diff --git a/arch/powerpc/kernel/head_64.S b/arch/powerpc/kernel/head_64.S
index ba30e3177b05..45d81976987f 100644
--- a/arch/powerpc/kernel/head_64.S
+++ b/arch/powerpc/kernel/head_64.S
@@ -1914,24 +1914,6 @@ _GLOBAL(hmt_start_secondary)
 	blr
 #endif
 
-#if defined(CONFIG_KEXEC) || defined(CONFIG_SMP)
-_GLOBAL(smp_release_cpus)
-	/* All secondary cpus are spinning on a common
-	 * spinloop, release them all now so they can start
-	 * to spin on their individual paca spinloops.
-	 * For non SMP kernels, the secondary cpus never
-	 * get out of the common spinloop.
-	 * XXX This does nothing useful on iSeries, secondaries are
-	 * already waiting on their paca.
-	 */
-	li	r3,1
-	LOADADDR(r5,__secondary_hold_spinloop)
-	std	r3,0(r5)
-	sync
-	blr
-#endif /* CONFIG_SMP */
-
-
 /*
  * We put a few things here that have to be page-aligned.
  * This stuff goes at the beginning of the bss, which is page-aligned.
diff --git a/arch/powerpc/kernel/setup_64.c b/arch/powerpc/kernel/setup_64.c
index d4a3c5dd1a21..275d86ddd612 100644
--- a/arch/powerpc/kernel/setup_64.c
+++ b/arch/powerpc/kernel/setup_64.c
@@ -103,8 +103,6 @@ extern void htab_initialize(void);
 extern void early_init_devtree(void *flat_dt);
 extern void unflatten_device_tree(void);
 
-extern void smp_release_cpus(void);
-
 int have_of = 1;
 int boot_cpuid = 0;
 int boot_cpuid_phys = 0;
@@ -400,6 +398,27 @@ void __init early_setup(unsigned long dt_ptr)
 }
 
 
+#if defined(CONFIG_SMP) || defined(CONFIG_KEXEC)
+void smp_release_cpus(void)
+{
+	extern unsigned long __secondary_hold_spinloop;
+
+	DBG(" -> smp_release_cpus()\n");
+
+	/* All secondary cpus are spinning on a common spinloop, release them
+	 * all now so they can start to spin on their individual paca
+	 * spinloops. For non SMP kernels, the secondary cpus never get out
+	 * of the common spinloop.
+	 * This is useless but harmless on iSeries, secondaries are already
+	 * waiting on their paca spinloops. */
+
+	__secondary_hold_spinloop = 1;
+	mb();
+
+	DBG(" <- smp_release_cpus()\n");
+}
+#endif /* CONFIG_SMP || CONFIG_KEXEC */
+
 /*
  * Initialize some remaining members of the ppc64_caches and systemcfg structures
  * (at least until we get rid of them completely). This is mostly some
diff --git a/arch/ppc64/kernel/head.S b/arch/ppc64/kernel/head.S
index 0b65efe1682e..db1cf397be2d 100644
--- a/arch/ppc64/kernel/head.S
+++ b/arch/ppc64/kernel/head.S
@@ -1914,24 +1914,6 @@ _GLOBAL(hmt_start_secondary)
 	blr
 #endif
 
-#if defined(CONFIG_KEXEC) || defined(CONFIG_SMP)
-_GLOBAL(smp_release_cpus)
-	/* All secondary cpus are spinning on a common
-	 * spinloop, release them all now so they can start
-	 * to spin on their individual paca spinloops.
-	 * For non SMP kernels, the secondary cpus never
-	 * get out of the common spinloop.
-	 * XXX This does nothing useful on iSeries, secondaries are
-	 * already waiting on their paca.
-	 */
-	li	r3,1
-	LOADADDR(r5,__secondary_hold_spinloop)
-	std	r3,0(r5)
-	sync
-	blr
-#endif /* CONFIG_SMP */
-
-
 /*
  * We put a few things here that have to be page-aligned.
  * This stuff goes at the beginning of the bss, which is page-aligned.
diff --git a/arch/ppc64/kernel/machine_kexec.c b/arch/ppc64/kernel/machine_kexec.c
index bf7cc4f8210f..ff8679f260f3 100644
--- a/arch/ppc64/kernel/machine_kexec.c
+++ b/arch/ppc64/kernel/machine_kexec.c
@@ -244,7 +244,6 @@ static void kexec_prepare_cpus(void)
 
 static void kexec_prepare_cpus(void)
 {
-	extern void smp_release_cpus(void);
 	/*
 	 * move the secondarys to us so that we can copy
 	 * the new kernel 0-0x100 safely
diff --git a/include/asm-ppc64/smp.h b/include/asm-ppc64/smp.h
index 46c04046952d..ba0f5c8bbb22 100644
--- a/include/asm-ppc64/smp.h
+++ b/include/asm-ppc64/smp.h
@@ -79,6 +79,7 @@ extern int smt_enabled_at_boot;
 extern int smp_mpic_probe(void);
 extern void smp_mpic_setup_cpu(int cpu);
 extern void smp_generic_kick_cpu(int nr);
+extern void smp_release_cpus(void);
 
 extern void smp_generic_give_timebase(void);
 extern void smp_generic_take_timebase(void);