| 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 2009 Sun Microsystems, Inc. All rights reserved. |
| 23 | * Use is subject to license terms. |
| 24 | */ |
| 25 | |
| 26 | /* |
| 27 | * Copyright (c) 2012, 2015 by Delphix. All rights reserved. |
| 28 | */ |
| 29 | |
| 30 | #ifndef _ZIO_IMPL_H |
| 31 | #define _ZIO_IMPL_H |
| 32 | |
| 33 | #include <sys/zfs_context.h> |
| 34 | #include <sys/zio.h> |
| 35 | |
| 36 | #ifdef __cplusplus |
| 37 | extern "C" { |
| 38 | #endif |
| 39 | |
| 40 | /* |
| 41 | * XXX -- Describe ZFS I/O pipeline here. Fill in as needed. |
| 42 | * |
| 43 | * The ZFS I/O pipeline is comprised of various stages which are defined |
| 44 | * in the zio_stage enum below. The individual stages are used to construct |
| 45 | * these basic I/O operations: Read, Write, Free, Claim, and Ioctl. |
| 46 | * |
| 47 | * I/O operations: (XXX - provide detail for each of the operations) |
| 48 | * |
| 49 | * Read: |
| 50 | * Write: |
| 51 | * Free: |
| 52 | * Claim: |
| 53 | * Ioctl: |
| 54 | * |
| 55 | * Although the most common pipeline are used by the basic I/O operations |
| 56 | * above, there are some helper pipelines (one could consider them |
| 57 | * sub-pipelines) which are used internally by the ZIO module and are |
| 58 | * explained below: |
| 59 | * |
| 60 | * Interlock Pipeline: |
| 61 | * The interlock pipeline is the most basic pipeline and is used by all |
| 62 | * of the I/O operations. The interlock pipeline does not perform any I/O |
| 63 | * and is used to coordinate the dependencies between I/Os that are being |
| 64 | * issued (i.e. the parent/child relationship). |
| 65 | * |
| 66 | * Vdev child Pipeline: |
| 67 | * The vdev child pipeline is responsible for performing the physical I/O. |
| 68 | * It is in this pipeline where the I/O are queued and possibly cached. |
| 69 | * |
| 70 | * In addition to performing I/O, the pipeline is also responsible for |
| 71 | * data transformations. The transformations performed are based on the |
| 72 | * specific properties that user may have selected and modify the |
| 73 | * behavior of the pipeline. Examples of supported transformations are |
| 74 | * compression, dedup, and nop writes. Transformations will either modify |
| 75 | * the data or the pipeline. This list below further describes each of |
| 76 | * the supported transformations: |
| 77 | * |
| 78 | * Compression: |
| 79 | * ZFS supports three different flavors of compression -- gzip, lzjb, and |
| 80 | * zle. Compression occurs as part of the write pipeline and is performed |
| 81 | * in the ZIO_STAGE_WRITE_BP_INIT stage. |
| 82 | * |
| 83 | * Dedup: |
| 84 | * Dedup reads are handled by the ZIO_STAGE_DDT_READ_START and |
| 85 | * ZIO_STAGE_DDT_READ_DONE stages. These stages are added to an existing |
| 86 | * read pipeline if the dedup bit is set on the block pointer. |
| 87 | * Writing a dedup block is performed by the ZIO_STAGE_DDT_WRITE stage |
| 88 | * and added to a write pipeline if a user has enabled dedup on that |
| 89 | * particular dataset. |
| 90 | * |
| 91 | * NOP Write: |
| 92 | * The NOP write feature is performed by the ZIO_STAGE_NOP_WRITE stage |
| 93 | * and is added to an existing write pipeline if a crypographically |
| 94 | * secure checksum (i.e. SHA256) is enabled and compression is turned on. |
| 95 | * The NOP write stage will compare the checksums of the current data |
| 96 | * on-disk (level-0 blocks only) and the data that is currently being written. |
| 97 | * If the checksum values are identical then the pipeline is converted to |
| 98 | * an interlock pipeline skipping block allocation and bypassing the |
| 99 | * physical I/O. The nop write feature can handle writes in either |
| 100 | * syncing or open context (i.e. zil writes) and as a result is mutually |
| 101 | * exclusive with dedup. |
| 102 | */ |
| 103 | |
| 104 | /* |
| 105 | * zio pipeline stage definitions |
| 106 | */ |
| 107 | enum zio_stage { |
| 108 | ZIO_STAGE_OPEN = 1 << 0, /* RWFCI */ |
| 109 | |
| 110 | ZIO_STAGE_READ_BP_INIT = 1 << 1, /* R---- */ |
| 111 | ZIO_STAGE_WRITE_BP_INIT = 1 << 2, /* -W--- */ |
| 112 | ZIO_STAGE_FREE_BP_INIT = 1 << 3, /* --F-- */ |
| 113 | ZIO_STAGE_ISSUE_ASYNC = 1 << 4, /* RWF-- */ |
| 114 | ZIO_STAGE_WRITE_COMPRESS = 1 << 5, /* -W--- */ |
| 115 | |
| 116 | ZIO_STAGE_CHECKSUM_GENERATE = 1 << 6, /* -W--- */ |
| 117 | |
| 118 | ZIO_STAGE_NOP_WRITE = 1 << 7, /* -W--- */ |
| 119 | |
| 120 | ZIO_STAGE_DDT_READ_START = 1 << 8, /* R---- */ |
| 121 | ZIO_STAGE_DDT_READ_DONE = 1 << 9, /* R---- */ |
| 122 | ZIO_STAGE_DDT_WRITE = 1 << 10, /* -W--- */ |
| 123 | ZIO_STAGE_DDT_FREE = 1 << 11, /* --F-- */ |
| 124 | |
| 125 | ZIO_STAGE_GANG_ASSEMBLE = 1 << 12, /* RWFC- */ |
| 126 | ZIO_STAGE_GANG_ISSUE = 1 << 13, /* RWFC- */ |
| 127 | |
| 128 | ZIO_STAGE_DVA_THROTTLE = 1 << 14, /* -W--- */ |
| 129 | ZIO_STAGE_DVA_ALLOCATE = 1 << 15, /* -W--- */ |
| 130 | ZIO_STAGE_DVA_FREE = 1 << 16, /* --F-- */ |
| 131 | ZIO_STAGE_DVA_CLAIM = 1 << 17, /* ---C- */ |
| 132 | |
| 133 | ZIO_STAGE_READY = 1 << 18, /* RWFCI */ |
| 134 | |
| 135 | ZIO_STAGE_VDEV_IO_START = 1 << 19, /* RWF-I */ |
| 136 | ZIO_STAGE_VDEV_IO_DONE = 1 << 20, /* RWF-I */ |
| 137 | ZIO_STAGE_VDEV_IO_ASSESS = 1 << 21, /* RWF-I */ |
| 138 | |
| 139 | ZIO_STAGE_CHECKSUM_VERIFY = 1 << 22, /* R---- */ |
| 140 | |
| 141 | ZIO_STAGE_DONE = 1 << 23 /* RWFCI */ |
| 142 | }; |
| 143 | |
| 144 | #define ZIO_INTERLOCK_STAGES \ |
| 145 | (ZIO_STAGE_READY | \ |
| 146 | ZIO_STAGE_DONE) |
| 147 | |
| 148 | #define ZIO_INTERLOCK_PIPELINE \ |
| 149 | ZIO_INTERLOCK_STAGES |
| 150 | |
| 151 | #define ZIO_VDEV_IO_STAGES \ |
| 152 | (ZIO_STAGE_VDEV_IO_START | \ |
| 153 | ZIO_STAGE_VDEV_IO_DONE | \ |
| 154 | ZIO_STAGE_VDEV_IO_ASSESS) |
| 155 | |
| 156 | #define ZIO_VDEV_CHILD_PIPELINE \ |
| 157 | (ZIO_VDEV_IO_STAGES | \ |
| 158 | ZIO_STAGE_DONE) |
| 159 | |
| 160 | #define ZIO_READ_COMMON_STAGES \ |
| 161 | (ZIO_INTERLOCK_STAGES | \ |
| 162 | ZIO_VDEV_IO_STAGES | \ |
| 163 | ZIO_STAGE_CHECKSUM_VERIFY) |
| 164 | |
| 165 | #define ZIO_READ_PHYS_PIPELINE \ |
| 166 | ZIO_READ_COMMON_STAGES |
| 167 | |
| 168 | #define ZIO_READ_PIPELINE \ |
| 169 | (ZIO_READ_COMMON_STAGES | \ |
| 170 | ZIO_STAGE_READ_BP_INIT) |
| 171 | |
| 172 | #define ZIO_DDT_CHILD_READ_PIPELINE \ |
| 173 | ZIO_READ_COMMON_STAGES |
| 174 | |
| 175 | #define ZIO_DDT_READ_PIPELINE \ |
| 176 | (ZIO_INTERLOCK_STAGES | \ |
| 177 | ZIO_STAGE_READ_BP_INIT | \ |
| 178 | ZIO_STAGE_DDT_READ_START | \ |
| 179 | ZIO_STAGE_DDT_READ_DONE) |
| 180 | |
| 181 | #define ZIO_WRITE_COMMON_STAGES \ |
| 182 | (ZIO_INTERLOCK_STAGES | \ |
| 183 | ZIO_VDEV_IO_STAGES | \ |
| 184 | ZIO_STAGE_ISSUE_ASYNC | \ |
| 185 | ZIO_STAGE_CHECKSUM_GENERATE) |
| 186 | |
| 187 | #define ZIO_WRITE_PHYS_PIPELINE \ |
| 188 | ZIO_WRITE_COMMON_STAGES |
| 189 | |
| 190 | #define ZIO_REWRITE_PIPELINE \ |
| 191 | (ZIO_WRITE_COMMON_STAGES | \ |
| 192 | ZIO_STAGE_WRITE_COMPRESS | \ |
| 193 | ZIO_STAGE_WRITE_BP_INIT) |
| 194 | |
| 195 | #define ZIO_WRITE_PIPELINE \ |
| 196 | (ZIO_WRITE_COMMON_STAGES | \ |
| 197 | ZIO_STAGE_WRITE_BP_INIT | \ |
| 198 | ZIO_STAGE_WRITE_COMPRESS | \ |
| 199 | ZIO_STAGE_DVA_THROTTLE | \ |
| 200 | ZIO_STAGE_DVA_ALLOCATE) |
| 201 | |
| 202 | #define ZIO_DDT_CHILD_WRITE_PIPELINE \ |
| 203 | (ZIO_INTERLOCK_STAGES | \ |
| 204 | ZIO_VDEV_IO_STAGES | \ |
| 205 | ZIO_STAGE_DVA_THROTTLE | \ |
| 206 | ZIO_STAGE_DVA_ALLOCATE) |
| 207 | |
| 208 | #define ZIO_DDT_WRITE_PIPELINE \ |
| 209 | (ZIO_INTERLOCK_STAGES | \ |
| 210 | ZIO_STAGE_WRITE_BP_INIT | \ |
| 211 | ZIO_STAGE_ISSUE_ASYNC | \ |
| 212 | ZIO_STAGE_WRITE_COMPRESS | \ |
| 213 | ZIO_STAGE_CHECKSUM_GENERATE | \ |
| 214 | ZIO_STAGE_DDT_WRITE) |
| 215 | |
| 216 | #define ZIO_GANG_STAGES \ |
| 217 | (ZIO_STAGE_GANG_ASSEMBLE | \ |
| 218 | ZIO_STAGE_GANG_ISSUE) |
| 219 | |
| 220 | #define ZIO_FREE_PIPELINE \ |
| 221 | (ZIO_INTERLOCK_STAGES | \ |
| 222 | ZIO_STAGE_FREE_BP_INIT | \ |
| 223 | ZIO_STAGE_DVA_FREE) |
| 224 | |
| 225 | #define ZIO_FREE_PHYS_PIPELINE \ |
| 226 | (ZIO_INTERLOCK_STAGES | \ |
| 227 | ZIO_VDEV_IO_STAGES) |
| 228 | |
| 229 | #define ZIO_DDT_FREE_PIPELINE \ |
| 230 | (ZIO_INTERLOCK_STAGES | \ |
| 231 | ZIO_STAGE_FREE_BP_INIT | \ |
| 232 | ZIO_STAGE_ISSUE_ASYNC | \ |
| 233 | ZIO_STAGE_DDT_FREE) |
| 234 | |
| 235 | #define ZIO_CLAIM_PIPELINE \ |
| 236 | (ZIO_INTERLOCK_STAGES | \ |
| 237 | ZIO_STAGE_DVA_CLAIM) |
| 238 | |
| 239 | #define ZIO_IOCTL_PIPELINE \ |
| 240 | (ZIO_INTERLOCK_STAGES | \ |
| 241 | ZIO_STAGE_VDEV_IO_START | \ |
| 242 | ZIO_STAGE_VDEV_IO_ASSESS) |
| 243 | |
| 244 | #define ZIO_BLOCKING_STAGES \ |
| 245 | (ZIO_STAGE_DVA_ALLOCATE | \ |
| 246 | ZIO_STAGE_DVA_CLAIM | \ |
| 247 | ZIO_STAGE_VDEV_IO_START) |
| 248 | |
| 249 | extern void zio_inject_init(void); |
| 250 | extern void zio_inject_fini(void); |
| 251 | |
| 252 | #ifdef __cplusplus |
| 253 | } |
| 254 | #endif |
| 255 | |
| 256 | #endif /* _ZIO_IMPL_H */ |
| 257 | |