/*
 * Fast SHA-256 implementation for SPE instruction set (PPC)
 *
 * This code makes use of the SPE SIMD instruction set as defined in
 * http://cache.freescale.com/files/32bit/doc/ref_manual/SPEPIM.pdf
 * Implementation is based on optimization guide notes from
 * http://cache.freescale.com/files/32bit/doc/app_note/AN2665.pdf
 *
 * Copyright (c) 2015 Markus Stockhausen <stockhausen@collogia.de>
 *
 * This program is free software; you can redistribute it and/or modify it
 * under the terms of the GNU General Public License as published by the Free
 * Software Foundation; either version 2 of the License, or (at your option)
 * any later version.
 *
 */

#include <asm/ppc_asm.h>
#include <asm/asm-offsets.h>

#define rHP	r3	/* pointer to hash values in memory		*/
#define rKP	r24	/* pointer to round constants			*/
#define rWP	r4	/* pointer to input data			*/

#define rH0	r5	/* 8 32 bit hash values in 8 registers		*/
#define rH1	r6
#define rH2	r7
#define rH3	r8
#define rH4	r9
#define rH5	r10
#define rH6	r11
#define rH7	r12

#define rW0	r14	/* 64 bit registers. 16 words in 8 registers	*/
#define rW1	r15
#define rW2	r16
#define rW3	r17
#define rW4	r18
#define rW5	r19
#define rW6	r20
#define rW7	r21

#define rT0	r22	/* 64 bit temporaries 				*/
#define rT1	r23
#define rT2	r0	/* 32 bit temporaries				*/
#define rT3	r25

#define CMP_KN_LOOP
#define CMP_KC_LOOP \
	cmpwi		rT1,0;

#define INITIALIZE \
	stwu		r1,-128(r1);	/* create stack frame		*/ \
	evstdw		r14,8(r1);	/* We must save non volatile	*/ \
	evstdw		r15,16(r1);	/* registers. Take the chance	*/ \
	evstdw		r16,24(r1);	/* and save the SPE part too	*/ \
	evstdw		r17,32(r1);					   \
	evstdw		r18,40(r1);					   \
	evstdw		r19,48(r1);					   \
	evstdw		r20,56(r1);					   \
	evstdw		r21,64(r1);					   \
	evstdw		r22,72(r1);					   \
	evstdw		r23,80(r1);					   \
	stw		r24,88(r1);	/* save normal registers	*/ \
	stw		r25,92(r1);


#define FINALIZE \
	evldw		r14,8(r1);	/* restore SPE registers	*/ \
	evldw		r15,16(r1);					   \
	evldw		r16,24(r1);					   \
	evldw		r17,32(r1);					   \
	evldw		r18,40(r1);					   \
	evldw		r19,48(r1);					   \
	evldw		r20,56(r1);					   \
	evldw		r21,64(r1);					   \
	evldw		r22,72(r1);					   \
	evldw		r23,80(r1);					   \
	lwz		r24,88(r1);	/* restore normal registers	*/ \
	lwz		r25,92(r1);					   \
	xor		r0,r0,r0;					   \
	stw		r0,8(r1);	/* Delete sensitive data	*/ \
	stw		r0,16(r1);	/* that we might have pushed	*/ \
	stw		r0,24(r1);	/* from other context that runs	*/ \
	stw		r0,32(r1);	/* the same code. Assume that	*/ \
	stw		r0,40(r1);	/* the lower part of the GPRs	*/ \
	stw		r0,48(r1);	/* was already overwritten on	*/ \
	stw		r0,56(r1);	/* the way down to here		*/ \
	stw		r0,64(r1);					   \
	stw		r0,72(r1);					   \
	stw		r0,80(r1);					   \
	addi		r1,r1,128;	/* cleanup stack frame		*/

#ifdef __BIG_ENDIAN__
#define LOAD_DATA(reg, off) \
	lwz		reg,off(rWP);	/* load data			*/
#define NEXT_BLOCK \
	addi		rWP,rWP,64;	/* increment per block		*/
#else
#define LOAD_DATA(reg, off) \
	lwbrx		reg,0,rWP; 	/* load data			*/ \
	addi		rWP,rWP,4;	/* increment per word		*/
#define NEXT_BLOCK			/* nothing to do		*/
#endif

#define R_LOAD_W(a, b, c, d, e, f, g, h, w, off) \
	LOAD_DATA(w, off)		/* 1: W				*/ \
	rotrwi		rT0,e,6;	/* 1: S1 = e rotr 6		*/ \
	rotrwi		rT1,e,11;	/* 1: S1' = e rotr 11		*/ \
	rotrwi		rT2,e,25;	/* 1: S1" = e rotr 25		*/ \
	xor		rT0,rT0,rT1;	/* 1: S1 = S1 xor S1'		*/ \
	and		rT3,e,f;	/* 1: ch = e and f		*/ \
	xor		rT0,rT0,rT2;	/* 1: S1 = S1 xor S1"		*/ \
	andc		rT1,g,e;	/* 1: ch' = ~e and g		*/ \
	lwz		rT2,off(rKP);	/* 1: K				*/ \
	xor		rT3,rT3,rT1;	/* 1: ch = ch xor ch'		*/ \
	add		h,h,rT0;	/* 1: temp1 = h + S1		*/ \
	add		rT3,rT3,w;	/* 1: temp1' = ch + w		*/ \
	rotrwi		rT0,a,2;	/* 1: S0 = a rotr 2		*/ \
	add		h,h,rT3;	/* 1: temp1 = temp1 + temp1'	*/ \
	rotrwi		rT1,a,13;	/* 1: S0' = a rotr 13		*/ \
	add		h,h,rT2;	/* 1: temp1 = temp1 + K		*/ \
	rotrwi		rT3,a,22;	/* 1: S0" = a rotr 22		*/ \
	xor		rT0,rT0,rT1;	/* 1: S0 = S0 xor S0'		*/ \
	add		d,d,h;		/* 1: d = d + temp1		*/ \
	xor		rT3,rT0,rT3;	/* 1: S0 = S0 xor S0"		*/ \
	evmergelo	w,w,w;		/*    shift W			*/ \
	or		rT2,a,b;	/* 1: maj = a or b		*/ \
	and		rT1,a,b;	/* 1: maj' = a and b		*/ \
	and		rT2,rT2,c;	/* 1: maj = maj and c		*/ \
	LOAD_DATA(w, off+4)		/* 2: W				*/ \
	or		rT2,rT1,rT2;	/* 1: maj = maj or maj'		*/ \
	rotrwi		rT0,d,6;	/* 2: S1 = e rotr 6		*/ \
	add		rT3,rT3,rT2;	/* 1: temp2 = S0 + maj		*/ \
	rotrwi		rT1,d,11;	/* 2: S1' = e rotr 11		*/ \
	add		h,h,rT3;	/* 1: h = temp1 + temp2		*/ \
	rotrwi		rT2,d,25;	/* 2: S1" = e rotr 25		*/ \
	xor		rT0,rT0,rT1;	/* 2: S1 = S1 xor S1'		*/ \
	and		rT3,d,e;	/* 2: ch = e and f		*/ \
	xor		rT0,rT0,rT2;	/* 2: S1 = S1 xor S1"		*/ \
	andc		rT1,f,d;	/* 2: ch' = ~e and g		*/ \
	lwz		rT2,off+4(rKP);	/* 2: K				*/ \
	xor		rT3,rT3,rT1;	/* 2: ch = ch xor ch'		*/ \
	add		g,g,rT0;	/* 2: temp1 = h + S1		*/ \
	add		rT3,rT3,w;	/* 2: temp1' = ch + w		*/ \
	rotrwi		rT0,h,2;	/* 2: S0 = a rotr 2		*/ \
	add		g,g,rT3;	/* 2: temp1 = temp1 + temp1'	*/ \
	rotrwi		rT1,h,13;	/* 2: S0' = a rotr 13		*/ \
	add		g,g,rT2;	/* 2: temp1 = temp1 + K		*/ \
	rotrwi		rT3,h,22;	/* 2: S0" = a rotr 22		*/ \
	xor		rT0,rT0,rT1;	/* 2: S0 = S0 xor S0'		*/ \
	or		rT2,h,a;	/* 2: maj = a or b		*/ \
	xor		rT3,rT0,rT3;	/* 2: S0 = S0 xor S0"		*/ \
	and		rT1,h,a;	/* 2: maj' = a and b		*/ \
	and		rT2,rT2,b;	/* 2: maj = maj and c		*/ \
	add		c,c,g;		/* 2: d = d + temp1		*/ \
	or		rT2,rT1,rT2;	/* 2: maj = maj or maj'		*/ \
	add		rT3,rT3,rT2;	/* 2: temp2 = S0 + maj		*/ \
	add		g,g,rT3		/* 2: h = temp1 + temp2		*/

#define R_CALC_W(a, b, c, d, e, f, g, h, w0, w1, w4, w5, w7, k, off) \
	rotrwi		rT2,e,6;	/* 1: S1 = e rotr 6		*/ \
	evmergelohi	rT0,w0,w1;	/*    w[-15]			*/ \
	rotrwi		rT3,e,11;	/* 1: S1' = e rotr 11		*/ \
	evsrwiu		rT1,rT0,3;	/*    s0 = w[-15] >> 3		*/ \
	xor		rT2,rT2,rT3;	/* 1: S1 = S1 xor S1'		*/ \
	evrlwi		rT0,rT0,25;	/*    s0' = w[-15] rotr	7	*/ \
	rotrwi		rT3,e,25;	/* 1: S1' = e rotr 25		*/ \
	evxor		rT1,rT1,rT0;	/*    s0 = s0 xor s0'		*/ \
	xor		rT2,rT2,rT3;	/* 1: S1 = S1 xor S1'		*/ \
	evrlwi		rT0,rT0,21;	/*    s0' = w[-15] rotr 18	*/ \
	add		h,h,rT2;	/* 1: temp1 = h + S1		*/ \
	evxor		rT0,rT0,rT1;	/*    s0 = s0 xor s0'		*/ \
	and		rT2,e,f;	/* 1: ch = e and f		*/ \
	evaddw		w0,w0,rT0;	/*    w = w[-16] + s0		*/ \
	andc		rT3,g,e;	/* 1: ch' = ~e and g		*/ \
	evsrwiu		rT0,w7,10;	/*    s1 = w[-2] >> 10		*/ \
	xor		rT2,rT2,rT3;	/* 1: ch = ch xor ch'		*/ \
	evrlwi		rT1,w7,15;	/*    s1' = w[-2] rotr 17	*/ \
	add		h,h,rT2;	/* 1: temp1 = temp1 + ch	*/ \
	evxor		rT0,rT0,rT1;	/*    s1 = s1 xor s1'		*/ \
	rotrwi		rT2,a,2;	/* 1: S0 = a rotr 2		*/ \
	evrlwi		rT1,w7,13;	/*    s1' = w[-2] rotr 19	*/ \
	rotrwi		rT3,a,13;	/* 1: S0' = a rotr 13		*/ \
	evxor		rT0,rT0,rT1;	/*    s1 = s1 xor s1'		*/ \
	xor		rT2,rT2,rT3;	/* 1: S0 = S0 xor S0'		*/ \
	evldw		rT1,off(rKP);	/*    k				*/ \
	rotrwi		rT3,a,22;	/* 1: S0' = a rotr 22		*/ \
	evaddw		w0,w0,rT0;	/*    w = w + s1		*/ \
	xor		rT2,rT2,rT3;	/* 1: S0 = S0 xor S0'		*/ \
	evmergelohi	rT0,w4,w5;	/*    w[-7]			*/ \
	and		rT3,a,b;	/* 1: maj = a and b		*/ \
	evaddw		w0,w0,rT0;	/*    w = w + w[-7]		*/ \
	CMP_K##k##_LOOP							   \
	add		rT2,rT2,rT3;	/* 1: temp2 = S0 + maj		*/ \
	evaddw		rT1,rT1,w0;	/*    wk = w + k		*/ \
	xor		rT3,a,b;	/* 1: maj = a xor b		*/ \
	evmergehi	rT0,rT1,rT1;	/*    wk1/wk2			*/ \
	and		rT3,rT3,c;	/* 1: maj = maj and c		*/ \
	add		h,h,rT0;	/* 1: temp1 = temp1 + wk	*/ \
	add		rT2,rT2,rT3;	/* 1: temp2 = temp2 + maj	*/ \
	add		g,g,rT1;	/* 2: temp1 = temp1 + wk	*/ \
	add		d,d,h;		/* 1: d = d + temp1		*/ \
	rotrwi		rT0,d,6;	/* 2: S1 = e rotr 6		*/ \
	add		h,h,rT2;	/* 1: h = temp1 + temp2		*/ \
	rotrwi		rT1,d,11;	/* 2: S1' = e rotr 11		*/ \
	rotrwi		rT2,d,25;	/* 2: S" = e rotr 25		*/ \
	xor		rT0,rT0,rT1;	/* 2: S1 = S1 xor S1'		*/ \
	and		rT3,d,e;	/* 2: ch = e and f		*/ \
	xor		rT0,rT0,rT2;	/* 2: S1 = S1 xor S1"		*/ \
	andc		rT1,f,d;	/* 2: ch' = ~e and g		*/ \
	add		g,g,rT0;	/* 2: temp1 = h + S1		*/ \
	xor		rT3,rT3,rT1;	/* 2: ch = ch xor ch'		*/ \
	rotrwi		rT0,h,2;	/* 2: S0 = a rotr 2		*/ \
	add		g,g,rT3;	/* 2: temp1 = temp1 + ch	*/ \
	rotrwi		rT1,h,13;	/* 2: S0' = a rotr 13		*/ \
	rotrwi		rT3,h,22;	/* 2: S0" = a rotr 22		*/ \
	xor		rT0,rT0,rT1;	/* 2: S0 = S0 xor S0'		*/ \
	or		rT2,h,a;	/* 2: maj = a or b		*/ \
	and		rT1,h,a;	/* 2: maj' = a and b		*/ \
	and		rT2,rT2,b;	/* 2: maj = maj and c		*/ \
	xor		rT3,rT0,rT3;	/* 2: S0 = S0 xor S0"		*/ \
	or		rT2,rT1,rT2;	/* 2: maj = maj or maj'		*/ \
	add		c,c,g;		/* 2: d = d + temp1		*/ \
	add		rT3,rT3,rT2;	/* 2: temp2 = S0 + maj		*/ \
	add		g,g,rT3		/* 2: h = temp1 + temp2		*/

_GLOBAL(ppc_spe_sha256_transform)
	INITIALIZE

	mtctr		r5
	lwz		rH0,0(rHP)
	lwz		rH1,4(rHP)
	lwz		rH2,8(rHP)
	lwz		rH3,12(rHP)
	lwz		rH4,16(rHP)
	lwz		rH5,20(rHP)
	lwz		rH6,24(rHP)
	lwz		rH7,28(rHP)

ppc_spe_sha256_main:
	lis		rKP,PPC_SPE_SHA256_K@ha
	addi		rKP,rKP,PPC_SPE_SHA256_K@l

	R_LOAD_W(rH0, rH1, rH2, rH3, rH4, rH5, rH6, rH7, rW0, 0)
	R_LOAD_W(rH6, rH7, rH0, rH1, rH2, rH3, rH4, rH5, rW1, 8)
	R_LOAD_W(rH4, rH5, rH6, rH7, rH0, rH1, rH2, rH3, rW2, 16)
	R_LOAD_W(rH2, rH3, rH4, rH5, rH6, rH7, rH0, rH1, rW3, 24)
	R_LOAD_W(rH0, rH1, rH2, rH3, rH4, rH5, rH6, rH7, rW4, 32)
	R_LOAD_W(rH6, rH7, rH0, rH1, rH2, rH3, rH4, rH5, rW5, 40)
	R_LOAD_W(rH4, rH5, rH6, rH7, rH0, rH1, rH2, rH3, rW6, 48)
	R_LOAD_W(rH2, rH3, rH4, rH5, rH6, rH7, rH0, rH1, rW7, 56)
ppc_spe_sha256_16_rounds:
	addi		rKP,rKP,64
	R_CALC_W(rH0, rH1, rH2, rH3, rH4, rH5, rH6, rH7,
		 rW0, rW1, rW4, rW5, rW7, N, 0)
	R_CALC_W(rH6, rH7, rH0, rH1, rH2, rH3, rH4, rH5,
		 rW1, rW2, rW5, rW6, rW0, N, 8)
	R_CALC_W(rH4, rH5, rH6, rH7, rH0, rH1, rH2, rH3,
		 rW2, rW3, rW6, rW7, rW1, N, 16)
	R_CALC_W(rH2, rH3, rH4, rH5, rH6, rH7, rH0, rH1,
		 rW3, rW4, rW7, rW0, rW2, N, 24)
	R_CALC_W(rH0, rH1, rH2, rH3, rH4, rH5, rH6, rH7,
		 rW4, rW5, rW0, rW1, rW3, N, 32)
	R_CALC_W(rH6, rH7, rH0, rH1, rH2, rH3, rH4, rH5,
		 rW5, rW6, rW1, rW2, rW4, N, 40)
	R_CALC_W(rH4, rH5, rH6, rH7, rH0, rH1, rH2, rH3,
		 rW6, rW7, rW2, rW3, rW5, N, 48)
	R_CALC_W(rH2, rH3, rH4, rH5, rH6, rH7, rH0, rH1,
		 rW7, rW0, rW3, rW4, rW6, C, 56)
	bt		gt,ppc_spe_sha256_16_rounds

	lwz		rW0,0(rHP)
	NEXT_BLOCK
	lwz		rW1,4(rHP)
	lwz		rW2,8(rHP)
	lwz		rW3,12(rHP)
	lwz		rW4,16(rHP)
	lwz		rW5,20(rHP)
	lwz		rW6,24(rHP)
	lwz		rW7,28(rHP)

	add		rH0,rH0,rW0
	stw		rH0,0(rHP)
	add		rH1,rH1,rW1
	stw		rH1,4(rHP)
	add		rH2,rH2,rW2
	stw		rH2,8(rHP)
	add		rH3,rH3,rW3
	stw		rH3,12(rHP)
	add		rH4,rH4,rW4
	stw		rH4,16(rHP)
	add		rH5,rH5,rW5
	stw		rH5,20(rHP)
	add		rH6,rH6,rW6
	stw		rH6,24(rHP)
	add		rH7,rH7,rW7
	stw		rH7,28(rHP)

	bdnz		ppc_spe_sha256_main

	FINALIZE
	blr

.data
.align 5
PPC_SPE_SHA256_K:
	.long 0x428a2f98,0x71374491,0xb5c0fbcf,0xe9b5dba5
	.long 0x3956c25b,0x59f111f1,0x923f82a4,0xab1c5ed5
	.long 0xd807aa98,0x12835b01,0x243185be,0x550c7dc3
	.long 0x72be5d74,0x80deb1fe,0x9bdc06a7,0xc19bf174
	.long 0xe49b69c1,0xefbe4786,0x0fc19dc6,0x240ca1cc
	.long 0x2de92c6f,0x4a7484aa,0x5cb0a9dc,0x76f988da
	.long 0x983e5152,0xa831c66d,0xb00327c8,0xbf597fc7
	.long 0xc6e00bf3,0xd5a79147,0x06ca6351,0x14292967
	.long 0x27b70a85,0x2e1b2138,0x4d2c6dfc,0x53380d13
	.long 0x650a7354,0x766a0abb,0x81c2c92e,0x92722c85
	.long 0xa2bfe8a1,0xa81a664b,0xc24b8b70,0xc76c51a3
	.long 0xd192e819,0xd6990624,0xf40e3585,0x106aa070
	.long 0x19a4c116,0x1e376c08,0x2748774c,0x34b0bcb5
	.long 0x391c0cb3,0x4ed8aa4a,0x5b9cca4f,0x682e6ff3
	.long 0x748f82ee,0x78a5636f,0x84c87814,0x8cc70208
	.long 0x90befffa,0xa4506ceb,0xbef9a3f7,0xc67178f2