[keccak-asm] rho & pi in asm

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

diff --git a/keccak/keccak-asm.S b/keccak/keccak-asm.S
index 9ccd79e0d8c9d3d7c6dafdb0c09b119620276347..95d8aa6289ff484ec781862a4f1d2208e0245f16 100644 (file)
--- a/keccak/keccak-asm.S
+++ b/keccak/keccak-asm.S
@@ -30,6 +30,16 @@
 .list
 
 .equ __zero_reg__, 1
+
+.global rho_pi_idx_table
+rho_pi_idx_table:
+       .irp i, 0, 1, 2, 3, 4
+               .irp j, 0, 1, 2, 3, 4
+                       .byte (((2 * \j + 3 * \i) % 5) * 5 + \i) * 8
+               .endr
+       .endr
+
+.align 2
 /*
        void keccak_theta (uint64_t *a, uint64_t *b){
        // uint64_t b[5][5];
@@ -128,8 +138,14 @@ theta_2b:
 
 .global keccak_theta
 keccak_theta:
+       push_range 2, 8
+       push r16
+       push_range 28, 29
+
        movw r30, r24 ; Z = a
        movw r26, r22 ; X = b
+       push_range 22, 25
+
        ldi r19, 5
 10:
        ldi r20, 8
@@ -208,5 +224,79 @@ keccak_theta:
        sbci XH, hi8(4 * 5 * 8 + 8)
        rcall theta_2b
 
+;      ret
+/*
+  rho & pi
+       for(i = 0; i < 5; ++i){
+               for(j = 0; j < 5; ++j){
+                       b[(2 * i + 3 * j) % 5][j] =
+              rotate64left_code(a[j][i], pgm_read_byte(&(keccak_rotate_codes[i][j])));
+               }
+       }
+
+   -- or --
+
+       const uint8_t* rot_code = (const uint8_t*)keccak_rotate_codes;
+    const uint8_t* idx_idx = (const uint8_t*)rho_pi_idx_table;
+    uint64_t *a_tmp = (uint64_t*)a;
+       for(i = 0; i < 25; ++i){
+                   *((uint64_t*)(((uint8_t*)b) + pgm_read_byte(idx_idx++))) =
+                rotate64left_code(*a_tmp++, pgm_read_byte(rot_code++));
+
+       }
+
+*/
+
+.equ B_REG_L, 6
+.equ B_REG_H, 7
+
+       ldi r18, lo8(keccak_rotate_codes)
+       ldi r19, hi8(keccak_rotate_codes)
+       movw r2, r18
+       ldi r18, lo8(rho_pi_idx_table)
+       ldi r19, hi8(rho_pi_idx_table)
+       movw r4, r18
+       ldi r16, 25
+       mov r8, r16
+
+       pop YH
+       pop YL
+       pop B_REG_H
+       pop B_REG_L
+
+10:
+       ld r18, Y+
+       ld r19, Y+
+       ld r20, Y+
+       ld r21, Y+
+       ld r22, Y+
+       ld r23, Y+
+       ld r24, Y+
+       ld r25, Y+
+       movw ZL, r2
+       lpm r16, Z+
+       movw r2, ZL
+       call rotate64left_code
+       movw ZL, r4
+       lpm r16, Z+
+       movw r4, ZL
+       movw XL, B_REG_L
+       add XL, r16
+       adc XH, __zero_reg__
+       st X+, r18
+       st X+, r19
+       st X+, r20
+       st X+, r21
+       st X+, r22
+       st X+, r23
+       st X+, r24
+       st X+, r25
+
+       dec r8
+       brne 10b
+
+       pop_range 28, 29
+       pop r16
+       pop_range 2, 8
 
        ret