1 | /* |
---|
2 | * Copyright (c) 2017 ARM Ltd |
---|
3 | * All rights reserved. |
---|
4 | * |
---|
5 | * Redistribution and use in source and binary forms, with or without |
---|
6 | * modification, are permitted provided that the following conditions |
---|
7 | * are met: |
---|
8 | * 1. Redistributions of source code must retain the above copyright |
---|
9 | * notice, this list of conditions and the following disclaimer. |
---|
10 | * 2. Redistributions in binary form must reproduce the above copyright |
---|
11 | * notice, this list of conditions and the following disclaimer in the |
---|
12 | * documentation and/or other materials provided with the distribution. |
---|
13 | * 3. The name of the company may not be used to endorse or promote |
---|
14 | * products derived from this software without specific prior written |
---|
15 | * permission. |
---|
16 | * |
---|
17 | * THIS SOFTWARE IS PROVIDED BY ARM LTD ``AS IS'' AND ANY EXPRESS OR IMPLIED |
---|
18 | * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF |
---|
19 | * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. |
---|
20 | * IN NO EVENT SHALL ARM LTD BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, |
---|
21 | * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED |
---|
22 | * TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR |
---|
23 | * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF |
---|
24 | * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING |
---|
25 | * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS |
---|
26 | * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
---|
27 | */ |
---|
28 | |
---|
29 | #if (defined (__OPTIMIZE_SIZE__) || defined (PREFER_SIZE_OVER_SPEED)) |
---|
30 | /* See memcmp-stub.c */ |
---|
31 | #else |
---|
32 | |
---|
33 | /* Assumptions: |
---|
34 | * |
---|
35 | * ARMv8-a, AArch64, unaligned accesses. |
---|
36 | */ |
---|
37 | |
---|
38 | /* Parameters and result. */ |
---|
39 | #define src1 x0 |
---|
40 | #define src2 x1 |
---|
41 | #define limit x2 |
---|
42 | #define result w0 |
---|
43 | |
---|
44 | /* Internal variables. */ |
---|
45 | #define data1 x3 |
---|
46 | #define data1w w3 |
---|
47 | #define data2 x4 |
---|
48 | #define data2w w4 |
---|
49 | #define tmp1 x5 |
---|
50 | |
---|
51 | .macro def_fn f p2align=0 |
---|
52 | .text |
---|
53 | .p2align \p2align |
---|
54 | .global \f |
---|
55 | .type \f, %function |
---|
56 | \f: |
---|
57 | .endm |
---|
58 | |
---|
59 | /* Small inputs of less than 8 bytes are handled separately. This allows the |
---|
60 | main code to be sped up using unaligned loads since there are now at least |
---|
61 | 8 bytes to be compared. If the first 8 bytes are equal, align src1. |
---|
62 | This ensures each iteration does at most one unaligned access even if both |
---|
63 | src1 and src2 are unaligned, and mutually aligned inputs behave as if |
---|
64 | aligned. After the main loop, process the last 8 bytes using unaligned |
---|
65 | accesses. */ |
---|
66 | |
---|
67 | def_fn memcmp p2align=6 |
---|
68 | subs limit, limit, 8 |
---|
69 | b.lo .Lless8 |
---|
70 | |
---|
71 | /* Limit >= 8, so check first 8 bytes using unaligned loads. */ |
---|
72 | ldr data1, [src1], 8 |
---|
73 | ldr data2, [src2], 8 |
---|
74 | and tmp1, src1, 7 |
---|
75 | add limit, limit, tmp1 |
---|
76 | cmp data1, data2 |
---|
77 | bne .Lreturn |
---|
78 | |
---|
79 | /* Align src1 and adjust src2 with bytes not yet done. */ |
---|
80 | sub src1, src1, tmp1 |
---|
81 | sub src2, src2, tmp1 |
---|
82 | |
---|
83 | subs limit, limit, 8 |
---|
84 | b.ls .Llast_bytes |
---|
85 | |
---|
86 | /* Loop performing 8 bytes per iteration using aligned src1. |
---|
87 | Limit is pre-decremented by 8 and must be larger than zero. |
---|
88 | Exit if <= 8 bytes left to do or if the data is not equal. */ |
---|
89 | .p2align 4 |
---|
90 | .Lloop8: |
---|
91 | ldr data1, [src1], 8 |
---|
92 | ldr data2, [src2], 8 |
---|
93 | subs limit, limit, 8 |
---|
94 | ccmp data1, data2, 0, hi /* NZCV = 0b0000. */ |
---|
95 | b.eq .Lloop8 |
---|
96 | |
---|
97 | cmp data1, data2 |
---|
98 | bne .Lreturn |
---|
99 | |
---|
100 | /* Compare last 1-8 bytes using unaligned access. */ |
---|
101 | .Llast_bytes: |
---|
102 | ldr data1, [src1, limit] |
---|
103 | ldr data2, [src2, limit] |
---|
104 | |
---|
105 | /* Compare data bytes and set return value to 0, -1 or 1. */ |
---|
106 | .Lreturn: |
---|
107 | #ifndef __AARCH64EB__ |
---|
108 | rev data1, data1 |
---|
109 | rev data2, data2 |
---|
110 | #endif |
---|
111 | cmp data1, data2 |
---|
112 | .Lret_eq: |
---|
113 | cset result, ne |
---|
114 | cneg result, result, lo |
---|
115 | ret |
---|
116 | |
---|
117 | .p2align 4 |
---|
118 | /* Compare up to 8 bytes. Limit is [-8..-1]. */ |
---|
119 | .Lless8: |
---|
120 | adds limit, limit, 4 |
---|
121 | b.lo .Lless4 |
---|
122 | ldr data1w, [src1], 4 |
---|
123 | ldr data2w, [src2], 4 |
---|
124 | cmp data1w, data2w |
---|
125 | b.ne .Lreturn |
---|
126 | sub limit, limit, 4 |
---|
127 | .Lless4: |
---|
128 | adds limit, limit, 4 |
---|
129 | beq .Lret_eq |
---|
130 | .Lbyte_loop: |
---|
131 | ldrb data1w, [src1], 1 |
---|
132 | ldrb data2w, [src2], 1 |
---|
133 | subs limit, limit, 1 |
---|
134 | ccmp data1w, data2w, 0, ne /* NZCV = 0b0000. */ |
---|
135 | b.eq .Lbyte_loop |
---|
136 | sub result, data1w, data2w |
---|
137 | ret |
---|
138 | |
---|
139 | .size memcmp, . - memcmp |
---|
140 | #endif |
---|