1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
|
/*
* CDDL HEADER START
*
* The contents of this file are subject to the terms of the
* Common Development and Distribution License (the "License").
* You may not use this file except in compliance with the License.
*
* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
* or http://www.opensolaris.org/os/licensing.
* See the License for the specific language governing permissions
* and limitations under the License.
*
* When distributing Covered Code, include this CDDL HEADER in each
* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
* If applicable, add the following below this CDDL HEADER, with the
* fields enclosed by brackets "[]" replaced with your own identifying
* information: Portions Copyright [yyyy] [name of copyright owner]
*
* CDDL HEADER END
*/
/*
* Copyright 2008 Sun Microsystems, Inc. All rights reserved.
* Use is subject to license terms.
*/
.file "__align_cpy_4.s"
/* __align_cpy_4(s1, s2, n)
*
* Copy 4-byte aligned source to 4-byte aligned target in multiples of 4 bytes.
*
* Input:
* o0 address of target
* o1 address of source
* o2 number of bytes to copy (must be a multiple of 4)
* Output:
* o0 address of target
* Caller's registers that have been changed by this function:
* o1-o5, g1, g5
*
* Note:
* This helper routine will not be used by any 32-bit compilations.
* To do so would break binary compatibility with previous versions of
* Solaris.
*
* Assumptions:
* Source and target addresses are 4-byte aligned.
* Bytes to be copied are non-overlapping or _exactly_ overlapping.
* The number of bytes to be copied is a multiple of 4.
* Call will usually be made with a byte count of more than 4*4 and
* less than a few hundred bytes. Legal values are 0 to MAX_SIZE_T.
*
* Optimization attempt:
* Reasonable speed for a generic v9.
*/
#include <sys/asm_linkage.h>
ENTRY(__align_cpy_4)
brz,pn %o2, .done ! Skip out if no bytes to copy.
cmp %o0, %o1
be,pn %xcc, .done ! Addresses are identical--done.
and %o0, 7, %o3 ! Is target 8-byte aligned?
and %o1, 7, %o4 ! Is source 8-byte aligned?
cmp %o3, %o4
bne,pt %icc, .noton8 ! Exactly one of source and target is
mov %o0, %g1 ! 8-byte aligned.
brz,pt %o3, .both8 ! Both are 8-byte aligned.
nop
ld [%o1], %o3 ! Neither is aligned, so do 4 bytes;
subcc %o2, 4, %o2 ! then both will be aligned.
st %o3, [%g1]
bz,pn %xcc, .done
add %g1, 4, %g1
b .both8
add %o1, 4, %o1
! Section of code dealing with case where source and target are both 8-byte
! aligned. Get and store 16 bytes at a time using ldx and stx.
.align 32
.both8: ! Both source and target are aligned.
cmp %o2, 16
bl,a,pn %xcc, .chkwd
cmp %o2, 8
sub %o2, 12, %o2
.loop16a: ! Load and store 16 bytes at a time.
ldx [%o1], %o3
ldx [%o1+8], %o4
subcc %o2, 16, %o2
stx %o3, [%g1]
stx %o4, [%g1+8]
add %o1, 16, %o1
bg,pt %xcc, .loop16a ! Have at least 16 bytes left.
add %g1, 16, %g1
addcc %o2, 12, %o2
bg,a,pt %xcc, .chkwd ! Have some remaining bytes.
cmp %o2, 8
retl
nop
.chkwd:
bl,a,pn %xcc, .wrword ! Only 4 bytes left.
ld [%o1], %o3
ldx [%o1], %o3 ! Have 8 or 12, so do 8.
stx %o3, [%g1]
add %o1, 8, %o1
add %g1, 8, %g1
subcc %o2, 8, %o2
bg,a,pn %xcc, .wrword ! Still have four to do.
ld [%o1], %o3
retl
nop
.wrword: ! Copy final word.
st %o3, [%g1]
.done:
retl
nop
! Section of code where either source or target, but not both, are 8-byte
! aligned. So, use ld and st instructions rather than trying to copy stuff
! around in registers.
.align 32 ! Ultra cache line boundary.
.noton8:
add %o1, %o2, %g5 ! Ending address of source.
andcc %o2, 15, %o3 ! Mod 16 of number of bytes to copy.
bz,pn %xcc, .loop16 ! Copy odd amounts first, then multiples of 16.
cmp %o3, 4
bz,pn %xcc, .mod4
cmp %o3, 8
bz,pn %xcc, .mod8
cmp %o3, 12
bz,pt %xcc, .mod12
nop
illtrap 0 ! Size not valid.
.mod4: ! Do first 4 bytes, then do multiples of 16.
lduw [%o1], %o2
add %o1, 4, %o1
st %o2, [%g1]
cmp %o1, %g5
bl,a,pt %xcc, .loop16
add %g1, 4, %g1
retl
nop
.mod8: ! Do first 8 bytes, then do multiples of 16.
lduw [%o1], %o2
lduw [%o1+4], %o3
add %o1, 8, %o1
st %o2, [%g1]
st %o3, [%g1+4]
cmp %o1, %g5
bl,a,pt %xcc, .loop16
add %g1, 8, %g1
retl
nop
.mod12: ! Do first 12 bytes, then do multiples of 16.
lduw [%o1], %o2
lduw [%o1+4], %o3
lduw [%o1+8], %o4
add %o1, 12, %o1
st %o2, [%g1]
st %o3, [%g1+4]
st %o4, [%g1+8]
cmp %o1, %g5
bl,a,pt %xcc, .loop16
add %g1, 12, %g1
retl
nop
.align 32 ! Ultra cache line boundary.
.loop16: ! Do multiples of 16 bytes.
lduw [%o1], %o2
lduw [%o1+4], %o3
lduw [%o1+8], %o4
lduw [%o1+12], %o5
add %o1, 16, %o1
st %o2, [%g1]
st %o3, [%g1+4]
cmp %o1, %g5
st %o4, [%g1+8]
st %o5, [%g1+12]
bl,a,pt %xcc, .loop16
add %g1, 16,%g1
retl ! Target address is already in o0.
nop
SET_SIZE(__align_cpy_4)
|
