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 * Additionally, if the task is not IPC active, we
1455 * must fail early. Other tasks can't yet look up
1456 * this task's task port to make Mach API calls, so
1457 * we shouldn't make such calls on their behalf.
1458 */
1459 task = proc_task(p);
1460 if (task == NULL || !task_is_ipc_active(task)) {
1461 error = EINVAL;
1462 goto finish;
1463 }
1464
1465 /*
1466 * Once we have a task reference we can convert that into a
1467 * map reference, which we will use in the calls below. The
1468 * task/process may change its map after we take this reference
1469 * (see execve), but the worst that will happen then is a return
1470 * of stale info (which is always a possibility).
1471 */
1472 task_reference(task);
1473 proc_rele(p);
1474 p = NULL;
1475 proc_map = get_task_map_reference(task);
1476 task_deallocate(task);
1477
1478 if (proc_map == NULL) {
1479 error = EINVAL;
1480 goto finish;
1481 }
1482
1483 effective_page_mask = vm_map_page_mask(proc_map);
1484
1485 arg_size = vm_map_round_page(argslen, effective_page_mask);
1486
1487 arg_addr = user_stack - arg_size;
1488
1489 ret = kmem_alloc(kernel_map, ©_start, arg_size,
1490 KMA_DATA | KMA_ZERO, VM_KERN_MEMORY_BSD);
1491 if (ret != KERN_SUCCESS) {
1492 error = ENOMEM;
1493 goto finish;
1494 }
1495
1496 copy_end = copy_start + arg_size;
1497
1498 if (vm_map_copyin(proc_map, (vm_map_address_t)arg_addr,
1499 (vm_map_size_t)arg_size, FALSE, &tmp) != KERN_SUCCESS) {
1500 error = EIO;
1501 goto finish;
1502 }
1503
1504 /*
1505 * Now that we've done the copyin from the process'
1506 * map, we can release the reference to it.
1507 */
1508 vm_map_deallocate(proc_map);
1509 proc_map = NULL;
1510
1511 if (vm_map_copy_overwrite(kernel_map,
1512 (vm_map_address_t)copy_start,
1513 tmp, (vm_map_size_t) arg_size, FALSE) != KERN_SUCCESS) {
1514 error = EIO;
1515 goto finish;
1516 }
1517 /* tmp was consumed */
1518 tmp = NULL;
1519
1520 if (omit_env_vars) {
1521 argvsize = 0;
1522
1523 /* Iterate over everything in argv, plus one for the bare executable path */
1524 for (current_arg_index = 0; current_arg_index < argc + 1 && argvsize < argslen; ++current_arg_index) {
1525 current_arg = (const char *)(copy_end - argslen) + argvsize;
1526 remaining = argslen - argvsize;
1527 current_arg_len = strnlen(current_arg, remaining);
1528 if (current_arg_len < remaining) {
1529 /* We have space for the null terminator */
1530 current_arg_len += 1;
1531
1532 if (current_arg_index == 0) {
1533 /* The bare executable path may have multiple null bytes after it for alignment */
1534 while (current_arg_len < remaining && current_arg[current_arg_len] == 0) {
1535 current_arg_len += 1;
1536 }
1537 }
1538 }
1539 argvsize += current_arg_len;
1540 }
1541 assert(argvsize <= argslen);
1542
1543 /* Adjust argslen and copy_end to make the copyout range extend to the end of argv */
1544 copy_end = copy_end - argslen + argvsize;
1545 argslen = argvsize;
1546 }
1547
1548 if (where == USER_ADDR_NULL) {
1549 /* Skip copyout */
1550 goto calculate_size;
1551 }
1552
1553 if (buflen >= argslen) {
1554 data = (caddr_t) (copy_end - argslen);
1555 size = argslen;
1556 } else {
1557 /*
1558 * Before rdar://25397314, this function contained incorrect logic when buflen is less
1559 * than argslen. The problem was that it copied in `buflen` bytes from the end of the target
1560 * process user stack into the beginning of a buffer of size round_page(buflen), and then
1561 * copied out `buflen` bytes from the end of this buffer. The effect of this was that
1562 * the caller of this sysctl would get zeros at the end of their buffer.
1563 *
1564 * To preserve this behavior, bzero everything from copy_end-round_page(buflen)+buflen to the
1565 * end of the buffer. This emulates copying in only `buflen` bytes.
1566 *
1567 *
1568 * In the old code:
1569 *
1570 * copy_start .... size: round_page(buflen) .... copy_end
1571 * [---copied in data (size: buflen)---|--- zeros ----------]
1572 * ^
1573 * data = copy_end - buflen
1574 *
1575 *
1576 * In the new code:
1577 * copy_start .... size: round_page(p->argslen) .... full copy_end
1578 * ^ ....................... p->argslen ...............................^
1579 * ^ ^ truncated copy_end ^
1580 * ^ ^ ^ ^
1581 * ^ ................ argslen ........................ ^
1582 * ^ ^ ^ ^
1583 * [-------copied in data (size: round_page(p->argslen))-------:----env vars---]
1584 * ^ ^
1585 * ^ data = copy_end - buflen
1586 * smallbuffer_start = max(copy_end - round_page(buflen), copy_start)
1587 *
1588 *
1589 * Full copy_end: copy_end calculated from copy_start + round_page(p->argslen)
1590 * Truncated copy_end: copy_end after truncation to remove environment variables.
1591 *
1592 * If environment variables were omitted, then we use the truncated copy_end, otherwise
1593 * we use full copy_end.
1594 *
1595 * smallbuffer_start: represents where copy_start would be in the old code.
1596 * data: The beginning of the region we copyout
1597 */
1598 smallbuffer_start = copy_end - vm_map_round_page(buflen, effective_page_mask);
1599 if (smallbuffer_start < copy_start) {
1600 smallbuffer_start = copy_start;
1601 }
1602 bzero((void *)(smallbuffer_start + buflen), copy_end - (smallbuffer_start + buflen));
1603 data = (caddr_t) (copy_end - buflen);
1604 size = buflen;
1605 }
1606
1607 if (argc_yes) {
1608 /* Put processes argc as the first word in the copyout buffer */
1609 suword(where, argc);
1610 error = copyout(data, (where + sizeof(int)), size);
1611 size += sizeof(int);
1612 } else {
1613 error = copyout(data, where, size);
1614
1615 /*
1616 * Make the old PROCARGS work to return the executable's path
1617 * But, only if there is enough space in the provided buffer
1618 *
1619 * on entry: data [possibily] points to the beginning of the path
1620 *
1621 * Note: we keep all pointers&sizes aligned to word boundries
1622 */
1623 if ((!error) && (buflen > 0 && (u_int)buflen > size)) {
1624 int binPath_sz, alignedBinPath_sz = 0;
1625 int extraSpaceNeeded, addThis;
1626 user_addr_t placeHere;
1627 char * str = (char *) data;
1628 size_t max_len = size;
1629
1630 /* Some apps are really bad about messing up their stacks
1631 * So, we have to be extra careful about getting the length
1632 * of the executing binary. If we encounter an error, we bail.
1633 */
1634
1635 /* Limit ourselves to PATH_MAX paths */
1636 if (max_len > PATH_MAX) {
1637 max_len = PATH_MAX;
1638 }
1639
1640 binPath_sz = 0;
1641
1642 while ((binPath_sz < max_len - 1) && (*str++ != 0)) {
1643 binPath_sz++;
1644 }
1645
1646 /* If we have a NUL terminator, copy it, too */
1647 if (binPath_sz < max_len - 1) {
1648 binPath_sz += 1;
1649 }
1650
1651 /* Pre-Flight the space requiremnts */
1652
1653 /* Account for the padding that fills out binPath to the next word */
1654 alignedBinPath_sz += (binPath_sz & (sizeof(int) - 1)) ? (sizeof(int) - (binPath_sz & (sizeof(int) - 1))) : 0;
1655
1656 placeHere = where + size;
1657
1658 /* Account for the bytes needed to keep placeHere word aligned */
1659 addThis = (placeHere & (sizeof(int) - 1)) ? (sizeof(int) - (placeHere & (sizeof(int) - 1))) : 0;
1660
1661 /* Add up all the space that is needed */
1662 extraSpaceNeeded = alignedBinPath_sz + addThis + binPath_sz + (4 * sizeof(int));
1663
1664 /* is there is room to tack on argv[0]? */
1665 if ((buflen & ~(sizeof(int) - 1)) >= (size + extraSpaceNeeded)) {
1666 placeHere += addThis;
1667 suword(placeHere, 0);
1668 placeHere += sizeof(int);
1669 suword(placeHere, 0xBFFF0000);
1670 placeHere += sizeof(int);
1671 suword(placeHere, 0);
1672 placeHere += sizeof(int);
1673 error = copyout(data, placeHere, binPath_sz);
1674 if (!error) {
1675 placeHere += binPath_sz;
1676 suword(placeHere, 0);
1677 size += extraSpaceNeeded;
1678 }
1679 }
1680 }
1681 }
1682
1683 calculate_size:
1684 /* Size has already been calculated for the where != NULL case */
1685 if (where == USER_ADDR_NULL) {
1686 size = argslen;
1687 if (argc_yes) {
1688 size += sizeof(int);
1689 } else {
1690 /*
1691 * old PROCARGS will return the executable's path and plus some
1692 * extra space for work alignment and data tags
1693 */
1694 size += PATH_MAX + (6 * sizeof(int));
1695 }
1696 size += (size & (sizeof(int) - 1)) ? (sizeof(int) - (size & (sizeof(int) - 1))) : 0;
1697 }
1698
1699 *sizep = size;
1700
1701 finish:
1702 if (p != NULL) {
1703 proc_rele(p);
1704 }
1705 if (tmp != NULL) {
1706 vm_map_copy_discard(tmp);
1707 }
1708 if (proc_map != NULL) {
1709 vm_map_deallocate(proc_map);
1710 }
1711 if (copy_start != (vm_offset_t) 0) {
1712 kmem_free(kernel_map, copy_start, arg_size);
1713 }
1714 return error;
1715 }
1716
1717
1718 /*
1719 * Max number of concurrent aio requests
1720 */
1721 STATIC int
1722 sysctl_aiomax
1723 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1724 {
1725 int new_value, changed;
1726 int error = sysctl_io_number(req, aio_max_requests, sizeof(int), &new_value, &changed);
1727 if (changed) {
1728 /* make sure the system-wide limit is greater than the per process limit */
1729 if (new_value >= aio_max_requests_per_process && new_value <= AIO_MAX_REQUESTS) {
1730 aio_max_requests = new_value;
1731 } else {
1732 error = EINVAL;
1733 }
1734 }
1735 return error;
1736 }
1737
1738
1739 /*
1740 * Max number of concurrent aio requests per process
1741 */
1742 STATIC int
1743 sysctl_aioprocmax
1744 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1745 {
1746 int new_value, changed;
1747 int error = sysctl_io_number(req, aio_max_requests_per_process, sizeof(int), &new_value, &changed);
1748 if (changed) {
1749 /* make sure per process limit is less than the system-wide limit */
1750 if (new_value <= aio_max_requests && new_value >= AIO_LISTIO_MAX) {
1751 aio_max_requests_per_process = new_value;
1752 } else {
1753 error = EINVAL;
1754 }
1755 }
1756 return error;
1757 }
1758
1759
1760 /*
1761 * Max number of async IO worker threads
1762 */
1763 STATIC int
1764 sysctl_aiothreads
1765 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1766 {
1767 int new_value, changed;
1768 int error = sysctl_io_number(req, aio_worker_threads, sizeof(int), &new_value, &changed);
1769 if (changed) {
1770 /* we only allow an increase in the number of worker threads */
1771 if (new_value > aio_worker_threads) {
1772 _aio_create_worker_threads((new_value - aio_worker_threads));
1773 aio_worker_threads = new_value;
1774 } else {
1775 error = EINVAL;
1776 }
1777 }
1778 return error;
1779 }
1780
1781
1782 /*
1783 * System-wide limit on the max number of processes
1784 */
1785 STATIC int
1786 sysctl_maxproc
1787 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1788 {
1789 int new_value, changed;
1790 int error = sysctl_io_number(req, maxproc, sizeof(int), &new_value, &changed);
1791 if (changed) {
1792 AUDIT_ARG(value32, new_value);
1793 /* make sure the system-wide limit is less than the configured hard
1794 * limit set at kernel compilation */
1795 if (new_value <= hard_maxproc && new_value > 0) {
1796 maxproc = new_value;
1797 } else {
1798 error = EINVAL;
1799 }
1800 }
1801 return error;
1802 }
1803
1804 #if CONFIG_SCHED_SMT
1805 STATIC int
1806 sysctl_sched_enable_smt
1807 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1808 {
1809 int new_value, changed;
1810 int error = sysctl_io_number(req, sched_enable_smt, sizeof(int), &new_value, &changed);
1811 if (error) {
1812 return error;
1813 }
1814 kern_return_t kret = KERN_SUCCESS;
1815 if (changed) {
1816 AUDIT_ARG(value32, new_value);
1817 if (new_value == 0) {
1818 sched_enable_smt = 0;
1819 kret = enable_smt_processors(false);
1820 } else {
1821 sched_enable_smt = 1;
1822 kret = enable_smt_processors(true);
1823 }
1824 }
1825 switch (kret) {
1826 case KERN_SUCCESS:
1827 error = 0;
1828 break;
1829 case KERN_INVALID_ARGUMENT:
1830 error = EINVAL;
1831 break;
1832 case KERN_FAILURE:
1833 error = EBUSY;
1834 break;
1835 default:
1836 error = ENOENT;
1837 break;
1838 }
1839 return error;
1840 }
1841 #else /* CONFIG_SCHED_SMT */
1842 STATIC int
1843 sysctl_sched_enable_smt
1844 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, __unused struct sysctl_req *req)
1845 {
1846 return 0;
1847 }
1848 #endif /* CONFIG_SCHED_SMT */
1849
1850 SYSCTL_STRING(_kern, KERN_OSTYPE, ostype,
1851 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
1852 ostype, 0, "");
1853 SYSCTL_STRING(_kern, KERN_OSRELEASE, osrelease,
1854 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
1855 osrelease, 0, "");
1856 SYSCTL_INT(_kern, KERN_OSREV, osrevision,
1857 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
1858 (int *)NULL, BSD, "");
1859 SYSCTL_STRING(_kern, KERN_VERSION, version,
1860 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
1861 version, 0, "");
1862 SYSCTL_STRING(_kern, OID_AUTO, uuid,
1863 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
1864 &kernel_uuid_string[0], 0, "");
1865
1866 SYSCTL_STRING(_kern, OID_AUTO, osbuildconfig,
1867 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED | CTLFLAG_MASKED,
1868 &osbuild_config[0], 0, "");
1869
1870 #if DEBUG
1871 #ifndef DKPR
1872 #define DKPR 1
1873 #endif
1874 #endif
1875
1876 #if DKPR
1877 int debug_kprint_syscall = 0;
1878 char debug_kprint_syscall_process[MAXCOMLEN + 1];
1879
1880 /* Thread safe: bits and string value are not used to reclaim state */
1881 SYSCTL_INT(_debug, OID_AUTO, kprint_syscall,
1882 CTLFLAG_RW | CTLFLAG_LOCKED, &debug_kprint_syscall, 0, "kprintf syscall tracing");
1883 SYSCTL_STRING(_debug, OID_AUTO, kprint_syscall_process,
1884 CTLFLAG_RW | CTLFLAG_LOCKED, debug_kprint_syscall_process, sizeof(debug_kprint_syscall_process),
1885 "name of process for kprintf syscall tracing");
1886
1887 int
1888 debug_kprint_current_process(const char **namep)
1889 {
1890 struct proc *p = current_proc();
1891
1892 if (p == NULL) {
1893 return 0;
1894 }
1895
1896 if (debug_kprint_syscall_process[0]) {
1897 /* user asked to scope tracing to a particular process name */
1898 if (0 == strncmp(debug_kprint_syscall_process,
1899 p->p_comm, sizeof(debug_kprint_syscall_process))) {
1900 /* no value in telling the user that we traced what they asked */
1901 if (namep) {
1902 *namep = NULL;
1903 }
1904
1905 return 1;
1906 } else {
1907 return 0;
1908 }
1909 }
1910
1911 /* trace all processes. Tell user what we traced */
1912 if (namep) {
1913 *namep = p->p_comm;
1914 }
1915
1916 return 1;
1917 }
1918 #endif
1919
1920 /* PR-5293665: need to use a callback function for kern.osversion to set
1921 * osversion in IORegistry */
1922
1923 STATIC int
1924 sysctl_osversion(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
1925 {
1926 int rval = 0;
1927
1928 rval = sysctl_handle_string(oidp, arg1, arg2, req);
1929
1930 if (req->newptr) {
1931 IORegistrySetOSBuildVersion((char *)arg1);
1932 }
1933
1934 return rval;
1935 }
1936
1937 SYSCTL_PROC(_kern, KERN_OSVERSION, osversion,
1938 CTLFLAG_RW | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
1939 osversion, 256 /* OSVERSIZE*/,
1940 sysctl_osversion, "A", "");
1941
1942 static bool
1943 _already_set_or_not_launchd(struct sysctl_req *req, char *val)
1944 {
1945 if (req->newptr != 0) {
1946 /*
1947 * Can only ever be set by launchd, and only once at boot.
1948 */
1949 if (proc_getpid(req->p) != 1 || val[0] != '\0') {
1950 return true;
1951 }
1952 }
1953 return false;
1954 }
1955
1956 #define kRootsInstalledReadWriteEntitlement "com.apple.private.roots-installed-read-write"
1957 #define kRootsInstalledReadOnlyEntitlement "com.apple.private.roots-installed-read-only"
1958 uint64_t roots_installed = 0;
1959
1960 static int
1961 sysctl_roots_installed
1962 (__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
1963 {
1964 int error = 0;
1965
1966 if (req->newptr != 0) {
1967 /* a ReadWrite entitlement is required for updating this syscl
1968 * meanwhile, only allow write once
1969 */
1970 if (!IOCurrentTaskHasEntitlement(kRootsInstalledReadWriteEntitlement) || (roots_installed != 0)) {
1971 return EPERM;
1972 }
1973 } else {
1974 /* for reader of this sysctl, need either ReadWrite or ReadOnly entitlement */
1975 if (!IOCurrentTaskHasEntitlement(kRootsInstalledReadWriteEntitlement) &&
1976 !IOCurrentTaskHasEntitlement(kRootsInstalledReadOnlyEntitlement)) {
1977 return EPERM;
1978 }
1979 }
1980
1981 error = sysctl_handle_quad(oidp, arg1, arg2, req);
1982
1983 return error;
1984 }
1985
1986 SYSCTL_PROC(_kern, OID_AUTO, roots_installed,
1987 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
1988 &roots_installed, sizeof(roots_installed),
1989 sysctl_roots_installed, "Q", "");
1990
1991 #if XNU_TARGET_OS_OSX
1992 static int
1993 sysctl_system_version_compat
1994 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1995 {
1996 int oldval = (task_has_system_version_compat_enabled(current_task()));
1997 int new_value = 0, changed = 0;
1998
1999 int error = sysctl_io_number(req, oldval, sizeof(int), &new_value, &changed);
2000 if (changed) {
2001 task_set_system_version_compat_enabled(current_task(), (new_value));
2002 }
2003 return error;
2004 }
2005
2006 SYSCTL_PROC(_kern, OID_AUTO, system_version_compat,
2007 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
2008 0, 0, sysctl_system_version_compat, "A", "");
2009
2010 char osproductversioncompat[48] = { '\0' };
2011
2012 static int
2013 sysctl_osproductversioncompat(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2014 {
2015 if (_already_set_or_not_launchd(req, osproductversioncompat)) {
2016 return EPERM;
2017 }
2018 return sysctl_handle_string(oidp, arg1, arg2, req);
2019 }
2020
2021
2022 SYSCTL_PROC(_kern, OID_AUTO, osproductversioncompat,
2023 CTLFLAG_RW | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2024 osproductversioncompat, sizeof(osproductversioncompat),
2025 sysctl_osproductversioncompat, "A", "The ProductVersion from SystemVersionCompat.plist");
2026 #endif
2027
2028 char osproductversion[48] = { '\0' };
2029
2030 static char iossupportversion_string[48] = { '\0' };
2031
2032 static int
2033 sysctl_osproductversion(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2034 {
2035 if (_already_set_or_not_launchd(req, osproductversion)) {
2036 return EPERM;
2037 }
2038
2039 #if XNU_TARGET_OS_OSX
2040 if (task_has_system_version_compat_enabled(current_task()) && (osproductversioncompat[0] != '\0')) {
2041 return sysctl_handle_string(oidp, osproductversioncompat, arg2, req);
2042 } else {
2043 return sysctl_handle_string(oidp, arg1, arg2, req);
2044 }
2045 #elif defined(XNU_TARGET_OS_XR)
2046 if (proc_platform(req->p) == PLATFORM_IOS && (iossupportversion_string[0] != '\0')) {
2047 return sysctl_handle_string(oidp, iossupportversion_string, arg2, req);
2048 } else {
2049 return sysctl_handle_string(oidp, arg1, arg2, req);
2050 }
2051 #else
2052 return sysctl_handle_string(oidp, arg1, arg2, req);
2053 #endif
2054 }
2055
2056 #if XNU_TARGET_OS_OSX
2057 static_assert(sizeof(osproductversioncompat) == sizeof(osproductversion),
2058 "osproductversion size matches osproductversioncompat size");
2059 #endif
2060
2061 SYSCTL_PROC(_kern, OID_AUTO, osproductversion,
2062 CTLFLAG_RW | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2063 osproductversion, sizeof(osproductversion),
2064 sysctl_osproductversion, "A", "The ProductVersion from SystemVersion.plist");
2065
2066 char osreleasetype[OSRELEASETYPE_SIZE] = { '\0' };
2067
2068 STATIC int
2069 sysctl_osreleasetype(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2070 {
2071 if (_already_set_or_not_launchd(req, osreleasetype)) {
2072 return EPERM;
2073 }
2074 return sysctl_handle_string(oidp, arg1, arg2, req);
2075 }
2076
2077 void reset_osreleasetype(void);
2078
2079 void
2080 reset_osreleasetype(void)
2081 {
2082 memset(osreleasetype, 0, sizeof(osreleasetype));
2083 }
2084
2085 SYSCTL_PROC(_kern, OID_AUTO, osreleasetype,
2086 CTLFLAG_RW | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2087 osreleasetype, sizeof(osreleasetype),
2088 sysctl_osreleasetype, "A", "The ReleaseType from SystemVersion.plist");
2089
2090 STATIC int
2091 sysctl_iossupportversion(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2092 {
2093 if (_already_set_or_not_launchd(req, iossupportversion_string)) {
2094 return EPERM;
2095 }
2096
2097 #if defined(XNU_TARGET_OS_XR)
2098 if (proc_platform(req->p) == PLATFORM_IOS) {
2099 /* return empty string for iOS processes to match how this would behave on iOS */
2100 return sysctl_handle_string(oidp, "", arg2, req);
2101 } else {
2102 /* native processes see the actual value */
2103 return sysctl_handle_string(oidp, arg1, arg2, req);
2104 }
2105 #else
2106 return sysctl_handle_string(oidp, arg1, arg2, req);
2107 #endif
2108 }
2109
2110 SYSCTL_PROC(_kern, OID_AUTO, iossupportversion,
2111 CTLFLAG_RW | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2112 iossupportversion_string, sizeof(iossupportversion_string),
2113 sysctl_iossupportversion, "A", "The iOSSupportVersion from SystemVersion.plist");
2114
2115 static uint64_t osvariant_status = 0;
2116
2117 STATIC int
2118 sysctl_osvariant_status(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2119 {
2120 if (req->newptr != 0) {
2121 /*
2122 * Can only ever be set by launchd, and only once.
2123 * Reset by usrctl() -> reset_osvariant_status() during
2124 * userspace reboot, since userspace could reboot into
2125 * a different variant.
2126 */
2127 if (proc_getpid(req->p) != 1 || osvariant_status != 0) {
2128 return EPERM;
2129 }
2130 }
2131
2132 int err = sysctl_handle_quad(oidp, arg1, arg2, req);
2133
2134 reset_debug_syscall_rejection_mode();
2135
2136 return err;
2137 }
2138
2139 SYSCTL_PROC(_kern, OID_AUTO, osvariant_status,
2140 CTLFLAG_RW | CTLTYPE_QUAD | CTLFLAG_LOCKED | CTLFLAG_MASKED,
2141 &osvariant_status, sizeof(osvariant_status),
2142 sysctl_osvariant_status, "Q", "Opaque flags used to cache OS variant information");
2143
2144 static bool
2145 _os_variant_check_disabled(enum os_variant_property property)
2146 {
2147 return (osvariant_status >> (32 + property)) & 0x1;
2148 }
2149
2150 static bool
2151 _os_variant_has(enum os_variant_status_flags_positions p)
2152 {
2153 return ((osvariant_status >> (p * OS_VARIANT_STATUS_BIT_WIDTH)) & OS_VARIANT_STATUS_MASK) == OS_VARIANT_S_YES;
2154 }
2155
2156 bool
2157 os_variant_has_internal_diagnostics(__unused const char *subsystem)
2158 {
2159 if (_os_variant_check_disabled(OS_VARIANT_PROPERTY_DIAGNOSTICS)) {
2160 return false;
2161 }
2162 #if XNU_TARGET_OS_OSX
2163 return _os_variant_has(OS_VARIANT_SFP_INTERNAL_CONTENT) || _os_variant_has(OS_VARIANT_SFP_INTERNAL_DIAGS_PROFILE);
2164 #else
2165 return _os_variant_has(OS_VARIANT_SFP_INTERNAL_RELEASE_TYPE);
2166 #endif /* XNU_TARGET_OS_OSX */
2167 }
2168
2169 void reset_osvariant_status(void);
2170
2171 void
2172 reset_osvariant_status(void)
2173 {
2174 osvariant_status = 0;
2175 reset_debug_syscall_rejection_mode();
2176 }
2177
2178 extern void commpage_update_dyld_flags(uint64_t);
2179 TUNABLE_WRITEABLE(uint64_t, dyld_flags, "dyld_flags", 0);
2180
2181 STATIC int
2182 sysctl_dyld_flags(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2183 {
2184 /*
2185 * Can only ever be set by launchd, possibly several times
2186 * as dyld may change its mind after a userspace reboot.
2187 */
2188 if (req->newptr != 0 && proc_getpid(req->p) != 1) {
2189 return EPERM;
2190 }
2191
2192 int res = sysctl_handle_quad(oidp, arg1, arg2, req);
2193 if (req->newptr && res == 0) {
2194 commpage_update_dyld_flags(dyld_flags);
2195 }
2196 return res;
2197 }
2198
2199 SYSCTL_PROC(_kern, OID_AUTO, dyld_flags,
2200 CTLFLAG_RW | CTLTYPE_QUAD | CTLFLAG_LOCKED | CTLFLAG_MASKED,
2201 &dyld_flags, sizeof(dyld_flags),
2202 sysctl_dyld_flags, "Q", "Opaque flags used to cache dyld system-wide configuration");
2203
2204 #if defined(XNU_TARGET_OS_BRIDGE)
2205 char macosproductversion[MACOS_VERS_LEN] = { '\0' };
2206
2207 SYSCTL_STRING(_kern, OID_AUTO, macosproductversion,
2208 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2209 &macosproductversion[0], MACOS_VERS_LEN, "The currently running macOS ProductVersion (from SystemVersion.plist on macOS)");
2210
2211 char macosversion[MACOS_VERS_LEN] = { '\0' };
2212
2213 SYSCTL_STRING(_kern, OID_AUTO, macosversion,
2214 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2215 &macosversion[0], MACOS_VERS_LEN, "The currently running macOS build version");
2216 #endif
2217
2218 STATIC int
2219 sysctl_sysctl_bootargs
2220 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2221 {
2222 int error;
2223 char buf[BOOT_LINE_LENGTH];
2224
2225 strlcpy(buf, PE_boot_args(), BOOT_LINE_LENGTH);
2226 error = sysctl_io_string(req, buf, BOOT_LINE_LENGTH, 0, NULL);
2227 return error;
2228 }
2229
2230 SYSCTL_PROC(_kern, OID_AUTO, bootargs,
2231 CTLFLAG_LOCKED | CTLFLAG_RD | CTLFLAG_KERN | CTLTYPE_STRING,
2232 NULL, 0,
2233 sysctl_sysctl_bootargs, "A", "bootargs");
2234
2235 STATIC int
2236 sysctl_kernelcacheuuid(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2237 {
2238 int rval = ENOENT;
2239 if (kernelcache_uuid_valid) {
2240 rval = sysctl_handle_string(oidp, arg1, arg2, req);
2241 }
2242 return rval;
2243 }
2244
2245 SYSCTL_PROC(_kern, OID_AUTO, kernelcacheuuid,
2246 CTLFLAG_RD | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2247 kernelcache_uuid_string, sizeof(kernelcache_uuid_string),
2248 sysctl_kernelcacheuuid, "A", "");
2249
2250 STATIC int
2251 sysctl_systemfilesetuuid(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2252 {
2253 int rval = ENOENT;
2254 if (pageablekc_uuid_valid) {
2255 rval = sysctl_handle_string(oidp, arg1, arg2, req);
2256 }
2257 return rval;
2258 }
2259
2260 SYSCTL_PROC(_kern, OID_AUTO, systemfilesetuuid,
2261 CTLFLAG_RD | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2262 pageablekc_uuid_string, sizeof(pageablekc_uuid_string),
2263 sysctl_systemfilesetuuid, "A", "");
2264
2265 STATIC int
2266 sysctl_auxiliaryfilesetuuid(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2267 {
2268 int rval = ENOENT;
2269 if (auxkc_uuid_valid) {
2270 rval = sysctl_handle_string(oidp, arg1, arg2, req);
2271 }
2272 return rval;
2273 }
2274
2275 SYSCTL_PROC(_kern, OID_AUTO, auxiliaryfilesetuuid,
2276 CTLFLAG_RD | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2277 auxkc_uuid_string, sizeof(auxkc_uuid_string),
2278 sysctl_auxiliaryfilesetuuid, "A", "");
2279
2280 STATIC int
2281 sysctl_filesetuuid(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2282 {
2283 int rval = ENOENT;
2284 kc_format_t kcformat;
2285 kernel_mach_header_t *mh;
2286 void *uuid = NULL;
2287 unsigned long uuidlen = 0;
2288 uuid_string_t uuid_str;
2289
2290 if (!PE_get_primary_kc_format(&kcformat) || kcformat != KCFormatFileset) {
2291 return rval;
2292 }
2293
2294 mh = (kernel_mach_header_t *)PE_get_kc_header(KCKindPrimary);
2295 uuid = getuuidfromheader(mh, &uuidlen);
2296
2297 if ((uuid != NULL) && (uuidlen == sizeof(uuid_t))) {
2298 uuid_unparse_upper(*(uuid_t *)uuid, uuid_str);
2299 rval = sysctl_io_string(req, (char *)uuid_str, sizeof(uuid_str), 0, NULL);
2300 }
2301
2302 return rval;
2303 }
2304
2305 SYSCTL_PROC(_kern, OID_AUTO, filesetuuid,
2306 CTLFLAG_RD | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2307 NULL, 0,
2308 sysctl_filesetuuid, "A", "");
2309
2310
2311 SYSCTL_INT(_kern, KERN_MAXFILES, maxfiles,
2312 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2313 &maxfiles, 0, "");
2314 SYSCTL_INT(_kern, KERN_ARGMAX, argmax,
2315 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2316 (int *)NULL, ARG_MAX, "");
2317 SYSCTL_INT(_kern, KERN_POSIX1, posix1version,
2318 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2319 (int *)NULL, _POSIX_VERSION, "");
2320 SYSCTL_INT(_kern, KERN_NGROUPS, ngroups,
2321 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2322 (int *)NULL, NGROUPS_MAX, "");
2323 SYSCTL_INT(_kern, KERN_JOB_CONTROL, job_control,
2324 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2325 (int *)NULL, 1, "");
2326 #if 1 /* _POSIX_SAVED_IDS from <unistd.h> */
2327 SYSCTL_INT(_kern, KERN_SAVED_IDS, saved_ids,
2328 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2329 (int *)NULL, 1, "");
2330 #else
2331 SYSCTL_INT(_kern, KERN_SAVED_IDS, saved_ids,
2332 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2333 NULL, 0, "");
2334 #endif
2335 SYSCTL_INT(_kern, OID_AUTO, num_files,
2336 CTLFLAG_RD | CTLFLAG_LOCKED,
2337 &nfiles, 0, "");
2338 SYSCTL_COMPAT_INT(_kern, OID_AUTO, num_vnodes,
2339 CTLFLAG_RD | CTLFLAG_LOCKED,
2340 &numvnodes, 0, "");
2341 SYSCTL_INT(_kern, OID_AUTO, num_tasks,
2342 CTLFLAG_RD | CTLFLAG_LOCKED,
2343 &task_max, 0, "");
2344 SYSCTL_INT(_kern, OID_AUTO, num_threads,
2345 CTLFLAG_RD | CTLFLAG_LOCKED,
2346 &thread_max, 0, "");
2347 SYSCTL_INT(_kern, OID_AUTO, num_taskthreads,
2348 CTLFLAG_RD | CTLFLAG_LOCKED,
2349 &task_threadmax, 0, "");
2350 SYSCTL_LONG(_kern, OID_AUTO, num_recycledvnodes,
2351 CTLFLAG_RD | CTLFLAG_LOCKED,
2352 &num_recycledvnodes, "");
2353 SYSCTL_COMPAT_INT(_kern, OID_AUTO, free_vnodes,
2354 CTLFLAG_RD | CTLFLAG_LOCKED,
2355 &freevnodes, 0, "");
2356
2357 STATIC int
2358 sysctl_maxvnodes(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2359 {
2360 int oldval = desiredvnodes;
2361 int error = sysctl_io_number(req, desiredvnodes, sizeof(int), &desiredvnodes, NULL);
2362
2363 if (oldval != desiredvnodes) {
2364 resize_namecache(desiredvnodes);
2365 }
2366
2367 return error;
2368 }
2369
2370 SYSCTL_INT(_kern, OID_AUTO, namecache_disabled,
2371 CTLFLAG_RW | CTLFLAG_LOCKED,
2372 &nc_disabled, 0, "");
2373
2374 SYSCTL_PROC(_kern, KERN_MAXVNODES, maxvnodes,
2375 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2376 0, 0, sysctl_maxvnodes, "I", "");
2377
2378 SYSCTL_PROC(_kern, KERN_MAXPROC, maxproc,
2379 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2380 0, 0, sysctl_maxproc, "I", "");
2381
2382 SYSCTL_PROC(_kern, KERN_AIOMAX, aiomax,
2383 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2384 0, 0, sysctl_aiomax, "I", "");
2385
2386 SYSCTL_PROC(_kern, KERN_AIOPROCMAX, aioprocmax,
2387 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2388 0, 0, sysctl_aioprocmax, "I", "");
2389
2390 SYSCTL_PROC(_kern, KERN_AIOTHREADS, aiothreads,
2391 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2392 0, 0, sysctl_aiothreads, "I", "");
2393
2394 SYSCTL_PROC(_kern, OID_AUTO, sched_enable_smt,
2395 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_KERN,
2396 0, 0, sysctl_sched_enable_smt, "I", "");
2397
2398 extern int sched_allow_NO_SMT_threads;
2399 SYSCTL_INT(_kern, OID_AUTO, sched_allow_NO_SMT_threads,
2400 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2401 &sched_allow_NO_SMT_threads, 0, "");
2402
2403 extern int sched_avoid_cpu0;
2404 SYSCTL_INT(_kern, OID_AUTO, sched_rt_avoid_cpu0,
2405 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2406 &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");
2407
2408 #if (DEVELOPMENT || DEBUG)
2409
2410 static int
2411 sysctl_kern_max_unsafe_rt_quanta(__unused struct sysctl_oid *oidp,
2412 __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2413 {
2414 extern void sched_set_max_unsafe_rt_quanta(int);
2415 extern int max_unsafe_rt_quanta;
2416
2417 int new_value, changed;
2418 int old_value = max_unsafe_rt_quanta;
2419 int error = sysctl_io_number(req, old_value, sizeof(int), &new_value,
2420 &changed);
2421 if (changed) {
2422 sched_set_max_unsafe_rt_quanta(new_value);
2423 }
2424
2425 return error;
2426 }
2427
2428 SYSCTL_PROC(_kern, OID_AUTO, max_unsafe_rt_quanta,
2429 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2430 0, 0, sysctl_kern_max_unsafe_rt_quanta, "I",
2431 "Number of quanta to allow a realtime "
2432 "thread to run before being penalized");
2433
2434 static int
2435 sysctl_kern_max_unsafe_fixed_quanta(__unused struct sysctl_oid *oidp,
2436 __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2437 {
2438 extern void sched_set_max_unsafe_fixed_quanta(int);
2439 extern int max_unsafe_fixed_quanta;
2440
2441 int new_value, changed;
2442 int old_value = max_unsafe_fixed_quanta;
2443 int error = sysctl_io_number(req, old_value, sizeof(int), &new_value,
2444 &changed);
2445 if (changed) {
2446 sched_set_max_unsafe_fixed_quanta(new_value);
2447 }
2448
2449 return error;
2450 }
2451
2452 SYSCTL_PROC(_kern, OID_AUTO, max_unsafe_fixed_quanta,
2453 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2454 0, 0, sysctl_kern_max_unsafe_fixed_quanta, "I",
2455 "Number of quanta to allow a fixed sched mode "
2456 "thread to run before being penalized");
2457
2458 static int
2459 sysctl_kern_quantum_us(__unused struct sysctl_oid *oidp, __unused void *arg1,
2460 __unused int arg2, struct sysctl_req *req)
2461 {
2462 const uint64_t quantum_us = sched_get_quantum_us();
2463
2464 return sysctl_io_number(req, quantum_us, sizeof(quantum_us), NULL, NULL);
2465 }
2466
2467 SYSCTL_PROC(_kern, OID_AUTO, quantum_us,
2468 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
2469 0, 0, sysctl_kern_quantum_us, "Q",
2470 "Length of scheduling quantum in microseconds");
2471
2472 extern int smt_sched_bonus_16ths;
2473 SYSCTL_INT(_kern, OID_AUTO, smt_sched_bonus_16ths,
2474 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2475 &smt_sched_bonus_16ths, 0, "");
2476
2477 extern int smt_timeshare_enabled;
2478 SYSCTL_INT(_kern, OID_AUTO, sched_smt_timeshare_enable,
2479 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2480 &smt_timeshare_enabled, 0, "");
2481
2482 extern int sched_smt_balance;
2483 SYSCTL_INT(_kern, OID_AUTO, sched_smt_balance,
2484 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2485 &sched_smt_balance, 0, "");
2486 extern int sched_allow_rt_smt;
2487 SYSCTL_INT(_kern, OID_AUTO, sched_allow_rt_smt,
2488 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2489 &sched_allow_rt_smt, 0, "");
2490 extern int sched_allow_rt_steal;
2491 SYSCTL_INT(_kern, OID_AUTO, sched_allow_rt_steal,
2492 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2493 &sched_allow_rt_steal, 0, "");
2494 extern int sched_backup_cpu_timeout_count;
2495 SYSCTL_INT(_kern, OID_AUTO, sched_backup_cpu_timeout_count,
2496 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2497 &sched_backup_cpu_timeout_count, 0, "The maximum number of 10us delays before allowing a backup cpu to select a thread");
2498 #if __arm64__
2499 /* Scheduler perfcontrol callouts sysctls */
2500 SYSCTL_DECL(_kern_perfcontrol_callout);
2501 SYSCTL_NODE(_kern, OID_AUTO, perfcontrol_callout, CTLFLAG_RW | CTLFLAG_LOCKED, 0,
2502 "scheduler perfcontrol callouts");
2503
2504 extern int perfcontrol_callout_stats_enabled;
2505 SYSCTL_INT(_kern_perfcontrol_callout, OID_AUTO, stats_enabled,
2506 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2507 &perfcontrol_callout_stats_enabled, 0, "");
2508
2509 extern uint64_t perfcontrol_callout_stat_avg(perfcontrol_callout_type_t type,
2510 perfcontrol_callout_stat_t stat);
2511
2512 /* On-Core Callout */
2513 STATIC int
2514 sysctl_perfcontrol_callout_stat
2515 (__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2516 {
2517 perfcontrol_callout_stat_t stat = (perfcontrol_callout_stat_t)arg1;
2518 perfcontrol_callout_type_t type = (perfcontrol_callout_type_t)arg2;
2519 return sysctl_io_number(req, (int)perfcontrol_callout_stat_avg(type, stat),
2520 sizeof(int), NULL, NULL);
2521 }
2522
2523 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, oncore_instr,
2524 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2525 (void *)PERFCONTROL_STAT_INSTRS, PERFCONTROL_CALLOUT_ON_CORE,
2526 sysctl_perfcontrol_callout_stat, "I", "");
2527 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, oncore_cycles,
2528 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2529 (void *)PERFCONTROL_STAT_CYCLES, PERFCONTROL_CALLOUT_ON_CORE,
2530 sysctl_perfcontrol_callout_stat, "I", "");
2531 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, offcore_instr,
2532 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2533 (void *)PERFCONTROL_STAT_INSTRS, PERFCONTROL_CALLOUT_OFF_CORE,
2534 sysctl_perfcontrol_callout_stat, "I", "");
2535 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, offcore_cycles,
2536 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2537 (void *)PERFCONTROL_STAT_CYCLES, PERFCONTROL_CALLOUT_OFF_CORE,
2538 sysctl_perfcontrol_callout_stat, "I", "");
2539 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, context_instr,
2540 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2541 (void *)PERFCONTROL_STAT_INSTRS, PERFCONTROL_CALLOUT_CONTEXT,
2542 sysctl_perfcontrol_callout_stat, "I", "");
2543 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, context_cycles,
2544 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2545 (void *)PERFCONTROL_STAT_CYCLES, PERFCONTROL_CALLOUT_CONTEXT,
2546 sysctl_perfcontrol_callout_stat, "I", "");
2547 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, update_instr,
2548 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2549 (void *)PERFCONTROL_STAT_INSTRS, PERFCONTROL_CALLOUT_STATE_UPDATE,
2550 sysctl_perfcontrol_callout_stat, "I", "");
2551 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, update_cycles,
2552 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2553 (void *)PERFCONTROL_STAT_CYCLES, PERFCONTROL_CALLOUT_STATE_UPDATE,
2554 sysctl_perfcontrol_callout_stat, "I", "");
2555
2556 #if __AMP__
2557 #if !CONFIG_CLUTCH
2558 extern int sched_amp_idle_steal;
2559 SYSCTL_INT(_kern, OID_AUTO, sched_amp_idle_steal,
2560 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2561 &sched_amp_idle_steal, 0, "");
2562 extern int sched_amp_spill_steal;
2563 SYSCTL_INT(_kern, OID_AUTO, sched_amp_spill_steal,
2564 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2565 &sched_amp_spill_steal, 0, "");
2566 extern int sched_amp_spill_count;
2567 SYSCTL_INT(_kern, OID_AUTO, sched_amp_spill_count,
2568 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2569 &sched_amp_spill_count, 0, "");
2570 #endif /* !CONFIG_CLUTCH */
2571 extern int sched_amp_spill_deferred_ipi;
2572 SYSCTL_INT(_kern, OID_AUTO, sched_amp_spill_deferred_ipi,
2573 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2574 &sched_amp_spill_deferred_ipi, 0, "");
2575 extern int sched_amp_pcores_preempt_immediate_ipi;
2576 SYSCTL_INT(_kern, OID_AUTO, sched_amp_pcores_preempt_immediate_ipi,
2577 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2578 &sched_amp_pcores_preempt_immediate_ipi, 0, "");
2579 #endif /* __AMP__ */
2580 #endif /* __arm64__ */
2581
2582 #if __arm64__
2583 extern int legacy_footprint_entitlement_mode;
2584 SYSCTL_INT(_kern, OID_AUTO, legacy_footprint_entitlement_mode,
2585 CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
2586 &legacy_footprint_entitlement_mode, 0, "");
2587 #endif /* __arm64__ */
2588
2589 /*
2590 * Realtime threads are ordered by highest priority first then,
2591 * for threads of the same priority, by earliest deadline first.
2592 * But if sched_rt_runq_strict_priority is false (the default),
2593 * a lower priority thread with an earlier deadline will be preferred
2594 * over a higher priority thread with a later deadline, as long as
2595 * both threads' computations will fit before the later deadline.
2596 */
2597 extern int sched_rt_runq_strict_priority;
2598 SYSCTL_INT(_kern, OID_AUTO, sched_rt_runq_strict_priority,
2599 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2600 &sched_rt_runq_strict_priority, 0, "");
2601
2602 static int
2603 sysctl_kern_sched_rt_n_backup_processors(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2604 {
2605 int new_value, changed;
2606 int old_value = sched_get_rt_n_backup_processors();
2607 int error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
2608 if (changed) {
2609 sched_set_rt_n_backup_processors(new_value);
2610 }
2611
2612 return error;
2613 }
2614
2615 SYSCTL_PROC(_kern, OID_AUTO, sched_rt_n_backup_processors,
2616 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2617 0, 0, sysctl_kern_sched_rt_n_backup_processors, "I", "");
2618
2619 static int
2620 sysctl_kern_sched_rt_deadline_epsilon_us(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2621 {
2622 int new_value, changed;
2623 int old_value = sched_get_rt_deadline_epsilon();
2624 int error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
2625 if (changed) {
2626 sched_set_rt_deadline_epsilon(new_value);
2627 }
2628
2629 return error;
2630 }
2631
2632 SYSCTL_PROC(_kern, OID_AUTO, sched_rt_deadline_epsilon_us,
2633 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2634 0, 0, sysctl_kern_sched_rt_deadline_epsilon_us, "I", "");
2635
2636 extern int sched_idle_delay_cpuid;
2637 SYSCTL_INT(_kern, OID_AUTO, sched_idle_delay_cpuid,
2638 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2639 &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");
2640
2641 static int
2642 sysctl_kern_sched_powered_cores(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2643 {
2644 int new_value, changed;
2645 int old_value = sched_get_powered_cores();
2646 int error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
2647 if (changed) {
2648 if (!PE_parse_boot_argn("enable_skstb", NULL, 0)) {
2649 return ENOTSUP;
2650 }
2651
2652 sched_set_powered_cores(new_value);
2653 }
2654
2655 return error;
2656 }
2657
2658 SYSCTL_PROC(_kern, OID_AUTO, sched_powered_cores,
2659 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2660 0, 0, sysctl_kern_sched_powered_cores, "I", "");
2661
2662 #if __arm64__
2663
2664 static int
2665 sysctl_kern_update_sched_recommended_cores(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2666 {
2667 uint64_t new_value;
2668 int changed;
2669 uint64_t old_value = sched_sysctl_get_recommended_cores();
2670 int error = sysctl_io_number(req, old_value, sizeof(uint64_t), &new_value, &changed);
2671 if (changed) {
2672 if (!PE_parse_boot_argn("enable_skstb", NULL, 0)) {
2673 return ENOTSUP;
2674 }
2675
2676 sched_perfcontrol_update_recommended_cores_reason(new_value, REASON_CLPC_USER, 0);
2677 }
2678
2679 return error;
2680 }
2681
2682 SYSCTL_PROC(_kern, OID_AUTO, sched_update_recommended_cores,
2683 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
2684 0, 0, sysctl_kern_update_sched_recommended_cores, "I", "");
2685
2686 #endif /* __arm64__ */
2687
2688 #endif /* (DEVELOPMENT || DEBUG) */
2689
2690 extern uint64_t sysctl_sched_recommended_cores;
2691 SYSCTL_QUAD(_kern, OID_AUTO, sched_recommended_cores,
2692 CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
2693 &sysctl_sched_recommended_cores, "");
2694
2695 static int
2696 sysctl_kern_suspend_cluster_powerdown(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2697 {
2698 int new_value, changed;
2699 int old_value = get_cluster_powerdown_user_suspended();
2700 int error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
2701 if (!error && changed) {
2702 if (new_value > 0) {
2703 error = suspend_cluster_powerdown_from_user();
2704 } else {
2705 error = resume_cluster_powerdown_from_user();
2706 }
2707 if (error) {
2708 error = EALREADY;
2709 }
2710 }
2711
2712 return error;
2713 }
2714
2715 SYSCTL_PROC(_kern, OID_AUTO, suspend_cluster_powerdown,
2716 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2717 0, 0, sysctl_kern_suspend_cluster_powerdown, "I", "");
2718
2719
2720 STATIC int
2721 sysctl_securelvl
2722 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2723 {
2724 int new_value, changed;
2725 int error = sysctl_io_number(req, securelevel, sizeof(int), &new_value, &changed);
2726 if (changed) {
2727 if (!(new_value < securelevel && proc_getpid(req->p) != 1)) {
2728 proc_list_lock();
2729 securelevel = new_value;
2730 proc_list_unlock();
2731 } else {
2732 error = EPERM;
2733 }
2734 }
2735 return error;
2736 }
2737
2738 SYSCTL_PROC(_kern, KERN_SECURELVL, securelevel,
2739 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2740 0, 0, sysctl_securelvl, "I", "");
2741
2742
2743 STATIC int
2744 sysctl_domainname
2745 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2746 {
2747 int error, changed;
2748 char tmpname[MAXHOSTNAMELEN] = {};
2749
2750 lck_mtx_lock(&domainname_lock);
2751 strlcpy(tmpname, domainname, sizeof(tmpname));
2752 lck_mtx_unlock(&domainname_lock);
2753
2754 error = sysctl_io_string(req, tmpname, sizeof(tmpname), 0, &changed);
2755 if (!error && changed) {
2756 lck_mtx_lock(&domainname_lock);
2757 strlcpy(domainname, tmpname, sizeof(domainname));
2758 lck_mtx_unlock(&domainname_lock);
2759 }
2760 return error;
2761 }
2762
2763 SYSCTL_PROC(_kern, KERN_DOMAINNAME, nisdomainname,
2764 CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_LOCKED,
2765 0, 0, sysctl_domainname, "A", "");
2766
2767 SYSCTL_COMPAT_INT(_kern, KERN_HOSTID, hostid,
2768 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2769 &hostid, 0, "");
2770
2771 STATIC int
2772 sysctl_hostname
2773 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2774 {
2775 int error, changed;
2776 char tmpname[MAXHOSTNAMELEN] = {};
2777 const char * name;
2778
2779 #if XNU_TARGET_OS_OSX
2780 name = hostname;
2781 #else /* XNU_TARGET_OS_OSX */
2782 #define ENTITLEMENT_USER_ASSIGNED_DEVICE_NAME \
2783 "com.apple.developer.device-information.user-assigned-device-name"
2784 if (csproc_get_platform_binary(current_proc()) ||
2785 IOCurrentTaskHasEntitlement(ENTITLEMENT_USER_ASSIGNED_DEVICE_NAME)) {
2786 name = hostname;
2787 } else {
2788 /* Deny writes if we don't pass entitlement check */
2789 if (req->newptr) {
2790 return EPERM;
2791 }
2792
2793 name = "localhost";
2794 }
2795 #endif /* ! XNU_TARGET_OS_OSX */
2796
2797 lck_mtx_lock(&hostname_lock);
2798 strlcpy(tmpname, name, sizeof(tmpname));
2799 lck_mtx_unlock(&hostname_lock);
2800
2801 error = sysctl_io_string(req, tmpname, sizeof(tmpname), 1, &changed);
2802 if (!error && changed) {
2803 lck_mtx_lock(&hostname_lock);
2804 strlcpy(hostname, tmpname, sizeof(hostname));
2805 lck_mtx_unlock(&hostname_lock);
2806 }
2807 return error;
2808 }
2809
2810 SYSCTL_PROC(_kern, KERN_HOSTNAME, hostname,
2811 CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_LOCKED,
2812 0, 0, sysctl_hostname, "A", "");
2813
2814 STATIC int
2815 sysctl_procname
2816 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2817 {
2818 /* Original code allowed writing, I'm copying this, although this all makes
2819 * no sense to me. Besides, this sysctl is never used. */
2820 return sysctl_io_string(req, &req->p->p_name[0], (2 * MAXCOMLEN + 1), 1, NULL);
2821 }
2822
2823 SYSCTL_PROC(_kern, KERN_PROCNAME, procname,
2824 CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
2825 0, 0, sysctl_procname, "A", "");
2826
2827 SYSCTL_INT(_kern, KERN_SPECULATIVE_READS, speculative_reads_disabled,
2828 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2829 &speculative_reads_disabled, 0, "");
2830
2831 SYSCTL_UINT(_kern, OID_AUTO, preheat_max_bytes,
2832 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2833 &preheat_max_bytes, 0, "");
2834
2835 SYSCTL_UINT(_kern, OID_AUTO, preheat_min_bytes,
2836 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2837 &preheat_min_bytes, 0, "");
2838
2839 SYSCTL_UINT(_kern, OID_AUTO, speculative_prefetch_max,
2840 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2841 &speculative_prefetch_max, 0, "");
2842
2843 SYSCTL_UINT(_kern, OID_AUTO, speculative_prefetch_max_iosize,
2844 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2845 &speculative_prefetch_max_iosize, 0, "");
2846
2847 SYSCTL_UINT(_kern, OID_AUTO, vm_page_free_target,
2848 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2849 &vm_page_free_target, 0, "");
2850
2851 SYSCTL_UINT(_kern, OID_AUTO, vm_page_free_min,
2852 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2853 &vm_page_free_min, 0, "");
2854
2855 SYSCTL_UINT(_kern, OID_AUTO, vm_page_free_reserved,
2856 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2857 &vm_page_free_reserved, 0, "");
2858
2859 SYSCTL_UINT(_kern, OID_AUTO, vm_page_speculative_percentage,
2860 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2861 &vm_pageout_state.vm_page_speculative_percentage, 0, "");
2862
2863 SYSCTL_UINT(_kern, OID_AUTO, vm_page_speculative_q_age_ms,
2864 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2865 &vm_pageout_state.vm_page_speculative_q_age_ms, 0, "");
2866
2867 #if (DEVELOPMENT || DEBUG)
2868 SYSCTL_UINT(_kern, OID_AUTO, vm_page_max_speculative_age_q,
2869 CTLFLAG_RD,
2870 &vm_page_max_speculative_age_q, 0, "");
2871 #endif /* (DEVELOPMENT || DEBUG) */
2872
2873 SYSCTL_UINT(_kern, OID_AUTO, vm_max_delayed_work_limit,
2874 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2875 &vm_max_delayed_work_limit, 0, "");
2876
2877 SYSCTL_UINT(_kern, OID_AUTO, vm_max_batch,
2878 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2879 &vm_max_batch, 0, "");
2880
2881 SYSCTL_STRING(_kern, OID_AUTO, bootsessionuuid,
2882 CTLFLAG_RD | CTLFLAG_LOCKED,
2883 &bootsessionuuid_string, sizeof(bootsessionuuid_string), "");
2884
2885
2886 STATIC int
2887 sysctl_boottime
2888 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2889 {
2890 struct timeval tv;
2891 boottime_timeval(&tv);
2892 struct proc *p = req->p;
2893
2894 if (proc_is64bit(p)) {
2895 struct user64_timeval t = {};
2896 t.tv_sec = tv.tv_sec;
2897 t.tv_usec = tv.tv_usec;
2898 return sysctl_io_opaque(req, &t, sizeof(t), NULL);
2899 } else {
2900 struct user32_timeval t = {};
2901 t.tv_sec = (user32_time_t)tv.tv_sec;
2902 t.tv_usec = tv.tv_usec;
2903 return sysctl_io_opaque(req, &t, sizeof(t), NULL);
2904 }
2905 }
2906
2907 SYSCTL_PROC(_kern, KERN_BOOTTIME, boottime,
2908 CTLTYPE_STRUCT | CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
2909 0, 0, sysctl_boottime, "S,timeval", "");
2910
2911 extern bool IOGetBootUUID(char *);
2912
2913 /* non-static: written by imageboot.c */
2914 uuid_string_t fake_bootuuid;
2915
2916 STATIC int
2917 sysctl_bootuuid
2918 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2919 {
2920 int error = ENOENT;
2921
2922 /* check the first byte to see if the string has been
2923 * populated. this is a uuid_STRING_t, this check would
2924 * not work with a uuid_t.
2925 */
2926 if (fake_bootuuid[0] != '\0') {
2927 error = sysctl_io_string(req, fake_bootuuid, 0, 0, NULL);
2928 goto out;
2929 }
2930
2931 uuid_string_t uuid_string;
2932 if (IOGetBootUUID(uuid_string)) {
2933 uuid_t boot_uuid;
2934 error = uuid_parse(uuid_string, boot_uuid);
2935 if (!error) {
2936 error = sysctl_io_string(req, __DECONST(char *, uuid_string), 0, 0, NULL);
2937 }
2938 }
2939
2940 out:
2941 return error;
2942 }
2943
2944 SYSCTL_PROC(_kern, OID_AUTO, bootuuid,
2945 CTLTYPE_STRING | CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
2946 0, 0, sysctl_bootuuid, "A", "");
2947
2948
2949 extern bool IOGetApfsPrebootUUID(char *);
2950 extern bool IOGetAssociatedApfsVolgroupUUID(char *);
2951
2952 STATIC int
2953 sysctl_apfsprebootuuid
2954 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2955 {
2956 int error = ENOENT;
2957
2958 uuid_string_t uuid_string;
2959 if (IOGetApfsPrebootUUID(uuid_string)) {
2960 uuid_t apfs_preboot_uuid;
2961 error = uuid_parse(uuid_string, apfs_preboot_uuid);
2962 if (!error) {
2963 error = sysctl_io_string(req, __DECONST(char *, uuid_string), 0, 0, NULL);
2964 }
2965 }
2966
2967 return error;
2968 }
2969
2970 SYSCTL_PROC(_kern, OID_AUTO, apfsprebootuuid,
2971 CTLTYPE_STRING | CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
2972 0, 0, sysctl_apfsprebootuuid, "A", "");
2973
2974 STATIC int
2975 sysctl_targetsystemvolgroupuuid
2976 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2977 {
2978 int error = ENOENT;
2979
2980 uuid_string_t uuid_string;
2981 if (IOGetApfsPrebootUUID(uuid_string)) {
2982 uuid_t apfs_preboot_uuid;
2983 error = uuid_parse(uuid_string, apfs_preboot_uuid);
2984 if (!error) {
2985 error = sysctl_io_string(req, __DECONST(char *, uuid_string), 0, 0, NULL);
2986 }
2987 } else {
2988 /*
2989 * In special boot modes, such as kcgen-mode, the
2990 * apfs-preboot-uuid property will not be set. Instead, a
2991 * different property, associated-volume-group, will be set
2992 * which indicates the UUID of the VolumeGroup containing the
2993 * system volume into which you will boot.
2994 */
2995 if (IOGetAssociatedApfsVolgroupUUID(uuid_string)) {
2996 uuid_t apfs_preboot_uuid;
2997 error = uuid_parse(uuid_string, apfs_preboot_uuid);
2998 if (!error) {
2999 error = sysctl_io_string(req, __DECONST(char *, uuid_string), 0, 0, NULL);
3000 }
3001 }
3002 }
3003
3004 return error;
3005 }
3006
3007 SYSCTL_PROC(_kern, OID_AUTO, targetsystemvolgroupuuid,
3008 CTLTYPE_STRING | CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_MASKED | CTLFLAG_LOCKED,
3009 0, 0, sysctl_targetsystemvolgroupuuid, "A", "");
3010
3011
3012 extern bool IOGetBootManifestHash(char *, size_t *);
3013 extern bool IOGetBootObjectsPath(char *);
3014
3015 STATIC int
3016 sysctl_bootobjectspath
3017 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3018 {
3019 int error = ENOENT;
3020
3021 #if defined(__x86_64__)
3022 /* auth-root-dmg is used for the Intel BaseSystem in some flows,
3023 * e.g. createinstallmedia and as part of upgrading from 10.15 or earlier
3024 * under these scenarios, set_fake_bootuuid will be called when pivoting to
3025 * the new root filesystem. need honor the fake bootuuid.
3026 */
3027 if (fake_bootuuid[0] != '\0') {
3028 error = sysctl_io_string(req, fake_bootuuid, 0, 0, NULL);
3029 } else {
3030 /* for intel mac, boot objects reside in [preboot volume]/[bootuuid]
3031 * bootuuid and apfsprebootuuid are populated by efiboot and they are alias.
3032 */
3033 uuid_string_t uuid_string;
3034 if (IOGetBootUUID(uuid_string)) {
3035 uuid_t boot_uuid;
3036 error = uuid_parse(uuid_string, boot_uuid);
3037 if (!error) {
3038 error = sysctl_io_string(req, (char *)uuid_string, 0, 0, NULL);
3039 }
3040 }
3041 }
3042 #else
3043 char boot_obj_path[MAXPATHLEN] = { "\0" };
3044 static const char kAsciiHexChars[] = "0123456789ABCDEF";
3045 unsigned int i, j;
3046
3047 /* Hashed with SHA2-384 or SHA1, boot manifest hash is 48 bytes or 20 bytes
3048 * hence, need a 97 bytes char array for the string.
3049 */
3050 size_t hash_data_size = CCSHA384_OUTPUT_SIZE;
3051 char hash_data[CCSHA384_OUTPUT_SIZE] = { "\0" };
3052 char boot_manifest_hash[CCSHA384_OUTPUT_SIZE * 2 + 1] = { "\0" };;
3053
3054 /* for Apple Silicon Macs, there is a boot-objects-path under IODeviceTree:/chosen
3055 * and boot objects reside in [preboot volume]/[boot-objects-path]
3056 * for embedded platforms, there would be a boot-manifest-hash under IODeviceTree:/chosen
3057 * and boot objects reside in [preboot volume]/[boot-manifest-hash]
3058 */
3059 if (IOGetBootObjectsPath(boot_obj_path)) {
3060 error = sysctl_io_string(req, (char *)boot_obj_path, 0, 0, NULL);
3061 } else if (IOGetBootManifestHash(hash_data, &hash_data_size)) {
3062 j = 0;
3063 for (i = 0; i < hash_data_size; ++i) {
3064 char octet = hash_data[i];
3065 boot_manifest_hash[j++] = kAsciiHexChars[((octet & 0xF0) >> 4)];
3066 boot_manifest_hash[j++] = kAsciiHexChars[(octet & 0x0F)];
3067 }
3068 /* make sure string has null termination */
3069 boot_manifest_hash[j] = '\0';
3070 error = sysctl_io_string(req, (char *)boot_manifest_hash, 0, 0, NULL);
3071 }
3072 #endif
3073 return error;
3074 }
3075
3076 SYSCTL_PROC(_kern, OID_AUTO, bootobjectspath,
3077 CTLTYPE_STRING | CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
3078 0, 0, sysctl_bootobjectspath, "A", "");
3079
3080
3081 STATIC int
3082 sysctl_symfile
3083 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3084 {
3085 char *str;
3086 int error = get_kernel_symfile(req->p, &str);
3087 if (error) {
3088 return error;
3089 }
3090 return sysctl_io_string(req, str, 0, 0, NULL);
3091 }
3092
3093
3094 SYSCTL_PROC(_kern, KERN_SYMFILE, symfile,
3095 CTLTYPE_STRING | CTLFLAG_RD | CTLFLAG_LOCKED,
3096 0, 0, sysctl_symfile, "A", "");
3097
3098 #if CONFIG_NETBOOT
3099 STATIC int
3100 sysctl_netboot
3101 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3102 {
3103 return sysctl_io_number(req, netboot_root(), sizeof(int), NULL, NULL);
3104 }
3105
3106 SYSCTL_PROC(_kern, KERN_NETBOOT, netboot,
3107 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
3108 0, 0, sysctl_netboot, "I", "");
3109 #endif
3110
3111 #ifdef CONFIG_IMGSRC_ACCESS
3112 /*
3113 * Legacy--act as if only one layer of nesting is possible.
3114 */
3115 STATIC int
3116 sysctl_imgsrcdev
3117 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3118 {
3119 vfs_context_t ctx = vfs_context_current();
3120 vnode_t devvp;
3121 int result;
3122
3123 if (!vfs_context_issuser(ctx)) {
3124 return EPERM;
3125 }
3126
3127 if (imgsrc_rootvnodes[0] == NULL) {
3128 return ENOENT;
3129 }
3130
3131 result = vnode_getwithref(imgsrc_rootvnodes[0]);
3132 if (result != 0) {
3133 return result;
3134 }
3135
3136 devvp = vnode_mount(imgsrc_rootvnodes[0])->mnt_devvp;
3137 result = vnode_getwithref(devvp);
3138 if (result != 0) {
3139 goto out;
3140 }
3141
3142 result = sysctl_io_number(req, vnode_specrdev(devvp), sizeof(dev_t), NULL, NULL);
3143
3144 vnode_put(devvp);
3145 out:
3146 vnode_put(imgsrc_rootvnodes[0]);
3147 return result;
3148 }
3149
3150 SYSCTL_PROC(_kern, OID_AUTO, imgsrcdev,
3151 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
3152 0, 0, sysctl_imgsrcdev, "I", "");
3153
3154 STATIC int
3155 sysctl_imgsrcinfo
3156 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3157 {
3158 int error;
3159 struct imgsrc_info info[MAX_IMAGEBOOT_NESTING] = {}; /* 2 for now, no problem */
3160 uint32_t i;
3161 vnode_t rvp, devvp;
3162
3163 if (imgsrc_rootvnodes[0] == NULLVP) {
3164 return ENXIO;
3165 }
3166
3167 for (i = 0; i < MAX_IMAGEBOOT_NESTING; i++) {
3168 /*
3169 * Go get the root vnode.
3170 */
3171 rvp = imgsrc_rootvnodes[i];
3172 if (rvp == NULLVP) {
3173 break;
3174 }
3175
3176 error = vnode_get(rvp);
3177 if (error != 0) {
3178 return error;
3179 }
3180
3181 /*
3182 * For now, no getting at a non-local volume.
3183 */
3184 devvp = vnode_mount(rvp)->mnt_devvp;
3185 if (devvp == NULL) {
3186 vnode_put(rvp);
3187 return EINVAL;
3188 }
3189
3190 error = vnode_getwithref(devvp);
3191 if (error != 0) {
3192 vnode_put(rvp);
3193 return error;
3194 }
3195
3196 /*
3197 * Fill in info.
3198 */
3199 info[i].ii_dev = vnode_specrdev(devvp);
3200 info[i].ii_flags = 0;
3201 info[i].ii_height = i;
3202 bzero(info[i].ii_reserved, sizeof(info[i].ii_reserved));
3203
3204 vnode_put(devvp);
3205 vnode_put(rvp);
3206 }
3207
3208 return sysctl_io_opaque(req, info, i * sizeof(info[0]), NULL);
3209 }
3210
3211 SYSCTL_PROC(_kern, OID_AUTO, imgsrcinfo,
3212 CTLTYPE_OPAQUE | CTLFLAG_RD | CTLFLAG_LOCKED,
3213 0, 0, sysctl_imgsrcinfo, "I", "");
3214
3215 #endif /* CONFIG_IMGSRC_ACCESS */
3216
3217
3218 SYSCTL_DECL(_kern_timer);
3219 SYSCTL_NODE(_kern, OID_AUTO, timer, CTLFLAG_RW | CTLFLAG_LOCKED, 0, "timer");
3220
3221
3222 SYSCTL_INT(_kern_timer, OID_AUTO, coalescing_enabled,
3223 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
3224 &mach_timer_coalescing_enabled, 0, "");
3225
3226 SYSCTL_QUAD(_kern_timer, OID_AUTO, deadline_tracking_bin_1,
3227 CTLFLAG_RW | CTLFLAG_LOCKED,
3228 &timer_deadline_tracking_bin_1, "");
3229 SYSCTL_QUAD(_kern_timer, OID_AUTO, deadline_tracking_bin_2,
3230 CTLFLAG_RW | CTLFLAG_LOCKED,
3231 &timer_deadline_tracking_bin_2, "");
3232
3233 SYSCTL_DECL(_kern_timer_longterm);
3234 SYSCTL_NODE(_kern_timer, OID_AUTO, longterm, CTLFLAG_RW | CTLFLAG_LOCKED, 0, "longterm");
3235
3236
3237 /* Must match definition in osfmk/kern/timer_call.c */
3238 enum {
3239 THRESHOLD, QCOUNT,
3240 ENQUEUES, DEQUEUES, ESCALATES, SCANS, PREEMPTS,
3241 LATENCY, LATENCY_MIN, LATENCY_MAX, LONG_TERM_SCAN_LIMIT,
3242 LONG_TERM_SCAN_INTERVAL, LONG_TERM_SCAN_PAUSES,
3243 SCAN_LIMIT, SCAN_INTERVAL, SCAN_PAUSES, SCAN_POSTPONES,
3244 };
3245 extern uint64_t timer_sysctl_get(int);
3246 extern kern_return_t timer_sysctl_set(int, uint64_t);
3247
3248 STATIC int
3249 sysctl_timer
3250 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3251 {
3252 int oid = (int)arg1;
3253 uint64_t value = timer_sysctl_get(oid);
3254 uint64_t new_value;
3255 int error;
3256 int changed;
3257
3258 error = sysctl_io_number(req, value, sizeof(value), &new_value, &changed);
3259 if (changed) {
3260 kern_return_t kr = timer_sysctl_set(oid, new_value);
3261 error = mach_to_bsd_errno(kr);
3262 }
3263
3264 return error;
3265 }
3266
3267 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, threshold,
3268 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
3269 (void *) THRESHOLD, 0, sysctl_timer, "Q", "");
3270 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, scan_limit,
3271 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
3272 (void *) LONG_TERM_SCAN_LIMIT, 0, sysctl_timer, "Q", "");
3273 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, scan_interval,
3274 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
3275 (void *) LONG_TERM_SCAN_INTERVAL, 0, sysctl_timer, "Q", "");
3276
3277 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, qlen,
3278 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3279 (void *) QCOUNT, 0, sysctl_timer, "Q", "");
3280 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, scan_pauses,
3281 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3282 (void *) LONG_TERM_SCAN_PAUSES, 0, sysctl_timer, "Q", "");
3283
3284 #if DEBUG
3285 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, enqueues,
3286 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3287 (void *) ENQUEUES, 0, sysctl_timer, "Q", "");
3288 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, dequeues,
3289 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3290 (void *) DEQUEUES, 0, sysctl_timer, "Q", "");
3291 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, escalates,
3292 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3293 (void *) ESCALATES, 0, sysctl_timer, "Q", "");
3294 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, scans,
3295 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3296 (void *) SCANS, 0, sysctl_timer, "Q", "");
3297 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, preempts,
3298 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3299 (void *) PREEMPTS, 0, sysctl_timer, "Q", "");
3300 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, latency,
3301 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3302 (void *) LATENCY, 0, sysctl_timer, "Q", "");
3303 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, latency_min,
3304 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3305 (void *) LATENCY_MIN, 0, sysctl_timer, "Q", "");
3306 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, latency_max,
3307 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3308 (void *) LATENCY_MAX, 0, sysctl_timer, "Q", "");
3309 #endif /* DEBUG */
3310
3311 SYSCTL_PROC(_kern_timer, OID_AUTO, scan_limit,
3312 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
3313 (void *) SCAN_LIMIT, 0, sysctl_timer, "Q", "");
3314 SYSCTL_PROC(_kern_timer, OID_AUTO, scan_interval,
3315 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
3316 (void *) SCAN_INTERVAL, 0, sysctl_timer, "Q", "");
3317 SYSCTL_PROC(_kern_timer, OID_AUTO, scan_pauses,
3318 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3319 (void *) SCAN_PAUSES, 0, sysctl_timer, "Q", "");
3320 SYSCTL_PROC(_kern_timer, OID_AUTO, scan_postpones,
3321 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3322 (void *) SCAN_POSTPONES, 0, sysctl_timer, "Q", "");
3323
3324 STATIC int
3325 sysctl_usrstack
3326 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3327 {
3328 return sysctl_io_number(req, (int)req->p->user_stack, sizeof(int), NULL, NULL);
3329 }
3330
3331 SYSCTL_PROC(_kern, KERN_USRSTACK32, usrstack,
3332 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
3333 0, 0, sysctl_usrstack, "I", "");
3334
3335 STATIC int
3336 sysctl_usrstack64
3337 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3338 {
3339 return sysctl_io_number(req, req->p->user_stack, sizeof(req->p->user_stack), NULL, NULL);
3340 }
3341
3342 SYSCTL_PROC(_kern, KERN_USRSTACK64, usrstack64,
3343 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3344 0, 0, sysctl_usrstack64, "Q", "");
3345
3346
3347 #if EXCLAVES_COREDUMP
3348
3349 /* secure kernel coredump support. */
3350 extern unsigned int sc_dump_mode;
3351 SYSCTL_UINT(_kern, OID_AUTO, secure_coredump, CTLFLAG_RD, &sc_dump_mode, 0, "secure_coredump");
3352
3353 #endif /* EXCLAVES_COREDUMP */
3354
3355
3356 #if CONFIG_COREDUMP
3357
3358 SYSCTL_STRING(_kern, KERN_COREFILE, corefile,
3359 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
3360 corefilename, sizeof(corefilename), "");
3361
3362 SYSCTL_STRING(_kern, OID_AUTO, drivercorefile,
3363 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
3364 drivercorefilename, sizeof(drivercorefilename), "");
3365
3366 STATIC int
3367 sysctl_coredump
3368 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3369 {
3370 #ifdef SECURE_KERNEL
3371 (void)req;
3372 return ENOTSUP;
3373 #else
3374 int new_value, changed;
3375 int error = sysctl_io_number(req, do_coredump, sizeof(int), &new_value, &changed);
3376 if (changed) {
3377 if ((new_value == 0) || (new_value == 1)) {
3378 do_coredump = new_value;
3379 } else {
3380 error = EINVAL;
3381 }
3382 }
3383 return error;
3384 #endif
3385 }
3386
3387 SYSCTL_PROC(_kern, KERN_COREDUMP, coredump,
3388 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
3389 0, 0, sysctl_coredump, "I", "");
3390
3391 STATIC int
3392 sysctl_suid_coredump
3393 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3394 {
3395 #ifdef SECURE_KERNEL
3396 (void)req;
3397 return ENOTSUP;
3398 #else
3399 int new_value, changed;
3400 int error = sysctl_io_number(req, sugid_coredump, sizeof(int), &new_value, &changed);
3401 if (changed) {
3402 if ((new_value == 0) || (new_value == 1)) {
3403 sugid_coredump = new_value;
3404 } else {
3405 error = EINVAL;
3406 }
3407 }
3408 return error;
3409 #endif
3410 }
3411
3412 SYSCTL_PROC(_kern, KERN_SUGID_COREDUMP, sugid_coredump,
3413 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
3414 0, 0, sysctl_suid_coredump, "I", "");
3415
3416 #endif /* CONFIG_COREDUMP */
3417
3418 #if CONFIG_KDP_INTERACTIVE_DEBUGGING
3419
3420 extern const char* kdp_corefile_path;
3421 STATIC int
3422 sysctl_kdp_corefile(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3423 {
3424 return SYSCTL_OUT(req, kdp_corefile_path, strlen(kdp_corefile_path) + 1);
3425 }
3426
3427 /* this needs to be a proc rather than a string since kdp_corefile_path is not a compile-time constant */
3428 SYSCTL_PROC(_kern, OID_AUTO, kdp_corefile,
3429 CTLTYPE_STRING | CTLFLAG_RD | CTLFLAG_LOCKED,
3430 0, 0, sysctl_kdp_corefile, "A", "");
3431
3432 #endif /* CONFIG_KDP_INTERACTIVE_DEBUGGING */
3433
3434 STATIC int
3435 sysctl_delayterm
3436 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3437 {
3438 struct proc *p = req->p;
3439 int new_value, changed;
3440 int error = sysctl_io_number(req, (req->p->p_lflag & P_LDELAYTERM)? 1: 0, sizeof(int), &new_value, &changed);
3441 if (changed) {
3442 proc_lock(p);
3443 if (new_value) {
3444 req->p->p_lflag |= P_LDELAYTERM;
3445 } else {
3446 req->p->p_lflag &= ~P_LDELAYTERM;
3447 }
3448 proc_unlock(p);
3449 }
3450 return error;
3451 }
3452
3453 SYSCTL_PROC(_kern, KERN_PROCDELAYTERM, delayterm,
3454 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
3455 0, 0, sysctl_delayterm, "I", "");
3456
3457
3458 STATIC int
3459 sysctl_rage_vnode
3460 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3461 {
3462 struct proc *p = req->p;
3463 struct uthread *ut;
3464 int new_value, old_value, changed;
3465 int error;
3466
3467 ut = current_uthread();
3468
3469 if (ut->uu_flag & UT_RAGE_VNODES) {
3470 old_value = KERN_RAGE_THREAD;
3471 } else if (p->p_lflag & P_LRAGE_VNODES) {
3472 old_value = KERN_RAGE_PROC;
3473 } else {
3474 old_value = 0;
3475 }
3476
3477 error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
3478
3479 if ((error == 0) && (changed != 0)) {
3480 switch (new_value) {
3481 case KERN_RAGE_PROC:
3482 proc_lock(p);
3483 p->p_lflag |= P_LRAGE_VNODES;
3484 proc_unlock(p);
3485 break;
3486 case KERN_UNRAGE_PROC:
3487 proc_lock(p);
3488 p->p_lflag &= ~P_LRAGE_VNODES;
3489 proc_unlock(p);
3490 break;
3491
3492 case KERN_RAGE_THREAD:
3493 ut->uu_flag |= UT_RAGE_VNODES;
3494 break;
3495 case KERN_UNRAGE_THREAD:
3496 ut = current_uthread();
3497 ut->uu_flag &= ~UT_RAGE_VNODES;
3498 break;
3499 }
3500 }
3501 return error;
3502 }
3503
3504 SYSCTL_PROC(_kern, KERN_RAGEVNODE, rage_vnode,
3505 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
3506 0, 0, sysctl_rage_vnode, "I", "");
3507
3508 /* XXX until filecoordinationd fixes a bit of inverted logic. */
3509 STATIC int
3510 sysctl_vfsnspace
3511 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3512 {
3513 int old_value = 0, new_value, changed;
3514
3515 return sysctl_io_number(req, old_value, sizeof(int), &new_value,
3516 &changed);
3517 }
3518
3519 SYSCTL_PROC(_kern, OID_AUTO, vfsnspace,
3520 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
3521 0, 0, sysctl_vfsnspace, "I", "");
3522
3523 /* XXX move this interface into libproc and remove this sysctl */
3524 STATIC int
3525 sysctl_setthread_cpupercent
3526 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3527 {
3528 int new_value, old_value;
3529 int error = 0;
3530 kern_return_t kret = KERN_SUCCESS;
3531 uint8_t percent = 0;
3532 int ms_refill = 0;
3533
3534 if (!req->newptr) {
3535 return 0;
3536 }
3537
3538 old_value = 0;
3539
3540 if ((error = sysctl_io_number(req, old_value, sizeof(old_value), &new_value, NULL)) != 0) {
3541 return error;
3542 }
3543
3544 percent = new_value & 0xff; /* low 8 bytes for perent */
3545 ms_refill = (new_value >> 8) & 0xffffff; /* upper 24bytes represent ms refill value */
3546 if (percent > 100) {
3547 return EINVAL;
3548 }
3549
3550 /*
3551 * If the caller is specifying a percentage of 0, this will unset the CPU limit, if present.
3552 */
3553 kret = percent == 0 ?
3554 thread_set_cpulimit(THREAD_CPULIMIT_DISABLE, 0, 0) :
3555 thread_set_cpulimit(THREAD_CPULIMIT_BLOCK, percent, ms_refill * (int)NSEC_PER_MSEC);
3556
3557 if (kret != 0) {
3558 return EIO;
3559 }
3560
3561 return 0;
3562 }
3563
3564 SYSCTL_PROC(_kern, OID_AUTO, setthread_cpupercent,
3565 CTLTYPE_INT | CTLFLAG_WR | CTLFLAG_ANYBODY,
3566 0, 0, sysctl_setthread_cpupercent, "I", "set thread cpu percentage limit");
3567
3568
3569 STATIC int
3570 sysctl_kern_check_openevt
3571 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3572 {
3573 struct proc *p = req->p;
3574 int new_value, old_value, changed;
3575 int error;
3576
3577 if (p->p_flag & P_CHECKOPENEVT) {
3578 old_value = KERN_OPENEVT_PROC;
3579 } else {
3580 old_value = 0;
3581 }
3582
3583 error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
3584
3585 if ((error == 0) && (changed != 0)) {
3586 switch (new_value) {
3587 case KERN_OPENEVT_PROC:
3588 OSBitOrAtomic(P_CHECKOPENEVT, &p->p_flag);
3589 break;
3590
3591 case KERN_UNOPENEVT_PROC:
3592 OSBitAndAtomic(~((uint32_t)P_CHECKOPENEVT), &p->p_flag);
3593 break;
3594
3595 default:
3596 error = EINVAL;
3597 }
3598 }
3599 return error;
3600 }
3601
3602 SYSCTL_PROC(_kern, KERN_CHECKOPENEVT, check_openevt, CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
3603 0, 0, sysctl_kern_check_openevt, "I", "set the per-process check-open-evt flag");
3604
3605
3606 #if DEVELOPMENT || DEBUG
3607 STATIC int
3608 sysctl_nx
3609 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3610 {
3611 #ifdef SECURE_KERNEL
3612 (void)req;
3613 return ENOTSUP;
3614 #else
3615 int new_value, changed;
3616 int error;
3617
3618 error = sysctl_io_number(req, nx_enabled, sizeof(nx_enabled), &new_value, &changed);
3619 if (error) {
3620 return error;
3621 }
3622
3623 if (changed) {
3624 #if defined(__x86_64__)
3625 /*
3626 * Only allow setting if NX is supported on the chip
3627 */
3628 if (!(cpuid_extfeatures() & CPUID_EXTFEATURE_XD)) {
3629 return ENOTSUP;
3630 }
3631 #endif
3632 nx_enabled = new_value;
3633 }
3634 return error;
3635 #endif /* SECURE_KERNEL */
3636 }
3637 #endif
3638
3639 #if DEVELOPMENT || DEBUG
3640 SYSCTL_PROC(_kern, KERN_NX_PROTECTION, nx,
3641 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
3642 0, 0, sysctl_nx, "I", "");
3643 #endif
3644
3645 STATIC int
3646 sysctl_loadavg
3647 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3648 {
3649 if (proc_is64bit(req->p)) {
3650 struct user64_loadavg loadinfo64 = {};
3651 fill_loadavg64(&averunnable, &loadinfo64);
3652 return sysctl_io_opaque(req, &loadinfo64, sizeof(loadinfo64), NULL);
3653 } else {
3654 struct user32_loadavg loadinfo32 = {};
3655 fill_loadavg32(&averunnable, &loadinfo32);
3656 return sysctl_io_opaque(req, &loadinfo32, sizeof(loadinfo32), NULL);
3657 }
3658 }
3659
3660 SYSCTL_PROC(_vm, VM_LOADAVG, loadavg,
3661 CTLTYPE_STRUCT | CTLFLAG_RD | CTLFLAG_LOCKED,
3662 0, 0, sysctl_loadavg, "S,loadavg", "");
3663
3664 /*
3665 * Note: Thread safe; vm_map_lock protects in vm_toggle_entry_reuse()
3666 */
3667 STATIC int
3668 sysctl_vm_toggle_address_reuse(__unused struct sysctl_oid *oidp, __unused void *arg1,
3669 __unused int arg2, struct sysctl_req *req)
3670 {
3671 int old_value = 0, new_value = 0, error = 0;
3672
3673 if (vm_toggle_entry_reuse( VM_TOGGLE_GETVALUE, &old_value )) {
3674 return error;
3675 }
3676 error = sysctl_io_number(req, old_value, sizeof(int), &new_value, NULL);
3677 if (!error) {
3678 return vm_toggle_entry_reuse(new_value, NULL);
3679 }
3680 return error;
3681 }
3682
3683 SYSCTL_PROC(_debug, OID_AUTO, toggle_address_reuse, CTLFLAG_ANYBODY | CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED, 0, 0, sysctl_vm_toggle_address_reuse, "I", "");
3684
3685 #ifdef CONFIG_XNUPOST
3686
3687 extern uint32_t xnupost_get_estimated_testdata_size(void);
3688 extern int xnupost_reset_all_tests(void);
3689
3690 STATIC int
3691 sysctl_handle_xnupost_get_tests SYSCTL_HANDLER_ARGS
3692 {
3693 /* fixup unused arguments warnings */
3694 __unused int _oa2 = arg2;
3695 __unused void * _oa1 = arg1;
3696 __unused struct sysctl_oid * _oidp = oidp;
3697
3698 int error = 0;
3699 user_addr_t oldp = 0;
3700 user_addr_t newp = 0;
3701 uint32_t usedbytes = 0;
3702
3703 oldp = req->oldptr;
3704 newp = req->newptr;
3705
3706 if (newp) {
3707 return ENOTSUP;
3708 }
3709
3710 if ((void *)oldp == NULL) {
3711 /* return estimated size for second call where info can be placed */
3712 req->oldidx = xnupost_get_estimated_testdata_size();
3713 } else {
3714 error = xnupost_export_testdata((void *)oldp, req->oldlen, &usedbytes);
3715 req->oldidx = usedbytes;
3716 }
3717
3718 return error;
3719 }
3720
3721 SYSCTL_PROC(_debug,
3722 OID_AUTO,
3723 xnupost_get_tests,
3724 CTLFLAG_MASKED | CTLFLAG_ANYBODY | CTLTYPE_OPAQUE | CTLFLAG_RD | CTLFLAG_LOCKED,
3725 0,
3726 0,
3727 sysctl_handle_xnupost_get_tests,
3728 "-",
3729 "read xnupost test data in kernel");
3730
3731 #if CONFIG_EXT_PANICLOG
3732 /*
3733 * Extensible panic log test hooks
3734 */
3735 static int
3736 sysctl_debug_ext_paniclog_test_hook SYSCTL_HANDLER_ARGS
3737 {
3738 #pragma unused(arg1, arg2)
3739 int rval = 0;
3740 uint32_t test_option = 0;
3741
3742 rval = sysctl_handle_int(oidp, &test_option, 0, req);
3743
3744 if (rval == 0 && req->newptr) {
3745 rval = ext_paniclog_test_hook(test_option);
3746 }
3747
3748 return rval;
3749 }
3750
3751 SYSCTL_PROC(_debug, OID_AUTO, ext_paniclog_test_hook,
3752 CTLTYPE_INT | CTLFLAG_RW,
3753 0, 0,
3754 sysctl_debug_ext_paniclog_test_hook, "A", "ext paniclog test hook");
3755
3756 #endif
3757
3758 STATIC int
3759 sysctl_debug_xnupost_ctl SYSCTL_HANDLER_ARGS
3760 {
3761 /* fixup unused arguments warnings */
3762 __unused int _oa2 = arg2;
3763 __unused void * _oa1 = arg1;
3764 __unused struct sysctl_oid * _oidp = oidp;
3765
3766 #define ARRCOUNT 4
3767 /*
3768 * INPUT: ACTION, PARAM1, PARAM2, PARAM3
3769 * OUTPUT: RESULTCODE, ADDITIONAL DATA
3770 */
3771 int32_t outval[ARRCOUNT] = {0};
3772 int32_t input[ARRCOUNT] = {0};
3773 int32_t out_size = sizeof(outval);
3774 int32_t in_size = sizeof(input);
3775 int error = 0;
3776
3777 /* if this is NULL call to find out size, send out size info */
3778 if (!req->newptr) {
3779 goto out;
3780 }
3781
3782 /* pull in provided value from userspace */
3783 error = SYSCTL_IN(req, &input[0], in_size);
3784 if (error) {
3785 return error;
3786 }
3787
3788 if (input[0] == XTCTL_RESET_TESTDATA) {
3789 outval[0] = xnupost_reset_all_tests();
3790 goto out;
3791 }
3792
3793 out:
3794 error = SYSCTL_OUT(req, &outval[0], out_size);
3795 return error;
3796 }
3797
3798 SYSCTL_PROC(_debug,
3799 OID_AUTO,
3800 xnupost_testctl,
3801 CTLFLAG_MASKED | CTLFLAG_ANYBODY | CTLTYPE_OPAQUE | CTLFLAG_RW | CTLFLAG_LOCKED,
3802 0,
3803 0,
3804 sysctl_debug_xnupost_ctl,
3805 "I",
3806 "xnupost control for kernel testing");
3807
3808 extern void test_oslog_handleOSLogCtl(int32_t * in, int32_t * out, int32_t arraycount);
3809
3810 STATIC int
3811 sysctl_debug_test_oslog_ctl(__unused struct sysctl_oid * oidp, __unused void * arg1, __unused int arg2, struct sysctl_req * req)
3812 {
3813 #define ARRCOUNT 4
3814 int32_t outval[ARRCOUNT] = {0};
3815 int32_t input[ARRCOUNT] = {0};
3816 int32_t size_outval = sizeof(outval);
3817 int32_t size_inval = sizeof(input);
3818 int32_t error;
3819
3820 /* if this is NULL call to find out size, send out size info */
3821 if (!req->newptr) {
3822 error = SYSCTL_OUT(req, &outval[0], size_outval);
3823 return error;
3824 }
3825
3826 /* pull in provided value from userspace */
3827 error = SYSCTL_IN(req, &input[0], size_inval);
3828 if (error) {
3829 return error;
3830 }
3831
3832 test_oslog_handleOSLogCtl(input, outval, ARRCOUNT);
3833
3834 error = SYSCTL_OUT(req, &outval[0], size_outval);
3835
3836 return error;
3837 }
3838
3839 SYSCTL_PROC(_debug,
3840 OID_AUTO,
3841 test_OSLogCtl,
3842 CTLFLAG_MASKED | CTLFLAG_ANYBODY | CTLTYPE_OPAQUE | CTLFLAG_RW | CTLFLAG_LOCKED,
3843 0,
3844 0,
3845 sysctl_debug_test_oslog_ctl,
3846 "I",
3847 "testing oslog in kernel");
3848
3849 #include <mach/task.h>
3850 #include <mach/semaphore.h>
3851
3852 static LCK_GRP_DECLARE(sysctl_debug_test_stackshot_owner_grp, "test-stackshot-owner-grp");
3853 static LCK_MTX_DECLARE(sysctl_debug_test_stackshot_owner_init_mtx,
3854 &sysctl_debug_test_stackshot_owner_grp);
3855
3856 /* This is a sysctl for testing collection of owner info on a lock in kernel space. A multi-threaded
3857 * test from userland sets this sysctl in such a way that a thread blocks in kernel mode, and a
3858 * stackshot is taken to see if the owner of the lock can be identified.
3859 *
3860 * We can't return to userland with a kernel lock held, so be sure to unlock before we leave.
3861 * the semaphores allow us to artificially create cases where the lock is being held and the
3862 * thread is hanging / taking a long time to do something. */
3863
3864 volatile char sysctl_debug_test_stackshot_mtx_inited = 0;
3865 semaphore_t sysctl_debug_test_stackshot_mutex_sem;
3866 lck_mtx_t sysctl_debug_test_stackshot_owner_lck;
3867
3868 #define SYSCTL_DEBUG_MTX_ACQUIRE_WAIT 1
3869 #define SYSCTL_DEBUG_MTX_ACQUIRE_NOWAIT 2
3870 #define SYSCTL_DEBUG_MTX_SIGNAL 3
3871 #define SYSCTL_DEBUG_MTX_TEARDOWN 4
3872
3873 STATIC int
3874 sysctl_debug_test_stackshot_mutex_owner(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3875 {
3876 long long option = -1;
3877 /* if the user tries to read the sysctl, we tell them what the address of the lock is (to test against stackshot's output) */
3878 long long mtx_unslid_addr = (long long)VM_KERNEL_UNSLIDE_OR_PERM(&sysctl_debug_test_stackshot_owner_lck);
3879 int error = sysctl_io_number(req, mtx_unslid_addr, sizeof(long long), (void*)&option, NULL);
3880
3881 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_init_mtx);
3882 if (!sysctl_debug_test_stackshot_mtx_inited) {
3883 lck_mtx_init(&sysctl_debug_test_stackshot_owner_lck,
3884 &sysctl_debug_test_stackshot_owner_grp,
3885 LCK_ATTR_NULL);
3886 semaphore_create(kernel_task,
3887 &sysctl_debug_test_stackshot_mutex_sem,
3888 SYNC_POLICY_FIFO, 0);
3889 sysctl_debug_test_stackshot_mtx_inited = 1;
3890 }
3891 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_init_mtx);
3892
3893 if (!error) {
3894 switch (option) {
3895 case SYSCTL_DEBUG_MTX_ACQUIRE_NOWAIT:
3896 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_lck);
3897 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_lck);
3898 break;
3899 case SYSCTL_DEBUG_MTX_ACQUIRE_WAIT:
3900 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_lck);
3901 semaphore_wait(sysctl_debug_test_stackshot_mutex_sem);
3902 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_lck);
3903 break;
3904 case SYSCTL_DEBUG_MTX_SIGNAL:
3905 semaphore_signal(sysctl_debug_test_stackshot_mutex_sem);
3906 break;
3907 case SYSCTL_DEBUG_MTX_TEARDOWN:
3908 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_init_mtx);
3909
3910 lck_mtx_destroy(&sysctl_debug_test_stackshot_owner_lck,
3911 &sysctl_debug_test_stackshot_owner_grp);
3912 semaphore_destroy(kernel_task,
3913 sysctl_debug_test_stackshot_mutex_sem);
3914 sysctl_debug_test_stackshot_mtx_inited = 0;
3915
3916 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_init_mtx);
3917 break;
3918 case -1: /* user just wanted to read the value, so do nothing */
3919 break;
3920 default:
3921 error = EINVAL;
3922 break;
3923 }
3924 }
3925 return error;
3926 }
3927
3928 /* we can't return to userland with a kernel rwlock held, so be sure to unlock before we leave.
3929 * the semaphores allow us to artificially create cases where the lock is being held and the
3930 * thread is hanging / taking a long time to do something. */
3931
3932 SYSCTL_PROC(_debug,
3933 OID_AUTO,
3934 test_MutexOwnerCtl,
3935 CTLFLAG_MASKED | CTLFLAG_ANYBODY | CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
3936 0,
3937 0,
3938 sysctl_debug_test_stackshot_mutex_owner,
3939 "-",
3940 "Testing mutex owner in kernel");
3941
3942 volatile char sysctl_debug_test_stackshot_rwlck_inited = 0;
3943 lck_rw_t sysctl_debug_test_stackshot_owner_rwlck;
3944 semaphore_t sysctl_debug_test_stackshot_rwlck_sem;
3945
3946 #define SYSCTL_DEBUG_KRWLCK_RACQUIRE_NOWAIT 1
3947 #define SYSCTL_DEBUG_KRWLCK_RACQUIRE_WAIT 2
3948 #define SYSCTL_DEBUG_KRWLCK_WACQUIRE_NOWAIT 3
3949 #define SYSCTL_DEBUG_KRWLCK_WACQUIRE_WAIT 4
3950 #define SYSCTL_DEBUG_KRWLCK_SIGNAL 5
3951 #define SYSCTL_DEBUG_KRWLCK_TEARDOWN 6
3952
3953 STATIC int
3954 sysctl_debug_test_stackshot_rwlck_owner(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3955 {
3956 long long option = -1;
3957 /* if the user tries to read the sysctl, we tell them what the address of the lock is
3958 * (to test against stackshot's output) */
3959 long long rwlck_unslid_addr = (long long)VM_KERNEL_UNSLIDE_OR_PERM(&sysctl_debug_test_stackshot_owner_rwlck);
3960 int error = sysctl_io_number(req, rwlck_unslid_addr, sizeof(long long), (void*)&option, NULL);
3961
3962 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_init_mtx);
3963 if (!sysctl_debug_test_stackshot_rwlck_inited) {
3964 lck_rw_init(&sysctl_debug_test_stackshot_owner_rwlck,
3965 &sysctl_debug_test_stackshot_owner_grp,
3966 LCK_ATTR_NULL);
3967 semaphore_create(kernel_task,
3968 &sysctl_debug_test_stackshot_rwlck_sem,
3969 SYNC_POLICY_FIFO,
3970 0);
3971 sysctl_debug_test_stackshot_rwlck_inited = 1;
3972 }
3973 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_init_mtx);
3974
3975 if (!error) {
3976 switch (option) {
3977 case SYSCTL_DEBUG_KRWLCK_RACQUIRE_NOWAIT:
3978 lck_rw_lock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_SHARED);
3979 lck_rw_unlock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_SHARED);
3980 break;
3981 case SYSCTL_DEBUG_KRWLCK_RACQUIRE_WAIT:
3982 lck_rw_lock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_SHARED);
3983 semaphore_wait(sysctl_debug_test_stackshot_rwlck_sem);
3984 lck_rw_unlock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_SHARED);
3985 break;
3986 case SYSCTL_DEBUG_KRWLCK_WACQUIRE_NOWAIT:
3987 lck_rw_lock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_EXCLUSIVE);
3988 lck_rw_unlock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_EXCLUSIVE);
3989 break;
3990 case SYSCTL_DEBUG_KRWLCK_WACQUIRE_WAIT:
3991 lck_rw_lock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_EXCLUSIVE);
3992 semaphore_wait(sysctl_debug_test_stackshot_rwlck_sem);
3993 lck_rw_unlock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_EXCLUSIVE);
3994 break;
3995 case SYSCTL_DEBUG_KRWLCK_SIGNAL:
3996 semaphore_signal(sysctl_debug_test_stackshot_rwlck_sem);
3997 break;
3998 case SYSCTL_DEBUG_KRWLCK_TEARDOWN:
3999 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_init_mtx);
4000
4001 lck_rw_destroy(&sysctl_debug_test_stackshot_owner_rwlck,
4002 &sysctl_debug_test_stackshot_owner_grp);
4003 semaphore_destroy(kernel_task,
4004 sysctl_debug_test_stackshot_rwlck_sem);
4005 sysctl_debug_test_stackshot_rwlck_inited = 0;
4006
4007 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_init_mtx);
4008 break;
4009 case -1: /* user just wanted to read the value, so do nothing */
4010 break;
4011 default:
4012 error = EINVAL;
4013 break;
4014 }
4015 }
4016 return error;
4017 }
4018
4019
4020 SYSCTL_PROC(_debug,
4021 OID_AUTO,
4022 test_RWLockOwnerCtl,
4023 CTLFLAG_MASKED | CTLFLAG_ANYBODY | CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
4024 0,
4025 0,
4026 sysctl_debug_test_stackshot_rwlck_owner,
4027 "-",
4028 "Testing rwlock owner in kernel");
4029 #endif /* !CONFIG_XNUPOST */
4030
4031 STATIC int
4032 sysctl_swapusage
4033 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4034 {
4035 int error;
4036 uint64_t swap_total;
4037 uint64_t swap_avail;
4038 vm_size_t swap_pagesize;
4039 boolean_t swap_encrypted;
4040 struct xsw_usage xsu = {};
4041
4042 error = macx_swapinfo(&swap_total,
4043 &swap_avail,
4044 &swap_pagesize,
4045 &swap_encrypted);
4046 if (error) {
4047 return error;
4048 }
4049
4050 xsu.xsu_total = swap_total;
4051 xsu.xsu_avail = swap_avail;
4052 xsu.xsu_used = swap_total - swap_avail;
4053 xsu.xsu_pagesize = (u_int32_t)MIN(swap_pagesize, UINT32_MAX);
4054 xsu.xsu_encrypted = swap_encrypted;
4055 return sysctl_io_opaque(req, &xsu, sizeof(xsu), NULL);
4056 }
4057
4058
4059
4060 SYSCTL_PROC(_vm, VM_SWAPUSAGE, swapusage,
4061 CTLTYPE_STRUCT | CTLFLAG_RD | CTLFLAG_LOCKED,
4062 0, 0, sysctl_swapusage, "S,xsw_usage", "");
4063
4064 extern int vm_swap_enabled;
4065 SYSCTL_INT(_vm, OID_AUTO, swap_enabled, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_swap_enabled, 0, "");
4066
4067 #if DEVELOPMENT || DEBUG
4068 extern int vm_num_swap_files_config;
4069 extern int vm_num_swap_files;
4070 extern lck_mtx_t vm_swap_data_lock;
4071 #define VM_MAX_SWAP_FILE_NUM 100
4072
4073 static int
4074 sysctl_vm_config_num_swap_files SYSCTL_HANDLER_ARGS
4075 {
4076 #pragma unused(arg1, arg2)
4077 int error = 0, val = vm_num_swap_files_config;
4078
4079 error = sysctl_handle_int(oidp, &val, 0, req);
4080 if (error || !req->newptr) {
4081 goto out;
4082 }
4083
4084 if (!VM_CONFIG_SWAP_IS_ACTIVE && !VM_CONFIG_FREEZER_SWAP_IS_ACTIVE) {
4085 printf("Swap is disabled\n");
4086 error = EINVAL;
4087 goto out;
4088 }
4089
4090 lck_mtx_lock(&vm_swap_data_lock);
4091
4092 if (val < vm_num_swap_files) {
4093 printf("Cannot configure fewer swap files than already exist.\n");
4094 error = EINVAL;
4095 lck_mtx_unlock(&vm_swap_data_lock);
4096 goto out;
4097 }
4098
4099 if (val > VM_MAX_SWAP_FILE_NUM) {
4100 printf("Capping number of swap files to upper bound.\n");
4101 val = VM_MAX_SWAP_FILE_NUM;
4102 }
4103
4104 vm_num_swap_files_config = val;
4105 lck_mtx_unlock(&vm_swap_data_lock);
4106 out:
4107
4108 return 0;
4109 }
4110
4111 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", "");
4112 #endif /* DEVELOPMENT || DEBUG */
4113
4114 /* this kernel does NOT implement shared_region_make_private_np() */
4115 SYSCTL_INT(_kern, KERN_SHREG_PRIVATIZABLE, shreg_private,
4116 CTLFLAG_RD | CTLFLAG_LOCKED,
4117 (int *)NULL, 0, "");
4118
4119 STATIC int
4120 fetch_process_cputype(
4121 proc_t cur_proc,
4122 int *name,
4123 u_int namelen,
4124 cpu_type_t *cputype)
4125 {
4126 proc_t p = PROC_NULL;
4127 int refheld = 0;
4128 cpu_type_t ret = 0;
4129 int error = 0;
4130
4131 if (namelen == 0) {
4132 p = cur_proc;
4133 } else if (namelen == 1) {
4134 p = proc_find(name[0]);
4135 if (p == NULL) {
4136 return EINVAL;
4137 }
4138 refheld = 1;
4139 } else {
4140 error = EINVAL;
4141 goto out;
4142 }
4143
4144 ret = cpu_type() & ~CPU_ARCH_MASK;
4145 if (IS_64BIT_PROCESS(p)) {
4146 ret |= CPU_ARCH_ABI64;
4147 }
4148
4149 *cputype = ret;
4150
4151 if (refheld != 0) {
4152 proc_rele(p);
4153 }
4154 out:
4155 return error;
4156 }
4157
4158 #if CONFIG_ROSETTA
4159 STATIC int
4160 sysctl_sysctl_translated(
4161 __unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, __unused struct sysctl_req *req)
4162 {
4163 int res = 0;
4164 if (proc_is_translated(req->p)) {
4165 res = 1;
4166 }
4167 return SYSCTL_OUT(req, &res, sizeof(res));
4168 }
4169 SYSCTL_PROC(_sysctl, OID_AUTO, proc_translated, CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED, 0, 0, sysctl_sysctl_translated, "I", "proc_translated");
4170 #endif /* CONFIG_ROSETTA */
4171
4172 STATIC int
4173 sysctl_sysctl_native(__unused struct sysctl_oid *oidp, void *arg1, int arg2,
4174 struct sysctl_req *req)
4175 {
4176 int error;
4177 cpu_type_t proc_cputype = 0;
4178 if ((error = fetch_process_cputype(req->p, (int *)arg1, arg2, &proc_cputype)) != 0) {
4179 return error;
4180 }
4181 int res = 1;
4182 if ((proc_cputype & ~CPU_ARCH_MASK) != (cpu_type() & ~CPU_ARCH_MASK)) {
4183 res = 0;
4184 }
4185 return SYSCTL_OUT(req, &res, sizeof(res));
4186 }
4187 SYSCTL_PROC(_sysctl, OID_AUTO, proc_native, CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED, 0, 0, sysctl_sysctl_native, "I", "proc_native");
4188
4189 STATIC int
4190 sysctl_sysctl_cputype(__unused struct sysctl_oid *oidp, void *arg1, int arg2,
4191 struct sysctl_req *req)
4192 {
4193 int error;
4194 cpu_type_t proc_cputype = 0;
4195 if ((error = fetch_process_cputype(req->p, (int *)arg1, arg2, &proc_cputype)) != 0) {
4196 return error;
4197 }
4198 return SYSCTL_OUT(req, &proc_cputype, sizeof(proc_cputype));
4199 }
4200 SYSCTL_PROC(_sysctl, OID_AUTO, proc_cputype, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED, 0, 0, sysctl_sysctl_cputype, "I", "proc_cputype");
4201
4202 STATIC int
4203 sysctl_safeboot
4204 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4205 {
4206 return sysctl_io_number(req, boothowto & RB_SAFEBOOT ? 1 : 0, sizeof(int), NULL, NULL);
4207 }
4208
4209 SYSCTL_PROC(_kern, KERN_SAFEBOOT, safeboot,
4210 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
4211 0, 0, sysctl_safeboot, "I", "");
4212
4213 STATIC int
4214 sysctl_singleuser
4215 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4216 {
4217 return sysctl_io_number(req, boothowto & RB_SINGLE ? 1 : 0, sizeof(int), NULL, NULL);
4218 }
4219
4220 SYSCTL_PROC(_kern, OID_AUTO, singleuser,
4221 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
4222 0, 0, sysctl_singleuser, "I", "");
4223
4224 STATIC int
4225 sysctl_minimalboot
4226 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4227 {
4228 return sysctl_io_number(req, minimalboot, sizeof(int), NULL, NULL);
4229 }
4230
4231 SYSCTL_PROC(_kern, OID_AUTO, minimalboot,
4232 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
4233 0, 0, sysctl_minimalboot, "I", "");
4234
4235 /*
4236 * Controls for debugging affinity sets - see osfmk/kern/affinity.c
4237 */
4238 extern boolean_t affinity_sets_enabled;
4239 extern int affinity_sets_mapping;
4240
4241 SYSCTL_INT(_kern, OID_AUTO, affinity_sets_enabled,
4242 CTLFLAG_RW | CTLFLAG_LOCKED, (int *) &affinity_sets_enabled, 0, "hinting enabled");
4243 SYSCTL_INT(_kern, OID_AUTO, affinity_sets_mapping,
4244 CTLFLAG_RW | CTLFLAG_LOCKED, &affinity_sets_mapping, 0, "mapping policy");
4245
4246 /*
4247 * Boolean indicating if KASLR is active.
4248 */
4249 STATIC int
4250 sysctl_slide
4251 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4252 {
4253 uint32_t slide;
4254
4255 slide = vm_kernel_slide ? 1 : 0;
4256
4257 return sysctl_io_number( req, slide, sizeof(int), NULL, NULL);
4258 }
4259
4260 SYSCTL_PROC(_kern, OID_AUTO, slide,
4261 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
4262 0, 0, sysctl_slide, "I", "");
4263
4264 #if DEBUG || DEVELOPMENT
4265 #if defined(__arm64__)
4266 extern vm_offset_t segTEXTEXECB;
4267
4268 static int
4269 sysctl_kernel_text_exec_base_slide SYSCTL_HANDLER_ARGS
4270 {
4271 #pragma unused(arg1, arg2, oidp)
4272 unsigned long slide = 0;
4273 kc_format_t kc_format;
4274
4275 PE_get_primary_kc_format(&kc_format);
4276
4277 if (kc_format == KCFormatFileset) {
4278 void *kch = PE_get_kc_header(KCKindPrimary);
4279 slide = (unsigned long)segTEXTEXECB - (unsigned long)kch + vm_kernel_slide;
4280 }
4281 return SYSCTL_OUT(req, &slide, sizeof(slide));
4282 }
4283
4284 SYSCTL_QUAD(_kern, OID_AUTO, kernel_slide, CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED, &vm_kernel_slide, "");
4285 SYSCTL_QUAD(_kern, OID_AUTO, kernel_text_exec_base, CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED, &segTEXTEXECB, "");
4286 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", "");
4287 #endif /* defined(__arm64__) */
4288
4289 /* User address of the PFZ */
4290 extern user32_addr_t commpage_text32_location;
4291 extern user64_addr_t commpage_text64_location;
4292
4293 STATIC int
4294 sysctl_pfz_start SYSCTL_HANDLER_ARGS
4295 {
4296 #pragma unused(oidp, arg1, arg2)
4297
4298 #ifdef __LP64__
4299 return sysctl_io_number(req, commpage_text64_location, sizeof(user64_addr_t), NULL, NULL);
4300 #else
4301 return sysctl_io_number(req, commpage_text32_location, sizeof(user32_addr_t), NULL, NULL);
4302 #endif
4303 }
4304
4305 SYSCTL_PROC(_kern, OID_AUTO, pfz,
4306 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_MASKED,
4307 0, 0, sysctl_pfz_start, "I", "");
4308 #endif
4309
4310
4311 /*
4312 * Limit on total memory users can wire.
4313 *
4314 * vm_global_user_wire_limit - system wide limit on wired memory from all processes combined.
4315 *
4316 * vm_per_task_user_wire_limit - per address space limit on wired memory. This puts a cap on the process's rlimit value.
4317 *
4318 * These values are initialized to reasonable defaults at boot time based on the available physical memory in
4319 * kmem_init().
4320 *
4321 * All values are in bytes.
4322 */
4323
4324 vm_map_size_t vm_global_user_wire_limit;
4325 vm_map_size_t vm_per_task_user_wire_limit;
4326 extern uint64_t max_mem_actual, max_mem;
4327
4328 uint64_t vm_add_wire_count_over_global_limit;
4329 uint64_t vm_add_wire_count_over_user_limit;
4330 /*
4331 * We used to have a global in the kernel called vm_global_no_user_wire_limit which was the inverse
4332 * of vm_global_user_wire_limit. But maintaining both of those is silly, and vm_global_user_wire_limit is the
4333 * real limit.
4334 * This function is for backwards compatibility with userspace
4335 * since we exposed the old global via a sysctl.
4336 */
4337 STATIC int
4338 sysctl_global_no_user_wire_amount(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4339 {
4340 vm_map_size_t old_value;
4341 vm_map_size_t new_value;
4342 int changed;
4343 int error;
4344 uint64_t config_memsize = max_mem;
4345 #if defined(XNU_TARGET_OS_OSX)
4346 config_memsize = max_mem_actual;
4347 #endif /* defined(XNU_TARGET_OS_OSX) */
4348
4349 old_value = (vm_map_size_t)(config_memsize - vm_global_user_wire_limit);
4350 error = sysctl_io_number(req, old_value, sizeof(vm_map_size_t), &new_value, &changed);
4351 if (changed) {
4352 if ((uint64_t)new_value > config_memsize) {
4353 error = EINVAL;
4354 } else {
4355 vm_global_user_wire_limit = (vm_map_size_t)(config_memsize - new_value);
4356 }
4357 }
4358 return error;
4359 }
4360 /*
4361 * There needs to be a more automatic/elegant way to do this
4362 */
4363 SYSCTL_QUAD(_vm, OID_AUTO, global_user_wire_limit, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_global_user_wire_limit, "");
4364 SYSCTL_QUAD(_vm, OID_AUTO, user_wire_limit, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_per_task_user_wire_limit, "");
4365 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", "");
4366
4367 /*
4368 * Relaxed atomic RW of a 64bit value via sysctl.
4369 */
4370 STATIC int
4371 sysctl_r_64bit_atomic(uint64_t *ptr, struct sysctl_req *req)
4372 {
4373 uint64_t old_value;
4374 uint64_t new_value;
4375 int error;
4376
4377 old_value = os_atomic_load_wide(ptr, relaxed);
4378 error = sysctl_io_number(req, old_value, sizeof(vm_map_size_t), &new_value, NULL);
4379 return error;
4380 }
4381 STATIC int
4382 sysctl_add_wire_count_over_global_limit(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4383 {
4384 return sysctl_r_64bit_atomic(&vm_add_wire_count_over_global_limit, req);
4385 }
4386 STATIC int
4387 sysctl_add_wire_count_over_user_limit(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4388 {
4389 return sysctl_r_64bit_atomic(&vm_add_wire_count_over_user_limit, req);
4390 }
4391
4392 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", "");
4393 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", "");
4394
4395 #if DEVELOPMENT || DEBUG
4396 /* These sysctls are used to test the wired limit. */
4397 extern unsigned int vm_page_wire_count;
4398 extern uint32_t vm_lopage_free_count;
4399 extern unsigned int vm_page_stolen_count;
4400 SYSCTL_INT(_vm, OID_AUTO, page_wire_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_wire_count, 0, "");
4401 SYSCTL_INT(_vm, OID_AUTO, lopage_free_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_lopage_free_count, 0, "");
4402 SYSCTL_INT(_vm, OID_AUTO, page_stolen_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_stolen_count, 0, "");
4403
4404 /*
4405 * Setting the per task variable exclude_physfootprint_ledger to 1 will allow the calling task to exclude memory entries that are
4406 * tagged by VM_LEDGER_TAG_DEFAULT and flagged by VM_LEDGER_FLAG_EXCLUDE_FOOTPRINT_DEBUG from its phys_footprint ledger.
4407 */
4408
4409 STATIC int
4410 sysctl_rw_task_no_footprint_for_debug(struct sysctl_oid *oidp __unused, void *arg1 __unused, int arg2 __unused, struct sysctl_req *req)
4411 {
4412 int error;
4413 int value;
4414 proc_t p = current_proc();
4415
4416 if (req->newptr) {
4417 // Write request
4418 error = SYSCTL_IN(req, &value, sizeof(value));
4419 if (!error) {
4420 if (value == 1) {
4421 task_set_no_footprint_for_debug(proc_task(p), TRUE);
4422 } else if (value == 0) {
4423 task_set_no_footprint_for_debug(proc_task(p), FALSE);
4424 } else {
4425 error = EINVAL;
4426 }
4427 }
4428 } else {
4429 // Read request
4430 value = task_get_no_footprint_for_debug(proc_task(p));
4431 error = SYSCTL_OUT(req, &value, sizeof(value));
4432 }
4433 return error;
4434 }
4435
4436 SYSCTL_PROC(_vm, OID_AUTO, task_no_footprint_for_debug,
4437 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_ANYBODY,
4438 0, 0, &sysctl_rw_task_no_footprint_for_debug, "I", "Allow debug memory to be excluded from this task's memory footprint (debug only)");
4439
4440 #endif /* DEVELOPMENT || DEBUG */
4441
4442
4443 extern int vm_map_copy_overwrite_aligned_src_not_internal;
4444 extern int vm_map_copy_overwrite_aligned_src_not_symmetric;
4445 extern int vm_map_copy_overwrite_aligned_src_large;
4446 SYSCTL_INT(_vm, OID_AUTO, vm_copy_src_not_internal, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_map_copy_overwrite_aligned_src_not_internal, 0, "");
4447 SYSCTL_INT(_vm, OID_AUTO, vm_copy_src_not_symmetric, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_map_copy_overwrite_aligned_src_not_symmetric, 0, "");
4448 SYSCTL_INT(_vm, OID_AUTO, vm_copy_src_large, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_map_copy_overwrite_aligned_src_large, 0, "");
4449
4450
4451 extern uint32_t vm_page_external_count;
4452
4453 SYSCTL_INT(_vm, OID_AUTO, vm_page_external_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_external_count, 0, "");
4454
4455 SYSCTL_INT(_vm, OID_AUTO, vm_page_filecache_min, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_state.vm_page_filecache_min, 0, "");
4456 SYSCTL_INT(_vm, OID_AUTO, vm_page_xpmapped_min, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_state.vm_page_xpmapped_min, 0, "");
4457
4458 #if DEVELOPMENT || DEBUG
4459 SYSCTL_INT(_vm, OID_AUTO, vm_page_filecache_min_divisor, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_state.vm_page_filecache_min_divisor, 0, "");
4460 SYSCTL_INT(_vm, OID_AUTO, vm_page_xpmapped_min_divisor, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_state.vm_page_xpmapped_min_divisor, 0, "");
4461 extern boolean_t vps_yield_for_pgqlockwaiters;
4462 SYSCTL_INT(_vm, OID_AUTO, vm_pageoutscan_yields_for_pageQlockwaiters, CTLFLAG_RW | CTLFLAG_LOCKED, &vps_yield_for_pgqlockwaiters, 0, "");
4463 #endif
4464
4465 extern int vm_compressor_mode;
4466 extern int vm_compressor_is_active;
4467 extern int vm_compressor_available;
4468 extern uint32_t c_seg_bufsize;
4469 extern uint32_t c_seg_allocsize;
4470 extern int c_seg_fixed_array_len;
4471 extern uint32_t c_segments_limit;
4472 extern uint32_t c_segment_pages_compressed_limit;
4473 extern uint64_t compressor_pool_size;
4474 extern uint32_t compressor_pool_multiplier;
4475 extern uint32_t vm_ripe_target_age;
4476 extern uint32_t swapout_target_age;
4477 extern _Atomic uint64_t compressor_bytes_used;
4478 extern _Atomic uint64_t c_segment_input_bytes;
4479 extern _Atomic uint64_t c_segment_compressed_bytes;
4480 extern uint32_t c_segment_pages_compressed;
4481 extern uint32_t compressor_eval_period_in_msecs;
4482 extern uint32_t compressor_sample_min_in_msecs;
4483 extern uint32_t compressor_sample_max_in_msecs;
4484 extern uint32_t compressor_thrashing_threshold_per_10msecs;
4485 extern uint32_t compressor_thrashing_min_per_10msecs;
4486 extern uint32_t vm_compressor_time_thread;
4487 extern uint32_t c_segment_svp_in_hash;
4488 extern uint32_t c_segment_svp_hash_succeeded;
4489 extern uint32_t c_segment_svp_hash_failed;
4490
4491 #if DEVELOPMENT || DEBUG
4492 extern uint32_t vm_compressor_minorcompact_threshold_divisor;
4493 extern uint32_t vm_compressor_majorcompact_threshold_divisor;
4494 extern uint32_t vm_compressor_unthrottle_threshold_divisor;
4495 extern uint32_t vm_compressor_catchup_threshold_divisor;
4496
4497 extern uint32_t vm_compressor_minorcompact_threshold_divisor_overridden;
4498 extern uint32_t vm_compressor_majorcompact_threshold_divisor_overridden;
4499 extern uint32_t vm_compressor_unthrottle_threshold_divisor_overridden;
4500 extern uint32_t vm_compressor_catchup_threshold_divisor_overridden;
4501
4502 extern vmct_stats_t vmct_stats;
4503
4504
4505 STATIC int
4506 sysctl_minorcompact_threshold_divisor(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4507 {
4508 int new_value, changed;
4509 int error = sysctl_io_number(req, vm_compressor_minorcompact_threshold_divisor, sizeof(int), &new_value, &changed);
4510
4511 if (changed) {
4512 vm_compressor_minorcompact_threshold_divisor = new_value;
4513 vm_compressor_minorcompact_threshold_divisor_overridden = 1;
4514 }
4515 return error;
4516 }
4517
4518 SYSCTL_PROC(_vm, OID_AUTO, compressor_minorcompact_threshold_divisor,
4519 CTLTYPE_INT | CTLFLAG_LOCKED | CTLFLAG_RW,
4520 0, 0, sysctl_minorcompact_threshold_divisor, "I", "");
4521
4522
4523 STATIC int
4524 sysctl_majorcompact_threshold_divisor(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4525 {
4526 int new_value, changed;
4527 int error = sysctl_io_number(req, vm_compressor_majorcompact_threshold_divisor, sizeof(int), &new_value, &changed);
4528
4529 if (changed) {
4530 vm_compressor_majorcompact_threshold_divisor = new_value;
4531 vm_compressor_majorcompact_threshold_divisor_overridden = 1;
4532 }
4533 return error;
4534 }
4535
4536 SYSCTL_PROC(_vm, OID_AUTO, compressor_majorcompact_threshold_divisor,
4537 CTLTYPE_INT | CTLFLAG_LOCKED | CTLFLAG_RW,
4538 0, 0, sysctl_majorcompact_threshold_divisor, "I", "");
4539
4540
4541 STATIC int
4542 sysctl_unthrottle_threshold_divisor(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4543 {
4544 int new_value, changed;
4545 int error = sysctl_io_number(req, vm_compressor_unthrottle_threshold_divisor, sizeof(int), &new_value, &changed);
4546
4547 if (changed) {
4548 vm_compressor_unthrottle_threshold_divisor = new_value;
4549 vm_compressor_unthrottle_threshold_divisor_overridden = 1;
4550 }
4551 return error;
4552 }
4553
4554 SYSCTL_PROC(_vm, OID_AUTO, compressor_unthrottle_threshold_divisor,
4555 CTLTYPE_INT | CTLFLAG_LOCKED | CTLFLAG_RW,
4556 0, 0, sysctl_unthrottle_threshold_divisor, "I", "");
4557
4558
4559 STATIC int
4560 sysctl_catchup_threshold_divisor(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4561 {
4562 int new_value, changed;
4563 int error = sysctl_io_number(req, vm_compressor_catchup_threshold_divisor, sizeof(int), &new_value, &changed);
4564
4565 if (changed) {
4566 vm_compressor_catchup_threshold_divisor = new_value;
4567 vm_compressor_catchup_threshold_divisor_overridden = 1;
4568 }
4569 return error;
4570 }
4571
4572 SYSCTL_PROC(_vm, OID_AUTO, compressor_catchup_threshold_divisor,
4573 CTLTYPE_INT | CTLFLAG_LOCKED | CTLFLAG_RW,
4574 0, 0, sysctl_catchup_threshold_divisor, "I", "");
4575 #endif
4576
4577
4578 SYSCTL_QUAD(_vm, OID_AUTO, compressor_input_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, ((uint64_t *)&c_segment_input_bytes), "");
4579 SYSCTL_QUAD(_vm, OID_AUTO, compressor_compressed_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, ((uint64_t *)&c_segment_compressed_bytes), "");
4580 SYSCTL_QUAD(_vm, OID_AUTO, compressor_bytes_used, CTLFLAG_RD | CTLFLAG_LOCKED, ((uint64_t *)&compressor_bytes_used), "");
4581
4582 SYSCTL_INT(_vm, OID_AUTO, compressor_mode, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_compressor_mode, 0, "");
4583 SYSCTL_INT(_vm, OID_AUTO, compressor_is_active, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_compressor_is_active, 0, "");
4584 SYSCTL_INT(_vm, OID_AUTO, compressor_swapout_target_age, CTLFLAG_RD | CTLFLAG_LOCKED, &swapout_target_age, 0, "");
4585 SYSCTL_INT(_vm, OID_AUTO, compressor_available, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_compressor_available, 0, "");
4586 SYSCTL_INT(_vm, OID_AUTO, compressor_segment_buffer_size, CTLFLAG_RD | CTLFLAG_LOCKED, &c_seg_bufsize, 0, "");
4587 SYSCTL_QUAD(_vm, OID_AUTO, compressor_pool_size, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_pool_size, "");
4588 SYSCTL_UINT(_vm, OID_AUTO, compressor_pool_multiplier, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_pool_multiplier, 0, "");
4589 SYSCTL_INT(_vm, OID_AUTO, compressor_segment_slots_fixed_array_len, CTLFLAG_RD | CTLFLAG_LOCKED, &c_seg_fixed_array_len, 0, "");
4590 SYSCTL_UINT(_vm, OID_AUTO, compressor_segment_limit, CTLFLAG_RD | CTLFLAG_LOCKED, &c_segments_limit, 0, "");
4591 SYSCTL_UINT(_vm, OID_AUTO, compressor_segment_pages_compressed_limit, CTLFLAG_RD | CTLFLAG_LOCKED, &c_segment_pages_compressed_limit, 0, "");
4592 SYSCTL_UINT(_vm, OID_AUTO, compressor_segment_alloc_size, CTLFLAG_RD | CTLFLAG_LOCKED, &c_seg_allocsize, 0, "");
4593 SYSCTL_UINT(_vm, OID_AUTO, compressor_segment_pages_compressed, CTLFLAG_RD | CTLFLAG_LOCKED, &c_segment_pages_compressed, 0, "");
4594 SYSCTL_UINT(_vm, OID_AUTO, compressor_segment_svp_in_hash, CTLFLAG_RD | CTLFLAG_LOCKED, &c_segment_svp_in_hash, 0, "");
4595 SYSCTL_UINT(_vm, OID_AUTO, compressor_segment_svp_hash_succeeded, CTLFLAG_RD | CTLFLAG_LOCKED, &c_segment_svp_hash_succeeded, 0, "");
4596 SYSCTL_UINT(_vm, OID_AUTO, compressor_segment_svp_hash_failed, CTLFLAG_RD | CTLFLAG_LOCKED, &c_segment_svp_hash_failed, 0, "");
4597
4598 #if CONFIG_TRACK_UNMODIFIED_ANON_PAGES
4599 extern uint64_t compressor_ro_uncompressed;
4600 extern uint64_t compressor_ro_uncompressed_total_returned;
4601 extern uint64_t compressor_ro_uncompressed_skip_returned;
4602 extern uint64_t compressor_ro_uncompressed_get;
4603 extern uint64_t compressor_ro_uncompressed_put;
4604 extern uint64_t compressor_ro_uncompressed_swap_usage;
4605
4606 SYSCTL_QUAD(_vm, OID_AUTO, compressor_ro_uncompressed_total_returned, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_ro_uncompressed_total_returned, "");
4607 SYSCTL_QUAD(_vm, OID_AUTO, compressor_ro_uncompressed_writes_saved, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_ro_uncompressed_skip_returned, "");
4608 SYSCTL_QUAD(_vm, OID_AUTO, compressor_ro_uncompressed_candidates, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_ro_uncompressed, "");
4609 SYSCTL_QUAD(_vm, OID_AUTO, compressor_ro_uncompressed_rereads, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_ro_uncompressed_get, "");
4610 SYSCTL_QUAD(_vm, OID_AUTO, compressor_ro_uncompressed_swap_pages_on_disk, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_ro_uncompressed_swap_usage, "");
4611 #endif /* CONFIG_TRACK_UNMODIFIED_ANON_PAGES */
4612
4613 extern int min_csegs_per_major_compaction;
4614 SYSCTL_INT(_vm, OID_AUTO, compressor_min_csegs_per_major_compaction, CTLFLAG_RW | CTLFLAG_LOCKED, &min_csegs_per_major_compaction, 0, "");
4615
4616 SYSCTL_INT(_vm, OID_AUTO, vm_ripe_target_age_in_secs, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_ripe_target_age, 0, "");
4617
4618 SYSCTL_INT(_vm, OID_AUTO, compressor_eval_period_in_msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &compressor_eval_period_in_msecs, 0, "");
4619 SYSCTL_INT(_vm, OID_AUTO, compressor_sample_min_in_msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &compressor_sample_min_in_msecs, 0, "");
4620 SYSCTL_INT(_vm, OID_AUTO, compressor_sample_max_in_msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &compressor_sample_max_in_msecs, 0, "");
4621 SYSCTL_INT(_vm, OID_AUTO, compressor_thrashing_threshold_per_10msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &compressor_thrashing_threshold_per_10msecs, 0, "");
4622 SYSCTL_INT(_vm, OID_AUTO, compressor_thrashing_min_per_10msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &compressor_thrashing_min_per_10msecs, 0, "");
4623
4624 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapouts_under_30s, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.unripe_under_30s, "");
4625 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapouts_under_60s, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.unripe_under_60s, "");
4626 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapouts_under_300s, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.unripe_under_300s, "");
4627 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_reclaim_swapins, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.reclaim_swapins, "");
4628 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_defrag_swapins, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.defrag_swapins, "");
4629 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_swapout_threshold_exceeded, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.compressor_swap_threshold_exceeded, "");
4630 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_swapout_fileq_throttled, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.external_q_throttled, "");
4631 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_swapout_free_count_low, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.free_count_below_reserve, "");
4632 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_swapout_thrashing_detected, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.thrashing_detected, "");
4633 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_swapout_fragmentation_detected, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.fragmentation_detected, "");
4634
4635 SYSCTL_STRING(_vm, OID_AUTO, swapfileprefix, CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED, swapfilename, sizeof(swapfilename) - SWAPFILENAME_INDEX_LEN, "");
4636
4637 SYSCTL_INT(_vm, OID_AUTO, compressor_timing_enabled, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_compressor_time_thread, 0, "");
4638
4639 #if DEVELOPMENT || DEBUG
4640 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_runtime0, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_runtimes[0], "");
4641 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_runtime1, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_runtimes[1], "");
4642
4643 SYSCTL_QUAD(_vm, OID_AUTO, compressor_threads_total_execution_time, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_cthreads_total, "");
4644
4645 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_pages0, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_pages[0], "");
4646 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_pages1, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_pages[1], "");
4647
4648 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_iterations0, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_iterations[0], "");
4649 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_iterations1, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_iterations[1], "");
4650
4651 SYSCTL_INT(_vm, OID_AUTO, compressor_thread_minpages0, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_minpages[0], 0, "");
4652 SYSCTL_INT(_vm, OID_AUTO, compressor_thread_minpages1, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_minpages[1], 0, "");
4653
4654 SYSCTL_INT(_vm, OID_AUTO, compressor_thread_maxpages0, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_maxpages[0], 0, "");
4655 SYSCTL_INT(_vm, OID_AUTO, compressor_thread_maxpages1, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_maxpages[1], 0, "");
4656
4657 int vm_compressor_injected_error_count;
4658
4659 SYSCTL_INT(_vm, OID_AUTO, compressor_injected_error_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_compressor_injected_error_count, 0, "");
4660
4661 static int
4662 sysctl_compressor_inject_error(__unused struct sysctl_oid *oidp,
4663 __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4664 {
4665 int result;
4666 vm_address_t va = 0;
4667 int changed;
4668
4669 result = sysctl_io_number(req, va, sizeof(va), &va, &changed);
4670 if (result == 0 && changed) {
4671 result = vm_map_inject_error(current_map(), va);
4672 if (result == 0) {
4673 /*
4674 * Count the number of errors injected successfully to detect
4675 * situations where corruption was caused by improper use of this
4676 * sysctl.
4677 */
4678 os_atomic_inc(&vm_compressor_injected_error_count, relaxed);
4679 }
4680 }
4681 return result;
4682 }
4683
4684 SYSCTL_PROC(_vm, OID_AUTO, compressor_inject_error, CTLTYPE_QUAD | CTLFLAG_LOCKED | CTLFLAG_RW,
4685 0, 0, sysctl_compressor_inject_error, "Q", "flips a bit in a compressed page for the current task");
4686
4687 /*
4688 * Opt a process in/out of self donation mode.
4689 */
4690 static int
4691 sysctl_vm_pid_toggle_selfdonate_pages SYSCTL_HANDLER_ARGS
4692 {
4693 #pragma unused(arg1, arg2)
4694 int error, pid = 0;
4695 proc_t p;
4696
4697 error = sysctl_handle_int(oidp, &pid, 0, req);
4698 if (error || !req->newptr) {
4699 return error;
4700 }
4701
4702 p = proc_find(pid);
4703 if (p != NULL) {
4704 (void) vm_toggle_task_selfdonate_pages(proc_task(p));
4705 proc_rele(p);
4706 return error;
4707 } else {
4708 printf("sysctl_vm_pid_selfdonate_pages: Invalid process\n");
4709 }
4710
4711 return EINVAL;
4712 }
4713 SYSCTL_PROC(_vm, OID_AUTO, pid_toggle_selfdonate_pages, CTLTYPE_INT | CTLFLAG_WR | CTLFLAG_LOCKED | CTLFLAG_MASKED,
4714 0, 0, &sysctl_vm_pid_toggle_selfdonate_pages, "I", "");
4715 #endif /* DEVELOPMENT || DEBUG */
4716 extern uint32_t vm_page_donate_mode;
4717 extern uint32_t vm_page_donate_target_high, vm_page_donate_target_low;
4718 SYSCTL_INT(_vm, OID_AUTO, vm_page_donate_mode, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_page_donate_mode, 0, "");
4719 SYSCTL_INT(_vm, OID_AUTO, vm_page_donate_target_high, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_page_donate_target_high, 0, "");
4720 SYSCTL_INT(_vm, OID_AUTO, vm_page_donate_target_low, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_page_donate_target_low, 0, "");
4721
4722 SYSCTL_QUAD(_vm, OID_AUTO, lz4_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_compressions, "");
4723 SYSCTL_QUAD(_vm, OID_AUTO, lz4_compression_failures, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_compression_failures, "");
4724 SYSCTL_QUAD(_vm, OID_AUTO, lz4_compressed_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_compressed_bytes, "");
4725 SYSCTL_QUAD(_vm, OID_AUTO, lz4_wk_compression_delta, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_wk_compression_delta, "");
4726 SYSCTL_QUAD(_vm, OID_AUTO, lz4_wk_compression_negative_delta, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_wk_compression_negative_delta, "");
4727
4728 SYSCTL_QUAD(_vm, OID_AUTO, lz4_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_decompressions, "");
4729 SYSCTL_QUAD(_vm, OID_AUTO, lz4_decompressed_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_decompressed_bytes, "");
4730
4731 SYSCTL_QUAD(_vm, OID_AUTO, uc_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.uc_decompressions, "");
4732
4733 SYSCTL_QUAD(_vm, OID_AUTO, wk_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_compressions, "");
4734
4735 SYSCTL_QUAD(_vm, OID_AUTO, wk_catime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_cabstime, "");
4736
4737 SYSCTL_QUAD(_vm, OID_AUTO, wkh_catime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wkh_cabstime, "");
4738 SYSCTL_QUAD(_vm, OID_AUTO, wkh_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wkh_compressions, "");
4739
4740 SYSCTL_QUAD(_vm, OID_AUTO, wks_catime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_cabstime, "");
4741 SYSCTL_QUAD(_vm, OID_AUTO, wks_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_compressions, "");
4742
4743 SYSCTL_QUAD(_vm, OID_AUTO, wk_compressions_exclusive, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_compressions_exclusive, "");
4744 SYSCTL_QUAD(_vm, OID_AUTO, wk_sv_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_sv_compressions, "");
4745 SYSCTL_QUAD(_vm, OID_AUTO, wk_mzv_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_mzv_compressions, "");
4746 SYSCTL_QUAD(_vm, OID_AUTO, wk_compression_failures, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_compression_failures, "");
4747 SYSCTL_QUAD(_vm, OID_AUTO, wk_compressed_bytes_exclusive, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_compressed_bytes_exclusive, "");
4748 SYSCTL_QUAD(_vm, OID_AUTO, wk_compressed_bytes_total, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_compressed_bytes_total, "");
4749
4750 SYSCTL_QUAD(_vm, OID_AUTO, wks_compressed_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_compressed_bytes, "");
4751 SYSCTL_QUAD(_vm, OID_AUTO, wks_compression_failures, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_compression_failures, "");
4752 SYSCTL_QUAD(_vm, OID_AUTO, wks_sv_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_sv_compressions, "");
4753
4754
4755 SYSCTL_QUAD(_vm, OID_AUTO, wk_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_decompressions, "");
4756
4757 SYSCTL_QUAD(_vm, OID_AUTO, wk_datime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_dabstime, "");
4758
4759 SYSCTL_QUAD(_vm, OID_AUTO, wkh_datime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wkh_dabstime, "");
4760 SYSCTL_QUAD(_vm, OID_AUTO, wkh_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wkh_decompressions, "");
4761
4762 SYSCTL_QUAD(_vm, OID_AUTO, wks_datime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_dabstime, "");
4763 SYSCTL_QUAD(_vm, OID_AUTO, wks_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_decompressions, "");
4764
4765 SYSCTL_QUAD(_vm, OID_AUTO, wk_decompressed_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_decompressed_bytes, "");
4766 SYSCTL_QUAD(_vm, OID_AUTO, wk_sv_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_sv_decompressions, "");
4767
4768 SYSCTL_INT(_vm, OID_AUTO, lz4_threshold, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_threshold, 0, "");
4769 SYSCTL_INT(_vm, OID_AUTO, wkdm_reeval_threshold, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.wkdm_reeval_threshold, 0, "");
4770 SYSCTL_INT(_vm, OID_AUTO, lz4_max_failure_skips, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_max_failure_skips, 0, "");
4771 SYSCTL_INT(_vm, OID_AUTO, lz4_max_failure_run_length, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_max_failure_run_length, 0, "");
4772 SYSCTL_INT(_vm, OID_AUTO, lz4_max_preselects, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_max_preselects, 0, "");
4773 SYSCTL_INT(_vm, OID_AUTO, lz4_run_preselection_threshold, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_run_preselection_threshold, 0, "");
4774 SYSCTL_INT(_vm, OID_AUTO, lz4_run_continue_bytes, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_run_continue_bytes, 0, "");
4775 SYSCTL_INT(_vm, OID_AUTO, lz4_profitable_bytes, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_profitable_bytes, 0, "");
4776 #if DEVELOPMENT || DEBUG
4777 extern int vm_compressor_current_codec;
4778 extern int vm_compressor_test_seg_wp;
4779 extern boolean_t vm_compressor_force_sw_wkdm;
4780 SYSCTL_INT(_vm, OID_AUTO, compressor_codec, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_compressor_current_codec, 0, "");
4781 SYSCTL_INT(_vm, OID_AUTO, compressor_test_wp, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_compressor_test_seg_wp, 0, "");
4782
4783 SYSCTL_INT(_vm, OID_AUTO, wksw_force, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_compressor_force_sw_wkdm, 0, "");
4784 extern int precompy, wkswhw;
4785
4786 SYSCTL_INT(_vm, OID_AUTO, precompy, CTLFLAG_RW | CTLFLAG_LOCKED, &precompy, 0, "");
4787 SYSCTL_INT(_vm, OID_AUTO, wkswhw, CTLFLAG_RW | CTLFLAG_LOCKED, &wkswhw, 0, "");
4788 extern unsigned int vm_ktrace_enabled;
4789 SYSCTL_INT(_vm, OID_AUTO, vm_ktrace, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_ktrace_enabled, 0, "");
4790 #endif /* DEVELOPMENT || DEBUG */
4791
4792 #if CONFIG_PHANTOM_CACHE
4793 extern uint32_t phantom_cache_thrashing_threshold;
4794 extern uint32_t phantom_cache_eval_period_in_msecs;
4795 extern uint32_t phantom_cache_thrashing_threshold_ssd;
4796
4797
4798 SYSCTL_INT(_vm, OID_AUTO, phantom_cache_eval_period_in_msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &phantom_cache_eval_period_in_msecs, 0, "");
4799 SYSCTL_INT(_vm, OID_AUTO, phantom_cache_thrashing_threshold, CTLFLAG_RW | CTLFLAG_LOCKED, &phantom_cache_thrashing_threshold, 0, "");
4800 SYSCTL_INT(_vm, OID_AUTO, phantom_cache_thrashing_threshold_ssd, CTLFLAG_RW | CTLFLAG_LOCKED, &phantom_cache_thrashing_threshold_ssd, 0, "");
4801 #endif
4802
4803 #if defined(__LP64__)
4804 extern uint32_t vm_page_background_count;
4805 extern uint32_t vm_page_background_target;
4806 extern uint32_t vm_page_background_internal_count;
4807 extern uint32_t vm_page_background_external_count;
4808 extern uint32_t vm_page_background_mode;
4809 extern uint32_t vm_page_background_exclude_external;
4810 extern uint64_t vm_page_background_promoted_count;
4811 extern uint64_t vm_pageout_rejected_bq_internal;
4812 extern uint64_t vm_pageout_rejected_bq_external;
4813
4814 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_mode, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_page_background_mode, 0, "");
4815 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_exclude_external, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_page_background_exclude_external, 0, "");
4816 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_target, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_page_background_target, 0, "");
4817 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_background_count, 0, "");
4818 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_internal_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_background_internal_count, 0, "");
4819 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_external_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_background_external_count, 0, "");
4820
4821 SYSCTL_QUAD(_vm, OID_AUTO, vm_page_background_promoted_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_background_promoted_count, "");
4822 SYSCTL_QUAD(_vm, OID_AUTO, vm_pageout_considered_bq_internal, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_vminfo.vm_pageout_considered_bq_internal, "");
4823 SYSCTL_QUAD(_vm, OID_AUTO, vm_pageout_considered_bq_external, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_vminfo.vm_pageout_considered_bq_external, "");
4824 SYSCTL_QUAD(_vm, OID_AUTO, vm_pageout_rejected_bq_internal, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_rejected_bq_internal, "");
4825 SYSCTL_QUAD(_vm, OID_AUTO, vm_pageout_rejected_bq_external, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_rejected_bq_external, "");
4826
4827 #endif /* __LP64__ */
4828
4829 extern boolean_t vm_darkwake_mode;
4830
4831 STATIC int
4832 sysctl_toggle_darkwake_mode(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4833 {
4834 int new_value, changed;
4835 int error = sysctl_io_number(req, vm_darkwake_mode, sizeof(int), &new_value, &changed);
4836
4837 if (!error && changed) {
4838 if (new_value != 0 && new_value != 1) {
4839 printf("Error: Invalid value passed to darkwake sysctl. Acceptable: 0 or 1.\n");
4840 error = EINVAL;
4841 } else {
4842 vm_update_darkwake_mode((boolean_t) new_value);
4843 }
4844 }
4845
4846 return error;
4847 }
4848
4849 SYSCTL_PROC(_vm, OID_AUTO, darkwake_mode,
4850 CTLTYPE_INT | CTLFLAG_LOCKED | CTLFLAG_RW,
4851 0, 0, sysctl_toggle_darkwake_mode, "I", "");
4852
4853 #if (DEVELOPMENT || DEBUG)
4854
4855 SYSCTL_UINT(_vm, OID_AUTO, vm_page_creation_throttled_hard,
4856 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
4857 &vm_page_creation_throttled_hard, 0, "");
4858
4859 SYSCTL_UINT(_vm, OID_AUTO, vm_page_creation_throttled_soft,
4860 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
4861 &vm_page_creation_throttled_soft, 0, "");
4862
4863 extern uint32_t vm_pageout_memorystatus_fb_factor_nr;
4864 extern uint32_t vm_pageout_memorystatus_fb_factor_dr;
4865 SYSCTL_INT(_vm, OID_AUTO, vm_pageout_memorystatus_fb_factor_nr, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_memorystatus_fb_factor_nr, 0, "");
4866 SYSCTL_INT(_vm, OID_AUTO, vm_pageout_memorystatus_fb_factor_dr, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_memorystatus_fb_factor_dr, 0, "");
4867
4868 extern uint32_t vm_grab_anon_nops;
4869
4870 SYSCTL_INT(_vm, OID_AUTO, vm_grab_anon_overrides, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_debug.vm_grab_anon_overrides, 0, "");
4871 SYSCTL_INT(_vm, OID_AUTO, vm_grab_anon_nops, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_debug.vm_grab_anon_nops, 0, "");
4872 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, "");
4873
4874
4875 extern int vm_page_delayed_work_ctx_needed;
4876 SYSCTL_INT(_vm, OID_AUTO, vm_page_needed_delayed_work_ctx, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_delayed_work_ctx_needed, 0, "");
4877
4878
4879 /* log message counters for persistence mode */
4880 SCALABLE_COUNTER_DECLARE(oslog_p_total_msgcount);
4881 SCALABLE_COUNTER_DECLARE(oslog_p_metadata_saved_msgcount);
4882 SCALABLE_COUNTER_DECLARE(oslog_p_metadata_dropped_msgcount);
4883 SCALABLE_COUNTER_DECLARE(oslog_p_signpost_saved_msgcount);
4884 SCALABLE_COUNTER_DECLARE(oslog_p_signpost_dropped_msgcount);
4885 SCALABLE_COUNTER_DECLARE(oslog_p_error_count);
4886 SCALABLE_COUNTER_DECLARE(oslog_p_error_count);
4887 SCALABLE_COUNTER_DECLARE(oslog_p_saved_msgcount);
4888 SCALABLE_COUNTER_DECLARE(oslog_p_dropped_msgcount);
4889 SCALABLE_COUNTER_DECLARE(oslog_p_boot_dropped_msgcount);
4890 SCALABLE_COUNTER_DECLARE(oslog_p_coprocessor_total_msgcount);
4891 SCALABLE_COUNTER_DECLARE(oslog_p_coprocessor_dropped_msgcount);
4892 SCALABLE_COUNTER_DECLARE(oslog_p_unresolved_kc_msgcount);
4893 SCALABLE_COUNTER_DECLARE(oslog_p_fmt_invalid_msgcount);
4894 SCALABLE_COUNTER_DECLARE(oslog_p_fmt_max_args_msgcount);
4895 SCALABLE_COUNTER_DECLARE(oslog_p_truncated_msgcount);
4896
4897 SCALABLE_COUNTER_DECLARE(oslog_subsystem_count);
4898 SCALABLE_COUNTER_DECLARE(oslog_subsystem_found);
4899 SCALABLE_COUNTER_DECLARE(oslog_subsystem_dropped);
4900
4901 SCALABLE_COUNTER_DECLARE(log_queue_cnt_received);
4902 SCALABLE_COUNTER_DECLARE(log_queue_cnt_rejected_fh);
4903 SCALABLE_COUNTER_DECLARE(log_queue_cnt_sent);
4904 SCALABLE_COUNTER_DECLARE(log_queue_cnt_dropped_nomem);
4905 SCALABLE_COUNTER_DECLARE(log_queue_cnt_queued);
4906 SCALABLE_COUNTER_DECLARE(log_queue_cnt_dropped_off);
4907 SCALABLE_COUNTER_DECLARE(log_queue_cnt_mem_active);
4908 SCALABLE_COUNTER_DECLARE(log_queue_cnt_mem_allocated);
4909 SCALABLE_COUNTER_DECLARE(log_queue_cnt_mem_released);
4910 SCALABLE_COUNTER_DECLARE(log_queue_cnt_mem_failed);
4911
4912 /* log message counters for streaming mode */
4913 SCALABLE_COUNTER_DECLARE(oslog_s_total_msgcount);
4914 SCALABLE_COUNTER_DECLARE(oslog_s_metadata_msgcount);
4915 SCALABLE_COUNTER_DECLARE(oslog_s_error_count);
4916 SCALABLE_COUNTER_DECLARE(oslog_s_streamed_msgcount);
4917 SCALABLE_COUNTER_DECLARE(oslog_s_dropped_msgcount);
4918
4919 /* log message counters for msgbuf logging */
4920 SCALABLE_COUNTER_DECLARE(oslog_msgbuf_msgcount);
4921 SCALABLE_COUNTER_DECLARE(oslog_msgbuf_dropped_msgcount);
4922 extern uint32_t oslog_msgbuf_dropped_charcount;
4923
4924 #if CONFIG_EXCLAVES
4925 /* log message counters for exclaves logging */
4926 SCALABLE_COUNTER_DECLARE(oslog_e_log_count);
4927 SCALABLE_COUNTER_DECLARE(oslog_e_log_dropped_count);
4928 SCALABLE_COUNTER_DECLARE(oslog_e_metadata_count);
4929 SCALABLE_COUNTER_DECLARE(oslog_e_metadata_dropped_count);
4930 SCALABLE_COUNTER_DECLARE(oslog_e_signpost_count);
4931 SCALABLE_COUNTER_DECLARE(oslog_e_signpost_dropped_count);
4932 SCALABLE_COUNTER_DECLARE(oslog_e_query_count);
4933 SCALABLE_COUNTER_DECLARE(oslog_e_query_error_count);
4934 SCALABLE_COUNTER_DECLARE(oslog_e_trace_mode_set_count);
4935 SCALABLE_COUNTER_DECLARE(oslog_e_trace_mode_error_count);
4936 #endif // CONFIG_EXCLAVES
4937
4938 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_total_msgcount, oslog_p_total_msgcount, "");
4939 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_metadata_saved_msgcount, oslog_p_metadata_saved_msgcount, "");
4940 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_metadata_dropped_msgcount, oslog_p_metadata_dropped_msgcount, "");
4941 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_signpost_saved_msgcount, oslog_p_signpost_saved_msgcount, "");
4942 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_signpost_dropped_msgcount, oslog_p_signpost_dropped_msgcount, "");
4943 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_error_count, oslog_p_error_count, "");
4944 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_saved_msgcount, oslog_p_saved_msgcount, "");
4945 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_dropped_msgcount, oslog_p_dropped_msgcount, "");
4946 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_boot_dropped_msgcount, oslog_p_boot_dropped_msgcount, "");
4947 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_coprocessor_total_msgcount, oslog_p_coprocessor_total_msgcount, "");
4948 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_coprocessor_dropped_msgcount, oslog_p_coprocessor_dropped_msgcount, "");
4949 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_unresolved_kc_msgcount, oslog_p_unresolved_kc_msgcount, "");
4950
4951 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_fmt_invalid_msgcount, oslog_p_fmt_invalid_msgcount, "");
4952 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_fmt_max_args_msgcount, oslog_p_fmt_max_args_msgcount, "");
4953 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_truncated_msgcount, oslog_p_truncated_msgcount, "");
4954
4955 SYSCTL_SCALABLE_COUNTER(_debug, oslog_s_total_msgcount, oslog_s_total_msgcount, "Number of logs sent to streaming");
4956 SYSCTL_SCALABLE_COUNTER(_debug, oslog_s_metadata_msgcount, oslog_s_metadata_msgcount, "Number of metadata sent to streaming");
4957 SYSCTL_SCALABLE_COUNTER(_debug, oslog_s_error_count, oslog_s_error_count, "Number of invalid stream logs");
4958 SYSCTL_SCALABLE_COUNTER(_debug, oslog_s_streamed_msgcount, oslog_s_streamed_msgcount, "Number of streamed logs");
4959 SYSCTL_SCALABLE_COUNTER(_debug, oslog_s_dropped_msgcount, oslog_s_dropped_msgcount, "Number of logs dropped from stream");
4960
4961 SYSCTL_SCALABLE_COUNTER(_debug, oslog_msgbuf_msgcount, oslog_msgbuf_msgcount, "Number of dmesg log messages");
4962 SYSCTL_SCALABLE_COUNTER(_debug, oslog_msgbuf_dropped_msgcount, oslog_msgbuf_dropped_msgcount, "Number of dropped dmesg log messages");
4963 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");
4964
4965 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_received, log_queue_cnt_received, "Number of received logs");
4966 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_rejected_fh, log_queue_cnt_rejected_fh, "Number of logs initially rejected by FH");
4967 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_sent, log_queue_cnt_sent, "Number of logs successfully saved in FH");
4968 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_dropped_nomem, log_queue_cnt_dropped_nomem, "Number of logs dropped due to lack of queue memory");
4969 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_queued, log_queue_cnt_queued, "Current number of logs stored in log queues");
4970 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_dropped_off, log_queue_cnt_dropped_off, "Number of logs dropped due to disabled log queues");
4971 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_mem_allocated, log_queue_cnt_mem_allocated, "Number of memory allocations");
4972 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_mem_released, log_queue_cnt_mem_released, "Number of memory releases");
4973 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_mem_failed, log_queue_cnt_mem_failed, "Number of failed memory allocations");
4974
4975 SYSCTL_SCALABLE_COUNTER(_debug, oslog_subsystem_count, oslog_subsystem_count, "Number of registered log subsystems");
4976 SYSCTL_SCALABLE_COUNTER(_debug, oslog_subsystem_found, oslog_subsystem_found, "Number of sucessful log subsystem lookups");
4977 SYSCTL_SCALABLE_COUNTER(_debug, oslog_subsystem_dropped, oslog_subsystem_dropped, "Number of dropped log subsystem registrations");
4978
4979 #if CONFIG_EXCLAVES
4980 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_metadata_count, oslog_e_metadata_count,
4981 "Number of metadata messages retrieved from the exclaves log server");
4982 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_metadata_dropped_count, oslog_e_metadata_dropped_count,
4983 "Number of dropped metadata messages retrieved from the exclaves log server");
4984 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_log_count, oslog_e_log_count,
4985 "Number of logs retrieved from the exclaves log server");
4986 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_log_dropped_count, oslog_e_log_dropped_count,
4987 "Number of dropeed logs retrieved from the exclaves log server");
4988 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_signpost_count, oslog_e_signpost_count,
4989 "Number of signposts retrieved from the exclaves log server");
4990 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_signpost_dropped_count, oslog_e_signpost_dropped_count,
4991 "Number of dropped signposts retrieved from the exclaves log server");
4992 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_query_count, oslog_e_query_count,
4993 "Number of sucessful queries to the exclaves log server");
4994 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_query_error_count, oslog_e_query_error_count,
4995 "Number of failed queries to the exclaves log server");
4996 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_trace_mode_set_count, oslog_e_trace_mode_set_count,
4997 "Number of exclaves trace mode updates");
4998 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_trace_mode_error_count, oslog_e_trace_mode_error_count,
4999 "Number of failed exclaves trace mode updates");
5000 #endif // CONFIG_EXCLAVES
5001
5002 #endif /* DEVELOPMENT || DEBUG */
5003
5004 /*
5005 * Enable tracing of voucher contents
5006 */
5007 extern uint32_t ipc_voucher_trace_contents;
5008
5009 SYSCTL_INT(_kern, OID_AUTO, ipc_voucher_trace_contents,
5010 CTLFLAG_RW | CTLFLAG_LOCKED, &ipc_voucher_trace_contents, 0, "Enable tracing voucher contents");
5011
5012 /*
5013 * Kernel stack size and depth
5014 */
5015 SYSCTL_INT(_kern, OID_AUTO, stack_size,
5016 CTLFLAG_RD | CTLFLAG_LOCKED, (int *) &kernel_stack_size, 0, "Kernel stack size");
5017 SYSCTL_INT(_kern, OID_AUTO, stack_depth_max,
5018 CTLFLAG_RD | CTLFLAG_LOCKED, (int *) &kernel_stack_depth_max, 0, "Max kernel stack depth at interrupt or context switch");
5019
5020 extern unsigned int kern_feature_overrides;
5021 SYSCTL_INT(_kern, OID_AUTO, kern_feature_overrides,
5022 CTLFLAG_RD | CTLFLAG_LOCKED, &kern_feature_overrides, 0, "Kernel feature override mask");
5023
5024 /*
5025 * enable back trace for port allocations
5026 */
5027 extern int ipc_portbt;
5028
5029 SYSCTL_INT(_kern, OID_AUTO, ipc_portbt,
5030 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
5031 &ipc_portbt, 0, "");
5032
5033 /*
5034 * Scheduler sysctls
5035 */
5036
5037 SYSCTL_STRING(_kern, OID_AUTO, sched,
5038 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
5039 sched_string, sizeof(sched_string),
5040 "Timeshare scheduler implementation");
5041
5042 static int
5043 sysctl_cpu_quiescent_counter_interval SYSCTL_HANDLER_ARGS
5044 {
5045 #pragma unused(arg1, arg2)
5046
5047 uint32_t local_min_interval_us = smr_cpu_checkin_get_min_interval_us();
5048
5049 int error = sysctl_handle_int(oidp, &local_min_interval_us, 0, req);
5050 if (error || !req->newptr) {
5051 return error;
5052 }
5053
5054 smr_cpu_checkin_set_min_interval_us(local_min_interval_us);
5055
5056 return 0;
5057 }
5058
5059 SYSCTL_PROC(_kern, OID_AUTO, cpu_checkin_interval,
5060 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
5061 0, 0,
5062 sysctl_cpu_quiescent_counter_interval, "I",
5063 "Quiescent CPU checkin interval (microseconds)");
5064
5065 /*
5066 * Allow the precise user/kernel time sysctl to be set, but don't allow it to
5067 * affect anything. Some tools expect to be able to set this, even though
5068 * runtime configuration is no longer supported.
5069 */
5070
5071 static int
5072 sysctl_precise_user_kernel_time SYSCTL_HANDLER_ARGS
5073 {
5074 #if PRECISE_USER_KERNEL_TIME
5075 int dummy_set = 1;
5076 #else /* PRECISE_USER_KERNEL_TIME */
5077 int dummy_set = 0;
5078 #endif /* !PRECISE_USER_KERNEL_TIME */
5079 return sysctl_handle_int(oidp, &dummy_set, 0, req);
5080 }
5081
5082 SYSCTL_PROC(_kern, OID_AUTO, precise_user_kernel_time,
5083 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
5084 0, 0, sysctl_precise_user_kernel_time, "I",
5085 "Precise accounting of kernel vs. user time (deprecated)");
5086
5087 #if CONFIG_PERVASIVE_ENERGY && HAS_CPU_DPE_COUNTER
5088 __security_const_late static int pervasive_energy = 1;
5089 #else /* CONFIG_PERVASIVE_ENERGY && HAS_CPU_DPE_COUNTER */
5090 __security_const_late static int pervasive_energy = 0;
5091 #endif /* !CONFIG_PERVASIVE_ENERGY || !HAS_CPU_DPE_COUNTER */
5092
5093 SYSCTL_INT(_kern, OID_AUTO, pervasive_energy,
5094 CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED, &pervasive_energy, 0, "");
5095
5096 /* Parameters related to timer coalescing tuning, to be replaced
5097 * with a dedicated systemcall in the future.
5098 */
5099 /* Enable processing pending timers in the context of any other interrupt
5100 * Coalescing tuning parameters for various thread/task attributes */
5101 STATIC int
5102 sysctl_timer_user_us_kernel_abstime SYSCTL_HANDLER_ARGS
5103 {
5104 #pragma unused(oidp)
5105 int size = arg2; /* subcommand*/
5106 int error;
5107 int changed = 0;
5108 uint64_t old_value_ns;
5109 uint64_t new_value_ns;
5110 uint64_t value_abstime;
5111 if (size == sizeof(uint32_t)) {
5112 value_abstime = *((uint32_t *)arg1);
5113 } else if (size == sizeof(uint64_t)) {
5114 value_abstime = *((uint64_t *)arg1);
5115 } else {
5116 return ENOTSUP;
5117 }
5118
5119 absolutetime_to_nanoseconds(value_abstime, &old_value_ns);
5120 error = sysctl_io_number(req, old_value_ns, sizeof(old_value_ns), &new_value_ns, &changed);
5121 if ((error) || (!changed)) {
5122 return error;
5123 }
5124
5125 nanoseconds_to_absolutetime(new_value_ns, &value_abstime);
5126 if (size == sizeof(uint32_t)) {
5127 *((uint32_t *)arg1) = (uint32_t)value_abstime;
5128 } else {
5129 *((uint64_t *)arg1) = value_abstime;
5130 }
5131 return error;
5132 }
5133
5134 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_bg_scale,
5135 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5136 &tcoal_prio_params.timer_coalesce_bg_shift, 0, "");
5137 SYSCTL_PROC(_kern, OID_AUTO, timer_resort_threshold_ns,
5138 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5139 &tcoal_prio_params.timer_resort_threshold_abstime,
5140 sizeof(tcoal_prio_params.timer_resort_threshold_abstime),
5141 sysctl_timer_user_us_kernel_abstime,
5142 "Q", "");
5143 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_bg_ns_max,
5144 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5145 &tcoal_prio_params.timer_coalesce_bg_abstime_max,
5146 sizeof(tcoal_prio_params.timer_coalesce_bg_abstime_max),
5147 sysctl_timer_user_us_kernel_abstime,
5148 "Q", "");
5149
5150 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_kt_scale,
5151 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5152 &tcoal_prio_params.timer_coalesce_kt_shift, 0, "");
5153
5154 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_kt_ns_max,
5155 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5156 &tcoal_prio_params.timer_coalesce_kt_abstime_max,
5157 sizeof(tcoal_prio_params.timer_coalesce_kt_abstime_max),
5158 sysctl_timer_user_us_kernel_abstime,
5159 "Q", "");
5160
5161 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_fp_scale,
5162 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5163 &tcoal_prio_params.timer_coalesce_fp_shift, 0, "");
5164
5165 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_fp_ns_max,
5166 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5167 &tcoal_prio_params.timer_coalesce_fp_abstime_max,
5168 sizeof(tcoal_prio_params.timer_coalesce_fp_abstime_max),
5169 sysctl_timer_user_us_kernel_abstime,
5170 "Q", "");
5171
5172 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_ts_scale,
5173 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5174 &tcoal_prio_params.timer_coalesce_ts_shift, 0, "");
5175
5176 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_ts_ns_max,
5177 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5178 &tcoal_prio_params.timer_coalesce_ts_abstime_max,
5179 sizeof(tcoal_prio_params.timer_coalesce_ts_abstime_max),
5180 sysctl_timer_user_us_kernel_abstime,
5181 "Q", "");
5182
5183 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier0_scale,
5184 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5185 &tcoal_prio_params.latency_qos_scale[0], 0, "");
5186
5187 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier0_ns_max,
5188 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5189 &tcoal_prio_params.latency_qos_abstime_max[0],
5190 sizeof(tcoal_prio_params.latency_qos_abstime_max[0]),
5191 sysctl_timer_user_us_kernel_abstime,
5192 "Q", "");
5193
5194 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier1_scale,
5195 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5196 &tcoal_prio_params.latency_qos_scale[1], 0, "");
5197
5198 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier1_ns_max,
5199 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5200 &tcoal_prio_params.latency_qos_abstime_max[1],
5201 sizeof(tcoal_prio_params.latency_qos_abstime_max[1]),
5202 sysctl_timer_user_us_kernel_abstime,
5203 "Q", "");
5204
5205 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier2_scale,
5206 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5207 &tcoal_prio_params.latency_qos_scale[2], 0, "");
5208
5209 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier2_ns_max,
5210 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5211 &tcoal_prio_params.latency_qos_abstime_max[2],
5212 sizeof(tcoal_prio_params.latency_qos_abstime_max[2]),
5213 sysctl_timer_user_us_kernel_abstime,
5214 "Q", "");
5215
5216 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier3_scale,
5217 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5218 &tcoal_prio_params.latency_qos_scale[3], 0, "");
5219
5220 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier3_ns_max,
5221 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5222 &tcoal_prio_params.latency_qos_abstime_max[3],
5223 sizeof(tcoal_prio_params.latency_qos_abstime_max[3]),
5224 sysctl_timer_user_us_kernel_abstime,
5225 "Q", "");
5226
5227 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier4_scale,
5228 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5229 &tcoal_prio_params.latency_qos_scale[4], 0, "");
5230
5231 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier4_ns_max,
5232 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5233 &tcoal_prio_params.latency_qos_abstime_max[4],
5234 sizeof(tcoal_prio_params.latency_qos_abstime_max[4]),
5235 sysctl_timer_user_us_kernel_abstime,
5236 "Q", "");
5237
5238 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier5_scale,
5239 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5240 &tcoal_prio_params.latency_qos_scale[5], 0, "");
5241
5242 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier5_ns_max,
5243 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5244 &tcoal_prio_params.latency_qos_abstime_max[5],
5245 sizeof(tcoal_prio_params.latency_qos_abstime_max[5]),
5246 sysctl_timer_user_us_kernel_abstime,
5247 "Q", "");
5248
5249 /* Communicate the "user idle level" heuristic to the timer layer, and
5250 * potentially other layers in the future.
5251 */
5252
5253 static int
5254 timer_user_idle_level(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
5255 {
5256 int new_value = 0, old_value = 0, changed = 0, error;
5257
5258 old_value = timer_get_user_idle_level();
5259
5260 error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
5261
5262 if (error == 0 && changed) {
5263 if (timer_set_user_idle_level(new_value) != KERN_SUCCESS) {
5264 error = ERANGE;
5265 }
5266 }
5267
5268 return error;
5269 }
5270
5271 SYSCTL_PROC(_machdep, OID_AUTO, user_idle_level,
5272 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
5273 0, 0,
5274 timer_user_idle_level, "I", "User idle level heuristic, 0-128");
5275
5276 #if DEVELOPMENT || DEBUG
5277 /*
5278 * Basic console mode for games; used for development purposes only.
5279 * Final implementation for this feature (with possible removal of
5280 * sysctl) tracked via rdar://101215873.
5281 */
5282 static int console_mode = 0;
5283 SYSCTL_INT(_kern, OID_AUTO, console_mode,
5284 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_MASKED,
5285 &console_mode, 0, "Game Console Mode");
5286 #endif /* DEVELOPMENT || DEBUG */
5287
5288
5289 #if HYPERVISOR
5290 SYSCTL_INT(_kern, OID_AUTO, hv_support,
5291 CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
5292 &hv_support_available, 0, "");
5293
5294 SYSCTL_INT(_kern, OID_AUTO, hv_disable,
5295 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5296 &hv_disable, 0, "");
5297
5298 #endif /* HYPERVISOR */
5299
5300 #if DEVELOPMENT || DEBUG
5301 extern uint64_t driverkit_checkin_timed_out;
5302 SYSCTL_QUAD(_kern, OID_AUTO, driverkit_checkin_timed_out,
5303 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
5304 &driverkit_checkin_timed_out, "timestamp of dext checkin timeout");
5305 #endif
5306
5307 #if CONFIG_DARKBOOT
5308 STATIC int
5309 sysctl_darkboot SYSCTL_HANDLER_ARGS
5310 {
5311 int err = 0, value = 0;
5312 #pragma unused(oidp, arg1, arg2, err, value, req)
5313
5314 /*
5315 * Handle the sysctl request.
5316 *
5317 * If this is a read, the function will set the value to the current darkboot value. Otherwise,
5318 * we'll get the request identifier into "value" and then we can honor it.
5319 */
5320 if ((err = sysctl_io_number(req, darkboot, sizeof(int), &value, NULL)) != 0) {
5321 goto exit;
5322 }
5323
5324 /* writing requested, let's process the request */
5325 if (req->newptr) {
5326 /* writing is protected by an entitlement */
5327 if (priv_check_cred(kauth_cred_get(), PRIV_DARKBOOT, 0) != 0) {
5328 err = EPERM;
5329 goto exit;
5330 }
5331
5332 switch (value) {
5333 case MEMORY_MAINTENANCE_DARK_BOOT_UNSET:
5334 /*
5335 * If the darkboot sysctl is unset, the NVRAM variable
5336 * must be unset too. If that's not the case, it means
5337 * someone is doing something crazy and not supported.
5338 */
5339 if (darkboot != 0) {
5340 int ret = PERemoveNVRAMProperty(MEMORY_MAINTENANCE_DARK_BOOT_NVRAM_NAME);
5341 if (ret) {
5342 darkboot = 0;
5343 } else {
5344 err = EINVAL;
5345 }
5346 }
5347 break;
5348 case MEMORY_MAINTENANCE_DARK_BOOT_SET:
5349 darkboot = 1;
5350 break;
5351 case MEMORY_MAINTENANCE_DARK_BOOT_SET_PERSISTENT: {
5352 /*
5353 * Set the NVRAM and update 'darkboot' in case
5354 * of success. Otherwise, do not update
5355 * 'darkboot' and report the failure.
5356 */
5357 if (PEWriteNVRAMBooleanProperty(MEMORY_MAINTENANCE_DARK_BOOT_NVRAM_NAME, TRUE)) {
5358 darkboot = 1;
5359 } else {
5360 err = EINVAL;
5361 }
5362
5363 break;
5364 }
5365 default:
5366 err = EINVAL;
5367 }
5368 }
5369
5370 exit:
5371 return err;
5372 }
5373
5374 SYSCTL_PROC(_kern, OID_AUTO, darkboot,
5375 CTLFLAG_KERN | CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_ANYBODY,
5376 0, 0, sysctl_darkboot, "I", "");
5377 #endif /* CONFIG_DARKBOOT */
5378
5379 #if DEVELOPMENT || DEBUG
5380 #include <sys/sysent.h>
5381 /* This should result in a fatal exception, verifying that "sysent" is
5382 * write-protected.
5383 */
5384 static int
5385 kern_sysent_write(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
5386 {
5387 uint64_t new_value = 0, old_value = 0;
5388 int changed = 0, error;
5389
5390 error = sysctl_io_number(req, old_value, sizeof(uint64_t), &new_value, &changed);
5391 if ((error == 0) && changed) {
5392 volatile uint32_t *wraddr = __DECONST(uint32_t *, &sysent[0]);
5393 *wraddr = 0;
5394 printf("sysent[0] write succeeded\n");
5395 }
5396 return error;
5397 }
5398
5399 SYSCTL_PROC(_kern, OID_AUTO, sysent_const_check,
5400 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
5401 0, 0,
5402 kern_sysent_write, "I", "Attempt sysent[0] write");
5403
5404 #endif
5405
5406 #if DEVELOPMENT || DEBUG
5407 SYSCTL_COMPAT_INT(_kern, OID_AUTO, development, CTLFLAG_RD | CTLFLAG_MASKED | CTLFLAG_KERN, NULL, 1, "");
5408 #else
5409 SYSCTL_COMPAT_INT(_kern, OID_AUTO, development, CTLFLAG_RD | CTLFLAG_MASKED, NULL, 0, "");
5410 #endif
5411
5412 SYSCTL_INT(_kern, OID_AUTO, serverperfmode, CTLFLAG_RD, &serverperfmode, 0, "");
5413
5414 #if DEVELOPMENT || DEBUG
5415
5416 decl_lck_spin_data(, spinlock_panic_test_lock);
5417
5418 __attribute__((noreturn))
5419 static void
5420 spinlock_panic_test_acquire_spinlock(void * arg __unused, wait_result_t wres __unused)
5421 {
5422 lck_spin_lock(&spinlock_panic_test_lock);
5423 while (1) {
5424 ;
5425 }
5426 }
5427
5428 static int
5429 sysctl_spinlock_panic_test SYSCTL_HANDLER_ARGS
5430 {
5431 #pragma unused(oidp, arg1, arg2)
5432 if (req->newlen == 0) {
5433 return EINVAL;
5434 }
5435
5436 thread_t panic_spinlock_thread;
5437 /* Initialize panic spinlock */
5438 lck_grp_t * panic_spinlock_grp;
5439 lck_grp_attr_t * panic_spinlock_grp_attr;
5440 lck_attr_t * panic_spinlock_attr;
5441
5442 panic_spinlock_grp_attr = lck_grp_attr_alloc_init();
5443 panic_spinlock_grp = lck_grp_alloc_init("panic_spinlock", panic_spinlock_grp_attr);
5444 panic_spinlock_attr = lck_attr_alloc_init();
5445
5446 lck_spin_init(&spinlock_panic_test_lock, panic_spinlock_grp, panic_spinlock_attr);
5447
5448
5449 /* Create thread to acquire spinlock */
5450 if (kernel_thread_start(spinlock_panic_test_acquire_spinlock, NULL, &panic_spinlock_thread) != KERN_SUCCESS) {
5451 return EBUSY;
5452 }
5453
5454 /* Try to acquire spinlock -- should panic eventually */
5455 lck_spin_lock(&spinlock_panic_test_lock);
5456 while (1) {
5457 ;
5458 }
5459 }
5460
5461 __attribute__((noreturn))
5462 static void
5463 simultaneous_panic_worker
5464 (void * arg, wait_result_t wres __unused)
5465 {
5466 atomic_int *start_panic = (atomic_int *)arg;
5467
5468 while (!atomic_load(start_panic)) {
5469 ;
5470 }
5471 panic("SIMULTANEOUS PANIC TEST: INITIATING PANIC FROM CPU %d", cpu_number());
5472 __builtin_unreachable();
5473 }
5474
5475 static int
5476 sysctl_simultaneous_panic_test SYSCTL_HANDLER_ARGS
5477 {
5478 #pragma unused(oidp, arg1, arg2)
5479 if (req->newlen == 0) {
5480 return EINVAL;
5481 }
5482
5483 int i = 0, threads_to_create = 2 * processor_count;
5484 atomic_int start_panic = 0;
5485 unsigned int threads_created = 0;
5486 thread_t new_panic_thread;
5487
5488 for (i = threads_to_create; i > 0; i--) {
5489 if (kernel_thread_start(simultaneous_panic_worker, (void *) &start_panic, &new_panic_thread) == KERN_SUCCESS) {
5490 threads_created++;
5491 }
5492 }
5493
5494 /* FAIL if we couldn't create at least processor_count threads */
5495 if (threads_created < processor_count) {
5496 panic("SIMULTANEOUS PANIC TEST: FAILED TO CREATE ENOUGH THREADS, ONLY CREATED %d (of %d)",
5497 threads_created, threads_to_create);
5498 }
5499
5500 atomic_exchange(&start_panic, 1);
5501 while (1) {
5502 ;
5503 }
5504 }
5505
5506 extern unsigned int panic_test_failure_mode;
5507 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");
5508
5509 extern unsigned int panic_test_action_count;
5510 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");
5511
5512 extern unsigned int panic_test_case;
5513 SYSCTL_INT(_debug, OID_AUTO, xnu_panic_test_case, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_KERN, &panic_test_case, 0, "panic/debugger testcase");
5514
5515 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");
5516 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");
5517
5518 extern int exc_resource_threads_enabled;
5519 SYSCTL_INT(_kern, OID_AUTO, exc_resource_threads_enabled, CTLFLAG_RW | CTLFLAG_LOCKED, &exc_resource_threads_enabled, 0, "exc_resource thread limit enabled");
5520
5521 extern unsigned int verbose_panic_flow_logging;
5522 SYSCTL_INT(_debug, OID_AUTO, verbose_panic_flow_logging, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_KERN, &verbose_panic_flow_logging, 0, "verbose logging during panic");
5523
5524 #endif /* DEVELOPMENT || DEBUG */
5525
5526 #if BUILT_LTO
5527 static int _built_lto = 1;
5528 #else // BUILT_LTO
5529 static int _built_lto = 0;
5530 #endif // !BUILT_LTO
5531
5532 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");
5533
5534 #if CONFIG_THREAD_GROUPS
5535 #if DEVELOPMENT || DEBUG
5536
5537 static int
5538 sysctl_get_thread_group_id SYSCTL_HANDLER_ARGS
5539 {
5540 #pragma unused(arg1, arg2, oidp)
5541 uint64_t thread_group_id = thread_group_get_id(thread_group_get(current_thread()));
5542 return SYSCTL_OUT(req, &thread_group_id, sizeof(thread_group_id));
5543 }
5544
5545 SYSCTL_PROC(_kern, OID_AUTO, thread_group_id, CTLFLAG_RD | CTLFLAG_LOCKED | CTLTYPE_QUAD,
5546 0, 0, &sysctl_get_thread_group_id, "I", "thread group id of the thread");
5547
5548 STATIC int
5549 sysctl_thread_group_count(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
5550 {
5551 int value = thread_group_count();
5552 return sysctl_io_number(req, value, sizeof(value), NULL, NULL);
5553 }
5554
5555 SYSCTL_PROC(_kern, OID_AUTO, thread_group_count, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5556 0, 0, &sysctl_thread_group_count, "I", "count of thread groups");
5557
5558 #endif /* DEVELOPMENT || DEBUG */
5559 const uint32_t thread_groups_supported = 1;
5560 #else /* CONFIG_THREAD_GROUPS */
5561 const uint32_t thread_groups_supported = 0;
5562 #endif /* CONFIG_THREAD_GROUPS */
5563
5564 STATIC int
5565 sysctl_thread_groups_supported(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
5566 {
5567 int value = thread_groups_supported;
5568 return sysctl_io_number(req, value, sizeof(value), NULL, NULL);
5569 }
5570
5571 SYSCTL_PROC(_kern, OID_AUTO, thread_groups_supported, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5572 0, 0, &sysctl_thread_groups_supported, "I", "thread groups supported");
5573
5574 static int
5575 sysctl_grade_cputype SYSCTL_HANDLER_ARGS
5576 {
5577 #pragma unused(arg1, arg2, oidp)
5578 int error = 0;
5579 int type_tuple[2] = {};
5580 int return_value = 0;
5581
5582 error = SYSCTL_IN(req, &type_tuple, sizeof(type_tuple));
5583
5584 if (error) {
5585 return error;
5586 }
5587
5588 return_value = grade_binary(type_tuple[0], type_tuple[1] & ~CPU_SUBTYPE_MASK, type_tuple[1] & CPU_SUBTYPE_MASK, FALSE);
5589
5590 error = SYSCTL_OUT(req, &return_value, sizeof(return_value));
5591
5592 if (error) {
5593 return error;
5594 }
5595
5596 return error;
5597 }
5598
5599 SYSCTL_PROC(_kern, OID_AUTO, grade_cputype,
5600 CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_MASKED | CTLFLAG_LOCKED | CTLTYPE_OPAQUE,
5601 0, 0, &sysctl_grade_cputype, "S",
5602 "grade value of cpu_type_t+cpu_sub_type_t");
5603
5604 extern boolean_t allow_direct_handoff;
5605 SYSCTL_INT(_kern, OID_AUTO, direct_handoff,
5606 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5607 &allow_direct_handoff, 0, "Enable direct handoff for realtime threads");
5608
5609 #if DEVELOPMENT || DEBUG
5610
5611 SYSCTL_QUAD(_kern, OID_AUTO, phys_carveout_pa, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5612 &phys_carveout_pa,
5613 "base physical address of the phys_carveout_mb boot-arg region");
5614 SYSCTL_QUAD(_kern, OID_AUTO, phys_carveout_va, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5615 &phys_carveout,
5616 "base virtual address of the phys_carveout_mb boot-arg region");
5617 SYSCTL_QUAD(_kern, OID_AUTO, phys_carveout_size, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5618 &phys_carveout_size,
5619 "size in bytes of the phys_carveout_mb boot-arg region");
5620
5621
5622
5623 static int
5624 cseg_wedge_thread SYSCTL_HANDLER_ARGS
5625 {
5626 #pragma unused(arg1, arg2)
5627
5628 int error, val = 0;
5629 error = sysctl_handle_int(oidp, &val, 0, req);
5630 if (error || val == 0) {
5631 return error;
5632 }
5633
5634 do_cseg_wedge_thread();
5635 return 0;
5636 }
5637 SYSCTL_PROC(_kern, OID_AUTO, cseg_wedge_thread, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_MASKED, 0, 0, cseg_wedge_thread, "I", "wedge c_seg thread");
5638
5639 static int
5640 cseg_unwedge_thread SYSCTL_HANDLER_ARGS
5641 {
5642 #pragma unused(arg1, arg2)
5643
5644 int error, val = 0;
5645 error = sysctl_handle_int(oidp, &val, 0, req);
5646 if (error || val == 0) {
5647 return error;
5648 }
5649
5650 do_cseg_unwedge_thread();
5651 return 0;
5652 }
5653 SYSCTL_PROC(_kern, OID_AUTO, cseg_unwedge_thread, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_MASKED, 0, 0, cseg_unwedge_thread, "I", "unstuck c_seg thread");
5654
5655 static atomic_int wedge_thread_should_wake = 0;
5656
5657 static int
5658 unwedge_thread SYSCTL_HANDLER_ARGS
5659 {
5660 #pragma unused(arg1, arg2)
5661 int error, val = 0;
5662 error = sysctl_handle_int(oidp, &val, 0, req);
5663 if (error || val == 0) {
5664 return error;
5665 }
5666
5667 atomic_store(&wedge_thread_should_wake, 1);
5668 return 0;
5669 }
5670
5671 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");
5672
5673 static int
5674 wedge_thread SYSCTL_HANDLER_ARGS
5675 {
5676 #pragma unused(arg1, arg2)
5677
5678 int error, val = 0;
5679 error = sysctl_handle_int(oidp, &val, 0, req);
5680 if (error || val == 0) {
5681 return error;
5682 }
5683
5684 uint64_t interval = 1;
5685 nanoseconds_to_absolutetime(1000 * 1000 * 50, &interval);
5686
5687 atomic_store(&wedge_thread_should_wake, 0);
5688 while (!atomic_load(&wedge_thread_should_wake)) {
5689 tsleep1(NULL, 0, "wedge_thread", mach_absolute_time() + interval, NULL);
5690 }
5691
5692 return 0;
5693 }
5694
5695 SYSCTL_PROC(_kern, OID_AUTO, wedge_thread,
5696 CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED, 0, 0, wedge_thread, "I",
5697 "wedge this thread so it cannot be cleaned up");
5698
5699 static int
5700 sysctl_total_corpses_count SYSCTL_HANDLER_ARGS
5701 {
5702 #pragma unused(oidp, arg1, arg2)
5703 extern unsigned long total_corpses_count(void);
5704
5705 unsigned long corpse_count_long = total_corpses_count();
5706 unsigned int corpse_count = (unsigned int)MIN(corpse_count_long, UINT_MAX);
5707 return sysctl_io_opaque(req, &corpse_count, sizeof(corpse_count), NULL);
5708 }
5709
5710 SYSCTL_PROC(_kern, OID_AUTO, total_corpses_count,
5711 CTLFLAG_RD | CTLFLAG_ANYBODY | CTLFLAG_LOCKED, 0, 0,
5712 sysctl_total_corpses_count, "I", "total corpses on the system");
5713
5714 static int
5715 sysctl_turnstile_test_prim_lock SYSCTL_HANDLER_ARGS;
5716 static int
5717 sysctl_turnstile_test_prim_unlock SYSCTL_HANDLER_ARGS;
5718 int
5719 tstile_test_prim_lock(boolean_t use_hashtable);
5720 int
5721 tstile_test_prim_unlock(boolean_t use_hashtable);
5722
5723 static int
5724 sysctl_turnstile_test_prim_lock SYSCTL_HANDLER_ARGS
5725 {
5726 #pragma unused(arg1, arg2)
5727 int error, val = 0;
5728 error = sysctl_handle_int(oidp, &val, 0, req);
5729 if (error || val == 0) {
5730 return error;
5731 }
5732 switch (val) {
5733 case SYSCTL_TURNSTILE_TEST_USER_DEFAULT:
5734 case SYSCTL_TURNSTILE_TEST_USER_HASHTABLE:
5735 case SYSCTL_TURNSTILE_TEST_KERNEL_DEFAULT:
5736 case SYSCTL_TURNSTILE_TEST_KERNEL_HASHTABLE:
5737 return tstile_test_prim_lock(val);
5738 default:
5739 return error;
5740 }
5741 }
5742
5743 static int
5744 sysctl_turnstile_test_prim_unlock SYSCTL_HANDLER_ARGS
5745 {
5746 #pragma unused(arg1, arg2)
5747 int error, val = 0;
5748 error = sysctl_handle_int(oidp, &val, 0, req);
5749 if (error || val == 0) {
5750 return error;
5751 }
5752 switch (val) {
5753 case SYSCTL_TURNSTILE_TEST_USER_DEFAULT:
5754 case SYSCTL_TURNSTILE_TEST_USER_HASHTABLE:
5755 case SYSCTL_TURNSTILE_TEST_KERNEL_DEFAULT:
5756 case SYSCTL_TURNSTILE_TEST_KERNEL_HASHTABLE:
5757 return tstile_test_prim_unlock(val);
5758 default:
5759 return error;
5760 }
5761 }
5762
5763 SYSCTL_PROC(_kern, OID_AUTO, turnstiles_test_lock, CTLFLAG_WR | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED,
5764 0, 0, sysctl_turnstile_test_prim_lock, "I", "turnstiles test lock");
5765
5766 SYSCTL_PROC(_kern, OID_AUTO, turnstiles_test_unlock, CTLFLAG_WR | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED,
5767 0, 0, sysctl_turnstile_test_prim_unlock, "I", "turnstiles test unlock");
5768
5769 int
5770 turnstile_get_boost_stats_sysctl(void *req);
5771 int
5772 turnstile_get_unboost_stats_sysctl(void *req);
5773 static int
5774 sysctl_turnstile_boost_stats SYSCTL_HANDLER_ARGS;
5775 static int
5776 sysctl_turnstile_unboost_stats SYSCTL_HANDLER_ARGS;
5777 extern uint64_t thread_block_on_turnstile_count;
5778 extern uint64_t thread_block_on_regular_waitq_count;
5779
5780 static int
5781 sysctl_turnstile_boost_stats SYSCTL_HANDLER_ARGS
5782 {
5783 #pragma unused(arg1, arg2, oidp)
5784 return turnstile_get_boost_stats_sysctl(req);
5785 }
5786
5787 static int
5788 sysctl_turnstile_unboost_stats SYSCTL_HANDLER_ARGS
5789 {
5790 #pragma unused(arg1, arg2, oidp)
5791 return turnstile_get_unboost_stats_sysctl(req);
5792 }
5793
5794 SYSCTL_PROC(_kern, OID_AUTO, turnstile_boost_stats, CTLFLAG_RD | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED | CTLTYPE_STRUCT,
5795 0, 0, sysctl_turnstile_boost_stats, "S", "turnstiles boost stats");
5796 SYSCTL_PROC(_kern, OID_AUTO, turnstile_unboost_stats, CTLFLAG_RD | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED | CTLTYPE_STRUCT,
5797 0, 0, sysctl_turnstile_unboost_stats, "S", "turnstiles unboost stats");
5798 SYSCTL_QUAD(_kern, OID_AUTO, thread_block_count_on_turnstile,
5799 CTLFLAG_RD | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED,
5800 &thread_block_on_turnstile_count, "thread blocked on turnstile count");
5801 SYSCTL_QUAD(_kern, OID_AUTO, thread_block_count_on_reg_waitq,
5802 CTLFLAG_RD | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED,
5803 &thread_block_on_regular_waitq_count, "thread blocked on regular waitq count");
5804
5805 #if CONFIG_PV_TICKET
5806
5807 extern int ticket_lock_spins;
5808 SYSCTL_INT(_kern, OID_AUTO, ticket_lock_spins,
5809 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5810 &ticket_lock_spins, 0, "loops before hypercall");
5811
5812 #if (DEBUG || DEVELOPMENT)
5813
5814 /* PV ticket lock stats */
5815
5816 SYSCTL_SCALABLE_COUNTER(_kern, ticket_lock_kicks, ticket_kick_count,
5817 "ticket lock kicks");
5818 SYSCTL_SCALABLE_COUNTER(_kern, ticket_lock_waits, ticket_wait_count,
5819 "ticket lock waits");
5820 SYSCTL_SCALABLE_COUNTER(_kern, ticket_lock_already, ticket_already_count,
5821 "ticket lock already unlocked");
5822 SYSCTL_SCALABLE_COUNTER(_kern, ticket_lock_just_unlock, ticket_just_unlock,
5823 "ticket unlock without kick");
5824 SYSCTL_SCALABLE_COUNTER(_kern, ticket_lock_wflag_cleared, ticket_wflag_cleared,
5825 "ticket lock wait flag cleared");
5826 SYSCTL_SCALABLE_COUNTER(_kern, ticket_lock_wflag_still, ticket_wflag_still,
5827 "ticket lock wait flag not cleared");
5828 SYSCTL_SCALABLE_COUNTER(_kern, ticket_lock_spin_count, ticket_spin_count,
5829 "ticket lock spin count");
5830
5831 /* sysctl kern.hcall_probe=n -- does hypercall #n exist? */
5832
5833 static int
5834 sysctl_hcall_probe SYSCTL_HANDLER_ARGS
5835 {
5836 char instr[20];
5837
5838 if (!req->newptr) {
5839 return 0;
5840 }
5841 if (req->newlen >= sizeof(instr)) {
5842 return EOVERFLOW;
5843 }
5844
5845 int error = SYSCTL_IN(req, instr, req->newlen);
5846 if (error) {
5847 return error;
5848 }
5849 instr[req->newlen] = '\0';
5850
5851 int hcall = 0;
5852 error = sscanf(instr, "%d", &hcall);
5853 if (error != 1 || hcall < 0) {
5854 return EINVAL;
5855 }
5856 uprintf("%savailable\n",
5857 hvg_is_hcall_available((hvg_hcall_code_t)hcall) ? "" : "not ");
5858 return 0;
5859 }
5860
5861 SYSCTL_PROC(_kern, OID_AUTO, hcall_probe,
5862 CTLTYPE_STRING | CTLFLAG_WR | CTLFLAG_LOCKED | CTLFLAG_MASKED,
5863 0, 0, sysctl_hcall_probe, "A", "probe hypercall by id");
5864
5865 #endif /* (DEBUG || DEVELOPMENT) */
5866 #endif /* CONFIG_PV_TICKET */
5867
5868 #if defined(__x86_64__)
5869 extern uint64_t MutexSpin;
5870
5871 SYSCTL_QUAD(_kern, OID_AUTO, mutex_spin_abs, CTLFLAG_RW, &MutexSpin,
5872 "Spin time in abs for acquiring a kernel mutex");
5873 #else
5874 extern machine_timeout_t MutexSpin;
5875
5876 SYSCTL_QUAD(_kern, OID_AUTO, mutex_spin_abs, CTLFLAG_RW, &MutexSpin,
5877 "Spin time in abs for acquiring a kernel mutex");
5878 #endif
5879
5880 extern uint64_t low_MutexSpin;
5881 extern int64_t high_MutexSpin;
5882 extern unsigned int real_ncpus;
5883
5884 SYSCTL_QUAD(_kern, OID_AUTO, low_mutex_spin_abs, CTLFLAG_RW, &low_MutexSpin,
5885 "Low spin threshold in abs for acquiring a kernel mutex");
5886
5887 static int
5888 sysctl_high_mutex_spin_ns SYSCTL_HANDLER_ARGS
5889 {
5890 #pragma unused(oidp, arg1, arg2)
5891 int error;
5892 int64_t val = 0;
5893 int64_t res;
5894
5895 /* Check if the user is writing to high_MutexSpin, or just reading it */
5896 if (req->newptr) {
5897 error = SYSCTL_IN(req, &val, sizeof(val));
5898 if (error || (val < 0 && val != -1)) {
5899 return error;
5900 }
5901 high_MutexSpin = val;
5902 }
5903
5904 if (high_MutexSpin >= 0) {
5905 res = high_MutexSpin;
5906 } else {
5907 res = low_MutexSpin * real_ncpus;
5908 }
5909 return SYSCTL_OUT(req, &res, sizeof(res));
5910 }
5911 SYSCTL_PROC(_kern, OID_AUTO, high_mutex_spin_abs, CTLFLAG_RW | CTLTYPE_QUAD, 0, 0, sysctl_high_mutex_spin_ns, "I",
5912 "High spin threshold in abs for acquiring a kernel mutex");
5913
5914 #if defined (__x86_64__)
5915
5916 semaphore_t sysctl_test_panic_with_thread_sem;
5917
5918 #pragma clang diagnostic push
5919 #pragma clang diagnostic ignored "-Winfinite-recursion" /* rdar://38801963 */
5920 __attribute__((noreturn))
5921 static void
5922 panic_thread_test_child_spin(void * arg, wait_result_t wres)
5923 {
5924 static int panic_thread_recurse_count = 5;
5925
5926 if (panic_thread_recurse_count > 0) {
5927 panic_thread_recurse_count--;
5928 panic_thread_test_child_spin(arg, wres);
5929 }
5930
5931 semaphore_signal(sysctl_test_panic_with_thread_sem);
5932 while (1) {
5933 ;
5934 }
5935 }
5936 #pragma clang diagnostic pop
5937
5938 static void
5939 panic_thread_test_child_park(void * arg __unused, wait_result_t wres __unused)
5940 {
5941 int event;
5942
5943 assert_wait(&event, THREAD_UNINT);
5944 semaphore_signal(sysctl_test_panic_with_thread_sem);
5945 thread_block(panic_thread_test_child_park);
5946 }
5947
5948 static int
5949 sysctl_test_panic_with_thread SYSCTL_HANDLER_ARGS
5950 {
5951 #pragma unused(arg1, arg2)
5952 int rval = 0;
5953 char str[16] = { '\0' };
5954 thread_t child_thread = THREAD_NULL;
5955
5956 rval = sysctl_handle_string(oidp, str, sizeof(str), req);
5957 if (rval != 0 || !req->newptr) {
5958 return EINVAL;
5959 }
5960
5961 semaphore_create(kernel_task, &sysctl_test_panic_with_thread_sem, SYNC_POLICY_FIFO, 0);
5962
5963 /* Create thread to spin or park in continuation */
5964 if (strncmp("spin", str, strlen("spin")) == 0) {
5965 if (kernel_thread_start(panic_thread_test_child_spin, NULL, &child_thread) != KERN_SUCCESS) {
5966 semaphore_destroy(kernel_task, sysctl_test_panic_with_thread_sem);
5967 return EBUSY;
5968 }
5969 } else if (strncmp("continuation", str, strlen("continuation")) == 0) {
5970 if (kernel_thread_start(panic_thread_test_child_park, NULL, &child_thread) != KERN_SUCCESS) {
5971 semaphore_destroy(kernel_task, sysctl_test_panic_with_thread_sem);
5972 return EBUSY;
5973 }
5974 } else {
5975 semaphore_destroy(kernel_task, sysctl_test_panic_with_thread_sem);
5976 return EINVAL;
5977 }
5978
5979 semaphore_wait(sysctl_test_panic_with_thread_sem);
5980
5981 panic_with_thread_context(0, NULL, 0, child_thread, "testing panic_with_thread_context for thread %p", child_thread);
5982
5983 /* Not reached */
5984 return EINVAL;
5985 }
5986
5987 SYSCTL_PROC(_kern, OID_AUTO, test_panic_with_thread,
5988 CTLFLAG_MASKED | CTLFLAG_KERN | CTLFLAG_LOCKED | CTLFLAG_WR | CTLTYPE_STRING,
5989 0, 0, sysctl_test_panic_with_thread, "A", "test panic flow for backtracing a different thread");
5990 #endif /* defined (__x86_64__) */
5991
5992 static int
5993 sysctl_generate_file_permissions_guard_exception SYSCTL_HANDLER_ARGS
5994 {
5995 #pragma unused(arg1, arg2)
5996 int error, val = 0;
5997 error = sysctl_handle_int(oidp, &val, 0, req);
5998 if (error || val == 0) {
5999 return error;
6000 }
6001 generate_file_permissions_guard_exception(0, val);
6002 return 0;
6003 }
6004
6005 SYSCTL_PROC(_kern, OID_AUTO, file_perm_guard_exception, CTLFLAG_WR | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED,
6006 0, 0, sysctl_generate_file_permissions_guard_exception, "I", "Test File Permission Guard exception");
6007
6008 #endif /* DEVELOPMENT || DEBUG */
6009
6010 extern const int copysize_limit_panic;
6011 static int
6012 sysctl_get_owned_vmobjects SYSCTL_HANDLER_ARGS
6013 {
6014 #pragma unused(oidp, arg1, arg2)
6015
6016 /* validate */
6017 if (req->newlen != sizeof(mach_port_name_t) || req->newptr == USER_ADDR_NULL ||
6018 req->oldidx != 0 || req->newidx != 0 || req->p == NULL ||
6019 (req->oldlen == 0 && req->oldptr != USER_ADDR_NULL)) {
6020 return EINVAL;
6021 }
6022
6023 int error;
6024 mach_port_name_t task_port_name;
6025 task_t task;
6026 size_t buffer_size = (req->oldptr != USER_ADDR_NULL) ? req->oldlen : 0;
6027 vmobject_list_output_t buffer = NULL;
6028 size_t output_size;
6029 size_t entries;
6030 bool free_buffer = false;
6031
6032 /* we have a "newptr" (for write) we get a task port name from the caller. */
6033 error = SYSCTL_IN(req, &task_port_name, sizeof(mach_port_name_t));
6034
6035 if (error != 0) {
6036 goto sysctl_get_vmobject_list_exit;
6037 }
6038
6039 task = port_name_to_task_read(task_port_name);
6040 if (task == TASK_NULL) {
6041 error = ESRCH;
6042 goto sysctl_get_vmobject_list_exit;
6043 }
6044
6045 bool corpse = task_is_a_corpse(task);
6046
6047 /* get the current size */
6048 size_t max_size;
6049 task_get_owned_vmobjects(task, 0, NULL, &max_size, &entries);
6050
6051 if (buffer_size && (buffer_size < sizeof(*buffer) + sizeof(vm_object_query_data_t))) {
6052 error = ENOMEM;
6053 goto sysctl_get_vmobject_list_deallocate_and_exit;
6054 }
6055
6056 if (corpse == false) {
6057 /* copy the vmobjects and vmobject data out of the task */
6058 if (buffer_size == 0) {
6059 output_size = max_size;
6060 } else {
6061 buffer_size = (buffer_size > max_size) ? max_size : buffer_size;
6062 buffer = (struct _vmobject_list_output_ *)kalloc_data(buffer_size, Z_WAITOK);
6063
6064 if (!buffer) {
6065 error = ENOMEM;
6066 goto sysctl_get_vmobject_list_deallocate_and_exit;
6067 }
6068 free_buffer = true;
6069
6070 task_get_owned_vmobjects(task, buffer_size, buffer, &output_size, &entries);
6071 }
6072 } else {
6073 vmobject_list_output_t list;
6074
6075 task_get_corpse_vmobject_list(task, &list, &max_size);
6076 assert(buffer == NULL);
6077
6078 /* copy corpse_vmobject_list to output buffer to avoid double copy */
6079 if (buffer_size) {
6080 size_t temp_size;
6081
6082 temp_size = buffer_size > max_size ? max_size : buffer_size;
6083 output_size = temp_size - sizeof(*buffer);
6084 /* whole multiple of vm_object_query_data_t */
6085 output_size = (output_size / sizeof(vm_object_query_data_t)) * sizeof(vm_object_query_data_t) + sizeof(*buffer);
6086 buffer = list;
6087 } else {
6088 output_size = max_size;
6089 }
6090 }
6091
6092 /* req->oldptr should be USER_ADDR_NULL if buffer == NULL and return the current size */
6093 /* otherwise copy buffer to oldptr and return the bytes copied */
6094 size_t num_copied, chunk_size;
6095 for (num_copied = 0, chunk_size = 0;
6096 num_copied < output_size;
6097 num_copied += chunk_size) {
6098 chunk_size = MIN(output_size - num_copied, copysize_limit_panic);
6099 error = SYSCTL_OUT(req, (char *)buffer + num_copied, chunk_size);
6100 if (error) {
6101 break;
6102 }
6103 }
6104
6105 sysctl_get_vmobject_list_deallocate_and_exit:
6106 task_deallocate(task);
6107
6108 sysctl_get_vmobject_list_exit:
6109 if (free_buffer) {
6110 kfree_data(buffer, buffer_size);
6111 }
6112
6113 return error;
6114 }
6115
6116 SYSCTL_PROC(_vm, OID_AUTO, get_owned_vmobjects,
6117 CTLTYPE_OPAQUE | CTLFLAG_RD | CTLFLAG_WR | CTLFLAG_MASKED | CTLFLAG_KERN | CTLFLAG_LOCKED | CTLFLAG_ANYBODY,
6118 0, 0, sysctl_get_owned_vmobjects, "A", "get owned vmobjects in task");
6119
6120 extern uint64_t num_static_scalable_counters;
6121 SYSCTL_QUAD(_kern, OID_AUTO, num_static_scalable_counters, CTLFLAG_RD | CTLFLAG_LOCKED, &num_static_scalable_counters, "");
6122
6123 #if SCHED_HYGIENE_DEBUG
6124 TUNABLE_DT(bool, sched_hygiene_nonspec_tb, "machine-timeouts", "nonspec-tb", "sched-hygiene-nonspec-tb", false, TUNABLE_DT_NONE);
6125 static SECURITY_READ_ONLY_LATE(int) sched_hygiene_debug_available = 1;
6126 #else
6127 static SECURITY_READ_ONLY_LATE(int) sched_hygiene_debug_available = 0;
6128 #endif /* SCHED_HYGIENE_DEBUG */
6129
6130 SYSCTL_INT(_debug, OID_AUTO, sched_hygiene_debug_available,
6131 CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
6132 &sched_hygiene_debug_available, 0, "");
6133
6134 uuid_string_t trial_treatment_id;
6135 uuid_string_t trial_experiment_id;
6136 int trial_deployment_id = -1;
6137
6138 SYSCTL_STRING(_kern, OID_AUTO, trial_treatment_id, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_ANYBODY | CTLFLAG_EXPERIMENT, trial_treatment_id, sizeof(trial_treatment_id), "");
6139 SYSCTL_STRING(_kern, OID_AUTO, trial_experiment_id, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_ANYBODY | CTLFLAG_EXPERIMENT, trial_experiment_id, sizeof(trial_experiment_id), "");
6140 SYSCTL_INT(_kern, OID_AUTO, trial_deployment_id, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_ANYBODY | CTLFLAG_EXPERIMENT, &trial_deployment_id, 0, "");
6141
6142 #if (DEVELOPMENT || DEBUG)
6143 /* For unit testing setting factors & limits. */
6144 unsigned int testing_experiment_factor;
6145 EXPERIMENT_FACTOR_UINT(_kern, testing_experiment_factor, &testing_experiment_factor, 5, 10, "");
6146
6147 extern int exception_log_max_pid;
6148 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");
6149 #endif /* (DEVELOPMENT || DEBUG) */
6150
6151 #if DEVELOPMENT || DEBUG
6152 static int
6153 unlink_kernelcore_sysctl SYSCTL_HANDLER_ARGS
6154 {
6155 if (!req->newptr) {
6156 return EINVAL;
6157 }
6158 void IOBSDLowSpaceUnlinkKernelCore(void);
6159 IOBSDLowSpaceUnlinkKernelCore();
6160 return 0;
6161 }
6162
6163 SYSCTL_PROC(_kern, OID_AUTO, unlink_kernelcore,
6164 CTLTYPE_INT | CTLFLAG_WR | CTLFLAG_LOCKED | CTLFLAG_MASKED, 0, 0,
6165 unlink_kernelcore_sysctl, "-", "unlink the kernelcore file");
6166 #endif /* DEVELOPMENT || DEBUG */
6167
6168 #if CONFIG_IOTRACE
6169 #pragma clang diagnostic push
6170 #pragma clang diagnostic ignored "-Wcast-qual"
6171 SYSCTL_INT(_debug, OID_AUTO, MMIOtrace,
6172 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
6173 (int *)&mmiotrace_enabled, 0, "");
6174 #pragma clang diagnostic pop
6175 #endif /* CONFIG_IOTRACE */
6176
6177 static int
6178 sysctl_page_protection_type SYSCTL_HANDLER_ARGS
6179 {
6180 #pragma unused(oidp, arg1, arg2)
6181 int value = ml_page_protection_type();
6182 return SYSCTL_OUT(req, &value, sizeof(value));
6183 }
6184
6185 SYSCTL_PROC(_kern, OID_AUTO, page_protection_type,
6186 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
6187 0, 0, sysctl_page_protection_type, "I", "Type of page protection that the system supports");
6188
6189 TUNABLE_DT(int, gpu_pmem_selector, "defaults", "kern.gpu_pmem_selector", "gpu-pmem-selector", 0, TUNABLE_DT_NONE);
6190
6191 #if CONFIG_EXCLAVES
6192
6193 static int
6194 sysctl_task_conclave SYSCTL_HANDLER_ARGS
6195 {
6196 extern const char *exclaves_resource_name(void *);
6197
6198 #pragma unused(arg2)
6199 void *conclave = task_get_conclave(current_task());
6200 if (conclave != NULL) {
6201 const char *name = exclaves_resource_name(conclave);
6202 assert3u(strlen(name), >, 0);
6203
6204 /*
6205 * This is a RO operation already and the string is never
6206 * written to.
6207 */
6208 #pragma clang diagnostic push
6209 #pragma clang diagnostic ignored "-Wcast-qual"
6210 return sysctl_handle_string(oidp, (char *)name, 0, req);
6211 #pragma clang diagnostic pop
6212 }
6213 return sysctl_handle_string(oidp, arg1, MAXCONCLAVENAME, req);
6214 }
6215
6216 SYSCTL_PROC(_kern, OID_AUTO, task_conclave,
6217 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
6218 "", 0, sysctl_task_conclave, "A", "Conclave string for the task");
6219
6220
6221 void task_set_conclave_untaintable(task_t task);
6222
6223 static int
6224 sysctl_task_conclave_untaintable SYSCTL_HANDLER_ARGS
6225 {
6226 #pragma unused(arg1, arg2)
6227 int error, val = 0;
6228 error = sysctl_handle_int(oidp, &val, 0, req);
6229 if (error || val == 0) {
6230 return error;
6231 }
6232
6233 task_set_conclave_untaintable(current_task());
6234 return 0;
6235 }
6236
6237 SYSCTL_PROC(_kern, OID_AUTO, task_conclave_untaintable,
6238 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
6239 "", 0, sysctl_task_conclave_untaintable, "A", "Task could not be tainted by talking to conclaves");
6240
6241 extern exclaves_requirement_t exclaves_relaxed_requirements;
6242 SYSCTL_QUAD(_kern, OID_AUTO, exclaves_relaxed_requirements,
6243 CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
6244 &exclaves_relaxed_requirements, "Exclaves requirements which have been relaxed");
6245
6246 #endif /* CONFIG_EXCLAVES */
6247
6248 #if (DEVELOPMENT || DEBUG)
6249 SYSCTL_INT(_kern, OID_AUTO, gpu_pmem_selector,
6250 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
6251 &gpu_pmem_selector, 0, "GPU wire down limit selector");
6252 #else /* !(DEVELOPMENT || DEBUG) */
6253 SYSCTL_INT(_kern, OID_AUTO, gpu_pmem_selector,
6254 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN | CTLFLAG_MASKED,
6255 &gpu_pmem_selector, 0, "GPU wire down limit selector");
6256 #endif /* (DEVELOPMENT || DEBUG) */
6257
6258 static int
6259 sysctl_exclaves_status SYSCTL_HANDLER_ARGS
6260 {
6261 int value = exclaves_get_status();
6262 return sysctl_io_number(req, value, sizeof(value), NULL, NULL);
6263 }
6264
6265 SYSCTL_PROC(_kern, OID_AUTO, exclaves_status,
6266 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
6267 0, 0, sysctl_exclaves_status, "I", "Running status of Exclaves");
6268
6269
6270 static int
6271 sysctl_exclaves_boot_stage SYSCTL_HANDLER_ARGS
6272 {
6273 int value = exclaves_get_boot_stage();
6274 return sysctl_io_number(req, value, sizeof(value), NULL, NULL);
6275 }
6276
6277 SYSCTL_PROC(_kern, OID_AUTO, exclaves_boot_stage,
6278 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
6279 0, 0, sysctl_exclaves_boot_stage, "I", "Boot stage of Exclaves");
6280
6281 #if CONFIG_EXCLAVES && (DEVELOPMENT || DEBUG)
6282 extern unsigned int exclaves_debug;
6283 SYSCTL_UINT(_kern, OID_AUTO, exclaves_debug, CTLFLAG_RW | CTLFLAG_LOCKED,
6284 &exclaves_debug, 0, "Exclaves debug flags");
6285
6286 static int
6287 sysctl_exclaves_inspection_status SYSCTL_HANDLER_ARGS
6288 {
6289 #pragma unused(oidp, arg1, arg2)
6290 int value = (int)exclaves_inspection_is_initialized();
6291 return sysctl_io_number(req, value, sizeof(value), NULL, NULL);
6292 }
6293 SYSCTL_PROC(_kern, OID_AUTO, exclaves_inspection_status,
6294 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
6295 0, 0, sysctl_exclaves_inspection_status, "I", "Exclaves debug inspection status");
6296 #endif /* CONFIG_EXCLAVES && (DEVELOPMENT || DEBUG) */
6297
6298 #if (DEBUG || DEVELOPMENT)
6299 extern uint32_t disable_vm_sanitize_telemetry;
6300 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");
6301 #endif
6302