projects / thirdparty / glibc.git / blame
| Commit | Line | Data |
|---|---|---|
| f964490f | 1 | /* s_ceill.S IBM extended format long double version. |
| 568035b7 | 2 | Copyright (C) 2004-2013 Free Software Foundation, Inc. |
| f964490f RM |
3 | This file is part of the GNU C Library. |
| 4 | ||
| 5 | The GNU C Library is free software; you can redistribute it and/or | |
| 6 | modify it under the terms of the GNU Lesser General Public | |
| 7 | License as published by the Free Software Foundation; either | |
| 8 | version 2.1 of the License, or (at your option) any later version. | |
| 9 | ||
| 10 | The GNU C Library is distributed in the hope that it will be useful, | |
| 11 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
| 12 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | |
| 13 | Lesser General Public License for more details. | |
| 14 | ||
| 15 | You should have received a copy of the GNU Lesser General Public | |
| 59ba27a6 PE |
16 | License along with the GNU C Library; if not, see |
| 17 | <http://www.gnu.org/licenses/>. */ | |
| f964490f RM |
18 | |
| 19 | #include <sysdep.h> | |
| 20 | #include <math_ldbl_opt.h> | |
| 21 | ||
| 22 | .section ".toc","aw" | |
| 23 | .LC0: /* 2**52 */ | |
| 24 | .tc FD_43300000_0[TC],0x4330000000000000 | |
| 25 | ||
| 26 | .section ".text" | |
| 27 | ||
| 28 | /* long double [fp1,fp2] ceill (long double x [fp1,fp2]) | |
| 29 | IEEE 1003.1 ceil function. | |
| 30 | ||
| 31 | PowerPC64 long double uses the IBM extended format which is | |
| 32 | represented two 64-floating point double values. The values are | |
| 33 | non-overlapping giving an effective precision of 106 bits. The first | |
| 34 | double contains the high order bits of mantisa and is always ceiled | |
| 35 | to represent a normal ceiling of long double to double. Since the | |
| 36 | long double value is sum of the high and low values, the low double | |
| 37 | normally has the opposite sign to compensate for the this ceiling. | |
| 38 | ||
| 39 | For long double there are two cases: | |
| 40 | 1) |x| < 2**52, all the integer bits are in the high double. | |
| 41 | ceil the high double and set the low double to -0.0. | |
| 42 | 2) |x| >= 2**52, ceiling involves both doubles. | |
| 43 | See the comment before lable .L2 for details. | |
| 44 | */ | |
| 45 | ||
| 46 | ENTRY (__ceill) | |
| 47 | mffs fp11 /* Save current FPU rounding mode. */ | |
| 48 | lfd fp13,.LC0@toc(2) | |
| 49 | fabs fp0,fp1 | |
| 50 | fabs fp9,fp2 | |
| 51 | fsub fp12,fp13,fp13 /* generate 0.0 */ | |
| 52 | fcmpu cr7,fp0,fp13 /* if (fabs(x) > TWO52) */ | |
| 1e832e37 | 53 | fcmpu cr6,fp1,fp12 /* if (x > 0.0) */ |
| f964490f RM |
54 | bnl- cr7,.L2 |
| 55 | mtfsfi 7,2 /* Set rounding mode toward +inf. */ | |
| 56 | fneg fp2,fp12 | |
| 57 | ble- cr6,.L1 | |
| 58 | fadd fp1,fp1,fp13 /* x+= TWO52; */ | |
| 59 | fsub fp1,fp1,fp13 /* x-= TWO52; */ | |
| 60 | fabs fp1,fp1 /* if (x == 0.0) */ | |
| 61 | .L0: | |
| 62 | mtfsf 0x01,fp11 /* restore previous rounding mode. */ | |
| 63 | blr /* x = 0.0; */ | |
| 64 | .L1: | |
| 65 | bge- cr6,.L0 /* if (x < 0.0) */ | |
| 66 | fsub fp1,fp1,fp13 /* x-= TWO52; */ | |
| 67 | fadd fp1,fp1,fp13 /* x+= TWO52; */ | |
| 68 | fcmpu cr5,fp1,fp12 /* if (x > 0.0) */ | |
| 69 | mtfsf 0x01,fp11 /* restore previous rounding mode. */ | |
| 70 | fnabs fp1,fp1 /* if (x == 0.0) */ | |
| 71 | blr /* x = -0.0; */ | |
| 72 | ||
| 73 | /* The high double is > TWO52 so we need to round the low double and | |
| 74 | perhaps the high double. In this case we have to round the low | |
| 75 | double and handle any adjustment to the high double that may be | |
| 76 | caused by rounding (up). This is complicated by the fact that the | |
| 77 | high double may already be rounded and the low double may have the | |
| 78 | opposite sign to compensate.This gets a bit tricky so we use the | |
| 79 | following algorithm: | |
| 80 | ||
| 81 | tau = floor(x_high/TWO52); | |
| 82 | x0 = x_high - tau; | |
| 83 | x1 = x_low + tau; | |
| 84 | r1 = rint(x1); | |
| 85 | y_high = x0 + r1; | |
| 86 | y_low = x0 - y_high + r1; | |
| 87 | return y; */ | |
| 88 | .L2: | |
| 89 | fcmpu cr7,fp9,fp13 /* if (|x_low| > TWO52) */ | |
| 90 | fcmpu cr0,fp9,fp12 /* || (|x_low| == 0.0) */ | |
| 91 | fcmpu cr5,fp2,fp12 /* if (x_low > 0.0) */ | |
| 92 | bgelr- cr7 /* return x; */ | |
| 93 | beqlr- cr0 | |
| 94 | mtfsfi 7,2 /* Set rounding mode toward +inf. */ | |
| 95 | fdiv fp8,fp1,fp13 /* x_high/TWO52 */ | |
| 96 | ||
| 97 | bng- cr6,.L6 /* if (x > 0.0) */ | |
| 98 | fctidz fp0,fp8 | |
| 99 | fcfid fp8,fp0 /* tau = floor(x_high/TWO52); */ | |
| 100 | bng cr5,.L4 /* if (x_low > 0.0) */ | |
| 101 | fmr fp3,fp1 | |
| 102 | fmr fp4,fp2 | |
| 103 | b .L5 | |
| 104 | .L4: /* if (x_low < 0.0) */ | |
| 105 | fsub fp3,fp1,fp8 /* x0 = x_high - tau; */ | |
| 106 | fadd fp4,fp2,fp8 /* x1 = x_low + tau; */ | |
| 107 | .L5: | |
| 108 | fadd fp5,fp4,fp13 /* r1 = r1 + TWO52; */ | |
| 109 | fsub fp5,fp5,fp13 /* r1 = r1 - TWO52; */ | |
| 110 | b .L9 | |
| 111 | .L6: /* if (x < 0.0) */ | |
| 112 | fctidz fp0,fp8 | |
| 113 | fcfid fp8,fp0 /* tau = floor(x_high/TWO52); */ | |
| 114 | bnl cr5,.L7 /* if (x_low < 0.0) */ | |
| 115 | fmr fp3,fp1 | |
| 116 | fmr fp4,fp2 | |
| 117 | b .L8 | |
| 118 | .L7: /* if (x_low > 0.0) */ | |
| 119 | fsub fp3,fp1,fp8 /* x0 = x_high - tau; */ | |
| 120 | fadd fp4,fp2,fp8 /* x1 = x_low + tau; */ | |
| 121 | .L8: | |
| 122 | fsub fp5,fp4,fp13 /* r1-= TWO52; */ | |
| 123 | fadd fp5,fp5,fp13 /* r1+= TWO52; */ | |
| 124 | .L9: | |
| 125 | mtfsf 0x01,fp11 /* restore previous rounding mode. */ | |
| 126 | fadd fp1,fp3,fp5 /* y_high = x0 + r1; */ | |
| 127 | fsub fp2,fp3,fp1 /* y_low = x0 - y_high + r1; */ | |
| 128 | fadd fp2,fp2,fp5 | |
| 129 | blr | |
| 130 | END (__ceill) | |
| 131 | ||
| 132 | long_double_symbol (libm, __ceill, ceill) |