From 162650303c3b317f41a9734abf7aeff6836cad7f Mon Sep 17 00:00:00 2001 From: Andrea Mazzoleni Date: Sat, 26 Oct 2013 16:35:35 +0200 Subject: [PATCH] Uses more generic recovering functions with vectors instead of variables. --- raid.c | 1621 +++++++++++++++++++++----------------------------------- raid.h | 2 +- 2 files changed, 619 insertions(+), 1004 deletions(-) diff --git a/raid.c b/raid.c index a50d2cca..f58adc1a 100644 --- a/raid.c +++ b/raid.c @@ -153,17 +153,17 @@ * R = sum(A[2,i] * Di) * S = sum(A[3,i] * Di) * T = sum(A[4,i] * Di) - * O = sum(A[5,i] * Di) with 0<=i=0;--d) { - p0 ^= v_32(vbuf[d][off]); - p1 ^= v_32(vbuf[d][off+4]); + p0 ^= v_32(vbuf[d][i]); + p1 ^= v_32(vbuf[d][i+4]); } - v_32(p[off]) = p0; - v_32(p[off+4]) = p1; + v_32(p[i]) = p0; + v_32(p[i+4]) = p1; } } @@ -252,7 +252,7 @@ void raid5_int64(unsigned char** vbuf, unsigned data, unsigned size) { unsigned char* p; int d, l; - unsigned off; + unsigned i; uint64_t p0; uint64_t p1; @@ -260,15 +260,15 @@ void raid5_int64(unsigned char** vbuf, unsigned data, unsigned size) l = data - 1; p = vbuf[data]; - for(off=0;off=0;--d) { - p0 ^= v_64(vbuf[d][off]); - p1 ^= v_64(vbuf[d][off+8]); + p0 ^= v_64(vbuf[d][i]); + p1 ^= v_64(vbuf[d][i+8]); } - v_64(p[off]) = p0; - v_64(p[off+8]) = p1; + v_64(p[i]) = p0; + v_64(p[i+8]) = p1; } } @@ -285,30 +285,30 @@ void raid5_sse2(unsigned char** vbuf, unsigned data, unsigned size) { unsigned char* p; int d, l; - unsigned off; + unsigned i; l = data - 1; p = vbuf[data]; - for(off=0;off=0;--d) { - asm volatile("movdqa %0,%%xmm4" : : "m" (vbuf[d][off])); - asm volatile("movdqa %0,%%xmm5" : : "m" (vbuf[d][off+16])); - asm volatile("movdqa %0,%%xmm6" : : "m" (vbuf[d][off+32])); - asm volatile("movdqa %0,%%xmm7" : : "m" (vbuf[d][off+48])); + asm volatile("movdqa %0,%%xmm4" : : "m" (vbuf[d][i])); + asm volatile("movdqa %0,%%xmm5" : : "m" (vbuf[d][i+16])); + asm volatile("movdqa %0,%%xmm6" : : "m" (vbuf[d][i+32])); + asm volatile("movdqa %0,%%xmm7" : : "m" (vbuf[d][i+48])); asm volatile("pxor %xmm4,%xmm0"); asm volatile("pxor %xmm5,%xmm1"); asm volatile("pxor %xmm6,%xmm2"); asm volatile("pxor %xmm7,%xmm3"); } - asm volatile("movntdq %%xmm0,%0" : "=m" (p[off])); - asm volatile("movntdq %%xmm1,%0" : "=m" (p[off+16])); - asm volatile("movntdq %%xmm2,%0" : "=m" (p[off+32])); - asm volatile("movntdq %%xmm3,%0" : "=m" (p[off+48])); + asm volatile("movntdq %%xmm0,%0" : "=m" (p[i])); + asm volatile("movntdq %%xmm1,%0" : "=m" (p[i+16])); + asm volatile("movntdq %%xmm2,%0" : "=m" (p[i+32])); + asm volatile("movntdq %%xmm3,%0" : "=m" (p[i+48])); } asm volatile("sfence" : : : "memory"); @@ -347,7 +347,7 @@ void raid6_int32(unsigned char** vbuf, unsigned data, unsigned size) unsigned char* p; unsigned char* q; int d, l; - unsigned off; + unsigned i; uint32_t d0, q0, p0; uint32_t d1, q1, p1; @@ -356,12 +356,12 @@ void raid6_int32(unsigned char** vbuf, unsigned data, unsigned size) p = vbuf[data]; q = vbuf[data+1]; - for(off=0;off=0;--d) { - d0 = v_32(vbuf[d][off]); - d1 = v_32(vbuf[d][off+4]); + d0 = v_32(vbuf[d][i]); + d1 = v_32(vbuf[d][i+4]); p0 ^= d0; p1 ^= d1; @@ -372,10 +372,10 @@ void raid6_int32(unsigned char** vbuf, unsigned data, unsigned size) q0 ^= d0; q1 ^= d1; } - v_32(p[off]) = p0; - v_32(p[off+4]) = p1; - v_32(q[off]) = q0; - v_32(q[off+4]) = q1; + v_32(p[i]) = p0; + v_32(p[i+4]) = p1; + v_32(q[i]) = q0; + v_32(q[i+4]) = q1; } } @@ -387,7 +387,7 @@ void raid6_int64(unsigned char** vbuf, unsigned data, unsigned size) unsigned char* p; unsigned char* q; int d, l; - unsigned off; + unsigned i; uint64_t d0, q0, p0; uint64_t d1, q1, p1; @@ -396,12 +396,12 @@ void raid6_int64(unsigned char** vbuf, unsigned data, unsigned size) p = vbuf[data]; q = vbuf[data+1]; - for(off=0;off=0;--d) { - d0 = v_64(vbuf[d][off]); - d1 = v_64(vbuf[d][off+8]); + d0 = v_64(vbuf[d][i]); + d1 = v_64(vbuf[d][i+8]); p0 ^= d0; p1 ^= d1; @@ -412,10 +412,10 @@ void raid6_int64(unsigned char** vbuf, unsigned data, unsigned size) q0 ^= d0; q1 ^= d1; } - v_64(p[off]) = p0; - v_64(p[off+8]) = p1; - v_64(q[off]) = q0; - v_64(q[off+8]) = q1; + v_64(p[i]) = p0; + v_64(p[i+8]) = p1; + v_64(q[i]) = q0; + v_64(q[i+8]) = q1; } } @@ -436,7 +436,7 @@ void raid6_sse2(unsigned char** vbuf, unsigned data, unsigned size) unsigned char* p; unsigned char* q; int d, l; - unsigned off; + unsigned i; l = data - 1; p = vbuf[data]; @@ -444,9 +444,9 @@ void raid6_sse2(unsigned char** vbuf, unsigned data, unsigned size) asm volatile("movdqa %0,%%xmm7" : : "m" (raid_16_const.poly[0])); - for(off=0;off=0;--d) { @@ -461,17 +461,17 @@ void raid6_sse2(unsigned char** vbuf, unsigned data, unsigned size) asm volatile("pxor %xmm4,%xmm2"); asm volatile("pxor %xmm5,%xmm3"); - asm volatile("movdqa %0,%%xmm4" : : "m" (vbuf[d][off])); - asm volatile("movdqa %0,%%xmm5" : : "m" (vbuf[d][off+16])); + asm volatile("movdqa %0,%%xmm4" : : "m" (vbuf[d][i])); + asm volatile("movdqa %0,%%xmm5" : : "m" (vbuf[d][i+16])); asm volatile("pxor %xmm4,%xmm0"); asm volatile("pxor %xmm5,%xmm1"); asm volatile("pxor %xmm4,%xmm2"); asm volatile("pxor %xmm5,%xmm3"); } - asm volatile("movntdq %%xmm0,%0" : "=m" (p[off])); - asm volatile("movntdq %%xmm1,%0" : "=m" (p[off+16])); - asm volatile("movntdq %%xmm2,%0" : "=m" (q[off])); - asm volatile("movntdq %%xmm3,%0" : "=m" (q[off+16])); + asm volatile("movntdq %%xmm0,%0" : "=m" (p[i])); + asm volatile("movntdq %%xmm1,%0" : "=m" (p[i+16])); + asm volatile("movntdq %%xmm2,%0" : "=m" (q[i])); + asm volatile("movntdq %%xmm3,%0" : "=m" (q[i+16])); } asm volatile("sfence" : : : "memory"); @@ -489,7 +489,7 @@ void raid6_sse2ext(unsigned char** vbuf, unsigned data, unsigned size) unsigned char* p; unsigned char* q; int d, l; - unsigned off; + unsigned i; l = data - 1; p = vbuf[data]; @@ -497,11 +497,11 @@ void raid6_sse2ext(unsigned char** vbuf, unsigned data, unsigned size) asm volatile("movdqa %0,%%xmm15" : : "m" (raid_16_const.poly[0])); - for(off=0;off=0;--d) { - d0 = v_8(vbuf[d][off]); + d0 = v_8(vbuf[d][i]); p0 ^= d0; q0 ^= gfmul[d0][gfmatrix[1][d]]; r0 ^= gfmul[d0][gfmatrix[2][d]]; } - v_8(p[off]) = p0; - v_8(q[off]) = q0; - v_8(r[off]) = r0; + v_8(p[i]) = p0; + v_8(q[i]) = q0; + v_8(r[i]) = r0; } } @@ -602,7 +602,7 @@ void raidTP_ssse3(unsigned char** vbuf, unsigned data, unsigned size) unsigned char* q; unsigned char* r; int d, l; - unsigned off; + unsigned i; l = data - 1; p = vbuf[data]; @@ -612,12 +612,12 @@ void raidTP_ssse3(unsigned char** vbuf, unsigned data, unsigned size) asm volatile("movdqa %0,%%xmm3" : : "m" (raid_16_const.poly[0])); asm volatile("movdqa %0,%%xmm7" : : "m" (raid_16_const.mask[0])); - for(off=0;off=0;--d) { - asm volatile("movdqa %0,%%xmm4" : : "m" (vbuf[d][off])); + asm volatile("movdqa %0,%%xmm4" : : "m" (vbuf[d][i])); asm volatile("pxor %xmm5,%xmm5"); asm volatile("pcmpgtb %xmm1,%xmm5"); @@ -640,9 +640,9 @@ void raidTP_ssse3(unsigned char** vbuf, unsigned data, unsigned size) asm volatile("pshufb %xmm5,%xmm6"); asm volatile("pxor %xmm6,%xmm2"); } - asm volatile("movntdq %%xmm0,%0" : "=m" (p[off])); - asm volatile("movntdq %%xmm1,%0" : "=m" (q[off])); - asm volatile("movntdq %%xmm2,%0" : "=m" (r[off])); + asm volatile("movntdq %%xmm0,%0" : "=m" (p[i])); + asm volatile("movntdq %%xmm1,%0" : "=m" (q[i])); + asm volatile("movntdq %%xmm2,%0" : "=m" (r[i])); } asm volatile("sfence" : : : "memory"); @@ -661,7 +661,7 @@ void raidTP_ssse3ext(unsigned char** vbuf, unsigned data, unsigned size) unsigned char* q; unsigned char* r; int d, l; - unsigned off; + unsigned i; l = data - 1; p = vbuf[data]; @@ -671,7 +671,7 @@ void raidTP_ssse3ext(unsigned char** vbuf, unsigned data, unsigned size) asm volatile("movdqa %0,%%xmm3" : : "m" (raid_16_const.poly[0])); asm volatile("movdqa %0,%%xmm11" : : "m" (raid_16_const.mask[0])); - for(off=0;off=0;--d) { - asm volatile("movdqa %0,%%xmm4" : : "m" (vbuf[d][off])); - asm volatile("movdqa %0,%%xmm12" : : "m" (vbuf[d][off+16])); + asm volatile("movdqa %0,%%xmm4" : : "m" (vbuf[d][i])); + asm volatile("movdqa %0,%%xmm12" : : "m" (vbuf[d][i+16])); asm volatile("pxor %xmm5,%xmm5"); asm volatile("pxor %xmm13,%xmm13"); @@ -720,12 +720,12 @@ void raidTP_ssse3ext(unsigned char** vbuf, unsigned data, unsigned size) asm volatile("pxor %xmm7,%xmm2"); asm volatile("pxor %xmm15,%xmm10"); } - asm volatile("movntdq %%xmm0,%0" : "=m" (p[off])); - asm volatile("movntdq %%xmm8,%0" : "=m" (p[off+16])); - asm volatile("movntdq %%xmm1,%0" : "=m" (q[off])); - asm volatile("movntdq %%xmm9,%0" : "=m" (q[off+16])); - asm volatile("movntdq %%xmm2,%0" : "=m" (r[off])); - asm volatile("movntdq %%xmm10,%0" : "=m" (r[off+16])); + asm volatile("movntdq %%xmm0,%0" : "=m" (p[i])); + asm volatile("movntdq %%xmm8,%0" : "=m" (p[i+16])); + asm volatile("movntdq %%xmm1,%0" : "=m" (q[i])); + asm volatile("movntdq %%xmm9,%0" : "=m" (q[i+16])); + asm volatile("movntdq %%xmm2,%0" : "=m" (r[i])); + asm volatile("movntdq %%xmm10,%0" : "=m" (r[i+16])); } asm volatile("sfence" : : : "memory"); @@ -746,7 +746,7 @@ void raidQP_int8(unsigned char** vbuf, unsigned data, unsigned size) unsigned char* r; unsigned char* s; int d, l; - unsigned off; + unsigned i; uint8_t d0, s0, r0, q0, p0; @@ -756,20 +756,20 @@ void raidQP_int8(unsigned char** vbuf, unsigned data, unsigned size) r = vbuf[data+2]; s = vbuf[data+3]; - for(off=0;off=0;--d) { - d0 = v_8(vbuf[d][off]); + d0 = v_8(vbuf[d][i]); p0 ^= d0; q0 ^= gfmul[d0][gfmatrix[1][d]]; r0 ^= gfmul[d0][gfmatrix[2][d]]; s0 ^= gfmul[d0][gfmatrix[3][d]]; } - v_8(p[off]) = p0; - v_8(q[off]) = q0; - v_8(r[off]) = r0; - v_8(s[off]) = s0; + v_8(p[i]) = p0; + v_8(q[i]) = q0; + v_8(r[i]) = r0; + v_8(s[i]) = s0; } } @@ -784,7 +784,7 @@ void raidQP_ssse3(unsigned char** vbuf, unsigned data, unsigned size) unsigned char* r; unsigned char* s; int d, l; - unsigned off; + unsigned i; l = data - 1; p = vbuf[data]; @@ -792,14 +792,14 @@ void raidQP_ssse3(unsigned char** vbuf, unsigned data, unsigned size) r = vbuf[data+2]; s = vbuf[data+3]; - for(off=0;off=0;--d) { asm volatile("movdqa %0,%%xmm7" : : "m" (raid_16_const.poly[0])); - asm volatile("movdqa %0,%%xmm4" : : "m" (vbuf[d][off])); + asm volatile("movdqa %0,%%xmm4" : : "m" (vbuf[d][i])); asm volatile("pxor %xmm5,%xmm5"); asm volatile("pcmpgtb %xmm1,%xmm5"); @@ -831,10 +831,10 @@ void raidQP_ssse3(unsigned char** vbuf, unsigned data, unsigned size) asm volatile("pxor %xmm6,%xmm3"); asm volatile("pxor %xmm7,%xmm3"); } - asm volatile("movntdq %%xmm0,%0" : "=m" (p[off])); - asm volatile("movntdq %%xmm1,%0" : "=m" (q[off])); - asm volatile("movntdq %%xmm2,%0" : "=m" (r[off])); - asm volatile("movntdq %%xmm3,%0" : "=m" (s[off])); + asm volatile("movntdq %%xmm0,%0" : "=m" (p[i])); + asm volatile("movntdq %%xmm1,%0" : "=m" (q[i])); + asm volatile("movntdq %%xmm2,%0" : "=m" (r[i])); + asm volatile("movntdq %%xmm3,%0" : "=m" (s[i])); } asm volatile("sfence" : : : "memory"); @@ -854,7 +854,7 @@ void raidQP_ssse3ext(unsigned char** vbuf, unsigned data, unsigned size) unsigned char* r; unsigned char* s; int d, l; - unsigned off; + unsigned i; l = data - 1; p = vbuf[data]; @@ -862,7 +862,7 @@ void raidQP_ssse3ext(unsigned char** vbuf, unsigned data, unsigned size) r = vbuf[data+2]; s = vbuf[data+3]; - for(off=0;off=0;--d) { asm volatile("movdqa %0,%%xmm7" : : "m" (raid_16_const.poly[0])); asm volatile("movdqa %0,%%xmm15" : : "m" (raid_16_const.mask[0])); - asm volatile("movdqa %0,%%xmm4" : : "m" (vbuf[d][off])); - asm volatile("movdqa %0,%%xmm12" : : "m" (vbuf[d][off+16])); + asm volatile("movdqa %0,%%xmm4" : : "m" (vbuf[d][i])); + asm volatile("movdqa %0,%%xmm12" : : "m" (vbuf[d][i+16])); asm volatile("pxor %xmm5,%xmm5"); asm volatile("pxor %xmm13,%xmm13"); @@ -928,14 +928,14 @@ void raidQP_ssse3ext(unsigned char** vbuf, unsigned data, unsigned size) asm volatile("pxor %xmm7,%xmm3"); asm volatile("pxor %xmm15,%xmm11"); } - asm volatile("movntdq %%xmm0,%0" : "=m" (p[off])); - asm volatile("movntdq %%xmm8,%0" : "=m" (p[off+16])); - asm volatile("movntdq %%xmm1,%0" : "=m" (q[off])); - asm volatile("movntdq %%xmm9,%0" : "=m" (q[off+16])); - asm volatile("movntdq %%xmm2,%0" : "=m" (r[off])); - asm volatile("movntdq %%xmm10,%0" : "=m" (r[off+16])); - asm volatile("movntdq %%xmm3,%0" : "=m" (s[off])); - asm volatile("movntdq %%xmm11,%0" : "=m" (s[off+16])); + asm volatile("movntdq %%xmm0,%0" : "=m" (p[i])); + asm volatile("movntdq %%xmm8,%0" : "=m" (p[i+16])); + asm volatile("movntdq %%xmm1,%0" : "=m" (q[i])); + asm volatile("movntdq %%xmm9,%0" : "=m" (q[i+16])); + asm volatile("movntdq %%xmm2,%0" : "=m" (r[i])); + asm volatile("movntdq %%xmm10,%0" : "=m" (r[i+16])); + asm volatile("movntdq %%xmm3,%0" : "=m" (s[i])); + asm volatile("movntdq %%xmm11,%0" : "=m" (s[i+16])); } asm volatile("sfence" : : : "memory"); @@ -957,7 +957,7 @@ void raidPP_int8(unsigned char** vbuf, unsigned data, unsigned size) unsigned char* s; unsigned char* t; int d, l; - unsigned off; + unsigned i; uint8_t d0, t0, s0, r0, q0, p0; @@ -968,10 +968,10 @@ void raidPP_int8(unsigned char** vbuf, unsigned data, unsigned size) s = vbuf[data+3]; t = vbuf[data+4]; - for(off=0;off=0;--d) { - d0 = v_8(vbuf[d][off]); + d0 = v_8(vbuf[d][i]); p0 ^= d0; q0 ^= gfmul[d0][gfmatrix[1][d]]; @@ -979,11 +979,11 @@ void raidPP_int8(unsigned char** vbuf, unsigned data, unsigned size) s0 ^= gfmul[d0][gfmatrix[3][d]]; t0 ^= gfmul[d0][gfmatrix[4][d]]; } - v_8(p[off]) = p0; - v_8(q[off]) = q0; - v_8(r[off]) = r0; - v_8(s[off]) = s0; - v_8(t[off]) = t0; + v_8(p[i]) = p0; + v_8(q[i]) = q0; + v_8(r[i]) = r0; + v_8(s[i]) = s0; + v_8(t[i]) = t0; } } @@ -1004,7 +1004,7 @@ void raidPP_ssse3(unsigned char** vbuf, unsigned data, unsigned size) unsigned char* s; unsigned char* t; int d, l; - unsigned off; + unsigned i; unsigned char p0[16] __attribute__((aligned(16))); l = data - 1; @@ -1014,14 +1014,14 @@ void raidPP_ssse3(unsigned char** vbuf, unsigned data, unsigned size) s = vbuf[data+3]; t = vbuf[data+4]; - for(off=0;off=0;--d) { - asm volatile("movdqa %0,%%xmm4" : : "m" (vbuf[d][off])); + asm volatile("movdqa %0,%%xmm4" : : "m" (vbuf[d][i])); asm volatile("movdqa %0,%%xmm6" : : "m" (p0[0])); asm volatile("movdqa %0,%%xmm7" : : "m" (raid_16_const.poly[0])); @@ -1063,11 +1063,11 @@ void raidPP_ssse3(unsigned char** vbuf, unsigned data, unsigned size) asm volatile("pxor %xmm7,%xmm3"); } asm volatile("movdqa %0,%%xmm6" : : "m" (p0[0])); - asm volatile("movntdq %%xmm6,%0" : "=m" (p[off])); - asm volatile("movntdq %%xmm0,%0" : "=m" (q[off])); - asm volatile("movntdq %%xmm1,%0" : "=m" (r[off])); - asm volatile("movntdq %%xmm2,%0" : "=m" (s[off])); - asm volatile("movntdq %%xmm3,%0" : "=m" (t[off])); + asm volatile("movntdq %%xmm6,%0" : "=m" (p[i])); + asm volatile("movntdq %%xmm0,%0" : "=m" (q[i])); + asm volatile("movntdq %%xmm1,%0" : "=m" (r[i])); + asm volatile("movntdq %%xmm2,%0" : "=m" (s[i])); + asm volatile("movntdq %%xmm3,%0" : "=m" (t[i])); } asm volatile("sfence" : : : "memory"); @@ -1088,7 +1088,7 @@ void raidPP_ssse3ext(unsigned char** vbuf, unsigned data, unsigned size) unsigned char* s; unsigned char* t; int d, l; - unsigned off; + unsigned i; l = data - 1; p = vbuf[data]; @@ -1100,14 +1100,14 @@ void raidPP_ssse3ext(unsigned char** vbuf, unsigned data, unsigned size) asm volatile("movdqa %0,%%xmm14" : : "m" (raid_16_const.poly[0])); asm volatile("movdqa %0,%%xmm15" : : "m" (raid_16_const.mask[0])); - for(off=0;off=0;--d) { - asm volatile("movdqa %0,%%xmm10" : : "m" (vbuf[d][off])); + asm volatile("movdqa %0,%%xmm10" : : "m" (vbuf[d][i])); asm volatile("pxor %xmm11,%xmm11"); asm volatile("pcmpgtb %xmm1,%xmm11"); @@ -1144,11 +1144,11 @@ void raidPP_ssse3ext(unsigned char** vbuf, unsigned data, unsigned size) asm volatile("pxor %xmm12,%xmm4"); asm volatile("pxor %xmm13,%xmm4"); } - asm volatile("movntdq %%xmm0,%0" : "=m" (p[off])); - asm volatile("movntdq %%xmm1,%0" : "=m" (q[off])); - asm volatile("movntdq %%xmm2,%0" : "=m" (r[off])); - asm volatile("movntdq %%xmm3,%0" : "=m" (s[off])); - asm volatile("movntdq %%xmm4,%0" : "=m" (t[off])); + asm volatile("movntdq %%xmm0,%0" : "=m" (p[i])); + asm volatile("movntdq %%xmm1,%0" : "=m" (q[i])); + asm volatile("movntdq %%xmm2,%0" : "=m" (r[i])); + asm volatile("movntdq %%xmm3,%0" : "=m" (s[i])); + asm volatile("movntdq %%xmm4,%0" : "=m" (t[i])); } asm volatile("sfence" : : : "memory"); @@ -1169,9 +1169,9 @@ void raidHP_int8(unsigned char** vbuf, unsigned data, unsigned size) unsigned char* r; unsigned char* s; unsigned char* t; - unsigned char* o; + unsigned char* u; int d, l; - unsigned off; + unsigned i; uint8_t d0, u0, t0, s0, r0, q0, p0; @@ -1181,12 +1181,12 @@ void raidHP_int8(unsigned char** vbuf, unsigned data, unsigned size) r = vbuf[data+2]; s = vbuf[data+3]; t = vbuf[data+4]; - o = vbuf[data+5]; + u = vbuf[data+5]; - for(off=0;off=0;--d) { - d0 = v_8(vbuf[d][off]); + d0 = v_8(vbuf[d][i]); p0 ^= d0; q0 ^= gfmul[d0][gfmatrix[1][d]]; @@ -1195,12 +1195,12 @@ void raidHP_int8(unsigned char** vbuf, unsigned data, unsigned size) t0 ^= gfmul[d0][gfmatrix[4][d]]; u0 ^= gfmul[d0][gfmatrix[5][d]]; } - v_8(p[off]) = p0; - v_8(q[off]) = q0; - v_8(r[off]) = r0; - v_8(s[off]) = s0; - v_8(t[off]) = t0; - v_8(o[off]) = u0; + v_8(p[i]) = p0; + v_8(q[i]) = q0; + v_8(r[i]) = r0; + v_8(s[i]) = s0; + v_8(t[i]) = t0; + v_8(u[i]) = u0; } } @@ -1219,9 +1219,9 @@ void raidHP_ssse3(unsigned char** vbuf, unsigned data, unsigned size) unsigned char* r; unsigned char* s; unsigned char* t; - unsigned char* o; + unsigned char* u; int d, l; - unsigned off; + unsigned i; unsigned char p0[16] __attribute__((aligned(16))); unsigned char q0[16] __attribute__((aligned(16))); @@ -1231,9 +1231,9 @@ void raidHP_ssse3(unsigned char** vbuf, unsigned data, unsigned size) r = vbuf[data+2]; s = vbuf[data+3]; t = vbuf[data+4]; - o = vbuf[data+5]; + u = vbuf[data+5]; - for(off=0;off=0;--d) { - asm volatile("movdqa %0,%%xmm10" : : "m" (vbuf[d][off])); + asm volatile("movdqa %0,%%xmm10" : : "m" (vbuf[d][i])); asm volatile("pxor %xmm11,%xmm11"); asm volatile("pcmpgtb %xmm1,%xmm11"); @@ -1386,12 +1386,12 @@ void raidHP_ssse3ext(unsigned char** vbuf, unsigned data, unsigned size) asm volatile("pxor %xmm12,%xmm5"); asm volatile("pxor %xmm13,%xmm5"); } - asm volatile("movntdq %%xmm0,%0" : "=m" (p[off])); - asm volatile("movntdq %%xmm1,%0" : "=m" (q[off])); - asm volatile("movntdq %%xmm2,%0" : "=m" (r[off])); - asm volatile("movntdq %%xmm3,%0" : "=m" (s[off])); - asm volatile("movntdq %%xmm4,%0" : "=m" (t[off])); - asm volatile("movntdq %%xmm5,%0" : "=m" (o[off])); + asm volatile("movntdq %%xmm0,%0" : "=m" (p[i])); + asm volatile("movntdq %%xmm1,%0" : "=m" (q[i])); + asm volatile("movntdq %%xmm2,%0" : "=m" (r[i])); + asm volatile("movntdq %%xmm3,%0" : "=m" (s[i])); + asm volatile("movntdq %%xmm4,%0" : "=m" (t[i])); + asm volatile("movntdq %%xmm5,%0" : "=m" (u[i])); } asm volatile("sfence" : : : "memory"); @@ -1474,36 +1474,36 @@ static inline const unsigned char* table(unsigned char v) * Computes the parity without the missing data blocks * and store it in the buffers of such data blocks. */ -static void raid_delta_1data(int x, int i, unsigned char** vbuf, unsigned data, unsigned char* zero, unsigned size) +static void raid_delta_1data(const int* d, const int* e, unsigned char** vbuf, unsigned data, unsigned char* zero, unsigned size) { unsigned char* p; unsigned char* pa; /* parity buffer */ - p = vbuf[data+i]; + p = vbuf[data+e[0]]; /* missing data blcks */ - pa = vbuf[x]; + pa = vbuf[d[0]]; /* set at zero the missing data blocks */ - vbuf[x] = zero; + vbuf[d[0]] = zero; /* compute only for the level we really need */ - vbuf[data+i] = pa; + vbuf[data+e[0]] = pa; /* compute the parity */ - raid_gen(i + 1, vbuf, data, size); + raid_gen(e[0] + 1, vbuf, data, size); /* restore as before */ - vbuf[x] = pa; - vbuf[data+i] = p; + vbuf[d[0]] = pa; + vbuf[data+e[0]] = p; } /** * Computes the parity without the missing data blocks * and store it in the buffers of such data blocks. */ -static void raid_delta_2data(unsigned int x, int y, int i, int j, unsigned char** vbuf, unsigned data, unsigned char* zero, unsigned size) +static void raid_delta_2data(const int* d, const int* e, unsigned char** vbuf, unsigned data, unsigned char* zero, unsigned size) { unsigned char* p; unsigned char* q; @@ -1511,36 +1511,36 @@ static void raid_delta_2data(unsigned int x, int y, int i, int j, unsigned char* unsigned char* qa; /* parity buffer */ - p = vbuf[data+i]; - q = vbuf[data+j]; + p = vbuf[data+e[0]]; + q = vbuf[data+e[1]]; /* missing data blocks */ - pa = vbuf[x]; - qa = vbuf[y]; + pa = vbuf[d[0]]; + qa = vbuf[d[1]]; /* set at zero the missing data blocks */ - vbuf[x] = zero; - vbuf[y] = zero; + vbuf[d[0]] = zero; + vbuf[d[1]] = zero; /* compute the parity over the missing data blocks */ - vbuf[data+i] = pa; - vbuf[data+j] = qa; + vbuf[data+e[0]] = pa; + vbuf[data+e[1]] = qa; /* compute only for the level we really need */ - raid_gen(j + 1, vbuf, data, size); + raid_gen(e[1] + 1, vbuf, data, size); /* restore as before */ - vbuf[x] = pa; - vbuf[y] = qa; - vbuf[data+i] = p; - vbuf[data+j] = q; + vbuf[d[0]] = pa; + vbuf[d[1]] = qa; + vbuf[data+e[0]] = p; + vbuf[data+e[1]] = q; } /** * Computes the parity without the missing data blocks * and store it in the buffers of such data blocks. */ -static void raid_delta_3data(unsigned int x, int y, int z, int i, int j, int k, unsigned char** vbuf, unsigned data, unsigned char* zero, unsigned size) +static void raid_delta_3data(const int* d, const int* e, unsigned char** vbuf, unsigned data, unsigned char* zero, unsigned size) { unsigned char* p; unsigned char* q; @@ -1550,42 +1550,42 @@ static void raid_delta_3data(unsigned int x, int y, int z, int i, int j, int k, unsigned char* ra; /* parity buffer */ - p = vbuf[data+i]; - q = vbuf[data+j]; - r = vbuf[data+k]; + p = vbuf[data+e[0]]; + q = vbuf[data+e[1]]; + r = vbuf[data+e[2]]; /* missing data blocks */ - pa = vbuf[x]; - qa = vbuf[y]; - ra = vbuf[z]; + pa = vbuf[d[0]]; + qa = vbuf[d[1]]; + ra = vbuf[d[2]]; /* set at zero the missing data blocks */ - vbuf[x] = zero; - vbuf[y] = zero; - vbuf[z] = zero; + vbuf[d[0]] = zero; + vbuf[d[1]] = zero; + vbuf[d[2]] = zero; /* compute the parity over the missing data blocks */ - vbuf[data+i] = pa; - vbuf[data+j] = qa; - vbuf[data+k] = ra; + vbuf[data+e[0]] = pa; + vbuf[data+e[1]] = qa; + vbuf[data+e[2]] = ra; /* compute only for the level we really need */ - raid_gen(k + 1, vbuf, data, size); + raid_gen(e[2] + 1, vbuf, data, size); /* restore as before */ - vbuf[x] = pa; - vbuf[y] = qa; - vbuf[z] = ra; - vbuf[data+i] = p; - vbuf[data+j] = q; - vbuf[data+k] = r; + vbuf[d[0]] = pa; + vbuf[d[1]] = qa; + vbuf[d[2]] = ra; + vbuf[data+e[0]] = p; + vbuf[data+e[1]] = q; + vbuf[data+e[2]] = r; } /** * Computes the parity without the missing data blocks * and store it in the buffers of such data blocks. */ -static void raid_delta_4data(unsigned int x, int y, int z, int v, int i, int j, int k, int l, unsigned char** vbuf, unsigned data, unsigned char* zero, unsigned size) +static void raid_delta_4data(const int* d, const int* e, unsigned char** vbuf, unsigned data, unsigned char* zero, unsigned size) { unsigned char* p; unsigned char* q; @@ -1597,48 +1597,48 @@ static void raid_delta_4data(unsigned int x, int y, int z, int v, int i, int j, unsigned char* sa; /* parity buffer */ - p = vbuf[data+i]; - q = vbuf[data+j]; - r = vbuf[data+k]; - s = vbuf[data+l]; + p = vbuf[data+e[0]]; + q = vbuf[data+e[1]]; + r = vbuf[data+e[2]]; + s = vbuf[data+e[3]]; /* missing data blocks */ - pa = vbuf[x]; - qa = vbuf[y]; - ra = vbuf[z]; - sa = vbuf[v]; + pa = vbuf[d[0]]; + qa = vbuf[d[1]]; + ra = vbuf[d[2]]; + sa = vbuf[d[3]]; /* set at zero the missing data blocks */ - vbuf[x] = zero; - vbuf[y] = zero; - vbuf[z] = zero; - vbuf[v] = zero; + vbuf[d[0]] = zero; + vbuf[d[1]] = zero; + vbuf[d[2]] = zero; + vbuf[d[3]] = zero; /* compute the parity over the missing data blocks */ - vbuf[data+i] = pa; - vbuf[data+j] = qa; - vbuf[data+k] = ra; - vbuf[data+l] = sa; + vbuf[data+e[0]] = pa; + vbuf[data+e[1]] = qa; + vbuf[data+e[2]] = ra; + vbuf[data+e[3]] = sa; /* compute only for the level we really need */ - raid_gen(l + 1, vbuf, data, size); + raid_gen(e[3] + 1, vbuf, data, size); /* restore as before */ - vbuf[x] = pa; - vbuf[y] = qa; - vbuf[z] = ra; - vbuf[v] = sa; - vbuf[data+i] = p; - vbuf[data+j] = q; - vbuf[data+k] = r; - vbuf[data+l] = s; + vbuf[d[0]] = pa; + vbuf[d[1]] = qa; + vbuf[d[2]] = ra; + vbuf[d[3]] = sa; + vbuf[data+e[0]] = p; + vbuf[data+e[1]] = q; + vbuf[data+e[2]] = r; + vbuf[data+e[3]] = s; } /** * Computes the parity without the missing data blocks * and store it in the buffers of such data blocks. */ -static void raid_delta_5data(unsigned int x, int y, int z, int v, int u, int i, int j, int k, int l, int m, unsigned char** vbuf, unsigned data, unsigned char* zero, unsigned size) +static void raid_delta_5data(const int* d, const int* e, unsigned char** vbuf, unsigned data, unsigned char* zero, unsigned size) { unsigned char* p; unsigned char* q; @@ -1652,116 +1652,116 @@ static void raid_delta_5data(unsigned int x, int y, int z, int v, int u, int i, unsigned char* ta; /* parity buffer */ - p = vbuf[data+i]; - q = vbuf[data+j]; - r = vbuf[data+k]; - s = vbuf[data+l]; - t = vbuf[data+m]; + p = vbuf[data+e[0]]; + q = vbuf[data+e[1]]; + r = vbuf[data+e[2]]; + s = vbuf[data+e[3]]; + t = vbuf[data+e[4]]; /* missing data blocks */ - pa = vbuf[x]; - qa = vbuf[y]; - ra = vbuf[z]; - sa = vbuf[v]; - ta = vbuf[u]; + pa = vbuf[d[0]]; + qa = vbuf[d[1]]; + ra = vbuf[d[2]]; + sa = vbuf[d[3]]; + ta = vbuf[d[4]]; /* set at zero the missing data blocks */ - vbuf[x] = zero; - vbuf[y] = zero; - vbuf[z] = zero; - vbuf[v] = zero; - vbuf[u] = zero; + vbuf[d[0]] = zero; + vbuf[d[1]] = zero; + vbuf[d[2]] = zero; + vbuf[d[3]] = zero; + vbuf[d[4]] = zero; /* compute the parity over the missing data blocks */ - vbuf[data+i] = pa; - vbuf[data+j] = qa; - vbuf[data+k] = ra; - vbuf[data+l] = sa; - vbuf[data+m] = ta; + vbuf[data+e[0]] = pa; + vbuf[data+e[1]] = qa; + vbuf[data+e[2]] = ra; + vbuf[data+e[3]] = sa; + vbuf[data+e[4]] = ta; /* compute only for the level we really need */ - raid_gen(m + 1, vbuf, data, size); + raid_gen(e[4] + 1, vbuf, data, size); /* restore as before */ - vbuf[x] = pa; - vbuf[y] = qa; - vbuf[z] = ra; - vbuf[v] = sa; - vbuf[u] = ta; - vbuf[data+i] = p; - vbuf[data+j] = q; - vbuf[data+k] = r; - vbuf[data+l] = s; - vbuf[data+m] = t; + vbuf[d[0]] = pa; + vbuf[d[1]] = qa; + vbuf[d[2]] = ra; + vbuf[d[3]] = sa; + vbuf[d[4]] = ta; + vbuf[data+e[0]] = p; + vbuf[data+e[1]] = q; + vbuf[data+e[2]] = r; + vbuf[data+e[3]] = s; + vbuf[data+e[4]] = t; } /** * Computes the parity without the missing data blocks * and store it in the buffers of such data blocks. */ -static void raid_delta_6data(unsigned int x, int y, int z, int v, int u, int w, int i, int j, int k, int l, int m, int n, unsigned char** vbuf, unsigned data, unsigned char* zero, unsigned size) +static void raid_delta_6data(const int* d, const int* e, unsigned char** vbuf, unsigned data, unsigned char* zero, unsigned size) { unsigned char* p; unsigned char* q; unsigned char* r; unsigned char* s; unsigned char* t; - unsigned char* o; + unsigned char* u; unsigned char* pa; unsigned char* qa; unsigned char* ra; unsigned char* sa; unsigned char* ta; - unsigned char* oa; + unsigned char* ua; /* parity buffer */ - p = vbuf[data+i]; - q = vbuf[data+j]; - r = vbuf[data+k]; - s = vbuf[data+l]; - t = vbuf[data+m]; - o = vbuf[data+n]; + p = vbuf[data+e[0]]; + q = vbuf[data+e[1]]; + r = vbuf[data+e[2]]; + s = vbuf[data+e[3]]; + t = vbuf[data+e[4]]; + u = vbuf[data+e[5]]; /* missing data blocks */ - pa = vbuf[x]; - qa = vbuf[y]; - ra = vbuf[z]; - sa = vbuf[v]; - ta = vbuf[u]; - oa = vbuf[w]; + pa = vbuf[d[0]]; + qa = vbuf[d[1]]; + ra = vbuf[d[2]]; + sa = vbuf[d[3]]; + ta = vbuf[d[4]]; + ua = vbuf[d[5]]; /* set at zero the missing data blocks */ - vbuf[x] = zero; - vbuf[y] = zero; - vbuf[z] = zero; - vbuf[v] = zero; - vbuf[u] = zero; - vbuf[w] = zero; + vbuf[d[0]] = zero; + vbuf[d[1]] = zero; + vbuf[d[2]] = zero; + vbuf[d[3]] = zero; + vbuf[d[4]] = zero; + vbuf[d[5]] = zero; /* compute the parity over the missing data blocks */ - vbuf[data+i] = pa; - vbuf[data+j] = qa; - vbuf[data+k] = ra; - vbuf[data+l] = sa; - vbuf[data+m] = ta; - vbuf[data+n] = oa; + vbuf[data+e[0]] = pa; + vbuf[data+e[1]] = qa; + vbuf[data+e[2]] = ra; + vbuf[data+e[3]] = sa; + vbuf[data+e[4]] = ta; + vbuf[data+e[5]] = ua; /* compute only for the level we really need */ - raid_gen(n + 1, vbuf, data, size); + raid_gen(e[5] + 1, vbuf, data, size); /* restore as before */ - vbuf[x] = pa; - vbuf[y] = qa; - vbuf[z] = ra; - vbuf[v] = sa; - vbuf[u] = ta; - vbuf[w] = oa; - vbuf[data+i] = p; - vbuf[data+j] = q; - vbuf[data+k] = r; - vbuf[data+l] = s; - vbuf[data+m] = t; - vbuf[data+n] = o; + vbuf[d[0]] = pa; + vbuf[d[1]] = qa; + vbuf[d[2]] = ra; + vbuf[d[3]] = sa; + vbuf[d[4]] = ta; + vbuf[d[5]] = ua; + vbuf[data+e[0]] = p; + vbuf[data+e[1]] = q; + vbuf[data+e[2]] = r; + vbuf[data+e[3]] = s; + vbuf[data+e[4]] = t; + vbuf[data+e[5]] = u; } /****************************************************************************/ @@ -1778,7 +1778,7 @@ static void raid_delta_6data(unsigned int x, int y, int z, int v, int u, int w, * * Dx = Pd (this one is easy :D) */ -static void raid5_recov_data(int x, unsigned char** vbuf, unsigned data, unsigned size) +static void raid5_recov_data(const int* d, unsigned char** vbuf, unsigned data, unsigned size) { unsigned char* p; unsigned char* pa; @@ -1786,10 +1786,10 @@ static void raid5_recov_data(int x, unsigned char** vbuf, unsigned data, unsigne /* for RAID5 we can directly compute the mssing block */ /* and we don't need to use the zero buffer */ p = vbuf[data]; - pa = vbuf[x]; + pa = vbuf[d[0]]; /* use the parity as missing data block */ - vbuf[x] = p; + vbuf[d[0]] = p; /* compute the parity over the missing data block */ vbuf[data] = pa; @@ -1798,7 +1798,7 @@ static void raid5_recov_data(int x, unsigned char** vbuf, unsigned data, unsigne raid5_gen(vbuf, data, size); /* restore as before */ - vbuf[x] = pa; + vbuf[d[0]] = pa; vbuf[data] = p; } @@ -1808,26 +1808,26 @@ static void raid5_recov_data(int x, unsigned char** vbuf, unsigned data, unsigne * Starting from the equations: * * Pd = Dx + Dy - * Qd = 2^x * Dx + 2^y * Dy + * Qd = 2^d[0] * Dx + 2^d[1] * Dy * * and solving we get: * - * 1 2^(-x) + * 1 2^(-d[0]) * Dy = ----------- * Pd + ----------- * Qd - * 2^(y-x) + 1 2^(y-x) + 1 + * 2^(d[1]-d[0]) + 1 2^(d[1]-d[0]) + 1 * * Dx = Dy + Pd * * with conditions: * - * 2^x != 0 - * 2^(y-x) + 1 != 0 + * 2^d[0] != 0 + * 2^(d[1]-d[0]) + 1 != 0 * - * That are always satisfied for any 0<=x