/* * Copyright (c) 2006-2024, RT-Thread Development Team * * SPDX-License-Identifier: Apache-2.0 * * Change Logs: * Date Author Notes * 2024-08-10 RV the first version */ #include #include "utest.h" /** * @brief Binding core affinity testcase. * * @note Create RT_CPUS_NR threads, thread 0 is bound to core 0, other threads are not bound to specific cores, * after running for a set number of times, count the number of times each core is run on the corresponding core, * thread 0 should always be run on core 0, other threads will be run on different cores. * * Test Case Name: [smp_bind_affinity] * * Test Objectives: * - Verify that threads bound to specific cores run on those cores. * * Test Scenarios: * - RT_CPUS_NR threads (T0~T(RT_CPUS_NR-1)) are created, with only T0 bound to core 0. When thread Tx is running, * - thread_tic[x] increments by 1; if x is equal to the core ID, thread_inc[x] also increments by 1. After the * - program runs for a period of time, observe the value relationship between the thread_inc and thread_tic arrays. * - If thread_inc[x] is equal to thread_tic[x], it indicates that thread Tx is correctly bound to core x. In this test case, * - only thread_inc[0] will be equal to thread_tic[0]. * * Verification Metrics: * - Output message: [I/utest] [ PASSED ] [ result ] testcase (core.smp_bind_affinity) * * Dependencies: * - Enable RT_USING_SMP, set RT_THREAD_PRIORITY_MAX = 256. * * Expected Results: * - You will see the relevant PASS message. Additionally, RT_CPUS_NR lines of printed information will be displayed, * - indicating whether each thread Tx has been running on core x all the time. The results show that only T0 has been * - running on core 0 consistently. */ /* Number of thread runs */ static int run_num = 100; #define THREAD_STACK_SIZE UTEST_THR_STACK_SIZE #define THREAD_PRIORITY 20 static rt_thread_t threads[RT_CPUS_NR]; static struct rt_spinlock lock; static int thread_inc[RT_CPUS_NR] = {0}; static int thread_tic[RT_CPUS_NR] = {0}; static int finsh_flag = 0; static int num = 0; static void thread_entry(void *parameter) { int id = 0; int para = *(int *)parameter; while (1) { thread_tic[para]++; id = rt_hw_cpu_id(); if (para == id) { thread_inc[para]++; } if (thread_tic[para] == run_num) { if (para == 0) uassert_int_equal(thread_inc[para], thread_tic[para]); else uassert_int_not_equal(thread_inc[para], thread_tic[para]); finsh_flag ++; } rt_thread_delay(5); } } static void thread_bind_affinity_tc(void) { static int params[RT_CPUS_NR] = {0}; char thread_name[8]; int i, j; for (i = 0; i < RT_CPUS_NR; i++) { params[i] = i; } /* Create RT_CPUS_NR threads Thread 0 is bound to core 0 Other threads are not bound */ for (i = 0; i < RT_CPUS_NR; i++) { rt_snprintf(thread_name, sizeof(thread_name), "thread%d", i); threads[i] = rt_thread_create(thread_name, thread_entry, (int *)¶ms[i], THREAD_STACK_SIZE, THREAD_PRIORITY, 20); if (i == 0) { rt_thread_control(threads[0], RT_THREAD_CTRL_BIND_CPU, (void *)0); } if (threads[i] != RT_NULL) { rt_thread_startup(threads[i]); } } while (finsh_flag != RT_CPUS_NR); /* Displays the number of times a thread was executed on the relevant core */ for (j = 0; j < RT_CPUS_NR; j++) { rt_spin_lock(&lock); rt_kprintf("Total runs[%d], Number of times thread[%d] run on [core%d]: [%4d], always run at core%d ? %s \r\n", run_num, j, j, thread_inc[j], j, (thread_inc[j] == run_num) ? "yes" : "no"); rt_spin_unlock(&lock); } } static rt_err_t utest_tc_init(void) { rt_spin_lock_init(&lock); return RT_EOK; } static rt_err_t utest_tc_cleanup(void) { for (num = 0; num < RT_CPUS_NR; num++) { rt_thread_delete(threads[num]); } return RT_EOK; } static void testcase(void) { UTEST_UNIT_RUN(thread_bind_affinity_tc); } UTEST_TC_EXPORT(testcase, "core.smp_bind_affinity", utest_tc_init, utest_tc_cleanup, 10);