bsps: Move clock drivers to bsps

This patch is a part of the BSP source reorganization.

Update #3285.

bsps/shared/dev/clock/clock-simidle.c

@@ -0,0 +1,59 @@
+/*
+ *  Instantiate the clock driver shell.
+ *
+ *  Since there is no clock source on the simulator, we fake
+ *  it with a special IDLE task.
+ */
+
+#include <rtems.h>
+#include <rtems/score/percpu.h>
+#include <rtems/score/threaddispatch.h>
+
+#define CLOCK_VECTOR 0
+
+volatile bool clock_driver_enabled;
+
+#define Clock_driver_support_initialize_hardware() \
+  do { \
+    clock_driver_enabled = true; \
+  } while (0)
+
+#define Clock_driver_support_shutdown_hardware() \
+  do { \
+    clock_driver_enabled = false; \
+  } while (0)
+
+#define CLOCK_DRIVER_USE_DUMMY_TIMECOUNTER
+
+#include "clockimpl.h"
+
+/*
+ * If this is defined, then the BSP has defined a delay of some sort so
+ * time passage appears somewhat correctly. Otherwise, it runs extremely
+ * fast with no delays.
+ */
+#ifndef BSP_CLOCK_DRIVER_DELAY
+#define BSP_CLOCK_DRIVER_DELAY()
+#endif
+
+/*
+ *  Since there is no interrupt on this simulator, let's just
+ *  fake time passing.  This will not let preemption from an
+ *  interrupt work but is enough for many tests.
+ */
+void *clock_driver_sim_idle_body(
+  uintptr_t   ignored
+)
+{
+  for( ; ; ) {
+    if ( clock_driver_enabled ) {
+      Per_CPU_Control *cpu = _Thread_Dispatch_disable();
+      _ISR_Nest_level++;
+      rtems_clock_tick();
+      _ISR_Nest_level--;
+      _Thread_Dispatch_enable( cpu );
+      BSP_CLOCK_DRIVER_DELAY();
+    }
+  }
+  return 0;   /* to avoid warning */
+}