1 // SPDX-License-Identifier: BSD-2-Clause
2 /*
3 * Copyright (c) 2020 iXsystems, Inc.
4 * All rights reserved.
5 *
6 * Redistribution and use in source and binary forms, with or without
7 * modification, are permitted provided that the following conditions
8 * are met:
9 * 1. Redistributions of source code must retain the above copyright
10 * notice, this list of conditions and the following disclaimer.
11 * 2. Redistributions in binary form must reproduce the above copyright
12 * notice, this list of conditions and the following disclaimer in the
13 * documentation and/or other materials provided with the distribution.
14 *
15 * THIS SOFTWARE IS PROVIDED BY THE AUTHORS AND CONTRIBUTORS ``AS IS'' AND
16 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
17 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
18 * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHORS OR CONTRIBUTORS BE LIABLE
19 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
20 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
21 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
22 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
23 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
24 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
25 * SUCH DAMAGE.
26 *
27 */
28
29 #include <sys/types.h>
30 #include <sys/param.h>
31 #include <sys/systm.h>
32 #include <sys/conf.h>
33 #include <sys/kernel.h>
34 #include <sys/lock.h>
35 #include <sys/malloc.h>
36 #include <sys/mutex.h>
37 #include <sys/proc.h>
38 #include <sys/errno.h>
39 #include <sys/uio.h>
40 #include <sys/buf.h>
41 #include <sys/file.h>
42 #include <sys/kmem.h>
43 #include <sys/conf.h>
44 #include <sys/cmn_err.h>
45 #include <sys/stat.h>
46 #include <sys/zfs_ioctl.h>
47 #include <sys/zfs_vfsops.h>
48 #include <sys/zfs_znode.h>
49 #include <sys/zap.h>
50 #include <sys/spa.h>
51 #include <sys/spa_impl.h>
52 #include <sys/vdev.h>
53 #include <sys/vdev_impl.h>
54 #include <sys/arc_os.h>
55 #include <sys/dmu.h>
56 #include <sys/dsl_dir.h>
57 #include <sys/dsl_dataset.h>
58 #include <sys/dsl_prop.h>
59 #include <sys/dsl_deleg.h>
60 #include <sys/dmu_objset.h>
61 #include <sys/dmu_impl.h>
62 #include <sys/dmu_tx.h>
63 #include <sys/sunddi.h>
64 #include <sys/policy.h>
65 #include <sys/zone.h>
66 #include <sys/nvpair.h>
67 #include <sys/mount.h>
68 #include <sys/taskqueue.h>
69 #include <sys/sdt.h>
70 #include <sys/fs/zfs.h>
71 #include <sys/zfs_ctldir.h>
72 #include <sys/zfs_dir.h>
73 #include <sys/zfs_onexit.h>
74 #include <sys/zvol.h>
75 #include <sys/dsl_scan.h>
76 #include <sys/dmu_objset.h>
77 #include <sys/dmu_send.h>
78 #include <sys/dsl_destroy.h>
79 #include <sys/dsl_bookmark.h>
80 #include <sys/dsl_userhold.h>
81 #include <sys/zfeature.h>
82 #include <sys/zcp.h>
83 #include <sys/zio_checksum.h>
84 #include <sys/vdev_removal.h>
85 #include <sys/dsl_crypt.h>
86
87 #include <sys/zfs_ioctl_compat.h>
88 #include <sys/zfs_context.h>
89
90 #include <sys/arc_impl.h>
91 #include <sys/dsl_pool.h>
92
93 #include <sys/vmmeter.h>
94
95 SYSCTL_DECL(_vfs_zfs);
96 SYSCTL_NODE(_vfs_zfs, OID_AUTO, arc, CTLFLAG_RW, 0,
97 "ZFS adaptive replacement cache");
98 SYSCTL_NODE(_vfs_zfs, OID_AUTO, brt, CTLFLAG_RW, 0,
99 "ZFS Block Reference Table");
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, file, CTLFLAG_RW, 0, "ZFS VDEV file");
129 SYSCTL_NODE(_vfs_zfs_vdev, OID_AUTO, mirror, CTLFLAG_RD, 0,
130 "ZFS VDEV mirror");
131
132 SYSCTL_DECL(_vfs_zfs_version);
133 SYSCTL_CONST_STRING(_vfs_zfs_version, OID_AUTO, module, CTLFLAG_RD,
134 (ZFS_META_VERSION "-" ZFS_META_RELEASE), "OpenZFS module version");
135
136 /* arc.c */
137
138 int
param_set_arc_u64(SYSCTL_HANDLER_ARGS)139 param_set_arc_u64(SYSCTL_HANDLER_ARGS)
140 {
141 int err;
142
143 err = sysctl_handle_64(oidp, arg1, 0, req);
144 if (err != 0 || req->newptr == NULL)
145 return (err);
146
147 arc_tuning_update(B_TRUE);
148
149 return (0);
150 }
151
152 int
param_set_arc_int(SYSCTL_HANDLER_ARGS)153 param_set_arc_int(SYSCTL_HANDLER_ARGS)
154 {
155 int err;
156
157 err = sysctl_handle_int(oidp, arg1, 0, req);
158 if (err != 0 || req->newptr == NULL)
159 return (err);
160
161 arc_tuning_update(B_TRUE);
162
163 return (0);
164 }
165
166 int
param_set_arc_max(SYSCTL_HANDLER_ARGS)167 param_set_arc_max(SYSCTL_HANDLER_ARGS)
168 {
169 unsigned long val;
170 int err;
171
172 val = zfs_arc_max;
173 err = sysctl_handle_64(oidp, &val, 0, req);
174 if (err != 0 || req->newptr == NULL)
175 return (SET_ERROR(err));
176
177 if (val != 0 && (val < MIN_ARC_MAX || val <= arc_c_min ||
178 val >= arc_all_memory()))
179 return (SET_ERROR(EINVAL));
180
181 zfs_arc_max = val;
182 arc_tuning_update(B_TRUE);
183
184 /* Update the sysctl to the tuned value */
185 if (val != 0)
186 zfs_arc_max = arc_c_max;
187
188 return (0);
189 }
190
191 SYSCTL_PROC(_vfs_zfs, OID_AUTO, arc_max,
192 CTLTYPE_ULONG | CTLFLAG_RWTUN | CTLFLAG_MPSAFE,
193 NULL, 0, param_set_arc_max, "LU",
194 "Maximum ARC size in bytes (LEGACY)");
195
196 int
param_set_arc_min(SYSCTL_HANDLER_ARGS)197 param_set_arc_min(SYSCTL_HANDLER_ARGS)
198 {
199 unsigned long val;
200 int err;
201
202 val = zfs_arc_min;
203 err = sysctl_handle_64(oidp, &val, 0, req);
204 if (err != 0 || req->newptr == NULL)
205 return (SET_ERROR(err));
206
207 if (val != 0 && (val < 2ULL << SPA_MAXBLOCKSHIFT || val > arc_c_max))
208 return (SET_ERROR(EINVAL));
209
210 zfs_arc_min = val;
211 arc_tuning_update(B_TRUE);
212
213 /* Update the sysctl to the tuned value */
214 if (val != 0)
215 zfs_arc_min = arc_c_min;
216
217 return (0);
218 }
219
220 SYSCTL_PROC(_vfs_zfs, OID_AUTO, arc_min,
221 CTLTYPE_ULONG | CTLFLAG_RWTUN | CTLFLAG_MPSAFE,
222 NULL, 0, param_set_arc_min, "LU",
223 "Minimum ARC size in bytes (LEGACY)");
224
225 extern uint_t zfs_arc_free_target;
226
227 int
param_set_arc_free_target(SYSCTL_HANDLER_ARGS)228 param_set_arc_free_target(SYSCTL_HANDLER_ARGS)
229 {
230 uint_t val;
231 int err;
232
233 val = zfs_arc_free_target;
234 err = sysctl_handle_int(oidp, &val, 0, req);
235 if (err != 0 || req->newptr == NULL)
236 return (err);
237
238 if (val < minfree)
239 return (EINVAL);
240 if (val > vm_cnt.v_page_count)
241 return (EINVAL);
242
243 zfs_arc_free_target = val;
244
245 return (0);
246 }
247
248 /*
249 * NOTE: This sysctl is CTLFLAG_RW not CTLFLAG_RWTUN due to its dependency on
250 * pagedaemon initialization.
251 */
252 SYSCTL_PROC(_vfs_zfs, OID_AUTO, arc_free_target,
253 CTLTYPE_UINT | CTLFLAG_RW | CTLFLAG_MPSAFE,
254 NULL, 0, param_set_arc_free_target, "IU",
255 "Desired number of free pages below which ARC triggers reclaim"
256 " (LEGACY)");
257
258 int
param_set_arc_no_grow_shift(SYSCTL_HANDLER_ARGS)259 param_set_arc_no_grow_shift(SYSCTL_HANDLER_ARGS)
260 {
261 int err, val;
262
263 val = arc_no_grow_shift;
264 err = sysctl_handle_int(oidp, &val, 0, req);
265 if (err != 0 || req->newptr == NULL)
266 return (err);
267
268 if (val < 0 || val >= arc_shrink_shift)
269 return (EINVAL);
270
271 arc_no_grow_shift = val;
272
273 return (0);
274 }
275
276 SYSCTL_PROC(_vfs_zfs, OID_AUTO, arc_no_grow_shift,
277 CTLTYPE_INT | CTLFLAG_RWTUN | CTLFLAG_MPSAFE,
278 NULL, 0, param_set_arc_no_grow_shift, "I",
279 "log2(fraction of ARC which must be free to allow growing) (LEGACY)");
280
281 extern uint64_t l2arc_write_max;
282
283 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_write_max,
284 CTLFLAG_RWTUN, &l2arc_write_max, 0,
285 "Max write bytes per interval (LEGACY)");
286
287 extern uint64_t l2arc_write_boost;
288
289 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_write_boost,
290 CTLFLAG_RWTUN, &l2arc_write_boost, 0,
291 "Extra write bytes during device warmup (LEGACY)");
292
293 extern uint64_t l2arc_headroom;
294
295 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_headroom,
296 CTLFLAG_RWTUN, &l2arc_headroom, 0,
297 "Number of max device writes to precache (LEGACY)");
298
299 extern uint64_t l2arc_headroom_boost;
300
301 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_headroom_boost,
302 CTLFLAG_RWTUN, &l2arc_headroom_boost, 0,
303 "Compressed l2arc_headroom multiplier (LEGACY)");
304
305 extern uint64_t l2arc_feed_secs;
306
307 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_feed_secs,
308 CTLFLAG_RWTUN, &l2arc_feed_secs, 0,
309 "Seconds between L2ARC writing (LEGACY)");
310
311 extern uint64_t l2arc_feed_min_ms;
312
313 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_feed_min_ms,
314 CTLFLAG_RWTUN, &l2arc_feed_min_ms, 0,
315 "Min feed interval in milliseconds (LEGACY)");
316
317 extern int l2arc_noprefetch;
318
319 SYSCTL_INT(_vfs_zfs, OID_AUTO, l2arc_noprefetch,
320 CTLFLAG_RWTUN, &l2arc_noprefetch, 0,
321 "Skip caching prefetched buffers (LEGACY)");
322
323 extern int l2arc_feed_again;
324
325 SYSCTL_INT(_vfs_zfs, OID_AUTO, l2arc_feed_again,
326 CTLFLAG_RWTUN, &l2arc_feed_again, 0,
327 "Turbo L2ARC warmup (LEGACY)");
328
329 extern int l2arc_norw;
330
331 SYSCTL_INT(_vfs_zfs, OID_AUTO, l2arc_norw,
332 CTLFLAG_RWTUN, &l2arc_norw, 0,
333 "No reads during writes (LEGACY)");
334
335 static int
param_get_arc_state_size(SYSCTL_HANDLER_ARGS)336 param_get_arc_state_size(SYSCTL_HANDLER_ARGS)
337 {
338 arc_state_t *state = (arc_state_t *)arg1;
339 int64_t val;
340
341 val = zfs_refcount_count(&state->arcs_size[ARC_BUFC_DATA]) +
342 zfs_refcount_count(&state->arcs_size[ARC_BUFC_METADATA]);
343 return (sysctl_handle_64(oidp, &val, 0, req));
344 }
345
346 extern arc_state_t ARC_anon;
347
348 SYSCTL_PROC(_vfs_zfs, OID_AUTO, anon_size,
349 CTLTYPE_S64 | CTLFLAG_RD | CTLFLAG_MPSAFE,
350 &ARC_anon, 0, param_get_arc_state_size, "Q",
351 "size of anonymous state");
352 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, anon_metadata_esize, CTLFLAG_RD,
353 &ARC_anon.arcs_esize[ARC_BUFC_METADATA].rc_count, 0,
354 "size of evictable metadata in anonymous state");
355 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, anon_data_esize, CTLFLAG_RD,
356 &ARC_anon.arcs_esize[ARC_BUFC_DATA].rc_count, 0,
357 "size of evictable data in anonymous state");
358
359 extern arc_state_t ARC_mru;
360
361 SYSCTL_PROC(_vfs_zfs, OID_AUTO, mru_size,
362 CTLTYPE_S64 | CTLFLAG_RD | CTLFLAG_MPSAFE,
363 &ARC_mru, 0, param_get_arc_state_size, "Q",
364 "size of mru state");
365 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mru_metadata_esize, CTLFLAG_RD,
366 &ARC_mru.arcs_esize[ARC_BUFC_METADATA].rc_count, 0,
367 "size of evictable metadata in mru state");
368 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mru_data_esize, CTLFLAG_RD,
369 &ARC_mru.arcs_esize[ARC_BUFC_DATA].rc_count, 0,
370 "size of evictable data in mru state");
371
372 extern arc_state_t ARC_mru_ghost;
373
374 SYSCTL_PROC(_vfs_zfs, OID_AUTO, mru_ghost_size,
375 CTLTYPE_S64 | CTLFLAG_RD | CTLFLAG_MPSAFE,
376 &ARC_mru_ghost, 0, param_get_arc_state_size, "Q",
377 "size of mru ghost state");
378 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mru_ghost_metadata_esize, CTLFLAG_RD,
379 &ARC_mru_ghost.arcs_esize[ARC_BUFC_METADATA].rc_count, 0,
380 "size of evictable metadata in mru ghost state");
381 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mru_ghost_data_esize, CTLFLAG_RD,
382 &ARC_mru_ghost.arcs_esize[ARC_BUFC_DATA].rc_count, 0,
383 "size of evictable data in mru ghost state");
384
385 extern arc_state_t ARC_mfu;
386
387 SYSCTL_PROC(_vfs_zfs, OID_AUTO, mfu_size,
388 CTLTYPE_S64 | CTLFLAG_RD | CTLFLAG_MPSAFE,
389 &ARC_mfu, 0, param_get_arc_state_size, "Q",
390 "size of mfu state");
391 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mfu_metadata_esize, CTLFLAG_RD,
392 &ARC_mfu.arcs_esize[ARC_BUFC_METADATA].rc_count, 0,
393 "size of evictable metadata in mfu state");
394 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mfu_data_esize, CTLFLAG_RD,
395 &ARC_mfu.arcs_esize[ARC_BUFC_DATA].rc_count, 0,
396 "size of evictable data in mfu state");
397
398 extern arc_state_t ARC_mfu_ghost;
399
400 SYSCTL_PROC(_vfs_zfs, OID_AUTO, mfu_ghost_size,
401 CTLTYPE_S64 | CTLFLAG_RD | CTLFLAG_MPSAFE,
402 &ARC_mfu_ghost, 0, param_get_arc_state_size, "Q",
403 "size of mfu ghost state");
404 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mfu_ghost_metadata_esize, CTLFLAG_RD,
405 &ARC_mfu_ghost.arcs_esize[ARC_BUFC_METADATA].rc_count, 0,
406 "size of evictable metadata in mfu ghost state");
407 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mfu_ghost_data_esize, CTLFLAG_RD,
408 &ARC_mfu_ghost.arcs_esize[ARC_BUFC_DATA].rc_count, 0,
409 "size of evictable data in mfu ghost state");
410
411 extern arc_state_t ARC_uncached;
412
413 SYSCTL_PROC(_vfs_zfs, OID_AUTO, uncached_size,
414 CTLTYPE_S64 | CTLFLAG_RD | CTLFLAG_MPSAFE,
415 &ARC_uncached, 0, param_get_arc_state_size, "Q",
416 "size of uncached state");
417 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, uncached_metadata_esize, CTLFLAG_RD,
418 &ARC_uncached.arcs_esize[ARC_BUFC_METADATA].rc_count, 0,
419 "size of evictable metadata in uncached state");
420 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, uncached_data_esize, CTLFLAG_RD,
421 &ARC_uncached.arcs_esize[ARC_BUFC_DATA].rc_count, 0,
422 "size of evictable data in uncached state");
423
424 extern arc_state_t ARC_l2c_only;
425
426 SYSCTL_PROC(_vfs_zfs, OID_AUTO, l2c_only_size,
427 CTLTYPE_S64 | CTLFLAG_RD | CTLFLAG_MPSAFE,
428 &ARC_l2c_only, 0, param_get_arc_state_size, "Q",
429 "size of l2c_only state");
430
431 /* dbuf.c */
432
433 /* dmu.c */
434
435 /* dmu_zfetch.c */
436
437 SYSCTL_NODE(_vfs_zfs, OID_AUTO, zfetch, CTLFLAG_RW, 0, "ZFS ZFETCH (LEGACY)");
438
439 extern uint32_t zfetch_max_distance;
440
441 SYSCTL_UINT(_vfs_zfs_zfetch, OID_AUTO, max_distance,
442 CTLFLAG_RWTUN, &zfetch_max_distance, 0,
443 "Max bytes to prefetch per stream (LEGACY)");
444
445 extern uint32_t zfetch_max_idistance;
446
447 SYSCTL_UINT(_vfs_zfs_zfetch, OID_AUTO, max_idistance,
448 CTLFLAG_RWTUN, &zfetch_max_idistance, 0,
449 "Max bytes to prefetch indirects for per stream (LEGACY)");
450
451 /* dsl_pool.c */
452
453 /* dnode.c */
454
455 /* dsl_scan.c */
456
457 /* metaslab.c */
458
459 int
param_set_active_allocator(SYSCTL_HANDLER_ARGS)460 param_set_active_allocator(SYSCTL_HANDLER_ARGS)
461 {
462 char buf[16];
463 int rc;
464
465 if (req->newptr == NULL)
466 strlcpy(buf, zfs_active_allocator, sizeof (buf));
467
468 rc = sysctl_handle_string(oidp, buf, sizeof (buf), req);
469 if (rc || req->newptr == NULL)
470 return (rc);
471 if (strcmp(buf, zfs_active_allocator) == 0)
472 return (0);
473
474 return (param_set_active_allocator_common(buf));
475 }
476
477 /*
478 * In pools where the log space map feature is not enabled we touch
479 * multiple metaslabs (and their respective space maps) with each
480 * transaction group. Thus, we benefit from having a small space map
481 * block size since it allows us to issue more I/O operations scattered
482 * around the disk. So a sane default for the space map block size
483 * is 8~16K.
484 */
485 extern int zfs_metaslab_sm_blksz_no_log;
486
487 SYSCTL_INT(_vfs_zfs_metaslab, OID_AUTO, sm_blksz_no_log,
488 CTLFLAG_RDTUN, &zfs_metaslab_sm_blksz_no_log, 0,
489 "Block size for space map in pools with log space map disabled. "
490 "Power of 2 greater than 4096.");
491
492 /*
493 * When the log space map feature is enabled, we accumulate a lot of
494 * changes per metaslab that are flushed once in a while so we benefit
495 * from a bigger block size like 128K for the metaslab space maps.
496 */
497 extern int zfs_metaslab_sm_blksz_with_log;
498
499 SYSCTL_INT(_vfs_zfs_metaslab, OID_AUTO, sm_blksz_with_log,
500 CTLFLAG_RDTUN, &zfs_metaslab_sm_blksz_with_log, 0,
501 "Block size for space map in pools with log space map enabled. "
502 "Power of 2 greater than 4096.");
503
504 /*
505 * The in-core space map representation is more compact than its on-disk form.
506 * The zfs_condense_pct determines how much more compact the in-core
507 * space map representation must be before we compact it on-disk.
508 * Values should be greater than or equal to 100.
509 */
510 extern uint_t zfs_condense_pct;
511
512 SYSCTL_UINT(_vfs_zfs, OID_AUTO, condense_pct,
513 CTLFLAG_RWTUN, &zfs_condense_pct, 0,
514 "Condense on-disk spacemap when it is more than this many percents"
515 " of in-memory counterpart");
516
517 extern uint_t zfs_remove_max_segment;
518
519 SYSCTL_UINT(_vfs_zfs, OID_AUTO, remove_max_segment,
520 CTLFLAG_RWTUN, &zfs_remove_max_segment, 0,
521 "Largest contiguous segment ZFS will attempt to allocate when removing"
522 " a device");
523
524 extern int zfs_removal_suspend_progress;
525
526 SYSCTL_INT(_vfs_zfs, OID_AUTO, removal_suspend_progress,
527 CTLFLAG_RWTUN, &zfs_removal_suspend_progress, 0,
528 "Ensures certain actions can happen while in the middle of a removal");
529
530 /*
531 * Minimum size which forces the dynamic allocator to change
532 * it's allocation strategy. Once the space map cannot satisfy
533 * an allocation of this size then it switches to using more
534 * aggressive strategy (i.e search by size rather than offset).
535 */
536 extern uint64_t metaslab_df_alloc_threshold;
537
538 SYSCTL_QUAD(_vfs_zfs_metaslab, OID_AUTO, df_alloc_threshold,
539 CTLFLAG_RWTUN, &metaslab_df_alloc_threshold, 0,
540 "Minimum size which forces the dynamic allocator to change its"
541 " allocation strategy");
542
543 /*
544 * The minimum free space, in percent, which must be available
545 * in a space map to continue allocations in a first-fit fashion.
546 * Once the space map's free space drops below this level we dynamically
547 * switch to using best-fit allocations.
548 */
549 extern uint_t metaslab_df_free_pct;
550
551 SYSCTL_UINT(_vfs_zfs_metaslab, OID_AUTO, df_free_pct,
552 CTLFLAG_RWTUN, &metaslab_df_free_pct, 0,
553 "The minimum free space, in percent, which must be available in a"
554 " space map to continue allocations in a first-fit fashion");
555
556 /* mmp.c */
557
558 int
param_set_multihost_interval(SYSCTL_HANDLER_ARGS)559 param_set_multihost_interval(SYSCTL_HANDLER_ARGS)
560 {
561 int err;
562
563 err = sysctl_handle_64(oidp, &zfs_multihost_interval, 0, req);
564 if (err != 0 || req->newptr == NULL)
565 return (err);
566
567 if (spa_mode_global != SPA_MODE_UNINIT)
568 mmp_signal_all_threads();
569
570 return (0);
571 }
572
573 /* spa.c */
574
575 extern int zfs_ccw_retry_interval;
576
577 SYSCTL_INT(_vfs_zfs, OID_AUTO, ccw_retry_interval,
578 CTLFLAG_RWTUN, &zfs_ccw_retry_interval, 0,
579 "Configuration cache file write, retry after failure, interval"
580 " (seconds)");
581
582 extern uint64_t zfs_max_missing_tvds_cachefile;
583
584 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, max_missing_tvds_cachefile,
585 CTLFLAG_RWTUN, &zfs_max_missing_tvds_cachefile, 0,
586 "Allow importing pools with missing top-level vdevs in cache file");
587
588 extern uint64_t zfs_max_missing_tvds_scan;
589
590 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, max_missing_tvds_scan,
591 CTLFLAG_RWTUN, &zfs_max_missing_tvds_scan, 0,
592 "Allow importing pools with missing top-level vdevs during scan");
593
594 /* spa_misc.c */
595
596 extern int zfs_flags;
597
598 static int
sysctl_vfs_zfs_debug_flags(SYSCTL_HANDLER_ARGS)599 sysctl_vfs_zfs_debug_flags(SYSCTL_HANDLER_ARGS)
600 {
601 int err, val;
602
603 val = zfs_flags;
604 err = sysctl_handle_int(oidp, &val, 0, req);
605 if (err != 0 || req->newptr == NULL)
606 return (err);
607
608 /*
609 * ZFS_DEBUG_MODIFY must be enabled prior to boot so all
610 * arc buffers in the system have the necessary additional
611 * checksum data. However, it is safe to disable at any
612 * time.
613 */
614 if (!(zfs_flags & ZFS_DEBUG_MODIFY))
615 val &= ~ZFS_DEBUG_MODIFY;
616 zfs_flags = val;
617
618 return (0);
619 }
620
621 SYSCTL_PROC(_vfs_zfs, OID_AUTO, debugflags,
622 CTLTYPE_UINT | CTLFLAG_MPSAFE | CTLFLAG_RWTUN, NULL, 0,
623 sysctl_vfs_zfs_debug_flags, "IU", "Debug flags for ZFS testing.");
624
625 int
param_set_deadman_synctime(SYSCTL_HANDLER_ARGS)626 param_set_deadman_synctime(SYSCTL_HANDLER_ARGS)
627 {
628 unsigned long val;
629 int err;
630
631 val = zfs_deadman_synctime_ms;
632 err = sysctl_handle_64(oidp, &val, 0, req);
633 if (err != 0 || req->newptr == NULL)
634 return (err);
635 zfs_deadman_synctime_ms = val;
636
637 spa_set_deadman_synctime(MSEC2NSEC(zfs_deadman_synctime_ms));
638
639 return (0);
640 }
641
642 int
param_set_deadman_ziotime(SYSCTL_HANDLER_ARGS)643 param_set_deadman_ziotime(SYSCTL_HANDLER_ARGS)
644 {
645 unsigned long val;
646 int err;
647
648 val = zfs_deadman_ziotime_ms;
649 err = sysctl_handle_64(oidp, &val, 0, req);
650 if (err != 0 || req->newptr == NULL)
651 return (err);
652 zfs_deadman_ziotime_ms = val;
653
654 spa_set_deadman_ziotime(MSEC2NSEC(zfs_deadman_synctime_ms));
655
656 return (0);
657 }
658
659 int
param_set_deadman_failmode(SYSCTL_HANDLER_ARGS)660 param_set_deadman_failmode(SYSCTL_HANDLER_ARGS)
661 {
662 char buf[16];
663 int rc;
664
665 if (req->newptr == NULL)
666 strlcpy(buf, zfs_deadman_failmode, sizeof (buf));
667
668 rc = sysctl_handle_string(oidp, buf, sizeof (buf), req);
669 if (rc || req->newptr == NULL)
670 return (rc);
671 if (strcmp(buf, zfs_deadman_failmode) == 0)
672 return (0);
673 if (strcmp(buf, "wait") == 0)
674 zfs_deadman_failmode = "wait";
675 if (strcmp(buf, "continue") == 0)
676 zfs_deadman_failmode = "continue";
677 if (strcmp(buf, "panic") == 0)
678 zfs_deadman_failmode = "panic";
679
680 return (-param_set_deadman_failmode_common(buf));
681 }
682
683 int
param_set_raidz_impl(SYSCTL_HANDLER_ARGS)684 param_set_raidz_impl(SYSCTL_HANDLER_ARGS)
685 {
686 const size_t bufsize = 128;
687 char *buf;
688 int rc;
689
690 buf = malloc(bufsize, M_SOLARIS, M_WAITOK | M_ZERO);
691 if (req->newptr == NULL)
692 vdev_raidz_impl_get(buf, bufsize);
693
694 rc = sysctl_handle_string(oidp, buf, bufsize, req);
695 if (rc || req->newptr == NULL) {
696 free(buf, M_SOLARIS);
697 return (rc);
698 }
699 rc = vdev_raidz_impl_set(buf);
700 free(buf, M_SOLARIS);
701 return (rc);
702 }
703
704 int
param_set_slop_shift(SYSCTL_HANDLER_ARGS)705 param_set_slop_shift(SYSCTL_HANDLER_ARGS)
706 {
707 int val;
708 int err;
709
710 val = spa_slop_shift;
711 err = sysctl_handle_int(oidp, &val, 0, req);
712 if (err != 0 || req->newptr == NULL)
713 return (err);
714
715 if (val < 1 || val > 31)
716 return (EINVAL);
717
718 spa_slop_shift = val;
719
720 return (0);
721 }
722
723 /* spacemap.c */
724
725 extern int space_map_ibs;
726
727 SYSCTL_INT(_vfs_zfs, OID_AUTO, space_map_ibs, CTLFLAG_RWTUN,
728 &space_map_ibs, 0, "Space map indirect block shift");
729
730
731 /* vdev.c */
732
733 int
param_set_min_auto_ashift(SYSCTL_HANDLER_ARGS)734 param_set_min_auto_ashift(SYSCTL_HANDLER_ARGS)
735 {
736 int val;
737 int err;
738
739 val = zfs_vdev_min_auto_ashift;
740 err = sysctl_handle_int(oidp, &val, 0, req);
741 if (err != 0 || req->newptr == NULL)
742 return (SET_ERROR(err));
743
744 if (val < ASHIFT_MIN || val > zfs_vdev_max_auto_ashift)
745 return (SET_ERROR(EINVAL));
746
747 zfs_vdev_min_auto_ashift = val;
748
749 return (0);
750 }
751
752 SYSCTL_PROC(_vfs_zfs, OID_AUTO, min_auto_ashift,
753 CTLTYPE_UINT | CTLFLAG_RWTUN | CTLFLAG_MPSAFE,
754 &zfs_vdev_min_auto_ashift, sizeof (zfs_vdev_min_auto_ashift),
755 param_set_min_auto_ashift, "IU",
756 "Min ashift used when creating new top-level vdev. (LEGACY)");
757
758 int
param_set_max_auto_ashift(SYSCTL_HANDLER_ARGS)759 param_set_max_auto_ashift(SYSCTL_HANDLER_ARGS)
760 {
761 int val;
762 int err;
763
764 val = zfs_vdev_max_auto_ashift;
765 err = sysctl_handle_int(oidp, &val, 0, req);
766 if (err != 0 || req->newptr == NULL)
767 return (SET_ERROR(err));
768
769 if (val > ASHIFT_MAX || val < zfs_vdev_min_auto_ashift)
770 return (SET_ERROR(EINVAL));
771
772 zfs_vdev_max_auto_ashift = val;
773
774 return (0);
775 }
776
777 SYSCTL_PROC(_vfs_zfs, OID_AUTO, max_auto_ashift,
778 CTLTYPE_UINT | CTLFLAG_RWTUN | CTLFLAG_MPSAFE,
779 &zfs_vdev_max_auto_ashift, sizeof (zfs_vdev_max_auto_ashift),
780 param_set_max_auto_ashift, "IU",
781 "Max ashift used when optimizing for logical -> physical sector size on"
782 " new top-level vdevs. (LEGACY)");
783
784 /*
785 * Since the DTL space map of a vdev is not expected to have a lot of
786 * entries, we default its block size to 4K.
787 */
788 extern int zfs_vdev_dtl_sm_blksz;
789
790 SYSCTL_INT(_vfs_zfs, OID_AUTO, dtl_sm_blksz,
791 CTLFLAG_RDTUN, &zfs_vdev_dtl_sm_blksz, 0,
792 "Block size for DTL space map. Power of 2 greater than 4096.");
793
794 /*
795 * vdev-wide space maps that have lots of entries written to them at
796 * the end of each transaction can benefit from a higher I/O bandwidth
797 * (e.g. vdev_obsolete_sm), thus we default their block size to 128K.
798 */
799 extern int zfs_vdev_standard_sm_blksz;
800
801 SYSCTL_INT(_vfs_zfs, OID_AUTO, standard_sm_blksz,
802 CTLFLAG_RDTUN, &zfs_vdev_standard_sm_blksz, 0,
803 "Block size for standard space map. Power of 2 greater than 4096.");
804
805 extern int vdev_validate_skip;
806
807 SYSCTL_INT(_vfs_zfs, OID_AUTO, validate_skip,
808 CTLFLAG_RDTUN, &vdev_validate_skip, 0,
809 "Enable to bypass vdev_validate().");
810
811 /* vdev_mirror.c */
812
813 /* vdev_queue.c */
814
815 extern uint_t zfs_vdev_max_active;
816
817 SYSCTL_UINT(_vfs_zfs, OID_AUTO, top_maxinflight,
818 CTLFLAG_RWTUN, &zfs_vdev_max_active, 0,
819 "The maximum number of I/Os of all types active for each device."
820 " (LEGACY)");
821
822 /* zio.c */
823
824 SYSCTL_INT(_vfs_zfs_zio, OID_AUTO, exclude_metadata,
825 CTLFLAG_RDTUN, &zio_exclude_metadata, 0,
826 "Exclude metadata buffers from dumps as well");
827