1dnl  S/390-64 mpn_add_n and mpn_sub_n.
2
3dnl  Copyright 2011 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
31include(`../config.m4')
32
33C            cycles/limb
34C z900		 5.5
35C z990		 3
36C z9		 ?
37C z10		 6
38C z196		 ?
39
40C TODO
41C  * Optimise for small n
42C  * Use r0 and save/restore one less register
43C  * Using logops_n's v1 inner loop operand order make the loop about 20%
44C    faster, at the expense of highly alignment-dependent performance.
45
46C INPUT PARAMETERS
47define(`rp',	`%r2')
48define(`up',	`%r3')
49define(`vp',	`%r4')
50define(`n',	`%r5')
51
52ifdef(`OPERATION_add_n', `
53  define(ADSB,		alg)
54  define(ADSBCR,	alcgr)
55  define(ADSBC,		alcg)
56  define(RETVAL,`dnl
57	lghi	%r2, 0
58	alcgr	%r2, %r2')
59  define(func,		mpn_add_n)
60  define(func_nc,	mpn_add_nc)')
61ifdef(`OPERATION_sub_n', `
62  define(ADSB,		slg)
63  define(ADSBCR,	slbgr)
64  define(ADSBC,		slbg)
65  define(RETVAL,`dnl
66	slbgr	%r2, %r2
67	lcgr	%r2, %r2')
68  define(func,		mpn_sub_n)
69  define(func_nc,	mpn_sub_nc)')
70
71MULFUNC_PROLOGUE(mpn_add_n mpn_sub_n)
72
73ASM_START()
74PROLOGUE(func)
75	stmg	%r6, %r8, 48(%r15)
76
77	aghi	n, 3
78	lghi	%r7, 3
79	srlg	%r1, n, 2
80	ngr	%r7, n			C n mod 4
81	je	L(b1)
82	cghi	%r7, 2
83	jl	L(b2)
84	jne	L(b0)
85
86L(b3):	lmg	%r5, %r7, 0(up)
87	la	up, 24(up)
88	ADSB	%r5, 0(vp)
89	ADSBC	%r6, 8(vp)
90	ADSBC	%r7, 16(vp)
91	la	vp, 24(vp)
92	stmg	%r5, %r7, 0(rp)
93	la	rp, 24(rp)
94	brctg	%r1, L(top)
95	j	L(end)
96
97L(b0):	lmg	%r5, %r8, 0(up)		C This redundant insns is no mistake,
98	la	up, 32(up)		C it is needed to make main loop run
99	ADSB	%r5, 0(vp)		C fast for n = 0 (mod 4).
100	ADSBC	%r6, 8(vp)
101	j	L(m0)
102
103L(b1):	lg	%r5, 0(up)
104	la	up, 8(up)
105	ADSB	%r5, 0(vp)
106	la	vp, 8(vp)
107	stg	%r5, 0(rp)
108	la	rp, 8(rp)
109	brctg	%r1, L(top)
110	j	L(end)
111
112L(b2):	lmg	%r5, %r6, 0(up)
113	la	up, 16(up)
114	ADSB	%r5, 0(vp)
115	ADSBC	%r6, 8(vp)
116	la	vp, 16(vp)
117	stmg	%r5, %r6, 0(rp)
118	la	rp, 16(rp)
119	brctg	%r1, L(top)
120	j	L(end)
121
122L(top):	lmg	%r5, %r8, 0(up)
123	la	up, 32(up)
124	ADSBC	%r5, 0(vp)
125	ADSBC	%r6, 8(vp)
126L(m0):	ADSBC	%r7, 16(vp)
127	ADSBC	%r8, 24(vp)
128	la	vp, 32(vp)
129	stmg	%r5, %r8, 0(rp)
130	la	rp, 32(rp)
131	brctg	%r1, L(top)
132
133L(end):	RETVAL
134	lmg	%r6, %r8, 48(%r15)
135	br	%r14
136EPILOGUE()
137