projects / ceph.git / blame
| Commit | Line | Data |
|---|---|---|
| f91f0fd5 TL |
1 | #include "ec_base_vsx.h" |
| 2 | ||
| 3 | void gf_6vect_mad_vsx(int len, int vec, int vec_i, unsigned char *gftbls, | |
| 4 | unsigned char *src, unsigned char **dest) | |
| 5 | { | |
| 6 | unsigned char *s, *t0, *t1, *t2, *t3, *t4, *t5; | |
| 7 | vector unsigned char vX1, vX2, vX3, vX4; | |
| 8 | vector unsigned char vY1, vY2, vY3, vY4, vY5, vY6, vY7, vY8, vY9, vYA, vYB, vYC; | |
| 9 | vector unsigned char vYD, vYE, vYF, vYG, vYH, vYI, vYJ, vYK, vYL, vYM, vYN, vYO; | |
| 10 | vector unsigned char vhi0, vlo0, vhi1, vlo1, vhi2, vlo2; | |
| 11 | vector unsigned char vhi3, vlo3, vhi4, vlo4, vhi5, vlo5; | |
| 12 | int i, head; | |
| 13 | ||
| 14 | s = (unsigned char *)src; | |
| 15 | t0 = (unsigned char *)dest[0]; | |
| 16 | t1 = (unsigned char *)dest[1]; | |
| 17 | t2 = (unsigned char *)dest[2]; | |
| 18 | t3 = (unsigned char *)dest[3]; | |
| 19 | t4 = (unsigned char *)dest[4]; | |
| 20 | t5 = (unsigned char *)dest[5]; | |
| 21 | ||
| 22 | head = len % 64; | |
| 23 | if (head != 0) { | |
| 24 | gf_vect_mad_base(head, vec, vec_i, &gftbls[0 * 32 * vec], src, t0); | |
| 25 | gf_vect_mad_base(head, vec, vec_i, &gftbls[1 * 32 * vec], src, t1); | |
| 26 | gf_vect_mad_base(head, vec, vec_i, &gftbls[2 * 32 * vec], src, t2); | |
| 27 | gf_vect_mad_base(head, vec, vec_i, &gftbls[3 * 32 * vec], src, t3); | |
| 28 | gf_vect_mad_base(head, vec, vec_i, &gftbls[4 * 32 * vec], src, t4); | |
| 29 | gf_vect_mad_base(head, vec, vec_i, &gftbls[5 * 32 * vec], src, t5); | |
| 30 | } | |
| 31 | ||
| 32 | vlo0 = EC_vec_xl(0, gftbls + (((0 * vec) << 5) + (vec_i << 5))); | |
| 33 | vhi0 = EC_vec_xl(16, gftbls + (((0 * vec) << 5) + (vec_i << 5))); | |
| 34 | vlo1 = EC_vec_xl(0, gftbls + (((1 * vec) << 5) + (vec_i << 5))); | |
| 35 | vhi1 = EC_vec_xl(16, gftbls + (((1 * vec) << 5) + (vec_i << 5))); | |
| 36 | vlo2 = EC_vec_xl(0, gftbls + (((2 * vec) << 5) + (vec_i << 5))); | |
| 37 | vhi2 = EC_vec_xl(16, gftbls + (((2 * vec) << 5) + (vec_i << 5))); | |
| 38 | vlo3 = EC_vec_xl(0, gftbls + (((3 * vec) << 5) + (vec_i << 5))); | |
| 39 | vhi3 = EC_vec_xl(16, gftbls + (((3 * vec) << 5) + (vec_i << 5))); | |
| 40 | vlo4 = EC_vec_xl(0, gftbls + (((4 * vec) << 5) + (vec_i << 5))); | |
| 41 | vhi4 = EC_vec_xl(16, gftbls + (((4 * vec) << 5) + (vec_i << 5))); | |
| 42 | vlo5 = EC_vec_xl(0, gftbls + (((5 * vec) << 5) + (vec_i << 5))); | |
| 43 | vhi5 = EC_vec_xl(16, gftbls + (((5 * vec) << 5) + (vec_i << 5))); | |
| 44 | ||
| 45 | for (i = head; i < len - 63; i += 64) { | |
| 46 | vX1 = vec_xl(0, s + i); | |
| 47 | vX2 = vec_xl(16, s + i); | |
| 48 | vX3 = vec_xl(32, s + i); | |
| 49 | vX4 = vec_xl(48, s + i); | |
| 50 | ||
| 51 | vY1 = vec_xl(0, t0 + i); | |
| 52 | vY2 = vec_xl(16, t0 + i); | |
| 53 | vYD = vec_xl(32, t0 + i); | |
| 54 | vYE = vec_xl(48, t0 + i); | |
| 55 | ||
| 56 | vY1 = vY1 ^ EC_vec_permxor(vhi0, vlo0, vX1); | |
| 57 | vY2 = vY2 ^ EC_vec_permxor(vhi0, vlo0, vX2); | |
| 58 | vYD = vYD ^ EC_vec_permxor(vhi0, vlo0, vX3); | |
| 59 | vYE = vYE ^ EC_vec_permxor(vhi0, vlo0, vX4); | |
| 60 | ||
| 61 | vec_xst(vY1, 0, t0 + i); | |
| 62 | vec_xst(vY2, 16, t0 + i); | |
| 63 | vec_xst(vYD, 32, t0 + i); | |
| 64 | vec_xst(vYE, 48, t0 + i); | |
| 65 | ||
| 66 | vY3 = vec_xl(0, t1 + i); | |
| 67 | vY4 = vec_xl(16, t1 + i); | |
| 68 | vYF = vec_xl(32, t1 + i); | |
| 69 | vYG = vec_xl(48, t1 + i); | |
| 70 | ||
| 71 | vY3 = vY3 ^ EC_vec_permxor(vhi1, vlo1, vX1); | |
| 72 | vY4 = vY4 ^ EC_vec_permxor(vhi1, vlo1, vX2); | |
| 73 | vYF = vYF ^ EC_vec_permxor(vhi1, vlo1, vX3); | |
| 74 | vYG = vYG ^ EC_vec_permxor(vhi1, vlo1, vX4); | |
| 75 | ||
| 76 | vec_xst(vY3, 0, t1 + i); | |
| 77 | vec_xst(vY4, 16, t1 + i); | |
| 78 | vec_xst(vYF, 32, t1 + i); | |
| 79 | vec_xst(vYG, 48, t1 + i); | |
| 80 | ||
| 81 | vY5 = vec_xl(0, t2 + i); | |
| 82 | vY6 = vec_xl(16, t2 + i); | |
| 83 | vYH = vec_xl(32, t2 + i); | |
| 84 | vYI = vec_xl(48, t2 + i); | |
| 85 | ||
| 86 | vY5 = vY5 ^ EC_vec_permxor(vhi2, vlo2, vX1); | |
| 87 | vY6 = vY6 ^ EC_vec_permxor(vhi2, vlo2, vX2); | |
| 88 | vYH = vYH ^ EC_vec_permxor(vhi2, vlo2, vX3); | |
| 89 | vYI = vYI ^ EC_vec_permxor(vhi2, vlo2, vX4); | |
| 90 | ||
| 91 | vY7 = vec_xl(0, t3 + i); | |
| 92 | vY8 = vec_xl(16, t3 + i); | |
| 93 | vYJ = vec_xl(32, t3 + i); | |
| 94 | vYK = vec_xl(48, t3 + i); | |
| 95 | ||
| 96 | vec_xst(vY5, 0, t2 + i); | |
| 97 | vec_xst(vY6, 16, t2 + i); | |
| 98 | vec_xst(vYH, 32, t2 + i); | |
| 99 | vec_xst(vYI, 48, t2 + i); | |
| 100 | ||
| 101 | vY7 = vY7 ^ EC_vec_permxor(vhi3, vlo3, vX1); | |
| 102 | vY8 = vY8 ^ EC_vec_permxor(vhi3, vlo3, vX2); | |
| 103 | vYJ = vYJ ^ EC_vec_permxor(vhi3, vlo3, vX3); | |
| 104 | vYK = vYK ^ EC_vec_permxor(vhi3, vlo3, vX4); | |
| 105 | ||
| 106 | vY9 = vec_xl(0, t4 + i); | |
| 107 | vYA = vec_xl(16, t4 + i); | |
| 108 | vYL = vec_xl(32, t4 + i); | |
| 109 | vYM = vec_xl(48, t4 + i); | |
| 110 | ||
| 111 | vec_xst(vY7, 0, t3 + i); | |
| 112 | vec_xst(vY8, 16, t3 + i); | |
| 113 | vec_xst(vYJ, 32, t3 + i); | |
| 114 | vec_xst(vYK, 48, t3 + i); | |
| 115 | ||
| 116 | vY9 = vY9 ^ EC_vec_permxor(vhi4, vlo4, vX1); | |
| 117 | vYA = vYA ^ EC_vec_permxor(vhi4, vlo4, vX2); | |
| 118 | vYL = vYL ^ EC_vec_permxor(vhi4, vlo4, vX3); | |
| 119 | vYM = vYM ^ EC_vec_permxor(vhi4, vlo4, vX4); | |
| 120 | ||
| 121 | vYB = vec_xl(0, t5 + i); | |
| 122 | vYC = vec_xl(16, t5 + i); | |
| 123 | vYN = vec_xl(32, t5 + i); | |
| 124 | vYO = vec_xl(48, t5 + i); | |
| 125 | ||
| 126 | vec_xst(vY9, 0, t4 + i); | |
| 127 | vec_xst(vYA, 16, t4 + i); | |
| 128 | vec_xst(vYL, 32, t4 + i); | |
| 129 | vec_xst(vYM, 48, t4 + i); | |
| 130 | ||
| 131 | vYB = vYB ^ EC_vec_permxor(vhi5, vlo5, vX1); | |
| 132 | vYC = vYC ^ EC_vec_permxor(vhi5, vlo5, vX2); | |
| 133 | vYN = vYN ^ EC_vec_permxor(vhi5, vlo5, vX3); | |
| 134 | vYO = vYO ^ EC_vec_permxor(vhi5, vlo5, vX4); | |
| 135 | ||
| 136 | vec_xst(vYB, 0, t5 + i); | |
| 137 | vec_xst(vYC, 16, t5 + i); | |
| 138 | vec_xst(vYN, 32, t5 + i); | |
| 139 | vec_xst(vYO, 48, t5 + i); | |
| 140 | } | |
| 141 | return; | |
| 142 | } |