1 /*
2 * Copyright (c) 2000-2025 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_page.h>
160 #include <vm/vm_pageout_xnu.h>
161 #include <vm/vm_compressor_algorithms_xnu.h>
162 #include <vm/vm_compressor_xnu.h>
163 #include <sys/imgsrc.h>
164 #include <kern/timer_call.h>
165 #include <sys/codesign.h>
166 #include <IOKit/IOBSD.h>
167 #if CONFIG_CSR
168 #include <sys/csr.h>
169 #endif
170
171 #if defined(__i386__) || defined(__x86_64__)
172 #include <i386/cpuid.h>
173 #endif
174
175 #if CONFIG_FREEZE
176 #include <sys/kern_memorystatus.h>
177 #endif
178
179 #if KPERF
180 #include <kperf/kperf.h>
181 #endif
182
183 #if HYPERVISOR
184 #include <kern/hv_support.h>
185 #endif
186
187
188 #include <corecrypto/ccsha2.h>
189
190 /*
191 * deliberately setting max requests to really high number
192 * so that runaway settings do not cause MALLOC overflows
193 */
194 #define AIO_MAX_REQUESTS (128 * CONFIG_AIO_MAX)
195
196 extern int aio_max_requests;
197 extern int aio_max_requests_per_process;
198 extern int aio_worker_threads;
199 extern int lowpri_IO_window_msecs;
200 extern int lowpri_IO_delay_msecs;
201 #if DEVELOPMENT || DEBUG
202 extern int nx_enabled;
203 #endif
204 extern int speculative_reads_disabled;
205 extern unsigned int speculative_prefetch_max;
206 extern unsigned int speculative_prefetch_max_iosize;
207 extern unsigned int preheat_max_bytes;
208 extern unsigned int preheat_min_bytes;
209 extern long numvnodes;
210 extern long freevnodes;
211 extern long num_recycledvnodes;
212
213 extern uuid_string_t bootsessionuuid_string;
214
215 extern unsigned int vm_max_delayed_work_limit;
216 extern unsigned int vm_max_batch;
217
218 extern unsigned int vm_page_free_min;
219 extern unsigned int vm_page_free_target;
220 extern unsigned int vm_page_free_reserved;
221 extern unsigned int vm_page_max_speculative_age_q;
222
223 static uint64_t userspacereboottime = 0;
224 static unsigned int userspacerebootpurpose = 0;
225
226 #if (DEVELOPMENT || DEBUG)
227 extern uint32_t vm_page_creation_throttled_hard;
228 extern uint32_t vm_page_creation_throttled_soft;
229 #endif /* DEVELOPMENT || DEBUG */
230
231 #if DEVELOPMENT || DEBUG
232 extern bool bootarg_hide_process_traced;
233 #endif
234
235 /*
236 * Conditionally allow dtrace to see these functions for debugging purposes.
237 */
238 #ifdef STATIC
239 #undef STATIC
240 #endif
241 #if 0
242 #define STATIC
243 #else
244 #define STATIC static
245 #endif
246
247 extern boolean_t mach_timer_coalescing_enabled;
248
249 extern uint64_t timer_deadline_tracking_bin_1, timer_deadline_tracking_bin_2;
250
251 STATIC void
252 fill_user32_eproc(proc_t, struct user32_eproc *__restrict);
253 STATIC void
254 fill_user32_externproc(proc_t, struct user32_extern_proc *__restrict);
255 STATIC void
256 fill_user64_eproc(proc_t, struct user64_eproc *__restrict);
257 STATIC void
258 fill_user64_proc(proc_t, struct user64_kinfo_proc *__restrict);
259 STATIC void
260 fill_user64_externproc(proc_t, struct user64_extern_proc *__restrict);
261 STATIC void
262 fill_user32_proc(proc_t, struct user32_kinfo_proc *__restrict);
263
264 #if CONFIG_NETBOOT
265 extern int
266 netboot_root(void);
267 #endif
268 int
269 sysctl_procargs(int *name, u_int namelen, user_addr_t where,
270 size_t *sizep, proc_t cur_proc);
271 STATIC int
272 sysctl_procargsx(int *name, u_int namelen, user_addr_t where, size_t *sizep,
273 proc_t cur_proc, int argc_yes);
274 int
275 sysctl_struct(user_addr_t oldp, size_t *oldlenp, user_addr_t newp,
276 size_t newlen, void *sp, int len);
277
278 STATIC int sysdoproc_filt_KERN_PROC_PID(proc_t p, void * arg);
279 STATIC int sysdoproc_filt_KERN_PROC_PGRP(proc_t p, void * arg);
280 STATIC int sysdoproc_filt_KERN_PROC_TTY(proc_t p, void * arg);
281 STATIC int sysdoproc_filt_KERN_PROC_UID(proc_t p, void * arg);
282 STATIC int sysdoproc_filt_KERN_PROC_RUID(proc_t p, void * arg);
283 int sysdoproc_callback(proc_t p, void *arg);
284
285 #if CONFIG_THREAD_GROUPS && (DEVELOPMENT || DEBUG)
286 STATIC int sysctl_get_thread_group_id SYSCTL_HANDLER_ARGS;
287 #endif
288
289 /* forward declarations for non-static STATIC */
290 STATIC void fill_loadavg64(struct loadavg *la, struct user64_loadavg *la64);
291 STATIC void fill_loadavg32(struct loadavg *la, struct user32_loadavg *la32);
292 STATIC int sysctl_handle_kern_threadname(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
293 STATIC int sysctl_sched_stats(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
294 STATIC int sysctl_sched_stats_enable(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
295 #if COUNT_SYSCALLS
296 STATIC int sysctl_docountsyscalls SYSCTL_HANDLER_ARGS;
297 #endif /* COUNT_SYSCALLS */
298 #if defined(XNU_TARGET_OS_OSX)
299 STATIC int sysctl_doprocargs SYSCTL_HANDLER_ARGS;
300 #endif /* defined(XNU_TARGET_OS_OSX) */
301 STATIC int sysctl_doprocargs2 SYSCTL_HANDLER_ARGS;
302 STATIC int sysctl_prochandle SYSCTL_HANDLER_ARGS;
303 STATIC int sysctl_aiomax(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
304 STATIC int sysctl_aioprocmax(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
305 STATIC int sysctl_aiothreads(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
306 STATIC int sysctl_maxproc(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
307 STATIC int sysctl_osversion(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
308 STATIC int sysctl_sysctl_bootargs(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
309 STATIC int sysctl_maxvnodes(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
310 STATIC int sysctl_securelvl(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
311 STATIC int sysctl_domainname(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
312 STATIC int sysctl_hostname(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
313 STATIC int sysctl_procname(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
314 STATIC int sysctl_boottime(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
315 STATIC int sysctl_bootuuid(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
316 STATIC int sysctl_symfile(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
317 #if CONFIG_NETBOOT
318 STATIC int sysctl_netboot(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
319 #endif
320 #ifdef CONFIG_IMGSRC_ACCESS
321 STATIC int sysctl_imgsrcdev(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
322 #endif
323 STATIC int sysctl_usrstack(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
324 STATIC int sysctl_usrstack64(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
325 #if CONFIG_COREDUMP || CONFIG_UCOREDUMP
326 STATIC int sysctl_coredump(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
327 STATIC int sysctl_suid_coredump(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
328 #if CONFIG_UCOREDUMP
329 STATIC int sysctl_ucoredump(struct sysctl_oid *, void *, int, struct sysctl_req *);
330 #endif
331 #endif
332 STATIC int sysctl_delayterm(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
333 STATIC int sysctl_rage_vnode(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
334 STATIC int sysctl_kern_check_openevt(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
335 #if DEVELOPMENT || DEBUG
336 STATIC int sysctl_nx(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
337 #endif
338 STATIC int sysctl_loadavg(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
339 STATIC int sysctl_vm_toggle_address_reuse(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
340 STATIC int sysctl_swapusage(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
341 STATIC int fetch_process_cputype( proc_t cur_proc, int *name, u_int namelen, cpu_type_t *cputype);
342 STATIC int sysctl_sysctl_native(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
343 STATIC int sysctl_sysctl_cputype(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
344 STATIC int sysctl_safeboot(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
345 STATIC int sysctl_singleuser(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
346 STATIC int sysctl_minimalboot(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
347 STATIC int sysctl_slide(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
348
349 #ifdef CONFIG_XNUPOST
350 #include <tests/xnupost.h>
351
352 STATIC int sysctl_debug_test_oslog_ctl(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
353 STATIC int sysctl_debug_test_stackshot_mutex_owner(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
354 STATIC int sysctl_debug_test_stackshot_rwlck_owner(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
355 #endif
356
357 extern void IORegistrySetOSBuildVersion(char * build_version);
358 extern int IOParseWorkloadConfig(workload_config_ctx_t *ctx, const char * buffer, size_t size);
359 extern int IOUnparseWorkloadConfig(char *buffer, size_t *size);
360
361 STATIC void
fill_loadavg64(struct loadavg * la,struct user64_loadavg * la64)362 fill_loadavg64(struct loadavg *la, struct user64_loadavg *la64)
363 {
364 la64->ldavg[0] = la->ldavg[0];
365 la64->ldavg[1] = la->ldavg[1];
366 la64->ldavg[2] = la->ldavg[2];
367 la64->fscale = (user64_long_t)la->fscale;
368 }
369
370 STATIC void
fill_loadavg32(struct loadavg * la,struct user32_loadavg * la32)371 fill_loadavg32(struct loadavg *la, struct user32_loadavg *la32)
372 {
373 la32->ldavg[0] = la->ldavg[0];
374 la32->ldavg[1] = la->ldavg[1];
375 la32->ldavg[2] = la->ldavg[2];
376 la32->fscale = (user32_long_t)la->fscale;
377 }
378
379 #if COUNT_SYSCALLS
380 extern int do_count_syscalls;
381 #endif
382
383 #ifdef INSECURE
384 int securelevel = -1;
385 #else
386 int securelevel;
387 #endif
388
389 STATIC int
sysctl_handle_kern_threadname(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)390 sysctl_handle_kern_threadname( __unused struct sysctl_oid *oidp, __unused void *arg1,
391 __unused int arg2, struct sysctl_req *req)
392 {
393 int error;
394 struct uthread *ut = current_uthread();
395 user_addr_t oldp = 0, newp = 0;
396 size_t *oldlenp = NULL;
397 size_t newlen = 0;
398
399 oldp = req->oldptr;
400 oldlenp = &(req->oldlen);
401 newp = req->newptr;
402 newlen = req->newlen;
403
404 /* We want the current length, and maybe the string itself */
405 if (oldlenp) {
406 /* if we have no thread name yet tell'em we want MAXTHREADNAMESIZE - 1 */
407 size_t currlen = MAXTHREADNAMESIZE - 1;
408
409 if (ut->pth_name) {
410 /* use length of current thread name */
411 currlen = strlen(ut->pth_name);
412 }
413 if (oldp) {
414 if (*oldlenp < currlen) {
415 return ENOMEM;
416 }
417 /* NOTE - we do not copy the NULL terminator */
418 if (ut->pth_name) {
419 error = copyout(ut->pth_name, oldp, currlen);
420 if (error) {
421 return error;
422 }
423 }
424 }
425 /* return length of thread name minus NULL terminator (just like strlen) */
426 req->oldidx = currlen;
427 }
428
429 /* We want to set the name to something */
430 if (newp) {
431 if (newlen > (MAXTHREADNAMESIZE - 1)) {
432 return ENAMETOOLONG;
433 }
434 if (!ut->pth_name) {
435 char *tmp_pth_name = (char *)kalloc_data(MAXTHREADNAMESIZE,
436 Z_WAITOK | Z_ZERO);
437 if (!tmp_pth_name) {
438 return ENOMEM;
439 }
440 if (!OSCompareAndSwapPtr(NULL, tmp_pth_name, &ut->pth_name)) {
441 kfree_data(tmp_pth_name, MAXTHREADNAMESIZE);
442 return EBUSY;
443 }
444 } else {
445 kernel_debug_string_simple(TRACE_STRING_THREADNAME_PREV, ut->pth_name);
446 bzero(ut->pth_name, MAXTHREADNAMESIZE);
447 }
448 error = copyin(newp, ut->pth_name, newlen);
449 if (error) {
450 return error;
451 }
452
453 kernel_debug_string_simple(TRACE_STRING_THREADNAME, ut->pth_name);
454 }
455
456 return 0;
457 }
458
459 SYSCTL_PROC(_kern, KERN_THREADNAME, threadname, CTLFLAG_ANYBODY | CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_LOCKED, 0, 0, sysctl_handle_kern_threadname, "A", "");
460
461 #define WORKLOAD_CONFIG_MAX_SIZE (128 * 1024 * 1024)
462
463 /* Called locked - sysctl defined without CTLFLAG_LOCKED. */
464 static int
465 sysctl_workload_config SYSCTL_HANDLER_ARGS
466 {
467 #pragma unused(arg1, arg2)
468
469 char *plist_blob = NULL;
470 kern_return_t ret = KERN_FAILURE;
471 int error = -1;
472
473 /* Only allow reading of workload config on non-RELEASE kernels. */
474 #if DEVELOPMENT || DEBUG
475
476 const size_t buf_size = req->oldlen;
477
478 if (!req->oldptr) {
479 /* Just looking for the size to allocate. */
480 size_t size = 0;
481 ret = IOUnparseWorkloadConfig(NULL, &size);
482 if (ret != KERN_SUCCESS) {
483 return ENOMEM;
484 }
485
486 error = SYSCTL_OUT(req, NULL, size);
487 if (error) {
488 return error;
489 }
490 } else {
491 if (buf_size > (WORKLOAD_CONFIG_MAX_SIZE - 1) ||
492 buf_size == 0) {
493 return EINVAL;
494 }
495
496 plist_blob = kalloc_data(buf_size, Z_WAITOK | Z_ZERO);
497 if (!plist_blob) {
498 return ENOMEM;
499 }
500
501 size_t size = buf_size;
502 ret = IOUnparseWorkloadConfig(plist_blob, &size);
503 if (ret != KERN_SUCCESS) {
504 kfree_data(plist_blob, buf_size);
505 return ENOMEM;
506 }
507
508 error = SYSCTL_OUT(req, plist_blob, MIN(buf_size, size));
509
510 /* If the buffer was too small to fit the entire config. */
511 if (buf_size < size) {
512 error = ENOMEM;
513 }
514
515 kfree_data(plist_blob, buf_size);
516 if (error) {
517 return error;
518 }
519 }
520 #endif /* DEVELOPMENT || DEBUG */
521
522 if (req->newptr) {
523 size_t newlen = req->newlen;
524 if (newlen > (WORKLOAD_CONFIG_MAX_SIZE - 1)) {
525 return EINVAL;
526 }
527
528
529 workload_config_ctx_t *ctx = NULL;
530 /*
531 * Only allow workload_config_boot to be loaded once at boot by launchd.
532 */
533 if (current_proc() == initproc &&
534 !workload_config_initialized(&workload_config_boot)) {
535 ctx = &workload_config_boot;
536 } else {
537 #if DEVELOPMENT || DEBUG
538 /*
539 * Use the devel config context otherwise. If a devel config has been
540 * initialized it will be used for lookups in place of the boot config.
541 */
542 ctx = &workload_config_devel;
543 if (workload_config_initialized(ctx)) {
544 workload_config_free(ctx);
545 }
546
547 /* The devel context can be explicitly cleared by an empty string. */
548 if (newlen == 1) {
549 return 0;
550 }
551 #else
552 return EINVAL;
553 #endif
554 }
555
556 plist_blob = kalloc_data(newlen + 1, Z_WAITOK | Z_ZERO);
557 if (!plist_blob) {
558 return ENOMEM;
559 }
560 error = copyin(req->newptr, plist_blob, newlen);
561 if (error) {
562 kfree_data(plist_blob, newlen + 1);
563 return error;
564 }
565 plist_blob[newlen] = '\0';
566 ret = IOParseWorkloadConfig(ctx, plist_blob, newlen + 1);
567
568 kfree_data(plist_blob, newlen + 1);
569 return ret == KERN_SUCCESS ? 0 : EINVAL;
570 }
571
572 return 0;
573 }
574
575 SYSCTL_PROC(_kern, OID_AUTO, workload_config, CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_MASKED,
576 0, 0, sysctl_workload_config, "A", "global workgroup configuration plist load/unload");
577
578 #define BSD_HOST 1
579 STATIC int
sysctl_sched_stats(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)580 sysctl_sched_stats(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
581 {
582 host_basic_info_data_t hinfo;
583 kern_return_t kret;
584 uint32_t size;
585 uint32_t buf_size = 0;
586 int changed;
587 mach_msg_type_number_t count = HOST_BASIC_INFO_COUNT;
588 struct _processor_statistics_np *buf;
589 int error;
590
591 kret = host_info((host_t)BSD_HOST, HOST_BASIC_INFO, (host_info_t)&hinfo, &count);
592 if (kret != KERN_SUCCESS) {
593 return EINVAL;
594 }
595
596 size = sizeof(struct _processor_statistics_np) * (hinfo.logical_cpu_max + 2); /* One for RT Queue, One for Fair Share Queue */
597
598 if (req->oldlen < size) {
599 return EINVAL;
600 }
601
602 buf_size = size;
603 buf = (struct _processor_statistics_np *)kalloc_data(buf_size, Z_ZERO | Z_WAITOK);
604
605 kret = get_sched_statistics(buf, &size);
606 if (kret != KERN_SUCCESS) {
607 error = EINVAL;
608 goto out;
609 }
610
611 error = sysctl_io_opaque(req, buf, size, &changed);
612 if (error) {
613 goto out;
614 }
615
616 if (changed) {
617 panic("Sched info changed?!");
618 }
619 out:
620 kfree_data(buf, buf_size);
621 return error;
622 }
623
624 SYSCTL_PROC(_kern, OID_AUTO, sched_stats, CTLFLAG_LOCKED, 0, 0, sysctl_sched_stats, "-", "");
625
626 STATIC int
sysctl_sched_stats_enable(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,__unused struct sysctl_req * req)627 sysctl_sched_stats_enable(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, __unused struct sysctl_req *req)
628 {
629 boolean_t active;
630 int res;
631
632 if (req->newlen != sizeof(active)) {
633 return EINVAL;
634 }
635
636 res = copyin(req->newptr, &active, sizeof(active));
637 if (res != 0) {
638 return res;
639 }
640
641 return set_sched_stats_active(active);
642 }
643
644 SYSCTL_PROC(_kern, OID_AUTO, sched_stats_enable, CTLFLAG_LOCKED | CTLFLAG_WR, 0, 0, sysctl_sched_stats_enable, "-", "");
645
646 extern uint32_t sched_debug_flags;
647 SYSCTL_INT(_debug, OID_AUTO, sched, CTLFLAG_RW | CTLFLAG_LOCKED, &sched_debug_flags, 0, "scheduler debug");
648
649 #if (DEBUG || DEVELOPMENT)
650 extern boolean_t doprnt_hide_pointers;
651 SYSCTL_INT(_debug, OID_AUTO, hide_kernel_pointers, CTLFLAG_RW | CTLFLAG_LOCKED, &doprnt_hide_pointers, 0, "hide kernel pointers from log");
652 #endif
653
654
655 extern int get_kernel_symfile(proc_t, char **);
656
657 #if COUNT_SYSCALLS
658 #define KERN_COUNT_SYSCALLS (KERN_OSTYPE + 1000)
659
660 extern const unsigned int nsysent;
661 extern int syscalls_log[];
662 extern const char *syscallnames[];
663
664 STATIC int
665 sysctl_docountsyscalls SYSCTL_HANDLER_ARGS
666 {
667 __unused int cmd = oidp->oid_arg2; /* subcommand*/
668 __unused int *name = arg1; /* oid element argument vector */
669 __unused int namelen = arg2; /* number of oid element arguments */
670 int error, changed;
671
672 int tmp;
673
674 /* valid values passed in:
675 * = 0 means don't keep called counts for each bsd syscall
676 * > 0 means keep called counts for each bsd syscall
677 * = 2 means dump current counts to the system log
678 * = 3 means reset all counts
679 * for example, to dump current counts:
680 * sysctl -w kern.count_calls=2
681 */
682 error = sysctl_io_number(req, do_count_syscalls,
683 sizeof(do_count_syscalls), &tmp, &changed);
684
685 if (error != 0 || !changed) {
686 return error;
687 }
688
689 if (tmp == 1) {
690 do_count_syscalls = 1;
691 } else if (tmp == 0 || tmp == 2 || tmp == 3) {
692 for (int i = 0; i < nsysent; i++) {
693 if (syscalls_log[i] != 0) {
694 if (tmp == 2) {
695 printf("%d calls - name %s \n", syscalls_log[i], syscallnames[i]);
696 } else {
697 syscalls_log[i] = 0;
698 }
699 }
700 }
701 do_count_syscalls = (tmp != 0);
702 }
703
704 return error;
705 }
706 SYSCTL_PROC(_kern, KERN_COUNT_SYSCALLS, count_syscalls, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
707 0, /* Pointer argument (arg1) */
708 0, /* Integer argument (arg2) */
709 sysctl_docountsyscalls, /* Handler function */
710 NULL, /* Data pointer */
711 "");
712 #endif /* COUNT_SYSCALLS */
713
714 /*
715 * The following sysctl_* functions should not be used
716 * any more, as they can only cope with callers in
717 * user mode: Use new-style
718 * sysctl_io_number()
719 * sysctl_io_string()
720 * sysctl_io_opaque()
721 * instead.
722 */
723
724 STATIC int
sysdoproc_filt_KERN_PROC_PID(proc_t p,void * arg)725 sysdoproc_filt_KERN_PROC_PID(proc_t p, void * arg)
726 {
727 if (proc_getpid(p) != (pid_t)*(int*)arg) {
728 return 0;
729 } else {
730 return 1;
731 }
732 }
733
734 STATIC int
sysdoproc_filt_KERN_PROC_PGRP(proc_t p,void * arg)735 sysdoproc_filt_KERN_PROC_PGRP(proc_t p, void * arg)
736 {
737 if (p->p_pgrpid != (pid_t)*(int*)arg) {
738 return 0;
739 } else {
740 return 1;
741 }
742 }
743
744 STATIC int
sysdoproc_filt_KERN_PROC_TTY(proc_t p,void * arg)745 sysdoproc_filt_KERN_PROC_TTY(proc_t p, void * arg)
746 {
747 struct pgrp *pg;
748 dev_t dev = NODEV;
749
750 if ((p->p_flag & P_CONTROLT) && (pg = proc_pgrp(p, NULL)) != PGRP_NULL) {
751 dev = os_atomic_load(&pg->pg_session->s_ttydev, relaxed);
752 pgrp_rele(pg);
753 }
754
755 return dev != NODEV && dev == (dev_t)*(int *)arg;
756 }
757
758 STATIC int
sysdoproc_filt_KERN_PROC_UID(proc_t p,void * arg)759 sysdoproc_filt_KERN_PROC_UID(proc_t p, void * arg)
760 {
761 uid_t uid;
762
763 smr_proc_task_enter();
764 uid = kauth_cred_getuid(proc_ucred_smr(p));
765 smr_proc_task_leave();
766
767 if (uid != (uid_t)*(int*)arg) {
768 return 0;
769 } else {
770 return 1;
771 }
772 }
773
774
775 STATIC int
sysdoproc_filt_KERN_PROC_RUID(proc_t p,void * arg)776 sysdoproc_filt_KERN_PROC_RUID(proc_t p, void * arg)
777 {
778 uid_t ruid;
779
780 smr_proc_task_enter();
781 ruid = kauth_cred_getruid(proc_ucred_smr(p));
782 smr_proc_task_leave();
783
784 if (ruid != (uid_t)*(int*)arg) {
785 return 0;
786 } else {
787 return 1;
788 }
789 }
790
791 /*
792 * try over estimating by 5 procs
793 */
794 #define KERN_PROCSLOP (5 * sizeof(struct kinfo_proc))
795 struct sysdoproc_args {
796 size_t buflen;
797 void *kprocp;
798 boolean_t is_64_bit;
799 user_addr_t dp;
800 size_t needed;
801 unsigned int sizeof_kproc;
802 int *errorp;
803 int uidcheck;
804 int ruidcheck;
805 int ttycheck;
806 int uidval;
807 };
808
809 int
sysdoproc_callback(proc_t p,void * arg)810 sysdoproc_callback(proc_t p, void *arg)
811 {
812 struct sysdoproc_args *args = arg;
813
814 if (args->buflen >= args->sizeof_kproc) {
815 if ((args->ruidcheck != 0) && (sysdoproc_filt_KERN_PROC_RUID(p, &args->uidval) == 0)) {
816 return PROC_RETURNED;
817 }
818 if ((args->uidcheck != 0) && (sysdoproc_filt_KERN_PROC_UID(p, &args->uidval) == 0)) {
819 return PROC_RETURNED;
820 }
821 if ((args->ttycheck != 0) && (sysdoproc_filt_KERN_PROC_TTY(p, &args->uidval) == 0)) {
822 return PROC_RETURNED;
823 }
824
825 bzero(args->kprocp, args->sizeof_kproc);
826 if (args->is_64_bit) {
827 fill_user64_proc(p, args->kprocp);
828 } else {
829 fill_user32_proc(p, args->kprocp);
830 }
831 int error = copyout(args->kprocp, args->dp, args->sizeof_kproc);
832 if (error) {
833 *args->errorp = error;
834 return PROC_RETURNED_DONE;
835 }
836 args->dp += args->sizeof_kproc;
837 args->buflen -= args->sizeof_kproc;
838 }
839 args->needed += args->sizeof_kproc;
840 return PROC_RETURNED;
841 }
842
843 SYSCTL_NODE(_kern, KERN_PROC, proc, CTLFLAG_RD | CTLFLAG_LOCKED, 0, "");
844 STATIC int
845 sysctl_prochandle SYSCTL_HANDLER_ARGS
846 {
847 int cmd = oidp->oid_arg2; /* subcommand for multiple nodes */
848 int *name = arg1; /* oid element argument vector */
849 int namelen = arg2; /* number of oid element arguments */
850 user_addr_t where = req->oldptr;/* user buffer copy out address */
851
852 user_addr_t dp = where;
853 size_t needed = 0;
854 size_t buflen = where != USER_ADDR_NULL ? req->oldlen : 0;
855 int error = 0;
856 boolean_t is_64_bit = proc_is64bit(current_proc());
857 struct user32_kinfo_proc user32_kproc;
858 struct user64_kinfo_proc user_kproc;
859 int sizeof_kproc;
860 void *kprocp;
861 int (*filterfn)(proc_t, void *) = 0;
862 struct sysdoproc_args args;
863 int uidcheck = 0;
864 int ruidcheck = 0;
865 int ttycheck = 0;
866
867 if (namelen != 1 && !(namelen == 0 && cmd == KERN_PROC_ALL)) {
868 return EINVAL;
869 }
870
871 if (is_64_bit) {
872 sizeof_kproc = sizeof(user_kproc);
873 kprocp = &user_kproc;
874 } else {
875 sizeof_kproc = sizeof(user32_kproc);
876 kprocp = &user32_kproc;
877 }
878
879 switch (cmd) {
880 case KERN_PROC_PID:
881 filterfn = sysdoproc_filt_KERN_PROC_PID;
882 break;
883
884 case KERN_PROC_PGRP:
885 filterfn = sysdoproc_filt_KERN_PROC_PGRP;
886 break;
887
888 case KERN_PROC_TTY:
889 ttycheck = 1;
890 break;
891
892 case KERN_PROC_UID:
893 uidcheck = 1;
894 break;
895
896 case KERN_PROC_RUID:
897 ruidcheck = 1;
898 break;
899
900 case KERN_PROC_ALL:
901 break;
902
903 default:
904 /* must be kern.proc.<unknown> */
905 return ENOTSUP;
906 }
907
908 error = 0;
909 args.buflen = buflen;
910 args.kprocp = kprocp;
911 args.is_64_bit = is_64_bit;
912 args.dp = dp;
913 args.needed = needed;
914 args.errorp = &error;
915 args.uidcheck = uidcheck;
916 args.ruidcheck = ruidcheck;
917 args.ttycheck = ttycheck;
918 args.sizeof_kproc = sizeof_kproc;
919 if (namelen) {
920 args.uidval = name[0];
921 }
922
923 proc_iterate((PROC_ALLPROCLIST | PROC_ZOMBPROCLIST),
924 sysdoproc_callback, &args, filterfn, name);
925
926 if (error) {
927 return error;
928 }
929
930 dp = args.dp;
931 needed = args.needed;
932
933 if (where != USER_ADDR_NULL) {
934 req->oldlen = dp - where;
935 if (needed > req->oldlen) {
936 return ENOMEM;
937 }
938 } else {
939 needed += KERN_PROCSLOP;
940 req->oldlen = needed;
941 }
942 /* adjust index so we return the right required/consumed amount */
943 req->oldidx += req->oldlen;
944 return 0;
945 }
946
947
948 /*
949 * We specify the subcommand code for multiple nodes as the 'req->arg2' value
950 * in the sysctl declaration itself, which comes into the handler function
951 * as 'oidp->oid_arg2'.
952 *
953 * For these particular sysctls, since they have well known OIDs, we could
954 * have just obtained it from the '((int *)arg1)[0]' parameter, but that would
955 * not demonstrate how to handle multiple sysctls that used OID_AUTO instead
956 * of a well known value with a common handler function. This is desirable,
957 * because we want well known values to "go away" at some future date.
958 *
959 * It should be noted that the value of '((int *)arg1)[1]' is used for many
960 * an integer parameter to the subcommand for many of these sysctls; we'd
961 * rather have used '((int *)arg1)[0]' for that, or even better, an element
962 * in a structure passed in as the the 'newp' argument to sysctlbyname(3),
963 * and then use leaf-node permissions enforcement, but that would have
964 * necessitated modifying user space code to correspond to the interface
965 * change, and we are striving for binary backward compatibility here; even
966 * though these are SPI, and not intended for use by user space applications
967 * which are not themselves system tools or libraries, some applications
968 * have erroneously used them.
969 */
970 SYSCTL_PROC(_kern_proc, KERN_PROC_ALL, all, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
971 0, /* Pointer argument (arg1) */
972 KERN_PROC_ALL, /* Integer argument (arg2) */
973 sysctl_prochandle, /* Handler function */
974 NULL, /* Data is size variant on ILP32/LP64 */
975 "");
976 SYSCTL_PROC(_kern_proc, KERN_PROC_PID, pid, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
977 0, /* Pointer argument (arg1) */
978 KERN_PROC_PID, /* Integer argument (arg2) */
979 sysctl_prochandle, /* Handler function */
980 NULL, /* Data is size variant on ILP32/LP64 */
981 "");
982 SYSCTL_PROC(_kern_proc, KERN_PROC_TTY, tty, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
983 0, /* Pointer argument (arg1) */
984 KERN_PROC_TTY, /* Integer argument (arg2) */
985 sysctl_prochandle, /* Handler function */
986 NULL, /* Data is size variant on ILP32/LP64 */
987 "");
988 SYSCTL_PROC(_kern_proc, KERN_PROC_PGRP, pgrp, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
989 0, /* Pointer argument (arg1) */
990 KERN_PROC_PGRP, /* Integer argument (arg2) */
991 sysctl_prochandle, /* Handler function */
992 NULL, /* Data is size variant on ILP32/LP64 */
993 "");
994 SYSCTL_PROC(_kern_proc, KERN_PROC_UID, uid, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
995 0, /* Pointer argument (arg1) */
996 KERN_PROC_UID, /* Integer argument (arg2) */
997 sysctl_prochandle, /* Handler function */
998 NULL, /* Data is size variant on ILP32/LP64 */
999 "");
1000 SYSCTL_PROC(_kern_proc, KERN_PROC_RUID, ruid, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
1001 0, /* Pointer argument (arg1) */
1002 KERN_PROC_RUID, /* Integer argument (arg2) */
1003 sysctl_prochandle, /* Handler function */
1004 NULL, /* Data is size variant on ILP32/LP64 */
1005 "");
1006 SYSCTL_PROC(_kern_proc, KERN_PROC_LCID, lcid, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
1007 0, /* Pointer argument (arg1) */
1008 KERN_PROC_LCID, /* Integer argument (arg2) */
1009 sysctl_prochandle, /* Handler function */
1010 NULL, /* Data is size variant on ILP32/LP64 */
1011 "");
1012
1013
1014 /*
1015 * Fill in non-zero fields of an eproc structure for the specified process.
1016 */
1017 STATIC void
fill_user32_eproc(proc_t p,struct user32_eproc * __restrict ep)1018 fill_user32_eproc(proc_t p, struct user32_eproc *__restrict ep)
1019 {
1020 struct pgrp *pg;
1021 struct session *sessp;
1022 kauth_cred_t my_cred;
1023
1024 pg = proc_pgrp(p, &sessp);
1025
1026 if (pg != PGRP_NULL) {
1027 ep->e_pgid = p->p_pgrpid;
1028 ep->e_jobc = pg->pg_jobc;
1029 if (sessp->s_ttyvp) {
1030 ep->e_flag = EPROC_CTTY;
1031 }
1032 }
1033
1034 ep->e_ppid = p->p_ppid;
1035
1036 smr_proc_task_enter();
1037 my_cred = proc_ucred_smr(p);
1038
1039 /* A fake historical pcred */
1040 ep->e_pcred.p_ruid = kauth_cred_getruid(my_cred);
1041 ep->e_pcred.p_svuid = kauth_cred_getsvuid(my_cred);
1042 ep->e_pcred.p_rgid = kauth_cred_getrgid(my_cred);
1043 ep->e_pcred.p_svgid = kauth_cred_getsvgid(my_cred);
1044
1045 /* A fake historical *kauth_cred_t */
1046 unsigned long refcnt = os_atomic_load(&my_cred->cr_ref, relaxed);
1047 ep->e_ucred.cr_ref = (uint32_t)MIN(refcnt, UINT32_MAX);
1048 ep->e_ucred.cr_uid = kauth_cred_getuid(my_cred);
1049 ep->e_ucred.cr_ngroups = (short)posix_cred_get(my_cred)->cr_ngroups;
1050 bcopy(posix_cred_get(my_cred)->cr_groups,
1051 ep->e_ucred.cr_groups, NGROUPS * sizeof(gid_t));
1052
1053 my_cred = NOCRED;
1054 smr_proc_task_leave();
1055
1056 ep->e_tdev = NODEV;
1057 if (pg != PGRP_NULL) {
1058 if (p->p_flag & P_CONTROLT) {
1059 session_lock(sessp);
1060 ep->e_tdev = os_atomic_load(&sessp->s_ttydev, relaxed);
1061 ep->e_tpgid = sessp->s_ttypgrpid;
1062 session_unlock(sessp);
1063 }
1064 if (SESS_LEADER(p, sessp)) {
1065 ep->e_flag |= EPROC_SLEADER;
1066 }
1067 pgrp_rele(pg);
1068 }
1069 }
1070
1071 /*
1072 * Fill in non-zero fields of an LP64 eproc structure for the specified process.
1073 */
1074 STATIC void
fill_user64_eproc(proc_t p,struct user64_eproc * __restrict ep)1075 fill_user64_eproc(proc_t p, struct user64_eproc *__restrict ep)
1076 {
1077 struct pgrp *pg;
1078 struct session *sessp;
1079 kauth_cred_t my_cred;
1080
1081 pg = proc_pgrp(p, &sessp);
1082
1083 if (pg != PGRP_NULL) {
1084 ep->e_pgid = p->p_pgrpid;
1085 ep->e_jobc = pg->pg_jobc;
1086 if (sessp->s_ttyvp) {
1087 ep->e_flag = EPROC_CTTY;
1088 }
1089 }
1090
1091 ep->e_ppid = p->p_ppid;
1092
1093 smr_proc_task_enter();
1094 my_cred = proc_ucred_smr(p);
1095
1096 /* A fake historical pcred */
1097 ep->e_pcred.p_ruid = kauth_cred_getruid(my_cred);
1098 ep->e_pcred.p_svuid = kauth_cred_getsvuid(my_cred);
1099 ep->e_pcred.p_rgid = kauth_cred_getrgid(my_cred);
1100 ep->e_pcred.p_svgid = kauth_cred_getsvgid(my_cred);
1101
1102 /* A fake historical *kauth_cred_t */
1103 unsigned long refcnt = os_atomic_load(&my_cred->cr_ref, relaxed);
1104 ep->e_ucred.cr_ref = (uint32_t)MIN(refcnt, UINT32_MAX);
1105 ep->e_ucred.cr_uid = kauth_cred_getuid(my_cred);
1106 ep->e_ucred.cr_ngroups = (short)posix_cred_get(my_cred)->cr_ngroups;
1107 bcopy(posix_cred_get(my_cred)->cr_groups,
1108 ep->e_ucred.cr_groups, NGROUPS * sizeof(gid_t));
1109
1110 my_cred = NOCRED;
1111 smr_proc_task_leave();
1112
1113 ep->e_tdev = NODEV;
1114 if (pg != PGRP_NULL) {
1115 if (p->p_flag & P_CONTROLT) {
1116 session_lock(sessp);
1117 ep->e_tdev = os_atomic_load(&sessp->s_ttydev, relaxed);
1118 ep->e_tpgid = sessp->s_ttypgrpid;
1119 session_unlock(sessp);
1120 }
1121 if (SESS_LEADER(p, sessp)) {
1122 ep->e_flag |= EPROC_SLEADER;
1123 }
1124 pgrp_rele(pg);
1125 }
1126 }
1127
1128 /*
1129 * Fill in an eproc structure for the specified process.
1130 * bzeroed by our caller, so only set non-zero fields.
1131 */
1132 STATIC void
fill_user32_externproc(proc_t p,struct user32_extern_proc * __restrict exp)1133 fill_user32_externproc(proc_t p, struct user32_extern_proc *__restrict exp)
1134 {
1135 exp->p_starttime.tv_sec = (user32_time_t)p->p_start.tv_sec;
1136 exp->p_starttime.tv_usec = p->p_start.tv_usec;
1137 exp->p_flag = p->p_flag;
1138 #if DEVELOPMENT || DEBUG
1139 if (p->p_lflag & P_LTRACED && !bootarg_hide_process_traced) {
1140 #else
1141 if (p->p_lflag & P_LTRACED) {
1142 #endif
1143 exp->p_flag |= P_TRACED;
1144 }
1145 if (p->p_lflag & P_LPPWAIT) {
1146 exp->p_flag |= P_PPWAIT;
1147 }
1148 if (p->p_lflag & P_LEXIT) {
1149 exp->p_flag |= P_WEXIT;
1150 }
1151 exp->p_stat = p->p_stat;
1152 exp->p_pid = proc_getpid(p);
1153 #if DEVELOPMENT || DEBUG
1154 if (bootarg_hide_process_traced) {
1155 exp->p_oppid = 0;
1156 } else
1157 #endif
1158 {
1159 exp->p_oppid = p->p_oppid;
1160 }
1161 /* Mach related */
1162 exp->p_debugger = p->p_debugger;
1163 exp->sigwait = p->sigwait;
1164 /* scheduling */
1165 #ifdef _PROC_HAS_SCHEDINFO_
1166 exp->p_estcpu = p->p_estcpu;
1167 exp->p_pctcpu = p->p_pctcpu;
1168 exp->p_slptime = p->p_slptime;
1169 #endif
1170 exp->p_realtimer.it_interval.tv_sec =
1171 (user32_time_t)p->p_realtimer.it_interval.tv_sec;
1172 exp->p_realtimer.it_interval.tv_usec =
1173 (__int32_t)p->p_realtimer.it_interval.tv_usec;
1174
1175 exp->p_realtimer.it_value.tv_sec =
1176 (user32_time_t)p->p_realtimer.it_value.tv_sec;
1177 exp->p_realtimer.it_value.tv_usec =
1178 (__int32_t)p->p_realtimer.it_value.tv_usec;
1179
1180 exp->p_rtime.tv_sec = (user32_time_t)p->p_rtime.tv_sec;
1181 exp->p_rtime.tv_usec = (__int32_t)p->p_rtime.tv_usec;
1182
1183 exp->p_sigignore = p->p_sigignore;
1184 exp->p_sigcatch = p->p_sigcatch;
1185 exp->p_priority = p->p_priority;
1186 exp->p_nice = p->p_nice;
1187 bcopy(&p->p_comm, &exp->p_comm, MAXCOMLEN);
1188 exp->p_xstat = (u_short)MIN(p->p_xstat, USHRT_MAX);
1189 exp->p_acflag = p->p_acflag;
1190 }
1191
1192 /*
1193 * Fill in an LP64 version of extern_proc structure for the specified process.
1194 */
1195 STATIC void
1196 fill_user64_externproc(proc_t p, struct user64_extern_proc *__restrict exp)
1197 {
1198 exp->p_starttime.tv_sec = p->p_start.tv_sec;
1199 exp->p_starttime.tv_usec = p->p_start.tv_usec;
1200 exp->p_flag = p->p_flag;
1201 #if DEVELOPMENT || DEBUG
1202 if (p->p_lflag & P_LTRACED && !bootarg_hide_process_traced) {
1203 #else
1204 if (p->p_lflag & P_LTRACED) {
1205 #endif
1206 exp->p_flag |= P_TRACED;
1207 }
1208 if (p->p_lflag & P_LPPWAIT) {
1209 exp->p_flag |= P_PPWAIT;
1210 }
1211 if (p->p_lflag & P_LEXIT) {
1212 exp->p_flag |= P_WEXIT;
1213 }
1214 exp->p_stat = p->p_stat;
1215 exp->p_pid = proc_getpid(p);
1216 #if DEVELOPMENT || DEBUG
1217 if (bootarg_hide_process_traced) {
1218 exp->p_oppid = 0;
1219 } else
1220 #endif
1221 {
1222 exp->p_oppid = p->p_oppid;
1223 }
1224 /* Mach related */
1225 exp->p_debugger = p->p_debugger;
1226 exp->sigwait = p->sigwait;
1227 /* scheduling */
1228 #ifdef _PROC_HAS_SCHEDINFO_
1229 exp->p_estcpu = p->p_estcpu;
1230 exp->p_pctcpu = p->p_pctcpu;
1231 exp->p_slptime = p->p_slptime;
1232 #endif
1233 exp->p_realtimer.it_interval.tv_sec = p->p_realtimer.it_interval.tv_sec;
1234 exp->p_realtimer.it_interval.tv_usec = p->p_realtimer.it_interval.tv_usec;
1235
1236 exp->p_realtimer.it_value.tv_sec = p->p_realtimer.it_value.tv_sec;
1237 exp->p_realtimer.it_value.tv_usec = p->p_realtimer.it_value.tv_usec;
1238
1239 exp->p_rtime.tv_sec = p->p_rtime.tv_sec;
1240 exp->p_rtime.tv_usec = p->p_rtime.tv_usec;
1241
1242 exp->p_sigignore = p->p_sigignore;
1243 exp->p_sigcatch = p->p_sigcatch;
1244 exp->p_priority = p->p_priority;
1245 exp->p_nice = p->p_nice;
1246 bcopy(&p->p_comm, &exp->p_comm, MAXCOMLEN);
1247 exp->p_xstat = (u_short)MIN(p->p_xstat, USHRT_MAX);
1248 exp->p_acflag = p->p_acflag;
1249 }
1250
1251 STATIC void
1252 fill_user32_proc(proc_t p, struct user32_kinfo_proc *__restrict kp)
1253 {
1254 /* on a 64 bit kernel, 32 bit users get some truncated information */
1255 fill_user32_externproc(p, &kp->kp_proc);
1256 fill_user32_eproc(p, &kp->kp_eproc);
1257 }
1258
1259 STATIC void
1260 fill_user64_proc(proc_t p, struct user64_kinfo_proc *__restrict kp)
1261 {
1262 fill_user64_externproc(p, &kp->kp_proc);
1263 fill_user64_eproc(p, &kp->kp_eproc);
1264 }
1265
1266 #if defined(XNU_TARGET_OS_OSX)
1267 /*
1268 * Return the top *sizep bytes of the user stack, or the entire area of the
1269 * user stack down through the saved exec_path, whichever is smaller.
1270 */
1271 STATIC int
1272 sysctl_doprocargs SYSCTL_HANDLER_ARGS
1273 {
1274 __unused int cmd = oidp->oid_arg2; /* subcommand*/
1275 int *name = arg1; /* oid element argument vector */
1276 int namelen = arg2; /* number of oid element arguments */
1277 user_addr_t oldp = req->oldptr; /* user buffer copy out address */
1278 size_t *oldlenp = &req->oldlen; /* user buffer copy out size */
1279 // user_addr_t newp = req->newptr; /* user buffer copy in address */
1280 // size_t newlen = req->newlen; /* user buffer copy in size */
1281 int error;
1282
1283 error = sysctl_procargsx( name, namelen, oldp, oldlenp, current_proc(), 0);
1284
1285 /* adjust index so we return the right required/consumed amount */
1286 if (!error) {
1287 req->oldidx += req->oldlen;
1288 }
1289
1290 return error;
1291 }
1292 SYSCTL_PROC(_kern, KERN_PROCARGS, procargs, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
1293 0, /* Pointer argument (arg1) */
1294 0, /* Integer argument (arg2) */
1295 sysctl_doprocargs, /* Handler function */
1296 NULL, /* Data pointer */
1297 "");
1298 #endif /* defined(XNU_TARGET_OS_OSX) */
1299
1300 STATIC int
1301 sysctl_doprocargs2 SYSCTL_HANDLER_ARGS
1302 {
1303 __unused int cmd = oidp->oid_arg2; /* subcommand*/
1304 int *name = arg1; /* oid element argument vector */
1305 int namelen = arg2; /* number of oid element arguments */
1306 user_addr_t oldp = req->oldptr; /* user buffer copy out address */
1307 size_t *oldlenp = &req->oldlen; /* user buffer copy out size */
1308 // user_addr_t newp = req->newptr; /* user buffer copy in address */
1309 // size_t newlen = req->newlen; /* user buffer copy in size */
1310 int error;
1311
1312 error = sysctl_procargsx( name, namelen, oldp, oldlenp, current_proc(), 1);
1313
1314 /* adjust index so we return the right required/consumed amount */
1315 if (!error) {
1316 req->oldidx += req->oldlen;
1317 }
1318
1319 return error;
1320 }
1321 SYSCTL_PROC(_kern, KERN_PROCARGS2, procargs2, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
1322 0, /* Pointer argument (arg1) */
1323 0, /* Integer argument (arg2) */
1324 sysctl_doprocargs2, /* Handler function */
1325 NULL, /* Data pointer */
1326 "");
1327
1328 #define SYSCTL_PROCARGS_READ_ENVVARS_ENTITLEMENT "com.apple.private.read-environment-variables"
1329 STATIC int
1330 sysctl_procargsx(int *name, u_int namelen, user_addr_t where,
1331 size_t *sizep, proc_t cur_proc, int argc_yes)
1332 {
1333 assert(sizep != NULL);
1334 proc_t p = NULL;
1335 size_t buflen = where != USER_ADDR_NULL ? *sizep : 0;
1336 int error = 0;
1337 struct _vm_map *proc_map = NULL;
1338 struct task * task;
1339 vm_map_copy_t tmp = NULL;
1340 user_addr_t arg_addr;
1341 size_t arg_size;
1342 caddr_t data;
1343 size_t argslen = 0;
1344 size_t size = 0;
1345 vm_offset_t copy_start = 0, copy_end;
1346 vm_offset_t smallbuffer_start;
1347 kern_return_t ret;
1348 int pid;
1349 uid_t uid;
1350 int argc = -1;
1351 size_t argvsize;
1352 size_t remaining;
1353 size_t current_arg_index;
1354 size_t current_arg_len;
1355 const char * current_arg;
1356 bool omit_env_vars = true;
1357 user_addr_t user_stack;
1358 vm_map_offset_t effective_page_mask;
1359
1360 if (namelen < 1) {
1361 error = EINVAL;
1362 goto finish;
1363 }
1364
1365 if (argc_yes) {
1366 buflen -= sizeof(int); /* reserve first word to return argc */
1367 }
1368 /* we only care about buflen when where (oldp from sysctl) is not NULL. */
1369 /* when where (oldp from sysctl) is NULL and sizep (oldlenp from sysctl */
1370 /* is not NULL then the caller wants us to return the length needed to */
1371 /* hold the data we would return */
1372 if (where != USER_ADDR_NULL && (buflen <= 0 || buflen > ARG_MAX)) {
1373 error = EINVAL;
1374 goto finish;
1375 }
1376
1377 /*
1378 * Lookup process by pid
1379 */
1380 pid = name[0];
1381 p = proc_find(pid);
1382 if (p == NULL) {
1383 error = EINVAL;
1384 goto finish;
1385 }
1386
1387 /* Allow reading environment variables if any of the following are true:
1388 * - kernel is DEVELOPMENT || DEBUG
1389 * - target process is same as current_proc()
1390 * - target process is not cs_restricted
1391 * - SIP is off
1392 * - caller has an entitlement
1393 */
1394
1395 #if DEVELOPMENT || DEBUG
1396 omit_env_vars = false;
1397 #endif
1398 if (p == current_proc() ||
1399 !cs_restricted(p) ||
1400 #if CONFIG_CSR
1401 csr_check(CSR_ALLOW_UNRESTRICTED_DTRACE) == 0 ||
1402 #endif
1403 IOCurrentTaskHasEntitlement(SYSCTL_PROCARGS_READ_ENVVARS_ENTITLEMENT)
1404 ) {
1405 omit_env_vars = false;
1406 }
1407
1408 /*
1409 * Copy the top N bytes of the stack.
1410 * On all machines we have so far, the stack grows
1411 * downwards.
1412 *
1413 * If the user expects no more than N bytes of
1414 * argument list, use that as a guess for the
1415 * size.
1416 */
1417
1418 if (!p->user_stack) {
1419 error = EINVAL;
1420 goto finish;
1421 }
1422
1423 /* save off argc, argslen, user_stack before releasing the proc */
1424 argc = p->p_argc;
1425 argslen = p->p_argslen;
1426 user_stack = p->user_stack;
1427
1428 /*
1429 * When these sysctls were introduced, the first string in the strings
1430 * section was just the bare path of the executable. However, for security
1431 * reasons we now prefix this string with executable_path= so it can be
1432 * parsed getenv style. To avoid binary compatability issues with exising
1433 * callers of this sysctl, we strip it off here.
1434 * (rdar://problem/13746466)
1435 */
1436 #define EXECUTABLE_KEY "executable_path="
1437 argslen -= strlen(EXECUTABLE_KEY);
1438
1439 if (where == USER_ADDR_NULL && !omit_env_vars) {
1440 /* caller only wants to know length of proc args data.
1441 * If we don't need to omit environment variables, we can skip
1442 * copying the target process stack */
1443 goto calculate_size;
1444 }
1445
1446 smr_proc_task_enter();
1447 uid = kauth_cred_getuid(proc_ucred_smr(p));
1448 smr_proc_task_leave();
1449
1450 if ((uid != kauth_cred_getuid(kauth_cred_get()))
1451 && suser(kauth_cred_get(), &cur_proc->p_acflag)) {
1452 error = EINVAL;
1453 goto finish;
1454 }
1455
1456 /*
1457 * Before we can block (any VM code), make another
1458 * reference to the map to keep it alive. We do
1459 * that by getting a reference on the task itself.
1460 *
1461 * Additionally, if the task is not IPC active, we
1462 * must fail early. Other tasks can't yet look up
1463 * this task's task port to make Mach API calls, so
1464 * we shouldn't make such calls on their behalf.
1465 */
1466 task = proc_task(p);
1467 if (task == NULL || !task_is_ipc_active(task)) {
1468 error = EINVAL;
1469 goto finish;
1470 }
1471
1472 /*
1473 * Once we have a task reference we can convert that into a
1474 * map reference, which we will use in the calls below. The
1475 * task/process may change its map after we take this reference
1476 * (see execve), but the worst that will happen then is a return
1477 * of stale info (which is always a possibility).
1478 */
1479 task_reference(task);
1480 proc_rele(p);
1481 p = NULL;
1482 proc_map = get_task_map_reference(task);
1483 task_deallocate(task);
1484
1485 if (proc_map == NULL) {
1486 error = EINVAL;
1487 goto finish;
1488 }
1489
1490 effective_page_mask = vm_map_page_mask(proc_map);
1491
1492 arg_size = vm_map_round_page(argslen, effective_page_mask);
1493
1494 arg_addr = user_stack - arg_size;
1495
1496 ret = kmem_alloc(kernel_map, ©_start, arg_size,
1497 KMA_DATA | KMA_ZERO, VM_KERN_MEMORY_BSD);
1498 if (ret != KERN_SUCCESS) {
1499 error = ENOMEM;
1500 goto finish;
1501 }
1502
1503 copy_end = copy_start + arg_size;
1504
1505 if (vm_map_copyin(proc_map, (vm_map_address_t)arg_addr,
1506 (vm_map_size_t)arg_size, FALSE, &tmp) != KERN_SUCCESS) {
1507 error = EIO;
1508 goto finish;
1509 }
1510
1511 /*
1512 * Now that we've done the copyin from the process'
1513 * map, we can release the reference to it.
1514 */
1515 vm_map_deallocate(proc_map);
1516 proc_map = NULL;
1517
1518 if (vm_map_copy_overwrite(kernel_map,
1519 (vm_map_address_t)copy_start,
1520 tmp, (vm_map_size_t) arg_size,
1521 FALSE) != KERN_SUCCESS) {
1522 error = EIO;
1523 goto finish;
1524 }
1525 /* tmp was consumed */
1526 tmp = NULL;
1527
1528 if (omit_env_vars) {
1529 argvsize = 0;
1530
1531 /* Iterate over everything in argv, plus one for the bare executable path */
1532 for (current_arg_index = 0; current_arg_index < argc + 1 && argvsize < argslen; ++current_arg_index) {
1533 current_arg = (const char *)(copy_end - argslen) + argvsize;
1534 remaining = argslen - argvsize;
1535 current_arg_len = strnlen(current_arg, remaining);
1536 if (current_arg_len < remaining) {
1537 /* We have space for the null terminator */
1538 current_arg_len += 1;
1539
1540 if (current_arg_index == 0) {
1541 /* The bare executable path may have multiple null bytes after it for alignment */
1542 while (current_arg_len < remaining && current_arg[current_arg_len] == 0) {
1543 current_arg_len += 1;
1544 }
1545 }
1546 }
1547 argvsize += current_arg_len;
1548 }
1549 assert(argvsize <= argslen);
1550
1551 /* Adjust argslen and copy_end to make the copyout range extend to the end of argv */
1552 copy_end = copy_end - argslen + argvsize;
1553 argslen = argvsize;
1554 }
1555
1556 if (where == USER_ADDR_NULL) {
1557 /* Skip copyout */
1558 goto calculate_size;
1559 }
1560
1561 if (buflen >= argslen) {
1562 data = (caddr_t) (copy_end - argslen);
1563 size = argslen;
1564 } else {
1565 /*
1566 * Before rdar://25397314, this function contained incorrect logic when buflen is less
1567 * than argslen. The problem was that it copied in `buflen` bytes from the end of the target
1568 * process user stack into the beginning of a buffer of size round_page(buflen), and then
1569 * copied out `buflen` bytes from the end of this buffer. The effect of this was that
1570 * the caller of this sysctl would get zeros at the end of their buffer.
1571 *
1572 * To preserve this behavior, bzero everything from copy_end-round_page(buflen)+buflen to the
1573 * end of the buffer. This emulates copying in only `buflen` bytes.
1574 *
1575 *
1576 * In the old code:
1577 *
1578 * copy_start .... size: round_page(buflen) .... copy_end
1579 * [---copied in data (size: buflen)---|--- zeros ----------]
1580 * ^
1581 * data = copy_end - buflen
1582 *
1583 *
1584 * In the new code:
1585 * copy_start .... size: round_page(p->argslen) .... full copy_end
1586 * ^ ....................... p->argslen ...............................^
1587 * ^ ^ truncated copy_end ^
1588 * ^ ^ ^ ^
1589 * ^ ................ argslen ........................ ^
1590 * ^ ^ ^ ^
1591 * [-------copied in data (size: round_page(p->argslen))-------:----env vars---]
1592 * ^ ^
1593 * ^ data = copy_end - buflen
1594 * smallbuffer_start = max(copy_end - round_page(buflen), copy_start)
1595 *
1596 *
1597 * Full copy_end: copy_end calculated from copy_start + round_page(p->argslen)
1598 * Truncated copy_end: copy_end after truncation to remove environment variables.
1599 *
1600 * If environment variables were omitted, then we use the truncated copy_end, otherwise
1601 * we use full copy_end.
1602 *
1603 * smallbuffer_start: represents where copy_start would be in the old code.
1604 * data: The beginning of the region we copyout
1605 */
1606 smallbuffer_start = copy_end - vm_map_round_page(buflen, effective_page_mask);
1607 if (smallbuffer_start < copy_start) {
1608 smallbuffer_start = copy_start;
1609 }
1610 bzero((void *)(smallbuffer_start + buflen), copy_end - (smallbuffer_start + buflen));
1611 data = (caddr_t) (copy_end - buflen);
1612 size = buflen;
1613 }
1614
1615 if (argc_yes) {
1616 /* Put processes argc as the first word in the copyout buffer */
1617 suword(where, argc);
1618 error = copyout(data, (where + sizeof(int)), size);
1619 size += sizeof(int);
1620 } else {
1621 error = copyout(data, where, size);
1622
1623 /*
1624 * Make the old PROCARGS work to return the executable's path
1625 * But, only if there is enough space in the provided buffer
1626 *
1627 * on entry: data [possibily] points to the beginning of the path
1628 *
1629 * Note: we keep all pointers&sizes aligned to word boundries
1630 */
1631 if ((!error) && (buflen > 0 && (u_int)buflen > size)) {
1632 int binPath_sz, alignedBinPath_sz = 0;
1633 int extraSpaceNeeded, addThis;
1634 user_addr_t placeHere;
1635 char * str = (char *) data;
1636 size_t max_len = size;
1637
1638 /* Some apps are really bad about messing up their stacks
1639 * So, we have to be extra careful about getting the length
1640 * of the executing binary. If we encounter an error, we bail.
1641 */
1642
1643 /* Limit ourselves to PATH_MAX paths */
1644 if (max_len > PATH_MAX) {
1645 max_len = PATH_MAX;
1646 }
1647
1648 binPath_sz = 0;
1649
1650 while ((binPath_sz < max_len - 1) && (*str++ != 0)) {
1651 binPath_sz++;
1652 }
1653
1654 /* If we have a NUL terminator, copy it, too */
1655 if (binPath_sz < max_len - 1) {
1656 binPath_sz += 1;
1657 }
1658
1659 /* Pre-Flight the space requiremnts */
1660
1661 /* Account for the padding that fills out binPath to the next word */
1662 alignedBinPath_sz += (binPath_sz & (sizeof(int) - 1)) ? (sizeof(int) - (binPath_sz & (sizeof(int) - 1))) : 0;
1663
1664 placeHere = where + size;
1665
1666 /* Account for the bytes needed to keep placeHere word aligned */
1667 addThis = (placeHere & (sizeof(int) - 1)) ? (sizeof(int) - (placeHere & (sizeof(int) - 1))) : 0;
1668
1669 /* Add up all the space that is needed */
1670 extraSpaceNeeded = alignedBinPath_sz + addThis + binPath_sz + (4 * sizeof(int));
1671
1672 /* is there is room to tack on argv[0]? */
1673 if ((buflen & ~(sizeof(int) - 1)) >= (size + extraSpaceNeeded)) {
1674 placeHere += addThis;
1675 suword(placeHere, 0);
1676 placeHere += sizeof(int);
1677 suword(placeHere, 0xBFFF0000);
1678 placeHere += sizeof(int);
1679 suword(placeHere, 0);
1680 placeHere += sizeof(int);
1681 error = copyout(data, placeHere, binPath_sz);
1682 if (!error) {
1683 placeHere += binPath_sz;
1684 suword(placeHere, 0);
1685 size += extraSpaceNeeded;
1686 }
1687 }
1688 }
1689 }
1690
1691 calculate_size:
1692 /* Size has already been calculated for the where != NULL case */
1693 if (where == USER_ADDR_NULL) {
1694 size = argslen;
1695 if (argc_yes) {
1696 size += sizeof(int);
1697 } else {
1698 /*
1699 * old PROCARGS will return the executable's path and plus some
1700 * extra space for work alignment and data tags
1701 */
1702 size += PATH_MAX + (6 * sizeof(int));
1703 }
1704 size += (size & (sizeof(int) - 1)) ? (sizeof(int) - (size & (sizeof(int) - 1))) : 0;
1705 }
1706
1707 *sizep = size;
1708
1709 finish:
1710 if (p != NULL) {
1711 proc_rele(p);
1712 }
1713 if (tmp != NULL) {
1714 vm_map_copy_discard(tmp);
1715 }
1716 if (proc_map != NULL) {
1717 vm_map_deallocate(proc_map);
1718 }
1719 if (copy_start != (vm_offset_t) 0) {
1720 kmem_free(kernel_map, copy_start, arg_size);
1721 }
1722 return error;
1723 }
1724
1725
1726 /*
1727 * Max number of concurrent aio requests
1728 */
1729 STATIC int
1730 sysctl_aiomax
1731 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1732 {
1733 int new_value, changed;
1734 int error = sysctl_io_number(req, aio_max_requests, sizeof(int), &new_value, &changed);
1735 if (changed) {
1736 /* make sure the system-wide limit is greater than the per process limit */
1737 if (new_value >= aio_max_requests_per_process && new_value <= AIO_MAX_REQUESTS) {
1738 aio_max_requests = new_value;
1739 } else {
1740 error = EINVAL;
1741 }
1742 }
1743 return error;
1744 }
1745
1746
1747 /*
1748 * Max number of concurrent aio requests per process
1749 */
1750 STATIC int
1751 sysctl_aioprocmax
1752 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1753 {
1754 int new_value, changed;
1755 int error = sysctl_io_number(req, aio_max_requests_per_process, sizeof(int), &new_value, &changed);
1756 if (changed) {
1757 /* make sure per process limit is less than the system-wide limit */
1758 if (new_value <= aio_max_requests && new_value >= AIO_LISTIO_MAX) {
1759 aio_max_requests_per_process = new_value;
1760 } else {
1761 error = EINVAL;
1762 }
1763 }
1764 return error;
1765 }
1766
1767
1768 /*
1769 * Max number of async IO worker threads
1770 */
1771 STATIC int
1772 sysctl_aiothreads
1773 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1774 {
1775 int new_value, changed;
1776 int error = sysctl_io_number(req, aio_worker_threads, sizeof(int), &new_value, &changed);
1777 if (changed) {
1778 /* we only allow an increase in the number of worker threads */
1779 if (new_value > aio_worker_threads) {
1780 _aio_create_worker_threads((new_value - aio_worker_threads));
1781 aio_worker_threads = new_value;
1782 } else {
1783 error = EINVAL;
1784 }
1785 }
1786 return error;
1787 }
1788
1789
1790 /*
1791 * System-wide limit on the max number of processes
1792 */
1793 STATIC int
1794 sysctl_maxproc
1795 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1796 {
1797 int new_value, changed;
1798 int error = sysctl_io_number(req, maxproc, sizeof(int), &new_value, &changed);
1799 if (changed) {
1800 AUDIT_ARG(value32, new_value);
1801 /* make sure the system-wide limit is less than the configured hard
1802 * limit set at kernel compilation */
1803 if (new_value <= hard_maxproc && new_value > 0) {
1804 maxproc = new_value;
1805 } else {
1806 error = EINVAL;
1807 }
1808 }
1809 return error;
1810 }
1811
1812 #if CONFIG_SCHED_SMT
1813 STATIC int
1814 sysctl_sched_enable_smt
1815 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1816 {
1817 int new_value, changed;
1818 int error = sysctl_io_number(req, sched_enable_smt, sizeof(int), &new_value, &changed);
1819 if (error) {
1820 return error;
1821 }
1822 kern_return_t kret = KERN_SUCCESS;
1823 if (changed) {
1824 AUDIT_ARG(value32, new_value);
1825 if (new_value == 0) {
1826 sched_enable_smt = 0;
1827 kret = enable_smt_processors(false);
1828 } else {
1829 sched_enable_smt = 1;
1830 kret = enable_smt_processors(true);
1831 }
1832 }
1833 switch (kret) {
1834 case KERN_SUCCESS:
1835 error = 0;
1836 break;
1837 case KERN_INVALID_ARGUMENT:
1838 error = EINVAL;
1839 break;
1840 case KERN_FAILURE:
1841 error = EBUSY;
1842 break;
1843 default:
1844 error = ENOENT;
1845 break;
1846 }
1847 return error;
1848 }
1849 #else /* CONFIG_SCHED_SMT */
1850 STATIC int
1851 sysctl_sched_enable_smt
1852 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, __unused struct sysctl_req *req)
1853 {
1854 return 0;
1855 }
1856 #endif /* CONFIG_SCHED_SMT */
1857
1858 SYSCTL_STRING(_kern, KERN_OSTYPE, ostype,
1859 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
1860 ostype, 0, "");
1861 SYSCTL_STRING(_kern, KERN_OSRELEASE, osrelease,
1862 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
1863 osrelease, 0, "");
1864 SYSCTL_INT(_kern, KERN_OSREV, osrevision,
1865 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
1866 (int *)NULL, BSD, "");
1867 SYSCTL_STRING(_kern, KERN_VERSION, version,
1868 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
1869 version, 0, "");
1870 SYSCTL_STRING(_kern, OID_AUTO, uuid,
1871 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
1872 &kernel_uuid_string[0], 0, "");
1873
1874 SYSCTL_STRING(_kern, OID_AUTO, osbuildconfig,
1875 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED | CTLFLAG_MASKED,
1876 &osbuild_config[0], 0, "");
1877
1878 #if DEBUG
1879 #ifndef DKPR
1880 #define DKPR 1
1881 #endif
1882 #endif
1883
1884 #if DKPR
1885 int debug_kprint_syscall = 0;
1886 char debug_kprint_syscall_process[MAXCOMLEN + 1];
1887
1888 /* Thread safe: bits and string value are not used to reclaim state */
1889 SYSCTL_INT(_debug, OID_AUTO, kprint_syscall,
1890 CTLFLAG_RW | CTLFLAG_LOCKED, &debug_kprint_syscall, 0, "kprintf syscall tracing");
1891 SYSCTL_STRING(_debug, OID_AUTO, kprint_syscall_process,
1892 CTLFLAG_RW | CTLFLAG_LOCKED, debug_kprint_syscall_process, sizeof(debug_kprint_syscall_process),
1893 "name of process for kprintf syscall tracing");
1894
1895 int
1896 debug_kprint_current_process(const char **namep)
1897 {
1898 struct proc *p = current_proc();
1899
1900 if (p == NULL) {
1901 return 0;
1902 }
1903
1904 if (debug_kprint_syscall_process[0]) {
1905 /* user asked to scope tracing to a particular process name */
1906 if (0 == strncmp(debug_kprint_syscall_process,
1907 p->p_comm, sizeof(debug_kprint_syscall_process))) {
1908 /* no value in telling the user that we traced what they asked */
1909 if (namep) {
1910 *namep = NULL;
1911 }
1912
1913 return 1;
1914 } else {
1915 return 0;
1916 }
1917 }
1918
1919 /* trace all processes. Tell user what we traced */
1920 if (namep) {
1921 *namep = p->p_comm;
1922 }
1923
1924 return 1;
1925 }
1926 #endif
1927
1928 /* PR-5293665: need to use a callback function for kern.osversion to set
1929 * osversion in IORegistry */
1930
1931 STATIC int
1932 sysctl_osversion(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
1933 {
1934 int rval = 0;
1935
1936 rval = sysctl_handle_string(oidp, arg1, arg2, req);
1937
1938 if (req->newptr) {
1939 IORegistrySetOSBuildVersion((char *)arg1);
1940 }
1941
1942 return rval;
1943 }
1944
1945 SYSCTL_PROC(_kern, KERN_OSVERSION, osversion,
1946 CTLFLAG_RW | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
1947 osversion, 256 /* OSVERSIZE*/,
1948 sysctl_osversion, "A", "");
1949
1950 static bool
1951 _already_set_or_not_launchd(struct sysctl_req *req, char *val)
1952 {
1953 if (req->newptr != 0) {
1954 /*
1955 * Can only ever be set by launchd, and only once at boot.
1956 */
1957 if (proc_getpid(req->p) != 1 || val[0] != '\0') {
1958 return true;
1959 }
1960 }
1961 return false;
1962 }
1963
1964 #define kRootsInstalledReadWriteEntitlement "com.apple.private.roots-installed-read-write"
1965 #define kRootsInstalledReadOnlyEntitlement "com.apple.private.roots-installed-read-only"
1966 uint64_t roots_installed = 0;
1967
1968 static int
1969 sysctl_roots_installed
1970 (__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
1971 {
1972 int error = 0;
1973
1974 if (req->newptr != 0) {
1975 /* a ReadWrite entitlement is required for updating this syscl
1976 * meanwhile, only allow write once
1977 */
1978 if (!IOCurrentTaskHasEntitlement(kRootsInstalledReadWriteEntitlement) || (roots_installed != 0)) {
1979 return EPERM;
1980 }
1981 } else {
1982 /* for reader of this sysctl, need either ReadWrite or ReadOnly entitlement */
1983 if (!IOCurrentTaskHasEntitlement(kRootsInstalledReadWriteEntitlement) &&
1984 !IOCurrentTaskHasEntitlement(kRootsInstalledReadOnlyEntitlement)) {
1985 return EPERM;
1986 }
1987 }
1988
1989 error = sysctl_handle_quad(oidp, arg1, arg2, req);
1990
1991 return error;
1992 }
1993
1994 SYSCTL_PROC(_kern, OID_AUTO, roots_installed,
1995 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
1996 &roots_installed, sizeof(roots_installed),
1997 sysctl_roots_installed, "Q", "");
1998
1999 #if XNU_TARGET_OS_OSX
2000 static int
2001 sysctl_system_version_compat
2002 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2003 {
2004 int oldval = (task_has_system_version_compat_enabled(current_task()));
2005 int new_value = 0, changed = 0;
2006
2007 int error = sysctl_io_number(req, oldval, sizeof(int), &new_value, &changed);
2008 if (changed) {
2009 task_set_system_version_compat_enabled(current_task(), (new_value));
2010 }
2011 return error;
2012 }
2013
2014 SYSCTL_PROC(_kern, OID_AUTO, system_version_compat,
2015 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
2016 0, 0, sysctl_system_version_compat, "A", "");
2017 #endif /* XNU_TARGET_OS_OSX */
2018
2019 #if XNU_TARGET_OS_OSX || defined(XNU_EXPERIMENTAL_SYSTEM_VERSION_COMPAT)
2020 char osproductversioncompat[48] = { '\0' };
2021
2022 static int
2023 sysctl_osproductversioncompat(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2024 {
2025 if (_already_set_or_not_launchd(req, osproductversioncompat)) {
2026 return EPERM;
2027 }
2028 return sysctl_handle_string(oidp, arg1, arg2, req);
2029 }
2030
2031
2032 SYSCTL_PROC(_kern, OID_AUTO, osproductversioncompat,
2033 CTLFLAG_RW | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2034 osproductversioncompat, sizeof(osproductversioncompat),
2035 sysctl_osproductversioncompat, "A", "The ProductVersion from SystemVersionCompat.plist");
2036 #endif /* XNU_TARGET_OS_OSX || defined(XNU_EXPERIMENTAL_SYSTEM_VERSION_COMPAT) */
2037
2038 char osproductversion[48] = { '\0' };
2039
2040 static char iossupportversion_string[48] = { '\0' };
2041
2042 #if defined(XNU_EXPERIMENTAL_SYSTEM_VERSION_COMPAT)
2043 /*
2044 * Equivalent to dyld_program_sdk_at_least(dyld_fall_2025_os_versions).
2045 */
2046 static bool
2047 proc_2025_fall_os_sdk_or_later(struct proc *p)
2048 {
2049 const uint32_t proc_sdk_ver = proc_sdk(p);
2050
2051 switch (proc_platform(p)) {
2052 case PLATFORM_MACOS:
2053 return proc_sdk_ver >= 0x00100000; // DYLD_MACOSX_VERSION_16_0
2054 case PLATFORM_IOS:
2055 case PLATFORM_IOSSIMULATOR:
2056 case PLATFORM_MACCATALYST:
2057 return proc_sdk_ver >= 0x00130000; // DYLD_IOS_VERSION_19_0
2058 case PLATFORM_BRIDGEOS:
2059 return proc_sdk_ver >= 0x000a0000; // DYLD_BRIDGEOS_VERSION_10_0
2060 case PLATFORM_TVOS:
2061 case PLATFORM_TVOSSIMULATOR:
2062 return proc_sdk_ver >= 0x00130000; // DYLD_TVOS_VERSION_19_0
2063 case PLATFORM_WATCHOS:
2064 case PLATFORM_WATCHOSSIMULATOR:
2065 return proc_sdk_ver >= 0x000c0000; // DYLD_WATCHOS_VERSION_12_0
2066 default:
2067 return true;
2068 }
2069 }
2070 #endif /* defined(XNU_EXPERIMENTAL_SYSTEM_VERSION_COMPAT) */
2071
2072 static int
2073 sysctl_osproductversion(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2074 {
2075 if (_already_set_or_not_launchd(req, osproductversion)) {
2076 return EPERM;
2077 }
2078
2079 #if XNU_TARGET_OS_OSX
2080 if (task_has_system_version_compat_enabled(current_task()) && (osproductversioncompat[0] != '\0')) {
2081 return sysctl_handle_string(oidp, osproductversioncompat, arg2, req);
2082 }
2083 #endif /* XNU_TARGET_OS_OSX */
2084
2085 #if defined(XNU_TARGET_OS_XR)
2086 if (proc_platform(req->p) == PLATFORM_IOS && (iossupportversion_string[0] != '\0')) {
2087 return sysctl_handle_string(oidp, iossupportversion_string, arg2, req);
2088 }
2089 #endif /* defined(XNU_TARGET_OS_XR) */
2090
2091 #if defined(XNU_EXPERIMENTAL_SYSTEM_VERSION_COMPAT)
2092 if (!proc_2025_fall_os_sdk_or_later(req->p) && (osproductversioncompat[0] != '\0')) {
2093 return sysctl_handle_string(oidp, osproductversioncompat, arg2, req);
2094 }
2095 #endif /* defined(XNU_EXPERIMENTAL_SYSTEM_VERSION_COMPAT) */
2096
2097 return sysctl_handle_string(oidp, arg1, arg2, req);
2098 }
2099
2100 #if XNU_TARGET_OS_OSX
2101 static_assert(sizeof(osproductversioncompat) == sizeof(osproductversion),
2102 "osproductversion size matches osproductversioncompat size");
2103 #endif
2104
2105 SYSCTL_PROC(_kern, OID_AUTO, osproductversion,
2106 CTLFLAG_RW | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2107 osproductversion, sizeof(osproductversion),
2108 sysctl_osproductversion, "A", "The ProductVersion from SystemVersion.plist");
2109
2110 char osreleasetype[OSRELEASETYPE_SIZE] = { '\0' };
2111
2112 STATIC int
2113 sysctl_osreleasetype(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2114 {
2115 if (_already_set_or_not_launchd(req, osreleasetype)) {
2116 return EPERM;
2117 }
2118 return sysctl_handle_string(oidp, arg1, arg2, req);
2119 }
2120
2121 void reset_osreleasetype(void);
2122
2123 void
2124 reset_osreleasetype(void)
2125 {
2126 memset(osreleasetype, 0, sizeof(osreleasetype));
2127 }
2128
2129 SYSCTL_PROC(_kern, OID_AUTO, osreleasetype,
2130 CTLFLAG_RW | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2131 osreleasetype, sizeof(osreleasetype),
2132 sysctl_osreleasetype, "A", "The ReleaseType from SystemVersion.plist");
2133
2134 STATIC int
2135 sysctl_iossupportversion(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2136 {
2137 if (_already_set_or_not_launchd(req, iossupportversion_string)) {
2138 return EPERM;
2139 }
2140
2141 #if defined(XNU_TARGET_OS_XR)
2142 if (proc_platform(req->p) == PLATFORM_IOS) {
2143 /* return empty string for iOS processes to match how this would behave on iOS */
2144 return sysctl_handle_string(oidp, "", arg2, req);
2145 } else {
2146 /* native processes see the actual value */
2147 return sysctl_handle_string(oidp, arg1, arg2, req);
2148 }
2149 #else
2150 return sysctl_handle_string(oidp, arg1, arg2, req);
2151 #endif
2152 }
2153
2154 SYSCTL_PROC(_kern, OID_AUTO, iossupportversion,
2155 CTLFLAG_RW | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2156 iossupportversion_string, sizeof(iossupportversion_string),
2157 sysctl_iossupportversion, "A", "The iOSSupportVersion from SystemVersion.plist");
2158
2159 static uint64_t osvariant_status = 0;
2160
2161 STATIC int
2162 sysctl_osvariant_status(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2163 {
2164 if (req->newptr != 0) {
2165 /*
2166 * Can only ever be set by launchd, and only once.
2167 * Reset by usrctl() -> reset_osvariant_status() during
2168 * userspace reboot, since userspace could reboot into
2169 * a different variant.
2170 */
2171 if (proc_getpid(req->p) != 1 || osvariant_status != 0) {
2172 return EPERM;
2173 }
2174 }
2175
2176 int err = sysctl_handle_quad(oidp, arg1, arg2, req);
2177
2178 reset_debug_syscall_rejection_mode();
2179
2180 return err;
2181 }
2182
2183 SYSCTL_PROC(_kern, OID_AUTO, osvariant_status,
2184 CTLFLAG_RW | CTLTYPE_QUAD | CTLFLAG_LOCKED | CTLFLAG_MASKED,
2185 &osvariant_status, sizeof(osvariant_status),
2186 sysctl_osvariant_status, "Q", "Opaque flags used to cache OS variant information");
2187
2188 static bool
2189 _os_variant_check_disabled(enum os_variant_property property)
2190 {
2191 return (osvariant_status >> (32 + property)) & 0x1;
2192 }
2193
2194 static bool
2195 _os_variant_has(enum os_variant_status_flags_positions p)
2196 {
2197 return ((osvariant_status >> (p * OS_VARIANT_STATUS_BIT_WIDTH)) & OS_VARIANT_STATUS_MASK) == OS_VARIANT_S_YES;
2198 }
2199
2200 bool
2201 os_variant_has_internal_diagnostics(__unused const char *subsystem)
2202 {
2203 if (_os_variant_check_disabled(OS_VARIANT_PROPERTY_DIAGNOSTICS)) {
2204 return false;
2205 }
2206 #if XNU_TARGET_OS_OSX
2207 return _os_variant_has(OS_VARIANT_SFP_INTERNAL_CONTENT) || _os_variant_has(OS_VARIANT_SFP_INTERNAL_DIAGS_PROFILE);
2208 #else
2209 return _os_variant_has(OS_VARIANT_SFP_INTERNAL_RELEASE_TYPE);
2210 #endif /* XNU_TARGET_OS_OSX */
2211 }
2212
2213 void reset_osvariant_status(void);
2214
2215 void
2216 reset_osvariant_status(void)
2217 {
2218 osvariant_status = 0;
2219 reset_debug_syscall_rejection_mode();
2220 }
2221
2222 extern void commpage_update_dyld_flags(uint64_t);
2223 TUNABLE_WRITEABLE(uint64_t, dyld_flags, "dyld_flags", 0);
2224
2225 STATIC int
2226 sysctl_dyld_flags(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2227 {
2228 /*
2229 * Can only ever be set by launchd, possibly several times
2230 * as dyld may change its mind after a userspace reboot.
2231 */
2232 if (req->newptr != 0 && proc_getpid(req->p) != 1) {
2233 return EPERM;
2234 }
2235
2236 int res = sysctl_handle_quad(oidp, arg1, arg2, req);
2237 if (req->newptr && res == 0) {
2238 commpage_update_dyld_flags(dyld_flags);
2239 }
2240 return res;
2241 }
2242
2243 SYSCTL_PROC(_kern, OID_AUTO, dyld_flags,
2244 CTLFLAG_RW | CTLTYPE_QUAD | CTLFLAG_LOCKED | CTLFLAG_MASKED,
2245 &dyld_flags, sizeof(dyld_flags),
2246 sysctl_dyld_flags, "Q", "Opaque flags used to cache dyld system-wide configuration");
2247
2248 #if defined(XNU_TARGET_OS_BRIDGE)
2249 char macosproductversion[MACOS_VERS_LEN] = { '\0' };
2250
2251 SYSCTL_STRING(_kern, OID_AUTO, macosproductversion,
2252 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2253 &macosproductversion[0], MACOS_VERS_LEN, "The currently running macOS ProductVersion (from SystemVersion.plist on macOS)");
2254
2255 char macosversion[MACOS_VERS_LEN] = { '\0' };
2256
2257 SYSCTL_STRING(_kern, OID_AUTO, macosversion,
2258 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2259 &macosversion[0], MACOS_VERS_LEN, "The currently running macOS build version");
2260 #endif
2261
2262 STATIC int
2263 sysctl_sysctl_bootargs
2264 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2265 {
2266 int error;
2267 char buf[BOOT_LINE_LENGTH];
2268
2269 strlcpy(buf, PE_boot_args(), BOOT_LINE_LENGTH);
2270 error = sysctl_io_string(req, buf, BOOT_LINE_LENGTH, 0, NULL);
2271 return error;
2272 }
2273
2274 SYSCTL_PROC(_kern, OID_AUTO, bootargs,
2275 CTLFLAG_LOCKED | CTLFLAG_RD | CTLFLAG_KERN | CTLTYPE_STRING,
2276 NULL, 0,
2277 sysctl_sysctl_bootargs, "A", "bootargs");
2278
2279 STATIC int
2280 sysctl_kernelcacheuuid(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2281 {
2282 int rval = ENOENT;
2283 if (kernelcache_uuid_valid) {
2284 rval = sysctl_handle_string(oidp, arg1, arg2, req);
2285 }
2286 return rval;
2287 }
2288
2289 SYSCTL_PROC(_kern, OID_AUTO, kernelcacheuuid,
2290 CTLFLAG_RD | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2291 kernelcache_uuid_string, sizeof(kernelcache_uuid_string),
2292 sysctl_kernelcacheuuid, "A", "");
2293
2294 STATIC int
2295 sysctl_systemfilesetuuid(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2296 {
2297 int rval = ENOENT;
2298 if (pageablekc_uuid_valid) {
2299 rval = sysctl_handle_string(oidp, arg1, arg2, req);
2300 }
2301 return rval;
2302 }
2303
2304 SYSCTL_PROC(_kern, OID_AUTO, systemfilesetuuid,
2305 CTLFLAG_RD | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2306 pageablekc_uuid_string, sizeof(pageablekc_uuid_string),
2307 sysctl_systemfilesetuuid, "A", "");
2308
2309 STATIC int
2310 sysctl_auxiliaryfilesetuuid(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2311 {
2312 int rval = ENOENT;
2313 if (auxkc_uuid_valid) {
2314 rval = sysctl_handle_string(oidp, arg1, arg2, req);
2315 }
2316 return rval;
2317 }
2318
2319 SYSCTL_PROC(_kern, OID_AUTO, auxiliaryfilesetuuid,
2320 CTLFLAG_RD | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2321 auxkc_uuid_string, sizeof(auxkc_uuid_string),
2322 sysctl_auxiliaryfilesetuuid, "A", "");
2323
2324 STATIC int
2325 sysctl_filesetuuid(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2326 {
2327 int rval = ENOENT;
2328 kc_format_t kcformat;
2329 kernel_mach_header_t *mh;
2330 void *uuid = NULL;
2331 unsigned long uuidlen = 0;
2332 uuid_string_t uuid_str;
2333
2334 if (!PE_get_primary_kc_format(&kcformat) || kcformat != KCFormatFileset) {
2335 return rval;
2336 }
2337
2338 mh = (kernel_mach_header_t *)PE_get_kc_header(KCKindPrimary);
2339 uuid = getuuidfromheader(mh, &uuidlen);
2340
2341 if ((uuid != NULL) && (uuidlen == sizeof(uuid_t))) {
2342 uuid_unparse_upper(*(uuid_t *)uuid, uuid_str);
2343 rval = sysctl_io_string(req, (char *)uuid_str, sizeof(uuid_str), 0, NULL);
2344 }
2345
2346 return rval;
2347 }
2348
2349 SYSCTL_PROC(_kern, OID_AUTO, filesetuuid,
2350 CTLFLAG_RD | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2351 NULL, 0,
2352 sysctl_filesetuuid, "A", "");
2353
2354
2355 SYSCTL_INT(_kern, KERN_MAXFILES, maxfiles,
2356 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2357 &maxfiles, 0, "");
2358 SYSCTL_INT(_kern, KERN_ARGMAX, argmax,
2359 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2360 (int *)NULL, ARG_MAX, "");
2361 SYSCTL_INT(_kern, KERN_POSIX1, posix1version,
2362 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2363 (int *)NULL, _POSIX_VERSION, "");
2364 SYSCTL_INT(_kern, KERN_NGROUPS, ngroups,
2365 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2366 (int *)NULL, NGROUPS_MAX, "");
2367 SYSCTL_INT(_kern, KERN_JOB_CONTROL, job_control,
2368 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2369 (int *)NULL, 1, "");
2370 #if 1 /* _POSIX_SAVED_IDS from <unistd.h> */
2371 SYSCTL_INT(_kern, KERN_SAVED_IDS, saved_ids,
2372 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2373 (int *)NULL, 1, "");
2374 #else
2375 SYSCTL_INT(_kern, KERN_SAVED_IDS, saved_ids,
2376 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2377 NULL, 0, "");
2378 #endif
2379 SYSCTL_INT(_kern, OID_AUTO, num_files,
2380 CTLFLAG_RD | CTLFLAG_LOCKED,
2381 &nfiles, 0, "");
2382 SYSCTL_COMPAT_INT(_kern, OID_AUTO, num_vnodes,
2383 CTLFLAG_RD | CTLFLAG_LOCKED,
2384 &numvnodes, 0, "");
2385 SYSCTL_INT(_kern, OID_AUTO, num_tasks,
2386 CTLFLAG_RD | CTLFLAG_LOCKED,
2387 &task_max, 0, "");
2388 SYSCTL_INT(_kern, OID_AUTO, num_threads,
2389 CTLFLAG_RD | CTLFLAG_LOCKED,
2390 &thread_max, 0, "");
2391 SYSCTL_INT(_kern, OID_AUTO, num_taskthreads,
2392 CTLFLAG_RD | CTLFLAG_LOCKED,
2393 &task_threadmax, 0, "");
2394 SYSCTL_LONG(_kern, OID_AUTO, num_recycledvnodes,
2395 CTLFLAG_RD | CTLFLAG_LOCKED,
2396 &num_recycledvnodes, "");
2397 SYSCTL_COMPAT_INT(_kern, OID_AUTO, free_vnodes,
2398 CTLFLAG_RD | CTLFLAG_LOCKED,
2399 &freevnodes, 0, "");
2400
2401 STATIC int
2402 sysctl_maxvnodes(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2403 {
2404 int oldval = desiredvnodes;
2405 int error = sysctl_io_number(req, desiredvnodes, sizeof(int), &desiredvnodes, NULL);
2406
2407 if (oldval != desiredvnodes) {
2408 resize_namecache(desiredvnodes);
2409 }
2410
2411 return error;
2412 }
2413
2414 SYSCTL_INT(_kern, OID_AUTO, namecache_disabled,
2415 CTLFLAG_RW | CTLFLAG_LOCKED,
2416 &nc_disabled, 0, "");
2417
2418 SYSCTL_PROC(_kern, KERN_MAXVNODES, maxvnodes,
2419 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2420 0, 0, sysctl_maxvnodes, "I", "");
2421
2422 SYSCTL_PROC(_kern, KERN_MAXPROC, maxproc,
2423 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2424 0, 0, sysctl_maxproc, "I", "");
2425
2426 SYSCTL_PROC(_kern, KERN_AIOMAX, aiomax,
2427 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2428 0, 0, sysctl_aiomax, "I", "");
2429
2430 SYSCTL_PROC(_kern, KERN_AIOPROCMAX, aioprocmax,
2431 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2432 0, 0, sysctl_aioprocmax, "I", "");
2433
2434 SYSCTL_PROC(_kern, KERN_AIOTHREADS, aiothreads,
2435 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2436 0, 0, sysctl_aiothreads, "I", "");
2437
2438 SYSCTL_PROC(_kern, OID_AUTO, sched_enable_smt,
2439 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_KERN,
2440 0, 0, sysctl_sched_enable_smt, "I", "");
2441
2442 extern int sched_allow_NO_SMT_threads;
2443 SYSCTL_INT(_kern, OID_AUTO, sched_allow_NO_SMT_threads,
2444 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2445 &sched_allow_NO_SMT_threads, 0, "");
2446
2447 extern int sched_avoid_cpu0;
2448 SYSCTL_INT(_kern, OID_AUTO, sched_rt_avoid_cpu0,
2449 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2450 &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");
2451
2452 #if (DEVELOPMENT || DEBUG)
2453
2454 static int
2455 sysctl_kern_max_unsafe_rt_quanta(__unused struct sysctl_oid *oidp,
2456 __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2457 {
2458 extern void sched_set_max_unsafe_rt_quanta(int);
2459 extern int max_unsafe_rt_quanta;
2460
2461 int new_value, changed;
2462 int old_value = max_unsafe_rt_quanta;
2463 int error = sysctl_io_number(req, old_value, sizeof(int), &new_value,
2464 &changed);
2465 if (changed) {
2466 sched_set_max_unsafe_rt_quanta(new_value);
2467 }
2468
2469 return error;
2470 }
2471
2472 SYSCTL_PROC(_kern, OID_AUTO, max_unsafe_rt_quanta,
2473 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2474 0, 0, sysctl_kern_max_unsafe_rt_quanta, "I",
2475 "Number of quanta to allow a realtime "
2476 "thread to run before being penalized");
2477
2478 static int
2479 sysctl_kern_max_unsafe_fixed_quanta(__unused struct sysctl_oid *oidp,
2480 __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2481 {
2482 extern void sched_set_max_unsafe_fixed_quanta(int);
2483 extern int max_unsafe_fixed_quanta;
2484
2485 int new_value, changed;
2486 int old_value = max_unsafe_fixed_quanta;
2487 int error = sysctl_io_number(req, old_value, sizeof(int), &new_value,
2488 &changed);
2489 if (changed) {
2490 sched_set_max_unsafe_fixed_quanta(new_value);
2491 }
2492
2493 return error;
2494 }
2495
2496 SYSCTL_PROC(_kern, OID_AUTO, max_unsafe_fixed_quanta,
2497 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2498 0, 0, sysctl_kern_max_unsafe_fixed_quanta, "I",
2499 "Number of quanta to allow a fixed sched mode "
2500 "thread to run before being penalized");
2501
2502 static int
2503 sysctl_kern_quantum_us(__unused struct sysctl_oid *oidp, __unused void *arg1,
2504 __unused int arg2, struct sysctl_req *req)
2505 {
2506 const uint64_t quantum_us = sched_get_quantum_us();
2507
2508 return sysctl_io_number(req, quantum_us, sizeof(quantum_us), NULL, NULL);
2509 }
2510
2511 SYSCTL_PROC(_kern, OID_AUTO, quantum_us,
2512 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
2513 0, 0, sysctl_kern_quantum_us, "Q",
2514 "Length of scheduling quantum in microseconds");
2515
2516 extern int smt_sched_bonus_16ths;
2517 SYSCTL_INT(_kern, OID_AUTO, smt_sched_bonus_16ths,
2518 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2519 &smt_sched_bonus_16ths, 0, "");
2520
2521 extern int smt_timeshare_enabled;
2522 SYSCTL_INT(_kern, OID_AUTO, sched_smt_timeshare_enable,
2523 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2524 &smt_timeshare_enabled, 0, "");
2525
2526 extern int sched_smt_balance;
2527 SYSCTL_INT(_kern, OID_AUTO, sched_smt_balance,
2528 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2529 &sched_smt_balance, 0, "");
2530 extern int sched_allow_rt_smt;
2531 SYSCTL_INT(_kern, OID_AUTO, sched_allow_rt_smt,
2532 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2533 &sched_allow_rt_smt, 0, "");
2534 extern int sched_backup_cpu_timeout_count;
2535 SYSCTL_INT(_kern, OID_AUTO, sched_backup_cpu_timeout_count,
2536 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2537 &sched_backup_cpu_timeout_count, 0, "The maximum number of 10us delays before allowing a backup cpu to select a thread");
2538 #if __arm64__
2539 /* Scheduler perfcontrol callouts sysctls */
2540 SYSCTL_DECL(_kern_perfcontrol_callout);
2541 SYSCTL_NODE(_kern, OID_AUTO, perfcontrol_callout, CTLFLAG_RW | CTLFLAG_LOCKED, 0,
2542 "scheduler perfcontrol callouts");
2543
2544 extern int perfcontrol_callout_stats_enabled;
2545 SYSCTL_INT(_kern_perfcontrol_callout, OID_AUTO, stats_enabled,
2546 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2547 &perfcontrol_callout_stats_enabled, 0, "");
2548
2549 extern uint64_t perfcontrol_callout_stat_avg(perfcontrol_callout_type_t type,
2550 perfcontrol_callout_stat_t stat);
2551
2552 /* On-Core Callout */
2553 STATIC int
2554 sysctl_perfcontrol_callout_stat
2555 (__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2556 {
2557 perfcontrol_callout_stat_t stat = (perfcontrol_callout_stat_t)arg1;
2558 perfcontrol_callout_type_t type = (perfcontrol_callout_type_t)arg2;
2559 return sysctl_io_number(req, (int)perfcontrol_callout_stat_avg(type, stat),
2560 sizeof(int), NULL, NULL);
2561 }
2562
2563 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, oncore_instr,
2564 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2565 (void *)PERFCONTROL_STAT_INSTRS, PERFCONTROL_CALLOUT_ON_CORE,
2566 sysctl_perfcontrol_callout_stat, "I", "");
2567 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, oncore_cycles,
2568 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2569 (void *)PERFCONTROL_STAT_CYCLES, PERFCONTROL_CALLOUT_ON_CORE,
2570 sysctl_perfcontrol_callout_stat, "I", "");
2571 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, offcore_instr,
2572 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2573 (void *)PERFCONTROL_STAT_INSTRS, PERFCONTROL_CALLOUT_OFF_CORE,
2574 sysctl_perfcontrol_callout_stat, "I", "");
2575 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, offcore_cycles,
2576 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2577 (void *)PERFCONTROL_STAT_CYCLES, PERFCONTROL_CALLOUT_OFF_CORE,
2578 sysctl_perfcontrol_callout_stat, "I", "");
2579 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, context_instr,
2580 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2581 (void *)PERFCONTROL_STAT_INSTRS, PERFCONTROL_CALLOUT_CONTEXT,
2582 sysctl_perfcontrol_callout_stat, "I", "");
2583 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, context_cycles,
2584 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2585 (void *)PERFCONTROL_STAT_CYCLES, PERFCONTROL_CALLOUT_CONTEXT,
2586 sysctl_perfcontrol_callout_stat, "I", "");
2587 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, update_instr,
2588 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2589 (void *)PERFCONTROL_STAT_INSTRS, PERFCONTROL_CALLOUT_STATE_UPDATE,
2590 sysctl_perfcontrol_callout_stat, "I", "");
2591 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, update_cycles,
2592 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2593 (void *)PERFCONTROL_STAT_CYCLES, PERFCONTROL_CALLOUT_STATE_UPDATE,
2594 sysctl_perfcontrol_callout_stat, "I", "");
2595
2596 #if __AMP__
2597 #if !CONFIG_CLUTCH
2598 extern int sched_amp_idle_steal;
2599 SYSCTL_INT(_kern, OID_AUTO, sched_amp_idle_steal,
2600 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2601 &sched_amp_idle_steal, 0, "");
2602 extern int sched_amp_spill_steal;
2603 SYSCTL_INT(_kern, OID_AUTO, sched_amp_spill_steal,
2604 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2605 &sched_amp_spill_steal, 0, "");
2606 extern int sched_amp_spill_count;
2607 SYSCTL_INT(_kern, OID_AUTO, sched_amp_spill_count,
2608 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2609 &sched_amp_spill_count, 0, "");
2610 #endif /* !CONFIG_CLUTCH */
2611 extern int sched_amp_spill_deferred_ipi;
2612 SYSCTL_INT(_kern, OID_AUTO, sched_amp_spill_deferred_ipi,
2613 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2614 &sched_amp_spill_deferred_ipi, 0, "");
2615 extern int sched_amp_pcores_preempt_immediate_ipi;
2616 SYSCTL_INT(_kern, OID_AUTO, sched_amp_pcores_preempt_immediate_ipi,
2617 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2618 &sched_amp_pcores_preempt_immediate_ipi, 0, "");
2619 #endif /* __AMP__ */
2620 #endif /* __arm64__ */
2621
2622 #if __arm64__
2623 extern int legacy_footprint_entitlement_mode;
2624 SYSCTL_INT(_kern, OID_AUTO, legacy_footprint_entitlement_mode,
2625 CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
2626 &legacy_footprint_entitlement_mode, 0, "");
2627 #endif /* __arm64__ */
2628
2629 /*
2630 * Realtime threads are ordered by highest priority first then,
2631 * for threads of the same priority, by earliest deadline first.
2632 * But if sched_rt_runq_strict_priority is false (the default),
2633 * a lower priority thread with an earlier deadline will be preferred
2634 * over a higher priority thread with a later deadline, as long as
2635 * both threads' computations will fit before the later deadline.
2636 */
2637 extern int sched_rt_runq_strict_priority;
2638 SYSCTL_INT(_kern, OID_AUTO, sched_rt_runq_strict_priority,
2639 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2640 &sched_rt_runq_strict_priority, 0, "");
2641
2642 static int
2643 sysctl_kern_sched_rt_n_backup_processors(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2644 {
2645 int new_value, changed;
2646 int old_value = sched_get_rt_n_backup_processors();
2647 int error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
2648 if (changed) {
2649 sched_set_rt_n_backup_processors(new_value);
2650 }
2651
2652 return error;
2653 }
2654
2655 SYSCTL_PROC(_kern, OID_AUTO, sched_rt_n_backup_processors,
2656 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2657 0, 0, sysctl_kern_sched_rt_n_backup_processors, "I", "");
2658
2659 static int
2660 sysctl_kern_sched_rt_deadline_epsilon_us(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2661 {
2662 int new_value, changed;
2663 int old_value = sched_get_rt_deadline_epsilon();
2664 int error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
2665 if (changed) {
2666 sched_set_rt_deadline_epsilon(new_value);
2667 }
2668
2669 return error;
2670 }
2671
2672 SYSCTL_PROC(_kern, OID_AUTO, sched_rt_deadline_epsilon_us,
2673 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2674 0, 0, sysctl_kern_sched_rt_deadline_epsilon_us, "I", "");
2675
2676 extern int sched_idle_delay_cpuid;
2677 SYSCTL_INT(_kern, OID_AUTO, sched_idle_delay_cpuid,
2678 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2679 &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");
2680
2681 static int
2682 sysctl_kern_sched_powered_cores(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2683 {
2684 int new_value, changed;
2685 int old_value = sched_get_powered_cores();
2686 int error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
2687 if (changed) {
2688 if (!PE_parse_boot_argn("enable_skstb", NULL, 0)) {
2689 return ENOTSUP;
2690 }
2691
2692 sched_set_powered_cores(new_value);
2693 }
2694
2695 return error;
2696 }
2697
2698 SYSCTL_PROC(_kern, OID_AUTO, sched_powered_cores,
2699 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2700 0, 0, sysctl_kern_sched_powered_cores, "I", "");
2701
2702 #if __arm64__
2703
2704 static int
2705 sysctl_kern_update_sched_recommended_cores(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2706 {
2707 uint64_t new_value;
2708 int changed;
2709 uint64_t old_value = sched_sysctl_get_recommended_cores();
2710 int error = sysctl_io_number(req, old_value, sizeof(uint64_t), &new_value, &changed);
2711 if (changed) {
2712 if (!PE_parse_boot_argn("enable_skstb", NULL, 0)) {
2713 return ENOTSUP;
2714 }
2715
2716 sched_perfcontrol_update_recommended_cores_reason(new_value, REASON_CLPC_USER, 0);
2717 }
2718
2719 return error;
2720 }
2721
2722 SYSCTL_PROC(_kern, OID_AUTO, sched_update_recommended_cores,
2723 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
2724 0, 0, sysctl_kern_update_sched_recommended_cores, "I", "");
2725
2726 #endif /* __arm64__ */
2727
2728 #endif /* (DEVELOPMENT || DEBUG) */
2729
2730 extern uint64_t sysctl_sched_recommended_cores;
2731 SYSCTL_QUAD(_kern, OID_AUTO, sched_recommended_cores,
2732 CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
2733 &sysctl_sched_recommended_cores, "");
2734
2735 static int
2736 sysctl_kern_suspend_cluster_powerdown(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2737 {
2738 int new_value, changed;
2739 int old_value = get_cluster_powerdown_user_suspended();
2740 int error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
2741 if (!error && changed) {
2742 if (new_value > 0) {
2743 error = suspend_cluster_powerdown_from_user();
2744 } else {
2745 error = resume_cluster_powerdown_from_user();
2746 }
2747 if (error) {
2748 error = EALREADY;
2749 }
2750 }
2751
2752 return error;
2753 }
2754
2755 SYSCTL_PROC(_kern, OID_AUTO, suspend_cluster_powerdown,
2756 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2757 0, 0, sysctl_kern_suspend_cluster_powerdown, "I", "");
2758
2759
2760 STATIC int
2761 sysctl_securelvl
2762 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2763 {
2764 int new_value, changed;
2765 int error = sysctl_io_number(req, securelevel, sizeof(int), &new_value, &changed);
2766 if (changed) {
2767 if (!(new_value < securelevel && proc_getpid(req->p) != 1)) {
2768 proc_list_lock();
2769 securelevel = new_value;
2770 proc_list_unlock();
2771 } else {
2772 error = EPERM;
2773 }
2774 }
2775 return error;
2776 }
2777
2778 SYSCTL_PROC(_kern, KERN_SECURELVL, securelevel,
2779 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2780 0, 0, sysctl_securelvl, "I", "");
2781
2782
2783 STATIC int
2784 sysctl_domainname
2785 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2786 {
2787 int error, changed;
2788 char tmpname[MAXHOSTNAMELEN] = {};
2789
2790 lck_mtx_lock(&domainname_lock);
2791 strlcpy(tmpname, domainname, sizeof(tmpname));
2792 lck_mtx_unlock(&domainname_lock);
2793
2794 error = sysctl_io_string(req, tmpname, sizeof(tmpname), 0, &changed);
2795 if (!error && changed) {
2796 lck_mtx_lock(&domainname_lock);
2797 strlcpy(domainname, tmpname, sizeof(domainname));
2798 lck_mtx_unlock(&domainname_lock);
2799 }
2800 return error;
2801 }
2802
2803 SYSCTL_PROC(_kern, KERN_DOMAINNAME, nisdomainname,
2804 CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_LOCKED,
2805 0, 0, sysctl_domainname, "A", "");
2806
2807 SYSCTL_COMPAT_INT(_kern, KERN_HOSTID, hostid,
2808 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2809 &hostid, 0, "");
2810
2811 STATIC int
2812 sysctl_hostname
2813 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2814 {
2815 int error, changed;
2816 char tmpname[MAXHOSTNAMELEN] = {};
2817 const char * name;
2818
2819 #if XNU_TARGET_OS_OSX
2820 name = hostname;
2821 #else /* XNU_TARGET_OS_OSX */
2822 #define ENTITLEMENT_USER_ASSIGNED_DEVICE_NAME \
2823 "com.apple.developer.device-information.user-assigned-device-name"
2824 if (csproc_get_platform_binary(current_proc()) ||
2825 IOCurrentTaskHasEntitlement(ENTITLEMENT_USER_ASSIGNED_DEVICE_NAME)) {
2826 name = hostname;
2827 } else {
2828 /* Deny writes if we don't pass entitlement check */
2829 if (req->newptr) {
2830 return EPERM;
2831 }
2832
2833 name = "localhost";
2834 }
2835 #endif /* ! XNU_TARGET_OS_OSX */
2836
2837 lck_mtx_lock(&hostname_lock);
2838 strlcpy(tmpname, name, sizeof(tmpname));
2839 lck_mtx_unlock(&hostname_lock);
2840
2841 error = sysctl_io_string(req, tmpname, sizeof(tmpname), 1, &changed);
2842 if (!error && changed) {
2843 lck_mtx_lock(&hostname_lock);
2844 strlcpy(hostname, tmpname, sizeof(hostname));
2845 lck_mtx_unlock(&hostname_lock);
2846 }
2847 return error;
2848 }
2849
2850 SYSCTL_PROC(_kern, KERN_HOSTNAME, hostname,
2851 CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_LOCKED,
2852 0, 0, sysctl_hostname, "A", "");
2853
2854 STATIC int
2855 sysctl_procname
2856 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2857 {
2858 /* Original code allowed writing, I'm copying this, although this all makes
2859 * no sense to me. Besides, this sysctl is never used. */
2860 return sysctl_io_string(req, &req->p->p_name[0], (2 * MAXCOMLEN + 1), 1, NULL);
2861 }
2862
2863 SYSCTL_PROC(_kern, KERN_PROCNAME, procname,
2864 CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
2865 0, 0, sysctl_procname, "A", "");
2866
2867 SYSCTL_INT(_kern, KERN_SPECULATIVE_READS, speculative_reads_disabled,
2868 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2869 &speculative_reads_disabled, 0, "");
2870
2871 SYSCTL_UINT(_kern, OID_AUTO, preheat_max_bytes,
2872 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2873 &preheat_max_bytes, 0, "");
2874
2875 SYSCTL_UINT(_kern, OID_AUTO, preheat_min_bytes,
2876 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2877 &preheat_min_bytes, 0, "");
2878
2879 SYSCTL_UINT(_kern, OID_AUTO, speculative_prefetch_max,
2880 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2881 &speculative_prefetch_max, 0, "");
2882
2883 SYSCTL_UINT(_kern, OID_AUTO, speculative_prefetch_max_iosize,
2884 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2885 &speculative_prefetch_max_iosize, 0, "");
2886
2887 SYSCTL_UINT(_kern, OID_AUTO, vm_page_free_target,
2888 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2889 &vm_page_free_target, 0, "");
2890
2891 SYSCTL_UINT(_kern, OID_AUTO, vm_page_free_min,
2892 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2893 &vm_page_free_min, 0, "");
2894
2895 SYSCTL_UINT(_kern, OID_AUTO, vm_page_free_reserved,
2896 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2897 &vm_page_free_reserved, 0, "");
2898
2899 SYSCTL_UINT(_kern, OID_AUTO, vm_page_speculative_percentage,
2900 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2901 &vm_pageout_state.vm_page_speculative_percentage, 0, "");
2902
2903 SYSCTL_UINT(_kern, OID_AUTO, vm_page_speculative_q_age_ms,
2904 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2905 &vm_pageout_state.vm_page_speculative_q_age_ms, 0, "");
2906
2907 #if (DEVELOPMENT || DEBUG)
2908 SYSCTL_UINT(_kern, OID_AUTO, vm_page_max_speculative_age_q,
2909 CTLFLAG_RD,
2910 &vm_page_max_speculative_age_q, 0, "");
2911 #endif /* (DEVELOPMENT || DEBUG) */
2912
2913 SYSCTL_UINT(_kern, OID_AUTO, vm_max_delayed_work_limit,
2914 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2915 &vm_max_delayed_work_limit, 0, "");
2916
2917 SYSCTL_UINT(_kern, OID_AUTO, vm_max_batch,
2918 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2919 &vm_max_batch, 0, "");
2920
2921 SYSCTL_STRING(_kern, OID_AUTO, bootsessionuuid,
2922 CTLFLAG_RD | CTLFLAG_LOCKED,
2923 &bootsessionuuid_string, sizeof(bootsessionuuid_string), "");
2924
2925
2926 STATIC int
2927 sysctl_boottime
2928 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2929 {
2930 struct timeval tv;
2931 boottime_timeval(&tv);
2932 struct proc *p = req->p;
2933
2934 if (proc_is64bit(p)) {
2935 struct user64_timeval t = {};
2936 t.tv_sec = tv.tv_sec;
2937 t.tv_usec = tv.tv_usec;
2938 return sysctl_io_opaque(req, &t, sizeof(t), NULL);
2939 } else {
2940 struct user32_timeval t = {};
2941 t.tv_sec = (user32_time_t)tv.tv_sec;
2942 t.tv_usec = tv.tv_usec;
2943 return sysctl_io_opaque(req, &t, sizeof(t), NULL);
2944 }
2945 }
2946
2947 SYSCTL_PROC(_kern, KERN_BOOTTIME, boottime,
2948 CTLTYPE_STRUCT | CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
2949 0, 0, sysctl_boottime, "S,timeval", "");
2950
2951 extern bool IOGetBootUUID(char *);
2952
2953 /* non-static: written by imageboot.c */
2954 uuid_string_t fake_bootuuid;
2955
2956 STATIC int
2957 sysctl_bootuuid
2958 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2959 {
2960 int error = ENOENT;
2961
2962 /* check the first byte to see if the string has been
2963 * populated. this is a uuid_STRING_t, this check would
2964 * not work with a uuid_t.
2965 */
2966 if (fake_bootuuid[0] != '\0') {
2967 error = sysctl_io_string(req, fake_bootuuid, 0, 0, NULL);
2968 goto out;
2969 }
2970
2971 uuid_string_t uuid_string;
2972 if (IOGetBootUUID(uuid_string)) {
2973 uuid_t boot_uuid;
2974 error = uuid_parse(uuid_string, boot_uuid);
2975 if (!error) {
2976 error = sysctl_io_string(req, __DECONST(char *, uuid_string), 0, 0, NULL);
2977 }
2978 }
2979
2980 out:
2981 return error;
2982 }
2983
2984 SYSCTL_PROC(_kern, OID_AUTO, bootuuid,
2985 CTLTYPE_STRING | CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
2986 0, 0, sysctl_bootuuid, "A", "");
2987
2988
2989 extern bool IOGetApfsPrebootUUID(char *);
2990 extern bool IOGetAssociatedApfsVolgroupUUID(char *);
2991
2992 STATIC int
2993 sysctl_apfsprebootuuid
2994 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2995 {
2996 int error = ENOENT;
2997
2998 uuid_string_t uuid_string;
2999 if (IOGetApfsPrebootUUID(uuid_string)) {
3000 uuid_t apfs_preboot_uuid;
3001 error = uuid_parse(uuid_string, apfs_preboot_uuid);
3002 if (!error) {
3003 error = sysctl_io_string(req, __DECONST(char *, uuid_string), 0, 0, NULL);
3004 }
3005 }
3006
3007 return error;
3008 }
3009
3010 SYSCTL_PROC(_kern, OID_AUTO, apfsprebootuuid,
3011 CTLTYPE_STRING | CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
3012 0, 0, sysctl_apfsprebootuuid, "A", "");
3013
3014 STATIC int
3015 sysctl_targetsystemvolgroupuuid
3016 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3017 {
3018 int error = ENOENT;
3019
3020 uuid_string_t uuid_string;
3021 if (IOGetApfsPrebootUUID(uuid_string)) {
3022 uuid_t apfs_preboot_uuid;
3023 error = uuid_parse(uuid_string, apfs_preboot_uuid);
3024 if (!error) {
3025 error = sysctl_io_string(req, __DECONST(char *, uuid_string), 0, 0, NULL);
3026 }
3027 } else {
3028 /*
3029 * In special boot modes, such as kcgen-mode, the
3030 * apfs-preboot-uuid property will not be set. Instead, a
3031 * different property, associated-volume-group, will be set
3032 * which indicates the UUID of the VolumeGroup containing the
3033 * system volume into which you will boot.
3034 */
3035 if (IOGetAssociatedApfsVolgroupUUID(uuid_string)) {
3036 uuid_t apfs_preboot_uuid;
3037 error = uuid_parse(uuid_string, apfs_preboot_uuid);
3038 if (!error) {
3039 error = sysctl_io_string(req, __DECONST(char *, uuid_string), 0, 0, NULL);
3040 }
3041 }
3042 }
3043
3044 return error;
3045 }
3046
3047 SYSCTL_PROC(_kern, OID_AUTO, targetsystemvolgroupuuid,
3048 CTLTYPE_STRING | CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_MASKED | CTLFLAG_LOCKED,
3049 0, 0, sysctl_targetsystemvolgroupuuid, "A", "");
3050
3051
3052 extern bool IOGetBootManifestHash(char *, size_t *);
3053 extern bool IOGetBootObjectsPath(char *);
3054
3055 STATIC int
3056 sysctl_bootobjectspath
3057 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3058 {
3059 int error = ENOENT;
3060
3061 #if defined(__x86_64__)
3062 /* auth-root-dmg is used for the Intel BaseSystem in some flows,
3063 * e.g. createinstallmedia and as part of upgrading from 10.15 or earlier
3064 * under these scenarios, set_fake_bootuuid will be called when pivoting to
3065 * the new root filesystem. need honor the fake bootuuid.
3066 */
3067 if (fake_bootuuid[0] != '\0') {
3068 error = sysctl_io_string(req, fake_bootuuid, 0, 0, NULL);
3069 } else {
3070 /* for intel mac, boot objects reside in [preboot volume]/[bootuuid]
3071 * bootuuid and apfsprebootuuid are populated by efiboot and they are alias.
3072 */
3073 uuid_string_t uuid_string;
3074 if (IOGetBootUUID(uuid_string)) {
3075 uuid_t boot_uuid;
3076 error = uuid_parse(uuid_string, boot_uuid);
3077 if (!error) {
3078 error = sysctl_io_string(req, (char *)uuid_string, 0, 0, NULL);
3079 }
3080 }
3081 }
3082 #else
3083 char boot_obj_path[MAXPATHLEN] = { "\0" };
3084 static const char kAsciiHexChars[] = "0123456789ABCDEF";
3085 unsigned int i, j;
3086
3087 /* Hashed with SHA2-384 or SHA1, boot manifest hash is 48 bytes or 20 bytes
3088 * hence, need a 97 bytes char array for the string.
3089 */
3090 size_t hash_data_size = CCSHA384_OUTPUT_SIZE;
3091 char hash_data[CCSHA384_OUTPUT_SIZE] = { "\0" };
3092 char boot_manifest_hash[CCSHA384_OUTPUT_SIZE * 2 + 1] = { "\0" };;
3093
3094 /* for Apple Silicon Macs, there is a boot-objects-path under IODeviceTree:/chosen
3095 * and boot objects reside in [preboot volume]/[boot-objects-path]
3096 * for embedded platforms, there would be a boot-manifest-hash under IODeviceTree:/chosen
3097 * and boot objects reside in [preboot volume]/[boot-manifest-hash]
3098 */
3099 if (IOGetBootObjectsPath(boot_obj_path)) {
3100 error = sysctl_io_string(req, (char *)boot_obj_path, 0, 0, NULL);
3101 } else if (IOGetBootManifestHash(hash_data, &hash_data_size)) {
3102 j = 0;
3103 for (i = 0; i < hash_data_size; ++i) {
3104 char octet = hash_data[i];
3105 boot_manifest_hash[j++] = kAsciiHexChars[((octet & 0xF0) >> 4)];
3106 boot_manifest_hash[j++] = kAsciiHexChars[(octet & 0x0F)];
3107 }
3108 /* make sure string has null termination */
3109 boot_manifest_hash[j] = '\0';
3110 error = sysctl_io_string(req, (char *)boot_manifest_hash, 0, 0, NULL);
3111 }
3112 #endif
3113 return error;
3114 }
3115
3116 SYSCTL_PROC(_kern, OID_AUTO, bootobjectspath,
3117 CTLTYPE_STRING | CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
3118 0, 0, sysctl_bootobjectspath, "A", "");
3119
3120
3121 STATIC int
3122 sysctl_symfile
3123 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3124 {
3125 char *str;
3126 int error = get_kernel_symfile(req->p, &str);
3127 if (error) {
3128 return error;
3129 }
3130 return sysctl_io_string(req, str, 0, 0, NULL);
3131 }
3132
3133
3134 SYSCTL_PROC(_kern, KERN_SYMFILE, symfile,
3135 CTLTYPE_STRING | CTLFLAG_RD | CTLFLAG_LOCKED,
3136 0, 0, sysctl_symfile, "A", "");
3137
3138 #if CONFIG_NETBOOT
3139 STATIC int
3140 sysctl_netboot
3141 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3142 {
3143 return sysctl_io_number(req, netboot_root(), sizeof(int), NULL, NULL);
3144 }
3145
3146 SYSCTL_PROC(_kern, KERN_NETBOOT, netboot,
3147 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
3148 0, 0, sysctl_netboot, "I", "");
3149 #endif
3150
3151 #ifdef CONFIG_IMGSRC_ACCESS
3152 /*
3153 * Legacy--act as if only one layer of nesting is possible.
3154 */
3155 STATIC int
3156 sysctl_imgsrcdev
3157 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3158 {
3159 vfs_context_t ctx = vfs_context_current();
3160 vnode_t devvp;
3161 int result;
3162
3163 if (!vfs_context_issuser(ctx)) {
3164 return EPERM;
3165 }
3166
3167 if (imgsrc_rootvnodes[0] == NULL) {
3168 return ENOENT;
3169 }
3170
3171 result = vnode_getwithref(imgsrc_rootvnodes[0]);
3172 if (result != 0) {
3173 return result;
3174 }
3175
3176 devvp = vnode_mount(imgsrc_rootvnodes[0])->mnt_devvp;
3177 result = vnode_getwithref(devvp);
3178 if (result != 0) {
3179 goto out;
3180 }
3181
3182 result = sysctl_io_number(req, vnode_specrdev(devvp), sizeof(dev_t), NULL, NULL);
3183
3184 vnode_put(devvp);
3185 out:
3186 vnode_put(imgsrc_rootvnodes[0]);
3187 return result;
3188 }
3189
3190 SYSCTL_PROC(_kern, OID_AUTO, imgsrcdev,
3191 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
3192 0, 0, sysctl_imgsrcdev, "I", "");
3193
3194 STATIC int
3195 sysctl_imgsrcinfo
3196 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3197 {
3198 int error;
3199 struct imgsrc_info info[MAX_IMAGEBOOT_NESTING] = {}; /* 2 for now, no problem */
3200 uint32_t i;
3201 vnode_t rvp, devvp;
3202
3203 if (imgsrc_rootvnodes[0] == NULLVP) {
3204 return ENXIO;
3205 }
3206
3207 for (i = 0; i < MAX_IMAGEBOOT_NESTING; i++) {
3208 /*
3209 * Go get the root vnode.
3210 */
3211 rvp = imgsrc_rootvnodes[i];
3212 if (rvp == NULLVP) {
3213 break;
3214 }
3215
3216 error = vnode_get(rvp);
3217 if (error != 0) {
3218 return error;
3219 }
3220
3221 /*
3222 * For now, no getting at a non-local volume.
3223 */
3224 devvp = vnode_mount(rvp)->mnt_devvp;
3225 if (devvp == NULL) {
3226 vnode_put(rvp);
3227 return EINVAL;
3228 }
3229
3230 error = vnode_getwithref(devvp);
3231 if (error != 0) {
3232 vnode_put(rvp);
3233 return error;
3234 }
3235
3236 /*
3237 * Fill in info.
3238 */
3239 info[i].ii_dev = vnode_specrdev(devvp);
3240 info[i].ii_flags = 0;
3241 info[i].ii_height = i;
3242 bzero(info[i].ii_reserved, sizeof(info[i].ii_reserved));
3243
3244 vnode_put(devvp);
3245 vnode_put(rvp);
3246 }
3247
3248 return sysctl_io_opaque(req, info, i * sizeof(info[0]), NULL);
3249 }
3250
3251 SYSCTL_PROC(_kern, OID_AUTO, imgsrcinfo,
3252 CTLTYPE_OPAQUE | CTLFLAG_RD | CTLFLAG_LOCKED,
3253 0, 0, sysctl_imgsrcinfo, "I", "");
3254
3255 #endif /* CONFIG_IMGSRC_ACCESS */
3256
3257
3258 SYSCTL_DECL(_kern_timer);
3259 SYSCTL_NODE(_kern, OID_AUTO, timer, CTLFLAG_RW | CTLFLAG_LOCKED, 0, "timer");
3260
3261
3262 SYSCTL_INT(_kern_timer, OID_AUTO, coalescing_enabled,
3263 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
3264 &mach_timer_coalescing_enabled, 0, "");
3265
3266 SYSCTL_QUAD(_kern_timer, OID_AUTO, deadline_tracking_bin_1,
3267 CTLFLAG_RW | CTLFLAG_LOCKED,
3268 &timer_deadline_tracking_bin_1, "");
3269 SYSCTL_QUAD(_kern_timer, OID_AUTO, deadline_tracking_bin_2,
3270 CTLFLAG_RW | CTLFLAG_LOCKED,
3271 &timer_deadline_tracking_bin_2, "");
3272
3273 SYSCTL_DECL(_kern_timer_longterm);
3274 SYSCTL_NODE(_kern_timer, OID_AUTO, longterm, CTLFLAG_RW | CTLFLAG_LOCKED, 0, "longterm");
3275
3276
3277 /* Must match definition in osfmk/kern/timer_call.c */
3278 enum {
3279 THRESHOLD, QCOUNT,
3280 ENQUEUES, DEQUEUES, ESCALATES, SCANS, PREEMPTS,
3281 LATENCY, LATENCY_MIN, LATENCY_MAX, LONG_TERM_SCAN_LIMIT,
3282 LONG_TERM_SCAN_INTERVAL, LONG_TERM_SCAN_PAUSES,
3283 SCAN_LIMIT, SCAN_INTERVAL, SCAN_PAUSES, SCAN_POSTPONES,
3284 };
3285 extern uint64_t timer_sysctl_get(int);
3286 extern kern_return_t timer_sysctl_set(int, uint64_t);
3287
3288 STATIC int
3289 sysctl_timer
3290 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3291 {
3292 int oid = (int)arg1;
3293 uint64_t value = timer_sysctl_get(oid);
3294 uint64_t new_value;
3295 int error;
3296 int changed;
3297
3298 error = sysctl_io_number(req, value, sizeof(value), &new_value, &changed);
3299 if (changed) {
3300 kern_return_t kr = timer_sysctl_set(oid, new_value);
3301 error = mach_to_bsd_errno(kr);
3302 }
3303
3304 return error;
3305 }
3306
3307 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, threshold,
3308 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
3309 (void *) THRESHOLD, 0, sysctl_timer, "Q", "");
3310 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, scan_limit,
3311 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
3312 (void *) LONG_TERM_SCAN_LIMIT, 0, sysctl_timer, "Q", "");
3313 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, scan_interval,
3314 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
3315 (void *) LONG_TERM_SCAN_INTERVAL, 0, sysctl_timer, "Q", "");
3316
3317 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, qlen,
3318 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3319 (void *) QCOUNT, 0, sysctl_timer, "Q", "");
3320 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, scan_pauses,
3321 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3322 (void *) LONG_TERM_SCAN_PAUSES, 0, sysctl_timer, "Q", "");
3323
3324 #if DEBUG
3325 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, enqueues,
3326 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3327 (void *) ENQUEUES, 0, sysctl_timer, "Q", "");
3328 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, dequeues,
3329 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3330 (void *) DEQUEUES, 0, sysctl_timer, "Q", "");
3331 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, escalates,
3332 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3333 (void *) ESCALATES, 0, sysctl_timer, "Q", "");
3334 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, scans,
3335 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3336 (void *) SCANS, 0, sysctl_timer, "Q", "");
3337 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, preempts,
3338 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3339 (void *) PREEMPTS, 0, sysctl_timer, "Q", "");
3340 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, latency,
3341 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3342 (void *) LATENCY, 0, sysctl_timer, "Q", "");
3343 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, latency_min,
3344 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3345 (void *) LATENCY_MIN, 0, sysctl_timer, "Q", "");
3346 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, latency_max,
3347 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3348 (void *) LATENCY_MAX, 0, sysctl_timer, "Q", "");
3349 #endif /* DEBUG */
3350
3351 SYSCTL_PROC(_kern_timer, OID_AUTO, scan_limit,
3352 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
3353 (void *) SCAN_LIMIT, 0, sysctl_timer, "Q", "");
3354 SYSCTL_PROC(_kern_timer, OID_AUTO, scan_interval,
3355 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
3356 (void *) SCAN_INTERVAL, 0, sysctl_timer, "Q", "");
3357 SYSCTL_PROC(_kern_timer, OID_AUTO, scan_pauses,
3358 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3359 (void *) SCAN_PAUSES, 0, sysctl_timer, "Q", "");
3360 SYSCTL_PROC(_kern_timer, OID_AUTO, scan_postpones,
3361 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3362 (void *) SCAN_POSTPONES, 0, sysctl_timer, "Q", "");
3363
3364 STATIC int
3365 sysctl_usrstack
3366 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3367 {
3368 return sysctl_io_number(req, (int)req->p->user_stack, sizeof(int), NULL, NULL);
3369 }
3370
3371 SYSCTL_PROC(_kern, KERN_USRSTACK32, usrstack,
3372 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
3373 0, 0, sysctl_usrstack, "I", "");
3374
3375 STATIC int
3376 sysctl_usrstack64
3377 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3378 {
3379 return sysctl_io_number(req, req->p->user_stack, sizeof(req->p->user_stack), NULL, NULL);
3380 }
3381
3382 SYSCTL_PROC(_kern, KERN_USRSTACK64, usrstack64,
3383 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3384 0, 0, sysctl_usrstack64, "Q", "");
3385
3386
3387 #if EXCLAVES_COREDUMP
3388
3389 /* secure kernel coredump support. */
3390 extern unsigned int sc_dump_mode;
3391 SYSCTL_UINT(_kern, OID_AUTO, secure_coredump, CTLFLAG_RD, &sc_dump_mode, 0, "secure_coredump");
3392
3393 #endif /* EXCLAVES_COREDUMP */
3394
3395 #if CONFIG_COREDUMP || CONFIG_UCOREDUMP
3396
3397 SYSCTL_STRING(_kern, KERN_COREFILE, corefile,
3398 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
3399 corefilename, sizeof(corefilename), "");
3400
3401 SYSCTL_STRING(_kern, OID_AUTO, drivercorefile,
3402 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
3403 drivercorefilename, sizeof(drivercorefilename), "");
3404
3405 STATIC int
3406 sysctl_coredump
3407 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3408 {
3409 #ifdef SECURE_KERNEL
3410 (void)req;
3411 return ENOTSUP;
3412 #else
3413 int new_value, changed;
3414 int error = sysctl_io_number(req, do_coredump, sizeof(int), &new_value, &changed);
3415 if (changed) {
3416 if ((new_value == 0) || (new_value == 1)) {
3417 do_coredump = new_value;
3418 } else {
3419 error = EINVAL;
3420 }
3421 }
3422 return error;
3423 #endif
3424 }
3425
3426 SYSCTL_PROC(_kern, KERN_COREDUMP, coredump,
3427 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
3428 0, 0, sysctl_coredump, "I", "");
3429
3430 STATIC int
3431 sysctl_suid_coredump
3432 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3433 {
3434 #ifdef SECURE_KERNEL
3435 (void)req;
3436 return ENOTSUP;
3437 #else
3438 int new_value, changed;
3439 int error = sysctl_io_number(req, sugid_coredump, sizeof(int), &new_value, &changed);
3440 if (changed) {
3441 if ((new_value == 0) || (new_value == 1)) {
3442 sugid_coredump = new_value;
3443 } else {
3444 error = EINVAL;
3445 }
3446 }
3447 return error;
3448 #endif
3449 }
3450
3451 SYSCTL_PROC(_kern, KERN_SUGID_COREDUMP, sugid_coredump,
3452 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
3453 0, 0, sysctl_suid_coredump, "I", "");
3454
3455 #if CONFIG_UCOREDUMP
3456
3457 STATIC int
3458 sysctl_ucoredump
3459 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3460 {
3461 #ifdef SECURE_KERNEL
3462 (void)req;
3463 return ENOTSUP;
3464 #else
3465 int new_value, changed;
3466 int error = sysctl_io_number(req, do_ucoredump, sizeof(int), &new_value, &changed);
3467 if (changed) {
3468 if (new_value == 0 || new_value == 1) {
3469 do_ucoredump = new_value;
3470 } else {
3471 error = EINVAL;
3472 }
3473 }
3474 return error;
3475 #endif
3476 }
3477
3478 SYSCTL_PROC(_kern, OID_AUTO, ucoredump,
3479 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
3480 0, 0, sysctl_ucoredump, "I", "");
3481 #endif /* CONFIG_UCOREDUMP */
3482 #endif /* CONFIG_COREDUMP || CONFIG_UCOREDUMP */
3483
3484 #if CONFIG_KDP_INTERACTIVE_DEBUGGING
3485
3486 extern const char* kdp_corefile_path;
3487 STATIC int
3488 sysctl_kdp_corefile(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3489 {
3490 return SYSCTL_OUT(req, kdp_corefile_path, strlen(kdp_corefile_path) + 1);
3491 }
3492
3493 /* this needs to be a proc rather than a string since kdp_corefile_path is not a compile-time constant */
3494 SYSCTL_PROC(_kern, OID_AUTO, kdp_corefile,
3495 CTLTYPE_STRING | CTLFLAG_RD | CTLFLAG_LOCKED,
3496 0, 0, sysctl_kdp_corefile, "A", "");
3497
3498 #endif /* CONFIG_KDP_INTERACTIVE_DEBUGGING */
3499
3500 STATIC int
3501 sysctl_delayterm
3502 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3503 {
3504 struct proc *p = req->p;
3505 int new_value, changed;
3506 int error = sysctl_io_number(req, (req->p->p_lflag & P_LDELAYTERM)? 1: 0, sizeof(int), &new_value, &changed);
3507 if (changed) {
3508 proc_lock(p);
3509 if (new_value) {
3510 req->p->p_lflag |= P_LDELAYTERM;
3511 } else {
3512 req->p->p_lflag &= ~P_LDELAYTERM;
3513 }
3514 proc_unlock(p);
3515 }
3516 return error;
3517 }
3518
3519 SYSCTL_PROC(_kern, KERN_PROCDELAYTERM, delayterm,
3520 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
3521 0, 0, sysctl_delayterm, "I", "");
3522
3523
3524 STATIC int
3525 sysctl_rage_vnode
3526 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3527 {
3528 struct proc *p = req->p;
3529 struct uthread *ut;
3530 int new_value, old_value, changed;
3531 int error;
3532
3533 ut = current_uthread();
3534
3535 if (ut->uu_flag & UT_RAGE_VNODES) {
3536 old_value = KERN_RAGE_THREAD;
3537 } else if (p->p_lflag & P_LRAGE_VNODES) {
3538 old_value = KERN_RAGE_PROC;
3539 } else {
3540 old_value = 0;
3541 }
3542
3543 error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
3544
3545 if ((error == 0) && (changed != 0)) {
3546 switch (new_value) {
3547 case KERN_RAGE_PROC:
3548 proc_lock(p);
3549 p->p_lflag |= P_LRAGE_VNODES;
3550 proc_unlock(p);
3551 break;
3552 case KERN_UNRAGE_PROC:
3553 proc_lock(p);
3554 p->p_lflag &= ~P_LRAGE_VNODES;
3555 proc_unlock(p);
3556 break;
3557
3558 case KERN_RAGE_THREAD:
3559 ut->uu_flag |= UT_RAGE_VNODES;
3560 break;
3561 case KERN_UNRAGE_THREAD:
3562 ut = current_uthread();
3563 ut->uu_flag &= ~UT_RAGE_VNODES;
3564 break;
3565 }
3566 }
3567 return error;
3568 }
3569
3570 SYSCTL_PROC(_kern, KERN_RAGEVNODE, rage_vnode,
3571 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
3572 0, 0, sysctl_rage_vnode, "I", "");
3573
3574 /* XXX until filecoordinationd fixes a bit of inverted logic. */
3575 STATIC int
3576 sysctl_vfsnspace
3577 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3578 {
3579 int old_value = 0, new_value, changed;
3580
3581 return sysctl_io_number(req, old_value, sizeof(int), &new_value,
3582 &changed);
3583 }
3584
3585 SYSCTL_PROC(_kern, OID_AUTO, vfsnspace,
3586 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
3587 0, 0, sysctl_vfsnspace, "I", "");
3588
3589 /* XXX move this interface into libproc and remove this sysctl */
3590 STATIC int
3591 sysctl_setthread_cpupercent
3592 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3593 {
3594 int new_value, old_value;
3595 int error = 0;
3596 kern_return_t kret = KERN_SUCCESS;
3597 uint8_t percent = 0;
3598 int ms_refill = 0;
3599
3600 if (!req->newptr) {
3601 return 0;
3602 }
3603
3604 old_value = 0;
3605
3606 if ((error = sysctl_io_number(req, old_value, sizeof(old_value), &new_value, NULL)) != 0) {
3607 return error;
3608 }
3609
3610 percent = new_value & 0xff; /* low 8 bytes for perent */
3611 ms_refill = (new_value >> 8) & 0xffffff; /* upper 24bytes represent ms refill value */
3612 if (percent > 100) {
3613 return EINVAL;
3614 }
3615
3616 /*
3617 * If the caller is specifying a percentage of 0, this will unset the CPU limit, if present.
3618 */
3619 kret = percent == 0 ?
3620 thread_set_cpulimit(THREAD_CPULIMIT_DISABLE, 0, 0) :
3621 thread_set_cpulimit(THREAD_CPULIMIT_BLOCK, percent, ms_refill * (int)NSEC_PER_MSEC);
3622
3623 if (kret != 0) {
3624 return EIO;
3625 }
3626
3627 return 0;
3628 }
3629
3630 SYSCTL_PROC(_kern, OID_AUTO, setthread_cpupercent,
3631 CTLTYPE_INT | CTLFLAG_WR | CTLFLAG_ANYBODY,
3632 0, 0, sysctl_setthread_cpupercent, "I", "set thread cpu percentage limit");
3633
3634
3635 STATIC int
3636 sysctl_kern_check_openevt
3637 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3638 {
3639 struct proc *p = req->p;
3640 int new_value, old_value, changed;
3641 int error;
3642
3643 if (p->p_flag & P_CHECKOPENEVT) {
3644 old_value = KERN_OPENEVT_PROC;
3645 } else {
3646 old_value = 0;
3647 }
3648
3649 error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
3650
3651 if ((error == 0) && (changed != 0)) {
3652 switch (new_value) {
3653 case KERN_OPENEVT_PROC:
3654 OSBitOrAtomic(P_CHECKOPENEVT, &p->p_flag);
3655 break;
3656
3657 case KERN_UNOPENEVT_PROC:
3658 OSBitAndAtomic(~((uint32_t)P_CHECKOPENEVT), &p->p_flag);
3659 break;
3660
3661 default:
3662 error = EINVAL;
3663 }
3664 }
3665 return error;
3666 }
3667
3668 SYSCTL_PROC(_kern, KERN_CHECKOPENEVT, check_openevt, CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
3669 0, 0, sysctl_kern_check_openevt, "I", "set the per-process check-open-evt flag");
3670
3671
3672 #if DEVELOPMENT || DEBUG
3673 STATIC int
3674 sysctl_nx
3675 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3676 {
3677 #ifdef SECURE_KERNEL
3678 (void)req;
3679 return ENOTSUP;
3680 #else
3681 int new_value, changed;
3682 int error;
3683
3684 error = sysctl_io_number(req, nx_enabled, sizeof(nx_enabled), &new_value, &changed);
3685 if (error) {
3686 return error;
3687 }
3688
3689 if (changed) {
3690 #if defined(__x86_64__)
3691 /*
3692 * Only allow setting if NX is supported on the chip
3693 */
3694 if (!(cpuid_extfeatures() & CPUID_EXTFEATURE_XD)) {
3695 return ENOTSUP;
3696 }
3697 #endif
3698 nx_enabled = new_value;
3699 }
3700 return error;
3701 #endif /* SECURE_KERNEL */
3702 }
3703 #endif
3704
3705 #if DEVELOPMENT || DEBUG
3706 SYSCTL_PROC(_kern, KERN_NX_PROTECTION, nx,
3707 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
3708 0, 0, sysctl_nx, "I", "");
3709 #endif
3710
3711 STATIC int
3712 sysctl_loadavg
3713 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3714 {
3715 if (proc_is64bit(req->p)) {
3716 struct user64_loadavg loadinfo64 = {};
3717 fill_loadavg64(&averunnable, &loadinfo64);
3718 return sysctl_io_opaque(req, &loadinfo64, sizeof(loadinfo64), NULL);
3719 } else {
3720 struct user32_loadavg loadinfo32 = {};
3721 fill_loadavg32(&averunnable, &loadinfo32);
3722 return sysctl_io_opaque(req, &loadinfo32, sizeof(loadinfo32), NULL);
3723 }
3724 }
3725
3726 SYSCTL_PROC(_vm, VM_LOADAVG, loadavg,
3727 CTLTYPE_STRUCT | CTLFLAG_RD | CTLFLAG_LOCKED,
3728 0, 0, sysctl_loadavg, "S,loadavg", "");
3729
3730 /*
3731 * Note: Thread safe; vm_map_lock protects in vm_toggle_entry_reuse()
3732 */
3733 STATIC int
3734 sysctl_vm_toggle_address_reuse(__unused struct sysctl_oid *oidp, __unused void *arg1,
3735 __unused int arg2, struct sysctl_req *req)
3736 {
3737 int old_value = 0, new_value = 0, error = 0;
3738
3739 if (vm_toggle_entry_reuse( VM_TOGGLE_GETVALUE, &old_value )) {
3740 return error;
3741 }
3742 error = sysctl_io_number(req, old_value, sizeof(int), &new_value, NULL);
3743 if (!error) {
3744 return vm_toggle_entry_reuse(new_value, NULL);
3745 }
3746 return error;
3747 }
3748
3749 SYSCTL_PROC(_debug, OID_AUTO, toggle_address_reuse, CTLFLAG_ANYBODY | CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED, 0, 0, sysctl_vm_toggle_address_reuse, "I", "");
3750
3751 #ifdef CONFIG_XNUPOST
3752
3753 extern uint32_t xnupost_get_estimated_testdata_size(void);
3754 extern int xnupost_reset_all_tests(void);
3755
3756 STATIC int
3757 sysctl_handle_xnupost_get_tests SYSCTL_HANDLER_ARGS
3758 {
3759 /* fixup unused arguments warnings */
3760 __unused int _oa2 = arg2;
3761 __unused void * _oa1 = arg1;
3762 __unused struct sysctl_oid * _oidp = oidp;
3763
3764 int error = 0;
3765 user_addr_t oldp = 0;
3766 user_addr_t newp = 0;
3767 uint32_t usedbytes = 0;
3768
3769 oldp = req->oldptr;
3770 newp = req->newptr;
3771
3772 if (newp) {
3773 return ENOTSUP;
3774 }
3775
3776 if ((void *)oldp == NULL) {
3777 /* return estimated size for second call where info can be placed */
3778 req->oldidx = xnupost_get_estimated_testdata_size();
3779 } else {
3780 error = xnupost_export_testdata((void *)oldp, req->oldlen, &usedbytes);
3781 req->oldidx = usedbytes;
3782 }
3783
3784 return error;
3785 }
3786
3787 SYSCTL_PROC(_debug,
3788 OID_AUTO,
3789 xnupost_get_tests,
3790 CTLFLAG_MASKED | CTLFLAG_ANYBODY | CTLTYPE_OPAQUE | CTLFLAG_RD | CTLFLAG_LOCKED,
3791 0,
3792 0,
3793 sysctl_handle_xnupost_get_tests,
3794 "-",
3795 "read xnupost test data in kernel");
3796
3797 #if CONFIG_EXT_PANICLOG
3798 /*
3799 * Extensible panic log test hooks
3800 */
3801 static int
3802 sysctl_debug_ext_paniclog_test_hook SYSCTL_HANDLER_ARGS
3803 {
3804 #pragma unused(arg1, arg2)
3805 int rval = 0;
3806 uint32_t test_option = 0;
3807
3808 rval = sysctl_handle_int(oidp, &test_option, 0, req);
3809
3810 if (rval == 0 && req->newptr) {
3811 rval = ext_paniclog_test_hook(test_option);
3812 }
3813
3814 return rval;
3815 }
3816
3817 SYSCTL_PROC(_debug, OID_AUTO, ext_paniclog_test_hook,
3818 CTLTYPE_INT | CTLFLAG_RW,
3819 0, 0,
3820 sysctl_debug_ext_paniclog_test_hook, "A", "ext paniclog test hook");
3821
3822 #endif
3823
3824 STATIC int
3825 sysctl_debug_xnupost_ctl SYSCTL_HANDLER_ARGS
3826 {
3827 /* fixup unused arguments warnings */
3828 __unused int _oa2 = arg2;
3829 __unused void * _oa1 = arg1;
3830 __unused struct sysctl_oid * _oidp = oidp;
3831
3832 #define ARRCOUNT 4
3833 /*
3834 * INPUT: ACTION, PARAM1, PARAM2, PARAM3
3835 * OUTPUT: RESULTCODE, ADDITIONAL DATA
3836 */
3837 int32_t outval[ARRCOUNT] = {0};
3838 int32_t input[ARRCOUNT] = {0};
3839 int32_t out_size = sizeof(outval);
3840 int32_t in_size = sizeof(input);
3841 int error = 0;
3842
3843 /* if this is NULL call to find out size, send out size info */
3844 if (!req->newptr) {
3845 goto out;
3846 }
3847
3848 /* pull in provided value from userspace */
3849 error = SYSCTL_IN(req, &input[0], in_size);
3850 if (error) {
3851 return error;
3852 }
3853
3854 if (input[0] == XTCTL_RESET_TESTDATA) {
3855 outval[0] = xnupost_reset_all_tests();
3856 goto out;
3857 }
3858
3859 out:
3860 error = SYSCTL_OUT(req, &outval[0], out_size);
3861 return error;
3862 }
3863
3864 SYSCTL_PROC(_debug,
3865 OID_AUTO,
3866 xnupost_testctl,
3867 CTLFLAG_MASKED | CTLFLAG_ANYBODY | CTLTYPE_OPAQUE | CTLFLAG_RW | CTLFLAG_LOCKED,
3868 0,
3869 0,
3870 sysctl_debug_xnupost_ctl,
3871 "I",
3872 "xnupost control for kernel testing");
3873
3874 extern void test_oslog_handleOSLogCtl(int32_t * in, int32_t * out, int32_t arraycount);
3875
3876 STATIC int
3877 sysctl_debug_test_oslog_ctl(__unused struct sysctl_oid * oidp, __unused void * arg1, __unused int arg2, struct sysctl_req * req)
3878 {
3879 #define ARRCOUNT 4
3880 int32_t outval[ARRCOUNT] = {0};
3881 int32_t input[ARRCOUNT] = {0};
3882 int32_t size_outval = sizeof(outval);
3883 int32_t size_inval = sizeof(input);
3884 int32_t error;
3885
3886 /* if this is NULL call to find out size, send out size info */
3887 if (!req->newptr) {
3888 error = SYSCTL_OUT(req, &outval[0], size_outval);
3889 return error;
3890 }
3891
3892 /* pull in provided value from userspace */
3893 error = SYSCTL_IN(req, &input[0], size_inval);
3894 if (error) {
3895 return error;
3896 }
3897
3898 test_oslog_handleOSLogCtl(input, outval, ARRCOUNT);
3899
3900 error = SYSCTL_OUT(req, &outval[0], size_outval);
3901
3902 return error;
3903 }
3904
3905 SYSCTL_PROC(_debug,
3906 OID_AUTO,
3907 test_OSLogCtl,
3908 CTLFLAG_MASKED | CTLFLAG_ANYBODY | CTLTYPE_OPAQUE | CTLFLAG_RW | CTLFLAG_LOCKED,
3909 0,
3910 0,
3911 sysctl_debug_test_oslog_ctl,
3912 "I",
3913 "testing oslog in kernel");
3914
3915 #include <mach/task.h>
3916 #include <mach/semaphore.h>
3917
3918 static LCK_GRP_DECLARE(sysctl_debug_test_stackshot_owner_grp, "test-stackshot-owner-grp");
3919 static LCK_MTX_DECLARE(sysctl_debug_test_stackshot_owner_init_mtx,
3920 &sysctl_debug_test_stackshot_owner_grp);
3921
3922 /* This is a sysctl for testing collection of owner info on a lock in kernel space. A multi-threaded
3923 * test from userland sets this sysctl in such a way that a thread blocks in kernel mode, and a
3924 * stackshot is taken to see if the owner of the lock can be identified.
3925 *
3926 * We can't return to userland with a kernel lock held, so be sure to unlock before we leave.
3927 * the semaphores allow us to artificially create cases where the lock is being held and the
3928 * thread is hanging / taking a long time to do something. */
3929
3930 volatile char sysctl_debug_test_stackshot_mtx_inited = 0;
3931 semaphore_t sysctl_debug_test_stackshot_mutex_sem;
3932 lck_mtx_t sysctl_debug_test_stackshot_owner_lck;
3933
3934 #define SYSCTL_DEBUG_MTX_ACQUIRE_WAIT 1
3935 #define SYSCTL_DEBUG_MTX_ACQUIRE_NOWAIT 2
3936 #define SYSCTL_DEBUG_MTX_SIGNAL 3
3937 #define SYSCTL_DEBUG_MTX_TEARDOWN 4
3938
3939 STATIC int
3940 sysctl_debug_test_stackshot_mutex_owner(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3941 {
3942 long long option = -1;
3943 /* if the user tries to read the sysctl, we tell them what the address of the lock is (to test against stackshot's output) */
3944 long long mtx_unslid_addr = (long long)VM_KERNEL_UNSLIDE_OR_PERM(&sysctl_debug_test_stackshot_owner_lck);
3945 int error = sysctl_io_number(req, mtx_unslid_addr, sizeof(long long), (void*)&option, NULL);
3946
3947 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_init_mtx);
3948 if (!sysctl_debug_test_stackshot_mtx_inited) {
3949 lck_mtx_init(&sysctl_debug_test_stackshot_owner_lck,
3950 &sysctl_debug_test_stackshot_owner_grp,
3951 LCK_ATTR_NULL);
3952 semaphore_create(kernel_task,
3953 &sysctl_debug_test_stackshot_mutex_sem,
3954 SYNC_POLICY_FIFO, 0);
3955 sysctl_debug_test_stackshot_mtx_inited = 1;
3956 }
3957 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_init_mtx);
3958
3959 if (!error) {
3960 switch (option) {
3961 case SYSCTL_DEBUG_MTX_ACQUIRE_NOWAIT:
3962 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_lck);
3963 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_lck);
3964 break;
3965 case SYSCTL_DEBUG_MTX_ACQUIRE_WAIT:
3966 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_lck);
3967 semaphore_wait(sysctl_debug_test_stackshot_mutex_sem);
3968 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_lck);
3969 break;
3970 case SYSCTL_DEBUG_MTX_SIGNAL:
3971 semaphore_signal(sysctl_debug_test_stackshot_mutex_sem);
3972 break;
3973 case SYSCTL_DEBUG_MTX_TEARDOWN:
3974 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_init_mtx);
3975
3976 lck_mtx_destroy(&sysctl_debug_test_stackshot_owner_lck,
3977 &sysctl_debug_test_stackshot_owner_grp);
3978 semaphore_destroy(kernel_task,
3979 sysctl_debug_test_stackshot_mutex_sem);
3980 sysctl_debug_test_stackshot_mtx_inited = 0;
3981
3982 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_init_mtx);
3983 break;
3984 case -1: /* user just wanted to read the value, so do nothing */
3985 break;
3986 default:
3987 error = EINVAL;
3988 break;
3989 }
3990 }
3991 return error;
3992 }
3993
3994 /* we can't return to userland with a kernel rwlock held, so be sure to unlock before we leave.
3995 * the semaphores allow us to artificially create cases where the lock is being held and the
3996 * thread is hanging / taking a long time to do something. */
3997
3998 SYSCTL_PROC(_debug,
3999 OID_AUTO,
4000 test_MutexOwnerCtl,
4001 CTLFLAG_MASKED | CTLFLAG_ANYBODY | CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
4002 0,
4003 0,
4004 sysctl_debug_test_stackshot_mutex_owner,
4005 "-",
4006 "Testing mutex owner in kernel");
4007
4008 volatile char sysctl_debug_test_stackshot_rwlck_inited = 0;
4009 lck_rw_t sysctl_debug_test_stackshot_owner_rwlck;
4010 semaphore_t sysctl_debug_test_stackshot_rwlck_sem;
4011
4012 #define SYSCTL_DEBUG_KRWLCK_RACQUIRE_NOWAIT 1
4013 #define SYSCTL_DEBUG_KRWLCK_RACQUIRE_WAIT 2
4014 #define SYSCTL_DEBUG_KRWLCK_WACQUIRE_NOWAIT 3
4015 #define SYSCTL_DEBUG_KRWLCK_WACQUIRE_WAIT 4
4016 #define SYSCTL_DEBUG_KRWLCK_SIGNAL 5
4017 #define SYSCTL_DEBUG_KRWLCK_TEARDOWN 6
4018
4019 STATIC int
4020 sysctl_debug_test_stackshot_rwlck_owner(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4021 {
4022 long long option = -1;
4023 /* if the user tries to read the sysctl, we tell them what the address of the lock is
4024 * (to test against stackshot's output) */
4025 long long rwlck_unslid_addr = (long long)VM_KERNEL_UNSLIDE_OR_PERM(&sysctl_debug_test_stackshot_owner_rwlck);
4026 int error = sysctl_io_number(req, rwlck_unslid_addr, sizeof(long long), (void*)&option, NULL);
4027
4028 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_init_mtx);
4029 if (!sysctl_debug_test_stackshot_rwlck_inited) {
4030 lck_rw_init(&sysctl_debug_test_stackshot_owner_rwlck,
4031 &sysctl_debug_test_stackshot_owner_grp,
4032 LCK_ATTR_NULL);
4033 semaphore_create(kernel_task,
4034 &sysctl_debug_test_stackshot_rwlck_sem,
4035 SYNC_POLICY_FIFO,
4036 0);
4037 sysctl_debug_test_stackshot_rwlck_inited = 1;
4038 }
4039 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_init_mtx);
4040
4041 if (!error) {
4042 switch (option) {
4043 case SYSCTL_DEBUG_KRWLCK_RACQUIRE_NOWAIT:
4044 lck_rw_lock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_SHARED);
4045 lck_rw_unlock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_SHARED);
4046 break;
4047 case SYSCTL_DEBUG_KRWLCK_RACQUIRE_WAIT:
4048 lck_rw_lock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_SHARED);
4049 semaphore_wait(sysctl_debug_test_stackshot_rwlck_sem);
4050 lck_rw_unlock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_SHARED);
4051 break;
4052 case SYSCTL_DEBUG_KRWLCK_WACQUIRE_NOWAIT:
4053 lck_rw_lock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_EXCLUSIVE);
4054 lck_rw_unlock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_EXCLUSIVE);
4055 break;
4056 case SYSCTL_DEBUG_KRWLCK_WACQUIRE_WAIT:
4057 lck_rw_lock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_EXCLUSIVE);
4058 semaphore_wait(sysctl_debug_test_stackshot_rwlck_sem);
4059 lck_rw_unlock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_EXCLUSIVE);
4060 break;
4061 case SYSCTL_DEBUG_KRWLCK_SIGNAL:
4062 semaphore_signal(sysctl_debug_test_stackshot_rwlck_sem);
4063 break;
4064 case SYSCTL_DEBUG_KRWLCK_TEARDOWN:
4065 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_init_mtx);
4066
4067 lck_rw_destroy(&sysctl_debug_test_stackshot_owner_rwlck,
4068 &sysctl_debug_test_stackshot_owner_grp);
4069 semaphore_destroy(kernel_task,
4070 sysctl_debug_test_stackshot_rwlck_sem);
4071 sysctl_debug_test_stackshot_rwlck_inited = 0;
4072
4073 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_init_mtx);
4074 break;
4075 case -1: /* user just wanted to read the value, so do nothing */
4076 break;
4077 default:
4078 error = EINVAL;
4079 break;
4080 }
4081 }
4082 return error;
4083 }
4084
4085
4086 SYSCTL_PROC(_debug,
4087 OID_AUTO,
4088 test_RWLockOwnerCtl,
4089 CTLFLAG_MASKED | CTLFLAG_ANYBODY | CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
4090 0,
4091 0,
4092 sysctl_debug_test_stackshot_rwlck_owner,
4093 "-",
4094 "Testing rwlock owner in kernel");
4095 #endif /* !CONFIG_XNUPOST */
4096
4097 STATIC int
4098 sysctl_swapusage
4099 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4100 {
4101 int error;
4102 uint64_t swap_total;
4103 uint64_t swap_avail;
4104 vm_size_t swap_pagesize;
4105 boolean_t swap_encrypted;
4106 struct xsw_usage xsu = {};
4107
4108 error = macx_swapinfo(&swap_total,
4109 &swap_avail,
4110 &swap_pagesize,
4111 &swap_encrypted);
4112 if (error) {
4113 return error;
4114 }
4115
4116 xsu.xsu_total = swap_total;
4117 xsu.xsu_avail = swap_avail;
4118 xsu.xsu_used = swap_total - swap_avail;
4119 xsu.xsu_pagesize = (u_int32_t)MIN(swap_pagesize, UINT32_MAX);
4120 xsu.xsu_encrypted = swap_encrypted;
4121 return sysctl_io_opaque(req, &xsu, sizeof(xsu), NULL);
4122 }
4123
4124
4125
4126 SYSCTL_PROC(_vm, VM_SWAPUSAGE, swapusage,
4127 CTLTYPE_STRUCT | CTLFLAG_RD | CTLFLAG_LOCKED,
4128 0, 0, sysctl_swapusage, "S,xsw_usage", "");
4129
4130 extern int vm_swap_enabled;
4131 SYSCTL_INT(_vm, OID_AUTO, swap_enabled, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_swap_enabled, 0, "");
4132
4133 #if DEVELOPMENT || DEBUG
4134 extern int vm_num_swap_files_config;
4135 extern int vm_num_swap_files;
4136 extern lck_mtx_t vm_swap_data_lock;
4137 #define VM_MAX_SWAP_FILE_NUM 100
4138
4139 static int
4140 sysctl_vm_config_num_swap_files SYSCTL_HANDLER_ARGS
4141 {
4142 #pragma unused(arg1, arg2)
4143 int error = 0, val = vm_num_swap_files_config;
4144
4145 error = sysctl_handle_int(oidp, &val, 0, req);
4146 if (error || !req->newptr) {
4147 goto out;
4148 }
4149
4150 if (!VM_CONFIG_SWAP_IS_ACTIVE && !VM_CONFIG_FREEZER_SWAP_IS_ACTIVE) {
4151 printf("Swap is disabled\n");
4152 error = EINVAL;
4153 goto out;
4154 }
4155
4156 lck_mtx_lock(&vm_swap_data_lock);
4157
4158 if (val < vm_num_swap_files) {
4159 printf("Cannot configure fewer swap files than already exist.\n");
4160 error = EINVAL;
4161 lck_mtx_unlock(&vm_swap_data_lock);
4162 goto out;
4163 }
4164
4165 if (val > VM_MAX_SWAP_FILE_NUM) {
4166 printf("Capping number of swap files to upper bound.\n");
4167 val = VM_MAX_SWAP_FILE_NUM;
4168 }
4169
4170 vm_num_swap_files_config = val;
4171 lck_mtx_unlock(&vm_swap_data_lock);
4172 out:
4173
4174 return 0;
4175 }
4176
4177 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", "");
4178 #endif /* DEVELOPMENT || DEBUG */
4179
4180 /* this kernel does NOT implement shared_region_make_private_np() */
4181 SYSCTL_INT(_kern, KERN_SHREG_PRIVATIZABLE, shreg_private,
4182 CTLFLAG_RD | CTLFLAG_LOCKED,
4183 (int *)NULL, 0, "");
4184
4185 STATIC int
4186 fetch_process_cputype(
4187 proc_t cur_proc,
4188 int *name,
4189 u_int namelen,
4190 cpu_type_t *cputype)
4191 {
4192 proc_t p = PROC_NULL;
4193 int refheld = 0;
4194 cpu_type_t ret = 0;
4195 int error = 0;
4196
4197 if (namelen == 0) {
4198 p = cur_proc;
4199 } else if (namelen == 1) {
4200 p = proc_find(name[0]);
4201 if (p == NULL) {
4202 return EINVAL;
4203 }
4204 refheld = 1;
4205 } else {
4206 error = EINVAL;
4207 goto out;
4208 }
4209
4210 ret = cpu_type() & ~CPU_ARCH_MASK;
4211 if (IS_64BIT_PROCESS(p)) {
4212 ret |= CPU_ARCH_ABI64;
4213 }
4214
4215 *cputype = ret;
4216
4217 if (refheld != 0) {
4218 proc_rele(p);
4219 }
4220 out:
4221 return error;
4222 }
4223
4224 #if CONFIG_ROSETTA
4225 STATIC int
4226 sysctl_sysctl_translated(
4227 __unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, __unused struct sysctl_req *req)
4228 {
4229 int res = 0;
4230 if (proc_is_translated(req->p)) {
4231 res = 1;
4232 }
4233 return SYSCTL_OUT(req, &res, sizeof(res));
4234 }
4235 SYSCTL_PROC(_sysctl, OID_AUTO, proc_translated, CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED, 0, 0, sysctl_sysctl_translated, "I", "proc_translated");
4236 #endif /* CONFIG_ROSETTA */
4237
4238 STATIC int
4239 sysctl_sysctl_native(__unused struct sysctl_oid *oidp, void *arg1, int arg2,
4240 struct sysctl_req *req)
4241 {
4242 int error;
4243 cpu_type_t proc_cputype = 0;
4244 if ((error = fetch_process_cputype(req->p, (int *)arg1, arg2, &proc_cputype)) != 0) {
4245 return error;
4246 }
4247 int res = 1;
4248 if ((proc_cputype & ~CPU_ARCH_MASK) != (cpu_type() & ~CPU_ARCH_MASK)) {
4249 res = 0;
4250 }
4251 return SYSCTL_OUT(req, &res, sizeof(res));
4252 }
4253 SYSCTL_PROC(_sysctl, OID_AUTO, proc_native, CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED, 0, 0, sysctl_sysctl_native, "I", "proc_native");
4254
4255 STATIC int
4256 sysctl_sysctl_cputype(__unused struct sysctl_oid *oidp, void *arg1, int arg2,
4257 struct sysctl_req *req)
4258 {
4259 int error;
4260 cpu_type_t proc_cputype = 0;
4261 if ((error = fetch_process_cputype(req->p, (int *)arg1, arg2, &proc_cputype)) != 0) {
4262 return error;
4263 }
4264 return SYSCTL_OUT(req, &proc_cputype, sizeof(proc_cputype));
4265 }
4266 SYSCTL_PROC(_sysctl, OID_AUTO, proc_cputype, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED, 0, 0, sysctl_sysctl_cputype, "I", "proc_cputype");
4267
4268 STATIC int
4269 sysctl_safeboot
4270 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4271 {
4272 return sysctl_io_number(req, boothowto & RB_SAFEBOOT ? 1 : 0, sizeof(int), NULL, NULL);
4273 }
4274
4275 SYSCTL_PROC(_kern, KERN_SAFEBOOT, safeboot,
4276 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
4277 0, 0, sysctl_safeboot, "I", "");
4278
4279 STATIC int
4280 sysctl_singleuser
4281 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4282 {
4283 return sysctl_io_number(req, boothowto & RB_SINGLE ? 1 : 0, sizeof(int), NULL, NULL);
4284 }
4285
4286 SYSCTL_PROC(_kern, OID_AUTO, singleuser,
4287 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
4288 0, 0, sysctl_singleuser, "I", "");
4289
4290 STATIC int
4291 sysctl_minimalboot
4292 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4293 {
4294 return sysctl_io_number(req, minimalboot, sizeof(int), NULL, NULL);
4295 }
4296
4297 SYSCTL_PROC(_kern, OID_AUTO, minimalboot,
4298 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
4299 0, 0, sysctl_minimalboot, "I", "");
4300
4301 /*
4302 * Controls for debugging affinity sets - see osfmk/kern/affinity.c
4303 */
4304 extern boolean_t affinity_sets_enabled;
4305 extern int affinity_sets_mapping;
4306
4307 SYSCTL_INT(_kern, OID_AUTO, affinity_sets_enabled,
4308 CTLFLAG_RW | CTLFLAG_LOCKED, (int *) &affinity_sets_enabled, 0, "hinting enabled");
4309 SYSCTL_INT(_kern, OID_AUTO, affinity_sets_mapping,
4310 CTLFLAG_RW | CTLFLAG_LOCKED, &affinity_sets_mapping, 0, "mapping policy");
4311
4312 /*
4313 * Boolean indicating if KASLR is active.
4314 */
4315 STATIC int
4316 sysctl_slide
4317 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4318 {
4319 uint32_t slide;
4320
4321 slide = vm_kernel_slide ? 1 : 0;
4322
4323 return sysctl_io_number( req, slide, sizeof(int), NULL, NULL);
4324 }
4325
4326 SYSCTL_PROC(_kern, OID_AUTO, slide,
4327 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
4328 0, 0, sysctl_slide, "I", "");
4329
4330 #if DEBUG || DEVELOPMENT
4331 #if defined(__arm64__)
4332 extern vm_offset_t segTEXTEXECB;
4333
4334 static int
4335 sysctl_kernel_text_exec_base_slide SYSCTL_HANDLER_ARGS
4336 {
4337 #pragma unused(arg1, arg2, oidp)
4338 unsigned long slide = 0;
4339 kc_format_t kc_format;
4340
4341 PE_get_primary_kc_format(&kc_format);
4342
4343 if (kc_format == KCFormatFileset) {
4344 void *kch = PE_get_kc_header(KCKindPrimary);
4345 slide = (unsigned long)segTEXTEXECB - (unsigned long)kch + vm_kernel_slide;
4346 }
4347 return SYSCTL_OUT(req, &slide, sizeof(slide));
4348 }
4349
4350 SYSCTL_QUAD(_kern, OID_AUTO, kernel_slide, CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED, &vm_kernel_slide, "");
4351 SYSCTL_QUAD(_kern, OID_AUTO, kernel_text_exec_base, CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED, &segTEXTEXECB, "");
4352 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", "");
4353 #endif /* defined(__arm64__) */
4354
4355 /* User address of the PFZ */
4356 extern user32_addr_t commpage_text32_location;
4357 extern user64_addr_t commpage_text64_location;
4358
4359 STATIC int
4360 sysctl_pfz_start SYSCTL_HANDLER_ARGS
4361 {
4362 #pragma unused(oidp, arg1, arg2)
4363
4364 #ifdef __LP64__
4365 return sysctl_io_number(req, commpage_text64_location, sizeof(user64_addr_t), NULL, NULL);
4366 #else
4367 return sysctl_io_number(req, commpage_text32_location, sizeof(user32_addr_t), NULL, NULL);
4368 #endif
4369 }
4370
4371 SYSCTL_PROC(_kern, OID_AUTO, pfz,
4372 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_MASKED,
4373 0, 0, sysctl_pfz_start, "I", "");
4374 #endif
4375
4376
4377 /*
4378 * Limit on total memory users can wire.
4379 *
4380 * vm_global_user_wire_limit - system wide limit on wired memory from all processes combined.
4381 *
4382 * vm_per_task_user_wire_limit - per address space limit on wired memory. This puts a cap on the process's rlimit value.
4383 *
4384 * These values are initialized to reasonable defaults at boot time based on the available physical memory in
4385 * kmem_init().
4386 *
4387 * All values are in bytes.
4388 */
4389
4390 vm_map_size_t vm_global_user_wire_limit;
4391 vm_map_size_t vm_per_task_user_wire_limit;
4392 extern uint64_t max_mem_actual, max_mem;
4393
4394 uint64_t vm_add_wire_count_over_global_limit;
4395 uint64_t vm_add_wire_count_over_user_limit;
4396 /*
4397 * We used to have a global in the kernel called vm_global_no_user_wire_limit which was the inverse
4398 * of vm_global_user_wire_limit. But maintaining both of those is silly, and vm_global_user_wire_limit is the
4399 * real limit.
4400 * This function is for backwards compatibility with userspace
4401 * since we exposed the old global via a sysctl.
4402 */
4403 STATIC int
4404 sysctl_global_no_user_wire_amount(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4405 {
4406 vm_map_size_t old_value;
4407 vm_map_size_t new_value;
4408 int changed;
4409 int error;
4410 uint64_t config_memsize = max_mem;
4411 #if defined(XNU_TARGET_OS_OSX)
4412 config_memsize = max_mem_actual;
4413 #endif /* defined(XNU_TARGET_OS_OSX) */
4414
4415 old_value = (vm_map_size_t)(config_memsize - vm_global_user_wire_limit);
4416 error = sysctl_io_number(req, old_value, sizeof(vm_map_size_t), &new_value, &changed);
4417 if (changed) {
4418 if ((uint64_t)new_value > config_memsize) {
4419 error = EINVAL;
4420 } else {
4421 vm_global_user_wire_limit = (vm_map_size_t)(config_memsize - new_value);
4422 }
4423 }
4424 return error;
4425 }
4426 /*
4427 * There needs to be a more automatic/elegant way to do this
4428 */
4429 SYSCTL_QUAD(_vm, OID_AUTO, global_user_wire_limit, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_global_user_wire_limit, "");
4430 SYSCTL_QUAD(_vm, OID_AUTO, user_wire_limit, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_per_task_user_wire_limit, "");
4431 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", "");
4432
4433 /*
4434 * Relaxed atomic RW of a 64bit value via sysctl.
4435 */
4436 STATIC int
4437 sysctl_r_64bit_atomic(uint64_t *ptr, struct sysctl_req *req)
4438 {
4439 uint64_t old_value;
4440 uint64_t new_value;
4441 int error;
4442
4443 old_value = os_atomic_load_wide(ptr, relaxed);
4444 error = sysctl_io_number(req, old_value, sizeof(vm_map_size_t), &new_value, NULL);
4445 return error;
4446 }
4447 STATIC int
4448 sysctl_add_wire_count_over_global_limit(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4449 {
4450 return sysctl_r_64bit_atomic(&vm_add_wire_count_over_global_limit, req);
4451 }
4452 STATIC int
4453 sysctl_add_wire_count_over_user_limit(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4454 {
4455 return sysctl_r_64bit_atomic(&vm_add_wire_count_over_user_limit, req);
4456 }
4457
4458 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", "");
4459 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", "");
4460
4461 #if DEVELOPMENT || DEBUG
4462 /* These sysctls are used to test the wired limit. */
4463 SYSCTL_INT(_vm, OID_AUTO, page_wire_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_wire_count, 0,
4464 "The number of physical pages which are pinned and cannot be evicted");
4465 #if XNU_VM_HAS_LOPAGE
4466 SYSCTL_INT(_vm, OID_AUTO, lopage_free_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_lopage_free_count, 0, "");
4467 #endif
4468 SYSCTL_INT(_vm, OID_AUTO, page_stolen_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_stolen_count, 0, "");
4469 SYSCTL_UINT(_vm, OID_AUTO, page_swapped_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_swapped_count, 0,
4470 "The number of virtual pages whose contents are currently compressed and swapped to disk");
4471
4472 /*
4473 * Setting the per task variable exclude_physfootprint_ledger to 1 will allow the calling task to exclude memory entries that are
4474 * tagged by VM_LEDGER_TAG_DEFAULT and flagged by VM_LEDGER_FLAG_EXCLUDE_FOOTPRINT_DEBUG from its phys_footprint ledger.
4475 */
4476
4477 STATIC int
4478 sysctl_rw_task_no_footprint_for_debug(struct sysctl_oid *oidp __unused, void *arg1 __unused, int arg2 __unused, struct sysctl_req *req)
4479 {
4480 int error;
4481 int value;
4482 proc_t p = current_proc();
4483
4484 if (req->newptr) {
4485 // Write request
4486 error = SYSCTL_IN(req, &value, sizeof(value));
4487 if (!error) {
4488 if (value == 1) {
4489 task_set_no_footprint_for_debug(proc_task(p), TRUE);
4490 } else if (value == 0) {
4491 task_set_no_footprint_for_debug(proc_task(p), FALSE);
4492 } else {
4493 error = EINVAL;
4494 }
4495 }
4496 } else {
4497 // Read request
4498 value = task_get_no_footprint_for_debug(proc_task(p));
4499 error = SYSCTL_OUT(req, &value, sizeof(value));
4500 }
4501 return error;
4502 }
4503
4504 SYSCTL_PROC(_vm, OID_AUTO, task_no_footprint_for_debug,
4505 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_ANYBODY,
4506 0, 0, &sysctl_rw_task_no_footprint_for_debug, "I", "Allow debug memory to be excluded from this task's memory footprint (debug only)");
4507
4508 #endif /* DEVELOPMENT || DEBUG */
4509
4510
4511 extern int vm_map_copy_overwrite_aligned_src_not_internal;
4512 extern int vm_map_copy_overwrite_aligned_src_not_symmetric;
4513 extern int vm_map_copy_overwrite_aligned_src_large;
4514 SYSCTL_INT(_vm, OID_AUTO, vm_copy_src_not_internal, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_map_copy_overwrite_aligned_src_not_internal, 0, "");
4515 SYSCTL_INT(_vm, OID_AUTO, vm_copy_src_not_symmetric, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_map_copy_overwrite_aligned_src_not_symmetric, 0, "");
4516 SYSCTL_INT(_vm, OID_AUTO, vm_copy_src_large, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_map_copy_overwrite_aligned_src_large, 0, "");
4517
4518
4519 extern uint32_t vm_page_external_count;
4520
4521 SYSCTL_INT(_vm, OID_AUTO, vm_page_external_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_external_count, 0, "");
4522
4523 SYSCTL_INT(_vm, OID_AUTO, vm_page_filecache_min, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_state.vm_page_filecache_min, 0, "");
4524 SYSCTL_INT(_vm, OID_AUTO, vm_page_xpmapped_min, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_state.vm_page_xpmapped_min, 0, "");
4525
4526 #if DEVELOPMENT || DEBUG
4527 SYSCTL_INT(_vm, OID_AUTO, vm_page_filecache_min_divisor, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_state.vm_page_filecache_min_divisor, 0, "");
4528 SYSCTL_INT(_vm, OID_AUTO, vm_page_xpmapped_min_divisor, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_state.vm_page_xpmapped_min_divisor, 0, "");
4529 extern boolean_t vps_yield_for_pgqlockwaiters;
4530 SYSCTL_INT(_vm, OID_AUTO, vm_pageoutscan_yields_for_pageQlockwaiters, CTLFLAG_RW | CTLFLAG_LOCKED, &vps_yield_for_pgqlockwaiters, 0, "");
4531 #endif
4532
4533 extern int vm_compressor_mode;
4534 extern int vm_compressor_is_active;
4535 extern int vm_compressor_available;
4536 extern uint32_t c_seg_bufsize;
4537 extern uint32_t c_seg_allocsize;
4538 extern int c_seg_fixed_array_len;
4539 extern uint32_t c_segments_limit;
4540 extern uint32_t c_segment_pages_compressed_limit;
4541 extern uint64_t compressor_pool_size;
4542 extern uint32_t compressor_pool_multiplier;
4543 extern uint32_t vm_ripe_target_age;
4544 extern uint32_t swapout_target_age;
4545 extern _Atomic uint64_t compressor_bytes_used;
4546 extern _Atomic uint64_t c_segment_input_bytes;
4547 extern _Atomic uint64_t c_segment_compressed_bytes;
4548 extern uint32_t c_segment_pages_compressed;
4549 extern uint32_t compressor_eval_period_in_msecs;
4550 extern uint32_t compressor_sample_min_in_msecs;
4551 extern uint32_t compressor_sample_max_in_msecs;
4552 extern uint32_t compressor_thrashing_threshold_per_10msecs;
4553 extern uint32_t compressor_thrashing_min_per_10msecs;
4554 extern uint32_t vm_compressor_time_thread;
4555 extern uint32_t c_segment_svp_in_hash;
4556 extern uint32_t c_segment_svp_hash_succeeded;
4557 extern uint32_t c_segment_svp_hash_failed;
4558
4559 #if DEVELOPMENT || DEBUG
4560 extern uint32_t vm_compressor_minorcompact_threshold_divisor;
4561 extern uint32_t vm_compressor_majorcompact_threshold_divisor;
4562 extern uint32_t vm_compressor_unthrottle_threshold_divisor;
4563 extern uint32_t vm_compressor_catchup_threshold_divisor;
4564
4565 extern uint32_t vm_compressor_minorcompact_threshold_divisor_overridden;
4566 extern uint32_t vm_compressor_majorcompact_threshold_divisor_overridden;
4567 extern uint32_t vm_compressor_unthrottle_threshold_divisor_overridden;
4568 extern uint32_t vm_compressor_catchup_threshold_divisor_overridden;
4569
4570 extern vmct_stats_t vmct_stats;
4571
4572
4573 STATIC int
4574 sysctl_minorcompact_threshold_divisor(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4575 {
4576 int new_value, changed;
4577 int error = sysctl_io_number(req, vm_compressor_minorcompact_threshold_divisor, sizeof(int), &new_value, &changed);
4578
4579 if (changed) {
4580 vm_compressor_minorcompact_threshold_divisor = new_value;
4581 vm_compressor_minorcompact_threshold_divisor_overridden = 1;
4582 }
4583 return error;
4584 }
4585
4586 SYSCTL_PROC(_vm, OID_AUTO, compressor_minorcompact_threshold_divisor,
4587 CTLTYPE_INT | CTLFLAG_LOCKED | CTLFLAG_RW,
4588 0, 0, sysctl_minorcompact_threshold_divisor, "I", "");
4589
4590
4591 STATIC int
4592 sysctl_majorcompact_threshold_divisor(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4593 {
4594 int new_value, changed;
4595 int error = sysctl_io_number(req, vm_compressor_majorcompact_threshold_divisor, sizeof(int), &new_value, &changed);
4596
4597 if (changed) {
4598 vm_compressor_majorcompact_threshold_divisor = new_value;
4599 vm_compressor_majorcompact_threshold_divisor_overridden = 1;
4600 }
4601 return error;
4602 }
4603
4604 SYSCTL_PROC(_vm, OID_AUTO, compressor_majorcompact_threshold_divisor,
4605 CTLTYPE_INT | CTLFLAG_LOCKED | CTLFLAG_RW,
4606 0, 0, sysctl_majorcompact_threshold_divisor, "I", "");
4607
4608
4609 STATIC int
4610 sysctl_unthrottle_threshold_divisor(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4611 {
4612 int new_value, changed;
4613 int error = sysctl_io_number(req, vm_compressor_unthrottle_threshold_divisor, sizeof(int), &new_value, &changed);
4614
4615 if (changed) {
4616 vm_compressor_unthrottle_threshold_divisor = new_value;
4617 vm_compressor_unthrottle_threshold_divisor_overridden = 1;
4618 }
4619 return error;
4620 }
4621
4622 SYSCTL_PROC(_vm, OID_AUTO, compressor_unthrottle_threshold_divisor,
4623 CTLTYPE_INT | CTLFLAG_LOCKED | CTLFLAG_RW,
4624 0, 0, sysctl_unthrottle_threshold_divisor, "I", "");
4625
4626
4627 STATIC int
4628 sysctl_catchup_threshold_divisor(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4629 {
4630 int new_value, changed;
4631 int error = sysctl_io_number(req, vm_compressor_catchup_threshold_divisor, sizeof(int), &new_value, &changed);
4632
4633 if (changed) {
4634 vm_compressor_catchup_threshold_divisor = new_value;
4635 vm_compressor_catchup_threshold_divisor_overridden = 1;
4636 }
4637 return error;
4638 }
4639
4640 SYSCTL_PROC(_vm, OID_AUTO, compressor_catchup_threshold_divisor,
4641 CTLTYPE_INT | CTLFLAG_LOCKED | CTLFLAG_RW,
4642 0, 0, sysctl_catchup_threshold_divisor, "I", "");
4643 #endif
4644
4645
4646 SYSCTL_QUAD(_vm, OID_AUTO, compressor_input_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, ((uint64_t *)&c_segment_input_bytes), "");
4647 SYSCTL_QUAD(_vm, OID_AUTO, compressor_compressed_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, ((uint64_t *)&c_segment_compressed_bytes), "");
4648 SYSCTL_QUAD(_vm, OID_AUTO, compressor_bytes_used, CTLFLAG_RD | CTLFLAG_LOCKED, ((uint64_t *)&compressor_bytes_used), "");
4649
4650 SYSCTL_INT(_vm, OID_AUTO, compressor_mode, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_compressor_mode, 0, "");
4651 SYSCTL_INT(_vm, OID_AUTO, compressor_is_active, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_compressor_is_active, 0, "");
4652 SYSCTL_INT(_vm, OID_AUTO, compressor_swapout_target_age, CTLFLAG_RD | CTLFLAG_LOCKED, &swapout_target_age, 0, "");
4653 SYSCTL_INT(_vm, OID_AUTO, compressor_available, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_compressor_available, 0, "");
4654 SYSCTL_INT(_vm, OID_AUTO, compressor_segment_buffer_size, CTLFLAG_RD | CTLFLAG_LOCKED, &c_seg_bufsize, 0, "");
4655 SYSCTL_QUAD(_vm, OID_AUTO, compressor_pool_size, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_pool_size, "");
4656 SYSCTL_UINT(_vm, OID_AUTO, compressor_pool_multiplier, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_pool_multiplier, 0, "");
4657 SYSCTL_INT(_vm, OID_AUTO, compressor_segment_slots_fixed_array_len, CTLFLAG_RD | CTLFLAG_LOCKED, &c_seg_fixed_array_len, 0, "");
4658 SYSCTL_UINT(_vm, OID_AUTO, compressor_segment_limit, CTLFLAG_RD | CTLFLAG_LOCKED, &c_segments_limit, 0, "");
4659 SYSCTL_UINT(_vm, OID_AUTO, compressor_segment_pages_compressed_limit, CTLFLAG_RD | CTLFLAG_LOCKED, &c_segment_pages_compressed_limit, 0, "");
4660 SYSCTL_UINT(_vm, OID_AUTO, compressor_segment_alloc_size, CTLFLAG_RD | CTLFLAG_LOCKED, &c_seg_allocsize, 0, "");
4661 SYSCTL_UINT(_vm, OID_AUTO, compressor_segment_pages_compressed, CTLFLAG_RD | CTLFLAG_LOCKED, &c_segment_pages_compressed, 0, "");
4662 SYSCTL_UINT(_vm, OID_AUTO, compressor_segment_svp_in_hash, CTLFLAG_RD | CTLFLAG_LOCKED, &c_segment_svp_in_hash, 0, "");
4663 SYSCTL_UINT(_vm, OID_AUTO, compressor_segment_svp_hash_succeeded, CTLFLAG_RD | CTLFLAG_LOCKED, &c_segment_svp_hash_succeeded, 0, "");
4664 SYSCTL_UINT(_vm, OID_AUTO, compressor_segment_svp_hash_failed, CTLFLAG_RD | CTLFLAG_LOCKED, &c_segment_svp_hash_failed, 0, "");
4665
4666 #if CONFIG_TRACK_UNMODIFIED_ANON_PAGES
4667 extern uint64_t compressor_ro_uncompressed;
4668 extern uint64_t compressor_ro_uncompressed_total_returned;
4669 extern uint64_t compressor_ro_uncompressed_skip_returned;
4670 extern uint64_t compressor_ro_uncompressed_get;
4671 extern uint64_t compressor_ro_uncompressed_put;
4672 extern uint64_t compressor_ro_uncompressed_swap_usage;
4673
4674 SYSCTL_QUAD(_vm, OID_AUTO, compressor_ro_uncompressed_total_returned, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_ro_uncompressed_total_returned, "");
4675 SYSCTL_QUAD(_vm, OID_AUTO, compressor_ro_uncompressed_writes_saved, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_ro_uncompressed_skip_returned, "");
4676 SYSCTL_QUAD(_vm, OID_AUTO, compressor_ro_uncompressed_candidates, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_ro_uncompressed, "");
4677 SYSCTL_QUAD(_vm, OID_AUTO, compressor_ro_uncompressed_rereads, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_ro_uncompressed_get, "");
4678 SYSCTL_QUAD(_vm, OID_AUTO, compressor_ro_uncompressed_swap_pages_on_disk, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_ro_uncompressed_swap_usage, "");
4679 #endif /* CONFIG_TRACK_UNMODIFIED_ANON_PAGES */
4680
4681 extern int min_csegs_per_major_compaction;
4682 SYSCTL_INT(_vm, OID_AUTO, compressor_min_csegs_per_major_compaction, CTLFLAG_RW | CTLFLAG_LOCKED, &min_csegs_per_major_compaction, 0, "");
4683
4684 SYSCTL_INT(_vm, OID_AUTO, vm_ripe_target_age_in_secs, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_ripe_target_age, 0, "");
4685
4686 SYSCTL_INT(_vm, OID_AUTO, compressor_eval_period_in_msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &compressor_eval_period_in_msecs, 0, "");
4687 SYSCTL_INT(_vm, OID_AUTO, compressor_sample_min_in_msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &compressor_sample_min_in_msecs, 0, "");
4688 SYSCTL_INT(_vm, OID_AUTO, compressor_sample_max_in_msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &compressor_sample_max_in_msecs, 0, "");
4689 SYSCTL_INT(_vm, OID_AUTO, compressor_thrashing_threshold_per_10msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &compressor_thrashing_threshold_per_10msecs, 0, "");
4690 SYSCTL_INT(_vm, OID_AUTO, compressor_thrashing_min_per_10msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &compressor_thrashing_min_per_10msecs, 0, "");
4691
4692 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapouts_under_30s, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.unripe_under_30s, "");
4693 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapouts_under_60s, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.unripe_under_60s, "");
4694 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapouts_under_300s, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.unripe_under_300s, "");
4695 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_reclaim_swapins, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.reclaim_swapins, "");
4696 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_defrag_swapins, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.defrag_swapins, "");
4697 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_swapout_threshold_exceeded, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.compressor_swap_threshold_exceeded, "");
4698 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_swapout_fileq_throttled, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.external_q_throttled, "");
4699 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_swapout_free_count_low, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.free_count_below_reserve, "");
4700 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_swapout_thrashing_detected, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.thrashing_detected, "");
4701 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_swapout_fragmentation_detected, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.fragmentation_detected, "");
4702
4703 SYSCTL_STRING(_vm, OID_AUTO, swapfileprefix, CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED, swapfilename, sizeof(swapfilename) - SWAPFILENAME_INDEX_LEN, "");
4704
4705 SYSCTL_INT(_vm, OID_AUTO, compressor_timing_enabled, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_compressor_time_thread, 0, "");
4706
4707 #if DEVELOPMENT || DEBUG
4708 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_runtime0, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_runtimes[0], "");
4709 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_runtime1, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_runtimes[1], "");
4710
4711 SYSCTL_QUAD(_vm, OID_AUTO, compressor_threads_total_execution_time, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_cthreads_total, "");
4712
4713 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_pages0, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_pages[0], "");
4714 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_pages1, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_pages[1], "");
4715
4716 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_iterations0, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_iterations[0], "");
4717 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_iterations1, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_iterations[1], "");
4718
4719 SYSCTL_INT(_vm, OID_AUTO, compressor_thread_minpages0, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_minpages[0], 0, "");
4720 SYSCTL_INT(_vm, OID_AUTO, compressor_thread_minpages1, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_minpages[1], 0, "");
4721
4722 SYSCTL_INT(_vm, OID_AUTO, compressor_thread_maxpages0, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_maxpages[0], 0, "");
4723 SYSCTL_INT(_vm, OID_AUTO, compressor_thread_maxpages1, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_maxpages[1], 0, "");
4724
4725 int vm_compressor_injected_error_count;
4726
4727 SYSCTL_INT(_vm, OID_AUTO, compressor_injected_error_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_compressor_injected_error_count, 0, "");
4728
4729 static int
4730 sysctl_compressor_inject_error(__unused struct sysctl_oid *oidp,
4731 __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4732 {
4733 int result;
4734 vm_address_t va = 0;
4735 int changed;
4736
4737 result = sysctl_io_number(req, va, sizeof(va), &va, &changed);
4738 if (result == 0 && changed) {
4739 result = vm_map_inject_error(current_map(), va);
4740 if (result == 0) {
4741 /*
4742 * Count the number of errors injected successfully to detect
4743 * situations where corruption was caused by improper use of this
4744 * sysctl.
4745 */
4746 os_atomic_inc(&vm_compressor_injected_error_count, relaxed);
4747 }
4748 }
4749 return result;
4750 }
4751
4752 SYSCTL_PROC(_vm, OID_AUTO, compressor_inject_error, CTLTYPE_QUAD | CTLFLAG_LOCKED | CTLFLAG_RW,
4753 0, 0, sysctl_compressor_inject_error, "Q", "flips a bit in a compressed page for the current task");
4754
4755 /*
4756 * Opt a process in/out of self donation mode.
4757 */
4758 static int
4759 sysctl_vm_pid_toggle_selfdonate_pages SYSCTL_HANDLER_ARGS
4760 {
4761 #pragma unused(arg1, arg2)
4762 int error, pid = 0;
4763 proc_t p;
4764
4765 error = sysctl_handle_int(oidp, &pid, 0, req);
4766 if (error || !req->newptr) {
4767 return error;
4768 }
4769
4770 p = proc_find(pid);
4771 if (p != NULL) {
4772 (void) vm_toggle_task_selfdonate_pages(proc_task(p));
4773 proc_rele(p);
4774 return error;
4775 } else {
4776 printf("sysctl_vm_pid_selfdonate_pages: Invalid process\n");
4777 }
4778
4779 return EINVAL;
4780 }
4781 SYSCTL_PROC(_vm, OID_AUTO, pid_toggle_selfdonate_pages, CTLTYPE_INT | CTLFLAG_WR | CTLFLAG_LOCKED | CTLFLAG_MASKED,
4782 0, 0, &sysctl_vm_pid_toggle_selfdonate_pages, "I", "");
4783 #endif /* DEVELOPMENT || DEBUG */
4784 extern uint32_t vm_page_donate_mode;
4785 extern uint32_t vm_page_donate_target_high, vm_page_donate_target_low;
4786 SYSCTL_INT(_vm, OID_AUTO, vm_page_donate_mode, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_page_donate_mode, 0, "");
4787 SYSCTL_INT(_vm, OID_AUTO, vm_page_donate_target_high, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_page_donate_target_high, 0, "");
4788 SYSCTL_INT(_vm, OID_AUTO, vm_page_donate_target_low, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_page_donate_target_low, 0, "");
4789
4790 SYSCTL_QUAD(_vm, OID_AUTO, lz4_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_compressions, "");
4791 SYSCTL_QUAD(_vm, OID_AUTO, lz4_compression_failures, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_compression_failures, "");
4792 SYSCTL_QUAD(_vm, OID_AUTO, lz4_compressed_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_compressed_bytes, "");
4793 SYSCTL_QUAD(_vm, OID_AUTO, lz4_wk_compression_delta, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_wk_compression_delta, "");
4794 SYSCTL_QUAD(_vm, OID_AUTO, lz4_wk_compression_negative_delta, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_wk_compression_negative_delta, "");
4795
4796 SYSCTL_QUAD(_vm, OID_AUTO, lz4_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_decompressions, "");
4797 SYSCTL_QUAD(_vm, OID_AUTO, lz4_decompressed_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_decompressed_bytes, "");
4798
4799 SYSCTL_QUAD(_vm, OID_AUTO, uc_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.uc_decompressions, "");
4800
4801 SYSCTL_QUAD(_vm, OID_AUTO, wk_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_compressions, "");
4802
4803 SYSCTL_QUAD(_vm, OID_AUTO, wk_catime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_cabstime, "");
4804
4805 SYSCTL_QUAD(_vm, OID_AUTO, wkh_catime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wkh_cabstime, "");
4806 SYSCTL_QUAD(_vm, OID_AUTO, wkh_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wkh_compressions, "");
4807
4808 SYSCTL_QUAD(_vm, OID_AUTO, wks_catime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_cabstime, "");
4809 SYSCTL_QUAD(_vm, OID_AUTO, wks_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_compressions, "");
4810
4811 SYSCTL_QUAD(_vm, OID_AUTO, wk_compressions_exclusive, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_compressions_exclusive, "");
4812 SYSCTL_QUAD(_vm, OID_AUTO, wk_sv_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_sv_compressions, "");
4813 SYSCTL_QUAD(_vm, OID_AUTO, wk_mzv_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_mzv_compressions, "");
4814 SYSCTL_QUAD(_vm, OID_AUTO, wk_compression_failures, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_compression_failures, "");
4815 SYSCTL_QUAD(_vm, OID_AUTO, wk_compressed_bytes_exclusive, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_compressed_bytes_exclusive, "");
4816 SYSCTL_QUAD(_vm, OID_AUTO, wk_compressed_bytes_total, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_compressed_bytes_total, "");
4817
4818 SYSCTL_QUAD(_vm, OID_AUTO, wks_compressed_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_compressed_bytes, "");
4819 SYSCTL_QUAD(_vm, OID_AUTO, wks_compression_failures, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_compression_failures, "");
4820 SYSCTL_QUAD(_vm, OID_AUTO, wks_sv_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_sv_compressions, "");
4821
4822
4823 SYSCTL_QUAD(_vm, OID_AUTO, wk_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_decompressions, "");
4824
4825 SYSCTL_QUAD(_vm, OID_AUTO, wk_datime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_dabstime, "");
4826
4827 SYSCTL_QUAD(_vm, OID_AUTO, wkh_datime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wkh_dabstime, "");
4828 SYSCTL_QUAD(_vm, OID_AUTO, wkh_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wkh_decompressions, "");
4829
4830 SYSCTL_QUAD(_vm, OID_AUTO, wks_datime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_dabstime, "");
4831 SYSCTL_QUAD(_vm, OID_AUTO, wks_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_decompressions, "");
4832
4833 SYSCTL_QUAD(_vm, OID_AUTO, wk_decompressed_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_decompressed_bytes, "");
4834 SYSCTL_QUAD(_vm, OID_AUTO, wk_sv_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_sv_decompressions, "");
4835
4836 SYSCTL_INT(_vm, OID_AUTO, lz4_threshold, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_threshold, 0, "");
4837 SYSCTL_INT(_vm, OID_AUTO, wkdm_reeval_threshold, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.wkdm_reeval_threshold, 0, "");
4838 SYSCTL_INT(_vm, OID_AUTO, lz4_max_failure_skips, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_max_failure_skips, 0, "");
4839 SYSCTL_INT(_vm, OID_AUTO, lz4_max_failure_run_length, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_max_failure_run_length, 0, "");
4840 SYSCTL_INT(_vm, OID_AUTO, lz4_max_preselects, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_max_preselects, 0, "");
4841 SYSCTL_INT(_vm, OID_AUTO, lz4_run_preselection_threshold, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_run_preselection_threshold, 0, "");
4842 SYSCTL_INT(_vm, OID_AUTO, lz4_run_continue_bytes, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_run_continue_bytes, 0, "");
4843 SYSCTL_INT(_vm, OID_AUTO, lz4_profitable_bytes, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_profitable_bytes, 0, "");
4844 #if DEVELOPMENT || DEBUG
4845 extern int vm_compressor_current_codec;
4846 extern int vm_compressor_test_seg_wp;
4847 extern boolean_t vm_compressor_force_sw_wkdm;
4848 SYSCTL_INT(_vm, OID_AUTO, compressor_codec, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_compressor_current_codec, 0, "");
4849 SYSCTL_INT(_vm, OID_AUTO, compressor_test_wp, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_compressor_test_seg_wp, 0, "");
4850
4851 SYSCTL_INT(_vm, OID_AUTO, wksw_force, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_compressor_force_sw_wkdm, 0, "");
4852 extern int precompy, wkswhw;
4853
4854 SYSCTL_INT(_vm, OID_AUTO, precompy, CTLFLAG_RW | CTLFLAG_LOCKED, &precompy, 0, "");
4855 SYSCTL_INT(_vm, OID_AUTO, wkswhw, CTLFLAG_RW | CTLFLAG_LOCKED, &wkswhw, 0, "");
4856 extern unsigned int vm_ktrace_enabled;
4857 SYSCTL_INT(_vm, OID_AUTO, vm_ktrace, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_ktrace_enabled, 0, "");
4858 #endif /* DEVELOPMENT || DEBUG */
4859
4860 #if CONFIG_PHANTOM_CACHE
4861 extern uint32_t phantom_cache_thrashing_threshold;
4862 extern uint32_t phantom_cache_eval_period_in_msecs;
4863 extern uint32_t phantom_cache_thrashing_threshold_ssd;
4864
4865
4866 SYSCTL_INT(_vm, OID_AUTO, phantom_cache_eval_period_in_msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &phantom_cache_eval_period_in_msecs, 0, "");
4867 SYSCTL_INT(_vm, OID_AUTO, phantom_cache_thrashing_threshold, CTLFLAG_RW | CTLFLAG_LOCKED, &phantom_cache_thrashing_threshold, 0, "");
4868 SYSCTL_INT(_vm, OID_AUTO, phantom_cache_thrashing_threshold_ssd, CTLFLAG_RW | CTLFLAG_LOCKED, &phantom_cache_thrashing_threshold_ssd, 0, "");
4869 #endif
4870
4871 #if defined(__LP64__)
4872 extern uint32_t vm_page_background_count;
4873 extern uint32_t vm_page_background_target;
4874 extern uint32_t vm_page_background_internal_count;
4875 extern uint32_t vm_page_background_external_count;
4876 extern uint32_t vm_page_background_mode;
4877 extern uint32_t vm_page_background_exclude_external;
4878 extern uint64_t vm_page_background_promoted_count;
4879 extern uint64_t vm_pageout_rejected_bq_internal;
4880 extern uint64_t vm_pageout_rejected_bq_external;
4881
4882 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_mode, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_page_background_mode, 0, "");
4883 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_exclude_external, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_page_background_exclude_external, 0, "");
4884 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_target, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_page_background_target, 0, "");
4885 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_background_count, 0, "");
4886 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_internal_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_background_internal_count, 0, "");
4887 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_external_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_background_external_count, 0, "");
4888
4889 SYSCTL_QUAD(_vm, OID_AUTO, vm_page_background_promoted_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_background_promoted_count, "");
4890 SYSCTL_QUAD(_vm, OID_AUTO, vm_pageout_considered_bq_internal, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_vminfo.vm_pageout_considered_bq_internal, "");
4891 SYSCTL_QUAD(_vm, OID_AUTO, vm_pageout_considered_bq_external, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_vminfo.vm_pageout_considered_bq_external, "");
4892 SYSCTL_QUAD(_vm, OID_AUTO, vm_pageout_rejected_bq_internal, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_rejected_bq_internal, "");
4893 SYSCTL_QUAD(_vm, OID_AUTO, vm_pageout_rejected_bq_external, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_rejected_bq_external, "");
4894
4895 #endif /* __LP64__ */
4896
4897 extern boolean_t vm_darkwake_mode;
4898
4899 STATIC int
4900 sysctl_toggle_darkwake_mode(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4901 {
4902 int new_value, changed;
4903 int error = sysctl_io_number(req, vm_darkwake_mode, sizeof(int), &new_value, &changed);
4904
4905 if (!error && changed) {
4906 if (new_value != 0 && new_value != 1) {
4907 printf("Error: Invalid value passed to darkwake sysctl. Acceptable: 0 or 1.\n");
4908 error = EINVAL;
4909 } else {
4910 vm_update_darkwake_mode((boolean_t) new_value);
4911 }
4912 }
4913
4914 return error;
4915 }
4916
4917 SYSCTL_PROC(_vm, OID_AUTO, darkwake_mode,
4918 CTLTYPE_INT | CTLFLAG_LOCKED | CTLFLAG_RW,
4919 0, 0, sysctl_toggle_darkwake_mode, "I", "");
4920
4921 #if (DEVELOPMENT || DEBUG)
4922
4923 SYSCTL_UINT(_vm, OID_AUTO, vm_page_creation_throttled_hard,
4924 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
4925 &vm_page_creation_throttled_hard, 0, "");
4926
4927 SYSCTL_UINT(_vm, OID_AUTO, vm_page_creation_throttled_soft,
4928 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
4929 &vm_page_creation_throttled_soft, 0, "");
4930
4931 extern uint32_t vm_pageout_memorystatus_fb_factor_nr;
4932 extern uint32_t vm_pageout_memorystatus_fb_factor_dr;
4933 SYSCTL_INT(_vm, OID_AUTO, vm_pageout_memorystatus_fb_factor_nr, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_memorystatus_fb_factor_nr, 0, "");
4934 SYSCTL_INT(_vm, OID_AUTO, vm_pageout_memorystatus_fb_factor_dr, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_memorystatus_fb_factor_dr, 0, "");
4935
4936 extern uint32_t vm_grab_anon_nops;
4937
4938 SYSCTL_INT(_vm, OID_AUTO, vm_grab_anon_overrides, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_debug.vm_grab_anon_overrides, 0, "");
4939 SYSCTL_INT(_vm, OID_AUTO, vm_grab_anon_nops, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_debug.vm_grab_anon_nops, 0, "");
4940 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, "");
4941
4942
4943 extern int vm_page_delayed_work_ctx_needed;
4944 SYSCTL_INT(_vm, OID_AUTO, vm_page_needed_delayed_work_ctx, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_delayed_work_ctx_needed, 0, "");
4945
4946
4947 /* log message counters for persistence mode */
4948 SCALABLE_COUNTER_DECLARE(oslog_p_total_msgcount);
4949 SCALABLE_COUNTER_DECLARE(oslog_p_metadata_saved_msgcount);
4950 SCALABLE_COUNTER_DECLARE(oslog_p_metadata_dropped_msgcount);
4951 SCALABLE_COUNTER_DECLARE(oslog_p_signpost_saved_msgcount);
4952 SCALABLE_COUNTER_DECLARE(oslog_p_signpost_dropped_msgcount);
4953 SCALABLE_COUNTER_DECLARE(oslog_p_error_count);
4954 SCALABLE_COUNTER_DECLARE(oslog_p_error_count);
4955 SCALABLE_COUNTER_DECLARE(oslog_p_saved_msgcount);
4956 SCALABLE_COUNTER_DECLARE(oslog_p_dropped_msgcount);
4957 SCALABLE_COUNTER_DECLARE(oslog_p_boot_dropped_msgcount);
4958 SCALABLE_COUNTER_DECLARE(oslog_p_coprocessor_total_msgcount);
4959 SCALABLE_COUNTER_DECLARE(oslog_p_coprocessor_dropped_msgcount);
4960 SCALABLE_COUNTER_DECLARE(oslog_p_unresolved_kc_msgcount);
4961 SCALABLE_COUNTER_DECLARE(oslog_p_fmt_invalid_msgcount);
4962 SCALABLE_COUNTER_DECLARE(oslog_p_fmt_max_args_msgcount);
4963 SCALABLE_COUNTER_DECLARE(oslog_p_truncated_msgcount);
4964
4965 SCALABLE_COUNTER_DECLARE(oslog_subsystem_count);
4966 SCALABLE_COUNTER_DECLARE(oslog_subsystem_found);
4967 SCALABLE_COUNTER_DECLARE(oslog_subsystem_dropped);
4968
4969 SCALABLE_COUNTER_DECLARE(log_queue_cnt_received);
4970 SCALABLE_COUNTER_DECLARE(log_queue_cnt_rejected_fh);
4971 SCALABLE_COUNTER_DECLARE(log_queue_cnt_sent);
4972 SCALABLE_COUNTER_DECLARE(log_queue_cnt_dropped_nomem);
4973 SCALABLE_COUNTER_DECLARE(log_queue_cnt_queued);
4974 SCALABLE_COUNTER_DECLARE(log_queue_cnt_dropped_off);
4975 SCALABLE_COUNTER_DECLARE(log_queue_cnt_mem_active);
4976 SCALABLE_COUNTER_DECLARE(log_queue_cnt_mem_allocated);
4977 SCALABLE_COUNTER_DECLARE(log_queue_cnt_mem_released);
4978 SCALABLE_COUNTER_DECLARE(log_queue_cnt_mem_failed);
4979
4980 /* log message counters for streaming mode */
4981 SCALABLE_COUNTER_DECLARE(oslog_s_total_msgcount);
4982 SCALABLE_COUNTER_DECLARE(oslog_s_metadata_msgcount);
4983 SCALABLE_COUNTER_DECLARE(oslog_s_error_count);
4984 SCALABLE_COUNTER_DECLARE(oslog_s_streamed_msgcount);
4985 SCALABLE_COUNTER_DECLARE(oslog_s_dropped_msgcount);
4986
4987 /* log message counters for msgbuf logging */
4988 SCALABLE_COUNTER_DECLARE(oslog_msgbuf_msgcount);
4989 SCALABLE_COUNTER_DECLARE(oslog_msgbuf_dropped_msgcount);
4990 extern uint32_t oslog_msgbuf_dropped_charcount;
4991
4992 #if CONFIG_EXCLAVES
4993 /* log message counters for exclaves logging */
4994 SCALABLE_COUNTER_DECLARE(oslog_e_log_count);
4995 SCALABLE_COUNTER_DECLARE(oslog_e_log_dropped_count);
4996 SCALABLE_COUNTER_DECLARE(oslog_e_metadata_count);
4997 SCALABLE_COUNTER_DECLARE(oslog_e_metadata_dropped_count);
4998 SCALABLE_COUNTER_DECLARE(oslog_e_signpost_count);
4999 SCALABLE_COUNTER_DECLARE(oslog_e_signpost_dropped_count);
5000 SCALABLE_COUNTER_DECLARE(oslog_e_replay_failure_count);
5001 SCALABLE_COUNTER_DECLARE(oslog_e_query_count);
5002 SCALABLE_COUNTER_DECLARE(oslog_e_query_error_count);
5003 SCALABLE_COUNTER_DECLARE(oslog_e_trace_mode_set_count);
5004 SCALABLE_COUNTER_DECLARE(oslog_e_trace_mode_error_count);
5005 #endif // CONFIG_EXCLAVES
5006
5007 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_total_msgcount, oslog_p_total_msgcount, "");
5008 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_metadata_saved_msgcount, oslog_p_metadata_saved_msgcount, "");
5009 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_metadata_dropped_msgcount, oslog_p_metadata_dropped_msgcount, "");
5010 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_signpost_saved_msgcount, oslog_p_signpost_saved_msgcount, "");
5011 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_signpost_dropped_msgcount, oslog_p_signpost_dropped_msgcount, "");
5012 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_error_count, oslog_p_error_count, "");
5013 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_saved_msgcount, oslog_p_saved_msgcount, "");
5014 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_dropped_msgcount, oslog_p_dropped_msgcount, "");
5015 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_boot_dropped_msgcount, oslog_p_boot_dropped_msgcount, "");
5016 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_coprocessor_total_msgcount, oslog_p_coprocessor_total_msgcount, "");
5017 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_coprocessor_dropped_msgcount, oslog_p_coprocessor_dropped_msgcount, "");
5018 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_unresolved_kc_msgcount, oslog_p_unresolved_kc_msgcount, "");
5019
5020 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_fmt_invalid_msgcount, oslog_p_fmt_invalid_msgcount, "");
5021 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_fmt_max_args_msgcount, oslog_p_fmt_max_args_msgcount, "");
5022 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_truncated_msgcount, oslog_p_truncated_msgcount, "");
5023
5024 SYSCTL_SCALABLE_COUNTER(_debug, oslog_s_total_msgcount, oslog_s_total_msgcount, "Number of logs sent to streaming");
5025 SYSCTL_SCALABLE_COUNTER(_debug, oslog_s_metadata_msgcount, oslog_s_metadata_msgcount, "Number of metadata sent to streaming");
5026 SYSCTL_SCALABLE_COUNTER(_debug, oslog_s_error_count, oslog_s_error_count, "Number of invalid stream logs");
5027 SYSCTL_SCALABLE_COUNTER(_debug, oslog_s_streamed_msgcount, oslog_s_streamed_msgcount, "Number of streamed logs");
5028 SYSCTL_SCALABLE_COUNTER(_debug, oslog_s_dropped_msgcount, oslog_s_dropped_msgcount, "Number of logs dropped from stream");
5029
5030 SYSCTL_SCALABLE_COUNTER(_debug, oslog_msgbuf_msgcount, oslog_msgbuf_msgcount, "Number of dmesg log messages");
5031 SYSCTL_SCALABLE_COUNTER(_debug, oslog_msgbuf_dropped_msgcount, oslog_msgbuf_dropped_msgcount, "Number of dropped dmesg log messages");
5032 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");
5033
5034 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_received, log_queue_cnt_received, "Number of received logs");
5035 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_rejected_fh, log_queue_cnt_rejected_fh, "Number of logs initially rejected by FH");
5036 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_sent, log_queue_cnt_sent, "Number of logs successfully saved in FH");
5037 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_dropped_nomem, log_queue_cnt_dropped_nomem, "Number of logs dropped due to lack of queue memory");
5038 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_queued, log_queue_cnt_queued, "Current number of logs stored in log queues");
5039 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_dropped_off, log_queue_cnt_dropped_off, "Number of logs dropped due to disabled log queues");
5040 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_mem_allocated, log_queue_cnt_mem_allocated, "Number of memory allocations");
5041 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_mem_released, log_queue_cnt_mem_released, "Number of memory releases");
5042 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_mem_failed, log_queue_cnt_mem_failed, "Number of failed memory allocations");
5043
5044 SYSCTL_SCALABLE_COUNTER(_debug, oslog_subsystem_count, oslog_subsystem_count, "Number of registered log subsystems");
5045 SYSCTL_SCALABLE_COUNTER(_debug, oslog_subsystem_found, oslog_subsystem_found, "Number of sucessful log subsystem lookups");
5046 SYSCTL_SCALABLE_COUNTER(_debug, oslog_subsystem_dropped, oslog_subsystem_dropped, "Number of dropped log subsystem registrations");
5047
5048 #if CONFIG_EXCLAVES
5049 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_metadata_count, oslog_e_metadata_count,
5050 "Number of metadata messages retrieved from the exclaves log server");
5051 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_metadata_dropped_count, oslog_e_metadata_dropped_count,
5052 "Number of dropped metadata messages retrieved from the exclaves log server");
5053 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_log_count, oslog_e_log_count,
5054 "Number of logs retrieved from the exclaves log server");
5055 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_log_dropped_count, oslog_e_log_dropped_count,
5056 "Number of dropeed logs retrieved from the exclaves log server");
5057 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_signpost_count, oslog_e_signpost_count,
5058 "Number of signposts retrieved from the exclaves log server");
5059 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_signpost_dropped_count, oslog_e_signpost_dropped_count,
5060 "Number of dropped signposts retrieved from the exclaves log server");
5061 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_replay_failure_count, oslog_e_replay_failure_count,
5062 "Number of dropped messages that couldn't be replayed and failed generically");
5063 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_query_count, oslog_e_query_count,
5064 "Number of sucessful queries to the exclaves log server");
5065 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_query_error_count, oslog_e_query_error_count,
5066 "Number of failed queries to the exclaves log server");
5067 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_trace_mode_set_count, oslog_e_trace_mode_set_count,
5068 "Number of exclaves trace mode updates");
5069 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_trace_mode_error_count, oslog_e_trace_mode_error_count,
5070 "Number of failed exclaves trace mode updates");
5071 #endif // CONFIG_EXCLAVES
5072
5073 #endif /* DEVELOPMENT || DEBUG */
5074
5075 /*
5076 * Enable tracing of voucher contents
5077 */
5078 extern uint32_t ipc_voucher_trace_contents;
5079
5080 SYSCTL_INT(_kern, OID_AUTO, ipc_voucher_trace_contents,
5081 CTLFLAG_RW | CTLFLAG_LOCKED, &ipc_voucher_trace_contents, 0, "Enable tracing voucher contents");
5082
5083 /*
5084 * Kernel stack size and depth
5085 */
5086 SYSCTL_INT(_kern, OID_AUTO, stack_size,
5087 CTLFLAG_RD | CTLFLAG_LOCKED, (int *) &kernel_stack_size, 0, "Kernel stack size");
5088 SYSCTL_INT(_kern, OID_AUTO, stack_depth_max,
5089 CTLFLAG_RD | CTLFLAG_LOCKED, (int *) &kernel_stack_depth_max, 0, "Max kernel stack depth at interrupt or context switch");
5090
5091 extern unsigned int kern_feature_overrides;
5092 SYSCTL_INT(_kern, OID_AUTO, kern_feature_overrides,
5093 CTLFLAG_RD | CTLFLAG_LOCKED, &kern_feature_overrides, 0, "Kernel feature override mask");
5094
5095 /*
5096 * enable back trace for port allocations
5097 */
5098 extern int ipc_portbt;
5099
5100 SYSCTL_INT(_kern, OID_AUTO, ipc_portbt,
5101 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
5102 &ipc_portbt, 0, "");
5103
5104 /*
5105 * Scheduler sysctls
5106 */
5107
5108 SYSCTL_STRING(_kern, OID_AUTO, sched,
5109 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
5110 sched_string, sizeof(sched_string),
5111 "Timeshare scheduler implementation");
5112
5113 static int
5114 sysctl_cpu_quiescent_counter_interval SYSCTL_HANDLER_ARGS
5115 {
5116 #pragma unused(arg1, arg2)
5117
5118 uint32_t local_min_interval_us = smr_cpu_checkin_get_min_interval_us();
5119
5120 int error = sysctl_handle_int(oidp, &local_min_interval_us, 0, req);
5121 if (error || !req->newptr) {
5122 return error;
5123 }
5124
5125 smr_cpu_checkin_set_min_interval_us(local_min_interval_us);
5126
5127 return 0;
5128 }
5129
5130 SYSCTL_PROC(_kern, OID_AUTO, cpu_checkin_interval,
5131 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
5132 0, 0,
5133 sysctl_cpu_quiescent_counter_interval, "I",
5134 "Quiescent CPU checkin interval (microseconds)");
5135
5136 /*
5137 * Allow the precise user/kernel time sysctl to be set, but don't allow it to
5138 * affect anything. Some tools expect to be able to set this, even though
5139 * runtime configuration is no longer supported.
5140 */
5141
5142 static int
5143 sysctl_precise_user_kernel_time SYSCTL_HANDLER_ARGS
5144 {
5145 #if PRECISE_USER_KERNEL_TIME
5146 int dummy_set = 1;
5147 #else /* PRECISE_USER_KERNEL_TIME */
5148 int dummy_set = 0;
5149 #endif /* !PRECISE_USER_KERNEL_TIME */
5150 return sysctl_handle_int(oidp, &dummy_set, 0, req);
5151 }
5152
5153 SYSCTL_PROC(_kern, OID_AUTO, precise_user_kernel_time,
5154 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
5155 0, 0, sysctl_precise_user_kernel_time, "I",
5156 "Precise accounting of kernel vs. user time (deprecated)");
5157
5158 #if CONFIG_PERVASIVE_ENERGY && HAS_CPU_DPE_COUNTER
5159 __security_const_late static int pervasive_energy = 1;
5160 #else /* CONFIG_PERVASIVE_ENERGY && HAS_CPU_DPE_COUNTER */
5161 __security_const_late static int pervasive_energy = 0;
5162 #endif /* !CONFIG_PERVASIVE_ENERGY || !HAS_CPU_DPE_COUNTER */
5163
5164 SYSCTL_INT(_kern, OID_AUTO, pervasive_energy,
5165 CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED, &pervasive_energy, 0, "");
5166
5167 /* Parameters related to timer coalescing tuning, to be replaced
5168 * with a dedicated systemcall in the future.
5169 */
5170 /* Enable processing pending timers in the context of any other interrupt
5171 * Coalescing tuning parameters for various thread/task attributes */
5172 STATIC int
5173 sysctl_timer_user_us_kernel_abstime SYSCTL_HANDLER_ARGS
5174 {
5175 #pragma unused(oidp)
5176 int size = arg2; /* subcommand*/
5177 int error;
5178 int changed = 0;
5179 uint64_t old_value_ns;
5180 uint64_t new_value_ns;
5181 uint64_t value_abstime;
5182 if (size == sizeof(uint32_t)) {
5183 value_abstime = *((uint32_t *)arg1);
5184 } else if (size == sizeof(uint64_t)) {
5185 value_abstime = *((uint64_t *)arg1);
5186 } else {
5187 return ENOTSUP;
5188 }
5189
5190 absolutetime_to_nanoseconds(value_abstime, &old_value_ns);
5191 error = sysctl_io_number(req, old_value_ns, sizeof(old_value_ns), &new_value_ns, &changed);
5192 if ((error) || (!changed)) {
5193 return error;
5194 }
5195
5196 nanoseconds_to_absolutetime(new_value_ns, &value_abstime);
5197 if (size == sizeof(uint32_t)) {
5198 *((uint32_t *)arg1) = (uint32_t)value_abstime;
5199 } else {
5200 *((uint64_t *)arg1) = value_abstime;
5201 }
5202 return error;
5203 }
5204
5205 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_bg_scale,
5206 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5207 &tcoal_prio_params.timer_coalesce_bg_shift, 0, "");
5208 SYSCTL_PROC(_kern, OID_AUTO, timer_resort_threshold_ns,
5209 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5210 &tcoal_prio_params.timer_resort_threshold_abstime,
5211 sizeof(tcoal_prio_params.timer_resort_threshold_abstime),
5212 sysctl_timer_user_us_kernel_abstime,
5213 "Q", "");
5214 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_bg_ns_max,
5215 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5216 &tcoal_prio_params.timer_coalesce_bg_abstime_max,
5217 sizeof(tcoal_prio_params.timer_coalesce_bg_abstime_max),
5218 sysctl_timer_user_us_kernel_abstime,
5219 "Q", "");
5220
5221 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_kt_scale,
5222 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5223 &tcoal_prio_params.timer_coalesce_kt_shift, 0, "");
5224
5225 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_kt_ns_max,
5226 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5227 &tcoal_prio_params.timer_coalesce_kt_abstime_max,
5228 sizeof(tcoal_prio_params.timer_coalesce_kt_abstime_max),
5229 sysctl_timer_user_us_kernel_abstime,
5230 "Q", "");
5231
5232 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_fp_scale,
5233 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5234 &tcoal_prio_params.timer_coalesce_fp_shift, 0, "");
5235
5236 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_fp_ns_max,
5237 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5238 &tcoal_prio_params.timer_coalesce_fp_abstime_max,
5239 sizeof(tcoal_prio_params.timer_coalesce_fp_abstime_max),
5240 sysctl_timer_user_us_kernel_abstime,
5241 "Q", "");
5242
5243 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_ts_scale,
5244 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5245 &tcoal_prio_params.timer_coalesce_ts_shift, 0, "");
5246
5247 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_ts_ns_max,
5248 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5249 &tcoal_prio_params.timer_coalesce_ts_abstime_max,
5250 sizeof(tcoal_prio_params.timer_coalesce_ts_abstime_max),
5251 sysctl_timer_user_us_kernel_abstime,
5252 "Q", "");
5253
5254 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier0_scale,
5255 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5256 &tcoal_prio_params.latency_qos_scale[0], 0, "");
5257
5258 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier0_ns_max,
5259 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5260 &tcoal_prio_params.latency_qos_abstime_max[0],
5261 sizeof(tcoal_prio_params.latency_qos_abstime_max[0]),
5262 sysctl_timer_user_us_kernel_abstime,
5263 "Q", "");
5264
5265 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier1_scale,
5266 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5267 &tcoal_prio_params.latency_qos_scale[1], 0, "");
5268
5269 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier1_ns_max,
5270 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5271 &tcoal_prio_params.latency_qos_abstime_max[1],
5272 sizeof(tcoal_prio_params.latency_qos_abstime_max[1]),
5273 sysctl_timer_user_us_kernel_abstime,
5274 "Q", "");
5275
5276 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier2_scale,
5277 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5278 &tcoal_prio_params.latency_qos_scale[2], 0, "");
5279
5280 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier2_ns_max,
5281 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5282 &tcoal_prio_params.latency_qos_abstime_max[2],
5283 sizeof(tcoal_prio_params.latency_qos_abstime_max[2]),
5284 sysctl_timer_user_us_kernel_abstime,
5285 "Q", "");
5286
5287 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier3_scale,
5288 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5289 &tcoal_prio_params.latency_qos_scale[3], 0, "");
5290
5291 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier3_ns_max,
5292 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5293 &tcoal_prio_params.latency_qos_abstime_max[3],
5294 sizeof(tcoal_prio_params.latency_qos_abstime_max[3]),
5295 sysctl_timer_user_us_kernel_abstime,
5296 "Q", "");
5297
5298 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier4_scale,
5299 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5300 &tcoal_prio_params.latency_qos_scale[4], 0, "");
5301
5302 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier4_ns_max,
5303 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5304 &tcoal_prio_params.latency_qos_abstime_max[4],
5305 sizeof(tcoal_prio_params.latency_qos_abstime_max[4]),
5306 sysctl_timer_user_us_kernel_abstime,
5307 "Q", "");
5308
5309 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier5_scale,
5310 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5311 &tcoal_prio_params.latency_qos_scale[5], 0, "");
5312
5313 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier5_ns_max,
5314 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5315 &tcoal_prio_params.latency_qos_abstime_max[5],
5316 sizeof(tcoal_prio_params.latency_qos_abstime_max[5]),
5317 sysctl_timer_user_us_kernel_abstime,
5318 "Q", "");
5319
5320 /* Communicate the "user idle level" heuristic to the timer layer, and
5321 * potentially other layers in the future.
5322 */
5323
5324 static int
5325 timer_user_idle_level(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
5326 {
5327 int new_value = 0, old_value = 0, changed = 0, error;
5328
5329 old_value = timer_get_user_idle_level();
5330
5331 error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
5332
5333 if (error == 0 && changed) {
5334 if (timer_set_user_idle_level(new_value) != KERN_SUCCESS) {
5335 error = ERANGE;
5336 }
5337 }
5338
5339 return error;
5340 }
5341
5342 SYSCTL_PROC(_machdep, OID_AUTO, user_idle_level,
5343 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
5344 0, 0,
5345 timer_user_idle_level, "I", "User idle level heuristic, 0-128");
5346
5347 #if DEVELOPMENT || DEBUG
5348 /*
5349 * Basic console mode for games; used for development purposes only.
5350 * Final implementation for this feature (with possible removal of
5351 * sysctl) tracked via rdar://101215873.
5352 */
5353 static int console_mode = 0;
5354 SYSCTL_INT(_kern, OID_AUTO, console_mode,
5355 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_MASKED,
5356 &console_mode, 0, "Game Console Mode");
5357 #endif /* DEVELOPMENT || DEBUG */
5358
5359
5360 #if HYPERVISOR
5361 SYSCTL_INT(_kern, OID_AUTO, hv_support,
5362 CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
5363 &hv_support_available, 0, "");
5364
5365 SYSCTL_INT(_kern, OID_AUTO, hv_disable,
5366 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5367 &hv_disable, 0, "");
5368
5369 #endif /* HYPERVISOR */
5370
5371 #if DEVELOPMENT || DEBUG
5372 extern uint64_t driverkit_checkin_timed_out;
5373 SYSCTL_QUAD(_kern, OID_AUTO, driverkit_checkin_timed_out,
5374 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
5375 &driverkit_checkin_timed_out, "timestamp of dext checkin timeout");
5376 #endif
5377
5378 #if CONFIG_DARKBOOT
5379 STATIC int
5380 sysctl_darkboot SYSCTL_HANDLER_ARGS
5381 {
5382 int err = 0, value = 0;
5383 #pragma unused(oidp, arg1, arg2, err, value, req)
5384
5385 /*
5386 * Handle the sysctl request.
5387 *
5388 * If this is a read, the function will set the value to the current darkboot value. Otherwise,
5389 * we'll get the request identifier into "value" and then we can honor it.
5390 */
5391 if ((err = sysctl_io_number(req, darkboot, sizeof(int), &value, NULL)) != 0) {
5392 goto exit;
5393 }
5394
5395 /* writing requested, let's process the request */
5396 if (req->newptr) {
5397 /* writing is protected by an entitlement */
5398 if (priv_check_cred(kauth_cred_get(), PRIV_DARKBOOT, 0) != 0) {
5399 err = EPERM;
5400 goto exit;
5401 }
5402
5403 switch (value) {
5404 case MEMORY_MAINTENANCE_DARK_BOOT_UNSET:
5405 /*
5406 * If the darkboot sysctl is unset, the NVRAM variable
5407 * must be unset too. If that's not the case, it means
5408 * someone is doing something crazy and not supported.
5409 */
5410 if (darkboot != 0) {
5411 int ret = PERemoveNVRAMProperty(MEMORY_MAINTENANCE_DARK_BOOT_NVRAM_NAME);
5412 if (ret) {
5413 darkboot = 0;
5414 } else {
5415 err = EINVAL;
5416 }
5417 }
5418 break;
5419 case MEMORY_MAINTENANCE_DARK_BOOT_SET:
5420 darkboot = 1;
5421 break;
5422 case MEMORY_MAINTENANCE_DARK_BOOT_SET_PERSISTENT: {
5423 /*
5424 * Set the NVRAM and update 'darkboot' in case
5425 * of success. Otherwise, do not update
5426 * 'darkboot' and report the failure.
5427 */
5428 if (PEWriteNVRAMBooleanProperty(MEMORY_MAINTENANCE_DARK_BOOT_NVRAM_NAME, TRUE)) {
5429 darkboot = 1;
5430 } else {
5431 err = EINVAL;
5432 }
5433
5434 break;
5435 }
5436 default:
5437 err = EINVAL;
5438 }
5439 }
5440
5441 exit:
5442 return err;
5443 }
5444
5445 SYSCTL_PROC(_kern, OID_AUTO, darkboot,
5446 CTLFLAG_KERN | CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_ANYBODY,
5447 0, 0, sysctl_darkboot, "I", "");
5448 #endif /* CONFIG_DARKBOOT */
5449
5450 #if DEVELOPMENT || DEBUG
5451 #include <sys/sysent.h>
5452 /* This should result in a fatal exception, verifying that "sysent" is
5453 * write-protected.
5454 */
5455 static int
5456 kern_sysent_write(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
5457 {
5458 uint64_t new_value = 0, old_value = 0;
5459 int changed = 0, error;
5460
5461 error = sysctl_io_number(req, old_value, sizeof(uint64_t), &new_value, &changed);
5462 if ((error == 0) && changed) {
5463 volatile uint32_t *wraddr = __DECONST(uint32_t *, &sysent[0]);
5464 *wraddr = 0;
5465 printf("sysent[0] write succeeded\n");
5466 }
5467 return error;
5468 }
5469
5470 SYSCTL_PROC(_kern, OID_AUTO, sysent_const_check,
5471 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
5472 0, 0,
5473 kern_sysent_write, "I", "Attempt sysent[0] write");
5474
5475 #endif
5476
5477 #if DEVELOPMENT || DEBUG
5478 SYSCTL_COMPAT_INT(_kern, OID_AUTO, development, CTLFLAG_RD | CTLFLAG_MASKED | CTLFLAG_KERN, NULL, 1, "");
5479 #else
5480 SYSCTL_COMPAT_INT(_kern, OID_AUTO, development, CTLFLAG_RD | CTLFLAG_MASKED, NULL, 0, "");
5481 #endif
5482
5483 SYSCTL_INT(_kern, OID_AUTO, serverperfmode, CTLFLAG_RD, &serverperfmode, 0, "");
5484
5485 #if DEVELOPMENT || DEBUG
5486
5487 decl_lck_spin_data(, spinlock_panic_test_lock);
5488
5489 __attribute__((noreturn))
5490 static void
5491 spinlock_panic_test_acquire_spinlock(void * arg __unused, wait_result_t wres __unused)
5492 {
5493 lck_spin_lock(&spinlock_panic_test_lock);
5494 while (1) {
5495 ;
5496 }
5497 }
5498
5499 static int
5500 sysctl_spinlock_panic_test SYSCTL_HANDLER_ARGS
5501 {
5502 #pragma unused(oidp, arg1, arg2)
5503 if (req->newlen == 0) {
5504 return EINVAL;
5505 }
5506
5507 thread_t panic_spinlock_thread;
5508 /* Initialize panic spinlock */
5509 lck_grp_t * panic_spinlock_grp;
5510 lck_grp_attr_t * panic_spinlock_grp_attr;
5511 lck_attr_t * panic_spinlock_attr;
5512
5513 panic_spinlock_grp_attr = lck_grp_attr_alloc_init();
5514 panic_spinlock_grp = lck_grp_alloc_init("panic_spinlock", panic_spinlock_grp_attr);
5515 panic_spinlock_attr = lck_attr_alloc_init();
5516
5517 lck_spin_init(&spinlock_panic_test_lock, panic_spinlock_grp, panic_spinlock_attr);
5518
5519
5520 /* Create thread to acquire spinlock */
5521 if (kernel_thread_start(spinlock_panic_test_acquire_spinlock, NULL, &panic_spinlock_thread) != KERN_SUCCESS) {
5522 return EBUSY;
5523 }
5524
5525 /* Try to acquire spinlock -- should panic eventually */
5526 lck_spin_lock(&spinlock_panic_test_lock);
5527 while (1) {
5528 ;
5529 }
5530 }
5531
5532 __attribute__((noreturn))
5533 static void
5534 simultaneous_panic_worker
5535 (void * arg, wait_result_t wres __unused)
5536 {
5537 atomic_int *start_panic = (atomic_int *)arg;
5538
5539 while (!atomic_load(start_panic)) {
5540 ;
5541 }
5542 panic("SIMULTANEOUS PANIC TEST: INITIATING PANIC FROM CPU %d", cpu_number());
5543 __builtin_unreachable();
5544 }
5545
5546 static int
5547 sysctl_simultaneous_panic_test SYSCTL_HANDLER_ARGS
5548 {
5549 #pragma unused(oidp, arg1, arg2)
5550 if (req->newlen == 0) {
5551 return EINVAL;
5552 }
5553
5554 int i = 0, threads_to_create = 2 * processor_count;
5555 atomic_int start_panic = 0;
5556 unsigned int threads_created = 0;
5557 thread_t new_panic_thread;
5558
5559 for (i = threads_to_create; i > 0; i--) {
5560 if (kernel_thread_start(simultaneous_panic_worker, (void *) &start_panic, &new_panic_thread) == KERN_SUCCESS) {
5561 threads_created++;
5562 }
5563 }
5564
5565 /* FAIL if we couldn't create at least processor_count threads */
5566 if (threads_created < processor_count) {
5567 panic("SIMULTANEOUS PANIC TEST: FAILED TO CREATE ENOUGH THREADS, ONLY CREATED %d (of %d)",
5568 threads_created, threads_to_create);
5569 }
5570
5571 atomic_exchange(&start_panic, 1);
5572 while (1) {
5573 ;
5574 }
5575 }
5576
5577 extern unsigned int panic_test_failure_mode;
5578 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");
5579
5580 extern unsigned int panic_test_action_count;
5581 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");
5582
5583 extern unsigned int panic_test_case;
5584 SYSCTL_INT(_debug, OID_AUTO, xnu_panic_test_case, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_KERN, &panic_test_case, 0, "panic/debugger testcase");
5585
5586 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");
5587 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");
5588
5589 extern int exc_resource_threads_enabled;
5590 SYSCTL_INT(_kern, OID_AUTO, exc_resource_threads_enabled, CTLFLAG_RW | CTLFLAG_LOCKED, &exc_resource_threads_enabled, 0, "exc_resource thread limit enabled");
5591
5592 extern unsigned int verbose_panic_flow_logging;
5593 SYSCTL_INT(_debug, OID_AUTO, verbose_panic_flow_logging, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_KERN, &verbose_panic_flow_logging, 0, "verbose logging during panic");
5594
5595 #endif /* DEVELOPMENT || DEBUG */
5596
5597 #if BUILT_LTO
5598 static int _built_lto = 1;
5599 #else // BUILT_LTO
5600 static int _built_lto = 0;
5601 #endif // !BUILT_LTO
5602
5603 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");
5604
5605 #if CONFIG_THREAD_GROUPS
5606 #if DEVELOPMENT || DEBUG
5607
5608 static int
5609 sysctl_get_thread_group_id SYSCTL_HANDLER_ARGS
5610 {
5611 #pragma unused(arg1, arg2, oidp)
5612 uint64_t thread_group_id = thread_group_get_id(thread_group_get(current_thread()));
5613 return SYSCTL_OUT(req, &thread_group_id, sizeof(thread_group_id));
5614 }
5615
5616 SYSCTL_PROC(_kern, OID_AUTO, thread_group_id, CTLFLAG_RD | CTLFLAG_LOCKED | CTLTYPE_QUAD,
5617 0, 0, &sysctl_get_thread_group_id, "I", "thread group id of the thread");
5618
5619 extern kern_return_t sysctl_clutch_thread_group_cpu_time_for_thread(thread_t thread, int sched_bucket, uint64_t *cpu_stats);
5620
5621 static int
5622 sysctl_get_clutch_bucket_group_cpu_stats SYSCTL_HANDLER_ARGS
5623 {
5624 int error;
5625 kern_return_t kr;
5626 int sched_bucket = -1;
5627 error = SYSCTL_IN(req, &sched_bucket, sizeof(sched_bucket));
5628 if (error) {
5629 return error;
5630 }
5631 uint64_t cpu_stats[2];
5632 kr = sysctl_clutch_thread_group_cpu_time_for_thread(current_thread(), sched_bucket, cpu_stats);
5633 error = mach_to_bsd_errno(kr);
5634 if (error) {
5635 return error;
5636 }
5637 return SYSCTL_OUT(req, cpu_stats, sizeof(cpu_stats));
5638 }
5639
5640 SYSCTL_PROC(_kern, OID_AUTO, clutch_bucket_group_cpu_stats, CTLFLAG_RW | CTLFLAG_LOCKED | CTLTYPE_OPAQUE,
5641 0, 0, &sysctl_get_clutch_bucket_group_cpu_stats, "I",
5642 "CPU used and blocked time for the current thread group at a specified scheduling bucket");
5643
5644 STATIC int
5645 sysctl_thread_group_count(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
5646 {
5647 int value = thread_group_count();
5648 return sysctl_io_number(req, value, sizeof(value), NULL, NULL);
5649 }
5650
5651 SYSCTL_PROC(_kern, OID_AUTO, thread_group_count, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5652 0, 0, &sysctl_thread_group_count, "I", "count of thread groups");
5653
5654 #endif /* DEVELOPMENT || DEBUG */
5655 const uint32_t thread_groups_supported = 1;
5656 #else /* CONFIG_THREAD_GROUPS */
5657 const uint32_t thread_groups_supported = 0;
5658 #endif /* CONFIG_THREAD_GROUPS */
5659
5660 STATIC int
5661 sysctl_thread_groups_supported(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
5662 {
5663 int value = thread_groups_supported;
5664 return sysctl_io_number(req, value, sizeof(value), NULL, NULL);
5665 }
5666
5667 SYSCTL_PROC(_kern, OID_AUTO, thread_groups_supported, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5668 0, 0, &sysctl_thread_groups_supported, "I", "thread groups supported");
5669
5670 static int
5671 sysctl_grade_cputype SYSCTL_HANDLER_ARGS
5672 {
5673 #pragma unused(arg1, arg2, oidp)
5674 int error = 0;
5675 int type_tuple[2] = {};
5676 int return_value = 0;
5677
5678 error = SYSCTL_IN(req, &type_tuple, sizeof(type_tuple));
5679
5680 if (error) {
5681 return error;
5682 }
5683
5684 return_value = grade_binary(type_tuple[0], type_tuple[1] & ~CPU_SUBTYPE_MASK, type_tuple[1] & CPU_SUBTYPE_MASK, FALSE);
5685
5686 error = SYSCTL_OUT(req, &return_value, sizeof(return_value));
5687
5688 if (error) {
5689 return error;
5690 }
5691
5692 return error;
5693 }
5694
5695 SYSCTL_PROC(_kern, OID_AUTO, grade_cputype,
5696 CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_MASKED | CTLFLAG_LOCKED | CTLTYPE_OPAQUE,
5697 0, 0, &sysctl_grade_cputype, "S",
5698 "grade value of cpu_type_t+cpu_sub_type_t");
5699
5700
5701 #if DEVELOPMENT || DEBUG
5702 STATIC int
5703 sysctl_binary_grade_override( __unused struct sysctl_oid *oidp, __unused void *arg1,
5704 __unused int arg2, struct sysctl_req *req)
5705 {
5706 int error;
5707 user_addr_t oldp = 0, newp = 0;
5708 size_t *oldlenp = NULL;
5709 size_t newlen = 0;
5710
5711 oldp = req->oldptr;
5712 oldlenp = &(req->oldlen);
5713 newp = req->newptr;
5714 newlen = req->newlen;
5715
5716 /* We want the current length, and maybe the string itself */
5717 if (oldlenp) {
5718 char existing_overrides[256] = { 0 };
5719
5720 size_t currlen = bingrade_get_override_string(existing_overrides, sizeof(existing_overrides));
5721
5722 if (oldp && currlen > 0) {
5723 if (*oldlenp < currlen) {
5724 return ENOMEM;
5725 }
5726 /* NOTE - we do not copy the NULL terminator */
5727 error = copyout(existing_overrides, oldp, currlen);
5728 if (error) {
5729 return error;
5730 }
5731 }
5732 /* return length of overrides minus the NULL terminator (just like strlen) */
5733 req->oldidx = currlen;
5734 }
5735
5736 /* We want to set the override string to something */
5737 if (newp) {
5738 char *tmp_override = (char *)kalloc_data(newlen + 1, Z_WAITOK | Z_ZERO);
5739 if (!tmp_override) {
5740 return ENOMEM;
5741 }
5742
5743 error = copyin(newp, tmp_override, newlen);
5744 if (error) {
5745 kfree_data(tmp_override, newlen + 1);
5746 return error;
5747 }
5748
5749 tmp_override[newlen] = 0; /* Terminate string */
5750
5751 /* Set the binary grading overrides */
5752 if (binary_grade_overrides_update(tmp_override) == 0) {
5753 /* Nothing got set. */
5754 kfree_data(tmp_override, newlen + 1);
5755 return EINVAL;
5756 }
5757
5758 kfree_data(tmp_override, newlen + 1);
5759 }
5760
5761 return 0;
5762 }
5763
5764
5765 SYSCTL_PROC(_kern, OID_AUTO, grade_override,
5766 CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_LOCKED,
5767 0, 0, &sysctl_binary_grade_override, "A",
5768 "");
5769 #endif /* DEVELOPMENT || DEBUG */
5770
5771 extern boolean_t allow_direct_handoff;
5772 SYSCTL_INT(_kern, OID_AUTO, direct_handoff,
5773 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5774 &allow_direct_handoff, 0, "Enable direct handoff for realtime threads");
5775
5776 #if DEVELOPMENT || DEBUG
5777
5778 SYSCTL_QUAD(_kern, OID_AUTO, phys_carveout_pa, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5779 &phys_carveout_pa,
5780 "base physical address of the phys_carveout_mb boot-arg region");
5781 SYSCTL_QUAD(_kern, OID_AUTO, phys_carveout_va, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5782 &phys_carveout,
5783 "base virtual address of the phys_carveout_mb boot-arg region");
5784 SYSCTL_QUAD(_kern, OID_AUTO, phys_carveout_size, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5785 &phys_carveout_size,
5786 "size in bytes of the phys_carveout_mb boot-arg region");
5787
5788
5789
5790 static int
5791 cseg_wedge_thread SYSCTL_HANDLER_ARGS
5792 {
5793 #pragma unused(arg1, arg2)
5794
5795 int error, val = 0;
5796 error = sysctl_handle_int(oidp, &val, 0, req);
5797 if (error || val == 0) {
5798 return error;
5799 }
5800
5801 do_cseg_wedge_thread();
5802 return 0;
5803 }
5804 SYSCTL_PROC(_kern, OID_AUTO, cseg_wedge_thread, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_MASKED, 0, 0, cseg_wedge_thread, "I", "wedge c_seg thread");
5805
5806 static int
5807 cseg_unwedge_thread SYSCTL_HANDLER_ARGS
5808 {
5809 #pragma unused(arg1, arg2)
5810
5811 int error, val = 0;
5812 error = sysctl_handle_int(oidp, &val, 0, req);
5813 if (error || val == 0) {
5814 return error;
5815 }
5816
5817 do_cseg_unwedge_thread();
5818 return 0;
5819 }
5820 SYSCTL_PROC(_kern, OID_AUTO, cseg_unwedge_thread, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_MASKED, 0, 0, cseg_unwedge_thread, "I", "unstuck c_seg thread");
5821
5822 static atomic_int wedge_thread_should_wake = 0;
5823
5824 static int
5825 unwedge_thread SYSCTL_HANDLER_ARGS
5826 {
5827 #pragma unused(arg1, arg2)
5828 int error, val = 0;
5829 error = sysctl_handle_int(oidp, &val, 0, req);
5830 if (error || val == 0) {
5831 return error;
5832 }
5833
5834 atomic_store(&wedge_thread_should_wake, 1);
5835 return 0;
5836 }
5837
5838 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");
5839
5840 static int
5841 wedge_thread SYSCTL_HANDLER_ARGS
5842 {
5843 #pragma unused(arg1, arg2)
5844
5845 int error, val = 0;
5846 error = sysctl_handle_int(oidp, &val, 0, req);
5847 if (error || val == 0) {
5848 return error;
5849 }
5850
5851 uint64_t interval = 1;
5852 nanoseconds_to_absolutetime(1000 * 1000 * 50, &interval);
5853
5854 atomic_store(&wedge_thread_should_wake, 0);
5855 while (!atomic_load(&wedge_thread_should_wake)) {
5856 tsleep1(NULL, 0, "wedge_thread", mach_absolute_time() + interval, NULL);
5857 }
5858
5859 return 0;
5860 }
5861
5862 SYSCTL_PROC(_kern, OID_AUTO, wedge_thread,
5863 CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED, 0, 0, wedge_thread, "I",
5864 "wedge this thread so it cannot be cleaned up");
5865
5866 static int
5867 sysctl_total_corpses_count SYSCTL_HANDLER_ARGS
5868 {
5869 #pragma unused(oidp, arg1, arg2)
5870 extern unsigned long total_corpses_count(void);
5871
5872 unsigned long corpse_count_long = total_corpses_count();
5873 unsigned int corpse_count = (unsigned int)MIN(corpse_count_long, UINT_MAX);
5874 return sysctl_io_opaque(req, &corpse_count, sizeof(corpse_count), NULL);
5875 }
5876
5877 SYSCTL_PROC(_kern, OID_AUTO, total_corpses_count,
5878 CTLFLAG_RD | CTLFLAG_ANYBODY | CTLFLAG_LOCKED, 0, 0,
5879 sysctl_total_corpses_count, "I", "total corpses on the system");
5880
5881 static int
5882 sysctl_turnstile_test_prim_lock SYSCTL_HANDLER_ARGS;
5883 static int
5884 sysctl_turnstile_test_prim_unlock SYSCTL_HANDLER_ARGS;
5885 int
5886 tstile_test_prim_lock(boolean_t use_hashtable);
5887 int
5888 tstile_test_prim_unlock(boolean_t use_hashtable);
5889
5890 static int
5891 sysctl_turnstile_test_prim_lock SYSCTL_HANDLER_ARGS
5892 {
5893 #pragma unused(arg1, arg2)
5894 int error, val = 0;
5895 error = sysctl_handle_int(oidp, &val, 0, req);
5896 if (error || val == 0) {
5897 return error;
5898 }
5899 switch (val) {
5900 case SYSCTL_TURNSTILE_TEST_USER_DEFAULT:
5901 case SYSCTL_TURNSTILE_TEST_USER_HASHTABLE:
5902 case SYSCTL_TURNSTILE_TEST_KERNEL_DEFAULT:
5903 case SYSCTL_TURNSTILE_TEST_KERNEL_HASHTABLE:
5904 return tstile_test_prim_lock(val);
5905 default:
5906 return error;
5907 }
5908 }
5909
5910 static int
5911 sysctl_turnstile_test_prim_unlock SYSCTL_HANDLER_ARGS
5912 {
5913 #pragma unused(arg1, arg2)
5914 int error, val = 0;
5915 error = sysctl_handle_int(oidp, &val, 0, req);
5916 if (error || val == 0) {
5917 return error;
5918 }
5919 switch (val) {
5920 case SYSCTL_TURNSTILE_TEST_USER_DEFAULT:
5921 case SYSCTL_TURNSTILE_TEST_USER_HASHTABLE:
5922 case SYSCTL_TURNSTILE_TEST_KERNEL_DEFAULT:
5923 case SYSCTL_TURNSTILE_TEST_KERNEL_HASHTABLE:
5924 return tstile_test_prim_unlock(val);
5925 default:
5926 return error;
5927 }
5928 }
5929
5930 SYSCTL_PROC(_kern, OID_AUTO, turnstiles_test_lock, CTLFLAG_WR | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED,
5931 0, 0, sysctl_turnstile_test_prim_lock, "I", "turnstiles test lock");
5932
5933 SYSCTL_PROC(_kern, OID_AUTO, turnstiles_test_unlock, CTLFLAG_WR | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED,
5934 0, 0, sysctl_turnstile_test_prim_unlock, "I", "turnstiles test unlock");
5935
5936 int
5937 turnstile_get_boost_stats_sysctl(void *req);
5938 int
5939 turnstile_get_unboost_stats_sysctl(void *req);
5940 static int
5941 sysctl_turnstile_boost_stats SYSCTL_HANDLER_ARGS;
5942 static int
5943 sysctl_turnstile_unboost_stats SYSCTL_HANDLER_ARGS;
5944 extern uint64_t thread_block_on_turnstile_count;
5945 extern uint64_t thread_block_on_regular_waitq_count;
5946
5947 static int
5948 sysctl_turnstile_boost_stats SYSCTL_HANDLER_ARGS
5949 {
5950 #pragma unused(arg1, arg2, oidp)
5951 return turnstile_get_boost_stats_sysctl(req);
5952 }
5953
5954 static int
5955 sysctl_turnstile_unboost_stats SYSCTL_HANDLER_ARGS
5956 {
5957 #pragma unused(arg1, arg2, oidp)
5958 return turnstile_get_unboost_stats_sysctl(req);
5959 }
5960
5961 SYSCTL_PROC(_kern, OID_AUTO, turnstile_boost_stats, CTLFLAG_RD | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED | CTLTYPE_STRUCT,
5962 0, 0, sysctl_turnstile_boost_stats, "S", "turnstiles boost stats");
5963 SYSCTL_PROC(_kern, OID_AUTO, turnstile_unboost_stats, CTLFLAG_RD | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED | CTLTYPE_STRUCT,
5964 0, 0, sysctl_turnstile_unboost_stats, "S", "turnstiles unboost stats");
5965 SYSCTL_QUAD(_kern, OID_AUTO, thread_block_count_on_turnstile,
5966 CTLFLAG_RD | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED,
5967 &thread_block_on_turnstile_count, "thread blocked on turnstile count");
5968 SYSCTL_QUAD(_kern, OID_AUTO, thread_block_count_on_reg_waitq,
5969 CTLFLAG_RD | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED,
5970 &thread_block_on_regular_waitq_count, "thread blocked on regular waitq count");
5971
5972 #if CONFIG_PV_TICKET
5973
5974 extern int ticket_lock_spins;
5975 SYSCTL_INT(_kern, OID_AUTO, ticket_lock_spins,
5976 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5977 &ticket_lock_spins, 0, "loops before hypercall");
5978
5979 #if (DEBUG || DEVELOPMENT)
5980
5981 /* PV ticket lock stats */
5982
5983 SYSCTL_SCALABLE_COUNTER(_kern, ticket_lock_kicks, ticket_kick_count,
5984 "ticket lock kicks");
5985 SYSCTL_SCALABLE_COUNTER(_kern, ticket_lock_waits, ticket_wait_count,
5986 "ticket lock waits");
5987 SYSCTL_SCALABLE_COUNTER(_kern, ticket_lock_already, ticket_already_count,
5988 "ticket lock already unlocked");
5989 SYSCTL_SCALABLE_COUNTER(_kern, ticket_lock_just_unlock, ticket_just_unlock,
5990 "ticket unlock without kick");
5991 SYSCTL_SCALABLE_COUNTER(_kern, ticket_lock_wflag_cleared, ticket_wflag_cleared,
5992 "ticket lock wait flag cleared");
5993 SYSCTL_SCALABLE_COUNTER(_kern, ticket_lock_wflag_still, ticket_wflag_still,
5994 "ticket lock wait flag not cleared");
5995 SYSCTL_SCALABLE_COUNTER(_kern, ticket_lock_spin_count, ticket_spin_count,
5996 "ticket lock spin count");
5997
5998 /* sysctl kern.hcall_probe=n -- does hypercall #n exist? */
5999
6000 static int
6001 sysctl_hcall_probe SYSCTL_HANDLER_ARGS
6002 {
6003 char instr[20];
6004
6005 if (!req->newptr) {
6006 return 0;
6007 }
6008 if (req->newlen >= sizeof(instr)) {
6009 return EOVERFLOW;
6010 }
6011
6012 int error = SYSCTL_IN(req, instr, req->newlen);
6013 if (error) {
6014 return error;
6015 }
6016 instr[req->newlen] = '\0';
6017
6018 int hcall = 0;
6019 error = sscanf(instr, "%d", &hcall);
6020 if (error != 1 || hcall < 0) {
6021 return EINVAL;
6022 }
6023 uprintf("%savailable\n",
6024 hvg_is_hcall_available((hvg_hcall_code_t)hcall) ? "" : "not ");
6025 return 0;
6026 }
6027
6028 SYSCTL_PROC(_kern, OID_AUTO, hcall_probe,
6029 CTLTYPE_STRING | CTLFLAG_WR | CTLFLAG_LOCKED | CTLFLAG_MASKED,
6030 0, 0, sysctl_hcall_probe, "A", "probe hypercall by id");
6031
6032 #endif /* (DEBUG || DEVELOPMENT) */
6033 #endif /* CONFIG_PV_TICKET */
6034
6035 #if defined(__x86_64__)
6036 extern uint64_t MutexSpin;
6037
6038 SYSCTL_QUAD(_kern, OID_AUTO, mutex_spin_abs, CTLFLAG_RW, &MutexSpin,
6039 "Spin time in abs for acquiring a kernel mutex");
6040 #else
6041 extern machine_timeout_t MutexSpin;
6042
6043 SYSCTL_QUAD(_kern, OID_AUTO, mutex_spin_abs, CTLFLAG_RW, &MutexSpin,
6044 "Spin time in abs for acquiring a kernel mutex");
6045 #endif
6046
6047 extern uint64_t low_MutexSpin;
6048 extern int64_t high_MutexSpin;
6049 extern unsigned int real_ncpus;
6050
6051 SYSCTL_QUAD(_kern, OID_AUTO, low_mutex_spin_abs, CTLFLAG_RW, &low_MutexSpin,
6052 "Low spin threshold in abs for acquiring a kernel mutex");
6053
6054 static int
6055 sysctl_high_mutex_spin_ns SYSCTL_HANDLER_ARGS
6056 {
6057 #pragma unused(oidp, arg1, arg2)
6058 int error;
6059 int64_t val = 0;
6060 int64_t res;
6061
6062 /* Check if the user is writing to high_MutexSpin, or just reading it */
6063 if (req->newptr) {
6064 error = SYSCTL_IN(req, &val, sizeof(val));
6065 if (error || (val < 0 && val != -1)) {
6066 return error;
6067 }
6068 high_MutexSpin = val;
6069 }
6070
6071 if (high_MutexSpin >= 0) {
6072 res = high_MutexSpin;
6073 } else {
6074 res = low_MutexSpin * real_ncpus;
6075 }
6076 return SYSCTL_OUT(req, &res, sizeof(res));
6077 }
6078 SYSCTL_PROC(_kern, OID_AUTO, high_mutex_spin_abs, CTLFLAG_RW | CTLTYPE_QUAD, 0, 0, sysctl_high_mutex_spin_ns, "I",
6079 "High spin threshold in abs for acquiring a kernel mutex");
6080
6081 #if defined (__x86_64__)
6082
6083 semaphore_t sysctl_test_panic_with_thread_sem;
6084
6085 #pragma clang diagnostic push
6086 #pragma clang diagnostic ignored "-Winfinite-recursion" /* rdar://38801963 */
6087 __attribute__((noreturn))
6088 static void
6089 panic_thread_test_child_spin(void * arg, wait_result_t wres)
6090 {
6091 static int panic_thread_recurse_count = 5;
6092
6093 if (panic_thread_recurse_count > 0) {
6094 panic_thread_recurse_count--;
6095 panic_thread_test_child_spin(arg, wres);
6096 }
6097
6098 semaphore_signal(sysctl_test_panic_with_thread_sem);
6099 while (1) {
6100 ;
6101 }
6102 }
6103 #pragma clang diagnostic pop
6104
6105 static void
6106 panic_thread_test_child_park(void * arg __unused, wait_result_t wres __unused)
6107 {
6108 int event;
6109
6110 assert_wait(&event, THREAD_UNINT);
6111 semaphore_signal(sysctl_test_panic_with_thread_sem);
6112 thread_block(panic_thread_test_child_park);
6113 }
6114
6115 static int
6116 sysctl_test_panic_with_thread SYSCTL_HANDLER_ARGS
6117 {
6118 #pragma unused(arg1, arg2)
6119 int rval = 0;
6120 char str[16] = { '\0' };
6121 thread_t child_thread = THREAD_NULL;
6122
6123 rval = sysctl_handle_string(oidp, str, sizeof(str), req);
6124 if (rval != 0 || !req->newptr) {
6125 return EINVAL;
6126 }
6127
6128 semaphore_create(kernel_task, &sysctl_test_panic_with_thread_sem, SYNC_POLICY_FIFO, 0);
6129
6130 /* Create thread to spin or park in continuation */
6131 if (strncmp("spin", str, strlen("spin")) == 0) {
6132 if (kernel_thread_start(panic_thread_test_child_spin, NULL, &child_thread) != KERN_SUCCESS) {
6133 semaphore_destroy(kernel_task, sysctl_test_panic_with_thread_sem);
6134 return EBUSY;
6135 }
6136 } else if (strncmp("continuation", str, strlen("continuation")) == 0) {
6137 if (kernel_thread_start(panic_thread_test_child_park, NULL, &child_thread) != KERN_SUCCESS) {
6138 semaphore_destroy(kernel_task, sysctl_test_panic_with_thread_sem);
6139 return EBUSY;
6140 }
6141 } else {
6142 semaphore_destroy(kernel_task, sysctl_test_panic_with_thread_sem);
6143 return EINVAL;
6144 }
6145
6146 semaphore_wait(sysctl_test_panic_with_thread_sem);
6147
6148 panic_with_thread_context(0, NULL, 0, child_thread, "testing panic_with_thread_context for thread %p", child_thread);
6149
6150 /* Not reached */
6151 return EINVAL;
6152 }
6153
6154 SYSCTL_PROC(_kern, OID_AUTO, test_panic_with_thread,
6155 CTLFLAG_MASKED | CTLFLAG_KERN | CTLFLAG_LOCKED | CTLFLAG_WR | CTLTYPE_STRING,
6156 0, 0, sysctl_test_panic_with_thread, "A", "test panic flow for backtracing a different thread");
6157 #endif /* defined (__x86_64__) */
6158
6159 static int
6160 sysctl_generate_file_permissions_guard_exception SYSCTL_HANDLER_ARGS
6161 {
6162 #pragma unused(arg1, arg2)
6163 int error, val = 0;
6164 error = sysctl_handle_int(oidp, &val, 0, req);
6165 if (error || val == 0) {
6166 return error;
6167 }
6168 generate_file_permissions_guard_exception(0, val);
6169 return 0;
6170 }
6171
6172 SYSCTL_PROC(_kern, OID_AUTO, file_perm_guard_exception, CTLFLAG_WR | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED,
6173 0, 0, sysctl_generate_file_permissions_guard_exception, "I", "Test File Permission Guard exception");
6174
6175 #endif /* DEVELOPMENT || DEBUG */
6176
6177 extern const int copysize_limit_panic;
6178 static int
6179 sysctl_get_owned_vmobjects SYSCTL_HANDLER_ARGS
6180 {
6181 #pragma unused(oidp, arg1, arg2)
6182
6183 /* validate */
6184 if (req->newlen != sizeof(mach_port_name_t) || req->newptr == USER_ADDR_NULL ||
6185 req->oldidx != 0 || req->newidx != 0 || req->p == NULL ||
6186 (req->oldlen == 0 && req->oldptr != USER_ADDR_NULL)) {
6187 return EINVAL;
6188 }
6189
6190 int error;
6191 mach_port_name_t task_port_name;
6192 task_t task;
6193 size_t buffer_size = (req->oldptr != USER_ADDR_NULL) ? req->oldlen : 0;
6194 vmobject_list_output_t buffer = NULL;
6195 size_t output_size;
6196 size_t entries;
6197 bool free_buffer = false;
6198
6199 /* we have a "newptr" (for write) we get a task port name from the caller. */
6200 error = SYSCTL_IN(req, &task_port_name, sizeof(mach_port_name_t));
6201
6202 if (error != 0) {
6203 goto sysctl_get_vmobject_list_exit;
6204 }
6205
6206 task = port_name_to_task_read(task_port_name);
6207 if (task == TASK_NULL) {
6208 error = ESRCH;
6209 goto sysctl_get_vmobject_list_exit;
6210 }
6211
6212 bool corpse = task_is_a_corpse(task);
6213
6214 /* get the current size */
6215 size_t max_size;
6216 task_get_owned_vmobjects(task, 0, NULL, &max_size, &entries);
6217
6218 if (buffer_size && (buffer_size < sizeof(*buffer) + sizeof(vm_object_query_data_t))) {
6219 error = ENOMEM;
6220 goto sysctl_get_vmobject_list_deallocate_and_exit;
6221 }
6222
6223 if (corpse == false) {
6224 /* copy the vmobjects and vmobject data out of the task */
6225 if (buffer_size == 0) {
6226 output_size = max_size;
6227 } else {
6228 buffer_size = (buffer_size > max_size) ? max_size : buffer_size;
6229 buffer = (struct _vmobject_list_output_ *)kalloc_data(buffer_size, Z_WAITOK);
6230
6231 if (!buffer) {
6232 error = ENOMEM;
6233 goto sysctl_get_vmobject_list_deallocate_and_exit;
6234 }
6235 free_buffer = true;
6236
6237 task_get_owned_vmobjects(task, buffer_size, buffer, &output_size, &entries);
6238 }
6239 } else {
6240 vmobject_list_output_t list;
6241
6242 task_get_corpse_vmobject_list(task, &list, &max_size);
6243 assert(buffer == NULL);
6244
6245 /* copy corpse_vmobject_list to output buffer to avoid double copy */
6246 if (buffer_size) {
6247 size_t temp_size;
6248
6249 temp_size = buffer_size > max_size ? max_size : buffer_size;
6250 output_size = temp_size - sizeof(*buffer);
6251 /* whole multiple of vm_object_query_data_t */
6252 output_size = (output_size / sizeof(vm_object_query_data_t)) * sizeof(vm_object_query_data_t) + sizeof(*buffer);
6253 buffer = list;
6254 } else {
6255 output_size = max_size;
6256 }
6257 }
6258
6259 /* req->oldptr should be USER_ADDR_NULL if buffer == NULL and return the current size */
6260 /* otherwise copy buffer to oldptr and return the bytes copied */
6261 size_t num_copied, chunk_size;
6262 for (num_copied = 0, chunk_size = 0;
6263 num_copied < output_size;
6264 num_copied += chunk_size) {
6265 chunk_size = MIN(output_size - num_copied, copysize_limit_panic);
6266 error = SYSCTL_OUT(req, (char *)buffer + num_copied, chunk_size);
6267 if (error) {
6268 break;
6269 }
6270 }
6271
6272 sysctl_get_vmobject_list_deallocate_and_exit:
6273 task_deallocate(task);
6274
6275 sysctl_get_vmobject_list_exit:
6276 if (free_buffer) {
6277 kfree_data(buffer, buffer_size);
6278 }
6279
6280 return error;
6281 }
6282
6283 SYSCTL_PROC(_vm, OID_AUTO, get_owned_vmobjects,
6284 CTLTYPE_OPAQUE | CTLFLAG_RD | CTLFLAG_WR | CTLFLAG_MASKED | CTLFLAG_KERN | CTLFLAG_LOCKED | CTLFLAG_ANYBODY,
6285 0, 0, sysctl_get_owned_vmobjects, "A", "get owned vmobjects in task");
6286
6287 extern uint64_t num_static_scalable_counters;
6288 SYSCTL_QUAD(_kern, OID_AUTO, num_static_scalable_counters, CTLFLAG_RD | CTLFLAG_LOCKED, &num_static_scalable_counters, "");
6289
6290 #if SCHED_HYGIENE_DEBUG
6291 TUNABLE_DT(bool, sched_hygiene_nonspec_tb, "machine-timeouts", "nonspec-tb", "sched-hygiene-nonspec-tb", false, TUNABLE_DT_NONE);
6292 static SECURITY_READ_ONLY_LATE(int) sched_hygiene_debug_available = 1;
6293 #else
6294 static SECURITY_READ_ONLY_LATE(int) sched_hygiene_debug_available = 0;
6295 #endif /* SCHED_HYGIENE_DEBUG */
6296
6297 SYSCTL_INT(_debug, OID_AUTO, sched_hygiene_debug_available,
6298 CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
6299 &sched_hygiene_debug_available, 0, "");
6300
6301 uuid_string_t trial_treatment_id;
6302 uuid_string_t trial_experiment_id;
6303 int trial_deployment_id = -1;
6304
6305 SYSCTL_STRING(_kern, OID_AUTO, trial_treatment_id, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_ANYBODY | CTLFLAG_LEGACY_EXPERIMENT, trial_treatment_id, sizeof(trial_treatment_id), "");
6306 SYSCTL_STRING(_kern, OID_AUTO, trial_experiment_id, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_ANYBODY | CTLFLAG_LEGACY_EXPERIMENT, trial_experiment_id, sizeof(trial_experiment_id), "");
6307 SYSCTL_INT(_kern, OID_AUTO, trial_deployment_id, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_ANYBODY | CTLFLAG_LEGACY_EXPERIMENT, &trial_deployment_id, 0, "");
6308
6309 #if (DEVELOPMENT || DEBUG)
6310 /* For unit testing setting factors & limits. */
6311 unsigned int testing_experiment_factor;
6312 EXPERIMENT_FACTOR_LEGACY_UINT(_kern, testing_experiment_factor, &testing_experiment_factor, 5, 10, "");
6313
6314 static int32_t experiment_factor_test;
6315 EXPERIMENT_FACTOR_INT(test, &experiment_factor_test, 0, 32, "test factor");
6316
6317 #if MACH_ASSERT && __arm64__
6318 /* rdar://149041040 */
6319 extern unsigned int panic_on_jit_guard;
6320 EXPERIMENT_FACTOR_UINT(jitguard, &panic_on_jit_guard, 0, 7, "Panic on JIT guard failure");
6321 #endif /* MACH_ASSERT && __arm64__ */
6322
6323 extern int exception_log_max_pid;
6324 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");
6325 #endif /* (DEVELOPMENT || DEBUG) */
6326
6327 #if DEVELOPMENT || DEBUG
6328 static int
6329 unlink_kernelcore_sysctl SYSCTL_HANDLER_ARGS
6330 {
6331 if (!req->newptr) {
6332 return EINVAL;
6333 }
6334 void IOBSDLowSpaceUnlinkKernelCore(void);
6335 IOBSDLowSpaceUnlinkKernelCore();
6336 return 0;
6337 }
6338
6339 SYSCTL_PROC(_kern, OID_AUTO, unlink_kernelcore,
6340 CTLTYPE_INT | CTLFLAG_WR | CTLFLAG_LOCKED | CTLFLAG_MASKED, 0, 0,
6341 unlink_kernelcore_sysctl, "-", "unlink the kernelcore file");
6342 #endif /* DEVELOPMENT || DEBUG */
6343
6344 #if CONFIG_IOTRACE
6345 #pragma clang diagnostic push
6346 #pragma clang diagnostic ignored "-Wcast-qual"
6347 SYSCTL_INT(_debug, OID_AUTO, MMIOtrace,
6348 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
6349 (int *)&mmiotrace_enabled, 0, "");
6350 #pragma clang diagnostic pop
6351 #endif /* CONFIG_IOTRACE */
6352
6353 static int
6354 sysctl_page_protection_type SYSCTL_HANDLER_ARGS
6355 {
6356 #pragma unused(oidp, arg1, arg2)
6357 int value = ml_page_protection_type();
6358 return SYSCTL_OUT(req, &value, sizeof(value));
6359 }
6360
6361 SYSCTL_PROC(_kern, OID_AUTO, page_protection_type,
6362 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
6363 0, 0, sysctl_page_protection_type, "I", "Type of page protection that the system supports");
6364
6365 #if CONFIG_SPTM && HAS_SPTM_SYSCTL
6366 extern bool disarm_protected_io;
6367 static int sysctl_sptm_disarm_protected_io SYSCTL_HANDLER_ARGS
6368 {
6369 int error = 0;
6370
6371 uint64_t old_disarm_protected_io = (uint64_t) disarm_protected_io;
6372 error = SYSCTL_OUT(req, &old_disarm_protected_io, sizeof(old_disarm_protected_io));
6373
6374 if (error) {
6375 return error;
6376 }
6377
6378 uint64_t new_disarm_protected_io = old_disarm_protected_io;
6379 if (req->newptr) {
6380 error = SYSCTL_IN(req, &new_disarm_protected_io, sizeof(new_disarm_protected_io));
6381 if (!disarm_protected_io && new_disarm_protected_io) {
6382 sptm_sysctl(SPTM_SYSCTL_DISARM_PROTECTED_IO, SPTM_SYSCTL_SET, 1);
6383 os_atomic_thread_fence(release);
6384 disarm_protected_io = true;
6385 }
6386 }
6387
6388 return error;
6389 }
6390 SYSCTL_PROC(_kern, OID_AUTO, sptm_disarm_protected_io, CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED, 0, 0, sysctl_sptm_disarm_protected_io, "Q", "");
6391
6392 /**
6393 * Usage of kern.sptm_sysctl_poke
6394 *
6395 * This sysctl provides a convenient way to trigger the "getter" handler of a
6396 * specified SPTM sysctl. With this sysctl, you can trigger arbitrary SPTM
6397 * code without modifying xnu source code. All you need to do is define a
6398 * new SPTM sysctl and implement its "getter". After that, you can write
6399 * the SPTM sysctl number to this sysctl to trigger it.
6400 */
6401 static int sysctl_sptm_sysctl_poke SYSCTL_HANDLER_ARGS
6402 {
6403 int error = 0;
6404
6405 /* Always read-as-zero. */
6406 const uint64_t out = 0;
6407 error = SYSCTL_OUT(req, &out, sizeof(out));
6408
6409 if (error) {
6410 return error;
6411 }
6412
6413 uint64_t selector;
6414 if (req->newptr) {
6415 error = SYSCTL_IN(req, &selector, sizeof(selector));
6416 sptm_sysctl(selector, SPTM_SYSCTL_GET, 0);
6417 }
6418
6419 return error;
6420 }
6421 SYSCTL_PROC(_kern, OID_AUTO, sptm_sysctl_poke, CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED, 0, 0, sysctl_sptm_sysctl_poke, "Q", "");
6422 #endif /* CONFIG_SPTM && HAS_SPTM_SYSCTL */
6423
6424 #if CONFIG_SPTM && (DEVELOPMENT || DEBUG)
6425 /**
6426 * Sysctls to get SPTM allowed I/O ranges, pmap I/O ranges and I/O ranges by index.
6427 * Used by SEAR/LASER tools.
6428 */
6429 static int
6430 sysctl_sptm_allowed_io_ranges SYSCTL_HANDLER_ARGS
6431 {
6432 #pragma unused(oidp, arg1, arg2)
6433 sptm_io_range_t io_range = { 0 };
6434 unsigned int index = 0;
6435
6436 int error = SYSCTL_IN(req, &index, sizeof(index));
6437 if (error) {
6438 return error;
6439 }
6440
6441 libsptm_error_t ret = sptm_get_info(INFO_SPTM_ALLOWED_IO_RANGES, index, &io_range);
6442 if (__improbable(ret != LIBSPTM_SUCCESS)) {
6443 return EINVAL;
6444 }
6445
6446 return SYSCTL_OUT(req, &io_range, sizeof(io_range));
6447 }
6448 SYSCTL_PROC(_kern, OID_AUTO, sptm_allowed_io_ranges, CTLTYPE_STRUCT | CTLFLAG_RW | CTLFLAG_LOCKED,
6449 0, 0, sysctl_sptm_allowed_io_ranges, "S,sptm_io_range_t", "SPTM allowed I/O ranges by index");
6450
6451 static int
6452 sysctl_sptm_allowed_io_ranges_count SYSCTL_HANDLER_ARGS
6453 {
6454 #pragma unused(oidp, arg1, arg2)
6455 unsigned int count = 0;
6456
6457 libsptm_error_t ret = sptm_get_info(INFO_SPTM_ALLOWED_IO_RANGES_COUNT, 0, &count);
6458 if (__improbable(ret != LIBSPTM_SUCCESS)) {
6459 return EINVAL;
6460 }
6461
6462 return SYSCTL_OUT(req, &count, sizeof(count));
6463 }
6464 SYSCTL_PROC(_kern, OID_AUTO, sptm_allowed_io_ranges_count, CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
6465 0, 0, sysctl_sptm_allowed_io_ranges_count, "I", "SPTM allowed I/O ranges count");
6466
6467 static int
6468 sysctl_sptm_pmap_io_ranges SYSCTL_HANDLER_ARGS
6469 {
6470 #pragma unused(oidp, arg1, arg2)
6471 sptm_io_range_t io_range = { 0 };
6472 unsigned int index = 0;
6473
6474 int error = SYSCTL_IN(req, &index, sizeof(index));
6475 if (error) {
6476 return error;
6477 }
6478
6479 libsptm_error_t ret = sptm_get_info(INFO_SPTM_PMAP_IO_RANGES, index, &io_range);
6480 if (__improbable(ret != LIBSPTM_SUCCESS)) {
6481 return EINVAL;
6482 }
6483
6484 return SYSCTL_OUT(req, &io_range, sizeof(io_range));
6485 }
6486 SYSCTL_PROC(_kern, OID_AUTO, sptm_pmap_io_ranges, CTLTYPE_STRUCT | CTLFLAG_RW | CTLFLAG_LOCKED,
6487 0, 0, sysctl_sptm_pmap_io_ranges, "S,sptm_io_range_t", "SPTM pmap I/O ranges by index");
6488
6489 static int
6490 sysctl_sptm_pmap_io_ranges_count SYSCTL_HANDLER_ARGS
6491 {
6492 #pragma unused(oidp, arg1, arg2)
6493 unsigned int count = 0;
6494
6495 libsptm_error_t ret = sptm_get_info(INFO_SPTM_PMAP_IO_RANGES_COUNT, 0, &count);
6496 if (__improbable(ret != LIBSPTM_SUCCESS)) {
6497 return EINVAL;
6498 }
6499
6500 return SYSCTL_OUT(req, &count, sizeof(count));
6501 }
6502 SYSCTL_PROC(_kern, OID_AUTO, sptm_pmap_io_ranges_count, CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
6503 0, 0, sysctl_sptm_pmap_io_ranges_count, "I", "SPTM pmap I/O ranges count");
6504
6505 static int
6506 sysctl_sptm_io_ranges SYSCTL_HANDLER_ARGS
6507 {
6508 #pragma unused(oidp, arg1, arg2)
6509 sptm_io_range_t io_range = { 0 };
6510 unsigned int index = 0;
6511
6512 int error = SYSCTL_IN(req, &index, sizeof(index));
6513 if (error) {
6514 return error;
6515 }
6516
6517 libsptm_error_t ret = sptm_get_info(INFO_SPTM_IO_RANGES, index, &io_range);
6518 if (__improbable(ret != LIBSPTM_SUCCESS)) {
6519 return EINVAL;
6520 }
6521
6522 return SYSCTL_OUT(req, &io_range, sizeof(io_range));
6523 }
6524 SYSCTL_PROC(_kern, OID_AUTO, sptm_io_ranges, CTLTYPE_STRUCT | CTLFLAG_RW | CTLFLAG_LOCKED,
6525 0, 0, sysctl_sptm_io_ranges, "S,sptm_io_range_t", "SPTM I/O ranges by index");
6526
6527 static int
6528 sysctl_sptm_io_ranges_count SYSCTL_HANDLER_ARGS
6529 {
6530 #pragma unused(oidp, arg1, arg2)
6531 unsigned int count = 0;
6532
6533 libsptm_error_t ret = sptm_get_info(INFO_SPTM_IO_RANGES_COUNT, 0, &count);
6534 if (__improbable(ret != LIBSPTM_SUCCESS)) {
6535 return EINVAL;
6536 }
6537
6538 return SYSCTL_OUT(req, &count, sizeof(count));
6539 }
6540 SYSCTL_PROC(_kern, OID_AUTO, sptm_io_ranges_count, CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
6541 0, 0, sysctl_sptm_io_ranges_count, "I", "SPTM I/O ranges count");
6542 #endif /* CONFIG_SPTM && (DEVELOPMENT || DEBUG) */
6543
6544 #if __ARM64_PMAP_SUBPAGE_L1__ && CONFIG_SPTM
6545 extern bool surt_ready;
6546 static int
6547 sysctl_surt_ready SYSCTL_HANDLER_ARGS
6548 {
6549 #pragma unused(oidp, arg1, arg2)
6550 unsigned int surt_ready_uint = (unsigned int)surt_ready;
6551 return SYSCTL_OUT(req, &surt_ready_uint, sizeof(surt_ready_uint));
6552 }
6553 SYSCTL_PROC(_kern, OID_AUTO, surt_ready, CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
6554 0, 0, sysctl_surt_ready, "I", "SURT system readiness");
6555 #endif /* __ARM64_PMAP_SUBPAGE_L1__ && CONFIG_SPTM */
6556
6557 #if __arm64__ && (DEBUG || DEVELOPMENT)
6558 extern unsigned int pmap_wcrt_on_non_dram_count_get(void);
6559 static int
6560 sysctl_pmap_wcrt_on_non_dram_count SYSCTL_HANDLER_ARGS
6561 {
6562 #pragma unused(oidp, arg1, arg2)
6563 unsigned int count = pmap_wcrt_on_non_dram_count_get();
6564
6565 return SYSCTL_OUT(req, &count, sizeof(count));
6566 }
6567 SYSCTL_PROC(_kern, OID_AUTO, pmap_wcrt_on_non_dram_count, CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
6568 0, 0, sysctl_pmap_wcrt_on_non_dram_count, "I", "pmap WC/RT mapping request on non-DRAM count");
6569 #endif /* __arm64__ && (DEBUG || DEVELOPMENT) */
6570
6571 TUNABLE_DT(int, gpu_pmem_selector, "defaults", "kern.gpu_pmem_selector", "gpu-pmem-selector", 0, TUNABLE_DT_NONE);
6572
6573 #if CONFIG_EXCLAVES
6574
6575 static int
6576 sysctl_task_conclave SYSCTL_HANDLER_ARGS
6577 {
6578 extern const char *exclaves_resource_name(void *);
6579
6580 #pragma unused(arg2)
6581 void *conclave = task_get_conclave(current_task());
6582 if (conclave != NULL) {
6583 const char *name = exclaves_resource_name(conclave);
6584 assert3u(strlen(name), >, 0);
6585
6586 /*
6587 * This is a RO operation already and the string is never
6588 * written to.
6589 */
6590 #pragma clang diagnostic push
6591 #pragma clang diagnostic ignored "-Wcast-qual"
6592 return sysctl_handle_string(oidp, (char *)name, 0, req);
6593 #pragma clang diagnostic pop
6594 }
6595 return sysctl_handle_string(oidp, arg1, MAXCONCLAVENAME, req);
6596 }
6597
6598 SYSCTL_PROC(_kern, OID_AUTO, task_conclave,
6599 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
6600 "", 0, sysctl_task_conclave, "A", "Conclave string for the task");
6601
6602
6603 void task_set_conclave_untaintable(task_t task);
6604
6605 static int
6606 sysctl_task_conclave_untaintable SYSCTL_HANDLER_ARGS
6607 {
6608 #pragma unused(arg1, arg2)
6609 int error, val = 0;
6610 error = sysctl_handle_int(oidp, &val, 0, req);
6611 if (error || val == 0) {
6612 return error;
6613 }
6614
6615 task_set_conclave_untaintable(current_task());
6616 return 0;
6617 }
6618
6619 SYSCTL_PROC(_kern, OID_AUTO, task_conclave_untaintable,
6620 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
6621 "", 0, sysctl_task_conclave_untaintable, "A", "Task could not be tainted by talking to conclaves");
6622
6623 extern exclaves_requirement_t exclaves_relaxed_requirements;
6624 SYSCTL_QUAD(_kern, OID_AUTO, exclaves_relaxed_requirements,
6625 CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
6626 &exclaves_relaxed_requirements, "Exclaves requirements which have been relaxed");
6627
6628 #endif /* CONFIG_EXCLAVES */
6629
6630 #if (DEVELOPMENT || DEBUG)
6631 SYSCTL_INT(_kern, OID_AUTO, gpu_pmem_selector,
6632 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
6633 &gpu_pmem_selector, 0, "GPU wire down limit selector");
6634 #else /* !(DEVELOPMENT || DEBUG) */
6635 SYSCTL_INT(_kern, OID_AUTO, gpu_pmem_selector,
6636 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN | CTLFLAG_MASKED,
6637 &gpu_pmem_selector, 0, "GPU wire down limit selector");
6638 #endif /* (DEVELOPMENT || DEBUG) */
6639
6640 static int
6641 sysctl_exclaves_status SYSCTL_HANDLER_ARGS
6642 {
6643 int value = exclaves_get_status();
6644 return sysctl_io_number(req, value, sizeof(value), NULL, NULL);
6645 }
6646
6647 SYSCTL_PROC(_kern, OID_AUTO, exclaves_status,
6648 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
6649 0, 0, sysctl_exclaves_status, "I", "Running status of Exclaves");
6650
6651
6652 static int
6653 sysctl_exclaves_boot_stage SYSCTL_HANDLER_ARGS
6654 {
6655 int value = exclaves_get_boot_stage();
6656 return sysctl_io_number(req, value, sizeof(value), NULL, NULL);
6657 }
6658
6659 SYSCTL_PROC(_kern, OID_AUTO, exclaves_boot_stage,
6660 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
6661 0, 0, sysctl_exclaves_boot_stage, "I", "Boot stage of Exclaves");
6662
6663 #if CONFIG_EXCLAVES && (DEVELOPMENT || DEBUG)
6664 extern unsigned int exclaves_debug;
6665 SYSCTL_UINT(_kern, OID_AUTO, exclaves_debug, CTLFLAG_RW | CTLFLAG_LOCKED,
6666 &exclaves_debug, 0, "Exclaves debug flags");
6667
6668 static int
6669 sysctl_exclaves_inspection_status SYSCTL_HANDLER_ARGS
6670 {
6671 #pragma unused(oidp, arg1, arg2)
6672 int value = (int)exclaves_inspection_is_initialized();
6673 return sysctl_io_number(req, value, sizeof(value), NULL, NULL);
6674 }
6675 SYSCTL_PROC(_kern, OID_AUTO, exclaves_inspection_status,
6676 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
6677 0, 0, sysctl_exclaves_inspection_status, "I", "Exclaves debug inspection status");
6678 #endif /* CONFIG_EXCLAVES && (DEVELOPMENT || DEBUG) */
6679
6680 #if (DEBUG || DEVELOPMENT)
6681 extern uint32_t disable_vm_sanitize_telemetry;
6682 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");
6683 #endif
6684
6685 #define kReadUserspaceRebootInfoEntitlement "com.apple.private.kernel.userspacereboot-info-read-only"
6686 static int
6687 _sysctl_userspacereboot_info(struct sysctl_req *req, void *ptr, size_t ptr_size)
6688 {
6689 if (req->newptr != 0) {
6690 /* initproc is the only process that can write to these sysctls */
6691 if (proc_getpid(req->p) != 1) {
6692 return EPERM;
6693 }
6694 return SYSCTL_IN(req, ptr, ptr_size);
6695 } else {
6696 /* A read entitlement is required to read these sysctls */
6697 if (!IOCurrentTaskHasEntitlement(kReadUserspaceRebootInfoEntitlement)) {
6698 return EPERM;
6699 }
6700 return SYSCTL_OUT(req, ptr, ptr_size);
6701 }
6702 }
6703
6704 static int
6705 sysctl_userspacereboottime(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
6706 {
6707 return _sysctl_userspacereboot_info(req, &userspacereboottime, sizeof(userspacereboottime));
6708 }
6709
6710 static int
6711 sysctl_userspacerebootpurpose(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
6712 {
6713 return _sysctl_userspacereboot_info(req, &userspacerebootpurpose, sizeof(userspacerebootpurpose));
6714 }
6715
6716 SYSCTL_PROC(_kern, OID_AUTO, userspacereboottime, CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED, 0, 0, sysctl_userspacereboottime, "Q", "");
6717 SYSCTL_PROC(_kern, OID_AUTO, userspacerebootpurpose, CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED, 0, 0, sysctl_userspacerebootpurpose, "I", "");
6718
6719 #if XNU_TARGET_OS_IOS
6720
6721 static LCK_GRP_DECLARE(erm_config_lock_grp, "ERM sysctl");
6722 static LCK_RW_DECLARE(erm_config_lock, &erm_config_lock_grp);
6723 #define ERM_CONFIG_SYSCTL_WRITE_ENTITLEMENT "com.apple.private.security-research-device.extended-research-mode"
6724 #define ERM_CONFIG_SYSCTL_MAX_SIZE PAGE_SIZE
6725
6726 // This sysctl handler is only registered when Extended Research Mode (ERM) is active.
6727 static int
6728 sysctl_user_extended_research_mode_config_handler(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
6729 {
6730 // Pointer for the dynamically allocated buffer
6731 static void *extended_research_mode_config_data = NULL;
6732
6733 // Current size of the valid data stored in the buffer
6734 static size_t extended_research_mode_config_current_size = 0;
6735
6736 // Handle Read request (user wants to read the current config, before it is overwritten)
6737 if (req->oldptr != USER_ADDR_NULL) {
6738 int error = 0;
6739
6740 lck_rw_lock_shared(&erm_config_lock);
6741
6742 if (req->oldlen < extended_research_mode_config_current_size) {
6743 error = ENOMEM;
6744 } else {
6745 if (extended_research_mode_config_current_size > 0) {
6746 error = copyout(extended_research_mode_config_data,
6747 req->oldptr,
6748 extended_research_mode_config_current_size);
6749 }
6750 }
6751 // In all cases, report the total size of the currently stored config back to the user,
6752 req->oldlen = extended_research_mode_config_current_size;
6753 req->oldidx = req->oldlen;
6754
6755 lck_rw_unlock_shared(&erm_config_lock);
6756
6757 if (error != 0) {
6758 return error;
6759 }
6760 } else {
6761 // User just want to know the current buffer size.
6762 // All accesses to extended_research_mode_config* variables are expected
6763 // to be done under erm_config_lock.
6764 lck_rw_lock_shared(&erm_config_lock);
6765 req->oldidx = extended_research_mode_config_current_size;
6766 lck_rw_unlock_shared(&erm_config_lock);
6767 }
6768
6769
6770 // Handle Write request (new data provided by user)
6771 if (req->newptr != USER_ADDR_NULL) {
6772 if (!IOTaskHasEntitlement(proc_task(req->p), ERM_CONFIG_SYSCTL_WRITE_ENTITLEMENT)) {
6773 return EPERM;
6774 }
6775
6776 size_t requested_len = req->newlen;
6777
6778 if (requested_len > ERM_CONFIG_SYSCTL_MAX_SIZE) {
6779 // We ensure the config provided by user-space is not too big
6780 return EINVAL;
6781 }
6782
6783 // Allocate a new buffer for the incoming data
6784 void *new_buffer = (void *)kalloc_data(requested_len, Z_WAITOK | Z_ZERO);
6785
6786 if (new_buffer == NULL) {
6787 return ENOMEM; // Allocation failed
6788 }
6789
6790 // Copy data from user space into the newly allocated buffer
6791 int error = copyin(req->newptr, new_buffer, requested_len);
6792
6793 if (error == 0) {
6794 // Success: Replace the old buffer with the new one
6795 lck_rw_lock_exclusive(&erm_config_lock);
6796
6797 // Backup old buffer info for freeing it in a second step
6798 void *old_buffer_to_free = extended_research_mode_config_data;
6799 size_t old_buffer_size = extended_research_mode_config_current_size;
6800
6801 // Point to the new buffer and update size
6802 extended_research_mode_config_data = new_buffer;
6803 extended_research_mode_config_current_size = requested_len;
6804 lck_rw_unlock_exclusive(&erm_config_lock);
6805 new_buffer = NULL; // transferred to the static pointer
6806
6807 // Previous buffer is not referenced anymore, good to be deleted.
6808 kfree_data(old_buffer_to_free, old_buffer_size);
6809 } else {
6810 // Copyin failed, free the buffer we just allocated and keep the old data and size intact
6811 kfree_data(new_buffer, requested_len);
6812 return error;
6813 }
6814 }
6815
6816 return 0;
6817 }
6818
6819 // We don't register this sysctl handler automatically , but rather only register it only if the extended
6820 // research mode is active.
6821 SYSCTL_PROC(_user, // Parent node structure (_kern)
6822 OID_AUTO, // Automatically assign OID
6823 extended_research_mode_config, // Name of the node
6824 CTLFLAG_NOAUTO | // We will register this sysctl on our own
6825 CTLTYPE_OPAQUE | // Type: Opaque binary data
6826 CTLFLAG_WR | // Allow both read and write
6827 CTLFLAG_ANYBODY | // No user filtering
6828 CTLFLAG_LOCKED, // The handler manages its own locking.
6829 NULL, // arg1 (not used)
6830 0, // arg2 (not used)
6831 &sysctl_user_extended_research_mode_config_handler,
6832 "-", // don't print the content (as it is a blob)
6833 "Configuration blob for Extended Research Mode");
6834
6835 // This function is defined in kern_codesigning.c but don't worth include the whole .h just for it.
6836 bool extended_research_mode_state(void);
6837
6838 // Only register the research_mode_config sysctl if Extended Research Mode is active
6839 __startup_func
6840 static void
6841 extended_research_mode_config_sysctl_startup(void)
6842 {
6843 if (__improbable(extended_research_mode_state())) {
6844 // Register the sysctl handler
6845 sysctl_register_oid_early(&sysctl__user_extended_research_mode_config);
6846 }
6847 }
6848 STARTUP(SYSCTL, STARTUP_RANK_MIDDLE, extended_research_mode_config_sysctl_startup);
6849 #endif /* XNU_TARGET_OS_IOS */
6850
6851 #if DEBUG || DEVELOPMENT
6852 SCALABLE_COUNTER_DEFINE(mach_eventlink_handoff_success_count);
6853 SYSCTL_SCALABLE_COUNTER(_kern, mach_eventlink_handoff_success_count,
6854 mach_eventlink_handoff_success_count, "Number of successful handoffs");
6855 #endif /* DEBUG || DEVELOPMENT*/
6856