1/* alpha - left shift
2 *
3 *      Copyright (C) 1994, 1995, 1998, 2001,
4 *                    2002  Free Software Foundation, Inc.
5 *
6 *
7 * This file is part of Libgcrypt.
8 *
9 * Libgcrypt is free software; you can redistribute it and/or modify
10 * it under the terms of the GNU Lesser General Public License as
11 * published by the Free Software Foundation; either version 2.1 of
12 * the License, or (at your option) any later version.
13 *
14 * Libgcrypt is distributed in the hope that it will be useful,
15 * but WITHOUT ANY WARRANTY; without even the implied warranty of
16 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
17 * GNU Lesser General Public License for more details.
18 *
19 * You should have received a copy of the GNU Lesser General Public
20 * License along with this program; if not, write to the Free Software
21 * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA
22 *
23 * Note: This code is heavily based on the GNU MP Library.
24 *	 Actually it's the same code with only minor changes in the
25 *	 way the data is stored; this is to support the abstraction
26 *	 of an optional secure memory allocation which may be used
27 *	 to avoid revealing of sensitive data due to paging etc.
28 */
29
30
31
32/*******************
33 * mpi_limb_t
34 * _gcry_mpih_lshift( mpi_ptr_t wp,	(r16)
35 *		   mpi_ptr_t up,	(r17)
36 *		   mpi_size_t usize,	(r18)
37 *		   unsigned cnt)	(r19)
38 *
39 * This code runs at 4.8 cycles/limb on the 21064.  With infinite unrolling,
40 * it would take 4 cycles/limb.  It should be possible to get down to 3
41 * cycles/limb since both ldq and stq can be paired with the other used
42 * instructions.  But there are many restrictions in the 21064 pipeline that
43 * makes it hard, if not impossible, to get down to 3 cycles/limb:
44 *
45 * 1. ldq has a 3 cycle delay, srl and sll have a 2 cycle delay.
46 * 2. Only aligned instruction pairs can be paired.
47 * 3. The store buffer or silo might not be able to deal with the bandwidth.
48 */
49
50	.set	noreorder
51	.set	noat
52.text
53	.align	3
54	.globl	_gcry_mpih_lshift
55	.ent	_gcry_mpih_lshift
56_gcry_mpih_lshift:
57	.frame	$30,0,$26,0
58
59	s8addq	$18,$17,$17	# make r17 point at end of s1
60	ldq	$4,-8($17)	# load first limb
61	subq	$17,8,$17
62	subq	$31,$19,$7
63	s8addq	$18,$16,$16	# make r16 point at end of RES
64	subq	$18,1,$18
65	and	$18,4-1,$20	# number of limbs in first loop
66	srl	$4,$7,$0	# compute function result
67
68	beq	$20,.L0
69	subq	$18,$20,$18
70
71	.align	3
72.Loop0:
73	ldq	$3,-8($17)
74	subq	$16,8,$16
75	subq	$17,8,$17
76	subq	$20,1,$20
77	sll	$4,$19,$5
78	srl	$3,$7,$6
79	bis	$3,$3,$4
80	bis	$5,$6,$8
81	stq	$8,0($16)
82	bne	$20,.Loop0
83
84.L0:	beq	$18,.Lend
85
86	.align	3
87.Loop:	ldq	$3,-8($17)
88	subq	$16,32,$16
89	subq	$18,4,$18
90	sll	$4,$19,$5
91	srl	$3,$7,$6
92
93	ldq	$4,-16($17)
94	sll	$3,$19,$1
95	bis	$5,$6,$8
96	stq	$8,24($16)
97	srl	$4,$7,$2
98
99	ldq	$3,-24($17)
100	sll	$4,$19,$5
101	bis	$1,$2,$8
102	stq	$8,16($16)
103	srl	$3,$7,$6
104
105	ldq	$4,-32($17)
106	sll	$3,$19,$1
107	bis	$5,$6,$8
108	stq	$8,8($16)
109	srl	$4,$7,$2
110
111	subq	$17,32,$17
112	bis	$1,$2,$8
113	stq	$8,0($16)
114
115	bgt	$18,.Loop
116
117.Lend:	sll	$4,$19,$8
118	stq	$8,-8($16)
119	ret	$31,($26),1
120	.end	_gcry_mpih_lshift
121
122
123