diff --git a/arch/xtensa/src/common/xtensa_idle.c b/arch/xtensa/src/common/xtensa_idle.c
index bf6cad3ac1..ba6c415e13 100644
--- a/arch/xtensa/src/common/xtensa_idle.c
+++ b/arch/xtensa/src/common/xtensa_idle.c
@@ -69,36 +69,9 @@ void up_idle(void)
 
   nxsched_process_timer();
 #else
-
   /* This would be an appropriate place to put some MCU-specific logic to
    * sleep in a reduced power mode until an interrupt occurs to save power
    */
 
-  /* This is a kludge that I still don't understand.  The call to
-   * kmm_trysemaphore() in the nx_start.c IDLE loop seems necessary for the
-   * good health of the IDLE loop.  When the work queue is enabled, this
-   * logic is removed from the IDLE loop and it appears that we are somehow
-   * left idling with interrupts non-functional. The following should be
-   * no-op, it just disables then re-enables interrupts.  But it fixes the
-   * problem and will stay here until I understand the problem/fix better.
-   *
-   * And no, the contents of the CP0 status register are not incorrect.
-   * But for some reason the status register needs to be re-written again
-   * on this thread for it to take effect.  This might be a PIC32-only
-   * issue?
-   */
-
-#ifdef CONFIG_SCHED_WORKQUEUE
-  irqstate_t flags = enter_critical_section();
-  leave_critical_section(flags);
-#endif
-#endif
-
-  /* This would be an appropriate place to put some MCU-specific logic to
-   * sleep in a reduced power mode until an interrupt occurs to save power
-   */
-
-#if XCHAL_HAVE_INTERRUPTS
-  __asm__ __volatile__ ("waiti 0");
 #endif
 }