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 static void
warn_deprecated_sysctl(const char * old,const char * new)167 warn_deprecated_sysctl(const char *old, const char *new)
168 {
169 printf("WARNING: sysctl vfs.zfs.%s is deprecated. Use vfs.zfs.%s instead.\n",
170 old, new);
171 }
172
173 int
param_set_l2arc_dwpd_limit(SYSCTL_HANDLER_ARGS)174 param_set_l2arc_dwpd_limit(SYSCTL_HANDLER_ARGS)
175 {
176 uint64_t old_val = l2arc_dwpd_limit;
177 int err;
178
179 err = sysctl_handle_64(oidp, arg1, 0, req);
180 if (err != 0 || req->newptr == NULL)
181 return (err);
182
183 if (l2arc_dwpd_limit != old_val)
184 l2arc_dwpd_bump_reset();
185
186 return (0);
187 }
188
189 int
param_set_arc_max(SYSCTL_HANDLER_ARGS)190 param_set_arc_max(SYSCTL_HANDLER_ARGS)
191 {
192 unsigned long val;
193 int err;
194
195 val = zfs_arc_max;
196 err = sysctl_handle_64(oidp, &val, 0, req);
197 if (err != 0 || req->newptr == NULL)
198 return (SET_ERROR(err));
199
200 if (val != 0 && (val < MIN_ARC_MAX || val <= arc_c_min ||
201 val >= arc_all_memory()))
202 return (SET_ERROR(EINVAL));
203
204 zfs_arc_max = val;
205 arc_tuning_update(B_TRUE);
206
207 /* Update the sysctl to the tuned value */
208 if (val != 0)
209 zfs_arc_max = arc_c_max;
210
211 if (arg2 != 0)
212 warn_deprecated_sysctl("arc_max", "arc.max");
213
214 return (0);
215 }
216
217 SYSCTL_PROC(_vfs_zfs, OID_AUTO, arc_max,
218 CTLTYPE_ULONG | CTLFLAG_RWTUN | CTLFLAG_MPSAFE,
219 NULL, 1, param_set_arc_max, "LU",
220 "Maximum ARC size in bytes (LEGACY)");
221
222 int
param_set_arc_min(SYSCTL_HANDLER_ARGS)223 param_set_arc_min(SYSCTL_HANDLER_ARGS)
224 {
225 unsigned long val;
226 int err;
227
228 val = zfs_arc_min;
229 err = sysctl_handle_64(oidp, &val, 0, req);
230 if (err != 0 || req->newptr == NULL)
231 return (SET_ERROR(err));
232
233 if (val != 0 && (val < 2ULL << SPA_MAXBLOCKSHIFT || val > arc_c_max))
234 return (SET_ERROR(EINVAL));
235
236 zfs_arc_min = val;
237 arc_tuning_update(B_TRUE);
238
239 /* Update the sysctl to the tuned value */
240 if (val != 0)
241 zfs_arc_min = arc_c_min;
242
243 if (arg2 != 0)
244 warn_deprecated_sysctl("arc_min", "arc.min");
245
246 return (0);
247 }
248
249 SYSCTL_PROC(_vfs_zfs, OID_AUTO, arc_min,
250 CTLTYPE_ULONG | CTLFLAG_RWTUN | CTLFLAG_MPSAFE,
251 NULL, 1, param_set_arc_min, "LU",
252 "Minimum ARC size in bytes (LEGACY)");
253
254 extern uint_t zfs_arc_free_target;
255
256 int
param_set_arc_free_target(SYSCTL_HANDLER_ARGS)257 param_set_arc_free_target(SYSCTL_HANDLER_ARGS)
258 {
259 uint_t val;
260 int err;
261
262 val = zfs_arc_free_target;
263 err = sysctl_handle_int(oidp, &val, 0, req);
264 if (err != 0 || req->newptr == NULL)
265 return (err);
266
267 if (val < minfree)
268 return (EINVAL);
269 if (val > vm_cnt.v_page_count)
270 return (EINVAL);
271
272 zfs_arc_free_target = val;
273
274 if (arg2 != 0)
275 warn_deprecated_sysctl("arc_free_target", "arc.free_target");
276
277 return (0);
278 }
279
280 /*
281 * NOTE: This sysctl is CTLFLAG_RW not CTLFLAG_RWTUN due to its dependency on
282 * pagedaemon initialization.
283 */
284 SYSCTL_PROC(_vfs_zfs, OID_AUTO, arc_free_target,
285 CTLTYPE_UINT | CTLFLAG_RW | CTLFLAG_MPSAFE,
286 NULL, 1, param_set_arc_free_target, "IU",
287 "Desired number of free pages below which ARC triggers reclaim"
288 " (LEGACY)");
289
290 int
param_set_arc_no_grow_shift(SYSCTL_HANDLER_ARGS)291 param_set_arc_no_grow_shift(SYSCTL_HANDLER_ARGS)
292 {
293 int err, val;
294
295 val = arc_no_grow_shift;
296 err = sysctl_handle_int(oidp, &val, 0, req);
297 if (err != 0 || req->newptr == NULL)
298 return (err);
299
300 if (val < 0 || val >= arc_shrink_shift)
301 return (EINVAL);
302
303 arc_no_grow_shift = val;
304
305 if (arg2 != 0)
306 warn_deprecated_sysctl("arc_no_grow_shift", "arc.no_grow_shift");
307
308 return (0);
309 }
310
311 SYSCTL_PROC(_vfs_zfs, OID_AUTO, arc_no_grow_shift,
312 CTLTYPE_INT | CTLFLAG_RWTUN | CTLFLAG_MPSAFE,
313 NULL, 1, param_set_arc_no_grow_shift, "I",
314 "log2(fraction of ARC which must be free to allow growing) (LEGACY)");
315
316 #if 0
317 extern uint64_t l2arc_write_max;
318
319 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_write_max,
320 CTLFLAG_RWTUN, &l2arc_write_max, 0,
321 "Max write bytes per interval (LEGACY)");
322
323 extern uint64_t l2arc_write_boost;
324
325 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_write_boost,
326 CTLFLAG_RWTUN, &l2arc_write_boost, 0,
327 "Extra write bytes during device warmup (LEGACY)");
328
329 extern uint64_t l2arc_headroom;
330
331 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_headroom,
332 CTLFLAG_RWTUN, &l2arc_headroom, 0,
333 "Number of max device writes to precache (LEGACY)");
334
335 extern uint64_t l2arc_headroom_boost;
336
337 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_headroom_boost,
338 CTLFLAG_RWTUN, &l2arc_headroom_boost, 0,
339 "Compressed l2arc_headroom multiplier (LEGACY)");
340
341 extern uint64_t l2arc_feed_secs;
342
343 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_feed_secs,
344 CTLFLAG_RWTUN, &l2arc_feed_secs, 0,
345 "Seconds between L2ARC writing (LEGACY)");
346
347 extern uint64_t l2arc_feed_min_ms;
348
349 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, l2arc_feed_min_ms,
350 CTLFLAG_RWTUN, &l2arc_feed_min_ms, 0,
351 "Min feed interval in milliseconds (LEGACY)");
352
353 extern int l2arc_noprefetch;
354
355 SYSCTL_INT(_vfs_zfs, OID_AUTO, l2arc_noprefetch,
356 CTLFLAG_RWTUN, &l2arc_noprefetch, 0,
357 "Skip caching prefetched buffers (LEGACY)");
358
359 extern int l2arc_feed_again;
360
361 SYSCTL_INT(_vfs_zfs, OID_AUTO, l2arc_feed_again,
362 CTLFLAG_RWTUN, &l2arc_feed_again, 0,
363 "Turbo L2ARC warmup (LEGACY)");
364
365 extern int l2arc_norw;
366
367 SYSCTL_INT(_vfs_zfs, OID_AUTO, l2arc_norw,
368 CTLFLAG_RWTUN, &l2arc_norw, 0,
369 "No reads during writes (LEGACY)");
370
371 #endif
372
373 static int
param_get_arc_state_size(SYSCTL_HANDLER_ARGS)374 param_get_arc_state_size(SYSCTL_HANDLER_ARGS)
375 {
376 arc_state_t *state = (arc_state_t *)arg1;
377 int64_t val;
378
379 val = zfs_refcount_count(&state->arcs_size[ARC_BUFC_DATA]) +
380 zfs_refcount_count(&state->arcs_size[ARC_BUFC_METADATA]);
381 return (sysctl_handle_64(oidp, &val, 0, req));
382 }
383
384 extern arc_state_t ARC_anon;
385
386 SYSCTL_PROC(_vfs_zfs, OID_AUTO, anon_size,
387 CTLTYPE_S64 | CTLFLAG_RD | CTLFLAG_MPSAFE,
388 &ARC_anon, 0, param_get_arc_state_size, "Q",
389 "size of anonymous state");
390 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, anon_metadata_esize, CTLFLAG_RD,
391 &ARC_anon.arcs_esize[ARC_BUFC_METADATA].rc_count, 0,
392 "size of evictable metadata in anonymous state");
393 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, anon_data_esize, CTLFLAG_RD,
394 &ARC_anon.arcs_esize[ARC_BUFC_DATA].rc_count, 0,
395 "size of evictable data in anonymous state");
396
397 extern arc_state_t ARC_mru;
398
399 SYSCTL_PROC(_vfs_zfs, OID_AUTO, mru_size,
400 CTLTYPE_S64 | CTLFLAG_RD | CTLFLAG_MPSAFE,
401 &ARC_mru, 0, param_get_arc_state_size, "Q",
402 "size of mru state");
403 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mru_metadata_esize, CTLFLAG_RD,
404 &ARC_mru.arcs_esize[ARC_BUFC_METADATA].rc_count, 0,
405 "size of evictable metadata in mru state");
406 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mru_data_esize, CTLFLAG_RD,
407 &ARC_mru.arcs_esize[ARC_BUFC_DATA].rc_count, 0,
408 "size of evictable data in mru state");
409
410 extern arc_state_t ARC_mru_ghost;
411
412 SYSCTL_PROC(_vfs_zfs, OID_AUTO, mru_ghost_size,
413 CTLTYPE_S64 | CTLFLAG_RD | CTLFLAG_MPSAFE,
414 &ARC_mru_ghost, 0, param_get_arc_state_size, "Q",
415 "size of mru ghost state");
416 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mru_ghost_metadata_esize, CTLFLAG_RD,
417 &ARC_mru_ghost.arcs_esize[ARC_BUFC_METADATA].rc_count, 0,
418 "size of evictable metadata in mru ghost state");
419 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mru_ghost_data_esize, CTLFLAG_RD,
420 &ARC_mru_ghost.arcs_esize[ARC_BUFC_DATA].rc_count, 0,
421 "size of evictable data in mru ghost state");
422
423 extern arc_state_t ARC_mfu;
424
425 SYSCTL_PROC(_vfs_zfs, OID_AUTO, mfu_size,
426 CTLTYPE_S64 | CTLFLAG_RD | CTLFLAG_MPSAFE,
427 &ARC_mfu, 0, param_get_arc_state_size, "Q",
428 "size of mfu state");
429 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mfu_metadata_esize, CTLFLAG_RD,
430 &ARC_mfu.arcs_esize[ARC_BUFC_METADATA].rc_count, 0,
431 "size of evictable metadata in mfu state");
432 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mfu_data_esize, CTLFLAG_RD,
433 &ARC_mfu.arcs_esize[ARC_BUFC_DATA].rc_count, 0,
434 "size of evictable data in mfu state");
435
436 extern arc_state_t ARC_mfu_ghost;
437
438 SYSCTL_PROC(_vfs_zfs, OID_AUTO, mfu_ghost_size,
439 CTLTYPE_S64 | CTLFLAG_RD | CTLFLAG_MPSAFE,
440 &ARC_mfu_ghost, 0, param_get_arc_state_size, "Q",
441 "size of mfu ghost state");
442 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mfu_ghost_metadata_esize, CTLFLAG_RD,
443 &ARC_mfu_ghost.arcs_esize[ARC_BUFC_METADATA].rc_count, 0,
444 "size of evictable metadata in mfu ghost state");
445 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, mfu_ghost_data_esize, CTLFLAG_RD,
446 &ARC_mfu_ghost.arcs_esize[ARC_BUFC_DATA].rc_count, 0,
447 "size of evictable data in mfu ghost state");
448
449 extern arc_state_t ARC_uncached;
450
451 SYSCTL_PROC(_vfs_zfs, OID_AUTO, uncached_size,
452 CTLTYPE_S64 | CTLFLAG_RD | CTLFLAG_MPSAFE,
453 &ARC_uncached, 0, param_get_arc_state_size, "Q",
454 "size of uncached state");
455 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, uncached_metadata_esize, CTLFLAG_RD,
456 &ARC_uncached.arcs_esize[ARC_BUFC_METADATA].rc_count, 0,
457 "size of evictable metadata in uncached state");
458 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, uncached_data_esize, CTLFLAG_RD,
459 &ARC_uncached.arcs_esize[ARC_BUFC_DATA].rc_count, 0,
460 "size of evictable data in uncached state");
461
462 extern arc_state_t ARC_l2c_only;
463
464 SYSCTL_PROC(_vfs_zfs, OID_AUTO, l2c_only_size,
465 CTLTYPE_S64 | CTLFLAG_RD | CTLFLAG_MPSAFE,
466 &ARC_l2c_only, 0, param_get_arc_state_size, "Q",
467 "size of l2c_only state");
468
469 /* dbuf.c */
470
471 /* dmu.c */
472
473 /* dmu_zfetch.c */
474
475 SYSCTL_NODE(_vfs_zfs, OID_AUTO, zfetch, CTLFLAG_RW, 0, "ZFS ZFETCH (LEGACY)");
476
477 extern uint32_t zfetch_max_distance;
478
479 SYSCTL_UINT(_vfs_zfs_zfetch, OID_AUTO, max_distance,
480 CTLFLAG_RWTUN, &zfetch_max_distance, 0,
481 "Max bytes to prefetch per stream (LEGACY)");
482
483 extern uint32_t zfetch_max_idistance;
484
485 SYSCTL_UINT(_vfs_zfs_zfetch, OID_AUTO, max_idistance,
486 CTLFLAG_RWTUN, &zfetch_max_idistance, 0,
487 "Max bytes to prefetch indirects for per stream (LEGACY)");
488
489 /* dsl_pool.c */
490
491 /* dnode.c */
492
493 /* dsl_scan.c */
494
495 /* metaslab.c */
496
497 int
param_set_active_allocator(SYSCTL_HANDLER_ARGS)498 param_set_active_allocator(SYSCTL_HANDLER_ARGS)
499 {
500 char buf[16];
501 int rc;
502
503 if (req->newptr == NULL)
504 strlcpy(buf, zfs_active_allocator, sizeof (buf));
505
506 rc = sysctl_handle_string(oidp, buf, sizeof (buf), req);
507 if (rc || req->newptr == NULL)
508 return (rc);
509 if (strcmp(buf, zfs_active_allocator) == 0)
510 return (0);
511
512 return (param_set_active_allocator_common(buf));
513 }
514
515 /*
516 * In pools where the log space map feature is not enabled we touch
517 * multiple metaslabs (and their respective space maps) with each
518 * transaction group. Thus, we benefit from having a small space map
519 * block size since it allows us to issue more I/O operations scattered
520 * around the disk. So a sane default for the space map block size
521 * is 8~16K.
522 */
523 extern int zfs_metaslab_sm_blksz_no_log;
524
525 SYSCTL_INT(_vfs_zfs_metaslab, OID_AUTO, sm_blksz_no_log,
526 CTLFLAG_RDTUN, &zfs_metaslab_sm_blksz_no_log, 0,
527 "Block size for space map in pools with log space map disabled. "
528 "Power of 2 greater than 4096.");
529
530 /*
531 * When the log space map feature is enabled, we accumulate a lot of
532 * changes per metaslab that are flushed once in a while so we benefit
533 * from a bigger block size like 128K for the metaslab space maps.
534 */
535 extern int zfs_metaslab_sm_blksz_with_log;
536
537 SYSCTL_INT(_vfs_zfs_metaslab, OID_AUTO, sm_blksz_with_log,
538 CTLFLAG_RDTUN, &zfs_metaslab_sm_blksz_with_log, 0,
539 "Block size for space map in pools with log space map enabled. "
540 "Power of 2 greater than 4096.");
541
542 /*
543 * The in-core space map representation is more compact than its on-disk form.
544 * The zfs_condense_pct determines how much more compact the in-core
545 * space map representation must be before we compact it on-disk.
546 * Values should be greater than or equal to 100.
547 */
548 extern uint_t zfs_condense_pct;
549
550 SYSCTL_UINT(_vfs_zfs, OID_AUTO, condense_pct,
551 CTLFLAG_RWTUN, &zfs_condense_pct, 0,
552 "Condense on-disk spacemap when it is more than this many percents"
553 " of in-memory counterpart");
554
555 extern uint_t zfs_remove_max_segment;
556
557 SYSCTL_UINT(_vfs_zfs, OID_AUTO, remove_max_segment,
558 CTLFLAG_RWTUN, &zfs_remove_max_segment, 0,
559 "Largest contiguous segment ZFS will attempt to allocate when removing"
560 " a device");
561
562 extern int zfs_removal_suspend_progress;
563
564 SYSCTL_INT(_vfs_zfs, OID_AUTO, removal_suspend_progress,
565 CTLFLAG_RWTUN, &zfs_removal_suspend_progress, 0,
566 "Ensures certain actions can happen while in the middle of a removal");
567
568 /*
569 * Minimum size which forces the dynamic allocator to change
570 * it's allocation strategy. Once the space map cannot satisfy
571 * an allocation of this size then it switches to using more
572 * aggressive strategy (i.e search by size rather than offset).
573 */
574 extern uint64_t metaslab_df_alloc_threshold;
575
576 SYSCTL_QUAD(_vfs_zfs_metaslab, OID_AUTO, df_alloc_threshold,
577 CTLFLAG_RWTUN, &metaslab_df_alloc_threshold, 0,
578 "Minimum size which forces the dynamic allocator to change its"
579 " allocation strategy");
580
581 /*
582 * The minimum free space, in percent, which must be available
583 * in a space map to continue allocations in a first-fit fashion.
584 * Once the space map's free space drops below this level we dynamically
585 * switch to using best-fit allocations.
586 */
587 extern uint_t metaslab_df_free_pct;
588
589 SYSCTL_UINT(_vfs_zfs_metaslab, OID_AUTO, df_free_pct,
590 CTLFLAG_RWTUN, &metaslab_df_free_pct, 0,
591 "The minimum free space, in percent, which must be available in a"
592 " space map to continue allocations in a first-fit fashion");
593
594 /* mmp.c */
595
596 int
param_set_multihost_interval(SYSCTL_HANDLER_ARGS)597 param_set_multihost_interval(SYSCTL_HANDLER_ARGS)
598 {
599 int err;
600
601 err = sysctl_handle_64(oidp, &zfs_multihost_interval, 0, req);
602 if (err != 0 || req->newptr == NULL)
603 return (err);
604
605 if (spa_mode_global != SPA_MODE_UNINIT)
606 mmp_signal_all_threads();
607
608 return (0);
609 }
610
611 /* spa.c */
612
613 extern int zfs_ccw_retry_interval;
614
615 SYSCTL_INT(_vfs_zfs, OID_AUTO, ccw_retry_interval,
616 CTLFLAG_RWTUN, &zfs_ccw_retry_interval, 0,
617 "Configuration cache file write, retry after failure, interval"
618 " (seconds)");
619
620 extern uint64_t zfs_max_missing_tvds_cachefile;
621
622 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, max_missing_tvds_cachefile,
623 CTLFLAG_RWTUN, &zfs_max_missing_tvds_cachefile, 0,
624 "Allow importing pools with missing top-level vdevs in cache file");
625
626 extern uint64_t zfs_max_missing_tvds_scan;
627
628 SYSCTL_UQUAD(_vfs_zfs, OID_AUTO, max_missing_tvds_scan,
629 CTLFLAG_RWTUN, &zfs_max_missing_tvds_scan, 0,
630 "Allow importing pools with missing top-level vdevs during scan");
631
632 /* spa_misc.c */
633
634 extern int zfs_flags;
635
636 static int
sysctl_vfs_zfs_debug_flags(SYSCTL_HANDLER_ARGS)637 sysctl_vfs_zfs_debug_flags(SYSCTL_HANDLER_ARGS)
638 {
639 int err, val;
640
641 val = zfs_flags;
642 err = sysctl_handle_int(oidp, &val, 0, req);
643 if (err != 0 || req->newptr == NULL)
644 return (err);
645
646 /*
647 * ZFS_DEBUG_MODIFY must be enabled prior to boot so all
648 * arc buffers in the system have the necessary additional
649 * checksum data. However, it is safe to disable at any
650 * time.
651 */
652 if (!(zfs_flags & ZFS_DEBUG_MODIFY))
653 val &= ~ZFS_DEBUG_MODIFY;
654 zfs_flags = val;
655
656 return (0);
657 }
658
659 SYSCTL_PROC(_vfs_zfs, OID_AUTO, debugflags,
660 CTLTYPE_UINT | CTLFLAG_MPSAFE | CTLFLAG_RWTUN, NULL, 0,
661 sysctl_vfs_zfs_debug_flags, "IU", "Debug flags for ZFS testing.");
662
663 int
param_set_deadman_synctime(SYSCTL_HANDLER_ARGS)664 param_set_deadman_synctime(SYSCTL_HANDLER_ARGS)
665 {
666 unsigned long val;
667 int err;
668
669 val = zfs_deadman_synctime_ms;
670 err = sysctl_handle_64(oidp, &val, 0, req);
671 if (err != 0 || req->newptr == NULL)
672 return (err);
673 zfs_deadman_synctime_ms = val;
674
675 spa_set_deadman_synctime(MSEC2NSEC(zfs_deadman_synctime_ms));
676
677 return (0);
678 }
679
680 int
param_set_deadman_ziotime(SYSCTL_HANDLER_ARGS)681 param_set_deadman_ziotime(SYSCTL_HANDLER_ARGS)
682 {
683 unsigned long val;
684 int err;
685
686 val = zfs_deadman_ziotime_ms;
687 err = sysctl_handle_64(oidp, &val, 0, req);
688 if (err != 0 || req->newptr == NULL)
689 return (err);
690 zfs_deadman_ziotime_ms = val;
691
692 spa_set_deadman_ziotime(MSEC2NSEC(zfs_deadman_synctime_ms));
693
694 return (0);
695 }
696
697 int
param_set_deadman_failmode(SYSCTL_HANDLER_ARGS)698 param_set_deadman_failmode(SYSCTL_HANDLER_ARGS)
699 {
700 char buf[16];
701 int rc;
702
703 if (req->newptr == NULL)
704 strlcpy(buf, zfs_deadman_failmode, sizeof (buf));
705
706 rc = sysctl_handle_string(oidp, buf, sizeof (buf), req);
707 if (rc || req->newptr == NULL)
708 return (rc);
709 if (strcmp(buf, zfs_deadman_failmode) == 0)
710 return (0);
711 if (strcmp(buf, "wait") == 0)
712 zfs_deadman_failmode = "wait";
713 if (strcmp(buf, "continue") == 0)
714 zfs_deadman_failmode = "continue";
715 if (strcmp(buf, "panic") == 0)
716 zfs_deadman_failmode = "panic";
717
718 return (-param_set_deadman_failmode_common(buf));
719 }
720
721 int
param_set_raidz_impl(SYSCTL_HANDLER_ARGS)722 param_set_raidz_impl(SYSCTL_HANDLER_ARGS)
723 {
724 const size_t bufsize = 128;
725 char *buf;
726 int rc;
727
728 buf = malloc(bufsize, M_SOLARIS, M_WAITOK | M_ZERO);
729 if (req->newptr == NULL)
730 vdev_raidz_impl_get(buf, bufsize);
731
732 rc = sysctl_handle_string(oidp, buf, bufsize, req);
733 if (rc || req->newptr == NULL) {
734 free(buf, M_SOLARIS);
735 return (rc);
736 }
737 rc = vdev_raidz_impl_set(buf);
738 free(buf, M_SOLARIS);
739 return (rc);
740 }
741
742 int
param_set_slop_shift(SYSCTL_HANDLER_ARGS)743 param_set_slop_shift(SYSCTL_HANDLER_ARGS)
744 {
745 int val;
746 int err;
747
748 val = spa_slop_shift;
749 err = sysctl_handle_int(oidp, &val, 0, req);
750 if (err != 0 || req->newptr == NULL)
751 return (err);
752
753 if (val < 1 || val > 31)
754 return (EINVAL);
755
756 spa_slop_shift = val;
757
758 return (0);
759 }
760
761 /* spacemap.c */
762
763 extern int space_map_ibs;
764
765 SYSCTL_INT(_vfs_zfs, OID_AUTO, space_map_ibs, CTLFLAG_RWTUN,
766 &space_map_ibs, 0, "Space map indirect block shift");
767
768
769 /* vdev.c */
770
771 int
param_set_min_auto_ashift(SYSCTL_HANDLER_ARGS)772 param_set_min_auto_ashift(SYSCTL_HANDLER_ARGS)
773 {
774 int val;
775 int err;
776
777 val = zfs_vdev_min_auto_ashift;
778 err = sysctl_handle_int(oidp, &val, 0, req);
779 if (err != 0 || req->newptr == NULL)
780 return (SET_ERROR(err));
781
782 if (val < ASHIFT_MIN || val > zfs_vdev_max_auto_ashift)
783 return (SET_ERROR(EINVAL));
784
785 zfs_vdev_min_auto_ashift = val;
786
787 if (arg2 != 0)
788 warn_deprecated_sysctl("min_auto_ashift",
789 "vdev.min_auto_ashift");
790
791 return (0);
792 }
793
794 SYSCTL_PROC(_vfs_zfs, OID_AUTO, min_auto_ashift,
795 CTLTYPE_UINT | CTLFLAG_RWTUN | CTLFLAG_MPSAFE, NULL, 1,
796 param_set_min_auto_ashift, "IU",
797 "Min ashift used when creating new top-level vdev. (LEGACY)");
798
799 int
param_set_max_auto_ashift(SYSCTL_HANDLER_ARGS)800 param_set_max_auto_ashift(SYSCTL_HANDLER_ARGS)
801 {
802 int val;
803 int err;
804
805 val = zfs_vdev_max_auto_ashift;
806 err = sysctl_handle_int(oidp, &val, 0, req);
807 if (err != 0 || req->newptr == NULL)
808 return (SET_ERROR(err));
809
810 if (val > ASHIFT_MAX || val < zfs_vdev_min_auto_ashift)
811 return (SET_ERROR(EINVAL));
812
813 zfs_vdev_max_auto_ashift = val;
814
815 if (arg2 != 0)
816 warn_deprecated_sysctl("max_auto_ashift",
817 "vdev.max_auto_ashift");
818
819 return (0);
820 }
821
822 SYSCTL_PROC(_vfs_zfs, OID_AUTO, max_auto_ashift,
823 CTLTYPE_UINT | CTLFLAG_RWTUN | CTLFLAG_MPSAFE, NULL, 1,
824 param_set_max_auto_ashift, "IU",
825 "Max ashift used when optimizing for logical -> physical sector size on"
826 " new top-level vdevs. (LEGACY)");
827
828 /*
829 * Since the DTL space map of a vdev is not expected to have a lot of
830 * entries, we default its block size to 4K.
831 */
832 extern int zfs_vdev_dtl_sm_blksz;
833
834 SYSCTL_INT(_vfs_zfs, OID_AUTO, dtl_sm_blksz,
835 CTLFLAG_RDTUN, &zfs_vdev_dtl_sm_blksz, 0,
836 "Block size for DTL space map. Power of 2 greater than 4096.");
837
838 /*
839 * vdev-wide space maps that have lots of entries written to them at
840 * the end of each transaction can benefit from a higher I/O bandwidth
841 * (e.g. vdev_obsolete_sm), thus we default their block size to 128K.
842 */
843 extern int zfs_vdev_standard_sm_blksz;
844
845 SYSCTL_INT(_vfs_zfs, OID_AUTO, standard_sm_blksz,
846 CTLFLAG_RDTUN, &zfs_vdev_standard_sm_blksz, 0,
847 "Block size for standard space map. Power of 2 greater than 4096.");
848
849 extern int vdev_validate_skip;
850
851 SYSCTL_INT(_vfs_zfs, OID_AUTO, validate_skip,
852 CTLFLAG_RDTUN, &vdev_validate_skip, 0,
853 "Enable to bypass vdev_validate().");
854
855 /* vdev_mirror.c */
856
857 /* vdev_queue.c */
858
859 extern uint_t zfs_vdev_max_active;
860
861 SYSCTL_UINT(_vfs_zfs, OID_AUTO, top_maxinflight,
862 CTLFLAG_RWTUN, &zfs_vdev_max_active, 0,
863 "The maximum number of I/Os of all types active for each device."
864 " (LEGACY)");
865
866 /* zio.c */
867
868 SYSCTL_INT(_vfs_zfs_zio, OID_AUTO, exclude_metadata,
869 CTLFLAG_RDTUN, &zio_exclude_metadata, 0,
870 "Exclude metadata buffers from dumps as well");
871