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 case KERN_KDSET_EDM:
1156 case KERN_KDGET_EDM:
1157 ret = kdbg_control(name, namelen, oldp, oldlenp);
1158 break;
1159 default:
1160 ret = ENOTSUP;
1161 break;
1162 }
1163
1164 /* adjust index so we return the right required/consumed amount */
1165 if (!ret) {
1166 req->oldidx += req->oldlen;
1167 }
1168
1169 return ret;
1170 }
1171 SYSCTL_PROC(_kern, KERN_KDEBUG, kdebug, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
1172 0, /* Pointer argument (arg1) */
1173 0, /* Integer argument (arg2) */
1174 sysctl_kdebug_ops, /* Handler function */
1175 NULL, /* Data pointer */
1176 "");
1177
1178
1179 #if defined(XNU_TARGET_OS_OSX)
1180 /*
1181 * Return the top *sizep bytes of the user stack, or the entire area of the
1182 * user stack down through the saved exec_path, whichever is smaller.
1183 */
1184 STATIC int
1185 sysctl_doprocargs SYSCTL_HANDLER_ARGS
1186 {
1187 __unused int cmd = oidp->oid_arg2; /* subcommand*/
1188 int *name = arg1; /* oid element argument vector */
1189 int namelen = arg2; /* number of oid element arguments */
1190 user_addr_t oldp = req->oldptr; /* user buffer copy out address */
1191 size_t *oldlenp = &req->oldlen; /* user buffer copy out size */
1192 // user_addr_t newp = req->newptr; /* user buffer copy in address */
1193 // size_t newlen = req->newlen; /* user buffer copy in size */
1194 int error;
1195
1196 error = sysctl_procargsx( name, namelen, oldp, oldlenp, current_proc(), 0);
1197
1198 /* adjust index so we return the right required/consumed amount */
1199 if (!error) {
1200 req->oldidx += req->oldlen;
1201 }
1202
1203 return error;
1204 }
1205 SYSCTL_PROC(_kern, KERN_PROCARGS, procargs, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
1206 0, /* Pointer argument (arg1) */
1207 0, /* Integer argument (arg2) */
1208 sysctl_doprocargs, /* Handler function */
1209 NULL, /* Data pointer */
1210 "");
1211 #endif /* defined(XNU_TARGET_OS_OSX) */
1212
1213 STATIC int
1214 sysctl_doprocargs2 SYSCTL_HANDLER_ARGS
1215 {
1216 __unused int cmd = oidp->oid_arg2; /* subcommand*/
1217 int *name = arg1; /* oid element argument vector */
1218 int namelen = arg2; /* number of oid element arguments */
1219 user_addr_t oldp = req->oldptr; /* user buffer copy out address */
1220 size_t *oldlenp = &req->oldlen; /* user buffer copy out size */
1221 // user_addr_t newp = req->newptr; /* user buffer copy in address */
1222 // size_t newlen = req->newlen; /* user buffer copy in size */
1223 int error;
1224
1225 error = sysctl_procargsx( name, namelen, oldp, oldlenp, current_proc(), 1);
1226
1227 /* adjust index so we return the right required/consumed amount */
1228 if (!error) {
1229 req->oldidx += req->oldlen;
1230 }
1231
1232 return error;
1233 }
1234 SYSCTL_PROC(_kern, KERN_PROCARGS2, procargs2, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
1235 0, /* Pointer argument (arg1) */
1236 0, /* Integer argument (arg2) */
1237 sysctl_doprocargs2, /* Handler function */
1238 NULL, /* Data pointer */
1239 "");
1240
1241 #define SYSCTL_PROCARGS_READ_ENVVARS_ENTITLEMENT "com.apple.private.read-environment-variables"
1242 STATIC int
sysctl_procargsx(int * name,u_int namelen,user_addr_t where,size_t * sizep,proc_t cur_proc,int argc_yes)1243 sysctl_procargsx(int *name, u_int namelen, user_addr_t where,
1244 size_t *sizep, proc_t cur_proc, int argc_yes)
1245 {
1246 assert(sizep != NULL);
1247 proc_t p = NULL;
1248 size_t buflen = where != USER_ADDR_NULL ? *sizep : 0;
1249 int error = 0;
1250 struct _vm_map *proc_map = NULL;
1251 struct task * task;
1252 vm_map_copy_t tmp = NULL;
1253 user_addr_t arg_addr;
1254 size_t arg_size;
1255 caddr_t data;
1256 size_t argslen = 0;
1257 size_t size = 0;
1258 vm_offset_t copy_start = 0, copy_end;
1259 vm_offset_t smallbuffer_start;
1260 kern_return_t ret;
1261 int pid;
1262 kauth_cred_t my_cred;
1263 uid_t uid;
1264 int argc = -1;
1265 size_t argvsize;
1266 size_t remaining;
1267 size_t current_arg_index;
1268 size_t current_arg_len;
1269 const char * current_arg;
1270 bool omit_env_vars = true;
1271 user_addr_t user_stack;
1272 vm_map_offset_t effective_page_mask;
1273
1274 if (namelen < 1) {
1275 error = EINVAL;
1276 goto finish;
1277 }
1278
1279 if (argc_yes) {
1280 buflen -= sizeof(int); /* reserve first word to return argc */
1281 }
1282 /* we only care about buflen when where (oldp from sysctl) is not NULL. */
1283 /* when where (oldp from sysctl) is NULL and sizep (oldlenp from sysctl */
1284 /* is not NULL then the caller wants us to return the length needed to */
1285 /* hold the data we would return */
1286 if (where != USER_ADDR_NULL && (buflen <= 0 || buflen > ARG_MAX)) {
1287 error = EINVAL;
1288 goto finish;
1289 }
1290
1291 /*
1292 * Lookup process by pid
1293 */
1294 pid = name[0];
1295 p = proc_find(pid);
1296 if (p == NULL) {
1297 error = EINVAL;
1298 goto finish;
1299 }
1300
1301 /* Allow reading environment variables if any of the following are true:
1302 * - kernel is DEVELOPMENT || DEBUG
1303 * - target process is same as current_proc()
1304 * - target process is not cs_restricted
1305 * - SIP is off
1306 * - caller has an entitlement
1307 */
1308
1309 #if DEVELOPMENT || DEBUG
1310 omit_env_vars = false;
1311 #endif
1312 if (p == current_proc() ||
1313 !cs_restricted(p) ||
1314 #if CONFIG_CSR
1315 csr_check(CSR_ALLOW_UNRESTRICTED_DTRACE) == 0 ||
1316 #endif
1317 IOCurrentTaskHasEntitlement(SYSCTL_PROCARGS_READ_ENVVARS_ENTITLEMENT)
1318 ) {
1319 omit_env_vars = false;
1320 }
1321
1322 /*
1323 * Copy the top N bytes of the stack.
1324 * On all machines we have so far, the stack grows
1325 * downwards.
1326 *
1327 * If the user expects no more than N bytes of
1328 * argument list, use that as a guess for the
1329 * size.
1330 */
1331
1332 if (!p->user_stack) {
1333 error = EINVAL;
1334 goto finish;
1335 }
1336
1337 /* save off argc, argslen, user_stack before releasing the proc */
1338 argc = p->p_argc;
1339 argslen = p->p_argslen;
1340 user_stack = p->user_stack;
1341
1342 /*
1343 * When these sysctls were introduced, the first string in the strings
1344 * section was just the bare path of the executable. However, for security
1345 * reasons we now prefix this string with executable_path= so it can be
1346 * parsed getenv style. To avoid binary compatability issues with exising
1347 * callers of this sysctl, we strip it off here.
1348 * (rdar://problem/13746466)
1349 */
1350 #define EXECUTABLE_KEY "executable_path="
1351 argslen -= strlen(EXECUTABLE_KEY);
1352
1353 if (where == USER_ADDR_NULL && !omit_env_vars) {
1354 /* caller only wants to know length of proc args data.
1355 * If we don't need to omit environment variables, we can skip
1356 * copying the target process stack */
1357 goto calculate_size;
1358 }
1359
1360 my_cred = kauth_cred_proc_ref(p);
1361 uid = kauth_cred_getuid(my_cred);
1362 kauth_cred_unref(&my_cred);
1363
1364 if ((uid != kauth_cred_getuid(kauth_cred_get()))
1365 && suser(kauth_cred_get(), &cur_proc->p_acflag)) {
1366 error = EINVAL;
1367 goto finish;
1368 }
1369
1370 /*
1371 * Before we can block (any VM code), make another
1372 * reference to the map to keep it alive. We do
1373 * that by getting a reference on the task itself.
1374 */
1375 task = p->task;
1376 if (task == NULL) {
1377 error = EINVAL;
1378 goto finish;
1379 }
1380
1381 /*
1382 * Once we have a task reference we can convert that into a
1383 * map reference, which we will use in the calls below. The
1384 * task/process may change its map after we take this reference
1385 * (see execve), but the worst that will happen then is a return
1386 * of stale info (which is always a possibility).
1387 */
1388 task_reference(task);
1389 proc_rele(p);
1390 p = NULL;
1391 proc_map = get_task_map_reference(task);
1392 task_deallocate(task);
1393
1394 if (proc_map == NULL) {
1395 error = EINVAL;
1396 goto finish;
1397 }
1398
1399 effective_page_mask = vm_map_page_mask(proc_map);
1400
1401 arg_size = vm_map_round_page(argslen, effective_page_mask);
1402
1403 arg_addr = user_stack - arg_size;
1404
1405 ret = kmem_alloc(kernel_map, ©_start, arg_size,
1406 KMA_DATA | KMA_ZERO, VM_KERN_MEMORY_BSD);
1407 if (ret != KERN_SUCCESS) {
1408 error = ENOMEM;
1409 goto finish;
1410 }
1411
1412 copy_end = copy_start + arg_size;
1413
1414 if (vm_map_copyin(proc_map, (vm_map_address_t)arg_addr,
1415 (vm_map_size_t)arg_size, FALSE, &tmp) != KERN_SUCCESS) {
1416 error = EIO;
1417 goto finish;
1418 }
1419
1420 /*
1421 * Now that we've done the copyin from the process'
1422 * map, we can release the reference to it.
1423 */
1424 vm_map_deallocate(proc_map);
1425 proc_map = NULL;
1426
1427 if (vm_map_copy_overwrite(kernel_map,
1428 (vm_map_address_t)copy_start,
1429 tmp, (vm_map_size_t) arg_size, FALSE) != KERN_SUCCESS) {
1430 error = EIO;
1431 goto finish;
1432 }
1433 /* tmp was consumed */
1434 tmp = NULL;
1435
1436 if (omit_env_vars) {
1437 argvsize = 0;
1438
1439 /* Iterate over everything in argv, plus one for the bare executable path */
1440 for (current_arg_index = 0; current_arg_index < argc + 1 && argvsize < argslen; ++current_arg_index) {
1441 current_arg = (const char *)(copy_end - argslen) + argvsize;
1442 remaining = argslen - argvsize;
1443 current_arg_len = strnlen(current_arg, remaining);
1444 if (current_arg_len < remaining) {
1445 /* We have space for the null terminator */
1446 current_arg_len += 1;
1447
1448 if (current_arg_index == 0) {
1449 /* The bare executable path may have multiple null bytes after it for alignment */
1450 while (current_arg_len < remaining && current_arg[current_arg_len] == 0) {
1451 current_arg_len += 1;
1452 }
1453 }
1454 }
1455 argvsize += current_arg_len;
1456 }
1457 assert(argvsize <= argslen);
1458
1459 /* Adjust argslen and copy_end to make the copyout range extend to the end of argv */
1460 copy_end = copy_end - argslen + argvsize;
1461 argslen = argvsize;
1462 }
1463
1464 if (where == USER_ADDR_NULL) {
1465 /* Skip copyout */
1466 goto calculate_size;
1467 }
1468
1469 if (buflen >= argslen) {
1470 data = (caddr_t) (copy_end - argslen);
1471 size = argslen;
1472 } else {
1473 /*
1474 * Before rdar://25397314, this function contained incorrect logic when buflen is less
1475 * than argslen. The problem was that it copied in `buflen` bytes from the end of the target
1476 * process user stack into the beginning of a buffer of size round_page(buflen), and then
1477 * copied out `buflen` bytes from the end of this buffer. The effect of this was that
1478 * the caller of this sysctl would get zeros at the end of their buffer.
1479 *
1480 * To preserve this behavior, bzero everything from copy_end-round_page(buflen)+buflen to the
1481 * end of the buffer. This emulates copying in only `buflen` bytes.
1482 *
1483 *
1484 * In the old code:
1485 *
1486 * copy_start .... size: round_page(buflen) .... copy_end
1487 * [---copied in data (size: buflen)---|--- zeros ----------]
1488 * ^
1489 * data = copy_end - buflen
1490 *
1491 *
1492 * In the new code:
1493 * copy_start .... size: round_page(p->argslen) .... full copy_end
1494 * ^ ....................... p->argslen ...............................^
1495 * ^ ^ truncated copy_end ^
1496 * ^ ^ ^ ^
1497 * ^ ................ argslen ........................ ^
1498 * ^ ^ ^ ^
1499 * [-------copied in data (size: round_page(p->argslen))-------:----env vars---]
1500 * ^ ^
1501 * ^ data = copy_end - buflen
1502 * smallbuffer_start = max(copy_end - round_page(buflen), copy_start)
1503 *
1504 *
1505 * Full copy_end: copy_end calculated from copy_start + round_page(p->argslen)
1506 * Truncated copy_end: copy_end after truncation to remove environment variables.
1507 *
1508 * If environment variables were omitted, then we use the truncated copy_end, otherwise
1509 * we use full copy_end.
1510 *
1511 * smallbuffer_start: represents where copy_start would be in the old code.
1512 * data: The beginning of the region we copyout
1513 */
1514 smallbuffer_start = copy_end - vm_map_round_page(buflen, effective_page_mask);
1515 if (smallbuffer_start < copy_start) {
1516 smallbuffer_start = copy_start;
1517 }
1518 bzero((void *)(smallbuffer_start + buflen), copy_end - (smallbuffer_start + buflen));
1519 data = (caddr_t) (copy_end - buflen);
1520 size = buflen;
1521 }
1522
1523 if (argc_yes) {
1524 /* Put processes argc as the first word in the copyout buffer */
1525 suword(where, argc);
1526 error = copyout(data, (where + sizeof(int)), size);
1527 size += sizeof(int);
1528 } else {
1529 error = copyout(data, where, size);
1530
1531 /*
1532 * Make the old PROCARGS work to return the executable's path
1533 * But, only if there is enough space in the provided buffer
1534 *
1535 * on entry: data [possibily] points to the beginning of the path
1536 *
1537 * Note: we keep all pointers&sizes aligned to word boundries
1538 */
1539 if ((!error) && (buflen > 0 && (u_int)buflen > size)) {
1540 int binPath_sz, alignedBinPath_sz = 0;
1541 int extraSpaceNeeded, addThis;
1542 user_addr_t placeHere;
1543 char * str = (char *) data;
1544 size_t max_len = size;
1545
1546 /* Some apps are really bad about messing up their stacks
1547 * So, we have to be extra careful about getting the length
1548 * of the executing binary. If we encounter an error, we bail.
1549 */
1550
1551 /* Limit ourselves to PATH_MAX paths */
1552 if (max_len > PATH_MAX) {
1553 max_len = PATH_MAX;
1554 }
1555
1556 binPath_sz = 0;
1557
1558 while ((binPath_sz < max_len - 1) && (*str++ != 0)) {
1559 binPath_sz++;
1560 }
1561
1562 /* If we have a NUL terminator, copy it, too */
1563 if (binPath_sz < max_len - 1) {
1564 binPath_sz += 1;
1565 }
1566
1567 /* Pre-Flight the space requiremnts */
1568
1569 /* Account for the padding that fills out binPath to the next word */
1570 alignedBinPath_sz += (binPath_sz & (sizeof(int) - 1)) ? (sizeof(int) - (binPath_sz & (sizeof(int) - 1))) : 0;
1571
1572 placeHere = where + size;
1573
1574 /* Account for the bytes needed to keep placeHere word aligned */
1575 addThis = (placeHere & (sizeof(int) - 1)) ? (sizeof(int) - (placeHere & (sizeof(int) - 1))) : 0;
1576
1577 /* Add up all the space that is needed */
1578 extraSpaceNeeded = alignedBinPath_sz + addThis + binPath_sz + (4 * sizeof(int));
1579
1580 /* is there is room to tack on argv[0]? */
1581 if ((buflen & ~(sizeof(int) - 1)) >= (size + extraSpaceNeeded)) {
1582 placeHere += addThis;
1583 suword(placeHere, 0);
1584 placeHere += sizeof(int);
1585 suword(placeHere, 0xBFFF0000);
1586 placeHere += sizeof(int);
1587 suword(placeHere, 0);
1588 placeHere += sizeof(int);
1589 error = copyout(data, placeHere, binPath_sz);
1590 if (!error) {
1591 placeHere += binPath_sz;
1592 suword(placeHere, 0);
1593 size += extraSpaceNeeded;
1594 }
1595 }
1596 }
1597 }
1598
1599 calculate_size:
1600 /* Size has already been calculated for the where != NULL case */
1601 if (where == USER_ADDR_NULL) {
1602 size = argslen;
1603 if (argc_yes) {
1604 size += sizeof(int);
1605 } else {
1606 /*
1607 * old PROCARGS will return the executable's path and plus some
1608 * extra space for work alignment and data tags
1609 */
1610 size += PATH_MAX + (6 * sizeof(int));
1611 }
1612 size += (size & (sizeof(int) - 1)) ? (sizeof(int) - (size & (sizeof(int) - 1))) : 0;
1613 }
1614
1615 *sizep = size;
1616
1617 finish:
1618 if (p != NULL) {
1619 proc_rele(p);
1620 }
1621 if (tmp != NULL) {
1622 vm_map_copy_discard(tmp);
1623 }
1624 if (proc_map != NULL) {
1625 vm_map_deallocate(proc_map);
1626 }
1627 if (copy_start != (vm_offset_t) 0) {
1628 kmem_free(kernel_map, copy_start, arg_size);
1629 }
1630 return error;
1631 }
1632
1633
1634 /*
1635 * Max number of concurrent aio requests
1636 */
1637 STATIC int
sysctl_aiomax(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)1638 sysctl_aiomax
1639 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1640 {
1641 int new_value, changed;
1642 int error = sysctl_io_number(req, aio_max_requests, sizeof(int), &new_value, &changed);
1643 if (changed) {
1644 /* make sure the system-wide limit is greater than the per process limit */
1645 if (new_value >= aio_max_requests_per_process && new_value <= AIO_MAX_REQUESTS) {
1646 aio_max_requests = new_value;
1647 } else {
1648 error = EINVAL;
1649 }
1650 }
1651 return error;
1652 }
1653
1654
1655 /*
1656 * Max number of concurrent aio requests per process
1657 */
1658 STATIC int
sysctl_aioprocmax(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)1659 sysctl_aioprocmax
1660 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1661 {
1662 int new_value, changed;
1663 int error = sysctl_io_number(req, aio_max_requests_per_process, sizeof(int), &new_value, &changed);
1664 if (changed) {
1665 /* make sure per process limit is less than the system-wide limit */
1666 if (new_value <= aio_max_requests && new_value >= AIO_LISTIO_MAX) {
1667 aio_max_requests_per_process = new_value;
1668 } else {
1669 error = EINVAL;
1670 }
1671 }
1672 return error;
1673 }
1674
1675
1676 /*
1677 * Max number of async IO worker threads
1678 */
1679 STATIC int
sysctl_aiothreads(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)1680 sysctl_aiothreads
1681 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1682 {
1683 int new_value, changed;
1684 int error = sysctl_io_number(req, aio_worker_threads, sizeof(int), &new_value, &changed);
1685 if (changed) {
1686 /* we only allow an increase in the number of worker threads */
1687 if (new_value > aio_worker_threads) {
1688 _aio_create_worker_threads((new_value - aio_worker_threads));
1689 aio_worker_threads = new_value;
1690 } else {
1691 error = EINVAL;
1692 }
1693 }
1694 return error;
1695 }
1696
1697
1698 /*
1699 * System-wide limit on the max number of processes
1700 */
1701 STATIC int
sysctl_maxproc(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)1702 sysctl_maxproc
1703 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1704 {
1705 int new_value, changed;
1706 int error = sysctl_io_number(req, maxproc, sizeof(int), &new_value, &changed);
1707 if (changed) {
1708 AUDIT_ARG(value32, new_value);
1709 /* make sure the system-wide limit is less than the configured hard
1710 * limit set at kernel compilation */
1711 if (new_value <= hard_maxproc && new_value > 0) {
1712 maxproc = new_value;
1713 } else {
1714 error = EINVAL;
1715 }
1716 }
1717 return error;
1718 }
1719
1720 extern int sched_enable_smt;
1721 STATIC int
sysctl_sched_enable_smt(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)1722 sysctl_sched_enable_smt
1723 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1724 {
1725 int new_value, changed;
1726 int error = sysctl_io_number(req, sched_enable_smt, sizeof(int), &new_value, &changed);
1727 if (error) {
1728 return error;
1729 }
1730 kern_return_t kret = KERN_SUCCESS;
1731 if (changed) {
1732 AUDIT_ARG(value32, new_value);
1733 if (new_value == 0) {
1734 sched_enable_smt = 0;
1735 kret = enable_smt_processors(false);
1736 } else {
1737 sched_enable_smt = 1;
1738 kret = enable_smt_processors(true);
1739 }
1740 }
1741 switch (kret) {
1742 case KERN_SUCCESS:
1743 error = 0;
1744 break;
1745 case KERN_INVALID_ARGUMENT:
1746 error = EINVAL;
1747 break;
1748 case KERN_FAILURE:
1749 error = EBUSY;
1750 break;
1751 default:
1752 error = ENOENT;
1753 break;
1754 }
1755
1756 return error;
1757 }
1758
1759 SYSCTL_STRING(_kern, KERN_OSTYPE, ostype,
1760 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
1761 ostype, 0, "");
1762 SYSCTL_STRING(_kern, KERN_OSRELEASE, osrelease,
1763 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
1764 osrelease, 0, "");
1765 SYSCTL_INT(_kern, KERN_OSREV, osrevision,
1766 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
1767 (int *)NULL, BSD, "");
1768 SYSCTL_STRING(_kern, KERN_VERSION, version,
1769 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
1770 version, 0, "");
1771 SYSCTL_STRING(_kern, OID_AUTO, uuid,
1772 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
1773 &kernel_uuid_string[0], 0, "");
1774
1775 SYSCTL_STRING(_kern, OID_AUTO, osbuildconfig,
1776 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED | CTLFLAG_MASKED,
1777 &osbuild_config[0], 0, "");
1778
1779 #if DEBUG
1780 #ifndef DKPR
1781 #define DKPR 1
1782 #endif
1783 #endif
1784
1785 #if DKPR
1786 int debug_kprint_syscall = 0;
1787 char debug_kprint_syscall_process[MAXCOMLEN + 1];
1788
1789 /* Thread safe: bits and string value are not used to reclaim state */
1790 SYSCTL_INT(_debug, OID_AUTO, kprint_syscall,
1791 CTLFLAG_RW | CTLFLAG_LOCKED, &debug_kprint_syscall, 0, "kprintf syscall tracing");
1792 SYSCTL_STRING(_debug, OID_AUTO, kprint_syscall_process,
1793 CTLFLAG_RW | CTLFLAG_LOCKED, debug_kprint_syscall_process, sizeof(debug_kprint_syscall_process),
1794 "name of process for kprintf syscall tracing");
1795
1796 int
debug_kprint_current_process(const char ** namep)1797 debug_kprint_current_process(const char **namep)
1798 {
1799 struct proc *p = current_proc();
1800
1801 if (p == NULL) {
1802 return 0;
1803 }
1804
1805 if (debug_kprint_syscall_process[0]) {
1806 /* user asked to scope tracing to a particular process name */
1807 if (0 == strncmp(debug_kprint_syscall_process,
1808 p->p_comm, sizeof(debug_kprint_syscall_process))) {
1809 /* no value in telling the user that we traced what they asked */
1810 if (namep) {
1811 *namep = NULL;
1812 }
1813
1814 return 1;
1815 } else {
1816 return 0;
1817 }
1818 }
1819
1820 /* trace all processes. Tell user what we traced */
1821 if (namep) {
1822 *namep = p->p_comm;
1823 }
1824
1825 return 1;
1826 }
1827 #endif
1828
1829 /* PR-5293665: need to use a callback function for kern.osversion to set
1830 * osversion in IORegistry */
1831
1832 STATIC int
sysctl_osversion(__unused struct sysctl_oid * oidp,void * arg1,int arg2,struct sysctl_req * req)1833 sysctl_osversion(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
1834 {
1835 int rval = 0;
1836
1837 rval = sysctl_handle_string(oidp, arg1, arg2, req);
1838
1839 if (req->newptr) {
1840 IORegistrySetOSBuildVersion((char *)arg1);
1841 }
1842
1843 return rval;
1844 }
1845
1846 SYSCTL_PROC(_kern, KERN_OSVERSION, osversion,
1847 CTLFLAG_RW | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
1848 osversion, 256 /* OSVERSIZE*/,
1849 sysctl_osversion, "A", "");
1850
1851 static bool
_already_set_or_not_launchd(struct sysctl_req * req,char * val)1852 _already_set_or_not_launchd(struct sysctl_req *req, char *val)
1853 {
1854 if (req->newptr != 0) {
1855 /*
1856 * Can only ever be set by launchd, and only once at boot.
1857 */
1858 if (proc_getpid(req->p) != 1 || val[0] != '\0') {
1859 return true;
1860 }
1861 }
1862 return false;
1863 }
1864
1865 #if XNU_TARGET_OS_OSX
1866 static int
sysctl_system_version_compat(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)1867 sysctl_system_version_compat
1868 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1869 {
1870 int oldval = (task_has_system_version_compat_enabled(current_task()));
1871 int new_value = 0, changed = 0;
1872
1873 int error = sysctl_io_number(req, oldval, sizeof(int), &new_value, &changed);
1874 if (changed) {
1875 task_set_system_version_compat_enabled(current_task(), (new_value));
1876 }
1877 return error;
1878 }
1879
1880 SYSCTL_PROC(_kern, OID_AUTO, system_version_compat,
1881 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
1882 0, 0, sysctl_system_version_compat, "A", "");
1883
1884 char osproductversioncompat[48] = { '\0' };
1885
1886 static int
sysctl_osproductversioncompat(struct sysctl_oid * oidp,void * arg1,int arg2,struct sysctl_req * req)1887 sysctl_osproductversioncompat(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
1888 {
1889 if (_already_set_or_not_launchd(req, osproductversioncompat)) {
1890 return EPERM;
1891 }
1892 return sysctl_handle_string(oidp, arg1, arg2, req);
1893 }
1894
1895
1896 SYSCTL_PROC(_kern, OID_AUTO, osproductversioncompat,
1897 CTLFLAG_RW | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
1898 osproductversioncompat, sizeof(osproductversioncompat),
1899 sysctl_osproductversioncompat, "A", "The ProductVersion from SystemVersionCompat.plist");
1900 #endif
1901
1902 char osproductversion[48] = { '\0' };
1903
1904 static char iossupportversion_string[48] = { '\0' };
1905
1906 static int
sysctl_osproductversion(__unused struct sysctl_oid * oidp,void * arg1,int arg2,struct sysctl_req * req)1907 sysctl_osproductversion(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
1908 {
1909 if (_already_set_or_not_launchd(req, osproductversion)) {
1910 return EPERM;
1911 }
1912
1913 #if XNU_TARGET_OS_OSX
1914 if (task_has_system_version_compat_enabled(current_task()) && (osproductversioncompat[0] != '\0')) {
1915 return sysctl_handle_string(oidp, osproductversioncompat, arg2, req);
1916 } else {
1917 return sysctl_handle_string(oidp, arg1, arg2, req);
1918 }
1919 #else
1920 return sysctl_handle_string(oidp, arg1, arg2, req);
1921 #endif
1922 }
1923
1924 #if XNU_TARGET_OS_OSX
1925 static_assert(sizeof(osproductversioncompat) == sizeof(osproductversion),
1926 "osproductversion size matches osproductversioncompat size");
1927 #endif
1928
1929 SYSCTL_PROC(_kern, OID_AUTO, osproductversion,
1930 CTLFLAG_RW | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
1931 osproductversion, sizeof(osproductversion),
1932 sysctl_osproductversion, "A", "The ProductVersion from SystemVersion.plist");
1933
1934 char osreleasetype[48] = { '\0' };
1935
1936 STATIC int
sysctl_osreleasetype(__unused struct sysctl_oid * oidp,void * arg1,int arg2,struct sysctl_req * req)1937 sysctl_osreleasetype(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
1938 {
1939 if (_already_set_or_not_launchd(req, osreleasetype)) {
1940 return EPERM;
1941 }
1942 return sysctl_handle_string(oidp, arg1, arg2, req);
1943 }
1944
1945 void reset_osreleasetype(void);
1946
1947 void
reset_osreleasetype(void)1948 reset_osreleasetype(void)
1949 {
1950 memset(osreleasetype, 0, sizeof(osreleasetype));
1951 }
1952
1953 SYSCTL_PROC(_kern, OID_AUTO, osreleasetype,
1954 CTLFLAG_RW | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
1955 osreleasetype, sizeof(osreleasetype),
1956 sysctl_osreleasetype, "A", "The ReleaseType from SystemVersion.plist");
1957
1958 STATIC int
sysctl_iossupportversion(__unused struct sysctl_oid * oidp,void * arg1,int arg2,struct sysctl_req * req)1959 sysctl_iossupportversion(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
1960 {
1961 if (_already_set_or_not_launchd(req, iossupportversion_string)) {
1962 return EPERM;
1963 }
1964
1965 return sysctl_handle_string(oidp, arg1, arg2, req);
1966 }
1967
1968 SYSCTL_PROC(_kern, OID_AUTO, iossupportversion,
1969 CTLFLAG_RW | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
1970 iossupportversion_string, sizeof(iossupportversion_string),
1971 sysctl_iossupportversion, "A", "The iOSSupportVersion from SystemVersion.plist");
1972
1973 static uint64_t osvariant_status = 0;
1974
1975 STATIC int
sysctl_osvariant_status(__unused struct sysctl_oid * oidp,void * arg1,int arg2,struct sysctl_req * req)1976 sysctl_osvariant_status(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
1977 {
1978 if (req->newptr != 0) {
1979 /*
1980 * Can only ever be set by launchd, and only once.
1981 * Reset by usrctl() -> reset_osvariant_status() during
1982 * userspace reboot, since userspace could reboot into
1983 * a different variant.
1984 */
1985 if (proc_getpid(req->p) != 1 || osvariant_status != 0) {
1986 return EPERM;
1987 }
1988 }
1989
1990 return sysctl_handle_quad(oidp, arg1, arg2, req);
1991 }
1992
1993 SYSCTL_PROC(_kern, OID_AUTO, osvariant_status,
1994 CTLFLAG_RW | CTLTYPE_QUAD | CTLFLAG_LOCKED | CTLFLAG_MASKED,
1995 &osvariant_status, sizeof(osvariant_status),
1996 sysctl_osvariant_status, "Q", "Opaque flags used to cache OS variant information");
1997
1998 void reset_osvariant_status(void);
1999
2000 void
reset_osvariant_status(void)2001 reset_osvariant_status(void)
2002 {
2003 osvariant_status = 0;
2004 }
2005
2006 extern void commpage_update_dyld_flags(uint64_t);
2007 TUNABLE_WRITEABLE(uint64_t, dyld_flags, "dyld_flags", 0);
2008
2009 STATIC int
sysctl_dyld_flags(__unused struct sysctl_oid * oidp,void * arg1,int arg2,struct sysctl_req * req)2010 sysctl_dyld_flags(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2011 {
2012 /*
2013 * Can only ever be set by launchd, possibly several times
2014 * as dyld may change its mind after a userspace reboot.
2015 */
2016 if (req->newptr != 0 && proc_getpid(req->p) != 1) {
2017 return EPERM;
2018 }
2019
2020 int res = sysctl_handle_quad(oidp, arg1, arg2, req);
2021 if (req->newptr && res == 0) {
2022 commpage_update_dyld_flags(dyld_flags);
2023 }
2024 return res;
2025 }
2026
2027 SYSCTL_PROC(_kern, OID_AUTO, dyld_flags,
2028 CTLFLAG_RW | CTLTYPE_QUAD | CTLFLAG_LOCKED | CTLFLAG_MASKED,
2029 &dyld_flags, sizeof(dyld_flags),
2030 sysctl_dyld_flags, "Q", "Opaque flags used to cache dyld system-wide configuration");
2031
2032 #if defined(XNU_TARGET_OS_BRIDGE)
2033 char macosproductversion[MACOS_VERS_LEN] = { '\0' };
2034
2035 SYSCTL_STRING(_kern, OID_AUTO, macosproductversion,
2036 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2037 &macosproductversion[0], MACOS_VERS_LEN, "The currently running macOS ProductVersion (from SystemVersion.plist on macOS)");
2038
2039 char macosversion[MACOS_VERS_LEN] = { '\0' };
2040
2041 SYSCTL_STRING(_kern, OID_AUTO, macosversion,
2042 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2043 &macosversion[0], MACOS_VERS_LEN, "The currently running macOS build version");
2044 #endif
2045
2046 STATIC int
sysctl_sysctl_bootargs(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2047 sysctl_sysctl_bootargs
2048 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2049 {
2050 int error;
2051 char buf[BOOT_LINE_LENGTH];
2052
2053 strlcpy(buf, PE_boot_args(), BOOT_LINE_LENGTH);
2054 error = sysctl_io_string(req, buf, BOOT_LINE_LENGTH, 0, NULL);
2055 return error;
2056 }
2057
2058 SYSCTL_PROC(_kern, OID_AUTO, bootargs,
2059 CTLFLAG_LOCKED | CTLFLAG_RD | CTLFLAG_KERN | CTLTYPE_STRING,
2060 NULL, 0,
2061 sysctl_sysctl_bootargs, "A", "bootargs");
2062
2063 STATIC int
sysctl_kernelcacheuuid(struct sysctl_oid * oidp,void * arg1,int arg2,struct sysctl_req * req)2064 sysctl_kernelcacheuuid(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2065 {
2066 int rval = ENOENT;
2067 if (kernelcache_uuid_valid) {
2068 rval = sysctl_handle_string(oidp, arg1, arg2, req);
2069 }
2070 return rval;
2071 }
2072
2073 SYSCTL_PROC(_kern, OID_AUTO, kernelcacheuuid,
2074 CTLFLAG_RD | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2075 kernelcache_uuid_string, sizeof(kernelcache_uuid_string),
2076 sysctl_kernelcacheuuid, "A", "");
2077
2078 STATIC int
sysctl_systemfilesetuuid(struct sysctl_oid * oidp,void * arg1,int arg2,struct sysctl_req * req)2079 sysctl_systemfilesetuuid(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2080 {
2081 int rval = ENOENT;
2082 if (pageablekc_uuid_valid) {
2083 rval = sysctl_handle_string(oidp, arg1, arg2, req);
2084 }
2085 return rval;
2086 }
2087
2088 SYSCTL_PROC(_kern, OID_AUTO, systemfilesetuuid,
2089 CTLFLAG_RD | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2090 pageablekc_uuid_string, sizeof(pageablekc_uuid_string),
2091 sysctl_systemfilesetuuid, "A", "");
2092
2093 STATIC int
sysctl_auxiliaryfilesetuuid(struct sysctl_oid * oidp,void * arg1,int arg2,struct sysctl_req * req)2094 sysctl_auxiliaryfilesetuuid(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2095 {
2096 int rval = ENOENT;
2097 if (auxkc_uuid_valid) {
2098 rval = sysctl_handle_string(oidp, arg1, arg2, req);
2099 }
2100 return rval;
2101 }
2102
2103 SYSCTL_PROC(_kern, OID_AUTO, auxiliaryfilesetuuid,
2104 CTLFLAG_RD | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2105 auxkc_uuid_string, sizeof(auxkc_uuid_string),
2106 sysctl_auxiliaryfilesetuuid, "A", "");
2107
2108 STATIC int
sysctl_filesetuuid(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2109 sysctl_filesetuuid(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2110 {
2111 int rval = ENOENT;
2112 kc_format_t kcformat;
2113 kernel_mach_header_t *mh;
2114 void *uuid = NULL;
2115 unsigned long uuidlen = 0;
2116 uuid_string_t uuid_str;
2117
2118 if (!PE_get_primary_kc_format(&kcformat) || kcformat != KCFormatFileset) {
2119 return rval;
2120 }
2121
2122 mh = (kernel_mach_header_t *)PE_get_kc_header(KCKindPrimary);
2123 uuid = getuuidfromheader(mh, &uuidlen);
2124
2125 if ((uuid != NULL) && (uuidlen == sizeof(uuid_t))) {
2126 uuid_unparse_upper(*(uuid_t *)uuid, uuid_str);
2127 rval = sysctl_io_string(req, (char *)uuid_str, sizeof(uuid_str), 0, NULL);
2128 }
2129
2130 return rval;
2131 }
2132
2133 SYSCTL_PROC(_kern, OID_AUTO, filesetuuid,
2134 CTLFLAG_RD | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2135 NULL, 0,
2136 sysctl_filesetuuid, "A", "");
2137
2138
2139 SYSCTL_INT(_kern, KERN_MAXFILES, maxfiles,
2140 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2141 &maxfiles, 0, "");
2142 SYSCTL_INT(_kern, KERN_ARGMAX, argmax,
2143 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2144 (int *)NULL, ARG_MAX, "");
2145 SYSCTL_INT(_kern, KERN_POSIX1, posix1version,
2146 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2147 (int *)NULL, _POSIX_VERSION, "");
2148 SYSCTL_INT(_kern, KERN_NGROUPS, ngroups,
2149 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2150 (int *)NULL, NGROUPS_MAX, "");
2151 SYSCTL_INT(_kern, KERN_JOB_CONTROL, job_control,
2152 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2153 (int *)NULL, 1, "");
2154 #if 1 /* _POSIX_SAVED_IDS from <unistd.h> */
2155 SYSCTL_INT(_kern, KERN_SAVED_IDS, saved_ids,
2156 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2157 (int *)NULL, 1, "");
2158 #else
2159 SYSCTL_INT(_kern, KERN_SAVED_IDS, saved_ids,
2160 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2161 NULL, 0, "");
2162 #endif
2163 SYSCTL_INT(_kern, OID_AUTO, num_files,
2164 CTLFLAG_RD | CTLFLAG_LOCKED,
2165 &nfiles, 0, "");
2166 SYSCTL_COMPAT_INT(_kern, OID_AUTO, num_vnodes,
2167 CTLFLAG_RD | CTLFLAG_LOCKED,
2168 &numvnodes, 0, "");
2169 SYSCTL_INT(_kern, OID_AUTO, num_tasks,
2170 CTLFLAG_RD | CTLFLAG_LOCKED,
2171 &task_max, 0, "");
2172 SYSCTL_INT(_kern, OID_AUTO, num_threads,
2173 CTLFLAG_RD | CTLFLAG_LOCKED,
2174 &thread_max, 0, "");
2175 SYSCTL_INT(_kern, OID_AUTO, num_taskthreads,
2176 CTLFLAG_RD | CTLFLAG_LOCKED,
2177 &task_threadmax, 0, "");
2178 SYSCTL_LONG(_kern, OID_AUTO, num_recycledvnodes,
2179 CTLFLAG_RD | CTLFLAG_LOCKED,
2180 &num_recycledvnodes, "");
2181 SYSCTL_COMPAT_INT(_kern, OID_AUTO, free_vnodes,
2182 CTLFLAG_RD | CTLFLAG_LOCKED,
2183 &freevnodes, 0, "");
2184
2185 STATIC int
sysctl_maxvnodes(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2186 sysctl_maxvnodes(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2187 {
2188 int oldval = desiredvnodes;
2189 int error = sysctl_io_number(req, desiredvnodes, sizeof(int), &desiredvnodes, NULL);
2190
2191 if (oldval != desiredvnodes) {
2192 resize_namecache(desiredvnodes);
2193 }
2194
2195 return error;
2196 }
2197
2198 SYSCTL_INT(_kern, OID_AUTO, namecache_disabled,
2199 CTLFLAG_RW | CTLFLAG_LOCKED,
2200 &nc_disabled, 0, "");
2201
2202 SYSCTL_PROC(_kern, KERN_MAXVNODES, maxvnodes,
2203 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2204 0, 0, sysctl_maxvnodes, "I", "");
2205
2206 SYSCTL_PROC(_kern, KERN_MAXPROC, maxproc,
2207 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2208 0, 0, sysctl_maxproc, "I", "");
2209
2210 SYSCTL_PROC(_kern, KERN_AIOMAX, aiomax,
2211 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2212 0, 0, sysctl_aiomax, "I", "");
2213
2214 SYSCTL_PROC(_kern, KERN_AIOPROCMAX, aioprocmax,
2215 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2216 0, 0, sysctl_aioprocmax, "I", "");
2217
2218 SYSCTL_PROC(_kern, KERN_AIOTHREADS, aiothreads,
2219 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2220 0, 0, sysctl_aiothreads, "I", "");
2221
2222 SYSCTL_PROC(_kern, OID_AUTO, sched_enable_smt,
2223 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_KERN,
2224 0, 0, sysctl_sched_enable_smt, "I", "");
2225
2226 extern int sched_allow_NO_SMT_threads;
2227 SYSCTL_INT(_kern, OID_AUTO, sched_allow_NO_SMT_threads,
2228 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2229 &sched_allow_NO_SMT_threads, 0, "");
2230
2231 extern int sched_avoid_cpu0;
2232 SYSCTL_INT(_kern, OID_AUTO, sched_rt_avoid_cpu0,
2233 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2234 &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");
2235
2236 #if (DEVELOPMENT || DEBUG)
2237 extern int smt_sched_bonus_16ths;
2238 SYSCTL_INT(_kern, OID_AUTO, smt_sched_bonus_16ths,
2239 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2240 &smt_sched_bonus_16ths, 0, "");
2241
2242 extern int smt_timeshare_enabled;
2243 SYSCTL_INT(_kern, OID_AUTO, sched_smt_timeshare_enable,
2244 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2245 &smt_timeshare_enabled, 0, "");
2246
2247 extern int sched_smt_balance;
2248 SYSCTL_INT(_kern, OID_AUTO, sched_smt_balance,
2249 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2250 &sched_smt_balance, 0, "");
2251 extern int sched_allow_rt_smt;
2252 SYSCTL_INT(_kern, OID_AUTO, sched_allow_rt_smt,
2253 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2254 &sched_allow_rt_smt, 0, "");
2255 extern int sched_allow_rt_steal;
2256 SYSCTL_INT(_kern, OID_AUTO, sched_allow_rt_steal,
2257 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2258 &sched_allow_rt_steal, 0, "");
2259 extern int sched_choose_first_fd_processor;
2260 SYSCTL_INT(_kern, OID_AUTO, sched_choose_first_fd_processor,
2261 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2262 &sched_choose_first_fd_processor, 0, "");
2263 extern int sched_backup_cpu_timeout_count;
2264 SYSCTL_INT(_kern, OID_AUTO, sched_backup_cpu_timeout_count,
2265 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2266 &sched_backup_cpu_timeout_count, 0, "The maximum number of 10us delays before allowing a backup cpu to select a thread");
2267 #if __arm__ || __arm64__
2268 extern uint32_t perfcontrol_requested_recommended_cores;
2269 SYSCTL_UINT(_kern, OID_AUTO, sched_recommended_cores,
2270 CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
2271 &perfcontrol_requested_recommended_cores, 0, "");
2272
2273 /* Scheduler perfcontrol callouts sysctls */
2274 SYSCTL_DECL(_kern_perfcontrol_callout);
2275 SYSCTL_NODE(_kern, OID_AUTO, perfcontrol_callout, CTLFLAG_RW | CTLFLAG_LOCKED, 0,
2276 "scheduler perfcontrol callouts");
2277
2278 extern int perfcontrol_callout_stats_enabled;
2279 SYSCTL_INT(_kern_perfcontrol_callout, OID_AUTO, stats_enabled,
2280 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2281 &perfcontrol_callout_stats_enabled, 0, "");
2282
2283 extern uint64_t perfcontrol_callout_stat_avg(perfcontrol_callout_type_t type,
2284 perfcontrol_callout_stat_t stat);
2285
2286 /* On-Core Callout */
2287 STATIC int
sysctl_perfcontrol_callout_stat(__unused struct sysctl_oid * oidp,void * arg1,int arg2,struct sysctl_req * req)2288 sysctl_perfcontrol_callout_stat
2289 (__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2290 {
2291 perfcontrol_callout_stat_t stat = (perfcontrol_callout_stat_t)arg1;
2292 perfcontrol_callout_type_t type = (perfcontrol_callout_type_t)arg2;
2293 return sysctl_io_number(req, (int)perfcontrol_callout_stat_avg(type, stat),
2294 sizeof(int), NULL, NULL);
2295 }
2296
2297 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, oncore_instr,
2298 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2299 (void *)PERFCONTROL_STAT_INSTRS, PERFCONTROL_CALLOUT_ON_CORE,
2300 sysctl_perfcontrol_callout_stat, "I", "");
2301 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, oncore_cycles,
2302 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2303 (void *)PERFCONTROL_STAT_CYCLES, PERFCONTROL_CALLOUT_ON_CORE,
2304 sysctl_perfcontrol_callout_stat, "I", "");
2305 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, offcore_instr,
2306 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2307 (void *)PERFCONTROL_STAT_INSTRS, PERFCONTROL_CALLOUT_OFF_CORE,
2308 sysctl_perfcontrol_callout_stat, "I", "");
2309 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, offcore_cycles,
2310 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2311 (void *)PERFCONTROL_STAT_CYCLES, PERFCONTROL_CALLOUT_OFF_CORE,
2312 sysctl_perfcontrol_callout_stat, "I", "");
2313 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, context_instr,
2314 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2315 (void *)PERFCONTROL_STAT_INSTRS, PERFCONTROL_CALLOUT_CONTEXT,
2316 sysctl_perfcontrol_callout_stat, "I", "");
2317 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, context_cycles,
2318 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2319 (void *)PERFCONTROL_STAT_CYCLES, PERFCONTROL_CALLOUT_CONTEXT,
2320 sysctl_perfcontrol_callout_stat, "I", "");
2321 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, update_instr,
2322 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2323 (void *)PERFCONTROL_STAT_INSTRS, PERFCONTROL_CALLOUT_STATE_UPDATE,
2324 sysctl_perfcontrol_callout_stat, "I", "");
2325 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, update_cycles,
2326 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2327 (void *)PERFCONTROL_STAT_CYCLES, PERFCONTROL_CALLOUT_STATE_UPDATE,
2328 sysctl_perfcontrol_callout_stat, "I", "");
2329
2330 #if __AMP__
2331 extern int sched_amp_idle_steal;
2332 SYSCTL_INT(_kern, OID_AUTO, sched_amp_idle_steal,
2333 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2334 &sched_amp_idle_steal, 0, "");
2335 extern int sched_amp_spill_steal;
2336 SYSCTL_INT(_kern, OID_AUTO, sched_amp_spill_steal,
2337 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2338 &sched_amp_spill_steal, 0, "");
2339 extern int sched_amp_spill_count;
2340 SYSCTL_INT(_kern, OID_AUTO, sched_amp_spill_count,
2341 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2342 &sched_amp_spill_count, 0, "");
2343 extern int sched_amp_spill_deferred_ipi;
2344 SYSCTL_INT(_kern, OID_AUTO, sched_amp_spill_deferred_ipi,
2345 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2346 &sched_amp_spill_deferred_ipi, 0, "");
2347 extern int sched_amp_pcores_preempt_immediate_ipi;
2348 SYSCTL_INT(_kern, OID_AUTO, sched_amp_pcores_preempt_immediate_ipi,
2349 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2350 &sched_amp_pcores_preempt_immediate_ipi, 0, "");
2351 #endif /* __AMP__ */
2352 #endif /* __arm__ || __arm64__ */
2353
2354 #if __arm64__
2355 extern int legacy_footprint_entitlement_mode;
2356 SYSCTL_INT(_kern, OID_AUTO, legacy_footprint_entitlement_mode,
2357 CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
2358 &legacy_footprint_entitlement_mode, 0, "");
2359 #endif /* __arm64__ */
2360
2361 /*
2362 * Realtime threads are ordered by highest priority first then,
2363 * for threads of the same priority, by earliest deadline first.
2364 * But if sched_rt_runq_strict_priority is false (the default),
2365 * a lower priority thread with an earlier deadline will be preferred
2366 * over a higher priority thread with a later deadline, as long as
2367 * both threads' computations will fit before the later deadline.
2368 */
2369 extern int sched_rt_runq_strict_priority;
2370 SYSCTL_INT(_kern, OID_AUTO, sched_rt_runq_strict_priority,
2371 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2372 &sched_rt_runq_strict_priority, 0, "");
2373
2374 static int
sysctl_kern_sched_rt_n_backup_processors(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2375 sysctl_kern_sched_rt_n_backup_processors(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2376 {
2377 int new_value, changed;
2378 int old_value = sched_get_rt_n_backup_processors();
2379 int error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
2380 if (changed) {
2381 sched_set_rt_n_backup_processors(new_value);
2382 }
2383
2384 return error;
2385 }
2386
2387 SYSCTL_PROC(_kern, OID_AUTO, sched_rt_n_backup_processors,
2388 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2389 0, 0, sysctl_kern_sched_rt_n_backup_processors, "I", "");
2390
2391 static int
sysctl_kern_sched_rt_constraint_ll_us(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2392 sysctl_kern_sched_rt_constraint_ll_us(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2393 {
2394 int new_value, changed;
2395 int old_value = sched_get_rt_constraint_ll();
2396 int error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
2397 if (changed) {
2398 sched_set_rt_constraint_ll(new_value);
2399 }
2400
2401 return error;
2402 }
2403
2404 SYSCTL_PROC(_kern, OID_AUTO, sched_rt_constraint_ll_us,
2405 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2406 0, 0, sysctl_kern_sched_rt_constraint_ll_us, "I", "");
2407
2408 static int
sysctl_kern_sched_rt_deadline_epsilon_us(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2409 sysctl_kern_sched_rt_deadline_epsilon_us(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2410 {
2411 int new_value, changed;
2412 int old_value = sched_get_rt_deadline_epsilon();
2413 int error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
2414 if (changed) {
2415 sched_set_rt_deadline_epsilon(new_value);
2416 }
2417
2418 return error;
2419 }
2420
2421 SYSCTL_PROC(_kern, OID_AUTO, sched_rt_deadline_epsilon_us,
2422 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2423 0, 0, sysctl_kern_sched_rt_deadline_epsilon_us, "I", "");
2424
2425 extern int sched_idle_delay_cpuid;
2426 SYSCTL_INT(_kern, OID_AUTO, sched_idle_delay_cpuid,
2427 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2428 &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");
2429
2430 #endif /* (DEVELOPMENT || DEBUG) */
2431
2432 STATIC int
sysctl_securelvl(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2433 sysctl_securelvl
2434 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2435 {
2436 int new_value, changed;
2437 int error = sysctl_io_number(req, securelevel, sizeof(int), &new_value, &changed);
2438 if (changed) {
2439 if (!(new_value < securelevel && proc_getpid(req->p) != 1)) {
2440 proc_list_lock();
2441 securelevel = new_value;
2442 proc_list_unlock();
2443 } else {
2444 error = EPERM;
2445 }
2446 }
2447 return error;
2448 }
2449
2450 SYSCTL_PROC(_kern, KERN_SECURELVL, securelevel,
2451 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2452 0, 0, sysctl_securelvl, "I", "");
2453
2454
2455 STATIC int
sysctl_domainname(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2456 sysctl_domainname
2457 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2458 {
2459 int error, changed;
2460 char tmpname[MAXHOSTNAMELEN] = {};
2461
2462 lck_mtx_lock(&domainname_lock);
2463 strlcpy(tmpname, domainname, sizeof(tmpname));
2464 lck_mtx_unlock(&domainname_lock);
2465
2466 error = sysctl_io_string(req, tmpname, sizeof(tmpname), 0, &changed);
2467 if (!error && changed) {
2468 lck_mtx_lock(&hostname_lock);
2469 strlcpy(domainname, tmpname, sizeof(domainname));
2470 lck_mtx_unlock(&hostname_lock);
2471 }
2472 return error;
2473 }
2474
2475 SYSCTL_PROC(_kern, KERN_DOMAINNAME, nisdomainname,
2476 CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_LOCKED,
2477 0, 0, sysctl_domainname, "A", "");
2478
2479 SYSCTL_COMPAT_INT(_kern, KERN_HOSTID, hostid,
2480 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2481 &hostid, 0, "");
2482
2483 STATIC int
sysctl_hostname(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2484 sysctl_hostname
2485 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2486 {
2487 int error, changed;
2488 char tmpname[MAXHOSTNAMELEN] = {};
2489
2490 lck_mtx_lock(&hostname_lock);
2491 strlcpy(tmpname, hostname, sizeof(tmpname));
2492 lck_mtx_unlock(&hostname_lock);
2493
2494 error = sysctl_io_string(req, tmpname, sizeof(tmpname), 1, &changed);
2495 if (!error && changed) {
2496 lck_mtx_lock(&hostname_lock);
2497 strlcpy(hostname, tmpname, sizeof(hostname));
2498 lck_mtx_unlock(&hostname_lock);
2499 }
2500 return error;
2501 }
2502
2503 SYSCTL_PROC(_kern, KERN_HOSTNAME, hostname,
2504 CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_LOCKED,
2505 0, 0, sysctl_hostname, "A", "");
2506
2507 STATIC int
sysctl_procname(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2508 sysctl_procname
2509 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2510 {
2511 /* Original code allowed writing, I'm copying this, although this all makes
2512 * no sense to me. Besides, this sysctl is never used. */
2513 return sysctl_io_string(req, &req->p->p_name[0], (2 * MAXCOMLEN + 1), 1, NULL);
2514 }
2515
2516 SYSCTL_PROC(_kern, KERN_PROCNAME, procname,
2517 CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
2518 0, 0, sysctl_procname, "A", "");
2519
2520 SYSCTL_INT(_kern, KERN_SPECULATIVE_READS, speculative_reads_disabled,
2521 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2522 &speculative_reads_disabled, 0, "");
2523
2524 SYSCTL_UINT(_kern, OID_AUTO, preheat_max_bytes,
2525 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2526 &preheat_max_bytes, 0, "");
2527
2528 SYSCTL_UINT(_kern, OID_AUTO, preheat_min_bytes,
2529 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2530 &preheat_min_bytes, 0, "");
2531
2532 SYSCTL_UINT(_kern, OID_AUTO, speculative_prefetch_max,
2533 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2534 &speculative_prefetch_max, 0, "");
2535
2536 SYSCTL_UINT(_kern, OID_AUTO, speculative_prefetch_max_iosize,
2537 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2538 &speculative_prefetch_max_iosize, 0, "");
2539
2540 SYSCTL_UINT(_kern, OID_AUTO, vm_page_free_target,
2541 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2542 &vm_page_free_target, 0, "");
2543
2544 SYSCTL_UINT(_kern, OID_AUTO, vm_page_free_min,
2545 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2546 &vm_page_free_min, 0, "");
2547
2548 SYSCTL_UINT(_kern, OID_AUTO, vm_page_free_reserved,
2549 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2550 &vm_page_free_reserved, 0, "");
2551
2552 SYSCTL_UINT(_kern, OID_AUTO, vm_page_speculative_percentage,
2553 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2554 &vm_pageout_state.vm_page_speculative_percentage, 0, "");
2555
2556 SYSCTL_UINT(_kern, OID_AUTO, vm_page_speculative_q_age_ms,
2557 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2558 &vm_pageout_state.vm_page_speculative_q_age_ms, 0, "");
2559
2560 SYSCTL_UINT(_kern, OID_AUTO, vm_max_delayed_work_limit,
2561 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2562 &vm_max_delayed_work_limit, 0, "");
2563
2564 SYSCTL_UINT(_kern, OID_AUTO, vm_max_batch,
2565 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2566 &vm_max_batch, 0, "");
2567
2568 SYSCTL_STRING(_kern, OID_AUTO, bootsessionuuid,
2569 CTLFLAG_RD | CTLFLAG_LOCKED,
2570 &bootsessionuuid_string, sizeof(bootsessionuuid_string), "");
2571
2572
2573 STATIC int
sysctl_boottime(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2574 sysctl_boottime
2575 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2576 {
2577 struct timeval tv;
2578 boottime_timeval(&tv);
2579 struct proc *p = req->p;
2580
2581 if (proc_is64bit(p)) {
2582 struct user64_timeval t = {};
2583 t.tv_sec = tv.tv_sec;
2584 t.tv_usec = tv.tv_usec;
2585 return sysctl_io_opaque(req, &t, sizeof(t), NULL);
2586 } else {
2587 struct user32_timeval t = {};
2588 t.tv_sec = (user32_time_t)tv.tv_sec;
2589 t.tv_usec = tv.tv_usec;
2590 return sysctl_io_opaque(req, &t, sizeof(t), NULL);
2591 }
2592 }
2593
2594 SYSCTL_PROC(_kern, KERN_BOOTTIME, boottime,
2595 CTLTYPE_STRUCT | CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
2596 0, 0, sysctl_boottime, "S,timeval", "");
2597
2598 extern bool IOGetBootUUID(char *);
2599
2600 /* non-static: written by imageboot.c */
2601 uuid_string_t fake_bootuuid;
2602
2603 STATIC int
sysctl_bootuuid(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2604 sysctl_bootuuid
2605 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2606 {
2607 int error = ENOENT;
2608
2609 /* check the first byte to see if the string has been
2610 * populated. this is a uuid_STRING_t, this check would
2611 * not work with a uuid_t.
2612 */
2613 if (fake_bootuuid[0] != '\0') {
2614 error = sysctl_io_string(req, fake_bootuuid, 0, 0, NULL);
2615 goto out;
2616 }
2617
2618 uuid_string_t uuid_string;
2619 if (IOGetBootUUID(uuid_string)) {
2620 uuid_t boot_uuid;
2621 error = uuid_parse(uuid_string, boot_uuid);
2622 if (!error) {
2623 error = sysctl_io_string(req, __DECONST(char *, uuid_string), 0, 0, NULL);
2624 }
2625 }
2626
2627 out:
2628 return error;
2629 }
2630
2631 SYSCTL_PROC(_kern, OID_AUTO, bootuuid,
2632 CTLTYPE_STRING | CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
2633 0, 0, sysctl_bootuuid, "A", "");
2634
2635
2636 extern bool IOGetApfsPrebootUUID(char *);
2637 extern bool IOGetAssociatedApfsVolgroupUUID(char *);
2638
2639 STATIC int
sysctl_apfsprebootuuid(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2640 sysctl_apfsprebootuuid
2641 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2642 {
2643 int error = ENOENT;
2644
2645 uuid_string_t uuid_string;
2646 if (IOGetApfsPrebootUUID(uuid_string)) {
2647 uuid_t apfs_preboot_uuid;
2648 error = uuid_parse(uuid_string, apfs_preboot_uuid);
2649 if (!error) {
2650 error = sysctl_io_string(req, __DECONST(char *, uuid_string), 0, 0, NULL);
2651 }
2652 }
2653
2654 return error;
2655 }
2656
2657 SYSCTL_PROC(_kern, OID_AUTO, apfsprebootuuid,
2658 CTLTYPE_STRING | CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
2659 0, 0, sysctl_apfsprebootuuid, "A", "");
2660
2661 STATIC int
sysctl_targetsystemvolgroupuuid(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2662 sysctl_targetsystemvolgroupuuid
2663 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2664 {
2665 int error = ENOENT;
2666
2667 uuid_string_t uuid_string;
2668 if (IOGetApfsPrebootUUID(uuid_string)) {
2669 uuid_t apfs_preboot_uuid;
2670 error = uuid_parse(uuid_string, apfs_preboot_uuid);
2671 if (!error) {
2672 error = sysctl_io_string(req, __DECONST(char *, uuid_string), 0, 0, NULL);
2673 }
2674 } else {
2675 /*
2676 * In special boot modes, such as kcgen-mode, the
2677 * apfs-preboot-uuid property will not be set. Instead, a
2678 * different property, associated-volume-group, will be set
2679 * which indicates the UUID of the VolumeGroup containing the
2680 * system volume into which you will boot.
2681 */
2682 if (IOGetAssociatedApfsVolgroupUUID(uuid_string)) {
2683 uuid_t apfs_preboot_uuid;
2684 error = uuid_parse(uuid_string, apfs_preboot_uuid);
2685 if (!error) {
2686 error = sysctl_io_string(req, __DECONST(char *, uuid_string), 0, 0, NULL);
2687 }
2688 }
2689 }
2690
2691 return error;
2692 }
2693
2694 SYSCTL_PROC(_kern, OID_AUTO, targetsystemvolgroupuuid,
2695 CTLTYPE_STRING | CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_MASKED | CTLFLAG_LOCKED,
2696 0, 0, sysctl_targetsystemvolgroupuuid, "A", "");
2697
2698 STATIC int
sysctl_symfile(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2699 sysctl_symfile
2700 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2701 {
2702 char *str;
2703 int error = get_kernel_symfile(req->p, &str);
2704 if (error) {
2705 return error;
2706 }
2707 return sysctl_io_string(req, str, 0, 0, NULL);
2708 }
2709
2710
2711 SYSCTL_PROC(_kern, KERN_SYMFILE, symfile,
2712 CTLTYPE_STRING | CTLFLAG_RD | CTLFLAG_LOCKED,
2713 0, 0, sysctl_symfile, "A", "");
2714
2715 #if CONFIG_NFS_CLIENT
2716 STATIC int
sysctl_netboot(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2717 sysctl_netboot
2718 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2719 {
2720 return sysctl_io_number(req, netboot_root(), sizeof(int), NULL, NULL);
2721 }
2722
2723 SYSCTL_PROC(_kern, KERN_NETBOOT, netboot,
2724 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2725 0, 0, sysctl_netboot, "I", "");
2726 #endif
2727
2728 #ifdef CONFIG_IMGSRC_ACCESS
2729 /*
2730 * Legacy--act as if only one layer of nesting is possible.
2731 */
2732 STATIC int
sysctl_imgsrcdev(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2733 sysctl_imgsrcdev
2734 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2735 {
2736 vfs_context_t ctx = vfs_context_current();
2737 vnode_t devvp;
2738 int result;
2739
2740 if (!vfs_context_issuser(ctx)) {
2741 return EPERM;
2742 }
2743
2744 if (imgsrc_rootvnodes[0] == NULL) {
2745 return ENOENT;
2746 }
2747
2748 result = vnode_getwithref(imgsrc_rootvnodes[0]);
2749 if (result != 0) {
2750 return result;
2751 }
2752
2753 devvp = vnode_mount(imgsrc_rootvnodes[0])->mnt_devvp;
2754 result = vnode_getwithref(devvp);
2755 if (result != 0) {
2756 goto out;
2757 }
2758
2759 result = sysctl_io_number(req, vnode_specrdev(devvp), sizeof(dev_t), NULL, NULL);
2760
2761 vnode_put(devvp);
2762 out:
2763 vnode_put(imgsrc_rootvnodes[0]);
2764 return result;
2765 }
2766
2767 SYSCTL_PROC(_kern, OID_AUTO, imgsrcdev,
2768 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2769 0, 0, sysctl_imgsrcdev, "I", "");
2770
2771 STATIC int
sysctl_imgsrcinfo(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2772 sysctl_imgsrcinfo
2773 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2774 {
2775 int error;
2776 struct imgsrc_info info[MAX_IMAGEBOOT_NESTING] = {}; /* 2 for now, no problem */
2777 uint32_t i;
2778 vnode_t rvp, devvp;
2779
2780 if (imgsrc_rootvnodes[0] == NULLVP) {
2781 return ENXIO;
2782 }
2783
2784 for (i = 0; i < MAX_IMAGEBOOT_NESTING; i++) {
2785 /*
2786 * Go get the root vnode.
2787 */
2788 rvp = imgsrc_rootvnodes[i];
2789 if (rvp == NULLVP) {
2790 break;
2791 }
2792
2793 error = vnode_get(rvp);
2794 if (error != 0) {
2795 return error;
2796 }
2797
2798 /*
2799 * For now, no getting at a non-local volume.
2800 */
2801 devvp = vnode_mount(rvp)->mnt_devvp;
2802 if (devvp == NULL) {
2803 vnode_put(rvp);
2804 return EINVAL;
2805 }
2806
2807 error = vnode_getwithref(devvp);
2808 if (error != 0) {
2809 vnode_put(rvp);
2810 return error;
2811 }
2812
2813 /*
2814 * Fill in info.
2815 */
2816 info[i].ii_dev = vnode_specrdev(devvp);
2817 info[i].ii_flags = 0;
2818 info[i].ii_height = i;
2819 bzero(info[i].ii_reserved, sizeof(info[i].ii_reserved));
2820
2821 vnode_put(devvp);
2822 vnode_put(rvp);
2823 }
2824
2825 return sysctl_io_opaque(req, info, i * sizeof(info[0]), NULL);
2826 }
2827
2828 SYSCTL_PROC(_kern, OID_AUTO, imgsrcinfo,
2829 CTLTYPE_OPAQUE | CTLFLAG_RD | CTLFLAG_LOCKED,
2830 0, 0, sysctl_imgsrcinfo, "I", "");
2831
2832 #endif /* CONFIG_IMGSRC_ACCESS */
2833
2834
2835 SYSCTL_DECL(_kern_timer);
2836 SYSCTL_NODE(_kern, OID_AUTO, timer, CTLFLAG_RW | CTLFLAG_LOCKED, 0, "timer");
2837
2838
2839 SYSCTL_INT(_kern_timer, OID_AUTO, coalescing_enabled,
2840 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2841 &mach_timer_coalescing_enabled, 0, "");
2842
2843 SYSCTL_QUAD(_kern_timer, OID_AUTO, deadline_tracking_bin_1,
2844 CTLFLAG_RW | CTLFLAG_LOCKED,
2845 &timer_deadline_tracking_bin_1, "");
2846 SYSCTL_QUAD(_kern_timer, OID_AUTO, deadline_tracking_bin_2,
2847 CTLFLAG_RW | CTLFLAG_LOCKED,
2848 &timer_deadline_tracking_bin_2, "");
2849
2850 SYSCTL_DECL(_kern_timer_longterm);
2851 SYSCTL_NODE(_kern_timer, OID_AUTO, longterm, CTLFLAG_RW | CTLFLAG_LOCKED, 0, "longterm");
2852
2853
2854 /* Must match definition in osfmk/kern/timer_call.c */
2855 enum {
2856 THRESHOLD, QCOUNT,
2857 ENQUEUES, DEQUEUES, ESCALATES, SCANS, PREEMPTS,
2858 LATENCY, LATENCY_MIN, LATENCY_MAX, SCAN_LIMIT, SCAN_INTERVAL, PAUSES
2859 };
2860 extern uint64_t timer_sysctl_get(int);
2861 extern int timer_sysctl_set(int, uint64_t);
2862
2863 STATIC int
sysctl_timer(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2864 sysctl_timer
2865 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2866 {
2867 int oid = (int)arg1;
2868 uint64_t value = timer_sysctl_get(oid);
2869 uint64_t new_value;
2870 int error;
2871 int changed;
2872
2873 error = sysctl_io_number(req, value, sizeof(value), &new_value, &changed);
2874 if (changed) {
2875 error = timer_sysctl_set(oid, new_value);
2876 }
2877
2878 return error;
2879 }
2880
2881 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, threshold,
2882 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
2883 (void *) THRESHOLD, 0, sysctl_timer, "Q", "");
2884 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, scan_limit,
2885 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
2886 (void *) SCAN_LIMIT, 0, sysctl_timer, "Q", "");
2887 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, scan_interval,
2888 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
2889 (void *) SCAN_INTERVAL, 0, sysctl_timer, "Q", "");
2890
2891 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, qlen,
2892 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
2893 (void *) QCOUNT, 0, sysctl_timer, "Q", "");
2894 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, scan_pauses,
2895 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
2896 (void *) PAUSES, 0, sysctl_timer, "Q", "");
2897
2898 #if DEBUG
2899 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, enqueues,
2900 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
2901 (void *) ENQUEUES, 0, sysctl_timer, "Q", "");
2902 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, dequeues,
2903 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
2904 (void *) DEQUEUES, 0, sysctl_timer, "Q", "");
2905 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, escalates,
2906 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
2907 (void *) ESCALATES, 0, sysctl_timer, "Q", "");
2908 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, scans,
2909 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
2910 (void *) SCANS, 0, sysctl_timer, "Q", "");
2911 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, preempts,
2912 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
2913 (void *) PREEMPTS, 0, sysctl_timer, "Q", "");
2914 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, latency,
2915 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
2916 (void *) LATENCY, 0, sysctl_timer, "Q", "");
2917 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, latency_min,
2918 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
2919 (void *) LATENCY_MIN, 0, sysctl_timer, "Q", "");
2920 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, latency_max,
2921 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
2922 (void *) LATENCY_MAX, 0, sysctl_timer, "Q", "");
2923 #endif /* DEBUG */
2924
2925 STATIC int
sysctl_usrstack(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2926 sysctl_usrstack
2927 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2928 {
2929 return sysctl_io_number(req, (int)req->p->user_stack, sizeof(int), NULL, NULL);
2930 }
2931
2932 SYSCTL_PROC(_kern, KERN_USRSTACK32, usrstack,
2933 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2934 0, 0, sysctl_usrstack, "I", "");
2935
2936 STATIC int
sysctl_usrstack64(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2937 sysctl_usrstack64
2938 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2939 {
2940 return sysctl_io_number(req, req->p->user_stack, sizeof(req->p->user_stack), NULL, NULL);
2941 }
2942
2943 SYSCTL_PROC(_kern, KERN_USRSTACK64, usrstack64,
2944 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
2945 0, 0, sysctl_usrstack64, "Q", "");
2946
2947 #if CONFIG_COREDUMP
2948
2949 SYSCTL_STRING(_kern, KERN_COREFILE, corefile,
2950 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2951 corefilename, sizeof(corefilename), "");
2952
2953 STATIC int
sysctl_coredump(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2954 sysctl_coredump
2955 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2956 {
2957 #ifdef SECURE_KERNEL
2958 (void)req;
2959 return ENOTSUP;
2960 #else
2961 int new_value, changed;
2962 int error = sysctl_io_number(req, do_coredump, sizeof(int), &new_value, &changed);
2963 if (changed) {
2964 if ((new_value == 0) || (new_value == 1)) {
2965 do_coredump = new_value;
2966 } else {
2967 error = EINVAL;
2968 }
2969 }
2970 return error;
2971 #endif
2972 }
2973
2974 SYSCTL_PROC(_kern, KERN_COREDUMP, coredump,
2975 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2976 0, 0, sysctl_coredump, "I", "");
2977
2978 STATIC int
sysctl_suid_coredump(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)2979 sysctl_suid_coredump
2980 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2981 {
2982 #ifdef SECURE_KERNEL
2983 (void)req;
2984 return ENOTSUP;
2985 #else
2986 int new_value, changed;
2987 int error = sysctl_io_number(req, sugid_coredump, sizeof(int), &new_value, &changed);
2988 if (changed) {
2989 if ((new_value == 0) || (new_value == 1)) {
2990 sugid_coredump = new_value;
2991 } else {
2992 error = EINVAL;
2993 }
2994 }
2995 return error;
2996 #endif
2997 }
2998
2999 SYSCTL_PROC(_kern, KERN_SUGID_COREDUMP, sugid_coredump,
3000 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
3001 0, 0, sysctl_suid_coredump, "I", "");
3002
3003 #endif /* CONFIG_COREDUMP */
3004
3005 STATIC int
sysctl_delayterm(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3006 sysctl_delayterm
3007 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3008 {
3009 struct proc *p = req->p;
3010 int new_value, changed;
3011 int error = sysctl_io_number(req, (req->p->p_lflag & P_LDELAYTERM)? 1: 0, sizeof(int), &new_value, &changed);
3012 if (changed) {
3013 proc_lock(p);
3014 if (new_value) {
3015 req->p->p_lflag |= P_LDELAYTERM;
3016 } else {
3017 req->p->p_lflag &= ~P_LDELAYTERM;
3018 }
3019 proc_unlock(p);
3020 }
3021 return error;
3022 }
3023
3024 SYSCTL_PROC(_kern, KERN_PROCDELAYTERM, delayterm,
3025 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
3026 0, 0, sysctl_delayterm, "I", "");
3027
3028
3029 STATIC int
sysctl_rage_vnode(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3030 sysctl_rage_vnode
3031 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3032 {
3033 struct proc *p = req->p;
3034 struct uthread *ut;
3035 int new_value, old_value, changed;
3036 int error;
3037
3038 ut = current_uthread();
3039
3040 if (ut->uu_flag & UT_RAGE_VNODES) {
3041 old_value = KERN_RAGE_THREAD;
3042 } else if (p->p_lflag & P_LRAGE_VNODES) {
3043 old_value = KERN_RAGE_PROC;
3044 } else {
3045 old_value = 0;
3046 }
3047
3048 error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
3049
3050 if ((error == 0) && (changed != 0)) {
3051 switch (new_value) {
3052 case KERN_RAGE_PROC:
3053 proc_lock(p);
3054 p->p_lflag |= P_LRAGE_VNODES;
3055 proc_unlock(p);
3056 break;
3057 case KERN_UNRAGE_PROC:
3058 proc_lock(p);
3059 p->p_lflag &= ~P_LRAGE_VNODES;
3060 proc_unlock(p);
3061 break;
3062
3063 case KERN_RAGE_THREAD:
3064 ut->uu_flag |= UT_RAGE_VNODES;
3065 break;
3066 case KERN_UNRAGE_THREAD:
3067 ut = current_uthread();
3068 ut->uu_flag &= ~UT_RAGE_VNODES;
3069 break;
3070 }
3071 }
3072 return error;
3073 }
3074
3075 SYSCTL_PROC(_kern, KERN_RAGEVNODE, rage_vnode,
3076 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
3077 0, 0, sysctl_rage_vnode, "I", "");
3078
3079 /* XXX until filecoordinationd fixes a bit of inverted logic. */
3080 STATIC int
sysctl_vfsnspace(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3081 sysctl_vfsnspace
3082 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3083 {
3084 int old_value = 0, new_value, changed;
3085
3086 return sysctl_io_number(req, old_value, sizeof(int), &new_value,
3087 &changed);
3088 }
3089
3090 SYSCTL_PROC(_kern, OID_AUTO, vfsnspace,
3091 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
3092 0, 0, sysctl_vfsnspace, "I", "");
3093
3094 /* XXX move this interface into libproc and remove this sysctl */
3095 STATIC int
sysctl_setthread_cpupercent(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3096 sysctl_setthread_cpupercent
3097 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3098 {
3099 int new_value, old_value;
3100 int error = 0;
3101 kern_return_t kret = KERN_SUCCESS;
3102 uint8_t percent = 0;
3103 int ms_refill = 0;
3104
3105 if (!req->newptr) {
3106 return 0;
3107 }
3108
3109 old_value = 0;
3110
3111 if ((error = sysctl_io_number(req, old_value, sizeof(old_value), &new_value, NULL)) != 0) {
3112 return error;
3113 }
3114
3115 percent = new_value & 0xff; /* low 8 bytes for perent */
3116 ms_refill = (new_value >> 8) & 0xffffff; /* upper 24bytes represent ms refill value */
3117 if (percent > 100) {
3118 return EINVAL;
3119 }
3120
3121 /*
3122 * If the caller is specifying a percentage of 0, this will unset the CPU limit, if present.
3123 */
3124 if ((kret = thread_set_cpulimit(THREAD_CPULIMIT_BLOCK, percent, ms_refill * (int)NSEC_PER_MSEC)) != 0) {
3125 return EIO;
3126 }
3127
3128 return 0;
3129 }
3130
3131 SYSCTL_PROC(_kern, OID_AUTO, setthread_cpupercent,
3132 CTLTYPE_INT | CTLFLAG_WR | CTLFLAG_ANYBODY,
3133 0, 0, sysctl_setthread_cpupercent, "I", "set thread cpu percentage limit");
3134
3135
3136 STATIC int
sysctl_kern_check_openevt(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3137 sysctl_kern_check_openevt
3138 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3139 {
3140 struct proc *p = req->p;
3141 int new_value, old_value, changed;
3142 int error;
3143
3144 if (p->p_flag & P_CHECKOPENEVT) {
3145 old_value = KERN_OPENEVT_PROC;
3146 } else {
3147 old_value = 0;
3148 }
3149
3150 error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
3151
3152 if ((error == 0) && (changed != 0)) {
3153 switch (new_value) {
3154 case KERN_OPENEVT_PROC:
3155 OSBitOrAtomic(P_CHECKOPENEVT, &p->p_flag);
3156 break;
3157
3158 case KERN_UNOPENEVT_PROC:
3159 OSBitAndAtomic(~((uint32_t)P_CHECKOPENEVT), &p->p_flag);
3160 break;
3161
3162 default:
3163 error = EINVAL;
3164 }
3165 }
3166 return error;
3167 }
3168
3169 SYSCTL_PROC(_kern, KERN_CHECKOPENEVT, check_openevt, CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
3170 0, 0, sysctl_kern_check_openevt, "I", "set the per-process check-open-evt flag");
3171
3172
3173 #if DEVELOPMENT || DEBUG
3174 STATIC int
sysctl_nx(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3175 sysctl_nx
3176 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3177 {
3178 #ifdef SECURE_KERNEL
3179 (void)req;
3180 return ENOTSUP;
3181 #else
3182 int new_value, changed;
3183 int error;
3184
3185 error = sysctl_io_number(req, nx_enabled, sizeof(nx_enabled), &new_value, &changed);
3186 if (error) {
3187 return error;
3188 }
3189
3190 if (changed) {
3191 #if defined(__x86_64__)
3192 /*
3193 * Only allow setting if NX is supported on the chip
3194 */
3195 if (!(cpuid_extfeatures() & CPUID_EXTFEATURE_XD)) {
3196 return ENOTSUP;
3197 }
3198 #endif
3199 nx_enabled = new_value;
3200 }
3201 return error;
3202 #endif /* SECURE_KERNEL */
3203 }
3204 #endif
3205
3206 #if DEVELOPMENT || DEBUG
3207 SYSCTL_PROC(_kern, KERN_NX_PROTECTION, nx,
3208 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
3209 0, 0, sysctl_nx, "I", "");
3210 #endif
3211
3212 STATIC int
sysctl_loadavg(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3213 sysctl_loadavg
3214 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3215 {
3216 if (proc_is64bit(req->p)) {
3217 struct user64_loadavg loadinfo64 = {};
3218 fill_loadavg64(&averunnable, &loadinfo64);
3219 return sysctl_io_opaque(req, &loadinfo64, sizeof(loadinfo64), NULL);
3220 } else {
3221 struct user32_loadavg loadinfo32 = {};
3222 fill_loadavg32(&averunnable, &loadinfo32);
3223 return sysctl_io_opaque(req, &loadinfo32, sizeof(loadinfo32), NULL);
3224 }
3225 }
3226
3227 SYSCTL_PROC(_vm, VM_LOADAVG, loadavg,
3228 CTLTYPE_STRUCT | CTLFLAG_RD | CTLFLAG_LOCKED,
3229 0, 0, sysctl_loadavg, "S,loadavg", "");
3230
3231 /*
3232 * Note: Thread safe; vm_map_lock protects in vm_toggle_entry_reuse()
3233 */
3234 STATIC int
sysctl_vm_toggle_address_reuse(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3235 sysctl_vm_toggle_address_reuse(__unused struct sysctl_oid *oidp, __unused void *arg1,
3236 __unused int arg2, struct sysctl_req *req)
3237 {
3238 int old_value = 0, new_value = 0, error = 0;
3239
3240 if (vm_toggle_entry_reuse( VM_TOGGLE_GETVALUE, &old_value )) {
3241 return error;
3242 }
3243 error = sysctl_io_number(req, old_value, sizeof(int), &new_value, NULL);
3244 if (!error) {
3245 return vm_toggle_entry_reuse(new_value, NULL);
3246 }
3247 return error;
3248 }
3249
3250 SYSCTL_PROC(_debug, OID_AUTO, toggle_address_reuse, CTLFLAG_ANYBODY | CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED, 0, 0, sysctl_vm_toggle_address_reuse, "I", "");
3251
3252 #ifdef CONFIG_XNUPOST
3253
3254 extern uint32_t xnupost_get_estimated_testdata_size(void);
3255 extern int xnupost_reset_all_tests(void);
3256
3257 STATIC int
3258 sysctl_handle_xnupost_get_tests SYSCTL_HANDLER_ARGS
3259 {
3260 /* fixup unused arguments warnings */
3261 __unused int _oa2 = arg2;
3262 __unused void * _oa1 = arg1;
3263 __unused struct sysctl_oid * _oidp = oidp;
3264
3265 int error = 0;
3266 user_addr_t oldp = 0;
3267 user_addr_t newp = 0;
3268 uint32_t usedbytes = 0;
3269
3270 oldp = req->oldptr;
3271 newp = req->newptr;
3272
3273 if (newp) {
3274 return ENOTSUP;
3275 }
3276
3277 if ((void *)oldp == NULL) {
3278 /* return estimated size for second call where info can be placed */
3279 req->oldidx = xnupost_get_estimated_testdata_size();
3280 } else {
3281 error = xnupost_export_testdata((void *)oldp, req->oldlen, &usedbytes);
3282 req->oldidx = usedbytes;
3283 }
3284
3285 return error;
3286 }
3287
3288 SYSCTL_PROC(_debug,
3289 OID_AUTO,
3290 xnupost_get_tests,
3291 CTLFLAG_MASKED | CTLFLAG_ANYBODY | CTLTYPE_OPAQUE | CTLFLAG_RD | CTLFLAG_LOCKED,
3292 0,
3293 0,
3294 sysctl_handle_xnupost_get_tests,
3295 "-",
3296 "read xnupost test data in kernel");
3297
3298 STATIC int
3299 sysctl_debug_xnupost_ctl SYSCTL_HANDLER_ARGS
3300 {
3301 /* fixup unused arguments warnings */
3302 __unused int _oa2 = arg2;
3303 __unused void * _oa1 = arg1;
3304 __unused struct sysctl_oid * _oidp = oidp;
3305
3306 #define ARRCOUNT 4
3307 /*
3308 * INPUT: ACTION, PARAM1, PARAM2, PARAM3
3309 * OUTPUT: RESULTCODE, ADDITIONAL DATA
3310 */
3311 int32_t outval[ARRCOUNT] = {0};
3312 int32_t input[ARRCOUNT] = {0};
3313 int32_t out_size = sizeof(outval);
3314 int32_t in_size = sizeof(input);
3315 int error = 0;
3316
3317 /* if this is NULL call to find out size, send out size info */
3318 if (!req->newptr) {
3319 goto out;
3320 }
3321
3322 /* pull in provided value from userspace */
3323 error = SYSCTL_IN(req, &input[0], in_size);
3324 if (error) {
3325 return error;
3326 }
3327
3328 if (input[0] == XTCTL_RESET_TESTDATA) {
3329 outval[0] = xnupost_reset_all_tests();
3330 goto out;
3331 }
3332
3333 out:
3334 error = SYSCTL_OUT(req, &outval[0], out_size);
3335 return error;
3336 }
3337
3338 SYSCTL_PROC(_debug,
3339 OID_AUTO,
3340 xnupost_testctl,
3341 CTLFLAG_MASKED | CTLFLAG_ANYBODY | CTLTYPE_OPAQUE | CTLFLAG_RW | CTLFLAG_LOCKED,
3342 0,
3343 0,
3344 sysctl_debug_xnupost_ctl,
3345 "I",
3346 "xnupost control for kernel testing");
3347
3348 extern void test_oslog_handleOSLogCtl(int32_t * in, int32_t * out, int32_t arraycount);
3349
3350 STATIC int
sysctl_debug_test_oslog_ctl(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3351 sysctl_debug_test_oslog_ctl(__unused struct sysctl_oid * oidp, __unused void * arg1, __unused int arg2, struct sysctl_req * req)
3352 {
3353 #define ARRCOUNT 4
3354 int32_t outval[ARRCOUNT] = {0};
3355 int32_t input[ARRCOUNT] = {0};
3356 int32_t size_outval = sizeof(outval);
3357 int32_t size_inval = sizeof(input);
3358 int32_t error;
3359
3360 /* if this is NULL call to find out size, send out size info */
3361 if (!req->newptr) {
3362 error = SYSCTL_OUT(req, &outval[0], size_outval);
3363 return error;
3364 }
3365
3366 /* pull in provided value from userspace */
3367 error = SYSCTL_IN(req, &input[0], size_inval);
3368 if (error) {
3369 return error;
3370 }
3371
3372 test_oslog_handleOSLogCtl(input, outval, ARRCOUNT);
3373
3374 error = SYSCTL_OUT(req, &outval[0], size_outval);
3375
3376 return error;
3377 }
3378
3379 SYSCTL_PROC(_debug,
3380 OID_AUTO,
3381 test_OSLogCtl,
3382 CTLFLAG_MASKED | CTLFLAG_ANYBODY | CTLTYPE_OPAQUE | CTLFLAG_RW | CTLFLAG_LOCKED,
3383 0,
3384 0,
3385 sysctl_debug_test_oslog_ctl,
3386 "I",
3387 "testing oslog in kernel");
3388
3389 #include <mach/task.h>
3390 #include <mach/semaphore.h>
3391
3392 static LCK_GRP_DECLARE(sysctl_debug_test_stackshot_owner_grp, "test-stackshot-owner-grp");
3393 static LCK_MTX_DECLARE(sysctl_debug_test_stackshot_owner_init_mtx,
3394 &sysctl_debug_test_stackshot_owner_grp);
3395
3396 /* This is a sysctl for testing collection of owner info on a lock in kernel space. A multi-threaded
3397 * test from userland sets this sysctl in such a way that a thread blocks in kernel mode, and a
3398 * stackshot is taken to see if the owner of the lock can be identified.
3399 *
3400 * We can't return to userland with a kernel lock held, so be sure to unlock before we leave.
3401 * the semaphores allow us to artificially create cases where the lock is being held and the
3402 * thread is hanging / taking a long time to do something. */
3403
3404 volatile char sysctl_debug_test_stackshot_mtx_inited = 0;
3405 semaphore_t sysctl_debug_test_stackshot_mutex_sem;
3406 lck_mtx_t sysctl_debug_test_stackshot_owner_lck;
3407
3408 #define SYSCTL_DEBUG_MTX_ACQUIRE_WAIT 1
3409 #define SYSCTL_DEBUG_MTX_ACQUIRE_NOWAIT 2
3410 #define SYSCTL_DEBUG_MTX_SIGNAL 3
3411 #define SYSCTL_DEBUG_MTX_TEARDOWN 4
3412
3413 STATIC int
sysctl_debug_test_stackshot_mutex_owner(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3414 sysctl_debug_test_stackshot_mutex_owner(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3415 {
3416 long long option = -1;
3417 /* if the user tries to read the sysctl, we tell them what the address of the lock is (to test against stackshot's output) */
3418 long long mtx_unslid_addr = (long long)VM_KERNEL_UNSLIDE_OR_PERM(&sysctl_debug_test_stackshot_owner_lck);
3419 int error = sysctl_io_number(req, mtx_unslid_addr, sizeof(long long), (void*)&option, NULL);
3420
3421 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_init_mtx);
3422 if (!sysctl_debug_test_stackshot_mtx_inited) {
3423 lck_mtx_init(&sysctl_debug_test_stackshot_owner_lck,
3424 &sysctl_debug_test_stackshot_owner_grp,
3425 LCK_ATTR_NULL);
3426 semaphore_create(kernel_task,
3427 &sysctl_debug_test_stackshot_mutex_sem,
3428 SYNC_POLICY_FIFO, 0);
3429 sysctl_debug_test_stackshot_mtx_inited = 1;
3430 }
3431 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_init_mtx);
3432
3433 if (!error) {
3434 switch (option) {
3435 case SYSCTL_DEBUG_MTX_ACQUIRE_NOWAIT:
3436 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_lck);
3437 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_lck);
3438 break;
3439 case SYSCTL_DEBUG_MTX_ACQUIRE_WAIT:
3440 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_lck);
3441 semaphore_wait(sysctl_debug_test_stackshot_mutex_sem);
3442 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_lck);
3443 break;
3444 case SYSCTL_DEBUG_MTX_SIGNAL:
3445 semaphore_signal(sysctl_debug_test_stackshot_mutex_sem);
3446 break;
3447 case SYSCTL_DEBUG_MTX_TEARDOWN:
3448 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_init_mtx);
3449
3450 lck_mtx_destroy(&sysctl_debug_test_stackshot_owner_lck,
3451 &sysctl_debug_test_stackshot_owner_grp);
3452 semaphore_destroy(kernel_task,
3453 sysctl_debug_test_stackshot_mutex_sem);
3454 sysctl_debug_test_stackshot_mtx_inited = 0;
3455
3456 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_init_mtx);
3457 break;
3458 case -1: /* user just wanted to read the value, so do nothing */
3459 break;
3460 default:
3461 error = EINVAL;
3462 break;
3463 }
3464 }
3465 return error;
3466 }
3467
3468 /* we can't return to userland with a kernel rwlock held, so be sure to unlock before we leave.
3469 * the semaphores allow us to artificially create cases where the lock is being held and the
3470 * thread is hanging / taking a long time to do something. */
3471
3472 SYSCTL_PROC(_debug,
3473 OID_AUTO,
3474 test_MutexOwnerCtl,
3475 CTLFLAG_MASKED | CTLFLAG_ANYBODY | CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
3476 0,
3477 0,
3478 sysctl_debug_test_stackshot_mutex_owner,
3479 "-",
3480 "Testing mutex owner in kernel");
3481
3482 volatile char sysctl_debug_test_stackshot_rwlck_inited = 0;
3483 lck_rw_t sysctl_debug_test_stackshot_owner_rwlck;
3484 semaphore_t sysctl_debug_test_stackshot_rwlck_sem;
3485
3486 #define SYSCTL_DEBUG_KRWLCK_RACQUIRE_NOWAIT 1
3487 #define SYSCTL_DEBUG_KRWLCK_RACQUIRE_WAIT 2
3488 #define SYSCTL_DEBUG_KRWLCK_WACQUIRE_NOWAIT 3
3489 #define SYSCTL_DEBUG_KRWLCK_WACQUIRE_WAIT 4
3490 #define SYSCTL_DEBUG_KRWLCK_SIGNAL 5
3491 #define SYSCTL_DEBUG_KRWLCK_TEARDOWN 6
3492
3493 STATIC int
sysctl_debug_test_stackshot_rwlck_owner(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3494 sysctl_debug_test_stackshot_rwlck_owner(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3495 {
3496 long long option = -1;
3497 /* if the user tries to read the sysctl, we tell them what the address of the lock is
3498 * (to test against stackshot's output) */
3499 long long rwlck_unslid_addr = (long long)VM_KERNEL_UNSLIDE_OR_PERM(&sysctl_debug_test_stackshot_owner_rwlck);
3500 int error = sysctl_io_number(req, rwlck_unslid_addr, sizeof(long long), (void*)&option, NULL);
3501
3502 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_init_mtx);
3503 if (!sysctl_debug_test_stackshot_rwlck_inited) {
3504 lck_rw_init(&sysctl_debug_test_stackshot_owner_rwlck,
3505 &sysctl_debug_test_stackshot_owner_grp,
3506 LCK_ATTR_NULL);
3507 semaphore_create(kernel_task,
3508 &sysctl_debug_test_stackshot_rwlck_sem,
3509 SYNC_POLICY_FIFO,
3510 0);
3511 sysctl_debug_test_stackshot_rwlck_inited = 1;
3512 }
3513 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_init_mtx);
3514
3515 if (!error) {
3516 switch (option) {
3517 case SYSCTL_DEBUG_KRWLCK_RACQUIRE_NOWAIT:
3518 lck_rw_lock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_SHARED);
3519 lck_rw_unlock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_SHARED);
3520 break;
3521 case SYSCTL_DEBUG_KRWLCK_RACQUIRE_WAIT:
3522 lck_rw_lock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_SHARED);
3523 semaphore_wait(sysctl_debug_test_stackshot_rwlck_sem);
3524 lck_rw_unlock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_SHARED);
3525 break;
3526 case SYSCTL_DEBUG_KRWLCK_WACQUIRE_NOWAIT:
3527 lck_rw_lock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_EXCLUSIVE);
3528 lck_rw_unlock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_EXCLUSIVE);
3529 break;
3530 case SYSCTL_DEBUG_KRWLCK_WACQUIRE_WAIT:
3531 lck_rw_lock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_EXCLUSIVE);
3532 semaphore_wait(sysctl_debug_test_stackshot_rwlck_sem);
3533 lck_rw_unlock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_EXCLUSIVE);
3534 break;
3535 case SYSCTL_DEBUG_KRWLCK_SIGNAL:
3536 semaphore_signal(sysctl_debug_test_stackshot_rwlck_sem);
3537 break;
3538 case SYSCTL_DEBUG_KRWLCK_TEARDOWN:
3539 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_init_mtx);
3540
3541 lck_rw_destroy(&sysctl_debug_test_stackshot_owner_rwlck,
3542 &sysctl_debug_test_stackshot_owner_grp);
3543 semaphore_destroy(kernel_task,
3544 sysctl_debug_test_stackshot_rwlck_sem);
3545 sysctl_debug_test_stackshot_rwlck_inited = 0;
3546
3547 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_init_mtx);
3548 break;
3549 case -1: /* user just wanted to read the value, so do nothing */
3550 break;
3551 default:
3552 error = EINVAL;
3553 break;
3554 }
3555 }
3556 return error;
3557 }
3558
3559
3560 SYSCTL_PROC(_debug,
3561 OID_AUTO,
3562 test_RWLockOwnerCtl,
3563 CTLFLAG_MASKED | CTLFLAG_ANYBODY | CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
3564 0,
3565 0,
3566 sysctl_debug_test_stackshot_rwlck_owner,
3567 "-",
3568 "Testing rwlock owner in kernel");
3569 #endif /* !CONFIG_XNUPOST */
3570
3571 STATIC int
sysctl_swapusage(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3572 sysctl_swapusage
3573 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3574 {
3575 int error;
3576 uint64_t swap_total;
3577 uint64_t swap_avail;
3578 vm_size_t swap_pagesize;
3579 boolean_t swap_encrypted;
3580 struct xsw_usage xsu = {};
3581
3582 error = macx_swapinfo(&swap_total,
3583 &swap_avail,
3584 &swap_pagesize,
3585 &swap_encrypted);
3586 if (error) {
3587 return error;
3588 }
3589
3590 xsu.xsu_total = swap_total;
3591 xsu.xsu_avail = swap_avail;
3592 xsu.xsu_used = swap_total - swap_avail;
3593 xsu.xsu_pagesize = (u_int32_t)MIN(swap_pagesize, UINT32_MAX);
3594 xsu.xsu_encrypted = swap_encrypted;
3595 return sysctl_io_opaque(req, &xsu, sizeof(xsu), NULL);
3596 }
3597
3598
3599
3600 SYSCTL_PROC(_vm, VM_SWAPUSAGE, swapusage,
3601 CTLTYPE_STRUCT | CTLFLAG_RD | CTLFLAG_LOCKED,
3602 0, 0, sysctl_swapusage, "S,xsw_usage", "");
3603
3604 #if DEVELOPMENT || DEBUG
3605 extern int vm_num_swap_files_config;
3606 extern int vm_num_swap_files;
3607 extern lck_mtx_t vm_swap_data_lock;
3608 #define VM_MAX_SWAP_FILE_NUM 100
3609
3610 static int
3611 sysctl_vm_config_num_swap_files SYSCTL_HANDLER_ARGS
3612 {
3613 #pragma unused(arg1, arg2)
3614 int error = 0, val = vm_num_swap_files_config;
3615
3616 error = sysctl_handle_int(oidp, &val, 0, req);
3617 if (error || !req->newptr) {
3618 goto out;
3619 }
3620
3621 if (!VM_CONFIG_SWAP_IS_ACTIVE && !VM_CONFIG_FREEZER_SWAP_IS_ACTIVE) {
3622 printf("Swap is disabled\n");
3623 error = EINVAL;
3624 goto out;
3625 }
3626
3627 lck_mtx_lock(&vm_swap_data_lock);
3628
3629 if (val < vm_num_swap_files) {
3630 printf("Cannot configure fewer swap files than already exist.\n");
3631 error = EINVAL;
3632 lck_mtx_unlock(&vm_swap_data_lock);
3633 goto out;
3634 }
3635
3636 if (val > VM_MAX_SWAP_FILE_NUM) {
3637 printf("Capping number of swap files to upper bound.\n");
3638 val = VM_MAX_SWAP_FILE_NUM;
3639 }
3640
3641 vm_num_swap_files_config = val;
3642 lck_mtx_unlock(&vm_swap_data_lock);
3643 out:
3644
3645 return 0;
3646 }
3647
3648 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", "");
3649 #endif /* DEVELOPMENT || DEBUG */
3650
3651 /* this kernel does NOT implement shared_region_make_private_np() */
3652 SYSCTL_INT(_kern, KERN_SHREG_PRIVATIZABLE, shreg_private,
3653 CTLFLAG_RD | CTLFLAG_LOCKED,
3654 (int *)NULL, 0, "");
3655
3656 STATIC int
fetch_process_cputype(proc_t cur_proc,int * name,u_int namelen,cpu_type_t * cputype)3657 fetch_process_cputype(
3658 proc_t cur_proc,
3659 int *name,
3660 u_int namelen,
3661 cpu_type_t *cputype)
3662 {
3663 proc_t p = PROC_NULL;
3664 int refheld = 0;
3665 cpu_type_t ret = 0;
3666 int error = 0;
3667
3668 if (namelen == 0) {
3669 p = cur_proc;
3670 } else if (namelen == 1) {
3671 p = proc_find(name[0]);
3672 if (p == NULL) {
3673 return EINVAL;
3674 }
3675 refheld = 1;
3676 } else {
3677 error = EINVAL;
3678 goto out;
3679 }
3680
3681 ret = cpu_type() & ~CPU_ARCH_MASK;
3682 if (IS_64BIT_PROCESS(p)) {
3683 ret |= CPU_ARCH_ABI64;
3684 }
3685
3686 *cputype = ret;
3687
3688 if (refheld != 0) {
3689 proc_rele(p);
3690 }
3691 out:
3692 return error;
3693 }
3694
3695
3696 STATIC int
sysctl_sysctl_native(__unused struct sysctl_oid * oidp,void * arg1,int arg2,struct sysctl_req * req)3697 sysctl_sysctl_native(__unused struct sysctl_oid *oidp, void *arg1, int arg2,
3698 struct sysctl_req *req)
3699 {
3700 int error;
3701 cpu_type_t proc_cputype = 0;
3702 if ((error = fetch_process_cputype(req->p, (int *)arg1, arg2, &proc_cputype)) != 0) {
3703 return error;
3704 }
3705 int res = 1;
3706 if ((proc_cputype & ~CPU_ARCH_MASK) != (cpu_type() & ~CPU_ARCH_MASK)) {
3707 res = 0;
3708 }
3709 return SYSCTL_OUT(req, &res, sizeof(res));
3710 }
3711 SYSCTL_PROC(_sysctl, OID_AUTO, proc_native, CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED, 0, 0, sysctl_sysctl_native, "I", "proc_native");
3712
3713 STATIC int
sysctl_sysctl_cputype(__unused struct sysctl_oid * oidp,void * arg1,int arg2,struct sysctl_req * req)3714 sysctl_sysctl_cputype(__unused struct sysctl_oid *oidp, void *arg1, int arg2,
3715 struct sysctl_req *req)
3716 {
3717 int error;
3718 cpu_type_t proc_cputype = 0;
3719 if ((error = fetch_process_cputype(req->p, (int *)arg1, arg2, &proc_cputype)) != 0) {
3720 return error;
3721 }
3722 return SYSCTL_OUT(req, &proc_cputype, sizeof(proc_cputype));
3723 }
3724 SYSCTL_PROC(_sysctl, OID_AUTO, proc_cputype, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED, 0, 0, sysctl_sysctl_cputype, "I", "proc_cputype");
3725
3726 STATIC int
sysctl_safeboot(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3727 sysctl_safeboot
3728 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3729 {
3730 return sysctl_io_number(req, boothowto & RB_SAFEBOOT ? 1 : 0, sizeof(int), NULL, NULL);
3731 }
3732
3733 SYSCTL_PROC(_kern, KERN_SAFEBOOT, safeboot,
3734 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
3735 0, 0, sysctl_safeboot, "I", "");
3736
3737 STATIC int
sysctl_singleuser(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3738 sysctl_singleuser
3739 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3740 {
3741 return sysctl_io_number(req, boothowto & RB_SINGLE ? 1 : 0, sizeof(int), NULL, NULL);
3742 }
3743
3744 SYSCTL_PROC(_kern, OID_AUTO, singleuser,
3745 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
3746 0, 0, sysctl_singleuser, "I", "");
3747
3748 STATIC int
sysctl_minimalboot(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3749 sysctl_minimalboot
3750 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3751 {
3752 return sysctl_io_number(req, minimalboot, sizeof(int), NULL, NULL);
3753 }
3754
3755 SYSCTL_PROC(_kern, OID_AUTO, minimalboot,
3756 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
3757 0, 0, sysctl_minimalboot, "I", "");
3758
3759 /*
3760 * Controls for debugging affinity sets - see osfmk/kern/affinity.c
3761 */
3762 extern boolean_t affinity_sets_enabled;
3763 extern int affinity_sets_mapping;
3764
3765 SYSCTL_INT(_kern, OID_AUTO, affinity_sets_enabled,
3766 CTLFLAG_RW | CTLFLAG_LOCKED, (int *) &affinity_sets_enabled, 0, "hinting enabled");
3767 SYSCTL_INT(_kern, OID_AUTO, affinity_sets_mapping,
3768 CTLFLAG_RW | CTLFLAG_LOCKED, &affinity_sets_mapping, 0, "mapping policy");
3769
3770 /*
3771 * Boolean indicating if KASLR is active.
3772 */
3773 STATIC int
sysctl_slide(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3774 sysctl_slide
3775 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3776 {
3777 uint32_t slide;
3778
3779 slide = vm_kernel_slide ? 1 : 0;
3780
3781 return sysctl_io_number( req, slide, sizeof(int), NULL, NULL);
3782 }
3783
3784 SYSCTL_PROC(_kern, OID_AUTO, slide,
3785 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
3786 0, 0, sysctl_slide, "I", "");
3787
3788 #if DEBUG || DEVELOPMENT
3789 #if defined(__arm64__)
3790 extern vm_offset_t segTEXTEXECB;
3791
3792 static int
3793 sysctl_kernel_text_exec_base_slide SYSCTL_HANDLER_ARGS
3794 {
3795 #pragma unused(arg1, arg2, oidp)
3796 unsigned long slide = 0;
3797 kc_format_t kc_format;
3798
3799 PE_get_primary_kc_format(&kc_format);
3800
3801 if (kc_format == KCFormatFileset) {
3802 void *kch = PE_get_kc_header(KCKindPrimary);
3803 slide = (unsigned long)segTEXTEXECB - (unsigned long)kch + vm_kernel_slide;
3804 }
3805 return SYSCTL_OUT(req, &slide, sizeof(slide));
3806 }
3807
3808 SYSCTL_QUAD(_kern, OID_AUTO, kernel_slide, CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED, &vm_kernel_slide, "");
3809 SYSCTL_QUAD(_kern, OID_AUTO, kernel_text_exec_base, CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED, &segTEXTEXECB, "");
3810 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", "");
3811 #endif /* defined(__arm64__) */
3812
3813 /* User address of the PFZ */
3814 extern user32_addr_t commpage_text32_location;
3815 extern user64_addr_t commpage_text64_location;
3816
3817 STATIC int
3818 sysctl_pfz_start SYSCTL_HANDLER_ARGS
3819 {
3820 #pragma unused(oidp, arg1, arg2)
3821
3822 #ifdef __LP64__
3823 return sysctl_io_number(req, commpage_text64_location, sizeof(user64_addr_t), NULL, NULL);
3824 #else
3825 return sysctl_io_number(req, commpage_text32_location, sizeof(user32_addr_t), NULL, NULL);
3826 #endif
3827 }
3828
3829 SYSCTL_PROC(_kern, OID_AUTO, pfz,
3830 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_MASKED,
3831 0, 0, sysctl_pfz_start, "I", "");
3832 #endif
3833
3834
3835 /*
3836 * Limit on total memory users can wire.
3837 *
3838 * vm_global_user_wire_limit - system wide limit on wired memory from all processes combined.
3839 *
3840 * vm_per_task_user_wire_limit - per address space limit on wired memory. This puts a cap on the process's rlimit value.
3841 *
3842 * These values are initialized to reasonable defaults at boot time based on the available physical memory in
3843 * kmem_init().
3844 *
3845 * All values are in bytes.
3846 */
3847
3848 vm_map_size_t vm_global_user_wire_limit;
3849 vm_map_size_t vm_per_task_user_wire_limit;
3850 extern uint64_t max_mem_actual, max_mem;
3851
3852 uint64_t vm_add_wire_count_over_global_limit;
3853 uint64_t vm_add_wire_count_over_user_limit;
3854 /*
3855 * We used to have a global in the kernel called vm_global_no_user_wire_limit which was the inverse
3856 * of vm_global_user_wire_limit. But maintaining both of those is silly, and vm_global_user_wire_limit is the
3857 * real limit.
3858 * This function is for backwards compatibility with userspace
3859 * since we exposed the old global via a sysctl.
3860 */
3861 STATIC int
sysctl_global_no_user_wire_amount(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3862 sysctl_global_no_user_wire_amount(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3863 {
3864 vm_map_size_t old_value;
3865 vm_map_size_t new_value;
3866 int changed;
3867 int error;
3868 uint64_t config_memsize = max_mem;
3869 #if defined(XNU_TARGET_OS_OSX)
3870 config_memsize = max_mem_actual;
3871 #endif /* defined(XNU_TARGET_OS_OSX) */
3872
3873 old_value = (vm_map_size_t)(config_memsize - vm_global_user_wire_limit);
3874 error = sysctl_io_number(req, old_value, sizeof(vm_map_size_t), &new_value, &changed);
3875 if (changed) {
3876 if ((uint64_t)new_value > config_memsize) {
3877 error = EINVAL;
3878 } else {
3879 vm_global_user_wire_limit = (vm_map_size_t)(config_memsize - new_value);
3880 }
3881 }
3882 return error;
3883 }
3884 /*
3885 * There needs to be a more automatic/elegant way to do this
3886 */
3887 #if defined(__ARM__)
3888 SYSCTL_INT(_vm, OID_AUTO, global_user_wire_limit, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_global_user_wire_limit, 0, "");
3889 SYSCTL_INT(_vm, OID_AUTO, user_wire_limit, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_per_task_user_wire_limit, 0, "");
3890 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", "");
3891 #else
3892 SYSCTL_QUAD(_vm, OID_AUTO, global_user_wire_limit, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_global_user_wire_limit, "");
3893 SYSCTL_QUAD(_vm, OID_AUTO, user_wire_limit, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_per_task_user_wire_limit, "");
3894 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", "");
3895 #endif
3896
3897 /*
3898 * Relaxed atomic RW of a 64bit value via sysctl.
3899 */
3900 STATIC int
sysctl_r_64bit_atomic(uint64_t * ptr,struct sysctl_req * req)3901 sysctl_r_64bit_atomic(uint64_t *ptr, struct sysctl_req *req)
3902 {
3903 uint64_t old_value;
3904 uint64_t new_value;
3905 int error;
3906
3907 old_value = os_atomic_load_wide(ptr, relaxed);
3908 error = sysctl_io_number(req, old_value, sizeof(vm_map_size_t), &new_value, NULL);
3909 return error;
3910 }
3911 STATIC int
sysctl_add_wire_count_over_global_limit(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3912 sysctl_add_wire_count_over_global_limit(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3913 {
3914 return sysctl_r_64bit_atomic(&vm_add_wire_count_over_global_limit, req);
3915 }
3916 STATIC int
sysctl_add_wire_count_over_user_limit(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)3917 sysctl_add_wire_count_over_user_limit(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3918 {
3919 return sysctl_r_64bit_atomic(&vm_add_wire_count_over_user_limit, req);
3920 }
3921
3922 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", "");
3923 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", "");
3924
3925 #if DEVELOPMENT || DEBUG
3926 /* These sysctls are used to test the wired limit. */
3927 extern unsigned int vm_page_wire_count;
3928 extern uint32_t vm_lopage_free_count;
3929 SYSCTL_INT(_vm, OID_AUTO, page_wire_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_wire_count, 0, "");
3930 SYSCTL_INT(_vm, OID_AUTO, lopage_free_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_lopage_free_count, 0, "");
3931
3932 /*
3933 * Setting the per task variable exclude_physfootprint_ledger to 1 will allow the calling task to exclude memory entries that are
3934 * tagged by VM_LEDGER_TAG_DEFAULT and flagged by VM_LEDGER_FLAG_EXCLUDE_FOOTPRINT_DEBUG from its phys_footprint ledger.
3935 */
3936
3937 STATIC int
sysctl_rw_task_no_footprint_for_debug(struct sysctl_oid * oidp __unused,void * arg1 __unused,int arg2 __unused,struct sysctl_req * req)3938 sysctl_rw_task_no_footprint_for_debug(struct sysctl_oid *oidp __unused, void *arg1 __unused, int arg2 __unused, struct sysctl_req *req)
3939 {
3940 int error;
3941 int value;
3942 proc_t p = current_proc();
3943
3944 if (req->newptr) {
3945 // Write request
3946 error = SYSCTL_IN(req, &value, sizeof(value));
3947 if (!error) {
3948 if (value == 1) {
3949 task_set_no_footprint_for_debug(p->task, TRUE);
3950 } else if (value == 0) {
3951 task_set_no_footprint_for_debug(p->task, FALSE);
3952 } else {
3953 error = EINVAL;
3954 }
3955 }
3956 } else {
3957 // Read request
3958 value = task_get_no_footprint_for_debug(p->task);
3959 error = SYSCTL_OUT(req, &value, sizeof(value));
3960 }
3961 return error;
3962 }
3963
3964 SYSCTL_PROC(_vm, OID_AUTO, task_no_footprint_for_debug,
3965 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
3966 0, 0, &sysctl_rw_task_no_footprint_for_debug, "I", "Allow debug memory to be excluded from this task's memory footprint (debug only)");
3967
3968 #endif /* DEVELOPMENT || DEBUG */
3969
3970
3971 extern int vm_map_copy_overwrite_aligned_src_not_internal;
3972 extern int vm_map_copy_overwrite_aligned_src_not_symmetric;
3973 extern int vm_map_copy_overwrite_aligned_src_large;
3974 SYSCTL_INT(_vm, OID_AUTO, vm_copy_src_not_internal, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_map_copy_overwrite_aligned_src_not_internal, 0, "");
3975 SYSCTL_INT(_vm, OID_AUTO, vm_copy_src_not_symmetric, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_map_copy_overwrite_aligned_src_not_symmetric, 0, "");
3976 SYSCTL_INT(_vm, OID_AUTO, vm_copy_src_large, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_map_copy_overwrite_aligned_src_large, 0, "");
3977
3978
3979 extern uint32_t vm_page_external_count;
3980
3981 SYSCTL_INT(_vm, OID_AUTO, vm_page_external_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_external_count, 0, "");
3982
3983 SYSCTL_INT(_vm, OID_AUTO, vm_page_filecache_min, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_state.vm_page_filecache_min, 0, "");
3984 SYSCTL_INT(_vm, OID_AUTO, vm_page_xpmapped_min, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_state.vm_page_xpmapped_min, 0, "");
3985
3986 #if DEVELOPMENT || DEBUG
3987 SYSCTL_INT(_vm, OID_AUTO, vm_page_filecache_min_divisor, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_state.vm_page_filecache_min_divisor, 0, "");
3988 SYSCTL_INT(_vm, OID_AUTO, vm_page_xpmapped_min_divisor, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_state.vm_page_xpmapped_min_divisor, 0, "");
3989 #endif
3990
3991 extern int vm_compressor_mode;
3992 extern int vm_compressor_is_active;
3993 extern int vm_compressor_available;
3994 extern uint32_t c_seg_bufsize;
3995 extern uint64_t compressor_pool_size;
3996 extern uint32_t vm_ripe_target_age;
3997 extern uint32_t swapout_target_age;
3998 extern int64_t compressor_bytes_used;
3999 extern int64_t c_segment_input_bytes;
4000 extern int64_t c_segment_compressed_bytes;
4001 extern uint32_t compressor_eval_period_in_msecs;
4002 extern uint32_t compressor_sample_min_in_msecs;
4003 extern uint32_t compressor_sample_max_in_msecs;
4004 extern uint32_t compressor_thrashing_threshold_per_10msecs;
4005 extern uint32_t compressor_thrashing_min_per_10msecs;
4006 extern uint32_t vm_compressor_time_thread;
4007
4008 #if DEVELOPMENT || DEBUG
4009 extern uint32_t vm_compressor_minorcompact_threshold_divisor;
4010 extern uint32_t vm_compressor_majorcompact_threshold_divisor;
4011 extern uint32_t vm_compressor_unthrottle_threshold_divisor;
4012 extern uint32_t vm_compressor_catchup_threshold_divisor;
4013
4014 extern uint32_t vm_compressor_minorcompact_threshold_divisor_overridden;
4015 extern uint32_t vm_compressor_majorcompact_threshold_divisor_overridden;
4016 extern uint32_t vm_compressor_unthrottle_threshold_divisor_overridden;
4017 extern uint32_t vm_compressor_catchup_threshold_divisor_overridden;
4018
4019 extern vmct_stats_t vmct_stats;
4020
4021
4022 STATIC int
sysctl_minorcompact_threshold_divisor(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)4023 sysctl_minorcompact_threshold_divisor(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4024 {
4025 int new_value, changed;
4026 int error = sysctl_io_number(req, vm_compressor_minorcompact_threshold_divisor, sizeof(int), &new_value, &changed);
4027
4028 if (changed) {
4029 vm_compressor_minorcompact_threshold_divisor = new_value;
4030 vm_compressor_minorcompact_threshold_divisor_overridden = 1;
4031 }
4032 return error;
4033 }
4034
4035 SYSCTL_PROC(_vm, OID_AUTO, compressor_minorcompact_threshold_divisor,
4036 CTLTYPE_INT | CTLFLAG_LOCKED | CTLFLAG_RW,
4037 0, 0, sysctl_minorcompact_threshold_divisor, "I", "");
4038
4039
4040 STATIC int
sysctl_majorcompact_threshold_divisor(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)4041 sysctl_majorcompact_threshold_divisor(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4042 {
4043 int new_value, changed;
4044 int error = sysctl_io_number(req, vm_compressor_majorcompact_threshold_divisor, sizeof(int), &new_value, &changed);
4045
4046 if (changed) {
4047 vm_compressor_majorcompact_threshold_divisor = new_value;
4048 vm_compressor_majorcompact_threshold_divisor_overridden = 1;
4049 }
4050 return error;
4051 }
4052
4053 SYSCTL_PROC(_vm, OID_AUTO, compressor_majorcompact_threshold_divisor,
4054 CTLTYPE_INT | CTLFLAG_LOCKED | CTLFLAG_RW,
4055 0, 0, sysctl_majorcompact_threshold_divisor, "I", "");
4056
4057
4058 STATIC int
sysctl_unthrottle_threshold_divisor(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)4059 sysctl_unthrottle_threshold_divisor(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4060 {
4061 int new_value, changed;
4062 int error = sysctl_io_number(req, vm_compressor_unthrottle_threshold_divisor, sizeof(int), &new_value, &changed);
4063
4064 if (changed) {
4065 vm_compressor_unthrottle_threshold_divisor = new_value;
4066 vm_compressor_unthrottle_threshold_divisor_overridden = 1;
4067 }
4068 return error;
4069 }
4070
4071 SYSCTL_PROC(_vm, OID_AUTO, compressor_unthrottle_threshold_divisor,
4072 CTLTYPE_INT | CTLFLAG_LOCKED | CTLFLAG_RW,
4073 0, 0, sysctl_unthrottle_threshold_divisor, "I", "");
4074
4075
4076 STATIC int
sysctl_catchup_threshold_divisor(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)4077 sysctl_catchup_threshold_divisor(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4078 {
4079 int new_value, changed;
4080 int error = sysctl_io_number(req, vm_compressor_catchup_threshold_divisor, sizeof(int), &new_value, &changed);
4081
4082 if (changed) {
4083 vm_compressor_catchup_threshold_divisor = new_value;
4084 vm_compressor_catchup_threshold_divisor_overridden = 1;
4085 }
4086 return error;
4087 }
4088
4089 SYSCTL_PROC(_vm, OID_AUTO, compressor_catchup_threshold_divisor,
4090 CTLTYPE_INT | CTLFLAG_LOCKED | CTLFLAG_RW,
4091 0, 0, sysctl_catchup_threshold_divisor, "I", "");
4092 #endif
4093
4094
4095 SYSCTL_QUAD(_vm, OID_AUTO, compressor_input_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, &c_segment_input_bytes, "");
4096 SYSCTL_QUAD(_vm, OID_AUTO, compressor_compressed_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, &c_segment_compressed_bytes, "");
4097 SYSCTL_QUAD(_vm, OID_AUTO, compressor_bytes_used, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_bytes_used, "");
4098
4099 SYSCTL_INT(_vm, OID_AUTO, compressor_mode, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_compressor_mode, 0, "");
4100 SYSCTL_INT(_vm, OID_AUTO, compressor_is_active, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_compressor_is_active, 0, "");
4101 SYSCTL_INT(_vm, OID_AUTO, compressor_swapout_target_age, CTLFLAG_RD | CTLFLAG_LOCKED, &swapout_target_age, 0, "");
4102 SYSCTL_INT(_vm, OID_AUTO, compressor_available, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_compressor_available, 0, "");
4103 SYSCTL_INT(_vm, OID_AUTO, compressor_segment_buffer_size, CTLFLAG_RD | CTLFLAG_LOCKED, &c_seg_bufsize, 0, "");
4104 SYSCTL_QUAD(_vm, OID_AUTO, compressor_pool_size, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_pool_size, "");
4105
4106 extern int min_csegs_per_major_compaction;
4107 SYSCTL_INT(_vm, OID_AUTO, compressor_min_csegs_per_major_compaction, CTLFLAG_RW | CTLFLAG_LOCKED, &min_csegs_per_major_compaction, 0, "");
4108
4109 SYSCTL_INT(_vm, OID_AUTO, vm_ripe_target_age_in_secs, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_ripe_target_age, 0, "");
4110
4111 SYSCTL_INT(_vm, OID_AUTO, compressor_eval_period_in_msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &compressor_eval_period_in_msecs, 0, "");
4112 SYSCTL_INT(_vm, OID_AUTO, compressor_sample_min_in_msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &compressor_sample_min_in_msecs, 0, "");
4113 SYSCTL_INT(_vm, OID_AUTO, compressor_sample_max_in_msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &compressor_sample_max_in_msecs, 0, "");
4114 SYSCTL_INT(_vm, OID_AUTO, compressor_thrashing_threshold_per_10msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &compressor_thrashing_threshold_per_10msecs, 0, "");
4115 SYSCTL_INT(_vm, OID_AUTO, compressor_thrashing_min_per_10msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &compressor_thrashing_min_per_10msecs, 0, "");
4116
4117 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapouts_under_30s, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.unripe_under_30s, "");
4118 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapouts_under_60s, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.unripe_under_60s, "");
4119 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapouts_under_300s, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.unripe_under_300s, "");
4120 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_reclaim_swapins, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.reclaim_swapins, "");
4121 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_defrag_swapins, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.defrag_swapins, "");
4122 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_swapout_threshold_exceeded, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.compressor_swap_threshold_exceeded, "");
4123 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_swapout_fileq_throttled, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.external_q_throttled, "");
4124 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_swapout_free_count_low, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.free_count_below_reserve, "");
4125 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_swapout_thrashing_detected, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.thrashing_detected, "");
4126 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_swapout_fragmentation_detected, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.fragmentation_detected, "");
4127
4128 SYSCTL_STRING(_vm, OID_AUTO, swapfileprefix, CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED, swapfilename, sizeof(swapfilename) - SWAPFILENAME_INDEX_LEN, "");
4129
4130 SYSCTL_INT(_vm, OID_AUTO, compressor_timing_enabled, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_compressor_time_thread, 0, "");
4131
4132 #if DEVELOPMENT || DEBUG
4133 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_runtime0, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_runtimes[0], "");
4134 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_runtime1, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_runtimes[1], "");
4135
4136 SYSCTL_QUAD(_vm, OID_AUTO, compressor_threads_total, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_cthreads_total, "");
4137
4138 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_pages0, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_pages[0], "");
4139 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_pages1, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_pages[1], "");
4140
4141 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_iterations0, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_iterations[0], "");
4142 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_iterations1, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_iterations[1], "");
4143
4144 SYSCTL_INT(_vm, OID_AUTO, compressor_thread_minpages0, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_minpages[0], 0, "");
4145 SYSCTL_INT(_vm, OID_AUTO, compressor_thread_minpages1, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_minpages[1], 0, "");
4146
4147 SYSCTL_INT(_vm, OID_AUTO, compressor_thread_maxpages0, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_maxpages[0], 0, "");
4148 SYSCTL_INT(_vm, OID_AUTO, compressor_thread_maxpages1, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_maxpages[1], 0, "");
4149
4150 int vm_compressor_injected_error_count;
4151
4152 SYSCTL_INT(_vm, OID_AUTO, compressor_injected_error_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_compressor_injected_error_count, 0, "");
4153
4154 static int
sysctl_compressor_inject_error(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)4155 sysctl_compressor_inject_error(__unused struct sysctl_oid *oidp,
4156 __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4157 {
4158 int result;
4159 vm_address_t va = 0;
4160 int changed;
4161
4162 result = sysctl_io_number(req, va, sizeof(va), &va, &changed);
4163 if (result == 0 && changed) {
4164 result = vm_map_inject_error(current_map(), va);
4165 if (result == 0) {
4166 /*
4167 * Count the number of errors injected successfully to detect
4168 * situations where corruption was caused by improper use of this
4169 * sysctl.
4170 */
4171 os_atomic_inc(&vm_compressor_injected_error_count, relaxed);
4172 }
4173 }
4174 return result;
4175 }
4176
4177 SYSCTL_PROC(_vm, OID_AUTO, compressor_inject_error, CTLTYPE_QUAD | CTLFLAG_LOCKED | CTLFLAG_RW,
4178 0, 0, sysctl_compressor_inject_error, "Q", "flips a bit in a compressed page for the current task");
4179
4180 #endif
4181
4182 SYSCTL_QUAD(_vm, OID_AUTO, lz4_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_compressions, "");
4183 SYSCTL_QUAD(_vm, OID_AUTO, lz4_compression_failures, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_compression_failures, "");
4184 SYSCTL_QUAD(_vm, OID_AUTO, lz4_compressed_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_compressed_bytes, "");
4185 SYSCTL_QUAD(_vm, OID_AUTO, lz4_wk_compression_delta, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_wk_compression_delta, "");
4186 SYSCTL_QUAD(_vm, OID_AUTO, lz4_wk_compression_negative_delta, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_wk_compression_negative_delta, "");
4187
4188 SYSCTL_QUAD(_vm, OID_AUTO, lz4_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_decompressions, "");
4189 SYSCTL_QUAD(_vm, OID_AUTO, lz4_decompressed_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_decompressed_bytes, "");
4190
4191 SYSCTL_QUAD(_vm, OID_AUTO, uc_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.uc_decompressions, "");
4192
4193 SYSCTL_QUAD(_vm, OID_AUTO, wk_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_compressions, "");
4194
4195 SYSCTL_QUAD(_vm, OID_AUTO, wk_catime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_cabstime, "");
4196
4197 SYSCTL_QUAD(_vm, OID_AUTO, wkh_catime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wkh_cabstime, "");
4198 SYSCTL_QUAD(_vm, OID_AUTO, wkh_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wkh_compressions, "");
4199
4200 SYSCTL_QUAD(_vm, OID_AUTO, wks_catime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_cabstime, "");
4201 SYSCTL_QUAD(_vm, OID_AUTO, wks_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_compressions, "");
4202
4203 SYSCTL_QUAD(_vm, OID_AUTO, wk_compressions_exclusive, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_compressions_exclusive, "");
4204 SYSCTL_QUAD(_vm, OID_AUTO, wk_sv_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_sv_compressions, "");
4205 SYSCTL_QUAD(_vm, OID_AUTO, wk_mzv_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_mzv_compressions, "");
4206 SYSCTL_QUAD(_vm, OID_AUTO, wk_compression_failures, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_compression_failures, "");
4207 SYSCTL_QUAD(_vm, OID_AUTO, wk_compressed_bytes_exclusive, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_compressed_bytes_exclusive, "");
4208 SYSCTL_QUAD(_vm, OID_AUTO, wk_compressed_bytes_total, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_compressed_bytes_total, "");
4209
4210 SYSCTL_QUAD(_vm, OID_AUTO, wks_compressed_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_compressed_bytes, "");
4211 SYSCTL_QUAD(_vm, OID_AUTO, wks_compression_failures, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_compression_failures, "");
4212 SYSCTL_QUAD(_vm, OID_AUTO, wks_sv_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_sv_compressions, "");
4213
4214
4215 SYSCTL_QUAD(_vm, OID_AUTO, wk_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_decompressions, "");
4216
4217 SYSCTL_QUAD(_vm, OID_AUTO, wk_datime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_dabstime, "");
4218
4219 SYSCTL_QUAD(_vm, OID_AUTO, wkh_datime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wkh_dabstime, "");
4220 SYSCTL_QUAD(_vm, OID_AUTO, wkh_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wkh_decompressions, "");
4221
4222 SYSCTL_QUAD(_vm, OID_AUTO, wks_datime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_dabstime, "");
4223 SYSCTL_QUAD(_vm, OID_AUTO, wks_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_decompressions, "");
4224
4225 SYSCTL_QUAD(_vm, OID_AUTO, wk_decompressed_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_decompressed_bytes, "");
4226 SYSCTL_QUAD(_vm, OID_AUTO, wk_sv_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_sv_decompressions, "");
4227
4228 SYSCTL_INT(_vm, OID_AUTO, lz4_threshold, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_threshold, 0, "");
4229 SYSCTL_INT(_vm, OID_AUTO, wkdm_reeval_threshold, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.wkdm_reeval_threshold, 0, "");
4230 SYSCTL_INT(_vm, OID_AUTO, lz4_max_failure_skips, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_max_failure_skips, 0, "");
4231 SYSCTL_INT(_vm, OID_AUTO, lz4_max_failure_run_length, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_max_failure_run_length, 0, "");
4232 SYSCTL_INT(_vm, OID_AUTO, lz4_max_preselects, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_max_preselects, 0, "");
4233 SYSCTL_INT(_vm, OID_AUTO, lz4_run_preselection_threshold, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_run_preselection_threshold, 0, "");
4234 SYSCTL_INT(_vm, OID_AUTO, lz4_run_continue_bytes, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_run_continue_bytes, 0, "");
4235 SYSCTL_INT(_vm, OID_AUTO, lz4_profitable_bytes, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_profitable_bytes, 0, "");
4236 #if DEVELOPMENT || DEBUG
4237 extern int vm_compressor_current_codec;
4238 extern int vm_compressor_test_seg_wp;
4239 extern boolean_t vm_compressor_force_sw_wkdm;
4240 SYSCTL_INT(_vm, OID_AUTO, compressor_codec, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_compressor_current_codec, 0, "");
4241 SYSCTL_INT(_vm, OID_AUTO, compressor_test_wp, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_compressor_test_seg_wp, 0, "");
4242
4243 SYSCTL_INT(_vm, OID_AUTO, wksw_force, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_compressor_force_sw_wkdm, 0, "");
4244 extern int precompy, wkswhw;
4245
4246 SYSCTL_INT(_vm, OID_AUTO, precompy, CTLFLAG_RW | CTLFLAG_LOCKED, &precompy, 0, "");
4247 SYSCTL_INT(_vm, OID_AUTO, wkswhw, CTLFLAG_RW | CTLFLAG_LOCKED, &wkswhw, 0, "");
4248 extern unsigned int vm_ktrace_enabled;
4249 SYSCTL_INT(_vm, OID_AUTO, vm_ktrace, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_ktrace_enabled, 0, "");
4250 #endif
4251
4252 #if CONFIG_PHANTOM_CACHE
4253 extern uint32_t phantom_cache_thrashing_threshold;
4254 extern uint32_t phantom_cache_eval_period_in_msecs;
4255 extern uint32_t phantom_cache_thrashing_threshold_ssd;
4256
4257
4258 SYSCTL_INT(_vm, OID_AUTO, phantom_cache_eval_period_in_msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &phantom_cache_eval_period_in_msecs, 0, "");
4259 SYSCTL_INT(_vm, OID_AUTO, phantom_cache_thrashing_threshold, CTLFLAG_RW | CTLFLAG_LOCKED, &phantom_cache_thrashing_threshold, 0, "");
4260 SYSCTL_INT(_vm, OID_AUTO, phantom_cache_thrashing_threshold_ssd, CTLFLAG_RW | CTLFLAG_LOCKED, &phantom_cache_thrashing_threshold_ssd, 0, "");
4261 #endif
4262
4263 #if CONFIG_BACKGROUND_QUEUE
4264
4265 extern uint32_t vm_page_background_count;
4266 extern uint32_t vm_page_background_target;
4267 extern uint32_t vm_page_background_internal_count;
4268 extern uint32_t vm_page_background_external_count;
4269 extern uint32_t vm_page_background_mode;
4270 extern uint32_t vm_page_background_exclude_external;
4271 extern uint64_t vm_page_background_promoted_count;
4272 extern uint64_t vm_pageout_rejected_bq_internal;
4273 extern uint64_t vm_pageout_rejected_bq_external;
4274
4275 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_mode, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_page_background_mode, 0, "");
4276 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_exclude_external, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_page_background_exclude_external, 0, "");
4277 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_target, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_page_background_target, 0, "");
4278 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_background_count, 0, "");
4279 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_internal_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_background_internal_count, 0, "");
4280 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_external_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_background_external_count, 0, "");
4281
4282 SYSCTL_QUAD(_vm, OID_AUTO, vm_page_background_promoted_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_background_promoted_count, "");
4283 SYSCTL_QUAD(_vm, OID_AUTO, vm_pageout_considered_bq_internal, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_vminfo.vm_pageout_considered_bq_internal, "");
4284 SYSCTL_QUAD(_vm, OID_AUTO, vm_pageout_considered_bq_external, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_vminfo.vm_pageout_considered_bq_external, "");
4285 SYSCTL_QUAD(_vm, OID_AUTO, vm_pageout_rejected_bq_internal, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_rejected_bq_internal, "");
4286 SYSCTL_QUAD(_vm, OID_AUTO, vm_pageout_rejected_bq_external, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_rejected_bq_external, "");
4287
4288 #endif /* CONFIG_BACKGROUND_QUEUE */
4289
4290 extern void vm_update_darkwake_mode(boolean_t);
4291 extern boolean_t vm_darkwake_mode;
4292
4293 STATIC int
sysctl_toggle_darkwake_mode(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)4294 sysctl_toggle_darkwake_mode(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4295 {
4296 int new_value, changed;
4297 int error = sysctl_io_number(req, vm_darkwake_mode, sizeof(int), &new_value, &changed);
4298
4299 if (!error && changed) {
4300 if (new_value != 0 && new_value != 1) {
4301 printf("Error: Invalid value passed to darkwake sysctl. Acceptable: 0 or 1.\n");
4302 error = EINVAL;
4303 } else {
4304 vm_update_darkwake_mode((boolean_t) new_value);
4305 }
4306 }
4307
4308 return error;
4309 }
4310
4311 SYSCTL_PROC(_vm, OID_AUTO, darkwake_mode,
4312 CTLTYPE_INT | CTLFLAG_LOCKED | CTLFLAG_RW,
4313 0, 0, sysctl_toggle_darkwake_mode, "I", "");
4314
4315 #if (DEVELOPMENT || DEBUG)
4316
4317 SYSCTL_UINT(_vm, OID_AUTO, vm_page_creation_throttled_hard,
4318 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
4319 &vm_page_creation_throttled_hard, 0, "");
4320
4321 SYSCTL_UINT(_vm, OID_AUTO, vm_page_creation_throttled_soft,
4322 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
4323 &vm_page_creation_throttled_soft, 0, "");
4324
4325 extern uint32_t vm_pageout_memorystatus_fb_factor_nr;
4326 extern uint32_t vm_pageout_memorystatus_fb_factor_dr;
4327 SYSCTL_INT(_vm, OID_AUTO, vm_pageout_memorystatus_fb_factor_nr, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_memorystatus_fb_factor_nr, 0, "");
4328 SYSCTL_INT(_vm, OID_AUTO, vm_pageout_memorystatus_fb_factor_dr, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_memorystatus_fb_factor_dr, 0, "");
4329
4330 extern uint32_t vm_grab_anon_nops;
4331
4332 SYSCTL_INT(_vm, OID_AUTO, vm_grab_anon_overrides, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_debug.vm_grab_anon_overrides, 0, "");
4333 SYSCTL_INT(_vm, OID_AUTO, vm_grab_anon_nops, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_debug.vm_grab_anon_nops, 0, "");
4334
4335 extern int vm_page_delayed_work_ctx_needed;
4336 SYSCTL_INT(_vm, OID_AUTO, vm_page_needed_delayed_work_ctx, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_delayed_work_ctx_needed, 0, "");
4337
4338 /* log message counters for persistence mode */
4339 SCALABLE_COUNTER_DECLARE(oslog_p_total_msgcount);
4340 SCALABLE_COUNTER_DECLARE(oslog_p_metadata_saved_msgcount);
4341 SCALABLE_COUNTER_DECLARE(oslog_p_metadata_dropped_msgcount);
4342 SCALABLE_COUNTER_DECLARE(oslog_p_error_count);
4343 SCALABLE_COUNTER_DECLARE(oslog_p_saved_msgcount);
4344 SCALABLE_COUNTER_DECLARE(oslog_p_dropped_msgcount);
4345 SCALABLE_COUNTER_DECLARE(oslog_p_boot_dropped_msgcount);
4346 SCALABLE_COUNTER_DECLARE(oslog_p_coprocessor_total_msgcount);
4347 SCALABLE_COUNTER_DECLARE(oslog_p_coprocessor_dropped_msgcount);
4348 SCALABLE_COUNTER_DECLARE(oslog_p_unresolved_kc_msgcount);
4349 SCALABLE_COUNTER_DECLARE(oslog_p_fmt_invalid_msgcount);
4350 SCALABLE_COUNTER_DECLARE(oslog_p_fmt_max_args_msgcount);
4351 SCALABLE_COUNTER_DECLARE(oslog_p_truncated_msgcount);
4352
4353 SCALABLE_COUNTER_DECLARE(log_queue_cnt_received);
4354 SCALABLE_COUNTER_DECLARE(log_queue_cnt_rejected_fh);
4355 SCALABLE_COUNTER_DECLARE(log_queue_cnt_sent);
4356 SCALABLE_COUNTER_DECLARE(log_queue_cnt_dropped_nomem);
4357 SCALABLE_COUNTER_DECLARE(log_queue_cnt_queued);
4358 SCALABLE_COUNTER_DECLARE(log_queue_cnt_dropped_off);
4359 SCALABLE_COUNTER_DECLARE(log_queue_cnt_mem_active);
4360 SCALABLE_COUNTER_DECLARE(log_queue_cnt_mem_allocated);
4361 SCALABLE_COUNTER_DECLARE(log_queue_cnt_mem_released);
4362 SCALABLE_COUNTER_DECLARE(log_queue_cnt_mem_failed);
4363
4364 /* log message counters for streaming mode */
4365 SCALABLE_COUNTER_DECLARE(oslog_s_total_msgcount);
4366 SCALABLE_COUNTER_DECLARE(oslog_s_metadata_msgcount);
4367 SCALABLE_COUNTER_DECLARE(oslog_s_error_count);
4368 SCALABLE_COUNTER_DECLARE(oslog_s_streamed_msgcount);
4369 SCALABLE_COUNTER_DECLARE(oslog_s_dropped_msgcount);
4370
4371 /* log message counters for msgbuf logging */
4372 SCALABLE_COUNTER_DECLARE(oslog_msgbuf_msgcount);
4373 SCALABLE_COUNTER_DECLARE(oslog_msgbuf_dropped_msgcount);
4374 extern uint32_t oslog_msgbuf_dropped_charcount;
4375
4376 /* log message counters for vaddlog logging */
4377 extern uint32_t vaddlog_msgcount;
4378 extern uint32_t vaddlog_msgcount_dropped;
4379
4380 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_total_msgcount, oslog_p_total_msgcount, "");
4381 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_metadata_saved_msgcount, oslog_p_metadata_saved_msgcount, "");
4382 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_metadata_dropped_msgcount, oslog_p_metadata_dropped_msgcount, "");
4383 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_error_count, oslog_p_error_count, "");
4384 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_saved_msgcount, oslog_p_saved_msgcount, "");
4385 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_dropped_msgcount, oslog_p_dropped_msgcount, "");
4386 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_boot_dropped_msgcount, oslog_p_boot_dropped_msgcount, "");
4387 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_coprocessor_total_msgcount, oslog_p_coprocessor_total_msgcount, "");
4388 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_coprocessor_dropped_msgcount, oslog_p_coprocessor_dropped_msgcount, "");
4389 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_unresolved_kc_msgcount, oslog_p_unresolved_kc_msgcount, "");
4390
4391 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_fmt_invalid_msgcount, oslog_p_fmt_invalid_msgcount, "");
4392 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_fmt_max_args_msgcount, oslog_p_fmt_max_args_msgcount, "");
4393 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_truncated_msgcount, oslog_p_truncated_msgcount, "");
4394
4395 SYSCTL_SCALABLE_COUNTER(_debug, oslog_s_total_msgcount, oslog_s_total_msgcount, "Number of logs sent to streaming");
4396 SYSCTL_SCALABLE_COUNTER(_debug, oslog_s_metadata_msgcount, oslog_s_metadata_msgcount, "Number of metadata sent to streaming");
4397 SYSCTL_SCALABLE_COUNTER(_debug, oslog_s_error_count, oslog_s_error_count, "Number of invalid stream logs");
4398 SYSCTL_SCALABLE_COUNTER(_debug, oslog_s_streamed_msgcount, oslog_s_streamed_msgcount, "Number of streamed logs");
4399 SYSCTL_SCALABLE_COUNTER(_debug, oslog_s_dropped_msgcount, oslog_s_dropped_msgcount, "Number of logs dropped from stream");
4400
4401 SYSCTL_SCALABLE_COUNTER(_debug, oslog_msgbuf_msgcount, oslog_msgbuf_msgcount, "Number of dmesg log messages");
4402 SYSCTL_SCALABLE_COUNTER(_debug, oslog_msgbuf_dropped_msgcount, oslog_msgbuf_dropped_msgcount, "Number of dropped dmesg log messages");
4403 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");
4404
4405 SYSCTL_UINT(_debug, OID_AUTO, vaddlog_msgcount, CTLFLAG_ANYBODY | CTLFLAG_RD | CTLFLAG_LOCKED, &vaddlog_msgcount, 0, "");
4406 SYSCTL_UINT(_debug, OID_AUTO, vaddlog_msgcount_dropped, CTLFLAG_ANYBODY | CTLFLAG_RD | CTLFLAG_LOCKED, &vaddlog_msgcount_dropped, 0, "");
4407
4408 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_received, log_queue_cnt_received, "Number of received logs");
4409 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_rejected_fh, log_queue_cnt_rejected_fh, "Number of logs initially rejected by FH");
4410 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_sent, log_queue_cnt_sent, "Number of logs successfully saved in FH");
4411 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_dropped_nomem, log_queue_cnt_dropped_nomem, "Number of logs dropped due to lack of queue memory");
4412 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_queued, log_queue_cnt_queued, "Current number of logs stored in log queues");
4413 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_dropped_off, log_queue_cnt_dropped_off, "Number of logs dropped due to disabled log queues");
4414 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_mem_allocated, log_queue_cnt_mem_allocated, "Number of memory allocations");
4415 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_mem_released, log_queue_cnt_mem_released, "Number of memory releases");
4416 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_mem_failed, log_queue_cnt_mem_failed, "Number of failed memory allocations");
4417
4418 #endif /* DEVELOPMENT || DEBUG */
4419
4420 /*
4421 * Enable tracing of voucher contents
4422 */
4423 extern uint32_t ipc_voucher_trace_contents;
4424
4425 SYSCTL_INT(_kern, OID_AUTO, ipc_voucher_trace_contents,
4426 CTLFLAG_RW | CTLFLAG_LOCKED, &ipc_voucher_trace_contents, 0, "Enable tracing voucher contents");
4427
4428 /*
4429 * Kernel stack size and depth
4430 */
4431 SYSCTL_INT(_kern, OID_AUTO, stack_size,
4432 CTLFLAG_RD | CTLFLAG_LOCKED, (int *) &kernel_stack_size, 0, "Kernel stack size");
4433 SYSCTL_INT(_kern, OID_AUTO, stack_depth_max,
4434 CTLFLAG_RD | CTLFLAG_LOCKED, (int *) &kernel_stack_depth_max, 0, "Max kernel stack depth at interrupt or context switch");
4435
4436 extern unsigned int kern_feature_overrides;
4437 SYSCTL_INT(_kern, OID_AUTO, kern_feature_overrides,
4438 CTLFLAG_RD | CTLFLAG_LOCKED, &kern_feature_overrides, 0, "Kernel feature override mask");
4439
4440 /*
4441 * enable back trace for port allocations
4442 */
4443 extern int ipc_portbt;
4444
4445 SYSCTL_INT(_kern, OID_AUTO, ipc_portbt,
4446 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
4447 &ipc_portbt, 0, "");
4448
4449 /*
4450 * Mach message signature validation control and outputs
4451 */
4452 extern unsigned int ikm_signature_failures;
4453 SYSCTL_INT(_kern, OID_AUTO, ikm_signature_failures,
4454 CTLFLAG_RD | CTLFLAG_LOCKED, &ikm_signature_failures, 0, "Message signature failure count");
4455 extern unsigned int ikm_signature_failure_id;
4456 SYSCTL_INT(_kern, OID_AUTO, ikm_signature_failure_id,
4457 CTLFLAG_RD | CTLFLAG_LOCKED, &ikm_signature_failure_id, 0, "Message signature failure count");
4458
4459 #if (DEVELOPMENT || DEBUG)
4460 extern unsigned int ikm_signature_panic_disable;
4461 SYSCTL_INT(_kern, OID_AUTO, ikm_signature_panic_disable,
4462 CTLFLAG_RW | CTLFLAG_LOCKED, &ikm_signature_panic_disable, 0, "Message signature failure mode");
4463 extern unsigned int ikm_signature_header_failures;
4464 SYSCTL_INT(_kern, OID_AUTO, ikm_signature_header_failures,
4465 CTLFLAG_RD | CTLFLAG_LOCKED, &ikm_signature_header_failures, 0, "Message header signature failure count");
4466 extern unsigned int ikm_signature_trailer_failures;
4467 SYSCTL_INT(_kern, OID_AUTO, ikm_signature_trailer_failures,
4468 CTLFLAG_RD | CTLFLAG_LOCKED, &ikm_signature_trailer_failures, 0, "Message trailer signature failure count");
4469 #endif
4470
4471 /*
4472 * Scheduler sysctls
4473 */
4474
4475 SYSCTL_STRING(_kern, OID_AUTO, sched,
4476 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
4477 sched_string, sizeof(sched_string),
4478 "Timeshare scheduler implementation");
4479
4480 #if CONFIG_QUIESCE_COUNTER
4481 static int
4482 sysctl_cpu_quiescent_counter_interval SYSCTL_HANDLER_ARGS
4483 {
4484 #pragma unused(arg1, arg2)
4485
4486 uint32_t local_min_interval_us = cpu_quiescent_counter_get_min_interval_us();
4487
4488 int error = sysctl_handle_int(oidp, &local_min_interval_us, 0, req);
4489 if (error || !req->newptr) {
4490 return error;
4491 }
4492
4493 cpu_quiescent_counter_set_min_interval_us(local_min_interval_us);
4494
4495 return 0;
4496 }
4497
4498 SYSCTL_PROC(_kern, OID_AUTO, cpu_checkin_interval,
4499 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
4500 0, 0,
4501 sysctl_cpu_quiescent_counter_interval, "I",
4502 "Quiescent CPU checkin interval (microseconds)");
4503 #endif /* CONFIG_QUIESCE_COUNTER */
4504
4505
4506 /*
4507 * Only support runtime modification on development / debug
4508 */
4509 #if DEVELOPMENT || DEBUG
4510 extern int precise_user_kernel_time;
4511 SYSCTL_INT(_kern, OID_AUTO, precise_user_kernel_time,
4512 CTLFLAG_RW | CTLFLAG_LOCKED,
4513 &precise_user_kernel_time, 0, "Precise accounting of kernel vs. user time");
4514 #endif /* DEVELOPMENT || DEBUG */
4515
4516 /* Parameters related to timer coalescing tuning, to be replaced
4517 * with a dedicated systemcall in the future.
4518 */
4519 /* Enable processing pending timers in the context of any other interrupt
4520 * Coalescing tuning parameters for various thread/task attributes */
4521 STATIC int
4522 sysctl_timer_user_us_kernel_abstime SYSCTL_HANDLER_ARGS
4523 {
4524 #pragma unused(oidp)
4525 int size = arg2; /* subcommand*/
4526 int error;
4527 int changed = 0;
4528 uint64_t old_value_ns;
4529 uint64_t new_value_ns;
4530 uint64_t value_abstime;
4531 if (size == sizeof(uint32_t)) {
4532 value_abstime = *((uint32_t *)arg1);
4533 } else if (size == sizeof(uint64_t)) {
4534 value_abstime = *((uint64_t *)arg1);
4535 } else {
4536 return ENOTSUP;
4537 }
4538
4539 absolutetime_to_nanoseconds(value_abstime, &old_value_ns);
4540 error = sysctl_io_number(req, old_value_ns, sizeof(old_value_ns), &new_value_ns, &changed);
4541 if ((error) || (!changed)) {
4542 return error;
4543 }
4544
4545 nanoseconds_to_absolutetime(new_value_ns, &value_abstime);
4546 if (size == sizeof(uint32_t)) {
4547 *((uint32_t *)arg1) = (uint32_t)value_abstime;
4548 } else {
4549 *((uint64_t *)arg1) = value_abstime;
4550 }
4551 return error;
4552 }
4553
4554 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_bg_scale,
4555 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4556 &tcoal_prio_params.timer_coalesce_bg_shift, 0, "");
4557 SYSCTL_PROC(_kern, OID_AUTO, timer_resort_threshold_ns,
4558 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4559 &tcoal_prio_params.timer_resort_threshold_abstime,
4560 sizeof(tcoal_prio_params.timer_resort_threshold_abstime),
4561 sysctl_timer_user_us_kernel_abstime,
4562 "Q", "");
4563 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_bg_ns_max,
4564 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4565 &tcoal_prio_params.timer_coalesce_bg_abstime_max,
4566 sizeof(tcoal_prio_params.timer_coalesce_bg_abstime_max),
4567 sysctl_timer_user_us_kernel_abstime,
4568 "Q", "");
4569
4570 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_kt_scale,
4571 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4572 &tcoal_prio_params.timer_coalesce_kt_shift, 0, "");
4573
4574 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_kt_ns_max,
4575 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4576 &tcoal_prio_params.timer_coalesce_kt_abstime_max,
4577 sizeof(tcoal_prio_params.timer_coalesce_kt_abstime_max),
4578 sysctl_timer_user_us_kernel_abstime,
4579 "Q", "");
4580
4581 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_fp_scale,
4582 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4583 &tcoal_prio_params.timer_coalesce_fp_shift, 0, "");
4584
4585 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_fp_ns_max,
4586 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4587 &tcoal_prio_params.timer_coalesce_fp_abstime_max,
4588 sizeof(tcoal_prio_params.timer_coalesce_fp_abstime_max),
4589 sysctl_timer_user_us_kernel_abstime,
4590 "Q", "");
4591
4592 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_ts_scale,
4593 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4594 &tcoal_prio_params.timer_coalesce_ts_shift, 0, "");
4595
4596 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_ts_ns_max,
4597 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4598 &tcoal_prio_params.timer_coalesce_ts_abstime_max,
4599 sizeof(tcoal_prio_params.timer_coalesce_ts_abstime_max),
4600 sysctl_timer_user_us_kernel_abstime,
4601 "Q", "");
4602
4603 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier0_scale,
4604 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4605 &tcoal_prio_params.latency_qos_scale[0], 0, "");
4606
4607 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier0_ns_max,
4608 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4609 &tcoal_prio_params.latency_qos_abstime_max[0],
4610 sizeof(tcoal_prio_params.latency_qos_abstime_max[0]),
4611 sysctl_timer_user_us_kernel_abstime,
4612 "Q", "");
4613
4614 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier1_scale,
4615 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4616 &tcoal_prio_params.latency_qos_scale[1], 0, "");
4617
4618 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier1_ns_max,
4619 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4620 &tcoal_prio_params.latency_qos_abstime_max[1],
4621 sizeof(tcoal_prio_params.latency_qos_abstime_max[1]),
4622 sysctl_timer_user_us_kernel_abstime,
4623 "Q", "");
4624
4625 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier2_scale,
4626 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4627 &tcoal_prio_params.latency_qos_scale[2], 0, "");
4628
4629 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier2_ns_max,
4630 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4631 &tcoal_prio_params.latency_qos_abstime_max[2],
4632 sizeof(tcoal_prio_params.latency_qos_abstime_max[2]),
4633 sysctl_timer_user_us_kernel_abstime,
4634 "Q", "");
4635
4636 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier3_scale,
4637 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4638 &tcoal_prio_params.latency_qos_scale[3], 0, "");
4639
4640 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier3_ns_max,
4641 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4642 &tcoal_prio_params.latency_qos_abstime_max[3],
4643 sizeof(tcoal_prio_params.latency_qos_abstime_max[3]),
4644 sysctl_timer_user_us_kernel_abstime,
4645 "Q", "");
4646
4647 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier4_scale,
4648 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4649 &tcoal_prio_params.latency_qos_scale[4], 0, "");
4650
4651 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier4_ns_max,
4652 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4653 &tcoal_prio_params.latency_qos_abstime_max[4],
4654 sizeof(tcoal_prio_params.latency_qos_abstime_max[4]),
4655 sysctl_timer_user_us_kernel_abstime,
4656 "Q", "");
4657
4658 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier5_scale,
4659 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4660 &tcoal_prio_params.latency_qos_scale[5], 0, "");
4661
4662 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier5_ns_max,
4663 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4664 &tcoal_prio_params.latency_qos_abstime_max[5],
4665 sizeof(tcoal_prio_params.latency_qos_abstime_max[5]),
4666 sysctl_timer_user_us_kernel_abstime,
4667 "Q", "");
4668
4669 /* Communicate the "user idle level" heuristic to the timer layer, and
4670 * potentially other layers in the future.
4671 */
4672
4673 static int
timer_user_idle_level(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)4674 timer_user_idle_level(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4675 {
4676 int new_value = 0, old_value = 0, changed = 0, error;
4677
4678 old_value = timer_get_user_idle_level();
4679
4680 error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
4681
4682 if (error == 0 && changed) {
4683 if (timer_set_user_idle_level(new_value) != KERN_SUCCESS) {
4684 error = ERANGE;
4685 }
4686 }
4687
4688 return error;
4689 }
4690
4691 SYSCTL_PROC(_machdep, OID_AUTO, user_idle_level,
4692 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
4693 0, 0,
4694 timer_user_idle_level, "I", "User idle level heuristic, 0-128");
4695
4696 #if HYPERVISOR
4697 SYSCTL_INT(_kern, OID_AUTO, hv_support,
4698 CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
4699 &hv_support_available, 0, "");
4700
4701 SYSCTL_INT(_kern, OID_AUTO, hv_disable,
4702 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
4703 &hv_disable, 0, "");
4704 #endif
4705
4706 #if DEVELOPMENT || DEBUG
4707 extern uint64_t driverkit_checkin_timed_out;
4708 SYSCTL_QUAD(_kern, OID_AUTO, driverkit_checkin_timed_out,
4709 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
4710 &driverkit_checkin_timed_out, "timestamp of dext checkin timeout");
4711 #endif
4712
4713 extern int IOGetVMMPresent(void);
4714
4715 static int
4716 hv_vmm_present SYSCTL_HANDLER_ARGS
4717 {
4718 __unused struct sysctl_oid *unused_oidp = oidp;
4719 __unused void *unused_arg1 = arg1;
4720 __unused int unused_arg2 = arg2;
4721
4722 int hv_vmm_present = 0;
4723
4724 hv_vmm_present = IOGetVMMPresent();
4725
4726 return SYSCTL_OUT(req, &hv_vmm_present, sizeof(hv_vmm_present));
4727 }
4728
4729 SYSCTL_PROC(_kern, OID_AUTO, hv_vmm_present,
4730 CTLTYPE_INT | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED,
4731 0, 0,
4732 hv_vmm_present, "I", "");
4733
4734 #if CONFIG_DARKBOOT
4735 STATIC int
4736 sysctl_darkboot SYSCTL_HANDLER_ARGS
4737 {
4738 int err = 0, value = 0;
4739 #pragma unused(oidp, arg1, arg2, err, value, req)
4740
4741 /*
4742 * Handle the sysctl request.
4743 *
4744 * If this is a read, the function will set the value to the current darkboot value. Otherwise,
4745 * we'll get the request identifier into "value" and then we can honor it.
4746 */
4747 if ((err = sysctl_io_number(req, darkboot, sizeof(int), &value, NULL)) != 0) {
4748 goto exit;
4749 }
4750
4751 /* writing requested, let's process the request */
4752 if (req->newptr) {
4753 /* writing is protected by an entitlement */
4754 if (priv_check_cred(kauth_cred_get(), PRIV_DARKBOOT, 0) != 0) {
4755 err = EPERM;
4756 goto exit;
4757 }
4758
4759 switch (value) {
4760 case MEMORY_MAINTENANCE_DARK_BOOT_UNSET:
4761 /*
4762 * If the darkboot sysctl is unset, the NVRAM variable
4763 * must be unset too. If that's not the case, it means
4764 * someone is doing something crazy and not supported.
4765 */
4766 if (darkboot != 0) {
4767 int ret = PERemoveNVRAMProperty(MEMORY_MAINTENANCE_DARK_BOOT_NVRAM_NAME);
4768 if (ret) {
4769 darkboot = 0;
4770 } else {
4771 err = EINVAL;
4772 }
4773 }
4774 break;
4775 case MEMORY_MAINTENANCE_DARK_BOOT_SET:
4776 darkboot = 1;
4777 break;
4778 case MEMORY_MAINTENANCE_DARK_BOOT_SET_PERSISTENT: {
4779 /*
4780 * Set the NVRAM and update 'darkboot' in case
4781 * of success. Otherwise, do not update
4782 * 'darkboot' and report the failure.
4783 */
4784 if (PEWriteNVRAMBooleanProperty(MEMORY_MAINTENANCE_DARK_BOOT_NVRAM_NAME, TRUE)) {
4785 darkboot = 1;
4786 } else {
4787 err = EINVAL;
4788 }
4789
4790 break;
4791 }
4792 default:
4793 err = EINVAL;
4794 }
4795 }
4796
4797 exit:
4798 return err;
4799 }
4800
4801 SYSCTL_PROC(_kern, OID_AUTO, darkboot,
4802 CTLFLAG_KERN | CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_ANYBODY,
4803 0, 0, sysctl_darkboot, "I", "");
4804 #endif /* CONFIG_DARKBOOT */
4805
4806 #if DEVELOPMENT || DEBUG
4807 #include <sys/sysent.h>
4808 /* This should result in a fatal exception, verifying that "sysent" is
4809 * write-protected.
4810 */
4811 static int
kern_sysent_write(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)4812 kern_sysent_write(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4813 {
4814 uint64_t new_value = 0, old_value = 0;
4815 int changed = 0, error;
4816
4817 error = sysctl_io_number(req, old_value, sizeof(uint64_t), &new_value, &changed);
4818 if ((error == 0) && changed) {
4819 volatile uint32_t *wraddr = __DECONST(uint32_t *, &sysent[0]);
4820 *wraddr = 0;
4821 printf("sysent[0] write succeeded\n");
4822 }
4823 return error;
4824 }
4825
4826 SYSCTL_PROC(_kern, OID_AUTO, sysent_const_check,
4827 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
4828 0, 0,
4829 kern_sysent_write, "I", "Attempt sysent[0] write");
4830
4831 #endif
4832
4833 #if DEVELOPMENT || DEBUG
4834 SYSCTL_COMPAT_INT(_kern, OID_AUTO, development, CTLFLAG_RD | CTLFLAG_MASKED | CTLFLAG_KERN, NULL, 1, "");
4835 #else
4836 SYSCTL_COMPAT_INT(_kern, OID_AUTO, development, CTLFLAG_RD | CTLFLAG_MASKED, NULL, 0, "");
4837 #endif
4838
4839
4840 #if DEVELOPMENT || DEBUG
4841
4842 static int
4843 sysctl_panic_test SYSCTL_HANDLER_ARGS
4844 {
4845 #pragma unused(arg1, arg2)
4846 int rval = 0;
4847 char str[32] = "entry prelog postlog postcore";
4848
4849 rval = sysctl_handle_string(oidp, str, sizeof(str), req);
4850
4851 if (rval == 0 && req->newptr) {
4852 if (strncmp("entry", str, strlen("entry")) == 0) {
4853 panic_with_options(0, NULL, DEBUGGER_OPTION_RECURPANIC_ENTRY, "test recursive panic at entry");
4854 } else if (strncmp("prelog", str, strlen("prelog")) == 0) {
4855 panic_with_options(0, NULL, DEBUGGER_OPTION_RECURPANIC_PRELOG, "test recursive panic prior to writing a paniclog");
4856 } else if (strncmp("postlog", str, strlen("postlog")) == 0) {
4857 panic_with_options(0, NULL, DEBUGGER_OPTION_RECURPANIC_POSTLOG, "test recursive panic subsequent to paniclog");
4858 } else if (strncmp("postcore", str, strlen("postcore")) == 0) {
4859 panic_with_options(0, NULL, DEBUGGER_OPTION_RECURPANIC_POSTCORE, "test recursive panic subsequent to on-device core");
4860 }
4861 }
4862
4863 return rval;
4864 }
4865
4866 static int
4867 sysctl_debugger_test SYSCTL_HANDLER_ARGS
4868 {
4869 #pragma unused(arg1, arg2)
4870 int rval = 0;
4871 char str[32] = "entry prelog postlog postcore";
4872
4873 rval = sysctl_handle_string(oidp, str, sizeof(str), req);
4874
4875 if (rval == 0 && req->newptr) {
4876 if (strncmp("entry", str, strlen("entry")) == 0) {
4877 DebuggerWithContext(0, NULL, "test recursive panic via debugger at entry", DEBUGGER_OPTION_RECURPANIC_ENTRY, (unsigned long)(char *)__builtin_return_address(0));
4878 } else if (strncmp("prelog", str, strlen("prelog")) == 0) {
4879 DebuggerWithContext(0, NULL, "test recursive panic via debugger prior to writing a paniclog", DEBUGGER_OPTION_RECURPANIC_PRELOG, (unsigned long)(char *)__builtin_return_address(0));
4880 } else if (strncmp("postlog", str, strlen("postlog")) == 0) {
4881 DebuggerWithContext(0, NULL, "test recursive panic via debugger subsequent to paniclog", DEBUGGER_OPTION_RECURPANIC_POSTLOG, (unsigned long)(char *)__builtin_return_address(0));
4882 } else if (strncmp("postcore", str, strlen("postcore")) == 0) {
4883 DebuggerWithContext(0, NULL, "test recursive panic via debugger subsequent to on-device core", DEBUGGER_OPTION_RECURPANIC_POSTCORE, (unsigned long)(char *)__builtin_return_address(0));
4884 }
4885 }
4886
4887 return rval;
4888 }
4889
4890 decl_lck_spin_data(, spinlock_panic_test_lock);
4891
4892 __attribute__((noreturn))
4893 static void
spinlock_panic_test_acquire_spinlock(void * arg __unused,wait_result_t wres __unused)4894 spinlock_panic_test_acquire_spinlock(void * arg __unused, wait_result_t wres __unused)
4895 {
4896 lck_spin_lock(&spinlock_panic_test_lock);
4897 while (1) {
4898 ;
4899 }
4900 }
4901
4902 static int
4903 sysctl_spinlock_panic_test SYSCTL_HANDLER_ARGS
4904 {
4905 #pragma unused(oidp, arg1, arg2)
4906 if (req->newlen == 0) {
4907 return EINVAL;
4908 }
4909
4910 thread_t panic_spinlock_thread;
4911 /* Initialize panic spinlock */
4912 lck_grp_t * panic_spinlock_grp;
4913 lck_grp_attr_t * panic_spinlock_grp_attr;
4914 lck_attr_t * panic_spinlock_attr;
4915
4916 panic_spinlock_grp_attr = lck_grp_attr_alloc_init();
4917 panic_spinlock_grp = lck_grp_alloc_init("panic_spinlock", panic_spinlock_grp_attr);
4918 panic_spinlock_attr = lck_attr_alloc_init();
4919
4920 lck_spin_init(&spinlock_panic_test_lock, panic_spinlock_grp, panic_spinlock_attr);
4921
4922
4923 /* Create thread to acquire spinlock */
4924 if (kernel_thread_start(spinlock_panic_test_acquire_spinlock, NULL, &panic_spinlock_thread) != KERN_SUCCESS) {
4925 return EBUSY;
4926 }
4927
4928 /* Try to acquire spinlock -- should panic eventually */
4929 lck_spin_lock(&spinlock_panic_test_lock);
4930 while (1) {
4931 ;
4932 }
4933 }
4934
4935 __attribute__((noreturn))
4936 static void
simultaneous_panic_worker(void * arg,wait_result_t wres __unused)4937 simultaneous_panic_worker
4938 (void * arg, wait_result_t wres __unused)
4939 {
4940 atomic_int *start_panic = (atomic_int *)arg;
4941
4942 while (!atomic_load(start_panic)) {
4943 ;
4944 }
4945 panic("SIMULTANEOUS PANIC TEST: INITIATING PANIC FROM CPU %d", cpu_number());
4946 __builtin_unreachable();
4947 }
4948
4949 static int
4950 sysctl_simultaneous_panic_test SYSCTL_HANDLER_ARGS
4951 {
4952 #pragma unused(oidp, arg1, arg2)
4953 if (req->newlen == 0) {
4954 return EINVAL;
4955 }
4956
4957 int i = 0, threads_to_create = 2 * processor_count;
4958 atomic_int start_panic = 0;
4959 unsigned int threads_created = 0;
4960 thread_t new_panic_thread;
4961
4962 for (i = threads_to_create; i > 0; i--) {
4963 if (kernel_thread_start(simultaneous_panic_worker, (void *) &start_panic, &new_panic_thread) == KERN_SUCCESS) {
4964 threads_created++;
4965 }
4966 }
4967
4968 /* FAIL if we couldn't create at least processor_count threads */
4969 if (threads_created < processor_count) {
4970 panic("SIMULTANEOUS PANIC TEST: FAILED TO CREATE ENOUGH THREADS, ONLY CREATED %d (of %d)",
4971 threads_created, threads_to_create);
4972 }
4973
4974 atomic_exchange(&start_panic, 1);
4975 while (1) {
4976 ;
4977 }
4978 }
4979
4980 SYSCTL_PROC(_debug, OID_AUTO, panic_test, CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_MASKED, 0, 0, sysctl_panic_test, "A", "panic test");
4981 SYSCTL_PROC(_debug, OID_AUTO, debugger_test, CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_MASKED, 0, 0, sysctl_debugger_test, "A", "debugger test");
4982 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");
4983 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");
4984
4985 extern int exc_resource_threads_enabled;
4986
4987 SYSCTL_INT(_kern, OID_AUTO, exc_resource_threads_enabled, CTLFLAG_RD | CTLFLAG_LOCKED, &exc_resource_threads_enabled, 0, "exc_resource thread limit enabled");
4988
4989
4990 #endif /* DEVELOPMENT || DEBUG */
4991
4992 #if CONFIG_THREAD_GROUPS
4993 #if DEVELOPMENT || DEBUG
4994
4995 static int
4996 sysctl_get_thread_group_id SYSCTL_HANDLER_ARGS
4997 {
4998 #pragma unused(arg1, arg2, oidp)
4999 uint64_t thread_group_id = thread_group_get_id(thread_group_get(current_thread()));
5000 return SYSCTL_OUT(req, &thread_group_id, sizeof(thread_group_id));
5001 }
5002
5003 SYSCTL_PROC(_kern, OID_AUTO, thread_group_id, CTLFLAG_RD | CTLFLAG_LOCKED | CTLTYPE_QUAD,
5004 0, 0, &sysctl_get_thread_group_id, "I", "thread group id of the thread");
5005
5006 STATIC int
sysctl_thread_group_count(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)5007 sysctl_thread_group_count(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
5008 {
5009 int value = thread_group_count();
5010 return sysctl_io_number(req, value, sizeof(value), NULL, NULL);
5011 }
5012
5013 SYSCTL_PROC(_kern, OID_AUTO, thread_group_count, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5014 0, 0, &sysctl_thread_group_count, "I", "count of thread groups");
5015
5016 #endif /* DEVELOPMENT || DEBUG */
5017 const uint32_t thread_groups_supported = 1;
5018 #else /* CONFIG_THREAD_GROUPS */
5019 const uint32_t thread_groups_supported = 0;
5020 #endif /* CONFIG_THREAD_GROUPS */
5021
5022 STATIC int
sysctl_thread_groups_supported(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)5023 sysctl_thread_groups_supported(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
5024 {
5025 int value = thread_groups_supported;
5026 return sysctl_io_number(req, value, sizeof(value), NULL, NULL);
5027 }
5028
5029 SYSCTL_PROC(_kern, OID_AUTO, thread_groups_supported, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5030 0, 0, &sysctl_thread_groups_supported, "I", "thread groups supported");
5031
5032 static int
5033 sysctl_grade_cputype SYSCTL_HANDLER_ARGS
5034 {
5035 #pragma unused(arg1, arg2, oidp)
5036 int error = 0;
5037 int type_tuple[2] = {};
5038 int return_value = 0;
5039
5040 error = SYSCTL_IN(req, &type_tuple, sizeof(type_tuple));
5041
5042 if (error) {
5043 return error;
5044 }
5045
5046 return_value = grade_binary(type_tuple[0], type_tuple[1] & ~CPU_SUBTYPE_MASK, type_tuple[1] & CPU_SUBTYPE_MASK, FALSE);
5047
5048 error = SYSCTL_OUT(req, &return_value, sizeof(return_value));
5049
5050 if (error) {
5051 return error;
5052 }
5053
5054 return error;
5055 }
5056
5057 SYSCTL_PROC(_kern, OID_AUTO, grade_cputype,
5058 CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_MASKED | CTLFLAG_LOCKED | CTLTYPE_OPAQUE,
5059 0, 0, &sysctl_grade_cputype, "S",
5060 "grade value of cpu_type_t+cpu_sub_type_t");
5061
5062 extern boolean_t allow_direct_handoff;
5063 SYSCTL_INT(_kern, OID_AUTO, direct_handoff,
5064 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5065 &allow_direct_handoff, 0, "Enable direct handoff for realtime threads");
5066
5067 #if DEVELOPMENT || DEBUG
5068
5069 SYSCTL_LONG(_kern, OID_AUTO, phys_carveout_pa, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5070 &phys_carveout_pa,
5071 "base physical address of the phys_carveout_mb boot-arg region");
5072 SYSCTL_LONG(_kern, OID_AUTO, phys_carveout_size, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5073 &phys_carveout_size,
5074 "size in bytes of the phys_carveout_mb boot-arg region");
5075 SYSCTL_LONG(_kern, OID_AUTO, phys_carveout_metadata_pa, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5076 &phys_carveout_metadata_pa,
5077 "base physical address of the phys_carveout_metadata region");
5078 SYSCTL_LONG(_kern, OID_AUTO, phys_carveout_metadata_size, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5079 &phys_carveout_metadata_size,
5080 "size in bytes of the phys_carveout_metadata region");
5081
5082 extern void do_cseg_wedge_thread(void);
5083 extern void do_cseg_unwedge_thread(void);
5084
5085 static int
5086 cseg_wedge_thread SYSCTL_HANDLER_ARGS
5087 {
5088 #pragma unused(arg1, arg2)
5089
5090 int error, val = 0;
5091 error = sysctl_handle_int(oidp, &val, 0, req);
5092 if (error || val == 0) {
5093 return error;
5094 }
5095
5096 do_cseg_wedge_thread();
5097 return 0;
5098 }
5099 SYSCTL_PROC(_kern, OID_AUTO, cseg_wedge_thread, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_MASKED, 0, 0, cseg_wedge_thread, "I", "wedge c_seg thread");
5100
5101 static int
5102 cseg_unwedge_thread SYSCTL_HANDLER_ARGS
5103 {
5104 #pragma unused(arg1, arg2)
5105
5106 int error, val = 0;
5107 error = sysctl_handle_int(oidp, &val, 0, req);
5108 if (error || val == 0) {
5109 return error;
5110 }
5111
5112 do_cseg_unwedge_thread();
5113 return 0;
5114 }
5115 SYSCTL_PROC(_kern, OID_AUTO, cseg_unwedge_thread, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_MASKED, 0, 0, cseg_unwedge_thread, "I", "unstuck c_seg thread");
5116
5117 static atomic_int wedge_thread_should_wake = 0;
5118
5119 static int
5120 unwedge_thread SYSCTL_HANDLER_ARGS
5121 {
5122 #pragma unused(arg1, arg2)
5123 int error, val = 0;
5124 error = sysctl_handle_int(oidp, &val, 0, req);
5125 if (error || val == 0) {
5126 return error;
5127 }
5128
5129 atomic_store(&wedge_thread_should_wake, 1);
5130 return 0;
5131 }
5132
5133 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");
5134
5135 static int
5136 wedge_thread SYSCTL_HANDLER_ARGS
5137 {
5138 #pragma unused(arg1, arg2)
5139
5140 int error, val = 0;
5141 error = sysctl_handle_int(oidp, &val, 0, req);
5142 if (error || val == 0) {
5143 return error;
5144 }
5145
5146 uint64_t interval = 1;
5147 nanoseconds_to_absolutetime(1000 * 1000 * 50, &interval);
5148
5149 atomic_store(&wedge_thread_should_wake, 0);
5150 while (!atomic_load(&wedge_thread_should_wake)) {
5151 tsleep1(NULL, 0, "wedge_thread", mach_absolute_time() + interval, NULL);
5152 }
5153
5154 return 0;
5155 }
5156
5157 SYSCTL_PROC(_kern, OID_AUTO, wedge_thread,
5158 CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED, 0, 0, wedge_thread, "I",
5159 "wedge this thread so it cannot be cleaned up");
5160
5161 static int
5162 sysctl_total_corpses_count SYSCTL_HANDLER_ARGS
5163 {
5164 #pragma unused(oidp, arg1, arg2)
5165 extern unsigned long total_corpses_count(void);
5166
5167 unsigned long corpse_count_long = total_corpses_count();
5168 unsigned int corpse_count = (unsigned int)MIN(corpse_count_long, UINT_MAX);
5169 return sysctl_io_opaque(req, &corpse_count, sizeof(corpse_count), NULL);
5170 }
5171
5172 SYSCTL_PROC(_kern, OID_AUTO, total_corpses_count,
5173 CTLFLAG_RD | CTLFLAG_ANYBODY | CTLFLAG_LOCKED, 0, 0,
5174 sysctl_total_corpses_count, "I", "total corpses on the system");
5175
5176 static int
5177 sysctl_turnstile_test_prim_lock SYSCTL_HANDLER_ARGS;
5178 static int
5179 sysctl_turnstile_test_prim_unlock SYSCTL_HANDLER_ARGS;
5180 int
5181 tstile_test_prim_lock(boolean_t use_hashtable);
5182 int
5183 tstile_test_prim_unlock(boolean_t use_hashtable);
5184
5185 static int
5186 sysctl_turnstile_test_prim_lock SYSCTL_HANDLER_ARGS
5187 {
5188 #pragma unused(arg1, arg2)
5189 int error, val = 0;
5190 error = sysctl_handle_int(oidp, &val, 0, req);
5191 if (error || val == 0) {
5192 return error;
5193 }
5194 switch (val) {
5195 case SYSCTL_TURNSTILE_TEST_USER_DEFAULT:
5196 case SYSCTL_TURNSTILE_TEST_USER_HASHTABLE:
5197 case SYSCTL_TURNSTILE_TEST_KERNEL_DEFAULT:
5198 case SYSCTL_TURNSTILE_TEST_KERNEL_HASHTABLE:
5199 return tstile_test_prim_lock(val);
5200 default:
5201 return error;
5202 }
5203 }
5204
5205 static int
5206 sysctl_turnstile_test_prim_unlock SYSCTL_HANDLER_ARGS
5207 {
5208 #pragma unused(arg1, arg2)
5209 int error, val = 0;
5210 error = sysctl_handle_int(oidp, &val, 0, req);
5211 if (error || val == 0) {
5212 return error;
5213 }
5214 switch (val) {
5215 case SYSCTL_TURNSTILE_TEST_USER_DEFAULT:
5216 case SYSCTL_TURNSTILE_TEST_USER_HASHTABLE:
5217 case SYSCTL_TURNSTILE_TEST_KERNEL_DEFAULT:
5218 case SYSCTL_TURNSTILE_TEST_KERNEL_HASHTABLE:
5219 return tstile_test_prim_unlock(val);
5220 default:
5221 return error;
5222 }
5223 }
5224
5225 SYSCTL_PROC(_kern, OID_AUTO, turnstiles_test_lock, CTLFLAG_WR | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED,
5226 0, 0, sysctl_turnstile_test_prim_lock, "I", "turnstiles test lock");
5227
5228 SYSCTL_PROC(_kern, OID_AUTO, turnstiles_test_unlock, CTLFLAG_WR | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED,
5229 0, 0, sysctl_turnstile_test_prim_unlock, "I", "turnstiles test unlock");
5230
5231 int
5232 turnstile_get_boost_stats_sysctl(void *req);
5233 int
5234 turnstile_get_unboost_stats_sysctl(void *req);
5235 static int
5236 sysctl_turnstile_boost_stats SYSCTL_HANDLER_ARGS;
5237 static int
5238 sysctl_turnstile_unboost_stats SYSCTL_HANDLER_ARGS;
5239 extern uint64_t thread_block_on_turnstile_count;
5240 extern uint64_t thread_block_on_regular_waitq_count;
5241
5242 static int
5243 sysctl_turnstile_boost_stats SYSCTL_HANDLER_ARGS
5244 {
5245 #pragma unused(arg1, arg2, oidp)
5246 return turnstile_get_boost_stats_sysctl(req);
5247 }
5248
5249 static int
5250 sysctl_turnstile_unboost_stats SYSCTL_HANDLER_ARGS
5251 {
5252 #pragma unused(arg1, arg2, oidp)
5253 return turnstile_get_unboost_stats_sysctl(req);
5254 }
5255
5256 SYSCTL_PROC(_kern, OID_AUTO, turnstile_boost_stats, CTLFLAG_RD | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED | CTLTYPE_STRUCT,
5257 0, 0, sysctl_turnstile_boost_stats, "S", "turnstiles boost stats");
5258 SYSCTL_PROC(_kern, OID_AUTO, turnstile_unboost_stats, CTLFLAG_RD | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED | CTLTYPE_STRUCT,
5259 0, 0, sysctl_turnstile_unboost_stats, "S", "turnstiles unboost stats");
5260 SYSCTL_QUAD(_kern, OID_AUTO, thread_block_count_on_turnstile,
5261 CTLFLAG_RD | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED,
5262 &thread_block_on_turnstile_count, "thread blocked on turnstile count");
5263 SYSCTL_QUAD(_kern, OID_AUTO, thread_block_count_on_reg_waitq,
5264 CTLFLAG_RD | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED,
5265 &thread_block_on_regular_waitq_count, "thread blocked on regular waitq count");
5266
5267 #if defined(__x86_64__)
5268 extern uint64_t MutexSpin;
5269
5270 SYSCTL_QUAD(_kern, OID_AUTO, mutex_spin_abs, CTLFLAG_RW, &MutexSpin,
5271 "Spin time in abs for acquiring a kernel mutex");
5272 #else
5273 extern machine_timeout32_t MutexSpin;
5274
5275 SYSCTL_UINT(_kern, OID_AUTO, mutex_spin_abs, CTLFLAG_RW, &MutexSpin, 0,
5276 "Spin time in abs for acquiring a kernel mutex");
5277 #endif
5278
5279 extern uint64_t low_MutexSpin;
5280 extern int64_t high_MutexSpin;
5281 extern unsigned int real_ncpus;
5282
5283 SYSCTL_QUAD(_kern, OID_AUTO, low_mutex_spin_abs, CTLFLAG_RW, &low_MutexSpin,
5284 "Low spin threshold in abs for acquiring a kernel mutex");
5285
5286 static int
5287 sysctl_high_mutex_spin_ns SYSCTL_HANDLER_ARGS
5288 {
5289 #pragma unused(oidp, arg1, arg2)
5290 int error;
5291 int64_t val = 0;
5292 int64_t res;
5293
5294 /* Check if the user is writing to high_MutexSpin, or just reading it */
5295 if (req->newptr) {
5296 error = SYSCTL_IN(req, &val, sizeof(val));
5297 if (error || (val < 0 && val != -1)) {
5298 return error;
5299 }
5300 high_MutexSpin = val;
5301 }
5302
5303 if (high_MutexSpin >= 0) {
5304 res = high_MutexSpin;
5305 } else {
5306 res = low_MutexSpin * real_ncpus;
5307 }
5308 return SYSCTL_OUT(req, &res, sizeof(res));
5309 }
5310 SYSCTL_PROC(_kern, OID_AUTO, high_mutex_spin_abs, CTLFLAG_RW | CTLTYPE_QUAD, 0, 0, sysctl_high_mutex_spin_ns, "I",
5311 "High spin threshold in abs for acquiring a kernel mutex");
5312
5313
5314 #if defined (__x86_64__)
5315
5316 semaphore_t sysctl_test_panic_with_thread_sem;
5317
5318 #pragma clang diagnostic push
5319 #pragma clang diagnostic ignored "-Winfinite-recursion" /* rdar://38801963 */
5320 __attribute__((noreturn))
5321 static void
panic_thread_test_child_spin(void * arg,wait_result_t wres)5322 panic_thread_test_child_spin(void * arg, wait_result_t wres)
5323 {
5324 static int panic_thread_recurse_count = 5;
5325
5326 if (panic_thread_recurse_count > 0) {
5327 panic_thread_recurse_count--;
5328 panic_thread_test_child_spin(arg, wres);
5329 }
5330
5331 semaphore_signal(sysctl_test_panic_with_thread_sem);
5332 while (1) {
5333 ;
5334 }
5335 }
5336 #pragma clang diagnostic pop
5337
5338 static void
panic_thread_test_child_park(void * arg __unused,wait_result_t wres __unused)5339 panic_thread_test_child_park(void * arg __unused, wait_result_t wres __unused)
5340 {
5341 int event;
5342
5343 assert_wait(&event, THREAD_UNINT);
5344 semaphore_signal(sysctl_test_panic_with_thread_sem);
5345 thread_block(panic_thread_test_child_park);
5346 }
5347
5348 static int
5349 sysctl_test_panic_with_thread SYSCTL_HANDLER_ARGS
5350 {
5351 #pragma unused(arg1, arg2)
5352 int rval = 0;
5353 char str[16] = { '\0' };
5354 thread_t child_thread = THREAD_NULL;
5355
5356 rval = sysctl_handle_string(oidp, str, sizeof(str), req);
5357 if (rval != 0 || !req->newptr) {
5358 return EINVAL;
5359 }
5360
5361 semaphore_create(kernel_task, &sysctl_test_panic_with_thread_sem, SYNC_POLICY_FIFO, 0);
5362
5363 /* Create thread to spin or park in continuation */
5364 if (strncmp("spin", str, strlen("spin")) == 0) {
5365 if (kernel_thread_start(panic_thread_test_child_spin, NULL, &child_thread) != KERN_SUCCESS) {
5366 semaphore_destroy(kernel_task, sysctl_test_panic_with_thread_sem);
5367 return EBUSY;
5368 }
5369 } else if (strncmp("continuation", str, strlen("continuation")) == 0) {
5370 if (kernel_thread_start(panic_thread_test_child_park, NULL, &child_thread) != KERN_SUCCESS) {
5371 semaphore_destroy(kernel_task, sysctl_test_panic_with_thread_sem);
5372 return EBUSY;
5373 }
5374 } else {
5375 semaphore_destroy(kernel_task, sysctl_test_panic_with_thread_sem);
5376 return EINVAL;
5377 }
5378
5379 semaphore_wait(sysctl_test_panic_with_thread_sem);
5380
5381 panic_with_thread_context(0, NULL, 0, child_thread, "testing panic_with_thread_context for thread %p", child_thread);
5382
5383 /* Not reached */
5384 return EINVAL;
5385 }
5386
5387 SYSCTL_PROC(_kern, OID_AUTO, test_panic_with_thread,
5388 CTLFLAG_MASKED | CTLFLAG_KERN | CTLFLAG_LOCKED | CTLFLAG_WR | CTLTYPE_STRING,
5389 0, 0, sysctl_test_panic_with_thread, "A", "test panic flow for backtracing a different thread");
5390 #endif /* defined (__x86_64__) */
5391
5392 #endif /* DEVELOPMENT || DEBUG */
5393
5394 static int
5395 sysctl_get_owned_vmobjects SYSCTL_HANDLER_ARGS
5396 {
5397 #pragma unused(oidp, arg1, arg2)
5398
5399 /* validate */
5400 if (req->newlen != sizeof(mach_port_name_t) || req->newptr == USER_ADDR_NULL ||
5401 req->oldidx != 0 || req->newidx != 0 || req->p == NULL ||
5402 (req->oldlen == 0 && req->oldptr != USER_ADDR_NULL)) {
5403 return EINVAL;
5404 }
5405
5406 int error;
5407 mach_port_name_t task_port_name;
5408 task_t task;
5409 size_t buffer_size = (req->oldptr != USER_ADDR_NULL) ? req->oldlen : 0;
5410 vmobject_list_output_t buffer = NULL;
5411 size_t output_size;
5412 size_t entries;
5413
5414 /* we have a "newptr" (for write) we get a task port name from the caller. */
5415 error = SYSCTL_IN(req, &task_port_name, sizeof(mach_port_name_t));
5416
5417 if (error != 0) {
5418 goto sysctl_get_vmobject_list_exit;
5419 }
5420
5421 task = port_name_to_task_read(task_port_name);
5422 if (task == TASK_NULL) {
5423 error = ESRCH;
5424 goto sysctl_get_vmobject_list_exit;
5425 }
5426
5427 bool corpse = is_corpsetask(task);
5428
5429 /* get the current size */
5430 size_t max_size;
5431 task_get_owned_vmobjects(task, 0, NULL, &max_size, &entries);
5432
5433 if (buffer_size && (buffer_size < sizeof(*buffer) + sizeof(vm_object_query_data_t))) {
5434 error = ENOMEM;
5435 goto sysctl_get_vmobject_list_deallocate_and_exit;
5436 }
5437
5438 if (corpse == false) {
5439 /* copy the vmobjects and vmobject data out of the task */
5440 if (buffer_size == 0) {
5441 output_size = max_size;
5442 } else {
5443 buffer_size = (buffer_size > max_size) ? max_size : buffer_size;
5444 buffer = (struct _vmobject_list_output_ *)kalloc_data(buffer_size, Z_WAITOK);
5445
5446 if (!buffer) {
5447 error = ENOMEM;
5448 goto sysctl_get_vmobject_list_deallocate_and_exit;
5449 }
5450
5451 task_get_owned_vmobjects(task, buffer_size, buffer, &output_size, &entries);
5452 }
5453
5454 /* req->oldptr should be USER_ADDR_NULL if buffer == NULL and return the current size */
5455 /* otherwise copy buffer to oldptr and return the bytes copied */
5456 error = SYSCTL_OUT(req, (char *)buffer, output_size);
5457 } else {
5458 vmobject_list_output_t list;
5459
5460 task_get_corpse_vmobject_list(task, &list, &max_size);
5461 assert(buffer == NULL);
5462
5463 /* copy corpse_vmobject_list to output buffer to avoid double copy */
5464 if (buffer_size) {
5465 size_t temp_size;
5466
5467 temp_size = buffer_size > max_size ? max_size : buffer_size;
5468 output_size = temp_size - sizeof(*buffer);
5469 /* whole multiple of vm_object_query_data_t */
5470 output_size = (output_size / sizeof(vm_object_query_data_t)) * sizeof(vm_object_query_data_t) + sizeof(*buffer);
5471 buffer = list;
5472 } else {
5473 output_size = max_size;
5474 }
5475
5476 /* req->oldptr should be USER_ADDR_NULL if buffer == NULL and return the current size */
5477 /* otherwise copy buffer to oldptr and return the bytes copied */
5478 error = SYSCTL_OUT(req, (char*)buffer, output_size);
5479 buffer = NULL;
5480 }
5481
5482 sysctl_get_vmobject_list_deallocate_and_exit:
5483 task_deallocate(task);
5484
5485 sysctl_get_vmobject_list_exit:
5486 if (buffer) {
5487 kfree_data(buffer, buffer_size);
5488 }
5489
5490 return error;
5491 }
5492
5493 SYSCTL_PROC(_vm, OID_AUTO, get_owned_vmobjects,
5494 CTLTYPE_OPAQUE | CTLFLAG_RD | CTLFLAG_WR | CTLFLAG_MASKED | CTLFLAG_KERN | CTLFLAG_LOCKED | CTLFLAG_ANYBODY,
5495 0, 0, sysctl_get_owned_vmobjects, "A", "get owned vmobjects in task");
5496
5497 extern uint64_t num_static_scalable_counters;
5498 SYSCTL_QUAD(_kern, OID_AUTO, num_static_scalable_counters, CTLFLAG_RD | CTLFLAG_LOCKED, &num_static_scalable_counters, "");
5499
5500 uuid_string_t trial_treatment_id;
5501 uuid_string_t trial_experiment_id;
5502 int trial_deployment_id = -1;
5503
5504 SYSCTL_STRING(_kern, OID_AUTO, trial_treatment_id, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_ANYBODY | CTLFLAG_EXPERIMENT, trial_treatment_id, sizeof(trial_treatment_id), "");
5505 SYSCTL_STRING(_kern, OID_AUTO, trial_experiment_id, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_ANYBODY | CTLFLAG_EXPERIMENT, trial_experiment_id, sizeof(trial_experiment_id), "");
5506 SYSCTL_INT(_kern, OID_AUTO, trial_deployment_id, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_ANYBODY | CTLFLAG_EXPERIMENT, &trial_deployment_id, 0, "");
5507
5508 #if (DEVELOPMENT || DEBUG)
5509 /* For unit testing setting factors & limits. */
5510 unsigned int testing_experiment_factor;
5511 EXPERIMENT_FACTOR_UINT(_kern, testing_experiment_factor, &testing_experiment_factor, 5, 10, "");
5512
5513 extern int exception_log_max_pid;
5514 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");
5515 #endif /* (DEVELOPMENT || DEBUG) */
5516