1 /* Float ceil function, sparc64 vis3 version.
2 Copyright (C) 2012-2013 Free Software Foundation, Inc.
3 This file is part of the GNU C Library.
4 Contributed by David S. Miller <davem@davemloft.net>, 2012.
6 The GNU C Library is free software; you can redistribute it and/or
7 modify it under the terms of the GNU Lesser General Public
8 License as published by the Free Software Foundation; either
9 version 2.1 of the License, or (at your option) any later version.
11 The GNU C Library is distributed in the hope that it will be useful,
12 but WITHOUT ANY WARRANTY; without even the implied warranty of
13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 Lesser General Public License for more details.
16 You should have received a copy of the GNU Lesser General Public
17 License along with the GNU C Library; if not, see
18 <http://www.gnu.org/licenses/>. */
22 /* Since changing the rounding mode is extremely expensive, we
23 try to round up using a method that is rounding mode
26 We add then subtract (or subtract than add if the initial
27 value was negative) 2**23 to the value, then subtract it
30 This will clear out the fractional portion of the value.
31 One of two things will happen for non-whole initial values.
32 Either the rounding mode will round it up, or it will be
33 rounded down. If the value started out whole, it will be
34 equal after the addition and subtraction. This means we
35 can accurately detect with one test whether we need to add
36 another 1.0 to round it up properly.
38 VIS instructions are used to facilitate the formation of
39 easier constants, and the propagation of the sign bit. */
41 #define TWO_TWENTYTHREE 0x4b000000 /* 2**23 */
42 #define ONE_DOT_ZERO 0x3f800000 /* 1.0 */
44 #define ZERO %f10 /* 0.0 */
45 #define SIGN_BIT %f12 /* -0.0 */
48 sethi %hi(TWO_TWENTYTHREE), %o2
49 sethi %hi(ONE_DOT_ZERO), %o3
57 fcmps %fcc3, %f14, %f16
59 fmovsuge %fcc3, ZERO, %f16
60 fands %f1, SIGN_BIT, SIGN_BIT
62 fors %f16, SIGN_BIT, %f16
68 fmovsuge %fcc2, ZERO, %f9
72 fors %f0, SIGN_BIT, %f0