Context Navigation

source: git/src-cryptopp/sha.cpp

Last change on this file was e230cb0, checked in by David Stainton <dstainton415@…>, at 2016-10-12T13:27:29Z
Add cryptopp from tag CRYPTOPP_5_6_5
Property mode set to `100644`
File size: 27.6 KB

Line
1	// sha.cpp - modified by Wei Dai from Steve Reid's public domain sha1.c
2
3	// Steve Reid implemented SHA-1. Wei Dai implemented SHA-2.
4	// Both are in the public domain.
5
6	// use "cl /EP /P /DCRYPTOPP_GENERATE_X64_MASM sha.cpp" to generate MASM code
7
8	#include "pch.h"
9	#include "config.h"
10
11	#if CRYPTOPP_MSC_VERSION
12	# pragma warning(disable: 4100 4731)
13	#endif
14
15	#ifndef CRYPTOPP_IMPORTS
16	#ifndef CRYPTOPP_GENERATE_X64_MASM
17
18	#include "secblock.h"
19	#include "sha.h"
20	#include "misc.h"
21	#include "cpu.h"
22
23	#if defined(CRYPTOPP_DISABLE_SHA_ASM)
24	# undef CRYPTOPP_X86_ASM_AVAILABLE
25	# undef CRYPTOPP_X32_ASM_AVAILABLE
26	# undef CRYPTOPP_X64_ASM_AVAILABLE
27	# undef CRYPTOPP_BOOL_SSE2_ASM_AVAILABLE
28	#endif
29
30	NAMESPACE_BEGIN(CryptoPP)
31
32	// start of Steve Reid's code
33
34	#define blk0(i) (W[i] = data[i])
35	#define blk1(i) (W[i&15] = rotlFixed(W[(i+13)&15]^W[(i+8)&15]^W[(i+2)&15]^W[i&15],1))
36
37	void SHA1::InitState(HashWordType *state)
38	{
39	state[0] = 0x67452301L;
40	state[1] = 0xEFCDAB89L;
41	state[2] = 0x98BADCFEL;
42	state[3] = 0x10325476L;
43	state[4] = 0xC3D2E1F0L;
44	}
45
46	#define f1(x,y,z) (z^(x&(y^z)))
47	#define f2(x,y,z) (x^y^z)
48	#define f3(x,y,z) ((x&y)\|(z&(x\|y)))
49	#define f4(x,y,z) (x^y^z)
50
51	/* (R0+R1), R2, R3, R4 are the different operations used in SHA1 */
52	#define R0(v,w,x,y,z,i) z+=f1(w,x,y)+blk0(i)+0x5A827999+rotlFixed(v,5);w=rotlFixed(w,30);
53	#define R1(v,w,x,y,z,i) z+=f1(w,x,y)+blk1(i)+0x5A827999+rotlFixed(v,5);w=rotlFixed(w,30);
54	#define R2(v,w,x,y,z,i) z+=f2(w,x,y)+blk1(i)+0x6ED9EBA1+rotlFixed(v,5);w=rotlFixed(w,30);
55	#define R3(v,w,x,y,z,i) z+=f3(w,x,y)+blk1(i)+0x8F1BBCDC+rotlFixed(v,5);w=rotlFixed(w,30);
56	#define R4(v,w,x,y,z,i) z+=f4(w,x,y)+blk1(i)+0xCA62C1D6+rotlFixed(v,5);w=rotlFixed(w,30);
57
58	void SHA1::Transform(word32 state, const word32 data)
59	{
60	word32 W[16];
61	/* Copy context->state[] to working vars */
62	word32 a = state[0];
63	word32 b = state[1];
64	word32 c = state[2];
65	word32 d = state[3];
66	word32 e = state[4];
67	/* 4 rounds of 20 operations each. Loop unrolled. */
68	R0(a,b,c,d,e, 0); R0(e,a,b,c,d, 1); R0(d,e,a,b,c, 2); R0(c,d,e,a,b, 3);
69	R0(b,c,d,e,a, 4); R0(a,b,c,d,e, 5); R0(e,a,b,c,d, 6); R0(d,e,a,b,c, 7);
70	R0(c,d,e,a,b, 8); R0(b,c,d,e,a, 9); R0(a,b,c,d,e,10); R0(e,a,b,c,d,11);
71	R0(d,e,a,b,c,12); R0(c,d,e,a,b,13); R0(b,c,d,e,a,14); R0(a,b,c,d,e,15);
72	R1(e,a,b,c,d,16); R1(d,e,a,b,c,17); R1(c,d,e,a,b,18); R1(b,c,d,e,a,19);
73	R2(a,b,c,d,e,20); R2(e,a,b,c,d,21); R2(d,e,a,b,c,22); R2(c,d,e,a,b,23);
74	R2(b,c,d,e,a,24); R2(a,b,c,d,e,25); R2(e,a,b,c,d,26); R2(d,e,a,b,c,27);
75	R2(c,d,e,a,b,28); R2(b,c,d,e,a,29); R2(a,b,c,d,e,30); R2(e,a,b,c,d,31);
76	R2(d,e,a,b,c,32); R2(c,d,e,a,b,33); R2(b,c,d,e,a,34); R2(a,b,c,d,e,35);
77	R2(e,a,b,c,d,36); R2(d,e,a,b,c,37); R2(c,d,e,a,b,38); R2(b,c,d,e,a,39);
78	R3(a,b,c,d,e,40); R3(e,a,b,c,d,41); R3(d,e,a,b,c,42); R3(c,d,e,a,b,43);
79	R3(b,c,d,e,a,44); R3(a,b,c,d,e,45); R3(e,a,b,c,d,46); R3(d,e,a,b,c,47);
80	R3(c,d,e,a,b,48); R3(b,c,d,e,a,49); R3(a,b,c,d,e,50); R3(e,a,b,c,d,51);
81	R3(d,e,a,b,c,52); R3(c,d,e,a,b,53); R3(b,c,d,e,a,54); R3(a,b,c,d,e,55);
82	R3(e,a,b,c,d,56); R3(d,e,a,b,c,57); R3(c,d,e,a,b,58); R3(b,c,d,e,a,59);
83	R4(a,b,c,d,e,60); R4(e,a,b,c,d,61); R4(d,e,a,b,c,62); R4(c,d,e,a,b,63);
84	R4(b,c,d,e,a,64); R4(a,b,c,d,e,65); R4(e,a,b,c,d,66); R4(d,e,a,b,c,67);
85	R4(c,d,e,a,b,68); R4(b,c,d,e,a,69); R4(a,b,c,d,e,70); R4(e,a,b,c,d,71);
86	R4(d,e,a,b,c,72); R4(c,d,e,a,b,73); R4(b,c,d,e,a,74); R4(a,b,c,d,e,75);
87	R4(e,a,b,c,d,76); R4(d,e,a,b,c,77); R4(c,d,e,a,b,78); R4(b,c,d,e,a,79);
88	/* Add the working vars back into context.state[] */
89	state[0] += a;
90	state[1] += b;
91	state[2] += c;
92	state[3] += d;
93	state[4] += e;
94	}
95
96	// end of Steve Reid's code
97
98	// *************************************************************
99
100	void SHA224::InitState(HashWordType *state)
101	{
102	static const word32 s[8] = {0xc1059ed8, 0x367cd507, 0x3070dd17, 0xf70e5939, 0xffc00b31, 0x68581511, 0x64f98fa7, 0xbefa4fa4};
103	memcpy(state, s, sizeof(s));
104	}
105
106	void SHA256::InitState(HashWordType *state)
107	{
108	static const word32 s[8] = {0x6a09e667, 0xbb67ae85, 0x3c6ef372, 0xa54ff53a, 0x510e527f, 0x9b05688c, 0x1f83d9ab, 0x5be0cd19};
109	memcpy(state, s, sizeof(s));
110	}
111
112	#if CRYPTOPP_BOOL_SSE2_ASM_AVAILABLE
113	CRYPTOPP_ALIGN_DATA(16) extern const word32 SHA256_K[64] CRYPTOPP_SECTION_ALIGN16 = {
114	#else
115	extern const word32 SHA256_K[64] = {
116	#endif
117	0x428a2f98, 0x71374491, 0xb5c0fbcf, 0xe9b5dba5,
118	0x3956c25b, 0x59f111f1, 0x923f82a4, 0xab1c5ed5,
119	0xd807aa98, 0x12835b01, 0x243185be, 0x550c7dc3,
120	0x72be5d74, 0x80deb1fe, 0x9bdc06a7, 0xc19bf174,
121	0xe49b69c1, 0xefbe4786, 0x0fc19dc6, 0x240ca1cc,
122	0x2de92c6f, 0x4a7484aa, 0x5cb0a9dc, 0x76f988da,
123	0x983e5152, 0xa831c66d, 0xb00327c8, 0xbf597fc7,
124	0xc6e00bf3, 0xd5a79147, 0x06ca6351, 0x14292967,
125	0x27b70a85, 0x2e1b2138, 0x4d2c6dfc, 0x53380d13,
126	0x650a7354, 0x766a0abb, 0x81c2c92e, 0x92722c85,
127	0xa2bfe8a1, 0xa81a664b, 0xc24b8b70, 0xc76c51a3,
128	0xd192e819, 0xd6990624, 0xf40e3585, 0x106aa070,
129	0x19a4c116, 0x1e376c08, 0x2748774c, 0x34b0bcb5,
130	0x391c0cb3, 0x4ed8aa4a, 0x5b9cca4f, 0x682e6ff3,
131	0x748f82ee, 0x78a5636f, 0x84c87814, 0x8cc70208,
132	0x90befffa, 0xa4506ceb, 0xbef9a3f7, 0xc67178f2
133	};
134
135	#endif // #ifndef CRYPTOPP_GENERATE_X64_MASM
136
137	#if (defined(CRYPTOPP_X86_ASM_AVAILABLE) \|\| defined(CRYPTOPP_X32_ASM_AVAILABLE) \|\| defined(CRYPTOPP_GENERATE_X64_MASM))
138
139	static void CRYPTOPP_FASTCALL X86_SHA256_HashBlocks(word32 state, const word32 data, size_t len
140	#if defined(_MSC_VER) && (_MSC_VER == 1200)
141	, ... // VC60 workaround: prevent VC 6 from inlining this function
142	#endif
143	)
144	{
145	#if defined(_MSC_VER) && (_MSC_VER == 1200)
146	AS2(mov ecx, [state])
147	AS2(mov edx, [data])
148	#endif
149
150	#define LOCALS_SIZE 84 + 164 + 4*WORD_SZ
151	#define H(i) [BASE+ASM_MOD(1024+7-(i),8)*4]
152	#define G(i) H(i+1)
153	#define F(i) H(i+2)
154	#define E(i) H(i+3)
155	#define D(i) H(i+4)
156	#define C(i) H(i+5)
157	#define B(i) H(i+6)
158	#define A(i) H(i+7)
159	#define Wt(i) BASE+84+ASM_MOD(1024+15-(i),16)4
160	#define Wt_2(i) Wt((i)-2)
161	#define Wt_15(i) Wt((i)-15)
162	#define Wt_7(i) Wt((i)-7)
163	#define K_END [BASE+84+164+0*WORD_SZ]
164	#define STATE_SAVE [BASE+84+164+1*WORD_SZ]
165	#define DATA_SAVE [BASE+84+164+2*WORD_SZ]
166	#define DATA_END [BASE+84+164+3*WORD_SZ]
167	#define Kt(i) WORD_REG(si)+(i)*4
168	#if CRYPTOPP_BOOL_X32
169	#define BASE esp+8
170	#elif CRYPTOPP_BOOL_X86
171	#define BASE esp+4
172	#elif defined(__GNUC__)
173	#define BASE r8
174	#else
175	#define BASE rsp
176	#endif
177
178	#define RA0(i, edx, edi) \
179	AS2( add edx, [Kt(i)] )\
180	AS2( add edx, [Wt(i)] )\
181	AS2( add edx, H(i) )\
182
183	#define RA1(i, edx, edi)
184
185	#define RB0(i, edx, edi)
186
187	#define RB1(i, edx, edi) \
188	AS2( mov AS_REG_7d, [Wt_2(i)] )\
189	AS2( mov edi, [Wt_15(i)])\
190	AS2( mov ebx, AS_REG_7d )\
191	AS2( shr AS_REG_7d, 10 )\
192	AS2( ror ebx, 17 )\
193	AS2( xor AS_REG_7d, ebx )\
194	AS2( ror ebx, 2 )\
195	AS2( xor ebx, AS_REG_7d )/* s1(W_t-2) */\
196	AS2( add ebx, [Wt_7(i)])\
197	AS2( mov AS_REG_7d, edi )\
198	AS2( shr AS_REG_7d, 3 )\
199	AS2( ror edi, 7 )\
200	AS2( add ebx, [Wt(i)])/* s1(W_t-2) + W_t-7 + W_t-16 */\
201	AS2( xor AS_REG_7d, edi )\
202	AS2( add edx, [Kt(i)])\
203	AS2( ror edi, 11 )\
204	AS2( add edx, H(i) )\
205	AS2( xor AS_REG_7d, edi )/* s0(W_t-15) */\
206	AS2( add AS_REG_7d, ebx )/* W_t = s1(W_t-2) + W_t-7 + s0(W_t-15) W_t-16*/\
207	AS2( mov [Wt(i)], AS_REG_7d)\
208	AS2( add edx, AS_REG_7d )\
209
210	#define ROUND(i, r, eax, ecx, edi, edx)\
211	/* in: edi = E */\
212	/* unused: eax, ecx, temp: ebx, AS_REG_7d, out: edx = T1 */\
213	AS2( mov edx, F(i) )\
214	AS2( xor edx, G(i) )\
215	AS2( and edx, edi )\
216	AS2( xor edx, G(i) )/* Ch(E,F,G) = (G^(E&(F^G))) */\
217	AS2( mov AS_REG_7d, edi )\
218	AS2( ror edi, 6 )\
219	AS2( ror AS_REG_7d, 25 )\
220	RA##r(i, edx, edi )/* H + Wt + Kt + Ch(E,F,G) */\
221	AS2( xor AS_REG_7d, edi )\
222	AS2( ror edi, 5 )\
223	AS2( xor AS_REG_7d, edi )/* S1(E) */\
224	AS2( add edx, AS_REG_7d )/* T1 = S1(E) + Ch(E,F,G) + H + Wt + Kt */\
225	RB##r(i, edx, edi )/* H + Wt + Kt + Ch(E,F,G) */\
226	/* in: ecx = A, eax = B^C, edx = T1 */\
227	/* unused: edx, temp: ebx, AS_REG_7d, out: eax = A, ecx = B^C, edx = E */\
228	AS2( mov ebx, ecx )\
229	AS2( xor ecx, B(i) )/* A^B */\
230	AS2( and eax, ecx )\
231	AS2( xor eax, B(i) )/* Maj(A,B,C) = B^((A^B)&(B^C) */\
232	AS2( mov AS_REG_7d, ebx )\
233	AS2( ror ebx, 2 )\
234	AS2( add eax, edx )/* T1 + Maj(A,B,C) */\
235	AS2( add edx, D(i) )\
236	AS2( mov D(i), edx )\
237	AS2( ror AS_REG_7d, 22 )\
238	AS2( xor AS_REG_7d, ebx )\
239	AS2( ror ebx, 11 )\
240	AS2( xor AS_REG_7d, ebx )\
241	AS2( add eax, AS_REG_7d )/* T1 + S0(A) + Maj(A,B,C) */\
242	AS2( mov H(i), eax )\
243
244	// Unroll the use of CRYPTOPP_BOOL_X64 in assembler math. The GAS assembler on X32 (version 2.25)
245	// complains "Error: invalid operands (ABS and UND sections) for `*` and `-`"
246	#if CRYPTOPP_BOOL_X64
247	#define SWAP_COPY(i) \
248	AS2( mov WORD_REG(bx), [WORD_REG(dx)+i*WORD_SZ])\
249	AS1( bswap WORD_REG(bx))\
250	AS2( mov [Wt(i*2+1)], WORD_REG(bx))
251	#else // X86 and X32
252	#define SWAP_COPY(i) \
253	AS2( mov WORD_REG(bx), [WORD_REG(dx)+i*WORD_SZ])\
254	AS1( bswap WORD_REG(bx))\
255	AS2( mov [Wt(i)], WORD_REG(bx))
256	#endif
257
258	#if defined(__GNUC__)
259	#if CRYPTOPP_BOOL_X64
260	FixedSizeAlignedSecBlock<byte, LOCALS_SIZE> workspace;
261	#endif
262	__asm__ __volatile__
263	(
264	#if CRYPTOPP_BOOL_X64
265	"lea %4, %%r8;"
266	#endif
267	INTEL_NOPREFIX
268	#elif defined(CRYPTOPP_GENERATE_X64_MASM)
269	ALIGN 8
270	X86_SHA256_HashBlocks PROC FRAME
271	rex_push_reg rsi
272	push_reg rdi
273	push_reg rbx
274	push_reg rbp
275	alloc_stack(LOCALS_SIZE+8)
276	.endprolog
277	mov rdi, r8
278	lea rsi, [?SHA256_K@CryptoPP@@3QBIB + 48*4]
279	#endif
280
281	#if CRYPTOPP_BOOL_X86 \|\| CRYPTOPP_BOOL_X32
282	#ifndef __GNUC__
283	AS2( mov edi, [len])
284	AS2( lea WORD_REG(si), [SHA256_K+48*4])
285	#endif
286	#if !defined(_MSC_VER) \|\| (_MSC_VER < 1400)
287	AS_PUSH_IF86(bx)
288	#endif
289
290	AS_PUSH_IF86(bp)
291	AS2( mov ebx, esp)
292	AS2( and esp, -16)
293	AS2( sub WORD_REG(sp), LOCALS_SIZE)
294	AS_PUSH_IF86(bx)
295	#endif
296	AS2( mov STATE_SAVE, WORD_REG(cx))
297	AS2( mov DATA_SAVE, WORD_REG(dx))
298	AS2( lea WORD_REG(ax), [WORD_REG(di) + WORD_REG(dx)])
299	AS2( mov DATA_END, WORD_REG(ax))
300	AS2( mov K_END, WORD_REG(si))
301
302	#if CRYPTOPP_BOOL_SSE2_ASM_AVAILABLE
303	#if CRYPTOPP_BOOL_X86 \|\| CRYPTOPP_BOOL_X32
304	AS2( test edi, 1)
305	ASJ( jnz, 2, f)
306	AS1( dec DWORD PTR K_END)
307	#endif
308	AS2( movdqa xmm0, XMMWORD_PTR [WORD_REG(cx)+0*16])
309	AS2( movdqa xmm1, XMMWORD_PTR [WORD_REG(cx)+1*16])
310	#endif
311
312	#if CRYPTOPP_BOOL_X86 \|\| CRYPTOPP_BOOL_X32
313	#if CRYPTOPP_BOOL_SSE2_ASM_AVAILABLE
314	ASJ( jmp, 0, f)
315	#endif
316	ASL(2) // non-SSE2
317	AS2( mov esi, ecx)
318	AS2( lea edi, A(0))
319	AS2( mov ecx, 8)
320	ATT_NOPREFIX
321	AS1( rep movsd)
322	INTEL_NOPREFIX
323	AS2( mov esi, K_END)
324	ASJ( jmp, 3, f)
325	#endif
326
327	#if CRYPTOPP_BOOL_SSE2_ASM_AVAILABLE
328	ASL(0)
329	AS2( movdqa E(0), xmm1)
330	AS2( movdqa A(0), xmm0)
331	#endif
332	#if CRYPTOPP_BOOL_X86 \|\| CRYPTOPP_BOOL_X32
333	ASL(3)
334	#endif
335	AS2( sub WORD_REG(si), 48*4)
336	SWAP_COPY(0) SWAP_COPY(1) SWAP_COPY(2) SWAP_COPY(3)
337	SWAP_COPY(4) SWAP_COPY(5) SWAP_COPY(6) SWAP_COPY(7)
338	#if CRYPTOPP_BOOL_X86 \|\| CRYPTOPP_BOOL_X32
339	SWAP_COPY(8) SWAP_COPY(9) SWAP_COPY(10) SWAP_COPY(11)
340	SWAP_COPY(12) SWAP_COPY(13) SWAP_COPY(14) SWAP_COPY(15)
341	#endif
342	AS2( mov edi, E(0)) // E
343	AS2( mov eax, B(0)) // B
344	AS2( xor eax, C(0)) // B^C
345	AS2( mov ecx, A(0)) // A
346
347	ROUND(0, 0, eax, ecx, edi, edx)
348	ROUND(1, 0, ecx, eax, edx, edi)
349	ROUND(2, 0, eax, ecx, edi, edx)
350	ROUND(3, 0, ecx, eax, edx, edi)
351	ROUND(4, 0, eax, ecx, edi, edx)
352	ROUND(5, 0, ecx, eax, edx, edi)
353	ROUND(6, 0, eax, ecx, edi, edx)
354	ROUND(7, 0, ecx, eax, edx, edi)
355	ROUND(8, 0, eax, ecx, edi, edx)
356	ROUND(9, 0, ecx, eax, edx, edi)
357	ROUND(10, 0, eax, ecx, edi, edx)
358	ROUND(11, 0, ecx, eax, edx, edi)
359	ROUND(12, 0, eax, ecx, edi, edx)
360	ROUND(13, 0, ecx, eax, edx, edi)
361	ROUND(14, 0, eax, ecx, edi, edx)
362	ROUND(15, 0, ecx, eax, edx, edi)
363
364	ASL(1)
365	AS2(add WORD_REG(si), 4*16)
366	ROUND(0, 1, eax, ecx, edi, edx)
367	ROUND(1, 1, ecx, eax, edx, edi)
368	ROUND(2, 1, eax, ecx, edi, edx)
369	ROUND(3, 1, ecx, eax, edx, edi)
370	ROUND(4, 1, eax, ecx, edi, edx)
371	ROUND(5, 1, ecx, eax, edx, edi)
372	ROUND(6, 1, eax, ecx, edi, edx)
373	ROUND(7, 1, ecx, eax, edx, edi)
374	ROUND(8, 1, eax, ecx, edi, edx)
375	ROUND(9, 1, ecx, eax, edx, edi)
376	ROUND(10, 1, eax, ecx, edi, edx)
377	ROUND(11, 1, ecx, eax, edx, edi)
378	ROUND(12, 1, eax, ecx, edi, edx)
379	ROUND(13, 1, ecx, eax, edx, edi)
380	ROUND(14, 1, eax, ecx, edi, edx)
381	ROUND(15, 1, ecx, eax, edx, edi)
382	AS2( cmp WORD_REG(si), K_END)
383	ATT_NOPREFIX
384	ASJ( jb, 1, b)
385	INTEL_NOPREFIX
386
387	AS2( mov WORD_REG(dx), DATA_SAVE)
388	AS2( add WORD_REG(dx), 64)
389	AS2( mov AS_REG_7, STATE_SAVE)
390	AS2( mov DATA_SAVE, WORD_REG(dx))
391
392	#if CRYPTOPP_BOOL_SSE2_ASM_AVAILABLE
393	#if CRYPTOPP_BOOL_X86 \|\| CRYPTOPP_BOOL_X32
394	AS2( test DWORD PTR K_END, 1)
395	ASJ( jz, 4, f)
396	#endif
397	AS2( movdqa xmm1, XMMWORD_PTR [AS_REG_7+1*16])
398	AS2( movdqa xmm0, XMMWORD_PTR [AS_REG_7+0*16])
399	AS2( paddd xmm1, E(0))
400	AS2( paddd xmm0, A(0))
401	AS2( movdqa [AS_REG_7+1*16], xmm1)
402	AS2( movdqa [AS_REG_7+0*16], xmm0)
403	AS2( cmp WORD_REG(dx), DATA_END)
404	ATT_NOPREFIX
405	ASJ( jb, 0, b)
406	INTEL_NOPREFIX
407	#endif
408
409	#if CRYPTOPP_BOOL_X86 \|\| CRYPTOPP_BOOL_X32
410	#if CRYPTOPP_BOOL_SSE2_ASM_AVAILABLE
411	ASJ( jmp, 5, f)
412	ASL(4) // non-SSE2
413	#endif
414	AS2( add [AS_REG_7+0*4], ecx) // A
415	AS2( add [AS_REG_7+4*4], edi) // E
416	AS2( mov eax, B(0))
417	AS2( mov ebx, C(0))
418	AS2( mov ecx, D(0))
419	AS2( add [AS_REG_7+1*4], eax)
420	AS2( add [AS_REG_7+2*4], ebx)
421	AS2( add [AS_REG_7+3*4], ecx)
422	AS2( mov eax, F(0))
423	AS2( mov ebx, G(0))
424	AS2( mov ecx, H(0))
425	AS2( add [AS_REG_7+5*4], eax)
426	AS2( add [AS_REG_7+6*4], ebx)
427	AS2( add [AS_REG_7+7*4], ecx)
428	AS2( mov ecx, AS_REG_7d)
429	AS2( cmp WORD_REG(dx), DATA_END)
430	ASJ( jb, 2, b)
431	#if CRYPTOPP_BOOL_SSE2_ASM_AVAILABLE
432	ASL(5)
433	#endif
434	#endif
435
436	AS_POP_IF86(sp)
437	AS_POP_IF86(bp)
438	#if !defined(_MSC_VER) \|\| (_MSC_VER < 1400)
439	AS_POP_IF86(bx)
440	#endif
441
442	#ifdef CRYPTOPP_GENERATE_X64_MASM
443	add rsp, LOCALS_SIZE+8
444	pop rbp
445	pop rbx
446	pop rdi
447	pop rsi
448	ret
449	X86_SHA256_HashBlocks ENDP
450	#endif
451
452	#ifdef __GNUC__
453	ATT_PREFIX
454	:
455	: "c" (state), "d" (data), "S" (SHA256_K+48), "D" (len)
456	#if CRYPTOPP_BOOL_X64
457	, "m" (workspace[0])
458	#endif
459	: "memory", "cc", "%eax"
460	#if CRYPTOPP_BOOL_X64
461	, "%rbx", "%r8", "%r10"
462	#endif
463	);
464	#endif
465	}
466
467	#endif // (defined(CRYPTOPP_X86_ASM_AVAILABLE) \|\| defined(CRYPTOPP_GENERATE_X64_MASM))
468
469	#ifndef CRYPTOPP_GENERATE_X64_MASM
470
471	#ifdef CRYPTOPP_X64_MASM_AVAILABLE
472	extern "C" {
473	void CRYPTOPP_FASTCALL X86_SHA256_HashBlocks(word32 state, const word32 data, size_t len);
474	}
475	#endif
476
477	#if (defined(CRYPTOPP_X86_ASM_AVAILABLE) \|\| defined(CRYPTOPP_X32_ASM_AVAILABLE) \|\| defined(CRYPTOPP_X64_MASM_AVAILABLE)) && !defined(CRYPTOPP_DISABLE_SHA_ASM)
478
479	size_t SHA256::HashMultipleBlocks(const word32 *input, size_t length)
480	{
481	X86_SHA256_HashBlocks(m_state, input, (length&(size_t(0)-BLOCKSIZE)) - !HasSSE2());
482	return length % BLOCKSIZE;
483	}
484
485	size_t SHA224::HashMultipleBlocks(const word32 *input, size_t length)
486	{
487	X86_SHA256_HashBlocks(m_state, input, (length&(size_t(0)-BLOCKSIZE)) - !HasSSE2());
488	return length % BLOCKSIZE;
489	}
490
491	#endif
492
493	#define blk2(i) (W[i&15]+=s1(W[(i-2)&15])+W[(i-7)&15]+s0(W[(i-15)&15]))
494
495	#define Ch(x,y,z) (z^(x&(y^z)))
496	#define Maj(x,y,z) (y^((x^y)&(y^z)))
497
498	#define a(i) T[(0-i)&7]
499	#define b(i) T[(1-i)&7]
500	#define c(i) T[(2-i)&7]
501	#define d(i) T[(3-i)&7]
502	#define e(i) T[(4-i)&7]
503	#define f(i) T[(5-i)&7]
504	#define g(i) T[(6-i)&7]
505	#define h(i) T[(7-i)&7]
506
507	#define R(i) h(i)+=S1(e(i))+Ch(e(i),f(i),g(i))+SHA256_K[i+j]+(j?blk2(i):blk0(i));\
508	d(i)+=h(i);h(i)+=S0(a(i))+Maj(a(i),b(i),c(i))
509
510	// for SHA256
511	#define S0(x) (rotrFixed(x,2)^rotrFixed(x,13)^rotrFixed(x,22))
512	#define S1(x) (rotrFixed(x,6)^rotrFixed(x,11)^rotrFixed(x,25))
513	#define s0(x) (rotrFixed(x,7)^rotrFixed(x,18)^(x>>3))
514	#define s1(x) (rotrFixed(x,17)^rotrFixed(x,19)^(x>>10))
515
516	// Smaller but slower
517	#if defined(__OPTIMIZE_SIZE__)
518	void SHA256::Transform(word32 state, const word32 data)
519	{
520	word32 T[20];
521	word32 W[32];
522	unsigned int i = 0, j = 0;
523	word32 *t = T+8;
524
525	memcpy(t, state, 8*4);
526	word32 e = t[4], a = t[0];
527
528	do
529	{
530	word32 w = data[j];
531	W[j] = w;
532	w += SHA256_K[j];
533	w += t[7];
534	w += S1(e);
535	w += Ch(e, t[5], t[6]);
536	e = t[3] + w;
537	t[3] = t[3+8] = e;
538	w += S0(t[0]);
539	a = w + Maj(a, t[1], t[2]);
540	t[-1] = t[7] = a;
541	--t;
542	++j;
543	if (j%8 == 0)
544	t += 8;
545	} while (j<16);
546
547	do
548	{
549	i = j&0xf;
550	word32 w = s1(W[i+16-2]) + s0(W[i+16-15]) + W[i] + W[i+16-7];
551	W[i+16] = W[i] = w;
552	w += SHA256_K[j];
553	w += t[7];
554	w += S1(e);
555	w += Ch(e, t[5], t[6]);
556	e = t[3] + w;
557	t[3] = t[3+8] = e;
558	w += S0(t[0]);
559	a = w + Maj(a, t[1], t[2]);
560	t[-1] = t[7] = a;
561
562	w = s1(W[(i+1)+16-2]) + s0(W[(i+1)+16-15]) + W[(i+1)] + W[(i+1)+16-7];
563	W[(i+1)+16] = W[(i+1)] = w;
564	w += SHA256_K[j+1];
565	w += (t-1)[7];
566	w += S1(e);
567	w += Ch(e, (t-1)[5], (t-1)[6]);
568	e = (t-1)[3] + w;
569	(t-1)[3] = (t-1)[3+8] = e;
570	w += S0((t-1)[0]);
571	a = w + Maj(a, (t-1)[1], (t-1)[2]);
572	(t-1)[-1] = (t-1)[7] = a;
573
574	t-=2;
575	j+=2;
576	if (j%8 == 0)
577	t += 8;
578	} while (j<64);
579
580	state[0] += a;
581	state[1] += t[1];
582	state[2] += t[2];
583	state[3] += t[3];
584	state[4] += e;
585	state[5] += t[5];
586	state[6] += t[6];
587	state[7] += t[7];
588	}
589	#else
590	void SHA256::Transform(word32 state, const word32 data)
591	{
592	word32 W[16];
593	#if (defined(CRYPTOPP_X86_ASM_AVAILABLE) \|\| defined(CRYPTOPP_X32_ASM_AVAILABLE) \|\| defined(CRYPTOPP_X64_MASM_AVAILABLE)) && !defined(CRYPTOPP_DISABLE_SHA_ASM)
594	// this byte reverse is a waste of time, but this function is only called by MDC
595	ByteReverse(W, data, BLOCKSIZE);
596	X86_SHA256_HashBlocks(state, W, BLOCKSIZE - !HasSSE2());
597	#else
598	word32 T[8];
599	/* Copy context->state[] to working vars */
600	memcpy(T, state, sizeof(T));
601	/* 64 operations, partially loop unrolled */
602	for (unsigned int j=0; j<64; j+=16)
603	{
604	R( 0); R( 1); R( 2); R( 3);
605	R( 4); R( 5); R( 6); R( 7);
606	R( 8); R( 9); R(10); R(11);
607	R(12); R(13); R(14); R(15);
608	}
609	/* Add the working vars back into context.state[] */
610	state[0] += a(0);
611	state[1] += b(0);
612	state[2] += c(0);
613	state[3] += d(0);
614	state[4] += e(0);
615	state[5] += f(0);
616	state[6] += g(0);
617	state[7] += h(0);
618	#endif
619	}
620	#endif
621
622	#undef S0
623	#undef S1
624	#undef s0
625	#undef s1
626	#undef R
627
628	// *************************************************************
629
630	void SHA384::InitState(HashWordType *state)
631	{
632	static const word64 s[8] = {
633	W64LIT(0xcbbb9d5dc1059ed8), W64LIT(0x629a292a367cd507),
634	W64LIT(0x9159015a3070dd17), W64LIT(0x152fecd8f70e5939),
635	W64LIT(0x67332667ffc00b31), W64LIT(0x8eb44a8768581511),
636	W64LIT(0xdb0c2e0d64f98fa7), W64LIT(0x47b5481dbefa4fa4)};
637	memcpy(state, s, sizeof(s));
638	}
639
640	void SHA512::InitState(HashWordType *state)
641	{
642	static const word64 s[8] = {
643	W64LIT(0x6a09e667f3bcc908), W64LIT(0xbb67ae8584caa73b),
644	W64LIT(0x3c6ef372fe94f82b), W64LIT(0xa54ff53a5f1d36f1),
645	W64LIT(0x510e527fade682d1), W64LIT(0x9b05688c2b3e6c1f),
646	W64LIT(0x1f83d9abfb41bd6b), W64LIT(0x5be0cd19137e2179)};
647	memcpy(state, s, sizeof(s));
648	}
649
650	#if CRYPTOPP_BOOL_SSE2_ASM_AVAILABLE && (CRYPTOPP_BOOL_X86 \|\| CRYPTOPP_BOOL_X32)
651	CRYPTOPP_ALIGN_DATA(16) static const word64 SHA512_K[80] CRYPTOPP_SECTION_ALIGN16 = {
652	#else
653	CRYPTOPP_ALIGN_DATA(16) static const word64 SHA512_K[80] CRYPTOPP_SECTION_ALIGN16 = {
654	#endif
655	W64LIT(0x428a2f98d728ae22), W64LIT(0x7137449123ef65cd),
656	W64LIT(0xb5c0fbcfec4d3b2f), W64LIT(0xe9b5dba58189dbbc),
657	W64LIT(0x3956c25bf348b538), W64LIT(0x59f111f1b605d019),
658	W64LIT(0x923f82a4af194f9b), W64LIT(0xab1c5ed5da6d8118),
659	W64LIT(0xd807aa98a3030242), W64LIT(0x12835b0145706fbe),
660	W64LIT(0x243185be4ee4b28c), W64LIT(0x550c7dc3d5ffb4e2),
661	W64LIT(0x72be5d74f27b896f), W64LIT(0x80deb1fe3b1696b1),
662	W64LIT(0x9bdc06a725c71235), W64LIT(0xc19bf174cf692694),
663	W64LIT(0xe49b69c19ef14ad2), W64LIT(0xefbe4786384f25e3),
664	W64LIT(0x0fc19dc68b8cd5b5), W64LIT(0x240ca1cc77ac9c65),
665	W64LIT(0x2de92c6f592b0275), W64LIT(0x4a7484aa6ea6e483),
666	W64LIT(0x5cb0a9dcbd41fbd4), W64LIT(0x76f988da831153b5),
667	W64LIT(0x983e5152ee66dfab), W64LIT(0xa831c66d2db43210),
668	W64LIT(0xb00327c898fb213f), W64LIT(0xbf597fc7beef0ee4),
669	W64LIT(0xc6e00bf33da88fc2), W64LIT(0xd5a79147930aa725),
670	W64LIT(0x06ca6351e003826f), W64LIT(0x142929670a0e6e70),
671	W64LIT(0x27b70a8546d22ffc), W64LIT(0x2e1b21385c26c926),
672	W64LIT(0x4d2c6dfc5ac42aed), W64LIT(0x53380d139d95b3df),
673	W64LIT(0x650a73548baf63de), W64LIT(0x766a0abb3c77b2a8),
674	W64LIT(0x81c2c92e47edaee6), W64LIT(0x92722c851482353b),
675	W64LIT(0xa2bfe8a14cf10364), W64LIT(0xa81a664bbc423001),
676	W64LIT(0xc24b8b70d0f89791), W64LIT(0xc76c51a30654be30),
677	W64LIT(0xd192e819d6ef5218), W64LIT(0xd69906245565a910),
678	W64LIT(0xf40e35855771202a), W64LIT(0x106aa07032bbd1b8),
679	W64LIT(0x19a4c116b8d2d0c8), W64LIT(0x1e376c085141ab53),
680	W64LIT(0x2748774cdf8eeb99), W64LIT(0x34b0bcb5e19b48a8),
681	W64LIT(0x391c0cb3c5c95a63), W64LIT(0x4ed8aa4ae3418acb),
682	W64LIT(0x5b9cca4f7763e373), W64LIT(0x682e6ff3d6b2b8a3),
683	W64LIT(0x748f82ee5defb2fc), W64LIT(0x78a5636f43172f60),
684	W64LIT(0x84c87814a1f0ab72), W64LIT(0x8cc702081a6439ec),
685	W64LIT(0x90befffa23631e28), W64LIT(0xa4506cebde82bde9),
686	W64LIT(0xbef9a3f7b2c67915), W64LIT(0xc67178f2e372532b),
687	W64LIT(0xca273eceea26619c), W64LIT(0xd186b8c721c0c207),
688	W64LIT(0xeada7dd6cde0eb1e), W64LIT(0xf57d4f7fee6ed178),
689	W64LIT(0x06f067aa72176fba), W64LIT(0x0a637dc5a2c898a6),
690	W64LIT(0x113f9804bef90dae), W64LIT(0x1b710b35131c471b),
691	W64LIT(0x28db77f523047d84), W64LIT(0x32caab7b40c72493),
692	W64LIT(0x3c9ebe0a15c9bebc), W64LIT(0x431d67c49c100d4c),
693	W64LIT(0x4cc5d4becb3e42b6), W64LIT(0x597f299cfc657e2a),
694	W64LIT(0x5fcb6fab3ad6faec), W64LIT(0x6c44198c4a475817)
695	};
696
697	#if CRYPTOPP_BOOL_SSE2_ASM_AVAILABLE && (CRYPTOPP_BOOL_X86 \|\| CRYPTOPP_BOOL_X32)
698	// put assembly version in separate function, otherwise MSVC 2005 SP1 doesn't generate correct code for the non-assembly version
699	CRYPTOPP_NAKED static void CRYPTOPP_FASTCALL SHA512_SSE2_Transform(word64 state, const word64 data)
700	{
701	#ifdef __GNUC__
702	__asm__ __volatile__
703	(
704	INTEL_NOPREFIX
705	AS_PUSH_IF86( bx)
706	AS2( mov ebx, eax)
707	#else
708	AS1( push ebx)
709	AS1( push esi)
710	AS1( push edi)
711	AS2( lea ebx, SHA512_K)
712	#endif
713
714	AS2( mov eax, esp)
715	AS2( and esp, 0xfffffff0)
716	AS2( sub esp, 2716) // 1716 for expanded data, 20*8 for state
717	AS_PUSH_IF86( ax)
718	AS2( xor eax, eax)
719
720	#if CRYPTOPP_BOOL_X32
721	AS2( lea edi, [esp+8+8*8]) // start at middle of state buffer. will decrement pointer each round to avoid copying
722	AS2( lea esi, [esp+8+20*8+8]) // 16-byte alignment, then add 8
723	#else
724	AS2( lea edi, [esp+4+8*8]) // start at middle of state buffer. will decrement pointer each round to avoid copying
725	AS2( lea esi, [esp+4+20*8+8]) // 16-byte alignment, then add 8
726	#endif
727
728	AS2( movdqa xmm0, [ecx+0*16])
729	AS2( movdq2q mm4, xmm0)
730	AS2( movdqa [edi+0*16], xmm0)
731	AS2( movdqa xmm0, [ecx+1*16])
732	AS2( movdqa [edi+1*16], xmm0)
733	AS2( movdqa xmm0, [ecx+2*16])
734	AS2( movdq2q mm5, xmm0)
735	AS2( movdqa [edi+2*16], xmm0)
736	AS2( movdqa xmm0, [ecx+3*16])
737	AS2( movdqa [edi+3*16], xmm0)
738	ASJ( jmp, 0, f)
739
740	#define SSE2_S0_S1(r, a, b, c) \
741	AS2( movq mm6, r)\
742	AS2( psrlq r, a)\
743	AS2( movq mm7, r)\
744	AS2( psllq mm6, 64-c)\
745	AS2( pxor mm7, mm6)\
746	AS2( psrlq r, b-a)\
747	AS2( pxor mm7, r)\
748	AS2( psllq mm6, c-b)\
749	AS2( pxor mm7, mm6)\
750	AS2( psrlq r, c-b)\
751	AS2( pxor r, mm7)\
752	AS2( psllq mm6, b-a)\
753	AS2( pxor r, mm6)
754
755	#define SSE2_s0(r, a, b, c) \
756	AS2( movdqa xmm6, r)\
757	AS2( psrlq r, a)\
758	AS2( movdqa xmm7, r)\
759	AS2( psllq xmm6, 64-c)\
760	AS2( pxor xmm7, xmm6)\
761	AS2( psrlq r, b-a)\
762	AS2( pxor xmm7, r)\
763	AS2( psrlq r, c-b)\
764	AS2( pxor r, xmm7)\
765	AS2( psllq xmm6, c-a)\
766	AS2( pxor r, xmm6)
767
768	#define SSE2_s1(r, a, b, c) \
769	AS2( movdqa xmm6, r)\
770	AS2( psrlq r, a)\
771	AS2( movdqa xmm7, r)\
772	AS2( psllq xmm6, 64-c)\
773	AS2( pxor xmm7, xmm6)\
774	AS2( psrlq r, b-a)\
775	AS2( pxor xmm7, r)\
776	AS2( psllq xmm6, c-b)\
777	AS2( pxor xmm7, xmm6)\
778	AS2( psrlq r, c-b)\
779	AS2( pxor r, xmm7)
780
781	ASL(SHA512_Round)
782	// k + w is in mm0, a is in mm4, e is in mm5
783	AS2( paddq mm0, [edi+7*8]) // h
784	AS2( movq mm2, [edi+5*8]) // f
785	AS2( movq mm3, [edi+6*8]) // g
786	AS2( pxor mm2, mm3)
787	AS2( pand mm2, mm5)
788	SSE2_S0_S1(mm5,14,18,41)
789	AS2( pxor mm2, mm3)
790	AS2( paddq mm0, mm2) // h += Ch(e,f,g)
791	AS2( paddq mm5, mm0) // h += S1(e)
792	AS2( movq mm2, [edi+1*8]) // b
793	AS2( movq mm1, mm2)
794	AS2( por mm2, mm4)
795	AS2( pand mm2, [edi+2*8]) // c
796	AS2( pand mm1, mm4)
797	AS2( por mm1, mm2)
798	AS2( paddq mm1, mm5) // temp = h + Maj(a,b,c)
799	AS2( paddq mm5, [edi+3*8]) // e = d + h
800	AS2( movq [edi+3*8], mm5)
801	AS2( movq [edi+11*8], mm5)
802	SSE2_S0_S1(mm4,28,34,39) // S0(a)
803	AS2( paddq mm4, mm1) // a = temp + S0(a)
804	AS2( movq [edi-8], mm4)
805	AS2( movq [edi+7*8], mm4)
806	AS1( ret)
807
808	// first 16 rounds
809	ASL(0)
810	AS2( movq mm0, [edx+eax*8])
811	AS2( movq [esi+eax*8], mm0)
812	AS2( movq [esi+eax8+168], mm0)
813	AS2( paddq mm0, [ebx+eax*8])
814	ASC( call, SHA512_Round)
815	AS1( inc eax)
816	AS2( sub edi, 8)
817	AS2( test eax, 7)
818	ASJ( jnz, 0, b)
819	AS2( add edi, 8*8)
820	AS2( cmp eax, 16)
821	ASJ( jne, 0, b)
822
823	// rest of the rounds
824	AS2( movdqu xmm0, [esi+(16-2)*8])
825	ASL(1)
826	// data expansion, W[i-2] already in xmm0
827	AS2( movdqu xmm3, [esi])
828	AS2( paddq xmm3, [esi+(16-7)*8])
829	AS2( movdqa xmm2, [esi+(16-15)*8])
830	SSE2_s1(xmm0, 6, 19, 61)
831	AS2( paddq xmm0, xmm3)
832	SSE2_s0(xmm2, 1, 7, 8)
833	AS2( paddq xmm0, xmm2)
834	AS2( movdq2q mm0, xmm0)
835	AS2( movhlps xmm1, xmm0)
836	AS2( paddq mm0, [ebx+eax*8])
837	AS2( movlps [esi], xmm0)
838	AS2( movlps [esi+8], xmm1)
839	AS2( movlps [esi+8*16], xmm0)
840	AS2( movlps [esi+8*17], xmm1)
841	// 2 rounds
842	ASC( call, SHA512_Round)
843	AS2( sub edi, 8)
844	AS2( movdq2q mm0, xmm1)
845	AS2( paddq mm0, [ebx+eax*8+8])
846	ASC( call, SHA512_Round)
847	// update indices and loop
848	AS2( add esi, 16)
849	AS2( add eax, 2)
850	AS2( sub edi, 8)
851	AS2( test eax, 7)
852	ASJ( jnz, 1, b)
853	// do housekeeping every 8 rounds
854	AS2( mov esi, 0xf)
855	AS2( and esi, eax)
856	#if CRYPTOPP_BOOL_X32
857	AS2( lea esi, [esp+8+208+8+esi8])
858	#else
859	AS2( lea esi, [esp+4+208+8+esi8])
860	#endif
861	AS2( add edi, 8*8)
862	AS2( cmp eax, 80)
863	ASJ( jne, 1, b)
864
865	#define SSE2_CombineState(i) \
866	AS2( movdqa xmm0, [edi+i*16])\
867	AS2( paddq xmm0, [ecx+i*16])\
868	AS2( movdqa [ecx+i*16], xmm0)
869
870	SSE2_CombineState(0)
871	SSE2_CombineState(1)
872	SSE2_CombineState(2)
873	SSE2_CombineState(3)
874
875	AS_POP_IF86( sp)
876	AS1( emms)
877
878	#if defined(__GNUC__)
879	AS_POP_IF86( bx)
880	ATT_PREFIX
881	:
882	: "a" (SHA512_K), "c" (state), "d" (data)
883	: "%esi", "%edi", "memory", "cc"
884	);
885	#else
886	AS1( pop edi)
887	AS1( pop esi)
888	AS1( pop ebx)
889	AS1( ret)
890	#endif
891	}
892	#endif // #if CRYPTOPP_BOOL_SSE2_ASM_AVAILABLE
893
894	void SHA512::Transform(word64 state, const word64 data)
895	{
896	CRYPTOPP_ASSERT(IsAlignedOn(state, GetAlignmentOf<word64>()));
897	CRYPTOPP_ASSERT(IsAlignedOn(data, GetAlignmentOf<word64>()));
898
899	#if CRYPTOPP_BOOL_SSE2_ASM_AVAILABLE && (CRYPTOPP_BOOL_X86 \|\| CRYPTOPP_BOOL_X32)
900	if (HasSSE2())
901	{
902	SHA512_SSE2_Transform(state, data);
903	return;
904	}
905	#endif
906
907	#define S0(x) (rotrFixed(x,28)^rotrFixed(x,34)^rotrFixed(x,39))
908	#define S1(x) (rotrFixed(x,14)^rotrFixed(x,18)^rotrFixed(x,41))
909	#define s0(x) (rotrFixed(x,1)^rotrFixed(x,8)^(x>>7))
910	#define s1(x) (rotrFixed(x,19)^rotrFixed(x,61)^(x>>6))
911
912	#define R(i) h(i)+=S1(e(i))+Ch(e(i),f(i),g(i))+SHA512_K[i+j]+(j?blk2(i):blk0(i));\
913	d(i)+=h(i);h(i)+=S0(a(i))+Maj(a(i),b(i),c(i))
914
915	word64 W[16];
916	word64 T[8];
917	/* Copy context->state[] to working vars */
918	memcpy(T, state, sizeof(T));
919	/* 80 operations, partially loop unrolled */
920	for (unsigned int j=0; j<80; j+=16)
921	{
922	R( 0); R( 1); R( 2); R( 3);
923	R( 4); R( 5); R( 6); R( 7);
924	R( 8); R( 9); R(10); R(11);
925	R(12); R(13); R(14); R(15);
926	}
927	/* Add the working vars back into context.state[] */
928	state[0] += a(0);
929	state[1] += b(0);
930	state[2] += c(0);
931	state[3] += d(0);
932	state[4] += e(0);
933	state[5] += f(0);
934	state[6] += g(0);
935	state[7] += h(0);
936	}
937
938	NAMESPACE_END
939
940	#endif // #ifndef CRYPTOPP_GENERATE_X64_MASM
941	#endif // #ifndef CRYPTOPP_IMPORTS

Note: See TracBrowser for help on using the repository browser.

Download in other formats: