1 /*
2 * Copyright (c) 2000-2024 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 <sys/ubc.h> /* mach_to_bsd_errno */
108
109 #include <stdatomic.h>
110 #include <uuid/uuid.h>
111
112 #include <security/audit/audit.h>
113 #include <kern/kalloc.h>
114
115 #include <machine/smp.h>
116 #include <machine/atomic.h>
117 #include <machine/config.h>
118 #include <mach/machine.h>
119 #include <mach/mach_host.h>
120 #include <mach/mach_types.h>
121 #include <mach/processor_info.h>
122 #include <mach/vm_param.h>
123 #include <kern/debug.h>
124 #include <kern/mach_param.h>
125 #include <kern/task.h>
126 #include <kern/thread.h>
127 #include <kern/thread_group.h>
128 #include <kern/processor.h>
129 #include <kern/cpu_number.h>
130 #include <kern/sched_prim.h>
131 #include <kern/workload_config.h>
132 #include <kern/iotrace.h>
133 #include <vm/vm_kern_xnu.h>
134 #include <vm/vm_map_xnu.h>
135 #include <mach/host_info.h>
136 #include <mach/exclaves.h>
137 #include <kern/hvg_hypercall.h>
138 #include <kdp/sk_core.h>
139
140 #if DEVELOPMENT || DEBUG
141 #include <kern/ext_paniclog.h>
142 #endif
143
144 #include <sys/mount_internal.h>
145 #include <sys/kdebug.h>
146 #include <sys/kern_debug.h>
147 #include <sys/kern_sysctl.h>
148 #include <sys/variant_internal.h>
149
150 #include <IOKit/IOPlatformExpert.h>
151 #include <pexpert/pexpert.h>
152
153 #include <machine/machine_routines.h>
154 #include <machine/exec.h>
155
156 #include <nfs/nfs_conf.h>
157
158 #include <vm/vm_protos.h>
159 #include <vm/vm_pageout_xnu.h>
160 #include <vm/vm_compressor_algorithms_xnu.h>
161 #include <vm/vm_compressor_xnu.h>
162 #include <sys/imgsrc.h>
163 #include <kern/timer_call.h>
164 #include <sys/codesign.h>
165 #include <IOKit/IOBSD.h>
166 #if CONFIG_CSR
167 #include <sys/csr.h>
168 #endif
169
170 #if defined(__i386__) || defined(__x86_64__)
171 #include <i386/cpuid.h>
172 #endif
173
174 #if CONFIG_FREEZE
175 #include <sys/kern_memorystatus.h>
176 #endif
177
178 #if KPERF
179 #include <kperf/kperf.h>
180 #endif
181
182 #if HYPERVISOR
183 #include <kern/hv_support.h>
184 #endif
185
186
187 #include <corecrypto/ccsha2.h>
188
189 /*
190 * deliberately setting max requests to really high number
191 * so that runaway settings do not cause MALLOC overflows
192 */
193 #define AIO_MAX_REQUESTS (128 * CONFIG_AIO_MAX)
194
195 extern int aio_max_requests;
196 extern int aio_max_requests_per_process;
197 extern int aio_worker_threads;
198 extern int lowpri_IO_window_msecs;
199 extern int lowpri_IO_delay_msecs;
200 #if DEVELOPMENT || DEBUG
201 extern int nx_enabled;
202 #endif
203 extern int speculative_reads_disabled;
204 extern unsigned int speculative_prefetch_max;
205 extern unsigned int speculative_prefetch_max_iosize;
206 extern unsigned int preheat_max_bytes;
207 extern unsigned int preheat_min_bytes;
208 extern long numvnodes;
209 extern long freevnodes;
210 extern long num_recycledvnodes;
211
212 extern uuid_string_t bootsessionuuid_string;
213
214 extern unsigned int vm_max_delayed_work_limit;
215 extern unsigned int vm_max_batch;
216
217 extern unsigned int vm_page_free_min;
218 extern unsigned int vm_page_free_target;
219 extern unsigned int vm_page_free_reserved;
220 extern unsigned int vm_page_max_speculative_age_q;
221
222 #if (DEVELOPMENT || DEBUG)
223 extern uint32_t vm_page_creation_throttled_hard;
224 extern uint32_t vm_page_creation_throttled_soft;
225 #endif /* DEVELOPMENT || DEBUG */
226
227 #if DEVELOPMENT || DEBUG
228 extern bool bootarg_hide_process_traced;
229 #endif
230
231 /*
232 * Conditionally allow dtrace to see these functions for debugging purposes.
233 */
234 #ifdef STATIC
235 #undef STATIC
236 #endif
237 #if 0
238 #define STATIC
239 #else
240 #define STATIC static
241 #endif
242
243 extern boolean_t mach_timer_coalescing_enabled;
244
245 extern uint64_t timer_deadline_tracking_bin_1, timer_deadline_tracking_bin_2;
246
247 STATIC void
248 fill_user32_eproc(proc_t, struct user32_eproc *__restrict);
249 STATIC void
250 fill_user32_externproc(proc_t, struct user32_extern_proc *__restrict);
251 STATIC void
252 fill_user64_eproc(proc_t, struct user64_eproc *__restrict);
253 STATIC void
254 fill_user64_proc(proc_t, struct user64_kinfo_proc *__restrict);
255 STATIC void
256 fill_user64_externproc(proc_t, struct user64_extern_proc *__restrict);
257 STATIC void
258 fill_user32_proc(proc_t, struct user32_kinfo_proc *__restrict);
259
260 #if CONFIG_NETBOOT
261 extern int
262 netboot_root(void);
263 #endif
264 int
265 sysctl_procargs(int *name, u_int namelen, user_addr_t where,
266 size_t *sizep, proc_t cur_proc);
267 STATIC int
268 sysctl_procargsx(int *name, u_int namelen, user_addr_t where, size_t *sizep,
269 proc_t cur_proc, int argc_yes);
270 int
271 sysctl_struct(user_addr_t oldp, size_t *oldlenp, user_addr_t newp,
272 size_t newlen, void *sp, int len);
273
274 STATIC int sysdoproc_filt_KERN_PROC_PID(proc_t p, void * arg);
275 STATIC int sysdoproc_filt_KERN_PROC_PGRP(proc_t p, void * arg);
276 STATIC int sysdoproc_filt_KERN_PROC_TTY(proc_t p, void * arg);
277 STATIC int sysdoproc_filt_KERN_PROC_UID(proc_t p, void * arg);
278 STATIC int sysdoproc_filt_KERN_PROC_RUID(proc_t p, void * arg);
279 int sysdoproc_callback(proc_t p, void *arg);
280
281 #if CONFIG_THREAD_GROUPS && (DEVELOPMENT || DEBUG)
282 STATIC int sysctl_get_thread_group_id SYSCTL_HANDLER_ARGS;
283 #endif
284
285 /* forward declarations for non-static STATIC */
286 STATIC void fill_loadavg64(struct loadavg *la, struct user64_loadavg *la64);
287 STATIC void fill_loadavg32(struct loadavg *la, struct user32_loadavg *la32);
288 STATIC int sysctl_handle_kern_threadname(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
289 STATIC int sysctl_sched_stats(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
290 STATIC int sysctl_sched_stats_enable(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
291 #if COUNT_SYSCALLS
292 STATIC int sysctl_docountsyscalls SYSCTL_HANDLER_ARGS;
293 #endif /* COUNT_SYSCALLS */
294 #if defined(XNU_TARGET_OS_OSX)
295 STATIC int sysctl_doprocargs SYSCTL_HANDLER_ARGS;
296 #endif /* defined(XNU_TARGET_OS_OSX) */
297 STATIC int sysctl_doprocargs2 SYSCTL_HANDLER_ARGS;
298 STATIC int sysctl_prochandle SYSCTL_HANDLER_ARGS;
299 STATIC int sysctl_aiomax(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
300 STATIC int sysctl_aioprocmax(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
301 STATIC int sysctl_aiothreads(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
302 STATIC int sysctl_maxproc(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
303 STATIC int sysctl_osversion(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
304 STATIC int sysctl_sysctl_bootargs(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
305 STATIC int sysctl_maxvnodes(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
306 STATIC int sysctl_securelvl(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
307 STATIC int sysctl_domainname(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
308 STATIC int sysctl_hostname(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
309 STATIC int sysctl_procname(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
310 STATIC int sysctl_boottime(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
311 STATIC int sysctl_bootuuid(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
312 STATIC int sysctl_symfile(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
313 #if CONFIG_NETBOOT
314 STATIC int sysctl_netboot(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
315 #endif
316 #ifdef CONFIG_IMGSRC_ACCESS
317 STATIC int sysctl_imgsrcdev(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
318 #endif
319 STATIC int sysctl_usrstack(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
320 STATIC int sysctl_usrstack64(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
321 #if CONFIG_COREDUMP
322 STATIC int sysctl_coredump(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
323 STATIC int sysctl_suid_coredump(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
324 #endif
325 STATIC int sysctl_delayterm(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
326 STATIC int sysctl_rage_vnode(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
327 STATIC int sysctl_kern_check_openevt(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
328 #if DEVELOPMENT || DEBUG
329 STATIC int sysctl_nx(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
330 #endif
331 STATIC int sysctl_loadavg(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
332 STATIC int sysctl_vm_toggle_address_reuse(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
333 STATIC int sysctl_swapusage(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
334 STATIC int fetch_process_cputype( proc_t cur_proc, int *name, u_int namelen, cpu_type_t *cputype);
335 STATIC int sysctl_sysctl_native(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
336 STATIC int sysctl_sysctl_cputype(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
337 STATIC int sysctl_safeboot(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
338 STATIC int sysctl_singleuser(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
339 STATIC int sysctl_minimalboot(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
340 STATIC int sysctl_slide(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
341
342 #ifdef CONFIG_XNUPOST
343 #include <tests/xnupost.h>
344
345 STATIC int sysctl_debug_test_oslog_ctl(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
346 STATIC int sysctl_debug_test_stackshot_mutex_owner(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
347 STATIC int sysctl_debug_test_stackshot_rwlck_owner(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
348 #endif
349
350 extern void IORegistrySetOSBuildVersion(char * build_version);
351 extern int IOParseWorkloadConfig(workload_config_ctx_t *ctx, const char * buffer, size_t size);
352 extern int IOUnparseWorkloadConfig(char *buffer, size_t *size);
353
354 STATIC void
fill_loadavg64(struct loadavg * la,struct user64_loadavg * la64)355 fill_loadavg64(struct loadavg *la, struct user64_loadavg *la64)
356 {
357 la64->ldavg[0] = la->ldavg[0];
358 la64->ldavg[1] = la->ldavg[1];
359 la64->ldavg[2] = la->ldavg[2];
360 la64->fscale = (user64_long_t)la->fscale;
361 }
362
363 STATIC void
fill_loadavg32(struct loadavg * la,struct user32_loadavg * la32)364 fill_loadavg32(struct loadavg *la, struct user32_loadavg *la32)
365 {
366 la32->ldavg[0] = la->ldavg[0];
367 la32->ldavg[1] = la->ldavg[1];
368 la32->ldavg[2] = la->ldavg[2];
369 la32->fscale = (user32_long_t)la->fscale;
370 }
371
372 #if COUNT_SYSCALLS
373 extern int do_count_syscalls;
374 #endif
375
376 #ifdef INSECURE
377 int securelevel = -1;
378 #else
379 int securelevel;
380 #endif
381
382 STATIC int
sysctl_handle_kern_threadname(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)383 sysctl_handle_kern_threadname( __unused struct sysctl_oid *oidp, __unused void *arg1,
384 __unused int arg2, struct sysctl_req *req)
385 {
386 int error;
387 struct uthread *ut = current_uthread();
388 user_addr_t oldp = 0, newp = 0;
389 size_t *oldlenp = NULL;
390 size_t newlen = 0;
391
392 oldp = req->oldptr;
393 oldlenp = &(req->oldlen);
394 newp = req->newptr;
395 newlen = req->newlen;
396
397 /* We want the current length, and maybe the string itself */
398 if (oldlenp) {
399 /* if we have no thread name yet tell'em we want MAXTHREADNAMESIZE - 1 */
400 size_t currlen = MAXTHREADNAMESIZE - 1;
401
402 if (ut->pth_name) {
403 /* use length of current thread name */
404 currlen = strlen(ut->pth_name);
405 }
406 if (oldp) {
407 if (*oldlenp < currlen) {
408 return ENOMEM;
409 }
410 /* NOTE - we do not copy the NULL terminator */
411 if (ut->pth_name) {
412 error = copyout(ut->pth_name, oldp, currlen);
413 if (error) {
414 return error;
415 }
416 }
417 }
418 /* return length of thread name minus NULL terminator (just like strlen) */
419 req->oldidx = currlen;
420 }
421
422 /* We want to set the name to something */
423 if (newp) {
424 if (newlen > (MAXTHREADNAMESIZE - 1)) {
425 return ENAMETOOLONG;
426 }
427 if (!ut->pth_name) {
428 char *tmp_pth_name = (char *)kalloc_data(MAXTHREADNAMESIZE,
429 Z_WAITOK | Z_ZERO);
430 if (!tmp_pth_name) {
431 return ENOMEM;
432 }
433 if (!OSCompareAndSwapPtr(NULL, tmp_pth_name, &ut->pth_name)) {
434 kfree_data(tmp_pth_name, MAXTHREADNAMESIZE);
435 return EBUSY;
436 }
437 } else {
438 kernel_debug_string_simple(TRACE_STRING_THREADNAME_PREV, ut->pth_name);
439 bzero(ut->pth_name, MAXTHREADNAMESIZE);
440 }
441 error = copyin(newp, ut->pth_name, newlen);
442 if (error) {
443 return error;
444 }
445
446 kernel_debug_string_simple(TRACE_STRING_THREADNAME, ut->pth_name);
447 }
448
449 return 0;
450 }
451
452 SYSCTL_PROC(_kern, KERN_THREADNAME, threadname, CTLFLAG_ANYBODY | CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_LOCKED, 0, 0, sysctl_handle_kern_threadname, "A", "");
453
454 #define WORKLOAD_CONFIG_MAX_SIZE (128 * 1024 * 1024)
455
456 /* Called locked - sysctl defined without CTLFLAG_LOCKED. */
457 static int
458 sysctl_workload_config SYSCTL_HANDLER_ARGS
459 {
460 #pragma unused(arg1, arg2)
461
462 char *plist_blob = NULL;
463 kern_return_t ret = KERN_FAILURE;
464 int error = -1;
465
466 /* Only allow reading of workload config on non-RELEASE kernels. */
467 #if DEVELOPMENT || DEBUG
468
469 const size_t buf_size = req->oldlen;
470
471 if (!req->oldptr) {
472 /* Just looking for the size to allocate. */
473 size_t size = 0;
474 ret = IOUnparseWorkloadConfig(NULL, &size);
475 if (ret != KERN_SUCCESS) {
476 return ENOMEM;
477 }
478
479 error = SYSCTL_OUT(req, NULL, size);
480 if (error) {
481 return error;
482 }
483 } else {
484 if (buf_size > (WORKLOAD_CONFIG_MAX_SIZE - 1) ||
485 buf_size == 0) {
486 return EINVAL;
487 }
488
489 plist_blob = kalloc_data(buf_size, Z_WAITOK | Z_ZERO);
490 if (!plist_blob) {
491 return ENOMEM;
492 }
493
494 size_t size = buf_size;
495 ret = IOUnparseWorkloadConfig(plist_blob, &size);
496 if (ret != KERN_SUCCESS) {
497 kfree_data(plist_blob, buf_size);
498 return ENOMEM;
499 }
500
501 error = SYSCTL_OUT(req, plist_blob, MIN(buf_size, size));
502
503 /* If the buffer was too small to fit the entire config. */
504 if (buf_size < size) {
505 error = ENOMEM;
506 }
507
508 kfree_data(plist_blob, buf_size);
509 if (error) {
510 return error;
511 }
512 }
513 #endif /* DEVELOPMENT || DEBUG */
514
515 if (req->newptr) {
516 size_t newlen = req->newlen;
517 if (newlen > (WORKLOAD_CONFIG_MAX_SIZE - 1)) {
518 return EINVAL;
519 }
520
521
522 workload_config_ctx_t *ctx = NULL;
523 /*
524 * Only allow workload_config_boot to be loaded once at boot by launchd.
525 */
526 if (current_proc() == initproc &&
527 !workload_config_initialized(&workload_config_boot)) {
528 ctx = &workload_config_boot;
529 } else {
530 #if DEVELOPMENT || DEBUG
531 /*
532 * Use the devel config context otherwise. If a devel config has been
533 * initialized it will be used for lookups in place of the boot config.
534 */
535 ctx = &workload_config_devel;
536 if (workload_config_initialized(ctx)) {
537 workload_config_free(ctx);
538 }
539
540 /* The devel context can be explicitly cleared by an empty string. */
541 if (newlen == 1) {
542 return 0;
543 }
544 #else
545 return EINVAL;
546 #endif
547 }
548
549 plist_blob = kalloc_data(newlen + 1, Z_WAITOK | Z_ZERO);
550 if (!plist_blob) {
551 return ENOMEM;
552 }
553 error = copyin(req->newptr, plist_blob, newlen);
554 if (error) {
555 kfree_data(plist_blob, newlen + 1);
556 return error;
557 }
558 plist_blob[newlen] = '\0';
559 ret = IOParseWorkloadConfig(ctx, plist_blob, newlen + 1);
560
561 kfree_data(plist_blob, newlen + 1);
562 return ret == KERN_SUCCESS ? 0 : EINVAL;
563 }
564
565 return 0;
566 }
567
568 SYSCTL_PROC(_kern, OID_AUTO, workload_config, CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_MASKED,
569 0, 0, sysctl_workload_config, "A", "global workgroup configuration plist load/unload");
570
571 #define BSD_HOST 1
572 STATIC int
sysctl_sched_stats(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)573 sysctl_sched_stats(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
574 {
575 host_basic_info_data_t hinfo;
576 kern_return_t kret;
577 uint32_t size;
578 uint32_t buf_size = 0;
579 int changed;
580 mach_msg_type_number_t count = HOST_BASIC_INFO_COUNT;
581 struct _processor_statistics_np *buf;
582 int error;
583
584 kret = host_info((host_t)BSD_HOST, HOST_BASIC_INFO, (host_info_t)&hinfo, &count);
585 if (kret != KERN_SUCCESS) {
586 return EINVAL;
587 }
588
589 size = sizeof(struct _processor_statistics_np) * (hinfo.logical_cpu_max + 2); /* One for RT Queue, One for Fair Share Queue */
590
591 if (req->oldlen < size) {
592 return EINVAL;
593 }
594
595 buf_size = size;
596 buf = (struct _processor_statistics_np *)kalloc_data(buf_size, Z_ZERO | Z_WAITOK);
597
598 kret = get_sched_statistics(buf, &size);
599 if (kret != KERN_SUCCESS) {
600 error = EINVAL;
601 goto out;
602 }
603
604 error = sysctl_io_opaque(req, buf, size, &changed);
605 if (error) {
606 goto out;
607 }
608
609 if (changed) {
610 panic("Sched info changed?!");
611 }
612 out:
613 kfree_data(buf, buf_size);
614 return error;
615 }
616
617 SYSCTL_PROC(_kern, OID_AUTO, sched_stats, CTLFLAG_LOCKED, 0, 0, sysctl_sched_stats, "-", "");
618
619 STATIC int
sysctl_sched_stats_enable(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,__unused struct sysctl_req * req)620 sysctl_sched_stats_enable(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, __unused struct sysctl_req *req)
621 {
622 boolean_t active;
623 int res;
624
625 if (req->newlen != sizeof(active)) {
626 return EINVAL;
627 }
628
629 res = copyin(req->newptr, &active, sizeof(active));
630 if (res != 0) {
631 return res;
632 }
633
634 return set_sched_stats_active(active);
635 }
636
637 SYSCTL_PROC(_kern, OID_AUTO, sched_stats_enable, CTLFLAG_LOCKED | CTLFLAG_WR, 0, 0, sysctl_sched_stats_enable, "-", "");
638
639 extern uint32_t sched_debug_flags;
640 SYSCTL_INT(_debug, OID_AUTO, sched, CTLFLAG_RW | CTLFLAG_LOCKED, &sched_debug_flags, 0, "scheduler debug");
641
642 #if (DEBUG || DEVELOPMENT)
643 extern boolean_t doprnt_hide_pointers;
644 SYSCTL_INT(_debug, OID_AUTO, hide_kernel_pointers, CTLFLAG_RW | CTLFLAG_LOCKED, &doprnt_hide_pointers, 0, "hide kernel pointers from log");
645 #endif
646
647
648 extern int get_kernel_symfile(proc_t, char **);
649
650 #if COUNT_SYSCALLS
651 #define KERN_COUNT_SYSCALLS (KERN_OSTYPE + 1000)
652
653 extern const unsigned int nsysent;
654 extern int syscalls_log[];
655 extern const char *syscallnames[];
656
657 STATIC int
658 sysctl_docountsyscalls SYSCTL_HANDLER_ARGS
659 {
660 __unused int cmd = oidp->oid_arg2; /* subcommand*/
661 __unused int *name = arg1; /* oid element argument vector */
662 __unused int namelen = arg2; /* number of oid element arguments */
663 int error, changed;
664
665 int tmp;
666
667 /* valid values passed in:
668 * = 0 means don't keep called counts for each bsd syscall
669 * > 0 means keep called counts for each bsd syscall
670 * = 2 means dump current counts to the system log
671 * = 3 means reset all counts
672 * for example, to dump current counts:
673 * sysctl -w kern.count_calls=2
674 */
675 error = sysctl_io_number(req, do_count_syscalls,
676 sizeof(do_count_syscalls), &tmp, &changed);
677
678 if (error != 0 || !changed) {
679 return error;
680 }
681
682 if (tmp == 1) {
683 do_count_syscalls = 1;
684 } else if (tmp == 0 || tmp == 2 || tmp == 3) {
685 for (int i = 0; i < nsysent; i++) {
686 if (syscalls_log[i] != 0) {
687 if (tmp == 2) {
688 printf("%d calls - name %s \n", syscalls_log[i], syscallnames[i]);
689 } else {
690 syscalls_log[i] = 0;
691 }
692 }
693 }
694 do_count_syscalls = (tmp != 0);
695 }
696
697 return error;
698 }
699 SYSCTL_PROC(_kern, KERN_COUNT_SYSCALLS, count_syscalls, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
700 0, /* Pointer argument (arg1) */
701 0, /* Integer argument (arg2) */
702 sysctl_docountsyscalls, /* Handler function */
703 NULL, /* Data pointer */
704 "");
705 #endif /* COUNT_SYSCALLS */
706
707 /*
708 * The following sysctl_* functions should not be used
709 * any more, as they can only cope with callers in
710 * user mode: Use new-style
711 * sysctl_io_number()
712 * sysctl_io_string()
713 * sysctl_io_opaque()
714 * instead.
715 */
716
717 STATIC int
sysdoproc_filt_KERN_PROC_PID(proc_t p,void * arg)718 sysdoproc_filt_KERN_PROC_PID(proc_t p, void * arg)
719 {
720 if (proc_getpid(p) != (pid_t)*(int*)arg) {
721 return 0;
722 } else {
723 return 1;
724 }
725 }
726
727 STATIC int
sysdoproc_filt_KERN_PROC_PGRP(proc_t p,void * arg)728 sysdoproc_filt_KERN_PROC_PGRP(proc_t p, void * arg)
729 {
730 if (p->p_pgrpid != (pid_t)*(int*)arg) {
731 return 0;
732 } else {
733 return 1;
734 }
735 }
736
737 STATIC int
sysdoproc_filt_KERN_PROC_TTY(proc_t p,void * arg)738 sysdoproc_filt_KERN_PROC_TTY(proc_t p, void * arg)
739 {
740 struct pgrp *pg;
741 dev_t dev = NODEV;
742
743 if ((p->p_flag & P_CONTROLT) && (pg = proc_pgrp(p, NULL)) != PGRP_NULL) {
744 dev = os_atomic_load(&pg->pg_session->s_ttydev, relaxed);
745 pgrp_rele(pg);
746 }
747
748 return dev != NODEV && dev == (dev_t)*(int *)arg;
749 }
750
751 STATIC int
sysdoproc_filt_KERN_PROC_UID(proc_t p,void * arg)752 sysdoproc_filt_KERN_PROC_UID(proc_t p, void * arg)
753 {
754 uid_t uid;
755
756 smr_proc_task_enter();
757 uid = kauth_cred_getuid(proc_ucred_smr(p));
758 smr_proc_task_leave();
759
760 if (uid != (uid_t)*(int*)arg) {
761 return 0;
762 } else {
763 return 1;
764 }
765 }
766
767
768 STATIC int
sysdoproc_filt_KERN_PROC_RUID(proc_t p,void * arg)769 sysdoproc_filt_KERN_PROC_RUID(proc_t p, void * arg)
770 {
771 uid_t ruid;
772
773 smr_proc_task_enter();
774 ruid = kauth_cred_getruid(proc_ucred_smr(p));
775 smr_proc_task_leave();
776
777 if (ruid != (uid_t)*(int*)arg) {
778 return 0;
779 } else {
780 return 1;
781 }
782 }
783
784 /*
785 * try over estimating by 5 procs
786 */
787 #define KERN_PROCSLOP (5 * sizeof(struct kinfo_proc))
788 struct sysdoproc_args {
789 size_t buflen;
790 void *kprocp;
791 boolean_t is_64_bit;
792 user_addr_t dp;
793 size_t needed;
794 unsigned int sizeof_kproc;
795 int *errorp;
796 int uidcheck;
797 int ruidcheck;
798 int ttycheck;
799 int uidval;
800 };
801
802 int
sysdoproc_callback(proc_t p,void * arg)803 sysdoproc_callback(proc_t p, void *arg)
804 {
805 struct sysdoproc_args *args = arg;
806
807 if (args->buflen >= args->sizeof_kproc) {
808 if ((args->ruidcheck != 0) && (sysdoproc_filt_KERN_PROC_RUID(p, &args->uidval) == 0)) {
809 return PROC_RETURNED;
810 }
811 if ((args->uidcheck != 0) && (sysdoproc_filt_KERN_PROC_UID(p, &args->uidval) == 0)) {
812 return PROC_RETURNED;
813 }
814 if ((args->ttycheck != 0) && (sysdoproc_filt_KERN_PROC_TTY(p, &args->uidval) == 0)) {
815 return PROC_RETURNED;
816 }
817
818 bzero(args->kprocp, args->sizeof_kproc);
819 if (args->is_64_bit) {
820 fill_user64_proc(p, args->kprocp);
821 } else {
822 fill_user32_proc(p, args->kprocp);
823 }
824 int error = copyout(args->kprocp, args->dp, args->sizeof_kproc);
825 if (error) {
826 *args->errorp = error;
827 return PROC_RETURNED_DONE;
828 }
829 args->dp += args->sizeof_kproc;
830 args->buflen -= args->sizeof_kproc;
831 }
832 args->needed += args->sizeof_kproc;
833 return PROC_RETURNED;
834 }
835
836 SYSCTL_NODE(_kern, KERN_PROC, proc, CTLFLAG_RD | CTLFLAG_LOCKED, 0, "");
837 STATIC int
838 sysctl_prochandle SYSCTL_HANDLER_ARGS
839 {
840 int cmd = oidp->oid_arg2; /* subcommand for multiple nodes */
841 int *name = arg1; /* oid element argument vector */
842 int namelen = arg2; /* number of oid element arguments */
843 user_addr_t where = req->oldptr;/* user buffer copy out address */
844
845 user_addr_t dp = where;
846 size_t needed = 0;
847 size_t buflen = where != USER_ADDR_NULL ? req->oldlen : 0;
848 int error = 0;
849 boolean_t is_64_bit = proc_is64bit(current_proc());
850 struct user32_kinfo_proc user32_kproc;
851 struct user64_kinfo_proc user_kproc;
852 int sizeof_kproc;
853 void *kprocp;
854 int (*filterfn)(proc_t, void *) = 0;
855 struct sysdoproc_args args;
856 int uidcheck = 0;
857 int ruidcheck = 0;
858 int ttycheck = 0;
859
860 if (namelen != 1 && !(namelen == 0 && cmd == KERN_PROC_ALL)) {
861 return EINVAL;
862 }
863
864 if (is_64_bit) {
865 sizeof_kproc = sizeof(user_kproc);
866 kprocp = &user_kproc;
867 } else {
868 sizeof_kproc = sizeof(user32_kproc);
869 kprocp = &user32_kproc;
870 }
871
872 switch (cmd) {
873 case KERN_PROC_PID:
874 filterfn = sysdoproc_filt_KERN_PROC_PID;
875 break;
876
877 case KERN_PROC_PGRP:
878 filterfn = sysdoproc_filt_KERN_PROC_PGRP;
879 break;
880
881 case KERN_PROC_TTY:
882 ttycheck = 1;
883 break;
884
885 case KERN_PROC_UID:
886 uidcheck = 1;
887 break;
888
889 case KERN_PROC_RUID:
890 ruidcheck = 1;
891 break;
892
893 case KERN_PROC_ALL:
894 break;
895
896 default:
897 /* must be kern.proc.<unknown> */
898 return ENOTSUP;
899 }
900
901 error = 0;
902 args.buflen = buflen;
903 args.kprocp = kprocp;
904 args.is_64_bit = is_64_bit;
905 args.dp = dp;
906 args.needed = needed;
907 args.errorp = &error;
908 args.uidcheck = uidcheck;
909 args.ruidcheck = ruidcheck;
910 args.ttycheck = ttycheck;
911 args.sizeof_kproc = sizeof_kproc;
912 if (namelen) {
913 args.uidval = name[0];
914 }
915
916 proc_iterate((PROC_ALLPROCLIST | PROC_ZOMBPROCLIST),
917 sysdoproc_callback, &args, filterfn, name);
918
919 if (error) {
920 return error;
921 }
922
923 dp = args.dp;
924 needed = args.needed;
925
926 if (where != USER_ADDR_NULL) {
927 req->oldlen = dp - where;
928 if (needed > req->oldlen) {
929 return ENOMEM;
930 }
931 } else {
932 needed += KERN_PROCSLOP;
933 req->oldlen = needed;
934 }
935 /* adjust index so we return the right required/consumed amount */
936 req->oldidx += req->oldlen;
937 return 0;
938 }
939
940
941 /*
942 * We specify the subcommand code for multiple nodes as the 'req->arg2' value
943 * in the sysctl declaration itself, which comes into the handler function
944 * as 'oidp->oid_arg2'.
945 *
946 * For these particular sysctls, since they have well known OIDs, we could
947 * have just obtained it from the '((int *)arg1)[0]' parameter, but that would
948 * not demonstrate how to handle multiple sysctls that used OID_AUTO instead
949 * of a well known value with a common handler function. This is desirable,
950 * because we want well known values to "go away" at some future date.
951 *
952 * It should be noted that the value of '((int *)arg1)[1]' is used for many
953 * an integer parameter to the subcommand for many of these sysctls; we'd
954 * rather have used '((int *)arg1)[0]' for that, or even better, an element
955 * in a structure passed in as the the 'newp' argument to sysctlbyname(3),
956 * and then use leaf-node permissions enforcement, but that would have
957 * necessitated modifying user space code to correspond to the interface
958 * change, and we are striving for binary backward compatibility here; even
959 * though these are SPI, and not intended for use by user space applications
960 * which are not themselves system tools or libraries, some applications
961 * have erroneously used them.
962 */
963 SYSCTL_PROC(_kern_proc, KERN_PROC_ALL, all, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
964 0, /* Pointer argument (arg1) */
965 KERN_PROC_ALL, /* Integer argument (arg2) */
966 sysctl_prochandle, /* Handler function */
967 NULL, /* Data is size variant on ILP32/LP64 */
968 "");
969 SYSCTL_PROC(_kern_proc, KERN_PROC_PID, pid, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
970 0, /* Pointer argument (arg1) */
971 KERN_PROC_PID, /* Integer argument (arg2) */
972 sysctl_prochandle, /* Handler function */
973 NULL, /* Data is size variant on ILP32/LP64 */
974 "");
975 SYSCTL_PROC(_kern_proc, KERN_PROC_TTY, tty, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
976 0, /* Pointer argument (arg1) */
977 KERN_PROC_TTY, /* Integer argument (arg2) */
978 sysctl_prochandle, /* Handler function */
979 NULL, /* Data is size variant on ILP32/LP64 */
980 "");
981 SYSCTL_PROC(_kern_proc, KERN_PROC_PGRP, pgrp, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
982 0, /* Pointer argument (arg1) */
983 KERN_PROC_PGRP, /* Integer argument (arg2) */
984 sysctl_prochandle, /* Handler function */
985 NULL, /* Data is size variant on ILP32/LP64 */
986 "");
987 SYSCTL_PROC(_kern_proc, KERN_PROC_UID, uid, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
988 0, /* Pointer argument (arg1) */
989 KERN_PROC_UID, /* Integer argument (arg2) */
990 sysctl_prochandle, /* Handler function */
991 NULL, /* Data is size variant on ILP32/LP64 */
992 "");
993 SYSCTL_PROC(_kern_proc, KERN_PROC_RUID, ruid, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
994 0, /* Pointer argument (arg1) */
995 KERN_PROC_RUID, /* Integer argument (arg2) */
996 sysctl_prochandle, /* Handler function */
997 NULL, /* Data is size variant on ILP32/LP64 */
998 "");
999 SYSCTL_PROC(_kern_proc, KERN_PROC_LCID, lcid, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
1000 0, /* Pointer argument (arg1) */
1001 KERN_PROC_LCID, /* Integer argument (arg2) */
1002 sysctl_prochandle, /* Handler function */
1003 NULL, /* Data is size variant on ILP32/LP64 */
1004 "");
1005
1006
1007 /*
1008 * Fill in non-zero fields of an eproc structure for the specified process.
1009 */
1010 STATIC void
fill_user32_eproc(proc_t p,struct user32_eproc * __restrict ep)1011 fill_user32_eproc(proc_t p, struct user32_eproc *__restrict ep)
1012 {
1013 struct pgrp *pg;
1014 struct session *sessp;
1015 kauth_cred_t my_cred;
1016
1017 pg = proc_pgrp(p, &sessp);
1018
1019 if (pg != PGRP_NULL) {
1020 ep->e_pgid = p->p_pgrpid;
1021 ep->e_jobc = pg->pg_jobc;
1022 if (sessp->s_ttyvp) {
1023 ep->e_flag = EPROC_CTTY;
1024 }
1025 }
1026
1027 ep->e_ppid = p->p_ppid;
1028
1029 smr_proc_task_enter();
1030 my_cred = proc_ucred_smr(p);
1031
1032 /* A fake historical pcred */
1033 ep->e_pcred.p_ruid = kauth_cred_getruid(my_cred);
1034 ep->e_pcred.p_svuid = kauth_cred_getsvuid(my_cred);
1035 ep->e_pcred.p_rgid = kauth_cred_getrgid(my_cred);
1036 ep->e_pcred.p_svgid = kauth_cred_getsvgid(my_cred);
1037
1038 /* A fake historical *kauth_cred_t */
1039 unsigned long refcnt = os_atomic_load(&my_cred->cr_ref, relaxed);
1040 ep->e_ucred.cr_ref = (uint32_t)MIN(refcnt, UINT32_MAX);
1041 ep->e_ucred.cr_uid = kauth_cred_getuid(my_cred);
1042 ep->e_ucred.cr_ngroups = (short)posix_cred_get(my_cred)->cr_ngroups;
1043 bcopy(posix_cred_get(my_cred)->cr_groups,
1044 ep->e_ucred.cr_groups, NGROUPS * sizeof(gid_t));
1045
1046 my_cred = NOCRED;
1047 smr_proc_task_leave();
1048
1049 ep->e_tdev = NODEV;
1050 if (pg != PGRP_NULL) {
1051 if (p->p_flag & P_CONTROLT) {
1052 session_lock(sessp);
1053 ep->e_tdev = os_atomic_load(&sessp->s_ttydev, relaxed);
1054 ep->e_tpgid = sessp->s_ttypgrpid;
1055 session_unlock(sessp);
1056 }
1057 if (SESS_LEADER(p, sessp)) {
1058 ep->e_flag |= EPROC_SLEADER;
1059 }
1060 pgrp_rele(pg);
1061 }
1062 }
1063
1064 /*
1065 * Fill in non-zero fields of an LP64 eproc structure for the specified process.
1066 */
1067 STATIC void
fill_user64_eproc(proc_t p,struct user64_eproc * __restrict ep)1068 fill_user64_eproc(proc_t p, struct user64_eproc *__restrict ep)
1069 {
1070 struct pgrp *pg;
1071 struct session *sessp;
1072 kauth_cred_t my_cred;
1073
1074 pg = proc_pgrp(p, &sessp);
1075
1076 if (pg != PGRP_NULL) {
1077 ep->e_pgid = p->p_pgrpid;
1078 ep->e_jobc = pg->pg_jobc;
1079 if (sessp->s_ttyvp) {
1080 ep->e_flag = EPROC_CTTY;
1081 }
1082 }
1083
1084 ep->e_ppid = p->p_ppid;
1085
1086 smr_proc_task_enter();
1087 my_cred = proc_ucred_smr(p);
1088
1089 /* A fake historical pcred */
1090 ep->e_pcred.p_ruid = kauth_cred_getruid(my_cred);
1091 ep->e_pcred.p_svuid = kauth_cred_getsvuid(my_cred);
1092 ep->e_pcred.p_rgid = kauth_cred_getrgid(my_cred);
1093 ep->e_pcred.p_svgid = kauth_cred_getsvgid(my_cred);
1094
1095 /* A fake historical *kauth_cred_t */
1096 unsigned long refcnt = os_atomic_load(&my_cred->cr_ref, relaxed);
1097 ep->e_ucred.cr_ref = (uint32_t)MIN(refcnt, UINT32_MAX);
1098 ep->e_ucred.cr_uid = kauth_cred_getuid(my_cred);
1099 ep->e_ucred.cr_ngroups = (short)posix_cred_get(my_cred)->cr_ngroups;
1100 bcopy(posix_cred_get(my_cred)->cr_groups,
1101 ep->e_ucred.cr_groups, NGROUPS * sizeof(gid_t));
1102
1103 my_cred = NOCRED;
1104 smr_proc_task_leave();
1105
1106 ep->e_tdev = NODEV;
1107 if (pg != PGRP_NULL) {
1108 if (p->p_flag & P_CONTROLT) {
1109 session_lock(sessp);
1110 ep->e_tdev = os_atomic_load(&sessp->s_ttydev, relaxed);
1111 ep->e_tpgid = sessp->s_ttypgrpid;
1112 session_unlock(sessp);
1113 }
1114 if (SESS_LEADER(p, sessp)) {
1115 ep->e_flag |= EPROC_SLEADER;
1116 }
1117 pgrp_rele(pg);
1118 }
1119 }
1120
1121 /*
1122 * Fill in an eproc structure for the specified process.
1123 * bzeroed by our caller, so only set non-zero fields.
1124 */
1125 STATIC void
fill_user32_externproc(proc_t p,struct user32_extern_proc * __restrict exp)1126 fill_user32_externproc(proc_t p, struct user32_extern_proc *__restrict exp)
1127 {
1128 exp->p_starttime.tv_sec = (user32_time_t)p->p_start.tv_sec;
1129 exp->p_starttime.tv_usec = p->p_start.tv_usec;
1130 exp->p_flag = p->p_flag;
1131 #if DEVELOPMENT || DEBUG
1132 if (p->p_lflag & P_LTRACED && !bootarg_hide_process_traced) {
1133 #else
1134 if (p->p_lflag & P_LTRACED) {
1135 #endif
1136 exp->p_flag |= P_TRACED;
1137 }
1138 if (p->p_lflag & P_LPPWAIT) {
1139 exp->p_flag |= P_PPWAIT;
1140 }
1141 if (p->p_lflag & P_LEXIT) {
1142 exp->p_flag |= P_WEXIT;
1143 }
1144 exp->p_stat = p->p_stat;
1145 exp->p_pid = proc_getpid(p);
1146 #if DEVELOPMENT || DEBUG
1147 if (bootarg_hide_process_traced) {
1148 exp->p_oppid = 0;
1149 } else
1150 #endif
1151 {
1152 exp->p_oppid = p->p_oppid;
1153 }
1154 /* Mach related */
1155 exp->p_debugger = p->p_debugger;
1156 exp->sigwait = p->sigwait;
1157 /* scheduling */
1158 #ifdef _PROC_HAS_SCHEDINFO_
1159 exp->p_estcpu = p->p_estcpu;
1160 exp->p_pctcpu = p->p_pctcpu;
1161 exp->p_slptime = p->p_slptime;
1162 #endif
1163 exp->p_realtimer.it_interval.tv_sec =
1164 (user32_time_t)p->p_realtimer.it_interval.tv_sec;
1165 exp->p_realtimer.it_interval.tv_usec =
1166 (__int32_t)p->p_realtimer.it_interval.tv_usec;
1167
1168 exp->p_realtimer.it_value.tv_sec =
1169 (user32_time_t)p->p_realtimer.it_value.tv_sec;
1170 exp->p_realtimer.it_value.tv_usec =
1171 (__int32_t)p->p_realtimer.it_value.tv_usec;
1172
1173 exp->p_rtime.tv_sec = (user32_time_t)p->p_rtime.tv_sec;
1174 exp->p_rtime.tv_usec = (__int32_t)p->p_rtime.tv_usec;
1175
1176 exp->p_sigignore = p->p_sigignore;
1177 exp->p_sigcatch = p->p_sigcatch;
1178 exp->p_priority = p->p_priority;
1179 exp->p_nice = p->p_nice;
1180 bcopy(&p->p_comm, &exp->p_comm, MAXCOMLEN);
1181 exp->p_xstat = (u_short)MIN(p->p_xstat, USHRT_MAX);
1182 exp->p_acflag = p->p_acflag;
1183 }
1184
1185 /*
1186 * Fill in an LP64 version of extern_proc structure for the specified process.
1187 */
1188 STATIC void
1189 fill_user64_externproc(proc_t p, struct user64_extern_proc *__restrict exp)
1190 {
1191 exp->p_starttime.tv_sec = p->p_start.tv_sec;
1192 exp->p_starttime.tv_usec = p->p_start.tv_usec;
1193 exp->p_flag = p->p_flag;
1194 #if DEVELOPMENT || DEBUG
1195 if (p->p_lflag & P_LTRACED && !bootarg_hide_process_traced) {
1196 #else
1197 if (p->p_lflag & P_LTRACED) {
1198 #endif
1199 exp->p_flag |= P_TRACED;
1200 }
1201 if (p->p_lflag & P_LPPWAIT) {
1202 exp->p_flag |= P_PPWAIT;
1203 }
1204 if (p->p_lflag & P_LEXIT) {
1205 exp->p_flag |= P_WEXIT;
1206 }
1207 exp->p_stat = p->p_stat;
1208 exp->p_pid = proc_getpid(p);
1209 #if DEVELOPMENT || DEBUG
1210 if (bootarg_hide_process_traced) {
1211 exp->p_oppid = 0;
1212 } else
1213 #endif
1214 {
1215 exp->p_oppid = p->p_oppid;
1216 }
1217 /* Mach related */
1218 exp->p_debugger = p->p_debugger;
1219 exp->sigwait = p->sigwait;
1220 /* scheduling */
1221 #ifdef _PROC_HAS_SCHEDINFO_
1222 exp->p_estcpu = p->p_estcpu;
1223 exp->p_pctcpu = p->p_pctcpu;
1224 exp->p_slptime = p->p_slptime;
1225 #endif
1226 exp->p_realtimer.it_interval.tv_sec = p->p_realtimer.it_interval.tv_sec;
1227 exp->p_realtimer.it_interval.tv_usec = p->p_realtimer.it_interval.tv_usec;
1228
1229 exp->p_realtimer.it_value.tv_sec = p->p_realtimer.it_value.tv_sec;
1230 exp->p_realtimer.it_value.tv_usec = p->p_realtimer.it_value.tv_usec;
1231
1232 exp->p_rtime.tv_sec = p->p_rtime.tv_sec;
1233 exp->p_rtime.tv_usec = p->p_rtime.tv_usec;
1234
1235 exp->p_sigignore = p->p_sigignore;
1236 exp->p_sigcatch = p->p_sigcatch;
1237 exp->p_priority = p->p_priority;
1238 exp->p_nice = p->p_nice;
1239 bcopy(&p->p_comm, &exp->p_comm, MAXCOMLEN);
1240 exp->p_xstat = (u_short)MIN(p->p_xstat, USHRT_MAX);
1241 exp->p_acflag = p->p_acflag;
1242 }
1243
1244 STATIC void
1245 fill_user32_proc(proc_t p, struct user32_kinfo_proc *__restrict kp)
1246 {
1247 /* on a 64 bit kernel, 32 bit users get some truncated information */
1248 fill_user32_externproc(p, &kp->kp_proc);
1249 fill_user32_eproc(p, &kp->kp_eproc);
1250 }
1251
1252 STATIC void
1253 fill_user64_proc(proc_t p, struct user64_kinfo_proc *__restrict kp)
1254 {
1255 fill_user64_externproc(p, &kp->kp_proc);
1256 fill_user64_eproc(p, &kp->kp_eproc);
1257 }
1258
1259 #if defined(XNU_TARGET_OS_OSX)
1260 /*
1261 * Return the top *sizep bytes of the user stack, or the entire area of the
1262 * user stack down through the saved exec_path, whichever is smaller.
1263 */
1264 STATIC int
1265 sysctl_doprocargs SYSCTL_HANDLER_ARGS
1266 {
1267 __unused int cmd = oidp->oid_arg2; /* subcommand*/
1268 int *name = arg1; /* oid element argument vector */
1269 int namelen = arg2; /* number of oid element arguments */
1270 user_addr_t oldp = req->oldptr; /* user buffer copy out address */
1271 size_t *oldlenp = &req->oldlen; /* user buffer copy out size */
1272 // user_addr_t newp = req->newptr; /* user buffer copy in address */
1273 // size_t newlen = req->newlen; /* user buffer copy in size */
1274 int error;
1275
1276 error = sysctl_procargsx( name, namelen, oldp, oldlenp, current_proc(), 0);
1277
1278 /* adjust index so we return the right required/consumed amount */
1279 if (!error) {
1280 req->oldidx += req->oldlen;
1281 }
1282
1283 return error;
1284 }
1285 SYSCTL_PROC(_kern, KERN_PROCARGS, procargs, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
1286 0, /* Pointer argument (arg1) */
1287 0, /* Integer argument (arg2) */
1288 sysctl_doprocargs, /* Handler function */
1289 NULL, /* Data pointer */
1290 "");
1291 #endif /* defined(XNU_TARGET_OS_OSX) */
1292
1293 STATIC int
1294 sysctl_doprocargs2 SYSCTL_HANDLER_ARGS
1295 {
1296 __unused int cmd = oidp->oid_arg2; /* subcommand*/
1297 int *name = arg1; /* oid element argument vector */
1298 int namelen = arg2; /* number of oid element arguments */
1299 user_addr_t oldp = req->oldptr; /* user buffer copy out address */
1300 size_t *oldlenp = &req->oldlen; /* user buffer copy out size */
1301 // user_addr_t newp = req->newptr; /* user buffer copy in address */
1302 // size_t newlen = req->newlen; /* user buffer copy in size */
1303 int error;
1304
1305 error = sysctl_procargsx( name, namelen, oldp, oldlenp, current_proc(), 1);
1306
1307 /* adjust index so we return the right required/consumed amount */
1308 if (!error) {
1309 req->oldidx += req->oldlen;
1310 }
1311
1312 return error;
1313 }
1314 SYSCTL_PROC(_kern, KERN_PROCARGS2, procargs2, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
1315 0, /* Pointer argument (arg1) */
1316 0, /* Integer argument (arg2) */
1317 sysctl_doprocargs2, /* Handler function */
1318 NULL, /* Data pointer */
1319 "");
1320
1321 #define SYSCTL_PROCARGS_READ_ENVVARS_ENTITLEMENT "com.apple.private.read-environment-variables"
1322 STATIC int
1323 sysctl_procargsx(int *name, u_int namelen, user_addr_t where,
1324 size_t *sizep, proc_t cur_proc, int argc_yes)
1325 {
1326 assert(sizep != NULL);
1327 proc_t p = NULL;
1328 size_t buflen = where != USER_ADDR_NULL ? *sizep : 0;
1329 int error = 0;
1330 struct _vm_map *proc_map = NULL;
1331 struct task * task;
1332 vm_map_copy_t tmp = NULL;
1333 user_addr_t arg_addr;
1334 size_t arg_size;
1335 caddr_t data;
1336 size_t argslen = 0;
1337 size_t size = 0;
1338 vm_offset_t copy_start = 0, copy_end;
1339 vm_offset_t smallbuffer_start;
1340 kern_return_t ret;
1341 int pid;
1342 uid_t uid;
1343 int argc = -1;
1344 size_t argvsize;
1345 size_t remaining;
1346 size_t current_arg_index;
1347 size_t current_arg_len;
1348 const char * current_arg;
1349 bool omit_env_vars = true;
1350 user_addr_t user_stack;
1351 vm_map_offset_t effective_page_mask;
1352
1353 if (namelen < 1) {
1354 error = EINVAL;
1355 goto finish;
1356 }
1357
1358 if (argc_yes) {
1359 buflen -= sizeof(int); /* reserve first word to return argc */
1360 }
1361 /* we only care about buflen when where (oldp from sysctl) is not NULL. */
1362 /* when where (oldp from sysctl) is NULL and sizep (oldlenp from sysctl */
1363 /* is not NULL then the caller wants us to return the length needed to */
1364 /* hold the data we would return */
1365 if (where != USER_ADDR_NULL && (buflen <= 0 || buflen > ARG_MAX)) {
1366 error = EINVAL;
1367 goto finish;
1368 }
1369
1370 /*
1371 * Lookup process by pid
1372 */
1373 pid = name[0];
1374 p = proc_find(pid);
1375 if (p == NULL) {
1376 error = EINVAL;
1377 goto finish;
1378 }
1379
1380 /* Allow reading environment variables if any of the following are true:
1381 * - kernel is DEVELOPMENT || DEBUG
1382 * - target process is same as current_proc()
1383 * - target process is not cs_restricted
1384 * - SIP is off
1385 * - caller has an entitlement
1386 */
1387
1388 #if DEVELOPMENT || DEBUG
1389 omit_env_vars = false;
1390 #endif
1391 if (p == current_proc() ||
1392 !cs_restricted(p) ||
1393 #if CONFIG_CSR
1394 csr_check(CSR_ALLOW_UNRESTRICTED_DTRACE) == 0 ||
1395 #endif
1396 IOCurrentTaskHasEntitlement(SYSCTL_PROCARGS_READ_ENVVARS_ENTITLEMENT)
1397 ) {
1398 omit_env_vars = false;
1399 }
1400
1401 /*
1402 * Copy the top N bytes of the stack.
1403 * On all machines we have so far, the stack grows
1404 * downwards.
1405 *
1406 * If the user expects no more than N bytes of
1407 * argument list, use that as a guess for the
1408 * size.
1409 */
1410
1411 if (!p->user_stack) {
1412 error = EINVAL;
1413 goto finish;
1414 }
1415
1416 /* save off argc, argslen, user_stack before releasing the proc */
1417 argc = p->p_argc;
1418 argslen = p->p_argslen;
1419 user_stack = p->user_stack;
1420
1421 /*
1422 * When these sysctls were introduced, the first string in the strings
1423 * section was just the bare path of the executable. However, for security
1424 * reasons we now prefix this string with executable_path= so it can be
1425 * parsed getenv style. To avoid binary compatability issues with exising
1426 * callers of this sysctl, we strip it off here.
1427 * (rdar://problem/13746466)
1428 */
1429 #define EXECUTABLE_KEY "executable_path="
1430 argslen -= strlen(EXECUTABLE_KEY);
1431
1432 if (where == USER_ADDR_NULL && !omit_env_vars) {
1433 /* caller only wants to know length of proc args data.
1434 * If we don't need to omit environment variables, we can skip
1435 * copying the target process stack */
1436 goto calculate_size;
1437 }
1438
1439 smr_proc_task_enter();
1440 uid = kauth_cred_getuid(proc_ucred_smr(p));
1441 smr_proc_task_leave();
1442
1443 if ((uid != kauth_cred_getuid(kauth_cred_get()))
1444 && suser(kauth_cred_get(), &cur_proc->p_acflag)) {
1445 error = EINVAL;
1446 goto finish;
1447 }
1448
1449 /*
1450 * Before we can block (any VM code), make another
1451 * reference to the map to keep it alive. We do
1452 * that by getting a reference on the task itself.
1453 */
1454 task = proc_task(p);
1455 if (task == NULL) {
1456 error = EINVAL;
1457 goto finish;
1458 }
1459
1460 /*
1461 * Once we have a task reference we can convert that into a
1462 * map reference, which we will use in the calls below. The
1463 * task/process may change its map after we take this reference
1464 * (see execve), but the worst that will happen then is a return
1465 * of stale info (which is always a possibility).
1466 */
1467 task_reference(task);
1468 proc_rele(p);
1469 p = NULL;
1470 proc_map = get_task_map_reference(task);
1471 task_deallocate(task);
1472
1473 if (proc_map == NULL) {
1474 error = EINVAL;
1475 goto finish;
1476 }
1477
1478 effective_page_mask = vm_map_page_mask(proc_map);
1479
1480 arg_size = vm_map_round_page(argslen, effective_page_mask);
1481
1482 arg_addr = user_stack - arg_size;
1483
1484 ret = kmem_alloc(kernel_map, ©_start, arg_size,
1485 KMA_DATA | KMA_ZERO, VM_KERN_MEMORY_BSD);
1486 if (ret != KERN_SUCCESS) {
1487 error = ENOMEM;
1488 goto finish;
1489 }
1490
1491 copy_end = copy_start + arg_size;
1492
1493 if (vm_map_copyin(proc_map, (vm_map_address_t)arg_addr,
1494 (vm_map_size_t)arg_size, FALSE, &tmp) != KERN_SUCCESS) {
1495 error = EIO;
1496 goto finish;
1497 }
1498
1499 /*
1500 * Now that we've done the copyin from the process'
1501 * map, we can release the reference to it.
1502 */
1503 vm_map_deallocate(proc_map);
1504 proc_map = NULL;
1505
1506 if (vm_map_copy_overwrite(kernel_map,
1507 (vm_map_address_t)copy_start,
1508 tmp, (vm_map_size_t) arg_size, FALSE) != KERN_SUCCESS) {
1509 error = EIO;
1510 goto finish;
1511 }
1512 /* tmp was consumed */
1513 tmp = NULL;
1514
1515 if (omit_env_vars) {
1516 argvsize = 0;
1517
1518 /* Iterate over everything in argv, plus one for the bare executable path */
1519 for (current_arg_index = 0; current_arg_index < argc + 1 && argvsize < argslen; ++current_arg_index) {
1520 current_arg = (const char *)(copy_end - argslen) + argvsize;
1521 remaining = argslen - argvsize;
1522 current_arg_len = strnlen(current_arg, remaining);
1523 if (current_arg_len < remaining) {
1524 /* We have space for the null terminator */
1525 current_arg_len += 1;
1526
1527 if (current_arg_index == 0) {
1528 /* The bare executable path may have multiple null bytes after it for alignment */
1529 while (current_arg_len < remaining && current_arg[current_arg_len] == 0) {
1530 current_arg_len += 1;
1531 }
1532 }
1533 }
1534 argvsize += current_arg_len;
1535 }
1536 assert(argvsize <= argslen);
1537
1538 /* Adjust argslen and copy_end to make the copyout range extend to the end of argv */
1539 copy_end = copy_end - argslen + argvsize;
1540 argslen = argvsize;
1541 }
1542
1543 if (where == USER_ADDR_NULL) {
1544 /* Skip copyout */
1545 goto calculate_size;
1546 }
1547
1548 if (buflen >= argslen) {
1549 data = (caddr_t) (copy_end - argslen);
1550 size = argslen;
1551 } else {
1552 /*
1553 * Before rdar://25397314, this function contained incorrect logic when buflen is less
1554 * than argslen. The problem was that it copied in `buflen` bytes from the end of the target
1555 * process user stack into the beginning of a buffer of size round_page(buflen), and then
1556 * copied out `buflen` bytes from the end of this buffer. The effect of this was that
1557 * the caller of this sysctl would get zeros at the end of their buffer.
1558 *
1559 * To preserve this behavior, bzero everything from copy_end-round_page(buflen)+buflen to the
1560 * end of the buffer. This emulates copying in only `buflen` bytes.
1561 *
1562 *
1563 * In the old code:
1564 *
1565 * copy_start .... size: round_page(buflen) .... copy_end
1566 * [---copied in data (size: buflen)---|--- zeros ----------]
1567 * ^
1568 * data = copy_end - buflen
1569 *
1570 *
1571 * In the new code:
1572 * copy_start .... size: round_page(p->argslen) .... full copy_end
1573 * ^ ....................... p->argslen ...............................^
1574 * ^ ^ truncated copy_end ^
1575 * ^ ^ ^ ^
1576 * ^ ................ argslen ........................ ^
1577 * ^ ^ ^ ^
1578 * [-------copied in data (size: round_page(p->argslen))-------:----env vars---]
1579 * ^ ^
1580 * ^ data = copy_end - buflen
1581 * smallbuffer_start = max(copy_end - round_page(buflen), copy_start)
1582 *
1583 *
1584 * Full copy_end: copy_end calculated from copy_start + round_page(p->argslen)
1585 * Truncated copy_end: copy_end after truncation to remove environment variables.
1586 *
1587 * If environment variables were omitted, then we use the truncated copy_end, otherwise
1588 * we use full copy_end.
1589 *
1590 * smallbuffer_start: represents where copy_start would be in the old code.
1591 * data: The beginning of the region we copyout
1592 */
1593 smallbuffer_start = copy_end - vm_map_round_page(buflen, effective_page_mask);
1594 if (smallbuffer_start < copy_start) {
1595 smallbuffer_start = copy_start;
1596 }
1597 bzero((void *)(smallbuffer_start + buflen), copy_end - (smallbuffer_start + buflen));
1598 data = (caddr_t) (copy_end - buflen);
1599 size = buflen;
1600 }
1601
1602 if (argc_yes) {
1603 /* Put processes argc as the first word in the copyout buffer */
1604 suword(where, argc);
1605 error = copyout(data, (where + sizeof(int)), size);
1606 size += sizeof(int);
1607 } else {
1608 error = copyout(data, where, size);
1609
1610 /*
1611 * Make the old PROCARGS work to return the executable's path
1612 * But, only if there is enough space in the provided buffer
1613 *
1614 * on entry: data [possibily] points to the beginning of the path
1615 *
1616 * Note: we keep all pointers&sizes aligned to word boundries
1617 */
1618 if ((!error) && (buflen > 0 && (u_int)buflen > size)) {
1619 int binPath_sz, alignedBinPath_sz = 0;
1620 int extraSpaceNeeded, addThis;
1621 user_addr_t placeHere;
1622 char * str = (char *) data;
1623 size_t max_len = size;
1624
1625 /* Some apps are really bad about messing up their stacks
1626 * So, we have to be extra careful about getting the length
1627 * of the executing binary. If we encounter an error, we bail.
1628 */
1629
1630 /* Limit ourselves to PATH_MAX paths */
1631 if (max_len > PATH_MAX) {
1632 max_len = PATH_MAX;
1633 }
1634
1635 binPath_sz = 0;
1636
1637 while ((binPath_sz < max_len - 1) && (*str++ != 0)) {
1638 binPath_sz++;
1639 }
1640
1641 /* If we have a NUL terminator, copy it, too */
1642 if (binPath_sz < max_len - 1) {
1643 binPath_sz += 1;
1644 }
1645
1646 /* Pre-Flight the space requiremnts */
1647
1648 /* Account for the padding that fills out binPath to the next word */
1649 alignedBinPath_sz += (binPath_sz & (sizeof(int) - 1)) ? (sizeof(int) - (binPath_sz & (sizeof(int) - 1))) : 0;
1650
1651 placeHere = where + size;
1652
1653 /* Account for the bytes needed to keep placeHere word aligned */
1654 addThis = (placeHere & (sizeof(int) - 1)) ? (sizeof(int) - (placeHere & (sizeof(int) - 1))) : 0;
1655
1656 /* Add up all the space that is needed */
1657 extraSpaceNeeded = alignedBinPath_sz + addThis + binPath_sz + (4 * sizeof(int));
1658
1659 /* is there is room to tack on argv[0]? */
1660 if ((buflen & ~(sizeof(int) - 1)) >= (size + extraSpaceNeeded)) {
1661 placeHere += addThis;
1662 suword(placeHere, 0);
1663 placeHere += sizeof(int);
1664 suword(placeHere, 0xBFFF0000);
1665 placeHere += sizeof(int);
1666 suword(placeHere, 0);
1667 placeHere += sizeof(int);
1668 error = copyout(data, placeHere, binPath_sz);
1669 if (!error) {
1670 placeHere += binPath_sz;
1671 suword(placeHere, 0);
1672 size += extraSpaceNeeded;
1673 }
1674 }
1675 }
1676 }
1677
1678 calculate_size:
1679 /* Size has already been calculated for the where != NULL case */
1680 if (where == USER_ADDR_NULL) {
1681 size = argslen;
1682 if (argc_yes) {
1683 size += sizeof(int);
1684 } else {
1685 /*
1686 * old PROCARGS will return the executable's path and plus some
1687 * extra space for work alignment and data tags
1688 */
1689 size += PATH_MAX + (6 * sizeof(int));
1690 }
1691 size += (size & (sizeof(int) - 1)) ? (sizeof(int) - (size & (sizeof(int) - 1))) : 0;
1692 }
1693
1694 *sizep = size;
1695
1696 finish:
1697 if (p != NULL) {
1698 proc_rele(p);
1699 }
1700 if (tmp != NULL) {
1701 vm_map_copy_discard(tmp);
1702 }
1703 if (proc_map != NULL) {
1704 vm_map_deallocate(proc_map);
1705 }
1706 if (copy_start != (vm_offset_t) 0) {
1707 kmem_free(kernel_map, copy_start, arg_size);
1708 }
1709 return error;
1710 }
1711
1712
1713 /*
1714 * Max number of concurrent aio requests
1715 */
1716 STATIC int
1717 sysctl_aiomax
1718 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1719 {
1720 int new_value, changed;
1721 int error = sysctl_io_number(req, aio_max_requests, sizeof(int), &new_value, &changed);
1722 if (changed) {
1723 /* make sure the system-wide limit is greater than the per process limit */
1724 if (new_value >= aio_max_requests_per_process && new_value <= AIO_MAX_REQUESTS) {
1725 aio_max_requests = new_value;
1726 } else {
1727 error = EINVAL;
1728 }
1729 }
1730 return error;
1731 }
1732
1733
1734 /*
1735 * Max number of concurrent aio requests per process
1736 */
1737 STATIC int
1738 sysctl_aioprocmax
1739 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1740 {
1741 int new_value, changed;
1742 int error = sysctl_io_number(req, aio_max_requests_per_process, sizeof(int), &new_value, &changed);
1743 if (changed) {
1744 /* make sure per process limit is less than the system-wide limit */
1745 if (new_value <= aio_max_requests && new_value >= AIO_LISTIO_MAX) {
1746 aio_max_requests_per_process = new_value;
1747 } else {
1748 error = EINVAL;
1749 }
1750 }
1751 return error;
1752 }
1753
1754
1755 /*
1756 * Max number of async IO worker threads
1757 */
1758 STATIC int
1759 sysctl_aiothreads
1760 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1761 {
1762 int new_value, changed;
1763 int error = sysctl_io_number(req, aio_worker_threads, sizeof(int), &new_value, &changed);
1764 if (changed) {
1765 /* we only allow an increase in the number of worker threads */
1766 if (new_value > aio_worker_threads) {
1767 _aio_create_worker_threads((new_value - aio_worker_threads));
1768 aio_worker_threads = new_value;
1769 } else {
1770 error = EINVAL;
1771 }
1772 }
1773 return error;
1774 }
1775
1776
1777 /*
1778 * System-wide limit on the max number of processes
1779 */
1780 STATIC int
1781 sysctl_maxproc
1782 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1783 {
1784 int new_value, changed;
1785 int error = sysctl_io_number(req, maxproc, sizeof(int), &new_value, &changed);
1786 if (changed) {
1787 AUDIT_ARG(value32, new_value);
1788 /* make sure the system-wide limit is less than the configured hard
1789 * limit set at kernel compilation */
1790 if (new_value <= hard_maxproc && new_value > 0) {
1791 maxproc = new_value;
1792 } else {
1793 error = EINVAL;
1794 }
1795 }
1796 return error;
1797 }
1798
1799 STATIC int
1800 sysctl_sched_enable_smt
1801 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1802 {
1803 int new_value, changed;
1804 int error = sysctl_io_number(req, sched_enable_smt, sizeof(int), &new_value, &changed);
1805 if (error) {
1806 return error;
1807 }
1808 kern_return_t kret = KERN_SUCCESS;
1809 if (changed) {
1810 AUDIT_ARG(value32, new_value);
1811 if (new_value == 0) {
1812 sched_enable_smt = 0;
1813 kret = enable_smt_processors(false);
1814 } else {
1815 sched_enable_smt = 1;
1816 kret = enable_smt_processors(true);
1817 }
1818 }
1819 switch (kret) {
1820 case KERN_SUCCESS:
1821 error = 0;
1822 break;
1823 case KERN_INVALID_ARGUMENT:
1824 error = EINVAL;
1825 break;
1826 case KERN_FAILURE:
1827 error = EBUSY;
1828 break;
1829 default:
1830 error = ENOENT;
1831 break;
1832 }
1833
1834 return error;
1835 }
1836
1837 SYSCTL_STRING(_kern, KERN_OSTYPE, ostype,
1838 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
1839 ostype, 0, "");
1840 SYSCTL_STRING(_kern, KERN_OSRELEASE, osrelease,
1841 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
1842 osrelease, 0, "");
1843 SYSCTL_INT(_kern, KERN_OSREV, osrevision,
1844 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
1845 (int *)NULL, BSD, "");
1846 SYSCTL_STRING(_kern, KERN_VERSION, version,
1847 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
1848 version, 0, "");
1849 SYSCTL_STRING(_kern, OID_AUTO, uuid,
1850 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
1851 &kernel_uuid_string[0], 0, "");
1852
1853 SYSCTL_STRING(_kern, OID_AUTO, osbuildconfig,
1854 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED | CTLFLAG_MASKED,
1855 &osbuild_config[0], 0, "");
1856
1857 #if DEBUG
1858 #ifndef DKPR
1859 #define DKPR 1
1860 #endif
1861 #endif
1862
1863 #if DKPR
1864 int debug_kprint_syscall = 0;
1865 char debug_kprint_syscall_process[MAXCOMLEN + 1];
1866
1867 /* Thread safe: bits and string value are not used to reclaim state */
1868 SYSCTL_INT(_debug, OID_AUTO, kprint_syscall,
1869 CTLFLAG_RW | CTLFLAG_LOCKED, &debug_kprint_syscall, 0, "kprintf syscall tracing");
1870 SYSCTL_STRING(_debug, OID_AUTO, kprint_syscall_process,
1871 CTLFLAG_RW | CTLFLAG_LOCKED, debug_kprint_syscall_process, sizeof(debug_kprint_syscall_process),
1872 "name of process for kprintf syscall tracing");
1873
1874 int
1875 debug_kprint_current_process(const char **namep)
1876 {
1877 struct proc *p = current_proc();
1878
1879 if (p == NULL) {
1880 return 0;
1881 }
1882
1883 if (debug_kprint_syscall_process[0]) {
1884 /* user asked to scope tracing to a particular process name */
1885 if (0 == strncmp(debug_kprint_syscall_process,
1886 p->p_comm, sizeof(debug_kprint_syscall_process))) {
1887 /* no value in telling the user that we traced what they asked */
1888 if (namep) {
1889 *namep = NULL;
1890 }
1891
1892 return 1;
1893 } else {
1894 return 0;
1895 }
1896 }
1897
1898 /* trace all processes. Tell user what we traced */
1899 if (namep) {
1900 *namep = p->p_comm;
1901 }
1902
1903 return 1;
1904 }
1905 #endif
1906
1907 /* PR-5293665: need to use a callback function for kern.osversion to set
1908 * osversion in IORegistry */
1909
1910 STATIC int
1911 sysctl_osversion(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
1912 {
1913 int rval = 0;
1914
1915 rval = sysctl_handle_string(oidp, arg1, arg2, req);
1916
1917 if (req->newptr) {
1918 IORegistrySetOSBuildVersion((char *)arg1);
1919 }
1920
1921 return rval;
1922 }
1923
1924 SYSCTL_PROC(_kern, KERN_OSVERSION, osversion,
1925 CTLFLAG_RW | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
1926 osversion, 256 /* OSVERSIZE*/,
1927 sysctl_osversion, "A", "");
1928
1929 static bool
1930 _already_set_or_not_launchd(struct sysctl_req *req, char *val)
1931 {
1932 if (req->newptr != 0) {
1933 /*
1934 * Can only ever be set by launchd, and only once at boot.
1935 */
1936 if (proc_getpid(req->p) != 1 || val[0] != '\0') {
1937 return true;
1938 }
1939 }
1940 return false;
1941 }
1942
1943 #define kRootsInstalledReadWriteEntitlement "com.apple.private.roots-installed-read-write"
1944 #define kRootsInstalledReadOnlyEntitlement "com.apple.private.roots-installed-read-only"
1945 uint64_t roots_installed = 0;
1946
1947 static int
1948 sysctl_roots_installed
1949 (__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
1950 {
1951 int error = 0;
1952
1953 if (req->newptr != 0) {
1954 /* a ReadWrite entitlement is required for updating this syscl
1955 * meanwhile, only allow write once
1956 */
1957 if (!IOCurrentTaskHasEntitlement(kRootsInstalledReadWriteEntitlement) || (roots_installed != 0)) {
1958 return EPERM;
1959 }
1960 } else {
1961 /* for reader of this sysctl, need either ReadWrite or ReadOnly entitlement */
1962 if (!IOCurrentTaskHasEntitlement(kRootsInstalledReadWriteEntitlement) &&
1963 !IOCurrentTaskHasEntitlement(kRootsInstalledReadOnlyEntitlement)) {
1964 return EPERM;
1965 }
1966 }
1967
1968 error = sysctl_handle_quad(oidp, arg1, arg2, req);
1969
1970 return error;
1971 }
1972
1973 SYSCTL_PROC(_kern, OID_AUTO, roots_installed,
1974 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
1975 &roots_installed, sizeof(roots_installed),
1976 sysctl_roots_installed, "Q", "");
1977
1978 #if XNU_TARGET_OS_OSX
1979 static int
1980 sysctl_system_version_compat
1981 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1982 {
1983 int oldval = (task_has_system_version_compat_enabled(current_task()));
1984 int new_value = 0, changed = 0;
1985
1986 int error = sysctl_io_number(req, oldval, sizeof(int), &new_value, &changed);
1987 if (changed) {
1988 task_set_system_version_compat_enabled(current_task(), (new_value));
1989 }
1990 return error;
1991 }
1992
1993 SYSCTL_PROC(_kern, OID_AUTO, system_version_compat,
1994 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
1995 0, 0, sysctl_system_version_compat, "A", "");
1996
1997 char osproductversioncompat[48] = { '\0' };
1998
1999 static int
2000 sysctl_osproductversioncompat(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2001 {
2002 if (_already_set_or_not_launchd(req, osproductversioncompat)) {
2003 return EPERM;
2004 }
2005 return sysctl_handle_string(oidp, arg1, arg2, req);
2006 }
2007
2008
2009 SYSCTL_PROC(_kern, OID_AUTO, osproductversioncompat,
2010 CTLFLAG_RW | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2011 osproductversioncompat, sizeof(osproductversioncompat),
2012 sysctl_osproductversioncompat, "A", "The ProductVersion from SystemVersionCompat.plist");
2013 #endif
2014
2015 char osproductversion[48] = { '\0' };
2016
2017 static char iossupportversion_string[48] = { '\0' };
2018
2019 static int
2020 sysctl_osproductversion(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2021 {
2022 if (_already_set_or_not_launchd(req, osproductversion)) {
2023 return EPERM;
2024 }
2025
2026 #if XNU_TARGET_OS_OSX
2027 if (task_has_system_version_compat_enabled(current_task()) && (osproductversioncompat[0] != '\0')) {
2028 return sysctl_handle_string(oidp, osproductversioncompat, arg2, req);
2029 } else {
2030 return sysctl_handle_string(oidp, arg1, arg2, req);
2031 }
2032 #elif defined(XNU_TARGET_OS_XR)
2033 if (proc_platform(req->p) == PLATFORM_IOS && (iossupportversion_string[0] != '\0')) {
2034 return sysctl_handle_string(oidp, iossupportversion_string, arg2, req);
2035 } else {
2036 return sysctl_handle_string(oidp, arg1, arg2, req);
2037 }
2038 #else
2039 return sysctl_handle_string(oidp, arg1, arg2, req);
2040 #endif
2041 }
2042
2043 #if XNU_TARGET_OS_OSX
2044 static_assert(sizeof(osproductversioncompat) == sizeof(osproductversion),
2045 "osproductversion size matches osproductversioncompat size");
2046 #endif
2047
2048 SYSCTL_PROC(_kern, OID_AUTO, osproductversion,
2049 CTLFLAG_RW | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2050 osproductversion, sizeof(osproductversion),
2051 sysctl_osproductversion, "A", "The ProductVersion from SystemVersion.plist");
2052
2053 char osreleasetype[OSRELEASETYPE_SIZE] = { '\0' };
2054
2055 STATIC int
2056 sysctl_osreleasetype(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2057 {
2058 if (_already_set_or_not_launchd(req, osreleasetype)) {
2059 return EPERM;
2060 }
2061 return sysctl_handle_string(oidp, arg1, arg2, req);
2062 }
2063
2064 void reset_osreleasetype(void);
2065
2066 void
2067 reset_osreleasetype(void)
2068 {
2069 memset(osreleasetype, 0, sizeof(osreleasetype));
2070 }
2071
2072 SYSCTL_PROC(_kern, OID_AUTO, osreleasetype,
2073 CTLFLAG_RW | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2074 osreleasetype, sizeof(osreleasetype),
2075 sysctl_osreleasetype, "A", "The ReleaseType from SystemVersion.plist");
2076
2077 STATIC int
2078 sysctl_iossupportversion(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2079 {
2080 if (_already_set_or_not_launchd(req, iossupportversion_string)) {
2081 return EPERM;
2082 }
2083
2084 #if defined(XNU_TARGET_OS_XR)
2085 if (proc_platform(req->p) == PLATFORM_IOS) {
2086 /* return empty string for iOS processes to match how this would behave on iOS */
2087 return sysctl_handle_string(oidp, "", arg2, req);
2088 } else {
2089 /* native processes see the actual value */
2090 return sysctl_handle_string(oidp, arg1, arg2, req);
2091 }
2092 #else
2093 return sysctl_handle_string(oidp, arg1, arg2, req);
2094 #endif
2095 }
2096
2097 SYSCTL_PROC(_kern, OID_AUTO, iossupportversion,
2098 CTLFLAG_RW | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2099 iossupportversion_string, sizeof(iossupportversion_string),
2100 sysctl_iossupportversion, "A", "The iOSSupportVersion from SystemVersion.plist");
2101
2102 static uint64_t osvariant_status = 0;
2103
2104 STATIC int
2105 sysctl_osvariant_status(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2106 {
2107 if (req->newptr != 0) {
2108 /*
2109 * Can only ever be set by launchd, and only once.
2110 * Reset by usrctl() -> reset_osvariant_status() during
2111 * userspace reboot, since userspace could reboot into
2112 * a different variant.
2113 */
2114 if (proc_getpid(req->p) != 1 || osvariant_status != 0) {
2115 return EPERM;
2116 }
2117 }
2118
2119 int err = sysctl_handle_quad(oidp, arg1, arg2, req);
2120
2121 reset_debug_syscall_rejection_mode();
2122
2123 return err;
2124 }
2125
2126 SYSCTL_PROC(_kern, OID_AUTO, osvariant_status,
2127 CTLFLAG_RW | CTLTYPE_QUAD | CTLFLAG_LOCKED | CTLFLAG_MASKED,
2128 &osvariant_status, sizeof(osvariant_status),
2129 sysctl_osvariant_status, "Q", "Opaque flags used to cache OS variant information");
2130
2131 static bool
2132 _os_variant_check_disabled(enum os_variant_property property)
2133 {
2134 return (osvariant_status >> (32 + property)) & 0x1;
2135 }
2136
2137 static bool
2138 _os_variant_has(enum os_variant_status_flags_positions p)
2139 {
2140 return ((osvariant_status >> (p * OS_VARIANT_STATUS_BIT_WIDTH)) & OS_VARIANT_STATUS_MASK) == OS_VARIANT_S_YES;
2141 }
2142
2143 bool
2144 os_variant_has_internal_diagnostics(__unused const char *subsystem)
2145 {
2146 if (_os_variant_check_disabled(OS_VARIANT_PROPERTY_DIAGNOSTICS)) {
2147 return false;
2148 }
2149 #if XNU_TARGET_OS_OSX
2150 return _os_variant_has(OS_VARIANT_SFP_INTERNAL_CONTENT) || _os_variant_has(OS_VARIANT_SFP_INTERNAL_DIAGS_PROFILE);
2151 #else
2152 return _os_variant_has(OS_VARIANT_SFP_INTERNAL_RELEASE_TYPE);
2153 #endif /* XNU_TARGET_OS_OSX */
2154 }
2155
2156 void reset_osvariant_status(void);
2157
2158 void
2159 reset_osvariant_status(void)
2160 {
2161 osvariant_status = 0;
2162 reset_debug_syscall_rejection_mode();
2163 }
2164
2165 extern void commpage_update_dyld_flags(uint64_t);
2166 TUNABLE_WRITEABLE(uint64_t, dyld_flags, "dyld_flags", 0);
2167
2168 STATIC int
2169 sysctl_dyld_flags(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2170 {
2171 /*
2172 * Can only ever be set by launchd, possibly several times
2173 * as dyld may change its mind after a userspace reboot.
2174 */
2175 if (req->newptr != 0 && proc_getpid(req->p) != 1) {
2176 return EPERM;
2177 }
2178
2179 int res = sysctl_handle_quad(oidp, arg1, arg2, req);
2180 if (req->newptr && res == 0) {
2181 commpage_update_dyld_flags(dyld_flags);
2182 }
2183 return res;
2184 }
2185
2186 SYSCTL_PROC(_kern, OID_AUTO, dyld_flags,
2187 CTLFLAG_RW | CTLTYPE_QUAD | CTLFLAG_LOCKED | CTLFLAG_MASKED,
2188 &dyld_flags, sizeof(dyld_flags),
2189 sysctl_dyld_flags, "Q", "Opaque flags used to cache dyld system-wide configuration");
2190
2191 #if defined(XNU_TARGET_OS_BRIDGE)
2192 char macosproductversion[MACOS_VERS_LEN] = { '\0' };
2193
2194 SYSCTL_STRING(_kern, OID_AUTO, macosproductversion,
2195 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2196 &macosproductversion[0], MACOS_VERS_LEN, "The currently running macOS ProductVersion (from SystemVersion.plist on macOS)");
2197
2198 char macosversion[MACOS_VERS_LEN] = { '\0' };
2199
2200 SYSCTL_STRING(_kern, OID_AUTO, macosversion,
2201 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2202 &macosversion[0], MACOS_VERS_LEN, "The currently running macOS build version");
2203 #endif
2204
2205 STATIC int
2206 sysctl_sysctl_bootargs
2207 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2208 {
2209 int error;
2210 char buf[BOOT_LINE_LENGTH];
2211
2212 strlcpy(buf, PE_boot_args(), BOOT_LINE_LENGTH);
2213 error = sysctl_io_string(req, buf, BOOT_LINE_LENGTH, 0, NULL);
2214 return error;
2215 }
2216
2217 SYSCTL_PROC(_kern, OID_AUTO, bootargs,
2218 CTLFLAG_LOCKED | CTLFLAG_RD | CTLFLAG_KERN | CTLTYPE_STRING,
2219 NULL, 0,
2220 sysctl_sysctl_bootargs, "A", "bootargs");
2221
2222 STATIC int
2223 sysctl_kernelcacheuuid(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2224 {
2225 int rval = ENOENT;
2226 if (kernelcache_uuid_valid) {
2227 rval = sysctl_handle_string(oidp, arg1, arg2, req);
2228 }
2229 return rval;
2230 }
2231
2232 SYSCTL_PROC(_kern, OID_AUTO, kernelcacheuuid,
2233 CTLFLAG_RD | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2234 kernelcache_uuid_string, sizeof(kernelcache_uuid_string),
2235 sysctl_kernelcacheuuid, "A", "");
2236
2237 STATIC int
2238 sysctl_systemfilesetuuid(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2239 {
2240 int rval = ENOENT;
2241 if (pageablekc_uuid_valid) {
2242 rval = sysctl_handle_string(oidp, arg1, arg2, req);
2243 }
2244 return rval;
2245 }
2246
2247 SYSCTL_PROC(_kern, OID_AUTO, systemfilesetuuid,
2248 CTLFLAG_RD | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2249 pageablekc_uuid_string, sizeof(pageablekc_uuid_string),
2250 sysctl_systemfilesetuuid, "A", "");
2251
2252 STATIC int
2253 sysctl_auxiliaryfilesetuuid(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2254 {
2255 int rval = ENOENT;
2256 if (auxkc_uuid_valid) {
2257 rval = sysctl_handle_string(oidp, arg1, arg2, req);
2258 }
2259 return rval;
2260 }
2261
2262 SYSCTL_PROC(_kern, OID_AUTO, auxiliaryfilesetuuid,
2263 CTLFLAG_RD | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2264 auxkc_uuid_string, sizeof(auxkc_uuid_string),
2265 sysctl_auxiliaryfilesetuuid, "A", "");
2266
2267 STATIC int
2268 sysctl_filesetuuid(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2269 {
2270 int rval = ENOENT;
2271 kc_format_t kcformat;
2272 kernel_mach_header_t *mh;
2273 void *uuid = NULL;
2274 unsigned long uuidlen = 0;
2275 uuid_string_t uuid_str;
2276
2277 if (!PE_get_primary_kc_format(&kcformat) || kcformat != KCFormatFileset) {
2278 return rval;
2279 }
2280
2281 mh = (kernel_mach_header_t *)PE_get_kc_header(KCKindPrimary);
2282 uuid = getuuidfromheader(mh, &uuidlen);
2283
2284 if ((uuid != NULL) && (uuidlen == sizeof(uuid_t))) {
2285 uuid_unparse_upper(*(uuid_t *)uuid, uuid_str);
2286 rval = sysctl_io_string(req, (char *)uuid_str, sizeof(uuid_str), 0, NULL);
2287 }
2288
2289 return rval;
2290 }
2291
2292 SYSCTL_PROC(_kern, OID_AUTO, filesetuuid,
2293 CTLFLAG_RD | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2294 NULL, 0,
2295 sysctl_filesetuuid, "A", "");
2296
2297
2298 SYSCTL_INT(_kern, KERN_MAXFILES, maxfiles,
2299 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2300 &maxfiles, 0, "");
2301 SYSCTL_INT(_kern, KERN_ARGMAX, argmax,
2302 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2303 (int *)NULL, ARG_MAX, "");
2304 SYSCTL_INT(_kern, KERN_POSIX1, posix1version,
2305 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2306 (int *)NULL, _POSIX_VERSION, "");
2307 SYSCTL_INT(_kern, KERN_NGROUPS, ngroups,
2308 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2309 (int *)NULL, NGROUPS_MAX, "");
2310 SYSCTL_INT(_kern, KERN_JOB_CONTROL, job_control,
2311 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2312 (int *)NULL, 1, "");
2313 #if 1 /* _POSIX_SAVED_IDS from <unistd.h> */
2314 SYSCTL_INT(_kern, KERN_SAVED_IDS, saved_ids,
2315 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2316 (int *)NULL, 1, "");
2317 #else
2318 SYSCTL_INT(_kern, KERN_SAVED_IDS, saved_ids,
2319 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2320 NULL, 0, "");
2321 #endif
2322 SYSCTL_INT(_kern, OID_AUTO, num_files,
2323 CTLFLAG_RD | CTLFLAG_LOCKED,
2324 &nfiles, 0, "");
2325 SYSCTL_COMPAT_INT(_kern, OID_AUTO, num_vnodes,
2326 CTLFLAG_RD | CTLFLAG_LOCKED,
2327 &numvnodes, 0, "");
2328 SYSCTL_INT(_kern, OID_AUTO, num_tasks,
2329 CTLFLAG_RD | CTLFLAG_LOCKED,
2330 &task_max, 0, "");
2331 SYSCTL_INT(_kern, OID_AUTO, num_threads,
2332 CTLFLAG_RD | CTLFLAG_LOCKED,
2333 &thread_max, 0, "");
2334 SYSCTL_INT(_kern, OID_AUTO, num_taskthreads,
2335 CTLFLAG_RD | CTLFLAG_LOCKED,
2336 &task_threadmax, 0, "");
2337 SYSCTL_LONG(_kern, OID_AUTO, num_recycledvnodes,
2338 CTLFLAG_RD | CTLFLAG_LOCKED,
2339 &num_recycledvnodes, "");
2340 SYSCTL_COMPAT_INT(_kern, OID_AUTO, free_vnodes,
2341 CTLFLAG_RD | CTLFLAG_LOCKED,
2342 &freevnodes, 0, "");
2343
2344 STATIC int
2345 sysctl_maxvnodes(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2346 {
2347 int oldval = desiredvnodes;
2348 int error = sysctl_io_number(req, desiredvnodes, sizeof(int), &desiredvnodes, NULL);
2349
2350 if (oldval != desiredvnodes) {
2351 resize_namecache(desiredvnodes);
2352 }
2353
2354 return error;
2355 }
2356
2357 SYSCTL_INT(_kern, OID_AUTO, namecache_disabled,
2358 CTLFLAG_RW | CTLFLAG_LOCKED,
2359 &nc_disabled, 0, "");
2360
2361 SYSCTL_PROC(_kern, KERN_MAXVNODES, maxvnodes,
2362 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2363 0, 0, sysctl_maxvnodes, "I", "");
2364
2365 SYSCTL_PROC(_kern, KERN_MAXPROC, maxproc,
2366 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2367 0, 0, sysctl_maxproc, "I", "");
2368
2369 SYSCTL_PROC(_kern, KERN_AIOMAX, aiomax,
2370 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2371 0, 0, sysctl_aiomax, "I", "");
2372
2373 SYSCTL_PROC(_kern, KERN_AIOPROCMAX, aioprocmax,
2374 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2375 0, 0, sysctl_aioprocmax, "I", "");
2376
2377 SYSCTL_PROC(_kern, KERN_AIOTHREADS, aiothreads,
2378 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2379 0, 0, sysctl_aiothreads, "I", "");
2380
2381 SYSCTL_PROC(_kern, OID_AUTO, sched_enable_smt,
2382 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_KERN,
2383 0, 0, sysctl_sched_enable_smt, "I", "");
2384
2385 extern int sched_allow_NO_SMT_threads;
2386 SYSCTL_INT(_kern, OID_AUTO, sched_allow_NO_SMT_threads,
2387 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2388 &sched_allow_NO_SMT_threads, 0, "");
2389
2390 extern int sched_avoid_cpu0;
2391 SYSCTL_INT(_kern, OID_AUTO, sched_rt_avoid_cpu0,
2392 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2393 &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");
2394
2395 #if (DEVELOPMENT || DEBUG)
2396
2397 static int
2398 sysctl_kern_max_unsafe_rt_quanta(__unused struct sysctl_oid *oidp,
2399 __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2400 {
2401 extern void sched_set_max_unsafe_rt_quanta(int);
2402 extern int max_unsafe_rt_quanta;
2403
2404 int new_value, changed;
2405 int old_value = max_unsafe_rt_quanta;
2406 int error = sysctl_io_number(req, old_value, sizeof(int), &new_value,
2407 &changed);
2408 if (changed) {
2409 sched_set_max_unsafe_rt_quanta(new_value);
2410 }
2411
2412 return error;
2413 }
2414
2415 SYSCTL_PROC(_kern, OID_AUTO, max_unsafe_rt_quanta,
2416 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2417 0, 0, sysctl_kern_max_unsafe_rt_quanta, "I",
2418 "Number of quanta to allow a realtime "
2419 "thread to run before being penalized");
2420
2421 static int
2422 sysctl_kern_max_unsafe_fixed_quanta(__unused struct sysctl_oid *oidp,
2423 __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2424 {
2425 extern void sched_set_max_unsafe_fixed_quanta(int);
2426 extern int max_unsafe_fixed_quanta;
2427
2428 int new_value, changed;
2429 int old_value = max_unsafe_fixed_quanta;
2430 int error = sysctl_io_number(req, old_value, sizeof(int), &new_value,
2431 &changed);
2432 if (changed) {
2433 sched_set_max_unsafe_fixed_quanta(new_value);
2434 }
2435
2436 return error;
2437 }
2438
2439 SYSCTL_PROC(_kern, OID_AUTO, max_unsafe_fixed_quanta,
2440 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2441 0, 0, sysctl_kern_max_unsafe_fixed_quanta, "I",
2442 "Number of quanta to allow a fixed sched mode "
2443 "thread to run before being penalized");
2444
2445 static int
2446 sysctl_kern_quantum_us(__unused struct sysctl_oid *oidp, __unused void *arg1,
2447 __unused int arg2, struct sysctl_req *req)
2448 {
2449 extern uint64_t sysctl_get_quantum_us(void);
2450 const uint64_t quantum_us = sysctl_get_quantum_us();
2451
2452 return sysctl_io_number(req, quantum_us, sizeof(quantum_us), NULL, NULL);
2453 }
2454
2455 SYSCTL_PROC(_kern, OID_AUTO, quantum_us,
2456 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
2457 0, 0, sysctl_kern_quantum_us, "Q",
2458 "Length of scheduling quantum in microseconds");
2459
2460 extern int smt_sched_bonus_16ths;
2461 SYSCTL_INT(_kern, OID_AUTO, smt_sched_bonus_16ths,
2462 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2463 &smt_sched_bonus_16ths, 0, "");
2464
2465 extern int smt_timeshare_enabled;
2466 SYSCTL_INT(_kern, OID_AUTO, sched_smt_timeshare_enable,
2467 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2468 &smt_timeshare_enabled, 0, "");
2469
2470 extern int sched_smt_balance;
2471 SYSCTL_INT(_kern, OID_AUTO, sched_smt_balance,
2472 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2473 &sched_smt_balance, 0, "");
2474 extern int sched_allow_rt_smt;
2475 SYSCTL_INT(_kern, OID_AUTO, sched_allow_rt_smt,
2476 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2477 &sched_allow_rt_smt, 0, "");
2478 extern int sched_allow_rt_steal;
2479 SYSCTL_INT(_kern, OID_AUTO, sched_allow_rt_steal,
2480 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2481 &sched_allow_rt_steal, 0, "");
2482 extern int sched_backup_cpu_timeout_count;
2483 SYSCTL_INT(_kern, OID_AUTO, sched_backup_cpu_timeout_count,
2484 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2485 &sched_backup_cpu_timeout_count, 0, "The maximum number of 10us delays before allowing a backup cpu to select a thread");
2486 #if __arm64__
2487 /* Scheduler perfcontrol callouts sysctls */
2488 SYSCTL_DECL(_kern_perfcontrol_callout);
2489 SYSCTL_NODE(_kern, OID_AUTO, perfcontrol_callout, CTLFLAG_RW | CTLFLAG_LOCKED, 0,
2490 "scheduler perfcontrol callouts");
2491
2492 extern int perfcontrol_callout_stats_enabled;
2493 SYSCTL_INT(_kern_perfcontrol_callout, OID_AUTO, stats_enabled,
2494 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2495 &perfcontrol_callout_stats_enabled, 0, "");
2496
2497 extern uint64_t perfcontrol_callout_stat_avg(perfcontrol_callout_type_t type,
2498 perfcontrol_callout_stat_t stat);
2499
2500 /* On-Core Callout */
2501 STATIC int
2502 sysctl_perfcontrol_callout_stat
2503 (__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2504 {
2505 perfcontrol_callout_stat_t stat = (perfcontrol_callout_stat_t)arg1;
2506 perfcontrol_callout_type_t type = (perfcontrol_callout_type_t)arg2;
2507 return sysctl_io_number(req, (int)perfcontrol_callout_stat_avg(type, stat),
2508 sizeof(int), NULL, NULL);
2509 }
2510
2511 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, oncore_instr,
2512 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2513 (void *)PERFCONTROL_STAT_INSTRS, PERFCONTROL_CALLOUT_ON_CORE,
2514 sysctl_perfcontrol_callout_stat, "I", "");
2515 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, oncore_cycles,
2516 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2517 (void *)PERFCONTROL_STAT_CYCLES, PERFCONTROL_CALLOUT_ON_CORE,
2518 sysctl_perfcontrol_callout_stat, "I", "");
2519 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, offcore_instr,
2520 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2521 (void *)PERFCONTROL_STAT_INSTRS, PERFCONTROL_CALLOUT_OFF_CORE,
2522 sysctl_perfcontrol_callout_stat, "I", "");
2523 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, offcore_cycles,
2524 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2525 (void *)PERFCONTROL_STAT_CYCLES, PERFCONTROL_CALLOUT_OFF_CORE,
2526 sysctl_perfcontrol_callout_stat, "I", "");
2527 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, context_instr,
2528 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2529 (void *)PERFCONTROL_STAT_INSTRS, PERFCONTROL_CALLOUT_CONTEXT,
2530 sysctl_perfcontrol_callout_stat, "I", "");
2531 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, context_cycles,
2532 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2533 (void *)PERFCONTROL_STAT_CYCLES, PERFCONTROL_CALLOUT_CONTEXT,
2534 sysctl_perfcontrol_callout_stat, "I", "");
2535 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, update_instr,
2536 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2537 (void *)PERFCONTROL_STAT_INSTRS, PERFCONTROL_CALLOUT_STATE_UPDATE,
2538 sysctl_perfcontrol_callout_stat, "I", "");
2539 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, update_cycles,
2540 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2541 (void *)PERFCONTROL_STAT_CYCLES, PERFCONTROL_CALLOUT_STATE_UPDATE,
2542 sysctl_perfcontrol_callout_stat, "I", "");
2543
2544 #if __AMP__
2545 #if !CONFIG_CLUTCH
2546 extern int sched_amp_idle_steal;
2547 SYSCTL_INT(_kern, OID_AUTO, sched_amp_idle_steal,
2548 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2549 &sched_amp_idle_steal, 0, "");
2550 extern int sched_amp_spill_steal;
2551 SYSCTL_INT(_kern, OID_AUTO, sched_amp_spill_steal,
2552 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2553 &sched_amp_spill_steal, 0, "");
2554 extern int sched_amp_spill_count;
2555 SYSCTL_INT(_kern, OID_AUTO, sched_amp_spill_count,
2556 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2557 &sched_amp_spill_count, 0, "");
2558 #endif /* !CONFIG_CLUTCH */
2559 extern int sched_amp_spill_deferred_ipi;
2560 SYSCTL_INT(_kern, OID_AUTO, sched_amp_spill_deferred_ipi,
2561 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2562 &sched_amp_spill_deferred_ipi, 0, "");
2563 extern int sched_amp_pcores_preempt_immediate_ipi;
2564 SYSCTL_INT(_kern, OID_AUTO, sched_amp_pcores_preempt_immediate_ipi,
2565 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2566 &sched_amp_pcores_preempt_immediate_ipi, 0, "");
2567 #endif /* __AMP__ */
2568 #endif /* __arm64__ */
2569
2570 #if __arm64__
2571 extern int legacy_footprint_entitlement_mode;
2572 SYSCTL_INT(_kern, OID_AUTO, legacy_footprint_entitlement_mode,
2573 CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
2574 &legacy_footprint_entitlement_mode, 0, "");
2575 #endif /* __arm64__ */
2576
2577 /*
2578 * Realtime threads are ordered by highest priority first then,
2579 * for threads of the same priority, by earliest deadline first.
2580 * But if sched_rt_runq_strict_priority is false (the default),
2581 * a lower priority thread with an earlier deadline will be preferred
2582 * over a higher priority thread with a later deadline, as long as
2583 * both threads' computations will fit before the later deadline.
2584 */
2585 extern int sched_rt_runq_strict_priority;
2586 SYSCTL_INT(_kern, OID_AUTO, sched_rt_runq_strict_priority,
2587 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2588 &sched_rt_runq_strict_priority, 0, "");
2589
2590 static int
2591 sysctl_kern_sched_rt_n_backup_processors(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2592 {
2593 int new_value, changed;
2594 int old_value = sched_get_rt_n_backup_processors();
2595 int error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
2596 if (changed) {
2597 sched_set_rt_n_backup_processors(new_value);
2598 }
2599
2600 return error;
2601 }
2602
2603 SYSCTL_PROC(_kern, OID_AUTO, sched_rt_n_backup_processors,
2604 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2605 0, 0, sysctl_kern_sched_rt_n_backup_processors, "I", "");
2606
2607 static int
2608 sysctl_kern_sched_rt_deadline_epsilon_us(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2609 {
2610 int new_value, changed;
2611 int old_value = sched_get_rt_deadline_epsilon();
2612 int error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
2613 if (changed) {
2614 sched_set_rt_deadline_epsilon(new_value);
2615 }
2616
2617 return error;
2618 }
2619
2620 SYSCTL_PROC(_kern, OID_AUTO, sched_rt_deadline_epsilon_us,
2621 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2622 0, 0, sysctl_kern_sched_rt_deadline_epsilon_us, "I", "");
2623
2624 extern int sched_idle_delay_cpuid;
2625 SYSCTL_INT(_kern, OID_AUTO, sched_idle_delay_cpuid,
2626 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2627 &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");
2628
2629 static int
2630 sysctl_kern_sched_powered_cores(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2631 {
2632 int new_value, changed;
2633 int old_value = sched_get_powered_cores();
2634 int error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
2635 if (changed) {
2636 if (!PE_parse_boot_argn("enable_skstb", NULL, 0)) {
2637 return ENOTSUP;
2638 }
2639
2640 sched_set_powered_cores(new_value);
2641 }
2642
2643 return error;
2644 }
2645
2646 SYSCTL_PROC(_kern, OID_AUTO, sched_powered_cores,
2647 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2648 0, 0, sysctl_kern_sched_powered_cores, "I", "");
2649
2650 #if __arm64__
2651
2652 static int
2653 sysctl_kern_update_sched_recommended_cores(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2654 {
2655 uint64_t new_value;
2656 int changed;
2657 uint64_t old_value = sched_sysctl_get_recommended_cores();
2658 int error = sysctl_io_number(req, old_value, sizeof(uint64_t), &new_value, &changed);
2659 if (changed) {
2660 if (!PE_parse_boot_argn("enable_skstb", NULL, 0)) {
2661 return ENOTSUP;
2662 }
2663
2664 sched_perfcontrol_update_recommended_cores_reason(new_value, REASON_CLPC_USER, 0);
2665 }
2666
2667 return error;
2668 }
2669
2670 SYSCTL_PROC(_kern, OID_AUTO, sched_update_recommended_cores,
2671 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
2672 0, 0, sysctl_kern_update_sched_recommended_cores, "I", "");
2673
2674 #endif /* __arm64__ */
2675
2676 #endif /* (DEVELOPMENT || DEBUG) */
2677
2678 extern uint64_t sysctl_sched_recommended_cores;
2679 SYSCTL_QUAD(_kern, OID_AUTO, sched_recommended_cores,
2680 CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
2681 &sysctl_sched_recommended_cores, "");
2682
2683 static int
2684 sysctl_kern_suspend_cluster_powerdown(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2685 {
2686 int new_value, changed;
2687 int old_value = get_cluster_powerdown_user_suspended();
2688 int error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
2689 if (!error && changed) {
2690 if (new_value > 0) {
2691 error = suspend_cluster_powerdown_from_user();
2692 } else {
2693 error = resume_cluster_powerdown_from_user();
2694 }
2695 if (error) {
2696 error = EALREADY;
2697 }
2698 }
2699
2700 return error;
2701 }
2702
2703 SYSCTL_PROC(_kern, OID_AUTO, suspend_cluster_powerdown,
2704 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2705 0, 0, sysctl_kern_suspend_cluster_powerdown, "I", "");
2706
2707
2708 STATIC int
2709 sysctl_securelvl
2710 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2711 {
2712 int new_value, changed;
2713 int error = sysctl_io_number(req, securelevel, sizeof(int), &new_value, &changed);
2714 if (changed) {
2715 if (!(new_value < securelevel && proc_getpid(req->p) != 1)) {
2716 proc_list_lock();
2717 securelevel = new_value;
2718 proc_list_unlock();
2719 } else {
2720 error = EPERM;
2721 }
2722 }
2723 return error;
2724 }
2725
2726 SYSCTL_PROC(_kern, KERN_SECURELVL, securelevel,
2727 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2728 0, 0, sysctl_securelvl, "I", "");
2729
2730
2731 STATIC int
2732 sysctl_domainname
2733 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2734 {
2735 int error, changed;
2736 char tmpname[MAXHOSTNAMELEN] = {};
2737
2738 lck_mtx_lock(&domainname_lock);
2739 strlcpy(tmpname, domainname, sizeof(tmpname));
2740 lck_mtx_unlock(&domainname_lock);
2741
2742 error = sysctl_io_string(req, tmpname, sizeof(tmpname), 0, &changed);
2743 if (!error && changed) {
2744 lck_mtx_lock(&domainname_lock);
2745 strlcpy(domainname, tmpname, sizeof(domainname));
2746 lck_mtx_unlock(&domainname_lock);
2747 }
2748 return error;
2749 }
2750
2751 SYSCTL_PROC(_kern, KERN_DOMAINNAME, nisdomainname,
2752 CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_LOCKED,
2753 0, 0, sysctl_domainname, "A", "");
2754
2755 SYSCTL_COMPAT_INT(_kern, KERN_HOSTID, hostid,
2756 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2757 &hostid, 0, "");
2758
2759 STATIC int
2760 sysctl_hostname
2761 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2762 {
2763 int error, changed;
2764 char tmpname[MAXHOSTNAMELEN] = {};
2765 const char * name;
2766
2767 #if XNU_TARGET_OS_OSX
2768 name = hostname;
2769 #else /* XNU_TARGET_OS_OSX */
2770 #define ENTITLEMENT_USER_ASSIGNED_DEVICE_NAME \
2771 "com.apple.developer.device-information.user-assigned-device-name"
2772 if (csproc_get_platform_binary(current_proc()) ||
2773 IOCurrentTaskHasEntitlement(ENTITLEMENT_USER_ASSIGNED_DEVICE_NAME)) {
2774 name = hostname;
2775 } else {
2776 /* Deny writes if we don't pass entitlement check */
2777 if (req->newptr) {
2778 return EPERM;
2779 }
2780
2781 name = "localhost";
2782 }
2783 #endif /* ! XNU_TARGET_OS_OSX */
2784
2785 lck_mtx_lock(&hostname_lock);
2786 strlcpy(tmpname, name, sizeof(tmpname));
2787 lck_mtx_unlock(&hostname_lock);
2788
2789 error = sysctl_io_string(req, tmpname, sizeof(tmpname), 1, &changed);
2790 if (!error && changed) {
2791 lck_mtx_lock(&hostname_lock);
2792 strlcpy(hostname, tmpname, sizeof(hostname));
2793 lck_mtx_unlock(&hostname_lock);
2794 }
2795 return error;
2796 }
2797
2798 SYSCTL_PROC(_kern, KERN_HOSTNAME, hostname,
2799 CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_LOCKED,
2800 0, 0, sysctl_hostname, "A", "");
2801
2802 STATIC int
2803 sysctl_procname
2804 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2805 {
2806 /* Original code allowed writing, I'm copying this, although this all makes
2807 * no sense to me. Besides, this sysctl is never used. */
2808 return sysctl_io_string(req, &req->p->p_name[0], (2 * MAXCOMLEN + 1), 1, NULL);
2809 }
2810
2811 SYSCTL_PROC(_kern, KERN_PROCNAME, procname,
2812 CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
2813 0, 0, sysctl_procname, "A", "");
2814
2815 SYSCTL_INT(_kern, KERN_SPECULATIVE_READS, speculative_reads_disabled,
2816 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2817 &speculative_reads_disabled, 0, "");
2818
2819 SYSCTL_UINT(_kern, OID_AUTO, preheat_max_bytes,
2820 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2821 &preheat_max_bytes, 0, "");
2822
2823 SYSCTL_UINT(_kern, OID_AUTO, preheat_min_bytes,
2824 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2825 &preheat_min_bytes, 0, "");
2826
2827 SYSCTL_UINT(_kern, OID_AUTO, speculative_prefetch_max,
2828 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2829 &speculative_prefetch_max, 0, "");
2830
2831 SYSCTL_UINT(_kern, OID_AUTO, speculative_prefetch_max_iosize,
2832 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2833 &speculative_prefetch_max_iosize, 0, "");
2834
2835 SYSCTL_UINT(_kern, OID_AUTO, vm_page_free_target,
2836 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2837 &vm_page_free_target, 0, "");
2838
2839 SYSCTL_UINT(_kern, OID_AUTO, vm_page_free_min,
2840 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2841 &vm_page_free_min, 0, "");
2842
2843 SYSCTL_UINT(_kern, OID_AUTO, vm_page_free_reserved,
2844 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2845 &vm_page_free_reserved, 0, "");
2846
2847 SYSCTL_UINT(_kern, OID_AUTO, vm_page_speculative_percentage,
2848 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2849 &vm_pageout_state.vm_page_speculative_percentage, 0, "");
2850
2851 SYSCTL_UINT(_kern, OID_AUTO, vm_page_speculative_q_age_ms,
2852 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2853 &vm_pageout_state.vm_page_speculative_q_age_ms, 0, "");
2854
2855 #if (DEVELOPMENT || DEBUG)
2856 SYSCTL_UINT(_kern, OID_AUTO, vm_page_max_speculative_age_q,
2857 CTLFLAG_RD,
2858 &vm_page_max_speculative_age_q, 0, "");
2859 #endif /* (DEVELOPMENT || DEBUG) */
2860
2861 SYSCTL_UINT(_kern, OID_AUTO, vm_max_delayed_work_limit,
2862 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2863 &vm_max_delayed_work_limit, 0, "");
2864
2865 SYSCTL_UINT(_kern, OID_AUTO, vm_max_batch,
2866 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2867 &vm_max_batch, 0, "");
2868
2869 SYSCTL_STRING(_kern, OID_AUTO, bootsessionuuid,
2870 CTLFLAG_RD | CTLFLAG_LOCKED,
2871 &bootsessionuuid_string, sizeof(bootsessionuuid_string), "");
2872
2873
2874 STATIC int
2875 sysctl_boottime
2876 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2877 {
2878 struct timeval tv;
2879 boottime_timeval(&tv);
2880 struct proc *p = req->p;
2881
2882 if (proc_is64bit(p)) {
2883 struct user64_timeval t = {};
2884 t.tv_sec = tv.tv_sec;
2885 t.tv_usec = tv.tv_usec;
2886 return sysctl_io_opaque(req, &t, sizeof(t), NULL);
2887 } else {
2888 struct user32_timeval t = {};
2889 t.tv_sec = (user32_time_t)tv.tv_sec;
2890 t.tv_usec = tv.tv_usec;
2891 return sysctl_io_opaque(req, &t, sizeof(t), NULL);
2892 }
2893 }
2894
2895 SYSCTL_PROC(_kern, KERN_BOOTTIME, boottime,
2896 CTLTYPE_STRUCT | CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
2897 0, 0, sysctl_boottime, "S,timeval", "");
2898
2899 extern bool IOGetBootUUID(char *);
2900
2901 /* non-static: written by imageboot.c */
2902 uuid_string_t fake_bootuuid;
2903
2904 STATIC int
2905 sysctl_bootuuid
2906 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2907 {
2908 int error = ENOENT;
2909
2910 /* check the first byte to see if the string has been
2911 * populated. this is a uuid_STRING_t, this check would
2912 * not work with a uuid_t.
2913 */
2914 if (fake_bootuuid[0] != '\0') {
2915 error = sysctl_io_string(req, fake_bootuuid, 0, 0, NULL);
2916 goto out;
2917 }
2918
2919 uuid_string_t uuid_string;
2920 if (IOGetBootUUID(uuid_string)) {
2921 uuid_t boot_uuid;
2922 error = uuid_parse(uuid_string, boot_uuid);
2923 if (!error) {
2924 error = sysctl_io_string(req, __DECONST(char *, uuid_string), 0, 0, NULL);
2925 }
2926 }
2927
2928 out:
2929 return error;
2930 }
2931
2932 SYSCTL_PROC(_kern, OID_AUTO, bootuuid,
2933 CTLTYPE_STRING | CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
2934 0, 0, sysctl_bootuuid, "A", "");
2935
2936
2937 extern bool IOGetApfsPrebootUUID(char *);
2938 extern bool IOGetAssociatedApfsVolgroupUUID(char *);
2939
2940 STATIC int
2941 sysctl_apfsprebootuuid
2942 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2943 {
2944 int error = ENOENT;
2945
2946 uuid_string_t uuid_string;
2947 if (IOGetApfsPrebootUUID(uuid_string)) {
2948 uuid_t apfs_preboot_uuid;
2949 error = uuid_parse(uuid_string, apfs_preboot_uuid);
2950 if (!error) {
2951 error = sysctl_io_string(req, __DECONST(char *, uuid_string), 0, 0, NULL);
2952 }
2953 }
2954
2955 return error;
2956 }
2957
2958 SYSCTL_PROC(_kern, OID_AUTO, apfsprebootuuid,
2959 CTLTYPE_STRING | CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
2960 0, 0, sysctl_apfsprebootuuid, "A", "");
2961
2962 STATIC int
2963 sysctl_targetsystemvolgroupuuid
2964 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2965 {
2966 int error = ENOENT;
2967
2968 uuid_string_t uuid_string;
2969 if (IOGetApfsPrebootUUID(uuid_string)) {
2970 uuid_t apfs_preboot_uuid;
2971 error = uuid_parse(uuid_string, apfs_preboot_uuid);
2972 if (!error) {
2973 error = sysctl_io_string(req, __DECONST(char *, uuid_string), 0, 0, NULL);
2974 }
2975 } else {
2976 /*
2977 * In special boot modes, such as kcgen-mode, the
2978 * apfs-preboot-uuid property will not be set. Instead, a
2979 * different property, associated-volume-group, will be set
2980 * which indicates the UUID of the VolumeGroup containing the
2981 * system volume into which you will boot.
2982 */
2983 if (IOGetAssociatedApfsVolgroupUUID(uuid_string)) {
2984 uuid_t apfs_preboot_uuid;
2985 error = uuid_parse(uuid_string, apfs_preboot_uuid);
2986 if (!error) {
2987 error = sysctl_io_string(req, __DECONST(char *, uuid_string), 0, 0, NULL);
2988 }
2989 }
2990 }
2991
2992 return error;
2993 }
2994
2995 SYSCTL_PROC(_kern, OID_AUTO, targetsystemvolgroupuuid,
2996 CTLTYPE_STRING | CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_MASKED | CTLFLAG_LOCKED,
2997 0, 0, sysctl_targetsystemvolgroupuuid, "A", "");
2998
2999
3000 extern bool IOGetBootManifestHash(char *, size_t *);
3001 extern bool IOGetBootObjectsPath(char *);
3002
3003 STATIC int
3004 sysctl_bootobjectspath
3005 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3006 {
3007 int error = ENOENT;
3008
3009 #if defined(__x86_64__)
3010 /* auth-root-dmg is used for the Intel BaseSystem in some flows,
3011 * e.g. createinstallmedia and as part of upgrading from 10.15 or earlier
3012 * under these scenarios, set_fake_bootuuid will be called when pivoting to
3013 * the new root filesystem. need honor the fake bootuuid.
3014 */
3015 if (fake_bootuuid[0] != '\0') {
3016 error = sysctl_io_string(req, fake_bootuuid, 0, 0, NULL);
3017 } else {
3018 /* for intel mac, boot objects reside in [preboot volume]/[bootuuid]
3019 * bootuuid and apfsprebootuuid are populated by efiboot and they are alias.
3020 */
3021 uuid_string_t uuid_string;
3022 if (IOGetBootUUID(uuid_string)) {
3023 uuid_t boot_uuid;
3024 error = uuid_parse(uuid_string, boot_uuid);
3025 if (!error) {
3026 error = sysctl_io_string(req, (char *)uuid_string, 0, 0, NULL);
3027 }
3028 }
3029 }
3030 #else
3031 char boot_obj_path[MAXPATHLEN] = { "\0" };
3032 static const char kAsciiHexChars[] = "0123456789ABCDEF";
3033 unsigned int i, j;
3034
3035 /* Hashed with SHA2-384 or SHA1, boot manifest hash is 48 bytes or 20 bytes
3036 * hence, need a 97 bytes char array for the string.
3037 */
3038 size_t hash_data_size = CCSHA384_OUTPUT_SIZE;
3039 char hash_data[CCSHA384_OUTPUT_SIZE] = { "\0" };
3040 char boot_manifest_hash[CCSHA384_OUTPUT_SIZE * 2 + 1] = { "\0" };;
3041
3042 /* for Apple Silicon Macs, there is a boot-objects-path under IODeviceTree:/chosen
3043 * and boot objects reside in [preboot volume]/[boot-objects-path]
3044 * for embedded platforms, there would be a boot-manifest-hash under IODeviceTree:/chosen
3045 * and boot objects reside in [preboot volume]/[boot-manifest-hash]
3046 */
3047 if (IOGetBootObjectsPath(boot_obj_path)) {
3048 error = sysctl_io_string(req, (char *)boot_obj_path, 0, 0, NULL);
3049 } else if (IOGetBootManifestHash(hash_data, &hash_data_size)) {
3050 j = 0;
3051 for (i = 0; i < hash_data_size; ++i) {
3052 char octet = hash_data[i];
3053 boot_manifest_hash[j++] = kAsciiHexChars[((octet & 0xF0) >> 4)];
3054 boot_manifest_hash[j++] = kAsciiHexChars[(octet & 0x0F)];
3055 }
3056 /* make sure string has null termination */
3057 boot_manifest_hash[j] = '\0';
3058 error = sysctl_io_string(req, (char *)boot_manifest_hash, 0, 0, NULL);
3059 }
3060 #endif
3061 return error;
3062 }
3063
3064 SYSCTL_PROC(_kern, OID_AUTO, bootobjectspath,
3065 CTLTYPE_STRING | CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
3066 0, 0, sysctl_bootobjectspath, "A", "");
3067
3068
3069 STATIC int
3070 sysctl_symfile
3071 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3072 {
3073 char *str;
3074 int error = get_kernel_symfile(req->p, &str);
3075 if (error) {
3076 return error;
3077 }
3078 return sysctl_io_string(req, str, 0, 0, NULL);
3079 }
3080
3081
3082 SYSCTL_PROC(_kern, KERN_SYMFILE, symfile,
3083 CTLTYPE_STRING | CTLFLAG_RD | CTLFLAG_LOCKED,
3084 0, 0, sysctl_symfile, "A", "");
3085
3086 #if CONFIG_NETBOOT
3087 STATIC int
3088 sysctl_netboot
3089 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3090 {
3091 return sysctl_io_number(req, netboot_root(), sizeof(int), NULL, NULL);
3092 }
3093
3094 SYSCTL_PROC(_kern, KERN_NETBOOT, netboot,
3095 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
3096 0, 0, sysctl_netboot, "I", "");
3097 #endif
3098
3099 #ifdef CONFIG_IMGSRC_ACCESS
3100 /*
3101 * Legacy--act as if only one layer of nesting is possible.
3102 */
3103 STATIC int
3104 sysctl_imgsrcdev
3105 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3106 {
3107 vfs_context_t ctx = vfs_context_current();
3108 vnode_t devvp;
3109 int result;
3110
3111 if (!vfs_context_issuser(ctx)) {
3112 return EPERM;
3113 }
3114
3115 if (imgsrc_rootvnodes[0] == NULL) {
3116 return ENOENT;
3117 }
3118
3119 result = vnode_getwithref(imgsrc_rootvnodes[0]);
3120 if (result != 0) {
3121 return result;
3122 }
3123
3124 devvp = vnode_mount(imgsrc_rootvnodes[0])->mnt_devvp;
3125 result = vnode_getwithref(devvp);
3126 if (result != 0) {
3127 goto out;
3128 }
3129
3130 result = sysctl_io_number(req, vnode_specrdev(devvp), sizeof(dev_t), NULL, NULL);
3131
3132 vnode_put(devvp);
3133 out:
3134 vnode_put(imgsrc_rootvnodes[0]);
3135 return result;
3136 }
3137
3138 SYSCTL_PROC(_kern, OID_AUTO, imgsrcdev,
3139 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
3140 0, 0, sysctl_imgsrcdev, "I", "");
3141
3142 STATIC int
3143 sysctl_imgsrcinfo
3144 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3145 {
3146 int error;
3147 struct imgsrc_info info[MAX_IMAGEBOOT_NESTING] = {}; /* 2 for now, no problem */
3148 uint32_t i;
3149 vnode_t rvp, devvp;
3150
3151 if (imgsrc_rootvnodes[0] == NULLVP) {
3152 return ENXIO;
3153 }
3154
3155 for (i = 0; i < MAX_IMAGEBOOT_NESTING; i++) {
3156 /*
3157 * Go get the root vnode.
3158 */
3159 rvp = imgsrc_rootvnodes[i];
3160 if (rvp == NULLVP) {
3161 break;
3162 }
3163
3164 error = vnode_get(rvp);
3165 if (error != 0) {
3166 return error;
3167 }
3168
3169 /*
3170 * For now, no getting at a non-local volume.
3171 */
3172 devvp = vnode_mount(rvp)->mnt_devvp;
3173 if (devvp == NULL) {
3174 vnode_put(rvp);
3175 return EINVAL;
3176 }
3177
3178 error = vnode_getwithref(devvp);
3179 if (error != 0) {
3180 vnode_put(rvp);
3181 return error;
3182 }
3183
3184 /*
3185 * Fill in info.
3186 */
3187 info[i].ii_dev = vnode_specrdev(devvp);
3188 info[i].ii_flags = 0;
3189 info[i].ii_height = i;
3190 bzero(info[i].ii_reserved, sizeof(info[i].ii_reserved));
3191
3192 vnode_put(devvp);
3193 vnode_put(rvp);
3194 }
3195
3196 return sysctl_io_opaque(req, info, i * sizeof(info[0]), NULL);
3197 }
3198
3199 SYSCTL_PROC(_kern, OID_AUTO, imgsrcinfo,
3200 CTLTYPE_OPAQUE | CTLFLAG_RD | CTLFLAG_LOCKED,
3201 0, 0, sysctl_imgsrcinfo, "I", "");
3202
3203 #endif /* CONFIG_IMGSRC_ACCESS */
3204
3205
3206 SYSCTL_DECL(_kern_timer);
3207 SYSCTL_NODE(_kern, OID_AUTO, timer, CTLFLAG_RW | CTLFLAG_LOCKED, 0, "timer");
3208
3209
3210 SYSCTL_INT(_kern_timer, OID_AUTO, coalescing_enabled,
3211 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
3212 &mach_timer_coalescing_enabled, 0, "");
3213
3214 SYSCTL_QUAD(_kern_timer, OID_AUTO, deadline_tracking_bin_1,
3215 CTLFLAG_RW | CTLFLAG_LOCKED,
3216 &timer_deadline_tracking_bin_1, "");
3217 SYSCTL_QUAD(_kern_timer, OID_AUTO, deadline_tracking_bin_2,
3218 CTLFLAG_RW | CTLFLAG_LOCKED,
3219 &timer_deadline_tracking_bin_2, "");
3220
3221 SYSCTL_DECL(_kern_timer_longterm);
3222 SYSCTL_NODE(_kern_timer, OID_AUTO, longterm, CTLFLAG_RW | CTLFLAG_LOCKED, 0, "longterm");
3223
3224
3225 /* Must match definition in osfmk/kern/timer_call.c */
3226 enum {
3227 THRESHOLD, QCOUNT,
3228 ENQUEUES, DEQUEUES, ESCALATES, SCANS, PREEMPTS,
3229 LATENCY, LATENCY_MIN, LATENCY_MAX, LONG_TERM_SCAN_LIMIT,
3230 LONG_TERM_SCAN_INTERVAL, LONG_TERM_SCAN_PAUSES,
3231 SCAN_LIMIT, SCAN_INTERVAL, SCAN_PAUSES, SCAN_POSTPONES,
3232 };
3233 extern uint64_t timer_sysctl_get(int);
3234 extern kern_return_t timer_sysctl_set(int, uint64_t);
3235
3236 STATIC int
3237 sysctl_timer
3238 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3239 {
3240 int oid = (int)arg1;
3241 uint64_t value = timer_sysctl_get(oid);
3242 uint64_t new_value;
3243 int error;
3244 int changed;
3245
3246 error = sysctl_io_number(req, value, sizeof(value), &new_value, &changed);
3247 if (changed) {
3248 kern_return_t kr = timer_sysctl_set(oid, new_value);
3249 error = mach_to_bsd_errno(kr);
3250 }
3251
3252 return error;
3253 }
3254
3255 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, threshold,
3256 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
3257 (void *) THRESHOLD, 0, sysctl_timer, "Q", "");
3258 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, scan_limit,
3259 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
3260 (void *) LONG_TERM_SCAN_LIMIT, 0, sysctl_timer, "Q", "");
3261 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, scan_interval,
3262 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
3263 (void *) LONG_TERM_SCAN_INTERVAL, 0, sysctl_timer, "Q", "");
3264
3265 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, qlen,
3266 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3267 (void *) QCOUNT, 0, sysctl_timer, "Q", "");
3268 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, scan_pauses,
3269 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3270 (void *) LONG_TERM_SCAN_PAUSES, 0, sysctl_timer, "Q", "");
3271
3272 #if DEBUG
3273 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, enqueues,
3274 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3275 (void *) ENQUEUES, 0, sysctl_timer, "Q", "");
3276 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, dequeues,
3277 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3278 (void *) DEQUEUES, 0, sysctl_timer, "Q", "");
3279 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, escalates,
3280 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3281 (void *) ESCALATES, 0, sysctl_timer, "Q", "");
3282 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, scans,
3283 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3284 (void *) SCANS, 0, sysctl_timer, "Q", "");
3285 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, preempts,
3286 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3287 (void *) PREEMPTS, 0, sysctl_timer, "Q", "");
3288 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, latency,
3289 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3290 (void *) LATENCY, 0, sysctl_timer, "Q", "");
3291 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, latency_min,
3292 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3293 (void *) LATENCY_MIN, 0, sysctl_timer, "Q", "");
3294 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, latency_max,
3295 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3296 (void *) LATENCY_MAX, 0, sysctl_timer, "Q", "");
3297 #endif /* DEBUG */
3298
3299 SYSCTL_PROC(_kern_timer, OID_AUTO, scan_limit,
3300 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
3301 (void *) SCAN_LIMIT, 0, sysctl_timer, "Q", "");
3302 SYSCTL_PROC(_kern_timer, OID_AUTO, scan_interval,
3303 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
3304 (void *) SCAN_INTERVAL, 0, sysctl_timer, "Q", "");
3305 SYSCTL_PROC(_kern_timer, OID_AUTO, scan_pauses,
3306 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3307 (void *) SCAN_PAUSES, 0, sysctl_timer, "Q", "");
3308 SYSCTL_PROC(_kern_timer, OID_AUTO, scan_postpones,
3309 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3310 (void *) SCAN_POSTPONES, 0, sysctl_timer, "Q", "");
3311
3312 STATIC int
3313 sysctl_usrstack
3314 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3315 {
3316 return sysctl_io_number(req, (int)req->p->user_stack, sizeof(int), NULL, NULL);
3317 }
3318
3319 SYSCTL_PROC(_kern, KERN_USRSTACK32, usrstack,
3320 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
3321 0, 0, sysctl_usrstack, "I", "");
3322
3323 STATIC int
3324 sysctl_usrstack64
3325 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3326 {
3327 return sysctl_io_number(req, req->p->user_stack, sizeof(req->p->user_stack), NULL, NULL);
3328 }
3329
3330 SYSCTL_PROC(_kern, KERN_USRSTACK64, usrstack64,
3331 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3332 0, 0, sysctl_usrstack64, "Q", "");
3333
3334
3335 #if EXCLAVES_COREDUMP
3336
3337 /* secure kernel coredump support. */
3338 extern unsigned int sc_dump_mode;
3339 SYSCTL_UINT(_kern, OID_AUTO, secure_coredump, CTLFLAG_RD, &sc_dump_mode, 0, "secure_coredump");
3340
3341 #endif /* EXCLAVES_COREDUMP */
3342
3343
3344 #if CONFIG_COREDUMP
3345
3346 SYSCTL_STRING(_kern, KERN_COREFILE, corefile,
3347 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
3348 corefilename, sizeof(corefilename), "");
3349
3350 SYSCTL_STRING(_kern, OID_AUTO, drivercorefile,
3351 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
3352 drivercorefilename, sizeof(drivercorefilename), "");
3353
3354 STATIC int
3355 sysctl_coredump
3356 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3357 {
3358 #ifdef SECURE_KERNEL
3359 (void)req;
3360 return ENOTSUP;
3361 #else
3362 int new_value, changed;
3363 int error = sysctl_io_number(req, do_coredump, sizeof(int), &new_value, &changed);
3364 if (changed) {
3365 if ((new_value == 0) || (new_value == 1)) {
3366 do_coredump = new_value;
3367 } else {
3368 error = EINVAL;
3369 }
3370 }
3371 return error;
3372 #endif
3373 }
3374
3375 SYSCTL_PROC(_kern, KERN_COREDUMP, coredump,
3376 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
3377 0, 0, sysctl_coredump, "I", "");
3378
3379 STATIC int
3380 sysctl_suid_coredump
3381 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3382 {
3383 #ifdef SECURE_KERNEL
3384 (void)req;
3385 return ENOTSUP;
3386 #else
3387 int new_value, changed;
3388 int error = sysctl_io_number(req, sugid_coredump, sizeof(int), &new_value, &changed);
3389 if (changed) {
3390 if ((new_value == 0) || (new_value == 1)) {
3391 sugid_coredump = new_value;
3392 } else {
3393 error = EINVAL;
3394 }
3395 }
3396 return error;
3397 #endif
3398 }
3399
3400 SYSCTL_PROC(_kern, KERN_SUGID_COREDUMP, sugid_coredump,
3401 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
3402 0, 0, sysctl_suid_coredump, "I", "");
3403
3404 #endif /* CONFIG_COREDUMP */
3405
3406 #if CONFIG_KDP_INTERACTIVE_DEBUGGING
3407
3408 extern const char* kdp_corefile_path;
3409 STATIC int
3410 sysctl_kdp_corefile(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3411 {
3412 return SYSCTL_OUT(req, kdp_corefile_path, strlen(kdp_corefile_path) + 1);
3413 }
3414
3415 /* this needs to be a proc rather than a string since kdp_corefile_path is not a compile-time constant */
3416 SYSCTL_PROC(_kern, OID_AUTO, kdp_corefile,
3417 CTLTYPE_STRING | CTLFLAG_RD | CTLFLAG_LOCKED,
3418 0, 0, sysctl_kdp_corefile, "A", "");
3419
3420 #endif /* CONFIG_KDP_INTERACTIVE_DEBUGGING */
3421
3422 STATIC int
3423 sysctl_delayterm
3424 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3425 {
3426 struct proc *p = req->p;
3427 int new_value, changed;
3428 int error = sysctl_io_number(req, (req->p->p_lflag & P_LDELAYTERM)? 1: 0, sizeof(int), &new_value, &changed);
3429 if (changed) {
3430 proc_lock(p);
3431 if (new_value) {
3432 req->p->p_lflag |= P_LDELAYTERM;
3433 } else {
3434 req->p->p_lflag &= ~P_LDELAYTERM;
3435 }
3436 proc_unlock(p);
3437 }
3438 return error;
3439 }
3440
3441 SYSCTL_PROC(_kern, KERN_PROCDELAYTERM, delayterm,
3442 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
3443 0, 0, sysctl_delayterm, "I", "");
3444
3445
3446 STATIC int
3447 sysctl_rage_vnode
3448 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3449 {
3450 struct proc *p = req->p;
3451 struct uthread *ut;
3452 int new_value, old_value, changed;
3453 int error;
3454
3455 ut = current_uthread();
3456
3457 if (ut->uu_flag & UT_RAGE_VNODES) {
3458 old_value = KERN_RAGE_THREAD;
3459 } else if (p->p_lflag & P_LRAGE_VNODES) {
3460 old_value = KERN_RAGE_PROC;
3461 } else {
3462 old_value = 0;
3463 }
3464
3465 error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
3466
3467 if ((error == 0) && (changed != 0)) {
3468 switch (new_value) {
3469 case KERN_RAGE_PROC:
3470 proc_lock(p);
3471 p->p_lflag |= P_LRAGE_VNODES;
3472 proc_unlock(p);
3473 break;
3474 case KERN_UNRAGE_PROC:
3475 proc_lock(p);
3476 p->p_lflag &= ~P_LRAGE_VNODES;
3477 proc_unlock(p);
3478 break;
3479
3480 case KERN_RAGE_THREAD:
3481 ut->uu_flag |= UT_RAGE_VNODES;
3482 break;
3483 case KERN_UNRAGE_THREAD:
3484 ut = current_uthread();
3485 ut->uu_flag &= ~UT_RAGE_VNODES;
3486 break;
3487 }
3488 }
3489 return error;
3490 }
3491
3492 SYSCTL_PROC(_kern, KERN_RAGEVNODE, rage_vnode,
3493 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
3494 0, 0, sysctl_rage_vnode, "I", "");
3495
3496 /* XXX until filecoordinationd fixes a bit of inverted logic. */
3497 STATIC int
3498 sysctl_vfsnspace
3499 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3500 {
3501 int old_value = 0, new_value, changed;
3502
3503 return sysctl_io_number(req, old_value, sizeof(int), &new_value,
3504 &changed);
3505 }
3506
3507 SYSCTL_PROC(_kern, OID_AUTO, vfsnspace,
3508 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
3509 0, 0, sysctl_vfsnspace, "I", "");
3510
3511 /* XXX move this interface into libproc and remove this sysctl */
3512 STATIC int
3513 sysctl_setthread_cpupercent
3514 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3515 {
3516 int new_value, old_value;
3517 int error = 0;
3518 kern_return_t kret = KERN_SUCCESS;
3519 uint8_t percent = 0;
3520 int ms_refill = 0;
3521
3522 if (!req->newptr) {
3523 return 0;
3524 }
3525
3526 old_value = 0;
3527
3528 if ((error = sysctl_io_number(req, old_value, sizeof(old_value), &new_value, NULL)) != 0) {
3529 return error;
3530 }
3531
3532 percent = new_value & 0xff; /* low 8 bytes for perent */
3533 ms_refill = (new_value >> 8) & 0xffffff; /* upper 24bytes represent ms refill value */
3534 if (percent > 100) {
3535 return EINVAL;
3536 }
3537
3538 /*
3539 * If the caller is specifying a percentage of 0, this will unset the CPU limit, if present.
3540 */
3541 kret = percent == 0 ?
3542 thread_set_cpulimit(THREAD_CPULIMIT_DISABLE, 0, 0) :
3543 thread_set_cpulimit(THREAD_CPULIMIT_BLOCK, percent, ms_refill * (int)NSEC_PER_MSEC);
3544
3545 if (kret != 0) {
3546 return EIO;
3547 }
3548
3549 return 0;
3550 }
3551
3552 SYSCTL_PROC(_kern, OID_AUTO, setthread_cpupercent,
3553 CTLTYPE_INT | CTLFLAG_WR | CTLFLAG_ANYBODY,
3554 0, 0, sysctl_setthread_cpupercent, "I", "set thread cpu percentage limit");
3555
3556
3557 STATIC int
3558 sysctl_kern_check_openevt
3559 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3560 {
3561 struct proc *p = req->p;
3562 int new_value, old_value, changed;
3563 int error;
3564
3565 if (p->p_flag & P_CHECKOPENEVT) {
3566 old_value = KERN_OPENEVT_PROC;
3567 } else {
3568 old_value = 0;
3569 }
3570
3571 error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
3572
3573 if ((error == 0) && (changed != 0)) {
3574 switch (new_value) {
3575 case KERN_OPENEVT_PROC:
3576 OSBitOrAtomic(P_CHECKOPENEVT, &p->p_flag);
3577 break;
3578
3579 case KERN_UNOPENEVT_PROC:
3580 OSBitAndAtomic(~((uint32_t)P_CHECKOPENEVT), &p->p_flag);
3581 break;
3582
3583 default:
3584 error = EINVAL;
3585 }
3586 }
3587 return error;
3588 }
3589
3590 SYSCTL_PROC(_kern, KERN_CHECKOPENEVT, check_openevt, CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
3591 0, 0, sysctl_kern_check_openevt, "I", "set the per-process check-open-evt flag");
3592
3593
3594 #if DEVELOPMENT || DEBUG
3595 STATIC int
3596 sysctl_nx
3597 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3598 {
3599 #ifdef SECURE_KERNEL
3600 (void)req;
3601 return ENOTSUP;
3602 #else
3603 int new_value, changed;
3604 int error;
3605
3606 error = sysctl_io_number(req, nx_enabled, sizeof(nx_enabled), &new_value, &changed);
3607 if (error) {
3608 return error;
3609 }
3610
3611 if (changed) {
3612 #if defined(__x86_64__)
3613 /*
3614 * Only allow setting if NX is supported on the chip
3615 */
3616 if (!(cpuid_extfeatures() & CPUID_EXTFEATURE_XD)) {
3617 return ENOTSUP;
3618 }
3619 #endif
3620 nx_enabled = new_value;
3621 }
3622 return error;
3623 #endif /* SECURE_KERNEL */
3624 }
3625 #endif
3626
3627 #if DEVELOPMENT || DEBUG
3628 SYSCTL_PROC(_kern, KERN_NX_PROTECTION, nx,
3629 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
3630 0, 0, sysctl_nx, "I", "");
3631 #endif
3632
3633 STATIC int
3634 sysctl_loadavg
3635 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3636 {
3637 if (proc_is64bit(req->p)) {
3638 struct user64_loadavg loadinfo64 = {};
3639 fill_loadavg64(&averunnable, &loadinfo64);
3640 return sysctl_io_opaque(req, &loadinfo64, sizeof(loadinfo64), NULL);
3641 } else {
3642 struct user32_loadavg loadinfo32 = {};
3643 fill_loadavg32(&averunnable, &loadinfo32);
3644 return sysctl_io_opaque(req, &loadinfo32, sizeof(loadinfo32), NULL);
3645 }
3646 }
3647
3648 SYSCTL_PROC(_vm, VM_LOADAVG, loadavg,
3649 CTLTYPE_STRUCT | CTLFLAG_RD | CTLFLAG_LOCKED,
3650 0, 0, sysctl_loadavg, "S,loadavg", "");
3651
3652 /*
3653 * Note: Thread safe; vm_map_lock protects in vm_toggle_entry_reuse()
3654 */
3655 STATIC int
3656 sysctl_vm_toggle_address_reuse(__unused struct sysctl_oid *oidp, __unused void *arg1,
3657 __unused int arg2, struct sysctl_req *req)
3658 {
3659 int old_value = 0, new_value = 0, error = 0;
3660
3661 if (vm_toggle_entry_reuse( VM_TOGGLE_GETVALUE, &old_value )) {
3662 return error;
3663 }
3664 error = sysctl_io_number(req, old_value, sizeof(int), &new_value, NULL);
3665 if (!error) {
3666 return vm_toggle_entry_reuse(new_value, NULL);
3667 }
3668 return error;
3669 }
3670
3671 SYSCTL_PROC(_debug, OID_AUTO, toggle_address_reuse, CTLFLAG_ANYBODY | CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED, 0, 0, sysctl_vm_toggle_address_reuse, "I", "");
3672
3673 #ifdef CONFIG_XNUPOST
3674
3675 extern uint32_t xnupost_get_estimated_testdata_size(void);
3676 extern int xnupost_reset_all_tests(void);
3677
3678 STATIC int
3679 sysctl_handle_xnupost_get_tests SYSCTL_HANDLER_ARGS
3680 {
3681 /* fixup unused arguments warnings */
3682 __unused int _oa2 = arg2;
3683 __unused void * _oa1 = arg1;
3684 __unused struct sysctl_oid * _oidp = oidp;
3685
3686 int error = 0;
3687 user_addr_t oldp = 0;
3688 user_addr_t newp = 0;
3689 uint32_t usedbytes = 0;
3690
3691 oldp = req->oldptr;
3692 newp = req->newptr;
3693
3694 if (newp) {
3695 return ENOTSUP;
3696 }
3697
3698 if ((void *)oldp == NULL) {
3699 /* return estimated size for second call where info can be placed */
3700 req->oldidx = xnupost_get_estimated_testdata_size();
3701 } else {
3702 error = xnupost_export_testdata((void *)oldp, req->oldlen, &usedbytes);
3703 req->oldidx = usedbytes;
3704 }
3705
3706 return error;
3707 }
3708
3709 SYSCTL_PROC(_debug,
3710 OID_AUTO,
3711 xnupost_get_tests,
3712 CTLFLAG_MASKED | CTLFLAG_ANYBODY | CTLTYPE_OPAQUE | CTLFLAG_RD | CTLFLAG_LOCKED,
3713 0,
3714 0,
3715 sysctl_handle_xnupost_get_tests,
3716 "-",
3717 "read xnupost test data in kernel");
3718
3719 #if CONFIG_EXT_PANICLOG
3720 /*
3721 * Extensible panic log test hooks
3722 */
3723 static int
3724 sysctl_debug_ext_paniclog_test_hook SYSCTL_HANDLER_ARGS
3725 {
3726 #pragma unused(arg1, arg2)
3727 int rval = 0;
3728 uint32_t test_option = 0;
3729
3730 rval = sysctl_handle_int(oidp, &test_option, 0, req);
3731
3732 if (rval == 0 && req->newptr) {
3733 rval = ext_paniclog_test_hook(test_option);
3734 }
3735
3736 return rval;
3737 }
3738
3739 SYSCTL_PROC(_debug, OID_AUTO, ext_paniclog_test_hook,
3740 CTLTYPE_INT | CTLFLAG_RW,
3741 0, 0,
3742 sysctl_debug_ext_paniclog_test_hook, "A", "ext paniclog test hook");
3743
3744 #endif
3745
3746 STATIC int
3747 sysctl_debug_xnupost_ctl SYSCTL_HANDLER_ARGS
3748 {
3749 /* fixup unused arguments warnings */
3750 __unused int _oa2 = arg2;
3751 __unused void * _oa1 = arg1;
3752 __unused struct sysctl_oid * _oidp = oidp;
3753
3754 #define ARRCOUNT 4
3755 /*
3756 * INPUT: ACTION, PARAM1, PARAM2, PARAM3
3757 * OUTPUT: RESULTCODE, ADDITIONAL DATA
3758 */
3759 int32_t outval[ARRCOUNT] = {0};
3760 int32_t input[ARRCOUNT] = {0};
3761 int32_t out_size = sizeof(outval);
3762 int32_t in_size = sizeof(input);
3763 int error = 0;
3764
3765 /* if this is NULL call to find out size, send out size info */
3766 if (!req->newptr) {
3767 goto out;
3768 }
3769
3770 /* pull in provided value from userspace */
3771 error = SYSCTL_IN(req, &input[0], in_size);
3772 if (error) {
3773 return error;
3774 }
3775
3776 if (input[0] == XTCTL_RESET_TESTDATA) {
3777 outval[0] = xnupost_reset_all_tests();
3778 goto out;
3779 }
3780
3781 out:
3782 error = SYSCTL_OUT(req, &outval[0], out_size);
3783 return error;
3784 }
3785
3786 SYSCTL_PROC(_debug,
3787 OID_AUTO,
3788 xnupost_testctl,
3789 CTLFLAG_MASKED | CTLFLAG_ANYBODY | CTLTYPE_OPAQUE | CTLFLAG_RW | CTLFLAG_LOCKED,
3790 0,
3791 0,
3792 sysctl_debug_xnupost_ctl,
3793 "I",
3794 "xnupost control for kernel testing");
3795
3796 extern void test_oslog_handleOSLogCtl(int32_t * in, int32_t * out, int32_t arraycount);
3797
3798 STATIC int
3799 sysctl_debug_test_oslog_ctl(__unused struct sysctl_oid * oidp, __unused void * arg1, __unused int arg2, struct sysctl_req * req)
3800 {
3801 #define ARRCOUNT 4
3802 int32_t outval[ARRCOUNT] = {0};
3803 int32_t input[ARRCOUNT] = {0};
3804 int32_t size_outval = sizeof(outval);
3805 int32_t size_inval = sizeof(input);
3806 int32_t error;
3807
3808 /* if this is NULL call to find out size, send out size info */
3809 if (!req->newptr) {
3810 error = SYSCTL_OUT(req, &outval[0], size_outval);
3811 return error;
3812 }
3813
3814 /* pull in provided value from userspace */
3815 error = SYSCTL_IN(req, &input[0], size_inval);
3816 if (error) {
3817 return error;
3818 }
3819
3820 test_oslog_handleOSLogCtl(input, outval, ARRCOUNT);
3821
3822 error = SYSCTL_OUT(req, &outval[0], size_outval);
3823
3824 return error;
3825 }
3826
3827 SYSCTL_PROC(_debug,
3828 OID_AUTO,
3829 test_OSLogCtl,
3830 CTLFLAG_MASKED | CTLFLAG_ANYBODY | CTLTYPE_OPAQUE | CTLFLAG_RW | CTLFLAG_LOCKED,
3831 0,
3832 0,
3833 sysctl_debug_test_oslog_ctl,
3834 "I",
3835 "testing oslog in kernel");
3836
3837 #include <mach/task.h>
3838 #include <mach/semaphore.h>
3839
3840 static LCK_GRP_DECLARE(sysctl_debug_test_stackshot_owner_grp, "test-stackshot-owner-grp");
3841 static LCK_MTX_DECLARE(sysctl_debug_test_stackshot_owner_init_mtx,
3842 &sysctl_debug_test_stackshot_owner_grp);
3843
3844 /* This is a sysctl for testing collection of owner info on a lock in kernel space. A multi-threaded
3845 * test from userland sets this sysctl in such a way that a thread blocks in kernel mode, and a
3846 * stackshot is taken to see if the owner of the lock can be identified.
3847 *
3848 * We can't return to userland with a kernel lock held, so be sure to unlock before we leave.
3849 * the semaphores allow us to artificially create cases where the lock is being held and the
3850 * thread is hanging / taking a long time to do something. */
3851
3852 volatile char sysctl_debug_test_stackshot_mtx_inited = 0;
3853 semaphore_t sysctl_debug_test_stackshot_mutex_sem;
3854 lck_mtx_t sysctl_debug_test_stackshot_owner_lck;
3855
3856 #define SYSCTL_DEBUG_MTX_ACQUIRE_WAIT 1
3857 #define SYSCTL_DEBUG_MTX_ACQUIRE_NOWAIT 2
3858 #define SYSCTL_DEBUG_MTX_SIGNAL 3
3859 #define SYSCTL_DEBUG_MTX_TEARDOWN 4
3860
3861 STATIC int
3862 sysctl_debug_test_stackshot_mutex_owner(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3863 {
3864 long long option = -1;
3865 /* if the user tries to read the sysctl, we tell them what the address of the lock is (to test against stackshot's output) */
3866 long long mtx_unslid_addr = (long long)VM_KERNEL_UNSLIDE_OR_PERM(&sysctl_debug_test_stackshot_owner_lck);
3867 int error = sysctl_io_number(req, mtx_unslid_addr, sizeof(long long), (void*)&option, NULL);
3868
3869 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_init_mtx);
3870 if (!sysctl_debug_test_stackshot_mtx_inited) {
3871 lck_mtx_init(&sysctl_debug_test_stackshot_owner_lck,
3872 &sysctl_debug_test_stackshot_owner_grp,
3873 LCK_ATTR_NULL);
3874 semaphore_create(kernel_task,
3875 &sysctl_debug_test_stackshot_mutex_sem,
3876 SYNC_POLICY_FIFO, 0);
3877 sysctl_debug_test_stackshot_mtx_inited = 1;
3878 }
3879 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_init_mtx);
3880
3881 if (!error) {
3882 switch (option) {
3883 case SYSCTL_DEBUG_MTX_ACQUIRE_NOWAIT:
3884 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_lck);
3885 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_lck);
3886 break;
3887 case SYSCTL_DEBUG_MTX_ACQUIRE_WAIT:
3888 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_lck);
3889 semaphore_wait(sysctl_debug_test_stackshot_mutex_sem);
3890 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_lck);
3891 break;
3892 case SYSCTL_DEBUG_MTX_SIGNAL:
3893 semaphore_signal(sysctl_debug_test_stackshot_mutex_sem);
3894 break;
3895 case SYSCTL_DEBUG_MTX_TEARDOWN:
3896 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_init_mtx);
3897
3898 lck_mtx_destroy(&sysctl_debug_test_stackshot_owner_lck,
3899 &sysctl_debug_test_stackshot_owner_grp);
3900 semaphore_destroy(kernel_task,
3901 sysctl_debug_test_stackshot_mutex_sem);
3902 sysctl_debug_test_stackshot_mtx_inited = 0;
3903
3904 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_init_mtx);
3905 break;
3906 case -1: /* user just wanted to read the value, so do nothing */
3907 break;
3908 default:
3909 error = EINVAL;
3910 break;
3911 }
3912 }
3913 return error;
3914 }
3915
3916 /* we can't return to userland with a kernel rwlock held, so be sure to unlock before we leave.
3917 * the semaphores allow us to artificially create cases where the lock is being held and the
3918 * thread is hanging / taking a long time to do something. */
3919
3920 SYSCTL_PROC(_debug,
3921 OID_AUTO,
3922 test_MutexOwnerCtl,
3923 CTLFLAG_MASKED | CTLFLAG_ANYBODY | CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
3924 0,
3925 0,
3926 sysctl_debug_test_stackshot_mutex_owner,
3927 "-",
3928 "Testing mutex owner in kernel");
3929
3930 volatile char sysctl_debug_test_stackshot_rwlck_inited = 0;
3931 lck_rw_t sysctl_debug_test_stackshot_owner_rwlck;
3932 semaphore_t sysctl_debug_test_stackshot_rwlck_sem;
3933
3934 #define SYSCTL_DEBUG_KRWLCK_RACQUIRE_NOWAIT 1
3935 #define SYSCTL_DEBUG_KRWLCK_RACQUIRE_WAIT 2
3936 #define SYSCTL_DEBUG_KRWLCK_WACQUIRE_NOWAIT 3
3937 #define SYSCTL_DEBUG_KRWLCK_WACQUIRE_WAIT 4
3938 #define SYSCTL_DEBUG_KRWLCK_SIGNAL 5
3939 #define SYSCTL_DEBUG_KRWLCK_TEARDOWN 6
3940
3941 STATIC int
3942 sysctl_debug_test_stackshot_rwlck_owner(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3943 {
3944 long long option = -1;
3945 /* if the user tries to read the sysctl, we tell them what the address of the lock is
3946 * (to test against stackshot's output) */
3947 long long rwlck_unslid_addr = (long long)VM_KERNEL_UNSLIDE_OR_PERM(&sysctl_debug_test_stackshot_owner_rwlck);
3948 int error = sysctl_io_number(req, rwlck_unslid_addr, sizeof(long long), (void*)&option, NULL);
3949
3950 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_init_mtx);
3951 if (!sysctl_debug_test_stackshot_rwlck_inited) {
3952 lck_rw_init(&sysctl_debug_test_stackshot_owner_rwlck,
3953 &sysctl_debug_test_stackshot_owner_grp,
3954 LCK_ATTR_NULL);
3955 semaphore_create(kernel_task,
3956 &sysctl_debug_test_stackshot_rwlck_sem,
3957 SYNC_POLICY_FIFO,
3958 0);
3959 sysctl_debug_test_stackshot_rwlck_inited = 1;
3960 }
3961 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_init_mtx);
3962
3963 if (!error) {
3964 switch (option) {
3965 case SYSCTL_DEBUG_KRWLCK_RACQUIRE_NOWAIT:
3966 lck_rw_lock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_SHARED);
3967 lck_rw_unlock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_SHARED);
3968 break;
3969 case SYSCTL_DEBUG_KRWLCK_RACQUIRE_WAIT:
3970 lck_rw_lock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_SHARED);
3971 semaphore_wait(sysctl_debug_test_stackshot_rwlck_sem);
3972 lck_rw_unlock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_SHARED);
3973 break;
3974 case SYSCTL_DEBUG_KRWLCK_WACQUIRE_NOWAIT:
3975 lck_rw_lock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_EXCLUSIVE);
3976 lck_rw_unlock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_EXCLUSIVE);
3977 break;
3978 case SYSCTL_DEBUG_KRWLCK_WACQUIRE_WAIT:
3979 lck_rw_lock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_EXCLUSIVE);
3980 semaphore_wait(sysctl_debug_test_stackshot_rwlck_sem);
3981 lck_rw_unlock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_EXCLUSIVE);
3982 break;
3983 case SYSCTL_DEBUG_KRWLCK_SIGNAL:
3984 semaphore_signal(sysctl_debug_test_stackshot_rwlck_sem);
3985 break;
3986 case SYSCTL_DEBUG_KRWLCK_TEARDOWN:
3987 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_init_mtx);
3988
3989 lck_rw_destroy(&sysctl_debug_test_stackshot_owner_rwlck,
3990 &sysctl_debug_test_stackshot_owner_grp);
3991 semaphore_destroy(kernel_task,
3992 sysctl_debug_test_stackshot_rwlck_sem);
3993 sysctl_debug_test_stackshot_rwlck_inited = 0;
3994
3995 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_init_mtx);
3996 break;
3997 case -1: /* user just wanted to read the value, so do nothing */
3998 break;
3999 default:
4000 error = EINVAL;
4001 break;
4002 }
4003 }
4004 return error;
4005 }
4006
4007
4008 SYSCTL_PROC(_debug,
4009 OID_AUTO,
4010 test_RWLockOwnerCtl,
4011 CTLFLAG_MASKED | CTLFLAG_ANYBODY | CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
4012 0,
4013 0,
4014 sysctl_debug_test_stackshot_rwlck_owner,
4015 "-",
4016 "Testing rwlock owner in kernel");
4017 #endif /* !CONFIG_XNUPOST */
4018
4019 STATIC int
4020 sysctl_swapusage
4021 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4022 {
4023 int error;
4024 uint64_t swap_total;
4025 uint64_t swap_avail;
4026 vm_size_t swap_pagesize;
4027 boolean_t swap_encrypted;
4028 struct xsw_usage xsu = {};
4029
4030 error = macx_swapinfo(&swap_total,
4031 &swap_avail,
4032 &swap_pagesize,
4033 &swap_encrypted);
4034 if (error) {
4035 return error;
4036 }
4037
4038 xsu.xsu_total = swap_total;
4039 xsu.xsu_avail = swap_avail;
4040 xsu.xsu_used = swap_total - swap_avail;
4041 xsu.xsu_pagesize = (u_int32_t)MIN(swap_pagesize, UINT32_MAX);
4042 xsu.xsu_encrypted = swap_encrypted;
4043 return sysctl_io_opaque(req, &xsu, sizeof(xsu), NULL);
4044 }
4045
4046
4047
4048 SYSCTL_PROC(_vm, VM_SWAPUSAGE, swapusage,
4049 CTLTYPE_STRUCT | CTLFLAG_RD | CTLFLAG_LOCKED,
4050 0, 0, sysctl_swapusage, "S,xsw_usage", "");
4051
4052 extern int vm_swap_enabled;
4053 SYSCTL_INT(_vm, OID_AUTO, swap_enabled, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_swap_enabled, 0, "");
4054
4055 #if DEVELOPMENT || DEBUG
4056 extern int vm_num_swap_files_config;
4057 extern int vm_num_swap_files;
4058 extern lck_mtx_t vm_swap_data_lock;
4059 #define VM_MAX_SWAP_FILE_NUM 100
4060
4061 static int
4062 sysctl_vm_config_num_swap_files SYSCTL_HANDLER_ARGS
4063 {
4064 #pragma unused(arg1, arg2)
4065 int error = 0, val = vm_num_swap_files_config;
4066
4067 error = sysctl_handle_int(oidp, &val, 0, req);
4068 if (error || !req->newptr) {
4069 goto out;
4070 }
4071
4072 if (!VM_CONFIG_SWAP_IS_ACTIVE && !VM_CONFIG_FREEZER_SWAP_IS_ACTIVE) {
4073 printf("Swap is disabled\n");
4074 error = EINVAL;
4075 goto out;
4076 }
4077
4078 lck_mtx_lock(&vm_swap_data_lock);
4079
4080 if (val < vm_num_swap_files) {
4081 printf("Cannot configure fewer swap files than already exist.\n");
4082 error = EINVAL;
4083 lck_mtx_unlock(&vm_swap_data_lock);
4084 goto out;
4085 }
4086
4087 if (val > VM_MAX_SWAP_FILE_NUM) {
4088 printf("Capping number of swap files to upper bound.\n");
4089 val = VM_MAX_SWAP_FILE_NUM;
4090 }
4091
4092 vm_num_swap_files_config = val;
4093 lck_mtx_unlock(&vm_swap_data_lock);
4094 out:
4095
4096 return 0;
4097 }
4098
4099 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", "");
4100 #endif /* DEVELOPMENT || DEBUG */
4101
4102 /* this kernel does NOT implement shared_region_make_private_np() */
4103 SYSCTL_INT(_kern, KERN_SHREG_PRIVATIZABLE, shreg_private,
4104 CTLFLAG_RD | CTLFLAG_LOCKED,
4105 (int *)NULL, 0, "");
4106
4107 STATIC int
4108 fetch_process_cputype(
4109 proc_t cur_proc,
4110 int *name,
4111 u_int namelen,
4112 cpu_type_t *cputype)
4113 {
4114 proc_t p = PROC_NULL;
4115 int refheld = 0;
4116 cpu_type_t ret = 0;
4117 int error = 0;
4118
4119 if (namelen == 0) {
4120 p = cur_proc;
4121 } else if (namelen == 1) {
4122 p = proc_find(name[0]);
4123 if (p == NULL) {
4124 return EINVAL;
4125 }
4126 refheld = 1;
4127 } else {
4128 error = EINVAL;
4129 goto out;
4130 }
4131
4132 ret = cpu_type() & ~CPU_ARCH_MASK;
4133 if (IS_64BIT_PROCESS(p)) {
4134 ret |= CPU_ARCH_ABI64;
4135 }
4136
4137 *cputype = ret;
4138
4139 if (refheld != 0) {
4140 proc_rele(p);
4141 }
4142 out:
4143 return error;
4144 }
4145
4146
4147 STATIC int
4148 sysctl_sysctl_native(__unused struct sysctl_oid *oidp, void *arg1, int arg2,
4149 struct sysctl_req *req)
4150 {
4151 int error;
4152 cpu_type_t proc_cputype = 0;
4153 if ((error = fetch_process_cputype(req->p, (int *)arg1, arg2, &proc_cputype)) != 0) {
4154 return error;
4155 }
4156 int res = 1;
4157 if ((proc_cputype & ~CPU_ARCH_MASK) != (cpu_type() & ~CPU_ARCH_MASK)) {
4158 res = 0;
4159 }
4160 return SYSCTL_OUT(req, &res, sizeof(res));
4161 }
4162 SYSCTL_PROC(_sysctl, OID_AUTO, proc_native, CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED, 0, 0, sysctl_sysctl_native, "I", "proc_native");
4163
4164 STATIC int
4165 sysctl_sysctl_cputype(__unused struct sysctl_oid *oidp, void *arg1, int arg2,
4166 struct sysctl_req *req)
4167 {
4168 int error;
4169 cpu_type_t proc_cputype = 0;
4170 if ((error = fetch_process_cputype(req->p, (int *)arg1, arg2, &proc_cputype)) != 0) {
4171 return error;
4172 }
4173 return SYSCTL_OUT(req, &proc_cputype, sizeof(proc_cputype));
4174 }
4175 SYSCTL_PROC(_sysctl, OID_AUTO, proc_cputype, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED, 0, 0, sysctl_sysctl_cputype, "I", "proc_cputype");
4176
4177 STATIC int
4178 sysctl_safeboot
4179 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4180 {
4181 return sysctl_io_number(req, boothowto & RB_SAFEBOOT ? 1 : 0, sizeof(int), NULL, NULL);
4182 }
4183
4184 SYSCTL_PROC(_kern, KERN_SAFEBOOT, safeboot,
4185 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
4186 0, 0, sysctl_safeboot, "I", "");
4187
4188 STATIC int
4189 sysctl_singleuser
4190 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4191 {
4192 return sysctl_io_number(req, boothowto & RB_SINGLE ? 1 : 0, sizeof(int), NULL, NULL);
4193 }
4194
4195 SYSCTL_PROC(_kern, OID_AUTO, singleuser,
4196 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
4197 0, 0, sysctl_singleuser, "I", "");
4198
4199 STATIC int
4200 sysctl_minimalboot
4201 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4202 {
4203 return sysctl_io_number(req, minimalboot, sizeof(int), NULL, NULL);
4204 }
4205
4206 SYSCTL_PROC(_kern, OID_AUTO, minimalboot,
4207 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
4208 0, 0, sysctl_minimalboot, "I", "");
4209
4210 /*
4211 * Controls for debugging affinity sets - see osfmk/kern/affinity.c
4212 */
4213 extern boolean_t affinity_sets_enabled;
4214 extern int affinity_sets_mapping;
4215
4216 SYSCTL_INT(_kern, OID_AUTO, affinity_sets_enabled,
4217 CTLFLAG_RW | CTLFLAG_LOCKED, (int *) &affinity_sets_enabled, 0, "hinting enabled");
4218 SYSCTL_INT(_kern, OID_AUTO, affinity_sets_mapping,
4219 CTLFLAG_RW | CTLFLAG_LOCKED, &affinity_sets_mapping, 0, "mapping policy");
4220
4221 /*
4222 * Boolean indicating if KASLR is active.
4223 */
4224 STATIC int
4225 sysctl_slide
4226 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4227 {
4228 uint32_t slide;
4229
4230 slide = vm_kernel_slide ? 1 : 0;
4231
4232 return sysctl_io_number( req, slide, sizeof(int), NULL, NULL);
4233 }
4234
4235 SYSCTL_PROC(_kern, OID_AUTO, slide,
4236 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
4237 0, 0, sysctl_slide, "I", "");
4238
4239 #if DEBUG || DEVELOPMENT
4240 #if defined(__arm64__)
4241 extern vm_offset_t segTEXTEXECB;
4242
4243 static int
4244 sysctl_kernel_text_exec_base_slide SYSCTL_HANDLER_ARGS
4245 {
4246 #pragma unused(arg1, arg2, oidp)
4247 unsigned long slide = 0;
4248 kc_format_t kc_format;
4249
4250 PE_get_primary_kc_format(&kc_format);
4251
4252 if (kc_format == KCFormatFileset) {
4253 void *kch = PE_get_kc_header(KCKindPrimary);
4254 slide = (unsigned long)segTEXTEXECB - (unsigned long)kch + vm_kernel_slide;
4255 }
4256 return SYSCTL_OUT(req, &slide, sizeof(slide));
4257 }
4258
4259 SYSCTL_QUAD(_kern, OID_AUTO, kernel_slide, CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED, &vm_kernel_slide, "");
4260 SYSCTL_QUAD(_kern, OID_AUTO, kernel_text_exec_base, CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED, &segTEXTEXECB, "");
4261 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", "");
4262 #endif /* defined(__arm64__) */
4263
4264 /* User address of the PFZ */
4265 extern user32_addr_t commpage_text32_location;
4266 extern user64_addr_t commpage_text64_location;
4267
4268 STATIC int
4269 sysctl_pfz_start SYSCTL_HANDLER_ARGS
4270 {
4271 #pragma unused(oidp, arg1, arg2)
4272
4273 #ifdef __LP64__
4274 return sysctl_io_number(req, commpage_text64_location, sizeof(user64_addr_t), NULL, NULL);
4275 #else
4276 return sysctl_io_number(req, commpage_text32_location, sizeof(user32_addr_t), NULL, NULL);
4277 #endif
4278 }
4279
4280 SYSCTL_PROC(_kern, OID_AUTO, pfz,
4281 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_MASKED,
4282 0, 0, sysctl_pfz_start, "I", "");
4283 #endif
4284
4285
4286 /*
4287 * Limit on total memory users can wire.
4288 *
4289 * vm_global_user_wire_limit - system wide limit on wired memory from all processes combined.
4290 *
4291 * vm_per_task_user_wire_limit - per address space limit on wired memory. This puts a cap on the process's rlimit value.
4292 *
4293 * These values are initialized to reasonable defaults at boot time based on the available physical memory in
4294 * kmem_init().
4295 *
4296 * All values are in bytes.
4297 */
4298
4299 vm_map_size_t vm_global_user_wire_limit;
4300 vm_map_size_t vm_per_task_user_wire_limit;
4301 extern uint64_t max_mem_actual, max_mem;
4302
4303 uint64_t vm_add_wire_count_over_global_limit;
4304 uint64_t vm_add_wire_count_over_user_limit;
4305 /*
4306 * We used to have a global in the kernel called vm_global_no_user_wire_limit which was the inverse
4307 * of vm_global_user_wire_limit. But maintaining both of those is silly, and vm_global_user_wire_limit is the
4308 * real limit.
4309 * This function is for backwards compatibility with userspace
4310 * since we exposed the old global via a sysctl.
4311 */
4312 STATIC int
4313 sysctl_global_no_user_wire_amount(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4314 {
4315 vm_map_size_t old_value;
4316 vm_map_size_t new_value;
4317 int changed;
4318 int error;
4319 uint64_t config_memsize = max_mem;
4320 #if defined(XNU_TARGET_OS_OSX)
4321 config_memsize = max_mem_actual;
4322 #endif /* defined(XNU_TARGET_OS_OSX) */
4323
4324 old_value = (vm_map_size_t)(config_memsize - vm_global_user_wire_limit);
4325 error = sysctl_io_number(req, old_value, sizeof(vm_map_size_t), &new_value, &changed);
4326 if (changed) {
4327 if ((uint64_t)new_value > config_memsize) {
4328 error = EINVAL;
4329 } else {
4330 vm_global_user_wire_limit = (vm_map_size_t)(config_memsize - new_value);
4331 }
4332 }
4333 return error;
4334 }
4335 /*
4336 * There needs to be a more automatic/elegant way to do this
4337 */
4338 SYSCTL_QUAD(_vm, OID_AUTO, global_user_wire_limit, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_global_user_wire_limit, "");
4339 SYSCTL_QUAD(_vm, OID_AUTO, user_wire_limit, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_per_task_user_wire_limit, "");
4340 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", "");
4341
4342 /*
4343 * Relaxed atomic RW of a 64bit value via sysctl.
4344 */
4345 STATIC int
4346 sysctl_r_64bit_atomic(uint64_t *ptr, struct sysctl_req *req)
4347 {
4348 uint64_t old_value;
4349 uint64_t new_value;
4350 int error;
4351
4352 old_value = os_atomic_load_wide(ptr, relaxed);
4353 error = sysctl_io_number(req, old_value, sizeof(vm_map_size_t), &new_value, NULL);
4354 return error;
4355 }
4356 STATIC int
4357 sysctl_add_wire_count_over_global_limit(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4358 {
4359 return sysctl_r_64bit_atomic(&vm_add_wire_count_over_global_limit, req);
4360 }
4361 STATIC int
4362 sysctl_add_wire_count_over_user_limit(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4363 {
4364 return sysctl_r_64bit_atomic(&vm_add_wire_count_over_user_limit, req);
4365 }
4366
4367 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", "");
4368 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", "");
4369
4370 #if DEVELOPMENT || DEBUG
4371 /* These sysctls are used to test the wired limit. */
4372 extern unsigned int vm_page_wire_count;
4373 extern uint32_t vm_lopage_free_count;
4374 SYSCTL_INT(_vm, OID_AUTO, page_wire_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_wire_count, 0, "");
4375 SYSCTL_INT(_vm, OID_AUTO, lopage_free_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_lopage_free_count, 0, "");
4376
4377 /*
4378 * Setting the per task variable exclude_physfootprint_ledger to 1 will allow the calling task to exclude memory entries that are
4379 * tagged by VM_LEDGER_TAG_DEFAULT and flagged by VM_LEDGER_FLAG_EXCLUDE_FOOTPRINT_DEBUG from its phys_footprint ledger.
4380 */
4381
4382 STATIC int
4383 sysctl_rw_task_no_footprint_for_debug(struct sysctl_oid *oidp __unused, void *arg1 __unused, int arg2 __unused, struct sysctl_req *req)
4384 {
4385 int error;
4386 int value;
4387 proc_t p = current_proc();
4388
4389 if (req->newptr) {
4390 // Write request
4391 error = SYSCTL_IN(req, &value, sizeof(value));
4392 if (!error) {
4393 if (value == 1) {
4394 task_set_no_footprint_for_debug(proc_task(p), TRUE);
4395 } else if (value == 0) {
4396 task_set_no_footprint_for_debug(proc_task(p), FALSE);
4397 } else {
4398 error = EINVAL;
4399 }
4400 }
4401 } else {
4402 // Read request
4403 value = task_get_no_footprint_for_debug(proc_task(p));
4404 error = SYSCTL_OUT(req, &value, sizeof(value));
4405 }
4406 return error;
4407 }
4408
4409 SYSCTL_PROC(_vm, OID_AUTO, task_no_footprint_for_debug,
4410 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_ANYBODY,
4411 0, 0, &sysctl_rw_task_no_footprint_for_debug, "I", "Allow debug memory to be excluded from this task's memory footprint (debug only)");
4412
4413 #endif /* DEVELOPMENT || DEBUG */
4414
4415
4416 extern int vm_map_copy_overwrite_aligned_src_not_internal;
4417 extern int vm_map_copy_overwrite_aligned_src_not_symmetric;
4418 extern int vm_map_copy_overwrite_aligned_src_large;
4419 SYSCTL_INT(_vm, OID_AUTO, vm_copy_src_not_internal, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_map_copy_overwrite_aligned_src_not_internal, 0, "");
4420 SYSCTL_INT(_vm, OID_AUTO, vm_copy_src_not_symmetric, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_map_copy_overwrite_aligned_src_not_symmetric, 0, "");
4421 SYSCTL_INT(_vm, OID_AUTO, vm_copy_src_large, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_map_copy_overwrite_aligned_src_large, 0, "");
4422
4423
4424 extern uint32_t vm_page_external_count;
4425
4426 SYSCTL_INT(_vm, OID_AUTO, vm_page_external_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_external_count, 0, "");
4427
4428 SYSCTL_INT(_vm, OID_AUTO, vm_page_filecache_min, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_state.vm_page_filecache_min, 0, "");
4429 SYSCTL_INT(_vm, OID_AUTO, vm_page_xpmapped_min, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_state.vm_page_xpmapped_min, 0, "");
4430
4431 #if DEVELOPMENT || DEBUG
4432 SYSCTL_INT(_vm, OID_AUTO, vm_page_filecache_min_divisor, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_state.vm_page_filecache_min_divisor, 0, "");
4433 SYSCTL_INT(_vm, OID_AUTO, vm_page_xpmapped_min_divisor, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_state.vm_page_xpmapped_min_divisor, 0, "");
4434 extern boolean_t vps_yield_for_pgqlockwaiters;
4435 SYSCTL_INT(_vm, OID_AUTO, vm_pageoutscan_yields_for_pageQlockwaiters, CTLFLAG_RW | CTLFLAG_LOCKED, &vps_yield_for_pgqlockwaiters, 0, "");
4436 #endif
4437
4438 extern int vm_compressor_mode;
4439 extern int vm_compressor_is_active;
4440 extern int vm_compressor_available;
4441 extern uint32_t c_seg_bufsize;
4442 extern uint32_t c_seg_allocsize;
4443 extern int c_seg_fixed_array_len;
4444 extern uint32_t c_segments_limit;
4445 extern uint32_t c_segment_pages_compressed_limit;
4446 extern uint64_t compressor_pool_size;
4447 extern uint32_t compressor_pool_multiplier;
4448 extern uint32_t vm_ripe_target_age;
4449 extern uint32_t swapout_target_age;
4450 extern _Atomic uint64_t compressor_bytes_used;
4451 extern _Atomic uint64_t c_segment_input_bytes;
4452 extern _Atomic uint64_t c_segment_compressed_bytes;
4453 extern uint32_t c_segment_pages_compressed;
4454 extern uint32_t compressor_eval_period_in_msecs;
4455 extern uint32_t compressor_sample_min_in_msecs;
4456 extern uint32_t compressor_sample_max_in_msecs;
4457 extern uint32_t compressor_thrashing_threshold_per_10msecs;
4458 extern uint32_t compressor_thrashing_min_per_10msecs;
4459 extern uint32_t vm_compressor_time_thread;
4460 extern uint32_t c_segment_svp_in_hash;
4461 extern uint32_t c_segment_svp_hash_succeeded;
4462 extern uint32_t c_segment_svp_hash_failed;
4463
4464 #if DEVELOPMENT || DEBUG
4465 extern uint32_t vm_compressor_minorcompact_threshold_divisor;
4466 extern uint32_t vm_compressor_majorcompact_threshold_divisor;
4467 extern uint32_t vm_compressor_unthrottle_threshold_divisor;
4468 extern uint32_t vm_compressor_catchup_threshold_divisor;
4469
4470 extern uint32_t vm_compressor_minorcompact_threshold_divisor_overridden;
4471 extern uint32_t vm_compressor_majorcompact_threshold_divisor_overridden;
4472 extern uint32_t vm_compressor_unthrottle_threshold_divisor_overridden;
4473 extern uint32_t vm_compressor_catchup_threshold_divisor_overridden;
4474
4475 extern vmct_stats_t vmct_stats;
4476
4477
4478 STATIC int
4479 sysctl_minorcompact_threshold_divisor(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4480 {
4481 int new_value, changed;
4482 int error = sysctl_io_number(req, vm_compressor_minorcompact_threshold_divisor, sizeof(int), &new_value, &changed);
4483
4484 if (changed) {
4485 vm_compressor_minorcompact_threshold_divisor = new_value;
4486 vm_compressor_minorcompact_threshold_divisor_overridden = 1;
4487 }
4488 return error;
4489 }
4490
4491 SYSCTL_PROC(_vm, OID_AUTO, compressor_minorcompact_threshold_divisor,
4492 CTLTYPE_INT | CTLFLAG_LOCKED | CTLFLAG_RW,
4493 0, 0, sysctl_minorcompact_threshold_divisor, "I", "");
4494
4495
4496 STATIC int
4497 sysctl_majorcompact_threshold_divisor(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4498 {
4499 int new_value, changed;
4500 int error = sysctl_io_number(req, vm_compressor_majorcompact_threshold_divisor, sizeof(int), &new_value, &changed);
4501
4502 if (changed) {
4503 vm_compressor_majorcompact_threshold_divisor = new_value;
4504 vm_compressor_majorcompact_threshold_divisor_overridden = 1;
4505 }
4506 return error;
4507 }
4508
4509 SYSCTL_PROC(_vm, OID_AUTO, compressor_majorcompact_threshold_divisor,
4510 CTLTYPE_INT | CTLFLAG_LOCKED | CTLFLAG_RW,
4511 0, 0, sysctl_majorcompact_threshold_divisor, "I", "");
4512
4513
4514 STATIC int
4515 sysctl_unthrottle_threshold_divisor(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4516 {
4517 int new_value, changed;
4518 int error = sysctl_io_number(req, vm_compressor_unthrottle_threshold_divisor, sizeof(int), &new_value, &changed);
4519
4520 if (changed) {
4521 vm_compressor_unthrottle_threshold_divisor = new_value;
4522 vm_compressor_unthrottle_threshold_divisor_overridden = 1;
4523 }
4524 return error;
4525 }
4526
4527 SYSCTL_PROC(_vm, OID_AUTO, compressor_unthrottle_threshold_divisor,
4528 CTLTYPE_INT | CTLFLAG_LOCKED | CTLFLAG_RW,
4529 0, 0, sysctl_unthrottle_threshold_divisor, "I", "");
4530
4531
4532 STATIC int
4533 sysctl_catchup_threshold_divisor(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4534 {
4535 int new_value, changed;
4536 int error = sysctl_io_number(req, vm_compressor_catchup_threshold_divisor, sizeof(int), &new_value, &changed);
4537
4538 if (changed) {
4539 vm_compressor_catchup_threshold_divisor = new_value;
4540 vm_compressor_catchup_threshold_divisor_overridden = 1;
4541 }
4542 return error;
4543 }
4544
4545 SYSCTL_PROC(_vm, OID_AUTO, compressor_catchup_threshold_divisor,
4546 CTLTYPE_INT | CTLFLAG_LOCKED | CTLFLAG_RW,
4547 0, 0, sysctl_catchup_threshold_divisor, "I", "");
4548 #endif
4549
4550
4551 SYSCTL_QUAD(_vm, OID_AUTO, compressor_input_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, ((uint64_t *)&c_segment_input_bytes), "");
4552 SYSCTL_QUAD(_vm, OID_AUTO, compressor_compressed_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, ((uint64_t *)&c_segment_compressed_bytes), "");
4553 SYSCTL_QUAD(_vm, OID_AUTO, compressor_bytes_used, CTLFLAG_RD | CTLFLAG_LOCKED, ((uint64_t *)&compressor_bytes_used), "");
4554
4555 SYSCTL_INT(_vm, OID_AUTO, compressor_mode, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_compressor_mode, 0, "");
4556 SYSCTL_INT(_vm, OID_AUTO, compressor_is_active, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_compressor_is_active, 0, "");
4557 SYSCTL_INT(_vm, OID_AUTO, compressor_swapout_target_age, CTLFLAG_RD | CTLFLAG_LOCKED, &swapout_target_age, 0, "");
4558 SYSCTL_INT(_vm, OID_AUTO, compressor_available, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_compressor_available, 0, "");
4559 SYSCTL_INT(_vm, OID_AUTO, compressor_segment_buffer_size, CTLFLAG_RD | CTLFLAG_LOCKED, &c_seg_bufsize, 0, "");
4560 SYSCTL_QUAD(_vm, OID_AUTO, compressor_pool_size, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_pool_size, "");
4561 SYSCTL_UINT(_vm, OID_AUTO, compressor_pool_multiplier, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_pool_multiplier, 0, "");
4562 SYSCTL_INT(_vm, OID_AUTO, compressor_segment_slots_fixed_array_len, CTLFLAG_RD | CTLFLAG_LOCKED, &c_seg_fixed_array_len, 0, "");
4563 SYSCTL_UINT(_vm, OID_AUTO, compressor_segment_limit, CTLFLAG_RD | CTLFLAG_LOCKED, &c_segments_limit, 0, "");
4564 SYSCTL_UINT(_vm, OID_AUTO, compressor_segment_pages_compressed_limit, CTLFLAG_RD | CTLFLAG_LOCKED, &c_segment_pages_compressed_limit, 0, "");
4565 SYSCTL_UINT(_vm, OID_AUTO, compressor_segment_alloc_size, CTLFLAG_RD | CTLFLAG_LOCKED, &c_seg_allocsize, 0, "");
4566 SYSCTL_UINT(_vm, OID_AUTO, compressor_segment_pages_compressed, CTLFLAG_RD | CTLFLAG_LOCKED, &c_segment_pages_compressed, 0, "");
4567 SYSCTL_UINT(_vm, OID_AUTO, compressor_segment_svp_in_hash, CTLFLAG_RD | CTLFLAG_LOCKED, &c_segment_svp_in_hash, 0, "");
4568 SYSCTL_UINT(_vm, OID_AUTO, compressor_segment_svp_hash_succeeded, CTLFLAG_RD | CTLFLAG_LOCKED, &c_segment_svp_hash_succeeded, 0, "");
4569 SYSCTL_UINT(_vm, OID_AUTO, compressor_segment_svp_hash_failed, CTLFLAG_RD | CTLFLAG_LOCKED, &c_segment_svp_hash_failed, 0, "");
4570
4571 #if CONFIG_TRACK_UNMODIFIED_ANON_PAGES
4572 extern uint64_t compressor_ro_uncompressed;
4573 extern uint64_t compressor_ro_uncompressed_total_returned;
4574 extern uint64_t compressor_ro_uncompressed_skip_returned;
4575 extern uint64_t compressor_ro_uncompressed_get;
4576 extern uint64_t compressor_ro_uncompressed_put;
4577 extern uint64_t compressor_ro_uncompressed_swap_usage;
4578
4579 SYSCTL_QUAD(_vm, OID_AUTO, compressor_ro_uncompressed_total_returned, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_ro_uncompressed_total_returned, "");
4580 SYSCTL_QUAD(_vm, OID_AUTO, compressor_ro_uncompressed_writes_saved, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_ro_uncompressed_skip_returned, "");
4581 SYSCTL_QUAD(_vm, OID_AUTO, compressor_ro_uncompressed_candidates, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_ro_uncompressed, "");
4582 SYSCTL_QUAD(_vm, OID_AUTO, compressor_ro_uncompressed_rereads, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_ro_uncompressed_get, "");
4583 SYSCTL_QUAD(_vm, OID_AUTO, compressor_ro_uncompressed_swap_pages_on_disk, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_ro_uncompressed_swap_usage, "");
4584 #endif /* CONFIG_TRACK_UNMODIFIED_ANON_PAGES */
4585
4586 extern int min_csegs_per_major_compaction;
4587 SYSCTL_INT(_vm, OID_AUTO, compressor_min_csegs_per_major_compaction, CTLFLAG_RW | CTLFLAG_LOCKED, &min_csegs_per_major_compaction, 0, "");
4588
4589 SYSCTL_INT(_vm, OID_AUTO, vm_ripe_target_age_in_secs, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_ripe_target_age, 0, "");
4590
4591 SYSCTL_INT(_vm, OID_AUTO, compressor_eval_period_in_msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &compressor_eval_period_in_msecs, 0, "");
4592 SYSCTL_INT(_vm, OID_AUTO, compressor_sample_min_in_msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &compressor_sample_min_in_msecs, 0, "");
4593 SYSCTL_INT(_vm, OID_AUTO, compressor_sample_max_in_msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &compressor_sample_max_in_msecs, 0, "");
4594 SYSCTL_INT(_vm, OID_AUTO, compressor_thrashing_threshold_per_10msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &compressor_thrashing_threshold_per_10msecs, 0, "");
4595 SYSCTL_INT(_vm, OID_AUTO, compressor_thrashing_min_per_10msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &compressor_thrashing_min_per_10msecs, 0, "");
4596
4597 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapouts_under_30s, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.unripe_under_30s, "");
4598 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapouts_under_60s, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.unripe_under_60s, "");
4599 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapouts_under_300s, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.unripe_under_300s, "");
4600 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_reclaim_swapins, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.reclaim_swapins, "");
4601 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_defrag_swapins, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.defrag_swapins, "");
4602 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_swapout_threshold_exceeded, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.compressor_swap_threshold_exceeded, "");
4603 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_swapout_fileq_throttled, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.external_q_throttled, "");
4604 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_swapout_free_count_low, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.free_count_below_reserve, "");
4605 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_swapout_thrashing_detected, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.thrashing_detected, "");
4606 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_swapout_fragmentation_detected, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.fragmentation_detected, "");
4607
4608 SYSCTL_STRING(_vm, OID_AUTO, swapfileprefix, CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED, swapfilename, sizeof(swapfilename) - SWAPFILENAME_INDEX_LEN, "");
4609
4610 SYSCTL_INT(_vm, OID_AUTO, compressor_timing_enabled, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_compressor_time_thread, 0, "");
4611
4612 #if DEVELOPMENT || DEBUG
4613 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_runtime0, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_runtimes[0], "");
4614 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_runtime1, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_runtimes[1], "");
4615
4616 SYSCTL_QUAD(_vm, OID_AUTO, compressor_threads_total_execution_time, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_cthreads_total, "");
4617
4618 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_pages0, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_pages[0], "");
4619 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_pages1, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_pages[1], "");
4620
4621 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_iterations0, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_iterations[0], "");
4622 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_iterations1, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_iterations[1], "");
4623
4624 SYSCTL_INT(_vm, OID_AUTO, compressor_thread_minpages0, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_minpages[0], 0, "");
4625 SYSCTL_INT(_vm, OID_AUTO, compressor_thread_minpages1, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_minpages[1], 0, "");
4626
4627 SYSCTL_INT(_vm, OID_AUTO, compressor_thread_maxpages0, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_maxpages[0], 0, "");
4628 SYSCTL_INT(_vm, OID_AUTO, compressor_thread_maxpages1, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_maxpages[1], 0, "");
4629
4630 int vm_compressor_injected_error_count;
4631
4632 SYSCTL_INT(_vm, OID_AUTO, compressor_injected_error_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_compressor_injected_error_count, 0, "");
4633
4634 static int
4635 sysctl_compressor_inject_error(__unused struct sysctl_oid *oidp,
4636 __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4637 {
4638 int result;
4639 vm_address_t va = 0;
4640 int changed;
4641
4642 result = sysctl_io_number(req, va, sizeof(va), &va, &changed);
4643 if (result == 0 && changed) {
4644 result = vm_map_inject_error(current_map(), va);
4645 if (result == 0) {
4646 /*
4647 * Count the number of errors injected successfully to detect
4648 * situations where corruption was caused by improper use of this
4649 * sysctl.
4650 */
4651 os_atomic_inc(&vm_compressor_injected_error_count, relaxed);
4652 }
4653 }
4654 return result;
4655 }
4656
4657 SYSCTL_PROC(_vm, OID_AUTO, compressor_inject_error, CTLTYPE_QUAD | CTLFLAG_LOCKED | CTLFLAG_RW,
4658 0, 0, sysctl_compressor_inject_error, "Q", "flips a bit in a compressed page for the current task");
4659
4660 /*
4661 * Opt a process in/out of self donation mode.
4662 */
4663 static int
4664 sysctl_vm_pid_toggle_selfdonate_pages SYSCTL_HANDLER_ARGS
4665 {
4666 #pragma unused(arg1, arg2)
4667 int error, pid = 0;
4668 proc_t p;
4669
4670 error = sysctl_handle_int(oidp, &pid, 0, req);
4671 if (error || !req->newptr) {
4672 return error;
4673 }
4674
4675 p = proc_find(pid);
4676 if (p != NULL) {
4677 (void) vm_toggle_task_selfdonate_pages(proc_task(p));
4678 proc_rele(p);
4679 return error;
4680 } else {
4681 printf("sysctl_vm_pid_selfdonate_pages: Invalid process\n");
4682 }
4683
4684 return EINVAL;
4685 }
4686 SYSCTL_PROC(_vm, OID_AUTO, pid_toggle_selfdonate_pages, CTLTYPE_INT | CTLFLAG_WR | CTLFLAG_LOCKED | CTLFLAG_MASKED,
4687 0, 0, &sysctl_vm_pid_toggle_selfdonate_pages, "I", "");
4688 #endif
4689 extern uint32_t vm_page_donate_mode;
4690 extern uint32_t vm_page_donate_target_high, vm_page_donate_target_low;
4691 SYSCTL_INT(_vm, OID_AUTO, vm_page_donate_mode, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_page_donate_mode, 0, "");
4692 SYSCTL_INT(_vm, OID_AUTO, vm_page_donate_target_high, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_page_donate_target_high, 0, "");
4693 SYSCTL_INT(_vm, OID_AUTO, vm_page_donate_target_low, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_page_donate_target_low, 0, "");
4694
4695 SYSCTL_QUAD(_vm, OID_AUTO, lz4_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_compressions, "");
4696 SYSCTL_QUAD(_vm, OID_AUTO, lz4_compression_failures, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_compression_failures, "");
4697 SYSCTL_QUAD(_vm, OID_AUTO, lz4_compressed_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_compressed_bytes, "");
4698 SYSCTL_QUAD(_vm, OID_AUTO, lz4_wk_compression_delta, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_wk_compression_delta, "");
4699 SYSCTL_QUAD(_vm, OID_AUTO, lz4_wk_compression_negative_delta, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_wk_compression_negative_delta, "");
4700
4701 SYSCTL_QUAD(_vm, OID_AUTO, lz4_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_decompressions, "");
4702 SYSCTL_QUAD(_vm, OID_AUTO, lz4_decompressed_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_decompressed_bytes, "");
4703
4704 SYSCTL_QUAD(_vm, OID_AUTO, uc_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.uc_decompressions, "");
4705
4706 SYSCTL_QUAD(_vm, OID_AUTO, wk_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_compressions, "");
4707
4708 SYSCTL_QUAD(_vm, OID_AUTO, wk_catime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_cabstime, "");
4709
4710 SYSCTL_QUAD(_vm, OID_AUTO, wkh_catime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wkh_cabstime, "");
4711 SYSCTL_QUAD(_vm, OID_AUTO, wkh_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wkh_compressions, "");
4712
4713 SYSCTL_QUAD(_vm, OID_AUTO, wks_catime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_cabstime, "");
4714 SYSCTL_QUAD(_vm, OID_AUTO, wks_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_compressions, "");
4715
4716 SYSCTL_QUAD(_vm, OID_AUTO, wk_compressions_exclusive, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_compressions_exclusive, "");
4717 SYSCTL_QUAD(_vm, OID_AUTO, wk_sv_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_sv_compressions, "");
4718 SYSCTL_QUAD(_vm, OID_AUTO, wk_mzv_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_mzv_compressions, "");
4719 SYSCTL_QUAD(_vm, OID_AUTO, wk_compression_failures, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_compression_failures, "");
4720 SYSCTL_QUAD(_vm, OID_AUTO, wk_compressed_bytes_exclusive, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_compressed_bytes_exclusive, "");
4721 SYSCTL_QUAD(_vm, OID_AUTO, wk_compressed_bytes_total, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_compressed_bytes_total, "");
4722
4723 SYSCTL_QUAD(_vm, OID_AUTO, wks_compressed_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_compressed_bytes, "");
4724 SYSCTL_QUAD(_vm, OID_AUTO, wks_compression_failures, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_compression_failures, "");
4725 SYSCTL_QUAD(_vm, OID_AUTO, wks_sv_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_sv_compressions, "");
4726
4727
4728 SYSCTL_QUAD(_vm, OID_AUTO, wk_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_decompressions, "");
4729
4730 SYSCTL_QUAD(_vm, OID_AUTO, wk_datime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_dabstime, "");
4731
4732 SYSCTL_QUAD(_vm, OID_AUTO, wkh_datime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wkh_dabstime, "");
4733 SYSCTL_QUAD(_vm, OID_AUTO, wkh_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wkh_decompressions, "");
4734
4735 SYSCTL_QUAD(_vm, OID_AUTO, wks_datime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_dabstime, "");
4736 SYSCTL_QUAD(_vm, OID_AUTO, wks_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_decompressions, "");
4737
4738 SYSCTL_QUAD(_vm, OID_AUTO, wk_decompressed_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_decompressed_bytes, "");
4739 SYSCTL_QUAD(_vm, OID_AUTO, wk_sv_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_sv_decompressions, "");
4740
4741 SYSCTL_INT(_vm, OID_AUTO, lz4_threshold, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_threshold, 0, "");
4742 SYSCTL_INT(_vm, OID_AUTO, wkdm_reeval_threshold, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.wkdm_reeval_threshold, 0, "");
4743 SYSCTL_INT(_vm, OID_AUTO, lz4_max_failure_skips, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_max_failure_skips, 0, "");
4744 SYSCTL_INT(_vm, OID_AUTO, lz4_max_failure_run_length, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_max_failure_run_length, 0, "");
4745 SYSCTL_INT(_vm, OID_AUTO, lz4_max_preselects, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_max_preselects, 0, "");
4746 SYSCTL_INT(_vm, OID_AUTO, lz4_run_preselection_threshold, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_run_preselection_threshold, 0, "");
4747 SYSCTL_INT(_vm, OID_AUTO, lz4_run_continue_bytes, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_run_continue_bytes, 0, "");
4748 SYSCTL_INT(_vm, OID_AUTO, lz4_profitable_bytes, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_profitable_bytes, 0, "");
4749 #if DEVELOPMENT || DEBUG
4750 extern int vm_compressor_current_codec;
4751 extern int vm_compressor_test_seg_wp;
4752 extern boolean_t vm_compressor_force_sw_wkdm;
4753 SYSCTL_INT(_vm, OID_AUTO, compressor_codec, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_compressor_current_codec, 0, "");
4754 SYSCTL_INT(_vm, OID_AUTO, compressor_test_wp, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_compressor_test_seg_wp, 0, "");
4755
4756 SYSCTL_INT(_vm, OID_AUTO, wksw_force, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_compressor_force_sw_wkdm, 0, "");
4757 extern int precompy, wkswhw;
4758
4759 SYSCTL_INT(_vm, OID_AUTO, precompy, CTLFLAG_RW | CTLFLAG_LOCKED, &precompy, 0, "");
4760 SYSCTL_INT(_vm, OID_AUTO, wkswhw, CTLFLAG_RW | CTLFLAG_LOCKED, &wkswhw, 0, "");
4761 extern unsigned int vm_ktrace_enabled;
4762 SYSCTL_INT(_vm, OID_AUTO, vm_ktrace, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_ktrace_enabled, 0, "");
4763 #endif
4764
4765 #if CONFIG_PHANTOM_CACHE
4766 extern uint32_t phantom_cache_thrashing_threshold;
4767 extern uint32_t phantom_cache_eval_period_in_msecs;
4768 extern uint32_t phantom_cache_thrashing_threshold_ssd;
4769
4770
4771 SYSCTL_INT(_vm, OID_AUTO, phantom_cache_eval_period_in_msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &phantom_cache_eval_period_in_msecs, 0, "");
4772 SYSCTL_INT(_vm, OID_AUTO, phantom_cache_thrashing_threshold, CTLFLAG_RW | CTLFLAG_LOCKED, &phantom_cache_thrashing_threshold, 0, "");
4773 SYSCTL_INT(_vm, OID_AUTO, phantom_cache_thrashing_threshold_ssd, CTLFLAG_RW | CTLFLAG_LOCKED, &phantom_cache_thrashing_threshold_ssd, 0, "");
4774 #endif
4775
4776 #if defined(__LP64__)
4777 extern uint32_t vm_page_background_count;
4778 extern uint32_t vm_page_background_target;
4779 extern uint32_t vm_page_background_internal_count;
4780 extern uint32_t vm_page_background_external_count;
4781 extern uint32_t vm_page_background_mode;
4782 extern uint32_t vm_page_background_exclude_external;
4783 extern uint64_t vm_page_background_promoted_count;
4784 extern uint64_t vm_pageout_rejected_bq_internal;
4785 extern uint64_t vm_pageout_rejected_bq_external;
4786
4787 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_mode, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_page_background_mode, 0, "");
4788 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_exclude_external, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_page_background_exclude_external, 0, "");
4789 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_target, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_page_background_target, 0, "");
4790 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_background_count, 0, "");
4791 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_internal_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_background_internal_count, 0, "");
4792 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_external_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_background_external_count, 0, "");
4793
4794 SYSCTL_QUAD(_vm, OID_AUTO, vm_page_background_promoted_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_background_promoted_count, "");
4795 SYSCTL_QUAD(_vm, OID_AUTO, vm_pageout_considered_bq_internal, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_vminfo.vm_pageout_considered_bq_internal, "");
4796 SYSCTL_QUAD(_vm, OID_AUTO, vm_pageout_considered_bq_external, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_vminfo.vm_pageout_considered_bq_external, "");
4797 SYSCTL_QUAD(_vm, OID_AUTO, vm_pageout_rejected_bq_internal, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_rejected_bq_internal, "");
4798 SYSCTL_QUAD(_vm, OID_AUTO, vm_pageout_rejected_bq_external, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_rejected_bq_external, "");
4799
4800 #endif /* __LP64__ */
4801
4802 extern boolean_t vm_darkwake_mode;
4803
4804 STATIC int
4805 sysctl_toggle_darkwake_mode(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4806 {
4807 int new_value, changed;
4808 int error = sysctl_io_number(req, vm_darkwake_mode, sizeof(int), &new_value, &changed);
4809
4810 if (!error && changed) {
4811 if (new_value != 0 && new_value != 1) {
4812 printf("Error: Invalid value passed to darkwake sysctl. Acceptable: 0 or 1.\n");
4813 error = EINVAL;
4814 } else {
4815 vm_update_darkwake_mode((boolean_t) new_value);
4816 }
4817 }
4818
4819 return error;
4820 }
4821
4822 SYSCTL_PROC(_vm, OID_AUTO, darkwake_mode,
4823 CTLTYPE_INT | CTLFLAG_LOCKED | CTLFLAG_RW,
4824 0, 0, sysctl_toggle_darkwake_mode, "I", "");
4825
4826 #if (DEVELOPMENT || DEBUG)
4827
4828 SYSCTL_UINT(_vm, OID_AUTO, vm_page_creation_throttled_hard,
4829 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
4830 &vm_page_creation_throttled_hard, 0, "");
4831
4832 SYSCTL_UINT(_vm, OID_AUTO, vm_page_creation_throttled_soft,
4833 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
4834 &vm_page_creation_throttled_soft, 0, "");
4835
4836 extern uint32_t vm_pageout_memorystatus_fb_factor_nr;
4837 extern uint32_t vm_pageout_memorystatus_fb_factor_dr;
4838 SYSCTL_INT(_vm, OID_AUTO, vm_pageout_memorystatus_fb_factor_nr, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_memorystatus_fb_factor_nr, 0, "");
4839 SYSCTL_INT(_vm, OID_AUTO, vm_pageout_memorystatus_fb_factor_dr, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_memorystatus_fb_factor_dr, 0, "");
4840
4841 extern uint32_t vm_grab_anon_nops;
4842
4843 SYSCTL_INT(_vm, OID_AUTO, vm_grab_anon_overrides, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_debug.vm_grab_anon_overrides, 0, "");
4844 SYSCTL_INT(_vm, OID_AUTO, vm_grab_anon_nops, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_debug.vm_grab_anon_nops, 0, "");
4845 SYSCTL_INT(_vm, OID_AUTO, vm_pageout_yield_for_free_pages, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_debug.vm_pageout_yield_for_free_pages, 0, "");
4846
4847
4848 extern int vm_page_delayed_work_ctx_needed;
4849 SYSCTL_INT(_vm, OID_AUTO, vm_page_needed_delayed_work_ctx, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_delayed_work_ctx_needed, 0, "");
4850
4851 /* log message counters for persistence mode */
4852 SCALABLE_COUNTER_DECLARE(oslog_p_total_msgcount);
4853 SCALABLE_COUNTER_DECLARE(oslog_p_metadata_saved_msgcount);
4854 SCALABLE_COUNTER_DECLARE(oslog_p_metadata_dropped_msgcount);
4855 SCALABLE_COUNTER_DECLARE(oslog_p_signpost_saved_msgcount);
4856 SCALABLE_COUNTER_DECLARE(oslog_p_signpost_dropped_msgcount);
4857 SCALABLE_COUNTER_DECLARE(oslog_p_error_count);
4858 SCALABLE_COUNTER_DECLARE(oslog_p_error_count);
4859 SCALABLE_COUNTER_DECLARE(oslog_p_saved_msgcount);
4860 SCALABLE_COUNTER_DECLARE(oslog_p_dropped_msgcount);
4861 SCALABLE_COUNTER_DECLARE(oslog_p_boot_dropped_msgcount);
4862 SCALABLE_COUNTER_DECLARE(oslog_p_coprocessor_total_msgcount);
4863 SCALABLE_COUNTER_DECLARE(oslog_p_coprocessor_dropped_msgcount);
4864 SCALABLE_COUNTER_DECLARE(oslog_p_unresolved_kc_msgcount);
4865 SCALABLE_COUNTER_DECLARE(oslog_p_fmt_invalid_msgcount);
4866 SCALABLE_COUNTER_DECLARE(oslog_p_fmt_max_args_msgcount);
4867 SCALABLE_COUNTER_DECLARE(oslog_p_truncated_msgcount);
4868
4869 SCALABLE_COUNTER_DECLARE(oslog_subsystem_count);
4870 SCALABLE_COUNTER_DECLARE(oslog_subsystem_found);
4871 SCALABLE_COUNTER_DECLARE(oslog_subsystem_dropped);
4872
4873 SCALABLE_COUNTER_DECLARE(log_queue_cnt_received);
4874 SCALABLE_COUNTER_DECLARE(log_queue_cnt_rejected_fh);
4875 SCALABLE_COUNTER_DECLARE(log_queue_cnt_sent);
4876 SCALABLE_COUNTER_DECLARE(log_queue_cnt_dropped_nomem);
4877 SCALABLE_COUNTER_DECLARE(log_queue_cnt_queued);
4878 SCALABLE_COUNTER_DECLARE(log_queue_cnt_dropped_off);
4879 SCALABLE_COUNTER_DECLARE(log_queue_cnt_mem_active);
4880 SCALABLE_COUNTER_DECLARE(log_queue_cnt_mem_allocated);
4881 SCALABLE_COUNTER_DECLARE(log_queue_cnt_mem_released);
4882 SCALABLE_COUNTER_DECLARE(log_queue_cnt_mem_failed);
4883
4884 /* log message counters for streaming mode */
4885 SCALABLE_COUNTER_DECLARE(oslog_s_total_msgcount);
4886 SCALABLE_COUNTER_DECLARE(oslog_s_metadata_msgcount);
4887 SCALABLE_COUNTER_DECLARE(oslog_s_error_count);
4888 SCALABLE_COUNTER_DECLARE(oslog_s_streamed_msgcount);
4889 SCALABLE_COUNTER_DECLARE(oslog_s_dropped_msgcount);
4890
4891 /* log message counters for msgbuf logging */
4892 SCALABLE_COUNTER_DECLARE(oslog_msgbuf_msgcount);
4893 SCALABLE_COUNTER_DECLARE(oslog_msgbuf_dropped_msgcount);
4894 extern uint32_t oslog_msgbuf_dropped_charcount;
4895
4896 #if CONFIG_EXCLAVES
4897 /* log message counters for exclaves logging */
4898 SCALABLE_COUNTER_DECLARE(oslog_e_log_count);
4899 SCALABLE_COUNTER_DECLARE(oslog_e_log_dropped_count);
4900 SCALABLE_COUNTER_DECLARE(oslog_e_metadata_count);
4901 SCALABLE_COUNTER_DECLARE(oslog_e_metadata_dropped_count);
4902 SCALABLE_COUNTER_DECLARE(oslog_e_signpost_count);
4903 SCALABLE_COUNTER_DECLARE(oslog_e_signpost_dropped_count);
4904 SCALABLE_COUNTER_DECLARE(oslog_e_query_count);
4905 SCALABLE_COUNTER_DECLARE(oslog_e_query_error_count);
4906 SCALABLE_COUNTER_DECLARE(oslog_e_trace_mode_set_count);
4907 SCALABLE_COUNTER_DECLARE(oslog_e_trace_mode_error_count);
4908 #endif // CONFIG_EXCLAVES
4909
4910 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_total_msgcount, oslog_p_total_msgcount, "");
4911 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_metadata_saved_msgcount, oslog_p_metadata_saved_msgcount, "");
4912 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_metadata_dropped_msgcount, oslog_p_metadata_dropped_msgcount, "");
4913 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_signpost_saved_msgcount, oslog_p_signpost_saved_msgcount, "");
4914 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_signpost_dropped_msgcount, oslog_p_signpost_dropped_msgcount, "");
4915 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_error_count, oslog_p_error_count, "");
4916 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_saved_msgcount, oslog_p_saved_msgcount, "");
4917 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_dropped_msgcount, oslog_p_dropped_msgcount, "");
4918 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_boot_dropped_msgcount, oslog_p_boot_dropped_msgcount, "");
4919 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_coprocessor_total_msgcount, oslog_p_coprocessor_total_msgcount, "");
4920 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_coprocessor_dropped_msgcount, oslog_p_coprocessor_dropped_msgcount, "");
4921 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_unresolved_kc_msgcount, oslog_p_unresolved_kc_msgcount, "");
4922
4923 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_fmt_invalid_msgcount, oslog_p_fmt_invalid_msgcount, "");
4924 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_fmt_max_args_msgcount, oslog_p_fmt_max_args_msgcount, "");
4925 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_truncated_msgcount, oslog_p_truncated_msgcount, "");
4926
4927 SYSCTL_SCALABLE_COUNTER(_debug, oslog_s_total_msgcount, oslog_s_total_msgcount, "Number of logs sent to streaming");
4928 SYSCTL_SCALABLE_COUNTER(_debug, oslog_s_metadata_msgcount, oslog_s_metadata_msgcount, "Number of metadata sent to streaming");
4929 SYSCTL_SCALABLE_COUNTER(_debug, oslog_s_error_count, oslog_s_error_count, "Number of invalid stream logs");
4930 SYSCTL_SCALABLE_COUNTER(_debug, oslog_s_streamed_msgcount, oslog_s_streamed_msgcount, "Number of streamed logs");
4931 SYSCTL_SCALABLE_COUNTER(_debug, oslog_s_dropped_msgcount, oslog_s_dropped_msgcount, "Number of logs dropped from stream");
4932
4933 SYSCTL_SCALABLE_COUNTER(_debug, oslog_msgbuf_msgcount, oslog_msgbuf_msgcount, "Number of dmesg log messages");
4934 SYSCTL_SCALABLE_COUNTER(_debug, oslog_msgbuf_dropped_msgcount, oslog_msgbuf_dropped_msgcount, "Number of dropped dmesg log messages");
4935 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");
4936
4937 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_received, log_queue_cnt_received, "Number of received logs");
4938 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_rejected_fh, log_queue_cnt_rejected_fh, "Number of logs initially rejected by FH");
4939 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_sent, log_queue_cnt_sent, "Number of logs successfully saved in FH");
4940 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_dropped_nomem, log_queue_cnt_dropped_nomem, "Number of logs dropped due to lack of queue memory");
4941 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_queued, log_queue_cnt_queued, "Current number of logs stored in log queues");
4942 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_dropped_off, log_queue_cnt_dropped_off, "Number of logs dropped due to disabled log queues");
4943 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_mem_allocated, log_queue_cnt_mem_allocated, "Number of memory allocations");
4944 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_mem_released, log_queue_cnt_mem_released, "Number of memory releases");
4945 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_mem_failed, log_queue_cnt_mem_failed, "Number of failed memory allocations");
4946
4947 SYSCTL_SCALABLE_COUNTER(_debug, oslog_subsystem_count, oslog_subsystem_count, "Number of registered log subsystems");
4948 SYSCTL_SCALABLE_COUNTER(_debug, oslog_subsystem_found, oslog_subsystem_found, "Number of sucessful log subsystem lookups");
4949 SYSCTL_SCALABLE_COUNTER(_debug, oslog_subsystem_dropped, oslog_subsystem_dropped, "Number of dropped log subsystem registrations");
4950
4951 #if CONFIG_EXCLAVES
4952 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_metadata_count, oslog_e_metadata_count,
4953 "Number of metadata messages retrieved from the exclaves log server");
4954 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_metadata_dropped_count, oslog_e_metadata_dropped_count,
4955 "Number of dropped metadata messages retrieved from the exclaves log server");
4956 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_log_count, oslog_e_log_count,
4957 "Number of logs retrieved from the exclaves log server");
4958 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_log_dropped_count, oslog_e_log_dropped_count,
4959 "Number of dropeed logs retrieved from the exclaves log server");
4960 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_signpost_count, oslog_e_signpost_count,
4961 "Number of signposts retrieved from the exclaves log server");
4962 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_signpost_dropped_count, oslog_e_signpost_dropped_count,
4963 "Number of dropped signposts retrieved from the exclaves log server");
4964 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_query_count, oslog_e_query_count,
4965 "Number of sucessful queries to the exclaves log server");
4966 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_query_error_count, oslog_e_query_error_count,
4967 "Number of failed queries to the exclaves log server");
4968 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_trace_mode_set_count, oslog_e_trace_mode_set_count,
4969 "Number of exclaves trace mode updates");
4970 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_trace_mode_error_count, oslog_e_trace_mode_error_count,
4971 "Number of failed exclaves trace mode updates");
4972 #endif // CONFIG_EXCLAVES
4973
4974 #endif /* DEVELOPMENT || DEBUG */
4975
4976 /*
4977 * Enable tracing of voucher contents
4978 */
4979 extern uint32_t ipc_voucher_trace_contents;
4980
4981 SYSCTL_INT(_kern, OID_AUTO, ipc_voucher_trace_contents,
4982 CTLFLAG_RW | CTLFLAG_LOCKED, &ipc_voucher_trace_contents, 0, "Enable tracing voucher contents");
4983
4984 /*
4985 * Kernel stack size and depth
4986 */
4987 SYSCTL_INT(_kern, OID_AUTO, stack_size,
4988 CTLFLAG_RD | CTLFLAG_LOCKED, (int *) &kernel_stack_size, 0, "Kernel stack size");
4989 SYSCTL_INT(_kern, OID_AUTO, stack_depth_max,
4990 CTLFLAG_RD | CTLFLAG_LOCKED, (int *) &kernel_stack_depth_max, 0, "Max kernel stack depth at interrupt or context switch");
4991
4992 extern unsigned int kern_feature_overrides;
4993 SYSCTL_INT(_kern, OID_AUTO, kern_feature_overrides,
4994 CTLFLAG_RD | CTLFLAG_LOCKED, &kern_feature_overrides, 0, "Kernel feature override mask");
4995
4996 /*
4997 * enable back trace for port allocations
4998 */
4999 extern int ipc_portbt;
5000
5001 SYSCTL_INT(_kern, OID_AUTO, ipc_portbt,
5002 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
5003 &ipc_portbt, 0, "");
5004
5005 /*
5006 * Scheduler sysctls
5007 */
5008
5009 SYSCTL_STRING(_kern, OID_AUTO, sched,
5010 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
5011 sched_string, sizeof(sched_string),
5012 "Timeshare scheduler implementation");
5013
5014 static int
5015 sysctl_cpu_quiescent_counter_interval SYSCTL_HANDLER_ARGS
5016 {
5017 #pragma unused(arg1, arg2)
5018
5019 uint32_t local_min_interval_us = smr_cpu_checkin_get_min_interval_us();
5020
5021 int error = sysctl_handle_int(oidp, &local_min_interval_us, 0, req);
5022 if (error || !req->newptr) {
5023 return error;
5024 }
5025
5026 smr_cpu_checkin_set_min_interval_us(local_min_interval_us);
5027
5028 return 0;
5029 }
5030
5031 SYSCTL_PROC(_kern, OID_AUTO, cpu_checkin_interval,
5032 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
5033 0, 0,
5034 sysctl_cpu_quiescent_counter_interval, "I",
5035 "Quiescent CPU checkin interval (microseconds)");
5036
5037 /*
5038 * Allow the precise user/kernel time sysctl to be set, but don't allow it to
5039 * affect anything. Some tools expect to be able to set this, even though
5040 * runtime configuration is no longer supported.
5041 */
5042
5043 static int
5044 sysctl_precise_user_kernel_time SYSCTL_HANDLER_ARGS
5045 {
5046 #if PRECISE_USER_KERNEL_TIME
5047 int dummy_set = 1;
5048 #else /* PRECISE_USER_KERNEL_TIME */
5049 int dummy_set = 0;
5050 #endif /* !PRECISE_USER_KERNEL_TIME */
5051 return sysctl_handle_int(oidp, &dummy_set, 0, req);
5052 }
5053
5054 SYSCTL_PROC(_kern, OID_AUTO, precise_user_kernel_time,
5055 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
5056 0, 0, sysctl_precise_user_kernel_time, "I",
5057 "Precise accounting of kernel vs. user time (deprecated)");
5058
5059 #if CONFIG_PERVASIVE_ENERGY && HAS_CPU_DPE_COUNTER
5060 __security_const_late static int pervasive_energy = 1;
5061 #else /* CONFIG_PERVASIVE_ENERGY && HAS_CPU_DPE_COUNTER */
5062 __security_const_late static int pervasive_energy = 0;
5063 #endif /* !CONFIG_PERVASIVE_ENERGY || !HAS_CPU_DPE_COUNTER */
5064
5065 SYSCTL_INT(_kern, OID_AUTO, pervasive_energy,
5066 CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED, &pervasive_energy, 0, "");
5067
5068 /* Parameters related to timer coalescing tuning, to be replaced
5069 * with a dedicated systemcall in the future.
5070 */
5071 /* Enable processing pending timers in the context of any other interrupt
5072 * Coalescing tuning parameters for various thread/task attributes */
5073 STATIC int
5074 sysctl_timer_user_us_kernel_abstime SYSCTL_HANDLER_ARGS
5075 {
5076 #pragma unused(oidp)
5077 int size = arg2; /* subcommand*/
5078 int error;
5079 int changed = 0;
5080 uint64_t old_value_ns;
5081 uint64_t new_value_ns;
5082 uint64_t value_abstime;
5083 if (size == sizeof(uint32_t)) {
5084 value_abstime = *((uint32_t *)arg1);
5085 } else if (size == sizeof(uint64_t)) {
5086 value_abstime = *((uint64_t *)arg1);
5087 } else {
5088 return ENOTSUP;
5089 }
5090
5091 absolutetime_to_nanoseconds(value_abstime, &old_value_ns);
5092 error = sysctl_io_number(req, old_value_ns, sizeof(old_value_ns), &new_value_ns, &changed);
5093 if ((error) || (!changed)) {
5094 return error;
5095 }
5096
5097 nanoseconds_to_absolutetime(new_value_ns, &value_abstime);
5098 if (size == sizeof(uint32_t)) {
5099 *((uint32_t *)arg1) = (uint32_t)value_abstime;
5100 } else {
5101 *((uint64_t *)arg1) = value_abstime;
5102 }
5103 return error;
5104 }
5105
5106 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_bg_scale,
5107 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5108 &tcoal_prio_params.timer_coalesce_bg_shift, 0, "");
5109 SYSCTL_PROC(_kern, OID_AUTO, timer_resort_threshold_ns,
5110 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5111 &tcoal_prio_params.timer_resort_threshold_abstime,
5112 sizeof(tcoal_prio_params.timer_resort_threshold_abstime),
5113 sysctl_timer_user_us_kernel_abstime,
5114 "Q", "");
5115 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_bg_ns_max,
5116 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5117 &tcoal_prio_params.timer_coalesce_bg_abstime_max,
5118 sizeof(tcoal_prio_params.timer_coalesce_bg_abstime_max),
5119 sysctl_timer_user_us_kernel_abstime,
5120 "Q", "");
5121
5122 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_kt_scale,
5123 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5124 &tcoal_prio_params.timer_coalesce_kt_shift, 0, "");
5125
5126 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_kt_ns_max,
5127 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5128 &tcoal_prio_params.timer_coalesce_kt_abstime_max,
5129 sizeof(tcoal_prio_params.timer_coalesce_kt_abstime_max),
5130 sysctl_timer_user_us_kernel_abstime,
5131 "Q", "");
5132
5133 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_fp_scale,
5134 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5135 &tcoal_prio_params.timer_coalesce_fp_shift, 0, "");
5136
5137 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_fp_ns_max,
5138 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5139 &tcoal_prio_params.timer_coalesce_fp_abstime_max,
5140 sizeof(tcoal_prio_params.timer_coalesce_fp_abstime_max),
5141 sysctl_timer_user_us_kernel_abstime,
5142 "Q", "");
5143
5144 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_ts_scale,
5145 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5146 &tcoal_prio_params.timer_coalesce_ts_shift, 0, "");
5147
5148 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_ts_ns_max,
5149 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5150 &tcoal_prio_params.timer_coalesce_ts_abstime_max,
5151 sizeof(tcoal_prio_params.timer_coalesce_ts_abstime_max),
5152 sysctl_timer_user_us_kernel_abstime,
5153 "Q", "");
5154
5155 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier0_scale,
5156 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5157 &tcoal_prio_params.latency_qos_scale[0], 0, "");
5158
5159 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier0_ns_max,
5160 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5161 &tcoal_prio_params.latency_qos_abstime_max[0],
5162 sizeof(tcoal_prio_params.latency_qos_abstime_max[0]),
5163 sysctl_timer_user_us_kernel_abstime,
5164 "Q", "");
5165
5166 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier1_scale,
5167 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5168 &tcoal_prio_params.latency_qos_scale[1], 0, "");
5169
5170 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier1_ns_max,
5171 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5172 &tcoal_prio_params.latency_qos_abstime_max[1],
5173 sizeof(tcoal_prio_params.latency_qos_abstime_max[1]),
5174 sysctl_timer_user_us_kernel_abstime,
5175 "Q", "");
5176
5177 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier2_scale,
5178 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5179 &tcoal_prio_params.latency_qos_scale[2], 0, "");
5180
5181 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier2_ns_max,
5182 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5183 &tcoal_prio_params.latency_qos_abstime_max[2],
5184 sizeof(tcoal_prio_params.latency_qos_abstime_max[2]),
5185 sysctl_timer_user_us_kernel_abstime,
5186 "Q", "");
5187
5188 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier3_scale,
5189 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5190 &tcoal_prio_params.latency_qos_scale[3], 0, "");
5191
5192 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier3_ns_max,
5193 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5194 &tcoal_prio_params.latency_qos_abstime_max[3],
5195 sizeof(tcoal_prio_params.latency_qos_abstime_max[3]),
5196 sysctl_timer_user_us_kernel_abstime,
5197 "Q", "");
5198
5199 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier4_scale,
5200 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5201 &tcoal_prio_params.latency_qos_scale[4], 0, "");
5202
5203 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier4_ns_max,
5204 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5205 &tcoal_prio_params.latency_qos_abstime_max[4],
5206 sizeof(tcoal_prio_params.latency_qos_abstime_max[4]),
5207 sysctl_timer_user_us_kernel_abstime,
5208 "Q", "");
5209
5210 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier5_scale,
5211 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5212 &tcoal_prio_params.latency_qos_scale[5], 0, "");
5213
5214 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier5_ns_max,
5215 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5216 &tcoal_prio_params.latency_qos_abstime_max[5],
5217 sizeof(tcoal_prio_params.latency_qos_abstime_max[5]),
5218 sysctl_timer_user_us_kernel_abstime,
5219 "Q", "");
5220
5221 /* Communicate the "user idle level" heuristic to the timer layer, and
5222 * potentially other layers in the future.
5223 */
5224
5225 static int
5226 timer_user_idle_level(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
5227 {
5228 int new_value = 0, old_value = 0, changed = 0, error;
5229
5230 old_value = timer_get_user_idle_level();
5231
5232 error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
5233
5234 if (error == 0 && changed) {
5235 if (timer_set_user_idle_level(new_value) != KERN_SUCCESS) {
5236 error = ERANGE;
5237 }
5238 }
5239
5240 return error;
5241 }
5242
5243 SYSCTL_PROC(_machdep, OID_AUTO, user_idle_level,
5244 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
5245 0, 0,
5246 timer_user_idle_level, "I", "User idle level heuristic, 0-128");
5247
5248 #if DEVELOPMENT || DEBUG
5249 /*
5250 * Basic console mode for games; used for development purposes only.
5251 * Final implementation for this feature (with possible removal of
5252 * sysctl) tracked via rdar://101215873.
5253 */
5254 static int console_mode = 0;
5255 SYSCTL_INT(_kern, OID_AUTO, console_mode,
5256 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_MASKED,
5257 &console_mode, 0, "Game Console Mode");
5258 #endif /* DEVELOPMENT || DEBUG */
5259
5260
5261 #if HYPERVISOR
5262 SYSCTL_INT(_kern, OID_AUTO, hv_support,
5263 CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
5264 &hv_support_available, 0, "");
5265
5266 SYSCTL_INT(_kern, OID_AUTO, hv_disable,
5267 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5268 &hv_disable, 0, "");
5269
5270 #endif /* HYPERVISOR */
5271
5272 #if DEVELOPMENT || DEBUG
5273 extern uint64_t driverkit_checkin_timed_out;
5274 SYSCTL_QUAD(_kern, OID_AUTO, driverkit_checkin_timed_out,
5275 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
5276 &driverkit_checkin_timed_out, "timestamp of dext checkin timeout");
5277 #endif
5278
5279 #if CONFIG_DARKBOOT
5280 STATIC int
5281 sysctl_darkboot SYSCTL_HANDLER_ARGS
5282 {
5283 int err = 0, value = 0;
5284 #pragma unused(oidp, arg1, arg2, err, value, req)
5285
5286 /*
5287 * Handle the sysctl request.
5288 *
5289 * If this is a read, the function will set the value to the current darkboot value. Otherwise,
5290 * we'll get the request identifier into "value" and then we can honor it.
5291 */
5292 if ((err = sysctl_io_number(req, darkboot, sizeof(int), &value, NULL)) != 0) {
5293 goto exit;
5294 }
5295
5296 /* writing requested, let's process the request */
5297 if (req->newptr) {
5298 /* writing is protected by an entitlement */
5299 if (priv_check_cred(kauth_cred_get(), PRIV_DARKBOOT, 0) != 0) {
5300 err = EPERM;
5301 goto exit;
5302 }
5303
5304 switch (value) {
5305 case MEMORY_MAINTENANCE_DARK_BOOT_UNSET:
5306 /*
5307 * If the darkboot sysctl is unset, the NVRAM variable
5308 * must be unset too. If that's not the case, it means
5309 * someone is doing something crazy and not supported.
5310 */
5311 if (darkboot != 0) {
5312 int ret = PERemoveNVRAMProperty(MEMORY_MAINTENANCE_DARK_BOOT_NVRAM_NAME);
5313 if (ret) {
5314 darkboot = 0;
5315 } else {
5316 err = EINVAL;
5317 }
5318 }
5319 break;
5320 case MEMORY_MAINTENANCE_DARK_BOOT_SET:
5321 darkboot = 1;
5322 break;
5323 case MEMORY_MAINTENANCE_DARK_BOOT_SET_PERSISTENT: {
5324 /*
5325 * Set the NVRAM and update 'darkboot' in case
5326 * of success. Otherwise, do not update
5327 * 'darkboot' and report the failure.
5328 */
5329 if (PEWriteNVRAMBooleanProperty(MEMORY_MAINTENANCE_DARK_BOOT_NVRAM_NAME, TRUE)) {
5330 darkboot = 1;
5331 } else {
5332 err = EINVAL;
5333 }
5334
5335 break;
5336 }
5337 default:
5338 err = EINVAL;
5339 }
5340 }
5341
5342 exit:
5343 return err;
5344 }
5345
5346 SYSCTL_PROC(_kern, OID_AUTO, darkboot,
5347 CTLFLAG_KERN | CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_ANYBODY,
5348 0, 0, sysctl_darkboot, "I", "");
5349 #endif /* CONFIG_DARKBOOT */
5350
5351 #if DEVELOPMENT || DEBUG
5352 #include <sys/sysent.h>
5353 /* This should result in a fatal exception, verifying that "sysent" is
5354 * write-protected.
5355 */
5356 static int
5357 kern_sysent_write(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
5358 {
5359 uint64_t new_value = 0, old_value = 0;
5360 int changed = 0, error;
5361
5362 error = sysctl_io_number(req, old_value, sizeof(uint64_t), &new_value, &changed);
5363 if ((error == 0) && changed) {
5364 volatile uint32_t *wraddr = __DECONST(uint32_t *, &sysent[0]);
5365 *wraddr = 0;
5366 printf("sysent[0] write succeeded\n");
5367 }
5368 return error;
5369 }
5370
5371 SYSCTL_PROC(_kern, OID_AUTO, sysent_const_check,
5372 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
5373 0, 0,
5374 kern_sysent_write, "I", "Attempt sysent[0] write");
5375
5376 #endif
5377
5378 #if DEVELOPMENT || DEBUG
5379 SYSCTL_COMPAT_INT(_kern, OID_AUTO, development, CTLFLAG_RD | CTLFLAG_MASKED | CTLFLAG_KERN, NULL, 1, "");
5380 #else
5381 SYSCTL_COMPAT_INT(_kern, OID_AUTO, development, CTLFLAG_RD | CTLFLAG_MASKED, NULL, 0, "");
5382 #endif
5383
5384 SYSCTL_INT(_kern, OID_AUTO, serverperfmode, CTLFLAG_RD, &serverperfmode, 0, "");
5385
5386 #if DEVELOPMENT || DEBUG
5387
5388 decl_lck_spin_data(, spinlock_panic_test_lock);
5389
5390 __attribute__((noreturn))
5391 static void
5392 spinlock_panic_test_acquire_spinlock(void * arg __unused, wait_result_t wres __unused)
5393 {
5394 lck_spin_lock(&spinlock_panic_test_lock);
5395 while (1) {
5396 ;
5397 }
5398 }
5399
5400 static int
5401 sysctl_spinlock_panic_test SYSCTL_HANDLER_ARGS
5402 {
5403 #pragma unused(oidp, arg1, arg2)
5404 if (req->newlen == 0) {
5405 return EINVAL;
5406 }
5407
5408 thread_t panic_spinlock_thread;
5409 /* Initialize panic spinlock */
5410 lck_grp_t * panic_spinlock_grp;
5411 lck_grp_attr_t * panic_spinlock_grp_attr;
5412 lck_attr_t * panic_spinlock_attr;
5413
5414 panic_spinlock_grp_attr = lck_grp_attr_alloc_init();
5415 panic_spinlock_grp = lck_grp_alloc_init("panic_spinlock", panic_spinlock_grp_attr);
5416 panic_spinlock_attr = lck_attr_alloc_init();
5417
5418 lck_spin_init(&spinlock_panic_test_lock, panic_spinlock_grp, panic_spinlock_attr);
5419
5420
5421 /* Create thread to acquire spinlock */
5422 if (kernel_thread_start(spinlock_panic_test_acquire_spinlock, NULL, &panic_spinlock_thread) != KERN_SUCCESS) {
5423 return EBUSY;
5424 }
5425
5426 /* Try to acquire spinlock -- should panic eventually */
5427 lck_spin_lock(&spinlock_panic_test_lock);
5428 while (1) {
5429 ;
5430 }
5431 }
5432
5433 __attribute__((noreturn))
5434 static void
5435 simultaneous_panic_worker
5436 (void * arg, wait_result_t wres __unused)
5437 {
5438 atomic_int *start_panic = (atomic_int *)arg;
5439
5440 while (!atomic_load(start_panic)) {
5441 ;
5442 }
5443 panic("SIMULTANEOUS PANIC TEST: INITIATING PANIC FROM CPU %d", cpu_number());
5444 __builtin_unreachable();
5445 }
5446
5447 static int
5448 sysctl_simultaneous_panic_test SYSCTL_HANDLER_ARGS
5449 {
5450 #pragma unused(oidp, arg1, arg2)
5451 if (req->newlen == 0) {
5452 return EINVAL;
5453 }
5454
5455 int i = 0, threads_to_create = 2 * processor_count;
5456 atomic_int start_panic = 0;
5457 unsigned int threads_created = 0;
5458 thread_t new_panic_thread;
5459
5460 for (i = threads_to_create; i > 0; i--) {
5461 if (kernel_thread_start(simultaneous_panic_worker, (void *) &start_panic, &new_panic_thread) == KERN_SUCCESS) {
5462 threads_created++;
5463 }
5464 }
5465
5466 /* FAIL if we couldn't create at least processor_count threads */
5467 if (threads_created < processor_count) {
5468 panic("SIMULTANEOUS PANIC TEST: FAILED TO CREATE ENOUGH THREADS, ONLY CREATED %d (of %d)",
5469 threads_created, threads_to_create);
5470 }
5471
5472 atomic_exchange(&start_panic, 1);
5473 while (1) {
5474 ;
5475 }
5476 }
5477
5478 extern unsigned int panic_test_failure_mode;
5479 SYSCTL_INT(_debug, OID_AUTO, xnu_panic_failure_mode, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_KERN, &panic_test_failure_mode, 0, "panic/debugger test failure mode");
5480
5481 extern unsigned int panic_test_action_count;
5482 SYSCTL_INT(_debug, OID_AUTO, xnu_panic_action_count, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_KERN, &panic_test_action_count, 0, "panic/debugger test action count");
5483
5484 extern unsigned int panic_test_case;
5485 SYSCTL_INT(_debug, OID_AUTO, xnu_panic_test_case, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_KERN, &panic_test_case, 0, "panic/debugger testcase");
5486
5487 SYSCTL_PROC(_debug, OID_AUTO, xnu_spinlock_panic_test, CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_MASKED, 0, 0, sysctl_spinlock_panic_test, "A", "spinlock panic test");
5488 SYSCTL_PROC(_debug, OID_AUTO, xnu_simultaneous_panic_test, CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_MASKED, 0, 0, sysctl_simultaneous_panic_test, "A", "simultaneous panic test");
5489
5490 extern int exc_resource_threads_enabled;
5491 SYSCTL_INT(_kern, OID_AUTO, exc_resource_threads_enabled, CTLFLAG_RW | CTLFLAG_LOCKED, &exc_resource_threads_enabled, 0, "exc_resource thread limit enabled");
5492
5493
5494 #endif /* DEVELOPMENT || DEBUG */
5495
5496 #if BUILT_LTO
5497 static int _built_lto = 1;
5498 #else // BUILT_LTO
5499 static int _built_lto = 0;
5500 #endif // !BUILT_LTO
5501
5502 SYSCTL_INT(_kern, OID_AUTO, link_time_optimized, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN, &_built_lto, 0, "Whether the kernel was built with Link Time Optimization enabled");
5503
5504 #if CONFIG_THREAD_GROUPS
5505 #if DEVELOPMENT || DEBUG
5506
5507 static int
5508 sysctl_get_thread_group_id SYSCTL_HANDLER_ARGS
5509 {
5510 #pragma unused(arg1, arg2, oidp)
5511 uint64_t thread_group_id = thread_group_get_id(thread_group_get(current_thread()));
5512 return SYSCTL_OUT(req, &thread_group_id, sizeof(thread_group_id));
5513 }
5514
5515 SYSCTL_PROC(_kern, OID_AUTO, thread_group_id, CTLFLAG_RD | CTLFLAG_LOCKED | CTLTYPE_QUAD,
5516 0, 0, &sysctl_get_thread_group_id, "I", "thread group id of the thread");
5517
5518 STATIC int
5519 sysctl_thread_group_count(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
5520 {
5521 int value = thread_group_count();
5522 return sysctl_io_number(req, value, sizeof(value), NULL, NULL);
5523 }
5524
5525 SYSCTL_PROC(_kern, OID_AUTO, thread_group_count, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5526 0, 0, &sysctl_thread_group_count, "I", "count of thread groups");
5527
5528 #endif /* DEVELOPMENT || DEBUG */
5529 const uint32_t thread_groups_supported = 1;
5530 #else /* CONFIG_THREAD_GROUPS */
5531 const uint32_t thread_groups_supported = 0;
5532 #endif /* CONFIG_THREAD_GROUPS */
5533
5534 STATIC int
5535 sysctl_thread_groups_supported(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
5536 {
5537 int value = thread_groups_supported;
5538 return sysctl_io_number(req, value, sizeof(value), NULL, NULL);
5539 }
5540
5541 SYSCTL_PROC(_kern, OID_AUTO, thread_groups_supported, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5542 0, 0, &sysctl_thread_groups_supported, "I", "thread groups supported");
5543
5544 static int
5545 sysctl_grade_cputype SYSCTL_HANDLER_ARGS
5546 {
5547 #pragma unused(arg1, arg2, oidp)
5548 int error = 0;
5549 int type_tuple[2] = {};
5550 int return_value = 0;
5551
5552 error = SYSCTL_IN(req, &type_tuple, sizeof(type_tuple));
5553
5554 if (error) {
5555 return error;
5556 }
5557
5558 return_value = grade_binary(type_tuple[0], type_tuple[1] & ~CPU_SUBTYPE_MASK, type_tuple[1] & CPU_SUBTYPE_MASK, FALSE);
5559
5560 error = SYSCTL_OUT(req, &return_value, sizeof(return_value));
5561
5562 if (error) {
5563 return error;
5564 }
5565
5566 return error;
5567 }
5568
5569 SYSCTL_PROC(_kern, OID_AUTO, grade_cputype,
5570 CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_MASKED | CTLFLAG_LOCKED | CTLTYPE_OPAQUE,
5571 0, 0, &sysctl_grade_cputype, "S",
5572 "grade value of cpu_type_t+cpu_sub_type_t");
5573
5574 extern boolean_t allow_direct_handoff;
5575 SYSCTL_INT(_kern, OID_AUTO, direct_handoff,
5576 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5577 &allow_direct_handoff, 0, "Enable direct handoff for realtime threads");
5578
5579 #if DEVELOPMENT || DEBUG
5580
5581 SYSCTL_QUAD(_kern, OID_AUTO, phys_carveout_pa, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5582 &phys_carveout_pa,
5583 "base physical address of the phys_carveout_mb boot-arg region");
5584 SYSCTL_QUAD(_kern, OID_AUTO, phys_carveout_va, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5585 &phys_carveout,
5586 "base virtual address of the phys_carveout_mb boot-arg region");
5587 SYSCTL_QUAD(_kern, OID_AUTO, phys_carveout_size, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5588 &phys_carveout_size,
5589 "size in bytes of the phys_carveout_mb boot-arg region");
5590
5591
5592 static int
5593 cseg_wedge_thread SYSCTL_HANDLER_ARGS
5594 {
5595 #pragma unused(arg1, arg2)
5596
5597 int error, val = 0;
5598 error = sysctl_handle_int(oidp, &val, 0, req);
5599 if (error || val == 0) {
5600 return error;
5601 }
5602
5603 do_cseg_wedge_thread();
5604 return 0;
5605 }
5606 SYSCTL_PROC(_kern, OID_AUTO, cseg_wedge_thread, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_MASKED, 0, 0, cseg_wedge_thread, "I", "wedge c_seg thread");
5607
5608 static int
5609 cseg_unwedge_thread SYSCTL_HANDLER_ARGS
5610 {
5611 #pragma unused(arg1, arg2)
5612
5613 int error, val = 0;
5614 error = sysctl_handle_int(oidp, &val, 0, req);
5615 if (error || val == 0) {
5616 return error;
5617 }
5618
5619 do_cseg_unwedge_thread();
5620 return 0;
5621 }
5622 SYSCTL_PROC(_kern, OID_AUTO, cseg_unwedge_thread, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_MASKED, 0, 0, cseg_unwedge_thread, "I", "unstuck c_seg thread");
5623
5624 static atomic_int wedge_thread_should_wake = 0;
5625
5626 static int
5627 unwedge_thread SYSCTL_HANDLER_ARGS
5628 {
5629 #pragma unused(arg1, arg2)
5630 int error, val = 0;
5631 error = sysctl_handle_int(oidp, &val, 0, req);
5632 if (error || val == 0) {
5633 return error;
5634 }
5635
5636 atomic_store(&wedge_thread_should_wake, 1);
5637 return 0;
5638 }
5639
5640 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");
5641
5642 static int
5643 wedge_thread SYSCTL_HANDLER_ARGS
5644 {
5645 #pragma unused(arg1, arg2)
5646
5647 int error, val = 0;
5648 error = sysctl_handle_int(oidp, &val, 0, req);
5649 if (error || val == 0) {
5650 return error;
5651 }
5652
5653 uint64_t interval = 1;
5654 nanoseconds_to_absolutetime(1000 * 1000 * 50, &interval);
5655
5656 atomic_store(&wedge_thread_should_wake, 0);
5657 while (!atomic_load(&wedge_thread_should_wake)) {
5658 tsleep1(NULL, 0, "wedge_thread", mach_absolute_time() + interval, NULL);
5659 }
5660
5661 return 0;
5662 }
5663
5664 SYSCTL_PROC(_kern, OID_AUTO, wedge_thread,
5665 CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED, 0, 0, wedge_thread, "I",
5666 "wedge this thread so it cannot be cleaned up");
5667
5668 static int
5669 sysctl_total_corpses_count SYSCTL_HANDLER_ARGS
5670 {
5671 #pragma unused(oidp, arg1, arg2)
5672 extern unsigned long total_corpses_count(void);
5673
5674 unsigned long corpse_count_long = total_corpses_count();
5675 unsigned int corpse_count = (unsigned int)MIN(corpse_count_long, UINT_MAX);
5676 return sysctl_io_opaque(req, &corpse_count, sizeof(corpse_count), NULL);
5677 }
5678
5679 SYSCTL_PROC(_kern, OID_AUTO, total_corpses_count,
5680 CTLFLAG_RD | CTLFLAG_ANYBODY | CTLFLAG_LOCKED, 0, 0,
5681 sysctl_total_corpses_count, "I", "total corpses on the system");
5682
5683 static int
5684 sysctl_turnstile_test_prim_lock SYSCTL_HANDLER_ARGS;
5685 static int
5686 sysctl_turnstile_test_prim_unlock SYSCTL_HANDLER_ARGS;
5687 int
5688 tstile_test_prim_lock(boolean_t use_hashtable);
5689 int
5690 tstile_test_prim_unlock(boolean_t use_hashtable);
5691
5692 static int
5693 sysctl_turnstile_test_prim_lock SYSCTL_HANDLER_ARGS
5694 {
5695 #pragma unused(arg1, arg2)
5696 int error, val = 0;
5697 error = sysctl_handle_int(oidp, &val, 0, req);
5698 if (error || val == 0) {
5699 return error;
5700 }
5701 switch (val) {
5702 case SYSCTL_TURNSTILE_TEST_USER_DEFAULT:
5703 case SYSCTL_TURNSTILE_TEST_USER_HASHTABLE:
5704 case SYSCTL_TURNSTILE_TEST_KERNEL_DEFAULT:
5705 case SYSCTL_TURNSTILE_TEST_KERNEL_HASHTABLE:
5706 return tstile_test_prim_lock(val);
5707 default:
5708 return error;
5709 }
5710 }
5711
5712 static int
5713 sysctl_turnstile_test_prim_unlock SYSCTL_HANDLER_ARGS
5714 {
5715 #pragma unused(arg1, arg2)
5716 int error, val = 0;
5717 error = sysctl_handle_int(oidp, &val, 0, req);
5718 if (error || val == 0) {
5719 return error;
5720 }
5721 switch (val) {
5722 case SYSCTL_TURNSTILE_TEST_USER_DEFAULT:
5723 case SYSCTL_TURNSTILE_TEST_USER_HASHTABLE:
5724 case SYSCTL_TURNSTILE_TEST_KERNEL_DEFAULT:
5725 case SYSCTL_TURNSTILE_TEST_KERNEL_HASHTABLE:
5726 return tstile_test_prim_unlock(val);
5727 default:
5728 return error;
5729 }
5730 }
5731
5732 SYSCTL_PROC(_kern, OID_AUTO, turnstiles_test_lock, CTLFLAG_WR | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED,
5733 0, 0, sysctl_turnstile_test_prim_lock, "I", "turnstiles test lock");
5734
5735 SYSCTL_PROC(_kern, OID_AUTO, turnstiles_test_unlock, CTLFLAG_WR | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED,
5736 0, 0, sysctl_turnstile_test_prim_unlock, "I", "turnstiles test unlock");
5737
5738 int
5739 turnstile_get_boost_stats_sysctl(void *req);
5740 int
5741 turnstile_get_unboost_stats_sysctl(void *req);
5742 static int
5743 sysctl_turnstile_boost_stats SYSCTL_HANDLER_ARGS;
5744 static int
5745 sysctl_turnstile_unboost_stats SYSCTL_HANDLER_ARGS;
5746 extern uint64_t thread_block_on_turnstile_count;
5747 extern uint64_t thread_block_on_regular_waitq_count;
5748
5749 static int
5750 sysctl_turnstile_boost_stats SYSCTL_HANDLER_ARGS
5751 {
5752 #pragma unused(arg1, arg2, oidp)
5753 return turnstile_get_boost_stats_sysctl(req);
5754 }
5755
5756 static int
5757 sysctl_turnstile_unboost_stats SYSCTL_HANDLER_ARGS
5758 {
5759 #pragma unused(arg1, arg2, oidp)
5760 return turnstile_get_unboost_stats_sysctl(req);
5761 }
5762
5763 SYSCTL_PROC(_kern, OID_AUTO, turnstile_boost_stats, CTLFLAG_RD | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED | CTLTYPE_STRUCT,
5764 0, 0, sysctl_turnstile_boost_stats, "S", "turnstiles boost stats");
5765 SYSCTL_PROC(_kern, OID_AUTO, turnstile_unboost_stats, CTLFLAG_RD | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED | CTLTYPE_STRUCT,
5766 0, 0, sysctl_turnstile_unboost_stats, "S", "turnstiles unboost stats");
5767 SYSCTL_QUAD(_kern, OID_AUTO, thread_block_count_on_turnstile,
5768 CTLFLAG_RD | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED,
5769 &thread_block_on_turnstile_count, "thread blocked on turnstile count");
5770 SYSCTL_QUAD(_kern, OID_AUTO, thread_block_count_on_reg_waitq,
5771 CTLFLAG_RD | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED,
5772 &thread_block_on_regular_waitq_count, "thread blocked on regular waitq count");
5773
5774 #if CONFIG_PV_TICKET
5775
5776 extern int ticket_lock_spins;
5777 SYSCTL_INT(_kern, OID_AUTO, ticket_lock_spins,
5778 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5779 &ticket_lock_spins, 0, "loops before hypercall");
5780
5781 #if (DEBUG || DEVELOPMENT)
5782
5783 /* PV ticket lock stats */
5784
5785 SYSCTL_SCALABLE_COUNTER(_kern, ticket_lock_kicks, ticket_kick_count,
5786 "ticket lock kicks");
5787 SYSCTL_SCALABLE_COUNTER(_kern, ticket_lock_waits, ticket_wait_count,
5788 "ticket lock waits");
5789 SYSCTL_SCALABLE_COUNTER(_kern, ticket_lock_already, ticket_already_count,
5790 "ticket lock already unlocked");
5791 SYSCTL_SCALABLE_COUNTER(_kern, ticket_lock_just_unlock, ticket_just_unlock,
5792 "ticket unlock without kick");
5793 SYSCTL_SCALABLE_COUNTER(_kern, ticket_lock_wflag_cleared, ticket_wflag_cleared,
5794 "ticket lock wait flag cleared");
5795 SYSCTL_SCALABLE_COUNTER(_kern, ticket_lock_wflag_still, ticket_wflag_still,
5796 "ticket lock wait flag not cleared");
5797 SYSCTL_SCALABLE_COUNTER(_kern, ticket_lock_spin_count, ticket_spin_count,
5798 "ticket lock spin count");
5799
5800 /* sysctl kern.hcall_probe=n -- does hypercall #n exist? */
5801
5802 static int
5803 sysctl_hcall_probe SYSCTL_HANDLER_ARGS
5804 {
5805 char instr[20];
5806
5807 if (!req->newptr) {
5808 return 0;
5809 }
5810 if (req->newlen >= sizeof(instr)) {
5811 return EOVERFLOW;
5812 }
5813
5814 int error = SYSCTL_IN(req, instr, req->newlen);
5815 if (error) {
5816 return error;
5817 }
5818 instr[req->newlen] = '\0';
5819
5820 int hcall = 0;
5821 error = sscanf(instr, "%d", &hcall);
5822 if (error != 1 || hcall < 0) {
5823 return EINVAL;
5824 }
5825 uprintf("%savailable\n",
5826 hvg_is_hcall_available((hvg_hcall_code_t)hcall) ? "" : "not ");
5827 return 0;
5828 }
5829
5830 SYSCTL_PROC(_kern, OID_AUTO, hcall_probe,
5831 CTLTYPE_STRING | CTLFLAG_WR | CTLFLAG_LOCKED | CTLFLAG_MASKED,
5832 0, 0, sysctl_hcall_probe, "A", "probe hypercall by id");
5833
5834 #endif /* (DEBUG || DEVELOPMENT) */
5835 #endif /* CONFIG_PV_TICKET */
5836
5837 #if defined(__x86_64__)
5838 extern uint64_t MutexSpin;
5839
5840 SYSCTL_QUAD(_kern, OID_AUTO, mutex_spin_abs, CTLFLAG_RW, &MutexSpin,
5841 "Spin time in abs for acquiring a kernel mutex");
5842 #else
5843 extern machine_timeout_t MutexSpin;
5844
5845 SYSCTL_QUAD(_kern, OID_AUTO, mutex_spin_abs, CTLFLAG_RW, &MutexSpin,
5846 "Spin time in abs for acquiring a kernel mutex");
5847 #endif
5848
5849 extern uint64_t low_MutexSpin;
5850 extern int64_t high_MutexSpin;
5851 extern unsigned int real_ncpus;
5852
5853 SYSCTL_QUAD(_kern, OID_AUTO, low_mutex_spin_abs, CTLFLAG_RW, &low_MutexSpin,
5854 "Low spin threshold in abs for acquiring a kernel mutex");
5855
5856 static int
5857 sysctl_high_mutex_spin_ns SYSCTL_HANDLER_ARGS
5858 {
5859 #pragma unused(oidp, arg1, arg2)
5860 int error;
5861 int64_t val = 0;
5862 int64_t res;
5863
5864 /* Check if the user is writing to high_MutexSpin, or just reading it */
5865 if (req->newptr) {
5866 error = SYSCTL_IN(req, &val, sizeof(val));
5867 if (error || (val < 0 && val != -1)) {
5868 return error;
5869 }
5870 high_MutexSpin = val;
5871 }
5872
5873 if (high_MutexSpin >= 0) {
5874 res = high_MutexSpin;
5875 } else {
5876 res = low_MutexSpin * real_ncpus;
5877 }
5878 return SYSCTL_OUT(req, &res, sizeof(res));
5879 }
5880 SYSCTL_PROC(_kern, OID_AUTO, high_mutex_spin_abs, CTLFLAG_RW | CTLTYPE_QUAD, 0, 0, sysctl_high_mutex_spin_ns, "I",
5881 "High spin threshold in abs for acquiring a kernel mutex");
5882
5883 #if defined (__x86_64__)
5884
5885 semaphore_t sysctl_test_panic_with_thread_sem;
5886
5887 #pragma clang diagnostic push
5888 #pragma clang diagnostic ignored "-Winfinite-recursion" /* rdar://38801963 */
5889 __attribute__((noreturn))
5890 static void
5891 panic_thread_test_child_spin(void * arg, wait_result_t wres)
5892 {
5893 static int panic_thread_recurse_count = 5;
5894
5895 if (panic_thread_recurse_count > 0) {
5896 panic_thread_recurse_count--;
5897 panic_thread_test_child_spin(arg, wres);
5898 }
5899
5900 semaphore_signal(sysctl_test_panic_with_thread_sem);
5901 while (1) {
5902 ;
5903 }
5904 }
5905 #pragma clang diagnostic pop
5906
5907 static void
5908 panic_thread_test_child_park(void * arg __unused, wait_result_t wres __unused)
5909 {
5910 int event;
5911
5912 assert_wait(&event, THREAD_UNINT);
5913 semaphore_signal(sysctl_test_panic_with_thread_sem);
5914 thread_block(panic_thread_test_child_park);
5915 }
5916
5917 static int
5918 sysctl_test_panic_with_thread SYSCTL_HANDLER_ARGS
5919 {
5920 #pragma unused(arg1, arg2)
5921 int rval = 0;
5922 char str[16] = { '\0' };
5923 thread_t child_thread = THREAD_NULL;
5924
5925 rval = sysctl_handle_string(oidp, str, sizeof(str), req);
5926 if (rval != 0 || !req->newptr) {
5927 return EINVAL;
5928 }
5929
5930 semaphore_create(kernel_task, &sysctl_test_panic_with_thread_sem, SYNC_POLICY_FIFO, 0);
5931
5932 /* Create thread to spin or park in continuation */
5933 if (strncmp("spin", str, strlen("spin")) == 0) {
5934 if (kernel_thread_start(panic_thread_test_child_spin, NULL, &child_thread) != KERN_SUCCESS) {
5935 semaphore_destroy(kernel_task, sysctl_test_panic_with_thread_sem);
5936 return EBUSY;
5937 }
5938 } else if (strncmp("continuation", str, strlen("continuation")) == 0) {
5939 if (kernel_thread_start(panic_thread_test_child_park, NULL, &child_thread) != KERN_SUCCESS) {
5940 semaphore_destroy(kernel_task, sysctl_test_panic_with_thread_sem);
5941 return EBUSY;
5942 }
5943 } else {
5944 semaphore_destroy(kernel_task, sysctl_test_panic_with_thread_sem);
5945 return EINVAL;
5946 }
5947
5948 semaphore_wait(sysctl_test_panic_with_thread_sem);
5949
5950 panic_with_thread_context(0, NULL, 0, child_thread, "testing panic_with_thread_context for thread %p", child_thread);
5951
5952 /* Not reached */
5953 return EINVAL;
5954 }
5955
5956 SYSCTL_PROC(_kern, OID_AUTO, test_panic_with_thread,
5957 CTLFLAG_MASKED | CTLFLAG_KERN | CTLFLAG_LOCKED | CTLFLAG_WR | CTLTYPE_STRING,
5958 0, 0, sysctl_test_panic_with_thread, "A", "test panic flow for backtracing a different thread");
5959 #endif /* defined (__x86_64__) */
5960
5961 static int
5962 sysctl_generate_file_permissions_guard_exception SYSCTL_HANDLER_ARGS
5963 {
5964 #pragma unused(arg1, arg2)
5965 int error, val = 0;
5966 error = sysctl_handle_int(oidp, &val, 0, req);
5967 if (error || val == 0) {
5968 return error;
5969 }
5970 generate_file_permissions_guard_exception(0, val);
5971 return 0;
5972 }
5973
5974 SYSCTL_PROC(_kern, OID_AUTO, file_perm_guard_exception, CTLFLAG_WR | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED,
5975 0, 0, sysctl_generate_file_permissions_guard_exception, "I", "Test File Permission Guard exception");
5976
5977 #endif /* DEVELOPMENT || DEBUG */
5978
5979 extern const int copysize_limit_panic;
5980 static int
5981 sysctl_get_owned_vmobjects SYSCTL_HANDLER_ARGS
5982 {
5983 #pragma unused(oidp, arg1, arg2)
5984
5985 /* validate */
5986 if (req->newlen != sizeof(mach_port_name_t) || req->newptr == USER_ADDR_NULL ||
5987 req->oldidx != 0 || req->newidx != 0 || req->p == NULL ||
5988 (req->oldlen == 0 && req->oldptr != USER_ADDR_NULL)) {
5989 return EINVAL;
5990 }
5991
5992 int error;
5993 mach_port_name_t task_port_name;
5994 task_t task;
5995 size_t buffer_size = (req->oldptr != USER_ADDR_NULL) ? req->oldlen : 0;
5996 vmobject_list_output_t buffer = NULL;
5997 size_t output_size;
5998 size_t entries;
5999 bool free_buffer = false;
6000
6001 /* we have a "newptr" (for write) we get a task port name from the caller. */
6002 error = SYSCTL_IN(req, &task_port_name, sizeof(mach_port_name_t));
6003
6004 if (error != 0) {
6005 goto sysctl_get_vmobject_list_exit;
6006 }
6007
6008 task = port_name_to_task_read(task_port_name);
6009 if (task == TASK_NULL) {
6010 error = ESRCH;
6011 goto sysctl_get_vmobject_list_exit;
6012 }
6013
6014 bool corpse = task_is_a_corpse(task);
6015
6016 /* get the current size */
6017 size_t max_size;
6018 task_get_owned_vmobjects(task, 0, NULL, &max_size, &entries);
6019
6020 if (buffer_size && (buffer_size < sizeof(*buffer) + sizeof(vm_object_query_data_t))) {
6021 error = ENOMEM;
6022 goto sysctl_get_vmobject_list_deallocate_and_exit;
6023 }
6024
6025 if (corpse == false) {
6026 /* copy the vmobjects and vmobject data out of the task */
6027 if (buffer_size == 0) {
6028 output_size = max_size;
6029 } else {
6030 buffer_size = (buffer_size > max_size) ? max_size : buffer_size;
6031 buffer = (struct _vmobject_list_output_ *)kalloc_data(buffer_size, Z_WAITOK);
6032
6033 if (!buffer) {
6034 error = ENOMEM;
6035 goto sysctl_get_vmobject_list_deallocate_and_exit;
6036 }
6037 free_buffer = true;
6038
6039 task_get_owned_vmobjects(task, buffer_size, buffer, &output_size, &entries);
6040 }
6041 } else {
6042 vmobject_list_output_t list;
6043
6044 task_get_corpse_vmobject_list(task, &list, &max_size);
6045 assert(buffer == NULL);
6046
6047 /* copy corpse_vmobject_list to output buffer to avoid double copy */
6048 if (buffer_size) {
6049 size_t temp_size;
6050
6051 temp_size = buffer_size > max_size ? max_size : buffer_size;
6052 output_size = temp_size - sizeof(*buffer);
6053 /* whole multiple of vm_object_query_data_t */
6054 output_size = (output_size / sizeof(vm_object_query_data_t)) * sizeof(vm_object_query_data_t) + sizeof(*buffer);
6055 buffer = list;
6056 } else {
6057 output_size = max_size;
6058 }
6059 }
6060
6061 /* req->oldptr should be USER_ADDR_NULL if buffer == NULL and return the current size */
6062 /* otherwise copy buffer to oldptr and return the bytes copied */
6063 size_t num_copied, chunk_size;
6064 for (num_copied = 0, chunk_size = 0;
6065 num_copied < output_size;
6066 num_copied += chunk_size) {
6067 chunk_size = MIN(output_size - num_copied, copysize_limit_panic);
6068 error = SYSCTL_OUT(req, (char *)buffer + num_copied, chunk_size);
6069 if (error) {
6070 break;
6071 }
6072 }
6073
6074 sysctl_get_vmobject_list_deallocate_and_exit:
6075 task_deallocate(task);
6076
6077 sysctl_get_vmobject_list_exit:
6078 if (free_buffer) {
6079 kfree_data(buffer, buffer_size);
6080 }
6081
6082 return error;
6083 }
6084
6085 SYSCTL_PROC(_vm, OID_AUTO, get_owned_vmobjects,
6086 CTLTYPE_OPAQUE | CTLFLAG_RD | CTLFLAG_WR | CTLFLAG_MASKED | CTLFLAG_KERN | CTLFLAG_LOCKED | CTLFLAG_ANYBODY,
6087 0, 0, sysctl_get_owned_vmobjects, "A", "get owned vmobjects in task");
6088
6089 extern uint64_t num_static_scalable_counters;
6090 SYSCTL_QUAD(_kern, OID_AUTO, num_static_scalable_counters, CTLFLAG_RD | CTLFLAG_LOCKED, &num_static_scalable_counters, "");
6091
6092 #if SCHED_HYGIENE_DEBUG
6093 TUNABLE_DT(bool, sched_hygiene_nonspec_tb, "machine-timeouts", "nonspec-tb", "sched-hygiene-nonspec-tb", false, TUNABLE_DT_NONE);
6094 #endif /* SCHED_HYGIENE_DEBUG */
6095
6096 uuid_string_t trial_treatment_id;
6097 uuid_string_t trial_experiment_id;
6098 int trial_deployment_id = -1;
6099
6100 SYSCTL_STRING(_kern, OID_AUTO, trial_treatment_id, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_ANYBODY | CTLFLAG_EXPERIMENT, trial_treatment_id, sizeof(trial_treatment_id), "");
6101 SYSCTL_STRING(_kern, OID_AUTO, trial_experiment_id, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_ANYBODY | CTLFLAG_EXPERIMENT, trial_experiment_id, sizeof(trial_experiment_id), "");
6102 SYSCTL_INT(_kern, OID_AUTO, trial_deployment_id, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_ANYBODY | CTLFLAG_EXPERIMENT, &trial_deployment_id, 0, "");
6103
6104 #if (DEVELOPMENT || DEBUG)
6105 /* For unit testing setting factors & limits. */
6106 unsigned int testing_experiment_factor;
6107 EXPERIMENT_FACTOR_UINT(_kern, testing_experiment_factor, &testing_experiment_factor, 5, 10, "");
6108
6109 extern int exception_log_max_pid;
6110 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");
6111 #endif /* (DEVELOPMENT || DEBUG) */
6112
6113 #if DEVELOPMENT || DEBUG
6114 static int
6115 unlink_kernelcore_sysctl SYSCTL_HANDLER_ARGS
6116 {
6117 if (!req->newptr) {
6118 return EINVAL;
6119 }
6120 void IOBSDLowSpaceUnlinkKernelCore(void);
6121 IOBSDLowSpaceUnlinkKernelCore();
6122 return 0;
6123 }
6124
6125 SYSCTL_PROC(_kern, OID_AUTO, unlink_kernelcore,
6126 CTLTYPE_INT | CTLFLAG_WR | CTLFLAG_LOCKED | CTLFLAG_MASKED, 0, 0,
6127 unlink_kernelcore_sysctl, "-", "unlink the kernelcore file");
6128 #endif /* DEVELOPMENT || DEBUG */
6129
6130 #if CONFIG_IOTRACE
6131 #pragma clang diagnostic push
6132 #pragma clang diagnostic ignored "-Wcast-qual"
6133 SYSCTL_INT(_debug, OID_AUTO, MMIOtrace,
6134 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
6135 (int *)&mmiotrace_enabled, 0, "");
6136 #pragma clang diagnostic pop
6137 #endif /* CONFIG_IOTRACE */
6138
6139 static int
6140 sysctl_page_protection_type SYSCTL_HANDLER_ARGS
6141 {
6142 #pragma unused(oidp, arg1, arg2)
6143 int value = ml_page_protection_type();
6144 return SYSCTL_OUT(req, &value, sizeof(value));
6145 }
6146
6147 SYSCTL_PROC(_kern, OID_AUTO, page_protection_type,
6148 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
6149 0, 0, sysctl_page_protection_type, "I", "Type of page protection that the system supports");
6150
6151 TUNABLE_DT(int, gpu_pmem_selector, "defaults", "kern.gpu_pmem_selector", "gpu-pmem-selector", 0, TUNABLE_DT_NONE);
6152
6153 #if CONFIG_EXCLAVES
6154
6155 static int
6156 sysctl_task_conclave SYSCTL_HANDLER_ARGS
6157 {
6158 extern const char *exclaves_resource_name(void *);
6159
6160 #pragma unused(arg2)
6161 void *conclave = task_get_conclave(current_task());
6162 if (conclave != NULL) {
6163 const char *name = exclaves_resource_name(conclave);
6164 assert3u(strlen(name), >, 0);
6165
6166 /*
6167 * This is a RO operation already and the string is never
6168 * written to.
6169 */
6170 #pragma clang diagnostic push
6171 #pragma clang diagnostic ignored "-Wcast-qual"
6172 return sysctl_handle_string(oidp, (char *)name, 0, req);
6173 #pragma clang diagnostic pop
6174 }
6175 return sysctl_handle_string(oidp, arg1, MAXCONCLAVENAME, req);
6176 }
6177
6178 SYSCTL_PROC(_kern, OID_AUTO, task_conclave,
6179 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
6180 "", 0, sysctl_task_conclave, "A", "Conclave string for the task");
6181
6182
6183 void task_set_conclave_untaintable(task_t task);
6184
6185 static int
6186 sysctl_task_conclave_untaintable SYSCTL_HANDLER_ARGS
6187 {
6188 #pragma unused(arg1, arg2)
6189 int error, val = 0;
6190 error = sysctl_handle_int(oidp, &val, 0, req);
6191 if (error || val == 0) {
6192 return error;
6193 }
6194
6195 task_set_conclave_untaintable(current_task());
6196 return 0;
6197 }
6198
6199 SYSCTL_PROC(_kern, OID_AUTO, task_conclave_untaintable,
6200 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
6201 "", 0, sysctl_task_conclave_untaintable, "A", "Task could not be tainted by talking to conclaves");
6202
6203 extern exclaves_requirement_t exclaves_relaxed_requirements;
6204 SYSCTL_QUAD(_kern, OID_AUTO, exclaves_relaxed_requirements,
6205 CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
6206 &exclaves_relaxed_requirements, "Exclaves requirements which have been relaxed");
6207
6208 #endif /* CONFIG_EXCLAVES */
6209
6210 #if (DEVELOPMENT || DEBUG)
6211 SYSCTL_INT(_kern, OID_AUTO, gpu_pmem_selector,
6212 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
6213 &gpu_pmem_selector, 0, "GPU wire down limit selector");
6214 #else /* !(DEVELOPMENT || DEBUG) */
6215 SYSCTL_INT(_kern, OID_AUTO, gpu_pmem_selector,
6216 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN | CTLFLAG_MASKED,
6217 &gpu_pmem_selector, 0, "GPU wire down limit selector");
6218 #endif /* (DEVELOPMENT || DEBUG) */
6219
6220 static int
6221 sysctl_exclaves_status SYSCTL_HANDLER_ARGS
6222 {
6223 int value = exclaves_get_status();
6224 return sysctl_io_number(req, value, sizeof(value), NULL, NULL);
6225 }
6226
6227 SYSCTL_PROC(_kern, OID_AUTO, exclaves_status,
6228 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
6229 0, 0, sysctl_exclaves_status, "I", "Running status of Exclaves");
6230
6231
6232 static int
6233 sysctl_exclaves_boot_stage SYSCTL_HANDLER_ARGS
6234 {
6235 int value = exclaves_get_boot_stage();
6236 return sysctl_io_number(req, value, sizeof(value), NULL, NULL);
6237 }
6238
6239 SYSCTL_PROC(_kern, OID_AUTO, exclaves_boot_stage,
6240 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
6241 0, 0, sysctl_exclaves_boot_stage, "I", "Boot stage of Exclaves");
6242
6243 #if CONFIG_EXCLAVES && (DEVELOPMENT || DEBUG)
6244 extern unsigned int exclaves_debug;
6245 SYSCTL_UINT(_kern, OID_AUTO, exclaves_debug, CTLFLAG_RW | CTLFLAG_LOCKED,
6246 &exclaves_debug, 0, "Exclaves debug flags");
6247
6248 static int
6249 sysctl_exclaves_inspection_status SYSCTL_HANDLER_ARGS
6250 {
6251 #pragma unused(oidp, arg1, arg2)
6252 int value = (int)exclaves_inspection_is_initialized();
6253 return sysctl_io_number(req, value, sizeof(value), NULL, NULL);
6254 }
6255 SYSCTL_PROC(_kern, OID_AUTO, exclaves_inspection_status,
6256 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
6257 0, 0, sysctl_exclaves_inspection_status, "I", "Exclaves debug inspection status");
6258 #endif /* CONFIG_EXCLAVES && (DEVELOPMENT || DEBUG) */
6259
6260 #if (DEBUG || DEVELOPMENT)
6261 extern uint32_t disable_vm_sanitize_telemetry;
6262 SYSCTL_UINT(_debug, OID_AUTO, disable_vm_sanitize_telemetry, CTLFLAG_RW | CTLFLAG_LOCKED /*| CTLFLAG_MASKED*/, &disable_vm_sanitize_telemetry, 0, "disable VM API sanitization telemetry");
6263 #endif
6264