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) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
23 * Copyright (c) 2012, 2017 by Delphix. All rights reserved.
24 * Copyright (c) 2013 by Saso Kiselkov. All rights reserved.
25 */
26
27#ifndef	_SYS_ARC_H
28#define	_SYS_ARC_H
29
30#include <sys/zfs_context.h>
31
32#ifdef	__cplusplus
33extern "C" {
34#endif
35
36#include <sys/zio.h>
37#include <sys/dmu.h>
38#include <sys/spa.h>
39
40/*
41 * Used by arc_flush() to inform arc_evict_state() that it should evict
42 * all available buffers from the arc state being passed in.
43 */
44#define	ARC_EVICT_ALL	-1ULL
45
46#define	HDR_SET_LSIZE(hdr, x) do { \
47	ASSERT(IS_P2ALIGNED(x, 1U << SPA_MINBLOCKSHIFT)); \
48	(hdr)->b_lsize = ((x) >> SPA_MINBLOCKSHIFT); \
49_NOTE(CONSTCOND) } while (0)
50
51#define	HDR_SET_PSIZE(hdr, x) do { \
52	ASSERT(IS_P2ALIGNED((x), 1U << SPA_MINBLOCKSHIFT)); \
53	(hdr)->b_psize = ((x) >> SPA_MINBLOCKSHIFT); \
54_NOTE(CONSTCOND) } while (0)
55
56#define	HDR_GET_LSIZE(hdr)	((hdr)->b_lsize << SPA_MINBLOCKSHIFT)
57#define	HDR_GET_PSIZE(hdr)	((hdr)->b_psize << SPA_MINBLOCKSHIFT)
58
59typedef struct arc_buf_hdr arc_buf_hdr_t;
60typedef struct arc_buf arc_buf_t;
61
62/*
63 * Because the ARC can store encrypted data, errors (not due to bugs) may arise
64 * while transforming data into its desired format - specifically, when
65 * decrypting, the key may not be present, or the HMAC may not be correct
66 * which signifies deliberate tampering with the on-disk state
67 * (assuming that the checksum was correct). The "error" parameter will be
68 * nonzero in this case, even if there is no associated zio.
69 */
70typedef void arc_read_done_func_t(zio_t *zio, const zbookmark_phys_t *zb,
71    const blkptr_t *bp, arc_buf_t *buf, void *priv);
72typedef void arc_write_done_func_t(zio_t *zio, arc_buf_t *buf, void *priv);
73
74/* generic arc_done_func_t's which you can use */
75arc_read_done_func_t arc_bcopy_func;
76arc_read_done_func_t arc_getbuf_func;
77
78typedef enum arc_flags
79{
80	/*
81	 * Public flags that can be passed into the ARC by external consumers.
82	 */
83	ARC_FLAG_WAIT			= 1 << 0,	/* perform sync I/O */
84	ARC_FLAG_NOWAIT			= 1 << 1,	/* perform async I/O */
85	ARC_FLAG_PREFETCH		= 1 << 2,	/* I/O is a prefetch */
86	ARC_FLAG_CACHED			= 1 << 3,	/* I/O was in cache */
87	ARC_FLAG_L2CACHE		= 1 << 4,	/* cache in L2ARC */
88	ARC_FLAG_PREDICTIVE_PREFETCH	= 1 << 5,	/* I/O from zfetch */
89	ARC_FLAG_PRESCIENT_PREFETCH	= 1 << 6,	/* long min lifespan */
90
91	/*
92	 * Private ARC flags.  These flags are private ARC only flags that
93	 * will show up in b_flags in the arc_hdr_buf_t. These flags should
94	 * only be set by ARC code.
95	 */
96	ARC_FLAG_IN_HASH_TABLE		= 1 << 7,	/* buffer is hashed */
97	ARC_FLAG_IO_IN_PROGRESS		= 1 << 8,	/* I/O in progress */
98	ARC_FLAG_IO_ERROR		= 1 << 9,	/* I/O failed for buf */
99	ARC_FLAG_INDIRECT		= 1 << 10,	/* indirect block */
100	/* Indicates that block was read with ASYNC priority. */
101	ARC_FLAG_PRIO_ASYNC_READ	= 1 << 11,
102	ARC_FLAG_L2_WRITING		= 1 << 12,	/* write in progress */
103	ARC_FLAG_L2_EVICTED		= 1 << 13,	/* evicted during I/O */
104	ARC_FLAG_L2_WRITE_HEAD		= 1 << 14,	/* head of write list */
105	/*
106	 * Encrypted or authenticated on disk (may be plaintext in memory).
107	 * This header has b_crypt_hdr allocated. Does not include indirect
108	 * blocks with checksums of MACs which will also have their X
109	 * (encrypted) bit set in the bp.
110	 */
111	ARC_FLAG_PROTECTED		= 1 << 15,
112	/* data has not been authenticated yet */
113	ARC_FLAG_NOAUTH			= 1 << 16,
114	/* indicates that the buffer contains metadata (otherwise, data) */
115	ARC_FLAG_BUFC_METADATA		= 1 << 17,
116
117	/* Flags specifying whether optional hdr struct fields are defined */
118	ARC_FLAG_HAS_L1HDR		= 1 << 18,
119	ARC_FLAG_HAS_L2HDR		= 1 << 19,
120
121	/*
122	 * Indicates the arc_buf_hdr_t's b_pdata matches the on-disk data.
123	 * This allows the l2arc to use the blkptr's checksum to verify
124	 * the data without having to store the checksum in the hdr.
125	 */
126	ARC_FLAG_COMPRESSED_ARC		= 1 << 20,
127	ARC_FLAG_SHARED_DATA		= 1 << 21,
128
129	/*
130	 * The arc buffer's compression mode is stored in the top 7 bits of the
131	 * flags field, so these dummy flags are included so that MDB can
132	 * interpret the enum properly.
133	 */
134	ARC_FLAG_COMPRESS_0		= 1 << 24,
135	ARC_FLAG_COMPRESS_1		= 1 << 25,
136	ARC_FLAG_COMPRESS_2		= 1 << 26,
137	ARC_FLAG_COMPRESS_3		= 1 << 27,
138	ARC_FLAG_COMPRESS_4		= 1 << 28,
139	ARC_FLAG_COMPRESS_5		= 1 << 29,
140	ARC_FLAG_COMPRESS_6		= 1 << 30
141
142} arc_flags_t;
143
144typedef enum arc_buf_flags {
145	ARC_BUF_FLAG_SHARED		= 1 << 0,
146	ARC_BUF_FLAG_COMPRESSED		= 1 << 1,
147	/*
148	 * indicates whether this arc_buf_t is encrypted, regardless of
149	 * state on-disk
150	 */
151	ARC_BUF_FLAG_ENCRYPTED		= 1 << 2
152} arc_buf_flags_t;
153
154struct arc_buf {
155	arc_buf_hdr_t		*b_hdr;
156	arc_buf_t		*b_next;
157	kmutex_t		b_evict_lock;
158	void			*b_data;
159	arc_buf_flags_t		b_flags;
160};
161
162typedef enum arc_buf_contents {
163	ARC_BUFC_INVALID,			/* invalid type */
164	ARC_BUFC_DATA,				/* buffer contains data */
165	ARC_BUFC_METADATA,			/* buffer contains metadata */
166	ARC_BUFC_NUMTYPES
167} arc_buf_contents_t;
168
169/*
170 * The following breakdows of arc_size exist for kstat only.
171 */
172typedef enum arc_space_type {
173	ARC_SPACE_DATA,
174	ARC_SPACE_META,
175	ARC_SPACE_HDRS,
176	ARC_SPACE_L2HDRS,
177	ARC_SPACE_OTHER,
178	ARC_SPACE_BONUS,
179	ARC_SPACE_NUMTYPES
180} arc_space_type_t;
181
182void arc_space_consume(uint64_t space, arc_space_type_t type);
183void arc_space_return(uint64_t space, arc_space_type_t type);
184boolean_t arc_is_metadata(arc_buf_t *buf);
185boolean_t arc_is_encrypted(arc_buf_t *buf);
186boolean_t arc_is_unauthenticated(arc_buf_t *buf);
187enum zio_compress arc_get_compression(arc_buf_t *buf);
188void arc_get_raw_params(arc_buf_t *buf, boolean_t *byteorder, uint8_t *salt,
189    uint8_t *iv, uint8_t *mac);
190int arc_untransform(arc_buf_t *buf, spa_t *spa, const zbookmark_phys_t *zb,
191    boolean_t in_place);
192void arc_convert_to_raw(arc_buf_t *buf, uint64_t dsobj, boolean_t byteorder,
193    dmu_object_type_t ot, const uint8_t *salt, const uint8_t *iv,
194    const uint8_t *mac);
195arc_buf_t *arc_alloc_buf(spa_t *spa, void *tag, arc_buf_contents_t type,
196    int32_t size);
197arc_buf_t *arc_alloc_compressed_buf(spa_t *spa, void *tag,
198    uint64_t psize, uint64_t lsize, enum zio_compress compression_type);
199arc_buf_t *arc_alloc_raw_buf(spa_t *spa, void *tag, uint64_t dsobj,
200    boolean_t byteorder, const uint8_t *salt, const uint8_t *iv,
201    const uint8_t *mac, dmu_object_type_t ot, uint64_t psize, uint64_t lsize,
202    enum zio_compress compression_type);
203arc_buf_t *arc_loan_buf(spa_t *spa, boolean_t is_metadata, int size);
204arc_buf_t *arc_loan_compressed_buf(spa_t *spa, uint64_t psize, uint64_t lsize,
205    enum zio_compress compression_type);
206arc_buf_t *arc_loan_raw_buf(spa_t *spa, uint64_t dsobj, boolean_t byteorder,
207    const uint8_t *salt, const uint8_t *iv, const uint8_t *mac,
208    dmu_object_type_t ot, uint64_t psize, uint64_t lsize,
209    enum zio_compress compression_type);
210void arc_return_buf(arc_buf_t *buf, void *tag);
211void arc_loan_inuse_buf(arc_buf_t *buf, void *tag);
212void arc_buf_destroy(arc_buf_t *buf, void *tag);
213int arc_buf_size(arc_buf_t *buf);
214int arc_buf_lsize(arc_buf_t *buf);
215void arc_buf_access(arc_buf_t *buf);
216void arc_release(arc_buf_t *buf, void *tag);
217int arc_released(arc_buf_t *buf);
218void arc_buf_freeze(arc_buf_t *buf);
219void arc_buf_thaw(arc_buf_t *buf);
220#ifdef ZFS_DEBUG
221int arc_referenced(arc_buf_t *buf);
222#endif
223
224int arc_read(zio_t *pio, spa_t *spa, const blkptr_t *bp,
225    arc_read_done_func_t *done, void *private, zio_priority_t priority,
226    int flags, arc_flags_t *arc_flags, const zbookmark_phys_t *zb);
227zio_t *arc_write(zio_t *pio, spa_t *spa, uint64_t txg,
228    blkptr_t *bp, arc_buf_t *buf, boolean_t l2arc, const zio_prop_t *zp,
229    arc_write_done_func_t *ready, arc_write_done_func_t *child_ready,
230    arc_write_done_func_t *physdone, arc_write_done_func_t *done,
231    void *private, zio_priority_t priority, int zio_flags,
232    const zbookmark_phys_t *zb);
233void arc_freed(spa_t *spa, const blkptr_t *bp);
234
235void arc_flush(spa_t *spa, boolean_t retry);
236void arc_tempreserve_clear(uint64_t reserve);
237int arc_tempreserve_space(spa_t *spa, uint64_t reserve, uint64_t txg);
238
239uint64_t arc_all_memory(void);
240uint64_t arc_max_bytes(void);
241void arc_init(void);
242void arc_fini(void);
243
244/*
245 * Level 2 ARC
246 */
247
248void l2arc_add_vdev(spa_t *spa, vdev_t *vd);
249void l2arc_remove_vdev(vdev_t *vd);
250boolean_t l2arc_vdev_present(vdev_t *vd);
251void l2arc_init(void);
252void l2arc_fini(void);
253void l2arc_start(void);
254void l2arc_stop(void);
255
256#ifndef _KERNEL
257extern boolean_t arc_watch;
258extern int arc_procfd;
259#endif
260
261#ifdef	__cplusplus
262}
263#endif
264
265#endif /* _SYS_ARC_H */
266