Merge pull request #1272 from dp-arm/dp/extensions

Refactor SPE/SVE code and fix some bugs in AMUv1 on AArch32

7 files changed, 247 insertions, 238 deletions

diff --git a/include/lib/extensions/spe.h b/include/lib/extensions/spe.h
index 8a74127a..b2b188ef 100644
--- a/include/lib/extensions/spe.h
+++ b/include/lib/extensions/spe.h
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2017, ARM Limited and Contributors. All rights reserved.
+ * Copyright (c) 2017-2018, ARM Limited and Contributors. All rights reserved.
  *
  * SPDX-License-Identifier: BSD-3-Clause
  */
@@ -7,6 +7,7 @@
 #ifndef __SPE_H__
 #define __SPE_H__
 
+int spe_supported(void);
 void spe_enable(int el2_unused);
 void spe_disable(void);
 
diff --git a/include/lib/extensions/sve.h b/include/lib/extensions/sve.h
index 28923e3f..9c7f37f2 100644
--- a/include/lib/extensions/sve.h
+++ b/include/lib/extensions/sve.h
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2017, ARM Limited and Contributors. All rights reserved.
+ * Copyright (c) 2017-2018, ARM Limited and Contributors. All rights reserved.
  *
  * SPDX-License-Identifier: BSD-3-Clause
  */
@@ -7,6 +7,7 @@
 #ifndef __SVE_H__
 #define __SVE_H__
 
+int sve_supported(void);
 void sve_enable(int el2_unused);
 
 #endif /* __SVE_H__ */
diff --git a/lib/extensions/amu/aarch32/amu.c b/lib/extensions/amu/aarch32/amu.c
index 68cc4b34..05c98f1c 100644
--- a/lib/extensions/amu/aarch32/amu.c
+++ b/lib/extensions/amu/aarch32/amu.c
@@ -30,7 +30,7 @@ int amu_supported(void)
 
 void amu_enable(int el2_unused)
 {
-	if (!amu_supported())
+	if (amu_supported() == 0)
 		return;
 
 	if (el2_unused) {
@@ -54,7 +54,7 @@ void amu_enable(int el2_unused)
 /* Read the group 0 counter identified by the given `idx`. */
 uint64_t amu_group0_cnt_read(int idx)
 {
-	assert(amu_supported());
+	assert(amu_supported() != 0);
 	assert(idx >= 0 && idx < AMU_GROUP0_NR_COUNTERS);
 
 	return amu_group0_cnt_read_internal(idx);
@@ -63,7 +63,7 @@ uint64_t amu_group0_cnt_read(int idx)
 /* Write the group 0 counter identified by the given `idx` with `val`. */
 void amu_group0_cnt_write(int idx, uint64_t val)
 {
-	assert(amu_supported());
+	assert(amu_supported() != 0);
 	assert(idx >= 0 && idx < AMU_GROUP0_NR_COUNTERS);
 
 	amu_group0_cnt_write_internal(idx, val);
@@ -73,7 +73,7 @@ void amu_group0_cnt_write(int idx, uint64_t val)
 /* Read the group 1 counter identified by the given `idx`. */
 uint64_t amu_group1_cnt_read(int idx)
 {
-	assert(amu_supported());
+	assert(amu_supported() != 0);
 	assert(idx >= 0 && idx < AMU_GROUP1_NR_COUNTERS);
 
 	return amu_group1_cnt_read_internal(idx);
@@ -82,7 +82,7 @@ uint64_t amu_group1_cnt_read(int idx)
 /* Write the group 1 counter identified by the given `idx` with `val`. */
 void amu_group1_cnt_write(int idx, uint64_t val)
 {
-	assert(amu_supported());
+	assert(amu_supported() != 0);
 	assert(idx >= 0 && idx < AMU_GROUP1_NR_COUNTERS);
 
 	amu_group1_cnt_write_internal(idx, val);
@@ -91,7 +91,7 @@ void amu_group1_cnt_write(int idx, uint64_t val)
 
 void amu_group1_set_evtype(int idx, unsigned int val)
 {
-	assert(amu_supported());
+	assert(amu_supported() != 0);
 	assert(idx >= 0 && idx < AMU_GROUP1_NR_COUNTERS);
 
 	amu_group1_set_evtype_internal(idx, val);
@@ -103,13 +103,14 @@ static void *amu_context_save(const void *arg)
 	struct amu_ctx *ctx;
 	int i;
 
-	if (!amu_supported())
+	if (amu_supported() == 0)
 		return (void *)-1;
 
 	ctx = &amu_ctxs[plat_my_core_pos()];
 
 	/* Assert that group 0 counter configuration is what we expect */
-	assert(read_amcntenset0() == AMU_GROUP0_COUNTERS_MASK);
+	assert(read_amcntenset0() == AMU_GROUP0_COUNTERS_MASK &&
+	       read_amcntenset1() == AMU_GROUP1_COUNTERS_MASK);
 
 	/*
 	 * Disable group 0 counters to avoid other observers like SCP sampling
@@ -131,17 +132,15 @@ static void *amu_context_save(const void *arg)
 static void *amu_context_restore(const void *arg)
 {
 	struct amu_ctx *ctx;
-	uint64_t features;
 	int i;
 
-	features = read_id_pfr0() >> ID_PFR0_AMU_SHIFT;
-	if ((features & ID_PFR0_AMU_MASK) != 1)
+	if (amu_supported() == 0)
 		return (void *)-1;
 
 	ctx = &amu_ctxs[plat_my_core_pos()];
 
 	/* Counters were disabled in `amu_context_save()` */
-	assert(read_amcntenset0() == 0);
+	assert(read_amcntenset0() == 0 && read_amcntenset1() == 0);
 
 	/* Restore group 0 counters */
 	for (i = 0; i < AMU_GROUP0_NR_COUNTERS; i++)
diff --git a/lib/extensions/amu/aarch32/amu_helpers.S b/lib/extensions/amu/aarch32/amu_helpers.S
index 84dca04c..effb8e50 100644
--- a/lib/extensions/amu/aarch32/amu_helpers.S
+++ b/lib/extensions/amu/aarch32/amu_helpers.S
@@ -18,7 +18,7 @@
  * uint64_t amu_group0_cnt_read_internal(int idx);
  *
  * Given `idx`, read the corresponding AMU counter
- * and return it in `r0`.
+ * and return it in `r0` and `r1`.
  */
 func amu_group0_cnt_read_internal
 #if ENABLE_ASSERTIONS
@@ -52,13 +52,15 @@ endfunc amu_group0_cnt_read_internal
  * void amu_group0_cnt_write_internal(int idx, uint64_t val);
  *
  * Given `idx`, write `val` to the corresponding AMU counter.
+ * `idx` is passed in `r0` and `val` is passed in `r2` and `r3`.
+ * `r1` is used as a scratch register.
  */
 func amu_group0_cnt_write_internal
 #if ENABLE_ASSERTIONS
 	/* `idx` should be between [0, 3] */
-	mov	r2, r0
-	lsr	r2, r2, #2
-	cmp	r2, #0
+	mov	r1, r0
+	lsr	r1, r1, #2
+	cmp	r1, #0
 	ASM_ASSERT(eq)
 #endif
 
@@ -66,19 +68,19 @@ func amu_group0_cnt_write_internal
 	 * Given `idx` calculate address of stcopr16/bx lr instruction pair
 	 * in the table below.
 	 */
-	adr	r2, 1f
+	adr	r1, 1f
 	lsl	r0, r0, #3	/* each stcopr16/bx lr sequence is 8 bytes */
-	add	r2, r2, r0
-	bx	r2
+	add	r1, r1, r0
+	bx	r1
 
 1:
-	stcopr16	r0,r1, AMEVCNTR00	/* index 0 */
+	stcopr16	r2, r3, AMEVCNTR00	/* index 0 */
 	bx 		lr
-	stcopr16	r0,r1, AMEVCNTR01	/* index 1 */
+	stcopr16	r2, r3, AMEVCNTR01	/* index 1 */
 	bx 		lr
-	stcopr16	r0,r1, AMEVCNTR02	/* index 2 */
+	stcopr16	r2, r3, AMEVCNTR02	/* index 2 */
 	bx 		lr
-	stcopr16	r0,r1, AMEVCNTR03	/* index 3 */
+	stcopr16	r2, r3, AMEVCNTR03	/* index 3 */
 	bx 		lr
 endfunc amu_group0_cnt_write_internal
 
@@ -86,14 +88,14 @@ endfunc amu_group0_cnt_write_internal
  * uint64_t amu_group1_cnt_read_internal(int idx);
  *
  * Given `idx`, read the corresponding AMU counter
- * and return it in `r0`.
+ * and return it in `r0` and `r1`.
  */
 func amu_group1_cnt_read_internal
 #if ENABLE_ASSERTIONS
 	/* `idx` should be between [0, 15] */
-	mov	r2, r0
-	lsr	r2, r2, #4
-	cmp	r2, #0
+	mov	r1, r0
+	lsr	r1, r1, #4
+	cmp	r1, #0
 	ASM_ASSERT(eq)
 #endif
 
@@ -107,51 +109,53 @@ func amu_group1_cnt_read_internal
 	bx	r1
 
 1:
-	ldcopr16	r0,r1, AMEVCNTR10	/* index 0 */
-	bx	lr
-	ldcopr16	r0,r1, AMEVCNTR11	/* index 1 */
-	bx	lr
-	ldcopr16	r0,r1, AMEVCNTR12	/* index 2 */
-	bx	lr
-	ldcopr16	r0,r1, AMEVCNTR13	/* index 3 */
-	bx	lr
-	ldcopr16	r0,r1, AMEVCNTR14	/* index 4 */
-	bx	lr
-	ldcopr16	r0,r1, AMEVCNTR15	/* index 5 */
-	bx	lr
-	ldcopr16	r0,r1, AMEVCNTR16	/* index 6 */
-	bx	lr
-	ldcopr16	r0,r1, AMEVCNTR17	/* index 7 */
-	bx	lr
-	ldcopr16	r0,r1, AMEVCNTR18	/* index 8 */
-	bx	lr
-	ldcopr16	r0,r1, AMEVCNTR19	/* index 9 */
-	bx	lr
-	ldcopr16	r0,r1, AMEVCNTR1A	/* index 10 */
-	bx	lr
-	ldcopr16	r0,r1, AMEVCNTR1B	/* index 11 */
-	bx	lr
-	ldcopr16	r0,r1, AMEVCNTR1C	/* index 12 */
-	bx	lr
-	ldcopr16	r0,r1, AMEVCNTR1D	/* index 13 */
-	bx	lr
-	ldcopr16	r0,r1, AMEVCNTR1E	/* index 14 */
-	bx	lr
-	ldcopr16	r0,r1, AMEVCNTR1F	/* index 15 */
-	bx	lr
+	ldcopr16	r0, r1, AMEVCNTR10	/* index 0 */
+	bx		lr
+	ldcopr16	r0, r1, AMEVCNTR11	/* index 1 */
+	bx		lr
+	ldcopr16	r0, r1, AMEVCNTR12	/* index 2 */
+	bx		lr
+	ldcopr16	r0, r1, AMEVCNTR13	/* index 3 */
+	bx		lr
+	ldcopr16	r0, r1, AMEVCNTR14	/* index 4 */
+	bx		lr
+	ldcopr16	r0, r1, AMEVCNTR15	/* index 5 */
+	bx		lr
+	ldcopr16	r0, r1, AMEVCNTR16	/* index 6 */
+	bx		lr
+	ldcopr16	r0, r1, AMEVCNTR17	/* index 7 */
+	bx		lr
+	ldcopr16	r0, r1, AMEVCNTR18	/* index 8 */
+	bx		lr
+	ldcopr16	r0, r1, AMEVCNTR19	/* index 9 */
+	bx		lr
+	ldcopr16	r0, r1, AMEVCNTR1A	/* index 10 */
+	bx		lr
+	ldcopr16	r0, r1, AMEVCNTR1B	/* index 11 */
+	bx		lr
+	ldcopr16	r0, r1, AMEVCNTR1C	/* index 12 */
+	bx		lr
+	ldcopr16	r0, r1, AMEVCNTR1D	/* index 13 */
+	bx		lr
+	ldcopr16	r0, r1, AMEVCNTR1E	/* index 14 */
+	bx		lr
+	ldcopr16	r0, r1, AMEVCNTR1F	/* index 15 */
+	bx		lr
 endfunc amu_group1_cnt_read_internal
 
 /*
  * void amu_group1_cnt_write_internal(int idx, uint64_t val);
  *
  * Given `idx`, write `val` to the corresponding AMU counter.
+ * `idx` is passed in `r0` and `val` is passed in `r2` and `r3`.
+ * `r1` is used as a scratch register.
  */
 func amu_group1_cnt_write_internal
 #if ENABLE_ASSERTIONS
 	/* `idx` should be between [0, 15] */
-	mov	r2, r0
-	lsr	r2, r2, #4
-	cmp	r2, #0
+	mov	r1, r0
+	lsr	r1, r1, #4
+	cmp	r1, #0
 	ASM_ASSERT(eq)
 #endif
 
@@ -159,43 +163,43 @@ func amu_group1_cnt_write_internal
 	 * Given `idx` calculate address of ldcopr16/bx lr instruction pair
 	 * in the table below.
 	 */
-	adr	r2, 1f
+	adr	r1, 1f
 	lsl	r0, r0, #3	/* each stcopr16/bx lr sequence is 8 bytes */
-	add	r2, r2, r0
-	bx	r2
+	add	r1, r1, r0
+	bx	r1
 
 1:
-	stcopr16	r0,r1,	AMEVCNTR10	/* index 0 */
+	stcopr16	r2, r3,	AMEVCNTR10	/* index 0 */
 	bx		lr
-	stcopr16	r0,r1,	AMEVCNTR11	/* index 1 */
+	stcopr16	r2, r3,	AMEVCNTR11	/* index 1 */
 	bx		lr
-	stcopr16	r0,r1,	AMEVCNTR12	/* index 2 */
+	stcopr16	r2, r3,	AMEVCNTR12	/* index 2 */
 	bx		lr
-	stcopr16	r0,r1,	AMEVCNTR13	/* index 3 */
+	stcopr16	r2, r3,	AMEVCNTR13	/* index 3 */
 	bx		lr
-	stcopr16	r0,r1,	AMEVCNTR14	/* index 4 */
+	stcopr16	r2, r3,	AMEVCNTR14	/* index 4 */
 	bx		lr
-	stcopr16	r0,r1,	AMEVCNTR15	/* index 5 */
+	stcopr16	r2, r3,	AMEVCNTR15	/* index 5 */
 	bx		lr
-	stcopr16	r0,r1,	AMEVCNTR16	/* index 6 */
+	stcopr16	r2, r3,	AMEVCNTR16	/* index 6 */
 	bx		lr
-	stcopr16	r0,r1,	AMEVCNTR17	/* index 7 */
+	stcopr16	r2, r3,	AMEVCNTR17	/* index 7 */
 	bx		lr
-	stcopr16	r0,r1,	AMEVCNTR18	/* index 8 */
+	stcopr16	r2, r3,	AMEVCNTR18	/* index 8 */
 	bx		lr
-	stcopr16	r0,r1,	AMEVCNTR19	/* index 9 */
+	stcopr16	r2, r3,	AMEVCNTR19	/* index 9 */
 	bx		lr
-	stcopr16	r0,r1,	AMEVCNTR1A	/* index 10 */
+	stcopr16	r2, r3,	AMEVCNTR1A	/* index 10 */
 	bx		lr
-	stcopr16	r0,r1,	AMEVCNTR1B	/* index 11 */
+	stcopr16	r2, r3,	AMEVCNTR1B	/* index 11 */
 	bx		lr
-	stcopr16	r0,r1,	AMEVCNTR1C	/* index 12 */
+	stcopr16	r2, r3,	AMEVCNTR1C	/* index 12 */
 	bx		lr
-	stcopr16	r0,r1,	AMEVCNTR1D	/* index 13 */
+	stcopr16	r2, r3,	AMEVCNTR1D	/* index 13 */
 	bx		lr
-	stcopr16	r0,r1,	AMEVCNTR1E	/* index 14 */
+	stcopr16	r2, r3,	AMEVCNTR1E	/* index 14 */
 	bx		lr
-	stcopr16	r0,r1,	AMEVCNTR1F	/* index 15 */
+	stcopr16	r2, r3,	AMEVCNTR1F	/* index 15 */
 	bx		lr
 endfunc amu_group1_cnt_write_internal
 
@@ -230,36 +234,36 @@ func amu_group1_set_evtype_internal
 	bx	r2
 
 1:
-	stcopr	r0,	AMEVTYPER10 /* index 0 */
+	stcopr	r1,	AMEVTYPER10 /* index 0 */
 	bx	lr
-	stcopr	r0,	AMEVTYPER11 /* index 1 */
+	stcopr	r1,	AMEVTYPER11 /* index 1 */
 	bx	lr
-	stcopr	r0,	AMEVTYPER12 /* index 2 */
+	stcopr	r1,	AMEVTYPER12 /* index 2 */
 	bx	lr
-	stcopr	r0,	AMEVTYPER13 /* index 3 */
+	stcopr	r1,	AMEVTYPER13 /* index 3 */
 	bx	lr
-	stcopr	r0,	AMEVTYPER14 /* index 4 */
+	stcopr	r1,	AMEVTYPER14 /* index 4 */
 	bx	lr
-	stcopr	r0,	AMEVTYPER15 /* index 5 */
+	stcopr	r1,	AMEVTYPER15 /* index 5 */
 	bx	lr
-	stcopr	r0,	AMEVTYPER16 /* index 6 */
+	stcopr	r1,	AMEVTYPER16 /* index 6 */
 	bx	lr
-	stcopr	r0,	AMEVTYPER17 /* index 7 */
+	stcopr	r1,	AMEVTYPER17 /* index 7 */
 	bx	lr
-	stcopr	r0,	AMEVTYPER18 /* index 8 */
+	stcopr	r1,	AMEVTYPER18 /* index 8 */
 	bx	lr
-	stcopr	r0,	AMEVTYPER19 /* index 9 */
+	stcopr	r1,	AMEVTYPER19 /* index 9 */
 	bx	lr
-	stcopr	r0,	AMEVTYPER1A /* index 10 */
+	stcopr	r1,	AMEVTYPER1A /* index 10 */
 	bx	lr
-	stcopr	r0,	AMEVTYPER1B /* index 11 */
+	stcopr	r1,	AMEVTYPER1B /* index 11 */
 	bx	lr
-	stcopr	r0,	AMEVTYPER1C /* index 12 */
+	stcopr	r1,	AMEVTYPER1C /* index 12 */
 	bx	lr
-	stcopr	r0,	AMEVTYPER1D /* index 13 */
+	stcopr	r1,	AMEVTYPER1D /* index 13 */
 	bx	lr
-	stcopr	r0,	AMEVTYPER1E /* index 14 */
+	stcopr	r1,	AMEVTYPER1E /* index 14 */
 	bx	lr
-	stcopr	r0,	AMEVTYPER1F /* index 15 */
+	stcopr	r1,	AMEVTYPER1F /* index 15 */
 	bx	lr
 endfunc amu_group1_set_evtype_internal
diff --git a/lib/extensions/amu/aarch64/amu.c b/lib/extensions/amu/aarch64/amu.c
index 7d39f35c..5d556e5d 100644
--- a/lib/extensions/amu/aarch64/amu.c
+++ b/lib/extensions/amu/aarch64/amu.c
@@ -37,7 +37,7 @@ void amu_enable(int el2_unused)
 {
 	uint64_t v;
 
-	if (!amu_supported())
+	if (amu_supported() == 0)
 		return;
 
 	if (el2_unused) {
@@ -67,7 +67,7 @@ void amu_enable(int el2_unused)
 /* Read the group 0 counter identified by the given `idx`. */
 uint64_t amu_group0_cnt_read(int idx)
 {
-	assert(amu_supported());
+	assert(amu_supported() != 0);
 	assert(idx >= 0 && idx < AMU_GROUP0_NR_COUNTERS);
 
 	return amu_group0_cnt_read_internal(idx);
@@ -76,7 +76,7 @@ uint64_t amu_group0_cnt_read(int idx)
 /* Write the group 0 counter identified by the given `idx` with `val`. */
 void amu_group0_cnt_write(int idx, uint64_t val)
 {
-	assert(amu_supported());
+	assert(amu_supported() != 0);
 	assert(idx >= 0 && idx < AMU_GROUP0_NR_COUNTERS);
 
 	amu_group0_cnt_write_internal(idx, val);
@@ -86,7 +86,7 @@ void amu_group0_cnt_write(int idx, uint64_t val)
 /* Read the group 1 counter identified by the given `idx`. */
 uint64_t amu_group1_cnt_read(int idx)
 {
-	assert(amu_supported());
+	assert(amu_supported() != 0);
 	assert(idx >= 0 && idx < AMU_GROUP1_NR_COUNTERS);
 
 	return amu_group1_cnt_read_internal(idx);
@@ -95,7 +95,7 @@ uint64_t amu_group1_cnt_read(int idx)
 /* Write the group 1 counter identified by the given `idx` with `val`. */
 void amu_group1_cnt_write(int idx, uint64_t val)
 {
-	assert(amu_supported());
+	assert(amu_supported() != 0);
 	assert(idx >= 0 && idx < AMU_GROUP1_NR_COUNTERS);
 
 	amu_group1_cnt_write_internal(idx, val);
@@ -108,7 +108,7 @@ void amu_group1_cnt_write(int idx, uint64_t val)
  */
 void amu_group1_set_evtype(int idx, unsigned int val)
 {
-	assert(amu_supported());
+	assert(amu_supported() != 0);
 	assert (idx >= 0 && idx < AMU_GROUP1_NR_COUNTERS);
 
 	amu_group1_set_evtype_internal(idx, val);
@@ -120,7 +120,7 @@ static void *amu_context_save(const void *arg)
 	struct amu_ctx *ctx = &amu_ctxs[plat_my_core_pos()];
 	int i;
 
-	if (!amu_supported())
+	if (amu_supported() == 0)
 		return (void *)-1;
 
 	/* Assert that group 0/1 counter configuration is what we expect */
@@ -154,7 +154,7 @@ static void *amu_context_restore(const void *arg)
 	struct amu_ctx *ctx = &amu_ctxs[plat_my_core_pos()];
 	int i;
 
-	if (!amu_supported())
+	if (amu_supported() == 0)
 		return (void *)-1;
 
 	/* Counters were disabled in `amu_context_save()` */
diff --git a/lib/extensions/spe/spe.c b/lib/extensions/spe/spe.c
index 3b297f21..a9bed490 100644
--- a/lib/extensions/spe/spe.c
+++ b/lib/extensions/spe/spe.c
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2017, ARM Limited and Contributors. All rights reserved.
+ * Copyright (c) 2017-2018, ARM Limited and Contributors. All rights reserved.
  *
  * SPDX-License-Identifier: BSD-3-Clause
  */
@@ -14,71 +14,72 @@
  */
 #define psb_csync()	asm volatile("hint #17")
 
-void spe_enable(int el2_unused)
+int spe_supported(void)
 {
 	uint64_t features;
 
 	features = read_id_aa64dfr0_el1() >> ID_AA64DFR0_PMS_SHIFT;
-	if ((features & ID_AA64DFR0_PMS_MASK) == 1) {
-		uint64_t v;
+	return (features & ID_AA64DFR0_PMS_MASK) == 1;
+}
 
-		if (el2_unused) {
-			/*
-			 * MDCR_EL2.TPMS (ARM v8.2): Do not trap statistical
-			 * profiling controls to EL2.
-			 *
-			 * MDCR_EL2.E2PB (ARM v8.2): SPE enabled in Non-secure
-			 * state. Accesses to profiling buffer controls at
-			 * Non-secure EL1 are not trapped to EL2.
-			 */
-			v = read_mdcr_el2();
-			v &= ~MDCR_EL2_TPMS;
-			v |= MDCR_EL2_E2PB(MDCR_EL2_E2PB_EL1);
-			write_mdcr_el2(v);
-		}
+void spe_enable(int el2_unused)
+{
+	uint64_t v;
 
+	if (spe_supported() == 0)
+		return;
+
+	if (el2_unused) {
 		/*
-		 * MDCR_EL2.NSPB (ARM v8.2): SPE enabled in Non-secure state
-		 * and disabled in secure state. Accesses to SPE registers at
-		 * S-EL1 generate trap exceptions to EL3.
+		 * MDCR_EL2.TPMS (ARM v8.2): Do not trap statistical
+		 * profiling controls to EL2.
+		 *
+		 * MDCR_EL2.E2PB (ARM v8.2): SPE enabled in Non-secure
+		 * state. Accesses to profiling buffer controls at
+		 * Non-secure EL1 are not trapped to EL2.
 		 */
-		v = read_mdcr_el3();
-		v |= MDCR_NSPB(MDCR_NSPB_EL1);
-		write_mdcr_el3(v);
+		v = read_mdcr_el2();
+		v &= ~MDCR_EL2_TPMS;
+		v |= MDCR_EL2_E2PB(MDCR_EL2_E2PB_EL1);
+		write_mdcr_el2(v);
 	}
+
+	/*
+	 * MDCR_EL2.NSPB (ARM v8.2): SPE enabled in Non-secure state
+	 * and disabled in secure state. Accesses to SPE registers at
+	 * S-EL1 generate trap exceptions to EL3.
+	 */
+	v = read_mdcr_el3();
+	v |= MDCR_NSPB(MDCR_NSPB_EL1);
+	write_mdcr_el3(v);
 }
 
 void spe_disable(void)
 {
-	uint64_t features;
+	uint64_t v;
 
-	features = read_id_aa64dfr0_el1() >> ID_AA64DFR0_PMS_SHIFT;
-	if ((features & ID_AA64DFR0_PMS_MASK) == 1) {
-		uint64_t v;
+	if (spe_supported() == 0)
+		return;
 
-		/* Drain buffered data */
-		psb_csync();
-		dsbnsh();
+	/* Drain buffered data */
+	psb_csync();
+	dsbnsh();
 
-		/* Disable profiling buffer */
-		v = read_pmblimitr_el1();
-		v &= ~(1ULL << 0);
-		write_pmblimitr_el1(v);
-		isb();
-	}
+	/* Disable profiling buffer */
+	v = read_pmblimitr_el1();
+	v &= ~(1ULL << 0);
+	write_pmblimitr_el1(v);
+	isb();
 }
 
 static void *spe_drain_buffers_hook(const void *arg)
 {
-	uint64_t features;
-
-	features = read_id_aa64dfr0_el1() >> ID_AA64DFR0_PMS_SHIFT;
-	if ((features & ID_AA64DFR0_PMS_MASK) == 1) {
-		/* Drain buffered data */
-		psb_csync();
-		dsbnsh();
-	}
+	if (spe_supported() == 0)
+		return (void *)-1;
 
+	/* Drain buffered data */
+	psb_csync();
+	dsbnsh();
 	return 0;
 }
 
diff --git a/lib/extensions/sve/sve.c b/lib/extensions/sve/sve.c
index 14e51bd8..64424878 100644
--- a/lib/extensions/sve/sve.c
+++ b/lib/extensions/sve/sve.c
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2017, ARM Limited and Contributors. All rights reserved.
+ * Copyright (c) 2017-2018, ARM Limited and Contributors. All rights reserved.
  *
  * SPDX-License-Identifier: BSD-3-Clause
  */
@@ -9,117 +9,120 @@
 #include <pubsub.h>
 #include <sve.h>
 
-static void *disable_sve_hook(const void *arg)
+int sve_supported(void)
 {
 	uint64_t features;
 
 	features = read_id_aa64pfr0_el1() >> ID_AA64PFR0_SVE_SHIFT;
-	if ((features & ID_AA64PFR0_SVE_MASK) == 1) {
-		uint64_t cptr;
-
-		/*
-		 * Disable SVE, SIMD and FP access for the Secure world.
-		 * As the SIMD/FP registers are part of the SVE Z-registers, any
-		 * use of SIMD/FP functionality will corrupt the SVE registers.
-		 * Therefore it is necessary to prevent use of SIMD/FP support
-		 * in the Secure world as well as SVE functionality.
-		 */
-		cptr = read_cptr_el3();
-		cptr = (cptr | TFP_BIT) & ~(CPTR_EZ_BIT);
-		write_cptr_el3(cptr);
+	return (features & ID_AA64PFR0_SVE_MASK) == 1;
+}
 
-		/*
-		 * No explicit ISB required here as ERET to switch to Secure
-		 * world covers it
-		 */
-	}
+static void *disable_sve_hook(const void *arg)
+{
+	uint64_t cptr;
+
+	if (sve_supported() == 0)
+		return (void *)-1;
+
+	/*
+	 * Disable SVE, SIMD and FP access for the Secure world.
+	 * As the SIMD/FP registers are part of the SVE Z-registers, any
+	 * use of SIMD/FP functionality will corrupt the SVE registers.
+	 * Therefore it is necessary to prevent use of SIMD/FP support
+	 * in the Secure world as well as SVE functionality.
+	 */
+	cptr = read_cptr_el3();
+	cptr = (cptr | TFP_BIT) & ~(CPTR_EZ_BIT);
+	write_cptr_el3(cptr);
+
+	/*
+	 * No explicit ISB required here as ERET to switch to Secure
+	 * world covers it
+	 */
 	return 0;
 }
 
 static void *enable_sve_hook(const void *arg)
 {
-	uint64_t features;
-
-	features = read_id_aa64pfr0_el1() >> ID_AA64PFR0_SVE_SHIFT;
-	if ((features & ID_AA64PFR0_SVE_MASK) == 1) {
-		uint64_t cptr;
-
-		/*
-		 * Enable SVE, SIMD and FP access for the Non-secure world.
-		 */
-		cptr = read_cptr_el3();
-		cptr = (cptr | CPTR_EZ_BIT) & ~(TFP_BIT);
-		write_cptr_el3(cptr);
-
-		/*
-		 * No explicit ISB required here as ERET to switch to Non-secure
-		 * world covers it
-		 */
-	}
+	uint64_t cptr;
+
+	if (sve_supported() == 0)
+		return (void *)-1;
+
+	/*
+	 * Enable SVE, SIMD and FP access for the Non-secure world.
+	 */
+	cptr = read_cptr_el3();
+	cptr = (cptr | CPTR_EZ_BIT) & ~(TFP_BIT);
+	write_cptr_el3(cptr);
+
+	/*
+	 * No explicit ISB required here as ERET to switch to Non-secure
+	 * world covers it
+	 */
 	return 0;
 }
 
 void sve_enable(int el2_unused)
 {
-	uint64_t features;
+	uint64_t cptr;
+
+	if (sve_supported() == 0)
+		return;
 
-	features = read_id_aa64pfr0_el1() >> ID_AA64PFR0_SVE_SHIFT;
-	if ((features & ID_AA64PFR0_SVE_MASK) == 1) {
-		uint64_t cptr;
 #if CTX_INCLUDE_FPREGS
-		/*
-		 * CTX_INCLUDE_FPREGS is not supported on SVE enabled systems.
-		 */
-		assert(0);
+	/*
+	 * CTX_INCLUDE_FPREGS is not supported on SVE enabled systems.
+	 */
+	assert(0);
 #endif
+	/*
+	 * Update CPTR_EL3 to enable access to SVE functionality for the
+	 * Non-secure world.
+	 * NOTE - assumed that CPTR_EL3.TFP is set to allow access to
+	 * the SIMD, floating-point and SVE support.
+	 *
+	 * CPTR_EL3.EZ: Set to 1 to enable access to SVE  functionality
+	 *  in the Non-secure world.
+	 */
+	cptr = read_cptr_el3();
+	cptr |= CPTR_EZ_BIT;
+	write_cptr_el3(cptr);
+
+	/*
+	 * Need explicit ISB here to guarantee that update to ZCR_ELx
+	 * and CPTR_EL2.TZ do not result in trap to EL3.
+	 */
+	isb();
+
+	/*
+	 * Ensure lower ELs have access to full vector length.
+	 */
+	write_zcr_el3(ZCR_EL3_LEN_MASK);
+
+	if (el2_unused) {
 		/*
-		 * Update CPTR_EL3 to enable access to SVE functionality for the
-		 * Non-secure world.
-		 * NOTE - assumed that CPTR_EL3.TFP is set to allow access to
-		 * the SIMD, floating-point and SVE support.
+		 * Update CPTR_EL2 to enable access to SVE functionality
+		 * for Non-secure world, EL2 and Non-secure EL1 and EL0.
+		 * NOTE - assumed that CPTR_EL2.TFP is set to allow
+		 * access to the SIMD, floating-point and SVE support.
 		 *
-		 * CPTR_EL3.EZ: Set to 1 to enable access to SVE  functionality
-		 *  in the Non-secure world.
+		 * CPTR_EL2.TZ: Set to 0 to enable access to SVE support
+		 *  for EL2 and Non-secure EL1 and EL0.
 		 */
-		cptr = read_cptr_el3();
-		cptr |= CPTR_EZ_BIT;
-		write_cptr_el3(cptr);
-
-		/*
-		 * Need explicit ISB here to guarantee that update to ZCR_ELx
-		 * and CPTR_EL2.TZ do not result in trap to EL3.
-		 */
-		isb();
+		cptr = read_cptr_el2();
+		cptr &= ~(CPTR_EL2_TZ_BIT);
+		write_cptr_el2(cptr);
 
 		/*
 		 * Ensure lower ELs have access to full vector length.
 		 */
-		write_zcr_el3(ZCR_EL3_LEN_MASK);
-
-		if (el2_unused) {
-			/*
-			 * Update CPTR_EL2 to enable access to SVE functionality
-			 * for Non-secure world, EL2 and Non-secure EL1 and EL0.
-			 * NOTE - assumed that CPTR_EL2.TFP is set to allow
-			 * access to the SIMD, floating-point and SVE support.
-			 *
-			 * CPTR_EL2.TZ: Set to 0 to enable access to SVE support
-			 *  for EL2 and Non-secure EL1 and EL0.
-			 */
-			cptr = read_cptr_el2();
-			cptr &= ~(CPTR_EL2_TZ_BIT);
-			write_cptr_el2(cptr);
-
-			/*
-			 * Ensure lower ELs have access to full vector length.
-			 */
-			write_zcr_el2(ZCR_EL2_LEN_MASK);
-		}
-		/*
-		 * No explicit ISB required here as ERET to switch to
-		 * Non-secure world covers it.
-		 */
+		write_zcr_el2(ZCR_EL2_LEN_MASK);
 	}
+	/*
+	 * No explicit ISB required here as ERET to switch to
+	 * Non-secure world covers it.
+	 */
 }
 
 SUBSCRIBE_TO_EVENT(cm_exited_normal_world, disable_sve_hook);