sched: replace sync pause with async pause for nxsched_suspend

reason:
In the kernel, we are planning to remove all occurrences of up_cpu_pause as one of the steps to
simplify the implementation of critical sections. The goal is to enable spin_lock_irqsave to encapsulate critical sections,
thereby facilitating the replacement of critical sections(big lock) with smaller spin_lock_irqsave(small lock)

Configuring NuttX and compile:
$ ./tools/configure.sh -l qemu-armv8a:nsh_smp
$ make
Running with qemu
$ qemu-system-aarch64 -cpu cortex-a53 -smp 4 -nographic
-machine virt,virtualization=on,gic-version=3
-net none -chardev stdio,id=con,mux=on -serial chardev:con
-mon chardev=con,mode=readline -kernel ./nuttx

Signed-off-by: hujun5 <hujun5@xiaomi.com>

sched/sched/sched_suspend.c

@@ -36,6 +36,57 @@
 
 #include "sched/sched.h"
 
+#ifdef CONFIG_SMP
+/****************************************************************************
+ * Private Type Declarations
+ ****************************************************************************/
+
+struct suspend_arg_s
+{
+  pid_t pid;
+  cpu_set_t saved_affinity;
+  uint16_t saved_flags;
+  bool need_restore;
+};
+
+/****************************************************************************
+ * Private Functions
+ ****************************************************************************/
+
+static int nxsched_suspend_handler(FAR void *cookie)
+{
+  FAR struct suspend_arg_s *arg = cookie;
+  FAR struct tcb_s *tcb;
+  irqstate_t flags;
+
+  flags = enter_critical_section();
+  tcb = nxsched_get_tcb(arg->pid);
+
+  if (!tcb || tcb->task_state == TSTATE_TASK_INVALID ||
+      (tcb->flags & TCB_FLAG_EXIT_PROCESSING) != 0)
+    {
+      /* There is no TCB with this pid or, if there is, it is not a task. */
+
+      leave_critical_section(flags);
+      return OK;
+    }
+
+  if (arg->need_restore)
+    {
+      tcb->affinity = arg->saved_affinity;
+      tcb->flags = arg->saved_flags;
+    }
+
+  nxsched_remove_readytorun(tcb, false);
+
+  tcb->task_state = TSTATE_TASK_STOPPED;
+  dq_addlast((FAR dq_entry_t *)tcb, &g_stoppedtasks);
+
+  leave_critical_section(flags);
+  return OK;
+}
+#endif
+
 /****************************************************************************
  * Public Functions
  ****************************************************************************/
@@ -94,18 +145,46 @@ void nxsched_suspend(FAR struct tcb_s *tcb)
 
       /* Remove the tcb task from the ready-to-run list. */
 
-      switch_needed = nxsched_remove_readytorun(tcb, true);
+#ifdef CONFIG_SMP
-
+      if (tcb->task_state == TSTATE_TASK_RUNNING && tcb->cpu != this_cpu())
-      /* Add the task to the specified blocked task list */
-
-      tcb->task_state = TSTATE_TASK_STOPPED;
-      dq_addlast((FAR dq_entry_t *)tcb, list_stoppedtasks());
-
-      /* Now, perform the context switch if one is needed */
-
-      if (switch_needed)
         {
-          up_switch_context(this_task(), rtcb);
+          struct suspend_arg_s arg;
+
+          if ((tcb->flags & TCB_FLAG_CPU_LOCKED) != 0)
+            {
+              arg.pid = tcb->pid;
+              arg.need_restore = false;
+            }
+          else
+            {
+              arg.pid = tcb->pid;
+              arg.saved_flags = tcb->flags;
+              arg.saved_affinity = tcb->affinity;
+              arg.need_restore = true;
+
+              tcb->flags |= TCB_FLAG_CPU_LOCKED;
+              CPU_SET(tcb->cpu, &tcb->affinity);
+            }
+
+          nxsched_smp_call_single(tcb->cpu, nxsched_suspend_handler,
+                                  &arg, true);
+        }
+      else
+#endif
+        {
+          switch_needed = nxsched_remove_readytorun(tcb, true);
+
+          /* Add the task to the specified blocked task list */
+
+          tcb->task_state = TSTATE_TASK_STOPPED;
+          dq_addlast((FAR dq_entry_t *)tcb, list_stoppedtasks());
+
+          /* Now, perform the context switch if one is needed */
+
+          if (switch_needed)
+            {
+              up_switch_context(this_task(), rtcb);
+            }
         }
     }