xref: /netbsd-src/external/lgpl3/gmp/dist/mpn/power/add_n.asm (revision ce54336801cf28877c3414aa2fcb251dddd543a2)
1dnl  IBM POWER mpn_add_n -- Add two limb vectors of equal, non-zero length.
2
3dnl  Copyright 1992, 1994-1996, 1999-2001, 2005 Free Software Foundation, Inc.
4
5dnl  This file is part of the GNU MP Library.
6dnl
7dnl  The GNU MP Library is free software; you can redistribute it and/or modify
8dnl  it under the terms of either:
9dnl
10dnl    * the GNU Lesser General Public License as published by the Free
11dnl      Software Foundation; either version 3 of the License, or (at your
12dnl      option) any later version.
13dnl
14dnl  or
15dnl
16dnl    * the GNU General Public License as published by the Free Software
17dnl      Foundation; either version 2 of the License, or (at your option) any
18dnl      later version.
19dnl
20dnl  or both in parallel, as here.
21dnl
22dnl  The GNU MP Library is distributed in the hope that it will be useful, but
23dnl  WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
24dnl  or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
25dnl  for more details.
26dnl
27dnl  You should have received copies of the GNU General Public License and the
28dnl  GNU Lesser General Public License along with the GNU MP Library.  If not,
29dnl  see https://www.gnu.org/licenses/.
30
31
32dnl  INPUT PARAMETERS
33dnl  res_ptr	r3
34dnl  s1_ptr	r4
35dnl  s2_ptr	r5
36dnl  size	r6
37
38include(`../config.m4')
39
40ASM_START()
41PROLOGUE(mpn_add_n)
42	andil.	10,6,1		C odd or even number of limbs?
43	l	8,0(4)		C load least significant s1 limb
44	l	0,0(5)		C load least significant s2 limb
45	cal	3,-4(3)		C offset res_ptr, it's updated before it's used
46	sri	10,6,1		C count for unrolled loop
47	a	7,0,8		C add least significant limbs, set cy
48	mtctr	10		C copy count into CTR
49	beq	0,Leven		C branch if even # of limbs (# of limbs >= 2)
50
51C We have an odd # of limbs.  Add the first limbs separately.
52	cmpi	1,10,0		C is count for unrolled loop zero?
53	bc	4,6,L1		C bne cr1,L1 (misassembled by gas)
54	st	7,4(3)
55	aze	3,10		C use the fact that r10 is zero...
56	br			C return
57
58C We added least significant limbs.  Now reload the next limbs to enter loop.
59L1:	lu	8,4(4)		C load s1 limb and update s1_ptr
60	lu	0,4(5)		C load s2 limb and update s2_ptr
61	stu	7,4(3)
62	ae	7,0,8		C add limbs, set cy
63Leven:	lu	9,4(4)		C load s1 limb and update s1_ptr
64	lu	10,4(5)		C load s2 limb and update s2_ptr
65	bdz	Lend		C If done, skip loop
66
67Loop:	lu	8,4(4)		C load s1 limb and update s1_ptr
68	lu	0,4(5)		C load s2 limb and update s2_ptr
69	ae	11,10,9		C add previous limbs with cy, set cy
70	stu	7,4(3)		C
71	lu	9,4(4)		C load s1 limb and update s1_ptr
72	lu	10,4(5)		C load s2 limb and update s2_ptr
73	ae	7,0,8		C add previous limbs with cy, set cy
74	stu	11,4(3)		C
75	bdn	Loop		C decrement CTR and loop back
76
77Lend:	ae	11,10,9		C add limbs with cy, set cy
78	st	7,4(3)		C
79	st	11,8(3)		C
80	lil	3,0		C load cy into ...
81	aze	3,3		C ... return value register
82	br
83EPILOGUE(mpn_add_n)
84