Update ARM mod and redc functions

5 files changed, 99 insertions, 88 deletions

diff --git a/arm/ecc-secp192r1-modp.asm b/arm/ecc-secp192r1-modp.asm
index 72a81a54..e500bc85 100644
--- a/arm/ecc-secp192r1-modp.asm
+++ b/arm/ecc-secp192r1-modp.asm
@@ -35,15 +35,16 @@ ifelse(`
 
 define(`HP', `r0') C Overlaps unused modulo argument
 define(`RP', `r1')
-
-define(`T0', `r2')
-define(`T1', `r3')
-define(`T2', `r4')
-define(`T3', `r5')
-define(`T4', `r6')
-define(`T5', `r7')
-define(`T6', `r8')
-define(`T7', `r10')
+define(`XP', `r2')
+
+define(`T0', `r3')
+define(`T1', `r4')
+define(`T2', `r5')
+define(`T3', `r6')
+define(`T4', `r7')
+define(`T5', `r8')
+define(`T6', `r10')
+define(`T7', `r11')
 define(`H0', `T0') C Overlaps T0 and T1
 define(`H1', `T1')
 define(`C2', `HP')
@@ -54,12 +55,12 @@ define(`C4', `r12')
 	.align 2
 
 PROLOGUE(_nettle_ecc_secp192r1_modp)
-	push	{r4,r5,r6,r7,r8,r10}
+	push	{r4,r5,r6,r7,r8,r10,r11}
 	C Reduce two words at a time
-	add	HP, RP, #48
-	add	RP, RP, #8
+	add	HP, XP, #48
+	add	XP, XP, #8
 	ldmdb	HP!, {H0,H1}
-	ldm	RP, {T2,T3,T4,T5,T6,T7}
+	ldm	XP, {T2,T3,T4,T5,T6,T7}
 	mov	C4, #0
 	adds	T4, T4, H0
 	adcs	T5, T5, H1
@@ -77,7 +78,7 @@ PROLOGUE(_nettle_ecc_secp192r1_modp)
 	C Need to add carry to T0 and T2, do T2 later
 	adc	C2, C2, #0
 
-	ldmdb	RP!, {T0, T1}
+	ldmdb	XP!, {T0, T1}
 	adcs	T0, T0, T6
 	adcs	T1, T1, T7
 	adcs	T2, T2, T6
@@ -101,6 +102,6 @@ PROLOGUE(_nettle_ecc_secp192r1_modp)
 
 	stm	RP, {T0,T1,T2,T3,T4,T5}
 
-	pop	{r4,r5,r6,r7,r8,r10}
+	pop	{r4,r5,r6,r7,r8,r10,r11}
 	bx	lr
 EPILOGUE(_nettle_ecc_secp192r1_modp)
diff --git a/arm/ecc-secp224r1-modp.asm b/arm/ecc-secp224r1-modp.asm
index 3256601c..4b3b24e5 100644
--- a/arm/ecc-secp224r1-modp.asm
+++ b/arm/ecc-secp224r1-modp.asm
@@ -33,10 +33,11 @@ ifelse(`
 	.file "ecc-secp224r1-modp.asm"
 	.arm
 
-define(`RP', `r1')
-define(`H', `r0') C Overlaps unused modulo argument
+define(`RP', `r1')	C Overlaps T0
+define(`XP', `r2')
+define(`H', `r0')	C Overlaps unused modulo argument
 
-define(`T0', `r2')
+define(`T0', `r1')
 define(`T1', `r3')
 define(`T2', `r4')
 define(`T3', `r5')
@@ -53,9 +54,10 @@ define(`L2', `lr')
 	.align 2
 
 PROLOGUE(_nettle_ecc_secp224r1_modp)
-	push	{r4,r5,r6,r7,r8,r10,r11,lr}
+	C Pushes RP last
+	push	{r1,r4,r5,r6,r7,r8,r10,r11,lr}
 
-	add	L2, RP, #28
+	add	L2, XP, #28
 	ldm	L2, {T0,T1,T2,T3,T4,T5,T6}
 	mov	H, #0
 
@@ -80,15 +82,15 @@ PROLOGUE(_nettle_ecc_secp224r1_modp)
 	sbc	H, #0
 
 	C Now subtract from low half
-	ldm	RP!, {L0,L1,L2}
+	ldm	XP!, {L0,L1,L2}
 
 	C Clear carry, with the sbcs, this is the 1.
-	adds	RP, #0
+	adds	XP, #0
 
 	sbcs	T0, L0, T0
 	sbcs	T1, L1, T1
 	sbcs	T2, L2, T2
-	ldm	RP!, {T3,L0,L1,L2}
+	ldm	XP!, {T3,L0,L1,L2}
 	sbcs	T3, T3, N3
 	sbcs	T4, L0, T4
 	sbcs	T5, L1, T5
@@ -109,6 +111,8 @@ PROLOGUE(_nettle_ecc_secp224r1_modp)
 	sbcs	T6, T6, #0
 	sbcs	H, H, H
 
+	pop	{XP}	C Original RP
+
 	C Final borrow, subtract (B^3 - 1) |H|
 	subs	T0, T0, H
 	sbcs	T1, T1, H
@@ -118,7 +122,7 @@ PROLOGUE(_nettle_ecc_secp224r1_modp)
 	sbcs	T5, T5, #0
 	sbcs	T6, T6, #0
 
-	stmdb	RP, {T0,T1,T2,T3,T4,T5,T6}
+	stm	XP, {T0,T1,T2,T3,T4,T5,T6}
 
 	pop	{r4,r5,r6,r7,r8,r10,r11,pc}
 EPILOGUE(_nettle_ecc_secp224r1_modp)
diff --git a/arm/ecc-secp256r1-redc.asm b/arm/ecc-secp256r1-redc.asm
index e127a2f2..da574398 100644
--- a/arm/ecc-secp256r1-redc.asm
+++ b/arm/ecc-secp256r1-redc.asm
@@ -33,10 +33,11 @@ ifelse(`
 	.file "ecc-secp256r1-redc.asm"
 	.arm
 
-define(`RP', `r1')
+define(`RP', `r1')	C Overlaps T1 below
+define(`XP', `r2')
 
-define(`T0', `r0') C Overlaps unused modulo argument
-define(`T1', `r2')
+define(`T0', `r0')	C Overlaps unused modulo argument
+define(`T1', `r1')
 define(`T2', `r3')
 define(`T3', `r4')
 define(`T4', `r5')
@@ -53,9 +54,10 @@ define(`F3', `lr')
 	.align 2
 
 PROLOGUE(_nettle_ecc_secp256r1_redc)
-	push	{r4,r5,r6,r7,r8,r10,r11,lr}
+	C Pushes RP last
+	push	{r1, r4,r5,r6,r7,r8,r10,r11,lr}
 
-	ldm	RP!, {T0,T1,T2,T3,T4,T5,T6,T7}
+	ldm	XP!, {T0,T1,T2,T3,T4,T5,T6,T7}
 
 	C Set <F3,F2,F1> to the high 4 limbs of (B^2-B+1)<T2,T1,T0>
 	C         T2 T1
@@ -88,7 +90,7 @@ PROLOGUE(_nettle_ecc_secp256r1_redc)
 	mov	T3, T6
 	adcs	T4, T7, F0
 
-	ldm	RP!, {T5,T6,T7}
+	ldm	XP!, {T5,T6,T7}
 	adcs	T5, T5, F1
 	adcs	T6, T6, F2
 	adcs	T7, T7, F3
@@ -112,7 +114,7 @@ PROLOGUE(_nettle_ecc_secp256r1_redc)
 	mov	T3, T6
 	adcs	T4, T7, F0
 
-	ldm	RP!, {T5,T6,T7}
+	ldm	XP!, {T5,T6,T7}
 	adcs	T5, T5, F1
 	adcs	T6, T6, F2
 	adcs	T7, T7, F3
@@ -143,7 +145,7 @@ PROLOGUE(_nettle_ecc_secp256r1_redc)
 	adcs	T5, T5, #0
 	adcs	T6, T6, T0
 	adcs	T7, T7, F0
-	ldm	RP!, {T0, T1}
+	ldm	XP!, {T0, T1}
 	mov	F3, #0
 	adcs	F1, F1, T0
 	adcs	F2, F2, T1
@@ -156,6 +158,8 @@ PROLOGUE(_nettle_ecc_secp256r1_redc)
 	adc	F3, F3, #0
 	rsb	F3, F3, #0
 
+	pop	{XP}	C Original RP
+
 	adcs	T0, T2, #0
 	adcs	T1, T3, #0
 	adcs	T2, T4, #0
@@ -166,8 +170,7 @@ PROLOGUE(_nettle_ecc_secp256r1_redc)
 	adcs	T6, F1, F3
 	adcs	T7, F2, #0
 
-	sub	RP, RP, #64
-	stm	RP, {T0,T1,T2,T3,T4,T5,T6,T7}
+	stm	XP, {T0,T1,T2,T3,T4,T5,T6,T7}
 
 	pop	{r4,r5,r6,r7,r8,r10,r11,pc}
 EPILOGUE(_nettle_ecc_secp256r1_redc)
diff --git a/arm/ecc-secp384r1-modp.asm b/arm/ecc-secp384r1-modp.asm
index 96744ee9..840eead8 100644
--- a/arm/ecc-secp384r1-modp.asm
+++ b/arm/ecc-secp384r1-modp.asm
@@ -34,15 +34,17 @@ ifelse(`
 	.arm
 
 define(`RP', `r1')
+define(`XP', `r2')
+
 define(`T0', `r0')
-define(`T1', `r2')
-define(`T2', `r3')
-define(`T3', `r4')
-define(`F0', `r5')
-define(`F1', `r6')
-define(`F2', `r7')
-define(`F3', `r8')
-define(`F4', `r10')
+define(`T1', `r3')
+define(`T2', `r4')
+define(`T3', `r5')
+define(`F0', `r6')
+define(`F1', `r7')
+define(`F2', `r8')
+define(`F3', `r10')
+define(`F4', `r11')
 define(`N', `r12')
 define(`H', `lr')
 	
@@ -51,10 +53,10 @@ define(`H', `lr')
 	.align 2
 
 PROLOGUE(_nettle_ecc_secp384r1_modp)
-	push	{r4,r5,r6,r7,r8,r10,lr}
+	push	{r4,r5,r6,r7,r8,r10,r11,lr}
 
-	add	RP, RP, #80
-	ldm	RP, {T0, T1, T2, T3}	C 20-23
+	add	XP, XP, #80
+	ldm	XP, {T0, T1, T2, T3}	C 20-23
 
 	C First get top 4 limbs, which need folding twice, as
 	C
@@ -91,8 +93,8 @@ PROLOGUE(_nettle_ecc_secp384r1_modp)
 	adcs	F4, F4, #0
 
 	C Add in to high part
-	sub	RP, RP, #32
-	ldm	RP, {T0, T1, T2, T3}	C 12-15
+	sub	XP, XP, #32
+	ldm	XP, {T0, T1, T2, T3}	C 12-15
 	mov	H, #0
 	adds	F0, T0, F0
 	adcs	F1, T1, F1
@@ -101,8 +103,8 @@ PROLOGUE(_nettle_ecc_secp384r1_modp)
 	adcs	F4, F4, #0			C Do F4 later
 
 	C Add to low part, keeping carry (positive or negative) in H
-	sub	RP, RP, #48
-	ldm	RP, {T0, T1, T2, T3}	C 0-3
+	sub	XP, XP, #48
+	ldm	XP, {T0, T1, T2, T3}	C 0-3
 	mov	H, #0
 	adds	T0, T0, F0
 	adcs	T1, T1, F1
@@ -116,10 +118,10 @@ PROLOGUE(_nettle_ecc_secp384r1_modp)
 	adds	T3, T3, F0
 	adc	H, H, #0
 	
-	stm	RP!, {T0,T1,T2,T3}	C 0-3
+	stm	XP!, {T0,T1,T2,T3}	C 0-3
 	mov	N, #2
 .Loop:
-	ldm	RP, {T0,T1,T2,T3}	C 4-7
+	ldm	XP, {T0,T1,T2,T3}	C 4-7
 
 	C First, propagate carry
 	adds	T0, T0, H
@@ -137,7 +139,7 @@ PROLOGUE(_nettle_ecc_secp384r1_modp)
 	adc	H, H, #0
 
 	C +B^3 terms
-	ldr	F0, [RP, #+48]		C 16
+	ldr	F0, [XP, #+48]		C 16
 	adds	T0, T0, F1
 	adcs	T1, T1, F2
 	adcs	T2, T2, F3
@@ -145,8 +147,8 @@ PROLOGUE(_nettle_ecc_secp384r1_modp)
 	adc	H, H, #0
 
 	C -B
-	ldr	F1, [RP, #+52]		C 17-18
-	ldr	F2, [RP, #+56]
+	ldr	F1, [XP, #+52]		C 17-18
+	ldr	F2, [XP, #+56]
 	subs	T0, T0, F3
 	sbcs	T1, T1, F0
 	sbcs	T2, T2, F1
@@ -154,14 +156,14 @@ PROLOGUE(_nettle_ecc_secp384r1_modp)
 	sbcs	H, H, #0
 
 	C +1
-	ldr	F3, [RP, #+60]		C 19
+	ldr	F3, [XP, #+60]		C 19
 	adds	T0, T0, F0
 	adcs	T1, T1, F1
 	adcs	T2, T2, F2
 	adcs	T3, T3, F3
 	adc	H, H, #0
 	subs	N, N, #1
-	stm	RP!, {T0,T1,T2,T3}
+	stm	XP!, {T0,T1,T2,T3}
 	bne	.Loop
 
 	C Fold high limbs, we need to add in
@@ -170,9 +172,9 @@ PROLOGUE(_nettle_ecc_secp384r1_modp)
 	C
 	C We always have F4 >= 0, but we can have H < 0.
 	C Sign extension gets tricky when F4 = 0 and H < 0.
-	sub	RP, RP, #48
+	sub	XP, XP, #48
 
-	ldm	RP, {T0,T1,T2,T3}	C 0-3
+	ldm	XP, {T0,T1,T2,T3}	C 0-3
 
 	C     H  H  0 -H  H
 	C  ----------------
@@ -201,8 +203,8 @@ PROLOGUE(_nettle_ecc_secp384r1_modp)
 	adcs	T3, T3, F3
 	adc	H, H, F0	C	0+cy	H+cy	-2+cy
 
-	stm	RP!, {T0,T1,T2,T3}	C 0-3
-	ldm	RP, {T0,T1,T2,T3}	C 4-7
+	stm	XP!, {T0,T1,T2,T3}	C 0-3
+	ldm	XP, {T0,T1,T2,T3}	C 4-7
 	
 	C   F4  0 -F4
 	C ---------
@@ -226,8 +228,8 @@ PROLOGUE(_nettle_ecc_secp384r1_modp)
 	adcs	T2, T2, F2
 	adcs	T3, T3, F3
 
-	stm	RP!, {T0,T1,T2,T3}	C 4-7
-	ldm	RP, {T0,T1,T2,T3}	C 8-11
+	stm	XP!, {T0,T1,T2,T3}	C 4-7
+	ldm	XP, {T0,T1,T2,T3}	C 8-11
 
 	adcs	T0, T0, F4
 	adcs	T1, T1, H
@@ -235,11 +237,11 @@ PROLOGUE(_nettle_ecc_secp384r1_modp)
 	adcs	T3, T3, H
 	adc	H, H, #0
 	
-	stm	RP, {T0,T1,T2,T3}	C 8-11
+	stm	XP, {T0,T1,T2,T3}	C 8-11
 
 	C Final (unlikely) carry
-	sub	RP, RP, #32
-	ldm	RP, {T0,T1,T2,T3}	C 0-3
+	sub	XP, XP, #32
+	ldm	XP!, {T0,T1,T2,T3}	C 0-3
 	C Fold H into F0-F4
 	mov	F0, H
 	asr	H, #31
@@ -254,17 +256,17 @@ PROLOGUE(_nettle_ecc_secp384r1_modp)
 	adcs	T3, T3, F3
 	
 	stm	RP!, {T0,T1,T2,T3}	C 0-3
-	ldm	RP, {T0,T1,T2,T3}	C 4-7
+	ldm	XP!, {T0,T1,T2,T3}	C 4-7
 	adcs	T0, T0, F4
 	adcs	T1, T1, H
 	adcs	T2, T2, H
 	adcs	T3, T3, H
 	stm	RP!, {T0,T1,T2,T3}	C 4-7
-	ldm	RP, {T0,T1,T2,T3}	C 8-11
+	ldm	XP, {T0,T1,T2,T3}	C 8-11
 	adcs	T0, T0, H
 	adcs	T1, T1, H
 	adcs	T2, T2, H
 	adcs	T3, T3, H
-	stm	RP!, {T0,T1,T2,T3}	C 8-11
-	pop	{r4,r5,r6,r7,r8,r10,pc}
+	stm	RP, {T0,T1,T2,T3}	C 8-11
+	pop	{r4,r5,r6,r7,r8,r10,r11,pc}
 EPILOGUE(_nettle_ecc_secp384r1_modp)
diff --git a/arm/ecc-secp521r1-modp.asm b/arm/ecc-secp521r1-modp.asm
index 22e8dd4e..d1952173 100644
--- a/arm/ecc-secp521r1-modp.asm
+++ b/arm/ecc-secp521r1-modp.asm
@@ -35,13 +35,14 @@ ifelse(`
 
 define(`HP', `r0')
 define(`RP', `r1')
-define(`T0', `r2')
-define(`T1', `r3')
-define(`T2', `r4')
-define(`F0', `r5')
-define(`F1', `r6')
-define(`F2', `r7')
-define(`F3', `r8')
+define(`XP', `r2')
+define(`T0', `r3')
+define(`T1', `r4')
+define(`T2', `r5')
+define(`F0', `r6')
+define(`F1', `r7')
+define(`F2', `r8')
+define(`F3', `r10')
 define(`H', `r12')
 define(`N', `lr')
 
@@ -53,20 +54,20 @@ define(`N', `lr')
 	.align 2
 
 PROLOGUE(_nettle_ecc_secp521r1_modp)
-	push	{r4,r5,r6,r7,r8,lr}
+	push	{r4,r5,r6,r7,r8,r10,lr}
 
 	C Use that B^17 = 2^23 (mod p)
-	ldr	F3, [RP, #+68]		C 17
-	add	HP, RP, #72		C 18
-	ldr	T0, [RP]		C 0
+	ldr	F3, [XP, #+68]		C 17
+	add	HP, XP, #72		C 18
+	ldr	T0, [XP]		C 0
 	adds	T0, T0, F3, lsl	#23
-	str	T0, [RP], #+4
+	str	T0, [XP], #+4
 	mov	N, #5
 
 	C 5 iterations, reading limbs 18-20, 21-23, 24-26, 27-29, 30-32
 	C and adding to limbs          1-3,    4-6,   7-9, 19-12, 13-15
 .Loop:
-	ldm	RP, {T0,T1,T2}		C  1+3*k --  3+3*k
+	ldm	XP, {T0,T1,T2}		C  1+3*k --  3+3*k
 	lsr	F0, F3, #9
 	ldm	HP!, {F1,F2,F3}		C 18+3*k -- 20+3*k
 	orr	F0, F0, F1, lsl #23
@@ -78,11 +79,11 @@ PROLOGUE(_nettle_ecc_secp521r1_modp)
 	adcs	T1, T1, F1
 	adcs	T2, T2, F2
 	sub	N, N, #1
-	stm	RP!,{T0,T1,T2}
+	stm	XP!,{T0,T1,T2}
 	teq	N, #0
 	bne	.Loop
 
-	ldr	F0, [RP], #-64		C 16
+	ldr	F0, [XP], #-64		C 16
 	ldr	F1, [HP]		C 33
 	ldr	T0, .Lc511
 
@@ -98,12 +99,12 @@ PROLOGUE(_nettle_ecc_secp521r1_modp)
 	lsr	F1, F1, #18
 	adc	F1, F1, #0
 
-	ldm	RP, {T0, T1}		C 0-1
+	ldm	XP!, {T0, T1}		C 0-1
 	adds	T0, T0, F0
 	adcs	T1, T1, F1
 	stm	RP!, {T0, T1}
 
-	ldm	RP, {T0,T1,T2,F0,F1,F2,F3}	C 2-8
+	ldm	XP!, {T0,T1,T2,F0,F1,F2,F3}	C 2-8
 	adcs	T0, T0, #0
 	adcs	T1, T1, #0
 	adcs	T2, T2, #0
@@ -112,7 +113,7 @@ PROLOGUE(_nettle_ecc_secp521r1_modp)
 	adcs	F2, F2, #0
 	adcs	F3, F3, #0
 	stm	RP!, {T0,T1,T2,F0,F1,F2,F3}	C 2-8
-	ldm	RP, {T0,T1,T2,F0,F1,F2,F3}	C 9-15
+	ldm	XP, {T0,T1,T2,F0,F1,F2,F3}	C 9-15
 	adcs	T0, T0, #0
 	adcs	T1, T1, #0
 	adcs	T2, T2, #0
@@ -123,5 +124,5 @@ PROLOGUE(_nettle_ecc_secp521r1_modp)
 	adcs	H, H, #0
 	stm	RP, {T0,T1,T2,F0,F1,F2,F3,H}	C 9-16
 
-	pop	{r4,r5,r6,r7,r8,pc}
+	pop	{r4,r5,r6,r7,r8,r10,pc}
 EPILOGUE(_nettle_ecc_secp521r1_modp)