1 /*
2 * Copyright (c) 2000-2019 Apple Inc. All rights reserved.
3 *
4 * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
5 *
6 * This file contains Original Code and/or Modifications of Original Code
7 * as defined in and that are subject to the Apple Public Source License
8 * Version 2.0 (the 'License'). You may not use this file except in
9 * compliance with the License. The rights granted to you under the License
10 * may not be used to create, or enable the creation or redistribution of,
11 * unlawful or unlicensed copies of an Apple operating system, or to
12 * circumvent, violate, or enable the circumvention or violation of, any
13 * terms of an Apple operating system software license agreement.
14 *
15 * Please obtain a copy of the License at
16 * http://www.opensource.apple.com/apsl/ and read it before using this file.
17 *
18 * The Original Code and all software distributed under the License are
19 * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
20 * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
21 * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
22 * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
23 * Please see the License for the specific language governing rights and
24 * limitations under the License.
25 *
26 * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
27 */
28 /* Copyright (c) 1995 NeXT Computer, Inc. All Rights Reserved */
29 /*-
30 * Copyright (c) 1982, 1986, 1989, 1993
31 * The Regents of the University of California. All rights reserved.
32 *
33 * This code is derived from software contributed to Berkeley by
34 * Mike Karels at Berkeley Software Design, Inc.
35 *
36 * Redistribution and use in source and binary forms, with or without
37 * modification, are permitted provided that the following conditions
38 * are met:
39 * 1. Redistributions of source code must retain the above copyright
40 * notice, this list of conditions and the following disclaimer.
41 * 2. Redistributions in binary form must reproduce the above copyright
42 * notice, this list of conditions and the following disclaimer in the
43 * documentation and/or other materials provided with the distribution.
44 * 3. All advertising materials mentioning features or use of this software
45 * must display the following acknowledgement:
46 * This product includes software developed by the University of
47 * California, Berkeley and its contributors.
48 * 4. Neither the name of the University nor the names of its contributors
49 * may be used to endorse or promote products derived from this software
50 * without specific prior written permission.
51 *
52 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
53 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
54 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
55 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
56 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
57 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
58 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
59 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
60 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
61 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
62 * SUCH DAMAGE.
63 *
64 * @(#)kern_sysctl.c 8.4 (Berkeley) 4/14/94
65 */
66 /*
67 * NOTICE: This file was modified by SPARTA, Inc. in 2005 to introduce
68 * support for mandatory and extensible security protections. This notice
69 * is included in support of clause 2.2 (b) of the Apple Public License,
70 * Version 2.0.
71 */
72
73 /*
74 * DEPRECATED sysctl system call code
75 *
76 * Everything in this file is deprecated. Sysctls should be handled
77 * by the code in kern_newsysctl.c.
78 * The remaining "case" sections are supposed to be converted into
79 * SYSCTL_*-style definitions, and as soon as all of them are gone,
80 * this source file is supposed to die.
81 *
82 * DO NOT ADD ANY MORE "case" SECTIONS TO THIS FILE, instead define
83 * your sysctl with SYSCTL_INT, SYSCTL_PROC etc. in your source file.
84 */
85
86 #include <sys/param.h>
87 #include <sys/systm.h>
88 #include <sys/kernel.h>
89 #include <sys/malloc.h>
90 #include <sys/proc_internal.h>
91 #include <sys/kauth.h>
92 #include <sys/file_internal.h>
93 #include <sys/vnode_internal.h>
94 #include <sys/unistd.h>
95 #include <sys/buf.h>
96 #include <sys/ioctl.h>
97 #include <sys/namei.h>
98 #include <sys/tty.h>
99 #include <sys/disklabel.h>
100 #include <sys/vm.h>
101 #include <sys/sysctl.h>
102 #include <sys/user.h>
103 #include <sys/aio_kern.h>
104 #include <sys/reboot.h>
105 #include <sys/memory_maintenance.h>
106 #include <sys/priv.h>
107 #include <stdatomic.h>
108 #include <uuid/uuid.h>
109
110 #include <security/audit/audit.h>
111 #include <kern/kalloc.h>
112
113 #include <machine/smp.h>
114 #include <machine/atomic.h>
115 #include <machine/config.h>
116 #include <mach/machine.h>
117 #include <mach/mach_host.h>
118 #include <mach/mach_types.h>
119 #include <mach/processor_info.h>
120 #include <mach/vm_param.h>
121 #include <kern/debug.h>
122 #include <kern/mach_param.h>
123 #include <kern/task.h>
124 #include <kern/thread.h>
125 #include <kern/thread_group.h>
126 #include <kern/processor.h>
127 #include <kern/cpu_number.h>
128 #include <kern/cpu_quiesce.h>
129 #include <kern/sched_prim.h>
130 #include <vm/vm_kern.h>
131 #include <vm/vm_map.h>
132 #include <mach/host_info.h>
133
134 #include <sys/mount_internal.h>
135 #include <sys/kdebug.h>
136 #include <sys/kern_sysctl.h>
137
138 #include <IOKit/IOPlatformExpert.h>
139 #include <pexpert/pexpert.h>
140
141 #include <machine/machine_routines.h>
142 #include <machine/exec.h>
143
144 #include <nfs/nfs_conf.h>
145
146 #include <vm/vm_protos.h>
147 #include <vm/vm_pageout.h>
148 #include <vm/vm_compressor_algorithms.h>
149 #include <sys/imgsrc.h>
150 #include <kern/timer_call.h>
151 #include <sys/codesign.h>
152 #include <IOKit/IOBSD.h>
153 #if CONFIG_CSR
154 #include <sys/csr.h>
155 #endif
156
157 #if defined(__i386__) || defined(__x86_64__)
158 #include <i386/cpuid.h>
159 #endif
160
161 #if CONFIG_FREEZE
162 #include <sys/kern_memorystatus.h>
163 #endif
164
165 #if KPERF
166 #include <kperf/kperf.h>
167 #endif
168
169 #if HYPERVISOR
170 #include <kern/hv_support.h>
171 #endif
172
173 /*
174 * deliberately setting max requests to really high number
175 * so that runaway settings do not cause MALLOC overflows
176 */
177 #define AIO_MAX_REQUESTS (128 * CONFIG_AIO_MAX)
178
179 extern int aio_max_requests;
180 extern int aio_max_requests_per_process;
181 extern int aio_worker_threads;
182 extern int lowpri_IO_window_msecs;
183 extern int lowpri_IO_delay_msecs;
184 #if DEVELOPMENT || DEBUG
185 extern int nx_enabled;
186 #endif
187 extern int speculative_reads_disabled;
188 extern unsigned int speculative_prefetch_max;
189 extern unsigned int speculative_prefetch_max_iosize;
190 extern unsigned int preheat_max_bytes;
191 extern unsigned int preheat_min_bytes;
192 extern long numvnodes;
193 extern long freevnodes;
194 extern long num_recycledvnodes;
195
196 extern uuid_string_t bootsessionuuid_string;
197
198 extern unsigned int vm_max_delayed_work_limit;
199 extern unsigned int vm_max_batch;
200
201 extern unsigned int vm_page_free_min;
202 extern unsigned int vm_page_free_target;
203 extern unsigned int vm_page_free_reserved;
204
205 #if (DEVELOPMENT || DEBUG)
206 extern uint32_t vm_page_creation_throttled_hard;
207 extern uint32_t vm_page_creation_throttled_soft;
208 #endif /* DEVELOPMENT || DEBUG */
209
210 /*
211 * Conditionally allow dtrace to see these functions for debugging purposes.
212 */
213 #ifdef STATIC
214 #undef STATIC
215 #endif
216 #if 0
217 #define STATIC
218 #else
219 #define STATIC static
220 #endif
221
222 extern boolean_t mach_timer_coalescing_enabled;
223
224 extern uint64_t timer_deadline_tracking_bin_1, timer_deadline_tracking_bin_2;
225
226 STATIC void
227 fill_user32_eproc(proc_t, struct user32_eproc *__restrict);
228 STATIC void
229 fill_user32_externproc(proc_t, struct user32_extern_proc *__restrict);
230 STATIC void
231 fill_user64_eproc(proc_t, struct user64_eproc *__restrict);
232 STATIC void
233 fill_user64_proc(proc_t, struct user64_kinfo_proc *__restrict);
234 STATIC void
235 fill_user64_externproc(proc_t, struct user64_extern_proc *__restrict);
236 STATIC void
237 fill_user32_proc(proc_t, struct user32_kinfo_proc *__restrict);
238
239 extern int
240 kdbg_control(int *name, u_int namelen, user_addr_t where, size_t * sizep);
241 #if CONFIG_NFS_CLIENT
242 extern int
243 netboot_root(void);
244 #endif
245 int
246 sysctl_procargs(int *name, u_int namelen, user_addr_t where,
247 size_t *sizep, proc_t cur_proc);
248 STATIC int
249 sysctl_procargsx(int *name, u_int namelen, user_addr_t where, size_t *sizep,
250 proc_t cur_proc, int argc_yes);
251 int
252 sysctl_struct(user_addr_t oldp, size_t *oldlenp, user_addr_t newp,
253 size_t newlen, void *sp, int len);
254
255 STATIC int sysdoproc_filt_KERN_PROC_PID(proc_t p, void * arg);
256 STATIC int sysdoproc_filt_KERN_PROC_PGRP(proc_t p, void * arg);
257 STATIC int sysdoproc_filt_KERN_PROC_TTY(proc_t p, void * arg);
258 STATIC int sysdoproc_filt_KERN_PROC_UID(proc_t p, void * arg);
259 STATIC int sysdoproc_filt_KERN_PROC_RUID(proc_t p, void * arg);
260 int sysdoproc_callback(proc_t p, void *arg);
261
262 #if CONFIG_THREAD_GROUPS && (DEVELOPMENT || DEBUG)
263 STATIC int sysctl_get_thread_group_id SYSCTL_HANDLER_ARGS;
264 #endif
265
266 /* forward declarations for non-static STATIC */
267 STATIC void fill_loadavg64(struct loadavg *la, struct user64_loadavg *la64);
268 STATIC void fill_loadavg32(struct loadavg *la, struct user32_loadavg *la32);
269 STATIC int sysctl_handle_kern_threadname(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
270 STATIC int sysctl_sched_stats(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
271 STATIC int sysctl_sched_stats_enable(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
272 STATIC int sysctl_kdebug_ops SYSCTL_HANDLER_ARGS;
273 #if COUNT_SYSCALLS
274 STATIC int sysctl_docountsyscalls SYSCTL_HANDLER_ARGS;
275 #endif /* COUNT_SYSCALLS */
276 #if defined(XNU_TARGET_OS_OSX)
277 STATIC int sysctl_doprocargs SYSCTL_HANDLER_ARGS;
278 #endif /* defined(XNU_TARGET_OS_OSX) */
279 STATIC int sysctl_doprocargs2 SYSCTL_HANDLER_ARGS;
280 STATIC int sysctl_prochandle SYSCTL_HANDLER_ARGS;
281 STATIC int sysctl_aiomax(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
282 STATIC int sysctl_aioprocmax(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
283 STATIC int sysctl_aiothreads(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
284 STATIC int sysctl_maxproc(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
285 STATIC int sysctl_osversion(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
286 STATIC int sysctl_sysctl_bootargs(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
287 STATIC int sysctl_maxvnodes(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
288 STATIC int sysctl_securelvl(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
289 STATIC int sysctl_domainname(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
290 STATIC int sysctl_hostname(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
291 STATIC int sysctl_procname(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
292 STATIC int sysctl_boottime(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
293 STATIC int sysctl_bootuuid(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
294 STATIC int sysctl_symfile(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
295 #if CONFIG_NFS_CLIENT
296 STATIC int sysctl_netboot(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
297 #endif
298 #ifdef CONFIG_IMGSRC_ACCESS
299 STATIC int sysctl_imgsrcdev(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
300 #endif
301 STATIC int sysctl_usrstack(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
302 STATIC int sysctl_usrstack64(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
303 #if CONFIG_COREDUMP
304 STATIC int sysctl_coredump(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
305 STATIC int sysctl_suid_coredump(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
306 #endif
307 STATIC int sysctl_delayterm(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
308 STATIC int sysctl_rage_vnode(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
309 STATIC int sysctl_kern_check_openevt(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
310 #if DEVELOPMENT || DEBUG
311 STATIC int sysctl_nx(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
312 #endif
313 STATIC int sysctl_loadavg(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
314 STATIC int sysctl_vm_toggle_address_reuse(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
315 STATIC int sysctl_swapusage(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
316 STATIC int fetch_process_cputype( proc_t cur_proc, int *name, u_int namelen, cpu_type_t *cputype);
317 STATIC int sysctl_sysctl_native(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
318 STATIC int sysctl_sysctl_cputype(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
319 STATIC int sysctl_safeboot(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
320 STATIC int sysctl_singleuser(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
321 STATIC int sysctl_minimalboot(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
322 STATIC int sysctl_slide(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
323
324 #ifdef CONFIG_XNUPOST
325 #include <tests/xnupost.h>
326
327 STATIC int sysctl_debug_test_oslog_ctl(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
328 STATIC int sysctl_debug_test_stackshot_mutex_owner(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
329 STATIC int sysctl_debug_test_stackshot_rwlck_owner(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
330 #endif
331
332 extern void IORegistrySetOSBuildVersion(char * build_version);
333
334 STATIC void
fill_loadavg64(struct loadavg * la,struct user64_loadavg * la64)335 fill_loadavg64(struct loadavg *la, struct user64_loadavg *la64)
336 {
337 la64->ldavg[0] = la->ldavg[0];
338 la64->ldavg[1] = la->ldavg[1];
339 la64->ldavg[2] = la->ldavg[2];
340 la64->fscale = (user64_long_t)la->fscale;
341 }
342
343 STATIC void
fill_loadavg32(struct loadavg * la,struct user32_loadavg * la32)344 fill_loadavg32(struct loadavg *la, struct user32_loadavg *la32)
345 {
346 la32->ldavg[0] = la->ldavg[0];
347 la32->ldavg[1] = la->ldavg[1];
348 la32->ldavg[2] = la->ldavg[2];
349 la32->fscale = (user32_long_t)la->fscale;
350 }
351
352 #if CONFIG_COREDUMP
353 /*
354 * Attributes stored in the kernel.
355 */
356 extern char corefilename[MAXPATHLEN + 1];
357 extern int do_coredump;
358 extern int sugid_coredump;
359 #endif
360
361 #if COUNT_SYSCALLS
362 extern int do_count_syscalls;
363 #endif
364
365 #ifdef INSECURE
366 int securelevel = -1;
367 #else
368 int securelevel;
369 #endif
370
371 STATIC int
sysctl_handle_kern_threadname(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)372 sysctl_handle_kern_threadname( __unused struct sysctl_oid *oidp, __unused void *arg1,
373 __unused int arg2, struct sysctl_req *req)
374 {
375 int error;
376 struct uthread *ut = current_uthread();
377 user_addr_t oldp = 0, newp = 0;
378 size_t *oldlenp = NULL;
379 size_t newlen = 0;
380
381 oldp = req->oldptr;
382 oldlenp = &(req->oldlen);
383 newp = req->newptr;
384 newlen = req->newlen;
385
386 /* We want the current length, and maybe the string itself */
387 if (oldlenp) {
388 /* if we have no thread name yet tell'em we want MAXTHREADNAMESIZE - 1 */
389 size_t currlen = MAXTHREADNAMESIZE - 1;
390
391 if (ut->pth_name) {
392 /* use length of current thread name */
393 currlen = strlen(ut->pth_name);
394 }
395 if (oldp) {
396 if (*oldlenp < currlen) {
397 return ENOMEM;
398 }
399 /* NOTE - we do not copy the NULL terminator */
400 if (ut->pth_name) {
401 error = copyout(ut->pth_name, oldp, currlen);
402 if (error) {
403 return error;
404 }
405 }
406 }
407 /* return length of thread name minus NULL terminator (just like strlen) */
408 req->oldidx = currlen;
409 }
410
411 /* We want to set the name to something */
412 if (newp) {
413 if (newlen > (MAXTHREADNAMESIZE - 1)) {
414 return ENAMETOOLONG;
415 }
416 if (!ut->pth_name) {
417 char *tmp_pth_name = (char *)kalloc_data(MAXTHREADNAMESIZE,
418 Z_WAITOK | Z_ZERO);
419 if (!tmp_pth_name) {
420 return ENOMEM;
421 }
422 if (!OSCompareAndSwapPtr(NULL, tmp_pth_name, &ut->pth_name)) {
423 kfree_data(tmp_pth_name, MAXTHREADNAMESIZE);
424 return EBUSY;
425 }
426 } else {
427 kernel_debug_string_simple(TRACE_STRING_THREADNAME_PREV, ut->pth_name);
428 bzero(ut->pth_name, MAXTHREADNAMESIZE);
429 }
430 error = copyin(newp, ut->pth_name, newlen);
431 if (error) {
432 return error;
433 }
434
435 kernel_debug_string_simple(TRACE_STRING_THREADNAME, ut->pth_name);
436 }
437
438 return 0;
439 }
440
441 SYSCTL_PROC(_kern, KERN_THREADNAME, threadname, CTLFLAG_ANYBODY | CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_LOCKED, 0, 0, sysctl_handle_kern_threadname, "A", "");
442
443 #define BSD_HOST 1
444 STATIC int
sysctl_sched_stats(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)445 sysctl_sched_stats(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
446 {
447 host_basic_info_data_t hinfo;
448 kern_return_t kret;
449 uint32_t size;
450 uint32_t buf_size = 0;
451 int changed;
452 mach_msg_type_number_t count = HOST_BASIC_INFO_COUNT;
453 struct _processor_statistics_np *buf;
454 int error;
455
456 kret = host_info((host_t)BSD_HOST, HOST_BASIC_INFO, (host_info_t)&hinfo, &count);
457 if (kret != KERN_SUCCESS) {
458 return EINVAL;
459 }
460
461 size = sizeof(struct _processor_statistics_np) * (hinfo.logical_cpu_max + 2); /* One for RT Queue, One for Fair Share Queue */
462
463 if (req->oldlen < size) {
464 return EINVAL;
465 }
466
467 buf_size = size;
468 buf = (struct _processor_statistics_np *)kalloc_data(buf_size, Z_ZERO | Z_WAITOK);
469
470 kret = get_sched_statistics(buf, &size);
471 if (kret != KERN_SUCCESS) {
472 error = EINVAL;
473 goto out;
474 }
475
476 error = sysctl_io_opaque(req, buf, size, &changed);
477 if (error) {
478 goto out;
479 }
480
481 if (changed) {
482 panic("Sched info changed?!");
483 }
484 out:
485 kfree_data(buf, buf_size);
486 return error;
487 }
488
489 SYSCTL_PROC(_kern, OID_AUTO, sched_stats, CTLFLAG_LOCKED, 0, 0, sysctl_sched_stats, "-", "");
490
491 STATIC int
sysctl_sched_stats_enable(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,__unused struct sysctl_req * req)492 sysctl_sched_stats_enable(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, __unused struct sysctl_req *req)
493 {
494 boolean_t active;
495 int res;
496
497 if (req->newlen != sizeof(active)) {
498 return EINVAL;
499 }
500
501 res = copyin(req->newptr, &active, sizeof(active));
502 if (res != 0) {
503 return res;
504 }
505
506 return set_sched_stats_active(active);
507 }
508
509 SYSCTL_PROC(_kern, OID_AUTO, sched_stats_enable, CTLFLAG_LOCKED | CTLFLAG_WR, 0, 0, sysctl_sched_stats_enable, "-", "");
510
511 extern uint32_t sched_debug_flags;
512 SYSCTL_INT(_debug, OID_AUTO, sched, CTLFLAG_RW | CTLFLAG_LOCKED, &sched_debug_flags, 0, "scheduler debug");
513
514 #if (DEBUG || DEVELOPMENT)
515 extern boolean_t doprnt_hide_pointers;
516 SYSCTL_INT(_debug, OID_AUTO, hide_kernel_pointers, CTLFLAG_RW | CTLFLAG_LOCKED, &doprnt_hide_pointers, 0, "hide kernel pointers from log");
517 #endif
518
519
520 extern int get_kernel_symfile(proc_t, char **);
521
522 #if COUNT_SYSCALLS
523 #define KERN_COUNT_SYSCALLS (KERN_OSTYPE + 1000)
524
525 extern const unsigned int nsysent;
526 extern int syscalls_log[];
527 extern const char *syscallnames[];
528
529 STATIC int
530 sysctl_docountsyscalls SYSCTL_HANDLER_ARGS
531 {
532 __unused int cmd = oidp->oid_arg2; /* subcommand*/
533 __unused int *name = arg1; /* oid element argument vector */
534 __unused int namelen = arg2; /* number of oid element arguments */
535 int error, changed;
536
537 int tmp;
538
539 /* valid values passed in:
540 * = 0 means don't keep called counts for each bsd syscall
541 * > 0 means keep called counts for each bsd syscall
542 * = 2 means dump current counts to the system log
543 * = 3 means reset all counts
544 * for example, to dump current counts:
545 * sysctl -w kern.count_calls=2
546 */
547 error = sysctl_io_number(req, do_count_syscalls,
548 sizeof(do_count_syscalls), &tmp, &changed);
549
550 if (error != 0 || !changed) {
551 return error;
552 }
553
554 if (tmp == 1) {
555 do_count_syscalls = 1;
556 } else if (tmp == 0 || tmp == 2 || tmp == 3) {
557 for (int i = 0; i < nsysent; i++) {
558 if (syscalls_log[i] != 0) {
559 if (tmp == 2) {
560 printf("%d calls - name %s \n", syscalls_log[i], syscallnames[i]);
561 } else {
562 syscalls_log[i] = 0;
563 }
564 }
565 }
566 do_count_syscalls = (tmp != 0);
567 }
568
569 return error;
570 }
571 SYSCTL_PROC(_kern, KERN_COUNT_SYSCALLS, count_syscalls, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
572 0, /* Pointer argument (arg1) */
573 0, /* Integer argument (arg2) */
574 sysctl_docountsyscalls, /* Handler function */
575 NULL, /* Data pointer */
576 "");
577 #endif /* COUNT_SYSCALLS */
578
579 /*
580 * The following sysctl_* functions should not be used
581 * any more, as they can only cope with callers in
582 * user mode: Use new-style
583 * sysctl_io_number()
584 * sysctl_io_string()
585 * sysctl_io_opaque()
586 * instead.
587 */
588
589 STATIC int
sysdoproc_filt_KERN_PROC_PID(proc_t p,void * arg)590 sysdoproc_filt_KERN_PROC_PID(proc_t p, void * arg)
591 {
592 if (proc_getpid(p) != (pid_t)*(int*)arg) {
593 return 0;
594 } else {
595 return 1;
596 }
597 }
598
599 STATIC int
sysdoproc_filt_KERN_PROC_PGRP(proc_t p,void * arg)600 sysdoproc_filt_KERN_PROC_PGRP(proc_t p, void * arg)
601 {
602 if (p->p_pgrpid != (pid_t)*(int*)arg) {
603 return 0;
604 } else {
605 return 1;
606 }
607 }
608
609 STATIC int
sysdoproc_filt_KERN_PROC_TTY(proc_t p,void * arg)610 sysdoproc_filt_KERN_PROC_TTY(proc_t p, void * arg)
611 {
612 struct pgrp *pg;
613 dev_t dev = NODEV;
614
615 if ((p->p_flag & P_CONTROLT) && (pg = proc_pgrp(p, NULL)) != PGRP_NULL) {
616 dev = os_atomic_load(&pg->pg_session->s_ttydev, relaxed);
617 pgrp_rele(pg);
618 }
619
620 return dev != NODEV && dev == (dev_t)*(int *)arg;
621 }
622
623 STATIC int
sysdoproc_filt_KERN_PROC_UID(proc_t p,void * arg)624 sysdoproc_filt_KERN_PROC_UID(proc_t p, void * arg)
625 {
626 kauth_cred_t my_cred;
627 uid_t uid;
628
629 if (proc_ucred(p) == NULL) {
630 return 0;
631 }
632 my_cred = kauth_cred_proc_ref(p);
633 uid = kauth_cred_getuid(my_cred);
634 kauth_cred_unref(&my_cred);
635
636 if (uid != (uid_t)*(int*)arg) {
637 return 0;
638 } else {
639 return 1;
640 }
641 }
642
643
644 STATIC int
sysdoproc_filt_KERN_PROC_RUID(proc_t p,void * arg)645 sysdoproc_filt_KERN_PROC_RUID(proc_t p, void * arg)
646 {
647 kauth_cred_t my_cred;
648 uid_t ruid;
649
650 if (proc_ucred(p) == NULL) {
651 return 0;
652 }
653 my_cred = kauth_cred_proc_ref(p);
654 ruid = kauth_cred_getruid(my_cred);
655 kauth_cred_unref(&my_cred);
656
657 if (ruid != (uid_t)*(int*)arg) {
658 return 0;
659 } else {
660 return 1;
661 }
662 }
663
664 /*
665 * try over estimating by 5 procs
666 */
667 #define KERN_PROCSLOP (5 * sizeof(struct kinfo_proc))
668 struct sysdoproc_args {
669 size_t buflen;
670 void *kprocp;
671 boolean_t is_64_bit;
672 user_addr_t dp;
673 size_t needed;
674 unsigned int sizeof_kproc;
675 int *errorp;
676 int uidcheck;
677 int ruidcheck;
678 int ttycheck;
679 int uidval;
680 };
681
682 int
sysdoproc_callback(proc_t p,void * arg)683 sysdoproc_callback(proc_t p, void *arg)
684 {
685 struct sysdoproc_args *args = arg;
686
687 if (args->buflen >= args->sizeof_kproc) {
688 if ((args->ruidcheck != 0) && (sysdoproc_filt_KERN_PROC_RUID(p, &args->uidval) == 0)) {
689 return PROC_RETURNED;
690 }
691 if ((args->uidcheck != 0) && (sysdoproc_filt_KERN_PROC_UID(p, &args->uidval) == 0)) {
692 return PROC_RETURNED;
693 }
694 if ((args->ttycheck != 0) && (sysdoproc_filt_KERN_PROC_TTY(p, &args->uidval) == 0)) {
695 return PROC_RETURNED;
696 }
697
698 bzero(args->kprocp, args->sizeof_kproc);
699 if (args->is_64_bit) {
700 fill_user64_proc(p, args->kprocp);
701 } else {
702 fill_user32_proc(p, args->kprocp);
703 }
704 int error = copyout(args->kprocp, args->dp, args->sizeof_kproc);
705 if (error) {
706 *args->errorp = error;
707 return PROC_RETURNED_DONE;
708 }
709 args->dp += args->sizeof_kproc;
710 args->buflen -= args->sizeof_kproc;
711 }
712 args->needed += args->sizeof_kproc;
713 return PROC_RETURNED;
714 }
715
716 SYSCTL_NODE(_kern, KERN_PROC, proc, CTLFLAG_RD | CTLFLAG_LOCKED, 0, "");
717 STATIC int
718 sysctl_prochandle SYSCTL_HANDLER_ARGS
719 {
720 int cmd = oidp->oid_arg2; /* subcommand for multiple nodes */
721 int *name = arg1; /* oid element argument vector */
722 int namelen = arg2; /* number of oid element arguments */
723 user_addr_t where = req->oldptr;/* user buffer copy out address */
724
725 user_addr_t dp = where;
726 size_t needed = 0;
727 size_t buflen = where != USER_ADDR_NULL ? req->oldlen : 0;
728 int error = 0;
729 boolean_t is_64_bit = proc_is64bit(current_proc());
730 struct user32_kinfo_proc user32_kproc;
731 struct user64_kinfo_proc user_kproc;
732 int sizeof_kproc;
733 void *kprocp;
734 int (*filterfn)(proc_t, void *) = 0;
735 struct sysdoproc_args args;
736 int uidcheck = 0;
737 int ruidcheck = 0;
738 int ttycheck = 0;
739
740 if (namelen != 1 && !(namelen == 0 && cmd == KERN_PROC_ALL)) {
741 return EINVAL;
742 }
743
744 if (is_64_bit) {
745 sizeof_kproc = sizeof(user_kproc);
746 kprocp = &user_kproc;
747 } else {
748 sizeof_kproc = sizeof(user32_kproc);
749 kprocp = &user32_kproc;
750 }
751
752 switch (cmd) {
753 case KERN_PROC_PID:
754 filterfn = sysdoproc_filt_KERN_PROC_PID;
755 break;
756
757 case KERN_PROC_PGRP:
758 filterfn = sysdoproc_filt_KERN_PROC_PGRP;
759 break;
760
761 case KERN_PROC_TTY:
762 ttycheck = 1;
763 break;
764
765 case KERN_PROC_UID:
766 uidcheck = 1;
767 break;
768
769 case KERN_PROC_RUID:
770 ruidcheck = 1;
771 break;
772
773 case KERN_PROC_ALL:
774 break;
775
776 default:
777 /* must be kern.proc.<unknown> */
778 return ENOTSUP;
779 }
780
781 error = 0;
782 args.buflen = buflen;
783 args.kprocp = kprocp;
784 args.is_64_bit = is_64_bit;
785 args.dp = dp;
786 args.needed = needed;
787 args.errorp = &error;
788 args.uidcheck = uidcheck;
789 args.ruidcheck = ruidcheck;
790 args.ttycheck = ttycheck;
791 args.sizeof_kproc = sizeof_kproc;
792 if (namelen) {
793 args.uidval = name[0];
794 }
795
796 proc_iterate((PROC_ALLPROCLIST | PROC_ZOMBPROCLIST),
797 sysdoproc_callback, &args, filterfn, name);
798
799 if (error) {
800 return error;
801 }
802
803 dp = args.dp;
804 needed = args.needed;
805
806 if (where != USER_ADDR_NULL) {
807 req->oldlen = dp - where;
808 if (needed > req->oldlen) {
809 return ENOMEM;
810 }
811 } else {
812 needed += KERN_PROCSLOP;
813 req->oldlen = needed;
814 }
815 /* adjust index so we return the right required/consumed amount */
816 req->oldidx += req->oldlen;
817 return 0;
818 }
819
820
821 /*
822 * We specify the subcommand code for multiple nodes as the 'req->arg2' value
823 * in the sysctl declaration itself, which comes into the handler function
824 * as 'oidp->oid_arg2'.
825 *
826 * For these particular sysctls, since they have well known OIDs, we could
827 * have just obtained it from the '((int *)arg1)[0]' parameter, but that would
828 * not demonstrate how to handle multiple sysctls that used OID_AUTO instead
829 * of a well known value with a common handler function. This is desirable,
830 * because we want well known values to "go away" at some future date.
831 *
832 * It should be noted that the value of '((int *)arg1)[1]' is used for many
833 * an integer parameter to the subcommand for many of these sysctls; we'd
834 * rather have used '((int *)arg1)[0]' for that, or even better, an element
835 * in a structure passed in as the the 'newp' argument to sysctlbyname(3),
836 * and then use leaf-node permissions enforcement, but that would have
837 * necessitated modifying user space code to correspond to the interface
838 * change, and we are striving for binary backward compatibility here; even
839 * though these are SPI, and not intended for use by user space applications
840 * which are not themselves system tools or libraries, some applications
841 * have erroneously used them.
842 */
843 SYSCTL_PROC(_kern_proc, KERN_PROC_ALL, all, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
844 0, /* Pointer argument (arg1) */
845 KERN_PROC_ALL, /* Integer argument (arg2) */
846 sysctl_prochandle, /* Handler function */
847 NULL, /* Data is size variant on ILP32/LP64 */
848 "");
849 SYSCTL_PROC(_kern_proc, KERN_PROC_PID, pid, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
850 0, /* Pointer argument (arg1) */
851 KERN_PROC_PID, /* Integer argument (arg2) */
852 sysctl_prochandle, /* Handler function */
853 NULL, /* Data is size variant on ILP32/LP64 */
854 "");
855 SYSCTL_PROC(_kern_proc, KERN_PROC_TTY, tty, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
856 0, /* Pointer argument (arg1) */
857 KERN_PROC_TTY, /* Integer argument (arg2) */
858 sysctl_prochandle, /* Handler function */
859 NULL, /* Data is size variant on ILP32/LP64 */
860 "");
861 SYSCTL_PROC(_kern_proc, KERN_PROC_PGRP, pgrp, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
862 0, /* Pointer argument (arg1) */
863 KERN_PROC_PGRP, /* Integer argument (arg2) */
864 sysctl_prochandle, /* Handler function */
865 NULL, /* Data is size variant on ILP32/LP64 */
866 "");
867 SYSCTL_PROC(_kern_proc, KERN_PROC_UID, uid, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
868 0, /* Pointer argument (arg1) */
869 KERN_PROC_UID, /* Integer argument (arg2) */
870 sysctl_prochandle, /* Handler function */
871 NULL, /* Data is size variant on ILP32/LP64 */
872 "");
873 SYSCTL_PROC(_kern_proc, KERN_PROC_RUID, ruid, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
874 0, /* Pointer argument (arg1) */
875 KERN_PROC_RUID, /* Integer argument (arg2) */
876 sysctl_prochandle, /* Handler function */
877 NULL, /* Data is size variant on ILP32/LP64 */
878 "");
879 SYSCTL_PROC(_kern_proc, KERN_PROC_LCID, lcid, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
880 0, /* Pointer argument (arg1) */
881 KERN_PROC_LCID, /* Integer argument (arg2) */
882 sysctl_prochandle, /* Handler function */
883 NULL, /* Data is size variant on ILP32/LP64 */
884 "");
885
886
887 /*
888 * Fill in non-zero fields of an eproc structure for the specified process.
889 */
890 STATIC void
fill_user32_eproc(proc_t p,struct user32_eproc * __restrict ep)891 fill_user32_eproc(proc_t p, struct user32_eproc *__restrict ep)
892 {
893 struct pgrp *pg;
894 struct session *sessp;
895 kauth_cred_t my_cred;
896
897 pg = proc_pgrp(p, &sessp);
898
899 if (pg != PGRP_NULL) {
900 ep->e_pgid = p->p_pgrpid;
901 ep->e_jobc = pg->pg_jobc;
902 if (sessp->s_ttyvp) {
903 ep->e_flag = EPROC_CTTY;
904 }
905 }
906
907 ep->e_ppid = p->p_ppid;
908 if (proc_ucred(p)) {
909 my_cred = kauth_cred_proc_ref(p);
910
911 /* A fake historical pcred */
912 ep->e_pcred.p_ruid = kauth_cred_getruid(my_cred);
913 ep->e_pcred.p_svuid = kauth_cred_getsvuid(my_cred);
914 ep->e_pcred.p_rgid = kauth_cred_getrgid(my_cred);
915 ep->e_pcred.p_svgid = kauth_cred_getsvgid(my_cred);
916
917 /* A fake historical *kauth_cred_t */
918 unsigned long refcnt = os_atomic_load(&my_cred->cr_ref, relaxed);
919 ep->e_ucred.cr_ref = (uint32_t)MIN(refcnt, UINT32_MAX);
920 ep->e_ucred.cr_uid = kauth_cred_getuid(my_cred);
921 ep->e_ucred.cr_ngroups = (short)posix_cred_get(my_cred)->cr_ngroups;
922 bcopy(posix_cred_get(my_cred)->cr_groups,
923 ep->e_ucred.cr_groups, NGROUPS * sizeof(gid_t));
924
925 kauth_cred_unref(&my_cred);
926 }
927
928 ep->e_tdev = NODEV;
929 if (pg != PGRP_NULL) {
930 if (p->p_flag & P_CONTROLT) {
931 session_lock(sessp);
932 ep->e_tdev = os_atomic_load(&sessp->s_ttydev, relaxed);
933 ep->e_tpgid = sessp->s_ttypgrpid;
934 session_unlock(sessp);
935 }
936 if (SESS_LEADER(p, sessp)) {
937 ep->e_flag |= EPROC_SLEADER;
938 }
939 pgrp_rele(pg);
940 }
941 }
942
943 /*
944 * Fill in non-zero fields of an LP64 eproc structure for the specified process.
945 */
946 STATIC void
fill_user64_eproc(proc_t p,struct user64_eproc * __restrict ep)947 fill_user64_eproc(proc_t p, struct user64_eproc *__restrict ep)
948 {
949 struct pgrp *pg;
950 struct session *sessp;
951 kauth_cred_t my_cred;
952
953 pg = proc_pgrp(p, &sessp);
954
955 if (pg != PGRP_NULL) {
956 ep->e_pgid = p->p_pgrpid;
957 ep->e_jobc = pg->pg_jobc;
958 if (sessp->s_ttyvp) {
959 ep->e_flag = EPROC_CTTY;
960 }
961 }
962
963 ep->e_ppid = p->p_ppid;
964 if (proc_ucred(p)) {
965 my_cred = kauth_cred_proc_ref(p);
966
967 /* A fake historical pcred */
968 ep->e_pcred.p_ruid = kauth_cred_getruid(my_cred);
969 ep->e_pcred.p_svuid = kauth_cred_getsvuid(my_cred);
970 ep->e_pcred.p_rgid = kauth_cred_getrgid(my_cred);
971 ep->e_pcred.p_svgid = kauth_cred_getsvgid(my_cred);
972
973 /* A fake historical *kauth_cred_t */
974 unsigned long refcnt = os_atomic_load(&my_cred->cr_ref, relaxed);
975 ep->e_ucred.cr_ref = (uint32_t)MIN(refcnt, UINT32_MAX);
976 ep->e_ucred.cr_uid = kauth_cred_getuid(my_cred);
977 ep->e_ucred.cr_ngroups = (short)posix_cred_get(my_cred)->cr_ngroups;
978 bcopy(posix_cred_get(my_cred)->cr_groups,
979 ep->e_ucred.cr_groups, NGROUPS * sizeof(gid_t));
980
981 kauth_cred_unref(&my_cred);
982 }
983
984 ep->e_tdev = NODEV;
985 if (pg != PGRP_NULL) {
986 if (p->p_flag & P_CONTROLT) {
987 session_lock(sessp);
988 ep->e_tdev = os_atomic_load(&sessp->s_ttydev, relaxed);
989 ep->e_tpgid = sessp->s_ttypgrpid;
990 session_unlock(sessp);
991 }
992 if (SESS_LEADER(p, sessp)) {
993 ep->e_flag |= EPROC_SLEADER;
994 }
995 pgrp_rele(pg);
996 }
997 }
998
999 /*
1000 * Fill in an eproc structure for the specified process.
1001 * bzeroed by our caller, so only set non-zero fields.
1002 */
1003 STATIC void
fill_user32_externproc(proc_t p,struct user32_extern_proc * __restrict exp)1004 fill_user32_externproc(proc_t p, struct user32_extern_proc *__restrict exp)
1005 {
1006 exp->p_starttime.tv_sec = (user32_time_t)p->p_start.tv_sec;
1007 exp->p_starttime.tv_usec = p->p_start.tv_usec;
1008 exp->p_flag = p->p_flag;
1009 if (p->p_lflag & P_LTRACED) {
1010 exp->p_flag |= P_TRACED;
1011 }
1012 if (p->p_lflag & P_LPPWAIT) {
1013 exp->p_flag |= P_PPWAIT;
1014 }
1015 if (p->p_lflag & P_LEXIT) {
1016 exp->p_flag |= P_WEXIT;
1017 }
1018 exp->p_stat = p->p_stat;
1019 exp->p_pid = proc_getpid(p);
1020 exp->p_oppid = p->p_oppid;
1021 /* Mach related */
1022 exp->p_debugger = p->p_debugger;
1023 exp->sigwait = p->sigwait;
1024 /* scheduling */
1025 #ifdef _PROC_HAS_SCHEDINFO_
1026 exp->p_estcpu = p->p_estcpu;
1027 exp->p_pctcpu = p->p_pctcpu;
1028 exp->p_slptime = p->p_slptime;
1029 #endif
1030 exp->p_realtimer.it_interval.tv_sec =
1031 (user32_time_t)p->p_realtimer.it_interval.tv_sec;
1032 exp->p_realtimer.it_interval.tv_usec =
1033 (__int32_t)p->p_realtimer.it_interval.tv_usec;
1034
1035 exp->p_realtimer.it_value.tv_sec =
1036 (user32_time_t)p->p_realtimer.it_value.tv_sec;
1037 exp->p_realtimer.it_value.tv_usec =
1038 (__int32_t)p->p_realtimer.it_value.tv_usec;
1039
1040 exp->p_rtime.tv_sec = (user32_time_t)p->p_rtime.tv_sec;
1041 exp->p_rtime.tv_usec = (__int32_t)p->p_rtime.tv_usec;
1042
1043 exp->p_sigignore = p->p_sigignore;
1044 exp->p_sigcatch = p->p_sigcatch;
1045 exp->p_priority = p->p_priority;
1046 exp->p_nice = p->p_nice;
1047 bcopy(&p->p_comm, &exp->p_comm, MAXCOMLEN);
1048 exp->p_xstat = (u_short)MIN(p->p_xstat, USHRT_MAX);
1049 exp->p_acflag = p->p_acflag;
1050 }
1051
1052 /*
1053 * Fill in an LP64 version of extern_proc structure for the specified process.
1054 */
1055 STATIC void
fill_user64_externproc(proc_t p,struct user64_extern_proc * __restrict exp)1056 fill_user64_externproc(proc_t p, struct user64_extern_proc *__restrict exp)
1057 {
1058 exp->p_starttime.tv_sec = p->p_start.tv_sec;
1059 exp->p_starttime.tv_usec = p->p_start.tv_usec;
1060 exp->p_flag = p->p_flag;
1061 if (p->p_lflag & P_LTRACED) {
1062 exp->p_flag |= P_TRACED;
1063 }
1064 if (p->p_lflag & P_LPPWAIT) {
1065 exp->p_flag |= P_PPWAIT;
1066 }
1067 if (p->p_lflag & P_LEXIT) {
1068 exp->p_flag |= P_WEXIT;
1069 }
1070 exp->p_stat = p->p_stat;
1071 exp->p_pid = proc_getpid(p);
1072 exp->p_oppid = p->p_oppid;
1073 /* Mach related */
1074 exp->p_debugger = p->p_debugger;
1075 exp->sigwait = p->sigwait;
1076 /* scheduling */
1077 #ifdef _PROC_HAS_SCHEDINFO_
1078 exp->p_estcpu = p->p_estcpu;
1079 exp->p_pctcpu = p->p_pctcpu;
1080 exp->p_slptime = p->p_slptime;
1081 #endif
1082 exp->p_realtimer.it_interval.tv_sec = p->p_realtimer.it_interval.tv_sec;
1083 exp->p_realtimer.it_interval.tv_usec = p->p_realtimer.it_interval.tv_usec;
1084
1085 exp->p_realtimer.it_value.tv_sec = p->p_realtimer.it_value.tv_sec;
1086 exp->p_realtimer.it_value.tv_usec = p->p_realtimer.it_value.tv_usec;
1087
1088 exp->p_rtime.tv_sec = p->p_rtime.tv_sec;
1089 exp->p_rtime.tv_usec = p->p_rtime.tv_usec;
1090
1091 exp->p_sigignore = p->p_sigignore;
1092 exp->p_sigcatch = p->p_sigcatch;
1093 exp->p_priority = p->p_priority;
1094 exp->p_nice = p->p_nice;
1095 bcopy(&p->p_comm, &exp->p_comm, MAXCOMLEN);
1096 exp->p_xstat = (u_short)MIN(p->p_xstat, USHRT_MAX);
1097 exp->p_acflag = p->p_acflag;
1098 }
1099
1100 STATIC void
fill_user32_proc(proc_t p,struct user32_kinfo_proc * __restrict kp)1101 fill_user32_proc(proc_t p, struct user32_kinfo_proc *__restrict kp)
1102 {
1103 /* on a 64 bit kernel, 32 bit users get some truncated information */
1104 fill_user32_externproc(p, &kp->kp_proc);
1105 fill_user32_eproc(p, &kp->kp_eproc);
1106 }
1107
1108 STATIC void
fill_user64_proc(proc_t p,struct user64_kinfo_proc * __restrict kp)1109 fill_user64_proc(proc_t p, struct user64_kinfo_proc *__restrict kp)
1110 {
1111 fill_user64_externproc(p, &kp->kp_proc);
1112 fill_user64_eproc(p, &kp->kp_eproc);
1113 }
1114
1115 STATIC int
1116 sysctl_kdebug_ops SYSCTL_HANDLER_ARGS
1117 {
1118 __unused int cmd = oidp->oid_arg2; /* subcommand*/
1119 int *name = arg1; /* oid element argument vector */
1120 int namelen = arg2; /* number of oid element arguments */
1121 user_addr_t oldp = req->oldptr; /* user buffer copy out address */
1122 size_t *oldlenp = &req->oldlen; /* user buffer copy out size */
1123 // user_addr_t newp = req->newptr; /* user buffer copy in address */
1124 // size_t newlen = req->newlen; /* user buffer copy in size */
1125
1126 int ret = 0;
1127
1128 if (namelen == 0) {
1129 return ENOTSUP;
1130 }
1131
1132 switch (name[0]) {
1133 case KERN_KDEFLAGS:
1134 case KERN_KDDFLAGS:
1135 case KERN_KDENABLE:
1136 case KERN_KDGETBUF:
1137 case KERN_KDSETUP:
1138 case KERN_KDREMOVE:
1139 case KERN_KDSETREG:
1140 case KERN_KDGETREG:
1141 case KERN_KDREADTR:
1142 case KERN_KDWRITETR:
1143 case KERN_KDWRITEMAP:
1144 case KERN_KDTEST:
1145 case KERN_KDPIDTR:
1146 case KERN_KDTHRMAP:
1147 case KERN_KDPIDEX:
1148 case KERN_KDSETBUF:
1149 case KERN_KDREADCURTHRMAP:
1150 case KERN_KDSET_TYPEFILTER:
1151 case KERN_KDBUFWAIT:
1152 case KERN_KDCPUMAP:
1153 case KERN_KDCPUMAP_EXT:
1154 case KERN_KDWRITETR_V3:
1155 ret = kdbg_control(name, namelen, oldp, oldlenp);
1156 break;
1157 default:
1158 ret = ENOTSUP;
1159 break;
1160 }
1161
1162 /* adjust index so we return the right required/consumed amount */
1163 if (!ret) {
1164 req->oldidx += req->oldlen;
1165 }
1166
1167 return ret;
1168 }
1169 SYSCTL_PROC(_kern, KERN_KDEBUG, kdebug, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
1170 0, /* Pointer argument (arg1) */
1171 0, /* Integer argument (arg2) */
1172 sysctl_kdebug_ops, /* Handler function */
1173 NULL, /* Data pointer */
1174 "");
1175
1176
1177 #if defined(XNU_TARGET_OS_OSX)
1178 /*
1179 * Return the top *sizep bytes of the user stack, or the entire area of the
1180 * user stack down through the saved exec_path, whichever is smaller.
1181 */
1182 STATIC int
1183 sysctl_doprocargs SYSCTL_HANDLER_ARGS
1184 {
1185 __unused int cmd = oidp->oid_arg2; /* subcommand*/
1186 int *name = arg1; /* oid element argument vector */
1187 int namelen = arg2; /* number of oid element arguments */
1188 user_addr_t oldp = req->oldptr; /* user buffer copy out address */
1189 size_t *oldlenp = &req->oldlen; /* user buffer copy out size */
1190 // user_addr_t newp = req->newptr; /* user buffer copy in address */
1191 // size_t newlen = req->newlen; /* user buffer copy in size */
1192 int error;
1193
1194 error = sysctl_procargsx( name, namelen, oldp, oldlenp, current_proc(), 0);
1195
1196 /* adjust index so we return the right required/consumed amount */
1197 if (!error) {
1198 req->oldidx += req->oldlen;
1199 }
1200
1201 return error;
1202 }
1203 SYSCTL_PROC(_kern, KERN_PROCARGS, procargs, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
1204 0, /* Pointer argument (arg1) */
1205 0, /* Integer argument (arg2) */
1206 sysctl_doprocargs, /* Handler function */
1207 NULL, /* Data pointer */
1208 "");
1209 #endif /* defined(XNU_TARGET_OS_OSX) */
1210
1211 STATIC int
1212 sysctl_doprocargs2 SYSCTL_HANDLER_ARGS
1213 {
1214 __unused int cmd = oidp->oid_arg2; /* subcommand*/
1215 int *name = arg1; /* oid element argument vector */
1216 int namelen = arg2; /* number of oid element arguments */
1217 user_addr_t oldp = req->oldptr; /* user buffer copy out address */
1218 size_t *oldlenp = &req->oldlen; /* user buffer copy out size */
1219 // user_addr_t newp = req->newptr; /* user buffer copy in address */
1220 // size_t newlen = req->newlen; /* user buffer copy in size */
1221 int error;
1222
1223 error = sysctl_procargsx( name, namelen, oldp, oldlenp, current_proc(), 1);
1224
1225 /* adjust index so we return the right required/consumed amount */
1226 if (!error) {
1227 req->oldidx += req->oldlen;
1228 }
1229
1230 return error;
1231 }
1232 SYSCTL_PROC(_kern, KERN_PROCARGS2, procargs2, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
1233 0, /* Pointer argument (arg1) */
1234 0, /* Integer argument (arg2) */
1235 sysctl_doprocargs2, /* Handler function */
1236 NULL, /* Data pointer */
1237 "");
1238
1239 #define SYSCTL_PROCARGS_READ_ENVVARS_ENTITLEMENT "com.apple.private.read-environment-variables"
1240 STATIC int
sysctl_procargsx(int * name,u_int namelen,user_addr_t where,size_t * sizep,proc_t cur_proc,int argc_yes)1241 sysctl_procargsx(int *name, u_int namelen, user_addr_t where,
1242 size_t *sizep, proc_t cur_proc, int argc_yes)
1243 {
1244 assert(sizep != NULL);
1245 proc_t p = NULL;
1246 size_t buflen = where != USER_ADDR_NULL ? *sizep : 0;
1247 int error = 0;
1248 struct _vm_map *proc_map = NULL;
1249 struct task * task;
1250 vm_map_copy_t tmp = NULL;
1251 user_addr_t arg_addr;
1252 size_t arg_size;
1253 caddr_t data;
1254 size_t argslen = 0;
1255 size_t size = 0;
1256 vm_offset_t copy_start = 0, copy_end;
1257 vm_offset_t smallbuffer_start;
1258 kern_return_t ret;
1259 int pid;
1260 kauth_cred_t my_cred;
1261 uid_t uid;
1262 int argc = -1;
1263 size_t argvsize;
1264 size_t remaining;
1265 size_t current_arg_index;
1266 size_t current_arg_len;
1267 const char * current_arg;
1268 bool omit_env_vars = true;
1269 user_addr_t user_stack;
1270 vm_map_offset_t effective_page_mask;
1271
1272 if (namelen < 1) {
1273 error = EINVAL;
1274 goto finish;
1275 }
1276
1277 if (argc_yes) {
1278 buflen -= sizeof(int); /* reserve first word to return argc */
1279 }
1280 /* we only care about buflen when where (oldp from sysctl) is not NULL. */
1281 /* when where (oldp from sysctl) is NULL and sizep (oldlenp from sysctl */
1282 /* is not NULL then the caller wants us to return the length needed to */
1283 /* hold the data we would return */
1284 if (where != USER_ADDR_NULL && (buflen <= 0 || buflen > ARG_MAX)) {
1285 error = EINVAL;
1286 goto finish;
1287 }
1288
1289 /*
1290 * Lookup process by pid
1291 */
1292 pid = name[0];
1293 p = proc_find(pid);
1294 if (p == NULL) {
1295 error = EINVAL;
1296 goto finish;
1297 }
1298
1299 /* Allow reading environment variables if any of the following are true:
1300 * - kernel is DEVELOPMENT || DEBUG
1301 * - target process is same as current_proc()
1302 * - target process is not cs_restricted
1303 * - SIP is off
1304 * - caller has an entitlement
1305 */
1306
1307 #if DEVELOPMENT || DEBUG
1308 omit_env_vars = false;
1309 #endif
1310 if (p == current_proc() ||
1311 !cs_restricted(p) ||
1312 #if CONFIG_CSR
1313 csr_check(CSR_ALLOW_UNRESTRICTED_DTRACE) == 0 ||
1314 #endif
1315 IOCurrentTaskHasEntitlement(SYSCTL_PROCARGS_READ_ENVVARS_ENTITLEMENT)
1316 ) {
1317 omit_env_vars = false;
1318 }
1319
1320 /*
1321 * Copy the top N bytes of the stack.
1322 * On all machines we have so far, the stack grows
1323 * downwards.
1324 *
1325 * If the user expects no more than N bytes of
1326 * argument list, use that as a guess for the
1327 * size.
1328 */
1329
1330 if (!p->user_stack) {
1331 error = EINVAL;
1332 goto finish;
1333 }
1334
1335 /* save off argc, argslen, user_stack before releasing the proc */
1336 argc = p->p_argc;
1337 argslen = p->p_argslen;
1338 user_stack = p->user_stack;
1339
1340 /*
1341 * When these sysctls were introduced, the first string in the strings
1342 * section was just the bare path of the executable. However, for security
1343 * reasons we now prefix this string with executable_path= so it can be
1344 * parsed getenv style. To avoid binary compatability issues with exising
1345 * callers of this sysctl, we strip it off here.
1346 * (rdar://problem/13746466)
1347 */
1348 #define EXECUTABLE_KEY "executable_path="
1349 argslen -= strlen(EXECUTABLE_KEY);
1350
1351 if (where == USER_ADDR_NULL && !omit_env_vars) {
1352 /* caller only wants to know length of proc args data.
1353 * If we don't need to omit environment variables, we can skip
1354 * copying the target process stack */
1355 goto calculate_size;
1356 }
1357
1358 my_cred = kauth_cred_proc_ref(p);
1359 uid = kauth_cred_getuid(my_cred);
1360 kauth_cred_unref(&my_cred);
1361
1362 if ((uid != kauth_cred_getuid(kauth_cred_get()))
1363 && suser(kauth_cred_get(), &cur_proc->p_acflag)) {
1364 error = EINVAL;
1365 goto finish;
1366 }
1367
1368 /*
1369 * Before we can block (any VM code), make another
1370 * reference to the map to keep it alive. We do
1371 * that by getting a reference on the task itself.
1372 */
1373 task = p->task;
1374 if (task == NULL) {
1375 error = EINVAL;
1376 goto finish;
1377 }
1378
1379 /*
1380 * Once we have a task reference we can convert that into a
1381 * map reference, which we will use in the calls below. The
1382 * task/process may change its map after we take this reference
1383 * (see execve), but the worst that will happen then is a return
1384 * of stale info (which is always a possibility).
1385 */
1386 task_reference(task);
1387 proc_rele(p);
1388 p = NULL;
1389 proc_map = get_task_map_reference(task);
1390 task_deallocate(task);
1391
1392 if (proc_map == NULL) {
1393 error = EINVAL;
1394 goto finish;
1395 }
1396
1397 effective_page_mask = vm_map_page_mask(proc_map);
1398
1399 arg_size = vm_map_round_page(argslen, effective_page_mask);
1400
1401 arg_addr = user_stack - arg_size;
1402
1403 ret = kmem_alloc_flags(kernel_map, ©_start, arg_size, VM_KERN_MEMORY_BSD, KMA_ZERO);
1404 if (ret != KERN_SUCCESS) {
1405 error = ENOMEM;
1406 goto finish;
1407 }
1408
1409 copy_end = copy_start + arg_size;
1410
1411 if (vm_map_copyin(proc_map, (vm_map_address_t)arg_addr,
1412 (vm_map_size_t)arg_size, FALSE, &tmp) != KERN_SUCCESS) {
1413 error = EIO;
1414 goto finish;
1415 }
1416
1417 /*
1418 * Now that we've done the copyin from the process'
1419 * map, we can release the reference to it.
1420 */
1421 vm_map_deallocate(proc_map);
1422 proc_map = NULL;
1423
1424 if (vm_map_copy_overwrite(kernel_map,
1425 (vm_map_address_t)copy_start,
1426 tmp, (vm_map_size_t) arg_size, FALSE) != KERN_SUCCESS) {
1427 error = EIO;
1428 goto finish;
1429 }
1430 /* tmp was consumed */
1431 tmp = NULL;
1432
1433 if (omit_env_vars) {
1434 argvsize = 0;
1435
1436 /* Iterate over everything in argv, plus one for the bare executable path */
1437 for (current_arg_index = 0; current_arg_index < argc + 1 && argvsize < argslen; ++current_arg_index) {
1438 current_arg = (const char *)(copy_end - argslen) + argvsize;
1439 remaining = argslen - argvsize;
1440 current_arg_len = strnlen(current_arg, remaining);
1441 if (current_arg_len < remaining) {
1442 /* We have space for the null terminator */
1443 current_arg_len += 1;
1444
1445 if (current_arg_index == 0) {
1446 /* The bare executable path may have multiple null bytes after it for alignment */
1447 while (current_arg_len < remaining && current_arg[current_arg_len] == 0) {
1448 current_arg_len += 1;
1449 }
1450 }
1451 }
1452 argvsize += current_arg_len;
1453 }
1454 assert(argvsize <= argslen);
1455
1456 /* Adjust argslen and copy_end to make the copyout range extend to the end of argv */
1457 copy_end = copy_end - argslen + argvsize;
1458 argslen = argvsize;
1459 }
1460
1461 if (where == USER_ADDR_NULL) {
1462 /* Skip copyout */
1463 goto calculate_size;
1464 }
1465
1466 if (buflen >= argslen) {
1467 data = (caddr_t) (copy_end - argslen);
1468 size = argslen;
1469 } else {
1470 /*
1471 * Before rdar://25397314, this function contained incorrect logic when buflen is less
1472 * than argslen. The problem was that it copied in `buflen` bytes from the end of the target
1473 * process user stack into the beginning of a buffer of size round_page(buflen), and then
1474 * copied out `buflen` bytes from the end of this buffer. The effect of this was that
1475 * the caller of this sysctl would get zeros at the end of their buffer.
1476 *
1477 * To preserve this behavior, bzero everything from copy_end-round_page(buflen)+buflen to the
1478 * end of the buffer. This emulates copying in only `buflen` bytes.
1479 *
1480 *
1481 * In the old code:
1482 *
1483 * copy_start .... size: round_page(buflen) .... copy_end
1484 * [---copied in data (size: buflen)---|--- zeros ----------]
1485 * ^
1486 * data = copy_end - buflen
1487 *
1488 *
1489 * In the new code:
1490 * copy_start .... size: round_page(p->argslen) .... full copy_end
1491 * ^ ....................... p->argslen ...............................^
1492 * ^ ^ truncated copy_end ^
1493 * ^ ^ ^ ^
1494 * ^ ................ argslen ........................ ^
1495 * ^ ^ ^ ^
1496 * [-------copied in data (size: round_page(p->argslen))-------:----env vars---]
1497 * ^ ^
1498 * ^ data = copy_end - buflen
1499 * smallbuffer_start = max(copy_end - round_page(buflen), copy_start)
1500 *
1501 *
1502 * Full copy_end: copy_end calculated from copy_start + round_page(p->argslen)
1503 * Truncated copy_end: copy_end after truncation to remove environment variables.
1504 *
1505 * If environment variables were omitted, then we use the truncated copy_end, otherwise
1506 * we use full copy_end.
1507 *
1508 * smallbuffer_start: represents where copy_start would be in the old code.
1509 * data: The beginning of the region we copyout
1510 */
1511 smallbuffer_start = copy_end - vm_map_round_page(buflen, effective_page_mask);
1512 if (smallbuffer_start < copy_start) {
1513 smallbuffer_start = copy_start;
1514 }
1515 bzero((void *)(smallbuffer_start + buflen), copy_end - (smallbuffer_start + buflen));
1516 data = (caddr_t) (copy_end - buflen);
1517 size = buflen;
1518 }
1519
1520 if (argc_yes) {
1521 /* Put processes argc as the first word in the copyout buffer */
1522 suword(where, argc);
1523 error = copyout(data, (where + sizeof(int)), size);
1524 size += sizeof(int);
1525 } else {
1526 error = copyout(data, where, size);
1527
1528 /*
1529 * Make the old PROCARGS work to return the executable's path
1530 * But, only if there is enough space in the provided buffer
1531 *
1532 * on entry: data [possibily] points to the beginning of the path
1533 *
1534 * Note: we keep all pointers&sizes aligned to word boundries
1535 */
1536 if ((!error) && (buflen > 0 && (u_int)buflen > size)) {
1537 int binPath_sz, alignedBinPath_sz = 0;
1538 int extraSpaceNeeded, addThis;
1539 user_addr_t placeHere;
1540 char * str = (char *) data;
1541 size_t max_len = size;
1542
1543 /* Some apps are really bad about messing up their stacks
1544 * So, we have to be extra careful about getting the length
1545 * of the executing binary. If we encounter an error, we bail.
1546 */
1547
1548 /* Limit ourselves to PATH_MAX paths */
1549 if (max_len > PATH_MAX) {
1550 max_len = PATH_MAX;
1551 }
1552
1553 binPath_sz = 0;
1554
1555 while ((binPath_sz < max_len - 1) && (*str++ != 0)) {
1556 binPath_sz++;
1557 }
1558
1559 /* If we have a NUL terminator, copy it, too */
1560 if (binPath_sz < max_len - 1) {
1561 binPath_sz += 1;
1562 }
1563
1564 /* Pre-Flight the space requiremnts */
1565
1566 /* Account for the padding that fills out binPath to the next word */
1567 alignedBinPath_sz += (binPath_sz & (sizeof(int) - 1)) ? (sizeof(int) - (binPath_sz & (sizeof(int) - 1))) : 0;
1568
1569 placeHere = where + size;
1570
1571 /* Account for the bytes needed to keep placeHere word aligned */
1572 addThis = (placeHere & (sizeof(int) - 1)) ? (sizeof(int) - (placeHere & (sizeof(int) - 1))) : 0;
1573
1574 /* Add up all the space that is needed */
1575 extraSpaceNeeded = alignedBinPath_sz + addThis + binPath_sz + (4 * sizeof(int));
1576
1577 /* is there is room to tack on argv[0]? */
1578 if ((buflen & ~(sizeof(int) - 1)) >= (size + extraSpaceNeeded)) {
1579 placeHere += addThis;
1580 suword(placeHere, 0);
1581 placeHere += sizeof(int);
1582 suword(placeHere, 0xBFFF0000);
1583 placeHere += sizeof(int);
1584 suword(placeHere, 0);
1585 placeHere += sizeof(int);
1586 error = copyout(data, placeHere, binPath_sz);
1587 if (!error) {
1588 placeHere += binPath_sz;
1589 suword(placeHere, 0);
1590 size += extraSpaceNeeded;
1591 }
1592 }
1593 }
1594 }
1595
1596 calculate_size:
1597 /* Size has already been calculated for the where != NULL case */
1598 if (where == USER_ADDR_NULL) {
1599 size = argslen;
1600 if (argc_yes) {
1601 size += sizeof(int);
1602 } else {
1603 /*
1604 * old PROCARGS will return the executable's path and plus some
1605 * extra space for work alignment and data tags
1606 */
1607 size += PATH_MAX + (6 * sizeof(int));
1608 }
1609 size += (size & (sizeof(int) - 1)) ? (sizeof(int) - (size & (sizeof(int) - 1))) : 0;
1610 }
1611
1612 *sizep = size;
1613
1614 finish:
1615 if (p != NULL) {
1616 proc_rele(p);
1617 }
1618 if (tmp != NULL) {
1619 vm_map_copy_discard(tmp);
1620 }
1621 if (proc_map != NULL) {
1622 vm_map_deallocate(proc_map);
1623 }
1624 if (copy_start != (vm_offset_t) 0) {
1625 kmem_free(kernel_map, copy_start, arg_size);
1626 }
1627 return error;
1628 }
1629
1630
1631 /*
1632 * Max number of concurrent aio requests
1633 */
1634 STATIC int
sysctl_aiomax(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)1635 sysctl_aiomax
1636 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1637 {
1638 int new_value, changed;
1639 int error = sysctl_io_number(req, aio_max_requests, sizeof(int), &new_value, &changed);
1640 if (changed) {
1641 /* make sure the system-wide limit is greater than the per process limit */
1642 if (new_value >= aio_max_requests_per_process && new_value <= AIO_MAX_REQUESTS) {
1643 aio_max_requests = new_value;
1644 } else {
1645 error = EINVAL;
1646 }
1647 }
1648 return error;
1649 }
1650
1651
1652 /*
1653 * Max number of concurrent aio requests per process
1654 */
1655 STATIC int
sysctl_aioprocmax(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)1656 sysctl_aioprocmax
1657 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1658 {
1659 int new_value, changed;
1660 int error = sysctl_io_number(req, aio_max_requests_per_process, sizeof(int), &new_value, &changed);
1661 if (changed) {
1662 /* make sure per process limit is less than the system-wide limit */
1663 if (new_value <= aio_max_requests && new_value >= AIO_LISTIO_MAX) {
1664 aio_max_requests_per_process = new_value;
1665 } else {
1666 error = EINVAL;
1667 }
1668 }
1669 return error;
1670 }
1671
1672
1673 /*
1674 * Max number of async IO worker threads
1675 */
1676 STATIC int
sysctl_aiothreads(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)1677 sysctl_aiothreads
1678 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1679 {
1680 int new_value, changed;
1681 int error = sysctl_io_number(req, aio_worker_threads, sizeof(int), &new_value, &changed);
1682 if (changed) {
1683 /* we only allow an increase in the number of worker threads */
1684 if (new_value > aio_worker_threads) {
1685 _aio_create_worker_threads((new_value - aio_worker_threads));
1686 aio_worker_threads = new_value;
1687 } else {
1688 error = EINVAL;
1689 }
1690 }
1691 return error;
1692 }
1693
1694
1695 /*
1696 * System-wide limit on the max number of processes
1697 */
1698 STATIC int
sysctl_maxproc(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)1699 sysctl_maxproc
1700 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1701 {
1702 int new_value, changed;
1703 int error = sysctl_io_number(req, maxproc, sizeof(int), &new_value, &changed);
1704 if (changed) {
1705 AUDIT_ARG(value32, new_value);
1706 /* make sure the system-wide limit is less than the configured hard
1707 * limit set at kernel compilation */
1708 if (new_value <= hard_maxproc && new_value > 0) {
1709 maxproc = new_value;
1710 } else {
1711 error = EINVAL;
1712 }
1713 }
1714 return error;
1715 }
1716
1717 extern int sched_enable_smt;
1718 STATIC int
sysctl_sched_enable_smt(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)1719 sysctl_sched_enable_smt
1720 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1721 {
1722 int new_value, changed;
1723 int error = sysctl_io_number(req, sched_enable_smt, sizeof(int), &new_value, &changed);
1724 if (error) {
1725 return error;
1726 }
1727 kern_return_t kret = KERN_SUCCESS;
1728 if (changed) {
1729 AUDIT_ARG(value32, new_value);
1730 if (new_value == 0) {
1731 sched_enable_smt = 0;
1732 kret = enable_smt_processors(false);
1733 } else {
1734 sched_enable_smt = 1;
1735 kret = enable_smt_processors(true);
1736 }
1737 }
1738 switch (kret) {
1739 case KERN_SUCCESS:
1740 error = 0;
1741 break;
1742 case KERN_INVALID_ARGUMENT:
1743 error = EINVAL;
1744 break;
1745 case KERN_FAILURE:
1746 error = EBUSY;
1747 break;
1748 default:
1749 error = ENOENT;
1750 break;
1751 }
1752
1753 return error;
1754 }
1755
1756 SYSCTL_STRING(_kern, KERN_OSTYPE, ostype,
1757 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
1758 ostype, 0, "");
1759 SYSCTL_STRING(_kern, KERN_OSRELEASE, osrelease,
1760 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
1761 osrelease, 0, "");
1762 SYSCTL_INT(_kern, KERN_OSREV, osrevision,
1763 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
1764 (int *)NULL, BSD, "");
1765 SYSCTL_STRING(_kern, KERN_VERSION, version,
1766 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
1767 version, 0, "");
1768 SYSCTL_STRING(_kern, OID_AUTO, uuid,
1769 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
1770 &kernel_uuid_string[0], 0, "");
1771
1772 SYSCTL_STRING(_kern, OID_AUTO, osbuildconfig,
1773 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED | CTLFLAG_MASKED,
1774 &osbuild_config[0], 0, "");
1775
1776 #if DEBUG
1777 #ifndef DKPR
1778 #define DKPR 1
1779 #endif
1780 #endif
1781
1782 #if DKPR
1783 int debug_kprint_syscall = 0;
1784 char debug_kprint_syscall_process[MAXCOMLEN + 1];
1785
1786 /* Thread safe: bits and string value are not used to reclaim state */
1787 SYSCTL_INT(_debug, OID_AUTO, kprint_syscall,
1788 CTLFLAG_RW | CTLFLAG_LOCKED, &debug_kprint_syscall, 0, "kprintf syscall tracing");
1789 SYSCTL_STRING(_debug, OID_AUTO, kprint_syscall_process,
1790 CTLFLAG_RW | CTLFLAG_LOCKED, debug_kprint_syscall_process, sizeof(debug_kprint_syscall_process),
1791 "name of process for kprintf syscall tracing");
1792
1793 int
debug_kprint_current_process(const char ** namep)1794 debug_kprint_current_process(const char **namep)
1795 {
1796 struct proc *p = current_proc();
1797
1798 if (p == NULL) {
1799 return 0;
1800 }
1801
1802 if (debug_kprint_syscall_process[0]) {
1803 /* user asked to scope tracing to a particular process name */
1804 if (0 == strncmp(debug_kprint_syscall_process,
1805 p->p_comm, sizeof(debug_kprint_syscall_process))) {
1806 /* no value in telling the user that we traced what they asked */
1807 if (namep) {
1808 *namep = NULL;
1809 }
1810
1811 return 1;
1812 } else {
1813 return 0;
1814 }
1815 }
1816
1817 /* trace all processes. Tell user what we traced */
1818 if (namep) {
1819 *namep = p->p_comm;
1820 }
1821
1822 return 1;
1823 }
1824 #endif
1825
1826 /* PR-5293665: need to use a callback function for kern.osversion to set
1827 * osversion in IORegistry */
1828
1829 STATIC int
sysctl_osversion(__unused struct sysctl_oid * oidp,void * arg1,int arg2,struct sysctl_req * req)1830 sysctl_osversion(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
1831 {
1832 int rval = 0;
1833
1834 rval = sysctl_handle_string(oidp, arg1, arg2, req);
1835
1836 if (req->newptr) {
1837 IORegistrySetOSBuildVersion((char *)arg1);
1838 }
1839
1840 return rval;
1841 }
1842
1843 SYSCTL_PROC(_kern, KERN_OSVERSION, osversion,
1844 CTLFLAG_RW | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
1845 osversion, 256 /* OSVERSIZE*/,
1846 sysctl_osversion, "A", "");
1847
1848 static bool
_already_set_or_not_launchd(struct sysctl_req * req,char * val)1849 _already_set_or_not_launchd(struct sysctl_req *req, char *val)
1850 {
1851 if (req->newptr != 0) {
1852 /*
1853 * Can only ever be set by launchd, and only once at boot.
1854 */
1855 if (proc_getpid(req->p) != 1 || val[0] != '\0') {
1856 return true;
1857 }
1858 }
1859 return false;
1860 }
1861
1862 #if XNU_TARGET_OS_OSX
1863 static int
sysctl_system_version_compat(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)1864 sysctl_system_version_compat
1865 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1866 {
1867 int oldval = (task_has_system_version_compat_enabled(current_task()));
1868 int new_value = 0, changed = 0;
1869
1870 int error = sysctl_io_number(req, oldval, sizeof(int), &new_value, &changed);
1871 if (changed) {
1872 task_set_system_version_compat_enabled(current_task(), (new_value));
1873 }
1874 return error;
1875 }
1876
1877 SYSCTL_PROC(_kern, OID_AUTO, system_version_compat,
1878 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
1879 0, 0, sysctl_system_version_compat, "A", "");
1880
1881 char osproductversioncompat[48] = { '\0' };
1882
1883 static int
sysctl_osproductversioncompat(struct sysctl_oid * oidp,void * arg1,int arg2,struct sysctl_req * req)1884 sysctl_osproductversioncompat(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
1885 {
1886 if (_already_set_or_not_launchd(req, osproductversioncompat)) {
1887 return EPERM;
1888 }
1889 return sysctl_handle_string(oidp, arg1, arg2, req);
1890 }
1891
1892
1893 SYSCTL_PROC(_kern, OID_AUTO, osproductversioncompat,
1894 CTLFLAG_RW | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
1895 osproductversioncompat, sizeof(osproductversioncompat),
1896 sysctl_osproductversioncompat, "A", "The ProductVersion from SystemVersionCompat.plist");
1897 #endif
1898
1899 char osproductversion[48] = { '\0' };
1900
1901 static char iossupportversion_string[48] = { '\0' };
1902
1903 static int
sysctl_osproductversion(__unused struct sysctl_oid * oidp,void * arg1,int arg2,struct sysctl_req * req)1904 sysctl_osproductversion(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
1905 {
1906 if (_already_set_or_not_launchd(req, osproductversion)) {
1907 return EPERM;
1908 }
1909
1910 #if XNU_TARGET_OS_OSX
1911 if (task_has_system_version_compat_enabled(current_task()) && (osproductversioncompat[0] != '\0')) {
1912 return sysctl_handle_string(oidp, osproductversioncompat, arg2, req);
1913 } else {
1914 return sysctl_handle_string(oidp, arg1, arg2, req);
1915 }
1916 #else
1917 return sysctl_handle_string(oidp, arg1, arg2, req);
1918 #endif
1919 }
1920
1921 #if XNU_TARGET_OS_OSX
1922 static_assert(sizeof(osproductversioncompat) == sizeof(osproductversion),
1923 "osproductversion size matches osproductversioncompat size");
1924 #endif
1925
1926 SYSCTL_PROC(_kern, OID_AUTO, osproductversion,
1927 CTLFLAG_RW | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
1928 osproductversion, sizeof(osproductversion),
1929 sysctl_osproductversion, "A", "The ProductVersion from SystemVersion.plist");
1930
1931 char osreleasetype[48] = { '\0' };
1932
1933 STATIC int
sysctl_osreleasetype(__unused struct sysctl_oid * oidp,void * arg1,int arg2,struct sysctl_req * req)1934 sysctl_osreleasetype(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
1935 {
1936 if (_already_set_or_not_launchd(req, osreleasetype)) {
1937 return EPERM;
1938 }
1939 return sysctl_handle_string(oidp, arg1, arg2, req);
1940 }
1941
1942 void reset_osreleasetype(void);
1943
1944 void
reset_osreleasetype(void)1945 reset_osreleasetype(void)
1946 {
1947 memset(osreleasetype, 0, sizeof(osreleasetype));
1948 }
1949
1950 SYSCTL_PROC(_kern, OID_AUTO, osreleasetype,
1951 CTLFLAG_RW | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
1952 osreleasetype, sizeof(osreleasetype),
1953 sysctl_osreleasetype, "A", "The ReleaseType from SystemVersion.plist");
1954
1955 STATIC int
sysctl_iossupportversion(__unused struct sysctl_oid * oidp,void * arg1,int arg2,struct sysctl_req * req)1956 sysctl_iossupportversion(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
1957 {
1958 if (_already_set_or_not_launchd(req, iossupportversion_string)) {
1959 return EPERM;
1960 }
1961
1962 return sysctl_handle_string(oidp, arg1, arg2, req);
1963 }
1964
1965 SYSCTL_PROC(_kern, OID_AUTO, iossupportversion,
1966 CTLFLAG_RW | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
1967 iossupportversion_string, sizeof(iossupportversion_string),
1968 sysctl_iossupportversion, "A", "The iOSSupportVersion from SystemVersion.plist");
1969
1970 static uint64_t osvariant_status = 0;
1971
1972 STATIC int
sysctl_osvariant_status(__unused struct sysctl_oid * oidp,void * arg1,int arg2,struct sysctl_req * req)1973 sysctl_osvariant_status(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
1974 {
1975 if (req->newptr != 0) {
1976 /*
1977 * Can only ever be set by launchd, and only once.
1978 * Reset by usrctl() -> reset_osvariant_status() during
1979 * userspace reboot, since userspace could reboot into
1980 * a different variant.
1981 */
1982 if (proc_getpid(req->p) != 1 || osvariant_status != 0) {
1983 return EPERM;
1984 }
1985 }
1986
1987 return sysctl_handle_quad(oidp, arg1, arg2, req);
1988 }
1989
1990 SYSCTL_PROC(_kern, OID_AUTO, osvariant_status,
1991 CTLFLAG_RW | CTLTYPE_QUAD | CTLFLAG_LOCKED | CTLFLAG_MASKED,
1992 &osvariant_status, sizeof(osvariant_status),
1993 sysctl_osvariant_status, "Q", "Opaque flags used to cache OS variant information");
1994
1995 void reset_osvariant_status(void);
1996
1997 void
reset_osvariant_status(void)1998 reset_osvariant_status(void)
1999 {
2000 osvariant_status = 0;
2001 }
2002
2003 extern void commpage_update_dyld_flags(uint64_t);
2004 TUNABLE_WRITEABLE(uint64_t, dyld_flags, "dyld_flags", 0);
2005
2006 STATIC int
sysctl_dyld_flags(__unused struct sysctl_oid * oidp,void * arg1,int arg2,struct sysctl_req * req)2007 sysctl_dyld_flags(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2008 {
2009 /*
2010 * Can only ever be set by launchd, possibly several times
2011 * as dyld may change its mind after a userspace reboot.
2012 */
2013 if (req->newptr != 0 && proc_getpid(req->p) != 1) {
2014 return EPERM;
2015 }
2016
2017 int res = sysctl_handle_quad(oidp, arg1, arg2, req);
2018 if (req->newptr && res == 0) {
2019 commpage_update_dyld_flags(dyld_flags);
2020 }
2021 return res;
2022 }
2023
2024 SYSCTL_PROC(_kern, OID_AUTO, dyld_flags,
2025 CTLFLAG_RW | CTLTYPE_QUAD | CTLFLAG_LOCKED | CTLFLAG_MASKED,
2026 &dyld_flags, sizeof(dyld_flags),
2027 sysctl_dyld_flags, "Q", "Opaque flags used to cache dyld system-wide configuration");
2028
2029 #if defined(XNU_TARGET_OS_BRIDGE)
2030 char macosproductversion[MACOS_VERS_LEN] = { '\0' };
2031
2032 SYSCTL_STRING(_kern, OID_AUTO, macosproductversion,
2033 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2034 &macosproductversion[0], MACOS_VERS_LEN, "The currently running macOS ProductVersion (from SystemVersion.plist on macOS)");
2035
2036 char macosversion[MACOS_VERS_LEN] = { '\0' };
2037
2038 SYSCTL_STRING(_kern, OID_AUTO, macosversion,
2039 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2040 &macosversion[0], MACOS_VERS_LEN, "The currently running macOS build version");
2041 #endif
2042
2043 STATIC int
sysctl_sysctl_bootargs(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2044 sysctl_sysctl_bootargs
2045 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2046 {
2047 int error;
2048 char buf[BOOT_LINE_LENGTH];
2049
2050 strlcpy(buf, PE_boot_args(), BOOT_LINE_LENGTH);
2051 error = sysctl_io_string(req, buf, BOOT_LINE_LENGTH, 0, NULL);
2052 return error;
2053 }
2054
2055 SYSCTL_PROC(_kern, OID_AUTO, bootargs,
2056 CTLFLAG_LOCKED | CTLFLAG_RD | CTLFLAG_KERN | CTLTYPE_STRING,
2057 NULL, 0,
2058 sysctl_sysctl_bootargs, "A", "bootargs");
2059
2060 STATIC int
sysctl_kernelcacheuuid(struct sysctl_oid * oidp,void * arg1,int arg2,struct sysctl_req * req)2061 sysctl_kernelcacheuuid(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2062 {
2063 int rval = ENOENT;
2064 if (kernelcache_uuid_valid) {
2065 rval = sysctl_handle_string(oidp, arg1, arg2, req);
2066 }
2067 return rval;
2068 }
2069
2070 SYSCTL_PROC(_kern, OID_AUTO, kernelcacheuuid,
2071 CTLFLAG_RD | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2072 kernelcache_uuid_string, sizeof(kernelcache_uuid_string),
2073 sysctl_kernelcacheuuid, "A", "");
2074
2075 STATIC int
sysctl_systemfilesetuuid(struct sysctl_oid * oidp,void * arg1,int arg2,struct sysctl_req * req)2076 sysctl_systemfilesetuuid(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2077 {
2078 int rval = ENOENT;
2079 if (pageablekc_uuid_valid) {
2080 rval = sysctl_handle_string(oidp, arg1, arg2, req);
2081 }
2082 return rval;
2083 }
2084
2085 SYSCTL_PROC(_kern, OID_AUTO, systemfilesetuuid,
2086 CTLFLAG_RD | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2087 pageablekc_uuid_string, sizeof(pageablekc_uuid_string),
2088 sysctl_systemfilesetuuid, "A", "");
2089
2090 STATIC int
sysctl_auxiliaryfilesetuuid(struct sysctl_oid * oidp,void * arg1,int arg2,struct sysctl_req * req)2091 sysctl_auxiliaryfilesetuuid(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2092 {
2093 int rval = ENOENT;
2094 if (auxkc_uuid_valid) {
2095 rval = sysctl_handle_string(oidp, arg1, arg2, req);
2096 }
2097 return rval;
2098 }
2099
2100 SYSCTL_PROC(_kern, OID_AUTO, auxiliaryfilesetuuid,
2101 CTLFLAG_RD | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2102 auxkc_uuid_string, sizeof(auxkc_uuid_string),
2103 sysctl_auxiliaryfilesetuuid, "A", "");
2104
2105 STATIC int
sysctl_filesetuuid(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2106 sysctl_filesetuuid(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2107 {
2108 int rval = ENOENT;
2109 kc_format_t kcformat;
2110 kernel_mach_header_t *mh;
2111 void *uuid = NULL;
2112 unsigned long uuidlen = 0;
2113 uuid_string_t uuid_str;
2114
2115 if (!PE_get_primary_kc_format(&kcformat) || kcformat != KCFormatFileset) {
2116 return rval;
2117 }
2118
2119 mh = (kernel_mach_header_t *)PE_get_kc_header(KCKindPrimary);
2120 uuid = getuuidfromheader(mh, &uuidlen);
2121
2122 if ((uuid != NULL) && (uuidlen == sizeof(uuid_t))) {
2123 uuid_unparse_upper(*(uuid_t *)uuid, uuid_str);
2124 rval = sysctl_io_string(req, (char *)uuid_str, sizeof(uuid_str), 0, NULL);
2125 }
2126
2127 return rval;
2128 }
2129
2130 SYSCTL_PROC(_kern, OID_AUTO, filesetuuid,
2131 CTLFLAG_RD | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2132 NULL, 0,
2133 sysctl_filesetuuid, "A", "");
2134
2135
2136 SYSCTL_INT(_kern, KERN_MAXFILES, maxfiles,
2137 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2138 &maxfiles, 0, "");
2139 SYSCTL_INT(_kern, KERN_ARGMAX, argmax,
2140 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2141 (int *)NULL, ARG_MAX, "");
2142 SYSCTL_INT(_kern, KERN_POSIX1, posix1version,
2143 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2144 (int *)NULL, _POSIX_VERSION, "");
2145 SYSCTL_INT(_kern, KERN_NGROUPS, ngroups,
2146 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2147 (int *)NULL, NGROUPS_MAX, "");
2148 SYSCTL_INT(_kern, KERN_JOB_CONTROL, job_control,
2149 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2150 (int *)NULL, 1, "");
2151 #if 1 /* _POSIX_SAVED_IDS from <unistd.h> */
2152 SYSCTL_INT(_kern, KERN_SAVED_IDS, saved_ids,
2153 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2154 (int *)NULL, 1, "");
2155 #else
2156 SYSCTL_INT(_kern, KERN_SAVED_IDS, saved_ids,
2157 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2158 NULL, 0, "");
2159 #endif
2160 SYSCTL_INT(_kern, OID_AUTO, num_files,
2161 CTLFLAG_RD | CTLFLAG_LOCKED,
2162 &nfiles, 0, "");
2163 SYSCTL_COMPAT_INT(_kern, OID_AUTO, num_vnodes,
2164 CTLFLAG_RD | CTLFLAG_LOCKED,
2165 &numvnodes, 0, "");
2166 SYSCTL_INT(_kern, OID_AUTO, num_tasks,
2167 CTLFLAG_RD | CTLFLAG_LOCKED,
2168 &task_max, 0, "");
2169 SYSCTL_INT(_kern, OID_AUTO, num_threads,
2170 CTLFLAG_RD | CTLFLAG_LOCKED,
2171 &thread_max, 0, "");
2172 SYSCTL_INT(_kern, OID_AUTO, num_taskthreads,
2173 CTLFLAG_RD | CTLFLAG_LOCKED,
2174 &task_threadmax, 0, "");
2175 SYSCTL_LONG(_kern, OID_AUTO, num_recycledvnodes,
2176 CTLFLAG_RD | CTLFLAG_LOCKED,
2177 &num_recycledvnodes, "");
2178 SYSCTL_COMPAT_INT(_kern, OID_AUTO, free_vnodes,
2179 CTLFLAG_RD | CTLFLAG_LOCKED,
2180 &freevnodes, 0, "");
2181
2182 STATIC int
sysctl_maxvnodes(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2183 sysctl_maxvnodes(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2184 {
2185 int oldval = desiredvnodes;
2186 int error = sysctl_io_number(req, desiredvnodes, sizeof(int), &desiredvnodes, NULL);
2187
2188 if (oldval != desiredvnodes) {
2189 resize_namecache(desiredvnodes);
2190 }
2191
2192 return error;
2193 }
2194
2195 SYSCTL_INT(_kern, OID_AUTO, namecache_disabled,
2196 CTLFLAG_RW | CTLFLAG_LOCKED,
2197 &nc_disabled, 0, "");
2198
2199 SYSCTL_PROC(_kern, KERN_MAXVNODES, maxvnodes,
2200 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2201 0, 0, sysctl_maxvnodes, "I", "");
2202
2203 SYSCTL_PROC(_kern, KERN_MAXPROC, maxproc,
2204 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2205 0, 0, sysctl_maxproc, "I", "");
2206
2207 SYSCTL_PROC(_kern, KERN_AIOMAX, aiomax,
2208 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2209 0, 0, sysctl_aiomax, "I", "");
2210
2211 SYSCTL_PROC(_kern, KERN_AIOPROCMAX, aioprocmax,
2212 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2213 0, 0, sysctl_aioprocmax, "I", "");
2214
2215 SYSCTL_PROC(_kern, KERN_AIOTHREADS, aiothreads,
2216 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2217 0, 0, sysctl_aiothreads, "I", "");
2218
2219 SYSCTL_PROC(_kern, OID_AUTO, sched_enable_smt,
2220 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_KERN,
2221 0, 0, sysctl_sched_enable_smt, "I", "");
2222
2223 extern int sched_allow_NO_SMT_threads;
2224 SYSCTL_INT(_kern, OID_AUTO, sched_allow_NO_SMT_threads,
2225 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2226 &sched_allow_NO_SMT_threads, 0, "");
2227
2228 extern int sched_avoid_cpu0;
2229 SYSCTL_INT(_kern, OID_AUTO, sched_rt_avoid_cpu0,
2230 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2231 &sched_avoid_cpu0, 0, "If 1, choose cpu0 after all other primaries; if 2, choose cpu0 and cpu1 last, after all other cpus including secondaries");
2232
2233 #if (DEVELOPMENT || DEBUG)
2234 extern int smt_sched_bonus_16ths;
2235 SYSCTL_INT(_kern, OID_AUTO, smt_sched_bonus_16ths,
2236 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2237 &smt_sched_bonus_16ths, 0, "");
2238
2239 extern int smt_timeshare_enabled;
2240 SYSCTL_INT(_kern, OID_AUTO, sched_smt_timeshare_enable,
2241 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2242 &smt_timeshare_enabled, 0, "");
2243
2244 extern int sched_smt_balance;
2245 SYSCTL_INT(_kern, OID_AUTO, sched_smt_balance,
2246 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2247 &sched_smt_balance, 0, "");
2248 extern int sched_allow_rt_smt;
2249 SYSCTL_INT(_kern, OID_AUTO, sched_allow_rt_smt,
2250 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2251 &sched_allow_rt_smt, 0, "");
2252 extern int sched_allow_rt_steal;
2253 SYSCTL_INT(_kern, OID_AUTO, sched_allow_rt_steal,
2254 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2255 &sched_allow_rt_steal, 0, "");
2256 extern int sched_choose_first_fd_processor;
2257 SYSCTL_INT(_kern, OID_AUTO, sched_choose_first_fd_processor,
2258 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2259 &sched_choose_first_fd_processor, 0, "");
2260 extern int sched_backup_cpu_timeout_count;
2261 SYSCTL_INT(_kern, OID_AUTO, sched_backup_cpu_timeout_count,
2262 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2263 &sched_backup_cpu_timeout_count, 0, "The maximum number of 10us delays before allowing a backup cpu to select a thread");
2264 #if __arm__ || __arm64__
2265 extern uint32_t perfcontrol_requested_recommended_cores;
2266 SYSCTL_UINT(_kern, OID_AUTO, sched_recommended_cores,
2267 CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
2268 &perfcontrol_requested_recommended_cores, 0, "");
2269
2270 /* Scheduler perfcontrol callouts sysctls */
2271 SYSCTL_DECL(_kern_perfcontrol_callout);
2272 SYSCTL_NODE(_kern, OID_AUTO, perfcontrol_callout, CTLFLAG_RW | CTLFLAG_LOCKED, 0,
2273 "scheduler perfcontrol callouts");
2274
2275 extern int perfcontrol_callout_stats_enabled;
2276 SYSCTL_INT(_kern_perfcontrol_callout, OID_AUTO, stats_enabled,
2277 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2278 &perfcontrol_callout_stats_enabled, 0, "");
2279
2280 extern uint64_t perfcontrol_callout_stat_avg(perfcontrol_callout_type_t type,
2281 perfcontrol_callout_stat_t stat);
2282
2283 /* On-Core Callout */
2284 STATIC int
sysctl_perfcontrol_callout_stat(__unused struct sysctl_oid * oidp,void * arg1,int arg2,struct sysctl_req * req)2285 sysctl_perfcontrol_callout_stat
2286 (__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2287 {
2288 perfcontrol_callout_stat_t stat = (perfcontrol_callout_stat_t)arg1;
2289 perfcontrol_callout_type_t type = (perfcontrol_callout_type_t)arg2;
2290 return sysctl_io_number(req, (int)perfcontrol_callout_stat_avg(type, stat),
2291 sizeof(int), NULL, NULL);
2292 }
2293
2294 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, oncore_instr,
2295 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2296 (void *)PERFCONTROL_STAT_INSTRS, PERFCONTROL_CALLOUT_ON_CORE,
2297 sysctl_perfcontrol_callout_stat, "I", "");
2298 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, oncore_cycles,
2299 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2300 (void *)PERFCONTROL_STAT_CYCLES, PERFCONTROL_CALLOUT_ON_CORE,
2301 sysctl_perfcontrol_callout_stat, "I", "");
2302 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, offcore_instr,
2303 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2304 (void *)PERFCONTROL_STAT_INSTRS, PERFCONTROL_CALLOUT_OFF_CORE,
2305 sysctl_perfcontrol_callout_stat, "I", "");
2306 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, offcore_cycles,
2307 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2308 (void *)PERFCONTROL_STAT_CYCLES, PERFCONTROL_CALLOUT_OFF_CORE,
2309 sysctl_perfcontrol_callout_stat, "I", "");
2310 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, context_instr,
2311 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2312 (void *)PERFCONTROL_STAT_INSTRS, PERFCONTROL_CALLOUT_CONTEXT,
2313 sysctl_perfcontrol_callout_stat, "I", "");
2314 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, context_cycles,
2315 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2316 (void *)PERFCONTROL_STAT_CYCLES, PERFCONTROL_CALLOUT_CONTEXT,
2317 sysctl_perfcontrol_callout_stat, "I", "");
2318 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, update_instr,
2319 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2320 (void *)PERFCONTROL_STAT_INSTRS, PERFCONTROL_CALLOUT_STATE_UPDATE,
2321 sysctl_perfcontrol_callout_stat, "I", "");
2322 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, update_cycles,
2323 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2324 (void *)PERFCONTROL_STAT_CYCLES, PERFCONTROL_CALLOUT_STATE_UPDATE,
2325 sysctl_perfcontrol_callout_stat, "I", "");
2326
2327 #if __AMP__
2328 extern int sched_amp_idle_steal;
2329 SYSCTL_INT(_kern, OID_AUTO, sched_amp_idle_steal,
2330 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2331 &sched_amp_idle_steal, 0, "");
2332 extern int sched_amp_spill_steal;
2333 SYSCTL_INT(_kern, OID_AUTO, sched_amp_spill_steal,
2334 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2335 &sched_amp_spill_steal, 0, "");
2336 extern int sched_amp_spill_count;
2337 SYSCTL_INT(_kern, OID_AUTO, sched_amp_spill_count,
2338 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2339 &sched_amp_spill_count, 0, "");
2340 extern int sched_amp_spill_deferred_ipi;
2341 SYSCTL_INT(_kern, OID_AUTO, sched_amp_spill_deferred_ipi,
2342 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2343 &sched_amp_spill_deferred_ipi, 0, "");
2344 extern int sched_amp_pcores_preempt_immediate_ipi;
2345 SYSCTL_INT(_kern, OID_AUTO, sched_amp_pcores_preempt_immediate_ipi,
2346 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2347 &sched_amp_pcores_preempt_immediate_ipi, 0, "");
2348 #endif /* __AMP__ */
2349 #endif /* __arm__ || __arm64__ */
2350
2351 #if __arm64__
2352 extern int legacy_footprint_entitlement_mode;
2353 SYSCTL_INT(_kern, OID_AUTO, legacy_footprint_entitlement_mode,
2354 CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
2355 &legacy_footprint_entitlement_mode, 0, "");
2356 #endif /* __arm64__ */
2357
2358 /*
2359 * Realtime threads are ordered by highest priority first then,
2360 * for threads of the same priority, by earliest deadline first.
2361 * But if sched_rt_runq_strict_priority is false (the default),
2362 * a lower priority thread with an earlier deadline will be preferred
2363 * over a higher priority thread with a later deadline, as long as
2364 * both threads' computations will fit before the later deadline.
2365 */
2366 extern int sched_rt_runq_strict_priority;
2367 SYSCTL_INT(_kern, OID_AUTO, sched_rt_runq_strict_priority,
2368 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2369 &sched_rt_runq_strict_priority, 0, "");
2370
2371 static int
sysctl_kern_sched_rt_n_backup_processors(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2372 sysctl_kern_sched_rt_n_backup_processors(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2373 {
2374 int new_value, changed;
2375 int old_value = sched_get_rt_n_backup_processors();
2376 int error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
2377 if (changed) {
2378 sched_set_rt_n_backup_processors(new_value);
2379 }
2380
2381 return error;
2382 }
2383
2384 SYSCTL_PROC(_kern, OID_AUTO, sched_rt_n_backup_processors,
2385 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2386 0, 0, sysctl_kern_sched_rt_n_backup_processors, "I", "");
2387
2388 static int
sysctl_kern_sched_rt_constraint_ll_us(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2389 sysctl_kern_sched_rt_constraint_ll_us(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2390 {
2391 int new_value, changed;
2392 int old_value = sched_get_rt_constraint_ll();
2393 int error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
2394 if (changed) {
2395 sched_set_rt_constraint_ll(new_value);
2396 }
2397
2398 return error;
2399 }
2400
2401 SYSCTL_PROC(_kern, OID_AUTO, sched_rt_constraint_ll_us,
2402 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2403 0, 0, sysctl_kern_sched_rt_constraint_ll_us, "I", "");
2404
2405 static int
sysctl_kern_sched_rt_deadline_epsilon_us(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2406 sysctl_kern_sched_rt_deadline_epsilon_us(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2407 {
2408 int new_value, changed;
2409 int old_value = sched_get_rt_deadline_epsilon();
2410 int error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
2411 if (changed) {
2412 sched_set_rt_deadline_epsilon(new_value);
2413 }
2414
2415 return error;
2416 }
2417
2418 SYSCTL_PROC(_kern, OID_AUTO, sched_rt_deadline_epsilon_us,
2419 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2420 0, 0, sysctl_kern_sched_rt_deadline_epsilon_us, "I", "");
2421
2422 extern int sched_idle_delay_cpuid;
2423 SYSCTL_INT(_kern, OID_AUTO, sched_idle_delay_cpuid,
2424 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2425 &sched_idle_delay_cpuid, 0, "This cpuid will be delayed by 500us on exiting idle, to simulate interrupt or preemption delays when testing the scheduler");
2426
2427 #endif /* (DEVELOPMENT || DEBUG) */
2428
2429 STATIC int
sysctl_securelvl(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2430 sysctl_securelvl
2431 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2432 {
2433 int new_value, changed;
2434 int error = sysctl_io_number(req, securelevel, sizeof(int), &new_value, &changed);
2435 if (changed) {
2436 if (!(new_value < securelevel && proc_getpid(req->p) != 1)) {
2437 proc_list_lock();
2438 securelevel = new_value;
2439 proc_list_unlock();
2440 } else {
2441 error = EPERM;
2442 }
2443 }
2444 return error;
2445 }
2446
2447 SYSCTL_PROC(_kern, KERN_SECURELVL, securelevel,
2448 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2449 0, 0, sysctl_securelvl, "I", "");
2450
2451
2452 STATIC int
sysctl_domainname(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2453 sysctl_domainname
2454 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2455 {
2456 int error, changed;
2457 char tmpname[MAXHOSTNAMELEN] = {};
2458
2459 lck_mtx_lock(&domainname_lock);
2460 strlcpy(tmpname, domainname, sizeof(tmpname));
2461 lck_mtx_unlock(&domainname_lock);
2462
2463 error = sysctl_io_string(req, tmpname, sizeof(tmpname), 0, &changed);
2464 if (!error && changed) {
2465 lck_mtx_lock(&hostname_lock);
2466 strlcpy(domainname, tmpname, sizeof(domainname));
2467 lck_mtx_unlock(&hostname_lock);
2468 }
2469 return error;
2470 }
2471
2472 SYSCTL_PROC(_kern, KERN_DOMAINNAME, nisdomainname,
2473 CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_LOCKED,
2474 0, 0, sysctl_domainname, "A", "");
2475
2476 SYSCTL_COMPAT_INT(_kern, KERN_HOSTID, hostid,
2477 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2478 &hostid, 0, "");
2479
2480 STATIC int
sysctl_hostname(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2481 sysctl_hostname
2482 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2483 {
2484 int error, changed;
2485 char tmpname[MAXHOSTNAMELEN] = {};
2486
2487 lck_mtx_lock(&hostname_lock);
2488 strlcpy(tmpname, hostname, sizeof(tmpname));
2489 lck_mtx_unlock(&hostname_lock);
2490
2491 error = sysctl_io_string(req, tmpname, sizeof(tmpname), 1, &changed);
2492 if (!error && changed) {
2493 lck_mtx_lock(&hostname_lock);
2494 strlcpy(hostname, tmpname, sizeof(hostname));
2495 lck_mtx_unlock(&hostname_lock);
2496 }
2497 return error;
2498 }
2499
2500 SYSCTL_PROC(_kern, KERN_HOSTNAME, hostname,
2501 CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_LOCKED,
2502 0, 0, sysctl_hostname, "A", "");
2503
2504 STATIC int
sysctl_procname(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2505 sysctl_procname
2506 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2507 {
2508 /* Original code allowed writing, I'm copying this, although this all makes
2509 * no sense to me. Besides, this sysctl is never used. */
2510 return sysctl_io_string(req, &req->p->p_name[0], (2 * MAXCOMLEN + 1), 1, NULL);
2511 }
2512
2513 SYSCTL_PROC(_kern, KERN_PROCNAME, procname,
2514 CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
2515 0, 0, sysctl_procname, "A", "");
2516
2517 SYSCTL_INT(_kern, KERN_SPECULATIVE_READS, speculative_reads_disabled,
2518 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2519 &speculative_reads_disabled, 0, "");
2520
2521 SYSCTL_UINT(_kern, OID_AUTO, preheat_max_bytes,
2522 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2523 &preheat_max_bytes, 0, "");
2524
2525 SYSCTL_UINT(_kern, OID_AUTO, preheat_min_bytes,
2526 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2527 &preheat_min_bytes, 0, "");
2528
2529 SYSCTL_UINT(_kern, OID_AUTO, speculative_prefetch_max,
2530 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2531 &speculative_prefetch_max, 0, "");
2532
2533 SYSCTL_UINT(_kern, OID_AUTO, speculative_prefetch_max_iosize,
2534 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2535 &speculative_prefetch_max_iosize, 0, "");
2536
2537 SYSCTL_UINT(_kern, OID_AUTO, vm_page_free_target,
2538 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2539 &vm_page_free_target, 0, "");
2540
2541 SYSCTL_UINT(_kern, OID_AUTO, vm_page_free_min,
2542 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2543 &vm_page_free_min, 0, "");
2544
2545 SYSCTL_UINT(_kern, OID_AUTO, vm_page_free_reserved,
2546 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2547 &vm_page_free_reserved, 0, "");
2548
2549 SYSCTL_UINT(_kern, OID_AUTO, vm_page_speculative_percentage,
2550 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2551 &vm_pageout_state.vm_page_speculative_percentage, 0, "");
2552
2553 SYSCTL_UINT(_kern, OID_AUTO, vm_page_speculative_q_age_ms,
2554 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2555 &vm_pageout_state.vm_page_speculative_q_age_ms, 0, "");
2556
2557 SYSCTL_UINT(_kern, OID_AUTO, vm_max_delayed_work_limit,
2558 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2559 &vm_max_delayed_work_limit, 0, "");
2560
2561 SYSCTL_UINT(_kern, OID_AUTO, vm_max_batch,
2562 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2563 &vm_max_batch, 0, "");
2564
2565 SYSCTL_STRING(_kern, OID_AUTO, bootsessionuuid,
2566 CTLFLAG_RD | CTLFLAG_LOCKED,
2567 &bootsessionuuid_string, sizeof(bootsessionuuid_string), "");
2568
2569
2570 STATIC int
sysctl_boottime(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2571 sysctl_boottime
2572 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2573 {
2574 struct timeval tv;
2575 boottime_timeval(&tv);
2576 struct proc *p = req->p;
2577
2578 if (proc_is64bit(p)) {
2579 struct user64_timeval t = {};
2580 t.tv_sec = tv.tv_sec;
2581 t.tv_usec = tv.tv_usec;
2582 return sysctl_io_opaque(req, &t, sizeof(t), NULL);
2583 } else {
2584 struct user32_timeval t = {};
2585 t.tv_sec = (user32_time_t)tv.tv_sec;
2586 t.tv_usec = tv.tv_usec;
2587 return sysctl_io_opaque(req, &t, sizeof(t), NULL);
2588 }
2589 }
2590
2591 SYSCTL_PROC(_kern, KERN_BOOTTIME, boottime,
2592 CTLTYPE_STRUCT | CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
2593 0, 0, sysctl_boottime, "S,timeval", "");
2594
2595 extern bool IOGetBootUUID(char *);
2596
2597 /* non-static: written by imageboot.c */
2598 uuid_string_t fake_bootuuid;
2599
2600 STATIC int
sysctl_bootuuid(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2601 sysctl_bootuuid
2602 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2603 {
2604 int error = ENOENT;
2605
2606 /* check the first byte to see if the string has been
2607 * populated. this is a uuid_STRING_t, this check would
2608 * not work with a uuid_t.
2609 */
2610 if (fake_bootuuid[0] != '\0') {
2611 error = sysctl_io_string(req, fake_bootuuid, 0, 0, NULL);
2612 goto out;
2613 }
2614
2615 uuid_string_t uuid_string;
2616 if (IOGetBootUUID(uuid_string)) {
2617 uuid_t boot_uuid;
2618 error = uuid_parse(uuid_string, boot_uuid);
2619 if (!error) {
2620 error = sysctl_io_string(req, __DECONST(char *, uuid_string), 0, 0, NULL);
2621 }
2622 }
2623
2624 out:
2625 return error;
2626 }
2627
2628 SYSCTL_PROC(_kern, OID_AUTO, bootuuid,
2629 CTLTYPE_STRING | CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
2630 0, 0, sysctl_bootuuid, "A", "");
2631
2632
2633 extern bool IOGetApfsPrebootUUID(char *);
2634 extern bool IOGetAssociatedApfsVolgroupUUID(char *);
2635
2636 STATIC int
sysctl_apfsprebootuuid(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2637 sysctl_apfsprebootuuid
2638 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2639 {
2640 int error = ENOENT;
2641
2642 uuid_string_t uuid_string;
2643 if (IOGetApfsPrebootUUID(uuid_string)) {
2644 uuid_t apfs_preboot_uuid;
2645 error = uuid_parse(uuid_string, apfs_preboot_uuid);
2646 if (!error) {
2647 error = sysctl_io_string(req, __DECONST(char *, uuid_string), 0, 0, NULL);
2648 }
2649 }
2650
2651 return error;
2652 }
2653
2654 SYSCTL_PROC(_kern, OID_AUTO, apfsprebootuuid,
2655 CTLTYPE_STRING | CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
2656 0, 0, sysctl_apfsprebootuuid, "A", "");
2657
2658 STATIC int
sysctl_targetsystemvolgroupuuid(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2659 sysctl_targetsystemvolgroupuuid
2660 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2661 {
2662 int error = ENOENT;
2663
2664 uuid_string_t uuid_string;
2665 if (IOGetApfsPrebootUUID(uuid_string)) {
2666 uuid_t apfs_preboot_uuid;
2667 error = uuid_parse(uuid_string, apfs_preboot_uuid);
2668 if (!error) {
2669 error = sysctl_io_string(req, __DECONST(char *, uuid_string), 0, 0, NULL);
2670 }
2671 } else {
2672 /*
2673 * In special boot modes, such as kcgen-mode, the
2674 * apfs-preboot-uuid property will not be set. Instead, a
2675 * different property, associated-volume-group, will be set
2676 * which indicates the UUID of the VolumeGroup containing the
2677 * system volume into which you will boot.
2678 */
2679 if (IOGetAssociatedApfsVolgroupUUID(uuid_string)) {
2680 uuid_t apfs_preboot_uuid;
2681 error = uuid_parse(uuid_string, apfs_preboot_uuid);
2682 if (!error) {
2683 error = sysctl_io_string(req, __DECONST(char *, uuid_string), 0, 0, NULL);
2684 }
2685 }
2686 }
2687
2688 return error;
2689 }
2690
2691 SYSCTL_PROC(_kern, OID_AUTO, targetsystemvolgroupuuid,
2692 CTLTYPE_STRING | CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_MASKED | CTLFLAG_LOCKED,
2693 0, 0, sysctl_targetsystemvolgroupuuid, "A", "");
2694
2695 STATIC int
sysctl_symfile(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2696 sysctl_symfile
2697 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2698 {
2699 char *str;
2700 int error = get_kernel_symfile(req->p, &str);
2701 if (error) {
2702 return error;
2703 }
2704 return sysctl_io_string(req, str, 0, 0, NULL);
2705 }
2706
2707
2708 SYSCTL_PROC(_kern, KERN_SYMFILE, symfile,
2709 CTLTYPE_STRING | CTLFLAG_RD | CTLFLAG_LOCKED,
2710 0, 0, sysctl_symfile, "A", "");
2711
2712 #if CONFIG_NFS_CLIENT
2713 STATIC int
sysctl_netboot(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2714 sysctl_netboot
2715 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2716 {
2717 return sysctl_io_number(req, netboot_root(), sizeof(int), NULL, NULL);
2718 }
2719
2720 SYSCTL_PROC(_kern, KERN_NETBOOT, netboot,
2721 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2722 0, 0, sysctl_netboot, "I", "");
2723 #endif
2724
2725 #ifdef CONFIG_IMGSRC_ACCESS
2726 /*
2727 * Legacy--act as if only one layer of nesting is possible.
2728 */
2729 STATIC int
sysctl_imgsrcdev(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2730 sysctl_imgsrcdev
2731 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2732 {
2733 vfs_context_t ctx = vfs_context_current();
2734 vnode_t devvp;
2735 int result;
2736
2737 if (!vfs_context_issuser(ctx)) {
2738 return EPERM;
2739 }
2740
2741 if (imgsrc_rootvnodes[0] == NULL) {
2742 return ENOENT;
2743 }
2744
2745 result = vnode_getwithref(imgsrc_rootvnodes[0]);
2746 if (result != 0) {
2747 return result;
2748 }
2749
2750 devvp = vnode_mount(imgsrc_rootvnodes[0])->mnt_devvp;
2751 result = vnode_getwithref(devvp);
2752 if (result != 0) {
2753 goto out;
2754 }
2755
2756 result = sysctl_io_number(req, vnode_specrdev(devvp), sizeof(dev_t), NULL, NULL);
2757
2758 vnode_put(devvp);
2759 out:
2760 vnode_put(imgsrc_rootvnodes[0]);
2761 return result;
2762 }
2763
2764 SYSCTL_PROC(_kern, OID_AUTO, imgsrcdev,
2765 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2766 0, 0, sysctl_imgsrcdev, "I", "");
2767
2768 STATIC int
sysctl_imgsrcinfo(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2769 sysctl_imgsrcinfo
2770 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2771 {
2772 int error;
2773 struct imgsrc_info info[MAX_IMAGEBOOT_NESTING] = {}; /* 2 for now, no problem */
2774 uint32_t i;
2775 vnode_t rvp, devvp;
2776
2777 if (imgsrc_rootvnodes[0] == NULLVP) {
2778 return ENXIO;
2779 }
2780
2781 for (i = 0; i < MAX_IMAGEBOOT_NESTING; i++) {
2782 /*
2783 * Go get the root vnode.
2784 */
2785 rvp = imgsrc_rootvnodes[i];
2786 if (rvp == NULLVP) {
2787 break;
2788 }
2789
2790 error = vnode_get(rvp);
2791 if (error != 0) {
2792 return error;
2793 }
2794
2795 /*
2796 * For now, no getting at a non-local volume.
2797 */
2798 devvp = vnode_mount(rvp)->mnt_devvp;
2799 if (devvp == NULL) {
2800 vnode_put(rvp);
2801 return EINVAL;
2802 }
2803
2804 error = vnode_getwithref(devvp);
2805 if (error != 0) {
2806 vnode_put(rvp);
2807 return error;
2808 }
2809
2810 /*
2811 * Fill in info.
2812 */
2813 info[i].ii_dev = vnode_specrdev(devvp);
2814 info[i].ii_flags = 0;
2815 info[i].ii_height = i;
2816 bzero(info[i].ii_reserved, sizeof(info[i].ii_reserved));
2817
2818 vnode_put(devvp);
2819 vnode_put(rvp);
2820 }
2821
2822 return sysctl_io_opaque(req, info, i * sizeof(info[0]), NULL);
2823 }
2824
2825 SYSCTL_PROC(_kern, OID_AUTO, imgsrcinfo,
2826 CTLTYPE_OPAQUE | CTLFLAG_RD | CTLFLAG_LOCKED,
2827 0, 0, sysctl_imgsrcinfo, "I", "");
2828
2829 #endif /* CONFIG_IMGSRC_ACCESS */
2830
2831
2832 SYSCTL_DECL(_kern_timer);
2833 SYSCTL_NODE(_kern, OID_AUTO, timer, CTLFLAG_RW | CTLFLAG_LOCKED, 0, "timer");
2834
2835
2836 SYSCTL_INT(_kern_timer, OID_AUTO, coalescing_enabled,
2837 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2838 &mach_timer_coalescing_enabled, 0, "");
2839
2840 SYSCTL_QUAD(_kern_timer, OID_AUTO, deadline_tracking_bin_1,
2841 CTLFLAG_RW | CTLFLAG_LOCKED,
2842 &timer_deadline_tracking_bin_1, "");
2843 SYSCTL_QUAD(_kern_timer, OID_AUTO, deadline_tracking_bin_2,
2844 CTLFLAG_RW | CTLFLAG_LOCKED,
2845 &timer_deadline_tracking_bin_2, "");
2846
2847 SYSCTL_DECL(_kern_timer_longterm);
2848 SYSCTL_NODE(_kern_timer, OID_AUTO, longterm, CTLFLAG_RW | CTLFLAG_LOCKED, 0, "longterm");
2849
2850
2851 /* Must match definition in osfmk/kern/timer_call.c */
2852 enum {
2853 THRESHOLD, QCOUNT,
2854 ENQUEUES, DEQUEUES, ESCALATES, SCANS, PREEMPTS,
2855 LATENCY, LATENCY_MIN, LATENCY_MAX, SCAN_LIMIT, SCAN_INTERVAL, PAUSES
2856 };
2857 extern uint64_t timer_sysctl_get(int);
2858 extern int timer_sysctl_set(int, uint64_t);
2859
2860 STATIC int
sysctl_timer(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2861 sysctl_timer
2862 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2863 {
2864 int oid = (int)arg1;
2865 uint64_t value = timer_sysctl_get(oid);
2866 uint64_t new_value;
2867 int error;
2868 int changed;
2869
2870 error = sysctl_io_number(req, value, sizeof(value), &new_value, &changed);
2871 if (changed) {
2872 error = timer_sysctl_set(oid, new_value);
2873 }
2874
2875 return error;
2876 }
2877
2878 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, threshold,
2879 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
2880 (void *) THRESHOLD, 0, sysctl_timer, "Q", "");
2881 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, scan_limit,
2882 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
2883 (void *) SCAN_LIMIT, 0, sysctl_timer, "Q", "");
2884 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, scan_interval,
2885 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
2886 (void *) SCAN_INTERVAL, 0, sysctl_timer, "Q", "");
2887
2888 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, qlen,
2889 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
2890 (void *) QCOUNT, 0, sysctl_timer, "Q", "");
2891 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, scan_pauses,
2892 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
2893 (void *) PAUSES, 0, sysctl_timer, "Q", "");
2894
2895 #if DEBUG
2896 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, enqueues,
2897 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
2898 (void *) ENQUEUES, 0, sysctl_timer, "Q", "");
2899 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, dequeues,
2900 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
2901 (void *) DEQUEUES, 0, sysctl_timer, "Q", "");
2902 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, escalates,
2903 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
2904 (void *) ESCALATES, 0, sysctl_timer, "Q", "");
2905 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, scans,
2906 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
2907 (void *) SCANS, 0, sysctl_timer, "Q", "");
2908 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, preempts,
2909 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
2910 (void *) PREEMPTS, 0, sysctl_timer, "Q", "");
2911 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, latency,
2912 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
2913 (void *) LATENCY, 0, sysctl_timer, "Q", "");
2914 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, latency_min,
2915 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
2916 (void *) LATENCY_MIN, 0, sysctl_timer, "Q", "");
2917 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, latency_max,
2918 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
2919 (void *) LATENCY_MAX, 0, sysctl_timer, "Q", "");
2920 #endif /* DEBUG */
2921
2922 STATIC int
sysctl_usrstack(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2923 sysctl_usrstack
2924 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2925 {
2926 return sysctl_io_number(req, (int)req->p->user_stack, sizeof(int), NULL, NULL);
2927 }
2928
2929 SYSCTL_PROC(_kern, KERN_USRSTACK32, usrstack,
2930 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2931 0, 0, sysctl_usrstack, "I", "");
2932
2933 STATIC int
sysctl_usrstack64(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2934 sysctl_usrstack64
2935 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2936 {
2937 return sysctl_io_number(req, req->p->user_stack, sizeof(req->p->user_stack), NULL, NULL);
2938 }
2939
2940 SYSCTL_PROC(_kern, KERN_USRSTACK64, usrstack64,
2941 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
2942 0, 0, sysctl_usrstack64, "Q", "");
2943
2944 #if CONFIG_COREDUMP
2945
2946 SYSCTL_STRING(_kern, KERN_COREFILE, corefile,
2947 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2948 corefilename, sizeof(corefilename), "");
2949
2950 STATIC int
sysctl_coredump(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2951 sysctl_coredump
2952 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2953 {
2954 #ifdef SECURE_KERNEL
2955 (void)req;
2956 return ENOTSUP;
2957 #else
2958 int new_value, changed;
2959 int error = sysctl_io_number(req, do_coredump, sizeof(int), &new_value, &changed);
2960 if (changed) {
2961 if ((new_value == 0) || (new_value == 1)) {
2962 do_coredump = new_value;
2963 } else {
2964 error = EINVAL;
2965 }
2966 }
2967 return error;
2968 #endif
2969 }
2970
2971 SYSCTL_PROC(_kern, KERN_COREDUMP, coredump,
2972 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2973 0, 0, sysctl_coredump, "I", "");
2974
2975 STATIC int
sysctl_suid_coredump(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2976 sysctl_suid_coredump
2977 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2978 {
2979 #ifdef SECURE_KERNEL
2980 (void)req;
2981 return ENOTSUP;
2982 #else
2983 int new_value, changed;
2984 int error = sysctl_io_number(req, sugid_coredump, sizeof(int), &new_value, &changed);
2985 if (changed) {
2986 if ((new_value == 0) || (new_value == 1)) {
2987 sugid_coredump = new_value;
2988 } else {
2989 error = EINVAL;
2990 }
2991 }
2992 return error;
2993 #endif
2994 }
2995
2996 SYSCTL_PROC(_kern, KERN_SUGID_COREDUMP, sugid_coredump,
2997 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2998 0, 0, sysctl_suid_coredump, "I", "");
2999
3000 #endif /* CONFIG_COREDUMP */
3001
3002 STATIC int
sysctl_delayterm(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3003 sysctl_delayterm
3004 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3005 {
3006 struct proc *p = req->p;
3007 int new_value, changed;
3008 int error = sysctl_io_number(req, (req->p->p_lflag & P_LDELAYTERM)? 1: 0, sizeof(int), &new_value, &changed);
3009 if (changed) {
3010 proc_lock(p);
3011 if (new_value) {
3012 req->p->p_lflag |= P_LDELAYTERM;
3013 } else {
3014 req->p->p_lflag &= ~P_LDELAYTERM;
3015 }
3016 proc_unlock(p);
3017 }
3018 return error;
3019 }
3020
3021 SYSCTL_PROC(_kern, KERN_PROCDELAYTERM, delayterm,
3022 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
3023 0, 0, sysctl_delayterm, "I", "");
3024
3025
3026 STATIC int
sysctl_rage_vnode(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3027 sysctl_rage_vnode
3028 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3029 {
3030 struct proc *p = req->p;
3031 struct uthread *ut;
3032 int new_value, old_value, changed;
3033 int error;
3034
3035 ut = current_uthread();
3036
3037 if (ut->uu_flag & UT_RAGE_VNODES) {
3038 old_value = KERN_RAGE_THREAD;
3039 } else if (p->p_lflag & P_LRAGE_VNODES) {
3040 old_value = KERN_RAGE_PROC;
3041 } else {
3042 old_value = 0;
3043 }
3044
3045 error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
3046
3047 if ((error == 0) && (changed != 0)) {
3048 switch (new_value) {
3049 case KERN_RAGE_PROC:
3050 proc_lock(p);
3051 p->p_lflag |= P_LRAGE_VNODES;
3052 proc_unlock(p);
3053 break;
3054 case KERN_UNRAGE_PROC:
3055 proc_lock(p);
3056 p->p_lflag &= ~P_LRAGE_VNODES;
3057 proc_unlock(p);
3058 break;
3059
3060 case KERN_RAGE_THREAD:
3061 ut->uu_flag |= UT_RAGE_VNODES;
3062 break;
3063 case KERN_UNRAGE_THREAD:
3064 ut = current_uthread();
3065 ut->uu_flag &= ~UT_RAGE_VNODES;
3066 break;
3067 }
3068 }
3069 return error;
3070 }
3071
3072 SYSCTL_PROC(_kern, KERN_RAGEVNODE, rage_vnode,
3073 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
3074 0, 0, sysctl_rage_vnode, "I", "");
3075
3076 /* XXX until filecoordinationd fixes a bit of inverted logic. */
3077 STATIC int
sysctl_vfsnspace(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3078 sysctl_vfsnspace
3079 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3080 {
3081 int old_value = 0, new_value, changed;
3082
3083 return sysctl_io_number(req, old_value, sizeof(int), &new_value,
3084 &changed);
3085 }
3086
3087 SYSCTL_PROC(_kern, OID_AUTO, vfsnspace,
3088 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
3089 0, 0, sysctl_vfsnspace, "I", "");
3090
3091 /* XXX move this interface into libproc and remove this sysctl */
3092 STATIC int
sysctl_setthread_cpupercent(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3093 sysctl_setthread_cpupercent
3094 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3095 {
3096 int new_value, old_value;
3097 int error = 0;
3098 kern_return_t kret = KERN_SUCCESS;
3099 uint8_t percent = 0;
3100 int ms_refill = 0;
3101
3102 if (!req->newptr) {
3103 return 0;
3104 }
3105
3106 old_value = 0;
3107
3108 if ((error = sysctl_io_number(req, old_value, sizeof(old_value), &new_value, NULL)) != 0) {
3109 return error;
3110 }
3111
3112 percent = new_value & 0xff; /* low 8 bytes for perent */
3113 ms_refill = (new_value >> 8) & 0xffffff; /* upper 24bytes represent ms refill value */
3114 if (percent > 100) {
3115 return EINVAL;
3116 }
3117
3118 /*
3119 * If the caller is specifying a percentage of 0, this will unset the CPU limit, if present.
3120 */
3121 if ((kret = thread_set_cpulimit(THREAD_CPULIMIT_BLOCK, percent, ms_refill * (int)NSEC_PER_MSEC)) != 0) {
3122 return EIO;
3123 }
3124
3125 return 0;
3126 }
3127
3128 SYSCTL_PROC(_kern, OID_AUTO, setthread_cpupercent,
3129 CTLTYPE_INT | CTLFLAG_WR | CTLFLAG_ANYBODY,
3130 0, 0, sysctl_setthread_cpupercent, "I", "set thread cpu percentage limit");
3131
3132
3133 STATIC int
sysctl_kern_check_openevt(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3134 sysctl_kern_check_openevt
3135 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3136 {
3137 struct proc *p = req->p;
3138 int new_value, old_value, changed;
3139 int error;
3140
3141 if (p->p_flag & P_CHECKOPENEVT) {
3142 old_value = KERN_OPENEVT_PROC;
3143 } else {
3144 old_value = 0;
3145 }
3146
3147 error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
3148
3149 if ((error == 0) && (changed != 0)) {
3150 switch (new_value) {
3151 case KERN_OPENEVT_PROC:
3152 OSBitOrAtomic(P_CHECKOPENEVT, &p->p_flag);
3153 break;
3154
3155 case KERN_UNOPENEVT_PROC:
3156 OSBitAndAtomic(~((uint32_t)P_CHECKOPENEVT), &p->p_flag);
3157 break;
3158
3159 default:
3160 error = EINVAL;
3161 }
3162 }
3163 return error;
3164 }
3165
3166 SYSCTL_PROC(_kern, KERN_CHECKOPENEVT, check_openevt, CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
3167 0, 0, sysctl_kern_check_openevt, "I", "set the per-process check-open-evt flag");
3168
3169
3170 #if DEVELOPMENT || DEBUG
3171 STATIC int
sysctl_nx(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3172 sysctl_nx
3173 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3174 {
3175 #ifdef SECURE_KERNEL
3176 (void)req;
3177 return ENOTSUP;
3178 #else
3179 int new_value, changed;
3180 int error;
3181
3182 error = sysctl_io_number(req, nx_enabled, sizeof(nx_enabled), &new_value, &changed);
3183 if (error) {
3184 return error;
3185 }
3186
3187 if (changed) {
3188 #if defined(__x86_64__)
3189 /*
3190 * Only allow setting if NX is supported on the chip
3191 */
3192 if (!(cpuid_extfeatures() & CPUID_EXTFEATURE_XD)) {
3193 return ENOTSUP;
3194 }
3195 #endif
3196 nx_enabled = new_value;
3197 }
3198 return error;
3199 #endif /* SECURE_KERNEL */
3200 }
3201 #endif
3202
3203 #if DEVELOPMENT || DEBUG
3204 SYSCTL_PROC(_kern, KERN_NX_PROTECTION, nx,
3205 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
3206 0, 0, sysctl_nx, "I", "");
3207 #endif
3208
3209 STATIC int
sysctl_loadavg(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3210 sysctl_loadavg
3211 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3212 {
3213 if (proc_is64bit(req->p)) {
3214 struct user64_loadavg loadinfo64 = {};
3215 fill_loadavg64(&averunnable, &loadinfo64);
3216 return sysctl_io_opaque(req, &loadinfo64, sizeof(loadinfo64), NULL);
3217 } else {
3218 struct user32_loadavg loadinfo32 = {};
3219 fill_loadavg32(&averunnable, &loadinfo32);
3220 return sysctl_io_opaque(req, &loadinfo32, sizeof(loadinfo32), NULL);
3221 }
3222 }
3223
3224 SYSCTL_PROC(_vm, VM_LOADAVG, loadavg,
3225 CTLTYPE_STRUCT | CTLFLAG_RD | CTLFLAG_LOCKED,
3226 0, 0, sysctl_loadavg, "S,loadavg", "");
3227
3228 /*
3229 * Note: Thread safe; vm_map_lock protects in vm_toggle_entry_reuse()
3230 */
3231 STATIC int
sysctl_vm_toggle_address_reuse(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3232 sysctl_vm_toggle_address_reuse(__unused struct sysctl_oid *oidp, __unused void *arg1,
3233 __unused int arg2, struct sysctl_req *req)
3234 {
3235 int old_value = 0, new_value = 0, error = 0;
3236
3237 if (vm_toggle_entry_reuse( VM_TOGGLE_GETVALUE, &old_value )) {
3238 return error;
3239 }
3240 error = sysctl_io_number(req, old_value, sizeof(int), &new_value, NULL);
3241 if (!error) {
3242 return vm_toggle_entry_reuse(new_value, NULL);
3243 }
3244 return error;
3245 }
3246
3247 SYSCTL_PROC(_debug, OID_AUTO, toggle_address_reuse, CTLFLAG_ANYBODY | CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED, 0, 0, sysctl_vm_toggle_address_reuse, "I", "");
3248
3249 #ifdef CONFIG_XNUPOST
3250
3251 extern uint32_t xnupost_get_estimated_testdata_size(void);
3252 extern int xnupost_reset_all_tests(void);
3253
3254 STATIC int
3255 sysctl_handle_xnupost_get_tests SYSCTL_HANDLER_ARGS
3256 {
3257 /* fixup unused arguments warnings */
3258 __unused int _oa2 = arg2;
3259 __unused void * _oa1 = arg1;
3260 __unused struct sysctl_oid * _oidp = oidp;
3261
3262 int error = 0;
3263 user_addr_t oldp = 0;
3264 user_addr_t newp = 0;
3265 uint32_t usedbytes = 0;
3266
3267 oldp = req->oldptr;
3268 newp = req->newptr;
3269
3270 if (newp) {
3271 return ENOTSUP;
3272 }
3273
3274 if ((void *)oldp == NULL) {
3275 /* return estimated size for second call where info can be placed */
3276 req->oldidx = xnupost_get_estimated_testdata_size();
3277 } else {
3278 error = xnupost_export_testdata((void *)oldp, req->oldlen, &usedbytes);
3279 req->oldidx = usedbytes;
3280 }
3281
3282 return error;
3283 }
3284
3285 SYSCTL_PROC(_debug,
3286 OID_AUTO,
3287 xnupost_get_tests,
3288 CTLFLAG_MASKED | CTLFLAG_ANYBODY | CTLTYPE_OPAQUE | CTLFLAG_RD | CTLFLAG_LOCKED,
3289 0,
3290 0,
3291 sysctl_handle_xnupost_get_tests,
3292 "-",
3293 "read xnupost test data in kernel");
3294
3295 STATIC int
3296 sysctl_debug_xnupost_ctl SYSCTL_HANDLER_ARGS
3297 {
3298 /* fixup unused arguments warnings */
3299 __unused int _oa2 = arg2;
3300 __unused void * _oa1 = arg1;
3301 __unused struct sysctl_oid * _oidp = oidp;
3302
3303 #define ARRCOUNT 4
3304 /*
3305 * INPUT: ACTION, PARAM1, PARAM2, PARAM3
3306 * OUTPUT: RESULTCODE, ADDITIONAL DATA
3307 */
3308 int32_t outval[ARRCOUNT] = {0};
3309 int32_t input[ARRCOUNT] = {0};
3310 int32_t out_size = sizeof(outval);
3311 int32_t in_size = sizeof(input);
3312 int error = 0;
3313
3314 /* if this is NULL call to find out size, send out size info */
3315 if (!req->newptr) {
3316 goto out;
3317 }
3318
3319 /* pull in provided value from userspace */
3320 error = SYSCTL_IN(req, &input[0], in_size);
3321 if (error) {
3322 return error;
3323 }
3324
3325 if (input[0] == XTCTL_RESET_TESTDATA) {
3326 outval[0] = xnupost_reset_all_tests();
3327 goto out;
3328 }
3329
3330 out:
3331 error = SYSCTL_OUT(req, &outval[0], out_size);
3332 return error;
3333 }
3334
3335 SYSCTL_PROC(_debug,
3336 OID_AUTO,
3337 xnupost_testctl,
3338 CTLFLAG_MASKED | CTLFLAG_ANYBODY | CTLTYPE_OPAQUE | CTLFLAG_RW | CTLFLAG_LOCKED,
3339 0,
3340 0,
3341 sysctl_debug_xnupost_ctl,
3342 "I",
3343 "xnupost control for kernel testing");
3344
3345 extern void test_oslog_handleOSLogCtl(int32_t * in, int32_t * out, int32_t arraycount);
3346
3347 STATIC int
sysctl_debug_test_oslog_ctl(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3348 sysctl_debug_test_oslog_ctl(__unused struct sysctl_oid * oidp, __unused void * arg1, __unused int arg2, struct sysctl_req * req)
3349 {
3350 #define ARRCOUNT 4
3351 int32_t outval[ARRCOUNT] = {0};
3352 int32_t input[ARRCOUNT] = {0};
3353 int32_t size_outval = sizeof(outval);
3354 int32_t size_inval = sizeof(input);
3355 int32_t error;
3356
3357 /* if this is NULL call to find out size, send out size info */
3358 if (!req->newptr) {
3359 error = SYSCTL_OUT(req, &outval[0], size_outval);
3360 return error;
3361 }
3362
3363 /* pull in provided value from userspace */
3364 error = SYSCTL_IN(req, &input[0], size_inval);
3365 if (error) {
3366 return error;
3367 }
3368
3369 test_oslog_handleOSLogCtl(input, outval, ARRCOUNT);
3370
3371 error = SYSCTL_OUT(req, &outval[0], size_outval);
3372
3373 return error;
3374 }
3375
3376 SYSCTL_PROC(_debug,
3377 OID_AUTO,
3378 test_OSLogCtl,
3379 CTLFLAG_MASKED | CTLFLAG_ANYBODY | CTLTYPE_OPAQUE | CTLFLAG_RW | CTLFLAG_LOCKED,
3380 0,
3381 0,
3382 sysctl_debug_test_oslog_ctl,
3383 "I",
3384 "testing oslog in kernel");
3385
3386 #include <mach/task.h>
3387 #include <mach/semaphore.h>
3388
3389 static LCK_GRP_DECLARE(sysctl_debug_test_stackshot_owner_grp, "test-stackshot-owner-grp");
3390 static LCK_MTX_DECLARE(sysctl_debug_test_stackshot_owner_init_mtx,
3391 &sysctl_debug_test_stackshot_owner_grp);
3392
3393 /* This is a sysctl for testing collection of owner info on a lock in kernel space. A multi-threaded
3394 * test from userland sets this sysctl in such a way that a thread blocks in kernel mode, and a
3395 * stackshot is taken to see if the owner of the lock can be identified.
3396 *
3397 * We can't return to userland with a kernel lock held, so be sure to unlock before we leave.
3398 * the semaphores allow us to artificially create cases where the lock is being held and the
3399 * thread is hanging / taking a long time to do something. */
3400
3401 volatile char sysctl_debug_test_stackshot_mtx_inited = 0;
3402 semaphore_t sysctl_debug_test_stackshot_mutex_sem;
3403 lck_mtx_t sysctl_debug_test_stackshot_owner_lck;
3404
3405 #define SYSCTL_DEBUG_MTX_ACQUIRE_WAIT 1
3406 #define SYSCTL_DEBUG_MTX_ACQUIRE_NOWAIT 2
3407 #define SYSCTL_DEBUG_MTX_SIGNAL 3
3408 #define SYSCTL_DEBUG_MTX_TEARDOWN 4
3409
3410 STATIC int
sysctl_debug_test_stackshot_mutex_owner(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3411 sysctl_debug_test_stackshot_mutex_owner(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3412 {
3413 long long option = -1;
3414 /* if the user tries to read the sysctl, we tell them what the address of the lock is (to test against stackshot's output) */
3415 long long mtx_unslid_addr = (long long)VM_KERNEL_UNSLIDE_OR_PERM(&sysctl_debug_test_stackshot_owner_lck);
3416 int error = sysctl_io_number(req, mtx_unslid_addr, sizeof(long long), (void*)&option, NULL);
3417
3418 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_init_mtx);
3419 if (!sysctl_debug_test_stackshot_mtx_inited) {
3420 lck_mtx_init(&sysctl_debug_test_stackshot_owner_lck,
3421 &sysctl_debug_test_stackshot_owner_grp,
3422 LCK_ATTR_NULL);
3423 semaphore_create(kernel_task,
3424 &sysctl_debug_test_stackshot_mutex_sem,
3425 SYNC_POLICY_FIFO, 0);
3426 sysctl_debug_test_stackshot_mtx_inited = 1;
3427 }
3428 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_init_mtx);
3429
3430 if (!error) {
3431 switch (option) {
3432 case SYSCTL_DEBUG_MTX_ACQUIRE_NOWAIT:
3433 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_lck);
3434 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_lck);
3435 break;
3436 case SYSCTL_DEBUG_MTX_ACQUIRE_WAIT:
3437 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_lck);
3438 semaphore_wait(sysctl_debug_test_stackshot_mutex_sem);
3439 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_lck);
3440 break;
3441 case SYSCTL_DEBUG_MTX_SIGNAL:
3442 semaphore_signal(sysctl_debug_test_stackshot_mutex_sem);
3443 break;
3444 case SYSCTL_DEBUG_MTX_TEARDOWN:
3445 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_init_mtx);
3446
3447 lck_mtx_destroy(&sysctl_debug_test_stackshot_owner_lck,
3448 &sysctl_debug_test_stackshot_owner_grp);
3449 semaphore_destroy(kernel_task,
3450 sysctl_debug_test_stackshot_mutex_sem);
3451 sysctl_debug_test_stackshot_mtx_inited = 0;
3452
3453 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_init_mtx);
3454 break;
3455 case -1: /* user just wanted to read the value, so do nothing */
3456 break;
3457 default:
3458 error = EINVAL;
3459 break;
3460 }
3461 }
3462 return error;
3463 }
3464
3465 /* we can't return to userland with a kernel rwlock held, so be sure to unlock before we leave.
3466 * the semaphores allow us to artificially create cases where the lock is being held and the
3467 * thread is hanging / taking a long time to do something. */
3468
3469 SYSCTL_PROC(_debug,
3470 OID_AUTO,
3471 test_MutexOwnerCtl,
3472 CTLFLAG_MASKED | CTLFLAG_ANYBODY | CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
3473 0,
3474 0,
3475 sysctl_debug_test_stackshot_mutex_owner,
3476 "-",
3477 "Testing mutex owner in kernel");
3478
3479 volatile char sysctl_debug_test_stackshot_rwlck_inited = 0;
3480 lck_rw_t sysctl_debug_test_stackshot_owner_rwlck;
3481 semaphore_t sysctl_debug_test_stackshot_rwlck_sem;
3482
3483 #define SYSCTL_DEBUG_KRWLCK_RACQUIRE_NOWAIT 1
3484 #define SYSCTL_DEBUG_KRWLCK_RACQUIRE_WAIT 2
3485 #define SYSCTL_DEBUG_KRWLCK_WACQUIRE_NOWAIT 3
3486 #define SYSCTL_DEBUG_KRWLCK_WACQUIRE_WAIT 4
3487 #define SYSCTL_DEBUG_KRWLCK_SIGNAL 5
3488 #define SYSCTL_DEBUG_KRWLCK_TEARDOWN 6
3489
3490 STATIC int
sysctl_debug_test_stackshot_rwlck_owner(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3491 sysctl_debug_test_stackshot_rwlck_owner(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3492 {
3493 long long option = -1;
3494 /* if the user tries to read the sysctl, we tell them what the address of the lock is
3495 * (to test against stackshot's output) */
3496 long long rwlck_unslid_addr = (long long)VM_KERNEL_UNSLIDE_OR_PERM(&sysctl_debug_test_stackshot_owner_rwlck);
3497 int error = sysctl_io_number(req, rwlck_unslid_addr, sizeof(long long), (void*)&option, NULL);
3498
3499 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_init_mtx);
3500 if (!sysctl_debug_test_stackshot_rwlck_inited) {
3501 lck_rw_init(&sysctl_debug_test_stackshot_owner_rwlck,
3502 &sysctl_debug_test_stackshot_owner_grp,
3503 LCK_ATTR_NULL);
3504 semaphore_create(kernel_task,
3505 &sysctl_debug_test_stackshot_rwlck_sem,
3506 SYNC_POLICY_FIFO,
3507 0);
3508 sysctl_debug_test_stackshot_rwlck_inited = 1;
3509 }
3510 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_init_mtx);
3511
3512 if (!error) {
3513 switch (option) {
3514 case SYSCTL_DEBUG_KRWLCK_RACQUIRE_NOWAIT:
3515 lck_rw_lock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_SHARED);
3516 lck_rw_unlock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_SHARED);
3517 break;
3518 case SYSCTL_DEBUG_KRWLCK_RACQUIRE_WAIT:
3519 lck_rw_lock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_SHARED);
3520 semaphore_wait(sysctl_debug_test_stackshot_rwlck_sem);
3521 lck_rw_unlock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_SHARED);
3522 break;
3523 case SYSCTL_DEBUG_KRWLCK_WACQUIRE_NOWAIT:
3524 lck_rw_lock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_EXCLUSIVE);
3525 lck_rw_unlock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_EXCLUSIVE);
3526 break;
3527 case SYSCTL_DEBUG_KRWLCK_WACQUIRE_WAIT:
3528 lck_rw_lock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_EXCLUSIVE);
3529 semaphore_wait(sysctl_debug_test_stackshot_rwlck_sem);
3530 lck_rw_unlock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_EXCLUSIVE);
3531 break;
3532 case SYSCTL_DEBUG_KRWLCK_SIGNAL:
3533 semaphore_signal(sysctl_debug_test_stackshot_rwlck_sem);
3534 break;
3535 case SYSCTL_DEBUG_KRWLCK_TEARDOWN:
3536 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_init_mtx);
3537
3538 lck_rw_destroy(&sysctl_debug_test_stackshot_owner_rwlck,
3539 &sysctl_debug_test_stackshot_owner_grp);
3540 semaphore_destroy(kernel_task,
3541 sysctl_debug_test_stackshot_rwlck_sem);
3542 sysctl_debug_test_stackshot_rwlck_inited = 0;
3543
3544 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_init_mtx);
3545 break;
3546 case -1: /* user just wanted to read the value, so do nothing */
3547 break;
3548 default:
3549 error = EINVAL;
3550 break;
3551 }
3552 }
3553 return error;
3554 }
3555
3556
3557 SYSCTL_PROC(_debug,
3558 OID_AUTO,
3559 test_RWLockOwnerCtl,
3560 CTLFLAG_MASKED | CTLFLAG_ANYBODY | CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
3561 0,
3562 0,
3563 sysctl_debug_test_stackshot_rwlck_owner,
3564 "-",
3565 "Testing rwlock owner in kernel");
3566 #endif /* !CONFIG_XNUPOST */
3567
3568 STATIC int
sysctl_swapusage(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3569 sysctl_swapusage
3570 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3571 {
3572 int error;
3573 uint64_t swap_total;
3574 uint64_t swap_avail;
3575 vm_size_t swap_pagesize;
3576 boolean_t swap_encrypted;
3577 struct xsw_usage xsu = {};
3578
3579 error = macx_swapinfo(&swap_total,
3580 &swap_avail,
3581 &swap_pagesize,
3582 &swap_encrypted);
3583 if (error) {
3584 return error;
3585 }
3586
3587 xsu.xsu_total = swap_total;
3588 xsu.xsu_avail = swap_avail;
3589 xsu.xsu_used = swap_total - swap_avail;
3590 xsu.xsu_pagesize = (u_int32_t)MIN(swap_pagesize, UINT32_MAX);
3591 xsu.xsu_encrypted = swap_encrypted;
3592 return sysctl_io_opaque(req, &xsu, sizeof(xsu), NULL);
3593 }
3594
3595
3596
3597 SYSCTL_PROC(_vm, VM_SWAPUSAGE, swapusage,
3598 CTLTYPE_STRUCT | CTLFLAG_RD | CTLFLAG_LOCKED,
3599 0, 0, sysctl_swapusage, "S,xsw_usage", "");
3600
3601 #if DEVELOPMENT || DEBUG
3602 extern int vm_num_swap_files_config;
3603 extern int vm_num_swap_files;
3604 extern lck_mtx_t vm_swap_data_lock;
3605 #define VM_MAX_SWAP_FILE_NUM 100
3606
3607 static int
3608 sysctl_vm_config_num_swap_files SYSCTL_HANDLER_ARGS
3609 {
3610 #pragma unused(arg1, arg2)
3611 int error = 0, val = vm_num_swap_files_config;
3612
3613 error = sysctl_handle_int(oidp, &val, 0, req);
3614 if (error || !req->newptr) {
3615 goto out;
3616 }
3617
3618 if (!VM_CONFIG_SWAP_IS_ACTIVE && !VM_CONFIG_FREEZER_SWAP_IS_ACTIVE) {
3619 printf("Swap is disabled\n");
3620 error = EINVAL;
3621 goto out;
3622 }
3623
3624 lck_mtx_lock(&vm_swap_data_lock);
3625
3626 if (val < vm_num_swap_files) {
3627 printf("Cannot configure fewer swap files than already exist.\n");
3628 error = EINVAL;
3629 lck_mtx_unlock(&vm_swap_data_lock);
3630 goto out;
3631 }
3632
3633 if (val > VM_MAX_SWAP_FILE_NUM) {
3634 printf("Capping number of swap files to upper bound.\n");
3635 val = VM_MAX_SWAP_FILE_NUM;
3636 }
3637
3638 vm_num_swap_files_config = val;
3639 lck_mtx_unlock(&vm_swap_data_lock);
3640 out:
3641
3642 return 0;
3643 }
3644
3645 SYSCTL_PROC(_debug, OID_AUTO, num_swap_files_configured, CTLFLAG_ANYBODY | CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED, 0, 0, sysctl_vm_config_num_swap_files, "I", "");
3646 #endif /* DEVELOPMENT || DEBUG */
3647
3648 /* this kernel does NOT implement shared_region_make_private_np() */
3649 SYSCTL_INT(_kern, KERN_SHREG_PRIVATIZABLE, shreg_private,
3650 CTLFLAG_RD | CTLFLAG_LOCKED,
3651 (int *)NULL, 0, "");
3652
3653 STATIC int
fetch_process_cputype(proc_t cur_proc,int * name,u_int namelen,cpu_type_t * cputype)3654 fetch_process_cputype(
3655 proc_t cur_proc,
3656 int *name,
3657 u_int namelen,
3658 cpu_type_t *cputype)
3659 {
3660 proc_t p = PROC_NULL;
3661 int refheld = 0;
3662 cpu_type_t ret = 0;
3663 int error = 0;
3664
3665 if (namelen == 0) {
3666 p = cur_proc;
3667 } else if (namelen == 1) {
3668 p = proc_find(name[0]);
3669 if (p == NULL) {
3670 return EINVAL;
3671 }
3672 refheld = 1;
3673 } else {
3674 error = EINVAL;
3675 goto out;
3676 }
3677
3678 ret = cpu_type() & ~CPU_ARCH_MASK;
3679 if (IS_64BIT_PROCESS(p)) {
3680 ret |= CPU_ARCH_ABI64;
3681 }
3682
3683 *cputype = ret;
3684
3685 if (refheld != 0) {
3686 proc_rele(p);
3687 }
3688 out:
3689 return error;
3690 }
3691
3692
3693 STATIC int
sysctl_sysctl_native(__unused struct sysctl_oid * oidp,void * arg1,int arg2,struct sysctl_req * req)3694 sysctl_sysctl_native(__unused struct sysctl_oid *oidp, void *arg1, int arg2,
3695 struct sysctl_req *req)
3696 {
3697 int error;
3698 cpu_type_t proc_cputype = 0;
3699 if ((error = fetch_process_cputype(req->p, (int *)arg1, arg2, &proc_cputype)) != 0) {
3700 return error;
3701 }
3702 int res = 1;
3703 if ((proc_cputype & ~CPU_ARCH_MASK) != (cpu_type() & ~CPU_ARCH_MASK)) {
3704 res = 0;
3705 }
3706 return SYSCTL_OUT(req, &res, sizeof(res));
3707 }
3708 SYSCTL_PROC(_sysctl, OID_AUTO, proc_native, CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED, 0, 0, sysctl_sysctl_native, "I", "proc_native");
3709
3710 STATIC int
sysctl_sysctl_cputype(__unused struct sysctl_oid * oidp,void * arg1,int arg2,struct sysctl_req * req)3711 sysctl_sysctl_cputype(__unused struct sysctl_oid *oidp, void *arg1, int arg2,
3712 struct sysctl_req *req)
3713 {
3714 int error;
3715 cpu_type_t proc_cputype = 0;
3716 if ((error = fetch_process_cputype(req->p, (int *)arg1, arg2, &proc_cputype)) != 0) {
3717 return error;
3718 }
3719 return SYSCTL_OUT(req, &proc_cputype, sizeof(proc_cputype));
3720 }
3721 SYSCTL_PROC(_sysctl, OID_AUTO, proc_cputype, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED, 0, 0, sysctl_sysctl_cputype, "I", "proc_cputype");
3722
3723 STATIC int
sysctl_safeboot(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3724 sysctl_safeboot
3725 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3726 {
3727 return sysctl_io_number(req, boothowto & RB_SAFEBOOT ? 1 : 0, sizeof(int), NULL, NULL);
3728 }
3729
3730 SYSCTL_PROC(_kern, KERN_SAFEBOOT, safeboot,
3731 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
3732 0, 0, sysctl_safeboot, "I", "");
3733
3734 STATIC int
sysctl_singleuser(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3735 sysctl_singleuser
3736 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3737 {
3738 return sysctl_io_number(req, boothowto & RB_SINGLE ? 1 : 0, sizeof(int), NULL, NULL);
3739 }
3740
3741 SYSCTL_PROC(_kern, OID_AUTO, singleuser,
3742 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
3743 0, 0, sysctl_singleuser, "I", "");
3744
3745 STATIC int
sysctl_minimalboot(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3746 sysctl_minimalboot
3747 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3748 {
3749 return sysctl_io_number(req, minimalboot, sizeof(int), NULL, NULL);
3750 }
3751
3752 SYSCTL_PROC(_kern, OID_AUTO, minimalboot,
3753 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
3754 0, 0, sysctl_minimalboot, "I", "");
3755
3756 /*
3757 * Controls for debugging affinity sets - see osfmk/kern/affinity.c
3758 */
3759 extern boolean_t affinity_sets_enabled;
3760 extern int affinity_sets_mapping;
3761
3762 SYSCTL_INT(_kern, OID_AUTO, affinity_sets_enabled,
3763 CTLFLAG_RW | CTLFLAG_LOCKED, (int *) &affinity_sets_enabled, 0, "hinting enabled");
3764 SYSCTL_INT(_kern, OID_AUTO, affinity_sets_mapping,
3765 CTLFLAG_RW | CTLFLAG_LOCKED, &affinity_sets_mapping, 0, "mapping policy");
3766
3767 /*
3768 * Boolean indicating if KASLR is active.
3769 */
3770 STATIC int
sysctl_slide(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3771 sysctl_slide
3772 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3773 {
3774 uint32_t slide;
3775
3776 slide = vm_kernel_slide ? 1 : 0;
3777
3778 return sysctl_io_number( req, slide, sizeof(int), NULL, NULL);
3779 }
3780
3781 SYSCTL_PROC(_kern, OID_AUTO, slide,
3782 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
3783 0, 0, sysctl_slide, "I", "");
3784
3785 #if DEBUG || DEVELOPMENT
3786 #if defined(__arm64__)
3787 extern vm_offset_t segTEXTEXECB;
3788
3789 static int
3790 sysctl_kernel_text_exec_base_slide SYSCTL_HANDLER_ARGS
3791 {
3792 #pragma unused(arg1, arg2, oidp)
3793 unsigned long slide = 0;
3794 kc_format_t kc_format;
3795
3796 PE_get_primary_kc_format(&kc_format);
3797
3798 if (kc_format == KCFormatFileset) {
3799 void *kch = PE_get_kc_header(KCKindPrimary);
3800 slide = (unsigned long)segTEXTEXECB - (unsigned long)kch + vm_kernel_slide;
3801 }
3802 return SYSCTL_OUT(req, &slide, sizeof(slide));
3803 }
3804
3805 SYSCTL_QUAD(_kern, OID_AUTO, kernel_slide, CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED, &vm_kernel_slide, "");
3806 SYSCTL_QUAD(_kern, OID_AUTO, kernel_text_exec_base, CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED, &segTEXTEXECB, "");
3807 SYSCTL_PROC(_kern, OID_AUTO, kernel_text_exec_base_slide, CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED, 0, 0, sysctl_kernel_text_exec_base_slide, "Q", "");
3808 #endif /* defined(__arm64__) */
3809
3810 /* User address of the PFZ */
3811 extern user32_addr_t commpage_text32_location;
3812 extern user64_addr_t commpage_text64_location;
3813
3814 STATIC int
3815 sysctl_pfz_start SYSCTL_HANDLER_ARGS
3816 {
3817 #pragma unused(oidp, arg1, arg2)
3818
3819 #ifdef __LP64__
3820 return sysctl_io_number(req, commpage_text64_location, sizeof(user64_addr_t), NULL, NULL);
3821 #else
3822 return sysctl_io_number(req, commpage_text32_location, sizeof(user32_addr_t), NULL, NULL);
3823 #endif
3824 }
3825
3826 SYSCTL_PROC(_kern, OID_AUTO, pfz,
3827 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_MASKED,
3828 0, 0, sysctl_pfz_start, "I", "");
3829 #endif
3830
3831
3832 /*
3833 * Limit on total memory users can wire.
3834 *
3835 * vm_global_user_wire_limit - system wide limit on wired memory from all processes combined.
3836 *
3837 * vm_per_task_user_wire_limit - per address space limit on wired memory. This puts a cap on the process's rlimit value.
3838 *
3839 * These values are initialized to reasonable defaults at boot time based on the available physical memory in
3840 * kmem_init().
3841 *
3842 * All values are in bytes.
3843 */
3844
3845 vm_map_size_t vm_global_user_wire_limit;
3846 vm_map_size_t vm_per_task_user_wire_limit;
3847 extern uint64_t max_mem_actual, max_mem;
3848
3849 uint64_t vm_add_wire_count_over_global_limit;
3850 uint64_t vm_add_wire_count_over_user_limit;
3851 /*
3852 * We used to have a global in the kernel called vm_global_no_user_wire_limit which was the inverse
3853 * of vm_global_user_wire_limit. But maintaining both of those is silly, and vm_global_user_wire_limit is the
3854 * real limit.
3855 * This function is for backwards compatibility with userspace
3856 * since we exposed the old global via a sysctl.
3857 */
3858 STATIC int
sysctl_global_no_user_wire_amount(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3859 sysctl_global_no_user_wire_amount(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3860 {
3861 vm_map_size_t old_value;
3862 vm_map_size_t new_value;
3863 int changed;
3864 int error;
3865 uint64_t config_memsize = max_mem;
3866 #if defined(XNU_TARGET_OS_OSX)
3867 config_memsize = max_mem_actual;
3868 #endif /* defined(XNU_TARGET_OS_OSX) */
3869
3870 old_value = (vm_map_size_t)(config_memsize - vm_global_user_wire_limit);
3871 error = sysctl_io_number(req, old_value, sizeof(vm_map_size_t), &new_value, &changed);
3872 if (changed) {
3873 if ((uint64_t)new_value > config_memsize) {
3874 error = EINVAL;
3875 } else {
3876 vm_global_user_wire_limit = (vm_map_size_t)(config_memsize - new_value);
3877 }
3878 }
3879 return error;
3880 }
3881 /*
3882 * There needs to be a more automatic/elegant way to do this
3883 */
3884 #if defined(__ARM__)
3885 SYSCTL_INT(_vm, OID_AUTO, global_user_wire_limit, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_global_user_wire_limit, 0, "");
3886 SYSCTL_INT(_vm, OID_AUTO, user_wire_limit, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_per_task_user_wire_limit, 0, "");
3887 SYSCTL_PROC(_vm, OID_AUTO, global_no_user_wire_amount, CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED, 0, 0, &sysctl_global_no_user_wire_amount, "I", "");
3888 #else
3889 SYSCTL_QUAD(_vm, OID_AUTO, global_user_wire_limit, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_global_user_wire_limit, "");
3890 SYSCTL_QUAD(_vm, OID_AUTO, user_wire_limit, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_per_task_user_wire_limit, "");
3891 SYSCTL_PROC(_vm, OID_AUTO, global_no_user_wire_amount, CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED, 0, 0, &sysctl_global_no_user_wire_amount, "Q", "");
3892 #endif
3893
3894 /*
3895 * Relaxed atomic RW of a 64bit value via sysctl.
3896 */
3897 STATIC int
sysctl_r_64bit_atomic(uint64_t * ptr,struct sysctl_req * req)3898 sysctl_r_64bit_atomic(uint64_t *ptr, struct sysctl_req *req)
3899 {
3900 uint64_t old_value;
3901 uint64_t new_value;
3902 int error;
3903
3904 old_value = os_atomic_load_wide(ptr, relaxed);
3905 error = sysctl_io_number(req, old_value, sizeof(vm_map_size_t), &new_value, NULL);
3906 return error;
3907 }
3908 STATIC int
sysctl_add_wire_count_over_global_limit(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3909 sysctl_add_wire_count_over_global_limit(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3910 {
3911 return sysctl_r_64bit_atomic(&vm_add_wire_count_over_global_limit, req);
3912 }
3913 STATIC int
sysctl_add_wire_count_over_user_limit(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3914 sysctl_add_wire_count_over_user_limit(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3915 {
3916 return sysctl_r_64bit_atomic(&vm_add_wire_count_over_user_limit, req);
3917 }
3918
3919 SYSCTL_PROC(_vm, OID_AUTO, add_wire_count_over_global_limit, CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED, 0, 0, &sysctl_add_wire_count_over_global_limit, "Q", "");
3920 SYSCTL_PROC(_vm, OID_AUTO, add_wire_count_over_user_limit, CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED, 0, 0, &sysctl_add_wire_count_over_user_limit, "Q", "");
3921
3922 #if DEVELOPMENT || DEBUG
3923 /* These sysctls are used to test the wired limit. */
3924 extern unsigned int vm_page_wire_count;
3925 extern uint32_t vm_lopage_free_count;
3926 SYSCTL_INT(_vm, OID_AUTO, page_wire_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_wire_count, 0, "");
3927 SYSCTL_INT(_vm, OID_AUTO, lopage_free_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_lopage_free_count, 0, "");
3928
3929 /*
3930 * Setting the per task variable exclude_physfootprint_ledger to 1 will allow the calling task to exclude memory entries that are
3931 * tagged by VM_LEDGER_TAG_DEFAULT and flagged by VM_LEDGER_FLAG_EXCLUDE_FOOTPRINT_DEBUG from its phys_footprint ledger.
3932 */
3933
3934 STATIC int
sysctl_rw_task_no_footprint_for_debug(struct sysctl_oid * oidp __unused,void * arg1 __unused,int arg2 __unused,struct sysctl_req * req)3935 sysctl_rw_task_no_footprint_for_debug(struct sysctl_oid *oidp __unused, void *arg1 __unused, int arg2 __unused, struct sysctl_req *req)
3936 {
3937 int error;
3938 int value;
3939 proc_t p = current_proc();
3940
3941 if (req->newptr) {
3942 // Write request
3943 error = SYSCTL_IN(req, &value, sizeof(value));
3944 if (!error) {
3945 if (value == 1) {
3946 task_set_no_footprint_for_debug(p->task, TRUE);
3947 } else if (value == 0) {
3948 task_set_no_footprint_for_debug(p->task, FALSE);
3949 } else {
3950 error = EINVAL;
3951 }
3952 }
3953 } else {
3954 // Read request
3955 value = task_get_no_footprint_for_debug(p->task);
3956 error = SYSCTL_OUT(req, &value, sizeof(value));
3957 }
3958 return error;
3959 }
3960
3961 SYSCTL_PROC(_vm, OID_AUTO, task_no_footprint_for_debug,
3962 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
3963 0, 0, &sysctl_rw_task_no_footprint_for_debug, "I", "Allow debug memory to be excluded from this task's memory footprint (debug only)");
3964
3965 #endif /* DEVELOPMENT || DEBUG */
3966
3967
3968 extern int vm_map_copy_overwrite_aligned_src_not_internal;
3969 extern int vm_map_copy_overwrite_aligned_src_not_symmetric;
3970 extern int vm_map_copy_overwrite_aligned_src_large;
3971 SYSCTL_INT(_vm, OID_AUTO, vm_copy_src_not_internal, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_map_copy_overwrite_aligned_src_not_internal, 0, "");
3972 SYSCTL_INT(_vm, OID_AUTO, vm_copy_src_not_symmetric, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_map_copy_overwrite_aligned_src_not_symmetric, 0, "");
3973 SYSCTL_INT(_vm, OID_AUTO, vm_copy_src_large, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_map_copy_overwrite_aligned_src_large, 0, "");
3974
3975
3976 extern uint32_t vm_page_external_count;
3977
3978 SYSCTL_INT(_vm, OID_AUTO, vm_page_external_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_external_count, 0, "");
3979
3980 SYSCTL_INT(_vm, OID_AUTO, vm_page_filecache_min, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_state.vm_page_filecache_min, 0, "");
3981 SYSCTL_INT(_vm, OID_AUTO, vm_page_xpmapped_min, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_state.vm_page_xpmapped_min, 0, "");
3982
3983 #if DEVELOPMENT || DEBUG
3984 SYSCTL_INT(_vm, OID_AUTO, vm_page_filecache_min_divisor, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_state.vm_page_filecache_min_divisor, 0, "");
3985 SYSCTL_INT(_vm, OID_AUTO, vm_page_xpmapped_min_divisor, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_state.vm_page_xpmapped_min_divisor, 0, "");
3986 #endif
3987
3988 extern int vm_compressor_mode;
3989 extern int vm_compressor_is_active;
3990 extern int vm_compressor_available;
3991 extern uint32_t c_seg_bufsize;
3992 extern uint32_t vm_ripe_target_age;
3993 extern uint32_t swapout_target_age;
3994 extern int64_t compressor_bytes_used;
3995 extern int64_t c_segment_input_bytes;
3996 extern int64_t c_segment_compressed_bytes;
3997 extern uint32_t compressor_eval_period_in_msecs;
3998 extern uint32_t compressor_sample_min_in_msecs;
3999 extern uint32_t compressor_sample_max_in_msecs;
4000 extern uint32_t compressor_thrashing_threshold_per_10msecs;
4001 extern uint32_t compressor_thrashing_min_per_10msecs;
4002 extern uint32_t vm_compressor_time_thread;
4003
4004 #if DEVELOPMENT || DEBUG
4005 extern uint32_t vm_compressor_minorcompact_threshold_divisor;
4006 extern uint32_t vm_compressor_majorcompact_threshold_divisor;
4007 extern uint32_t vm_compressor_unthrottle_threshold_divisor;
4008 extern uint32_t vm_compressor_catchup_threshold_divisor;
4009
4010 extern uint32_t vm_compressor_minorcompact_threshold_divisor_overridden;
4011 extern uint32_t vm_compressor_majorcompact_threshold_divisor_overridden;
4012 extern uint32_t vm_compressor_unthrottle_threshold_divisor_overridden;
4013 extern uint32_t vm_compressor_catchup_threshold_divisor_overridden;
4014
4015 extern vmct_stats_t vmct_stats;
4016
4017
4018 STATIC int
sysctl_minorcompact_threshold_divisor(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)4019 sysctl_minorcompact_threshold_divisor(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4020 {
4021 int new_value, changed;
4022 int error = sysctl_io_number(req, vm_compressor_minorcompact_threshold_divisor, sizeof(int), &new_value, &changed);
4023
4024 if (changed) {
4025 vm_compressor_minorcompact_threshold_divisor = new_value;
4026 vm_compressor_minorcompact_threshold_divisor_overridden = 1;
4027 }
4028 return error;
4029 }
4030
4031 SYSCTL_PROC(_vm, OID_AUTO, compressor_minorcompact_threshold_divisor,
4032 CTLTYPE_INT | CTLFLAG_LOCKED | CTLFLAG_RW,
4033 0, 0, sysctl_minorcompact_threshold_divisor, "I", "");
4034
4035
4036 STATIC int
sysctl_majorcompact_threshold_divisor(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)4037 sysctl_majorcompact_threshold_divisor(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4038 {
4039 int new_value, changed;
4040 int error = sysctl_io_number(req, vm_compressor_majorcompact_threshold_divisor, sizeof(int), &new_value, &changed);
4041
4042 if (changed) {
4043 vm_compressor_majorcompact_threshold_divisor = new_value;
4044 vm_compressor_majorcompact_threshold_divisor_overridden = 1;
4045 }
4046 return error;
4047 }
4048
4049 SYSCTL_PROC(_vm, OID_AUTO, compressor_majorcompact_threshold_divisor,
4050 CTLTYPE_INT | CTLFLAG_LOCKED | CTLFLAG_RW,
4051 0, 0, sysctl_majorcompact_threshold_divisor, "I", "");
4052
4053
4054 STATIC int
sysctl_unthrottle_threshold_divisor(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)4055 sysctl_unthrottle_threshold_divisor(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4056 {
4057 int new_value, changed;
4058 int error = sysctl_io_number(req, vm_compressor_unthrottle_threshold_divisor, sizeof(int), &new_value, &changed);
4059
4060 if (changed) {
4061 vm_compressor_unthrottle_threshold_divisor = new_value;
4062 vm_compressor_unthrottle_threshold_divisor_overridden = 1;
4063 }
4064 return error;
4065 }
4066
4067 SYSCTL_PROC(_vm, OID_AUTO, compressor_unthrottle_threshold_divisor,
4068 CTLTYPE_INT | CTLFLAG_LOCKED | CTLFLAG_RW,
4069 0, 0, sysctl_unthrottle_threshold_divisor, "I", "");
4070
4071
4072 STATIC int
sysctl_catchup_threshold_divisor(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)4073 sysctl_catchup_threshold_divisor(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4074 {
4075 int new_value, changed;
4076 int error = sysctl_io_number(req, vm_compressor_catchup_threshold_divisor, sizeof(int), &new_value, &changed);
4077
4078 if (changed) {
4079 vm_compressor_catchup_threshold_divisor = new_value;
4080 vm_compressor_catchup_threshold_divisor_overridden = 1;
4081 }
4082 return error;
4083 }
4084
4085 SYSCTL_PROC(_vm, OID_AUTO, compressor_catchup_threshold_divisor,
4086 CTLTYPE_INT | CTLFLAG_LOCKED | CTLFLAG_RW,
4087 0, 0, sysctl_catchup_threshold_divisor, "I", "");
4088 #endif
4089
4090
4091 SYSCTL_QUAD(_vm, OID_AUTO, compressor_input_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, &c_segment_input_bytes, "");
4092 SYSCTL_QUAD(_vm, OID_AUTO, compressor_compressed_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, &c_segment_compressed_bytes, "");
4093 SYSCTL_QUAD(_vm, OID_AUTO, compressor_bytes_used, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_bytes_used, "");
4094
4095 SYSCTL_INT(_vm, OID_AUTO, compressor_mode, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_compressor_mode, 0, "");
4096 SYSCTL_INT(_vm, OID_AUTO, compressor_is_active, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_compressor_is_active, 0, "");
4097 SYSCTL_INT(_vm, OID_AUTO, compressor_swapout_target_age, CTLFLAG_RD | CTLFLAG_LOCKED, &swapout_target_age, 0, "");
4098 SYSCTL_INT(_vm, OID_AUTO, compressor_available, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_compressor_available, 0, "");
4099 SYSCTL_INT(_vm, OID_AUTO, compressor_segment_buffer_size, CTLFLAG_RD | CTLFLAG_LOCKED, &c_seg_bufsize, 0, "");
4100
4101 extern int min_csegs_per_major_compaction;
4102 SYSCTL_INT(_vm, OID_AUTO, compressor_min_csegs_per_major_compaction, CTLFLAG_RW | CTLFLAG_LOCKED, &min_csegs_per_major_compaction, 0, "");
4103
4104 SYSCTL_INT(_vm, OID_AUTO, vm_ripe_target_age_in_secs, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_ripe_target_age, 0, "");
4105
4106 SYSCTL_INT(_vm, OID_AUTO, compressor_eval_period_in_msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &compressor_eval_period_in_msecs, 0, "");
4107 SYSCTL_INT(_vm, OID_AUTO, compressor_sample_min_in_msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &compressor_sample_min_in_msecs, 0, "");
4108 SYSCTL_INT(_vm, OID_AUTO, compressor_sample_max_in_msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &compressor_sample_max_in_msecs, 0, "");
4109 SYSCTL_INT(_vm, OID_AUTO, compressor_thrashing_threshold_per_10msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &compressor_thrashing_threshold_per_10msecs, 0, "");
4110 SYSCTL_INT(_vm, OID_AUTO, compressor_thrashing_min_per_10msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &compressor_thrashing_min_per_10msecs, 0, "");
4111
4112 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapouts_under_30s, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.unripe_under_30s, "");
4113 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapouts_under_60s, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.unripe_under_60s, "");
4114 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapouts_under_300s, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.unripe_under_300s, "");
4115 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_reclaim_swapins, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.reclaim_swapins, "");
4116 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_defrag_swapins, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.defrag_swapins, "");
4117 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_swapout_threshold_exceeded, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.compressor_swap_threshold_exceeded, "");
4118 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_swapout_fileq_throttled, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.external_q_throttled, "");
4119 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_swapout_free_count_low, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.free_count_below_reserve, "");
4120 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_swapout_thrashing_detected, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.thrashing_detected, "");
4121 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_swapout_fragmentation_detected, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.fragmentation_detected, "");
4122
4123 SYSCTL_STRING(_vm, OID_AUTO, swapfileprefix, CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED, swapfilename, sizeof(swapfilename) - SWAPFILENAME_INDEX_LEN, "");
4124
4125 SYSCTL_INT(_vm, OID_AUTO, compressor_timing_enabled, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_compressor_time_thread, 0, "");
4126
4127 #if DEVELOPMENT || DEBUG
4128 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_runtime0, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_runtimes[0], "");
4129 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_runtime1, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_runtimes[1], "");
4130
4131 SYSCTL_QUAD(_vm, OID_AUTO, compressor_threads_total, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_cthreads_total, "");
4132
4133 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_pages0, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_pages[0], "");
4134 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_pages1, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_pages[1], "");
4135
4136 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_iterations0, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_iterations[0], "");
4137 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_iterations1, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_iterations[1], "");
4138
4139 SYSCTL_INT(_vm, OID_AUTO, compressor_thread_minpages0, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_minpages[0], 0, "");
4140 SYSCTL_INT(_vm, OID_AUTO, compressor_thread_minpages1, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_minpages[1], 0, "");
4141
4142 SYSCTL_INT(_vm, OID_AUTO, compressor_thread_maxpages0, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_maxpages[0], 0, "");
4143 SYSCTL_INT(_vm, OID_AUTO, compressor_thread_maxpages1, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_maxpages[1], 0, "");
4144
4145 int vm_compressor_injected_error_count;
4146
4147 SYSCTL_INT(_vm, OID_AUTO, compressor_injected_error_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_compressor_injected_error_count, 0, "");
4148
4149 static int
sysctl_compressor_inject_error(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)4150 sysctl_compressor_inject_error(__unused struct sysctl_oid *oidp,
4151 __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4152 {
4153 int result;
4154 vm_address_t va = 0;
4155 int changed;
4156
4157 result = sysctl_io_number(req, va, sizeof(va), &va, &changed);
4158 if (result == 0 && changed) {
4159 result = vm_map_inject_error(current_map(), va);
4160 if (result == 0) {
4161 /*
4162 * Count the number of errors injected successfully to detect
4163 * situations where corruption was caused by improper use of this
4164 * sysctl.
4165 */
4166 os_atomic_inc(&vm_compressor_injected_error_count, relaxed);
4167 }
4168 }
4169 return result;
4170 }
4171
4172 SYSCTL_PROC(_vm, OID_AUTO, compressor_inject_error, CTLTYPE_QUAD | CTLFLAG_LOCKED | CTLFLAG_RW,
4173 0, 0, sysctl_compressor_inject_error, "Q", "flips a bit in a compressed page for the current task");
4174
4175 #endif
4176
4177 SYSCTL_QUAD(_vm, OID_AUTO, lz4_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_compressions, "");
4178 SYSCTL_QUAD(_vm, OID_AUTO, lz4_compression_failures, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_compression_failures, "");
4179 SYSCTL_QUAD(_vm, OID_AUTO, lz4_compressed_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_compressed_bytes, "");
4180 SYSCTL_QUAD(_vm, OID_AUTO, lz4_wk_compression_delta, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_wk_compression_delta, "");
4181 SYSCTL_QUAD(_vm, OID_AUTO, lz4_wk_compression_negative_delta, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_wk_compression_negative_delta, "");
4182
4183 SYSCTL_QUAD(_vm, OID_AUTO, lz4_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_decompressions, "");
4184 SYSCTL_QUAD(_vm, OID_AUTO, lz4_decompressed_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_decompressed_bytes, "");
4185
4186 SYSCTL_QUAD(_vm, OID_AUTO, uc_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.uc_decompressions, "");
4187
4188 SYSCTL_QUAD(_vm, OID_AUTO, wk_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_compressions, "");
4189
4190 SYSCTL_QUAD(_vm, OID_AUTO, wk_catime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_cabstime, "");
4191
4192 SYSCTL_QUAD(_vm, OID_AUTO, wkh_catime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wkh_cabstime, "");
4193 SYSCTL_QUAD(_vm, OID_AUTO, wkh_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wkh_compressions, "");
4194
4195 SYSCTL_QUAD(_vm, OID_AUTO, wks_catime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_cabstime, "");
4196 SYSCTL_QUAD(_vm, OID_AUTO, wks_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_compressions, "");
4197
4198 SYSCTL_QUAD(_vm, OID_AUTO, wk_compressions_exclusive, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_compressions_exclusive, "");
4199 SYSCTL_QUAD(_vm, OID_AUTO, wk_sv_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_sv_compressions, "");
4200 SYSCTL_QUAD(_vm, OID_AUTO, wk_mzv_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_mzv_compressions, "");
4201 SYSCTL_QUAD(_vm, OID_AUTO, wk_compression_failures, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_compression_failures, "");
4202 SYSCTL_QUAD(_vm, OID_AUTO, wk_compressed_bytes_exclusive, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_compressed_bytes_exclusive, "");
4203 SYSCTL_QUAD(_vm, OID_AUTO, wk_compressed_bytes_total, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_compressed_bytes_total, "");
4204
4205 SYSCTL_QUAD(_vm, OID_AUTO, wks_compressed_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_compressed_bytes, "");
4206 SYSCTL_QUAD(_vm, OID_AUTO, wks_compression_failures, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_compression_failures, "");
4207 SYSCTL_QUAD(_vm, OID_AUTO, wks_sv_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_sv_compressions, "");
4208
4209
4210 SYSCTL_QUAD(_vm, OID_AUTO, wk_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_decompressions, "");
4211
4212 SYSCTL_QUAD(_vm, OID_AUTO, wk_datime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_dabstime, "");
4213
4214 SYSCTL_QUAD(_vm, OID_AUTO, wkh_datime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wkh_dabstime, "");
4215 SYSCTL_QUAD(_vm, OID_AUTO, wkh_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wkh_decompressions, "");
4216
4217 SYSCTL_QUAD(_vm, OID_AUTO, wks_datime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_dabstime, "");
4218 SYSCTL_QUAD(_vm, OID_AUTO, wks_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_decompressions, "");
4219
4220 SYSCTL_QUAD(_vm, OID_AUTO, wk_decompressed_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_decompressed_bytes, "");
4221 SYSCTL_QUAD(_vm, OID_AUTO, wk_sv_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_sv_decompressions, "");
4222
4223 SYSCTL_INT(_vm, OID_AUTO, lz4_threshold, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_threshold, 0, "");
4224 SYSCTL_INT(_vm, OID_AUTO, wkdm_reeval_threshold, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.wkdm_reeval_threshold, 0, "");
4225 SYSCTL_INT(_vm, OID_AUTO, lz4_max_failure_skips, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_max_failure_skips, 0, "");
4226 SYSCTL_INT(_vm, OID_AUTO, lz4_max_failure_run_length, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_max_failure_run_length, 0, "");
4227 SYSCTL_INT(_vm, OID_AUTO, lz4_max_preselects, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_max_preselects, 0, "");
4228 SYSCTL_INT(_vm, OID_AUTO, lz4_run_preselection_threshold, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_run_preselection_threshold, 0, "");
4229 SYSCTL_INT(_vm, OID_AUTO, lz4_run_continue_bytes, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_run_continue_bytes, 0, "");
4230 SYSCTL_INT(_vm, OID_AUTO, lz4_profitable_bytes, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_profitable_bytes, 0, "");
4231 #if DEVELOPMENT || DEBUG
4232 extern int vm_compressor_current_codec;
4233 extern int vm_compressor_test_seg_wp;
4234 extern boolean_t vm_compressor_force_sw_wkdm;
4235 SYSCTL_INT(_vm, OID_AUTO, compressor_codec, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_compressor_current_codec, 0, "");
4236 SYSCTL_INT(_vm, OID_AUTO, compressor_test_wp, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_compressor_test_seg_wp, 0, "");
4237
4238 SYSCTL_INT(_vm, OID_AUTO, wksw_force, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_compressor_force_sw_wkdm, 0, "");
4239 extern int precompy, wkswhw;
4240
4241 SYSCTL_INT(_vm, OID_AUTO, precompy, CTLFLAG_RW | CTLFLAG_LOCKED, &precompy, 0, "");
4242 SYSCTL_INT(_vm, OID_AUTO, wkswhw, CTLFLAG_RW | CTLFLAG_LOCKED, &wkswhw, 0, "");
4243 extern unsigned int vm_ktrace_enabled;
4244 SYSCTL_INT(_vm, OID_AUTO, vm_ktrace, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_ktrace_enabled, 0, "");
4245 #endif
4246
4247 #if CONFIG_PHANTOM_CACHE
4248 extern uint32_t phantom_cache_thrashing_threshold;
4249 extern uint32_t phantom_cache_eval_period_in_msecs;
4250 extern uint32_t phantom_cache_thrashing_threshold_ssd;
4251
4252
4253 SYSCTL_INT(_vm, OID_AUTO, phantom_cache_eval_period_in_msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &phantom_cache_eval_period_in_msecs, 0, "");
4254 SYSCTL_INT(_vm, OID_AUTO, phantom_cache_thrashing_threshold, CTLFLAG_RW | CTLFLAG_LOCKED, &phantom_cache_thrashing_threshold, 0, "");
4255 SYSCTL_INT(_vm, OID_AUTO, phantom_cache_thrashing_threshold_ssd, CTLFLAG_RW | CTLFLAG_LOCKED, &phantom_cache_thrashing_threshold_ssd, 0, "");
4256 #endif
4257
4258 #if CONFIG_BACKGROUND_QUEUE
4259
4260 extern uint32_t vm_page_background_count;
4261 extern uint32_t vm_page_background_target;
4262 extern uint32_t vm_page_background_internal_count;
4263 extern uint32_t vm_page_background_external_count;
4264 extern uint32_t vm_page_background_mode;
4265 extern uint32_t vm_page_background_exclude_external;
4266 extern uint64_t vm_page_background_promoted_count;
4267 extern uint64_t vm_pageout_rejected_bq_internal;
4268 extern uint64_t vm_pageout_rejected_bq_external;
4269
4270 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_mode, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_page_background_mode, 0, "");
4271 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_exclude_external, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_page_background_exclude_external, 0, "");
4272 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_target, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_page_background_target, 0, "");
4273 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_background_count, 0, "");
4274 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_internal_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_background_internal_count, 0, "");
4275 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_external_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_background_external_count, 0, "");
4276
4277 SYSCTL_QUAD(_vm, OID_AUTO, vm_page_background_promoted_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_background_promoted_count, "");
4278 SYSCTL_QUAD(_vm, OID_AUTO, vm_pageout_considered_bq_internal, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_vminfo.vm_pageout_considered_bq_internal, "");
4279 SYSCTL_QUAD(_vm, OID_AUTO, vm_pageout_considered_bq_external, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_vminfo.vm_pageout_considered_bq_external, "");
4280 SYSCTL_QUAD(_vm, OID_AUTO, vm_pageout_rejected_bq_internal, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_rejected_bq_internal, "");
4281 SYSCTL_QUAD(_vm, OID_AUTO, vm_pageout_rejected_bq_external, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_rejected_bq_external, "");
4282
4283 #endif /* CONFIG_BACKGROUND_QUEUE */
4284
4285 extern void vm_update_darkwake_mode(boolean_t);
4286 extern boolean_t vm_darkwake_mode;
4287
4288 STATIC int
sysctl_toggle_darkwake_mode(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)4289 sysctl_toggle_darkwake_mode(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4290 {
4291 int new_value, changed;
4292 int error = sysctl_io_number(req, vm_darkwake_mode, sizeof(int), &new_value, &changed);
4293
4294 if (!error && changed) {
4295 if (new_value != 0 && new_value != 1) {
4296 printf("Error: Invalid value passed to darkwake sysctl. Acceptable: 0 or 1.\n");
4297 error = EINVAL;
4298 } else {
4299 vm_update_darkwake_mode((boolean_t) new_value);
4300 }
4301 }
4302
4303 return error;
4304 }
4305
4306 SYSCTL_PROC(_vm, OID_AUTO, darkwake_mode,
4307 CTLTYPE_INT | CTLFLAG_LOCKED | CTLFLAG_RW,
4308 0, 0, sysctl_toggle_darkwake_mode, "I", "");
4309
4310 #if (DEVELOPMENT || DEBUG)
4311
4312 SYSCTL_UINT(_vm, OID_AUTO, vm_page_creation_throttled_hard,
4313 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
4314 &vm_page_creation_throttled_hard, 0, "");
4315
4316 SYSCTL_UINT(_vm, OID_AUTO, vm_page_creation_throttled_soft,
4317 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
4318 &vm_page_creation_throttled_soft, 0, "");
4319
4320 extern uint32_t vm_pageout_memorystatus_fb_factor_nr;
4321 extern uint32_t vm_pageout_memorystatus_fb_factor_dr;
4322 SYSCTL_INT(_vm, OID_AUTO, vm_pageout_memorystatus_fb_factor_nr, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_memorystatus_fb_factor_nr, 0, "");
4323 SYSCTL_INT(_vm, OID_AUTO, vm_pageout_memorystatus_fb_factor_dr, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_memorystatus_fb_factor_dr, 0, "");
4324
4325 extern uint32_t vm_grab_anon_nops;
4326
4327 SYSCTL_INT(_vm, OID_AUTO, vm_grab_anon_overrides, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_debug.vm_grab_anon_overrides, 0, "");
4328 SYSCTL_INT(_vm, OID_AUTO, vm_grab_anon_nops, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_debug.vm_grab_anon_nops, 0, "");
4329
4330 extern int vm_page_delayed_work_ctx_needed;
4331 SYSCTL_INT(_vm, OID_AUTO, vm_page_needed_delayed_work_ctx, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_delayed_work_ctx_needed, 0, "");
4332
4333 /* log message counters for persistence mode */
4334 SCALABLE_COUNTER_DECLARE(oslog_p_total_msgcount);
4335 SCALABLE_COUNTER_DECLARE(oslog_p_metadata_saved_msgcount);
4336 SCALABLE_COUNTER_DECLARE(oslog_p_metadata_dropped_msgcount);
4337 SCALABLE_COUNTER_DECLARE(oslog_p_error_count);
4338 SCALABLE_COUNTER_DECLARE(oslog_p_saved_msgcount);
4339 SCALABLE_COUNTER_DECLARE(oslog_p_dropped_msgcount);
4340 SCALABLE_COUNTER_DECLARE(oslog_p_boot_dropped_msgcount);
4341 SCALABLE_COUNTER_DECLARE(oslog_p_coprocessor_total_msgcount);
4342 SCALABLE_COUNTER_DECLARE(oslog_p_coprocessor_dropped_msgcount);
4343 SCALABLE_COUNTER_DECLARE(oslog_p_unresolved_kc_msgcount);
4344 SCALABLE_COUNTER_DECLARE(oslog_p_fmt_invalid_msgcount);
4345 SCALABLE_COUNTER_DECLARE(oslog_p_fmt_max_args_msgcount);
4346 SCALABLE_COUNTER_DECLARE(oslog_p_truncated_msgcount);
4347
4348 SCALABLE_COUNTER_DECLARE(log_queue_cnt_received);
4349 SCALABLE_COUNTER_DECLARE(log_queue_cnt_rejected_fh);
4350 SCALABLE_COUNTER_DECLARE(log_queue_cnt_sent);
4351 SCALABLE_COUNTER_DECLARE(log_queue_cnt_dropped_nomem);
4352 SCALABLE_COUNTER_DECLARE(log_queue_cnt_queued);
4353 SCALABLE_COUNTER_DECLARE(log_queue_cnt_dropped_off);
4354 SCALABLE_COUNTER_DECLARE(log_queue_cnt_mem_active);
4355 SCALABLE_COUNTER_DECLARE(log_queue_cnt_mem_allocated);
4356 SCALABLE_COUNTER_DECLARE(log_queue_cnt_mem_released);
4357 SCALABLE_COUNTER_DECLARE(log_queue_cnt_mem_failed);
4358
4359 /* log message counters for streaming mode */
4360 SCALABLE_COUNTER_DECLARE(oslog_s_total_msgcount);
4361 SCALABLE_COUNTER_DECLARE(oslog_s_metadata_msgcount);
4362 SCALABLE_COUNTER_DECLARE(oslog_s_error_count);
4363 SCALABLE_COUNTER_DECLARE(oslog_s_streamed_msgcount);
4364 SCALABLE_COUNTER_DECLARE(oslog_s_dropped_msgcount);
4365
4366 /* log message counters for msgbuf logging */
4367 SCALABLE_COUNTER_DECLARE(oslog_msgbuf_msgcount);
4368 SCALABLE_COUNTER_DECLARE(oslog_msgbuf_dropped_msgcount);
4369 extern uint32_t oslog_msgbuf_dropped_charcount;
4370
4371 /* log message counters for vaddlog logging */
4372 extern uint32_t vaddlog_msgcount;
4373 extern uint32_t vaddlog_msgcount_dropped;
4374
4375 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_total_msgcount, oslog_p_total_msgcount, "");
4376 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_metadata_saved_msgcount, oslog_p_metadata_saved_msgcount, "");
4377 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_metadata_dropped_msgcount, oslog_p_metadata_dropped_msgcount, "");
4378 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_error_count, oslog_p_error_count, "");
4379 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_saved_msgcount, oslog_p_saved_msgcount, "");
4380 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_dropped_msgcount, oslog_p_dropped_msgcount, "");
4381 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_boot_dropped_msgcount, oslog_p_boot_dropped_msgcount, "");
4382 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_coprocessor_total_msgcount, oslog_p_coprocessor_total_msgcount, "");
4383 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_coprocessor_dropped_msgcount, oslog_p_coprocessor_dropped_msgcount, "");
4384 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_unresolved_kc_msgcount, oslog_p_unresolved_kc_msgcount, "");
4385
4386 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_fmt_invalid_msgcount, oslog_p_fmt_invalid_msgcount, "");
4387 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_fmt_max_args_msgcount, oslog_p_fmt_max_args_msgcount, "");
4388 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_truncated_msgcount, oslog_p_truncated_msgcount, "");
4389
4390 SYSCTL_SCALABLE_COUNTER(_debug, oslog_s_total_msgcount, oslog_s_total_msgcount, "Number of logs sent to streaming");
4391 SYSCTL_SCALABLE_COUNTER(_debug, oslog_s_metadata_msgcount, oslog_s_metadata_msgcount, "Number of metadata sent to streaming");
4392 SYSCTL_SCALABLE_COUNTER(_debug, oslog_s_error_count, oslog_s_error_count, "Number of invalid stream logs");
4393 SYSCTL_SCALABLE_COUNTER(_debug, oslog_s_streamed_msgcount, oslog_s_streamed_msgcount, "Number of streamed logs");
4394 SYSCTL_SCALABLE_COUNTER(_debug, oslog_s_dropped_msgcount, oslog_s_dropped_msgcount, "Number of logs dropped from stream");
4395
4396 SYSCTL_SCALABLE_COUNTER(_debug, oslog_msgbuf_msgcount, oslog_msgbuf_msgcount, "Number of dmesg log messages");
4397 SYSCTL_SCALABLE_COUNTER(_debug, oslog_msgbuf_dropped_msgcount, oslog_msgbuf_dropped_msgcount, "Number of dropped dmesg log messages");
4398 SYSCTL_UINT(_debug, OID_AUTO, oslog_msgbuf_dropped_charcount, CTLFLAG_ANYBODY | CTLFLAG_RD | CTLFLAG_LOCKED, &oslog_msgbuf_dropped_charcount, 0, "Number of dropped dmesg log chars");
4399
4400 SYSCTL_UINT(_debug, OID_AUTO, vaddlog_msgcount, CTLFLAG_ANYBODY | CTLFLAG_RD | CTLFLAG_LOCKED, &vaddlog_msgcount, 0, "");
4401 SYSCTL_UINT(_debug, OID_AUTO, vaddlog_msgcount_dropped, CTLFLAG_ANYBODY | CTLFLAG_RD | CTLFLAG_LOCKED, &vaddlog_msgcount_dropped, 0, "");
4402
4403 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_received, log_queue_cnt_received, "Number of received logs");
4404 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_rejected_fh, log_queue_cnt_rejected_fh, "Number of logs initially rejected by FH");
4405 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_sent, log_queue_cnt_sent, "Number of logs successfully saved in FH");
4406 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_dropped_nomem, log_queue_cnt_dropped_nomem, "Number of logs dropped due to lack of queue memory");
4407 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_queued, log_queue_cnt_queued, "Current number of logs stored in log queues");
4408 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_dropped_off, log_queue_cnt_dropped_off, "Number of logs dropped due to disabled log queues");
4409 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_mem_allocated, log_queue_cnt_mem_allocated, "Number of memory allocations");
4410 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_mem_released, log_queue_cnt_mem_released, "Number of memory releases");
4411 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_mem_failed, log_queue_cnt_mem_failed, "Number of failed memory allocations");
4412
4413 #endif /* DEVELOPMENT || DEBUG */
4414
4415 /*
4416 * Enable tracing of voucher contents
4417 */
4418 extern uint32_t ipc_voucher_trace_contents;
4419
4420 SYSCTL_INT(_kern, OID_AUTO, ipc_voucher_trace_contents,
4421 CTLFLAG_RW | CTLFLAG_LOCKED, &ipc_voucher_trace_contents, 0, "Enable tracing voucher contents");
4422
4423 /*
4424 * Kernel stack size and depth
4425 */
4426 SYSCTL_INT(_kern, OID_AUTO, stack_size,
4427 CTLFLAG_RD | CTLFLAG_LOCKED, (int *) &kernel_stack_size, 0, "Kernel stack size");
4428 SYSCTL_INT(_kern, OID_AUTO, stack_depth_max,
4429 CTLFLAG_RD | CTLFLAG_LOCKED, (int *) &kernel_stack_depth_max, 0, "Max kernel stack depth at interrupt or context switch");
4430
4431 extern unsigned int kern_feature_overrides;
4432 SYSCTL_INT(_kern, OID_AUTO, kern_feature_overrides,
4433 CTLFLAG_RD | CTLFLAG_LOCKED, &kern_feature_overrides, 0, "Kernel feature override mask");
4434
4435 /*
4436 * enable back trace for port allocations
4437 */
4438 extern int ipc_portbt;
4439
4440 SYSCTL_INT(_kern, OID_AUTO, ipc_portbt,
4441 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
4442 &ipc_portbt, 0, "");
4443
4444 /*
4445 * Mach message signature validation control and outputs
4446 */
4447 extern unsigned int ikm_signature_failures;
4448 SYSCTL_INT(_kern, OID_AUTO, ikm_signature_failures,
4449 CTLFLAG_RD | CTLFLAG_LOCKED, &ikm_signature_failures, 0, "Message signature failure count");
4450 extern unsigned int ikm_signature_failure_id;
4451 SYSCTL_INT(_kern, OID_AUTO, ikm_signature_failure_id,
4452 CTLFLAG_RD | CTLFLAG_LOCKED, &ikm_signature_failure_id, 0, "Message signature failure count");
4453
4454 #if (DEVELOPMENT || DEBUG)
4455 extern unsigned int ikm_signature_panic_disable;
4456 SYSCTL_INT(_kern, OID_AUTO, ikm_signature_panic_disable,
4457 CTLFLAG_RW | CTLFLAG_LOCKED, &ikm_signature_panic_disable, 0, "Message signature failure mode");
4458 extern unsigned int ikm_signature_header_failures;
4459 SYSCTL_INT(_kern, OID_AUTO, ikm_signature_header_failures,
4460 CTLFLAG_RD | CTLFLAG_LOCKED, &ikm_signature_header_failures, 0, "Message header signature failure count");
4461 extern unsigned int ikm_signature_trailer_failures;
4462 SYSCTL_INT(_kern, OID_AUTO, ikm_signature_trailer_failures,
4463 CTLFLAG_RD | CTLFLAG_LOCKED, &ikm_signature_trailer_failures, 0, "Message trailer signature failure count");
4464 #endif
4465
4466 /*
4467 * Scheduler sysctls
4468 */
4469
4470 SYSCTL_STRING(_kern, OID_AUTO, sched,
4471 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
4472 sched_string, sizeof(sched_string),
4473 "Timeshare scheduler implementation");
4474
4475 #if CONFIG_QUIESCE_COUNTER
4476 static int
4477 sysctl_cpu_quiescent_counter_interval SYSCTL_HANDLER_ARGS
4478 {
4479 #pragma unused(arg1, arg2)
4480
4481 uint32_t local_min_interval_us = cpu_quiescent_counter_get_min_interval_us();
4482
4483 int error = sysctl_handle_int(oidp, &local_min_interval_us, 0, req);
4484 if (error || !req->newptr) {
4485 return error;
4486 }
4487
4488 cpu_quiescent_counter_set_min_interval_us(local_min_interval_us);
4489
4490 return 0;
4491 }
4492
4493 SYSCTL_PROC(_kern, OID_AUTO, cpu_checkin_interval,
4494 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
4495 0, 0,
4496 sysctl_cpu_quiescent_counter_interval, "I",
4497 "Quiescent CPU checkin interval (microseconds)");
4498 #endif /* CONFIG_QUIESCE_COUNTER */
4499
4500
4501 /*
4502 * Only support runtime modification on development / debug
4503 */
4504 #if DEVELOPMENT || DEBUG
4505 extern int precise_user_kernel_time;
4506 SYSCTL_INT(_kern, OID_AUTO, precise_user_kernel_time,
4507 CTLFLAG_RW | CTLFLAG_LOCKED,
4508 &precise_user_kernel_time, 0, "Precise accounting of kernel vs. user time");
4509 #endif /* DEVELOPMENT || DEBUG */
4510
4511 /* Parameters related to timer coalescing tuning, to be replaced
4512 * with a dedicated systemcall in the future.
4513 */
4514 /* Enable processing pending timers in the context of any other interrupt
4515 * Coalescing tuning parameters for various thread/task attributes */
4516 STATIC int
4517 sysctl_timer_user_us_kernel_abstime SYSCTL_HANDLER_ARGS
4518 {
4519 #pragma unused(oidp)
4520 int size = arg2; /* subcommand*/
4521 int error;
4522 int changed = 0;
4523 uint64_t old_value_ns;
4524 uint64_t new_value_ns;
4525 uint64_t value_abstime;
4526 if (size == sizeof(uint32_t)) {
4527 value_abstime = *((uint32_t *)arg1);
4528 } else if (size == sizeof(uint64_t)) {
4529 value_abstime = *((uint64_t *)arg1);
4530 } else {
4531 return ENOTSUP;
4532 }
4533
4534 absolutetime_to_nanoseconds(value_abstime, &old_value_ns);
4535 error = sysctl_io_number(req, old_value_ns, sizeof(old_value_ns), &new_value_ns, &changed);
4536 if ((error) || (!changed)) {
4537 return error;
4538 }
4539
4540 nanoseconds_to_absolutetime(new_value_ns, &value_abstime);
4541 if (size == sizeof(uint32_t)) {
4542 *((uint32_t *)arg1) = (uint32_t)value_abstime;
4543 } else {
4544 *((uint64_t *)arg1) = value_abstime;
4545 }
4546 return error;
4547 }
4548
4549 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_bg_scale,
4550 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4551 &tcoal_prio_params.timer_coalesce_bg_shift, 0, "");
4552 SYSCTL_PROC(_kern, OID_AUTO, timer_resort_threshold_ns,
4553 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4554 &tcoal_prio_params.timer_resort_threshold_abstime,
4555 sizeof(tcoal_prio_params.timer_resort_threshold_abstime),
4556 sysctl_timer_user_us_kernel_abstime,
4557 "Q", "");
4558 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_bg_ns_max,
4559 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4560 &tcoal_prio_params.timer_coalesce_bg_abstime_max,
4561 sizeof(tcoal_prio_params.timer_coalesce_bg_abstime_max),
4562 sysctl_timer_user_us_kernel_abstime,
4563 "Q", "");
4564
4565 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_kt_scale,
4566 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4567 &tcoal_prio_params.timer_coalesce_kt_shift, 0, "");
4568
4569 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_kt_ns_max,
4570 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4571 &tcoal_prio_params.timer_coalesce_kt_abstime_max,
4572 sizeof(tcoal_prio_params.timer_coalesce_kt_abstime_max),
4573 sysctl_timer_user_us_kernel_abstime,
4574 "Q", "");
4575
4576 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_fp_scale,
4577 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4578 &tcoal_prio_params.timer_coalesce_fp_shift, 0, "");
4579
4580 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_fp_ns_max,
4581 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4582 &tcoal_prio_params.timer_coalesce_fp_abstime_max,
4583 sizeof(tcoal_prio_params.timer_coalesce_fp_abstime_max),
4584 sysctl_timer_user_us_kernel_abstime,
4585 "Q", "");
4586
4587 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_ts_scale,
4588 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4589 &tcoal_prio_params.timer_coalesce_ts_shift, 0, "");
4590
4591 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_ts_ns_max,
4592 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4593 &tcoal_prio_params.timer_coalesce_ts_abstime_max,
4594 sizeof(tcoal_prio_params.timer_coalesce_ts_abstime_max),
4595 sysctl_timer_user_us_kernel_abstime,
4596 "Q", "");
4597
4598 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier0_scale,
4599 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4600 &tcoal_prio_params.latency_qos_scale[0], 0, "");
4601
4602 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier0_ns_max,
4603 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4604 &tcoal_prio_params.latency_qos_abstime_max[0],
4605 sizeof(tcoal_prio_params.latency_qos_abstime_max[0]),
4606 sysctl_timer_user_us_kernel_abstime,
4607 "Q", "");
4608
4609 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier1_scale,
4610 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4611 &tcoal_prio_params.latency_qos_scale[1], 0, "");
4612
4613 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier1_ns_max,
4614 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4615 &tcoal_prio_params.latency_qos_abstime_max[1],
4616 sizeof(tcoal_prio_params.latency_qos_abstime_max[1]),
4617 sysctl_timer_user_us_kernel_abstime,
4618 "Q", "");
4619
4620 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier2_scale,
4621 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4622 &tcoal_prio_params.latency_qos_scale[2], 0, "");
4623
4624 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier2_ns_max,
4625 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4626 &tcoal_prio_params.latency_qos_abstime_max[2],
4627 sizeof(tcoal_prio_params.latency_qos_abstime_max[2]),
4628 sysctl_timer_user_us_kernel_abstime,
4629 "Q", "");
4630
4631 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier3_scale,
4632 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4633 &tcoal_prio_params.latency_qos_scale[3], 0, "");
4634
4635 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier3_ns_max,
4636 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4637 &tcoal_prio_params.latency_qos_abstime_max[3],
4638 sizeof(tcoal_prio_params.latency_qos_abstime_max[3]),
4639 sysctl_timer_user_us_kernel_abstime,
4640 "Q", "");
4641
4642 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier4_scale,
4643 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4644 &tcoal_prio_params.latency_qos_scale[4], 0, "");
4645
4646 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier4_ns_max,
4647 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4648 &tcoal_prio_params.latency_qos_abstime_max[4],
4649 sizeof(tcoal_prio_params.latency_qos_abstime_max[4]),
4650 sysctl_timer_user_us_kernel_abstime,
4651 "Q", "");
4652
4653 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier5_scale,
4654 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4655 &tcoal_prio_params.latency_qos_scale[5], 0, "");
4656
4657 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier5_ns_max,
4658 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4659 &tcoal_prio_params.latency_qos_abstime_max[5],
4660 sizeof(tcoal_prio_params.latency_qos_abstime_max[5]),
4661 sysctl_timer_user_us_kernel_abstime,
4662 "Q", "");
4663
4664 /* Communicate the "user idle level" heuristic to the timer layer, and
4665 * potentially other layers in the future.
4666 */
4667
4668 static int
timer_user_idle_level(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)4669 timer_user_idle_level(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4670 {
4671 int new_value = 0, old_value = 0, changed = 0, error;
4672
4673 old_value = timer_get_user_idle_level();
4674
4675 error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
4676
4677 if (error == 0 && changed) {
4678 if (timer_set_user_idle_level(new_value) != KERN_SUCCESS) {
4679 error = ERANGE;
4680 }
4681 }
4682
4683 return error;
4684 }
4685
4686 SYSCTL_PROC(_machdep, OID_AUTO, user_idle_level,
4687 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
4688 0, 0,
4689 timer_user_idle_level, "I", "User idle level heuristic, 0-128");
4690
4691 #if HYPERVISOR
4692 SYSCTL_INT(_kern, OID_AUTO, hv_support,
4693 CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
4694 &hv_support_available, 0, "");
4695
4696 SYSCTL_INT(_kern, OID_AUTO, hv_disable,
4697 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4698 &hv_disable, 0, "");
4699 #endif
4700
4701 #if DEVELOPMENT || DEBUG
4702 extern uint64_t driverkit_checkin_timed_out;
4703 SYSCTL_QUAD(_kern, OID_AUTO, driverkit_checkin_timed_out,
4704 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
4705 &driverkit_checkin_timed_out, "timestamp of dext checkin timeout");
4706 #endif
4707
4708 extern int IOGetVMMPresent(void);
4709
4710 static int
4711 hv_vmm_present SYSCTL_HANDLER_ARGS
4712 {
4713 __unused struct sysctl_oid *unused_oidp = oidp;
4714 __unused void *unused_arg1 = arg1;
4715 __unused int unused_arg2 = arg2;
4716
4717 int hv_vmm_present = 0;
4718
4719 hv_vmm_present = IOGetVMMPresent();
4720
4721 return SYSCTL_OUT(req, &hv_vmm_present, sizeof(hv_vmm_present));
4722 }
4723
4724 SYSCTL_PROC(_kern, OID_AUTO, hv_vmm_present,
4725 CTLTYPE_INT | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED,
4726 0, 0,
4727 hv_vmm_present, "I", "");
4728
4729 #if CONFIG_DARKBOOT
4730 STATIC int
4731 sysctl_darkboot SYSCTL_HANDLER_ARGS
4732 {
4733 int err = 0, value = 0;
4734 #pragma unused(oidp, arg1, arg2, err, value, req)
4735
4736 /*
4737 * Handle the sysctl request.
4738 *
4739 * If this is a read, the function will set the value to the current darkboot value. Otherwise,
4740 * we'll get the request identifier into "value" and then we can honor it.
4741 */
4742 if ((err = sysctl_io_number(req, darkboot, sizeof(int), &value, NULL)) != 0) {
4743 goto exit;
4744 }
4745
4746 /* writing requested, let's process the request */
4747 if (req->newptr) {
4748 /* writing is protected by an entitlement */
4749 if (priv_check_cred(kauth_cred_get(), PRIV_DARKBOOT, 0) != 0) {
4750 err = EPERM;
4751 goto exit;
4752 }
4753
4754 switch (value) {
4755 case MEMORY_MAINTENANCE_DARK_BOOT_UNSET:
4756 /*
4757 * If the darkboot sysctl is unset, the NVRAM variable
4758 * must be unset too. If that's not the case, it means
4759 * someone is doing something crazy and not supported.
4760 */
4761 if (darkboot != 0) {
4762 int ret = PERemoveNVRAMProperty(MEMORY_MAINTENANCE_DARK_BOOT_NVRAM_NAME);
4763 if (ret) {
4764 darkboot = 0;
4765 } else {
4766 err = EINVAL;
4767 }
4768 }
4769 break;
4770 case MEMORY_MAINTENANCE_DARK_BOOT_SET:
4771 darkboot = 1;
4772 break;
4773 case MEMORY_MAINTENANCE_DARK_BOOT_SET_PERSISTENT: {
4774 /*
4775 * Set the NVRAM and update 'darkboot' in case
4776 * of success. Otherwise, do not update
4777 * 'darkboot' and report the failure.
4778 */
4779 if (PEWriteNVRAMBooleanProperty(MEMORY_MAINTENANCE_DARK_BOOT_NVRAM_NAME, TRUE)) {
4780 darkboot = 1;
4781 } else {
4782 err = EINVAL;
4783 }
4784
4785 break;
4786 }
4787 default:
4788 err = EINVAL;
4789 }
4790 }
4791
4792 exit:
4793 return err;
4794 }
4795
4796 SYSCTL_PROC(_kern, OID_AUTO, darkboot,
4797 CTLFLAG_KERN | CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_ANYBODY,
4798 0, 0, sysctl_darkboot, "I", "");
4799 #endif /* CONFIG_DARKBOOT */
4800
4801 #if DEVELOPMENT || DEBUG
4802 #include <sys/sysent.h>
4803 /* This should result in a fatal exception, verifying that "sysent" is
4804 * write-protected.
4805 */
4806 static int
kern_sysent_write(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)4807 kern_sysent_write(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4808 {
4809 uint64_t new_value = 0, old_value = 0;
4810 int changed = 0, error;
4811
4812 error = sysctl_io_number(req, old_value, sizeof(uint64_t), &new_value, &changed);
4813 if ((error == 0) && changed) {
4814 volatile uint32_t *wraddr = __DECONST(uint32_t *, &sysent[0]);
4815 *wraddr = 0;
4816 printf("sysent[0] write succeeded\n");
4817 }
4818 return error;
4819 }
4820
4821 SYSCTL_PROC(_kern, OID_AUTO, sysent_const_check,
4822 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
4823 0, 0,
4824 kern_sysent_write, "I", "Attempt sysent[0] write");
4825
4826 #endif
4827
4828 #if DEVELOPMENT || DEBUG
4829 SYSCTL_COMPAT_INT(_kern, OID_AUTO, development, CTLFLAG_RD | CTLFLAG_MASKED | CTLFLAG_KERN, NULL, 1, "");
4830 #else
4831 SYSCTL_COMPAT_INT(_kern, OID_AUTO, development, CTLFLAG_RD | CTLFLAG_MASKED, NULL, 0, "");
4832 #endif
4833
4834
4835 #if DEVELOPMENT || DEBUG
4836
4837 static int
4838 sysctl_panic_test SYSCTL_HANDLER_ARGS
4839 {
4840 #pragma unused(arg1, arg2)
4841 int rval = 0;
4842 char str[32] = "entry prelog postlog postcore";
4843
4844 rval = sysctl_handle_string(oidp, str, sizeof(str), req);
4845
4846 if (rval == 0 && req->newptr) {
4847 if (strncmp("entry", str, strlen("entry")) == 0) {
4848 panic_with_options(0, NULL, DEBUGGER_OPTION_RECURPANIC_ENTRY, "test recursive panic at entry");
4849 } else if (strncmp("prelog", str, strlen("prelog")) == 0) {
4850 panic_with_options(0, NULL, DEBUGGER_OPTION_RECURPANIC_PRELOG, "test recursive panic prior to writing a paniclog");
4851 } else if (strncmp("postlog", str, strlen("postlog")) == 0) {
4852 panic_with_options(0, NULL, DEBUGGER_OPTION_RECURPANIC_POSTLOG, "test recursive panic subsequent to paniclog");
4853 } else if (strncmp("postcore", str, strlen("postcore")) == 0) {
4854 panic_with_options(0, NULL, DEBUGGER_OPTION_RECURPANIC_POSTCORE, "test recursive panic subsequent to on-device core");
4855 }
4856 }
4857
4858 return rval;
4859 }
4860
4861 static int
4862 sysctl_debugger_test SYSCTL_HANDLER_ARGS
4863 {
4864 #pragma unused(arg1, arg2)
4865 int rval = 0;
4866 char str[32] = "entry prelog postlog postcore";
4867
4868 rval = sysctl_handle_string(oidp, str, sizeof(str), req);
4869
4870 if (rval == 0 && req->newptr) {
4871 if (strncmp("entry", str, strlen("entry")) == 0) {
4872 DebuggerWithContext(0, NULL, "test recursive panic via debugger at entry", DEBUGGER_OPTION_RECURPANIC_ENTRY, (unsigned long)(char *)__builtin_return_address(0));
4873 } else if (strncmp("prelog", str, strlen("prelog")) == 0) {
4874 DebuggerWithContext(0, NULL, "test recursive panic via debugger prior to writing a paniclog", DEBUGGER_OPTION_RECURPANIC_PRELOG, (unsigned long)(char *)__builtin_return_address(0));
4875 } else if (strncmp("postlog", str, strlen("postlog")) == 0) {
4876 DebuggerWithContext(0, NULL, "test recursive panic via debugger subsequent to paniclog", DEBUGGER_OPTION_RECURPANIC_POSTLOG, (unsigned long)(char *)__builtin_return_address(0));
4877 } else if (strncmp("postcore", str, strlen("postcore")) == 0) {
4878 DebuggerWithContext(0, NULL, "test recursive panic via debugger subsequent to on-device core", DEBUGGER_OPTION_RECURPANIC_POSTCORE, (unsigned long)(char *)__builtin_return_address(0));
4879 }
4880 }
4881
4882 return rval;
4883 }
4884
4885 decl_lck_spin_data(, spinlock_panic_test_lock);
4886
4887 __attribute__((noreturn))
4888 static void
spinlock_panic_test_acquire_spinlock(void * arg __unused,wait_result_t wres __unused)4889 spinlock_panic_test_acquire_spinlock(void * arg __unused, wait_result_t wres __unused)
4890 {
4891 lck_spin_lock(&spinlock_panic_test_lock);
4892 while (1) {
4893 ;
4894 }
4895 }
4896
4897 static int
4898 sysctl_spinlock_panic_test SYSCTL_HANDLER_ARGS
4899 {
4900 #pragma unused(oidp, arg1, arg2)
4901 if (req->newlen == 0) {
4902 return EINVAL;
4903 }
4904
4905 thread_t panic_spinlock_thread;
4906 /* Initialize panic spinlock */
4907 lck_grp_t * panic_spinlock_grp;
4908 lck_grp_attr_t * panic_spinlock_grp_attr;
4909 lck_attr_t * panic_spinlock_attr;
4910
4911 panic_spinlock_grp_attr = lck_grp_attr_alloc_init();
4912 panic_spinlock_grp = lck_grp_alloc_init("panic_spinlock", panic_spinlock_grp_attr);
4913 panic_spinlock_attr = lck_attr_alloc_init();
4914
4915 lck_spin_init(&spinlock_panic_test_lock, panic_spinlock_grp, panic_spinlock_attr);
4916
4917
4918 /* Create thread to acquire spinlock */
4919 if (kernel_thread_start(spinlock_panic_test_acquire_spinlock, NULL, &panic_spinlock_thread) != KERN_SUCCESS) {
4920 return EBUSY;
4921 }
4922
4923 /* Try to acquire spinlock -- should panic eventually */
4924 lck_spin_lock(&spinlock_panic_test_lock);
4925 while (1) {
4926 ;
4927 }
4928 }
4929
4930 __attribute__((noreturn))
4931 static void
simultaneous_panic_worker(void * arg,wait_result_t wres __unused)4932 simultaneous_panic_worker
4933 (void * arg, wait_result_t wres __unused)
4934 {
4935 atomic_int *start_panic = (atomic_int *)arg;
4936
4937 while (!atomic_load(start_panic)) {
4938 ;
4939 }
4940 panic("SIMULTANEOUS PANIC TEST: INITIATING PANIC FROM CPU %d", cpu_number());
4941 __builtin_unreachable();
4942 }
4943
4944 static int
4945 sysctl_simultaneous_panic_test SYSCTL_HANDLER_ARGS
4946 {
4947 #pragma unused(oidp, arg1, arg2)
4948 if (req->newlen == 0) {
4949 return EINVAL;
4950 }
4951
4952 int i = 0, threads_to_create = 2 * processor_count;
4953 atomic_int start_panic = 0;
4954 unsigned int threads_created = 0;
4955 thread_t new_panic_thread;
4956
4957 for (i = threads_to_create; i > 0; i--) {
4958 if (kernel_thread_start(simultaneous_panic_worker, (void *) &start_panic, &new_panic_thread) == KERN_SUCCESS) {
4959 threads_created++;
4960 }
4961 }
4962
4963 /* FAIL if we couldn't create at least processor_count threads */
4964 if (threads_created < processor_count) {
4965 panic("SIMULTANEOUS PANIC TEST: FAILED TO CREATE ENOUGH THREADS, ONLY CREATED %d (of %d)",
4966 threads_created, threads_to_create);
4967 }
4968
4969 atomic_exchange(&start_panic, 1);
4970 while (1) {
4971 ;
4972 }
4973 }
4974
4975 SYSCTL_PROC(_debug, OID_AUTO, panic_test, CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_MASKED, 0, 0, sysctl_panic_test, "A", "panic test");
4976 SYSCTL_PROC(_debug, OID_AUTO, debugger_test, CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_MASKED, 0, 0, sysctl_debugger_test, "A", "debugger test");
4977 SYSCTL_PROC(_debug, OID_AUTO, spinlock_panic_test, CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_MASKED, 0, 0, sysctl_spinlock_panic_test, "A", "spinlock panic test");
4978 SYSCTL_PROC(_debug, OID_AUTO, simultaneous_panic_test, CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_MASKED, 0, 0, sysctl_simultaneous_panic_test, "A", "simultaneous panic test");
4979
4980 extern int exc_resource_threads_enabled;
4981
4982 SYSCTL_INT(_kern, OID_AUTO, exc_resource_threads_enabled, CTLFLAG_RD | CTLFLAG_LOCKED, &exc_resource_threads_enabled, 0, "exc_resource thread limit enabled");
4983
4984
4985 #endif /* DEVELOPMENT || DEBUG */
4986
4987 #if CONFIG_THREAD_GROUPS
4988 #if DEVELOPMENT || DEBUG
4989
4990 static int
4991 sysctl_get_thread_group_id SYSCTL_HANDLER_ARGS
4992 {
4993 #pragma unused(arg1, arg2, oidp)
4994 uint64_t thread_group_id = thread_group_get_id(thread_group_get(current_thread()));
4995 return SYSCTL_OUT(req, &thread_group_id, sizeof(thread_group_id));
4996 }
4997
4998 SYSCTL_PROC(_kern, OID_AUTO, thread_group_id, CTLFLAG_RD | CTLFLAG_LOCKED | CTLTYPE_QUAD,
4999 0, 0, &sysctl_get_thread_group_id, "I", "thread group id of the thread");
5000
5001 STATIC int
sysctl_thread_group_count(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)5002 sysctl_thread_group_count(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
5003 {
5004 int value = thread_group_count();
5005 return sysctl_io_number(req, value, sizeof(value), NULL, NULL);
5006 }
5007
5008 SYSCTL_PROC(_kern, OID_AUTO, thread_group_count, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5009 0, 0, &sysctl_thread_group_count, "I", "count of thread groups");
5010
5011 #endif /* DEVELOPMENT || DEBUG */
5012 const uint32_t thread_groups_supported = 1;
5013 #else /* CONFIG_THREAD_GROUPS */
5014 const uint32_t thread_groups_supported = 0;
5015 #endif /* CONFIG_THREAD_GROUPS */
5016
5017 STATIC int
sysctl_thread_groups_supported(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)5018 sysctl_thread_groups_supported(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
5019 {
5020 int value = thread_groups_supported;
5021 return sysctl_io_number(req, value, sizeof(value), NULL, NULL);
5022 }
5023
5024 SYSCTL_PROC(_kern, OID_AUTO, thread_groups_supported, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5025 0, 0, &sysctl_thread_groups_supported, "I", "thread groups supported");
5026
5027 static int
5028 sysctl_grade_cputype SYSCTL_HANDLER_ARGS
5029 {
5030 #pragma unused(arg1, arg2, oidp)
5031 int error = 0;
5032 int type_tuple[2] = {};
5033 int return_value = 0;
5034
5035 error = SYSCTL_IN(req, &type_tuple, sizeof(type_tuple));
5036
5037 if (error) {
5038 return error;
5039 }
5040
5041 return_value = grade_binary(type_tuple[0], type_tuple[1] & ~CPU_SUBTYPE_MASK, type_tuple[1] & CPU_SUBTYPE_MASK, FALSE);
5042
5043 error = SYSCTL_OUT(req, &return_value, sizeof(return_value));
5044
5045 if (error) {
5046 return error;
5047 }
5048
5049 return error;
5050 }
5051
5052 SYSCTL_PROC(_kern, OID_AUTO, grade_cputype,
5053 CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_MASKED | CTLFLAG_LOCKED | CTLTYPE_OPAQUE,
5054 0, 0, &sysctl_grade_cputype, "S",
5055 "grade value of cpu_type_t+cpu_sub_type_t");
5056
5057 extern boolean_t allow_direct_handoff;
5058 SYSCTL_INT(_kern, OID_AUTO, direct_handoff,
5059 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5060 &allow_direct_handoff, 0, "Enable direct handoff for realtime threads");
5061
5062 #if DEVELOPMENT || DEBUG
5063
5064 SYSCTL_LONG(_kern, OID_AUTO, phys_carveout_pa, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5065 &phys_carveout_pa,
5066 "base physical address of the phys_carveout_mb boot-arg region");
5067 SYSCTL_LONG(_kern, OID_AUTO, phys_carveout_size, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5068 &phys_carveout_size,
5069 "size in bytes of the phys_carveout_mb boot-arg region");
5070 SYSCTL_LONG(_kern, OID_AUTO, phys_carveout_metadata_pa, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5071 &phys_carveout_metadata_pa,
5072 "base physical address of the phys_carveout_metadata region");
5073 SYSCTL_LONG(_kern, OID_AUTO, phys_carveout_metadata_size, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5074 &phys_carveout_metadata_size,
5075 "size in bytes of the phys_carveout_metadata region");
5076
5077 extern void do_cseg_wedge_thread(void);
5078 extern void do_cseg_unwedge_thread(void);
5079
5080 static int
5081 cseg_wedge_thread SYSCTL_HANDLER_ARGS
5082 {
5083 #pragma unused(arg1, arg2)
5084
5085 int error, val = 0;
5086 error = sysctl_handle_int(oidp, &val, 0, req);
5087 if (error || val == 0) {
5088 return error;
5089 }
5090
5091 do_cseg_wedge_thread();
5092 return 0;
5093 }
5094 SYSCTL_PROC(_kern, OID_AUTO, cseg_wedge_thread, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_MASKED, 0, 0, cseg_wedge_thread, "I", "wedge c_seg thread");
5095
5096 static int
5097 cseg_unwedge_thread SYSCTL_HANDLER_ARGS
5098 {
5099 #pragma unused(arg1, arg2)
5100
5101 int error, val = 0;
5102 error = sysctl_handle_int(oidp, &val, 0, req);
5103 if (error || val == 0) {
5104 return error;
5105 }
5106
5107 do_cseg_unwedge_thread();
5108 return 0;
5109 }
5110 SYSCTL_PROC(_kern, OID_AUTO, cseg_unwedge_thread, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_MASKED, 0, 0, cseg_unwedge_thread, "I", "unstuck c_seg thread");
5111
5112 static atomic_int wedge_thread_should_wake = 0;
5113
5114 static int
5115 unwedge_thread SYSCTL_HANDLER_ARGS
5116 {
5117 #pragma unused(arg1, arg2)
5118 int error, val = 0;
5119 error = sysctl_handle_int(oidp, &val, 0, req);
5120 if (error || val == 0) {
5121 return error;
5122 }
5123
5124 atomic_store(&wedge_thread_should_wake, 1);
5125 return 0;
5126 }
5127
5128 SYSCTL_PROC(_kern, OID_AUTO, unwedge_thread, CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED, 0, 0, unwedge_thread, "I", "unwedge the thread wedged by kern.wedge_thread");
5129
5130 static int
5131 wedge_thread SYSCTL_HANDLER_ARGS
5132 {
5133 #pragma unused(arg1, arg2)
5134
5135 int error, val = 0;
5136 error = sysctl_handle_int(oidp, &val, 0, req);
5137 if (error || val == 0) {
5138 return error;
5139 }
5140
5141 uint64_t interval = 1;
5142 nanoseconds_to_absolutetime(1000 * 1000 * 50, &interval);
5143
5144 atomic_store(&wedge_thread_should_wake, 0);
5145 while (!atomic_load(&wedge_thread_should_wake)) {
5146 tsleep1(NULL, 0, "wedge_thread", mach_absolute_time() + interval, NULL);
5147 }
5148
5149 return 0;
5150 }
5151
5152 SYSCTL_PROC(_kern, OID_AUTO, wedge_thread,
5153 CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED, 0, 0, wedge_thread, "I",
5154 "wedge this thread so it cannot be cleaned up");
5155
5156 static int
5157 sysctl_total_corpses_count SYSCTL_HANDLER_ARGS
5158 {
5159 #pragma unused(oidp, arg1, arg2)
5160 extern unsigned long total_corpses_count(void);
5161
5162 unsigned long corpse_count_long = total_corpses_count();
5163 unsigned int corpse_count = (unsigned int)MIN(corpse_count_long, UINT_MAX);
5164 return sysctl_io_opaque(req, &corpse_count, sizeof(corpse_count), NULL);
5165 }
5166
5167 SYSCTL_PROC(_kern, OID_AUTO, total_corpses_count,
5168 CTLFLAG_RD | CTLFLAG_ANYBODY | CTLFLAG_LOCKED, 0, 0,
5169 sysctl_total_corpses_count, "I", "total corpses on the system");
5170
5171 static int
5172 sysctl_turnstile_test_prim_lock SYSCTL_HANDLER_ARGS;
5173 static int
5174 sysctl_turnstile_test_prim_unlock SYSCTL_HANDLER_ARGS;
5175 int
5176 tstile_test_prim_lock(boolean_t use_hashtable);
5177 int
5178 tstile_test_prim_unlock(boolean_t use_hashtable);
5179
5180 static int
5181 sysctl_turnstile_test_prim_lock SYSCTL_HANDLER_ARGS
5182 {
5183 #pragma unused(arg1, arg2)
5184 int error, val = 0;
5185 error = sysctl_handle_int(oidp, &val, 0, req);
5186 if (error || val == 0) {
5187 return error;
5188 }
5189 switch (val) {
5190 case SYSCTL_TURNSTILE_TEST_USER_DEFAULT:
5191 case SYSCTL_TURNSTILE_TEST_USER_HASHTABLE:
5192 case SYSCTL_TURNSTILE_TEST_KERNEL_DEFAULT:
5193 case SYSCTL_TURNSTILE_TEST_KERNEL_HASHTABLE:
5194 return tstile_test_prim_lock(val);
5195 default:
5196 return error;
5197 }
5198 }
5199
5200 static int
5201 sysctl_turnstile_test_prim_unlock SYSCTL_HANDLER_ARGS
5202 {
5203 #pragma unused(arg1, arg2)
5204 int error, val = 0;
5205 error = sysctl_handle_int(oidp, &val, 0, req);
5206 if (error || val == 0) {
5207 return error;
5208 }
5209 switch (val) {
5210 case SYSCTL_TURNSTILE_TEST_USER_DEFAULT:
5211 case SYSCTL_TURNSTILE_TEST_USER_HASHTABLE:
5212 case SYSCTL_TURNSTILE_TEST_KERNEL_DEFAULT:
5213 case SYSCTL_TURNSTILE_TEST_KERNEL_HASHTABLE:
5214 return tstile_test_prim_unlock(val);
5215 default:
5216 return error;
5217 }
5218 }
5219
5220 SYSCTL_PROC(_kern, OID_AUTO, turnstiles_test_lock, CTLFLAG_WR | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED,
5221 0, 0, sysctl_turnstile_test_prim_lock, "I", "turnstiles test lock");
5222
5223 SYSCTL_PROC(_kern, OID_AUTO, turnstiles_test_unlock, CTLFLAG_WR | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED,
5224 0, 0, sysctl_turnstile_test_prim_unlock, "I", "turnstiles test unlock");
5225
5226 int
5227 turnstile_get_boost_stats_sysctl(void *req);
5228 int
5229 turnstile_get_unboost_stats_sysctl(void *req);
5230 static int
5231 sysctl_turnstile_boost_stats SYSCTL_HANDLER_ARGS;
5232 static int
5233 sysctl_turnstile_unboost_stats SYSCTL_HANDLER_ARGS;
5234 extern uint64_t thread_block_on_turnstile_count;
5235 extern uint64_t thread_block_on_regular_waitq_count;
5236
5237 static int
5238 sysctl_turnstile_boost_stats SYSCTL_HANDLER_ARGS
5239 {
5240 #pragma unused(arg1, arg2, oidp)
5241 return turnstile_get_boost_stats_sysctl(req);
5242 }
5243
5244 static int
5245 sysctl_turnstile_unboost_stats SYSCTL_HANDLER_ARGS
5246 {
5247 #pragma unused(arg1, arg2, oidp)
5248 return turnstile_get_unboost_stats_sysctl(req);
5249 }
5250
5251 SYSCTL_PROC(_kern, OID_AUTO, turnstile_boost_stats, CTLFLAG_RD | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED | CTLTYPE_STRUCT,
5252 0, 0, sysctl_turnstile_boost_stats, "S", "turnstiles boost stats");
5253 SYSCTL_PROC(_kern, OID_AUTO, turnstile_unboost_stats, CTLFLAG_RD | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED | CTLTYPE_STRUCT,
5254 0, 0, sysctl_turnstile_unboost_stats, "S", "turnstiles unboost stats");
5255 SYSCTL_QUAD(_kern, OID_AUTO, thread_block_count_on_turnstile,
5256 CTLFLAG_RD | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED,
5257 &thread_block_on_turnstile_count, "thread blocked on turnstile count");
5258 SYSCTL_QUAD(_kern, OID_AUTO, thread_block_count_on_reg_waitq,
5259 CTLFLAG_RD | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED,
5260 &thread_block_on_regular_waitq_count, "thread blocked on regular waitq count");
5261
5262 #if defined(__x86_64__)
5263 extern uint64_t MutexSpin;
5264
5265 SYSCTL_QUAD(_kern, OID_AUTO, mutex_spin_abs, CTLFLAG_RW, &MutexSpin,
5266 "Spin time in abs for acquiring a kernel mutex");
5267 #else
5268 extern machine_timeout32_t MutexSpin;
5269
5270 SYSCTL_UINT(_kern, OID_AUTO, mutex_spin_abs, CTLFLAG_RW, &MutexSpin, 0,
5271 "Spin time in abs for acquiring a kernel mutex");
5272 #endif
5273
5274 extern uint64_t low_MutexSpin;
5275 extern int64_t high_MutexSpin;
5276 extern unsigned int real_ncpus;
5277
5278 SYSCTL_QUAD(_kern, OID_AUTO, low_mutex_spin_abs, CTLFLAG_RW, &low_MutexSpin,
5279 "Low spin threshold in abs for acquiring a kernel mutex");
5280
5281 static int
5282 sysctl_high_mutex_spin_ns SYSCTL_HANDLER_ARGS
5283 {
5284 #pragma unused(oidp, arg1, arg2)
5285 int error;
5286 int64_t val = 0;
5287 int64_t res;
5288
5289 /* Check if the user is writing to high_MutexSpin, or just reading it */
5290 if (req->newptr) {
5291 error = SYSCTL_IN(req, &val, sizeof(val));
5292 if (error || (val < 0 && val != -1)) {
5293 return error;
5294 }
5295 high_MutexSpin = val;
5296 }
5297
5298 if (high_MutexSpin >= 0) {
5299 res = high_MutexSpin;
5300 } else {
5301 res = low_MutexSpin * real_ncpus;
5302 }
5303 return SYSCTL_OUT(req, &res, sizeof(res));
5304 }
5305 SYSCTL_PROC(_kern, OID_AUTO, high_mutex_spin_abs, CTLFLAG_RW | CTLTYPE_QUAD, 0, 0, sysctl_high_mutex_spin_ns, "I",
5306 "High spin threshold in abs for acquiring a kernel mutex");
5307
5308
5309 #if defined (__x86_64__)
5310
5311 semaphore_t sysctl_test_panic_with_thread_sem;
5312
5313 #pragma clang diagnostic push
5314 #pragma clang diagnostic ignored "-Winfinite-recursion" /* rdar://38801963 */
5315 __attribute__((noreturn))
5316 static void
panic_thread_test_child_spin(void * arg,wait_result_t wres)5317 panic_thread_test_child_spin(void * arg, wait_result_t wres)
5318 {
5319 static int panic_thread_recurse_count = 5;
5320
5321 if (panic_thread_recurse_count > 0) {
5322 panic_thread_recurse_count--;
5323 panic_thread_test_child_spin(arg, wres);
5324 }
5325
5326 semaphore_signal(sysctl_test_panic_with_thread_sem);
5327 while (1) {
5328 ;
5329 }
5330 }
5331 #pragma clang diagnostic pop
5332
5333 static void
panic_thread_test_child_park(void * arg __unused,wait_result_t wres __unused)5334 panic_thread_test_child_park(void * arg __unused, wait_result_t wres __unused)
5335 {
5336 int event;
5337
5338 assert_wait(&event, THREAD_UNINT);
5339 semaphore_signal(sysctl_test_panic_with_thread_sem);
5340 thread_block(panic_thread_test_child_park);
5341 }
5342
5343 static int
5344 sysctl_test_panic_with_thread SYSCTL_HANDLER_ARGS
5345 {
5346 #pragma unused(arg1, arg2)
5347 int rval = 0;
5348 char str[16] = { '\0' };
5349 thread_t child_thread = THREAD_NULL;
5350
5351 rval = sysctl_handle_string(oidp, str, sizeof(str), req);
5352 if (rval != 0 || !req->newptr) {
5353 return EINVAL;
5354 }
5355
5356 semaphore_create(kernel_task, &sysctl_test_panic_with_thread_sem, SYNC_POLICY_FIFO, 0);
5357
5358 /* Create thread to spin or park in continuation */
5359 if (strncmp("spin", str, strlen("spin")) == 0) {
5360 if (kernel_thread_start(panic_thread_test_child_spin, NULL, &child_thread) != KERN_SUCCESS) {
5361 semaphore_destroy(kernel_task, sysctl_test_panic_with_thread_sem);
5362 return EBUSY;
5363 }
5364 } else if (strncmp("continuation", str, strlen("continuation")) == 0) {
5365 if (kernel_thread_start(panic_thread_test_child_park, NULL, &child_thread) != KERN_SUCCESS) {
5366 semaphore_destroy(kernel_task, sysctl_test_panic_with_thread_sem);
5367 return EBUSY;
5368 }
5369 } else {
5370 semaphore_destroy(kernel_task, sysctl_test_panic_with_thread_sem);
5371 return EINVAL;
5372 }
5373
5374 semaphore_wait(sysctl_test_panic_with_thread_sem);
5375
5376 panic_with_thread_context(0, NULL, 0, child_thread, "testing panic_with_thread_context for thread %p", child_thread);
5377
5378 /* Not reached */
5379 return EINVAL;
5380 }
5381
5382 SYSCTL_PROC(_kern, OID_AUTO, test_panic_with_thread,
5383 CTLFLAG_MASKED | CTLFLAG_KERN | CTLFLAG_LOCKED | CTLFLAG_WR | CTLTYPE_STRING,
5384 0, 0, sysctl_test_panic_with_thread, "A", "test panic flow for backtracing a different thread");
5385 #endif /* defined (__x86_64__) */
5386
5387 #endif /* DEVELOPMENT || DEBUG */
5388
5389 static int
5390 sysctl_get_owned_vmobjects SYSCTL_HANDLER_ARGS
5391 {
5392 #pragma unused(oidp, arg1, arg2)
5393
5394 /* validate */
5395 if (req->newlen != sizeof(mach_port_name_t) || req->newptr == USER_ADDR_NULL ||
5396 req->oldidx != 0 || req->newidx != 0 || req->p == NULL ||
5397 (req->oldlen == 0 && req->oldptr != USER_ADDR_NULL)) {
5398 return EINVAL;
5399 }
5400
5401 int error;
5402 mach_port_name_t task_port_name;
5403 task_t task;
5404 size_t buffer_size = (req->oldptr != USER_ADDR_NULL) ? req->oldlen : 0;
5405 vmobject_list_output_t buffer = NULL;
5406 size_t output_size;
5407 size_t entries;
5408
5409 /* we have a "newptr" (for write) we get a task port name from the caller. */
5410 error = SYSCTL_IN(req, &task_port_name, sizeof(mach_port_name_t));
5411
5412 if (error != 0) {
5413 goto sysctl_get_vmobject_list_exit;
5414 }
5415
5416 task = port_name_to_task_read(task_port_name);
5417 if (task == TASK_NULL) {
5418 error = ESRCH;
5419 goto sysctl_get_vmobject_list_exit;
5420 }
5421
5422 bool corpse = is_corpsetask(task);
5423
5424 /* get the current size */
5425 size_t max_size;
5426 task_get_owned_vmobjects(task, 0, NULL, &max_size, &entries);
5427
5428 if (buffer_size && (buffer_size < sizeof(*buffer) + sizeof(vm_object_query_data_t))) {
5429 error = ENOMEM;
5430 goto sysctl_get_vmobject_list_deallocate_and_exit;
5431 }
5432
5433 if (corpse == false) {
5434 /* copy the vmobjects and vmobject data out of the task */
5435 if (buffer_size == 0) {
5436 output_size = max_size;
5437 } else {
5438 buffer_size = (buffer_size > max_size) ? max_size : buffer_size;
5439 buffer = (struct _vmobject_list_output_ *)kalloc_data(buffer_size, Z_WAITOK);
5440
5441 if (!buffer) {
5442 error = ENOMEM;
5443 goto sysctl_get_vmobject_list_deallocate_and_exit;
5444 }
5445
5446 task_get_owned_vmobjects(task, buffer_size, buffer, &output_size, &entries);
5447 }
5448
5449 /* req->oldptr should be USER_ADDR_NULL if buffer == NULL and return the current size */
5450 /* otherwise copy buffer to oldptr and return the bytes copied */
5451 error = SYSCTL_OUT(req, (char *)buffer, output_size);
5452 } else {
5453 vmobject_list_output_t list;
5454
5455 task_get_corpse_vmobject_list(task, &list, &max_size);
5456 assert(buffer == NULL);
5457
5458 /* copy corpse_vmobject_list to output buffer to avoid double copy */
5459 if (buffer_size) {
5460 size_t temp_size;
5461
5462 temp_size = buffer_size > max_size ? max_size : buffer_size;
5463 output_size = temp_size - sizeof(*buffer);
5464 /* whole multiple of vm_object_query_data_t */
5465 output_size = (output_size / sizeof(vm_object_query_data_t)) * sizeof(vm_object_query_data_t) + sizeof(*buffer);
5466 buffer = list;
5467 } else {
5468 output_size = max_size;
5469 }
5470
5471 /* req->oldptr should be USER_ADDR_NULL if buffer == NULL and return the current size */
5472 /* otherwise copy buffer to oldptr and return the bytes copied */
5473 error = SYSCTL_OUT(req, (char*)buffer, output_size);
5474 buffer = NULL;
5475 }
5476
5477 sysctl_get_vmobject_list_deallocate_and_exit:
5478 task_deallocate(task);
5479
5480 sysctl_get_vmobject_list_exit:
5481 if (buffer) {
5482 kfree_data(buffer, buffer_size);
5483 }
5484
5485 return error;
5486 }
5487
5488 SYSCTL_PROC(_vm, OID_AUTO, get_owned_vmobjects,
5489 CTLTYPE_OPAQUE | CTLFLAG_RD | CTLFLAG_WR | CTLFLAG_MASKED | CTLFLAG_KERN | CTLFLAG_LOCKED | CTLFLAG_ANYBODY,
5490 0, 0, sysctl_get_owned_vmobjects, "A", "get owned vmobjects in task");
5491
5492 extern uint64_t num_static_scalable_counters;
5493 SYSCTL_QUAD(_kern, OID_AUTO, num_static_scalable_counters, CTLFLAG_RD | CTLFLAG_LOCKED, &num_static_scalable_counters, "");
5494
5495 uuid_string_t trial_treatment_id;
5496 uuid_string_t trial_experiment_id;
5497 int trial_deployment_id = -1;
5498
5499 SYSCTL_STRING(_kern, OID_AUTO, trial_treatment_id, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_ANYBODY | CTLFLAG_EXPERIMENT, trial_treatment_id, sizeof(trial_treatment_id), "");
5500 SYSCTL_STRING(_kern, OID_AUTO, trial_experiment_id, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_ANYBODY | CTLFLAG_EXPERIMENT, trial_experiment_id, sizeof(trial_experiment_id), "");
5501 SYSCTL_INT(_kern, OID_AUTO, trial_deployment_id, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_ANYBODY | CTLFLAG_EXPERIMENT, &trial_deployment_id, 0, "");
5502
5503 #if (DEVELOPMENT || DEBUG)
5504 /* For unit testing setting factors & limits. */
5505 unsigned int testing_experiment_factor;
5506 EXPERIMENT_FACTOR_UINT(_kern, testing_experiment_factor, &testing_experiment_factor, 5, 10, "");
5507
5508 extern int exception_log_max_pid;
5509 SYSCTL_INT(_debug, OID_AUTO, exception_log_max_pid, CTLFLAG_RW | CTLFLAG_LOCKED, &exception_log_max_pid, 0, "Log exceptions for all processes up to this pid");
5510 #endif /* (DEVELOPMENT || DEBUG) */
5511