fix: stabilize smp_bind_affinity_tc on SMP CI

This change fixes intermittent failures of the `core.smp_bind_affinity`
utest across SMP CI targets. The original failure was a flaky assertion on
unbound threads (`thread_inc[x] != thread_tic[x]` in thread_entry), not
incorrect bind_cpu behavior. CI logs already showed T0 binding worked
(thread_inc[0] == thread_tic[0] == 100) while the not_equal check failed.
The root issue is that unbound-thread CPU placement is platform-dependent
and outside the bind_cpu contract. On dual-core RISC-V/ARMv7 CI, an
unbound thread may legitimately run all iterations on core 0 while T0
delays on the same core, so any assertion requiring cross-core migration
(thread_inc != thread_tic, or core_mask with multiple bits) remains flaky.
This patch narrows the test to what bind_affinity actually verifies: T0
bound to core 0 must always execute on core 0. Unbound threads only add
scheduling pressure; their core_mask is printed for observation but not
asserted.
Changes:
- Sample `rt_hw_cpu_id()` and update counters under `rt_sched_lock` to
  avoid migration skew between counting and CPU sampling.
- Track `thread_core_mask` for diagnostics; assert only T0 binding via
  `thread_inc[0] == thread_tic[0]` and `thread_core_mask[0] == 1`.
- Remove flaky unbound-thread assertions (`thread_inc != thread_tic` and
  multi-core core_mask checks).
- Move T0 assertions to the main test thread after all workers finish.
- Use atomic `finsh_flag`; reset counters and `threads[]` in `utest_tc_init`.
- Spin in worker loops after `run_num` until cleanup deletes them (do not
  return from thread_entry or use `RT_WAITING_FOREVER` with `rt_thread_delay`).
- Guard `rt_thread_delete` in cleanup with non-NULL checks.

Signed-off-by: GuEe-GUI <2991707448@qq.com>

src/utest/smp/smp_bind_affinity_tc.c

@@ -24,11 +24,10 @@
  * - 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].
+ * - RT_CPUS_NR threads (T0~T(RT_CPUS_NR-1)) are created, with only T0 bound to core 0.
+ * - Each thread samples rt_hw_cpu_id() while running; T0 must stay on core 0.
+ * - Other threads are unbound and only used to add scheduling pressure; their CPU
+ * - placement is printed for observation but not asserted (platform-dependent).
  *
  * Verification Metrics:
  * - Output message: [I/utest] [ PASSED ] [ result ] testcase (core.smp_bind_affinity)
@@ -50,29 +49,39 @@ 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 rt_uint32_t thread_core_mask[RT_CPUS_NR] = {0};
+static rt_atomic_t finsh_flag;
 static int                num        = 0;
 
 static void thread_entry(void *parameter)
 {
-    int id   = 0;
+    int id;
     int para = *(int *)parameter;
+
     while (1)
     {
-        thread_tic[para]++;
+        if (thread_tic[para] >= run_num)
+        {
+            /* Spin until cleanup deletes this thread (same as other smp utests) */
+            while (1);
+        }
+
+        rt_sched_lock_level_t slvl;
+
+        /* Sample CPU and counters atomically to avoid migration skew */
+        rt_sched_lock(&slvl);
         id = rt_hw_cpu_id();
+        thread_tic[para]++;
+        thread_core_mask[para] |= (1u << id);
         if (para == id)
         {
             thread_inc[para]++;
         }
+        rt_sched_unlock(slvl);
 
         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_atomic_add(&finsh_flag, 1);
         }
         rt_thread_delay(5);
     }
@@ -104,20 +113,36 @@ static void thread_bind_affinity_tc(void)
         }
     }
 
-    while (finsh_flag != RT_CPUS_NR);
+    while (rt_atomic_load(&finsh_flag) != RT_CPUS_NR);
 
-    /* Displays the number of times a thread was executed on the relevant core */
+    /* Bound thread must always run on core 0 */
+    uassert_int_equal(thread_inc[0], thread_tic[0]);
+    uassert_int_equal(thread_core_mask[0], 1u);
+
+    /* Displays per-thread CPU statistics (unbound threads: observe only) */
     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_kprintf("Total runs[%d], Number of times thread[%d] run on [core%d]: [%4d], core mask: 0x%x, always run at core%d ? %s \r\n",
+                   run_num, j, j, thread_inc[j], thread_core_mask[j], j,
+                   (thread_inc[j] == run_num) ? "yes" : "no");
         rt_spin_unlock(&lock);
     }
 }
 
 static rt_err_t utest_tc_init(void)
 {
+    int i;
+
     rt_spin_lock_init(&lock);
+    rt_atomic_store(&finsh_flag, 0);
+    for (i = 0; i < RT_CPUS_NR; i++)
+    {
+        threads[i] = RT_NULL;
+        thread_inc[i] = 0;
+        thread_tic[i] = 0;
+        thread_core_mask[i] = 0;
+    }
     return RT_EOK;
 }
 
@@ -125,7 +150,11 @@ static rt_err_t utest_tc_cleanup(void)
 {
     for (num = 0; num < RT_CPUS_NR; num++)
     {
-        rt_thread_delete(threads[num]);
+        if (threads[num] != RT_NULL)
+        {
+            rt_thread_delete(threads[num]);
+            threads[num] = RT_NULL;
+        }
     }
     return RT_EOK;
 }