diff --git a/bl1/aarch64/bl1_arch_setup.c b/bl1/aarch64/bl1_arch_setup.c
index 758b8e8..a1ebbdb 100644
--- a/bl1/aarch64/bl1_arch_setup.c
+++ b/bl1/aarch64/bl1_arch_setup.c
@@ -44,6 +44,7 @@
 	tmp_reg |= (SCTLR_A_BIT | SCTLR_SA_BIT);
 	tmp_reg &= ~SCTLR_EE_BIT;
 	write_sctlr_el3(tmp_reg);
+	isb();
 
 	/*
 	 * Enable HVCs, route FIQs to EL3, set the next EL to be AArch64, route
diff --git a/bl1/aarch64/bl1_entrypoint.S b/bl1/aarch64/bl1_entrypoint.S
index 012b779..e25386f 100644
--- a/bl1/aarch64/bl1_entrypoint.S
+++ b/bl1/aarch64/bl1_entrypoint.S
@@ -86,7 +86,6 @@
 	mrs	x0, sctlr_el3
 	orr	x0, x0, #SCTLR_I_BIT
 	msr	sctlr_el3, x0
-
 	isb
 
 _wait_for_entrypoint:
diff --git a/bl1/aarch64/bl1_exceptions.S b/bl1/aarch64/bl1_exceptions.S
index 68d088b..7f930d8 100644
--- a/bl1/aarch64/bl1_exceptions.S
+++ b/bl1/aarch64/bl1_exceptions.S
@@ -221,6 +221,7 @@
 	bl	read_sctlr_el3
 	bic	x0, x0, x1
 	bl	write_sctlr_el3
+	isb
 	mov	x0, #DCCISW
 	bl	dcsw_op_all
 	bl	tlbialle3
diff --git a/bl2/aarch64/bl2_entrypoint.S b/bl2/aarch64/bl2_entrypoint.S
index b8af9a5..cd07aa9 100644
--- a/bl2/aarch64/bl2_entrypoint.S
+++ b/bl2/aarch64/bl2_entrypoint.S
@@ -73,7 +73,6 @@
 	mrs	x0, sctlr_el1
 	orr	x0, x0, #SCTLR_I_BIT
 	msr	sctlr_el1, x0
-
 	isb
 
 	/* ---------------------------------------------
diff --git a/bl31/aarch64/bl31_entrypoint.S b/bl31/aarch64/bl31_entrypoint.S
index 39fa605..1b79421 100644
--- a/bl31/aarch64/bl31_entrypoint.S
+++ b/bl31/aarch64/bl31_entrypoint.S
@@ -89,7 +89,6 @@
 	mrs	x1, sctlr_el3
 	orr	x1, x1, #SCTLR_I_BIT
 	msr	sctlr_el3, x1
-
 	isb
 
 	/* ---------------------------------------------
diff --git a/bl31/bl31_main.c b/bl31/bl31_main.c
index 01f00f2..755320d 100644
--- a/bl31/bl31_main.c
+++ b/bl31/bl31_main.c
@@ -100,6 +100,7 @@
 	assert(cm_get_context(mpidr, NON_SECURE));
 	cm_set_next_eret_context(NON_SECURE);
 	write_vbar_el3((uint64_t) runtime_exceptions);
+	isb();
 	next_image_type = NON_SECURE;
 
 	/*
diff --git a/drivers/arm/gic/aarch64/gic_v3_sysregs.S b/drivers/arm/gic/aarch64/gic_v3_sysregs.S
index 2a96da7..ddf85a8 100644
--- a/drivers/arm/gic/aarch64/gic_v3_sysregs.S
+++ b/drivers/arm/gic/aarch64/gic_v3_sysregs.S
@@ -67,23 +67,19 @@
 
 func write_icc_sre_el1
 	msr	ICC_SRE_EL1, x0
-	isb
 	ret
 
 
 func write_icc_sre_el2
 	msr	ICC_SRE_EL2, x0
-	isb
 	ret
 
 
 func write_icc_sre_el3
 	msr	ICC_SRE_EL3, x0
-	isb
 	ret
 
 
 func write_icc_pmr_el1
 	msr	ICC_PMR_EL1, x0
-	isb
 	ret
diff --git a/lib/aarch64/cache_helpers.S b/lib/aarch64/cache_helpers.S
index 2649ad0..dc91975 100644
--- a/lib/aarch64/cache_helpers.S
+++ b/lib/aarch64/cache_helpers.S
@@ -46,57 +46,41 @@
 
 func dcisw
 	dc	isw, x0
-	dsb	sy
-	isb
 	ret
 
 
 func dccisw
 	dc	cisw, x0
-	dsb	sy
-	isb
 	ret
 
 
 func dccsw
 	dc	csw, x0
-	dsb	sy
-	isb
 	ret
 
 
 func dccvac
 	dc	cvac, x0
-	dsb	sy
-	isb
 	ret
 
 
 func dcivac
 	dc	ivac, x0
-	dsb	sy
-	isb
 	ret
 
 
 func dccivac
 	dc	civac, x0
-	dsb	sy
-	isb
 	ret
 
 
 func dccvau
 	dc	cvau, x0
-	dsb	sy
-	isb
 	ret
 
 
 func dczva
 	dc	zva, x0
-	dsb	sy
-	isb
 	ret
 
 
diff --git a/lib/aarch64/cpu_helpers.S b/lib/aarch64/cpu_helpers.S
index 573d0b8..4e5eb5b 100644
--- a/lib/aarch64/cpu_helpers.S
+++ b/lib/aarch64/cpu_helpers.S
@@ -52,5 +52,6 @@
 	bl	read_cpuectlr
 	orr	x0, x0, #CPUECTLR_SMP_BIT
 	bl	write_cpuectlr
+	isb
 smp_setup_end:
 	ret	x19
diff --git a/lib/aarch64/misc_helpers.S b/lib/aarch64/misc_helpers.S
index e7b2331..c33ade2 100644
--- a/lib/aarch64/misc_helpers.S
+++ b/lib/aarch64/misc_helpers.S
@@ -187,19 +187,16 @@
 
 func write_spsr_el1
 	msr	spsr_el1, x0
-	isb
 	ret
 
 
 func write_spsr_el2
 	msr	spsr_el2, x0
-	isb
 	ret
 
 
 func write_spsr_el3
 	msr	spsr_el3, x0
-	isb
 	ret
 
 
@@ -240,19 +237,16 @@
 
 func write_elr_el1
 	msr	elr_el1, x0
-	isb
 	ret
 
 
 func write_elr_el2
 	msr	elr_el2, x0
-	isb
 	ret
 
 
 func write_elr_el3
 	msr	elr_el3, x0
-	isb
 	ret
 
 
diff --git a/lib/aarch64/sysreg_helpers.S b/lib/aarch64/sysreg_helpers.S
index 61468f9..1d73ba9 100644
--- a/lib/aarch64/sysreg_helpers.S
+++ b/lib/aarch64/sysreg_helpers.S
@@ -201,19 +201,16 @@
 
 func write_vbar_el1
 	msr	vbar_el1, x0
-	isb
 	ret
 
 
 func write_vbar_el2
 	msr	vbar_el2, x0
-	isb
 	ret
 
 
 func write_vbar_el3
 	msr	vbar_el3, x0
-	isb
 	ret
 
 
@@ -238,19 +235,16 @@
 
 func write_afsr0_el1
 	msr	afsr0_el1, x0
-	isb
 	ret
 
 
 func write_afsr0_el2
 	msr	afsr0_el2, x0
-	isb
 	ret
 
 
 func write_afsr0_el3
 	msr	afsr0_el3, x0
-	isb
 	ret
 
 
@@ -275,19 +269,16 @@
 
 func write_far_el1
 	msr	far_el1, x0
-	isb
 	ret
 
 
 func write_far_el2
 	msr	far_el2, x0
-	isb
 	ret
 
 
 func write_far_el3
 	msr	far_el3, x0
-	isb
 	ret
 
 
@@ -312,19 +303,16 @@
 
 func write_mair_el1
 	msr	mair_el1, x0
-	isb
 	ret
 
 
 func write_mair_el2
 	msr	mair_el2, x0
-	isb
 	ret
 
 
 func write_mair_el3
 	msr	mair_el3, x0
-	isb
 	ret
 
 
@@ -349,19 +337,16 @@
 
 func write_amair_el1
 	msr	amair_el1, x0
-	isb
 	ret
 
 
 func write_amair_el2
 	msr	amair_el2, x0
-	isb
 	ret
 
 
 func write_amair_el3
 	msr	amair_el3, x0
-	isb
 	ret
 
 
@@ -405,19 +390,16 @@
 
 func write_rmr_el1
 	msr	rmr_el1, x0
-	isb
 	ret
 
 
 func write_rmr_el2
 	msr	rmr_el2, x0
-	isb
 	ret
 
 
 func write_rmr_el3
 	msr	rmr_el3, x0
-	isb
 	ret
 
 
@@ -442,19 +424,16 @@
 
 func write_afsr1_el1
 	msr	afsr1_el1, x0
-	isb
 	ret
 
 
 func write_afsr1_el2
 	msr	afsr1_el2, x0
-	isb
 	ret
 
 
 func write_afsr1_el3
 	msr	afsr1_el3, x0
-	isb
 	ret
 
 
@@ -479,22 +458,16 @@
 
 func write_sctlr_el1
 	msr	sctlr_el1, x0
-	dsb	sy
-	isb
 	ret
 
 
 func write_sctlr_el2
 	msr	sctlr_el2, x0
-	dsb	sy
-	isb
 	ret
 
 
 func write_sctlr_el3
 	msr	sctlr_el3, x0
-	dsb	sy
-	isb
 	ret
 
 
@@ -519,22 +492,16 @@
 
 func write_actlr_el1
 	msr	actlr_el1, x0
-	dsb	sy
-	isb
 	ret
 
 
 func write_actlr_el2
 	msr	actlr_el2, x0
-	dsb	sy
-	isb
 	ret
 
 
 func write_actlr_el3
 	msr	actlr_el3, x0
-	dsb	sy
-	isb
 	ret
 
 
@@ -559,22 +526,16 @@
 
 func write_esr_el1
 	msr	esr_el1, x0
-	dsb	sy
-	isb
 	ret
 
 
 func write_esr_el2
 	msr	esr_el2, x0
-	dsb	sy
-	isb
 	ret
 
 
 func write_esr_el3
 	msr	esr_el3, x0
-	dsb	sy
-	isb
 	ret
 
 
@@ -599,22 +560,16 @@
 
 func write_tcr_el1
 	msr	tcr_el1, x0
-	dsb	sy
-	isb
 	ret
 
 
 func write_tcr_el2
 	msr	tcr_el2, x0
-	dsb	sy
-	isb
 	ret
 
 
 func write_tcr_el3
 	msr	tcr_el3, x0
-	dsb	sy
-	isb
 	ret
 
 
@@ -643,15 +598,11 @@
 
 func write_cptr_el2
 	msr	cptr_el2, x0
-	dsb	sy
-	isb
 	ret
 
 
 func write_cptr_el3
 	msr	cptr_el3, x0
-	dsb	sy
-	isb
 	ret
 
 
@@ -676,19 +627,16 @@
 
 func write_ttbr0_el1
 	msr	ttbr0_el1, x0
-	isb
 	ret
 
 
 func write_ttbr0_el2
 	msr	ttbr0_el2, x0
-	isb
 	ret
 
 
 func write_ttbr0_el3
 	msr	ttbr0_el3, x0
-	isb
 	ret
 
 
@@ -711,7 +659,6 @@
 
 func write_ttbr1_el1
 	msr	ttbr1_el1, x0
-	isb
 	ret
 
 
@@ -730,8 +677,6 @@
 
 func write_hcr
 	msr	hcr_el2, x0
-	dsb	sy
-	isb
 	ret
 
 
@@ -762,8 +707,6 @@
 
 func write_cpuectlr
 	msr	CPUECTLR_EL1, x0
-	dsb	sy
-	isb
 	ret
 
 
@@ -789,8 +732,6 @@
 
 func write_scr
 	msr	scr_el3, x0
-	dsb	sy
-	isb
 	ret
 
 
@@ -818,6 +759,7 @@
 	mov	x1, #AARCH64_CPTR_TFP
 	bic	x0, x0, x1
 	msr	cptr_el3, x0
+	isb
 	ret
 
 
diff --git a/lib/aarch64/tlb_helpers.S b/lib/aarch64/tlb_helpers.S
index ec1558b..8dfae12 100644
--- a/lib/aarch64/tlb_helpers.S
+++ b/lib/aarch64/tlb_helpers.S
@@ -41,47 +41,33 @@
 
 func tlbialle1
 	tlbi	alle1
-	dsb	sy
-	isb
 	ret
 
 
 func tlbialle1is
 	tlbi	alle1is
-	dsb	sy
-	isb
 	ret
 
 
 func tlbialle2
 	tlbi	alle2
-	dsb	sy
-	isb
 	ret
 
 
 func tlbialle2is
 	tlbi	alle2is
-	dsb	sy
-	isb
 	ret
 
 
 func tlbialle3
 	tlbi	alle3
-	dsb	sy
-	isb
 	ret
 
 
 func tlbialle3is
 	tlbi	alle3is
-	dsb	sy
-	isb
 	ret
 
 func tlbivmalle1
 	tlbi	vmalle1
-	dsb	sy
-	isb
 	ret
diff --git a/plat/fvp/aarch64/plat_common.c b/plat/fvp/aarch64/plat_common.c
index c8e529d..e2f2343 100644
--- a/plat/fvp/aarch64/plat_common.c
+++ b/plat/fvp/aarch64/plat_common.c
@@ -69,6 +69,8 @@
 	ttbr = (unsigned long) l1_xlation_table;
 
 	if (GET_EL(current_el) == MODE_EL3) {
+		assert((read_sctlr_el3() & SCTLR_M_BIT) == 0);
+
 		write_mair_el3(mair);
 		tcr |= TCR_EL3_RES1;
 		/* Invalidate EL3 TLBs */
@@ -77,11 +79,19 @@
 		write_tcr_el3(tcr);
 		write_ttbr0_el3(ttbr);
 
+		/* ensure all translation table writes have drained into memory,
+		 * the TLB invalidation is complete, and translation register
+		 * writes are committed before enabling the MMU
+		 */
+		dsb();
+		isb();
+
 		sctlr = read_sctlr_el3();
 		sctlr |= SCTLR_WXN_BIT | SCTLR_M_BIT | SCTLR_I_BIT;
 		sctlr |= SCTLR_A_BIT | SCTLR_C_BIT;
 		write_sctlr_el3(sctlr);
 	} else {
+		assert((read_sctlr_el1() & SCTLR_M_BIT) == 0);
 
 		write_mair_el1(mair);
 		/* Invalidate EL1 TLBs */
@@ -90,11 +100,20 @@
 		write_tcr_el1(tcr);
 		write_ttbr0_el1(ttbr);
 
+		/* ensure all translation table writes have drained into memory,
+		 * the TLB invalidation is complete, and translation register
+		 * writes are committed before enabling the MMU
+		 */
+		dsb();
+		isb();
+
 		sctlr = read_sctlr_el1();
 		sctlr |= SCTLR_WXN_BIT | SCTLR_M_BIT | SCTLR_I_BIT;
 		sctlr |= SCTLR_A_BIT | SCTLR_C_BIT;
 		write_sctlr_el1(sctlr);
 	}
+	/* ensure the MMU enable takes effect immediately */
+	isb();
 
 	return;
 }
@@ -113,6 +132,8 @@
 		sctlr = sctlr & ~(SCTLR_M_BIT | SCTLR_C_BIT);
 		write_sctlr_el1(sctlr);
 	}
+	/* ensure the MMU disable takes effect immediately */
+	isb();
 
 	/* Flush the caches */
 	dcsw_op_all(DCCISW);
diff --git a/plat/fvp/plat_gic.c b/plat/fvp/plat_gic.c
index 8457af1..db3c9cf 100644
--- a/plat/fvp/plat_gic.c
+++ b/plat/fvp/plat_gic.c
@@ -86,6 +86,7 @@
 	 */
 	scr_val = read_scr();
 	write_scr(scr_val | SCR_NS_BIT);
+	isb();	/* ensure NS=1 takes effect before accessing ICC_SRE_EL2 */
 
 	/*
 	 * By default EL2 and NS-EL1 software should be able to enable GICv3
@@ -103,9 +104,11 @@
 	write_icc_sre_el2(val | ICC_SRE_EN | ICC_SRE_SRE);
 
 	write_icc_pmr_el1(GIC_PRI_MASK);
+	isb();	/* commite ICC_* changes before setting NS=0 */
 
 	/* Restore SCR_EL3 */
 	write_scr(scr_val);
+	isb();	/* ensure NS=0 takes effect immediately */
 }
 
 /*******************************************************************************
diff --git a/plat/fvp/plat_pm.c b/plat/fvp/plat_pm.c
index 5430fff..f80e2d7 100644
--- a/plat/fvp/plat_pm.c
+++ b/plat/fvp/plat_pm.c
@@ -54,7 +54,11 @@
 	if (target_afflvl != MPIDR_AFFLVL0)
 		return PSCI_E_INVALID_PARAMS;
 
-	/* Enter standby state */
+	/*
+	 * Enter standby state
+	 * dsb is good practice before using wfi to enter low power states
+	 */
+	dsb();
 	wfi();
 
 	return PSCI_E_SUCCESS;
diff --git a/services/std_svc/psci/psci_afflvl_off.c b/services/std_svc/psci/psci_afflvl_off.c
index e007bc3..21a4d1a 100644
--- a/services/std_svc/psci/psci_afflvl_off.c
+++ b/services/std_svc/psci/psci_afflvl_off.c
@@ -82,6 +82,7 @@
 	sctlr = read_sctlr_el3();
 	sctlr &= ~SCTLR_C_BIT;
 	write_sctlr_el3(sctlr);
+	isb();	/* ensure MMU disable takes immediate effect */
 
 	/*
 	 * CAUTION: This flush to the level of unification makes an assumption
diff --git a/services/std_svc/psci/psci_afflvl_suspend.c b/services/std_svc/psci/psci_afflvl_suspend.c
index dc12f7a..534e4a9 100644
--- a/services/std_svc/psci/psci_afflvl_suspend.c
+++ b/services/std_svc/psci/psci_afflvl_suspend.c
@@ -198,6 +198,7 @@
 	sctlr = read_sctlr_el3();
 	sctlr &= ~SCTLR_C_BIT;
 	write_sctlr_el3(sctlr);
+	isb();	/* ensure MMU disable takes immediate effect */
 
 	/*
 	 * CAUTION: This flush to the level of unification makes an assumption
diff --git a/services/std_svc/psci/psci_entry.S b/services/std_svc/psci/psci_entry.S
index e2c690d..ec55a81 100644
--- a/services/std_svc/psci/psci_entry.S
+++ b/services/std_svc/psci/psci_entry.S
@@ -75,7 +75,6 @@
 	 * ---------------------------------------------
 	 */
 	msr	spsel, #0
-	isb
 
 	bl	read_mpidr
 	mov	x19, x0
@@ -158,7 +157,7 @@
 	ret
 
 func final_wfi
-	dsb	sy
+	dsb	sy		// ensure write buffer empty
 	wfi
 wfi_spill:
 	b	wfi_spill