further shrinked bmw tiny to 1648 bytes

[avr-crypto-lib.git] / bmw / bmw_small-tinyasm.S

diff --git a/bmw/bmw_small-tinyasm.S b/bmw/bmw_small-tinyasm.S
index fca301a39e093fa83d89ba06059d4e1697c55352..50e110d82d9635c14a735ee176cc30d6e919b92f 100644 (file)
--- a/bmw/bmw_small-tinyasm.S
+++ b/bmw/bmw_small-tinyasm.S
@@ -28,6 +28,11 @@
 
 #include "avr-asm-macros.S"
 
+acc2 =  8
+acc3 =  9
+acc0 = 14
+acc1 = 15
+
 /******************************************************************************/
 /*
   param a: r22:r23:r24:r25
@@ -91,6 +96,7 @@ bitrotateleft_1:
        breq 20f
 10:
        lsl r0
+rol32:
        rol r22
        rol r23
        rol r24
@@ -102,6 +108,18 @@ bitrotateleft_1:
 
 /******************************************************************************/
 
+sn_stub:
+       movw r22, r2
+       movw r24, r4
+       lpm r20, Z+
+       rcall rotateleft32
+eor32_to_acc:
+       eor acc0, r22
+       eor acc1, r23
+       eor acc2, r24
+       eor acc3, r25
+       ret
+
 s_table:
 s0:  .byte 1, 3, 4,19
 s1:  .byte 1, 2, 8,23
@@ -109,22 +127,11 @@ s2:  .byte 2, 1,12,25
 s3:  .byte 2, 2,15,29
 s4:  .byte 1, 0, 0, 0
 s5:  .byte 2, 0, 0, 0
-/*
-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
@@ -146,8 +153,7 @@ sn:
        movw r4, r24
        lpm r20, Z+
        rcall shiftright32
-       movw acc0, r22
-       movw acc2, r24
+       rcall mov32_to_acc
 ;---
        movw r22, r2
        movw r24, r4
@@ -155,25 +161,16 @@ sn:
        rcall shiftleft32
        rcall eor32_to_acc
 ;---
-       movw r22, r2
-       movw r24, r4
-       lpm r20, Z+
-       rcall rotateleft32
-       rcall eor32_to_acc
-;---
-       movw r22, r2
-       movw r24, r4
-       lpm r20, Z+
-       rcall rotateleft32
-       rcall eor32_to_acc
+       rcall sn_stub
+       rcall sn_stub
+
        movw r22, acc0
        movw r24, acc2
        pop acc3
        pop acc2
        pop acc1
        pop acc0
-       pop_range 2, 5
-       ret
+       rjmp pop5
 
 /******************************************************************************/
 /*
@@ -236,13 +233,6 @@ add_X_to_32:
        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
@@ -250,6 +240,16 @@ store32_to_X:
        st X+, r25
        ret
 
+mov32_to_acc:
+       movw acc0, r22
+       movw acc2, r24
+       ret
+
+eor_acc_from_Y_add_to_Z:
+       rcall load32_from_Y
+       rcall eor32_to_acc
+       rjmp  add_acc_to_Z
+
 /******************************************************************************/
 /*
   param q:  r28:r29 (Y)
@@ -290,20 +290,6 @@ m1   = 13
 acc0 = 14
 acc1 = 15
 
-add32_to_acc:
-       add acc0, r22
-       adc acc1, r23
-       adc acc2, r24
-       adc acc3, r25
-       ret
-
-eor32_to_acc:
-       eor acc0, r22
-       eor acc1, r23
-       eor acc2, r24
-       eor acc3, r25
-       ret
-
 load_acc_from_X:
        ld acc0, X+
        ld acc1, X+
@@ -327,6 +313,7 @@ add_acc_to_Z:
        ret
 
 load_rotate_add_M:
+       mov r20, j
        andi r20, 0x0f
        mov r0, r20
        lsl r0
@@ -338,14 +325,38 @@ load_rotate_add_M:
        inc r20
        rcall rotateleft32
        brts 10f
-       rcall add32_to_acc
-       ret
+       rjmp add32_to_acc
+;      ret
 10:    sub acc0, r22
        sbc acc1, r23
        sbc acc2, r24
        sbc acc3, r25
        ret
 
+
+;---
+
+/******************************************************************************/
+load_sn_add:
+       rcall load32_from_X
+       rcall sn
+add32_to_acc:
+       add acc0, r22
+       adc acc1, r23
+       adc acc2, r24
+       adc acc3, r25
+       ret
+
+/*
+  param q: r26:r27
+  param m: r22:r23
+  param h: r20:r21
+  param j: r24
+*/
+
+expand_intro:
+       push_range 26, 27
+       push r24
 addelement:
        mov j, r24
        movw h0, r20
@@ -361,47 +372,24 @@ addelement:
        rcall store_acc_to_dec_X
        adiw r26, 4
        clt
-       mov r20, j
        rcall load_rotate_add_M
-       mov r20, j
-       subi r20, -3
+       subi j, -3
        rcall load_rotate_add_M
-       mov r20, j
        set
-       subi r20, -10
+       subi j, -7
        rcall load_rotate_add_M
        lsl j
        lsl j
-       subi j, -7*4
+       subi j, -7*4+10*4
        andi j, 0x3f
        movw r26, h0
        add r26, j
        adc r27, r1
        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
-  param h: r20:r21
-  param j: r24
-*/
-
-expand_intro:
-       push_range 20, 27
-;      push r24
-       rcall addelement
-;      pop r24
-       pop_range 20, 27
+;--
+       pop r24
+       pop_range 26, 27
        lsl r24
        lsl r24
        add r26, r24
@@ -457,7 +445,11 @@ expand2:
        rcall load_sn_add
 expand2_exit:
        adiw r26, 4
-       rcall store_acc_to_dec_X
+store_acc_to_dec_X:
+       st -X, acc3
+       st -X, acc2
+       st -X, acc1
+       st -X, acc0
        ret
 
 /******************************************************************************/
@@ -506,16 +498,28 @@ h0 = 4
 h1 = 5
 m0 = 6
 m1 = 7
+ctx0 =   2
+ctx1 =   3
+msg0 =   4
+msg1 =   5
 
-
+restore_f1:
+       movw r26, r2
+       movw r22, r4
+    movw r20, r6
+       ret
+bmw_small_nextBlock_early:
+       movw r24, ctx0
+       movw r22, msg0
 .global bmw_small_nextBlock
 .global bmw224_nextBlock
 .global bmw256_nextBlock
 bmw_small_nextBlock:
 bmw224_nextBlock:
 bmw256_nextBlock:
+       push_range  2, 7
        push_range 28, 29
-       push_range  2, 17
+       push_range  8, 17
        stack_alloc_large 32*4, r28, r29
        ldi r16, 0x4f
        push r16
@@ -653,23 +657,18 @@ f1:
        movw r26, r2
        clr r24
        rcall expand1
-       movw r26, r2
-       movw r22, r4
-    movw r20, r6
+       rcall restore_f1
        ldi r24, 1
        rcall expand1
        ldi r17, 2
-10:    movw r26, r2
-       movw r22, r4
-       movw r20, r6
+10: rcall restore_f1
        mov r24, r17
        rcall expand2
        inc r17
        sbrs r17, 4
        rjmp 10b
+       rcall restore_f1
        movw r24, r2
-       movw r22, r4
-       movw r20, r6
 
 
        /* call f2 */
@@ -703,8 +702,7 @@ f2:
        movw h0, r20
        movw r28, r22
        rcall load32_from_X
-       movw acc0, r22
-       movw acc2, r24
+       rcall mov32_to_acc
        ldi r17, 15
 10:    rcall load32_from_X
        rcall eor32_to_acc
@@ -732,10 +730,8 @@ f2:
        movw r26, q16_0
        ldi r17, 16
 10:
-       ld acc0, Y+
-       ld acc1, Y+
-       ld acc2, Y+
-       ld acc3, Y+
+       rcall load32_from_Y
+       rcall mov32_to_acc
 ;---
        movw r22, xh0
        movw r24, xh2
@@ -794,9 +790,7 @@ f2:
        movw acc2, xl2
        rcall load32_from_X
        rcall eor32_to_acc
-       rcall load32_from_Y
-       rcall eor32_to_acc
-       rcall add_acc_to_Z
+       rcall eor_acc_from_Y_add_to_Z
        dec r17
        brne 10b
        sbiw r26, 9*4 /* X points to q[23] */
@@ -804,9 +798,7 @@ f2:
        eor acc1, xl0
        eor acc2, xl1
        eor acc3, xl2
-       rcall load32_from_Y
-       rcall eor32_to_acc
-       rcall add_acc_to_Z
+       rcall eor_acc_from_Y_add_to_Z
 ;---
        sbiw r26, 8*4 /* X points to q[16] */
        mov h0, r30
@@ -826,11 +818,9 @@ f2:
        rjmp 21f
 20:    rcall shiftright32
 21:
-       rcall eor32_to_acc
-       rcall load32_from_Y
-       rcall eor32_to_acc
        movw r30, h0
-       rcall add_acc_to_Z
+       rcall eor32_to_acc
+       rcall eor_acc_from_Y_add_to_Z
        movw h0, r30
        dec r17
        brne 10b
@@ -844,8 +834,7 @@ f2:
        rcall load32_from_X
        mov r20, r18
        rcall rotateleft32
-       movw acc0, r22
-       movw acc2, r24
+       rcall mov32_to_acc
        rcall add_acc_to_Z
        inc r18
        cpi r17, 5
@@ -861,8 +850,15 @@ f2:
 ;      rcall printX
 ;--- END DBG
        stack_free_large3 32*4+4
-       pop_range  2, 17
+       pop_range 10, 17
+pop9:
+       pop_range 8, 9
+pop28:
        pop_range 28, 29
+pop7:
+       pop_range 6, 7
+pop5:
+       pop_range 2, 5
        ret
 
 /******************************************************************************/
@@ -911,9 +907,7 @@ bmw256_lastBlock:
 1:
        cpi len1, hi8(512)
        brlo 2f
-       movw r24, ctx0
-       movw r22, blc0
-       rcall bmw_small_nextBlock
+       rcall bmw_small_nextBlock_early
        ldi r24, 64
        add blc0, r24
        adc blc1, r1
@@ -932,15 +926,14 @@ bmw256_lastBlock:
        memcpy(pctx.buffer, block, (length_b+7)/8);
        pctx.buffer[length_b>>3] |= 0x80 >> (length_b&0x07);
 */     movw r24, len0
+       ldi r23, 63
+       movw r26, blc0
        lsr r25
        ror r24
        lsr r24
        lsr r24
-       ldi r23, 63
-       sub r23, r24
-       movw r26, blc0
-       tst r24
        breq 301f
+       sub r23, r24
        /* copy (#r24) bytes to stack buffer */
 30: ld r20, X+
        st Z+, r20
@@ -976,9 +969,8 @@ bmw256_lastBlock:
        breq 400f
        cpi len0, 192
        brlo 400f
-       movw r24, ctx0
-       movw r22, buf0
-       rcall bmw_small_nextBlock
+       movw blc0, buf0
+       rcall bmw_small_nextBlock_early
        movw r26, buf0
        ldi r20, 64-8
 350:
@@ -998,30 +990,24 @@ bmw256_lastBlock:
        rcall load32_from_Z_stub
 410:
        clr r25
+       ldi r20, 1
        lsl r21
-       rol r22
-       rol r23
-       rol r24
-       rol r25
+       rcall rol32
        mov r20, len0
        add r21, len1
        adc r22, r1
        adc r23, r1
        adc r24, r1
        adc r25, r1
-       movw r30, buf0
-       adiw r30, 64-8
-       st Z+, r20
-       st Z+, r21
-       st Z+, r22
-       st Z+, r23
-       st Z+, r24
-       st Z+, r25
-       st Z+, r1
-       st Z+, r1
-       movw r24, ctx0
-       movw r22, buf0
-       rcall bmw_small_nextBlock
+       movw r26, buf0
+       adiw r26, 64-8
+       st X+, r20
+       st X+, r21
+       rcall store32_to_X
+       st X+, r1
+       st X+, r1
+       movw blc0, buf0
+       rcall bmw_small_nextBlock_early
 /*     memset(pctx.buffer, 0xaa, 64);
        for(i=0; i<16;++i){
                pctx.buffer[i*4] = i+0xa0;
@@ -1040,9 +1026,9 @@ bmw256_lastBlock:
 /*     bmw_small_nextBlock((bmw_small_ctx_t*)&pctx, ctx->h);
        memcpy(ctx->h, pctx.buffer, 64);
 */
-       movw r24, buf0
-       movw r22, ctx0
-       rcall bmw_small_nextBlock
+    movw r24, buf0
+    movw r22, ctx0
+    rcall bmw_small_nextBlock
        ldi r18, 64
        movw r26, ctx0
        movw r30, buf0
@@ -1053,9 +1039,7 @@ bmw256_lastBlock:
        brne 600b
 
        stack_free_large 68
-       pop_range 28, 29
-       pop_range 2, 7
-       ret
+       rjmp pop28
 
 
 /*******************************************************************************
@@ -1068,10 +1052,9 @@ bmw256_lastBlock:
 */
 .global bmw224_ctx2hash
 bmw224_ctx2hash:
-       movw r26, r24
        movw r30, r22
        adiw r30, 9*4
-       ldi r22, 28
+       ldi r18, 28
        rjmp 1f
 
 /*******************************************************************************
@@ -1084,14 +1067,13 @@ bmw224_ctx2hash:
 */
 .global bmw256_ctx2hash
 bmw256_ctx2hash:
-       movw r26, r24
        movw r30, r22
        adiw r30, 8*4
-       ldi r22, 32
-1:
-       ld r23, Z+
+       ldi r18, 32
+1:     movw r26, r24
+1:  ld r23, Z+
        st X+, r23
-       dec r22
+       dec r18
        brne 1b
        ret
 
@@ -1124,10 +1106,10 @@ dst0 =  10
 dst1 =  11
 .global bmw256
 bmw256:
-       push r16
-       ldi r16, 1
+       set
        rjmp bmw_small_all
 
+
 /*******************************************************************************
 * void bmw224(void* dest, const void* msg, uint32_t length_b){
 *      bmw_small_ctx_t ctx;
@@ -1157,15 +1139,21 @@ dst0 =   6
 dst1 =   7
 .global bmw224
 bmw224:
-       push r16
-       clr r16
+       clt
+       rjmp bmw_small_all
+
 
 bmw_small_all:
-       push_range 2, 9
+       push_range 2, 7
        push_range 28, 29
+       push_range 8, 9
+       push r16
        stack_alloc_large 64+4
        adiw r30, 1
-       movw ctx0, r30
+       clr r16
+       brtc 10f
+       inc r16
+10:    movw ctx0, r30
        movw dst0, r24
        movw msg0, r22
        movw len0, r18
@@ -1180,9 +1168,7 @@ bmw_small_all:
        mov r18, len2
        or  r18, len3
        breq 50f
-       movw r24, ctx0
-       movw r22, msg0
-       rcall bmw_small_nextBlock
+       rcall bmw_small_nextBlock_early
        subi len1, 2
        sbc len2, r1
        sbc len3, r1
@@ -1203,10 +1189,8 @@ bmw_small_all:
        adc r31, r1
        icall
        stack_free_large 64+4
-       pop_range 28, 29
-       pop_range 2, 9
        pop r16
-       ret
+       rjmp pop9
 
 init_lut:
        rjmp bmw224_init
@@ -1229,36 +1213,30 @@ c2h_lut:
 */
 .global bmw224_init
 bmw224_init:
-       movw r26, r24
-       ldi r22, 0x03
-       ldi r23, 0x02
-       ldi r24, 0x01
-       ldi r25, 0x00
+       ldi r22, 0x00
+       ldi r23, 0x40
 bmw_small_init:
-       rcall store32_to_X
-       ldi r18, 16-1
-       ldi r20, 0x04
-1:
-       add r22, r20
-       adc r23, r20
-       adc r24, r20
-       adc r25, r20
-       rcall store32_to_X
-       dec r18
-       brne 1b
-       st X+, r1
-       st X+, r1
-       st X+, r1
-       st X+, r1
+       movw r26, r24
+       adiw r26, 4
+10:
+       st -X, r22
+       inc r22
+       mov r20, r22
+       andi r20, 0x3
+       brne 10b
+       adiw r26, 8
+20: cp r22, r23
+       brne 10b
+       st -X, r1
+       st -X, r1
+       st -X, r1
+       st -X, r1
        ret
 
 .global bmw256_init
 bmw256_init:
-       movw r26, r24
-       ldi r22, 0x43
-       ldi r23, 0x42
-       ldi r24, 0x41
-       ldi r25, 0x40
+       ldi r22, 0x40
+       ldi r23, 0x80
        rjmp bmw_small_init