Linux kernel mirror (for testing)
git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
tjh.dev
kernel
os
linux
1// SPDX-License-Identifier: GPL-2.0-only
2/*
3 * linux/lib/raid6/neon.c - RAID6 syndrome calculation using ARM NEON intrinsics
4 *
5 * Copyright (C) 2013 Linaro Ltd <ard.biesheuvel@linaro.org>
6 */
7
8#include <linux/raid/pq.h>
9
10#ifdef __KERNEL__
11#include <asm/simd.h>
12#else
13#define scoped_ksimd()
14#define cpu_has_neon() (1)
15#endif
16
17/*
18 * There are 2 reasons these wrappers are kept in a separate compilation unit
19 * from the actual implementations in neonN.c (generated from neon.uc by
20 * unroll.awk):
21 * - the actual implementations use NEON intrinsics, and the GCC support header
22 * (arm_neon.h) is not fully compatible (type wise) with the kernel;
23 * - the neonN.c files are compiled with -mfpu=neon and optimization enabled,
24 * and we have to make sure that we never use *any* NEON/VFP instructions
25 * outside a kernel_neon_begin()/kernel_neon_end() pair.
26 */
27
28#define RAID6_NEON_WRAPPER(_n) \
29 static void raid6_neon ## _n ## _gen_syndrome(int disks, \
30 size_t bytes, void **ptrs) \
31 { \
32 void raid6_neon ## _n ## _gen_syndrome_real(int, \
33 unsigned long, void**); \
34 scoped_ksimd() \
35 raid6_neon ## _n ## _gen_syndrome_real(disks, \
36 (unsigned long)bytes, ptrs); \
37 } \
38 static void raid6_neon ## _n ## _xor_syndrome(int disks, \
39 int start, int stop, \
40 size_t bytes, void **ptrs) \
41 { \
42 void raid6_neon ## _n ## _xor_syndrome_real(int, \
43 int, int, unsigned long, void**); \
44 scoped_ksimd() \
45 raid6_neon ## _n ## _xor_syndrome_real(disks, \
46 start, stop, (unsigned long)bytes, ptrs);\
47 } \
48 struct raid6_calls const raid6_neonx ## _n = { \
49 raid6_neon ## _n ## _gen_syndrome, \
50 raid6_neon ## _n ## _xor_syndrome, \
51 raid6_have_neon, \
52 "neonx" #_n, \
53 0 \
54 }
55
56static int raid6_have_neon(void)
57{
58 return cpu_has_neon();
59}
60
61RAID6_NEON_WRAPPER(1);
62RAID6_NEON_WRAPPER(2);
63RAID6_NEON_WRAPPER(4);
64RAID6_NEON_WRAPPER(8);