1 /*
2 * Copyright (c) 2000-2024 Apple Inc. All rights reserved.
3 *
4 * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
5 *
6 * This file contains Original Code and/or Modifications of Original Code
7 * as defined in and that are subject to the Apple Public Source License
8 * Version 2.0 (the 'License'). You may not use this file except in
9 * compliance with the License. The rights granted to you under the License
10 * may not be used to create, or enable the creation or redistribution of,
11 * unlawful or unlicensed copies of an Apple operating system, or to
12 * circumvent, violate, or enable the circumvention or violation of, any
13 * terms of an Apple operating system software license agreement.
14 *
15 * Please obtain a copy of the License at
16 * http://www.opensource.apple.com/apsl/ and read it before using this file.
17 *
18 * The Original Code and all software distributed under the License are
19 * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
20 * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
21 * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
22 * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
23 * Please see the License for the specific language governing rights and
24 * limitations under the License.
25 *
26 * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
27 */
28 /* Copyright (c) 1995 NeXT Computer, Inc. All Rights Reserved */
29 /*-
30 * Copyright (c) 1982, 1986, 1989, 1993
31 * The Regents of the University of California. All rights reserved.
32 *
33 * This code is derived from software contributed to Berkeley by
34 * Mike Karels at Berkeley Software Design, Inc.
35 *
36 * Redistribution and use in source and binary forms, with or without
37 * modification, are permitted provided that the following conditions
38 * are met:
39 * 1. Redistributions of source code must retain the above copyright
40 * notice, this list of conditions and the following disclaimer.
41 * 2. Redistributions in binary form must reproduce the above copyright
42 * notice, this list of conditions and the following disclaimer in the
43 * documentation and/or other materials provided with the distribution.
44 * 3. All advertising materials mentioning features or use of this software
45 * must display the following acknowledgement:
46 * This product includes software developed by the University of
47 * California, Berkeley and its contributors.
48 * 4. Neither the name of the University nor the names of its contributors
49 * may be used to endorse or promote products derived from this software
50 * without specific prior written permission.
51 *
52 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
53 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
54 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
55 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
56 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
57 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
58 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
59 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
60 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
61 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
62 * SUCH DAMAGE.
63 *
64 * @(#)kern_sysctl.c 8.4 (Berkeley) 4/14/94
65 */
66 /*
67 * NOTICE: This file was modified by SPARTA, Inc. in 2005 to introduce
68 * support for mandatory and extensible security protections. This notice
69 * is included in support of clause 2.2 (b) of the Apple Public License,
70 * Version 2.0.
71 */
72
73 /*
74 * DEPRECATED sysctl system call code
75 *
76 * Everything in this file is deprecated. Sysctls should be handled
77 * by the code in kern_newsysctl.c.
78 * The remaining "case" sections are supposed to be converted into
79 * SYSCTL_*-style definitions, and as soon as all of them are gone,
80 * this source file is supposed to die.
81 *
82 * DO NOT ADD ANY MORE "case" SECTIONS TO THIS FILE, instead define
83 * your sysctl with SYSCTL_INT, SYSCTL_PROC etc. in your source file.
84 */
85
86 #include <sys/param.h>
87 #include <sys/systm.h>
88 #include <sys/kernel.h>
89 #include <sys/malloc.h>
90 #include <sys/proc_internal.h>
91 #include <sys/kauth.h>
92 #include <sys/file_internal.h>
93 #include <sys/vnode_internal.h>
94 #include <sys/unistd.h>
95 #include <sys/buf.h>
96 #include <sys/ioctl.h>
97 #include <sys/namei.h>
98 #include <sys/tty.h>
99 #include <sys/disklabel.h>
100 #include <sys/vm.h>
101 #include <sys/sysctl.h>
102 #include <sys/user.h>
103 #include <sys/aio_kern.h>
104 #include <sys/reboot.h>
105 #include <sys/memory_maintenance.h>
106 #include <sys/priv.h>
107 #include <sys/ubc.h> /* mach_to_bsd_errno */
108
109 #include <stdatomic.h>
110 #include <uuid/uuid.h>
111
112 #include <security/audit/audit.h>
113 #include <kern/kalloc.h>
114
115 #include <machine/smp.h>
116 #include <machine/atomic.h>
117 #include <machine/config.h>
118 #include <mach/machine.h>
119 #include <mach/mach_host.h>
120 #include <mach/mach_types.h>
121 #include <mach/processor_info.h>
122 #include <mach/vm_param.h>
123 #include <kern/debug.h>
124 #include <kern/mach_param.h>
125 #include <kern/task.h>
126 #include <kern/thread.h>
127 #include <kern/thread_group.h>
128 #include <kern/processor.h>
129 #include <kern/cpu_number.h>
130 #include <kern/sched_prim.h>
131 #include <kern/workload_config.h>
132 #include <kern/iotrace.h>
133 #include <vm/vm_kern_xnu.h>
134 #include <vm/vm_map_xnu.h>
135 #include <mach/host_info.h>
136 #include <mach/exclaves.h>
137 #include <kern/hvg_hypercall.h>
138 #include <kdp/sk_core.h>
139
140 #if DEVELOPMENT || DEBUG
141 #include <kern/ext_paniclog.h>
142 #endif
143
144 #include <sys/mount_internal.h>
145 #include <sys/kdebug.h>
146 #include <sys/kern_debug.h>
147 #include <sys/kern_sysctl.h>
148 #include <sys/variant_internal.h>
149
150 #include <IOKit/IOPlatformExpert.h>
151 #include <pexpert/pexpert.h>
152
153 #include <machine/machine_routines.h>
154 #include <machine/exec.h>
155
156 #include <nfs/nfs_conf.h>
157
158 #include <vm/vm_protos.h>
159 #include <vm/vm_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 #if (DEVELOPMENT || DEBUG)
224 extern uint32_t vm_page_creation_throttled_hard;
225 extern uint32_t vm_page_creation_throttled_soft;
226 #endif /* DEVELOPMENT || DEBUG */
227
228 #if DEVELOPMENT || DEBUG
229 extern bool bootarg_hide_process_traced;
230 #endif
231
232 /*
233 * Conditionally allow dtrace to see these functions for debugging purposes.
234 */
235 #ifdef STATIC
236 #undef STATIC
237 #endif
238 #if 0
239 #define STATIC
240 #else
241 #define STATIC static
242 #endif
243
244 extern boolean_t mach_timer_coalescing_enabled;
245
246 extern uint64_t timer_deadline_tracking_bin_1, timer_deadline_tracking_bin_2;
247
248 STATIC void
249 fill_user32_eproc(proc_t, struct user32_eproc *__restrict);
250 STATIC void
251 fill_user32_externproc(proc_t, struct user32_extern_proc *__restrict);
252 STATIC void
253 fill_user64_eproc(proc_t, struct user64_eproc *__restrict);
254 STATIC void
255 fill_user64_proc(proc_t, struct user64_kinfo_proc *__restrict);
256 STATIC void
257 fill_user64_externproc(proc_t, struct user64_extern_proc *__restrict);
258 STATIC void
259 fill_user32_proc(proc_t, struct user32_kinfo_proc *__restrict);
260
261 #if CONFIG_NETBOOT
262 extern int
263 netboot_root(void);
264 #endif
265 int
266 sysctl_procargs(int *name, u_int namelen, user_addr_t where,
267 size_t *sizep, proc_t cur_proc);
268 STATIC int
269 sysctl_procargsx(int *name, u_int namelen, user_addr_t where, size_t *sizep,
270 proc_t cur_proc, int argc_yes);
271 int
272 sysctl_struct(user_addr_t oldp, size_t *oldlenp, user_addr_t newp,
273 size_t newlen, void *sp, int len);
274
275 STATIC int sysdoproc_filt_KERN_PROC_PID(proc_t p, void * arg);
276 STATIC int sysdoproc_filt_KERN_PROC_PGRP(proc_t p, void * arg);
277 STATIC int sysdoproc_filt_KERN_PROC_TTY(proc_t p, void * arg);
278 STATIC int sysdoproc_filt_KERN_PROC_UID(proc_t p, void * arg);
279 STATIC int sysdoproc_filt_KERN_PROC_RUID(proc_t p, void * arg);
280 int sysdoproc_callback(proc_t p, void *arg);
281
282 #if CONFIG_THREAD_GROUPS && (DEVELOPMENT || DEBUG)
283 STATIC int sysctl_get_thread_group_id SYSCTL_HANDLER_ARGS;
284 #endif
285
286 /* forward declarations for non-static STATIC */
287 STATIC void fill_loadavg64(struct loadavg *la, struct user64_loadavg *la64);
288 STATIC void fill_loadavg32(struct loadavg *la, struct user32_loadavg *la32);
289 STATIC int sysctl_handle_kern_threadname(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
290 STATIC int sysctl_sched_stats(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
291 STATIC int sysctl_sched_stats_enable(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
292 #if COUNT_SYSCALLS
293 STATIC int sysctl_docountsyscalls SYSCTL_HANDLER_ARGS;
294 #endif /* COUNT_SYSCALLS */
295 #if defined(XNU_TARGET_OS_OSX)
296 STATIC int sysctl_doprocargs SYSCTL_HANDLER_ARGS;
297 #endif /* defined(XNU_TARGET_OS_OSX) */
298 STATIC int sysctl_doprocargs2 SYSCTL_HANDLER_ARGS;
299 STATIC int sysctl_prochandle SYSCTL_HANDLER_ARGS;
300 STATIC int sysctl_aiomax(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
301 STATIC int sysctl_aioprocmax(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
302 STATIC int sysctl_aiothreads(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
303 STATIC int sysctl_maxproc(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
304 STATIC int sysctl_osversion(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
305 STATIC int sysctl_sysctl_bootargs(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
306 STATIC int sysctl_maxvnodes(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
307 STATIC int sysctl_securelvl(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
308 STATIC int sysctl_domainname(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
309 STATIC int sysctl_hostname(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
310 STATIC int sysctl_procname(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
311 STATIC int sysctl_boottime(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
312 STATIC int sysctl_bootuuid(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
313 STATIC int sysctl_symfile(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
314 #if CONFIG_NETBOOT
315 STATIC int sysctl_netboot(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
316 #endif
317 #ifdef CONFIG_IMGSRC_ACCESS
318 STATIC int sysctl_imgsrcdev(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
319 #endif
320 STATIC int sysctl_usrstack(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
321 STATIC int sysctl_usrstack64(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
322 #if CONFIG_COREDUMP
323 STATIC int sysctl_coredump(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
324 STATIC int sysctl_suid_coredump(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
325 #endif
326 STATIC int sysctl_delayterm(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
327 STATIC int sysctl_rage_vnode(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
328 STATIC int sysctl_kern_check_openevt(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
329 #if DEVELOPMENT || DEBUG
330 STATIC int sysctl_nx(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
331 #endif
332 STATIC int sysctl_loadavg(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
333 STATIC int sysctl_vm_toggle_address_reuse(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
334 STATIC int sysctl_swapusage(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
335 STATIC int fetch_process_cputype( proc_t cur_proc, int *name, u_int namelen, cpu_type_t *cputype);
336 STATIC int sysctl_sysctl_native(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
337 STATIC int sysctl_sysctl_cputype(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
338 STATIC int sysctl_safeboot(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
339 STATIC int sysctl_singleuser(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
340 STATIC int sysctl_minimalboot(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
341 STATIC int sysctl_slide(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
342
343 #ifdef CONFIG_XNUPOST
344 #include <tests/xnupost.h>
345
346 STATIC int sysctl_debug_test_oslog_ctl(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
347 STATIC int sysctl_debug_test_stackshot_mutex_owner(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
348 STATIC int sysctl_debug_test_stackshot_rwlck_owner(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req);
349 #endif
350
351 extern void IORegistrySetOSBuildVersion(char * build_version);
352 extern int IOParseWorkloadConfig(workload_config_ctx_t *ctx, const char * buffer, size_t size);
353 extern int IOUnparseWorkloadConfig(char *buffer, size_t *size);
354
355 STATIC void
fill_loadavg64(struct loadavg * la,struct user64_loadavg * la64)356 fill_loadavg64(struct loadavg *la, struct user64_loadavg *la64)
357 {
358 la64->ldavg[0] = la->ldavg[0];
359 la64->ldavg[1] = la->ldavg[1];
360 la64->ldavg[2] = la->ldavg[2];
361 la64->fscale = (user64_long_t)la->fscale;
362 }
363
364 STATIC void
fill_loadavg32(struct loadavg * la,struct user32_loadavg * la32)365 fill_loadavg32(struct loadavg *la, struct user32_loadavg *la32)
366 {
367 la32->ldavg[0] = la->ldavg[0];
368 la32->ldavg[1] = la->ldavg[1];
369 la32->ldavg[2] = la->ldavg[2];
370 la32->fscale = (user32_long_t)la->fscale;
371 }
372
373 #if COUNT_SYSCALLS
374 extern int do_count_syscalls;
375 #endif
376
377 #ifdef INSECURE
378 int securelevel = -1;
379 #else
380 int securelevel;
381 #endif
382
383 STATIC int
sysctl_handle_kern_threadname(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)384 sysctl_handle_kern_threadname( __unused struct sysctl_oid *oidp, __unused void *arg1,
385 __unused int arg2, struct sysctl_req *req)
386 {
387 int error;
388 struct uthread *ut = current_uthread();
389 user_addr_t oldp = 0, newp = 0;
390 size_t *oldlenp = NULL;
391 size_t newlen = 0;
392
393 oldp = req->oldptr;
394 oldlenp = &(req->oldlen);
395 newp = req->newptr;
396 newlen = req->newlen;
397
398 /* We want the current length, and maybe the string itself */
399 if (oldlenp) {
400 /* if we have no thread name yet tell'em we want MAXTHREADNAMESIZE - 1 */
401 size_t currlen = MAXTHREADNAMESIZE - 1;
402
403 if (ut->pth_name) {
404 /* use length of current thread name */
405 currlen = strlen(ut->pth_name);
406 }
407 if (oldp) {
408 if (*oldlenp < currlen) {
409 return ENOMEM;
410 }
411 /* NOTE - we do not copy the NULL terminator */
412 if (ut->pth_name) {
413 error = copyout(ut->pth_name, oldp, currlen);
414 if (error) {
415 return error;
416 }
417 }
418 }
419 /* return length of thread name minus NULL terminator (just like strlen) */
420 req->oldidx = currlen;
421 }
422
423 /* We want to set the name to something */
424 if (newp) {
425 if (newlen > (MAXTHREADNAMESIZE - 1)) {
426 return ENAMETOOLONG;
427 }
428 if (!ut->pth_name) {
429 char *tmp_pth_name = (char *)kalloc_data(MAXTHREADNAMESIZE,
430 Z_WAITOK | Z_ZERO);
431 if (!tmp_pth_name) {
432 return ENOMEM;
433 }
434 if (!OSCompareAndSwapPtr(NULL, tmp_pth_name, &ut->pth_name)) {
435 kfree_data(tmp_pth_name, MAXTHREADNAMESIZE);
436 return EBUSY;
437 }
438 } else {
439 kernel_debug_string_simple(TRACE_STRING_THREADNAME_PREV, ut->pth_name);
440 bzero(ut->pth_name, MAXTHREADNAMESIZE);
441 }
442 error = copyin(newp, ut->pth_name, newlen);
443 if (error) {
444 return error;
445 }
446
447 kernel_debug_string_simple(TRACE_STRING_THREADNAME, ut->pth_name);
448 }
449
450 return 0;
451 }
452
453 SYSCTL_PROC(_kern, KERN_THREADNAME, threadname, CTLFLAG_ANYBODY | CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_LOCKED, 0, 0, sysctl_handle_kern_threadname, "A", "");
454
455 #define WORKLOAD_CONFIG_MAX_SIZE (128 * 1024 * 1024)
456
457 /* Called locked - sysctl defined without CTLFLAG_LOCKED. */
458 static int
459 sysctl_workload_config SYSCTL_HANDLER_ARGS
460 {
461 #pragma unused(arg1, arg2)
462
463 char *plist_blob = NULL;
464 kern_return_t ret = KERN_FAILURE;
465 int error = -1;
466
467 /* Only allow reading of workload config on non-RELEASE kernels. */
468 #if DEVELOPMENT || DEBUG
469
470 const size_t buf_size = req->oldlen;
471
472 if (!req->oldptr) {
473 /* Just looking for the size to allocate. */
474 size_t size = 0;
475 ret = IOUnparseWorkloadConfig(NULL, &size);
476 if (ret != KERN_SUCCESS) {
477 return ENOMEM;
478 }
479
480 error = SYSCTL_OUT(req, NULL, size);
481 if (error) {
482 return error;
483 }
484 } else {
485 if (buf_size > (WORKLOAD_CONFIG_MAX_SIZE - 1) ||
486 buf_size == 0) {
487 return EINVAL;
488 }
489
490 plist_blob = kalloc_data(buf_size, Z_WAITOK | Z_ZERO);
491 if (!plist_blob) {
492 return ENOMEM;
493 }
494
495 size_t size = buf_size;
496 ret = IOUnparseWorkloadConfig(plist_blob, &size);
497 if (ret != KERN_SUCCESS) {
498 kfree_data(plist_blob, buf_size);
499 return ENOMEM;
500 }
501
502 error = SYSCTL_OUT(req, plist_blob, MIN(buf_size, size));
503
504 /* If the buffer was too small to fit the entire config. */
505 if (buf_size < size) {
506 error = ENOMEM;
507 }
508
509 kfree_data(plist_blob, buf_size);
510 if (error) {
511 return error;
512 }
513 }
514 #endif /* DEVELOPMENT || DEBUG */
515
516 if (req->newptr) {
517 size_t newlen = req->newlen;
518 if (newlen > (WORKLOAD_CONFIG_MAX_SIZE - 1)) {
519 return EINVAL;
520 }
521
522
523 workload_config_ctx_t *ctx = NULL;
524 /*
525 * Only allow workload_config_boot to be loaded once at boot by launchd.
526 */
527 if (current_proc() == initproc &&
528 !workload_config_initialized(&workload_config_boot)) {
529 ctx = &workload_config_boot;
530 } else {
531 #if DEVELOPMENT || DEBUG
532 /*
533 * Use the devel config context otherwise. If a devel config has been
534 * initialized it will be used for lookups in place of the boot config.
535 */
536 ctx = &workload_config_devel;
537 if (workload_config_initialized(ctx)) {
538 workload_config_free(ctx);
539 }
540
541 /* The devel context can be explicitly cleared by an empty string. */
542 if (newlen == 1) {
543 return 0;
544 }
545 #else
546 return EINVAL;
547 #endif
548 }
549
550 plist_blob = kalloc_data(newlen + 1, Z_WAITOK | Z_ZERO);
551 if (!plist_blob) {
552 return ENOMEM;
553 }
554 error = copyin(req->newptr, plist_blob, newlen);
555 if (error) {
556 kfree_data(plist_blob, newlen + 1);
557 return error;
558 }
559 plist_blob[newlen] = '\0';
560 ret = IOParseWorkloadConfig(ctx, plist_blob, newlen + 1);
561
562 kfree_data(plist_blob, newlen + 1);
563 return ret == KERN_SUCCESS ? 0 : EINVAL;
564 }
565
566 return 0;
567 }
568
569 SYSCTL_PROC(_kern, OID_AUTO, workload_config, CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_MASKED,
570 0, 0, sysctl_workload_config, "A", "global workgroup configuration plist load/unload");
571
572 #define BSD_HOST 1
573 STATIC int
sysctl_sched_stats(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,struct sysctl_req * req)574 sysctl_sched_stats(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
575 {
576 host_basic_info_data_t hinfo;
577 kern_return_t kret;
578 uint32_t size;
579 uint32_t buf_size = 0;
580 int changed;
581 mach_msg_type_number_t count = HOST_BASIC_INFO_COUNT;
582 struct _processor_statistics_np *buf;
583 int error;
584
585 kret = host_info((host_t)BSD_HOST, HOST_BASIC_INFO, (host_info_t)&hinfo, &count);
586 if (kret != KERN_SUCCESS) {
587 return EINVAL;
588 }
589
590 size = sizeof(struct _processor_statistics_np) * (hinfo.logical_cpu_max + 2); /* One for RT Queue, One for Fair Share Queue */
591
592 if (req->oldlen < size) {
593 return EINVAL;
594 }
595
596 buf_size = size;
597 buf = (struct _processor_statistics_np *)kalloc_data(buf_size, Z_ZERO | Z_WAITOK);
598
599 kret = get_sched_statistics(buf, &size);
600 if (kret != KERN_SUCCESS) {
601 error = EINVAL;
602 goto out;
603 }
604
605 error = sysctl_io_opaque(req, buf, size, &changed);
606 if (error) {
607 goto out;
608 }
609
610 if (changed) {
611 panic("Sched info changed?!");
612 }
613 out:
614 kfree_data(buf, buf_size);
615 return error;
616 }
617
618 SYSCTL_PROC(_kern, OID_AUTO, sched_stats, CTLFLAG_LOCKED, 0, 0, sysctl_sched_stats, "-", "");
619
620 STATIC int
sysctl_sched_stats_enable(__unused struct sysctl_oid * oidp,__unused void * arg1,__unused int arg2,__unused struct sysctl_req * req)621 sysctl_sched_stats_enable(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, __unused struct sysctl_req *req)
622 {
623 boolean_t active;
624 int res;
625
626 if (req->newlen != sizeof(active)) {
627 return EINVAL;
628 }
629
630 res = copyin(req->newptr, &active, sizeof(active));
631 if (res != 0) {
632 return res;
633 }
634
635 return set_sched_stats_active(active);
636 }
637
638 SYSCTL_PROC(_kern, OID_AUTO, sched_stats_enable, CTLFLAG_LOCKED | CTLFLAG_WR, 0, 0, sysctl_sched_stats_enable, "-", "");
639
640 extern uint32_t sched_debug_flags;
641 SYSCTL_INT(_debug, OID_AUTO, sched, CTLFLAG_RW | CTLFLAG_LOCKED, &sched_debug_flags, 0, "scheduler debug");
642
643 #if (DEBUG || DEVELOPMENT)
644 extern boolean_t doprnt_hide_pointers;
645 SYSCTL_INT(_debug, OID_AUTO, hide_kernel_pointers, CTLFLAG_RW | CTLFLAG_LOCKED, &doprnt_hide_pointers, 0, "hide kernel pointers from log");
646 #endif
647
648
649 extern int get_kernel_symfile(proc_t, char **);
650
651 #if COUNT_SYSCALLS
652 #define KERN_COUNT_SYSCALLS (KERN_OSTYPE + 1000)
653
654 extern const unsigned int nsysent;
655 extern int syscalls_log[];
656 extern const char *syscallnames[];
657
658 STATIC int
659 sysctl_docountsyscalls SYSCTL_HANDLER_ARGS
660 {
661 __unused int cmd = oidp->oid_arg2; /* subcommand*/
662 __unused int *name = arg1; /* oid element argument vector */
663 __unused int namelen = arg2; /* number of oid element arguments */
664 int error, changed;
665
666 int tmp;
667
668 /* valid values passed in:
669 * = 0 means don't keep called counts for each bsd syscall
670 * > 0 means keep called counts for each bsd syscall
671 * = 2 means dump current counts to the system log
672 * = 3 means reset all counts
673 * for example, to dump current counts:
674 * sysctl -w kern.count_calls=2
675 */
676 error = sysctl_io_number(req, do_count_syscalls,
677 sizeof(do_count_syscalls), &tmp, &changed);
678
679 if (error != 0 || !changed) {
680 return error;
681 }
682
683 if (tmp == 1) {
684 do_count_syscalls = 1;
685 } else if (tmp == 0 || tmp == 2 || tmp == 3) {
686 for (int i = 0; i < nsysent; i++) {
687 if (syscalls_log[i] != 0) {
688 if (tmp == 2) {
689 printf("%d calls - name %s \n", syscalls_log[i], syscallnames[i]);
690 } else {
691 syscalls_log[i] = 0;
692 }
693 }
694 }
695 do_count_syscalls = (tmp != 0);
696 }
697
698 return error;
699 }
700 SYSCTL_PROC(_kern, KERN_COUNT_SYSCALLS, count_syscalls, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
701 0, /* Pointer argument (arg1) */
702 0, /* Integer argument (arg2) */
703 sysctl_docountsyscalls, /* Handler function */
704 NULL, /* Data pointer */
705 "");
706 #endif /* COUNT_SYSCALLS */
707
708 /*
709 * The following sysctl_* functions should not be used
710 * any more, as they can only cope with callers in
711 * user mode: Use new-style
712 * sysctl_io_number()
713 * sysctl_io_string()
714 * sysctl_io_opaque()
715 * instead.
716 */
717
718 STATIC int
sysdoproc_filt_KERN_PROC_PID(proc_t p,void * arg)719 sysdoproc_filt_KERN_PROC_PID(proc_t p, void * arg)
720 {
721 if (proc_getpid(p) != (pid_t)*(int*)arg) {
722 return 0;
723 } else {
724 return 1;
725 }
726 }
727
728 STATIC int
sysdoproc_filt_KERN_PROC_PGRP(proc_t p,void * arg)729 sysdoproc_filt_KERN_PROC_PGRP(proc_t p, void * arg)
730 {
731 if (p->p_pgrpid != (pid_t)*(int*)arg) {
732 return 0;
733 } else {
734 return 1;
735 }
736 }
737
738 STATIC int
sysdoproc_filt_KERN_PROC_TTY(proc_t p,void * arg)739 sysdoproc_filt_KERN_PROC_TTY(proc_t p, void * arg)
740 {
741 struct pgrp *pg;
742 dev_t dev = NODEV;
743
744 if ((p->p_flag & P_CONTROLT) && (pg = proc_pgrp(p, NULL)) != PGRP_NULL) {
745 dev = os_atomic_load(&pg->pg_session->s_ttydev, relaxed);
746 pgrp_rele(pg);
747 }
748
749 return dev != NODEV && dev == (dev_t)*(int *)arg;
750 }
751
752 STATIC int
sysdoproc_filt_KERN_PROC_UID(proc_t p,void * arg)753 sysdoproc_filt_KERN_PROC_UID(proc_t p, void * arg)
754 {
755 uid_t uid;
756
757 smr_proc_task_enter();
758 uid = kauth_cred_getuid(proc_ucred_smr(p));
759 smr_proc_task_leave();
760
761 if (uid != (uid_t)*(int*)arg) {
762 return 0;
763 } else {
764 return 1;
765 }
766 }
767
768
769 STATIC int
sysdoproc_filt_KERN_PROC_RUID(proc_t p,void * arg)770 sysdoproc_filt_KERN_PROC_RUID(proc_t p, void * arg)
771 {
772 uid_t ruid;
773
774 smr_proc_task_enter();
775 ruid = kauth_cred_getruid(proc_ucred_smr(p));
776 smr_proc_task_leave();
777
778 if (ruid != (uid_t)*(int*)arg) {
779 return 0;
780 } else {
781 return 1;
782 }
783 }
784
785 /*
786 * try over estimating by 5 procs
787 */
788 #define KERN_PROCSLOP (5 * sizeof(struct kinfo_proc))
789 struct sysdoproc_args {
790 size_t buflen;
791 void *kprocp;
792 boolean_t is_64_bit;
793 user_addr_t dp;
794 size_t needed;
795 unsigned int sizeof_kproc;
796 int *errorp;
797 int uidcheck;
798 int ruidcheck;
799 int ttycheck;
800 int uidval;
801 };
802
803 int
sysdoproc_callback(proc_t p,void * arg)804 sysdoproc_callback(proc_t p, void *arg)
805 {
806 struct sysdoproc_args *args = arg;
807
808 if (args->buflen >= args->sizeof_kproc) {
809 if ((args->ruidcheck != 0) && (sysdoproc_filt_KERN_PROC_RUID(p, &args->uidval) == 0)) {
810 return PROC_RETURNED;
811 }
812 if ((args->uidcheck != 0) && (sysdoproc_filt_KERN_PROC_UID(p, &args->uidval) == 0)) {
813 return PROC_RETURNED;
814 }
815 if ((args->ttycheck != 0) && (sysdoproc_filt_KERN_PROC_TTY(p, &args->uidval) == 0)) {
816 return PROC_RETURNED;
817 }
818
819 bzero(args->kprocp, args->sizeof_kproc);
820 if (args->is_64_bit) {
821 fill_user64_proc(p, args->kprocp);
822 } else {
823 fill_user32_proc(p, args->kprocp);
824 }
825 int error = copyout(args->kprocp, args->dp, args->sizeof_kproc);
826 if (error) {
827 *args->errorp = error;
828 return PROC_RETURNED_DONE;
829 }
830 args->dp += args->sizeof_kproc;
831 args->buflen -= args->sizeof_kproc;
832 }
833 args->needed += args->sizeof_kproc;
834 return PROC_RETURNED;
835 }
836
837 SYSCTL_NODE(_kern, KERN_PROC, proc, CTLFLAG_RD | CTLFLAG_LOCKED, 0, "");
838 STATIC int
839 sysctl_prochandle SYSCTL_HANDLER_ARGS
840 {
841 int cmd = oidp->oid_arg2; /* subcommand for multiple nodes */
842 int *name = arg1; /* oid element argument vector */
843 int namelen = arg2; /* number of oid element arguments */
844 user_addr_t where = req->oldptr;/* user buffer copy out address */
845
846 user_addr_t dp = where;
847 size_t needed = 0;
848 size_t buflen = where != USER_ADDR_NULL ? req->oldlen : 0;
849 int error = 0;
850 boolean_t is_64_bit = proc_is64bit(current_proc());
851 struct user32_kinfo_proc user32_kproc;
852 struct user64_kinfo_proc user_kproc;
853 int sizeof_kproc;
854 void *kprocp;
855 int (*filterfn)(proc_t, void *) = 0;
856 struct sysdoproc_args args;
857 int uidcheck = 0;
858 int ruidcheck = 0;
859 int ttycheck = 0;
860
861 if (namelen != 1 && !(namelen == 0 && cmd == KERN_PROC_ALL)) {
862 return EINVAL;
863 }
864
865 if (is_64_bit) {
866 sizeof_kproc = sizeof(user_kproc);
867 kprocp = &user_kproc;
868 } else {
869 sizeof_kproc = sizeof(user32_kproc);
870 kprocp = &user32_kproc;
871 }
872
873 switch (cmd) {
874 case KERN_PROC_PID:
875 filterfn = sysdoproc_filt_KERN_PROC_PID;
876 break;
877
878 case KERN_PROC_PGRP:
879 filterfn = sysdoproc_filt_KERN_PROC_PGRP;
880 break;
881
882 case KERN_PROC_TTY:
883 ttycheck = 1;
884 break;
885
886 case KERN_PROC_UID:
887 uidcheck = 1;
888 break;
889
890 case KERN_PROC_RUID:
891 ruidcheck = 1;
892 break;
893
894 case KERN_PROC_ALL:
895 break;
896
897 default:
898 /* must be kern.proc.<unknown> */
899 return ENOTSUP;
900 }
901
902 error = 0;
903 args.buflen = buflen;
904 args.kprocp = kprocp;
905 args.is_64_bit = is_64_bit;
906 args.dp = dp;
907 args.needed = needed;
908 args.errorp = &error;
909 args.uidcheck = uidcheck;
910 args.ruidcheck = ruidcheck;
911 args.ttycheck = ttycheck;
912 args.sizeof_kproc = sizeof_kproc;
913 if (namelen) {
914 args.uidval = name[0];
915 }
916
917 proc_iterate((PROC_ALLPROCLIST | PROC_ZOMBPROCLIST),
918 sysdoproc_callback, &args, filterfn, name);
919
920 if (error) {
921 return error;
922 }
923
924 dp = args.dp;
925 needed = args.needed;
926
927 if (where != USER_ADDR_NULL) {
928 req->oldlen = dp - where;
929 if (needed > req->oldlen) {
930 return ENOMEM;
931 }
932 } else {
933 needed += KERN_PROCSLOP;
934 req->oldlen = needed;
935 }
936 /* adjust index so we return the right required/consumed amount */
937 req->oldidx += req->oldlen;
938 return 0;
939 }
940
941
942 /*
943 * We specify the subcommand code for multiple nodes as the 'req->arg2' value
944 * in the sysctl declaration itself, which comes into the handler function
945 * as 'oidp->oid_arg2'.
946 *
947 * For these particular sysctls, since they have well known OIDs, we could
948 * have just obtained it from the '((int *)arg1)[0]' parameter, but that would
949 * not demonstrate how to handle multiple sysctls that used OID_AUTO instead
950 * of a well known value with a common handler function. This is desirable,
951 * because we want well known values to "go away" at some future date.
952 *
953 * It should be noted that the value of '((int *)arg1)[1]' is used for many
954 * an integer parameter to the subcommand for many of these sysctls; we'd
955 * rather have used '((int *)arg1)[0]' for that, or even better, an element
956 * in a structure passed in as the the 'newp' argument to sysctlbyname(3),
957 * and then use leaf-node permissions enforcement, but that would have
958 * necessitated modifying user space code to correspond to the interface
959 * change, and we are striving for binary backward compatibility here; even
960 * though these are SPI, and not intended for use by user space applications
961 * which are not themselves system tools or libraries, some applications
962 * have erroneously used them.
963 */
964 SYSCTL_PROC(_kern_proc, KERN_PROC_ALL, all, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
965 0, /* Pointer argument (arg1) */
966 KERN_PROC_ALL, /* Integer argument (arg2) */
967 sysctl_prochandle, /* Handler function */
968 NULL, /* Data is size variant on ILP32/LP64 */
969 "");
970 SYSCTL_PROC(_kern_proc, KERN_PROC_PID, pid, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
971 0, /* Pointer argument (arg1) */
972 KERN_PROC_PID, /* 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_TTY, tty, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
977 0, /* Pointer argument (arg1) */
978 KERN_PROC_TTY, /* 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_PGRP, pgrp, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
983 0, /* Pointer argument (arg1) */
984 KERN_PROC_PGRP, /* 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_UID, uid, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
989 0, /* Pointer argument (arg1) */
990 KERN_PROC_UID, /* 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_RUID, ruid, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
995 0, /* Pointer argument (arg1) */
996 KERN_PROC_RUID, /* 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_LCID, lcid, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
1001 0, /* Pointer argument (arg1) */
1002 KERN_PROC_LCID, /* Integer argument (arg2) */
1003 sysctl_prochandle, /* Handler function */
1004 NULL, /* Data is size variant on ILP32/LP64 */
1005 "");
1006
1007
1008 /*
1009 * Fill in non-zero fields of an eproc structure for the specified process.
1010 */
1011 STATIC void
fill_user32_eproc(proc_t p,struct user32_eproc * __restrict ep)1012 fill_user32_eproc(proc_t p, struct user32_eproc *__restrict ep)
1013 {
1014 struct pgrp *pg;
1015 struct session *sessp;
1016 kauth_cred_t my_cred;
1017
1018 pg = proc_pgrp(p, &sessp);
1019
1020 if (pg != PGRP_NULL) {
1021 ep->e_pgid = p->p_pgrpid;
1022 ep->e_jobc = pg->pg_jobc;
1023 if (sessp->s_ttyvp) {
1024 ep->e_flag = EPROC_CTTY;
1025 }
1026 }
1027
1028 ep->e_ppid = p->p_ppid;
1029
1030 smr_proc_task_enter();
1031 my_cred = proc_ucred_smr(p);
1032
1033 /* A fake historical pcred */
1034 ep->e_pcred.p_ruid = kauth_cred_getruid(my_cred);
1035 ep->e_pcred.p_svuid = kauth_cred_getsvuid(my_cred);
1036 ep->e_pcred.p_rgid = kauth_cred_getrgid(my_cred);
1037 ep->e_pcred.p_svgid = kauth_cred_getsvgid(my_cred);
1038
1039 /* A fake historical *kauth_cred_t */
1040 unsigned long refcnt = os_atomic_load(&my_cred->cr_ref, relaxed);
1041 ep->e_ucred.cr_ref = (uint32_t)MIN(refcnt, UINT32_MAX);
1042 ep->e_ucred.cr_uid = kauth_cred_getuid(my_cred);
1043 ep->e_ucred.cr_ngroups = (short)posix_cred_get(my_cred)->cr_ngroups;
1044 bcopy(posix_cred_get(my_cred)->cr_groups,
1045 ep->e_ucred.cr_groups, NGROUPS * sizeof(gid_t));
1046
1047 my_cred = NOCRED;
1048 smr_proc_task_leave();
1049
1050 ep->e_tdev = NODEV;
1051 if (pg != PGRP_NULL) {
1052 if (p->p_flag & P_CONTROLT) {
1053 session_lock(sessp);
1054 ep->e_tdev = os_atomic_load(&sessp->s_ttydev, relaxed);
1055 ep->e_tpgid = sessp->s_ttypgrpid;
1056 session_unlock(sessp);
1057 }
1058 if (SESS_LEADER(p, sessp)) {
1059 ep->e_flag |= EPROC_SLEADER;
1060 }
1061 pgrp_rele(pg);
1062 }
1063 }
1064
1065 /*
1066 * Fill in non-zero fields of an LP64 eproc structure for the specified process.
1067 */
1068 STATIC void
fill_user64_eproc(proc_t p,struct user64_eproc * __restrict ep)1069 fill_user64_eproc(proc_t p, struct user64_eproc *__restrict ep)
1070 {
1071 struct pgrp *pg;
1072 struct session *sessp;
1073 kauth_cred_t my_cred;
1074
1075 pg = proc_pgrp(p, &sessp);
1076
1077 if (pg != PGRP_NULL) {
1078 ep->e_pgid = p->p_pgrpid;
1079 ep->e_jobc = pg->pg_jobc;
1080 if (sessp->s_ttyvp) {
1081 ep->e_flag = EPROC_CTTY;
1082 }
1083 }
1084
1085 ep->e_ppid = p->p_ppid;
1086
1087 smr_proc_task_enter();
1088 my_cred = proc_ucred_smr(p);
1089
1090 /* A fake historical pcred */
1091 ep->e_pcred.p_ruid = kauth_cred_getruid(my_cred);
1092 ep->e_pcred.p_svuid = kauth_cred_getsvuid(my_cred);
1093 ep->e_pcred.p_rgid = kauth_cred_getrgid(my_cred);
1094 ep->e_pcred.p_svgid = kauth_cred_getsvgid(my_cred);
1095
1096 /* A fake historical *kauth_cred_t */
1097 unsigned long refcnt = os_atomic_load(&my_cred->cr_ref, relaxed);
1098 ep->e_ucred.cr_ref = (uint32_t)MIN(refcnt, UINT32_MAX);
1099 ep->e_ucred.cr_uid = kauth_cred_getuid(my_cred);
1100 ep->e_ucred.cr_ngroups = (short)posix_cred_get(my_cred)->cr_ngroups;
1101 bcopy(posix_cred_get(my_cred)->cr_groups,
1102 ep->e_ucred.cr_groups, NGROUPS * sizeof(gid_t));
1103
1104 my_cred = NOCRED;
1105 smr_proc_task_leave();
1106
1107 ep->e_tdev = NODEV;
1108 if (pg != PGRP_NULL) {
1109 if (p->p_flag & P_CONTROLT) {
1110 session_lock(sessp);
1111 ep->e_tdev = os_atomic_load(&sessp->s_ttydev, relaxed);
1112 ep->e_tpgid = sessp->s_ttypgrpid;
1113 session_unlock(sessp);
1114 }
1115 if (SESS_LEADER(p, sessp)) {
1116 ep->e_flag |= EPROC_SLEADER;
1117 }
1118 pgrp_rele(pg);
1119 }
1120 }
1121
1122 /*
1123 * Fill in an eproc structure for the specified process.
1124 * bzeroed by our caller, so only set non-zero fields.
1125 */
1126 STATIC void
fill_user32_externproc(proc_t p,struct user32_extern_proc * __restrict exp)1127 fill_user32_externproc(proc_t p, struct user32_extern_proc *__restrict exp)
1128 {
1129 exp->p_starttime.tv_sec = (user32_time_t)p->p_start.tv_sec;
1130 exp->p_starttime.tv_usec = p->p_start.tv_usec;
1131 exp->p_flag = p->p_flag;
1132 #if DEVELOPMENT || DEBUG
1133 if (p->p_lflag & P_LTRACED && !bootarg_hide_process_traced) {
1134 #else
1135 if (p->p_lflag & P_LTRACED) {
1136 #endif
1137 exp->p_flag |= P_TRACED;
1138 }
1139 if (p->p_lflag & P_LPPWAIT) {
1140 exp->p_flag |= P_PPWAIT;
1141 }
1142 if (p->p_lflag & P_LEXIT) {
1143 exp->p_flag |= P_WEXIT;
1144 }
1145 exp->p_stat = p->p_stat;
1146 exp->p_pid = proc_getpid(p);
1147 #if DEVELOPMENT || DEBUG
1148 if (bootarg_hide_process_traced) {
1149 exp->p_oppid = 0;
1150 } else
1151 #endif
1152 {
1153 exp->p_oppid = p->p_oppid;
1154 }
1155 /* Mach related */
1156 exp->p_debugger = p->p_debugger;
1157 exp->sigwait = p->sigwait;
1158 /* scheduling */
1159 #ifdef _PROC_HAS_SCHEDINFO_
1160 exp->p_estcpu = p->p_estcpu;
1161 exp->p_pctcpu = p->p_pctcpu;
1162 exp->p_slptime = p->p_slptime;
1163 #endif
1164 exp->p_realtimer.it_interval.tv_sec =
1165 (user32_time_t)p->p_realtimer.it_interval.tv_sec;
1166 exp->p_realtimer.it_interval.tv_usec =
1167 (__int32_t)p->p_realtimer.it_interval.tv_usec;
1168
1169 exp->p_realtimer.it_value.tv_sec =
1170 (user32_time_t)p->p_realtimer.it_value.tv_sec;
1171 exp->p_realtimer.it_value.tv_usec =
1172 (__int32_t)p->p_realtimer.it_value.tv_usec;
1173
1174 exp->p_rtime.tv_sec = (user32_time_t)p->p_rtime.tv_sec;
1175 exp->p_rtime.tv_usec = (__int32_t)p->p_rtime.tv_usec;
1176
1177 exp->p_sigignore = p->p_sigignore;
1178 exp->p_sigcatch = p->p_sigcatch;
1179 exp->p_priority = p->p_priority;
1180 exp->p_nice = p->p_nice;
1181 bcopy(&p->p_comm, &exp->p_comm, MAXCOMLEN);
1182 exp->p_xstat = (u_short)MIN(p->p_xstat, USHRT_MAX);
1183 exp->p_acflag = p->p_acflag;
1184 }
1185
1186 /*
1187 * Fill in an LP64 version of extern_proc structure for the specified process.
1188 */
1189 STATIC void
1190 fill_user64_externproc(proc_t p, struct user64_extern_proc *__restrict exp)
1191 {
1192 exp->p_starttime.tv_sec = p->p_start.tv_sec;
1193 exp->p_starttime.tv_usec = p->p_start.tv_usec;
1194 exp->p_flag = p->p_flag;
1195 #if DEVELOPMENT || DEBUG
1196 if (p->p_lflag & P_LTRACED && !bootarg_hide_process_traced) {
1197 #else
1198 if (p->p_lflag & P_LTRACED) {
1199 #endif
1200 exp->p_flag |= P_TRACED;
1201 }
1202 if (p->p_lflag & P_LPPWAIT) {
1203 exp->p_flag |= P_PPWAIT;
1204 }
1205 if (p->p_lflag & P_LEXIT) {
1206 exp->p_flag |= P_WEXIT;
1207 }
1208 exp->p_stat = p->p_stat;
1209 exp->p_pid = proc_getpid(p);
1210 #if DEVELOPMENT || DEBUG
1211 if (bootarg_hide_process_traced) {
1212 exp->p_oppid = 0;
1213 } else
1214 #endif
1215 {
1216 exp->p_oppid = p->p_oppid;
1217 }
1218 /* Mach related */
1219 exp->p_debugger = p->p_debugger;
1220 exp->sigwait = p->sigwait;
1221 /* scheduling */
1222 #ifdef _PROC_HAS_SCHEDINFO_
1223 exp->p_estcpu = p->p_estcpu;
1224 exp->p_pctcpu = p->p_pctcpu;
1225 exp->p_slptime = p->p_slptime;
1226 #endif
1227 exp->p_realtimer.it_interval.tv_sec = p->p_realtimer.it_interval.tv_sec;
1228 exp->p_realtimer.it_interval.tv_usec = p->p_realtimer.it_interval.tv_usec;
1229
1230 exp->p_realtimer.it_value.tv_sec = p->p_realtimer.it_value.tv_sec;
1231 exp->p_realtimer.it_value.tv_usec = p->p_realtimer.it_value.tv_usec;
1232
1233 exp->p_rtime.tv_sec = p->p_rtime.tv_sec;
1234 exp->p_rtime.tv_usec = p->p_rtime.tv_usec;
1235
1236 exp->p_sigignore = p->p_sigignore;
1237 exp->p_sigcatch = p->p_sigcatch;
1238 exp->p_priority = p->p_priority;
1239 exp->p_nice = p->p_nice;
1240 bcopy(&p->p_comm, &exp->p_comm, MAXCOMLEN);
1241 exp->p_xstat = (u_short)MIN(p->p_xstat, USHRT_MAX);
1242 exp->p_acflag = p->p_acflag;
1243 }
1244
1245 STATIC void
1246 fill_user32_proc(proc_t p, struct user32_kinfo_proc *__restrict kp)
1247 {
1248 /* on a 64 bit kernel, 32 bit users get some truncated information */
1249 fill_user32_externproc(p, &kp->kp_proc);
1250 fill_user32_eproc(p, &kp->kp_eproc);
1251 }
1252
1253 STATIC void
1254 fill_user64_proc(proc_t p, struct user64_kinfo_proc *__restrict kp)
1255 {
1256 fill_user64_externproc(p, &kp->kp_proc);
1257 fill_user64_eproc(p, &kp->kp_eproc);
1258 }
1259
1260 #if defined(XNU_TARGET_OS_OSX)
1261 /*
1262 * Return the top *sizep bytes of the user stack, or the entire area of the
1263 * user stack down through the saved exec_path, whichever is smaller.
1264 */
1265 STATIC int
1266 sysctl_doprocargs SYSCTL_HANDLER_ARGS
1267 {
1268 __unused int cmd = oidp->oid_arg2; /* subcommand*/
1269 int *name = arg1; /* oid element argument vector */
1270 int namelen = arg2; /* number of oid element arguments */
1271 user_addr_t oldp = req->oldptr; /* user buffer copy out address */
1272 size_t *oldlenp = &req->oldlen; /* user buffer copy out size */
1273 // user_addr_t newp = req->newptr; /* user buffer copy in address */
1274 // size_t newlen = req->newlen; /* user buffer copy in size */
1275 int error;
1276
1277 error = sysctl_procargsx( name, namelen, oldp, oldlenp, current_proc(), 0);
1278
1279 /* adjust index so we return the right required/consumed amount */
1280 if (!error) {
1281 req->oldidx += req->oldlen;
1282 }
1283
1284 return error;
1285 }
1286 SYSCTL_PROC(_kern, KERN_PROCARGS, procargs, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
1287 0, /* Pointer argument (arg1) */
1288 0, /* Integer argument (arg2) */
1289 sysctl_doprocargs, /* Handler function */
1290 NULL, /* Data pointer */
1291 "");
1292 #endif /* defined(XNU_TARGET_OS_OSX) */
1293
1294 STATIC int
1295 sysctl_doprocargs2 SYSCTL_HANDLER_ARGS
1296 {
1297 __unused int cmd = oidp->oid_arg2; /* subcommand*/
1298 int *name = arg1; /* oid element argument vector */
1299 int namelen = arg2; /* number of oid element arguments */
1300 user_addr_t oldp = req->oldptr; /* user buffer copy out address */
1301 size_t *oldlenp = &req->oldlen; /* user buffer copy out size */
1302 // user_addr_t newp = req->newptr; /* user buffer copy in address */
1303 // size_t newlen = req->newlen; /* user buffer copy in size */
1304 int error;
1305
1306 error = sysctl_procargsx( name, namelen, oldp, oldlenp, current_proc(), 1);
1307
1308 /* adjust index so we return the right required/consumed amount */
1309 if (!error) {
1310 req->oldidx += req->oldlen;
1311 }
1312
1313 return error;
1314 }
1315 SYSCTL_PROC(_kern, KERN_PROCARGS2, procargs2, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED,
1316 0, /* Pointer argument (arg1) */
1317 0, /* Integer argument (arg2) */
1318 sysctl_doprocargs2, /* Handler function */
1319 NULL, /* Data pointer */
1320 "");
1321
1322 #define SYSCTL_PROCARGS_READ_ENVVARS_ENTITLEMENT "com.apple.private.read-environment-variables"
1323 STATIC int
1324 sysctl_procargsx(int *name, u_int namelen, user_addr_t where,
1325 size_t *sizep, proc_t cur_proc, int argc_yes)
1326 {
1327 assert(sizep != NULL);
1328 proc_t p = NULL;
1329 size_t buflen = where != USER_ADDR_NULL ? *sizep : 0;
1330 int error = 0;
1331 struct _vm_map *proc_map = NULL;
1332 struct task * task;
1333 vm_map_copy_t tmp = NULL;
1334 user_addr_t arg_addr;
1335 size_t arg_size;
1336 caddr_t data;
1337 size_t argslen = 0;
1338 size_t size = 0;
1339 vm_offset_t copy_start = 0, copy_end;
1340 vm_offset_t smallbuffer_start;
1341 kern_return_t ret;
1342 int pid;
1343 uid_t uid;
1344 int argc = -1;
1345 size_t argvsize;
1346 size_t remaining;
1347 size_t current_arg_index;
1348 size_t current_arg_len;
1349 const char * current_arg;
1350 bool omit_env_vars = true;
1351 user_addr_t user_stack;
1352 vm_map_offset_t effective_page_mask;
1353
1354 if (namelen < 1) {
1355 error = EINVAL;
1356 goto finish;
1357 }
1358
1359 if (argc_yes) {
1360 buflen -= sizeof(int); /* reserve first word to return argc */
1361 }
1362 /* we only care about buflen when where (oldp from sysctl) is not NULL. */
1363 /* when where (oldp from sysctl) is NULL and sizep (oldlenp from sysctl */
1364 /* is not NULL then the caller wants us to return the length needed to */
1365 /* hold the data we would return */
1366 if (where != USER_ADDR_NULL && (buflen <= 0 || buflen > ARG_MAX)) {
1367 error = EINVAL;
1368 goto finish;
1369 }
1370
1371 /*
1372 * Lookup process by pid
1373 */
1374 pid = name[0];
1375 p = proc_find(pid);
1376 if (p == NULL) {
1377 error = EINVAL;
1378 goto finish;
1379 }
1380
1381 /* Allow reading environment variables if any of the following are true:
1382 * - kernel is DEVELOPMENT || DEBUG
1383 * - target process is same as current_proc()
1384 * - target process is not cs_restricted
1385 * - SIP is off
1386 * - caller has an entitlement
1387 */
1388
1389 #if DEVELOPMENT || DEBUG
1390 omit_env_vars = false;
1391 #endif
1392 if (p == current_proc() ||
1393 !cs_restricted(p) ||
1394 #if CONFIG_CSR
1395 csr_check(CSR_ALLOW_UNRESTRICTED_DTRACE) == 0 ||
1396 #endif
1397 IOCurrentTaskHasEntitlement(SYSCTL_PROCARGS_READ_ENVVARS_ENTITLEMENT)
1398 ) {
1399 omit_env_vars = false;
1400 }
1401
1402 /*
1403 * Copy the top N bytes of the stack.
1404 * On all machines we have so far, the stack grows
1405 * downwards.
1406 *
1407 * If the user expects no more than N bytes of
1408 * argument list, use that as a guess for the
1409 * size.
1410 */
1411
1412 if (!p->user_stack) {
1413 error = EINVAL;
1414 goto finish;
1415 }
1416
1417 /* save off argc, argslen, user_stack before releasing the proc */
1418 argc = p->p_argc;
1419 argslen = p->p_argslen;
1420 user_stack = p->user_stack;
1421
1422 /*
1423 * When these sysctls were introduced, the first string in the strings
1424 * section was just the bare path of the executable. However, for security
1425 * reasons we now prefix this string with executable_path= so it can be
1426 * parsed getenv style. To avoid binary compatability issues with exising
1427 * callers of this sysctl, we strip it off here.
1428 * (rdar://problem/13746466)
1429 */
1430 #define EXECUTABLE_KEY "executable_path="
1431 argslen -= strlen(EXECUTABLE_KEY);
1432
1433 if (where == USER_ADDR_NULL && !omit_env_vars) {
1434 /* caller only wants to know length of proc args data.
1435 * If we don't need to omit environment variables, we can skip
1436 * copying the target process stack */
1437 goto calculate_size;
1438 }
1439
1440 smr_proc_task_enter();
1441 uid = kauth_cred_getuid(proc_ucred_smr(p));
1442 smr_proc_task_leave();
1443
1444 if ((uid != kauth_cred_getuid(kauth_cred_get()))
1445 && suser(kauth_cred_get(), &cur_proc->p_acflag)) {
1446 error = EINVAL;
1447 goto finish;
1448 }
1449
1450 /*
1451 * Before we can block (any VM code), make another
1452 * reference to the map to keep it alive. We do
1453 * that by getting a reference on the task itself.
1454 *
1455 * Additionally, if the task is not IPC active, we
1456 * must fail early. Other tasks can't yet look up
1457 * this task's task port to make Mach API calls, so
1458 * we shouldn't make such calls on their behalf.
1459 */
1460 task = proc_task(p);
1461 if (task == NULL || !task_is_ipc_active(task)) {
1462 error = EINVAL;
1463 goto finish;
1464 }
1465
1466 /*
1467 * Once we have a task reference we can convert that into a
1468 * map reference, which we will use in the calls below. The
1469 * task/process may change its map after we take this reference
1470 * (see execve), but the worst that will happen then is a return
1471 * of stale info (which is always a possibility).
1472 */
1473 task_reference(task);
1474 proc_rele(p);
1475 p = NULL;
1476 proc_map = get_task_map_reference(task);
1477 task_deallocate(task);
1478
1479 if (proc_map == NULL) {
1480 error = EINVAL;
1481 goto finish;
1482 }
1483
1484 effective_page_mask = vm_map_page_mask(proc_map);
1485
1486 arg_size = vm_map_round_page(argslen, effective_page_mask);
1487
1488 arg_addr = user_stack - arg_size;
1489
1490 ret = kmem_alloc(kernel_map, ©_start, arg_size,
1491 KMA_DATA | KMA_ZERO, VM_KERN_MEMORY_BSD);
1492 if (ret != KERN_SUCCESS) {
1493 error = ENOMEM;
1494 goto finish;
1495 }
1496
1497 copy_end = copy_start + arg_size;
1498
1499 if (vm_map_copyin(proc_map, (vm_map_address_t)arg_addr,
1500 (vm_map_size_t)arg_size, FALSE, &tmp) != KERN_SUCCESS) {
1501 error = EIO;
1502 goto finish;
1503 }
1504
1505 /*
1506 * Now that we've done the copyin from the process'
1507 * map, we can release the reference to it.
1508 */
1509 vm_map_deallocate(proc_map);
1510 proc_map = NULL;
1511
1512 if (vm_map_copy_overwrite(kernel_map,
1513 (vm_map_address_t)copy_start,
1514 tmp, (vm_map_size_t) arg_size,
1515 FALSE) != KERN_SUCCESS) {
1516 error = EIO;
1517 goto finish;
1518 }
1519 /* tmp was consumed */
1520 tmp = NULL;
1521
1522 if (omit_env_vars) {
1523 argvsize = 0;
1524
1525 /* Iterate over everything in argv, plus one for the bare executable path */
1526 for (current_arg_index = 0; current_arg_index < argc + 1 && argvsize < argslen; ++current_arg_index) {
1527 current_arg = (const char *)(copy_end - argslen) + argvsize;
1528 remaining = argslen - argvsize;
1529 current_arg_len = strnlen(current_arg, remaining);
1530 if (current_arg_len < remaining) {
1531 /* We have space for the null terminator */
1532 current_arg_len += 1;
1533
1534 if (current_arg_index == 0) {
1535 /* The bare executable path may have multiple null bytes after it for alignment */
1536 while (current_arg_len < remaining && current_arg[current_arg_len] == 0) {
1537 current_arg_len += 1;
1538 }
1539 }
1540 }
1541 argvsize += current_arg_len;
1542 }
1543 assert(argvsize <= argslen);
1544
1545 /* Adjust argslen and copy_end to make the copyout range extend to the end of argv */
1546 copy_end = copy_end - argslen + argvsize;
1547 argslen = argvsize;
1548 }
1549
1550 if (where == USER_ADDR_NULL) {
1551 /* Skip copyout */
1552 goto calculate_size;
1553 }
1554
1555 if (buflen >= argslen) {
1556 data = (caddr_t) (copy_end - argslen);
1557 size = argslen;
1558 } else {
1559 /*
1560 * Before rdar://25397314, this function contained incorrect logic when buflen is less
1561 * than argslen. The problem was that it copied in `buflen` bytes from the end of the target
1562 * process user stack into the beginning of a buffer of size round_page(buflen), and then
1563 * copied out `buflen` bytes from the end of this buffer. The effect of this was that
1564 * the caller of this sysctl would get zeros at the end of their buffer.
1565 *
1566 * To preserve this behavior, bzero everything from copy_end-round_page(buflen)+buflen to the
1567 * end of the buffer. This emulates copying in only `buflen` bytes.
1568 *
1569 *
1570 * In the old code:
1571 *
1572 * copy_start .... size: round_page(buflen) .... copy_end
1573 * [---copied in data (size: buflen)---|--- zeros ----------]
1574 * ^
1575 * data = copy_end - buflen
1576 *
1577 *
1578 * In the new code:
1579 * copy_start .... size: round_page(p->argslen) .... full copy_end
1580 * ^ ....................... p->argslen ...............................^
1581 * ^ ^ truncated copy_end ^
1582 * ^ ^ ^ ^
1583 * ^ ................ argslen ........................ ^
1584 * ^ ^ ^ ^
1585 * [-------copied in data (size: round_page(p->argslen))-------:----env vars---]
1586 * ^ ^
1587 * ^ data = copy_end - buflen
1588 * smallbuffer_start = max(copy_end - round_page(buflen), copy_start)
1589 *
1590 *
1591 * Full copy_end: copy_end calculated from copy_start + round_page(p->argslen)
1592 * Truncated copy_end: copy_end after truncation to remove environment variables.
1593 *
1594 * If environment variables were omitted, then we use the truncated copy_end, otherwise
1595 * we use full copy_end.
1596 *
1597 * smallbuffer_start: represents where copy_start would be in the old code.
1598 * data: The beginning of the region we copyout
1599 */
1600 smallbuffer_start = copy_end - vm_map_round_page(buflen, effective_page_mask);
1601 if (smallbuffer_start < copy_start) {
1602 smallbuffer_start = copy_start;
1603 }
1604 bzero((void *)(smallbuffer_start + buflen), copy_end - (smallbuffer_start + buflen));
1605 data = (caddr_t) (copy_end - buflen);
1606 size = buflen;
1607 }
1608
1609 if (argc_yes) {
1610 /* Put processes argc as the first word in the copyout buffer */
1611 suword(where, argc);
1612 error = copyout(data, (where + sizeof(int)), size);
1613 size += sizeof(int);
1614 } else {
1615 error = copyout(data, where, size);
1616
1617 /*
1618 * Make the old PROCARGS work to return the executable's path
1619 * But, only if there is enough space in the provided buffer
1620 *
1621 * on entry: data [possibily] points to the beginning of the path
1622 *
1623 * Note: we keep all pointers&sizes aligned to word boundries
1624 */
1625 if ((!error) && (buflen > 0 && (u_int)buflen > size)) {
1626 int binPath_sz, alignedBinPath_sz = 0;
1627 int extraSpaceNeeded, addThis;
1628 user_addr_t placeHere;
1629 char * str = (char *) data;
1630 size_t max_len = size;
1631
1632 /* Some apps are really bad about messing up their stacks
1633 * So, we have to be extra careful about getting the length
1634 * of the executing binary. If we encounter an error, we bail.
1635 */
1636
1637 /* Limit ourselves to PATH_MAX paths */
1638 if (max_len > PATH_MAX) {
1639 max_len = PATH_MAX;
1640 }
1641
1642 binPath_sz = 0;
1643
1644 while ((binPath_sz < max_len - 1) && (*str++ != 0)) {
1645 binPath_sz++;
1646 }
1647
1648 /* If we have a NUL terminator, copy it, too */
1649 if (binPath_sz < max_len - 1) {
1650 binPath_sz += 1;
1651 }
1652
1653 /* Pre-Flight the space requiremnts */
1654
1655 /* Account for the padding that fills out binPath to the next word */
1656 alignedBinPath_sz += (binPath_sz & (sizeof(int) - 1)) ? (sizeof(int) - (binPath_sz & (sizeof(int) - 1))) : 0;
1657
1658 placeHere = where + size;
1659
1660 /* Account for the bytes needed to keep placeHere word aligned */
1661 addThis = (placeHere & (sizeof(int) - 1)) ? (sizeof(int) - (placeHere & (sizeof(int) - 1))) : 0;
1662
1663 /* Add up all the space that is needed */
1664 extraSpaceNeeded = alignedBinPath_sz + addThis + binPath_sz + (4 * sizeof(int));
1665
1666 /* is there is room to tack on argv[0]? */
1667 if ((buflen & ~(sizeof(int) - 1)) >= (size + extraSpaceNeeded)) {
1668 placeHere += addThis;
1669 suword(placeHere, 0);
1670 placeHere += sizeof(int);
1671 suword(placeHere, 0xBFFF0000);
1672 placeHere += sizeof(int);
1673 suword(placeHere, 0);
1674 placeHere += sizeof(int);
1675 error = copyout(data, placeHere, binPath_sz);
1676 if (!error) {
1677 placeHere += binPath_sz;
1678 suword(placeHere, 0);
1679 size += extraSpaceNeeded;
1680 }
1681 }
1682 }
1683 }
1684
1685 calculate_size:
1686 /* Size has already been calculated for the where != NULL case */
1687 if (where == USER_ADDR_NULL) {
1688 size = argslen;
1689 if (argc_yes) {
1690 size += sizeof(int);
1691 } else {
1692 /*
1693 * old PROCARGS will return the executable's path and plus some
1694 * extra space for work alignment and data tags
1695 */
1696 size += PATH_MAX + (6 * sizeof(int));
1697 }
1698 size += (size & (sizeof(int) - 1)) ? (sizeof(int) - (size & (sizeof(int) - 1))) : 0;
1699 }
1700
1701 *sizep = size;
1702
1703 finish:
1704 if (p != NULL) {
1705 proc_rele(p);
1706 }
1707 if (tmp != NULL) {
1708 vm_map_copy_discard(tmp);
1709 }
1710 if (proc_map != NULL) {
1711 vm_map_deallocate(proc_map);
1712 }
1713 if (copy_start != (vm_offset_t) 0) {
1714 kmem_free(kernel_map, copy_start, arg_size);
1715 }
1716 return error;
1717 }
1718
1719
1720 /*
1721 * Max number of concurrent aio requests
1722 */
1723 STATIC int
1724 sysctl_aiomax
1725 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1726 {
1727 int new_value, changed;
1728 int error = sysctl_io_number(req, aio_max_requests, sizeof(int), &new_value, &changed);
1729 if (changed) {
1730 /* make sure the system-wide limit is greater than the per process limit */
1731 if (new_value >= aio_max_requests_per_process && new_value <= AIO_MAX_REQUESTS) {
1732 aio_max_requests = new_value;
1733 } else {
1734 error = EINVAL;
1735 }
1736 }
1737 return error;
1738 }
1739
1740
1741 /*
1742 * Max number of concurrent aio requests per process
1743 */
1744 STATIC int
1745 sysctl_aioprocmax
1746 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1747 {
1748 int new_value, changed;
1749 int error = sysctl_io_number(req, aio_max_requests_per_process, sizeof(int), &new_value, &changed);
1750 if (changed) {
1751 /* make sure per process limit is less than the system-wide limit */
1752 if (new_value <= aio_max_requests && new_value >= AIO_LISTIO_MAX) {
1753 aio_max_requests_per_process = new_value;
1754 } else {
1755 error = EINVAL;
1756 }
1757 }
1758 return error;
1759 }
1760
1761
1762 /*
1763 * Max number of async IO worker threads
1764 */
1765 STATIC int
1766 sysctl_aiothreads
1767 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1768 {
1769 int new_value, changed;
1770 int error = sysctl_io_number(req, aio_worker_threads, sizeof(int), &new_value, &changed);
1771 if (changed) {
1772 /* we only allow an increase in the number of worker threads */
1773 if (new_value > aio_worker_threads) {
1774 _aio_create_worker_threads((new_value - aio_worker_threads));
1775 aio_worker_threads = new_value;
1776 } else {
1777 error = EINVAL;
1778 }
1779 }
1780 return error;
1781 }
1782
1783
1784 /*
1785 * System-wide limit on the max number of processes
1786 */
1787 STATIC int
1788 sysctl_maxproc
1789 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1790 {
1791 int new_value, changed;
1792 int error = sysctl_io_number(req, maxproc, sizeof(int), &new_value, &changed);
1793 if (changed) {
1794 AUDIT_ARG(value32, new_value);
1795 /* make sure the system-wide limit is less than the configured hard
1796 * limit set at kernel compilation */
1797 if (new_value <= hard_maxproc && new_value > 0) {
1798 maxproc = new_value;
1799 } else {
1800 error = EINVAL;
1801 }
1802 }
1803 return error;
1804 }
1805
1806 #if CONFIG_SCHED_SMT
1807 STATIC int
1808 sysctl_sched_enable_smt
1809 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1810 {
1811 int new_value, changed;
1812 int error = sysctl_io_number(req, sched_enable_smt, sizeof(int), &new_value, &changed);
1813 if (error) {
1814 return error;
1815 }
1816 kern_return_t kret = KERN_SUCCESS;
1817 if (changed) {
1818 AUDIT_ARG(value32, new_value);
1819 if (new_value == 0) {
1820 sched_enable_smt = 0;
1821 kret = enable_smt_processors(false);
1822 } else {
1823 sched_enable_smt = 1;
1824 kret = enable_smt_processors(true);
1825 }
1826 }
1827 switch (kret) {
1828 case KERN_SUCCESS:
1829 error = 0;
1830 break;
1831 case KERN_INVALID_ARGUMENT:
1832 error = EINVAL;
1833 break;
1834 case KERN_FAILURE:
1835 error = EBUSY;
1836 break;
1837 default:
1838 error = ENOENT;
1839 break;
1840 }
1841 return error;
1842 }
1843 #else /* CONFIG_SCHED_SMT */
1844 STATIC int
1845 sysctl_sched_enable_smt
1846 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, __unused struct sysctl_req *req)
1847 {
1848 return 0;
1849 }
1850 #endif /* CONFIG_SCHED_SMT */
1851
1852 SYSCTL_STRING(_kern, KERN_OSTYPE, ostype,
1853 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
1854 ostype, 0, "");
1855 SYSCTL_STRING(_kern, KERN_OSRELEASE, osrelease,
1856 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
1857 osrelease, 0, "");
1858 SYSCTL_INT(_kern, KERN_OSREV, osrevision,
1859 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
1860 (int *)NULL, BSD, "");
1861 SYSCTL_STRING(_kern, KERN_VERSION, version,
1862 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
1863 version, 0, "");
1864 SYSCTL_STRING(_kern, OID_AUTO, uuid,
1865 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
1866 &kernel_uuid_string[0], 0, "");
1867
1868 SYSCTL_STRING(_kern, OID_AUTO, osbuildconfig,
1869 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED | CTLFLAG_MASKED,
1870 &osbuild_config[0], 0, "");
1871
1872 #if DEBUG
1873 #ifndef DKPR
1874 #define DKPR 1
1875 #endif
1876 #endif
1877
1878 #if DKPR
1879 int debug_kprint_syscall = 0;
1880 char debug_kprint_syscall_process[MAXCOMLEN + 1];
1881
1882 /* Thread safe: bits and string value are not used to reclaim state */
1883 SYSCTL_INT(_debug, OID_AUTO, kprint_syscall,
1884 CTLFLAG_RW | CTLFLAG_LOCKED, &debug_kprint_syscall, 0, "kprintf syscall tracing");
1885 SYSCTL_STRING(_debug, OID_AUTO, kprint_syscall_process,
1886 CTLFLAG_RW | CTLFLAG_LOCKED, debug_kprint_syscall_process, sizeof(debug_kprint_syscall_process),
1887 "name of process for kprintf syscall tracing");
1888
1889 int
1890 debug_kprint_current_process(const char **namep)
1891 {
1892 struct proc *p = current_proc();
1893
1894 if (p == NULL) {
1895 return 0;
1896 }
1897
1898 if (debug_kprint_syscall_process[0]) {
1899 /* user asked to scope tracing to a particular process name */
1900 if (0 == strncmp(debug_kprint_syscall_process,
1901 p->p_comm, sizeof(debug_kprint_syscall_process))) {
1902 /* no value in telling the user that we traced what they asked */
1903 if (namep) {
1904 *namep = NULL;
1905 }
1906
1907 return 1;
1908 } else {
1909 return 0;
1910 }
1911 }
1912
1913 /* trace all processes. Tell user what we traced */
1914 if (namep) {
1915 *namep = p->p_comm;
1916 }
1917
1918 return 1;
1919 }
1920 #endif
1921
1922 /* PR-5293665: need to use a callback function for kern.osversion to set
1923 * osversion in IORegistry */
1924
1925 STATIC int
1926 sysctl_osversion(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
1927 {
1928 int rval = 0;
1929
1930 rval = sysctl_handle_string(oidp, arg1, arg2, req);
1931
1932 if (req->newptr) {
1933 IORegistrySetOSBuildVersion((char *)arg1);
1934 }
1935
1936 return rval;
1937 }
1938
1939 SYSCTL_PROC(_kern, KERN_OSVERSION, osversion,
1940 CTLFLAG_RW | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
1941 osversion, 256 /* OSVERSIZE*/,
1942 sysctl_osversion, "A", "");
1943
1944 static bool
1945 _already_set_or_not_launchd(struct sysctl_req *req, char *val)
1946 {
1947 if (req->newptr != 0) {
1948 /*
1949 * Can only ever be set by launchd, and only once at boot.
1950 */
1951 if (proc_getpid(req->p) != 1 || val[0] != '\0') {
1952 return true;
1953 }
1954 }
1955 return false;
1956 }
1957
1958 #define kRootsInstalledReadWriteEntitlement "com.apple.private.roots-installed-read-write"
1959 #define kRootsInstalledReadOnlyEntitlement "com.apple.private.roots-installed-read-only"
1960 uint64_t roots_installed = 0;
1961
1962 static int
1963 sysctl_roots_installed
1964 (__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
1965 {
1966 int error = 0;
1967
1968 if (req->newptr != 0) {
1969 /* a ReadWrite entitlement is required for updating this syscl
1970 * meanwhile, only allow write once
1971 */
1972 if (!IOCurrentTaskHasEntitlement(kRootsInstalledReadWriteEntitlement) || (roots_installed != 0)) {
1973 return EPERM;
1974 }
1975 } else {
1976 /* for reader of this sysctl, need either ReadWrite or ReadOnly entitlement */
1977 if (!IOCurrentTaskHasEntitlement(kRootsInstalledReadWriteEntitlement) &&
1978 !IOCurrentTaskHasEntitlement(kRootsInstalledReadOnlyEntitlement)) {
1979 return EPERM;
1980 }
1981 }
1982
1983 error = sysctl_handle_quad(oidp, arg1, arg2, req);
1984
1985 return error;
1986 }
1987
1988 SYSCTL_PROC(_kern, OID_AUTO, roots_installed,
1989 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
1990 &roots_installed, sizeof(roots_installed),
1991 sysctl_roots_installed, "Q", "");
1992
1993 #if XNU_TARGET_OS_OSX
1994 static int
1995 sysctl_system_version_compat
1996 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
1997 {
1998 int oldval = (task_has_system_version_compat_enabled(current_task()));
1999 int new_value = 0, changed = 0;
2000
2001 int error = sysctl_io_number(req, oldval, sizeof(int), &new_value, &changed);
2002 if (changed) {
2003 task_set_system_version_compat_enabled(current_task(), (new_value));
2004 }
2005 return error;
2006 }
2007
2008 SYSCTL_PROC(_kern, OID_AUTO, system_version_compat,
2009 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
2010 0, 0, sysctl_system_version_compat, "A", "");
2011
2012 char osproductversioncompat[48] = { '\0' };
2013
2014 static int
2015 sysctl_osproductversioncompat(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2016 {
2017 if (_already_set_or_not_launchd(req, osproductversioncompat)) {
2018 return EPERM;
2019 }
2020 return sysctl_handle_string(oidp, arg1, arg2, req);
2021 }
2022
2023
2024 SYSCTL_PROC(_kern, OID_AUTO, osproductversioncompat,
2025 CTLFLAG_RW | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2026 osproductversioncompat, sizeof(osproductversioncompat),
2027 sysctl_osproductversioncompat, "A", "The ProductVersion from SystemVersionCompat.plist");
2028 #endif
2029
2030 char osproductversion[48] = { '\0' };
2031
2032 static char iossupportversion_string[48] = { '\0' };
2033
2034 static int
2035 sysctl_osproductversion(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2036 {
2037 if (_already_set_or_not_launchd(req, osproductversion)) {
2038 return EPERM;
2039 }
2040
2041 #if XNU_TARGET_OS_OSX
2042 if (task_has_system_version_compat_enabled(current_task()) && (osproductversioncompat[0] != '\0')) {
2043 return sysctl_handle_string(oidp, osproductversioncompat, arg2, req);
2044 } else {
2045 return sysctl_handle_string(oidp, arg1, arg2, req);
2046 }
2047 #elif defined(XNU_TARGET_OS_XR)
2048 if (proc_platform(req->p) == PLATFORM_IOS && (iossupportversion_string[0] != '\0')) {
2049 return sysctl_handle_string(oidp, iossupportversion_string, arg2, req);
2050 } else {
2051 return sysctl_handle_string(oidp, arg1, arg2, req);
2052 }
2053 #else
2054 return sysctl_handle_string(oidp, arg1, arg2, req);
2055 #endif
2056 }
2057
2058 #if XNU_TARGET_OS_OSX
2059 static_assert(sizeof(osproductversioncompat) == sizeof(osproductversion),
2060 "osproductversion size matches osproductversioncompat size");
2061 #endif
2062
2063 SYSCTL_PROC(_kern, OID_AUTO, osproductversion,
2064 CTLFLAG_RW | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2065 osproductversion, sizeof(osproductversion),
2066 sysctl_osproductversion, "A", "The ProductVersion from SystemVersion.plist");
2067
2068 char osreleasetype[OSRELEASETYPE_SIZE] = { '\0' };
2069
2070 STATIC int
2071 sysctl_osreleasetype(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2072 {
2073 if (_already_set_or_not_launchd(req, osreleasetype)) {
2074 return EPERM;
2075 }
2076 return sysctl_handle_string(oidp, arg1, arg2, req);
2077 }
2078
2079 void reset_osreleasetype(void);
2080
2081 void
2082 reset_osreleasetype(void)
2083 {
2084 memset(osreleasetype, 0, sizeof(osreleasetype));
2085 }
2086
2087 SYSCTL_PROC(_kern, OID_AUTO, osreleasetype,
2088 CTLFLAG_RW | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2089 osreleasetype, sizeof(osreleasetype),
2090 sysctl_osreleasetype, "A", "The ReleaseType from SystemVersion.plist");
2091
2092 STATIC int
2093 sysctl_iossupportversion(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2094 {
2095 if (_already_set_or_not_launchd(req, iossupportversion_string)) {
2096 return EPERM;
2097 }
2098
2099 #if defined(XNU_TARGET_OS_XR)
2100 if (proc_platform(req->p) == PLATFORM_IOS) {
2101 /* return empty string for iOS processes to match how this would behave on iOS */
2102 return sysctl_handle_string(oidp, "", arg2, req);
2103 } else {
2104 /* native processes see the actual value */
2105 return sysctl_handle_string(oidp, arg1, arg2, req);
2106 }
2107 #else
2108 return sysctl_handle_string(oidp, arg1, arg2, req);
2109 #endif
2110 }
2111
2112 SYSCTL_PROC(_kern, OID_AUTO, iossupportversion,
2113 CTLFLAG_RW | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2114 iossupportversion_string, sizeof(iossupportversion_string),
2115 sysctl_iossupportversion, "A", "The iOSSupportVersion from SystemVersion.plist");
2116
2117 static uint64_t osvariant_status = 0;
2118
2119 STATIC int
2120 sysctl_osvariant_status(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2121 {
2122 if (req->newptr != 0) {
2123 /*
2124 * Can only ever be set by launchd, and only once.
2125 * Reset by usrctl() -> reset_osvariant_status() during
2126 * userspace reboot, since userspace could reboot into
2127 * a different variant.
2128 */
2129 if (proc_getpid(req->p) != 1 || osvariant_status != 0) {
2130 return EPERM;
2131 }
2132 }
2133
2134 int err = sysctl_handle_quad(oidp, arg1, arg2, req);
2135
2136 reset_debug_syscall_rejection_mode();
2137
2138 return err;
2139 }
2140
2141 SYSCTL_PROC(_kern, OID_AUTO, osvariant_status,
2142 CTLFLAG_RW | CTLTYPE_QUAD | CTLFLAG_LOCKED | CTLFLAG_MASKED,
2143 &osvariant_status, sizeof(osvariant_status),
2144 sysctl_osvariant_status, "Q", "Opaque flags used to cache OS variant information");
2145
2146 static bool
2147 _os_variant_check_disabled(enum os_variant_property property)
2148 {
2149 return (osvariant_status >> (32 + property)) & 0x1;
2150 }
2151
2152 static bool
2153 _os_variant_has(enum os_variant_status_flags_positions p)
2154 {
2155 return ((osvariant_status >> (p * OS_VARIANT_STATUS_BIT_WIDTH)) & OS_VARIANT_STATUS_MASK) == OS_VARIANT_S_YES;
2156 }
2157
2158 bool
2159 os_variant_has_internal_diagnostics(__unused const char *subsystem)
2160 {
2161 if (_os_variant_check_disabled(OS_VARIANT_PROPERTY_DIAGNOSTICS)) {
2162 return false;
2163 }
2164 #if XNU_TARGET_OS_OSX
2165 return _os_variant_has(OS_VARIANT_SFP_INTERNAL_CONTENT) || _os_variant_has(OS_VARIANT_SFP_INTERNAL_DIAGS_PROFILE);
2166 #else
2167 return _os_variant_has(OS_VARIANT_SFP_INTERNAL_RELEASE_TYPE);
2168 #endif /* XNU_TARGET_OS_OSX */
2169 }
2170
2171 void reset_osvariant_status(void);
2172
2173 void
2174 reset_osvariant_status(void)
2175 {
2176 osvariant_status = 0;
2177 reset_debug_syscall_rejection_mode();
2178 }
2179
2180 extern void commpage_update_dyld_flags(uint64_t);
2181 TUNABLE_WRITEABLE(uint64_t, dyld_flags, "dyld_flags", 0);
2182
2183 STATIC int
2184 sysctl_dyld_flags(__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2185 {
2186 /*
2187 * Can only ever be set by launchd, possibly several times
2188 * as dyld may change its mind after a userspace reboot.
2189 */
2190 if (req->newptr != 0 && proc_getpid(req->p) != 1) {
2191 return EPERM;
2192 }
2193
2194 int res = sysctl_handle_quad(oidp, arg1, arg2, req);
2195 if (req->newptr && res == 0) {
2196 commpage_update_dyld_flags(dyld_flags);
2197 }
2198 return res;
2199 }
2200
2201 SYSCTL_PROC(_kern, OID_AUTO, dyld_flags,
2202 CTLFLAG_RW | CTLTYPE_QUAD | CTLFLAG_LOCKED | CTLFLAG_MASKED,
2203 &dyld_flags, sizeof(dyld_flags),
2204 sysctl_dyld_flags, "Q", "Opaque flags used to cache dyld system-wide configuration");
2205
2206 #if defined(XNU_TARGET_OS_BRIDGE)
2207 char macosproductversion[MACOS_VERS_LEN] = { '\0' };
2208
2209 SYSCTL_STRING(_kern, OID_AUTO, macosproductversion,
2210 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2211 &macosproductversion[0], MACOS_VERS_LEN, "The currently running macOS ProductVersion (from SystemVersion.plist on macOS)");
2212
2213 char macosversion[MACOS_VERS_LEN] = { '\0' };
2214
2215 SYSCTL_STRING(_kern, OID_AUTO, macosversion,
2216 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2217 &macosversion[0], MACOS_VERS_LEN, "The currently running macOS build version");
2218 #endif
2219
2220 STATIC int
2221 sysctl_sysctl_bootargs
2222 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2223 {
2224 int error;
2225 char buf[BOOT_LINE_LENGTH];
2226
2227 strlcpy(buf, PE_boot_args(), BOOT_LINE_LENGTH);
2228 error = sysctl_io_string(req, buf, BOOT_LINE_LENGTH, 0, NULL);
2229 return error;
2230 }
2231
2232 SYSCTL_PROC(_kern, OID_AUTO, bootargs,
2233 CTLFLAG_LOCKED | CTLFLAG_RD | CTLFLAG_KERN | CTLTYPE_STRING,
2234 NULL, 0,
2235 sysctl_sysctl_bootargs, "A", "bootargs");
2236
2237 STATIC int
2238 sysctl_kernelcacheuuid(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2239 {
2240 int rval = ENOENT;
2241 if (kernelcache_uuid_valid) {
2242 rval = sysctl_handle_string(oidp, arg1, arg2, req);
2243 }
2244 return rval;
2245 }
2246
2247 SYSCTL_PROC(_kern, OID_AUTO, kernelcacheuuid,
2248 CTLFLAG_RD | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2249 kernelcache_uuid_string, sizeof(kernelcache_uuid_string),
2250 sysctl_kernelcacheuuid, "A", "");
2251
2252 STATIC int
2253 sysctl_systemfilesetuuid(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2254 {
2255 int rval = ENOENT;
2256 if (pageablekc_uuid_valid) {
2257 rval = sysctl_handle_string(oidp, arg1, arg2, req);
2258 }
2259 return rval;
2260 }
2261
2262 SYSCTL_PROC(_kern, OID_AUTO, systemfilesetuuid,
2263 CTLFLAG_RD | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2264 pageablekc_uuid_string, sizeof(pageablekc_uuid_string),
2265 sysctl_systemfilesetuuid, "A", "");
2266
2267 STATIC int
2268 sysctl_auxiliaryfilesetuuid(struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2269 {
2270 int rval = ENOENT;
2271 if (auxkc_uuid_valid) {
2272 rval = sysctl_handle_string(oidp, arg1, arg2, req);
2273 }
2274 return rval;
2275 }
2276
2277 SYSCTL_PROC(_kern, OID_AUTO, auxiliaryfilesetuuid,
2278 CTLFLAG_RD | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2279 auxkc_uuid_string, sizeof(auxkc_uuid_string),
2280 sysctl_auxiliaryfilesetuuid, "A", "");
2281
2282 STATIC int
2283 sysctl_filesetuuid(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2284 {
2285 int rval = ENOENT;
2286 kc_format_t kcformat;
2287 kernel_mach_header_t *mh;
2288 void *uuid = NULL;
2289 unsigned long uuidlen = 0;
2290 uuid_string_t uuid_str;
2291
2292 if (!PE_get_primary_kc_format(&kcformat) || kcformat != KCFormatFileset) {
2293 return rval;
2294 }
2295
2296 mh = (kernel_mach_header_t *)PE_get_kc_header(KCKindPrimary);
2297 uuid = getuuidfromheader(mh, &uuidlen);
2298
2299 if ((uuid != NULL) && (uuidlen == sizeof(uuid_t))) {
2300 uuid_unparse_upper(*(uuid_t *)uuid, uuid_str);
2301 rval = sysctl_io_string(req, (char *)uuid_str, sizeof(uuid_str), 0, NULL);
2302 }
2303
2304 return rval;
2305 }
2306
2307 SYSCTL_PROC(_kern, OID_AUTO, filesetuuid,
2308 CTLFLAG_RD | CTLFLAG_KERN | CTLTYPE_STRING | CTLFLAG_LOCKED,
2309 NULL, 0,
2310 sysctl_filesetuuid, "A", "");
2311
2312
2313 SYSCTL_INT(_kern, KERN_MAXFILES, maxfiles,
2314 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2315 &maxfiles, 0, "");
2316 SYSCTL_INT(_kern, KERN_ARGMAX, argmax,
2317 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2318 (int *)NULL, ARG_MAX, "");
2319 SYSCTL_INT(_kern, KERN_POSIX1, posix1version,
2320 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2321 (int *)NULL, _POSIX_VERSION, "");
2322 SYSCTL_INT(_kern, KERN_NGROUPS, ngroups,
2323 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2324 (int *)NULL, NGROUPS_MAX, "");
2325 SYSCTL_INT(_kern, KERN_JOB_CONTROL, job_control,
2326 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2327 (int *)NULL, 1, "");
2328 #if 1 /* _POSIX_SAVED_IDS from <unistd.h> */
2329 SYSCTL_INT(_kern, KERN_SAVED_IDS, saved_ids,
2330 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2331 (int *)NULL, 1, "");
2332 #else
2333 SYSCTL_INT(_kern, KERN_SAVED_IDS, saved_ids,
2334 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
2335 NULL, 0, "");
2336 #endif
2337 SYSCTL_INT(_kern, OID_AUTO, num_files,
2338 CTLFLAG_RD | CTLFLAG_LOCKED,
2339 &nfiles, 0, "");
2340 SYSCTL_COMPAT_INT(_kern, OID_AUTO, num_vnodes,
2341 CTLFLAG_RD | CTLFLAG_LOCKED,
2342 &numvnodes, 0, "");
2343 SYSCTL_INT(_kern, OID_AUTO, num_tasks,
2344 CTLFLAG_RD | CTLFLAG_LOCKED,
2345 &task_max, 0, "");
2346 SYSCTL_INT(_kern, OID_AUTO, num_threads,
2347 CTLFLAG_RD | CTLFLAG_LOCKED,
2348 &thread_max, 0, "");
2349 SYSCTL_INT(_kern, OID_AUTO, num_taskthreads,
2350 CTLFLAG_RD | CTLFLAG_LOCKED,
2351 &task_threadmax, 0, "");
2352 SYSCTL_LONG(_kern, OID_AUTO, num_recycledvnodes,
2353 CTLFLAG_RD | CTLFLAG_LOCKED,
2354 &num_recycledvnodes, "");
2355 SYSCTL_COMPAT_INT(_kern, OID_AUTO, free_vnodes,
2356 CTLFLAG_RD | CTLFLAG_LOCKED,
2357 &freevnodes, 0, "");
2358
2359 STATIC int
2360 sysctl_maxvnodes(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2361 {
2362 int oldval = desiredvnodes;
2363 int error = sysctl_io_number(req, desiredvnodes, sizeof(int), &desiredvnodes, NULL);
2364
2365 if (oldval != desiredvnodes) {
2366 resize_namecache(desiredvnodes);
2367 }
2368
2369 return error;
2370 }
2371
2372 SYSCTL_INT(_kern, OID_AUTO, namecache_disabled,
2373 CTLFLAG_RW | CTLFLAG_LOCKED,
2374 &nc_disabled, 0, "");
2375
2376 SYSCTL_PROC(_kern, KERN_MAXVNODES, maxvnodes,
2377 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2378 0, 0, sysctl_maxvnodes, "I", "");
2379
2380 SYSCTL_PROC(_kern, KERN_MAXPROC, maxproc,
2381 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2382 0, 0, sysctl_maxproc, "I", "");
2383
2384 SYSCTL_PROC(_kern, KERN_AIOMAX, aiomax,
2385 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2386 0, 0, sysctl_aiomax, "I", "");
2387
2388 SYSCTL_PROC(_kern, KERN_AIOPROCMAX, aioprocmax,
2389 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2390 0, 0, sysctl_aioprocmax, "I", "");
2391
2392 SYSCTL_PROC(_kern, KERN_AIOTHREADS, aiothreads,
2393 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2394 0, 0, sysctl_aiothreads, "I", "");
2395
2396 SYSCTL_PROC(_kern, OID_AUTO, sched_enable_smt,
2397 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_KERN,
2398 0, 0, sysctl_sched_enable_smt, "I", "");
2399
2400 extern int sched_allow_NO_SMT_threads;
2401 SYSCTL_INT(_kern, OID_AUTO, sched_allow_NO_SMT_threads,
2402 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2403 &sched_allow_NO_SMT_threads, 0, "");
2404
2405 extern int sched_avoid_cpu0;
2406 SYSCTL_INT(_kern, OID_AUTO, sched_rt_avoid_cpu0,
2407 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2408 &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");
2409
2410 #if (DEVELOPMENT || DEBUG)
2411
2412 static int
2413 sysctl_kern_max_unsafe_rt_quanta(__unused struct sysctl_oid *oidp,
2414 __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2415 {
2416 extern void sched_set_max_unsafe_rt_quanta(int);
2417 extern int max_unsafe_rt_quanta;
2418
2419 int new_value, changed;
2420 int old_value = max_unsafe_rt_quanta;
2421 int error = sysctl_io_number(req, old_value, sizeof(int), &new_value,
2422 &changed);
2423 if (changed) {
2424 sched_set_max_unsafe_rt_quanta(new_value);
2425 }
2426
2427 return error;
2428 }
2429
2430 SYSCTL_PROC(_kern, OID_AUTO, max_unsafe_rt_quanta,
2431 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2432 0, 0, sysctl_kern_max_unsafe_rt_quanta, "I",
2433 "Number of quanta to allow a realtime "
2434 "thread to run before being penalized");
2435
2436 static int
2437 sysctl_kern_max_unsafe_fixed_quanta(__unused struct sysctl_oid *oidp,
2438 __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2439 {
2440 extern void sched_set_max_unsafe_fixed_quanta(int);
2441 extern int max_unsafe_fixed_quanta;
2442
2443 int new_value, changed;
2444 int old_value = max_unsafe_fixed_quanta;
2445 int error = sysctl_io_number(req, old_value, sizeof(int), &new_value,
2446 &changed);
2447 if (changed) {
2448 sched_set_max_unsafe_fixed_quanta(new_value);
2449 }
2450
2451 return error;
2452 }
2453
2454 SYSCTL_PROC(_kern, OID_AUTO, max_unsafe_fixed_quanta,
2455 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2456 0, 0, sysctl_kern_max_unsafe_fixed_quanta, "I",
2457 "Number of quanta to allow a fixed sched mode "
2458 "thread to run before being penalized");
2459
2460 static int
2461 sysctl_kern_quantum_us(__unused struct sysctl_oid *oidp, __unused void *arg1,
2462 __unused int arg2, struct sysctl_req *req)
2463 {
2464 const uint64_t quantum_us = sched_get_quantum_us();
2465
2466 return sysctl_io_number(req, quantum_us, sizeof(quantum_us), NULL, NULL);
2467 }
2468
2469 SYSCTL_PROC(_kern, OID_AUTO, quantum_us,
2470 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
2471 0, 0, sysctl_kern_quantum_us, "Q",
2472 "Length of scheduling quantum in microseconds");
2473
2474 extern int smt_sched_bonus_16ths;
2475 SYSCTL_INT(_kern, OID_AUTO, smt_sched_bonus_16ths,
2476 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2477 &smt_sched_bonus_16ths, 0, "");
2478
2479 extern int smt_timeshare_enabled;
2480 SYSCTL_INT(_kern, OID_AUTO, sched_smt_timeshare_enable,
2481 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2482 &smt_timeshare_enabled, 0, "");
2483
2484 extern int sched_smt_balance;
2485 SYSCTL_INT(_kern, OID_AUTO, sched_smt_balance,
2486 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2487 &sched_smt_balance, 0, "");
2488 extern int sched_allow_rt_smt;
2489 SYSCTL_INT(_kern, OID_AUTO, sched_allow_rt_smt,
2490 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2491 &sched_allow_rt_smt, 0, "");
2492 extern int sched_allow_rt_steal;
2493 SYSCTL_INT(_kern, OID_AUTO, sched_allow_rt_steal,
2494 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2495 &sched_allow_rt_steal, 0, "");
2496 extern int sched_backup_cpu_timeout_count;
2497 SYSCTL_INT(_kern, OID_AUTO, sched_backup_cpu_timeout_count,
2498 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2499 &sched_backup_cpu_timeout_count, 0, "The maximum number of 10us delays before allowing a backup cpu to select a thread");
2500 #if __arm64__
2501 /* Scheduler perfcontrol callouts sysctls */
2502 SYSCTL_DECL(_kern_perfcontrol_callout);
2503 SYSCTL_NODE(_kern, OID_AUTO, perfcontrol_callout, CTLFLAG_RW | CTLFLAG_LOCKED, 0,
2504 "scheduler perfcontrol callouts");
2505
2506 extern int perfcontrol_callout_stats_enabled;
2507 SYSCTL_INT(_kern_perfcontrol_callout, OID_AUTO, stats_enabled,
2508 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2509 &perfcontrol_callout_stats_enabled, 0, "");
2510
2511 extern uint64_t perfcontrol_callout_stat_avg(perfcontrol_callout_type_t type,
2512 perfcontrol_callout_stat_t stat);
2513
2514 /* On-Core Callout */
2515 STATIC int
2516 sysctl_perfcontrol_callout_stat
2517 (__unused struct sysctl_oid *oidp, void *arg1, int arg2, struct sysctl_req *req)
2518 {
2519 perfcontrol_callout_stat_t stat = (perfcontrol_callout_stat_t)arg1;
2520 perfcontrol_callout_type_t type = (perfcontrol_callout_type_t)arg2;
2521 return sysctl_io_number(req, (int)perfcontrol_callout_stat_avg(type, stat),
2522 sizeof(int), NULL, NULL);
2523 }
2524
2525 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, oncore_instr,
2526 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2527 (void *)PERFCONTROL_STAT_INSTRS, PERFCONTROL_CALLOUT_ON_CORE,
2528 sysctl_perfcontrol_callout_stat, "I", "");
2529 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, oncore_cycles,
2530 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2531 (void *)PERFCONTROL_STAT_CYCLES, PERFCONTROL_CALLOUT_ON_CORE,
2532 sysctl_perfcontrol_callout_stat, "I", "");
2533 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, offcore_instr,
2534 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2535 (void *)PERFCONTROL_STAT_INSTRS, PERFCONTROL_CALLOUT_OFF_CORE,
2536 sysctl_perfcontrol_callout_stat, "I", "");
2537 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, offcore_cycles,
2538 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2539 (void *)PERFCONTROL_STAT_CYCLES, PERFCONTROL_CALLOUT_OFF_CORE,
2540 sysctl_perfcontrol_callout_stat, "I", "");
2541 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, context_instr,
2542 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2543 (void *)PERFCONTROL_STAT_INSTRS, PERFCONTROL_CALLOUT_CONTEXT,
2544 sysctl_perfcontrol_callout_stat, "I", "");
2545 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, context_cycles,
2546 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2547 (void *)PERFCONTROL_STAT_CYCLES, PERFCONTROL_CALLOUT_CONTEXT,
2548 sysctl_perfcontrol_callout_stat, "I", "");
2549 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, update_instr,
2550 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2551 (void *)PERFCONTROL_STAT_INSTRS, PERFCONTROL_CALLOUT_STATE_UPDATE,
2552 sysctl_perfcontrol_callout_stat, "I", "");
2553 SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, update_cycles,
2554 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
2555 (void *)PERFCONTROL_STAT_CYCLES, PERFCONTROL_CALLOUT_STATE_UPDATE,
2556 sysctl_perfcontrol_callout_stat, "I", "");
2557
2558 #if __AMP__
2559 #if !CONFIG_CLUTCH
2560 extern int sched_amp_idle_steal;
2561 SYSCTL_INT(_kern, OID_AUTO, sched_amp_idle_steal,
2562 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2563 &sched_amp_idle_steal, 0, "");
2564 extern int sched_amp_spill_steal;
2565 SYSCTL_INT(_kern, OID_AUTO, sched_amp_spill_steal,
2566 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2567 &sched_amp_spill_steal, 0, "");
2568 extern int sched_amp_spill_count;
2569 SYSCTL_INT(_kern, OID_AUTO, sched_amp_spill_count,
2570 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2571 &sched_amp_spill_count, 0, "");
2572 #endif /* !CONFIG_CLUTCH */
2573 extern int sched_amp_spill_deferred_ipi;
2574 SYSCTL_INT(_kern, OID_AUTO, sched_amp_spill_deferred_ipi,
2575 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2576 &sched_amp_spill_deferred_ipi, 0, "");
2577 extern int sched_amp_pcores_preempt_immediate_ipi;
2578 SYSCTL_INT(_kern, OID_AUTO, sched_amp_pcores_preempt_immediate_ipi,
2579 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2580 &sched_amp_pcores_preempt_immediate_ipi, 0, "");
2581 #endif /* __AMP__ */
2582 #endif /* __arm64__ */
2583
2584 #if __arm64__
2585 extern int legacy_footprint_entitlement_mode;
2586 SYSCTL_INT(_kern, OID_AUTO, legacy_footprint_entitlement_mode,
2587 CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
2588 &legacy_footprint_entitlement_mode, 0, "");
2589 #endif /* __arm64__ */
2590
2591 /*
2592 * Realtime threads are ordered by highest priority first then,
2593 * for threads of the same priority, by earliest deadline first.
2594 * But if sched_rt_runq_strict_priority is false (the default),
2595 * a lower priority thread with an earlier deadline will be preferred
2596 * over a higher priority thread with a later deadline, as long as
2597 * both threads' computations will fit before the later deadline.
2598 */
2599 extern int sched_rt_runq_strict_priority;
2600 SYSCTL_INT(_kern, OID_AUTO, sched_rt_runq_strict_priority,
2601 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2602 &sched_rt_runq_strict_priority, 0, "");
2603
2604 static int
2605 sysctl_kern_sched_rt_n_backup_processors(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2606 {
2607 int new_value, changed;
2608 int old_value = sched_get_rt_n_backup_processors();
2609 int error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
2610 if (changed) {
2611 sched_set_rt_n_backup_processors(new_value);
2612 }
2613
2614 return error;
2615 }
2616
2617 SYSCTL_PROC(_kern, OID_AUTO, sched_rt_n_backup_processors,
2618 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2619 0, 0, sysctl_kern_sched_rt_n_backup_processors, "I", "");
2620
2621 static int
2622 sysctl_kern_sched_rt_deadline_epsilon_us(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2623 {
2624 int new_value, changed;
2625 int old_value = sched_get_rt_deadline_epsilon();
2626 int error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
2627 if (changed) {
2628 sched_set_rt_deadline_epsilon(new_value);
2629 }
2630
2631 return error;
2632 }
2633
2634 SYSCTL_PROC(_kern, OID_AUTO, sched_rt_deadline_epsilon_us,
2635 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2636 0, 0, sysctl_kern_sched_rt_deadline_epsilon_us, "I", "");
2637
2638 extern int sched_idle_delay_cpuid;
2639 SYSCTL_INT(_kern, OID_AUTO, sched_idle_delay_cpuid,
2640 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
2641 &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");
2642
2643 static int
2644 sysctl_kern_sched_powered_cores(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2645 {
2646 int new_value, changed;
2647 int old_value = sched_get_powered_cores();
2648 int error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
2649 if (changed) {
2650 if (!PE_parse_boot_argn("enable_skstb", NULL, 0)) {
2651 return ENOTSUP;
2652 }
2653
2654 sched_set_powered_cores(new_value);
2655 }
2656
2657 return error;
2658 }
2659
2660 SYSCTL_PROC(_kern, OID_AUTO, sched_powered_cores,
2661 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2662 0, 0, sysctl_kern_sched_powered_cores, "I", "");
2663
2664 #if __arm64__
2665
2666 static int
2667 sysctl_kern_update_sched_recommended_cores(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2668 {
2669 uint64_t new_value;
2670 int changed;
2671 uint64_t old_value = sched_sysctl_get_recommended_cores();
2672 int error = sysctl_io_number(req, old_value, sizeof(uint64_t), &new_value, &changed);
2673 if (changed) {
2674 if (!PE_parse_boot_argn("enable_skstb", NULL, 0)) {
2675 return ENOTSUP;
2676 }
2677
2678 sched_perfcontrol_update_recommended_cores_reason(new_value, REASON_CLPC_USER, 0);
2679 }
2680
2681 return error;
2682 }
2683
2684 SYSCTL_PROC(_kern, OID_AUTO, sched_update_recommended_cores,
2685 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
2686 0, 0, sysctl_kern_update_sched_recommended_cores, "I", "");
2687
2688 #endif /* __arm64__ */
2689
2690 #endif /* (DEVELOPMENT || DEBUG) */
2691
2692 extern uint64_t sysctl_sched_recommended_cores;
2693 SYSCTL_QUAD(_kern, OID_AUTO, sched_recommended_cores,
2694 CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
2695 &sysctl_sched_recommended_cores, "");
2696
2697 static int
2698 sysctl_kern_suspend_cluster_powerdown(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2699 {
2700 int new_value, changed;
2701 int old_value = get_cluster_powerdown_user_suspended();
2702 int error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
2703 if (!error && changed) {
2704 if (new_value > 0) {
2705 error = suspend_cluster_powerdown_from_user();
2706 } else {
2707 error = resume_cluster_powerdown_from_user();
2708 }
2709 if (error) {
2710 error = EALREADY;
2711 }
2712 }
2713
2714 return error;
2715 }
2716
2717 SYSCTL_PROC(_kern, OID_AUTO, suspend_cluster_powerdown,
2718 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2719 0, 0, sysctl_kern_suspend_cluster_powerdown, "I", "");
2720
2721
2722 STATIC int
2723 sysctl_securelvl
2724 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2725 {
2726 int new_value, changed;
2727 int error = sysctl_io_number(req, securelevel, sizeof(int), &new_value, &changed);
2728 if (changed) {
2729 if (!(new_value < securelevel && proc_getpid(req->p) != 1)) {
2730 proc_list_lock();
2731 securelevel = new_value;
2732 proc_list_unlock();
2733 } else {
2734 error = EPERM;
2735 }
2736 }
2737 return error;
2738 }
2739
2740 SYSCTL_PROC(_kern, KERN_SECURELVL, securelevel,
2741 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
2742 0, 0, sysctl_securelvl, "I", "");
2743
2744
2745 STATIC int
2746 sysctl_domainname
2747 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2748 {
2749 int error, changed;
2750 char tmpname[MAXHOSTNAMELEN] = {};
2751
2752 lck_mtx_lock(&domainname_lock);
2753 strlcpy(tmpname, domainname, sizeof(tmpname));
2754 lck_mtx_unlock(&domainname_lock);
2755
2756 error = sysctl_io_string(req, tmpname, sizeof(tmpname), 0, &changed);
2757 if (!error && changed) {
2758 lck_mtx_lock(&domainname_lock);
2759 strlcpy(domainname, tmpname, sizeof(domainname));
2760 lck_mtx_unlock(&domainname_lock);
2761 }
2762 return error;
2763 }
2764
2765 SYSCTL_PROC(_kern, KERN_DOMAINNAME, nisdomainname,
2766 CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_LOCKED,
2767 0, 0, sysctl_domainname, "A", "");
2768
2769 SYSCTL_COMPAT_INT(_kern, KERN_HOSTID, hostid,
2770 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2771 &hostid, 0, "");
2772
2773 STATIC int
2774 sysctl_hostname
2775 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2776 {
2777 int error, changed;
2778 char tmpname[MAXHOSTNAMELEN] = {};
2779 const char * name;
2780
2781 #if XNU_TARGET_OS_OSX
2782 name = hostname;
2783 #else /* XNU_TARGET_OS_OSX */
2784 #define ENTITLEMENT_USER_ASSIGNED_DEVICE_NAME \
2785 "com.apple.developer.device-information.user-assigned-device-name"
2786 if (csproc_get_platform_binary(current_proc()) ||
2787 IOCurrentTaskHasEntitlement(ENTITLEMENT_USER_ASSIGNED_DEVICE_NAME)) {
2788 name = hostname;
2789 } else {
2790 /* Deny writes if we don't pass entitlement check */
2791 if (req->newptr) {
2792 return EPERM;
2793 }
2794
2795 name = "localhost";
2796 }
2797 #endif /* ! XNU_TARGET_OS_OSX */
2798
2799 lck_mtx_lock(&hostname_lock);
2800 strlcpy(tmpname, name, sizeof(tmpname));
2801 lck_mtx_unlock(&hostname_lock);
2802
2803 error = sysctl_io_string(req, tmpname, sizeof(tmpname), 1, &changed);
2804 if (!error && changed) {
2805 lck_mtx_lock(&hostname_lock);
2806 strlcpy(hostname, tmpname, sizeof(hostname));
2807 lck_mtx_unlock(&hostname_lock);
2808 }
2809 return error;
2810 }
2811
2812 SYSCTL_PROC(_kern, KERN_HOSTNAME, hostname,
2813 CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_LOCKED,
2814 0, 0, sysctl_hostname, "A", "");
2815
2816 STATIC int
2817 sysctl_procname
2818 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2819 {
2820 /* Original code allowed writing, I'm copying this, although this all makes
2821 * no sense to me. Besides, this sysctl is never used. */
2822 return sysctl_io_string(req, &req->p->p_name[0], (2 * MAXCOMLEN + 1), 1, NULL);
2823 }
2824
2825 SYSCTL_PROC(_kern, KERN_PROCNAME, procname,
2826 CTLTYPE_STRING | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
2827 0, 0, sysctl_procname, "A", "");
2828
2829 SYSCTL_INT(_kern, KERN_SPECULATIVE_READS, speculative_reads_disabled,
2830 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2831 &speculative_reads_disabled, 0, "");
2832
2833 SYSCTL_UINT(_kern, OID_AUTO, preheat_max_bytes,
2834 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2835 &preheat_max_bytes, 0, "");
2836
2837 SYSCTL_UINT(_kern, OID_AUTO, preheat_min_bytes,
2838 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2839 &preheat_min_bytes, 0, "");
2840
2841 SYSCTL_UINT(_kern, OID_AUTO, speculative_prefetch_max,
2842 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2843 &speculative_prefetch_max, 0, "");
2844
2845 SYSCTL_UINT(_kern, OID_AUTO, speculative_prefetch_max_iosize,
2846 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2847 &speculative_prefetch_max_iosize, 0, "");
2848
2849 SYSCTL_UINT(_kern, OID_AUTO, vm_page_free_target,
2850 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2851 &vm_page_free_target, 0, "");
2852
2853 SYSCTL_UINT(_kern, OID_AUTO, vm_page_free_min,
2854 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2855 &vm_page_free_min, 0, "");
2856
2857 SYSCTL_UINT(_kern, OID_AUTO, vm_page_free_reserved,
2858 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2859 &vm_page_free_reserved, 0, "");
2860
2861 SYSCTL_UINT(_kern, OID_AUTO, vm_page_speculative_percentage,
2862 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2863 &vm_pageout_state.vm_page_speculative_percentage, 0, "");
2864
2865 SYSCTL_UINT(_kern, OID_AUTO, vm_page_speculative_q_age_ms,
2866 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2867 &vm_pageout_state.vm_page_speculative_q_age_ms, 0, "");
2868
2869 #if (DEVELOPMENT || DEBUG)
2870 SYSCTL_UINT(_kern, OID_AUTO, vm_page_max_speculative_age_q,
2871 CTLFLAG_RD,
2872 &vm_page_max_speculative_age_q, 0, "");
2873 #endif /* (DEVELOPMENT || DEBUG) */
2874
2875 SYSCTL_UINT(_kern, OID_AUTO, vm_max_delayed_work_limit,
2876 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2877 &vm_max_delayed_work_limit, 0, "");
2878
2879 SYSCTL_UINT(_kern, OID_AUTO, vm_max_batch,
2880 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
2881 &vm_max_batch, 0, "");
2882
2883 SYSCTL_STRING(_kern, OID_AUTO, bootsessionuuid,
2884 CTLFLAG_RD | CTLFLAG_LOCKED,
2885 &bootsessionuuid_string, sizeof(bootsessionuuid_string), "");
2886
2887
2888 STATIC int
2889 sysctl_boottime
2890 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2891 {
2892 struct timeval tv;
2893 boottime_timeval(&tv);
2894 struct proc *p = req->p;
2895
2896 if (proc_is64bit(p)) {
2897 struct user64_timeval t = {};
2898 t.tv_sec = tv.tv_sec;
2899 t.tv_usec = tv.tv_usec;
2900 return sysctl_io_opaque(req, &t, sizeof(t), NULL);
2901 } else {
2902 struct user32_timeval t = {};
2903 t.tv_sec = (user32_time_t)tv.tv_sec;
2904 t.tv_usec = tv.tv_usec;
2905 return sysctl_io_opaque(req, &t, sizeof(t), NULL);
2906 }
2907 }
2908
2909 SYSCTL_PROC(_kern, KERN_BOOTTIME, boottime,
2910 CTLTYPE_STRUCT | CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
2911 0, 0, sysctl_boottime, "S,timeval", "");
2912
2913 extern bool IOGetBootUUID(char *);
2914
2915 /* non-static: written by imageboot.c */
2916 uuid_string_t fake_bootuuid;
2917
2918 STATIC int
2919 sysctl_bootuuid
2920 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2921 {
2922 int error = ENOENT;
2923
2924 /* check the first byte to see if the string has been
2925 * populated. this is a uuid_STRING_t, this check would
2926 * not work with a uuid_t.
2927 */
2928 if (fake_bootuuid[0] != '\0') {
2929 error = sysctl_io_string(req, fake_bootuuid, 0, 0, NULL);
2930 goto out;
2931 }
2932
2933 uuid_string_t uuid_string;
2934 if (IOGetBootUUID(uuid_string)) {
2935 uuid_t boot_uuid;
2936 error = uuid_parse(uuid_string, boot_uuid);
2937 if (!error) {
2938 error = sysctl_io_string(req, __DECONST(char *, uuid_string), 0, 0, NULL);
2939 }
2940 }
2941
2942 out:
2943 return error;
2944 }
2945
2946 SYSCTL_PROC(_kern, OID_AUTO, bootuuid,
2947 CTLTYPE_STRING | CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
2948 0, 0, sysctl_bootuuid, "A", "");
2949
2950
2951 extern bool IOGetApfsPrebootUUID(char *);
2952 extern bool IOGetAssociatedApfsVolgroupUUID(char *);
2953
2954 STATIC int
2955 sysctl_apfsprebootuuid
2956 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2957 {
2958 int error = ENOENT;
2959
2960 uuid_string_t uuid_string;
2961 if (IOGetApfsPrebootUUID(uuid_string)) {
2962 uuid_t apfs_preboot_uuid;
2963 error = uuid_parse(uuid_string, apfs_preboot_uuid);
2964 if (!error) {
2965 error = sysctl_io_string(req, __DECONST(char *, uuid_string), 0, 0, NULL);
2966 }
2967 }
2968
2969 return error;
2970 }
2971
2972 SYSCTL_PROC(_kern, OID_AUTO, apfsprebootuuid,
2973 CTLTYPE_STRING | CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
2974 0, 0, sysctl_apfsprebootuuid, "A", "");
2975
2976 STATIC int
2977 sysctl_targetsystemvolgroupuuid
2978 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
2979 {
2980 int error = ENOENT;
2981
2982 uuid_string_t uuid_string;
2983 if (IOGetApfsPrebootUUID(uuid_string)) {
2984 uuid_t apfs_preboot_uuid;
2985 error = uuid_parse(uuid_string, apfs_preboot_uuid);
2986 if (!error) {
2987 error = sysctl_io_string(req, __DECONST(char *, uuid_string), 0, 0, NULL);
2988 }
2989 } else {
2990 /*
2991 * In special boot modes, such as kcgen-mode, the
2992 * apfs-preboot-uuid property will not be set. Instead, a
2993 * different property, associated-volume-group, will be set
2994 * which indicates the UUID of the VolumeGroup containing the
2995 * system volume into which you will boot.
2996 */
2997 if (IOGetAssociatedApfsVolgroupUUID(uuid_string)) {
2998 uuid_t apfs_preboot_uuid;
2999 error = uuid_parse(uuid_string, apfs_preboot_uuid);
3000 if (!error) {
3001 error = sysctl_io_string(req, __DECONST(char *, uuid_string), 0, 0, NULL);
3002 }
3003 }
3004 }
3005
3006 return error;
3007 }
3008
3009 SYSCTL_PROC(_kern, OID_AUTO, targetsystemvolgroupuuid,
3010 CTLTYPE_STRING | CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_MASKED | CTLFLAG_LOCKED,
3011 0, 0, sysctl_targetsystemvolgroupuuid, "A", "");
3012
3013
3014 extern bool IOGetBootManifestHash(char *, size_t *);
3015 extern bool IOGetBootObjectsPath(char *);
3016
3017 STATIC int
3018 sysctl_bootobjectspath
3019 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3020 {
3021 int error = ENOENT;
3022
3023 #if defined(__x86_64__)
3024 /* auth-root-dmg is used for the Intel BaseSystem in some flows,
3025 * e.g. createinstallmedia and as part of upgrading from 10.15 or earlier
3026 * under these scenarios, set_fake_bootuuid will be called when pivoting to
3027 * the new root filesystem. need honor the fake bootuuid.
3028 */
3029 if (fake_bootuuid[0] != '\0') {
3030 error = sysctl_io_string(req, fake_bootuuid, 0, 0, NULL);
3031 } else {
3032 /* for intel mac, boot objects reside in [preboot volume]/[bootuuid]
3033 * bootuuid and apfsprebootuuid are populated by efiboot and they are alias.
3034 */
3035 uuid_string_t uuid_string;
3036 if (IOGetBootUUID(uuid_string)) {
3037 uuid_t boot_uuid;
3038 error = uuid_parse(uuid_string, boot_uuid);
3039 if (!error) {
3040 error = sysctl_io_string(req, (char *)uuid_string, 0, 0, NULL);
3041 }
3042 }
3043 }
3044 #else
3045 char boot_obj_path[MAXPATHLEN] = { "\0" };
3046 static const char kAsciiHexChars[] = "0123456789ABCDEF";
3047 unsigned int i, j;
3048
3049 /* Hashed with SHA2-384 or SHA1, boot manifest hash is 48 bytes or 20 bytes
3050 * hence, need a 97 bytes char array for the string.
3051 */
3052 size_t hash_data_size = CCSHA384_OUTPUT_SIZE;
3053 char hash_data[CCSHA384_OUTPUT_SIZE] = { "\0" };
3054 char boot_manifest_hash[CCSHA384_OUTPUT_SIZE * 2 + 1] = { "\0" };;
3055
3056 /* for Apple Silicon Macs, there is a boot-objects-path under IODeviceTree:/chosen
3057 * and boot objects reside in [preboot volume]/[boot-objects-path]
3058 * for embedded platforms, there would be a boot-manifest-hash under IODeviceTree:/chosen
3059 * and boot objects reside in [preboot volume]/[boot-manifest-hash]
3060 */
3061 if (IOGetBootObjectsPath(boot_obj_path)) {
3062 error = sysctl_io_string(req, (char *)boot_obj_path, 0, 0, NULL);
3063 } else if (IOGetBootManifestHash(hash_data, &hash_data_size)) {
3064 j = 0;
3065 for (i = 0; i < hash_data_size; ++i) {
3066 char octet = hash_data[i];
3067 boot_manifest_hash[j++] = kAsciiHexChars[((octet & 0xF0) >> 4)];
3068 boot_manifest_hash[j++] = kAsciiHexChars[(octet & 0x0F)];
3069 }
3070 /* make sure string has null termination */
3071 boot_manifest_hash[j] = '\0';
3072 error = sysctl_io_string(req, (char *)boot_manifest_hash, 0, 0, NULL);
3073 }
3074 #endif
3075 return error;
3076 }
3077
3078 SYSCTL_PROC(_kern, OID_AUTO, bootobjectspath,
3079 CTLTYPE_STRING | CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
3080 0, 0, sysctl_bootobjectspath, "A", "");
3081
3082
3083 STATIC int
3084 sysctl_symfile
3085 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3086 {
3087 char *str;
3088 int error = get_kernel_symfile(req->p, &str);
3089 if (error) {
3090 return error;
3091 }
3092 return sysctl_io_string(req, str, 0, 0, NULL);
3093 }
3094
3095
3096 SYSCTL_PROC(_kern, KERN_SYMFILE, symfile,
3097 CTLTYPE_STRING | CTLFLAG_RD | CTLFLAG_LOCKED,
3098 0, 0, sysctl_symfile, "A", "");
3099
3100 #if CONFIG_NETBOOT
3101 STATIC int
3102 sysctl_netboot
3103 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3104 {
3105 return sysctl_io_number(req, netboot_root(), sizeof(int), NULL, NULL);
3106 }
3107
3108 SYSCTL_PROC(_kern, KERN_NETBOOT, netboot,
3109 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
3110 0, 0, sysctl_netboot, "I", "");
3111 #endif
3112
3113 #ifdef CONFIG_IMGSRC_ACCESS
3114 /*
3115 * Legacy--act as if only one layer of nesting is possible.
3116 */
3117 STATIC int
3118 sysctl_imgsrcdev
3119 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3120 {
3121 vfs_context_t ctx = vfs_context_current();
3122 vnode_t devvp;
3123 int result;
3124
3125 if (!vfs_context_issuser(ctx)) {
3126 return EPERM;
3127 }
3128
3129 if (imgsrc_rootvnodes[0] == NULL) {
3130 return ENOENT;
3131 }
3132
3133 result = vnode_getwithref(imgsrc_rootvnodes[0]);
3134 if (result != 0) {
3135 return result;
3136 }
3137
3138 devvp = vnode_mount(imgsrc_rootvnodes[0])->mnt_devvp;
3139 result = vnode_getwithref(devvp);
3140 if (result != 0) {
3141 goto out;
3142 }
3143
3144 result = sysctl_io_number(req, vnode_specrdev(devvp), sizeof(dev_t), NULL, NULL);
3145
3146 vnode_put(devvp);
3147 out:
3148 vnode_put(imgsrc_rootvnodes[0]);
3149 return result;
3150 }
3151
3152 SYSCTL_PROC(_kern, OID_AUTO, imgsrcdev,
3153 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
3154 0, 0, sysctl_imgsrcdev, "I", "");
3155
3156 STATIC int
3157 sysctl_imgsrcinfo
3158 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3159 {
3160 int error;
3161 struct imgsrc_info info[MAX_IMAGEBOOT_NESTING] = {}; /* 2 for now, no problem */
3162 uint32_t i;
3163 vnode_t rvp, devvp;
3164
3165 if (imgsrc_rootvnodes[0] == NULLVP) {
3166 return ENXIO;
3167 }
3168
3169 for (i = 0; i < MAX_IMAGEBOOT_NESTING; i++) {
3170 /*
3171 * Go get the root vnode.
3172 */
3173 rvp = imgsrc_rootvnodes[i];
3174 if (rvp == NULLVP) {
3175 break;
3176 }
3177
3178 error = vnode_get(rvp);
3179 if (error != 0) {
3180 return error;
3181 }
3182
3183 /*
3184 * For now, no getting at a non-local volume.
3185 */
3186 devvp = vnode_mount(rvp)->mnt_devvp;
3187 if (devvp == NULL) {
3188 vnode_put(rvp);
3189 return EINVAL;
3190 }
3191
3192 error = vnode_getwithref(devvp);
3193 if (error != 0) {
3194 vnode_put(rvp);
3195 return error;
3196 }
3197
3198 /*
3199 * Fill in info.
3200 */
3201 info[i].ii_dev = vnode_specrdev(devvp);
3202 info[i].ii_flags = 0;
3203 info[i].ii_height = i;
3204 bzero(info[i].ii_reserved, sizeof(info[i].ii_reserved));
3205
3206 vnode_put(devvp);
3207 vnode_put(rvp);
3208 }
3209
3210 return sysctl_io_opaque(req, info, i * sizeof(info[0]), NULL);
3211 }
3212
3213 SYSCTL_PROC(_kern, OID_AUTO, imgsrcinfo,
3214 CTLTYPE_OPAQUE | CTLFLAG_RD | CTLFLAG_LOCKED,
3215 0, 0, sysctl_imgsrcinfo, "I", "");
3216
3217 #endif /* CONFIG_IMGSRC_ACCESS */
3218
3219
3220 SYSCTL_DECL(_kern_timer);
3221 SYSCTL_NODE(_kern, OID_AUTO, timer, CTLFLAG_RW | CTLFLAG_LOCKED, 0, "timer");
3222
3223
3224 SYSCTL_INT(_kern_timer, OID_AUTO, coalescing_enabled,
3225 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
3226 &mach_timer_coalescing_enabled, 0, "");
3227
3228 SYSCTL_QUAD(_kern_timer, OID_AUTO, deadline_tracking_bin_1,
3229 CTLFLAG_RW | CTLFLAG_LOCKED,
3230 &timer_deadline_tracking_bin_1, "");
3231 SYSCTL_QUAD(_kern_timer, OID_AUTO, deadline_tracking_bin_2,
3232 CTLFLAG_RW | CTLFLAG_LOCKED,
3233 &timer_deadline_tracking_bin_2, "");
3234
3235 SYSCTL_DECL(_kern_timer_longterm);
3236 SYSCTL_NODE(_kern_timer, OID_AUTO, longterm, CTLFLAG_RW | CTLFLAG_LOCKED, 0, "longterm");
3237
3238
3239 /* Must match definition in osfmk/kern/timer_call.c */
3240 enum {
3241 THRESHOLD, QCOUNT,
3242 ENQUEUES, DEQUEUES, ESCALATES, SCANS, PREEMPTS,
3243 LATENCY, LATENCY_MIN, LATENCY_MAX, LONG_TERM_SCAN_LIMIT,
3244 LONG_TERM_SCAN_INTERVAL, LONG_TERM_SCAN_PAUSES,
3245 SCAN_LIMIT, SCAN_INTERVAL, SCAN_PAUSES, SCAN_POSTPONES,
3246 };
3247 extern uint64_t timer_sysctl_get(int);
3248 extern kern_return_t timer_sysctl_set(int, uint64_t);
3249
3250 STATIC int
3251 sysctl_timer
3252 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3253 {
3254 int oid = (int)arg1;
3255 uint64_t value = timer_sysctl_get(oid);
3256 uint64_t new_value;
3257 int error;
3258 int changed;
3259
3260 error = sysctl_io_number(req, value, sizeof(value), &new_value, &changed);
3261 if (changed) {
3262 kern_return_t kr = timer_sysctl_set(oid, new_value);
3263 error = mach_to_bsd_errno(kr);
3264 }
3265
3266 return error;
3267 }
3268
3269 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, threshold,
3270 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
3271 (void *) THRESHOLD, 0, sysctl_timer, "Q", "");
3272 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, scan_limit,
3273 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
3274 (void *) LONG_TERM_SCAN_LIMIT, 0, sysctl_timer, "Q", "");
3275 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, scan_interval,
3276 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
3277 (void *) LONG_TERM_SCAN_INTERVAL, 0, sysctl_timer, "Q", "");
3278
3279 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, qlen,
3280 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3281 (void *) QCOUNT, 0, sysctl_timer, "Q", "");
3282 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, scan_pauses,
3283 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3284 (void *) LONG_TERM_SCAN_PAUSES, 0, sysctl_timer, "Q", "");
3285
3286 #if DEBUG
3287 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, enqueues,
3288 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3289 (void *) ENQUEUES, 0, sysctl_timer, "Q", "");
3290 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, dequeues,
3291 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3292 (void *) DEQUEUES, 0, sysctl_timer, "Q", "");
3293 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, escalates,
3294 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3295 (void *) ESCALATES, 0, sysctl_timer, "Q", "");
3296 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, scans,
3297 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3298 (void *) SCANS, 0, sysctl_timer, "Q", "");
3299 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, preempts,
3300 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3301 (void *) PREEMPTS, 0, sysctl_timer, "Q", "");
3302 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, latency,
3303 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3304 (void *) LATENCY, 0, sysctl_timer, "Q", "");
3305 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, latency_min,
3306 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3307 (void *) LATENCY_MIN, 0, sysctl_timer, "Q", "");
3308 SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, latency_max,
3309 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3310 (void *) LATENCY_MAX, 0, sysctl_timer, "Q", "");
3311 #endif /* DEBUG */
3312
3313 SYSCTL_PROC(_kern_timer, OID_AUTO, scan_limit,
3314 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
3315 (void *) SCAN_LIMIT, 0, sysctl_timer, "Q", "");
3316 SYSCTL_PROC(_kern_timer, OID_AUTO, scan_interval,
3317 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
3318 (void *) SCAN_INTERVAL, 0, sysctl_timer, "Q", "");
3319 SYSCTL_PROC(_kern_timer, OID_AUTO, scan_pauses,
3320 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3321 (void *) SCAN_PAUSES, 0, sysctl_timer, "Q", "");
3322 SYSCTL_PROC(_kern_timer, OID_AUTO, scan_postpones,
3323 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3324 (void *) SCAN_POSTPONES, 0, sysctl_timer, "Q", "");
3325
3326 STATIC int
3327 sysctl_usrstack
3328 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3329 {
3330 return sysctl_io_number(req, (int)req->p->user_stack, sizeof(int), NULL, NULL);
3331 }
3332
3333 SYSCTL_PROC(_kern, KERN_USRSTACK32, usrstack,
3334 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
3335 0, 0, sysctl_usrstack, "I", "");
3336
3337 STATIC int
3338 sysctl_usrstack64
3339 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3340 {
3341 return sysctl_io_number(req, req->p->user_stack, sizeof(req->p->user_stack), NULL, NULL);
3342 }
3343
3344 SYSCTL_PROC(_kern, KERN_USRSTACK64, usrstack64,
3345 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED,
3346 0, 0, sysctl_usrstack64, "Q", "");
3347
3348
3349 #if EXCLAVES_COREDUMP
3350
3351 /* secure kernel coredump support. */
3352 extern unsigned int sc_dump_mode;
3353 SYSCTL_UINT(_kern, OID_AUTO, secure_coredump, CTLFLAG_RD, &sc_dump_mode, 0, "secure_coredump");
3354
3355 #endif /* EXCLAVES_COREDUMP */
3356
3357
3358 #if CONFIG_COREDUMP
3359
3360 SYSCTL_STRING(_kern, KERN_COREFILE, corefile,
3361 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
3362 corefilename, sizeof(corefilename), "");
3363
3364 SYSCTL_STRING(_kern, OID_AUTO, drivercorefile,
3365 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
3366 drivercorefilename, sizeof(drivercorefilename), "");
3367
3368 STATIC int
3369 sysctl_coredump
3370 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3371 {
3372 #ifdef SECURE_KERNEL
3373 (void)req;
3374 return ENOTSUP;
3375 #else
3376 int new_value, changed;
3377 int error = sysctl_io_number(req, do_coredump, sizeof(int), &new_value, &changed);
3378 if (changed) {
3379 if ((new_value == 0) || (new_value == 1)) {
3380 do_coredump = new_value;
3381 } else {
3382 error = EINVAL;
3383 }
3384 }
3385 return error;
3386 #endif
3387 }
3388
3389 SYSCTL_PROC(_kern, KERN_COREDUMP, coredump,
3390 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
3391 0, 0, sysctl_coredump, "I", "");
3392
3393 STATIC int
3394 sysctl_suid_coredump
3395 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3396 {
3397 #ifdef SECURE_KERNEL
3398 (void)req;
3399 return ENOTSUP;
3400 #else
3401 int new_value, changed;
3402 int error = sysctl_io_number(req, sugid_coredump, sizeof(int), &new_value, &changed);
3403 if (changed) {
3404 if ((new_value == 0) || (new_value == 1)) {
3405 sugid_coredump = new_value;
3406 } else {
3407 error = EINVAL;
3408 }
3409 }
3410 return error;
3411 #endif
3412 }
3413
3414 SYSCTL_PROC(_kern, KERN_SUGID_COREDUMP, sugid_coredump,
3415 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
3416 0, 0, sysctl_suid_coredump, "I", "");
3417
3418 #endif /* CONFIG_COREDUMP */
3419
3420 #if CONFIG_KDP_INTERACTIVE_DEBUGGING
3421
3422 extern const char* kdp_corefile_path;
3423 STATIC int
3424 sysctl_kdp_corefile(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3425 {
3426 return SYSCTL_OUT(req, kdp_corefile_path, strlen(kdp_corefile_path) + 1);
3427 }
3428
3429 /* this needs to be a proc rather than a string since kdp_corefile_path is not a compile-time constant */
3430 SYSCTL_PROC(_kern, OID_AUTO, kdp_corefile,
3431 CTLTYPE_STRING | CTLFLAG_RD | CTLFLAG_LOCKED,
3432 0, 0, sysctl_kdp_corefile, "A", "");
3433
3434 #endif /* CONFIG_KDP_INTERACTIVE_DEBUGGING */
3435
3436 STATIC int
3437 sysctl_delayterm
3438 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3439 {
3440 struct proc *p = req->p;
3441 int new_value, changed;
3442 int error = sysctl_io_number(req, (req->p->p_lflag & P_LDELAYTERM)? 1: 0, sizeof(int), &new_value, &changed);
3443 if (changed) {
3444 proc_lock(p);
3445 if (new_value) {
3446 req->p->p_lflag |= P_LDELAYTERM;
3447 } else {
3448 req->p->p_lflag &= ~P_LDELAYTERM;
3449 }
3450 proc_unlock(p);
3451 }
3452 return error;
3453 }
3454
3455 SYSCTL_PROC(_kern, KERN_PROCDELAYTERM, delayterm,
3456 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
3457 0, 0, sysctl_delayterm, "I", "");
3458
3459
3460 STATIC int
3461 sysctl_rage_vnode
3462 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3463 {
3464 struct proc *p = req->p;
3465 struct uthread *ut;
3466 int new_value, old_value, changed;
3467 int error;
3468
3469 ut = current_uthread();
3470
3471 if (ut->uu_flag & UT_RAGE_VNODES) {
3472 old_value = KERN_RAGE_THREAD;
3473 } else if (p->p_lflag & P_LRAGE_VNODES) {
3474 old_value = KERN_RAGE_PROC;
3475 } else {
3476 old_value = 0;
3477 }
3478
3479 error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
3480
3481 if ((error == 0) && (changed != 0)) {
3482 switch (new_value) {
3483 case KERN_RAGE_PROC:
3484 proc_lock(p);
3485 p->p_lflag |= P_LRAGE_VNODES;
3486 proc_unlock(p);
3487 break;
3488 case KERN_UNRAGE_PROC:
3489 proc_lock(p);
3490 p->p_lflag &= ~P_LRAGE_VNODES;
3491 proc_unlock(p);
3492 break;
3493
3494 case KERN_RAGE_THREAD:
3495 ut->uu_flag |= UT_RAGE_VNODES;
3496 break;
3497 case KERN_UNRAGE_THREAD:
3498 ut = current_uthread();
3499 ut->uu_flag &= ~UT_RAGE_VNODES;
3500 break;
3501 }
3502 }
3503 return error;
3504 }
3505
3506 SYSCTL_PROC(_kern, KERN_RAGEVNODE, rage_vnode,
3507 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
3508 0, 0, sysctl_rage_vnode, "I", "");
3509
3510 /* XXX until filecoordinationd fixes a bit of inverted logic. */
3511 STATIC int
3512 sysctl_vfsnspace
3513 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3514 {
3515 int old_value = 0, new_value, changed;
3516
3517 return sysctl_io_number(req, old_value, sizeof(int), &new_value,
3518 &changed);
3519 }
3520
3521 SYSCTL_PROC(_kern, OID_AUTO, vfsnspace,
3522 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
3523 0, 0, sysctl_vfsnspace, "I", "");
3524
3525 /* XXX move this interface into libproc and remove this sysctl */
3526 STATIC int
3527 sysctl_setthread_cpupercent
3528 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3529 {
3530 int new_value, old_value;
3531 int error = 0;
3532 kern_return_t kret = KERN_SUCCESS;
3533 uint8_t percent = 0;
3534 int ms_refill = 0;
3535
3536 if (!req->newptr) {
3537 return 0;
3538 }
3539
3540 old_value = 0;
3541
3542 if ((error = sysctl_io_number(req, old_value, sizeof(old_value), &new_value, NULL)) != 0) {
3543 return error;
3544 }
3545
3546 percent = new_value & 0xff; /* low 8 bytes for perent */
3547 ms_refill = (new_value >> 8) & 0xffffff; /* upper 24bytes represent ms refill value */
3548 if (percent > 100) {
3549 return EINVAL;
3550 }
3551
3552 /*
3553 * If the caller is specifying a percentage of 0, this will unset the CPU limit, if present.
3554 */
3555 kret = percent == 0 ?
3556 thread_set_cpulimit(THREAD_CPULIMIT_DISABLE, 0, 0) :
3557 thread_set_cpulimit(THREAD_CPULIMIT_BLOCK, percent, ms_refill * (int)NSEC_PER_MSEC);
3558
3559 if (kret != 0) {
3560 return EIO;
3561 }
3562
3563 return 0;
3564 }
3565
3566 SYSCTL_PROC(_kern, OID_AUTO, setthread_cpupercent,
3567 CTLTYPE_INT | CTLFLAG_WR | CTLFLAG_ANYBODY,
3568 0, 0, sysctl_setthread_cpupercent, "I", "set thread cpu percentage limit");
3569
3570
3571 STATIC int
3572 sysctl_kern_check_openevt
3573 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3574 {
3575 struct proc *p = req->p;
3576 int new_value, old_value, changed;
3577 int error;
3578
3579 if (p->p_flag & P_CHECKOPENEVT) {
3580 old_value = KERN_OPENEVT_PROC;
3581 } else {
3582 old_value = 0;
3583 }
3584
3585 error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
3586
3587 if ((error == 0) && (changed != 0)) {
3588 switch (new_value) {
3589 case KERN_OPENEVT_PROC:
3590 OSBitOrAtomic(P_CHECKOPENEVT, &p->p_flag);
3591 break;
3592
3593 case KERN_UNOPENEVT_PROC:
3594 OSBitAndAtomic(~((uint32_t)P_CHECKOPENEVT), &p->p_flag);
3595 break;
3596
3597 default:
3598 error = EINVAL;
3599 }
3600 }
3601 return error;
3602 }
3603
3604 SYSCTL_PROC(_kern, KERN_CHECKOPENEVT, check_openevt, CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED,
3605 0, 0, sysctl_kern_check_openevt, "I", "set the per-process check-open-evt flag");
3606
3607
3608 #if DEVELOPMENT || DEBUG
3609 STATIC int
3610 sysctl_nx
3611 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3612 {
3613 #ifdef SECURE_KERNEL
3614 (void)req;
3615 return ENOTSUP;
3616 #else
3617 int new_value, changed;
3618 int error;
3619
3620 error = sysctl_io_number(req, nx_enabled, sizeof(nx_enabled), &new_value, &changed);
3621 if (error) {
3622 return error;
3623 }
3624
3625 if (changed) {
3626 #if defined(__x86_64__)
3627 /*
3628 * Only allow setting if NX is supported on the chip
3629 */
3630 if (!(cpuid_extfeatures() & CPUID_EXTFEATURE_XD)) {
3631 return ENOTSUP;
3632 }
3633 #endif
3634 nx_enabled = new_value;
3635 }
3636 return error;
3637 #endif /* SECURE_KERNEL */
3638 }
3639 #endif
3640
3641 #if DEVELOPMENT || DEBUG
3642 SYSCTL_PROC(_kern, KERN_NX_PROTECTION, nx,
3643 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
3644 0, 0, sysctl_nx, "I", "");
3645 #endif
3646
3647 STATIC int
3648 sysctl_loadavg
3649 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3650 {
3651 if (proc_is64bit(req->p)) {
3652 struct user64_loadavg loadinfo64 = {};
3653 fill_loadavg64(&averunnable, &loadinfo64);
3654 return sysctl_io_opaque(req, &loadinfo64, sizeof(loadinfo64), NULL);
3655 } else {
3656 struct user32_loadavg loadinfo32 = {};
3657 fill_loadavg32(&averunnable, &loadinfo32);
3658 return sysctl_io_opaque(req, &loadinfo32, sizeof(loadinfo32), NULL);
3659 }
3660 }
3661
3662 SYSCTL_PROC(_vm, VM_LOADAVG, loadavg,
3663 CTLTYPE_STRUCT | CTLFLAG_RD | CTLFLAG_LOCKED,
3664 0, 0, sysctl_loadavg, "S,loadavg", "");
3665
3666 /*
3667 * Note: Thread safe; vm_map_lock protects in vm_toggle_entry_reuse()
3668 */
3669 STATIC int
3670 sysctl_vm_toggle_address_reuse(__unused struct sysctl_oid *oidp, __unused void *arg1,
3671 __unused int arg2, struct sysctl_req *req)
3672 {
3673 int old_value = 0, new_value = 0, error = 0;
3674
3675 if (vm_toggle_entry_reuse( VM_TOGGLE_GETVALUE, &old_value )) {
3676 return error;
3677 }
3678 error = sysctl_io_number(req, old_value, sizeof(int), &new_value, NULL);
3679 if (!error) {
3680 return vm_toggle_entry_reuse(new_value, NULL);
3681 }
3682 return error;
3683 }
3684
3685 SYSCTL_PROC(_debug, OID_AUTO, toggle_address_reuse, CTLFLAG_ANYBODY | CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED, 0, 0, sysctl_vm_toggle_address_reuse, "I", "");
3686
3687 #ifdef CONFIG_XNUPOST
3688
3689 extern uint32_t xnupost_get_estimated_testdata_size(void);
3690 extern int xnupost_reset_all_tests(void);
3691
3692 STATIC int
3693 sysctl_handle_xnupost_get_tests SYSCTL_HANDLER_ARGS
3694 {
3695 /* fixup unused arguments warnings */
3696 __unused int _oa2 = arg2;
3697 __unused void * _oa1 = arg1;
3698 __unused struct sysctl_oid * _oidp = oidp;
3699
3700 int error = 0;
3701 user_addr_t oldp = 0;
3702 user_addr_t newp = 0;
3703 uint32_t usedbytes = 0;
3704
3705 oldp = req->oldptr;
3706 newp = req->newptr;
3707
3708 if (newp) {
3709 return ENOTSUP;
3710 }
3711
3712 if ((void *)oldp == NULL) {
3713 /* return estimated size for second call where info can be placed */
3714 req->oldidx = xnupost_get_estimated_testdata_size();
3715 } else {
3716 error = xnupost_export_testdata((void *)oldp, req->oldlen, &usedbytes);
3717 req->oldidx = usedbytes;
3718 }
3719
3720 return error;
3721 }
3722
3723 SYSCTL_PROC(_debug,
3724 OID_AUTO,
3725 xnupost_get_tests,
3726 CTLFLAG_MASKED | CTLFLAG_ANYBODY | CTLTYPE_OPAQUE | CTLFLAG_RD | CTLFLAG_LOCKED,
3727 0,
3728 0,
3729 sysctl_handle_xnupost_get_tests,
3730 "-",
3731 "read xnupost test data in kernel");
3732
3733 #if CONFIG_EXT_PANICLOG
3734 /*
3735 * Extensible panic log test hooks
3736 */
3737 static int
3738 sysctl_debug_ext_paniclog_test_hook SYSCTL_HANDLER_ARGS
3739 {
3740 #pragma unused(arg1, arg2)
3741 int rval = 0;
3742 uint32_t test_option = 0;
3743
3744 rval = sysctl_handle_int(oidp, &test_option, 0, req);
3745
3746 if (rval == 0 && req->newptr) {
3747 rval = ext_paniclog_test_hook(test_option);
3748 }
3749
3750 return rval;
3751 }
3752
3753 SYSCTL_PROC(_debug, OID_AUTO, ext_paniclog_test_hook,
3754 CTLTYPE_INT | CTLFLAG_RW,
3755 0, 0,
3756 sysctl_debug_ext_paniclog_test_hook, "A", "ext paniclog test hook");
3757
3758 #endif
3759
3760 STATIC int
3761 sysctl_debug_xnupost_ctl SYSCTL_HANDLER_ARGS
3762 {
3763 /* fixup unused arguments warnings */
3764 __unused int _oa2 = arg2;
3765 __unused void * _oa1 = arg1;
3766 __unused struct sysctl_oid * _oidp = oidp;
3767
3768 #define ARRCOUNT 4
3769 /*
3770 * INPUT: ACTION, PARAM1, PARAM2, PARAM3
3771 * OUTPUT: RESULTCODE, ADDITIONAL DATA
3772 */
3773 int32_t outval[ARRCOUNT] = {0};
3774 int32_t input[ARRCOUNT] = {0};
3775 int32_t out_size = sizeof(outval);
3776 int32_t in_size = sizeof(input);
3777 int error = 0;
3778
3779 /* if this is NULL call to find out size, send out size info */
3780 if (!req->newptr) {
3781 goto out;
3782 }
3783
3784 /* pull in provided value from userspace */
3785 error = SYSCTL_IN(req, &input[0], in_size);
3786 if (error) {
3787 return error;
3788 }
3789
3790 if (input[0] == XTCTL_RESET_TESTDATA) {
3791 outval[0] = xnupost_reset_all_tests();
3792 goto out;
3793 }
3794
3795 out:
3796 error = SYSCTL_OUT(req, &outval[0], out_size);
3797 return error;
3798 }
3799
3800 SYSCTL_PROC(_debug,
3801 OID_AUTO,
3802 xnupost_testctl,
3803 CTLFLAG_MASKED | CTLFLAG_ANYBODY | CTLTYPE_OPAQUE | CTLFLAG_RW | CTLFLAG_LOCKED,
3804 0,
3805 0,
3806 sysctl_debug_xnupost_ctl,
3807 "I",
3808 "xnupost control for kernel testing");
3809
3810 extern void test_oslog_handleOSLogCtl(int32_t * in, int32_t * out, int32_t arraycount);
3811
3812 STATIC int
3813 sysctl_debug_test_oslog_ctl(__unused struct sysctl_oid * oidp, __unused void * arg1, __unused int arg2, struct sysctl_req * req)
3814 {
3815 #define ARRCOUNT 4
3816 int32_t outval[ARRCOUNT] = {0};
3817 int32_t input[ARRCOUNT] = {0};
3818 int32_t size_outval = sizeof(outval);
3819 int32_t size_inval = sizeof(input);
3820 int32_t error;
3821
3822 /* if this is NULL call to find out size, send out size info */
3823 if (!req->newptr) {
3824 error = SYSCTL_OUT(req, &outval[0], size_outval);
3825 return error;
3826 }
3827
3828 /* pull in provided value from userspace */
3829 error = SYSCTL_IN(req, &input[0], size_inval);
3830 if (error) {
3831 return error;
3832 }
3833
3834 test_oslog_handleOSLogCtl(input, outval, ARRCOUNT);
3835
3836 error = SYSCTL_OUT(req, &outval[0], size_outval);
3837
3838 return error;
3839 }
3840
3841 SYSCTL_PROC(_debug,
3842 OID_AUTO,
3843 test_OSLogCtl,
3844 CTLFLAG_MASKED | CTLFLAG_ANYBODY | CTLTYPE_OPAQUE | CTLFLAG_RW | CTLFLAG_LOCKED,
3845 0,
3846 0,
3847 sysctl_debug_test_oslog_ctl,
3848 "I",
3849 "testing oslog in kernel");
3850
3851 #include <mach/task.h>
3852 #include <mach/semaphore.h>
3853
3854 static LCK_GRP_DECLARE(sysctl_debug_test_stackshot_owner_grp, "test-stackshot-owner-grp");
3855 static LCK_MTX_DECLARE(sysctl_debug_test_stackshot_owner_init_mtx,
3856 &sysctl_debug_test_stackshot_owner_grp);
3857
3858 /* This is a sysctl for testing collection of owner info on a lock in kernel space. A multi-threaded
3859 * test from userland sets this sysctl in such a way that a thread blocks in kernel mode, and a
3860 * stackshot is taken to see if the owner of the lock can be identified.
3861 *
3862 * We can't return to userland with a kernel lock held, so be sure to unlock before we leave.
3863 * the semaphores allow us to artificially create cases where the lock is being held and the
3864 * thread is hanging / taking a long time to do something. */
3865
3866 volatile char sysctl_debug_test_stackshot_mtx_inited = 0;
3867 semaphore_t sysctl_debug_test_stackshot_mutex_sem;
3868 lck_mtx_t sysctl_debug_test_stackshot_owner_lck;
3869
3870 #define SYSCTL_DEBUG_MTX_ACQUIRE_WAIT 1
3871 #define SYSCTL_DEBUG_MTX_ACQUIRE_NOWAIT 2
3872 #define SYSCTL_DEBUG_MTX_SIGNAL 3
3873 #define SYSCTL_DEBUG_MTX_TEARDOWN 4
3874
3875 STATIC int
3876 sysctl_debug_test_stackshot_mutex_owner(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3877 {
3878 long long option = -1;
3879 /* if the user tries to read the sysctl, we tell them what the address of the lock is (to test against stackshot's output) */
3880 long long mtx_unslid_addr = (long long)VM_KERNEL_UNSLIDE_OR_PERM(&sysctl_debug_test_stackshot_owner_lck);
3881 int error = sysctl_io_number(req, mtx_unslid_addr, sizeof(long long), (void*)&option, NULL);
3882
3883 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_init_mtx);
3884 if (!sysctl_debug_test_stackshot_mtx_inited) {
3885 lck_mtx_init(&sysctl_debug_test_stackshot_owner_lck,
3886 &sysctl_debug_test_stackshot_owner_grp,
3887 LCK_ATTR_NULL);
3888 semaphore_create(kernel_task,
3889 &sysctl_debug_test_stackshot_mutex_sem,
3890 SYNC_POLICY_FIFO, 0);
3891 sysctl_debug_test_stackshot_mtx_inited = 1;
3892 }
3893 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_init_mtx);
3894
3895 if (!error) {
3896 switch (option) {
3897 case SYSCTL_DEBUG_MTX_ACQUIRE_NOWAIT:
3898 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_lck);
3899 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_lck);
3900 break;
3901 case SYSCTL_DEBUG_MTX_ACQUIRE_WAIT:
3902 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_lck);
3903 semaphore_wait(sysctl_debug_test_stackshot_mutex_sem);
3904 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_lck);
3905 break;
3906 case SYSCTL_DEBUG_MTX_SIGNAL:
3907 semaphore_signal(sysctl_debug_test_stackshot_mutex_sem);
3908 break;
3909 case SYSCTL_DEBUG_MTX_TEARDOWN:
3910 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_init_mtx);
3911
3912 lck_mtx_destroy(&sysctl_debug_test_stackshot_owner_lck,
3913 &sysctl_debug_test_stackshot_owner_grp);
3914 semaphore_destroy(kernel_task,
3915 sysctl_debug_test_stackshot_mutex_sem);
3916 sysctl_debug_test_stackshot_mtx_inited = 0;
3917
3918 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_init_mtx);
3919 break;
3920 case -1: /* user just wanted to read the value, so do nothing */
3921 break;
3922 default:
3923 error = EINVAL;
3924 break;
3925 }
3926 }
3927 return error;
3928 }
3929
3930 /* we can't return to userland with a kernel rwlock held, so be sure to unlock before we leave.
3931 * the semaphores allow us to artificially create cases where the lock is being held and the
3932 * thread is hanging / taking a long time to do something. */
3933
3934 SYSCTL_PROC(_debug,
3935 OID_AUTO,
3936 test_MutexOwnerCtl,
3937 CTLFLAG_MASKED | CTLFLAG_ANYBODY | CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
3938 0,
3939 0,
3940 sysctl_debug_test_stackshot_mutex_owner,
3941 "-",
3942 "Testing mutex owner in kernel");
3943
3944 volatile char sysctl_debug_test_stackshot_rwlck_inited = 0;
3945 lck_rw_t sysctl_debug_test_stackshot_owner_rwlck;
3946 semaphore_t sysctl_debug_test_stackshot_rwlck_sem;
3947
3948 #define SYSCTL_DEBUG_KRWLCK_RACQUIRE_NOWAIT 1
3949 #define SYSCTL_DEBUG_KRWLCK_RACQUIRE_WAIT 2
3950 #define SYSCTL_DEBUG_KRWLCK_WACQUIRE_NOWAIT 3
3951 #define SYSCTL_DEBUG_KRWLCK_WACQUIRE_WAIT 4
3952 #define SYSCTL_DEBUG_KRWLCK_SIGNAL 5
3953 #define SYSCTL_DEBUG_KRWLCK_TEARDOWN 6
3954
3955 STATIC int
3956 sysctl_debug_test_stackshot_rwlck_owner(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
3957 {
3958 long long option = -1;
3959 /* if the user tries to read the sysctl, we tell them what the address of the lock is
3960 * (to test against stackshot's output) */
3961 long long rwlck_unslid_addr = (long long)VM_KERNEL_UNSLIDE_OR_PERM(&sysctl_debug_test_stackshot_owner_rwlck);
3962 int error = sysctl_io_number(req, rwlck_unslid_addr, sizeof(long long), (void*)&option, NULL);
3963
3964 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_init_mtx);
3965 if (!sysctl_debug_test_stackshot_rwlck_inited) {
3966 lck_rw_init(&sysctl_debug_test_stackshot_owner_rwlck,
3967 &sysctl_debug_test_stackshot_owner_grp,
3968 LCK_ATTR_NULL);
3969 semaphore_create(kernel_task,
3970 &sysctl_debug_test_stackshot_rwlck_sem,
3971 SYNC_POLICY_FIFO,
3972 0);
3973 sysctl_debug_test_stackshot_rwlck_inited = 1;
3974 }
3975 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_init_mtx);
3976
3977 if (!error) {
3978 switch (option) {
3979 case SYSCTL_DEBUG_KRWLCK_RACQUIRE_NOWAIT:
3980 lck_rw_lock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_SHARED);
3981 lck_rw_unlock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_SHARED);
3982 break;
3983 case SYSCTL_DEBUG_KRWLCK_RACQUIRE_WAIT:
3984 lck_rw_lock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_SHARED);
3985 semaphore_wait(sysctl_debug_test_stackshot_rwlck_sem);
3986 lck_rw_unlock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_SHARED);
3987 break;
3988 case SYSCTL_DEBUG_KRWLCK_WACQUIRE_NOWAIT:
3989 lck_rw_lock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_EXCLUSIVE);
3990 lck_rw_unlock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_EXCLUSIVE);
3991 break;
3992 case SYSCTL_DEBUG_KRWLCK_WACQUIRE_WAIT:
3993 lck_rw_lock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_EXCLUSIVE);
3994 semaphore_wait(sysctl_debug_test_stackshot_rwlck_sem);
3995 lck_rw_unlock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_EXCLUSIVE);
3996 break;
3997 case SYSCTL_DEBUG_KRWLCK_SIGNAL:
3998 semaphore_signal(sysctl_debug_test_stackshot_rwlck_sem);
3999 break;
4000 case SYSCTL_DEBUG_KRWLCK_TEARDOWN:
4001 lck_mtx_lock(&sysctl_debug_test_stackshot_owner_init_mtx);
4002
4003 lck_rw_destroy(&sysctl_debug_test_stackshot_owner_rwlck,
4004 &sysctl_debug_test_stackshot_owner_grp);
4005 semaphore_destroy(kernel_task,
4006 sysctl_debug_test_stackshot_rwlck_sem);
4007 sysctl_debug_test_stackshot_rwlck_inited = 0;
4008
4009 lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_init_mtx);
4010 break;
4011 case -1: /* user just wanted to read the value, so do nothing */
4012 break;
4013 default:
4014 error = EINVAL;
4015 break;
4016 }
4017 }
4018 return error;
4019 }
4020
4021
4022 SYSCTL_PROC(_debug,
4023 OID_AUTO,
4024 test_RWLockOwnerCtl,
4025 CTLFLAG_MASKED | CTLFLAG_ANYBODY | CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
4026 0,
4027 0,
4028 sysctl_debug_test_stackshot_rwlck_owner,
4029 "-",
4030 "Testing rwlock owner in kernel");
4031 #endif /* !CONFIG_XNUPOST */
4032
4033 STATIC int
4034 sysctl_swapusage
4035 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4036 {
4037 int error;
4038 uint64_t swap_total;
4039 uint64_t swap_avail;
4040 vm_size_t swap_pagesize;
4041 boolean_t swap_encrypted;
4042 struct xsw_usage xsu = {};
4043
4044 error = macx_swapinfo(&swap_total,
4045 &swap_avail,
4046 &swap_pagesize,
4047 &swap_encrypted);
4048 if (error) {
4049 return error;
4050 }
4051
4052 xsu.xsu_total = swap_total;
4053 xsu.xsu_avail = swap_avail;
4054 xsu.xsu_used = swap_total - swap_avail;
4055 xsu.xsu_pagesize = (u_int32_t)MIN(swap_pagesize, UINT32_MAX);
4056 xsu.xsu_encrypted = swap_encrypted;
4057 return sysctl_io_opaque(req, &xsu, sizeof(xsu), NULL);
4058 }
4059
4060
4061
4062 SYSCTL_PROC(_vm, VM_SWAPUSAGE, swapusage,
4063 CTLTYPE_STRUCT | CTLFLAG_RD | CTLFLAG_LOCKED,
4064 0, 0, sysctl_swapusage, "S,xsw_usage", "");
4065
4066 extern int vm_swap_enabled;
4067 SYSCTL_INT(_vm, OID_AUTO, swap_enabled, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_swap_enabled, 0, "");
4068
4069 #if DEVELOPMENT || DEBUG
4070 extern int vm_num_swap_files_config;
4071 extern int vm_num_swap_files;
4072 extern lck_mtx_t vm_swap_data_lock;
4073 #define VM_MAX_SWAP_FILE_NUM 100
4074
4075 static int
4076 sysctl_vm_config_num_swap_files SYSCTL_HANDLER_ARGS
4077 {
4078 #pragma unused(arg1, arg2)
4079 int error = 0, val = vm_num_swap_files_config;
4080
4081 error = sysctl_handle_int(oidp, &val, 0, req);
4082 if (error || !req->newptr) {
4083 goto out;
4084 }
4085
4086 if (!VM_CONFIG_SWAP_IS_ACTIVE && !VM_CONFIG_FREEZER_SWAP_IS_ACTIVE) {
4087 printf("Swap is disabled\n");
4088 error = EINVAL;
4089 goto out;
4090 }
4091
4092 lck_mtx_lock(&vm_swap_data_lock);
4093
4094 if (val < vm_num_swap_files) {
4095 printf("Cannot configure fewer swap files than already exist.\n");
4096 error = EINVAL;
4097 lck_mtx_unlock(&vm_swap_data_lock);
4098 goto out;
4099 }
4100
4101 if (val > VM_MAX_SWAP_FILE_NUM) {
4102 printf("Capping number of swap files to upper bound.\n");
4103 val = VM_MAX_SWAP_FILE_NUM;
4104 }
4105
4106 vm_num_swap_files_config = val;
4107 lck_mtx_unlock(&vm_swap_data_lock);
4108 out:
4109
4110 return 0;
4111 }
4112
4113 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", "");
4114 #endif /* DEVELOPMENT || DEBUG */
4115
4116 /* this kernel does NOT implement shared_region_make_private_np() */
4117 SYSCTL_INT(_kern, KERN_SHREG_PRIVATIZABLE, shreg_private,
4118 CTLFLAG_RD | CTLFLAG_LOCKED,
4119 (int *)NULL, 0, "");
4120
4121 STATIC int
4122 fetch_process_cputype(
4123 proc_t cur_proc,
4124 int *name,
4125 u_int namelen,
4126 cpu_type_t *cputype)
4127 {
4128 proc_t p = PROC_NULL;
4129 int refheld = 0;
4130 cpu_type_t ret = 0;
4131 int error = 0;
4132
4133 if (namelen == 0) {
4134 p = cur_proc;
4135 } else if (namelen == 1) {
4136 p = proc_find(name[0]);
4137 if (p == NULL) {
4138 return EINVAL;
4139 }
4140 refheld = 1;
4141 } else {
4142 error = EINVAL;
4143 goto out;
4144 }
4145
4146 ret = cpu_type() & ~CPU_ARCH_MASK;
4147 if (IS_64BIT_PROCESS(p)) {
4148 ret |= CPU_ARCH_ABI64;
4149 }
4150
4151 *cputype = ret;
4152
4153 if (refheld != 0) {
4154 proc_rele(p);
4155 }
4156 out:
4157 return error;
4158 }
4159
4160 #if CONFIG_ROSETTA
4161 STATIC int
4162 sysctl_sysctl_translated(
4163 __unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, __unused struct sysctl_req *req)
4164 {
4165 int res = 0;
4166 if (proc_is_translated(req->p)) {
4167 res = 1;
4168 }
4169 return SYSCTL_OUT(req, &res, sizeof(res));
4170 }
4171 SYSCTL_PROC(_sysctl, OID_AUTO, proc_translated, CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED, 0, 0, sysctl_sysctl_translated, "I", "proc_translated");
4172 #endif /* CONFIG_ROSETTA */
4173
4174 STATIC int
4175 sysctl_sysctl_native(__unused struct sysctl_oid *oidp, void *arg1, int arg2,
4176 struct sysctl_req *req)
4177 {
4178 int error;
4179 cpu_type_t proc_cputype = 0;
4180 if ((error = fetch_process_cputype(req->p, (int *)arg1, arg2, &proc_cputype)) != 0) {
4181 return error;
4182 }
4183 int res = 1;
4184 if ((proc_cputype & ~CPU_ARCH_MASK) != (cpu_type() & ~CPU_ARCH_MASK)) {
4185 res = 0;
4186 }
4187 return SYSCTL_OUT(req, &res, sizeof(res));
4188 }
4189 SYSCTL_PROC(_sysctl, OID_AUTO, proc_native, CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED, 0, 0, sysctl_sysctl_native, "I", "proc_native");
4190
4191 STATIC int
4192 sysctl_sysctl_cputype(__unused struct sysctl_oid *oidp, void *arg1, int arg2,
4193 struct sysctl_req *req)
4194 {
4195 int error;
4196 cpu_type_t proc_cputype = 0;
4197 if ((error = fetch_process_cputype(req->p, (int *)arg1, arg2, &proc_cputype)) != 0) {
4198 return error;
4199 }
4200 return SYSCTL_OUT(req, &proc_cputype, sizeof(proc_cputype));
4201 }
4202 SYSCTL_PROC(_sysctl, OID_AUTO, proc_cputype, CTLTYPE_NODE | CTLFLAG_RD | CTLFLAG_LOCKED, 0, 0, sysctl_sysctl_cputype, "I", "proc_cputype");
4203
4204 STATIC int
4205 sysctl_safeboot
4206 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4207 {
4208 return sysctl_io_number(req, boothowto & RB_SAFEBOOT ? 1 : 0, sizeof(int), NULL, NULL);
4209 }
4210
4211 SYSCTL_PROC(_kern, KERN_SAFEBOOT, safeboot,
4212 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
4213 0, 0, sysctl_safeboot, "I", "");
4214
4215 STATIC int
4216 sysctl_singleuser
4217 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4218 {
4219 return sysctl_io_number(req, boothowto & RB_SINGLE ? 1 : 0, sizeof(int), NULL, NULL);
4220 }
4221
4222 SYSCTL_PROC(_kern, OID_AUTO, singleuser,
4223 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
4224 0, 0, sysctl_singleuser, "I", "");
4225
4226 STATIC int
4227 sysctl_minimalboot
4228 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4229 {
4230 return sysctl_io_number(req, minimalboot, sizeof(int), NULL, NULL);
4231 }
4232
4233 SYSCTL_PROC(_kern, OID_AUTO, minimalboot,
4234 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
4235 0, 0, sysctl_minimalboot, "I", "");
4236
4237 /*
4238 * Controls for debugging affinity sets - see osfmk/kern/affinity.c
4239 */
4240 extern boolean_t affinity_sets_enabled;
4241 extern int affinity_sets_mapping;
4242
4243 SYSCTL_INT(_kern, OID_AUTO, affinity_sets_enabled,
4244 CTLFLAG_RW | CTLFLAG_LOCKED, (int *) &affinity_sets_enabled, 0, "hinting enabled");
4245 SYSCTL_INT(_kern, OID_AUTO, affinity_sets_mapping,
4246 CTLFLAG_RW | CTLFLAG_LOCKED, &affinity_sets_mapping, 0, "mapping policy");
4247
4248 /*
4249 * Boolean indicating if KASLR is active.
4250 */
4251 STATIC int
4252 sysctl_slide
4253 (__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4254 {
4255 uint32_t slide;
4256
4257 slide = vm_kernel_slide ? 1 : 0;
4258
4259 return sysctl_io_number( req, slide, sizeof(int), NULL, NULL);
4260 }
4261
4262 SYSCTL_PROC(_kern, OID_AUTO, slide,
4263 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
4264 0, 0, sysctl_slide, "I", "");
4265
4266 #if DEBUG || DEVELOPMENT
4267 #if defined(__arm64__)
4268 extern vm_offset_t segTEXTEXECB;
4269
4270 static int
4271 sysctl_kernel_text_exec_base_slide SYSCTL_HANDLER_ARGS
4272 {
4273 #pragma unused(arg1, arg2, oidp)
4274 unsigned long slide = 0;
4275 kc_format_t kc_format;
4276
4277 PE_get_primary_kc_format(&kc_format);
4278
4279 if (kc_format == KCFormatFileset) {
4280 void *kch = PE_get_kc_header(KCKindPrimary);
4281 slide = (unsigned long)segTEXTEXECB - (unsigned long)kch + vm_kernel_slide;
4282 }
4283 return SYSCTL_OUT(req, &slide, sizeof(slide));
4284 }
4285
4286 SYSCTL_QUAD(_kern, OID_AUTO, kernel_slide, CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED, &vm_kernel_slide, "");
4287 SYSCTL_QUAD(_kern, OID_AUTO, kernel_text_exec_base, CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED, &segTEXTEXECB, "");
4288 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", "");
4289 #endif /* defined(__arm64__) */
4290
4291 /* User address of the PFZ */
4292 extern user32_addr_t commpage_text32_location;
4293 extern user64_addr_t commpage_text64_location;
4294
4295 STATIC int
4296 sysctl_pfz_start SYSCTL_HANDLER_ARGS
4297 {
4298 #pragma unused(oidp, arg1, arg2)
4299
4300 #ifdef __LP64__
4301 return sysctl_io_number(req, commpage_text64_location, sizeof(user64_addr_t), NULL, NULL);
4302 #else
4303 return sysctl_io_number(req, commpage_text32_location, sizeof(user32_addr_t), NULL, NULL);
4304 #endif
4305 }
4306
4307 SYSCTL_PROC(_kern, OID_AUTO, pfz,
4308 CTLTYPE_QUAD | CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_MASKED,
4309 0, 0, sysctl_pfz_start, "I", "");
4310 #endif
4311
4312
4313 /*
4314 * Limit on total memory users can wire.
4315 *
4316 * vm_global_user_wire_limit - system wide limit on wired memory from all processes combined.
4317 *
4318 * vm_per_task_user_wire_limit - per address space limit on wired memory. This puts a cap on the process's rlimit value.
4319 *
4320 * These values are initialized to reasonable defaults at boot time based on the available physical memory in
4321 * kmem_init().
4322 *
4323 * All values are in bytes.
4324 */
4325
4326 vm_map_size_t vm_global_user_wire_limit;
4327 vm_map_size_t vm_per_task_user_wire_limit;
4328 extern uint64_t max_mem_actual, max_mem;
4329
4330 uint64_t vm_add_wire_count_over_global_limit;
4331 uint64_t vm_add_wire_count_over_user_limit;
4332 /*
4333 * We used to have a global in the kernel called vm_global_no_user_wire_limit which was the inverse
4334 * of vm_global_user_wire_limit. But maintaining both of those is silly, and vm_global_user_wire_limit is the
4335 * real limit.
4336 * This function is for backwards compatibility with userspace
4337 * since we exposed the old global via a sysctl.
4338 */
4339 STATIC int
4340 sysctl_global_no_user_wire_amount(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4341 {
4342 vm_map_size_t old_value;
4343 vm_map_size_t new_value;
4344 int changed;
4345 int error;
4346 uint64_t config_memsize = max_mem;
4347 #if defined(XNU_TARGET_OS_OSX)
4348 config_memsize = max_mem_actual;
4349 #endif /* defined(XNU_TARGET_OS_OSX) */
4350
4351 old_value = (vm_map_size_t)(config_memsize - vm_global_user_wire_limit);
4352 error = sysctl_io_number(req, old_value, sizeof(vm_map_size_t), &new_value, &changed);
4353 if (changed) {
4354 if ((uint64_t)new_value > config_memsize) {
4355 error = EINVAL;
4356 } else {
4357 vm_global_user_wire_limit = (vm_map_size_t)(config_memsize - new_value);
4358 }
4359 }
4360 return error;
4361 }
4362 /*
4363 * There needs to be a more automatic/elegant way to do this
4364 */
4365 SYSCTL_QUAD(_vm, OID_AUTO, global_user_wire_limit, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_global_user_wire_limit, "");
4366 SYSCTL_QUAD(_vm, OID_AUTO, user_wire_limit, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_per_task_user_wire_limit, "");
4367 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", "");
4368
4369 /*
4370 * Relaxed atomic RW of a 64bit value via sysctl.
4371 */
4372 STATIC int
4373 sysctl_r_64bit_atomic(uint64_t *ptr, struct sysctl_req *req)
4374 {
4375 uint64_t old_value;
4376 uint64_t new_value;
4377 int error;
4378
4379 old_value = os_atomic_load_wide(ptr, relaxed);
4380 error = sysctl_io_number(req, old_value, sizeof(vm_map_size_t), &new_value, NULL);
4381 return error;
4382 }
4383 STATIC int
4384 sysctl_add_wire_count_over_global_limit(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4385 {
4386 return sysctl_r_64bit_atomic(&vm_add_wire_count_over_global_limit, req);
4387 }
4388 STATIC int
4389 sysctl_add_wire_count_over_user_limit(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4390 {
4391 return sysctl_r_64bit_atomic(&vm_add_wire_count_over_user_limit, req);
4392 }
4393
4394 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", "");
4395 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", "");
4396
4397 #if DEVELOPMENT || DEBUG
4398 /* These sysctls are used to test the wired limit. */
4399 SYSCTL_INT(_vm, OID_AUTO, page_wire_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_wire_count, 0, "");
4400 #if XNU_VM_HAS_LOPAGE
4401 SYSCTL_INT(_vm, OID_AUTO, lopage_free_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_lopage_free_count, 0, "");
4402 #endif
4403 SYSCTL_INT(_vm, OID_AUTO, page_stolen_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_stolen_count, 0, "");
4404
4405 /*
4406 * Setting the per task variable exclude_physfootprint_ledger to 1 will allow the calling task to exclude memory entries that are
4407 * tagged by VM_LEDGER_TAG_DEFAULT and flagged by VM_LEDGER_FLAG_EXCLUDE_FOOTPRINT_DEBUG from its phys_footprint ledger.
4408 */
4409
4410 STATIC int
4411 sysctl_rw_task_no_footprint_for_debug(struct sysctl_oid *oidp __unused, void *arg1 __unused, int arg2 __unused, struct sysctl_req *req)
4412 {
4413 int error;
4414 int value;
4415 proc_t p = current_proc();
4416
4417 if (req->newptr) {
4418 // Write request
4419 error = SYSCTL_IN(req, &value, sizeof(value));
4420 if (!error) {
4421 if (value == 1) {
4422 task_set_no_footprint_for_debug(proc_task(p), TRUE);
4423 } else if (value == 0) {
4424 task_set_no_footprint_for_debug(proc_task(p), FALSE);
4425 } else {
4426 error = EINVAL;
4427 }
4428 }
4429 } else {
4430 // Read request
4431 value = task_get_no_footprint_for_debug(proc_task(p));
4432 error = SYSCTL_OUT(req, &value, sizeof(value));
4433 }
4434 return error;
4435 }
4436
4437 SYSCTL_PROC(_vm, OID_AUTO, task_no_footprint_for_debug,
4438 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_ANYBODY,
4439 0, 0, &sysctl_rw_task_no_footprint_for_debug, "I", "Allow debug memory to be excluded from this task's memory footprint (debug only)");
4440
4441 #endif /* DEVELOPMENT || DEBUG */
4442
4443
4444 extern int vm_map_copy_overwrite_aligned_src_not_internal;
4445 extern int vm_map_copy_overwrite_aligned_src_not_symmetric;
4446 extern int vm_map_copy_overwrite_aligned_src_large;
4447 SYSCTL_INT(_vm, OID_AUTO, vm_copy_src_not_internal, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_map_copy_overwrite_aligned_src_not_internal, 0, "");
4448 SYSCTL_INT(_vm, OID_AUTO, vm_copy_src_not_symmetric, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_map_copy_overwrite_aligned_src_not_symmetric, 0, "");
4449 SYSCTL_INT(_vm, OID_AUTO, vm_copy_src_large, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_map_copy_overwrite_aligned_src_large, 0, "");
4450
4451
4452 extern uint32_t vm_page_external_count;
4453
4454 SYSCTL_INT(_vm, OID_AUTO, vm_page_external_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_external_count, 0, "");
4455
4456 SYSCTL_INT(_vm, OID_AUTO, vm_page_filecache_min, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_state.vm_page_filecache_min, 0, "");
4457 SYSCTL_INT(_vm, OID_AUTO, vm_page_xpmapped_min, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_state.vm_page_xpmapped_min, 0, "");
4458
4459 #if DEVELOPMENT || DEBUG
4460 SYSCTL_INT(_vm, OID_AUTO, vm_page_filecache_min_divisor, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_state.vm_page_filecache_min_divisor, 0, "");
4461 SYSCTL_INT(_vm, OID_AUTO, vm_page_xpmapped_min_divisor, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_state.vm_page_xpmapped_min_divisor, 0, "");
4462 extern boolean_t vps_yield_for_pgqlockwaiters;
4463 SYSCTL_INT(_vm, OID_AUTO, vm_pageoutscan_yields_for_pageQlockwaiters, CTLFLAG_RW | CTLFLAG_LOCKED, &vps_yield_for_pgqlockwaiters, 0, "");
4464 #endif
4465
4466 extern int vm_compressor_mode;
4467 extern int vm_compressor_is_active;
4468 extern int vm_compressor_available;
4469 extern uint32_t c_seg_bufsize;
4470 extern uint32_t c_seg_allocsize;
4471 extern int c_seg_fixed_array_len;
4472 extern uint32_t c_segments_limit;
4473 extern uint32_t c_segment_pages_compressed_limit;
4474 extern uint64_t compressor_pool_size;
4475 extern uint32_t compressor_pool_multiplier;
4476 extern uint32_t vm_ripe_target_age;
4477 extern uint32_t swapout_target_age;
4478 extern _Atomic uint64_t compressor_bytes_used;
4479 extern _Atomic uint64_t c_segment_input_bytes;
4480 extern _Atomic uint64_t c_segment_compressed_bytes;
4481 extern uint32_t c_segment_pages_compressed;
4482 extern uint32_t compressor_eval_period_in_msecs;
4483 extern uint32_t compressor_sample_min_in_msecs;
4484 extern uint32_t compressor_sample_max_in_msecs;
4485 extern uint32_t compressor_thrashing_threshold_per_10msecs;
4486 extern uint32_t compressor_thrashing_min_per_10msecs;
4487 extern uint32_t vm_compressor_time_thread;
4488 extern uint32_t c_segment_svp_in_hash;
4489 extern uint32_t c_segment_svp_hash_succeeded;
4490 extern uint32_t c_segment_svp_hash_failed;
4491
4492 #if DEVELOPMENT || DEBUG
4493 extern uint32_t vm_compressor_minorcompact_threshold_divisor;
4494 extern uint32_t vm_compressor_majorcompact_threshold_divisor;
4495 extern uint32_t vm_compressor_unthrottle_threshold_divisor;
4496 extern uint32_t vm_compressor_catchup_threshold_divisor;
4497
4498 extern uint32_t vm_compressor_minorcompact_threshold_divisor_overridden;
4499 extern uint32_t vm_compressor_majorcompact_threshold_divisor_overridden;
4500 extern uint32_t vm_compressor_unthrottle_threshold_divisor_overridden;
4501 extern uint32_t vm_compressor_catchup_threshold_divisor_overridden;
4502
4503 extern vmct_stats_t vmct_stats;
4504
4505
4506 STATIC int
4507 sysctl_minorcompact_threshold_divisor(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4508 {
4509 int new_value, changed;
4510 int error = sysctl_io_number(req, vm_compressor_minorcompact_threshold_divisor, sizeof(int), &new_value, &changed);
4511
4512 if (changed) {
4513 vm_compressor_minorcompact_threshold_divisor = new_value;
4514 vm_compressor_minorcompact_threshold_divisor_overridden = 1;
4515 }
4516 return error;
4517 }
4518
4519 SYSCTL_PROC(_vm, OID_AUTO, compressor_minorcompact_threshold_divisor,
4520 CTLTYPE_INT | CTLFLAG_LOCKED | CTLFLAG_RW,
4521 0, 0, sysctl_minorcompact_threshold_divisor, "I", "");
4522
4523
4524 STATIC int
4525 sysctl_majorcompact_threshold_divisor(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4526 {
4527 int new_value, changed;
4528 int error = sysctl_io_number(req, vm_compressor_majorcompact_threshold_divisor, sizeof(int), &new_value, &changed);
4529
4530 if (changed) {
4531 vm_compressor_majorcompact_threshold_divisor = new_value;
4532 vm_compressor_majorcompact_threshold_divisor_overridden = 1;
4533 }
4534 return error;
4535 }
4536
4537 SYSCTL_PROC(_vm, OID_AUTO, compressor_majorcompact_threshold_divisor,
4538 CTLTYPE_INT | CTLFLAG_LOCKED | CTLFLAG_RW,
4539 0, 0, sysctl_majorcompact_threshold_divisor, "I", "");
4540
4541
4542 STATIC int
4543 sysctl_unthrottle_threshold_divisor(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4544 {
4545 int new_value, changed;
4546 int error = sysctl_io_number(req, vm_compressor_unthrottle_threshold_divisor, sizeof(int), &new_value, &changed);
4547
4548 if (changed) {
4549 vm_compressor_unthrottle_threshold_divisor = new_value;
4550 vm_compressor_unthrottle_threshold_divisor_overridden = 1;
4551 }
4552 return error;
4553 }
4554
4555 SYSCTL_PROC(_vm, OID_AUTO, compressor_unthrottle_threshold_divisor,
4556 CTLTYPE_INT | CTLFLAG_LOCKED | CTLFLAG_RW,
4557 0, 0, sysctl_unthrottle_threshold_divisor, "I", "");
4558
4559
4560 STATIC int
4561 sysctl_catchup_threshold_divisor(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4562 {
4563 int new_value, changed;
4564 int error = sysctl_io_number(req, vm_compressor_catchup_threshold_divisor, sizeof(int), &new_value, &changed);
4565
4566 if (changed) {
4567 vm_compressor_catchup_threshold_divisor = new_value;
4568 vm_compressor_catchup_threshold_divisor_overridden = 1;
4569 }
4570 return error;
4571 }
4572
4573 SYSCTL_PROC(_vm, OID_AUTO, compressor_catchup_threshold_divisor,
4574 CTLTYPE_INT | CTLFLAG_LOCKED | CTLFLAG_RW,
4575 0, 0, sysctl_catchup_threshold_divisor, "I", "");
4576 #endif
4577
4578
4579 SYSCTL_QUAD(_vm, OID_AUTO, compressor_input_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, ((uint64_t *)&c_segment_input_bytes), "");
4580 SYSCTL_QUAD(_vm, OID_AUTO, compressor_compressed_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, ((uint64_t *)&c_segment_compressed_bytes), "");
4581 SYSCTL_QUAD(_vm, OID_AUTO, compressor_bytes_used, CTLFLAG_RD | CTLFLAG_LOCKED, ((uint64_t *)&compressor_bytes_used), "");
4582
4583 SYSCTL_INT(_vm, OID_AUTO, compressor_mode, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_compressor_mode, 0, "");
4584 SYSCTL_INT(_vm, OID_AUTO, compressor_is_active, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_compressor_is_active, 0, "");
4585 SYSCTL_INT(_vm, OID_AUTO, compressor_swapout_target_age, CTLFLAG_RD | CTLFLAG_LOCKED, &swapout_target_age, 0, "");
4586 SYSCTL_INT(_vm, OID_AUTO, compressor_available, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_compressor_available, 0, "");
4587 SYSCTL_INT(_vm, OID_AUTO, compressor_segment_buffer_size, CTLFLAG_RD | CTLFLAG_LOCKED, &c_seg_bufsize, 0, "");
4588 SYSCTL_QUAD(_vm, OID_AUTO, compressor_pool_size, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_pool_size, "");
4589 SYSCTL_UINT(_vm, OID_AUTO, compressor_pool_multiplier, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_pool_multiplier, 0, "");
4590 SYSCTL_INT(_vm, OID_AUTO, compressor_segment_slots_fixed_array_len, CTLFLAG_RD | CTLFLAG_LOCKED, &c_seg_fixed_array_len, 0, "");
4591 SYSCTL_UINT(_vm, OID_AUTO, compressor_segment_limit, CTLFLAG_RD | CTLFLAG_LOCKED, &c_segments_limit, 0, "");
4592 SYSCTL_UINT(_vm, OID_AUTO, compressor_segment_pages_compressed_limit, CTLFLAG_RD | CTLFLAG_LOCKED, &c_segment_pages_compressed_limit, 0, "");
4593 SYSCTL_UINT(_vm, OID_AUTO, compressor_segment_alloc_size, CTLFLAG_RD | CTLFLAG_LOCKED, &c_seg_allocsize, 0, "");
4594 SYSCTL_UINT(_vm, OID_AUTO, compressor_segment_pages_compressed, CTLFLAG_RD | CTLFLAG_LOCKED, &c_segment_pages_compressed, 0, "");
4595 SYSCTL_UINT(_vm, OID_AUTO, compressor_segment_svp_in_hash, CTLFLAG_RD | CTLFLAG_LOCKED, &c_segment_svp_in_hash, 0, "");
4596 SYSCTL_UINT(_vm, OID_AUTO, compressor_segment_svp_hash_succeeded, CTLFLAG_RD | CTLFLAG_LOCKED, &c_segment_svp_hash_succeeded, 0, "");
4597 SYSCTL_UINT(_vm, OID_AUTO, compressor_segment_svp_hash_failed, CTLFLAG_RD | CTLFLAG_LOCKED, &c_segment_svp_hash_failed, 0, "");
4598
4599 #if CONFIG_TRACK_UNMODIFIED_ANON_PAGES
4600 extern uint64_t compressor_ro_uncompressed;
4601 extern uint64_t compressor_ro_uncompressed_total_returned;
4602 extern uint64_t compressor_ro_uncompressed_skip_returned;
4603 extern uint64_t compressor_ro_uncompressed_get;
4604 extern uint64_t compressor_ro_uncompressed_put;
4605 extern uint64_t compressor_ro_uncompressed_swap_usage;
4606
4607 SYSCTL_QUAD(_vm, OID_AUTO, compressor_ro_uncompressed_total_returned, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_ro_uncompressed_total_returned, "");
4608 SYSCTL_QUAD(_vm, OID_AUTO, compressor_ro_uncompressed_writes_saved, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_ro_uncompressed_skip_returned, "");
4609 SYSCTL_QUAD(_vm, OID_AUTO, compressor_ro_uncompressed_candidates, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_ro_uncompressed, "");
4610 SYSCTL_QUAD(_vm, OID_AUTO, compressor_ro_uncompressed_rereads, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_ro_uncompressed_get, "");
4611 SYSCTL_QUAD(_vm, OID_AUTO, compressor_ro_uncompressed_swap_pages_on_disk, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_ro_uncompressed_swap_usage, "");
4612 #endif /* CONFIG_TRACK_UNMODIFIED_ANON_PAGES */
4613
4614 extern int min_csegs_per_major_compaction;
4615 SYSCTL_INT(_vm, OID_AUTO, compressor_min_csegs_per_major_compaction, CTLFLAG_RW | CTLFLAG_LOCKED, &min_csegs_per_major_compaction, 0, "");
4616
4617 SYSCTL_INT(_vm, OID_AUTO, vm_ripe_target_age_in_secs, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_ripe_target_age, 0, "");
4618
4619 SYSCTL_INT(_vm, OID_AUTO, compressor_eval_period_in_msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &compressor_eval_period_in_msecs, 0, "");
4620 SYSCTL_INT(_vm, OID_AUTO, compressor_sample_min_in_msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &compressor_sample_min_in_msecs, 0, "");
4621 SYSCTL_INT(_vm, OID_AUTO, compressor_sample_max_in_msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &compressor_sample_max_in_msecs, 0, "");
4622 SYSCTL_INT(_vm, OID_AUTO, compressor_thrashing_threshold_per_10msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &compressor_thrashing_threshold_per_10msecs, 0, "");
4623 SYSCTL_INT(_vm, OID_AUTO, compressor_thrashing_min_per_10msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &compressor_thrashing_min_per_10msecs, 0, "");
4624
4625 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapouts_under_30s, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.unripe_under_30s, "");
4626 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapouts_under_60s, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.unripe_under_60s, "");
4627 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapouts_under_300s, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.unripe_under_300s, "");
4628 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_reclaim_swapins, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.reclaim_swapins, "");
4629 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_defrag_swapins, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.defrag_swapins, "");
4630 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_swapout_threshold_exceeded, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.compressor_swap_threshold_exceeded, "");
4631 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_swapout_fileq_throttled, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.external_q_throttled, "");
4632 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_swapout_free_count_low, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.free_count_below_reserve, "");
4633 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_swapout_thrashing_detected, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.thrashing_detected, "");
4634 SYSCTL_QUAD(_vm, OID_AUTO, compressor_swapper_swapout_fragmentation_detected, CTLFLAG_RD | CTLFLAG_LOCKED, &vmcs_stats.fragmentation_detected, "");
4635
4636 SYSCTL_STRING(_vm, OID_AUTO, swapfileprefix, CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED, swapfilename, sizeof(swapfilename) - SWAPFILENAME_INDEX_LEN, "");
4637
4638 SYSCTL_INT(_vm, OID_AUTO, compressor_timing_enabled, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_compressor_time_thread, 0, "");
4639
4640 #if DEVELOPMENT || DEBUG
4641 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_runtime0, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_runtimes[0], "");
4642 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_runtime1, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_runtimes[1], "");
4643
4644 SYSCTL_QUAD(_vm, OID_AUTO, compressor_threads_total_execution_time, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_cthreads_total, "");
4645
4646 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_pages0, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_pages[0], "");
4647 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_pages1, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_pages[1], "");
4648
4649 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_iterations0, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_iterations[0], "");
4650 SYSCTL_QUAD(_vm, OID_AUTO, compressor_thread_iterations1, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_iterations[1], "");
4651
4652 SYSCTL_INT(_vm, OID_AUTO, compressor_thread_minpages0, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_minpages[0], 0, "");
4653 SYSCTL_INT(_vm, OID_AUTO, compressor_thread_minpages1, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_minpages[1], 0, "");
4654
4655 SYSCTL_INT(_vm, OID_AUTO, compressor_thread_maxpages0, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_maxpages[0], 0, "");
4656 SYSCTL_INT(_vm, OID_AUTO, compressor_thread_maxpages1, CTLFLAG_RD | CTLFLAG_LOCKED, &vmct_stats.vmct_maxpages[1], 0, "");
4657
4658 int vm_compressor_injected_error_count;
4659
4660 SYSCTL_INT(_vm, OID_AUTO, compressor_injected_error_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_compressor_injected_error_count, 0, "");
4661
4662 static int
4663 sysctl_compressor_inject_error(__unused struct sysctl_oid *oidp,
4664 __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4665 {
4666 int result;
4667 vm_address_t va = 0;
4668 int changed;
4669
4670 result = sysctl_io_number(req, va, sizeof(va), &va, &changed);
4671 if (result == 0 && changed) {
4672 result = vm_map_inject_error(current_map(), va);
4673 if (result == 0) {
4674 /*
4675 * Count the number of errors injected successfully to detect
4676 * situations where corruption was caused by improper use of this
4677 * sysctl.
4678 */
4679 os_atomic_inc(&vm_compressor_injected_error_count, relaxed);
4680 }
4681 }
4682 return result;
4683 }
4684
4685 SYSCTL_PROC(_vm, OID_AUTO, compressor_inject_error, CTLTYPE_QUAD | CTLFLAG_LOCKED | CTLFLAG_RW,
4686 0, 0, sysctl_compressor_inject_error, "Q", "flips a bit in a compressed page for the current task");
4687
4688 /*
4689 * Opt a process in/out of self donation mode.
4690 */
4691 static int
4692 sysctl_vm_pid_toggle_selfdonate_pages SYSCTL_HANDLER_ARGS
4693 {
4694 #pragma unused(arg1, arg2)
4695 int error, pid = 0;
4696 proc_t p;
4697
4698 error = sysctl_handle_int(oidp, &pid, 0, req);
4699 if (error || !req->newptr) {
4700 return error;
4701 }
4702
4703 p = proc_find(pid);
4704 if (p != NULL) {
4705 (void) vm_toggle_task_selfdonate_pages(proc_task(p));
4706 proc_rele(p);
4707 return error;
4708 } else {
4709 printf("sysctl_vm_pid_selfdonate_pages: Invalid process\n");
4710 }
4711
4712 return EINVAL;
4713 }
4714 SYSCTL_PROC(_vm, OID_AUTO, pid_toggle_selfdonate_pages, CTLTYPE_INT | CTLFLAG_WR | CTLFLAG_LOCKED | CTLFLAG_MASKED,
4715 0, 0, &sysctl_vm_pid_toggle_selfdonate_pages, "I", "");
4716 #endif /* DEVELOPMENT || DEBUG */
4717 extern uint32_t vm_page_donate_mode;
4718 extern uint32_t vm_page_donate_target_high, vm_page_donate_target_low;
4719 SYSCTL_INT(_vm, OID_AUTO, vm_page_donate_mode, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_page_donate_mode, 0, "");
4720 SYSCTL_INT(_vm, OID_AUTO, vm_page_donate_target_high, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_page_donate_target_high, 0, "");
4721 SYSCTL_INT(_vm, OID_AUTO, vm_page_donate_target_low, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_page_donate_target_low, 0, "");
4722
4723 SYSCTL_QUAD(_vm, OID_AUTO, lz4_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_compressions, "");
4724 SYSCTL_QUAD(_vm, OID_AUTO, lz4_compression_failures, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_compression_failures, "");
4725 SYSCTL_QUAD(_vm, OID_AUTO, lz4_compressed_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_compressed_bytes, "");
4726 SYSCTL_QUAD(_vm, OID_AUTO, lz4_wk_compression_delta, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_wk_compression_delta, "");
4727 SYSCTL_QUAD(_vm, OID_AUTO, lz4_wk_compression_negative_delta, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_wk_compression_negative_delta, "");
4728
4729 SYSCTL_QUAD(_vm, OID_AUTO, lz4_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_decompressions, "");
4730 SYSCTL_QUAD(_vm, OID_AUTO, lz4_decompressed_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.lz4_decompressed_bytes, "");
4731
4732 SYSCTL_QUAD(_vm, OID_AUTO, uc_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.uc_decompressions, "");
4733
4734 SYSCTL_QUAD(_vm, OID_AUTO, wk_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_compressions, "");
4735
4736 SYSCTL_QUAD(_vm, OID_AUTO, wk_catime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_cabstime, "");
4737
4738 SYSCTL_QUAD(_vm, OID_AUTO, wkh_catime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wkh_cabstime, "");
4739 SYSCTL_QUAD(_vm, OID_AUTO, wkh_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wkh_compressions, "");
4740
4741 SYSCTL_QUAD(_vm, OID_AUTO, wks_catime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_cabstime, "");
4742 SYSCTL_QUAD(_vm, OID_AUTO, wks_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_compressions, "");
4743
4744 SYSCTL_QUAD(_vm, OID_AUTO, wk_compressions_exclusive, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_compressions_exclusive, "");
4745 SYSCTL_QUAD(_vm, OID_AUTO, wk_sv_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_sv_compressions, "");
4746 SYSCTL_QUAD(_vm, OID_AUTO, wk_mzv_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_mzv_compressions, "");
4747 SYSCTL_QUAD(_vm, OID_AUTO, wk_compression_failures, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_compression_failures, "");
4748 SYSCTL_QUAD(_vm, OID_AUTO, wk_compressed_bytes_exclusive, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_compressed_bytes_exclusive, "");
4749 SYSCTL_QUAD(_vm, OID_AUTO, wk_compressed_bytes_total, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_compressed_bytes_total, "");
4750
4751 SYSCTL_QUAD(_vm, OID_AUTO, wks_compressed_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_compressed_bytes, "");
4752 SYSCTL_QUAD(_vm, OID_AUTO, wks_compression_failures, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_compression_failures, "");
4753 SYSCTL_QUAD(_vm, OID_AUTO, wks_sv_compressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_sv_compressions, "");
4754
4755
4756 SYSCTL_QUAD(_vm, OID_AUTO, wk_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_decompressions, "");
4757
4758 SYSCTL_QUAD(_vm, OID_AUTO, wk_datime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_dabstime, "");
4759
4760 SYSCTL_QUAD(_vm, OID_AUTO, wkh_datime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wkh_dabstime, "");
4761 SYSCTL_QUAD(_vm, OID_AUTO, wkh_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wkh_decompressions, "");
4762
4763 SYSCTL_QUAD(_vm, OID_AUTO, wks_datime, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_dabstime, "");
4764 SYSCTL_QUAD(_vm, OID_AUTO, wks_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wks_decompressions, "");
4765
4766 SYSCTL_QUAD(_vm, OID_AUTO, wk_decompressed_bytes, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_decompressed_bytes, "");
4767 SYSCTL_QUAD(_vm, OID_AUTO, wk_sv_decompressions, CTLFLAG_RD | CTLFLAG_LOCKED, &compressor_stats.wk_sv_decompressions, "");
4768
4769 SYSCTL_INT(_vm, OID_AUTO, lz4_threshold, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_threshold, 0, "");
4770 SYSCTL_INT(_vm, OID_AUTO, wkdm_reeval_threshold, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.wkdm_reeval_threshold, 0, "");
4771 SYSCTL_INT(_vm, OID_AUTO, lz4_max_failure_skips, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_max_failure_skips, 0, "");
4772 SYSCTL_INT(_vm, OID_AUTO, lz4_max_failure_run_length, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_max_failure_run_length, 0, "");
4773 SYSCTL_INT(_vm, OID_AUTO, lz4_max_preselects, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_max_preselects, 0, "");
4774 SYSCTL_INT(_vm, OID_AUTO, lz4_run_preselection_threshold, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_run_preselection_threshold, 0, "");
4775 SYSCTL_INT(_vm, OID_AUTO, lz4_run_continue_bytes, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_run_continue_bytes, 0, "");
4776 SYSCTL_INT(_vm, OID_AUTO, lz4_profitable_bytes, CTLFLAG_RW | CTLFLAG_LOCKED, &vmctune.lz4_profitable_bytes, 0, "");
4777 #if DEVELOPMENT || DEBUG
4778 extern int vm_compressor_current_codec;
4779 extern int vm_compressor_test_seg_wp;
4780 extern boolean_t vm_compressor_force_sw_wkdm;
4781 SYSCTL_INT(_vm, OID_AUTO, compressor_codec, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_compressor_current_codec, 0, "");
4782 SYSCTL_INT(_vm, OID_AUTO, compressor_test_wp, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_compressor_test_seg_wp, 0, "");
4783
4784 SYSCTL_INT(_vm, OID_AUTO, wksw_force, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_compressor_force_sw_wkdm, 0, "");
4785 extern int precompy, wkswhw;
4786
4787 SYSCTL_INT(_vm, OID_AUTO, precompy, CTLFLAG_RW | CTLFLAG_LOCKED, &precompy, 0, "");
4788 SYSCTL_INT(_vm, OID_AUTO, wkswhw, CTLFLAG_RW | CTLFLAG_LOCKED, &wkswhw, 0, "");
4789 extern unsigned int vm_ktrace_enabled;
4790 SYSCTL_INT(_vm, OID_AUTO, vm_ktrace, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_ktrace_enabled, 0, "");
4791 #endif /* DEVELOPMENT || DEBUG */
4792
4793 #if CONFIG_PHANTOM_CACHE
4794 extern uint32_t phantom_cache_thrashing_threshold;
4795 extern uint32_t phantom_cache_eval_period_in_msecs;
4796 extern uint32_t phantom_cache_thrashing_threshold_ssd;
4797
4798
4799 SYSCTL_INT(_vm, OID_AUTO, phantom_cache_eval_period_in_msecs, CTLFLAG_RW | CTLFLAG_LOCKED, &phantom_cache_eval_period_in_msecs, 0, "");
4800 SYSCTL_INT(_vm, OID_AUTO, phantom_cache_thrashing_threshold, CTLFLAG_RW | CTLFLAG_LOCKED, &phantom_cache_thrashing_threshold, 0, "");
4801 SYSCTL_INT(_vm, OID_AUTO, phantom_cache_thrashing_threshold_ssd, CTLFLAG_RW | CTLFLAG_LOCKED, &phantom_cache_thrashing_threshold_ssd, 0, "");
4802 #endif
4803
4804 #if defined(__LP64__)
4805 extern uint32_t vm_page_background_count;
4806 extern uint32_t vm_page_background_target;
4807 extern uint32_t vm_page_background_internal_count;
4808 extern uint32_t vm_page_background_external_count;
4809 extern uint32_t vm_page_background_mode;
4810 extern uint32_t vm_page_background_exclude_external;
4811 extern uint64_t vm_page_background_promoted_count;
4812 extern uint64_t vm_pageout_rejected_bq_internal;
4813 extern uint64_t vm_pageout_rejected_bq_external;
4814
4815 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_mode, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_page_background_mode, 0, "");
4816 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_exclude_external, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_page_background_exclude_external, 0, "");
4817 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_target, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_page_background_target, 0, "");
4818 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_background_count, 0, "");
4819 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_internal_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_background_internal_count, 0, "");
4820 SYSCTL_INT(_vm, OID_AUTO, vm_page_background_external_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_background_external_count, 0, "");
4821
4822 SYSCTL_QUAD(_vm, OID_AUTO, vm_page_background_promoted_count, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_background_promoted_count, "");
4823 SYSCTL_QUAD(_vm, OID_AUTO, vm_pageout_considered_bq_internal, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_vminfo.vm_pageout_considered_bq_internal, "");
4824 SYSCTL_QUAD(_vm, OID_AUTO, vm_pageout_considered_bq_external, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_vminfo.vm_pageout_considered_bq_external, "");
4825 SYSCTL_QUAD(_vm, OID_AUTO, vm_pageout_rejected_bq_internal, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_rejected_bq_internal, "");
4826 SYSCTL_QUAD(_vm, OID_AUTO, vm_pageout_rejected_bq_external, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_pageout_rejected_bq_external, "");
4827
4828 #endif /* __LP64__ */
4829
4830 extern boolean_t vm_darkwake_mode;
4831
4832 STATIC int
4833 sysctl_toggle_darkwake_mode(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
4834 {
4835 int new_value, changed;
4836 int error = sysctl_io_number(req, vm_darkwake_mode, sizeof(int), &new_value, &changed);
4837
4838 if (!error && changed) {
4839 if (new_value != 0 && new_value != 1) {
4840 printf("Error: Invalid value passed to darkwake sysctl. Acceptable: 0 or 1.\n");
4841 error = EINVAL;
4842 } else {
4843 vm_update_darkwake_mode((boolean_t) new_value);
4844 }
4845 }
4846
4847 return error;
4848 }
4849
4850 SYSCTL_PROC(_vm, OID_AUTO, darkwake_mode,
4851 CTLTYPE_INT | CTLFLAG_LOCKED | CTLFLAG_RW,
4852 0, 0, sysctl_toggle_darkwake_mode, "I", "");
4853
4854 #if (DEVELOPMENT || DEBUG)
4855
4856 SYSCTL_UINT(_vm, OID_AUTO, vm_page_creation_throttled_hard,
4857 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
4858 &vm_page_creation_throttled_hard, 0, "");
4859
4860 SYSCTL_UINT(_vm, OID_AUTO, vm_page_creation_throttled_soft,
4861 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
4862 &vm_page_creation_throttled_soft, 0, "");
4863
4864 extern uint32_t vm_pageout_memorystatus_fb_factor_nr;
4865 extern uint32_t vm_pageout_memorystatus_fb_factor_dr;
4866 SYSCTL_INT(_vm, OID_AUTO, vm_pageout_memorystatus_fb_factor_nr, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_memorystatus_fb_factor_nr, 0, "");
4867 SYSCTL_INT(_vm, OID_AUTO, vm_pageout_memorystatus_fb_factor_dr, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_memorystatus_fb_factor_dr, 0, "");
4868
4869 extern uint32_t vm_grab_anon_nops;
4870
4871 SYSCTL_INT(_vm, OID_AUTO, vm_grab_anon_overrides, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_debug.vm_grab_anon_overrides, 0, "");
4872 SYSCTL_INT(_vm, OID_AUTO, vm_grab_anon_nops, CTLFLAG_RW | CTLFLAG_LOCKED, &vm_pageout_debug.vm_grab_anon_nops, 0, "");
4873 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, "");
4874
4875
4876 extern int vm_page_delayed_work_ctx_needed;
4877 SYSCTL_INT(_vm, OID_AUTO, vm_page_needed_delayed_work_ctx, CTLFLAG_RD | CTLFLAG_LOCKED, &vm_page_delayed_work_ctx_needed, 0, "");
4878
4879
4880 /* log message counters for persistence mode */
4881 SCALABLE_COUNTER_DECLARE(oslog_p_total_msgcount);
4882 SCALABLE_COUNTER_DECLARE(oslog_p_metadata_saved_msgcount);
4883 SCALABLE_COUNTER_DECLARE(oslog_p_metadata_dropped_msgcount);
4884 SCALABLE_COUNTER_DECLARE(oslog_p_signpost_saved_msgcount);
4885 SCALABLE_COUNTER_DECLARE(oslog_p_signpost_dropped_msgcount);
4886 SCALABLE_COUNTER_DECLARE(oslog_p_error_count);
4887 SCALABLE_COUNTER_DECLARE(oslog_p_error_count);
4888 SCALABLE_COUNTER_DECLARE(oslog_p_saved_msgcount);
4889 SCALABLE_COUNTER_DECLARE(oslog_p_dropped_msgcount);
4890 SCALABLE_COUNTER_DECLARE(oslog_p_boot_dropped_msgcount);
4891 SCALABLE_COUNTER_DECLARE(oslog_p_coprocessor_total_msgcount);
4892 SCALABLE_COUNTER_DECLARE(oslog_p_coprocessor_dropped_msgcount);
4893 SCALABLE_COUNTER_DECLARE(oslog_p_unresolved_kc_msgcount);
4894 SCALABLE_COUNTER_DECLARE(oslog_p_fmt_invalid_msgcount);
4895 SCALABLE_COUNTER_DECLARE(oslog_p_fmt_max_args_msgcount);
4896 SCALABLE_COUNTER_DECLARE(oslog_p_truncated_msgcount);
4897
4898 SCALABLE_COUNTER_DECLARE(oslog_subsystem_count);
4899 SCALABLE_COUNTER_DECLARE(oslog_subsystem_found);
4900 SCALABLE_COUNTER_DECLARE(oslog_subsystem_dropped);
4901
4902 SCALABLE_COUNTER_DECLARE(log_queue_cnt_received);
4903 SCALABLE_COUNTER_DECLARE(log_queue_cnt_rejected_fh);
4904 SCALABLE_COUNTER_DECLARE(log_queue_cnt_sent);
4905 SCALABLE_COUNTER_DECLARE(log_queue_cnt_dropped_nomem);
4906 SCALABLE_COUNTER_DECLARE(log_queue_cnt_queued);
4907 SCALABLE_COUNTER_DECLARE(log_queue_cnt_dropped_off);
4908 SCALABLE_COUNTER_DECLARE(log_queue_cnt_mem_active);
4909 SCALABLE_COUNTER_DECLARE(log_queue_cnt_mem_allocated);
4910 SCALABLE_COUNTER_DECLARE(log_queue_cnt_mem_released);
4911 SCALABLE_COUNTER_DECLARE(log_queue_cnt_mem_failed);
4912
4913 /* log message counters for streaming mode */
4914 SCALABLE_COUNTER_DECLARE(oslog_s_total_msgcount);
4915 SCALABLE_COUNTER_DECLARE(oslog_s_metadata_msgcount);
4916 SCALABLE_COUNTER_DECLARE(oslog_s_error_count);
4917 SCALABLE_COUNTER_DECLARE(oslog_s_streamed_msgcount);
4918 SCALABLE_COUNTER_DECLARE(oslog_s_dropped_msgcount);
4919
4920 /* log message counters for msgbuf logging */
4921 SCALABLE_COUNTER_DECLARE(oslog_msgbuf_msgcount);
4922 SCALABLE_COUNTER_DECLARE(oslog_msgbuf_dropped_msgcount);
4923 extern uint32_t oslog_msgbuf_dropped_charcount;
4924
4925 #if CONFIG_EXCLAVES
4926 /* log message counters for exclaves logging */
4927 SCALABLE_COUNTER_DECLARE(oslog_e_log_count);
4928 SCALABLE_COUNTER_DECLARE(oslog_e_log_dropped_count);
4929 SCALABLE_COUNTER_DECLARE(oslog_e_metadata_count);
4930 SCALABLE_COUNTER_DECLARE(oslog_e_metadata_dropped_count);
4931 SCALABLE_COUNTER_DECLARE(oslog_e_signpost_count);
4932 SCALABLE_COUNTER_DECLARE(oslog_e_signpost_dropped_count);
4933 SCALABLE_COUNTER_DECLARE(oslog_e_query_count);
4934 SCALABLE_COUNTER_DECLARE(oslog_e_query_error_count);
4935 SCALABLE_COUNTER_DECLARE(oslog_e_trace_mode_set_count);
4936 SCALABLE_COUNTER_DECLARE(oslog_e_trace_mode_error_count);
4937 #endif // CONFIG_EXCLAVES
4938
4939 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_total_msgcount, oslog_p_total_msgcount, "");
4940 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_metadata_saved_msgcount, oslog_p_metadata_saved_msgcount, "");
4941 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_metadata_dropped_msgcount, oslog_p_metadata_dropped_msgcount, "");
4942 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_signpost_saved_msgcount, oslog_p_signpost_saved_msgcount, "");
4943 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_signpost_dropped_msgcount, oslog_p_signpost_dropped_msgcount, "");
4944 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_error_count, oslog_p_error_count, "");
4945 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_saved_msgcount, oslog_p_saved_msgcount, "");
4946 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_dropped_msgcount, oslog_p_dropped_msgcount, "");
4947 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_boot_dropped_msgcount, oslog_p_boot_dropped_msgcount, "");
4948 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_coprocessor_total_msgcount, oslog_p_coprocessor_total_msgcount, "");
4949 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_coprocessor_dropped_msgcount, oslog_p_coprocessor_dropped_msgcount, "");
4950 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_unresolved_kc_msgcount, oslog_p_unresolved_kc_msgcount, "");
4951
4952 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_fmt_invalid_msgcount, oslog_p_fmt_invalid_msgcount, "");
4953 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_fmt_max_args_msgcount, oslog_p_fmt_max_args_msgcount, "");
4954 SYSCTL_SCALABLE_COUNTER(_debug, oslog_p_truncated_msgcount, oslog_p_truncated_msgcount, "");
4955
4956 SYSCTL_SCALABLE_COUNTER(_debug, oslog_s_total_msgcount, oslog_s_total_msgcount, "Number of logs sent to streaming");
4957 SYSCTL_SCALABLE_COUNTER(_debug, oslog_s_metadata_msgcount, oslog_s_metadata_msgcount, "Number of metadata sent to streaming");
4958 SYSCTL_SCALABLE_COUNTER(_debug, oslog_s_error_count, oslog_s_error_count, "Number of invalid stream logs");
4959 SYSCTL_SCALABLE_COUNTER(_debug, oslog_s_streamed_msgcount, oslog_s_streamed_msgcount, "Number of streamed logs");
4960 SYSCTL_SCALABLE_COUNTER(_debug, oslog_s_dropped_msgcount, oslog_s_dropped_msgcount, "Number of logs dropped from stream");
4961
4962 SYSCTL_SCALABLE_COUNTER(_debug, oslog_msgbuf_msgcount, oslog_msgbuf_msgcount, "Number of dmesg log messages");
4963 SYSCTL_SCALABLE_COUNTER(_debug, oslog_msgbuf_dropped_msgcount, oslog_msgbuf_dropped_msgcount, "Number of dropped dmesg log messages");
4964 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");
4965
4966 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_received, log_queue_cnt_received, "Number of received logs");
4967 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_rejected_fh, log_queue_cnt_rejected_fh, "Number of logs initially rejected by FH");
4968 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_sent, log_queue_cnt_sent, "Number of logs successfully saved in FH");
4969 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_dropped_nomem, log_queue_cnt_dropped_nomem, "Number of logs dropped due to lack of queue memory");
4970 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_queued, log_queue_cnt_queued, "Current number of logs stored in log queues");
4971 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_dropped_off, log_queue_cnt_dropped_off, "Number of logs dropped due to disabled log queues");
4972 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_mem_allocated, log_queue_cnt_mem_allocated, "Number of memory allocations");
4973 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_mem_released, log_queue_cnt_mem_released, "Number of memory releases");
4974 SYSCTL_SCALABLE_COUNTER(_debug, log_queue_cnt_mem_failed, log_queue_cnt_mem_failed, "Number of failed memory allocations");
4975
4976 SYSCTL_SCALABLE_COUNTER(_debug, oslog_subsystem_count, oslog_subsystem_count, "Number of registered log subsystems");
4977 SYSCTL_SCALABLE_COUNTER(_debug, oslog_subsystem_found, oslog_subsystem_found, "Number of sucessful log subsystem lookups");
4978 SYSCTL_SCALABLE_COUNTER(_debug, oslog_subsystem_dropped, oslog_subsystem_dropped, "Number of dropped log subsystem registrations");
4979
4980 #if CONFIG_EXCLAVES
4981 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_metadata_count, oslog_e_metadata_count,
4982 "Number of metadata messages retrieved from the exclaves log server");
4983 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_metadata_dropped_count, oslog_e_metadata_dropped_count,
4984 "Number of dropped metadata messages retrieved from the exclaves log server");
4985 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_log_count, oslog_e_log_count,
4986 "Number of logs retrieved from the exclaves log server");
4987 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_log_dropped_count, oslog_e_log_dropped_count,
4988 "Number of dropeed logs retrieved from the exclaves log server");
4989 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_signpost_count, oslog_e_signpost_count,
4990 "Number of signposts retrieved from the exclaves log server");
4991 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_signpost_dropped_count, oslog_e_signpost_dropped_count,
4992 "Number of dropped signposts retrieved from the exclaves log server");
4993 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_query_count, oslog_e_query_count,
4994 "Number of sucessful queries to the exclaves log server");
4995 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_query_error_count, oslog_e_query_error_count,
4996 "Number of failed queries to the exclaves log server");
4997 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_trace_mode_set_count, oslog_e_trace_mode_set_count,
4998 "Number of exclaves trace mode updates");
4999 SYSCTL_SCALABLE_COUNTER(_debug, oslog_e_trace_mode_error_count, oslog_e_trace_mode_error_count,
5000 "Number of failed exclaves trace mode updates");
5001 #endif // CONFIG_EXCLAVES
5002
5003 #endif /* DEVELOPMENT || DEBUG */
5004
5005 /*
5006 * Enable tracing of voucher contents
5007 */
5008 extern uint32_t ipc_voucher_trace_contents;
5009
5010 SYSCTL_INT(_kern, OID_AUTO, ipc_voucher_trace_contents,
5011 CTLFLAG_RW | CTLFLAG_LOCKED, &ipc_voucher_trace_contents, 0, "Enable tracing voucher contents");
5012
5013 /*
5014 * Kernel stack size and depth
5015 */
5016 SYSCTL_INT(_kern, OID_AUTO, stack_size,
5017 CTLFLAG_RD | CTLFLAG_LOCKED, (int *) &kernel_stack_size, 0, "Kernel stack size");
5018 SYSCTL_INT(_kern, OID_AUTO, stack_depth_max,
5019 CTLFLAG_RD | CTLFLAG_LOCKED, (int *) &kernel_stack_depth_max, 0, "Max kernel stack depth at interrupt or context switch");
5020
5021 extern unsigned int kern_feature_overrides;
5022 SYSCTL_INT(_kern, OID_AUTO, kern_feature_overrides,
5023 CTLFLAG_RD | CTLFLAG_LOCKED, &kern_feature_overrides, 0, "Kernel feature override mask");
5024
5025 /*
5026 * enable back trace for port allocations
5027 */
5028 extern int ipc_portbt;
5029
5030 SYSCTL_INT(_kern, OID_AUTO, ipc_portbt,
5031 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
5032 &ipc_portbt, 0, "");
5033
5034 /*
5035 * Scheduler sysctls
5036 */
5037
5038 SYSCTL_STRING(_kern, OID_AUTO, sched,
5039 CTLFLAG_RD | CTLFLAG_KERN | CTLFLAG_LOCKED,
5040 sched_string, sizeof(sched_string),
5041 "Timeshare scheduler implementation");
5042
5043 static int
5044 sysctl_cpu_quiescent_counter_interval SYSCTL_HANDLER_ARGS
5045 {
5046 #pragma unused(arg1, arg2)
5047
5048 uint32_t local_min_interval_us = smr_cpu_checkin_get_min_interval_us();
5049
5050 int error = sysctl_handle_int(oidp, &local_min_interval_us, 0, req);
5051 if (error || !req->newptr) {
5052 return error;
5053 }
5054
5055 smr_cpu_checkin_set_min_interval_us(local_min_interval_us);
5056
5057 return 0;
5058 }
5059
5060 SYSCTL_PROC(_kern, OID_AUTO, cpu_checkin_interval,
5061 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
5062 0, 0,
5063 sysctl_cpu_quiescent_counter_interval, "I",
5064 "Quiescent CPU checkin interval (microseconds)");
5065
5066 /*
5067 * Allow the precise user/kernel time sysctl to be set, but don't allow it to
5068 * affect anything. Some tools expect to be able to set this, even though
5069 * runtime configuration is no longer supported.
5070 */
5071
5072 static int
5073 sysctl_precise_user_kernel_time SYSCTL_HANDLER_ARGS
5074 {
5075 #if PRECISE_USER_KERNEL_TIME
5076 int dummy_set = 1;
5077 #else /* PRECISE_USER_KERNEL_TIME */
5078 int dummy_set = 0;
5079 #endif /* !PRECISE_USER_KERNEL_TIME */
5080 return sysctl_handle_int(oidp, &dummy_set, 0, req);
5081 }
5082
5083 SYSCTL_PROC(_kern, OID_AUTO, precise_user_kernel_time,
5084 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
5085 0, 0, sysctl_precise_user_kernel_time, "I",
5086 "Precise accounting of kernel vs. user time (deprecated)");
5087
5088 #if CONFIG_PERVASIVE_ENERGY && HAS_CPU_DPE_COUNTER
5089 __security_const_late static int pervasive_energy = 1;
5090 #else /* CONFIG_PERVASIVE_ENERGY && HAS_CPU_DPE_COUNTER */
5091 __security_const_late static int pervasive_energy = 0;
5092 #endif /* !CONFIG_PERVASIVE_ENERGY || !HAS_CPU_DPE_COUNTER */
5093
5094 SYSCTL_INT(_kern, OID_AUTO, pervasive_energy,
5095 CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED, &pervasive_energy, 0, "");
5096
5097 /* Parameters related to timer coalescing tuning, to be replaced
5098 * with a dedicated systemcall in the future.
5099 */
5100 /* Enable processing pending timers in the context of any other interrupt
5101 * Coalescing tuning parameters for various thread/task attributes */
5102 STATIC int
5103 sysctl_timer_user_us_kernel_abstime SYSCTL_HANDLER_ARGS
5104 {
5105 #pragma unused(oidp)
5106 int size = arg2; /* subcommand*/
5107 int error;
5108 int changed = 0;
5109 uint64_t old_value_ns;
5110 uint64_t new_value_ns;
5111 uint64_t value_abstime;
5112 if (size == sizeof(uint32_t)) {
5113 value_abstime = *((uint32_t *)arg1);
5114 } else if (size == sizeof(uint64_t)) {
5115 value_abstime = *((uint64_t *)arg1);
5116 } else {
5117 return ENOTSUP;
5118 }
5119
5120 absolutetime_to_nanoseconds(value_abstime, &old_value_ns);
5121 error = sysctl_io_number(req, old_value_ns, sizeof(old_value_ns), &new_value_ns, &changed);
5122 if ((error) || (!changed)) {
5123 return error;
5124 }
5125
5126 nanoseconds_to_absolutetime(new_value_ns, &value_abstime);
5127 if (size == sizeof(uint32_t)) {
5128 *((uint32_t *)arg1) = (uint32_t)value_abstime;
5129 } else {
5130 *((uint64_t *)arg1) = value_abstime;
5131 }
5132 return error;
5133 }
5134
5135 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_bg_scale,
5136 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5137 &tcoal_prio_params.timer_coalesce_bg_shift, 0, "");
5138 SYSCTL_PROC(_kern, OID_AUTO, timer_resort_threshold_ns,
5139 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5140 &tcoal_prio_params.timer_resort_threshold_abstime,
5141 sizeof(tcoal_prio_params.timer_resort_threshold_abstime),
5142 sysctl_timer_user_us_kernel_abstime,
5143 "Q", "");
5144 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_bg_ns_max,
5145 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5146 &tcoal_prio_params.timer_coalesce_bg_abstime_max,
5147 sizeof(tcoal_prio_params.timer_coalesce_bg_abstime_max),
5148 sysctl_timer_user_us_kernel_abstime,
5149 "Q", "");
5150
5151 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_kt_scale,
5152 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5153 &tcoal_prio_params.timer_coalesce_kt_shift, 0, "");
5154
5155 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_kt_ns_max,
5156 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5157 &tcoal_prio_params.timer_coalesce_kt_abstime_max,
5158 sizeof(tcoal_prio_params.timer_coalesce_kt_abstime_max),
5159 sysctl_timer_user_us_kernel_abstime,
5160 "Q", "");
5161
5162 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_fp_scale,
5163 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5164 &tcoal_prio_params.timer_coalesce_fp_shift, 0, "");
5165
5166 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_fp_ns_max,
5167 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5168 &tcoal_prio_params.timer_coalesce_fp_abstime_max,
5169 sizeof(tcoal_prio_params.timer_coalesce_fp_abstime_max),
5170 sysctl_timer_user_us_kernel_abstime,
5171 "Q", "");
5172
5173 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_ts_scale,
5174 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5175 &tcoal_prio_params.timer_coalesce_ts_shift, 0, "");
5176
5177 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_ts_ns_max,
5178 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5179 &tcoal_prio_params.timer_coalesce_ts_abstime_max,
5180 sizeof(tcoal_prio_params.timer_coalesce_ts_abstime_max),
5181 sysctl_timer_user_us_kernel_abstime,
5182 "Q", "");
5183
5184 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier0_scale,
5185 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5186 &tcoal_prio_params.latency_qos_scale[0], 0, "");
5187
5188 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier0_ns_max,
5189 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5190 &tcoal_prio_params.latency_qos_abstime_max[0],
5191 sizeof(tcoal_prio_params.latency_qos_abstime_max[0]),
5192 sysctl_timer_user_us_kernel_abstime,
5193 "Q", "");
5194
5195 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier1_scale,
5196 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5197 &tcoal_prio_params.latency_qos_scale[1], 0, "");
5198
5199 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier1_ns_max,
5200 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5201 &tcoal_prio_params.latency_qos_abstime_max[1],
5202 sizeof(tcoal_prio_params.latency_qos_abstime_max[1]),
5203 sysctl_timer_user_us_kernel_abstime,
5204 "Q", "");
5205
5206 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier2_scale,
5207 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5208 &tcoal_prio_params.latency_qos_scale[2], 0, "");
5209
5210 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier2_ns_max,
5211 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5212 &tcoal_prio_params.latency_qos_abstime_max[2],
5213 sizeof(tcoal_prio_params.latency_qos_abstime_max[2]),
5214 sysctl_timer_user_us_kernel_abstime,
5215 "Q", "");
5216
5217 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier3_scale,
5218 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5219 &tcoal_prio_params.latency_qos_scale[3], 0, "");
5220
5221 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier3_ns_max,
5222 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5223 &tcoal_prio_params.latency_qos_abstime_max[3],
5224 sizeof(tcoal_prio_params.latency_qos_abstime_max[3]),
5225 sysctl_timer_user_us_kernel_abstime,
5226 "Q", "");
5227
5228 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier4_scale,
5229 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5230 &tcoal_prio_params.latency_qos_scale[4], 0, "");
5231
5232 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier4_ns_max,
5233 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5234 &tcoal_prio_params.latency_qos_abstime_max[4],
5235 sizeof(tcoal_prio_params.latency_qos_abstime_max[4]),
5236 sysctl_timer_user_us_kernel_abstime,
5237 "Q", "");
5238
5239 SYSCTL_INT(_kern, OID_AUTO, timer_coalesce_tier5_scale,
5240 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5241 &tcoal_prio_params.latency_qos_scale[5], 0, "");
5242
5243 SYSCTL_PROC(_kern, OID_AUTO, timer_coalesce_tier5_ns_max,
5244 CTLTYPE_QUAD | CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5245 &tcoal_prio_params.latency_qos_abstime_max[5],
5246 sizeof(tcoal_prio_params.latency_qos_abstime_max[5]),
5247 sysctl_timer_user_us_kernel_abstime,
5248 "Q", "");
5249
5250 /* Communicate the "user idle level" heuristic to the timer layer, and
5251 * potentially other layers in the future.
5252 */
5253
5254 static int
5255 timer_user_idle_level(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
5256 {
5257 int new_value = 0, old_value = 0, changed = 0, error;
5258
5259 old_value = timer_get_user_idle_level();
5260
5261 error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
5262
5263 if (error == 0 && changed) {
5264 if (timer_set_user_idle_level(new_value) != KERN_SUCCESS) {
5265 error = ERANGE;
5266 }
5267 }
5268
5269 return error;
5270 }
5271
5272 SYSCTL_PROC(_machdep, OID_AUTO, user_idle_level,
5273 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
5274 0, 0,
5275 timer_user_idle_level, "I", "User idle level heuristic, 0-128");
5276
5277 #if DEVELOPMENT || DEBUG
5278 /*
5279 * Basic console mode for games; used for development purposes only.
5280 * Final implementation for this feature (with possible removal of
5281 * sysctl) tracked via rdar://101215873.
5282 */
5283 static int console_mode = 0;
5284 SYSCTL_INT(_kern, OID_AUTO, console_mode,
5285 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_MASKED,
5286 &console_mode, 0, "Game Console Mode");
5287 #endif /* DEVELOPMENT || DEBUG */
5288
5289
5290 #if HYPERVISOR
5291 SYSCTL_INT(_kern, OID_AUTO, hv_support,
5292 CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
5293 &hv_support_available, 0, "");
5294
5295 SYSCTL_INT(_kern, OID_AUTO, hv_disable,
5296 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5297 &hv_disable, 0, "");
5298
5299 #endif /* HYPERVISOR */
5300
5301 #if DEVELOPMENT || DEBUG
5302 extern uint64_t driverkit_checkin_timed_out;
5303 SYSCTL_QUAD(_kern, OID_AUTO, driverkit_checkin_timed_out,
5304 CTLFLAG_RW | CTLFLAG_KERN | CTLFLAG_LOCKED,
5305 &driverkit_checkin_timed_out, "timestamp of dext checkin timeout");
5306 #endif
5307
5308 #if CONFIG_DARKBOOT
5309 STATIC int
5310 sysctl_darkboot SYSCTL_HANDLER_ARGS
5311 {
5312 int err = 0, value = 0;
5313 #pragma unused(oidp, arg1, arg2, err, value, req)
5314
5315 /*
5316 * Handle the sysctl request.
5317 *
5318 * If this is a read, the function will set the value to the current darkboot value. Otherwise,
5319 * we'll get the request identifier into "value" and then we can honor it.
5320 */
5321 if ((err = sysctl_io_number(req, darkboot, sizeof(int), &value, NULL)) != 0) {
5322 goto exit;
5323 }
5324
5325 /* writing requested, let's process the request */
5326 if (req->newptr) {
5327 /* writing is protected by an entitlement */
5328 if (priv_check_cred(kauth_cred_get(), PRIV_DARKBOOT, 0) != 0) {
5329 err = EPERM;
5330 goto exit;
5331 }
5332
5333 switch (value) {
5334 case MEMORY_MAINTENANCE_DARK_BOOT_UNSET:
5335 /*
5336 * If the darkboot sysctl is unset, the NVRAM variable
5337 * must be unset too. If that's not the case, it means
5338 * someone is doing something crazy and not supported.
5339 */
5340 if (darkboot != 0) {
5341 int ret = PERemoveNVRAMProperty(MEMORY_MAINTENANCE_DARK_BOOT_NVRAM_NAME);
5342 if (ret) {
5343 darkboot = 0;
5344 } else {
5345 err = EINVAL;
5346 }
5347 }
5348 break;
5349 case MEMORY_MAINTENANCE_DARK_BOOT_SET:
5350 darkboot = 1;
5351 break;
5352 case MEMORY_MAINTENANCE_DARK_BOOT_SET_PERSISTENT: {
5353 /*
5354 * Set the NVRAM and update 'darkboot' in case
5355 * of success. Otherwise, do not update
5356 * 'darkboot' and report the failure.
5357 */
5358 if (PEWriteNVRAMBooleanProperty(MEMORY_MAINTENANCE_DARK_BOOT_NVRAM_NAME, TRUE)) {
5359 darkboot = 1;
5360 } else {
5361 err = EINVAL;
5362 }
5363
5364 break;
5365 }
5366 default:
5367 err = EINVAL;
5368 }
5369 }
5370
5371 exit:
5372 return err;
5373 }
5374
5375 SYSCTL_PROC(_kern, OID_AUTO, darkboot,
5376 CTLFLAG_KERN | CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_ANYBODY,
5377 0, 0, sysctl_darkboot, "I", "");
5378 #endif /* CONFIG_DARKBOOT */
5379
5380 #if DEVELOPMENT || DEBUG
5381 #include <sys/sysent.h>
5382 /* This should result in a fatal exception, verifying that "sysent" is
5383 * write-protected.
5384 */
5385 static int
5386 kern_sysent_write(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
5387 {
5388 uint64_t new_value = 0, old_value = 0;
5389 int changed = 0, error;
5390
5391 error = sysctl_io_number(req, old_value, sizeof(uint64_t), &new_value, &changed);
5392 if ((error == 0) && changed) {
5393 volatile uint32_t *wraddr = __DECONST(uint32_t *, &sysent[0]);
5394 *wraddr = 0;
5395 printf("sysent[0] write succeeded\n");
5396 }
5397 return error;
5398 }
5399
5400 SYSCTL_PROC(_kern, OID_AUTO, sysent_const_check,
5401 CTLTYPE_QUAD | CTLFLAG_RW | CTLFLAG_LOCKED,
5402 0, 0,
5403 kern_sysent_write, "I", "Attempt sysent[0] write");
5404
5405 #endif
5406
5407 #if DEVELOPMENT || DEBUG
5408 SYSCTL_COMPAT_INT(_kern, OID_AUTO, development, CTLFLAG_RD | CTLFLAG_MASKED | CTLFLAG_KERN, NULL, 1, "");
5409 #else
5410 SYSCTL_COMPAT_INT(_kern, OID_AUTO, development, CTLFLAG_RD | CTLFLAG_MASKED, NULL, 0, "");
5411 #endif
5412
5413 SYSCTL_INT(_kern, OID_AUTO, serverperfmode, CTLFLAG_RD, &serverperfmode, 0, "");
5414
5415 #if DEVELOPMENT || DEBUG
5416
5417 decl_lck_spin_data(, spinlock_panic_test_lock);
5418
5419 __attribute__((noreturn))
5420 static void
5421 spinlock_panic_test_acquire_spinlock(void * arg __unused, wait_result_t wres __unused)
5422 {
5423 lck_spin_lock(&spinlock_panic_test_lock);
5424 while (1) {
5425 ;
5426 }
5427 }
5428
5429 static int
5430 sysctl_spinlock_panic_test SYSCTL_HANDLER_ARGS
5431 {
5432 #pragma unused(oidp, arg1, arg2)
5433 if (req->newlen == 0) {
5434 return EINVAL;
5435 }
5436
5437 thread_t panic_spinlock_thread;
5438 /* Initialize panic spinlock */
5439 lck_grp_t * panic_spinlock_grp;
5440 lck_grp_attr_t * panic_spinlock_grp_attr;
5441 lck_attr_t * panic_spinlock_attr;
5442
5443 panic_spinlock_grp_attr = lck_grp_attr_alloc_init();
5444 panic_spinlock_grp = lck_grp_alloc_init("panic_spinlock", panic_spinlock_grp_attr);
5445 panic_spinlock_attr = lck_attr_alloc_init();
5446
5447 lck_spin_init(&spinlock_panic_test_lock, panic_spinlock_grp, panic_spinlock_attr);
5448
5449
5450 /* Create thread to acquire spinlock */
5451 if (kernel_thread_start(spinlock_panic_test_acquire_spinlock, NULL, &panic_spinlock_thread) != KERN_SUCCESS) {
5452 return EBUSY;
5453 }
5454
5455 /* Try to acquire spinlock -- should panic eventually */
5456 lck_spin_lock(&spinlock_panic_test_lock);
5457 while (1) {
5458 ;
5459 }
5460 }
5461
5462 __attribute__((noreturn))
5463 static void
5464 simultaneous_panic_worker
5465 (void * arg, wait_result_t wres __unused)
5466 {
5467 atomic_int *start_panic = (atomic_int *)arg;
5468
5469 while (!atomic_load(start_panic)) {
5470 ;
5471 }
5472 panic("SIMULTANEOUS PANIC TEST: INITIATING PANIC FROM CPU %d", cpu_number());
5473 __builtin_unreachable();
5474 }
5475
5476 static int
5477 sysctl_simultaneous_panic_test SYSCTL_HANDLER_ARGS
5478 {
5479 #pragma unused(oidp, arg1, arg2)
5480 if (req->newlen == 0) {
5481 return EINVAL;
5482 }
5483
5484 int i = 0, threads_to_create = 2 * processor_count;
5485 atomic_int start_panic = 0;
5486 unsigned int threads_created = 0;
5487 thread_t new_panic_thread;
5488
5489 for (i = threads_to_create; i > 0; i--) {
5490 if (kernel_thread_start(simultaneous_panic_worker, (void *) &start_panic, &new_panic_thread) == KERN_SUCCESS) {
5491 threads_created++;
5492 }
5493 }
5494
5495 /* FAIL if we couldn't create at least processor_count threads */
5496 if (threads_created < processor_count) {
5497 panic("SIMULTANEOUS PANIC TEST: FAILED TO CREATE ENOUGH THREADS, ONLY CREATED %d (of %d)",
5498 threads_created, threads_to_create);
5499 }
5500
5501 atomic_exchange(&start_panic, 1);
5502 while (1) {
5503 ;
5504 }
5505 }
5506
5507 extern unsigned int panic_test_failure_mode;
5508 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");
5509
5510 extern unsigned int panic_test_action_count;
5511 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");
5512
5513 extern unsigned int panic_test_case;
5514 SYSCTL_INT(_debug, OID_AUTO, xnu_panic_test_case, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_KERN, &panic_test_case, 0, "panic/debugger testcase");
5515
5516 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");
5517 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");
5518
5519 extern int exc_resource_threads_enabled;
5520 SYSCTL_INT(_kern, OID_AUTO, exc_resource_threads_enabled, CTLFLAG_RW | CTLFLAG_LOCKED, &exc_resource_threads_enabled, 0, "exc_resource thread limit enabled");
5521
5522 extern unsigned int verbose_panic_flow_logging;
5523 SYSCTL_INT(_debug, OID_AUTO, verbose_panic_flow_logging, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_KERN, &verbose_panic_flow_logging, 0, "verbose logging during panic");
5524
5525 #endif /* DEVELOPMENT || DEBUG */
5526
5527 #if BUILT_LTO
5528 static int _built_lto = 1;
5529 #else // BUILT_LTO
5530 static int _built_lto = 0;
5531 #endif // !BUILT_LTO
5532
5533 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");
5534
5535 #if CONFIG_THREAD_GROUPS
5536 #if DEVELOPMENT || DEBUG
5537
5538 static int
5539 sysctl_get_thread_group_id SYSCTL_HANDLER_ARGS
5540 {
5541 #pragma unused(arg1, arg2, oidp)
5542 uint64_t thread_group_id = thread_group_get_id(thread_group_get(current_thread()));
5543 return SYSCTL_OUT(req, &thread_group_id, sizeof(thread_group_id));
5544 }
5545
5546 SYSCTL_PROC(_kern, OID_AUTO, thread_group_id, CTLFLAG_RD | CTLFLAG_LOCKED | CTLTYPE_QUAD,
5547 0, 0, &sysctl_get_thread_group_id, "I", "thread group id of the thread");
5548
5549 STATIC int
5550 sysctl_thread_group_count(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
5551 {
5552 int value = thread_group_count();
5553 return sysctl_io_number(req, value, sizeof(value), NULL, NULL);
5554 }
5555
5556 SYSCTL_PROC(_kern, OID_AUTO, thread_group_count, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5557 0, 0, &sysctl_thread_group_count, "I", "count of thread groups");
5558
5559 #endif /* DEVELOPMENT || DEBUG */
5560 const uint32_t thread_groups_supported = 1;
5561 #else /* CONFIG_THREAD_GROUPS */
5562 const uint32_t thread_groups_supported = 0;
5563 #endif /* CONFIG_THREAD_GROUPS */
5564
5565 STATIC int
5566 sysctl_thread_groups_supported(__unused struct sysctl_oid *oidp, __unused void *arg1, __unused int arg2, struct sysctl_req *req)
5567 {
5568 int value = thread_groups_supported;
5569 return sysctl_io_number(req, value, sizeof(value), NULL, NULL);
5570 }
5571
5572 SYSCTL_PROC(_kern, OID_AUTO, thread_groups_supported, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5573 0, 0, &sysctl_thread_groups_supported, "I", "thread groups supported");
5574
5575 static int
5576 sysctl_grade_cputype SYSCTL_HANDLER_ARGS
5577 {
5578 #pragma unused(arg1, arg2, oidp)
5579 int error = 0;
5580 int type_tuple[2] = {};
5581 int return_value = 0;
5582
5583 error = SYSCTL_IN(req, &type_tuple, sizeof(type_tuple));
5584
5585 if (error) {
5586 return error;
5587 }
5588
5589 return_value = grade_binary(type_tuple[0], type_tuple[1] & ~CPU_SUBTYPE_MASK, type_tuple[1] & CPU_SUBTYPE_MASK, FALSE);
5590
5591 error = SYSCTL_OUT(req, &return_value, sizeof(return_value));
5592
5593 if (error) {
5594 return error;
5595 }
5596
5597 return error;
5598 }
5599
5600 SYSCTL_PROC(_kern, OID_AUTO, grade_cputype,
5601 CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_MASKED | CTLFLAG_LOCKED | CTLTYPE_OPAQUE,
5602 0, 0, &sysctl_grade_cputype, "S",
5603 "grade value of cpu_type_t+cpu_sub_type_t");
5604
5605 extern boolean_t allow_direct_handoff;
5606 SYSCTL_INT(_kern, OID_AUTO, direct_handoff,
5607 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5608 &allow_direct_handoff, 0, "Enable direct handoff for realtime threads");
5609
5610 #if DEVELOPMENT || DEBUG
5611
5612 SYSCTL_QUAD(_kern, OID_AUTO, phys_carveout_pa, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5613 &phys_carveout_pa,
5614 "base physical address of the phys_carveout_mb boot-arg region");
5615 SYSCTL_QUAD(_kern, OID_AUTO, phys_carveout_va, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5616 &phys_carveout,
5617 "base virtual address of the phys_carveout_mb boot-arg region");
5618 SYSCTL_QUAD(_kern, OID_AUTO, phys_carveout_size, CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
5619 &phys_carveout_size,
5620 "size in bytes of the phys_carveout_mb boot-arg region");
5621
5622
5623
5624 static int
5625 cseg_wedge_thread SYSCTL_HANDLER_ARGS
5626 {
5627 #pragma unused(arg1, arg2)
5628
5629 int error, val = 0;
5630 error = sysctl_handle_int(oidp, &val, 0, req);
5631 if (error || val == 0) {
5632 return error;
5633 }
5634
5635 do_cseg_wedge_thread();
5636 return 0;
5637 }
5638 SYSCTL_PROC(_kern, OID_AUTO, cseg_wedge_thread, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_MASKED, 0, 0, cseg_wedge_thread, "I", "wedge c_seg thread");
5639
5640 static int
5641 cseg_unwedge_thread SYSCTL_HANDLER_ARGS
5642 {
5643 #pragma unused(arg1, arg2)
5644
5645 int error, val = 0;
5646 error = sysctl_handle_int(oidp, &val, 0, req);
5647 if (error || val == 0) {
5648 return error;
5649 }
5650
5651 do_cseg_unwedge_thread();
5652 return 0;
5653 }
5654 SYSCTL_PROC(_kern, OID_AUTO, cseg_unwedge_thread, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_MASKED, 0, 0, cseg_unwedge_thread, "I", "unstuck c_seg thread");
5655
5656 static atomic_int wedge_thread_should_wake = 0;
5657
5658 static int
5659 unwedge_thread SYSCTL_HANDLER_ARGS
5660 {
5661 #pragma unused(arg1, arg2)
5662 int error, val = 0;
5663 error = sysctl_handle_int(oidp, &val, 0, req);
5664 if (error || val == 0) {
5665 return error;
5666 }
5667
5668 atomic_store(&wedge_thread_should_wake, 1);
5669 return 0;
5670 }
5671
5672 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");
5673
5674 static int
5675 wedge_thread SYSCTL_HANDLER_ARGS
5676 {
5677 #pragma unused(arg1, arg2)
5678
5679 int error, val = 0;
5680 error = sysctl_handle_int(oidp, &val, 0, req);
5681 if (error || val == 0) {
5682 return error;
5683 }
5684
5685 uint64_t interval = 1;
5686 nanoseconds_to_absolutetime(1000 * 1000 * 50, &interval);
5687
5688 atomic_store(&wedge_thread_should_wake, 0);
5689 while (!atomic_load(&wedge_thread_should_wake)) {
5690 tsleep1(NULL, 0, "wedge_thread", mach_absolute_time() + interval, NULL);
5691 }
5692
5693 return 0;
5694 }
5695
5696 SYSCTL_PROC(_kern, OID_AUTO, wedge_thread,
5697 CTLFLAG_RW | CTLFLAG_ANYBODY | CTLFLAG_LOCKED, 0, 0, wedge_thread, "I",
5698 "wedge this thread so it cannot be cleaned up");
5699
5700 static int
5701 sysctl_total_corpses_count SYSCTL_HANDLER_ARGS
5702 {
5703 #pragma unused(oidp, arg1, arg2)
5704 extern unsigned long total_corpses_count(void);
5705
5706 unsigned long corpse_count_long = total_corpses_count();
5707 unsigned int corpse_count = (unsigned int)MIN(corpse_count_long, UINT_MAX);
5708 return sysctl_io_opaque(req, &corpse_count, sizeof(corpse_count), NULL);
5709 }
5710
5711 SYSCTL_PROC(_kern, OID_AUTO, total_corpses_count,
5712 CTLFLAG_RD | CTLFLAG_ANYBODY | CTLFLAG_LOCKED, 0, 0,
5713 sysctl_total_corpses_count, "I", "total corpses on the system");
5714
5715 static int
5716 sysctl_turnstile_test_prim_lock SYSCTL_HANDLER_ARGS;
5717 static int
5718 sysctl_turnstile_test_prim_unlock SYSCTL_HANDLER_ARGS;
5719 int
5720 tstile_test_prim_lock(boolean_t use_hashtable);
5721 int
5722 tstile_test_prim_unlock(boolean_t use_hashtable);
5723
5724 static int
5725 sysctl_turnstile_test_prim_lock SYSCTL_HANDLER_ARGS
5726 {
5727 #pragma unused(arg1, arg2)
5728 int error, val = 0;
5729 error = sysctl_handle_int(oidp, &val, 0, req);
5730 if (error || val == 0) {
5731 return error;
5732 }
5733 switch (val) {
5734 case SYSCTL_TURNSTILE_TEST_USER_DEFAULT:
5735 case SYSCTL_TURNSTILE_TEST_USER_HASHTABLE:
5736 case SYSCTL_TURNSTILE_TEST_KERNEL_DEFAULT:
5737 case SYSCTL_TURNSTILE_TEST_KERNEL_HASHTABLE:
5738 return tstile_test_prim_lock(val);
5739 default:
5740 return error;
5741 }
5742 }
5743
5744 static int
5745 sysctl_turnstile_test_prim_unlock SYSCTL_HANDLER_ARGS
5746 {
5747 #pragma unused(arg1, arg2)
5748 int error, val = 0;
5749 error = sysctl_handle_int(oidp, &val, 0, req);
5750 if (error || val == 0) {
5751 return error;
5752 }
5753 switch (val) {
5754 case SYSCTL_TURNSTILE_TEST_USER_DEFAULT:
5755 case SYSCTL_TURNSTILE_TEST_USER_HASHTABLE:
5756 case SYSCTL_TURNSTILE_TEST_KERNEL_DEFAULT:
5757 case SYSCTL_TURNSTILE_TEST_KERNEL_HASHTABLE:
5758 return tstile_test_prim_unlock(val);
5759 default:
5760 return error;
5761 }
5762 }
5763
5764 SYSCTL_PROC(_kern, OID_AUTO, turnstiles_test_lock, CTLFLAG_WR | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED,
5765 0, 0, sysctl_turnstile_test_prim_lock, "I", "turnstiles test lock");
5766
5767 SYSCTL_PROC(_kern, OID_AUTO, turnstiles_test_unlock, CTLFLAG_WR | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED,
5768 0, 0, sysctl_turnstile_test_prim_unlock, "I", "turnstiles test unlock");
5769
5770 int
5771 turnstile_get_boost_stats_sysctl(void *req);
5772 int
5773 turnstile_get_unboost_stats_sysctl(void *req);
5774 static int
5775 sysctl_turnstile_boost_stats SYSCTL_HANDLER_ARGS;
5776 static int
5777 sysctl_turnstile_unboost_stats SYSCTL_HANDLER_ARGS;
5778 extern uint64_t thread_block_on_turnstile_count;
5779 extern uint64_t thread_block_on_regular_waitq_count;
5780
5781 static int
5782 sysctl_turnstile_boost_stats SYSCTL_HANDLER_ARGS
5783 {
5784 #pragma unused(arg1, arg2, oidp)
5785 return turnstile_get_boost_stats_sysctl(req);
5786 }
5787
5788 static int
5789 sysctl_turnstile_unboost_stats SYSCTL_HANDLER_ARGS
5790 {
5791 #pragma unused(arg1, arg2, oidp)
5792 return turnstile_get_unboost_stats_sysctl(req);
5793 }
5794
5795 SYSCTL_PROC(_kern, OID_AUTO, turnstile_boost_stats, CTLFLAG_RD | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED | CTLTYPE_STRUCT,
5796 0, 0, sysctl_turnstile_boost_stats, "S", "turnstiles boost stats");
5797 SYSCTL_PROC(_kern, OID_AUTO, turnstile_unboost_stats, CTLFLAG_RD | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED | CTLTYPE_STRUCT,
5798 0, 0, sysctl_turnstile_unboost_stats, "S", "turnstiles unboost stats");
5799 SYSCTL_QUAD(_kern, OID_AUTO, thread_block_count_on_turnstile,
5800 CTLFLAG_RD | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED,
5801 &thread_block_on_turnstile_count, "thread blocked on turnstile count");
5802 SYSCTL_QUAD(_kern, OID_AUTO, thread_block_count_on_reg_waitq,
5803 CTLFLAG_RD | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED,
5804 &thread_block_on_regular_waitq_count, "thread blocked on regular waitq count");
5805
5806 #if CONFIG_PV_TICKET
5807
5808 extern int ticket_lock_spins;
5809 SYSCTL_INT(_kern, OID_AUTO, ticket_lock_spins,
5810 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
5811 &ticket_lock_spins, 0, "loops before hypercall");
5812
5813 #if (DEBUG || DEVELOPMENT)
5814
5815 /* PV ticket lock stats */
5816
5817 SYSCTL_SCALABLE_COUNTER(_kern, ticket_lock_kicks, ticket_kick_count,
5818 "ticket lock kicks");
5819 SYSCTL_SCALABLE_COUNTER(_kern, ticket_lock_waits, ticket_wait_count,
5820 "ticket lock waits");
5821 SYSCTL_SCALABLE_COUNTER(_kern, ticket_lock_already, ticket_already_count,
5822 "ticket lock already unlocked");
5823 SYSCTL_SCALABLE_COUNTER(_kern, ticket_lock_just_unlock, ticket_just_unlock,
5824 "ticket unlock without kick");
5825 SYSCTL_SCALABLE_COUNTER(_kern, ticket_lock_wflag_cleared, ticket_wflag_cleared,
5826 "ticket lock wait flag cleared");
5827 SYSCTL_SCALABLE_COUNTER(_kern, ticket_lock_wflag_still, ticket_wflag_still,
5828 "ticket lock wait flag not cleared");
5829 SYSCTL_SCALABLE_COUNTER(_kern, ticket_lock_spin_count, ticket_spin_count,
5830 "ticket lock spin count");
5831
5832 /* sysctl kern.hcall_probe=n -- does hypercall #n exist? */
5833
5834 static int
5835 sysctl_hcall_probe SYSCTL_HANDLER_ARGS
5836 {
5837 char instr[20];
5838
5839 if (!req->newptr) {
5840 return 0;
5841 }
5842 if (req->newlen >= sizeof(instr)) {
5843 return EOVERFLOW;
5844 }
5845
5846 int error = SYSCTL_IN(req, instr, req->newlen);
5847 if (error) {
5848 return error;
5849 }
5850 instr[req->newlen] = '\0';
5851
5852 int hcall = 0;
5853 error = sscanf(instr, "%d", &hcall);
5854 if (error != 1 || hcall < 0) {
5855 return EINVAL;
5856 }
5857 uprintf("%savailable\n",
5858 hvg_is_hcall_available((hvg_hcall_code_t)hcall) ? "" : "not ");
5859 return 0;
5860 }
5861
5862 SYSCTL_PROC(_kern, OID_AUTO, hcall_probe,
5863 CTLTYPE_STRING | CTLFLAG_WR | CTLFLAG_LOCKED | CTLFLAG_MASKED,
5864 0, 0, sysctl_hcall_probe, "A", "probe hypercall by id");
5865
5866 #endif /* (DEBUG || DEVELOPMENT) */
5867 #endif /* CONFIG_PV_TICKET */
5868
5869 #if defined(__x86_64__)
5870 extern uint64_t MutexSpin;
5871
5872 SYSCTL_QUAD(_kern, OID_AUTO, mutex_spin_abs, CTLFLAG_RW, &MutexSpin,
5873 "Spin time in abs for acquiring a kernel mutex");
5874 #else
5875 extern machine_timeout_t MutexSpin;
5876
5877 SYSCTL_QUAD(_kern, OID_AUTO, mutex_spin_abs, CTLFLAG_RW, &MutexSpin,
5878 "Spin time in abs for acquiring a kernel mutex");
5879 #endif
5880
5881 extern uint64_t low_MutexSpin;
5882 extern int64_t high_MutexSpin;
5883 extern unsigned int real_ncpus;
5884
5885 SYSCTL_QUAD(_kern, OID_AUTO, low_mutex_spin_abs, CTLFLAG_RW, &low_MutexSpin,
5886 "Low spin threshold in abs for acquiring a kernel mutex");
5887
5888 static int
5889 sysctl_high_mutex_spin_ns SYSCTL_HANDLER_ARGS
5890 {
5891 #pragma unused(oidp, arg1, arg2)
5892 int error;
5893 int64_t val = 0;
5894 int64_t res;
5895
5896 /* Check if the user is writing to high_MutexSpin, or just reading it */
5897 if (req->newptr) {
5898 error = SYSCTL_IN(req, &val, sizeof(val));
5899 if (error || (val < 0 && val != -1)) {
5900 return error;
5901 }
5902 high_MutexSpin = val;
5903 }
5904
5905 if (high_MutexSpin >= 0) {
5906 res = high_MutexSpin;
5907 } else {
5908 res = low_MutexSpin * real_ncpus;
5909 }
5910 return SYSCTL_OUT(req, &res, sizeof(res));
5911 }
5912 SYSCTL_PROC(_kern, OID_AUTO, high_mutex_spin_abs, CTLFLAG_RW | CTLTYPE_QUAD, 0, 0, sysctl_high_mutex_spin_ns, "I",
5913 "High spin threshold in abs for acquiring a kernel mutex");
5914
5915 #if defined (__x86_64__)
5916
5917 semaphore_t sysctl_test_panic_with_thread_sem;
5918
5919 #pragma clang diagnostic push
5920 #pragma clang diagnostic ignored "-Winfinite-recursion" /* rdar://38801963 */
5921 __attribute__((noreturn))
5922 static void
5923 panic_thread_test_child_spin(void * arg, wait_result_t wres)
5924 {
5925 static int panic_thread_recurse_count = 5;
5926
5927 if (panic_thread_recurse_count > 0) {
5928 panic_thread_recurse_count--;
5929 panic_thread_test_child_spin(arg, wres);
5930 }
5931
5932 semaphore_signal(sysctl_test_panic_with_thread_sem);
5933 while (1) {
5934 ;
5935 }
5936 }
5937 #pragma clang diagnostic pop
5938
5939 static void
5940 panic_thread_test_child_park(void * arg __unused, wait_result_t wres __unused)
5941 {
5942 int event;
5943
5944 assert_wait(&event, THREAD_UNINT);
5945 semaphore_signal(sysctl_test_panic_with_thread_sem);
5946 thread_block(panic_thread_test_child_park);
5947 }
5948
5949 static int
5950 sysctl_test_panic_with_thread SYSCTL_HANDLER_ARGS
5951 {
5952 #pragma unused(arg1, arg2)
5953 int rval = 0;
5954 char str[16] = { '\0' };
5955 thread_t child_thread = THREAD_NULL;
5956
5957 rval = sysctl_handle_string(oidp, str, sizeof(str), req);
5958 if (rval != 0 || !req->newptr) {
5959 return EINVAL;
5960 }
5961
5962 semaphore_create(kernel_task, &sysctl_test_panic_with_thread_sem, SYNC_POLICY_FIFO, 0);
5963
5964 /* Create thread to spin or park in continuation */
5965 if (strncmp("spin", str, strlen("spin")) == 0) {
5966 if (kernel_thread_start(panic_thread_test_child_spin, NULL, &child_thread) != KERN_SUCCESS) {
5967 semaphore_destroy(kernel_task, sysctl_test_panic_with_thread_sem);
5968 return EBUSY;
5969 }
5970 } else if (strncmp("continuation", str, strlen("continuation")) == 0) {
5971 if (kernel_thread_start(panic_thread_test_child_park, NULL, &child_thread) != KERN_SUCCESS) {
5972 semaphore_destroy(kernel_task, sysctl_test_panic_with_thread_sem);
5973 return EBUSY;
5974 }
5975 } else {
5976 semaphore_destroy(kernel_task, sysctl_test_panic_with_thread_sem);
5977 return EINVAL;
5978 }
5979
5980 semaphore_wait(sysctl_test_panic_with_thread_sem);
5981
5982 panic_with_thread_context(0, NULL, 0, child_thread, "testing panic_with_thread_context for thread %p", child_thread);
5983
5984 /* Not reached */
5985 return EINVAL;
5986 }
5987
5988 SYSCTL_PROC(_kern, OID_AUTO, test_panic_with_thread,
5989 CTLFLAG_MASKED | CTLFLAG_KERN | CTLFLAG_LOCKED | CTLFLAG_WR | CTLTYPE_STRING,
5990 0, 0, sysctl_test_panic_with_thread, "A", "test panic flow for backtracing a different thread");
5991 #endif /* defined (__x86_64__) */
5992
5993 static int
5994 sysctl_generate_file_permissions_guard_exception SYSCTL_HANDLER_ARGS
5995 {
5996 #pragma unused(arg1, arg2)
5997 int error, val = 0;
5998 error = sysctl_handle_int(oidp, &val, 0, req);
5999 if (error || val == 0) {
6000 return error;
6001 }
6002 generate_file_permissions_guard_exception(0, val);
6003 return 0;
6004 }
6005
6006 SYSCTL_PROC(_kern, OID_AUTO, file_perm_guard_exception, CTLFLAG_WR | CTLFLAG_ANYBODY | CTLFLAG_KERN | CTLFLAG_LOCKED,
6007 0, 0, sysctl_generate_file_permissions_guard_exception, "I", "Test File Permission Guard exception");
6008
6009 #endif /* DEVELOPMENT || DEBUG */
6010
6011 extern const int copysize_limit_panic;
6012 static int
6013 sysctl_get_owned_vmobjects SYSCTL_HANDLER_ARGS
6014 {
6015 #pragma unused(oidp, arg1, arg2)
6016
6017 /* validate */
6018 if (req->newlen != sizeof(mach_port_name_t) || req->newptr == USER_ADDR_NULL ||
6019 req->oldidx != 0 || req->newidx != 0 || req->p == NULL ||
6020 (req->oldlen == 0 && req->oldptr != USER_ADDR_NULL)) {
6021 return EINVAL;
6022 }
6023
6024 int error;
6025 mach_port_name_t task_port_name;
6026 task_t task;
6027 size_t buffer_size = (req->oldptr != USER_ADDR_NULL) ? req->oldlen : 0;
6028 vmobject_list_output_t buffer = NULL;
6029 size_t output_size;
6030 size_t entries;
6031 bool free_buffer = false;
6032
6033 /* we have a "newptr" (for write) we get a task port name from the caller. */
6034 error = SYSCTL_IN(req, &task_port_name, sizeof(mach_port_name_t));
6035
6036 if (error != 0) {
6037 goto sysctl_get_vmobject_list_exit;
6038 }
6039
6040 task = port_name_to_task_read(task_port_name);
6041 if (task == TASK_NULL) {
6042 error = ESRCH;
6043 goto sysctl_get_vmobject_list_exit;
6044 }
6045
6046 bool corpse = task_is_a_corpse(task);
6047
6048 /* get the current size */
6049 size_t max_size;
6050 task_get_owned_vmobjects(task, 0, NULL, &max_size, &entries);
6051
6052 if (buffer_size && (buffer_size < sizeof(*buffer) + sizeof(vm_object_query_data_t))) {
6053 error = ENOMEM;
6054 goto sysctl_get_vmobject_list_deallocate_and_exit;
6055 }
6056
6057 if (corpse == false) {
6058 /* copy the vmobjects and vmobject data out of the task */
6059 if (buffer_size == 0) {
6060 output_size = max_size;
6061 } else {
6062 buffer_size = (buffer_size > max_size) ? max_size : buffer_size;
6063 buffer = (struct _vmobject_list_output_ *)kalloc_data(buffer_size, Z_WAITOK);
6064
6065 if (!buffer) {
6066 error = ENOMEM;
6067 goto sysctl_get_vmobject_list_deallocate_and_exit;
6068 }
6069 free_buffer = true;
6070
6071 task_get_owned_vmobjects(task, buffer_size, buffer, &output_size, &entries);
6072 }
6073 } else {
6074 vmobject_list_output_t list;
6075
6076 task_get_corpse_vmobject_list(task, &list, &max_size);
6077 assert(buffer == NULL);
6078
6079 /* copy corpse_vmobject_list to output buffer to avoid double copy */
6080 if (buffer_size) {
6081 size_t temp_size;
6082
6083 temp_size = buffer_size > max_size ? max_size : buffer_size;
6084 output_size = temp_size - sizeof(*buffer);
6085 /* whole multiple of vm_object_query_data_t */
6086 output_size = (output_size / sizeof(vm_object_query_data_t)) * sizeof(vm_object_query_data_t) + sizeof(*buffer);
6087 buffer = list;
6088 } else {
6089 output_size = max_size;
6090 }
6091 }
6092
6093 /* req->oldptr should be USER_ADDR_NULL if buffer == NULL and return the current size */
6094 /* otherwise copy buffer to oldptr and return the bytes copied */
6095 size_t num_copied, chunk_size;
6096 for (num_copied = 0, chunk_size = 0;
6097 num_copied < output_size;
6098 num_copied += chunk_size) {
6099 chunk_size = MIN(output_size - num_copied, copysize_limit_panic);
6100 error = SYSCTL_OUT(req, (char *)buffer + num_copied, chunk_size);
6101 if (error) {
6102 break;
6103 }
6104 }
6105
6106 sysctl_get_vmobject_list_deallocate_and_exit:
6107 task_deallocate(task);
6108
6109 sysctl_get_vmobject_list_exit:
6110 if (free_buffer) {
6111 kfree_data(buffer, buffer_size);
6112 }
6113
6114 return error;
6115 }
6116
6117 SYSCTL_PROC(_vm, OID_AUTO, get_owned_vmobjects,
6118 CTLTYPE_OPAQUE | CTLFLAG_RD | CTLFLAG_WR | CTLFLAG_MASKED | CTLFLAG_KERN | CTLFLAG_LOCKED | CTLFLAG_ANYBODY,
6119 0, 0, sysctl_get_owned_vmobjects, "A", "get owned vmobjects in task");
6120
6121 extern uint64_t num_static_scalable_counters;
6122 SYSCTL_QUAD(_kern, OID_AUTO, num_static_scalable_counters, CTLFLAG_RD | CTLFLAG_LOCKED, &num_static_scalable_counters, "");
6123
6124 #if SCHED_HYGIENE_DEBUG
6125 TUNABLE_DT(bool, sched_hygiene_nonspec_tb, "machine-timeouts", "nonspec-tb", "sched-hygiene-nonspec-tb", false, TUNABLE_DT_NONE);
6126 static SECURITY_READ_ONLY_LATE(int) sched_hygiene_debug_available = 1;
6127 #else
6128 static SECURITY_READ_ONLY_LATE(int) sched_hygiene_debug_available = 0;
6129 #endif /* SCHED_HYGIENE_DEBUG */
6130
6131 SYSCTL_INT(_debug, OID_AUTO, sched_hygiene_debug_available,
6132 CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
6133 &sched_hygiene_debug_available, 0, "");
6134
6135 uuid_string_t trial_treatment_id;
6136 uuid_string_t trial_experiment_id;
6137 int trial_deployment_id = -1;
6138
6139 SYSCTL_STRING(_kern, OID_AUTO, trial_treatment_id, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_ANYBODY | CTLFLAG_EXPERIMENT, trial_treatment_id, sizeof(trial_treatment_id), "");
6140 SYSCTL_STRING(_kern, OID_AUTO, trial_experiment_id, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_ANYBODY | CTLFLAG_EXPERIMENT, trial_experiment_id, sizeof(trial_experiment_id), "");
6141 SYSCTL_INT(_kern, OID_AUTO, trial_deployment_id, CTLFLAG_RW | CTLFLAG_LOCKED | CTLFLAG_ANYBODY | CTLFLAG_EXPERIMENT, &trial_deployment_id, 0, "");
6142
6143 #if (DEVELOPMENT || DEBUG)
6144 /* For unit testing setting factors & limits. */
6145 unsigned int testing_experiment_factor;
6146 EXPERIMENT_FACTOR_UINT(_kern, testing_experiment_factor, &testing_experiment_factor, 5, 10, "");
6147
6148 extern int exception_log_max_pid;
6149 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");
6150 #endif /* (DEVELOPMENT || DEBUG) */
6151
6152 #if DEVELOPMENT || DEBUG
6153 static int
6154 unlink_kernelcore_sysctl SYSCTL_HANDLER_ARGS
6155 {
6156 if (!req->newptr) {
6157 return EINVAL;
6158 }
6159 void IOBSDLowSpaceUnlinkKernelCore(void);
6160 IOBSDLowSpaceUnlinkKernelCore();
6161 return 0;
6162 }
6163
6164 SYSCTL_PROC(_kern, OID_AUTO, unlink_kernelcore,
6165 CTLTYPE_INT | CTLFLAG_WR | CTLFLAG_LOCKED | CTLFLAG_MASKED, 0, 0,
6166 unlink_kernelcore_sysctl, "-", "unlink the kernelcore file");
6167 #endif /* DEVELOPMENT || DEBUG */
6168
6169 #if CONFIG_IOTRACE
6170 #pragma clang diagnostic push
6171 #pragma clang diagnostic ignored "-Wcast-qual"
6172 SYSCTL_INT(_debug, OID_AUTO, MMIOtrace,
6173 CTLFLAG_KERN | CTLFLAG_RW | CTLFLAG_LOCKED,
6174 (int *)&mmiotrace_enabled, 0, "");
6175 #pragma clang diagnostic pop
6176 #endif /* CONFIG_IOTRACE */
6177
6178 static int
6179 sysctl_page_protection_type SYSCTL_HANDLER_ARGS
6180 {
6181 #pragma unused(oidp, arg1, arg2)
6182 int value = ml_page_protection_type();
6183 return SYSCTL_OUT(req, &value, sizeof(value));
6184 }
6185
6186 SYSCTL_PROC(_kern, OID_AUTO, page_protection_type,
6187 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
6188 0, 0, sysctl_page_protection_type, "I", "Type of page protection that the system supports");
6189
6190 TUNABLE_DT(int, gpu_pmem_selector, "defaults", "kern.gpu_pmem_selector", "gpu-pmem-selector", 0, TUNABLE_DT_NONE);
6191
6192 #if CONFIG_EXCLAVES
6193
6194 static int
6195 sysctl_task_conclave SYSCTL_HANDLER_ARGS
6196 {
6197 extern const char *exclaves_resource_name(void *);
6198
6199 #pragma unused(arg2)
6200 void *conclave = task_get_conclave(current_task());
6201 if (conclave != NULL) {
6202 const char *name = exclaves_resource_name(conclave);
6203 assert3u(strlen(name), >, 0);
6204
6205 /*
6206 * This is a RO operation already and the string is never
6207 * written to.
6208 */
6209 #pragma clang diagnostic push
6210 #pragma clang diagnostic ignored "-Wcast-qual"
6211 return sysctl_handle_string(oidp, (char *)name, 0, req);
6212 #pragma clang diagnostic pop
6213 }
6214 return sysctl_handle_string(oidp, arg1, MAXCONCLAVENAME, req);
6215 }
6216
6217 SYSCTL_PROC(_kern, OID_AUTO, task_conclave,
6218 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
6219 "", 0, sysctl_task_conclave, "A", "Conclave string for the task");
6220
6221
6222 void task_set_conclave_untaintable(task_t task);
6223
6224 static int
6225 sysctl_task_conclave_untaintable SYSCTL_HANDLER_ARGS
6226 {
6227 #pragma unused(arg1, arg2)
6228 int error, val = 0;
6229 error = sysctl_handle_int(oidp, &val, 0, req);
6230 if (error || val == 0) {
6231 return error;
6232 }
6233
6234 task_set_conclave_untaintable(current_task());
6235 return 0;
6236 }
6237
6238 SYSCTL_PROC(_kern, OID_AUTO, task_conclave_untaintable,
6239 CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_LOCKED,
6240 "", 0, sysctl_task_conclave_untaintable, "A", "Task could not be tainted by talking to conclaves");
6241
6242 extern exclaves_requirement_t exclaves_relaxed_requirements;
6243 SYSCTL_QUAD(_kern, OID_AUTO, exclaves_relaxed_requirements,
6244 CTLFLAG_KERN | CTLFLAG_RD | CTLFLAG_LOCKED,
6245 &exclaves_relaxed_requirements, "Exclaves requirements which have been relaxed");
6246
6247 #endif /* CONFIG_EXCLAVES */
6248
6249 #if (DEVELOPMENT || DEBUG)
6250 SYSCTL_INT(_kern, OID_AUTO, gpu_pmem_selector,
6251 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN,
6252 &gpu_pmem_selector, 0, "GPU wire down limit selector");
6253 #else /* !(DEVELOPMENT || DEBUG) */
6254 SYSCTL_INT(_kern, OID_AUTO, gpu_pmem_selector,
6255 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED | CTLFLAG_KERN | CTLFLAG_MASKED,
6256 &gpu_pmem_selector, 0, "GPU wire down limit selector");
6257 #endif /* (DEVELOPMENT || DEBUG) */
6258
6259 static int
6260 sysctl_exclaves_status SYSCTL_HANDLER_ARGS
6261 {
6262 int value = exclaves_get_status();
6263 return sysctl_io_number(req, value, sizeof(value), NULL, NULL);
6264 }
6265
6266 SYSCTL_PROC(_kern, OID_AUTO, exclaves_status,
6267 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
6268 0, 0, sysctl_exclaves_status, "I", "Running status of Exclaves");
6269
6270
6271 static int
6272 sysctl_exclaves_boot_stage SYSCTL_HANDLER_ARGS
6273 {
6274 int value = exclaves_get_boot_stage();
6275 return sysctl_io_number(req, value, sizeof(value), NULL, NULL);
6276 }
6277
6278 SYSCTL_PROC(_kern, OID_AUTO, exclaves_boot_stage,
6279 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
6280 0, 0, sysctl_exclaves_boot_stage, "I", "Boot stage of Exclaves");
6281
6282 #if CONFIG_EXCLAVES && (DEVELOPMENT || DEBUG)
6283 extern unsigned int exclaves_debug;
6284 SYSCTL_UINT(_kern, OID_AUTO, exclaves_debug, CTLFLAG_RW | CTLFLAG_LOCKED,
6285 &exclaves_debug, 0, "Exclaves debug flags");
6286
6287 static int
6288 sysctl_exclaves_inspection_status SYSCTL_HANDLER_ARGS
6289 {
6290 #pragma unused(oidp, arg1, arg2)
6291 int value = (int)exclaves_inspection_is_initialized();
6292 return sysctl_io_number(req, value, sizeof(value), NULL, NULL);
6293 }
6294 SYSCTL_PROC(_kern, OID_AUTO, exclaves_inspection_status,
6295 CTLTYPE_INT | CTLFLAG_RD | CTLFLAG_LOCKED,
6296 0, 0, sysctl_exclaves_inspection_status, "I", "Exclaves debug inspection status");
6297 #endif /* CONFIG_EXCLAVES && (DEVELOPMENT || DEBUG) */
6298
6299 #if (DEBUG || DEVELOPMENT)
6300 extern uint32_t disable_vm_sanitize_telemetry;
6301 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");
6302 #endif
6303