1/* alpha    rshift
2 *      Copyright (C) 1994, 1995, 1998, 1999,
3 *                    2000, 2001, 2002 Free Software Foundation, Inc.
4 *
5 * This file is part of Libgcrypt.
6 *
7 * Libgcrypt is free software; you can redistribute it and/or modify
8 * it under the terms of the GNU Lesser General Public License as
9 * published by the Free Software Foundation; either version 2.1 of
10 * the License, or (at your option) any later version.
11 *
12 * Libgcrypt is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
15 * GNU Lesser General Public License for more details.
16 *
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with this program; if not, write to the Free Software
19 * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA
20 *
21 * Note: This code is heavily based on the GNU MP Library.
22 *	 Actually it's the same code with only minor changes in the
23 *	 way the data is stored; this is to support the abstraction
24 *	 of an optional secure memory allocation which may be used
25 *	 to avoid revealing of sensitive data due to paging etc.
26 */
27
28
29
30
31/*******************
32 * mpi_limb_t
33 * _gcry_mpih_rshift( mpi_ptr_t wp,	(r16)
34 *		   mpi_ptr_t up,	(r17)
35 *		   mpi_size_t usize,	(r18)
36 *		   unsigned cnt)	(r19)
37 *
38 * This code runs at 4.8 cycles/limb on the 21064.  With infinite unrolling,
39 * it would take 4 cycles/limb.  It should be possible to get down to 3
40 * cycles/limb since both ldq and stq can be paired with the other used
41 * instructions.  But there are many restrictions in the 21064 pipeline that
42 * makes it hard, if not impossible, to get down to 3 cycles/limb:
43 *
44 * 1. ldq has a 3 cycle delay, srl and sll have a 2 cycle delay.
45 * 2. Only aligned instruction pairs can be paired.
46 * 3. The store buffer or silo might not be able to deal with the bandwidth.
47 */
48
49	.set	noreorder
50	.set	noat
51.text
52	.align	3
53	.globl	_gcry_mpih_rshift
54	.ent	_gcry_mpih_rshift
55_gcry_mpih_rshift:
56	.frame	$30,0,$26,0
57
58	ldq	$4,0($17)	# load first limb
59	addq	$17,8,$17
60	subq	$31,$19,$7
61	subq	$18,1,$18
62	and	$18,4-1,$20	# number of limbs in first loop
63	sll	$4,$7,$0	# compute function result
64
65	beq	$20,.R0
66	subq	$18,$20,$18
67
68	.align	3
69.Roop0:
70	ldq	$3,0($17)
71	addq	$16,8,$16
72	addq	$17,8,$17
73	subq	$20,1,$20
74	srl	$4,$19,$5
75	sll	$3,$7,$6
76	bis	$3,$3,$4
77	bis	$5,$6,$8
78	stq	$8,-8($16)
79	bne	$20,.Roop0
80
81.R0:	beq	$18,.Rend
82
83	.align	3
84.Roop:	ldq	$3,0($17)
85	addq	$16,32,$16
86	subq	$18,4,$18
87	srl	$4,$19,$5
88	sll	$3,$7,$6
89
90	ldq	$4,8($17)
91	srl	$3,$19,$1
92	bis	$5,$6,$8
93	stq	$8,-32($16)
94	sll	$4,$7,$2
95
96	ldq	$3,16($17)
97	srl	$4,$19,$5
98	bis	$1,$2,$8
99	stq	$8,-24($16)
100	sll	$3,$7,$6
101
102	ldq	$4,24($17)
103	srl	$3,$19,$1
104	bis	$5,$6,$8
105	stq	$8,-16($16)
106	sll	$4,$7,$2
107
108	addq	$17,32,$17
109	bis	$1,$2,$8
110	stq	$8,-8($16)
111
112	bgt	$18,.Roop
113
114.Rend:	srl	$4,$19,$8
115	stq	$8,0($16)
116	ret	$31,($26),1
117	.end	_gcry_mpih_rshift
118
119