X-Git-Url: https://git.cryptolib.org/?a=blobdiff_plain;f=keccak%2Fkeccak-asm.S;h=3b3a48818fdc84a47547b64d61e6b00ad1eca3a9;hb=29a44972ae3749a6a273d936f2e15327ecae8a94;hp=77dc9ce10bc99f92f67f5b7e43ddcf0155c5532c;hpb=4128060fcc4585ab8f40d01486baa4248b499160;p=avr-crypto-lib.git

diff --git a/keccak/keccak-asm.S b/keccak/keccak-asm.S
index 77dc9ce..3b3a488 100644
--- a/keccak/keccak-asm.S
+++ b/keccak/keccak-asm.S
@@ -31,7 +31,27 @@
 
 .equ __zero_reg__, 1
 
-.global rho_pi_idx_table
+/*
+typedef struct{
+	uint64_t a[5][5];
+	uint16_t r, c;
+	uint8_t  d, bs;
+} keccak_ctx_t;
+*/
+	.struct 0
+ctx_a:
+	.struct ctx_a + 8 * 5 * 5
+ctx_r:
+	.struct ctx_r + 2
+ctx_c:
+	.struct ctx_c + 2
+ctx_d:
+	.struct ctx_d + 1
+ctx_bs:
+
+	.section .text
+
+	.global rho_pi_idx_table
 rho_pi_idx_table:
 	.irp i, 0, 1, 2, 3, 4
 		.irp j, 0, 1, 2, 3, 4
@@ -39,7 +59,145 @@ rho_pi_idx_table:
 		.endr
 	.endr
 
-.align 2
+/*
+#define ROT_BIT(a) (( (a) <= 4) ? ((a) << 1) : (0x01 | ((8 - (a)) << 1)))
+#define ROT_CODE(a) ((((a) / 8 + ((((a) % 8) > 4) ? 1 : 0)) << 4) | ROT_BIT(((a) % 8)))
+
+const uint8_t keccak_rotate_codes[5][5] PROGMEM = {
+        { ROT_CODE( 0), ROT_CODE( 1), ROT_CODE(62), ROT_CODE(28), ROT_CODE(27) },
+        { ROT_CODE(36), ROT_CODE(44), ROT_CODE( 6), ROT_CODE(55), ROT_CODE(20) },
+        { ROT_CODE( 3), ROT_CODE(10), ROT_CODE(43), ROT_CODE(25), ROT_CODE(39) },
+        { ROT_CODE(41), ROT_CODE(45), ROT_CODE(15), ROT_CODE(21), ROT_CODE( 8) },
+        { ROT_CODE(18), ROT_CODE( 2), ROT_CODE(61), ROT_CODE(56), ROT_CODE(14) }
+};
+*/
+
+keccak_rotate_codes:
+.byte	0x00, 0x02, 0x85, 0x38, 0x36
+.byte	0x48, 0x58, 0x15, 0x73, 0x28
+.byte	0x06, 0x14, 0x56, 0x32, 0x53
+.byte	0x52, 0x67, 0x23, 0x37, 0x10
+.byte	0x24, 0x04, 0x87, 0x70, 0x25
+
+keccak_rc_comp:
+.byte	0x01, 0x92, 0xda, 0x70
+.byte	0x9b, 0x21, 0xf1, 0x59
+.byte	0x8a, 0x88, 0x39, 0x2a
+.byte	0xbb, 0xcb, 0xd9, 0x53
+.byte	0x52, 0xc0, 0x1a, 0x6a
+.byte	0xf1, 0xd0, 0x21, 0x78
+
+	.align 2
+
+rotate64_1bit_left:
+	bst r25, 7
+	rol r18
+	rol r19
+	rol r20
+	rol r21
+	rol r22
+	rol r23
+	rol r24
+	rol r25
+	bld r18, 0
+	ret
+
+rotate64_1bit_right:
+	bst r18, 0
+	ror r25
+	ror r24
+	ror r23
+	ror r22
+	ror r21
+	ror r20
+	ror r19
+	ror r18
+	bld r25, 7
+	ret
+
+rotate64_1byte_left:
+	mov r0, r25
+	mov r25, r24
+	mov r24, r23
+	mov r23, r22
+	mov r22, r21
+	mov r21, r20
+	mov r20, r19
+	mov r19, r18
+	mov r18, r0
+	ret
+
+rotate64_2byte_left:
+	movw r0, r24
+	movw r24, r22
+	movw r22, r20
+	movw r20, r18
+	movw r18, r0
+	ret
+
+rotate64_3byte_left:
+	mov r0, r25
+	mov r25, r22
+	mov r22, r19
+	mov r19, r24
+	mov r24, r21
+	mov r21, r18
+	mov r18, r23
+	mov r23, r20
+	mov r20, r0
+	ret
+
+rotate64_4byte_left:
+	movw r0, r24
+	movw r24, r20
+	movw r20, r0
+	movw r0, r22
+	movw r22, r18
+	movw r18, r0
+	ret
+
+rotate64_5byte_left:
+	mov r0, r25
+	mov r25, r20
+	mov r20, r23
+	mov r23, r18
+	mov r18, r21
+	mov r21, r24
+	mov r24, r19
+	mov r19, r22
+	mov r22, r0
+	ret
+
+rotate64_6byte_left:
+	movw r0, r18
+	movw r18, r20
+	movw r20, r22
+	movw r22, r24
+	movw r24, r0
+	ret
+
+rotate64_7byte_left:
+	mov r0, r18
+	mov r18, r19
+	mov r19, r20
+	mov r20, r21
+	mov r21, r22
+	mov r22, r23
+	mov r23, r24
+	mov r24, r25
+	mov r25, r0
+
+byte_rot_jmp_table:
+	ret
+	rjmp rotate64_1byte_left
+	rjmp rotate64_2byte_left
+	rjmp rotate64_3byte_left
+	rjmp rotate64_4byte_left
+	rjmp rotate64_5byte_left
+	rjmp rotate64_6byte_left
+	rjmp rotate64_7byte_left
+
+
 /*
 	void keccak_theta (uint64_t *a, uint64_t *b){
 	// uint64_t b[5][5];
@@ -166,14 +324,35 @@ chi_step:
 	brne 10b
 	ret
 
-.global keccak_theta
-keccak_theta:
-	push_range 2, 8
+	.global keccak_nextBlock
+	.func keccak_nextBlock
+keccak_nextBlock:
+	movw ZL, r24
+	subi ZL, lo8(-ctx_bs)
+	sbci ZL, hi8(-ctx_bs)
+	ld r20, Z
+	movw XL, r24
+	movw ZL, r22
+10:
+	ld r22, X
+	ld r23, Z+
+	eor r22, r23
+	st X+, r22
+	dec r20
+	brne 10b
+
+	.global keccak_f1600
+keccak_f1600:
+	push_range 2, 9
 	push r16
 	push_range 28, 29
 
+	stack_alloc_large 200, r26, r27
+	adiw XL, 1
+
+	clr r9
+5:
 	movw r30, r24 ; Z = a
-	movw r26, r22 ; X = b
 
 	ldi r19, 5
 10:
@@ -255,7 +434,7 @@ keccak_theta:
 
 ;	ret
 /*
-  rho & pi
+   -- rho & pi --
 	for(i = 0; i < 5; ++i){
 		for(j = 0; j < 5; ++j){
 			b[(2 * i + 3 * j) % 5][j] =
@@ -305,7 +484,34 @@ keccak_theta:
 	movw ZL, r2
 	lpm r16, Z+
 	movw r2, ZL
-	call rotate64left_code
+rotate64left_code:
+	ldi r30, pm_lo8(byte_rot_jmp_table)
+	ldi r31, pm_hi8(byte_rot_jmp_table)
+	mov r0, r16
+	andi r16, 0x70
+	swap r16
+	add r30, r16
+	adc r31, r1
+	mov r16, r0
+	andi r16, 0x0f
+	icall
+	clr r1
+rotate64_nbit_autodir:
+	lsr r16
+	brcc rotate64_nbit_left
+rotate64_nbit_right:
+	ldi r30, pm_lo8(rotate64_1bit_right)
+	ldi r31, pm_hi8(rotate64_1bit_right)
+	rjmp icall_r16_times
+rotate64_nbit_left:
+	ldi r30, pm_lo8(rotate64_1bit_left)
+	ldi r31, pm_hi8(rotate64_1bit_left)
+icall_r16_times:
+1:	dec r16
+	brmi 2f
+	icall
+	rjmp 1b
+2:
 	movw ZL, r4
 	lpm r16, Z+
 	movw r4, ZL
@@ -350,6 +556,7 @@ keccak_theta:
 
 	; Z points at b
 	movw XL, ZL
+	movw r4, ZL
 	adiw XL, 8
 	adiw ZL, 16
 	movw YL, r2
@@ -366,8 +573,51 @@ keccak_theta:
 	adiw ZL, 5 * 8
 	dec r18
 	brne 10b
+
+	/* -- iota -- */
+	ldi r30, lo8(keccak_rc_comp)
+	ldi r31, hi8(keccak_rc_comp)
+	add r30, r9
+	adc r31, __zero_reg__
+	lpm r20, Z+
+	movw YL, r2
+	ldi r21, 0x80
+	bst r20, 6
+	brtc 10f
+	ldd r22, Y+7
+	eor r22, r21
+	std Y+7, r22
+10:
+	bst r20, 5
+	brtc 10f
+	ldd r22, Y+3
+	eor r22, r21
+	std Y+3, r22
+10:
+	bst r20, 4
+	brtc 10f
+	ldd r22, Y+1
+	eor r22, r21
+	std Y+1, r22
+10:
+	andi r20, 0x8f
+	ld r22, Y
+	eor r22, r20
+	st Y, r22
+
+	inc r9
+	mov r16, r9
+	cpi r16, 24
+	breq 20f
+	movw r24, YL
+	movw r26, r4
+	rjmp 5b
+20:
+
+	stack_free_large3 200
+
 	pop_range 28, 29
 	pop r16
-	pop_range 2, 8
+	pop_range 2, 9
 
 	ret