[mirror_zfs.git] / module / zfs / vdev_raidz_math_aarch64_neonx2.c

/*
 * 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 (C) 2016 Romain Dolbeau. All rights reserved.
 */

#include <sys/isa_defs.h>

#if defined(__aarch64__)

#include "vdev_raidz_math_aarch64_neon_common.h"

#define	GEN_P_DEFINE() \
	GEN_X_DEFINE_0_3()	\
	GEN_X_DEFINE_4_5()	\
	GEN_X_DEFINE_6_7()
#define	GEN_P_STRIDE		8
#define	GEN_P_P			0, 1, 2, 3, 4, 5, 6, 7

#define	GEN_PQ_DEFINE()	\
	GEN_X_DEFINE_0_3()	\
	GEN_X_DEFINE_4_5()	\
	GEN_X_DEFINE_6_7()	\
	GEN_X_DEFINE_8_9()	\
	GEN_X_DEFINE_10_11()	\
	GEN_X_DEFINE_16()	\
	GEN_X_DEFINE_17()	\
	GEN_X_DEFINE_33_36()
#define	GEN_PQ_STRIDE		4
#define	GEN_PQ_D		0, 1, 2, 3
#define	GEN_PQ_P		4, 5, 6, 7
#define	GEN_PQ_Q		8, 9, 10, 11

#define	GEN_PQR_DEFINE() \
	GEN_X_DEFINE_0_3()	\
	GEN_X_DEFINE_4_5()	\
	GEN_X_DEFINE_6_7()	\
	GEN_X_DEFINE_8_9()	\
	GEN_X_DEFINE_22_23()	\
	GEN_X_DEFINE_24_27()	\
	GEN_X_DEFINE_16()	\
	GEN_X_DEFINE_17()	\
	GEN_X_DEFINE_33_36()
#define	GEN_PQR_STRIDE		4
#define	GEN_PQR_D		0, 1, 2, 3
#define	GEN_PQR_P		4, 5, 6, 7
#define	GEN_PQR_Q		8, 9, 22, 23
#define	GEN_PQR_R		24, 25, 26, 27

#define	REC_P_DEFINE() \
	GEN_X_DEFINE_0_3()	\
	GEN_X_DEFINE_33_36()
#define	REC_P_STRIDE		4
#define	REC_P_X			0, 1, 2, 3

#define	REC_Q_DEFINE() \
	GEN_X_DEFINE_0_3()	\
	GEN_X_DEFINE_16()	\
	GEN_X_DEFINE_17()	\
	GEN_X_DEFINE_33_36()
#define	REC_Q_STRIDE		4
#define	REC_Q_X			0, 1, 2, 3

#define	REC_R_DEFINE() \
	GEN_X_DEFINE_0_3()	\
	GEN_X_DEFINE_16()	\
	GEN_X_DEFINE_17()	\
	GEN_X_DEFINE_33_36()
#define	REC_R_STRIDE		4
#define	REC_R_X			0, 1, 2, 3

#define	REC_PQ_DEFINE() \
	GEN_X_DEFINE_0_3()	\
	GEN_X_DEFINE_4_5()	\
	GEN_X_DEFINE_6_7()	\
	GEN_X_DEFINE_8_9()	\
	GEN_X_DEFINE_16()	\
	GEN_X_DEFINE_17()	\
	GEN_X_DEFINE_22_23()	\
	GEN_X_DEFINE_33_36()
#define	REC_PQ_STRIDE		4
#define	REC_PQ_X		0, 1, 2, 3
#define	REC_PQ_Y		4, 5, 6, 7
#define	REC_PQ_D		8, 9, 22, 23

#define	REC_PR_DEFINE()	REC_PQ_DEFINE()
#define	REC_PR_STRIDE		4
#define	REC_PR_X		0, 1, 2, 3
#define	REC_PR_Y		4, 5, 6, 7
#define	REC_PR_D		8, 9, 22, 23

#define	REC_QR_DEFINE()	REC_PQ_DEFINE()
#define	REC_QR_STRIDE		4
#define	REC_QR_X		0, 1, 2, 3
#define	REC_QR_Y		4, 5, 6, 7
#define	REC_QR_D		8, 9, 22, 23

#define	REC_PQR_DEFINE() \
	GEN_X_DEFINE_0_3()	\
	GEN_X_DEFINE_4_5()	\
	GEN_X_DEFINE_6_7()	\
	GEN_X_DEFINE_8_9()	\
	GEN_X_DEFINE_16()	\
	GEN_X_DEFINE_17()	\
	GEN_X_DEFINE_22_23()	\
	GEN_X_DEFINE_24_27()	\
	GEN_X_DEFINE_28_30()	\
	GEN_X_DEFINE_31()	\
	GEN_X_DEFINE_33_36()
#define	REC_PQR_STRIDE		4
#define	REC_PQR_X		0, 1, 2, 3
#define	REC_PQR_Y		4, 5, 6, 7
#define	REC_PQR_Z		8, 9, 22, 23
#define	REC_PQR_D		24, 25, 26, 27
#define	REC_PQR_XS		24, 25, 26, 27
#define	REC_PQR_YS		28, 29, 30, 31


#include <sys/vdev_raidz_impl.h>
#include "vdev_raidz_math_impl.h"

DEFINE_GEN_METHODS(aarch64_neonx2);
/*
 * If compiled with -O0, gcc doesn't do any stack frame coalescing
 * and -Wframe-larger-than=1024 is triggered in debug mode.
 */
#pragma GCC diagnostic ignored "-Wframe-larger-than="
DEFINE_REC_METHODS(aarch64_neonx2);
#pragma GCC diagnostic pop

static boolean_t
raidz_will_aarch64_neonx2_work(void)
{
	return (B_TRUE); // __arch64__ requires NEON
}

const raidz_impl_ops_t vdev_raidz_aarch64_neonx2_impl = {
	.init = NULL,
	.fini = NULL,
	.gen = RAIDZ_GEN_METHODS(aarch64_neonx2),
	.rec = RAIDZ_REC_METHODS(aarch64_neonx2),
	.is_supported = &raidz_will_aarch64_neonx2_work,
	.name = "aarch64_neonx2"
};

#endif /* defined(__aarch64__) */
Commit	Line	Data
62a65a65 RD	1	/*
	2	* CDDL HEADER START
	3	*
	4	* The contents of this file are subject to the terms of the
	5	* Common Development and Distribution License (the "License").
	6	* You may not use this file except in compliance with the License.
	7	*
	8	* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
	9	* or http://www.opensolaris.org/os/licensing.
	10	* See the License for the specific language governing permissions
	11	* and limitations under the License.
	12	*
	13	* When distributing Covered Code, include this CDDL HEADER in each
	14	* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
	15	* If applicable, add the following below this CDDL HEADER, with the
	16	* fields enclosed by brackets "[]" replaced with your own identifying
	17	* information: Portions Copyright [yyyy] [name of copyright owner]
	18	*
	19	* CDDL HEADER END
	20	*/
	21	/*
	22	* Copyright (C) 2016 Romain Dolbeau. All rights reserved.
	23	*/
	24
	25	#include <sys/isa_defs.h>
	26
	27	#if defined(__aarch64__)
	28
	29	#include "vdev_raidz_math_aarch64_neon_common.h"
	30
	31	#define GEN_P_DEFINE() \
	32	GEN_X_DEFINE_0_3() \
	33	GEN_X_DEFINE_4_5() \
	34	GEN_X_DEFINE_6_7()
	35	#define GEN_P_STRIDE 8
	36	#define GEN_P_P 0, 1, 2, 3, 4, 5, 6, 7
	37
	38	#define GEN_PQ_DEFINE() \
	39	GEN_X_DEFINE_0_3() \
	40	GEN_X_DEFINE_4_5() \
	41	GEN_X_DEFINE_6_7() \
	42	GEN_X_DEFINE_8_9() \
	43	GEN_X_DEFINE_10_11() \
	44	GEN_X_DEFINE_16() \
	45	GEN_X_DEFINE_17() \
	46	GEN_X_DEFINE_33_36()
	47	#define GEN_PQ_STRIDE 4
	48	#define GEN_PQ_D 0, 1, 2, 3
	49	#define GEN_PQ_P 4, 5, 6, 7
	50	#define GEN_PQ_Q 8, 9, 10, 11
	51
	52	#define GEN_PQR_DEFINE() \
	53	GEN_X_DEFINE_0_3() \
	54	GEN_X_DEFINE_4_5() \
	55	GEN_X_DEFINE_6_7() \
	56	GEN_X_DEFINE_8_9() \
	57	GEN_X_DEFINE_22_23() \
	58	GEN_X_DEFINE_24_27() \
	59	GEN_X_DEFINE_16() \
	60	GEN_X_DEFINE_17() \
	61	GEN_X_DEFINE_33_36()
	62	#define GEN_PQR_STRIDE 4
	63	#define GEN_PQR_D 0, 1, 2, 3
	64	#define GEN_PQR_P 4, 5, 6, 7
65	#define GEN_PQR_Q 8, 9, 22, 23
66	#define GEN_PQR_R 24, 25, 26, 27
67
68	#define REC_P_DEFINE() \
69	GEN_X_DEFINE_0_3() \
70	GEN_X_DEFINE_33_36()
71	#define REC_P_STRIDE 4
72	#define REC_P_X 0, 1, 2, 3
73
74	#define REC_Q_DEFINE() \
75	GEN_X_DEFINE_0_3() \
76	GEN_X_DEFINE_16() \
77	GEN_X_DEFINE_17() \
78	GEN_X_DEFINE_33_36()
79	#define REC_Q_STRIDE 4
80	#define REC_Q_X 0, 1, 2, 3
81
82	#define REC_R_DEFINE() \
83	GEN_X_DEFINE_0_3() \
84	GEN_X_DEFINE_16() \
85	GEN_X_DEFINE_17() \
86	GEN_X_DEFINE_33_36()
87	#define REC_R_STRIDE 4
88	#define REC_R_X 0, 1, 2, 3
89
90	#define REC_PQ_DEFINE() \
91	GEN_X_DEFINE_0_3() \
92	GEN_X_DEFINE_4_5() \
93	GEN_X_DEFINE_6_7() \
94	GEN_X_DEFINE_8_9() \
95	GEN_X_DEFINE_16() \
96	GEN_X_DEFINE_17() \
97	GEN_X_DEFINE_22_23() \
98	GEN_X_DEFINE_33_36()
99	#define REC_PQ_STRIDE 4
100	#define REC_PQ_X 0, 1, 2, 3
101	#define REC_PQ_Y 4, 5, 6, 7
102	#define REC_PQ_D 8, 9, 22, 23
103
104	#define REC_PR_DEFINE() REC_PQ_DEFINE()
105	#define REC_PR_STRIDE 4
106	#define REC_PR_X 0, 1, 2, 3
107	#define REC_PR_Y 4, 5, 6, 7
108	#define REC_PR_D 8, 9, 22, 23
109
110	#define REC_QR_DEFINE() REC_PQ_DEFINE()
111	#define REC_QR_STRIDE 4
112	#define REC_QR_X 0, 1, 2, 3
113	#define REC_QR_Y 4, 5, 6, 7
114	#define REC_QR_D 8, 9, 22, 23
115
116	#define REC_PQR_DEFINE() \
117	GEN_X_DEFINE_0_3() \
118	GEN_X_DEFINE_4_5() \
119	GEN_X_DEFINE_6_7() \
120	GEN_X_DEFINE_8_9() \
121	GEN_X_DEFINE_16() \
122	GEN_X_DEFINE_17() \
123	GEN_X_DEFINE_22_23() \
124	GEN_X_DEFINE_24_27() \
125	GEN_X_DEFINE_28_30() \
126	GEN_X_DEFINE_31() \
127	GEN_X_DEFINE_33_36()
128	#define REC_PQR_STRIDE 4
129	#define REC_PQR_X 0, 1, 2, 3
130	#define REC_PQR_Y 4, 5, 6, 7
131	#define REC_PQR_Z 8, 9, 22, 23
132	#define REC_PQR_D 24, 25, 26, 27
133	#define REC_PQR_XS 24, 25, 26, 27
134	#define REC_PQR_YS 28, 29, 30, 31
135
136
137	#include <sys/vdev_raidz_impl.h>
138	#include "vdev_raidz_math_impl.h"
139
140	DEFINE_GEN_METHODS(aarch64_neonx2);
141	/*
142	* If compiled with -O0, gcc doesn't do any stack frame coalescing
143	* and -Wframe-larger-than=1024 is triggered in debug mode.
144	*/
145	#pragma GCC diagnostic ignored "-Wframe-larger-than="
146	DEFINE_REC_METHODS(aarch64_neonx2);
147	#pragma GCC diagnostic pop
148
149	static boolean_t
150	raidz_will_aarch64_neonx2_work(void)
151	{
152	return (B_TRUE); // __arch64__ requires NEON
153	}
154
155	const raidz_impl_ops_t vdev_raidz_aarch64_neonx2_impl = {
156	.init = NULL,
157	.fini = NULL,
158	.gen = RAIDZ_GEN_METHODS(aarch64_neonx2),
159	.rec = RAIDZ_REC_METHODS(aarch64_neonx2),
160	.is_supported = &raidz_will_aarch64_neonx2_work,
161	.name = "aarch64_neonx2"
162	};
163
164	#endif /* defined(__aarch64__) */