1/*
2   Copyright (c) 2015-2024, Synopsys, Inc. All rights reserved.
3
4   Redistribution and use in source and binary forms, with or without
5   modification, are permitted provided that the following conditions are met:
6
7   1) Redistributions of source code must retain the above copyright notice,
8   this list of conditions and the following disclaimer.
9
10   2) Redistributions in binary form must reproduce the above copyright notice,
11   this list of conditions and the following disclaimer in the documentation
12   and/or other materials provided with the distribution.
13
14   3) Neither the name of the Synopsys, Inc., nor the names of its contributors
15   may be used to endorse or promote products derived from this software
16   without specific prior written permission.
17
18   THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
19   AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
20   IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
21   ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE
22   LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
23   CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
24   SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
25   INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
26   CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
27   ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
28   POSSIBILITY OF SUCH DAMAGE.
29*/
30
31/* This implementation is optimized for performance.  For code size a generic
32   implementation of this function from newlib/libc/string/memcmp.c will be
33   used.  */
34#include <picolibc.h>
35
36#if !defined (__OPTIMIZE_SIZE__) && !defined (PREFER_SIZE_OVER_SPEED) \
37    && !defined (__ARC_RF16__)
38
39#include "asm.h"
40
41#if defined (__ARC601__) || !defined (__ARC_NORM__) \
42    || !defined (__ARC_BARREL_SHIFTER__)
43
44/* Addresses are unsigned, and at 0 is the vector table, so it's OK to assume
45   that we can subtract 8 from a source end address without underflow.  */
46
47ENTRY (memcmp)
48	or	r12,r0,r1
49	tst	r12,3
50	breq	r2,0,.Lnil
51	add_s	r3,r0,r2
52
53/* This algorithm for big endian targets sometimes works incorrectly
54   when sources are aligned. To be precise the last step is omitted.
55   Just use a simple bytewise variant until the algorithm is reviewed
56   and fixed.  */
57
58#ifdef __LITTLE_ENDIAN__
59	bne_s	.Lbytewise
60#else /* BIG ENDIAN */
61	b_s	.Lbytewise
62#endif /* ENDIAN */
63	sub	r6,r3,8
64	ld	r4,[r0,0]
65	ld	r5,[r1,0]
662:
67	brhs	r0,r6,.Loop_end
68	ld_s	r3,[r0,4]
69	ld_s	r12,[r1,4]
70	brne	r4,r5,.Leven
71	ld.a	r4,[r0,8]
72	breq.d	r3,r12,2b
73	ld.a	r5,[r1,8]
74#ifdef __LITTLE_ENDIAN__
75	mov_s	r4,r3
76	b.d	.Lodd
77	mov_s	r5,r12
78#else /* BIG ENDIAN */
79	cmp_s	r3,r12
80	j_s.d	[blink]
81	rrc	r0,2
82#endif /* ENDIAN */
83
84	.balign	4
85.Loop_end:
86	sub	r3,r0,r6
87	brhs	r3,4,.Last_cmp
88	brne	r4,r5,.Leven
89	ld	r4,[r0,4]
90	ld	r5,[r1,4]
91#ifdef __LITTLE_ENDIAN__
92	.balign	4
93.Last_cmp:
94	mov_l	r0,24
95	add3	r2,r0,r2
96	xor	r0,r4,r5
97	b.d	.Leven_cmp
98	bset	r0,r0,r2
99.Lodd:
100.Leven:
101	xor	r0,r4,r5
102.Leven_cmp:
103	mov_s	r1,0x80808080
104	; uses long immediate
105	sub_s	r12,r0,1
106	bic_s	r0,r0,r12
107	sub	r0,r1,r0
108	xor_s	r0,r0,r1
109	and	r1,r5,r0
110	and	r0,r4,r0
111#else /* BIG ENDIAN */
112.Last_cmp:
113	mov_s	r3,0
114	sub3	r2,r3,r2
115	sub_s	r3,r3,1
116	bclr	r3,r3,r2
117	add_l	r3,r3,1
118	and	r0,r4,r3
119	and	r1,r5,r3
120.Leven:
121#endif /* ENDIAN */
122	xor.f	0,r0,r1
123	sub_s	r0,r0,r1
124	j_s.d	[blink]
125	mov.mi	r0,r1
126	.balign	4
127.Lbytewise:
128	ldb	r4,[r0,0]
129	ldb	r5,[r1,0]
130	sub	r6,r3,2
1313:
132	brhs	r0,r6,.Lbyte_end
133	ldb_s	r3,[r0,1]
134	ldb_s	r12,[r1,1]
135	brne	r4,r5,.Lbyte_even
136	ldb.a	r4,[r0,2]
137	breq.d	r3,r12,3b
138	ldb.a	r5,[r1,2]
139.Lbyte_odd:
140	j_s.d	[blink]
141	sub	r0,r3,r12
142	.balign	4
143.Lbyte_end:
144	bbit1	r2,0,.Lbyte_even
145	brne	r4,r5,.Lbyte_even
146	ldb	r4,[r0,1]
147	ldb	r5,[r1,1]
148.Lbyte_even:
149	j_s.d	[blink]
150	sub	r0,r4,r5
151.Lnil:
152	j_s.d	[blink]
153	mov_s	r0,0
154ENDFUNC (memcmp)
155#endif /* __ARC601__ || !__ARC_NORM__ || !__ARC_BARREL_SHIFTER__ */
156
157#endif /* !__OPTIMIZE_SIZE__ && !PREFER_SIZE_OVER_SPEED */
158