further shrinked bmw tiny to 1778 bytes

diff --git a/bmw/bmw_large.c b/bmw/bmw_large.c
index 5b0039ab31663e81a3864a660a1ea311f2da1e26..719b5859c0199c154a2cc4a861f0a5be1ad00091 100644 (file)
--- a/bmw/bmw_large.c
+++ b/bmw/bmw_large.c
@@ -30,10 +30,10 @@
 #include <avr/pgmspace.h>
 #include "bmw_large.h"
 
-#define SHL64(a,n) ((a)<<(n))
-#define SHR64(a,n) ((a)>>(n))
-#define ROTL64(a,n) (((a)<<(n))|((a)>>(64-(n))))
-#define ROTR64(a,n) (((a)>>(n))|((a)<<(64-(n))))
+#define SHL64(a,n)  shiftl64(a,n)
+#define SHR64(a,n)  shiftr64(a,n)
+#define ROTL64(a,n) rotl64(a,n)
+#define ROTR64(a,n) rotr64(a,n)
 
 #define TWEAK   1
 #define BUG24   0
@@ -75,6 +75,26 @@
  #define dump_x(a,b,c)
 #endif
 
+static
+uint64_t rotl64(uint64_t a, uint8_t r){
+       return (a<<r)|(a>>(64-r));
+}
+
+static
+uint64_t rotr64(uint64_t a, uint8_t r){
+       return (a>>r)|(a<<(64-r));
+}
+
+static
+uint64_t shiftl64(uint64_t a, uint8_t r){
+       return (a<<r);
+}
+
+static
+uint64_t shiftr64(uint64_t a, uint8_t r){
+       return (a>>r);
+}
+
 static
 uint64_t bmw_large_s0(uint64_t x){
        uint64_t r;

diff --git a/bmw/bmw_small-tinyasm.S b/bmw/bmw_small-tinyasm.S
index 801331fbf3f9422c54debe4d0e590cd3131f2f3d..fca301a39e093fa83d89ba06059d4e1697c55352 100644 (file)
--- a/bmw/bmw_small-tinyasm.S
+++ b/bmw/bmw_small-tinyasm.S
@@ -109,13 +109,22 @@ s2:  .byte 2, 1,12,25
 s3:  .byte 2, 2,15,29
 s4:  .byte 1, 0, 0, 0
 s5:  .byte 2, 0, 0, 0
-
-eor_r22_in_r16:
-       eor r16, r22
-       eor r17, r23
-       eor r18, r24
-       eor r19, r25
-       ret
+/*
+s0:  .byte 0x34, 19
+s1:  .byte 0x28, 23
+s2:  .byte 0x9C, 25
+s3:  .byte 0xAF, 29
+s4:  .byte 0x00,  0
+s5:  .byte 0x80,  0
+*/
+acc2 =  8
+acc3 =  9
+h0   = 10
+h1   = 11
+m0   = 12
+m1   = 13
+acc0 = 14
+acc1 = 15
 
 /*
   param x: r22:r23:r24:25
@@ -123,8 +132,10 @@ eor_r22_in_r16:
 */
 sn:
        push_range 2, 5
-       push r17
-       push r19
+       push acc0
+       push acc1
+       push acc2
+       push acc3
        ldi r30, lo8(s_table)
        ldi r31, hi8(s_table)
        lsl r20
@@ -135,30 +146,32 @@ sn:
        movw r4, r24
        lpm r20, Z+
        rcall shiftright32
-       movw r16, r22
-       movw r18, r24
+       movw acc0, r22
+       movw acc2, r24
 ;---
        movw r22, r2
        movw r24, r4
        lpm r20, Z+
        rcall shiftleft32
-       rcall eor_r22_in_r16
+       rcall eor32_to_acc
 ;---
        movw r22, r2
        movw r24, r4
        lpm r20, Z+
        rcall rotateleft32
-       rcall eor_r22_in_r16
+       rcall eor32_to_acc
 ;---
        movw r22, r2
        movw r24, r4
        lpm r20, Z+
        rcall rotateleft32
-       rcall eor_r22_in_r16
-       movw r22, r16
-       movw r24, r18
-       pop r19
-       pop r17
+       rcall eor32_to_acc
+       movw r22, acc0
+       movw r24, acc2
+       pop acc3
+       pop acc2
+       pop acc1
+       pop acc0
        pop_range 2, 5
        ret
 
@@ -222,6 +235,21 @@ add_X_to_32:
        ld r0, X+
        adc r25, r0
        ret
+
+store_acc_to_dec_X:
+       st -X, acc3
+       st -X, acc2
+       st -X, acc1
+       st -X, acc0
+       ret
+
+store32_to_X:
+       st X+, r22
+       st X+, r23
+       st X+, r24
+       st X+, r25
+       ret
+
 /******************************************************************************/
 /*
   param q:  r28:r29 (Y)
@@ -230,28 +258,13 @@ add_X_to_32:
 */
 
 f0_hacktable:
-       .byte 0x03, 0x11
-       .byte 0xDD, 0xB3
-       .byte 0x2A, 0x79
-       .byte 0x07, 0xAA
-       .byte 0x51, 0xC2
-f0_indextable:
-       .byte 5*4,7*4,10*4,13*4,14*4
-;      .byte 0 ; just for alignment
-f0_s_table:
-       .byte 0,1,2,3,4
-       .byte 0,1,2,3,4
-       .byte 0,1,2,3,4
-;      .byte 0
-
-
-/******************************************************************************/
+       .byte 0x03, 0x11, 5*4
+       .byte 0xDD, 0xB3, 7*4
+       .byte 0x2A, 0x79, 10*4
+       .byte 0x07, 0xAA, 13*4
+       .byte 0x51, 0xC2, 14*4
+       .byte 0 ; just for alignment
 
-const_lut:
-       .long 0x55555550, 0x5aaaaaa5, 0x5ffffffa, 0x6555554f
-       .long 0x6aaaaaa4, 0x6ffffff9, 0x7555554e, 0x7aaaaaa3
-       .long 0x7ffffff8, 0x8555554d, 0x8aaaaaa2, 0x8ffffff7
-       .long 0x9555554c, 0x9aaaaaa1, 0x9ffffff6, 0xa555554b
 
 /*******************************************************************************
 * uint32_t addelment(uint8_t j, const uint32_t* m, const uint32_t* h){
@@ -337,17 +350,16 @@ addelement:
        mov j, r24
        movw h0, r20
        movw m0, r22
-       lsl r24
-       lsl r24
-       mov r28, r24
-       ldi r30, lo8(const_lut)
-       ldi r31, hi8(const_lut)
-       add r30, r24
-       adc r31, r1
-       lpm acc0, Z+
-       lpm acc1, Z+
-       lpm acc2, Z+
-       lpm acc3, Z+
+       sbiw r26, 4
+       rcall load_acc_from_X
+       ldi r24, 0x55
+       add acc0, r24
+       adc acc1, r24
+       adc acc2, r24
+       ldi r24, 5
+       adc acc3, r24
+       rcall store_acc_to_dec_X
+       adiw r26, 4
        clt
        mov r20, j
        rcall load_rotate_add_M
@@ -365,18 +377,18 @@ addelement:
        movw r26, h0
        add r26, j
        adc r27, r1
-       ld r0, X+
-       eor acc0, r0
-       ld r0, X+
-       eor acc1, r0
-       ld r0, X+
-       eor acc2, r0
-       ld r0, X+
-       eor acc3, r0
+       rcall load32_from_X
+       rcall eor32_to_acc
 ;---
        ret
 
 /******************************************************************************/
+load_sn_add:
+       rcall load32_from_X
+       rcall sn
+       rcall add32_to_acc
+       ret
+
 /*
   param q: r26:r27
   param m: r22:r23
@@ -399,11 +411,9 @@ expand1:
        rcall expand_intro
        ldi r19, 1
 10:
-       rcall load32_from_X
        mov r20, r19
        andi r20, 3
-       rcall sn
-       rcall add32_to_acc
+       rcall load_sn_add
        inc r19
        cpi r19, 17
        brne 10b
@@ -418,8 +428,14 @@ expand1:
   param j: r24
 */
 
+f2_1_shift_table:
+       .byte 0x2B, 0x64, 0x66, 0x03, 0x51, 0x55, 0x87, 0x55
+f2_2_shift_table:
+       .byte (2<<1), (7<<1), (4<<1), (3<<1), (4<<1)+1, (6<<1)+1, (6<<1)
+
 expand2_rot_table:
-       .byte 3,7,13,16,19,23,27,0
+       .byte 3,7,13,16,19,23,27
+;      .byte 0 ; just for alignment
 
 expand2:
        rcall expand_intro
@@ -436,18 +452,12 @@ expand2:
        dec r19
        brne 10b
        ldi r20, 4
-       rcall load32_from_X
-       rcall sn
-       rcall add32_to_acc
+       rcall load_sn_add
        ldi r20, 5
-       rcall load32_from_X
-       rcall sn
-       rcall add32_to_acc
+       rcall load_sn_add
 expand2_exit:
-       st X+, acc0
-       st X+, acc1
-       st X+, acc2
-       st X+, acc3
+       adiw r26, 4
+       rcall store_acc_to_dec_X
        ret
 
 /******************************************************************************/
@@ -469,12 +479,6 @@ expand2_exit:
   param m: r22:r23
   param h: r20:r21
 */
-f2_1_shift_table:
-       .byte 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00
-       .byte 0x2B, 0x64, 0x66, 0x03, 0x51, 0x55, 0x87, 0x55
-f2_2_shift_table:
-       .byte (2<<1), (7<<1), (4<<1), (3<<1), (4<<1)+1, (6<<1)+1, (6<<1)
-       .byte 0 ; just for alignment
 
 /******************************************************************************/
 /*
@@ -513,6 +517,13 @@ bmw256_nextBlock:
        push_range 28, 29
        push_range  2, 17
        stack_alloc_large 32*4, r28, r29
+       ldi r16, 0x4f
+       push r16
+       ldi r16, 0xff
+       push r16
+       push r16
+       ldi r16, 0xfb
+       push r16
        adiw r28, 1
 ;      push_range 28, 29 /* push Q */
 ;      push_range 22, 25 /* push M & H */
@@ -527,10 +538,7 @@ bmw256_nextBlock:
        adc acc1, r1
        adc acc2, r1
        adc acc3, r1
-       st -X, acc3
-       st -X, acc2
-       st -X, acc1
-       st -X, acc0
+       rcall store_acc_to_dec_X
        /* call f0 */
        movw r30, r22
        movw r26, r24
@@ -555,20 +563,18 @@ f0:
 30:
        ldi r18, 16
        /* load initial index */
-       ldi r30, lo8(f0_indextable-1)
-       ldi r31, hi8(f0_indextable-1)
-       add r30, r19
-       adc r31, r1
-       lpm r16, Z
+
        /* load values from hacktable */
-       ldi r30, lo8(f0_hacktable-2)
-       ldi r31, hi8(f0_hacktable-2)
-       lsl r19
-       add r30, r19
+       ldi r30, lo8(f0_hacktable-3)
+       ldi r31, hi8(f0_hacktable-3)
+       mov r16, r19
+       lsl r16
+       add r16, r19
+       add r30, r16
        adc r31, r1
-       lsr r19
        lpm r21, Z+
-       lpm r20, Z
+       lpm r20, Z+
+       lpm r16, Z+
 40:
        ;call add_hx_to_w
 add_hx_to_w:
@@ -584,14 +590,12 @@ add_hx_to_w:
        rcall add_X_to_32
        rjmp 500f
 300: /* substract */
-       ld r0, X+
-       sub r22, r0
-       ld r0, X+
-       sbc r23, r0
-       ld r0, X+
-       sbc r24, r0
-       ld r0, X+
-       sbc r25, r0
+       rcall load_acc_from_X
+       sub r22, acc0
+       sbc r23, acc1
+       sbc r24, acc2
+       sbc r25, acc3
+
 500:
        rcall store32_to_Y
        subi r16, -4
@@ -609,18 +613,19 @@ add_hx_to_w:
        rcall memxor_short
        sbiw r26, 60
 ;---
-       ldi r30, lo8(f0_s_table)
-       ldi r31, hi8(f0_s_table)
+       clr r17
        ldi r21, 15
        mov r8, r21
 50:
        rcall load32_from_Y
        sbiw r28, 4
-       lpm r20, Z+
-       movw r2, r30
+       mov r20, r17
        rcall sn
-       movw r30, r2
-
+       inc r17
+       cpi r17, 5
+       brne 52f
+       clr r17
+52:
        rcall add_X_to_32
        rcall store32_to_Y
 
@@ -630,22 +635,19 @@ add_hx_to_w:
        rcall load32_from_Y
        clr r20
        rcall sn
-       movw r30, r2
        movw r26, h0
        rcall add_X_to_32
        sbiw r26, 4
-       st -Y, r25
-       st -Y, r24
-       st -Y, r23
-       st -Y, r22
+       sbiw r28, 4
+       rcall store32_to_Y
+       sbiw r28, 4
        sbiw r28, 15*4
        movw r20, h0
        movw r22, m0
 
        /* call f1*/
-       movw r24, r28
+       movw r2, r28
 f1:
-       movw r2, r24
        movw r4, r22
        movw r6, r20
        movw r26, r2
@@ -735,10 +737,14 @@ f2:
        ld acc2, Y+
        ld acc3, Y+
 ;---
-       ldi r30, lo8(f2_1_shift_table-1)
-       ldi r31, hi8(f2_1_shift_table-1)
        movw r22, xh0
        movw r24, xh2
+       cpi r17, 9
+       brge 15f
+       clr r1
+       rjmp 26f
+15:    ldi r30, lo8(f2_1_shift_table-9)
+       ldi r31, hi8(f2_1_shift_table-9)
        add r30, r17
        adc r31, r1
        lpm r20, Z
@@ -854,7 +860,7 @@ f2:
 ;      ldi r22, 'H'
 ;      rcall printX
 ;--- END DBG
-       stack_free_large3 32*4
+       stack_free_large3 32*4+4
        pop_range  2, 17
        pop_range 28, 29
        ret
@@ -1021,16 +1027,15 @@ bmw256_lastBlock:
                pctx.buffer[i*4] = i+0xa0;
        }
 */
-       ldi r18, 0xa0
-       ldi r19, 0xaa
+       ldi r22, 0xa0
+       ldi r23, 0xaa
+       ldi r24, 0xaa
+       ldi r25, 0xaa
        movw r26, buf0
 500:
-       st X+, r18
-       st X+, r19
-       st X+, r19
-       st X+, r19
-       inc r18
-       sbrs r18, 4
+       rcall store32_to_X
+       inc r22
+       sbrs r22, 4
        rjmp 500b
 /*     bmw_small_nextBlock((bmw_small_ctx_t*)&pctx, ctx->h);
        memcpy(ctx->h, pctx.buffer, 64);
@@ -1230,10 +1235,7 @@ bmw224_init:
        ldi r24, 0x01
        ldi r25, 0x00
 bmw_small_init:
-       st X+, r22
-       st X+, r23
-       st X+, r24
-       st X+, r25
+       rcall store32_to_X
        ldi r18, 16-1
        ldi r20, 0x04
 1:
@@ -1241,10 +1243,7 @@ bmw_small_init:
        adc r23, r20
        adc r24, r20
        adc r25, r20
-       st X+, r22
-       st X+, r23
-       st X+, r24
-       st X+, r25
+       rcall store32_to_X
        dec r18
        brne 1b
        st X+, r1