Test for stack alignment.
[glibc.git] / sysdeps / rs6000 / mul_1.s
blobc0feef4b7248f188c674711f6a24cec55f1fdd6c
1 # IBM POWER __mpn_mul_1 -- Multiply a limb vector with a limb and store
2 # the result in a second limb vector.
4 # Copyright (C) 1992, 1994 Free Software Foundation, Inc.
6 # This file is part of the GNU MP Library.
8 # The GNU MP Library is free software; you can redistribute it and/or modify
9 # it under the terms of the GNU Lesser General Public License as published by
10 # the Free Software Foundation; either version 2.1 of the License, or (at your
11 # option) any later version.
13 # The GNU MP Library is distributed in the hope that it will be useful, but
14 # WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
15 # or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public
16 # License for more details.
18 # You should have received a copy of the GNU Lesser General Public License
19 # along with the GNU MP Library; see the file COPYING.LIB. If not, write to
20 # the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
21 # MA 02111-1307, USA.
24 # INPUT PARAMETERS
25 # res_ptr r3
26 # s1_ptr r4
27 # size r5
28 # s2_limb r6
30 # The RS/6000 has no unsigned 32x32->64 bit multiplication instruction. To
31 # obtain that operation, we have to use the 32x32->64 signed multiplication
32 # instruction, and add the appropriate compensation to the high limb of the
33 # result. We add the multiplicand if the multiplier has its most significant
34 # bit set, and we add the multiplier if the multiplicand has its most
35 # significant bit set. We need to preserve the carry flag between each
36 # iteration, so we have to compute the compensation carefully (the natural,
37 # srai+and doesn't work). Since the POWER architecture has a branch unit
38 # we can branch in zero cycles, so that's how we perform the additions.
40 .toc
41 .csect .__mpn_mul_1[PR]
42 .align 2
43 .globl __mpn_mul_1
44 .globl .__mpn_mul_1
45 .csect __mpn_mul_1[DS]
46 __mpn_mul_1:
47 .long .__mpn_mul_1[PR], TOC[tc0], 0
48 .csect .__mpn_mul_1[PR]
49 .__mpn_mul_1:
51 cal 3,-4(3)
52 l 0,0(4)
53 cmpi 0,6,0
54 mtctr 5
55 mul 9,0,6
56 srai 7,0,31
57 and 7,7,6
58 mfmq 8
59 ai 0,0,0 # reset carry
60 cax 9,9,7
61 blt Lneg
62 Lpos: bdz Lend
63 Lploop: lu 0,4(4)
64 stu 8,4(3)
65 cmpi 0,0,0
66 mul 10,0,6
67 mfmq 0
68 ae 8,0,9
69 bge Lp0
70 cax 10,10,6 # adjust high limb for negative limb from s1
71 Lp0: bdz Lend0
72 lu 0,4(4)
73 stu 8,4(3)
74 cmpi 0,0,0
75 mul 9,0,6
76 mfmq 0
77 ae 8,0,10
78 bge Lp1
79 cax 9,9,6 # adjust high limb for negative limb from s1
80 Lp1: bdn Lploop
81 b Lend
83 Lneg: cax 9,9,0
84 bdz Lend
85 Lnloop: lu 0,4(4)
86 stu 8,4(3)
87 cmpi 0,0,0
88 mul 10,0,6
89 cax 10,10,0 # adjust high limb for negative s2_limb
90 mfmq 0
91 ae 8,0,9
92 bge Ln0
93 cax 10,10,6 # adjust high limb for negative limb from s1
94 Ln0: bdz Lend0
95 lu 0,4(4)
96 stu 8,4(3)
97 cmpi 0,0,0
98 mul 9,0,6
99 cax 9,9,0 # adjust high limb for negative s2_limb
100 mfmq 0
101 ae 8,0,10
102 bge Ln1
103 cax 9,9,6 # adjust high limb for negative limb from s1
104 Ln1: bdn Lnloop
105 b Lend
107 Lend0: cal 9,0(10)
108 Lend: st 8,4(3)
109 aze 3,9