Subversion Repositories planix.SVN

Rev

Details | Last modification | View Log | RSS feed

Rev Author Line No. Line
2 - 1 
#define	BDNZ	BC	16,0,
2 
	TEXT	memmove(SB), $0
3 
	BR	move
4 
 
5 
	TEXT	memcpy(SB), $0
6 
move:
7 
 
8 
/*
9 
 * performance:
10 
 * (tba)
11 
 */
12 
 
13 
	MOVW	R3, s1+0(FP)
14 
	MOVW	n+8(FP), R9		/* R9 is count */
15 
	MOVW	R3, R10			/* R10 is to-pointer */
16 
	CMP	R9, $0
17 
	BEQ	ret
18 
	BLT	trap
19 
	MOVW	s2+4(FP), R11		/* R11 is from-pointer */
20 
 
21 
/*
22 
 * if no more than 16 bytes, just use one lsw/stsw
23 
 */
24 
	CMP	R9, $16
25 
	BLE	fout
26 
 
27 
	ADD	R9,R11, R13		/* R13 is end from-pointer */
28 
	ADD	R9,R10, R12		/* R12 is end to-pointer */
29 
 
30 
/*
31 
 * easiest test is copy backwards if
32 
 * destination string has higher mem address
33 
 */
34 
	CMPU	R10, R11
35 
	BGT	back
36 
 
37 
/*
38 
 * test if both pointers
39 
 * are similarly word aligned
40 
 */
41 
	XOR	R10,R11, R7
42 
	ANDCC	$3,R7
43 
	BNE	fbad
44 
 
45 
/*
46 
 * move a few bytes to align pointers
47 
 */
48 
	ANDCC	$3,R10,R7
49 
	BEQ	f2
50 
	SUBC	R7, $4, R7
51 
	SUB	R7, R9
52 
	MOVW	R7, XER
53 
	LSW	(R11), R16
54 
	ADD	R7, R11
55 
	STSW	R16, (R10)
56 
	ADD	R7, R10
57 
 
58 
/*
59 
 * turn R14 into doubleword count
60 
 * copy 16 bytes at a time while there's room.
61 
 */
62 
f2:
63 
	SRAWCC	$4, R9, R14
64 
	BLE	fout
65 
	MOVW	R14, CTR
66 
	SUB	$4, R11
67 
	SUB	$4, R10
68 
f3:
69 
	MOVWU	4(R11), R16
70 
	MOVWU	4(R11), R17
71 
	MOVWU	4(R11), R18
72 
	MOVWU	4(R11), R19
73 
	MOVWU	R16, 4(R10)
74 
	MOVWU	R17, 4(R10)
75 
	MOVWU	R18, 4(R10)
76 
	MOVWU	R19, 4(R10)
77 
	BDNZ	f3
78 
	RLWNMCC	$0, R9, $15, R9	/* residue */
79 
	BEQ	ret
80 
	ADD	$4, R11
81 
	ADD	$4, R10
82 
 
83 
/*
84 
 * move up to 16 bytes through R16 .. R19; aligned and unaligned
85 
 */
86 
fout:
87 
	MOVW	R9, XER
88 
	LSW	(R11), R16
89 
	STSW	R16, (R10)
90 
	BR	ret
91 
 
92 
/*
93 
 * loop for unaligned copy, then copy up to 15 remaining bytes
94 
 */
95 
fbad:
96 
	SRAWCC	$4, R9, R14
97 
	BLE	f6
98 
	MOVW	R14, CTR
99 
f5:
100 
	LSW	(R11), $16, R16
101 
	ADD	$16, R11
102 
	STSW	R16, $16, (R10)
103 
	ADD	$16, R10
104 
	BDNZ	f5
105 
	RLWNMCC	$0, R9, $15, R9	/* residue */
106 
	BEQ	ret
107 
f6:
108 
	MOVW	R9, XER
109 
	LSW	(R11), R16
110 
	STSW	R16, (R10)
111 
	BR	ret
112 
 
113 
/*
114 
 * whole thing repeated for backwards
115 
 */
116 
back:
117 
	CMP	R9, $4
118 
	BLT	bout
119 
 
120 
	XOR	R12,R13, R7
121 
	ANDCC	$3,R7
122 
	BNE	bout
123 
b1:
124 
	ANDCC	$3,R13, R7
125 
	BEQ	b2
126 
	MOVBZU	-1(R13), R16
127 
	MOVBZU	R16, -1(R12)
128 
	SUB	$1, R9
129 
	BR	b1
130 
b2:
131 
	SRAWCC	$4, R9, R14
132 
	BLE	b4
133 
	MOVW	R14, CTR
134 
b3:
135 
	MOVWU	-4(R13), R16
136 
	MOVWU	-4(R13), R17
137 
	MOVWU	-4(R13), R18
138 
	MOVWU	-4(R13), R19
139 
	MOVWU	R16, -4(R12)
140 
	MOVWU	R17, -4(R12)
141 
	MOVWU	R18, -4(R12)
142 
	MOVWU	R19, -4(R12)
143 
	BDNZ	b3
144 
	RLWNMCC	$0, R9, $15, R9	/* residue */
145 
	BEQ	ret
146 
b4:
147 
	SRAWCC	$2, R9, R14
148 
	BLE	bout
149 
	MOVW	R14, CTR
150 
b5:
151 
	MOVWU	-4(R13), R16
152 
	MOVWU	R16, -4(R12)
153 
	BDNZ	b5
154 
	RLWNMCC	$0, R9, $3, R9	/* residue */
155 
	BEQ	ret
156 
 
157 
bout:
158 
	CMPU	R13, R11
159 
	BLE	ret
160 
	MOVBZU	-1(R13), R16
161 
	MOVBZU	R16, -1(R12)
162 
	BR	bout
163 
 
164 
trap:
165 
	MOVW	$0, R0
166 
	MOVW	0(R0), R0
167 
 
168 
ret:
169 
	MOVW	s1+0(FP), R3
170 
	RETURN