[keccak-asm] keccak_f1600 in asm

[avr-crypto-lib.git] / keccak / keccak-asm.S

diff --git a/keccak/keccak-asm.S b/keccak/keccak-asm.S
index 95d8aa6289ff484ec781862a4f1d2208e0245f16..357c8bd028fd36d6a1baa56358c42978ff22157c 100644 (file)
--- a/keccak/keccak-asm.S
+++ b/keccak/keccak-asm.S
@@ -135,16 +135,49 @@ theta_2b:
        clr __zero_reg__
        ret
 
+;      a[i][j] =  b[i][j] ^ ((~(b[i][(j + 1) % 5])) & (b[i][(j + 2) % 5]));
 
-.global keccak_theta
-keccak_theta:
-       push_range 2, 8
+/*********************************************
+ * chi_step
+ *********************************************
+       input:
+               Y = a; uint8t *a;
+               X = b; uint8t *b;
+               Z = c; uint8t *c;
+       output:
+               a[0..7] ^= ~b[0..7] & c[0..7]
+               X += 8
+               Y += 8
+               Z += 8
+               r16 = 0
+               trash r21, r22, r23
+*/
+chi_step:
+       ldi r16, 8
+10:
+       ld r21, Y
+       ld r22, X+
+       ld r23, Z+
+       com r22
+       and r22, r23
+       eor r21, r22
+       st Y+, r21
+       dec r16
+       brne 10b
+       ret
+
+.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
-       push_range 22, 25
 
        ldi r19, 5
 10:
@@ -259,10 +292,10 @@ keccak_theta:
        ldi r16, 25
        mov r8, r16
 
-       pop YH
-       pop YL
-       pop B_REG_H
-       pop B_REG_L
+       sbiw r24, 5 * 8 + 1
+       movw YL, r24
+       sbiw XL, 8
+       movw B_REG_L, XL
 
 10:
        ld r18, Y+
@@ -294,9 +327,95 @@ keccak_theta:
 
        dec r8
        brne 10b
+/*
+       -- chi --
+       for(i = 0; i < 5; ++i){
+        a[i][0] ^= ((~(b[i][1])) & (b[i][2]));
+        a[i][1] ^= ((~(b[i][2])) & (b[i][3]));
+        a[i][2] ^= ((~(b[i][3])) & (b[i][4]));
+        a[i][3] ^= ((~(b[i][4])) & (b[i][0]));
+        a[i][4] ^= ((~(b[i][0])) & (b[i][1]));
+
+       }
+*/
+       ; memcpy(a, b, 200)
+       ; X points at b + 32 + 8 = b + 40 = b[1][0] has to point to b[0][0]
+       ldi r16, 200
+       sbiw XL, 5 * 8
+       movw ZL, XL
+       subi YL, lo8(5 * 5 * 8)
+       sbci YH, hi8(5 * 5 * 8)
+       movw r2, YL
+10:
+       ld r22, X+
+       st Y+, r22
+       dec r16
+       brne 10b
+
+       ; Z points at b
+       movw XL, ZL
+       movw r4, ZL
+       adiw XL, 8
+       adiw ZL, 16
+       movw YL, r2
+       ldi r18, 5
+10:
+       rcall chi_step
+       rcall chi_step
+       rcall chi_step
+       sbiw ZL, 5 * 8
+       rcall chi_step
+       sbiw XL, 5 * 8
+       rcall chi_step
+       adiw XL, 5 * 8
+       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