| Aurelien Jacquiot | 09831ca | 2011-10-04 11:15:51 -0400 | [diff] [blame] | 1 | ;; Copyright 2011 Free Software Foundation, Inc. |
| 2 | ;; Contributed by Bernd Schmidt <bernds@codesourcery.com>. |
| 3 | ;; |
| 4 | ;; This program is free software; you can redistribute it and/or modify |
| 5 | ;; it under the terms of the GNU General Public License as published by |
| 6 | ;; the Free Software Foundation; either version 2 of the License, or |
| 7 | ;; (at your option) any later version. |
| 8 | ;; |
| 9 | ;; This program is distributed in the hope that it will be useful, |
| 10 | ;; but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 11 | ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 12 | ;; GNU General Public License for more details. |
| 13 | ;; |
| 14 | ;; You should have received a copy of the GNU General Public License |
| 15 | ;; along with this program; if not, write to the Free Software |
| 16 | ;; Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. |
| 17 | |
| 18 | #include <linux/linkage.h> |
| 19 | |
| 20 | .text |
| 21 | ENTRY(__c6xabi_divremu) |
| 22 | ;; We use a series of up to 31 subc instructions. First, we find |
| 23 | ;; out how many leading zero bits there are in the divisor. This |
| 24 | ;; gives us both a shift count for aligning (shifting) the divisor |
| 25 | ;; to the, and the number of times we have to execute subc. |
| 26 | |
| 27 | ;; At the end, we have both the remainder and most of the quotient |
| 28 | ;; in A4. The top bit of the quotient is computed first and is |
| 29 | ;; placed in A2. |
| 30 | |
| 31 | ;; Return immediately if the dividend is zero. Setting B4 to 1 |
| 32 | ;; is a trick to allow us to leave the following insns in the jump |
| 33 | ;; delay slot without affecting the result. |
| 34 | mv .s2x A4, B1 |
| 35 | |
| 36 | [b1] lmbd .l2 1, B4, B1 |
| 37 | ||[!b1] b .s2 B3 ; RETURN A |
| 38 | ||[!b1] mvk .d2 1, B4 |
| 39 | |
| 40 | ||[!b1] zero .s1 A5 |
| 41 | mv .l1x B1, A6 |
| 42 | || shl .s2 B4, B1, B4 |
| 43 | |
| 44 | ;; The loop performs a maximum of 28 steps, so we do the |
| 45 | ;; first 3 here. |
| 46 | cmpltu .l1x A4, B4, A2 |
| 47 | [!A2] sub .l1x A4, B4, A4 |
| 48 | || shru .s2 B4, 1, B4 |
| 49 | || xor .s1 1, A2, A2 |
| 50 | |
| 51 | shl .s1 A2, 31, A2 |
| 52 | || [b1] subc .l1x A4,B4,A4 |
| 53 | || [b1] add .s2 -1, B1, B1 |
| 54 | [b1] subc .l1x A4,B4,A4 |
| 55 | || [b1] add .s2 -1, B1, B1 |
| 56 | |
| 57 | ;; RETURN A may happen here (note: must happen before the next branch) |
| 58 | __divremu0: |
| 59 | cmpgt .l2 B1, 7, B0 |
| 60 | || [b1] subc .l1x A4,B4,A4 |
| 61 | || [b1] add .s2 -1, B1, B1 |
| 62 | [b1] subc .l1x A4,B4,A4 |
| 63 | || [b1] add .s2 -1, B1, B1 |
| 64 | || [b0] b .s1 __divremu0 |
| 65 | [b1] subc .l1x A4,B4,A4 |
| 66 | || [b1] add .s2 -1, B1, B1 |
| 67 | [b1] subc .l1x A4,B4,A4 |
| 68 | || [b1] add .s2 -1, B1, B1 |
| 69 | [b1] subc .l1x A4,B4,A4 |
| 70 | || [b1] add .s2 -1, B1, B1 |
| 71 | [b1] subc .l1x A4,B4,A4 |
| 72 | || [b1] add .s2 -1, B1, B1 |
| 73 | [b1] subc .l1x A4,B4,A4 |
| 74 | || [b1] add .s2 -1, B1, B1 |
| 75 | ;; loop backwards branch happens here |
| 76 | |
| 77 | ret .s2 B3 |
| 78 | || mvk .s1 32, A1 |
| 79 | sub .l1 A1, A6, A6 |
| 80 | || extu .s1 A4, A6, A5 |
| 81 | shl .s1 A4, A6, A4 |
| 82 | shru .s1 A4, 1, A4 |
| 83 | || sub .l1 A6, 1, A6 |
| 84 | or .l1 A2, A4, A4 |
| 85 | shru .s1 A4, A6, A4 |
| 86 | nop |
| 87 | ENDPROC(__c6xabi_divremu) |