1 /* 2 * Copyright (c) 2020 iXsystems, Inc. 3 * All rights reserved. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 8 * 1. Redistributions of source code must retain the above copyright 9 * notice, this list of conditions and the following disclaimer. 10 * 2. Redistributions in binary form must reproduce the above copyright 11 * notice, this list of conditions and the following disclaimer in the 12 * documentation and/or other materials provided with the distribution. 13 * 14 * THIS SOFTWARE IS PROVIDED BY THE AUTHORS AND CONTRIBUTORS ``AS IS'' AND 15 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 16 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 17 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE LIABLE 18 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 19 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 20 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 21 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 22 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 23 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 24 * SUCH DAMAGE. 25 * 26 */ 27 28 #include <sys/cdefs.h> 29 __FBSDID("$FreeBSD$"); 30 31 #include <sys/types.h> 32 #include <sys/param.h> 33 #include <sys/systm.h> 34 #include <sys/conf.h> 35 #include <sys/kernel.h> 36 #include <sys/lock.h> 37 #include <sys/malloc.h> 38 #include <sys/mutex.h> 39 #include <sys/proc.h> 40 #include <sys/errno.h> 41 #include <sys/uio.h> 42 #include <sys/buf.h> 43 #include <sys/file.h> 44 #include <sys/kmem.h> 45 #include <sys/conf.h> 46 #include <sys/cmn_err.h> 47 #include <sys/stat.h> 48 #include <sys/zfs_ioctl.h> 49 #include <sys/zfs_vfsops.h> 50 #include <sys/zfs_znode.h> 51 #include <sys/zap.h> 52 #include <sys/spa.h> 53 #include <sys/spa_impl.h> 54 #include <sys/vdev.h> 55 #include <sys/vdev_impl.h> 56 #include <sys/arc_os.h> 57 #include <sys/dmu.h> 58 #include <sys/dsl_dir.h> 59 #include <sys/dsl_dataset.h> 60 #include <sys/dsl_prop.h> 61 #include <sys/dsl_deleg.h> 62 #include <sys/dmu_objset.h> 63 #include <sys/dmu_impl.h> 64 #include <sys/dmu_tx.h> 65 #include <sys/sunddi.h> 66 #include <sys/policy.h> 67 #include <sys/zone.h> 68 #include <sys/nvpair.h> 69 #include <sys/mount.h> 70 #include <sys/taskqueue.h> 71 #include <sys/sdt.h> 72 #include <sys/fs/zfs.h> 73 #include <sys/zfs_ctldir.h> 74 #include <sys/zfs_dir.h> 75 #include <sys/zfs_onexit.h> 76 #include <sys/zvol.h> 77 #include <sys/dsl_scan.h> 78 #include <sys/dmu_objset.h> 79 #include <sys/dmu_send.h> 80 #include <sys/dsl_destroy.h> 81 #include <sys/dsl_bookmark.h> 82 #include <sys/dsl_userhold.h> 83 #include <sys/zfeature.h> 84 #include <sys/zcp.h> 85 #include <sys/zio_checksum.h> 86 #include <sys/vdev_removal.h> 87 #include <sys/dsl_crypt.h> 88 89 #include <sys/zfs_ioctl_compat.h> 90 #include <sys/zfs_context.h> 91 92 #include <sys/arc_impl.h> 93 #include <sys/dsl_pool.h> 94 95 #include <sys/vmmeter.h> 96 97 SYSCTL_DECL(_vfs_zfs); 98 SYSCTL_NODE(_vfs_zfs, OID_AUTO, arc, CTLFLAG_RW, 0, 99 "ZFS adaptive replacement cache"); 100 SYSCTL_NODE(_vfs_zfs, OID_AUTO, condense, CTLFLAG_RW, 0, "ZFS condense"); 101 SYSCTL_NODE(_vfs_zfs, OID_AUTO, dbuf, CTLFLAG_RW, 0, "ZFS disk buf cache"); 102 SYSCTL_NODE(_vfs_zfs, OID_AUTO, dbuf_cache, CTLFLAG_RW, 0, 103 "ZFS disk buf cache"); 104 SYSCTL_NODE(_vfs_zfs, OID_AUTO, deadman, CTLFLAG_RW, 0, "ZFS deadman"); 105 SYSCTL_NODE(_vfs_zfs, OID_AUTO, dedup, CTLFLAG_RW, 0, "ZFS dedup"); 106 SYSCTL_NODE(_vfs_zfs, OID_AUTO, l2arc, CTLFLAG_RW, 0, "ZFS l2arc"); 107 SYSCTL_NODE(_vfs_zfs, OID_AUTO, livelist, CTLFLAG_RW, 0, "ZFS livelist"); 108 SYSCTL_NODE(_vfs_zfs, OID_AUTO, lua, CTLFLAG_RW, 0, "ZFS lua"); 109 SYSCTL_NODE(_vfs_zfs, OID_AUTO, metaslab, CTLFLAG_RW, 0, "ZFS metaslab"); 110 SYSCTL_NODE(_vfs_zfs, OID_AUTO, mg, CTLFLAG_RW, 0, "ZFS metaslab group"); 111 SYSCTL_NODE(_vfs_zfs, OID_AUTO, multihost, CTLFLAG_RW, 0, 112 "ZFS multihost protection"); 113 SYSCTL_NODE(_vfs_zfs, OID_AUTO, prefetch, CTLFLAG_RW, 0, "ZFS prefetch"); 114 SYSCTL_NODE(_vfs_zfs, OID_AUTO, reconstruct, CTLFLAG_RW, 0, "ZFS reconstruct"); 115 SYSCTL_NODE(_vfs_zfs, OID_AUTO, recv, CTLFLAG_RW, 0, "ZFS receive"); 116 SYSCTL_NODE(_vfs_zfs, OID_AUTO, send, CTLFLAG_RW, 0, "ZFS send"); 117 SYSCTL_NODE(_vfs_zfs, OID_AUTO, spa, CTLFLAG_RW, 0, "ZFS space allocation"); 118 SYSCTL_NODE(_vfs_zfs, OID_AUTO, trim, CTLFLAG_RW, 0, "ZFS TRIM"); 119 SYSCTL_NODE(_vfs_zfs, OID_AUTO, txg, CTLFLAG_RW, 0, "ZFS transaction group"); 120 SYSCTL_NODE(_vfs_zfs, OID_AUTO, vdev, CTLFLAG_RW, 0, "ZFS VDEV"); 121 SYSCTL_NODE(_vfs_zfs, OID_AUTO, vnops, CTLFLAG_RW, 0, "ZFS VNOPS"); 122 SYSCTL_NODE(_vfs_zfs, OID_AUTO, zevent, CTLFLAG_RW, 0, "ZFS event"); 123 SYSCTL_NODE(_vfs_zfs, OID_AUTO, zil, CTLFLAG_RW, 0, "ZFS ZIL"); 124 SYSCTL_NODE(_vfs_zfs, OID_AUTO, zio, CTLFLAG_RW, 0, "ZFS ZIO"); 125 126 SYSCTL_NODE(_vfs_zfs_livelist, OID_AUTO, condense, CTLFLAG_RW, 0, 127 "ZFS livelist condense"); 128 SYSCTL_NODE(_vfs_zfs_vdev, OID_AUTO, cache, CTLFLAG_RW, 0, "ZFS VDEV Cache"); 129 SYSCTL_NODE(_vfs_zfs_vdev, OID_AUTO, file, CTLFLAG_RW, 0, "ZFS VDEV file"); 130 SYSCTL_NODE(_vfs_zfs_vdev, OID_AUTO, mirror, CTLFLAG_RD, 0, 131 "ZFS VDEV mirror"); 132 133 SYSCTL_DECL(_vfs_zfs_version); 134 SYSCTL_CONST_STRING(_vfs_zfs_version, OID_AUTO, module, CTLFLAG_RD, 135 (ZFS_META_VERSION "-" ZFS_META_RELEASE), "OpenZFS module version"); 136 137 /* arc.c */ 138 139 int 140 param_set_arc_u64(SYSCTL_HANDLER_ARGS) 141 { 142 int err; 143 144 err = sysctl_handle_64(oidp, arg1, 0, req); 145 if (err != 0 || req->newptr == NULL) 146 return (err); 147 148 arc_tuning_update(B_TRUE); 149 150 return (0); 151 } 152 153 int 154 param_set_arc_int(SYSCTL_HANDLER_ARGS) 155 { 156 int err; 157 158 err = sysctl_handle_int(oidp, arg1, 0, req); 159 if (err != 0 || req->newptr == NULL) 160 return (err); 161 162 arc_tuning_update(B_TRUE); 163 164 return (0); 165 } 166 167 int 168 param_set_arc_max(SYSCTL_HANDLER_ARGS) 169 { 170 unsigned long val; 171 int err; 172 173 val = zfs_arc_max; 174 err = sysctl_handle_64(oidp, &val, 0, req); 175 if (err != 0 || req->newptr == NULL) 176 return (SET_ERROR(err)); 177 178 if (val != 0 && (val < MIN_ARC_MAX || val <= arc_c_min || 179 val >= arc_all_memory())) 180 return (SET_ERROR(EINVAL)); 181 182 zfs_arc_max = val; 183 arc_tuning_update(B_TRUE); 184 185 /* Update the sysctl to the tuned value */ 186 if (val != 0) 187 zfs_arc_max = arc_c_max; 188 189 return (0); 190 } 191 192 /* BEGIN CSTYLED */ 193 SYSCTL_PROC(_vfs_zfs, OID_AUTO, arc_max, 194 CTLTYPE_ULONG | CTLFLAG_RWTUN | CTLFLAG_MPSAFE, 195 NULL, 0, param_set_arc_max, "LU", 196 "Maximum ARC size in bytes (LEGACY)"); 197 /* END CSTYLED */ 198 199 int 200 param_set_arc_min(SYSCTL_HANDLER_ARGS) 201 { 202 unsigned long val; 203 int err; 204 205 val = zfs_arc_min; 206 err = sysctl_handle_64(oidp, &val, 0, req); 207 if (err != 0 || req->newptr == NULL) 208 return (SET_ERROR(err)); 209 210 if (val != 0 && (val < 2ULL << SPA_MAXBLOCKSHIFT || val > arc_c_max)) 211 return (SET_ERROR(EINVAL)); 212 213 zfs_arc_min = val; 214 arc_tuning_update(B_TRUE); 215 216 /* Update the sysctl to the tuned value */ 217 if (val != 0) 218 zfs_arc_min = arc_c_min; 219 220 return (0); 221 } 222 223 /* BEGIN CSTYLED */ 224 SYSCTL_PROC(_vfs_zfs, OID_AUTO, arc_min, 225 CTLTYPE_ULONG | CTLFLAG_RWTUN | CTLFLAG_MPSAFE, 226 NULL, 0, param_set_arc_min, "LU", 227 "Minimum ARC size in bytes (LEGACY)"); 228 /* END CSTYLED */ 229 230 extern uint_t zfs_arc_free_target; 231 232 int 233 param_set_arc_free_target(SYSCTL_HANDLER_ARGS) 234 { 235 uint_t val; 236 int err; 237 238 val = zfs_arc_free_target; 239 err = sysctl_handle_int(oidp, &val, 0, req); 240 if (err != 0 || req->newptr == NULL) 241 return (err); 242 243 if (val < minfree) 244 return (EINVAL); 245 if (val > vm_cnt.v_page_count) 246 return (EINVAL); 247 248 zfs_arc_free_target = val; 249 250 return (0); 251 } 252 253 /* 254 * NOTE: This sysctl is CTLFLAG_RW not CTLFLAG_RWTUN due to its dependency on 255 * pagedaemon initialization. 256 */ 257 /* BEGIN CSTYLED */ 258 SYSCTL_PROC(_vfs_zfs, OID_AUTO, arc_free_target, 259 CTLTYPE_UINT | CTLFLAG_RW | CTLFLAG_MPSAFE, 260 NULL, 0, param_set_arc_free_target, "IU", 261 "Desired number of free pages below which ARC triggers reclaim" 262 " (LEGACY)"); 263 /* END CSTYLED */ 264 265 int 266 param_set_arc_no_grow_shift(SYSCTL_HANDLER_ARGS) 267 { 268 int err, val; 269 270 val = arc_no_grow_shift; 271 err = sysctl_handle_int(oidp, &val, 0, req); 272 if (err != 0 || req->newptr == NULL) 273 return (err); 274 275 if (val < 0 || val >= arc_shrink_shift) 276 return (EINVAL); 277 278 arc_no_grow_shift = val; 279 280 return (0); 281 } 282 283 /* BEGIN CSTYLED */ 284 SYSCTL_PROC(_vfs_zfs, OID_AUTO, arc_no_grow_shift, 285 CTLTYPE_INT | CTLFLAG_RWTUN | CTLFLAG_MPSAFE, 286 NULL, 0, param_set_arc_no_grow_shift, "I", 287 "log2(fraction of ARC which must be free to allow growing) (LEGACY)"); 288 /* END CSTYLED */ 289 290 extern uint64_t l2arc_write_max; 291 292 /* BEGIN CSTYLED */ 293 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_write_max, 294 CTLFLAG_RWTUN, &l2arc_write_max, 0, 295 "Max write bytes per interval (LEGACY)"); 296 /* END CSTYLED */ 297 298 extern uint64_t l2arc_write_boost; 299 300 /* BEGIN CSTYLED */ 301 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_write_boost, 302 CTLFLAG_RWTUN, &l2arc_write_boost, 0, 303 "Extra write bytes during device warmup (LEGACY)"); 304 /* END CSTYLED */ 305 306 extern uint64_t l2arc_headroom; 307 308 /* BEGIN CSTYLED */ 309 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_headroom, 310 CTLFLAG_RWTUN, &l2arc_headroom, 0, 311 "Number of max device writes to precache (LEGACY)"); 312 /* END CSTYLED */ 313 314 extern uint64_t l2arc_headroom_boost; 315 316 /* BEGIN CSTYLED */ 317 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_headroom_boost, 318 CTLFLAG_RWTUN, &l2arc_headroom_boost, 0, 319 "Compressed l2arc_headroom multiplier (LEGACY)"); 320 /* END CSTYLED */ 321 322 extern uint64_t l2arc_feed_secs; 323 324 /* BEGIN CSTYLED */ 325 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_feed_secs, 326 CTLFLAG_RWTUN, &l2arc_feed_secs, 0, 327 "Seconds between L2ARC writing (LEGACY)"); 328 /* END CSTYLED */ 329 330 extern uint64_t l2arc_feed_min_ms; 331 332 /* BEGIN CSTYLED */ 333 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_feed_min_ms, 334 CTLFLAG_RWTUN, &l2arc_feed_min_ms, 0, 335 "Min feed interval in milliseconds (LEGACY)"); 336 /* END CSTYLED */ 337 338 extern int l2arc_noprefetch; 339 340 /* BEGIN CSTYLED */ 341 SYSCTL_INT(_vfs_zfs, OID_AUTO, l2arc_noprefetch, 342 CTLFLAG_RWTUN, &l2arc_noprefetch, 0, 343 "Skip caching prefetched buffers (LEGACY)"); 344 /* END CSTYLED */ 345 346 extern int l2arc_feed_again; 347 348 /* BEGIN CSTYLED */ 349 SYSCTL_INT(_vfs_zfs, OID_AUTO, l2arc_feed_again, 350 CTLFLAG_RWTUN, &l2arc_feed_again, 0, 351 "Turbo L2ARC warmup (LEGACY)"); 352 /* END CSTYLED */ 353 354 extern int l2arc_norw; 355 356 /* BEGIN CSTYLED */ 357 SYSCTL_INT(_vfs_zfs, OID_AUTO, l2arc_norw, 358 CTLFLAG_RWTUN, &l2arc_norw, 0, 359 "No reads during writes (LEGACY)"); 360 /* END CSTYLED */ 361 362 extern arc_state_t ARC_anon; 363 364 /* BEGIN CSTYLED */ 365 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, anon_size, CTLFLAG_RD, 366 &ARC_anon.arcs_size.rc_count, 0, "size of anonymous state"); 367 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, anon_metadata_esize, CTLFLAG_RD, 368 &ARC_anon.arcs_esize[ARC_BUFC_METADATA].rc_count, 0, 369 "size of anonymous state"); 370 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, anon_data_esize, CTLFLAG_RD, 371 &ARC_anon.arcs_esize[ARC_BUFC_DATA].rc_count, 0, 372 "size of anonymous state"); 373 /* END CSTYLED */ 374 375 extern arc_state_t ARC_mru; 376 377 /* BEGIN CSTYLED */ 378 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mru_size, CTLFLAG_RD, 379 &ARC_mru.arcs_size.rc_count, 0, "size of mru state"); 380 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mru_metadata_esize, CTLFLAG_RD, 381 &ARC_mru.arcs_esize[ARC_BUFC_METADATA].rc_count, 0, 382 "size of metadata in mru state"); 383 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mru_data_esize, CTLFLAG_RD, 384 &ARC_mru.arcs_esize[ARC_BUFC_DATA].rc_count, 0, 385 "size of data in mru state"); 386 /* END CSTYLED */ 387 388 extern arc_state_t ARC_mru_ghost; 389 390 /* BEGIN CSTYLED */ 391 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mru_ghost_size, CTLFLAG_RD, 392 &ARC_mru_ghost.arcs_size.rc_count, 0, "size of mru ghost state"); 393 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mru_ghost_metadata_esize, CTLFLAG_RD, 394 &ARC_mru_ghost.arcs_esize[ARC_BUFC_METADATA].rc_count, 0, 395 "size of metadata in mru ghost state"); 396 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mru_ghost_data_esize, CTLFLAG_RD, 397 &ARC_mru_ghost.arcs_esize[ARC_BUFC_DATA].rc_count, 0, 398 "size of data in mru ghost state"); 399 /* END CSTYLED */ 400 401 extern arc_state_t ARC_mfu; 402 403 /* BEGIN CSTYLED */ 404 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mfu_size, CTLFLAG_RD, 405 &ARC_mfu.arcs_size.rc_count, 0, "size of mfu state"); 406 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mfu_metadata_esize, CTLFLAG_RD, 407 &ARC_mfu.arcs_esize[ARC_BUFC_METADATA].rc_count, 0, 408 "size of metadata in mfu state"); 409 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mfu_data_esize, CTLFLAG_RD, 410 &ARC_mfu.arcs_esize[ARC_BUFC_DATA].rc_count, 0, 411 "size of data in mfu state"); 412 /* END CSTYLED */ 413 414 extern arc_state_t ARC_mfu_ghost; 415 416 /* BEGIN CSTYLED */ 417 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mfu_ghost_size, CTLFLAG_RD, 418 &ARC_mfu_ghost.arcs_size.rc_count, 0, "size of mfu ghost state"); 419 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mfu_ghost_metadata_esize, CTLFLAG_RD, 420 &ARC_mfu_ghost.arcs_esize[ARC_BUFC_METADATA].rc_count, 0, 421 "size of metadata in mfu ghost state"); 422 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mfu_ghost_data_esize, CTLFLAG_RD, 423 &ARC_mfu_ghost.arcs_esize[ARC_BUFC_DATA].rc_count, 0, 424 "size of data in mfu ghost state"); 425 /* END CSTYLED */ 426 427 extern arc_state_t ARC_l2c_only; 428 429 /* BEGIN CSTYLED */ 430 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2c_only_size, CTLFLAG_RD, 431 &ARC_l2c_only.arcs_size.rc_count, 0, "size of mru state"); 432 /* END CSTYLED */ 433 434 /* dbuf.c */ 435 436 /* dmu.c */ 437 438 /* dmu_zfetch.c */ 439 440 SYSCTL_NODE(_vfs_zfs, OID_AUTO, zfetch, CTLFLAG_RW, 0, "ZFS ZFETCH (LEGACY)"); 441 442 extern uint32_t zfetch_max_distance; 443 444 /* BEGIN CSTYLED */ 445 SYSCTL_UINT(_vfs_zfs_zfetch, OID_AUTO, max_distance, 446 CTLFLAG_RWTUN, &zfetch_max_distance, 0, 447 "Max bytes to prefetch per stream (LEGACY)"); 448 /* END CSTYLED */ 449 450 extern uint32_t zfetch_max_idistance; 451 452 /* BEGIN CSTYLED */ 453 SYSCTL_UINT(_vfs_zfs_zfetch, OID_AUTO, max_idistance, 454 CTLFLAG_RWTUN, &zfetch_max_idistance, 0, 455 "Max bytes to prefetch indirects for per stream (LEGACY)"); 456 /* END CSTYLED */ 457 458 /* dsl_pool.c */ 459 460 /* dnode.c */ 461 462 extern int zfs_default_bs; 463 464 /* BEGIN CSTYLED */ 465 SYSCTL_INT(_vfs_zfs, OID_AUTO, default_bs, CTLFLAG_RWTUN, 466 &zfs_default_bs, 0, "Default dnode block shift"); 467 /* END CSTYLED */ 468 469 extern int zfs_default_ibs; 470 471 /* BEGIN CSTYLED */ 472 SYSCTL_INT(_vfs_zfs, OID_AUTO, default_ibs, CTLFLAG_RWTUN, 473 &zfs_default_ibs, 0, "Default dnode indirect block shift"); 474 /* END CSTYLED */ 475 476 /* dsl_scan.c */ 477 478 /* metaslab.c */ 479 480 /* 481 * In pools where the log space map feature is not enabled we touch 482 * multiple metaslabs (and their respective space maps) with each 483 * transaction group. Thus, we benefit from having a small space map 484 * block size since it allows us to issue more I/O operations scattered 485 * around the disk. So a sane default for the space map block size 486 * is 8~16K. 487 */ 488 extern int zfs_metaslab_sm_blksz_no_log; 489 490 /* BEGIN CSTYLED */ 491 SYSCTL_INT(_vfs_zfs_metaslab, OID_AUTO, sm_blksz_no_log, 492 CTLFLAG_RDTUN, &zfs_metaslab_sm_blksz_no_log, 0, 493 "Block size for space map in pools with log space map disabled. " 494 "Power of 2 greater than 4096."); 495 /* END CSTYLED */ 496 497 /* 498 * When the log space map feature is enabled, we accumulate a lot of 499 * changes per metaslab that are flushed once in a while so we benefit 500 * from a bigger block size like 128K for the metaslab space maps. 501 */ 502 extern int zfs_metaslab_sm_blksz_with_log; 503 504 /* BEGIN CSTYLED */ 505 SYSCTL_INT(_vfs_zfs_metaslab, OID_AUTO, sm_blksz_with_log, 506 CTLFLAG_RDTUN, &zfs_metaslab_sm_blksz_with_log, 0, 507 "Block size for space map in pools with log space map enabled. " 508 "Power of 2 greater than 4096."); 509 /* END CSTYLED */ 510 511 /* 512 * The in-core space map representation is more compact than its on-disk form. 513 * The zfs_condense_pct determines how much more compact the in-core 514 * space map representation must be before we compact it on-disk. 515 * Values should be greater than or equal to 100. 516 */ 517 extern uint_t zfs_condense_pct; 518 519 /* BEGIN CSTYLED */ 520 SYSCTL_UINT(_vfs_zfs, OID_AUTO, condense_pct, 521 CTLFLAG_RWTUN, &zfs_condense_pct, 0, 522 "Condense on-disk spacemap when it is more than this many percents" 523 " of in-memory counterpart"); 524 /* END CSTYLED */ 525 526 extern uint_t zfs_remove_max_segment; 527 528 /* BEGIN CSTYLED */ 529 SYSCTL_UINT(_vfs_zfs, OID_AUTO, remove_max_segment, 530 CTLFLAG_RWTUN, &zfs_remove_max_segment, 0, 531 "Largest contiguous segment ZFS will attempt to allocate when removing" 532 " a device"); 533 /* END CSTYLED */ 534 535 extern int zfs_removal_suspend_progress; 536 537 /* BEGIN CSTYLED */ 538 SYSCTL_INT(_vfs_zfs, OID_AUTO, removal_suspend_progress, 539 CTLFLAG_RWTUN, &zfs_removal_suspend_progress, 0, 540 "Ensures certain actions can happen while in the middle of a removal"); 541 /* END CSTYLED */ 542 543 /* 544 * Minimum size which forces the dynamic allocator to change 545 * it's allocation strategy. Once the space map cannot satisfy 546 * an allocation of this size then it switches to using more 547 * aggressive strategy (i.e search by size rather than offset). 548 */ 549 extern uint64_t metaslab_df_alloc_threshold; 550 551 /* BEGIN CSTYLED */ 552 SYSCTL_QUAD(_vfs_zfs_metaslab, OID_AUTO, df_alloc_threshold, 553 CTLFLAG_RWTUN, &metaslab_df_alloc_threshold, 0, 554 "Minimum size which forces the dynamic allocator to change its" 555 " allocation strategy"); 556 /* END CSTYLED */ 557 558 /* 559 * The minimum free space, in percent, which must be available 560 * in a space map to continue allocations in a first-fit fashion. 561 * Once the space map's free space drops below this level we dynamically 562 * switch to using best-fit allocations. 563 */ 564 extern uint_t metaslab_df_free_pct; 565 566 /* BEGIN CSTYLED */ 567 SYSCTL_UINT(_vfs_zfs_metaslab, OID_AUTO, df_free_pct, 568 CTLFLAG_RWTUN, &metaslab_df_free_pct, 0, 569 "The minimum free space, in percent, which must be available in a" 570 " space map to continue allocations in a first-fit fashion"); 571 /* END CSTYLED */ 572 573 /* 574 * Percentage of all cpus that can be used by the metaslab taskq. 575 */ 576 extern int metaslab_load_pct; 577 578 /* BEGIN CSTYLED */ 579 SYSCTL_INT(_vfs_zfs_metaslab, OID_AUTO, load_pct, 580 CTLFLAG_RWTUN, &metaslab_load_pct, 0, 581 "Percentage of cpus that can be used by the metaslab taskq"); 582 /* END CSTYLED */ 583 584 /* 585 * Max number of metaslabs per group to preload. 586 */ 587 extern uint_t metaslab_preload_limit; 588 589 /* BEGIN CSTYLED */ 590 SYSCTL_UINT(_vfs_zfs_metaslab, OID_AUTO, preload_limit, 591 CTLFLAG_RWTUN, &metaslab_preload_limit, 0, 592 "Max number of metaslabs per group to preload"); 593 /* END CSTYLED */ 594 595 /* mmp.c */ 596 597 int 598 param_set_multihost_interval(SYSCTL_HANDLER_ARGS) 599 { 600 int err; 601 602 err = sysctl_handle_64(oidp, &zfs_multihost_interval, 0, req); 603 if (err != 0 || req->newptr == NULL) 604 return (err); 605 606 if (spa_mode_global != SPA_MODE_UNINIT) 607 mmp_signal_all_threads(); 608 609 return (0); 610 } 611 612 /* spa.c */ 613 614 extern int zfs_ccw_retry_interval; 615 616 /* BEGIN CSTYLED */ 617 SYSCTL_INT(_vfs_zfs, OID_AUTO, ccw_retry_interval, 618 CTLFLAG_RWTUN, &zfs_ccw_retry_interval, 0, 619 "Configuration cache file write, retry after failure, interval" 620 " (seconds)"); 621 /* END CSTYLED */ 622 623 extern uint64_t zfs_max_missing_tvds_cachefile; 624 625 /* BEGIN CSTYLED */ 626 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, max_missing_tvds_cachefile, 627 CTLFLAG_RWTUN, &zfs_max_missing_tvds_cachefile, 0, 628 "Allow importing pools with missing top-level vdevs in cache file"); 629 /* END CSTYLED */ 630 631 extern uint64_t zfs_max_missing_tvds_scan; 632 633 /* BEGIN CSTYLED */ 634 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, max_missing_tvds_scan, 635 CTLFLAG_RWTUN, &zfs_max_missing_tvds_scan, 0, 636 "Allow importing pools with missing top-level vdevs during scan"); 637 /* END CSTYLED */ 638 639 /* spa_misc.c */ 640 641 extern int zfs_flags; 642 643 static int 644 sysctl_vfs_zfs_debug_flags(SYSCTL_HANDLER_ARGS) 645 { 646 int err, val; 647 648 val = zfs_flags; 649 err = sysctl_handle_int(oidp, &val, 0, req); 650 if (err != 0 || req->newptr == NULL) 651 return (err); 652 653 /* 654 * ZFS_DEBUG_MODIFY must be enabled prior to boot so all 655 * arc buffers in the system have the necessary additional 656 * checksum data. However, it is safe to disable at any 657 * time. 658 */ 659 if (!(zfs_flags & ZFS_DEBUG_MODIFY)) 660 val &= ~ZFS_DEBUG_MODIFY; 661 zfs_flags = val; 662 663 return (0); 664 } 665 666 /* BEGIN CSTYLED */ 667 SYSCTL_PROC(_vfs_zfs, OID_AUTO, debugflags, 668 CTLTYPE_UINT | CTLFLAG_MPSAFE | CTLFLAG_RWTUN, NULL, 0, 669 sysctl_vfs_zfs_debug_flags, "IU", "Debug flags for ZFS testing."); 670 /* END CSTYLED */ 671 672 int 673 param_set_deadman_synctime(SYSCTL_HANDLER_ARGS) 674 { 675 unsigned long val; 676 int err; 677 678 val = zfs_deadman_synctime_ms; 679 err = sysctl_handle_64(oidp, &val, 0, req); 680 if (err != 0 || req->newptr == NULL) 681 return (err); 682 zfs_deadman_synctime_ms = val; 683 684 spa_set_deadman_synctime(MSEC2NSEC(zfs_deadman_synctime_ms)); 685 686 return (0); 687 } 688 689 int 690 param_set_deadman_ziotime(SYSCTL_HANDLER_ARGS) 691 { 692 unsigned long val; 693 int err; 694 695 val = zfs_deadman_ziotime_ms; 696 err = sysctl_handle_64(oidp, &val, 0, req); 697 if (err != 0 || req->newptr == NULL) 698 return (err); 699 zfs_deadman_ziotime_ms = val; 700 701 spa_set_deadman_ziotime(MSEC2NSEC(zfs_deadman_synctime_ms)); 702 703 return (0); 704 } 705 706 int 707 param_set_deadman_failmode(SYSCTL_HANDLER_ARGS) 708 { 709 char buf[16]; 710 int rc; 711 712 if (req->newptr == NULL) 713 strlcpy(buf, zfs_deadman_failmode, sizeof (buf)); 714 715 rc = sysctl_handle_string(oidp, buf, sizeof (buf), req); 716 if (rc || req->newptr == NULL) 717 return (rc); 718 if (strcmp(buf, zfs_deadman_failmode) == 0) 719 return (0); 720 if (strcmp(buf, "wait") == 0) 721 zfs_deadman_failmode = "wait"; 722 if (strcmp(buf, "continue") == 0) 723 zfs_deadman_failmode = "continue"; 724 if (strcmp(buf, "panic") == 0) 725 zfs_deadman_failmode = "panic"; 726 727 return (-param_set_deadman_failmode_common(buf)); 728 } 729 730 int 731 param_set_slop_shift(SYSCTL_HANDLER_ARGS) 732 { 733 int val; 734 int err; 735 736 val = spa_slop_shift; 737 err = sysctl_handle_int(oidp, &val, 0, req); 738 if (err != 0 || req->newptr == NULL) 739 return (err); 740 741 if (val < 1 || val > 31) 742 return (EINVAL); 743 744 spa_slop_shift = val; 745 746 return (0); 747 } 748 749 /* spacemap.c */ 750 751 extern int space_map_ibs; 752 753 /* BEGIN CSTYLED */ 754 SYSCTL_INT(_vfs_zfs, OID_AUTO, space_map_ibs, CTLFLAG_RWTUN, 755 &space_map_ibs, 0, "Space map indirect block shift"); 756 /* END CSTYLED */ 757 758 759 /* vdev.c */ 760 761 int 762 param_set_min_auto_ashift(SYSCTL_HANDLER_ARGS) 763 { 764 int val; 765 int err; 766 767 val = zfs_vdev_min_auto_ashift; 768 err = sysctl_handle_int(oidp, &val, 0, req); 769 if (err != 0 || req->newptr == NULL) 770 return (SET_ERROR(err)); 771 772 if (val < ASHIFT_MIN || val > zfs_vdev_max_auto_ashift) 773 return (SET_ERROR(EINVAL)); 774 775 zfs_vdev_min_auto_ashift = val; 776 777 return (0); 778 } 779 780 /* BEGIN CSTYLED */ 781 SYSCTL_PROC(_vfs_zfs, OID_AUTO, min_auto_ashift, 782 CTLTYPE_UINT | CTLFLAG_RWTUN | CTLFLAG_MPSAFE, 783 &zfs_vdev_min_auto_ashift, sizeof (zfs_vdev_min_auto_ashift), 784 param_set_min_auto_ashift, "IU", 785 "Min ashift used when creating new top-level vdev. (LEGACY)"); 786 /* END CSTYLED */ 787 788 int 789 param_set_max_auto_ashift(SYSCTL_HANDLER_ARGS) 790 { 791 int val; 792 int err; 793 794 val = zfs_vdev_max_auto_ashift; 795 err = sysctl_handle_int(oidp, &val, 0, req); 796 if (err != 0 || req->newptr == NULL) 797 return (SET_ERROR(err)); 798 799 if (val > ASHIFT_MAX || val < zfs_vdev_min_auto_ashift) 800 return (SET_ERROR(EINVAL)); 801 802 zfs_vdev_max_auto_ashift = val; 803 804 return (0); 805 } 806 807 /* BEGIN CSTYLED */ 808 SYSCTL_PROC(_vfs_zfs, OID_AUTO, max_auto_ashift, 809 CTLTYPE_UINT | CTLFLAG_RWTUN | CTLFLAG_MPSAFE, 810 &zfs_vdev_max_auto_ashift, sizeof (zfs_vdev_max_auto_ashift), 811 param_set_max_auto_ashift, "IU", 812 "Max ashift used when optimizing for logical -> physical sector size on" 813 " new top-level vdevs. (LEGACY)"); 814 /* END CSTYLED */ 815 816 /* 817 * Since the DTL space map of a vdev is not expected to have a lot of 818 * entries, we default its block size to 4K. 819 */ 820 extern int zfs_vdev_dtl_sm_blksz; 821 822 /* BEGIN CSTYLED */ 823 SYSCTL_INT(_vfs_zfs, OID_AUTO, dtl_sm_blksz, 824 CTLFLAG_RDTUN, &zfs_vdev_dtl_sm_blksz, 0, 825 "Block size for DTL space map. Power of 2 greater than 4096."); 826 /* END CSTYLED */ 827 828 /* 829 * vdev-wide space maps that have lots of entries written to them at 830 * the end of each transaction can benefit from a higher I/O bandwidth 831 * (e.g. vdev_obsolete_sm), thus we default their block size to 128K. 832 */ 833 extern int zfs_vdev_standard_sm_blksz; 834 835 /* BEGIN CSTYLED */ 836 SYSCTL_INT(_vfs_zfs, OID_AUTO, standard_sm_blksz, 837 CTLFLAG_RDTUN, &zfs_vdev_standard_sm_blksz, 0, 838 "Block size for standard space map. Power of 2 greater than 4096."); 839 /* END CSTYLED */ 840 841 extern int vdev_validate_skip; 842 843 /* BEGIN CSTYLED */ 844 SYSCTL_INT(_vfs_zfs, OID_AUTO, validate_skip, 845 CTLFLAG_RDTUN, &vdev_validate_skip, 0, 846 "Enable to bypass vdev_validate()."); 847 /* END CSTYLED */ 848 849 /* vdev_cache.c */ 850 851 /* vdev_mirror.c */ 852 853 /* vdev_queue.c */ 854 855 extern uint_t zfs_vdev_max_active; 856 857 /* BEGIN CSTYLED */ 858 SYSCTL_UINT(_vfs_zfs, OID_AUTO, top_maxinflight, 859 CTLFLAG_RWTUN, &zfs_vdev_max_active, 0, 860 "The maximum number of I/Os of all types active for each device." 861 " (LEGACY)"); 862 /* END CSTYLED */ 863 864 extern uint_t zfs_vdev_def_queue_depth; 865 866 /* BEGIN CSTYLED */ 867 SYSCTL_UINT(_vfs_zfs_vdev, OID_AUTO, def_queue_depth, 868 CTLFLAG_RWTUN, &zfs_vdev_def_queue_depth, 0, 869 "Default queue depth for each allocator"); 870 /* END CSTYLED */ 871 872 /* zio.c */ 873 874 /* BEGIN CSTYLED */ 875 SYSCTL_INT(_vfs_zfs_zio, OID_AUTO, exclude_metadata, 876 CTLFLAG_RDTUN, &zio_exclude_metadata, 0, 877 "Exclude metadata buffers from dumps as well"); 878 /* END CSTYLED */ 879