|
return;
|
|
}
|
|
|
|
// Phase 1. -- run the test code.
|
|
if (send_reset_notification(¤t_case, 2 * TIMEOUT_MS + SERIAL_FLUSH_TIME_MS) == false) {
|
|
TEST_ASSERT_MESSAGE(0, "Dev-host communication error.");
|
|
return;
|
|
}
|
|
wait_us(SERIAL_FLUSH_TIME_US); // Wait for the serial buffers to flush.
|
|
Watchdog &watchdog = Watchdog::get_instance();
|
|
TEST_ASSERT_FALSE(watchdog.is_running());
|
|
TEST_ASSERT_TRUE(watchdog.start(TIMEOUT_MS));
|
|
TEST_ASSERT_TRUE(watchdog.is_running());
|
|
if (!sleep_manager_can_deep_sleep()) {
|
|
TEST_ASSERT_MESSAGE(0, "Deepsleep should be allowed.");
|
|
}
|
|
|
|
// The Watchdog reset is allowed to be delayed up to twice the timeout
|
|
// value when the deepsleep mode is active.
|
|
// To make the test less sensitive to clock/wait accuracy, add 20% extra
|
|
// (making tha whole deepsleep wait equal to 2.2 * timeout).
|
|
ThisThread::sleep_for(220 * TIMEOUT_MS / 100); // Device reset expected.
|
|
|
|
// Watchdog reset should have occurred during the deepsleep above.
|
|
|
|
hal_watchdog_kick();
|
|
wdg_kicking_ticker.attach_us(mbed::callback(hal_watchdog_kick), 20000); // For testsuite failure handling.
|
|
TEST_ASSERT_MESSAGE(0, "Watchdog did not reset the device as expected.");
|
|
}
|
|
#endif
|
|
#endif
|
|
|
|
void test_restart_reset()
|
|
{
|
|
watchdog_features_t features = hal_watchdog_get_platform_features();
|
|
if (!features.disable_watchdog) {
|
|
TEST_IGNORE_MESSAGE("Disabling Watchdog not supported for this platform");
|
|
return;
|
|
}
|
|
|
|
// Phase 2. -- verify the test results.
|
|
if (current_case.received_data != CASE_DATA_INVALID) {
|
|
TEST_ASSERT_EQUAL(CASE_DATA_PHASE2_OK, current_case.received_data);
|
|
current_case.received_data = CASE_DATA_INVALID;
|
|
return;
|
|
}
|
|
|
|
// Phase 1. -- run the test code.
|
|
Watchdog &watchdog = Watchdog::get_instance();
|
|
TEST_ASSERT_FALSE(watchdog.is_running());
|
|
TEST_ASSERT_TRUE(watchdog.start(TIMEOUT_MS));
|
|
TEST_ASSERT_TRUE(watchdog.is_running());
|
|
wait_us(TIMEOUT_US / 2);
|
|
TEST_ASSERT_TRUE(watchdog.stop());
|
|
TEST_ASSERT_FALSE(watchdog.is_running());
|
|
// Check that stopping the Watchdog prevents a device reset.
|
|
// The watchdog should trigger at, or after the timeout value.
|
|
// The watchdog should trigger before twice the timeout value.
|
|
wait_us(TIMEOUT_US / 2 + TIMEOUT_US);
|
|
|
|
if (send_reset_notification(¤t_case, 2 * TIMEOUT_MS + SERIAL_FLUSH_TIME_MS) == false) {
|
|
TEST_ASSERT_MESSAGE(0, "Dev-host communication error.");
|
|
return;
|
|
}
|
|
wait_us(SERIAL_FLUSH_TIME_US); // Wait for the serial buffers to flush.
|
|
TEST_ASSERT_TRUE(watchdog.start(TIMEOUT_MS));
|
|
TEST_ASSERT_TRUE(watchdog.is_running());
|
|
// Watchdog should fire before twice the timeout value.
|
|
wait_us(2 * TIMEOUT_US); // Device reset expected.
|
|
|
|
// Watchdog reset should have occurred during a wait above.
|
|
|
|
hal_watchdog_kick();
|
|
wdg_kicking_ticker.attach_us(mbed::callback(hal_watchdog_kick), 20000); // For testsuite failure handling.
|
|
TEST_ASSERT_MESSAGE(0, "Watchdog did not reset the device as expected.");
|
|
}
|
|
|
|
void test_kick_reset()
|
|
{
|
|
// Phase 2. -- verify the test results.
|
|
if (current_case.received_data != CASE_DATA_INVALID) {
|
|
TEST_ASSERT_EQUAL(CASE_DATA_PHASE2_OK, current_case.received_data);
|
|
current_case.received_data = CASE_DATA_INVALID;
|
|
return;
|
|
}
|
|
|
|
// Phase 1. -- run the test code.
|
|
Watchdog &watchdog = Watchdog::get_instance();
|
|
TEST_ASSERT_FALSE(watchdog.is_running());
|
|
TEST_ASSERT_TRUE(watchdog.start(TIMEOUT_MS));
|
|
TEST_ASSERT_TRUE(watchdog.is_running());
|
|
for (int i = 3; i; i--) {
|
|
// The reset is prevented as long as the watchdog is kicked
|
|
// anytime before the timeout.
|
|
wait_us(TIMEOUT_US - KICK_ADVANCE_US);
|
|
watchdog.kick();
|
|
}
|
|
if (send_reset_notification(¤t_case, 2 * TIMEOUT_MS + SERIAL_FLUSH_TIME_MS) == false) {
|
|
TEST_ASSERT_MESSAGE(0, "Dev-host communication error.");
|
|
return;
|
|
}
|
|
wait_us(SERIAL_FLUSH_TIME_US); // Wait for the serial buffers to flush.
|
|
// Watchdog should fire before twice the timeout value.
|
|
wait_us(2 * TIMEOUT_US); // Device reset expected.
|
|
|
|
// Watchdog reset should have occurred during a wait above.
|
|
|
|
hal_watchdog_kick();
|
|
wdg_kicking_ticker.attach_us(mbed::callback(hal_watchdog_kick), 20000); // For testsuite failure handling.
|
|
TEST_ASSERT_MESSAGE(0, "Watchdog did not reset the device as expected.");
|
|
}
|
|
|
|
utest::v1::status_t case_setup(const Case *const source, const size_t index_of_case)
|
|
{
|
|
current_case.index = index_of_case;
|
|
return utest::v1::greentea_case_setup_handler(source, index_of_case);
|
|
}
|
|
|
|
int testsuite_setup(const size_t number_of_cases)
|
|
{
|
|
GREENTEA_SETUP(90, "watchdog_reset");
|
|
utest::v1::status_t status = utest::v1::greentea_test_setup_handler(number_of_cases);
|
|
if (status != utest::v1::STATUS_CONTINUE) {
|
|
return status;
|
|
}
|
|
|
|
char key[MSG_KEY_LEN + 1] = { };
|
|
char value[MSG_VALUE_LEN + 1] = { };
|
|
|
|
greentea_send_kv(MSG_KEY_DEVICE_READY, MSG_VALUE_DUMMY);
|
|
greentea_parse_kv(key, value, MSG_KEY_LEN, MSG_VALUE_LEN);
|
|
|
|
if (strcmp(key, MSG_KEY_START_CASE) != 0) {
|
|
utest_printf("Invalid message key.\n");
|
|
return utest::v1::STATUS_ABORT;
|
|
}
|
|
|
|
int num_args = sscanf(value, "%02x,%08lx", &(current_case.start_index), &(current_case.received_data));
|
|
if (num_args == 0 || num_args == EOF) {
|
|
utest_printf("Invalid data received from host\n");
|
|
return utest::v1::STATUS_ABORT;
|
|
}
|
|
|
|
utest_printf("This test suite is composed of %i test cases. Starting at index %i.\n", number_of_cases,
|
|
current_case.start_index);
|
|
return current_case.start_index;
|
|
}
|
|
|
|
Case cases[] = {
|
|
Case("Watchdog reset", case_setup, test_simple_reset),
|
|
#if DEVICE_SLEEP
|
|
Case("Watchdog reset in sleep mode", case_setup, test_sleep_reset),
|
|
#if DEVICE_LPTICKER
|
|
Case("Watchdog reset in deepsleep mode", case_setup, test_deepsleep_reset),
|
|
#endif
|
|
#endif
|
|
Case("Watchdog started again", case_setup, test_restart_reset),
|
|
Case("Kicking the Watchdog prevents reset", case_setup, test_kick_reset),
|
|
};
|
|
|
|
Specification specification((utest::v1::test_setup_handler_t) testsuite_setup, cases);
|
|
|
|
int main()
|
|
{
|
|
// Harness will start with a test case index provided by host script.
|
|
return !Harness::run(specification);
|
|
}
|
|
#endif // !DEVICE_WATCHDOG
|