arne
/
cryptopp
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Add XOP aware Salsa20

pull/730/head
Jeffrey Walton 2018-10-25 05:24:02 -04:00
parent daa07255ec
commit f33b19bf00
No known key found for this signature in database
GPG Key ID: B36AB348921B1838
1 changed files with 48 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

48
salsa.cpp
View File

@ -302,6 +302,13 @@ void Salsa20_Policy::OperateKeystream(KeystreamOperation operation, byte *output
AS2( mov dword ptr [REG_state + 8*4], REG_roundsLeft)
AS2( mov dword ptr [REG_state + 5*4], REG_temp32)
#ifdef __XOP__
#define SSE2_QUARTER_ROUND(a, b, d, i) \
AS2( movdqa xmm4, xmm##d) \
AS2( paddd xmm4, xmm##a) \
AS3( vprotd xmm4, xmm4, i) \
AS2( pxor xmm##b, xmm4)
#else
#define SSE2_QUARTER_ROUND(a, b, d, i) \
AS2( movdqa xmm4, xmm##d) \
AS2( paddd xmm4, xmm##a) \
@ -310,38 +317,79 @@ void Salsa20_Policy::OperateKeystream(KeystreamOperation operation, byte *output
AS2( psrld xmm5, 32-i) \
AS2( pxor xmm##b, xmm4) \
AS2( pxor xmm##b, xmm5)
#endif
#define L01(A,B,C,D,a,b,c,d,i) AS2( movdqa xmm##A, [SSE2_WORKSPACE + d*16 + i*256]) /* y3 */
#define L02(A,B,C,D,a,b,c,d,i) AS2( movdqa xmm##C, [SSE2_WORKSPACE + a*16 + i*256]) /* y0 */
#define L03(A,B,C,D,a,b,c,d,i) AS2( paddd xmm##A, xmm##C) /* y0+y3 */
#ifdef __XOP__
#define L04(A,B,C,D,a,b,c,d,i)
#define L05(A,B,C,D,a,b,c,d,i) AS3( vprotd xmm##A, xmm##A, 7)
#define L06(A,B,C,D,a,b,c,d,i)
#define L07(A,B,C,D,a,b,c,d,i) AS2( pxor xmm##A, [SSE2_WORKSPACE + b*16 + i*256])
#define L08(A,B,C,D,a,b,c,d,i)
#else
#define L04(A,B,C,D,a,b,c,d,i) AS2( movdqa xmm##B, xmm##A)
#define L05(A,B,C,D,a,b,c,d,i) AS2( pslld xmm##A, 7)
#define L06(A,B,C,D,a,b,c,d,i) AS2( psrld xmm##B, 32-7)
#define L07(A,B,C,D,a,b,c,d,i) AS2( pxor xmm##A, [SSE2_WORKSPACE + b*16 + i*256])
#define L08(A,B,C,D,a,b,c,d,i) AS2( pxor xmm##A, xmm##B) /* z1 */
#endif
#define L09(A,B,C,D,a,b,c,d,i) AS2( movdqa [SSE2_WORKSPACE + b*16], xmm##A)
#define L10(A,B,C,D,a,b,c,d,i) AS2( movdqa xmm##B, xmm##A)
#define L11(A,B,C,D,a,b,c,d,i) AS2( paddd xmm##A, xmm##C) /* z1+y0 */
#ifdef __XOP__
#define L12(A,B,C,D,a,b,c,d,i)
#define L13(A,B,C,D,a,b,c,d,i) AS3( vprotd xmm##A, xmm##A, 9)
#define L14(A,B,C,D,a,b,c,d,i)
#define L15(A,B,C,D,a,b,c,d,i) AS2( pxor xmm##A, [SSE2_WORKSPACE + c*16 + i*256])
#define L16(A,B,C,D,a,b,c,d,i)
#else
#define L12(A,B,C,D,a,b,c,d,i) AS2( movdqa xmm##D, xmm##A)
#define L13(A,B,C,D,a,b,c,d,i) AS2( pslld xmm##A, 9)
#define L14(A,B,C,D,a,b,c,d,i) AS2( psrld xmm##D, 32-9)
#define L15(A,B,C,D,a,b,c,d,i) AS2( pxor xmm##A, [SSE2_WORKSPACE + c*16 + i*256])
#define L16(A,B,C,D,a,b,c,d,i) AS2( pxor xmm##A, xmm##D) /* z2 */
#endif
#define L17(A,B,C,D,a,b,c,d,i) AS2( movdqa [SSE2_WORKSPACE + c*16], xmm##A)
#define L18(A,B,C,D,a,b,c,d,i) AS2( movdqa xmm##D, xmm##A)
#define L19(A,B,C,D,a,b,c,d,i) AS2( paddd xmm##A, xmm##B) /* z2+z1 */
#ifdef __XOP__
#define L20(A,B,C,D,a,b,c,d,i)
#define L21(A,B,C,D,a,b,c,d,i) AS3( vprotd xmm##A, xmm##A, 13)
#define L22(A,B,C,D,a,b,c,d,i)
#define L23(A,B,C,D,a,b,c,d,i) AS2( pxor xmm##A, [SSE2_WORKSPACE + d*16 + i*256])
#define L24(A,B,C,D,a,b,c,d,i)
#else
#define L20(A,B,C,D,a,b,c,d,i) AS2( movdqa xmm##B, xmm##A)
#define L21(A,B,C,D,a,b,c,d,i) AS2( pslld xmm##A, 13)
#define L22(A,B,C,D,a,b,c,d,i) AS2( psrld xmm##B, 32-13)
#define L23(A,B,C,D,a,b,c,d,i) AS2( pxor xmm##A, [SSE2_WORKSPACE + d*16 + i*256])
#define L24(A,B,C,D,a,b,c,d,i) AS2( pxor xmm##A, xmm##B) /* z3 */
#endif
#define L25(A,B,C,D,a,b,c,d,i) AS2( movdqa [SSE2_WORKSPACE + d*16], xmm##A)
#define L26(A,B,C,D,a,b,c,d,i) AS2( paddd xmm##A, xmm##D) /* z3+z2 */
#ifdef __XOP__
#define L27(A,B,C,D,a,b,c,d,i)
#define L28(A,B,C,D,a,b,c,d,i) AS3( vprotd xmm##A, xmm##A, 18)
#define L29(A,B,C,D,a,b,c,d,i)
#define L30(A,B,C,D,a,b,c,d,i) AS2( pxor xmm##A, xmm##C) /* xor y0 */
#define L31(A,B,C,D,a,b,c,d,i)
#else
#define L27(A,B,C,D,a,b,c,d,i) AS2( movdqa xmm##D, xmm##A)
#define L28(A,B,C,D,a,b,c,d,i) AS2( pslld xmm##A, 18)
#define L29(A,B,C,D,a,b,c,d,i) AS2( psrld xmm##D, 32-18)
#define L30(A,B,C,D,a,b,c,d,i) AS2( pxor xmm##A, xmm##C) /* xor y0 */
#define L31(A,B,C,D,a,b,c,d,i) AS2( pxor xmm##A, xmm##D) /* z0 */
#endif
#define L32(A,B,C,D,a,b,c,d,i) AS2( movdqa [SSE2_WORKSPACE + a*16], xmm##A)
#define SSE2_QUARTER_ROUND_X8(i, a, b, c, d, e, f, g, h) \