addmul_1.asm revision 1.1.1.1 
1dnl  ARM mpn_addmul_1.
2
3dnl  Copyright 2012 Free Software Foundation, Inc.
4
5dnl  This file is part of the GNU MP Library.
6
7dnl  The GNU MP Library is free software; you can redistribute it and/or modify
8dnl  it under the terms of the GNU Lesser General Public License as published
9dnl  by the Free Software Foundation; either version 3 of the License, or (at
10dnl  your option) any later version.
11
12dnl  The GNU MP Library is distributed in the hope that it will be useful, but
13dnl  WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
14dnl  or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU Lesser General Public
15dnl  License for more details.
16
17dnl  You should have received a copy of the GNU Lesser General Public License
18dnl  along with the GNU MP Library.  If not, see http://www.gnu.org/licenses/.
19
20include(`../config.m4')
21
22C	     cycles/limb
23C StrongARM:	 -
24C XScale	 -
25C Cortex-A8	 ?
26C Cortex-A9	 3.25
27C Cortex-A15	 4
28
29C TODO
30C  * Micro-optimise feed-in code.
31C  * Optimise for n=1,2 by delaying register saving.
32C  * Try using ldm/stm.
33
34define(`rp',`r0')
35define(`up',`r1')
36define(`n', `r2')
37define(`v0',`r3')
38
39ASM_START()
40PROLOGUE(mpn_addmul_1)
41	stmfd	sp!, { r4, r5, r6, r7 }
42
43	ands	r6, n, #3
44	mov	r12, #0
45	beq	L(fi0)
46	cmp	r6, #2
47	bcc	L(fi1)
48	beq	L(fi2)
49
50L(fi3):	ldr	r4, [up], #4
51	ldr	r6, [rp, #0]
52	ldr	r5, [up], #4
53	b	L(lo3)
54
55L(fi0):	ldr	r5, [up], #4
56	ldr	r7, [rp], #4
57	ldr	r4, [up], #4
58	b	L(lo0)
59
60L(fi1):	ldr	r4, [up], #4
61	ldr	r6, [rp], #8
62	subs	n, n, #1
63	beq	L(1)
64	ldr	r5, [up], #4
65	b	L(lo1)
66
67L(fi2):	ldr	r5, [up], #4
68	ldr	r7, [rp], #12
69	ldr	r4, [up], #4
70	b	L(lo2)
71
72	ALIGN(16)
73L(top):	ldr	r6, [rp, #-8]
74	ldr	r5, [up], #4
75	str	r7, [rp, #-12]
76L(lo1):	umaal	r6, r12, r4, v0
77	ldr	r7, [rp, #-4]
78	ldr	r4, [up], #4
79	str	r6, [rp, #-8]
80L(lo0):	umaal	r7, r12, r5, v0
81	ldr	r6, [rp, #0]
82	ldr	r5, [up], #4
83	str	r7, [rp, #-4]
84L(lo3):	umaal	r6, r12, r4, v0
85	ldr	r7, [rp, #4]
86	ldr	r4, [up], #4
87	str	r6, [rp], #16
88L(lo2):	umaal	r7, r12, r5, v0
89	subs	n, n, #4
90	bhi	L(top)
91
92	ldr	r6, [rp, #-8]
93	str	r7, [rp, #-12]
94L(1):	umaal	r6, r12, r4, v0
95	str	r6, [rp, #-8]
96	mov	r0, r12
97	ldmfd	sp!, { r4, r5, r6, r7 }
98	bx	lr
99EPILOGUE()
100