NEON fixes.

(Doesn't include the patch changes yet. They will be next.)


git-svn-id: http://src.chromium.org/svn/trunk/deps/third_party/openssl@234628 4ff67af0-8c30-449e-8e8b-ad334ec8d88c

5 files changed, 288 insertions, 285 deletions

diff --git a/openssl/crypto/chacha/chacha_enc.c b/openssl/crypto/chacha/chacha_enc.c
index e4b648f..2d6b076 100644
--- a/openssl/crypto/chacha/chacha_enc.c
+++ b/openssl/crypto/chacha/chacha_enc.c
@@ -135,7 +135,9 @@ void CRYPTO_chacha_20(unsigned char *out,
 	size_t todo, i;
 
 #if __arm__
-	if (CRYPTO_is_NEON_capable())
+	if (CRYPTO_is_NEON_capable() &&
+	    ((intptr_t)in & 15) == 0 &&
+	    ((intptr_t)out & 15) == 0)
 		{
 		CRYPTO_chacha_20_neon(out, in, in_len, key, nonce, counter);
 		return;
diff --git a/openssl/crypto/chacha/chacha_vec.c b/openssl/crypto/chacha/chacha_vec.c
index 1226c39..274345d 100644
--- a/openssl/crypto/chacha/chacha_vec.c
+++ b/openssl/crypto/chacha/chacha_vec.c
@@ -289,7 +289,7 @@ void CRYPTO_chacha_20(
 		op[9]  = REVW_BE(REVW_BE(ip[9])  ^ (x9  + kp[5]));
 		op[10] = REVW_BE(REVW_BE(ip[10]) ^ (x10 + kp[6]));
 		op[11] = REVW_BE(REVW_BE(ip[11]) ^ (x11 + kp[7]));
-		op[12] = REVW_BE(REVW_BE(ip[12]) ^ (x12 + BPI*iters+(BPI-1)));
+		op[12] = REVW_BE(REVW_BE(ip[12]) ^ (x12 + counter+BPI*iters+(BPI-1)));
 		op[13] = REVW_BE(REVW_BE(ip[13]) ^ (x13));
 		op[14] = REVW_BE(REVW_BE(ip[14]) ^ (x14 + np[0]));
 		op[15] = REVW_BE(REVW_BE(ip[15]) ^ (x15 + np[1]));
diff --git a/openssl/crypto/chacha/chacha_vec_arm.S b/openssl/crypto/chacha/chacha_vec_arm.S
index 35dc74d..f2a6961 100644
--- a/openssl/crypto/chacha/chacha_vec_arm.S
+++ b/openssl/crypto/chacha/chacha_vec_arm.S
@@ -5,7 +5,7 @@
 #
 # This file was generated by:
 #
-#     /opt/gcc-linaro-arm-linux-gnueabihf-4.7-2012.10-20121022_linux/bin/arm-linux-gnueabihf-gcc -O3 -mcpu=cortex-a8 -mfpu=neon -S chacha_vec.c  -I ../../include -fpic
+#     /opt/gcc-linaro-arm-linux-gnueabihf-4.7-2012.10-20121022_linux/bin/arm-linux-gnueabihf-gcc -O3 -mcpu=cortex-a8 -mfpu=neon -S chacha_vec.c -I ../../include -fpic -o chacha_vec_arm.S
 #
 # And then EABI attribute 28 was set to zero to allow linking with soft-float
 # code.
@@ -33,130 +33,133 @@
 	.thumb_func
 	.type	CRYPTO_chacha_20_neon, %function
 CRYPTO_chacha_20_neon:
-	@ args = 8, pretend = 0, frame = 296
+	@ args = 8, pretend = 0, frame = 304
 	@ frame_needed = 1, uses_anonymous_args = 0
 	@ link register save eliminated.
 	push	{r4, r5, r6, r7, r8, r9, sl, fp}
 	fstmfdd	sp!, {d8, d9, d10, d11, d12, d13, d14, d15}
-	sub	sp, sp, #296
+	sub	sp, sp, #304
 	add	r7, sp, #0
 	movw	ip, #43691
 	movt	ip, 43690
-	str	r2, [r7, #192]
+	str	r2, [r7, #196]
 	sub	sp, sp, #96
-	ldr	r4, [r7, #192]
-	ldr	r6, [r7, #392]
+	ldr	r4, [r7, #196]
+	ldr	r6, [r7, #400]
 	ldr	r2, .L38+16
 	umull	r4, ip, ip, r4
 	ldr	r6, [r6, #0]
-	ldr	r8, [r7, #392]
+	ldr	r8, [r7, #400]
 .LPIC24:
 	add	r2, pc
 	add	r4, sp, #15
-	str	r3, [r7, #236]
-	str	r6, [r7, #172]
+	str	r3, [r7, #244]
+	str	r6, [r7, #176]
 	bic	r4, r4, #15
-	str	r0, [r7, #184]
-	str	r4, [r7, #196]
+	str	r0, [r7, #188]
+	str	r4, [r7, #200]
 	lsrs	ip, ip, #7
-	str	r1, [r7, #180]
+	str	r1, [r7, #184]
 	ldmia	r2, {r0, r1, r2, r3}
 	ldr	r4, [r8, #4]
-	ldr	r5, [r7, #236]
+	ldr	r5, [r7, #244]
 	vld1.64	{d24-d25}, [r5:64]
 	vldr	d26, [r5, #16]
 	vldr	d27, [r5, #24]
-	ldr	fp, [r7, #196]
-	ldr	r8, [r7, #396]
-	ldr	r5, [r7, #172]
-	add	r6, fp, #64
-	str	r4, [r7, #292]
+	ldr	r9, [r7, #200]
+	ldr	r8, [r7, #404]
+	ldr	r5, [r7, #176]
+	add	r6, r9, #64
+	str	r4, [r7, #300]
 	mov	r4, #0
-	str	r8, [r7, #280]
-	str	r5, [r7, #288]
-	str	r4, [r7, #284]
+	str	r8, [r7, #288]
+	str	r5, [r7, #296]
+	str	r4, [r7, #292]
 	stmia	r6, {r0, r1, r2, r3}
-	vldr	d22, [fp, #64]
-	vldr	d23, [fp, #72]
-	vldr	d20, [r7, #280]
-	vldr	d21, [r7, #288]
-	str	ip, [r7, #188]
+	vldr	d22, [r9, #64]
+	vldr	d23, [r9, #72]
+	vldr	d20, [r7, #288]
+	vldr	d21, [r7, #296]
+	str	ip, [r7, #192]
 	beq	.L20
 	lsl	r6, ip, #1
-	ldr	r1, [fp, #68]
+	ldr	r1, [r9, #68]
 	add	r3, r6, ip
-	str	r6, [r7, #176]
-	ldr	r2, [fp, #72]
+	str	r6, [r7, #180]
+	ldr	r2, [r9, #72]
 	add	r8, r8, #2
-	ldr	r5, [fp, #76]
+	ldr	r5, [r9, #76]
 	vldr	d18, .L38
 	vldr	d19, .L38+8
-	str	r4, [r7, #232]
-	ldr	r6, [r7, #180]
-	ldr	r4, [r7, #184]
-	str	r0, [r7, #220]
-	str	r1, [r7, #216]
-	str	r8, [r7, #200]
-	str	r2, [r7, #212]
-	str	r3, [r7, #204]
-	str	r5, [r7, #208]
-	str	r6, [r7, #244]
 	str	r4, [r7, #240]
+	ldr	r6, [r7, #184]
+	ldr	r4, [r7, #188]
+	str	r0, [r7, #224]
+	str	r1, [r7, #220]
+	str	r8, [r7, #208]
+	str	r2, [r7, #216]
+	str	r3, [r7, #204]
+	str	r5, [r7, #212]
+	str	r6, [r7, #252]
+	str	r4, [r7, #248]
 .L4:
-	ldr	r6, [r7, #236]
+	ldr	r2, [r7, #244]
+	add	r9, r7, #216
+	ldr	r3, [r7, #244]
 	vadd.i32	q8, q10, q9
-	ldr	r5, [r7, #236]
+	ldr	r6, [r7, #208]
 	vmov	q15, q13  @ v4si
-	ldr	r8, [r7, #232]
+	ldr	r5, [r7, #240]
 	vmov	q3, q12  @ v4si
-	ldr	r6, [r6, #4]
+	ldr	r4, [r7, #244]
 	vmov	q2, q11  @ v4si
-	ldr	fp, [r7, #200]
+	adds	r5, r5, r6
+	ldr	r2, [r2, #8]
+	ldr	r6, [r7, #400]
 	vmov	q5, q10  @ v4si
-	ldr	r4, [r7, #236]
+	ldr	r3, [r3, #12]
 	vmov	q1, q13  @ v4si
-	add	ip, r8, fp
-	ldr	r5, [r5, #0]
-	ldr	r0, [r7, #236]
-	add	r8, r7, #208
-	ldr	r1, [r7, #236]
+	ldr	r0, [r7, #244]
 	vmov	q0, q12  @ v4si
-	str	r6, [r7, #260]
+	ldr	r1, [r7, #244]
 	vmov	q4, q11  @ v4si
-	ldr	r6, [r7, #392]
-	ldmia	r8, {r8, r9, sl, fp}
-	ldr	r0, [r0, #8]
-	ldr	r1, [r1, #12]
-	str	r5, [r7, #224]
+	ldmia	r9, {r9, sl, fp}
+	str	r5, [r7, #228]
 	ldr	r5, [r4, #24]
-	ldr	r3, [r4, #28]
+	ldr	r0, [r0, #0]
+	ldr	r1, [r1, #4]
+	str	r2, [r7, #264]
+	str	r3, [r7, #236]
 	ldr	r2, [r6, #4]
-	str	r0, [r7, #256]
-	str	r1, [r7, #228]
-	str	r5, [r7, #272]
+	ldr	r3, [r4, #28]
+	str	r5, [r7, #280]
 	ldr	r5, [r6, #0]
 	movs	r6, #0
+	ldr	ip, [r7, #228]
+	ldr	r8, [r7, #212]
+	str	r0, [r7, #232]
+	str	r1, [r7, #268]
 	ldr	r0, [r4, #16]
 	ldr	r1, [r4, #20]
 	movs	r4, #10
-	str	r2, [r7, #20]
-	str	r3, [r7, #276]
-	str	r9, [r7, #268]
+	str	r2, [r7, #24]
+	str	r3, [r7, #284]
+	str	r4, [r7, #256]
+	ldr	r2, [r7, #264]
+	str	r9, [r7, #276]
 	mov	r9, r6
-	str	r4, [r7, #248]
-	ldr	r2, [r7, #256]
-	ldr	r3, [r7, #228]
-	str	r8, [r7, #252]
+	ldr	r6, [r7, #280]
+	str	r8, [r7, #260]
 	mov	r8, sl
-	ldr	r6, [r7, #272]
+	str	r1, [r7, #272]
 	mov	sl, ip
-	str	r1, [r7, #264]
-	ldr	ip, [r7, #20]
-	str	r6, [r7, #256]
+	str	r6, [r7, #264]
 	mov	r6, r5
-	ldr	r1, [r7, #260]
+	ldr	r3, [r7, #236]
 	mov	r5, r0
-	ldr	r0, [r7, #224]
+	ldr	ip, [r7, #24]
+	ldr	r1, [r7, #268]
+	ldr	r0, [r7, #232]
 	b	.L39
 .L40:
 	.align	3
@@ -171,445 +174,445 @@ CRYPTO_chacha_20_neon:
 	vadd.i32	q4, q4, q0
 	add	r8, r8, r1
 	vadd.i32	q2, q2, q3
-	str	r8, [r7, #260]
+	str	r8, [r7, #268]
 	veor	q5, q5, q4
-	ldr	r8, [r7, #268]
+	ldr	r8, [r7, #276]
 	veor	q8, q8, q2
 	add	fp, fp, r0
-	str	fp, [r7, #272]
+	str	fp, [r7, #280]
 	add	r8, r8, r2
 	vrev32.16	q5, q5
-	str	r8, [r7, #268]
+	str	r8, [r7, #276]
 	vrev32.16	q8, q8
 	vadd.i32	q1, q1, q5
 	vadd.i32	q15, q15, q8
-	ldr	r8, [r7, #272]
+	ldr	r8, [r7, #280]
 	veor	q0, q1, q0
-	ldr	r4, [r7, #252]
+	ldr	r4, [r7, #260]
 	veor	q3, q15, q3
 	eor	sl, sl, r8
-	ldr	r8, [r7, #268]
+	ldr	r8, [r7, #276]
 	add	fp, r4, r3
 	vshl.i32	q7, q0, #12
-	ldr	r4, [r7, #260]
+	ldr	r4, [r7, #268]
 	vshl.i32	q6, q3, #12
 	eor	r6, r6, r8
 	eor	r9, r9, r4
-	ldr	r4, [r7, #264]
+	ldr	r4, [r7, #272]
 	vsri.32	q7, q0, #20
 	ror	r8, r6, #16
-	ldr	r6, [r7, #256]
+	ldr	r6, [r7, #264]
 	eor	ip, ip, fp
 	vsri.32	q6, q3, #20
 	ror	sl, sl, #16
 	ror	r9, r9, #16
 	add	r5, r5, sl
 	vadd.i32	q4, q4, q7
-	str	r5, [r7, #228]
+	str	r5, [r7, #236]
 	vadd.i32	q2, q2, q6
 	add	r5, r4, r9
 	add	r4, r6, r8
-	ldr	r6, [r7, #276]
+	ldr	r6, [r7, #284]
 	ror	ip, ip, #16
 	veor	q5, q4, q5
 	veor	q8, q2, q8
 	add	r6, r6, ip
-	str	r6, [r7, #256]
+	str	r6, [r7, #264]
 	eors	r1, r1, r5
-	ldr	r6, [r7, #228]
+	ldr	r6, [r7, #236]
 	vshl.i32	q3, q5, #8
 	vshl.i32	q14, q8, #8
 	eors	r2, r2, r4
 	eors	r0, r0, r6
-	ldr	r6, [r7, #256]
+	ldr	r6, [r7, #264]
 	vsri.32	q3, q5, #24
 	ror	r1, r1, #20
 	eors	r3, r3, r6
-	ldr	r6, [r7, #272]
+	ldr	r6, [r7, #280]
 	ror	r0, r0, #20
 	vsri.32	q14, q8, #24
 	adds	r6, r0, r6
-	str	r6, [r7, #276]
-	ldr	r6, [r7, #260]
+	str	r6, [r7, #284]
+	ldr	r6, [r7, #268]
 	vadd.i32	q1, q1, q3
 	vadd.i32	q15, q15, q14
 	ror	r2, r2, #20
 	adds	r6, r1, r6
-	str	r6, [r7, #252]
-	ldr	r6, [r7, #268]
+	str	r6, [r7, #260]
+	ldr	r6, [r7, #276]
 	veor	q6, q15, q6
 	veor	q7, q1, q7
 	ror	r3, r3, #20
 	adds	r6, r2, r6
-	str	r6, [r7, #272]
-	ldr	r6, [r7, #276]
+	str	r6, [r7, #280]
+	ldr	r6, [r7, #284]
 	vshl.i32	q0, q6, #7
 	vshl.i32	q5, q7, #7
 	add	fp, r3, fp
 	eor	sl, r6, sl
-	ldr	r6, [r7, #252]
+	ldr	r6, [r7, #260]
 	eor	ip, fp, ip
 	vsri.32	q0, q6, #25
 	eor	r9, r6, r9
-	ldr	r6, [r7, #272]
+	ldr	r6, [r7, #280]
 	ror	sl, sl, #24
 	vsri.32	q5, q7, #25
 	eor	r8, r6, r8
-	ldr	r6, [r7, #228]
+	ldr	r6, [r7, #236]
 	ror	r9, r9, #24
 	ror	ip, ip, #24
 	add	r6, sl, r6
-	str	r6, [r7, #268]
-	ldr	r6, [r7, #256]
+	str	r6, [r7, #276]
+	ldr	r6, [r7, #264]
 	add	r5, r9, r5
-	str	r5, [r7, #264]
+	str	r5, [r7, #272]
 	vext.32	q5, q5, q5, #1
 	add	r5, ip, r6
-	ldr	r6, [r7, #268]
+	ldr	r6, [r7, #276]
 	vext.32	q0, q0, q0, #1
 	vadd.i32	q4, q4, q5
 	eors	r0, r0, r6
-	ldr	r6, [r7, #264]
+	ldr	r6, [r7, #272]
 	vadd.i32	q2, q2, q0
 	vext.32	q3, q3, q3, #3
 	ror	r8, r8, #24
 	eors	r1, r1, r6
 	vext.32	q14, q14, q14, #3
 	add	r4, r8, r4
-	ldr	r6, [r7, #276]
+	ldr	r6, [r7, #284]
 	veor	q3, q4, q3
 	veor	q14, q2, q14
 	eors	r2, r2, r4
 	ror	r1, r1, #25
 	vext.32	q1, q1, q1, #2
 	adds	r6, r1, r6
-	str	r6, [r7, #276]
+	str	r6, [r7, #284]
 	vext.32	q15, q15, q15, #2
-	ldr	r6, [r7, #252]
+	ldr	r6, [r7, #260]
 	eors	r3, r3, r5
 	ror	r2, r2, #25
 	vrev32.16	q8, q14
 	adds	r6, r2, r6
 	vrev32.16	q3, q3
-	str	r6, [r7, #260]
+	str	r6, [r7, #268]
 	vadd.i32	q1, q1, q3
-	ldr	r6, [r7, #272]
+	ldr	r6, [r7, #280]
 	vadd.i32	q15, q15, q8
 	ror	r3, r3, #25
 	veor	q5, q1, q5
 	adds	r6, r3, r6
 	veor	q0, q15, q0
-	str	r6, [r7, #256]
-	ldr	r6, [r7, #260]
+	str	r6, [r7, #264]
+	ldr	r6, [r7, #268]
 	ror	r0, r0, #25
 	add	fp, r0, fp
 	vshl.i32	q6, q5, #12
 	eor	sl, r6, sl
-	ldr	r6, [r7, #276]
+	ldr	r6, [r7, #284]
 	vshl.i32	q14, q0, #12
 	eor	r8, fp, r8
 	eor	ip, r6, ip
-	ldr	r6, [r7, #256]
+	ldr	r6, [r7, #264]
 	vsri.32	q6, q5, #20
 	ror	sl, sl, #16
 	eor	r9, r6, r9
 	ror	r6, r8, #16
 	vsri.32	q14, q0, #20
-	ldr	r8, [r7, #264]
+	ldr	r8, [r7, #272]
 	ror	ip, ip, #16
 	add	r5, sl, r5
 	add	r8, r6, r8
 	add	r4, ip, r4
-	str	r4, [r7, #228]
+	str	r4, [r7, #236]
 	eor	r0, r8, r0
-	str	r5, [r7, #272]
+	str	r5, [r7, #280]
 	vadd.i32	q4, q4, q6
-	ldr	r5, [r7, #228]
+	ldr	r5, [r7, #236]
 	vadd.i32	q2, q2, q14
-	ldr	r4, [r7, #268]
+	ldr	r4, [r7, #276]
 	ror	r0, r0, #20
 	veor	q3, q4, q3
 	eors	r1, r1, r5
 	veor	q0, q2, q8
-	str	r8, [r7, #264]
-	str	r0, [r7, #20]
+	str	r8, [r7, #272]
+	str	r0, [r7, #24]
 	add	fp, r0, fp
-	ldr	r8, [r7, #272]
+	ldr	r8, [r7, #280]
 	ror	r9, r9, #16
-	ldr	r0, [r7, #276]
+	ldr	r0, [r7, #284]
 	add	r4, r9, r4
-	str	fp, [r7, #252]
+	str	fp, [r7, #260]
 	ror	r1, r1, #20
 	add	fp, r1, r0
 	eor	r2, r8, r2
-	ldr	r0, [r7, #252]
+	ldr	r0, [r7, #260]
 	eors	r3, r3, r4
 	vshl.i32	q5, q3, #8
-	str	r4, [r7, #224]
+	str	r4, [r7, #232]
 	vshl.i32	q8, q0, #8
-	ldr	r4, [r7, #260]
-	ldr	r5, [r7, #256]
+	ldr	r4, [r7, #268]
+	ldr	r5, [r7, #264]
 	ror	r2, r2, #20
 	ror	r3, r3, #20
 	eors	r6, r6, r0
 	adds	r5, r3, r5
 	add	r8, r2, r4
 	vsri.32	q5, q3, #24
-	ldr	r4, [r7, #264]
+	ldr	r4, [r7, #272]
 	eor	r9, r5, r9
 	eor	ip, fp, ip
 	vsri.32	q8, q0, #24
 	eor	sl, r8, sl
 	ror	r6, r6, #24
-	ldr	r0, [r7, #272]
-	str	r5, [r7, #268]
+	ldr	r0, [r7, #280]
+	str	r5, [r7, #276]
 	adds	r4, r6, r4
-	ldr	r5, [r7, #228]
+	ldr	r5, [r7, #236]
 	vadd.i32	q1, q1, q5
-	str	r4, [r7, #264]
+	str	r4, [r7, #272]
 	vadd.i32	q15, q15, q8
-	ldr	r4, [r7, #224]
+	ldr	r4, [r7, #232]
 	ror	ip, ip, #24
 	ror	sl, sl, #24
 	ror	r9, r9, #24
 	add	r5, ip, r5
 	add	r0, sl, r0
-	str	r5, [r7, #256]
+	str	r5, [r7, #264]
 	add	r5, r9, r4
-	str	r0, [r7, #276]
+	str	r0, [r7, #284]
 	veor	q6, q1, q6
-	ldr	r4, [r7, #20]
+	ldr	r4, [r7, #24]
 	veor	q14, q15, q14
-	ldr	r0, [r7, #264]
+	ldr	r0, [r7, #272]
 	eors	r3, r3, r5
 	vshl.i32	q0, q6, #7
 	vext.32	q1, q1, q1, #2
 	eors	r0, r0, r4
-	ldr	r4, [r7, #276]
-	str	r0, [r7, #272]
+	ldr	r4, [r7, #284]
+	str	r0, [r7, #280]
 	vshl.i32	q3, q14, #7
 	eors	r2, r2, r4
-	ldr	r4, [r7, #272]
-	ldr	r0, [r7, #256]
+	ldr	r4, [r7, #280]
+	ldr	r0, [r7, #264]
 	vsri.32	q0, q6, #25
 	ror	r2, r2, #25
 	ror	r3, r3, #25
 	eors	r1, r1, r0
 	vsri.32	q3, q14, #25
 	ror	r0, r4, #25
-	ldr	r4, [r7, #248]
+	ldr	r4, [r7, #256]
 	ror	r1, r1, #25
 	vext.32	q5, q5, q5, #1
 	subs	r4, r4, #1
-	str	r4, [r7, #248]
+	str	r4, [r7, #256]
 	vext.32	q15, q15, q15, #2
 	vext.32	q8, q8, q8, #1
 	vext.32	q0, q0, q0, #3
 	vext.32	q3, q3, q3, #3
 	bne	.L3
-	ldr	r4, [r7, #256]
-	vadd.i32	q4, q11, q4
-	str	r2, [r7, #256]
+	ldr	r4, [r7, #264]
 	vadd.i32	q14, q10, q9
-	ldr	r2, [r7, #244]
-	vld1.64	{d12-d13}, [r2:64]
-	str	r4, [r7, #272]
-	veor	q4, q4, q6
-	ldr	r4, [r7, #220]
+	str	r2, [r7, #264]
 	vadd.i32	q10, q10, q5
-	ldr	r2, [r7, #216]
-	vadd.i32	q0, q12, q0
-	add	fp, fp, r4
-	str	ip, [r7, #20]
-	ldr	r4, [r7, #212]
+	ldr	r2, [r7, #252]
+	vld1.64	{d12-d13}, [r2:64]
+	ldr	r2, [r7, #220]
+	vadd.i32	q4, q11, q4
+	str	ip, [r7, #24]
 	mov	ip, sl
-	str	r0, [r7, #224]
 	mov	sl, r8
+	ldr	r8, [r7, #260]
+	add	sl, sl, r2
+	ldr	r2, [r7, #212]
+	str	r4, [r7, #280]
+	vadd.i32	q0, q12, q0
+	ldr	r4, [r7, #224]
+	add	r8, r8, r2
+	ldr	r2, [r7, #240]
+	vadd.i32	q1, q13, q1
+	str	r0, [r7, #232]
+	add	fp, fp, r4
 	mov	r0, r5
-	ldr	r8, [r7, #252]
+	ldr	r4, [r7, #216]
 	mov	r5, r6
-	add	sl, sl, r2
 	mov	r6, r9
-	ldr	r2, [r7, #208]
-	ldr	r9, [r7, #268]
-	vadd.i32	q1, q13, q1
+	ldr	r9, [r7, #276]
+	adds	r2, r2, #3
+	str	r2, [r7, #240]
 	vadd.i32	q2, q11, q2
-	str	r1, [r7, #260]
+	ldr	r2, [r7, #252]
 	add	r9, r9, r4
-	add	r4, r8, r2
-	ldr	r8, [r7, #232]
 	vadd.i32	q3, q12, q3
+	ldr	r4, [r7, #228]
 	vadd.i32	q15, q13, q15
-	str	r3, [r7, #228]
-	add	r2, r8, #2
+	str	r1, [r7, #268]
 	vadd.i32	q8, q14, q8
-	add	ip, r2, ip
-	ldr	r2, [r7, #240]
-	vst1.64	{d8-d9}, [r2:64]
-	ldr	r2, [r7, #244]
-	ldr	r3, [r7, #276]
+	str	r3, [r7, #236]
+	veor	q4, q4, q6
+	ldr	r3, [r7, #284]
+	ldr	r1, [r7, #272]
+	add	ip, r4, ip
+	ldr	r4, [r7, #248]
+	vst1.64	{d8-d9}, [r4:64]
 	vldr	d8, [r2, #16]
 	vldr	d9, [r2, #24]
-	ldr	r1, [r7, #264]
 	veor	q0, q0, q4
-	add	r8, r8, #3
-	str	r8, [r7, #232]
-	ldr	r8, [r7, #240]
-	vstr	d0, [r8, #16]
-	vstr	d1, [r8, #24]
+	vstr	d0, [r4, #16]
+	vstr	d1, [r4, #24]
 	vldr	d0, [r2, #32]
 	vldr	d1, [r2, #40]
 	veor	q1, q1, q0
-	vstr	d2, [r8, #32]
-	vstr	d3, [r8, #40]
+	vstr	d2, [r4, #32]
+	vstr	d3, [r4, #40]
 	vldr	d2, [r2, #48]
 	vldr	d3, [r2, #56]
 	veor	q10, q10, q1
-	vstr	d20, [r8, #48]
-	vstr	d21, [r8, #56]
+	vstr	d20, [r4, #48]
+	vstr	d21, [r4, #56]
 	vldr	d8, [r2, #64]
 	vldr	d9, [r2, #72]
 	veor	q2, q2, q4
-	vstr	d4, [r8, #64]
-	vstr	d5, [r8, #72]
+	vstr	d4, [r4, #64]
+	vstr	d5, [r4, #72]
 	vldr	d10, [r2, #80]
 	vldr	d11, [r2, #88]
 	veor	q3, q3, q5
-	vstr	d6, [r8, #80]
-	vstr	d7, [r8, #88]
+	vstr	d6, [r4, #80]
+	vstr	d7, [r4, #88]
 	vldr	d12, [r2, #96]
 	vldr	d13, [r2, #104]
 	veor	q15, q15, q6
-	vstr	d30, [r8, #96]
-	vstr	d31, [r8, #104]
+	vstr	d30, [r4, #96]
+	vstr	d31, [r4, #104]
 	vldr	d20, [r2, #112]
 	vldr	d21, [r2, #120]
 	veor	q8, q8, q10
-	vstr	d16, [r8, #112]
-	vstr	d17, [r8, #120]
-	mov	r8, r2
-	ldr	r2, [r2, #128]
+	vstr	d16, [r4, #112]
+	vstr	d17, [r4, #120]
+	ldr	r4, [r2, #128]
+	ldr	r2, [r7, #248]
 	vadd.i32	q10, q14, q9
-	eor	r2, fp, r2
-	ldr	fp, [r7, #240]
+	eor	r4, fp, r4
 	vadd.i32	q10, q10, q9
-	str	r2, [fp, #128]
-	ldr	r2, [r8, #132]
+	str	r4, [r2, #128]
+	ldr	r4, [r7, #252]
+	ldr	r2, [r4, #132]
 	eor	r2, sl, r2
-	str	r2, [fp, #132]
-	ldr	r2, [r8, #136]
+	ldr	sl, [r7, #248]
+	str	r2, [sl, #132]
+	ldr	r2, [r4, #136]
 	eor	r2, r9, r2
-	str	r2, [fp, #136]
-	ldr	r2, [r8, #140]
-	eors	r2, r2, r4
-	str	r2, [fp, #140]
-	ldr	r2, [r7, #236]
-	ldr	r4, [r8, #144]
+	str	r2, [sl, #136]
+	ldr	r2, [r4, #140]
+	eor	r2, r8, r2
+	str	r2, [sl, #140]
+	ldr	r2, [r7, #244]
+	ldr	r4, [r4, #144]
 	ldr	r2, [r2, #0]
-	str	r4, [r7, #168]
-	ldr	r4, [r7, #224]
+	str	r4, [r7, #44]
+	ldr	r4, [r7, #232]
 	add	r8, r4, r2
-	ldr	r2, [r7, #168]
-	ldr	r4, [r7, #236]
+	ldr	r2, [r7, #44]
+	ldr	r4, [r7, #244]
 	eor	r8, r8, r2
-	ldr	r2, [r7, #244]
-	str	r8, [fp, #144]
+	ldr	r2, [r7, #252]
+	str	r8, [sl, #144]
 	ldr	r4, [r4, #4]
 	ldr	r2, [r2, #148]
-	str	r2, [r7, #36]
-	ldr	r2, [r7, #260]
+	str	r2, [r7, #40]
+	ldr	r2, [r7, #268]
 	add	r8, r2, r4
-	ldr	r4, [r7, #36]
-	ldr	r2, [r7, #236]
+	ldr	r4, [r7, #40]
+	ldr	r2, [r7, #244]
 	eor	r8, r8, r4
-	ldr	r4, [r7, #244]
-	str	r8, [fp, #148]
+	ldr	r4, [r7, #252]
+	str	r8, [sl, #148]
 	ldr	r2, [r2, #8]
 	ldr	r4, [r4, #152]
-	str	r4, [r7, #32]
-	ldr	r4, [r7, #256]
+	str	r4, [r7, #36]
+	ldr	r4, [r7, #264]
 	add	r8, r4, r2
-	ldr	r2, [r7, #32]
+	ldr	r2, [r7, #36]
 	eor	r8, r8, r2
-	str	r8, [fp, #152]
-	ldr	r2, [r7, #244]
-	ldr	r4, [r7, #236]
+	str	r8, [sl, #152]
+	ldr	r2, [r7, #252]
+	ldr	r4, [r7, #244]
 	ldr	r2, [r2, #156]
 	ldr	r4, [r4, #12]
-	str	r2, [r7, #28]
-	ldr	r2, [r7, #228]
+	str	r2, [r7, #32]
+	ldr	r2, [r7, #236]
 	add	r8, r2, r4
-	ldr	r4, [r7, #28]
-	ldr	r2, [r7, #244]
+	ldr	r4, [r7, #32]
+	ldr	r2, [r7, #252]
 	eor	r8, r8, r4
-	str	r8, [fp, #156]
-	ldr	r8, [r7, #236]
+	str	r8, [sl, #156]
+	ldr	r8, [r7, #244]
 	ldr	r2, [r2, #160]
 	ldr	r4, [r8, #16]
 	adds	r0, r0, r4
-	ldr	r4, [r7, #244]
+	ldr	r4, [r7, #252]
 	eors	r0, r0, r2
-	str	r0, [fp, #160]
+	str	r0, [sl, #160]
 	ldr	r0, [r8, #20]
 	ldr	r2, [r4, #164]
 	adds	r1, r1, r0
-	ldr	r0, [r7, #272]
+	ldr	r0, [r7, #280]
 	eors	r1, r1, r2
-	str	r1, [fp, #164]
+	str	r1, [sl, #164]
 	ldr	r2, [r8, #24]
 	ldr	r1, [r4, #168]
 	adds	r2, r0, r2
 	eors	r2, r2, r1
-	str	r2, [fp, #168]
+	str	r2, [sl, #168]
 	ldr	r1, [r8, #28]
 	ldr	r2, [r4, #172]
 	adds	r3, r3, r1
 	eors	r3, r3, r2
-	str	r3, [fp, #172]
+	str	r3, [sl, #172]
 	ldr	r3, [r4, #176]
 	eor	r3, ip, r3
-	str	r3, [fp, #176]
+	str	r3, [sl, #176]
 	ldr	r3, [r4, #180]
-	ldr	r4, [r7, #392]
+	ldr	r4, [r7, #400]
 	eors	r6, r6, r3
-	str	r6, [fp, #180]
-	ldr	r6, [r7, #244]
+	str	r6, [sl, #180]
+	ldr	r6, [r7, #252]
 	ldr	r2, [r4, #0]
 	ldr	r3, [r6, #184]
 	adds	r5, r5, r2
 	eors	r5, r5, r3
-	str	r5, [fp, #184]
+	str	r5, [sl, #184]
 	ldr	r2, [r6, #188]
 	adds	r6, r6, #192
 	ldr	r3, [r4, #4]
-	str	r6, [r7, #244]
-	ldr	r0, [r7, #20]
-	ldr	r1, [r7, #232]
+	str	r6, [r7, #252]
+	ldr	r0, [r7, #24]
+	ldr	r1, [r7, #240]
 	adds	r4, r0, r3
 	eors	r4, r4, r2
 	ldr	r2, [r7, #204]
-	str	r4, [fp, #188]
-	add	fp, fp, #192
+	str	r4, [sl, #188]
+	add	sl, sl, #192
 	cmp	r1, r2
-	str	fp, [r7, #240]
+	str	sl, [r7, #248]
 	bne	.L4
-	ldr	r4, [r7, #188]
-	ldr	r3, [r7, #176]
-	ldr	r6, [r7, #184]
+	ldr	r4, [r7, #192]
+	ldr	r3, [r7, #180]
+	ldr	r6, [r7, #188]
 	adds	r5, r3, r4
-	ldr	r8, [r7, #180]
+	ldr	r8, [r7, #184]
 	lsls	r5, r5, #6
 	adds	r4, r6, r5
 	add	r5, r8, r5
 .L2:
-	ldr	fp, [r7, #192]
+	ldr	r9, [r7, #196]
 	movw	r3, #43691
 	movt	r3, 43690
-	ldr	r6, [r7, #192]
-	umull	fp, r3, r3, fp
+	ldr	sl, [r7, #196]
+	umull	r9, r3, r3, r9
 	lsrs	r3, r3, #7
 	add	r3, r3, r3, lsl #1
-	sub	r3, r6, r3, lsl #6
+	sub	r3, sl, r3, lsl #6
 	lsrs	r6, r3, #6
 	beq	.L5
 	add	r1, r5, #16
@@ -693,7 +696,7 @@ CRYPTO_chacha_20_neon:
 	adds	r4, r4, r6
 	adds	r5, r5, r6
 .L5:
-	ldr	r6, [r7, #192]
+	ldr	r6, [r7, #196]
 	ands	ip, r6, #63
 	beq	.L1
 	vmov	q8, q10  @ v4si
@@ -743,11 +746,11 @@ CRYPTO_chacha_20_neon:
 	cmp	ip, #15
 	vadd.i32	q11, q11, q15
 	bhi	.L37
-	ldr	fp, [r7, #196]
-	vst1.64	{d22-d23}, [fp:128]
+	ldr	r9, [r7, #200]
+	vst1.64	{d22-d23}, [r9:128]
 .L14:
-	ldr	r6, [r7, #192]
-	and	r3, r6, #48
+	ldr	sl, [r7, #196]
+	and	r3, sl, #48
 	cmp	ip, r3
 	bls	.L1
 	adds	r0, r5, r3
@@ -772,7 +775,7 @@ CRYPTO_chacha_20_neon:
 	orreq	r2, r2, #1
 	lsl	sl, r8, #4
 	cbnz	r2, .L35
-	ldr	fp, [r7, #196]
+	ldr	fp, [r7, #200]
 	add	r6, fp, r3
 .L17:
 	vld1.8	{q8}, [r0]!
@@ -786,7 +789,7 @@ CRYPTO_chacha_20_neon:
 	add	r3, r3, sl
 	beq	.L1
 .L35:
-	ldr	r0, [r7, #196]
+	ldr	r0, [r7, #200]
 .L25:
 	ldrb	r2, [r5, r3]	@ zero_extendqisi2
 	ldrb	r1, [r3, r0]	@ zero_extendqisi2
@@ -796,7 +799,7 @@ CRYPTO_chacha_20_neon:
 	cmp	ip, r3
 	bhi	.L25
 .L1:
-	add	r7, r7, #296
+	add	r7, r7, #304
 	mov	sp, r7
 	fldmfdd	sp!, {d8, d9, d10, d11, d12, d13, d14, d15}
 	pop	{r4, r5, r6, r7, r8, r9, sl, fp}
@@ -819,7 +822,7 @@ CRYPTO_chacha_20_neon:
 	vadd.i32	q8, q8, q10
 	vldr	d0, [r5, #32]
 	vldr	d1, [r5, #40]
-	ldr	r6, [r7, #196]
+	ldr	r6, [r7, #200]
 	vstr	d16, [r6, #48]
 	vstr	d17, [r6, #56]
 	veor	q8, q13, q0
@@ -827,16 +830,16 @@ CRYPTO_chacha_20_neon:
 	vstr	d17, [r4, #40]
 	b	.L14
 .L12:
-	ldr	r8, [r7, #196]
+	ldr	r8, [r7, #200]
 	vstr	d18, [r8, #16]
 	vstr	d19, [r8, #24]
 	b	.L14
 .L20:
-	ldr	r5, [r7, #180]
-	ldr	r4, [r7, #184]
+	ldr	r5, [r7, #184]
+	ldr	r4, [r7, #188]
 	b	.L2
 .L13:
-	ldr	r6, [r7, #196]
+	ldr	r6, [r7, #200]
 	vstr	d26, [r6, #32]
 	vstr	d27, [r6, #40]
 	b	.L14
diff --git a/openssl/crypto/evp/e_chacha20poly1305.c b/openssl/crypto/evp/e_chacha20poly1305.c
index 1c0c0fb..c556ddc 100644
--- a/openssl/crypto/evp/e_chacha20poly1305.c
+++ b/openssl/crypto/evp/e_chacha20poly1305.c
@@ -120,6 +120,12 @@ static void poly1305_update_with_length(poly1305_state *poly1305,
 	CRYPTO_poly1305_update(poly1305, length_bytes, sizeof(length_bytes));
 }
 
+#if __arm__
+#define ALIGNED __attribute__((aligned(16)))
+#else
+#define ALIGNED
+#endif
+
 static ssize_t aead_chacha20_poly1305_seal(const EVP_AEAD_CTX *ctx,
 	unsigned char *out, size_t max_out_len,
 	const unsigned char *nonce, size_t nonce_len,
@@ -127,7 +133,7 @@ static ssize_t aead_chacha20_poly1305_seal(const EVP_AEAD_CTX *ctx,
 	const unsigned char *ad, size_t ad_len)
 	{
 	const struct aead_chacha20_poly1305_ctx *c20_ctx = ctx->aead_state;
-	unsigned char poly1305_key[32];
+	unsigned char poly1305_key[32] ALIGNED;
 	poly1305_state poly1305;
 	const uint64_t in_len_64 = in_len;
 
@@ -184,7 +190,7 @@ static ssize_t aead_chacha20_poly1305_open(const EVP_AEAD_CTX *ctx,
 	{
 	const struct aead_chacha20_poly1305_ctx *c20_ctx = ctx->aead_state;
 	unsigned char mac[POLY1305_TAG_LEN];
-	unsigned char poly1305_key[32];
+	unsigned char poly1305_key[32] ALIGNED;
 	size_t out_len;
 	poly1305_state poly1305;
 	const uint64_t in_len_64 = in_len;
diff --git a/openssl/crypto/poly1305/poly1305_arm.c b/openssl/crypto/poly1305/poly1305_arm.c
index 34e339d..c7d2991 100644
--- a/openssl/crypto/poly1305/poly1305_arm.c
+++ b/openssl/crypto/poly1305/poly1305_arm.c
@@ -238,8 +238,6 @@ void CRYPTO_poly1305_update_neon(poly1305_state *state, const unsigned char *in,
 	fe1305x2 *const c = h + 1;
 	fe1305x2 *const precomp = c + 1;
 	unsigned int i;
-	unsigned char data[sizeof(fe1305x2) + 16];
-	fe1305x2 *const r2r = (fe1305x2 *) (data + (15 & (-(int) data)));
 
 	if (st->buf_used)
 		{
@@ -252,20 +250,12 @@ void CRYPTO_poly1305_update_neon(poly1305_state *state, const unsigned char *in,
 		in_len -= todo;
 		in += todo;
 
-		if (st->buf_used == sizeof(st->buf))
+		if (st->buf_used == sizeof(st->buf) && in_len)
 			{
+			addmulmod(h,h,precomp,&zero);
 			fe1305x2_frombytearray(c, st->buf, sizeof(st->buf));
-			r2r->v[0] = precomp->v[0];
-			r2r->v[2] = precomp->v[2];
-			r2r->v[4] = precomp->v[4];
-			r2r->v[6] = precomp->v[6];
-			r2r->v[8] = precomp->v[8];
-			r2r->v[1] = r->v[1];
-			r2r->v[3] = r->v[3];
-			r2r->v[5] = r->v[5];
-			r2r->v[7] = r->v[7];
-			r2r->v[9] = r->v[9];
-			addmulmod(h,h,r2r,c);
+			for (i = 0; i < 10; i++)
+				h->v[i] += c->v[i];
 			st->buf_used = 0;
 			}
 		}
@@ -273,7 +263,7 @@ void CRYPTO_poly1305_update_neon(poly1305_state *state, const unsigned char *in,
 	while (in_len > 32)
 		{
 		unsigned int tlen = 1048576;
-		if (in_len < 1048576)
+		if (in_len < tlen)
 			tlen = in_len;
 		tlen -= blocks(h, precomp, in, tlen);
 		in_len -= tlen;
@@ -296,6 +286,8 @@ void CRYPTO_poly1305_finish_neon(poly1305_state* state, unsigned char mac[16])
 	fe1305x2 *const c = h + 1;
 	fe1305x2 *const precomp = c + 1;
 
+	addmulmod(h,h,precomp,&zero);
+
 	if (st->buf_used > 16)
 		{
 		fe1305x2_frombytearray(c, st->buf, st->buf_used);