kern_sysctl.c source code [xnu/bsd/kern/kern_sysctl.c]

1	/*
2	* Copyright (c) 2000-2023 Apple Inc. All rights reserved.
3	*
4	* @APPLE_OSREFERENCE_LICENSE_HEADER_START@
5	*
6	* This file contains Original Code and/or Modifications of Original Code
7	* as defined in and that are subject to the Apple Public Source License
8	* Version 2.0 (the 'License'). You may not use this file except in
9	* compliance with the License. The rights granted to you under the License
10	* may not be used to create, or enable the creation or redistribution of,
11	* unlawful or unlicensed copies of an Apple operating system, or to
12	* circumvent, violate, or enable the circumvention or violation of, any
13	* terms of an Apple operating system software license agreement.
14	*
15	* Please obtain a copy of the License at
16	* http://www.opensource.apple.com/apsl/ and read it before using this file.
17	*
18	* The Original Code and all software distributed under the License are
19	* distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
20	* EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
21	* INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
22	* FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
23	* Please see the License for the specific language governing rights and
24	* limitations under the License.
25	*
26	* @APPLE_OSREFERENCE_LICENSE_HEADER_END@
27	*/
28	/ Copyright (c) 1995 NeXT Computer, Inc. All Rights Reserved /
29	/-*
30	* Copyright (c) 1982, 1986, 1989, 1993
31	* The Regents of the University of California. All rights reserved.
32	*
33	* This code is derived from software contributed to Berkeley by
34	* Mike Karels at Berkeley Software Design, Inc.
35	*
36	* Redistribution and use in source and binary forms, with or without
37	* modification, are permitted provided that the following conditions
38	* are met:
39	* 1. Redistributions of source code must retain the above copyright
40	* notice, this list of conditions and the following disclaimer.
41	* 2. Redistributions in binary form must reproduce the above copyright
42	* notice, this list of conditions and the following disclaimer in the
43	* documentation and/or other materials provided with the distribution.
44	* 3. All advertising materials mentioning features or use of this software
45	* must display the following acknowledgement:
46	* This product includes software developed by the University of
47	* California, Berkeley and its contributors.
48	* 4. Neither the name of the University nor the names of its contributors
49	* may be used to endorse or promote products derived from this software
50	* without specific prior written permission.
51	*
52	* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
53	* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
54	* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
55	* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
56	* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
57	* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
58	* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
59	* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
60	* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
61	* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
62	* SUCH DAMAGE.
63	*
64	* @(#)kern_sysctl.c 8.4 (Berkeley) 4/14/94
65	*/
66	/*
67	* NOTICE: This file was modified by SPARTA, Inc. in 2005 to introduce
68	* support for mandatory and extensible security protections. This notice
69	* is included in support of clause 2.2 (b) of the Apple Public License,
70	* Version 2.0.
71	*/
72
73	/*
74	* DEPRECATED sysctl system call code
75	*
76	* Everything in this file is deprecated. Sysctls should be handled
77	* by the code in kern_newsysctl.c.
78	* The remaining "case" sections are supposed to be converted into
79	* SYSCTL_*-style definitions, and as soon as all of them are gone,
80	* this source file is supposed to die.
81	*
82	* DO NOT ADD ANY MORE "case" SECTIONS TO THIS FILE, instead define
83	* your sysctl with SYSCTL_INT, SYSCTL_PROC etc. in your source file.
84	*/
85
86	#include <sys/param.h>
87	#include <sys/systm.h>
88	#include <sys/kernel.h>
89	#include <sys/malloc.h>
90	#include <sys/proc_internal.h>
91	#include <sys/kauth.h>
92	#include <sys/file_internal.h>
93	#include <sys/vnode_internal.h>
94	#include <sys/unistd.h>
95	#include <sys/buf.h>
96	#include <sys/ioctl.h>
97	#include <sys/namei.h>
98	#include <sys/tty.h>
99	#include <sys/disklabel.h>
100	#include <sys/vm.h>
101	#include <sys/sysctl.h>
102	#include <sys/user.h>
103	#include <sys/aio_kern.h>
104	#include <sys/reboot.h>
105	#include <sys/memory_maintenance.h>
106	#include <sys/priv.h>
107	#include <stdatomic.h>
108	#include <uuid/uuid.h>
109
110	#include <security/audit/audit.h>
111	#include <kern/kalloc.h>
112
113	#include <machine/smp.h>
114	#include <machine/atomic.h>
115	#include <machine/config.h>
116	#include <mach/machine.h>
117	#include <mach/mach_host.h>
118	#include <mach/mach_types.h>
119	#include <mach/processor_info.h>
120	#include <mach/vm_param.h>
121	#include <kern/debug.h>
122	#include <kern/mach_param.h>
123	#include <kern/task.h>
124	#include <kern/thread.h>
125	#include <kern/thread_group.h>
126	#include <kern/processor.h>
127	#include <kern/cpu_number.h>
128	#include <kern/sched_prim.h>
129	#include <kern/workload_config.h>
130	#include <kern/iotrace.h>
131	#include <vm/vm_kern.h>
132	#include <vm/vm_map.h>
133	#include <mach/host_info.h>
134	#include <mach/exclaves.h>
135	#include <kern/hvg_hypercall.h>
136	#include <kdp/sk_core.h>
137
138	#if DEVELOPMENT \|\| DEBUG
139	#include <kern/ext_paniclog.h>
140	#endif
141
142	#include <sys/mount_internal.h>
143	#include <sys/kdebug.h>
144	#include <sys/kern_debug.h>
145	#include <sys/kern_sysctl.h>
146	#include <sys/variant_internal.h>
147
148	#include <IOKit/IOPlatformExpert.h>
149	#include <pexpert/pexpert.h>
150
151	#include <machine/machine_routines.h>
152	#include <machine/exec.h>
153
154	#include <nfs/nfs_conf.h>
155
156	#include <vm/vm_protos.h>
157	#include <vm/vm_pageout.h>
158	#include <vm/vm_compressor_algorithms.h>
159	#include <sys/imgsrc.h>
160	#include <kern/timer_call.h>
161	#include <sys/codesign.h>
162	#include <IOKit/IOBSD.h>
163	#if CONFIG_CSR
164	#include <sys/csr.h>
165	#endif
166
167	#if defined(__i386__) \|\| defined(__x86_64__)
168	#include <i386/cpuid.h>
169	#endif
170
171	#if CONFIG_FREEZE
172	#include <sys/kern_memorystatus.h>
173	#endif
174
175	#if KPERF
176	#include <kperf/kperf.h>
177	#endif
178
179	#if HYPERVISOR
180	#include <kern/hv_support.h>
181	#endif
182
183
184	#include <corecrypto/ccsha2.h>
185
186	/*
187	* deliberately setting max requests to really high number
188	* so that runaway settings do not cause MALLOC overflows
189	*/
190	#define AIO_MAX_REQUESTS (128 * CONFIG_AIO_MAX)
191
192	extern int aio_max_requests;
193	extern int aio_max_requests_per_process;
194	extern int aio_worker_threads;
195	extern int lowpri_IO_window_msecs;
196	extern int lowpri_IO_delay_msecs;
197	#if DEVELOPMENT \|\| DEBUG
198	extern int nx_enabled;
199	#endif
200	extern int speculative_reads_disabled;
201	extern unsigned int speculative_prefetch_max;
202	extern unsigned int speculative_prefetch_max_iosize;
203	extern unsigned int preheat_max_bytes;
204	extern unsigned int preheat_min_bytes;
205	extern long numvnodes;
206	extern long freevnodes;
207	extern long num_recycledvnodes;
208
209	extern uuid_string_t bootsessionuuid_string;
210
211	extern unsigned int vm_max_delayed_work_limit;
212	extern unsigned int vm_max_batch;
213
214	extern unsigned int vm_page_free_min;
215	extern unsigned int vm_page_free_target;
216	extern unsigned int vm_page_free_reserved;
217
218	#if (DEVELOPMENT \|\| DEBUG)
219	extern uint32_t vm_page_creation_throttled_hard;
220	extern uint32_t vm_page_creation_throttled_soft;
221	#endif /* DEVELOPMENT \|\| DEBUG */
222
223	#if DEVELOPMENT \|\| DEBUG
224	extern bool bootarg_hide_process_traced;
225	#endif
226
227	/*
228	* Conditionally allow dtrace to see these functions for debugging purposes.
229	*/
230	#ifdef STATIC
231	#undef STATIC
232	#endif
233	#if 0
234	#define STATIC
235	#else
236	#define STATIC static
237	#endif
238
239	extern boolean_t mach_timer_coalescing_enabled;
240
241	extern uint64_t timer_deadline_tracking_bin_1, timer_deadline_tracking_bin_2;
242
243	STATIC void
244	fill_user32_eproc(proc_t, struct user32_eproc *__restrict);
245	STATIC void
246	fill_user32_externproc(proc_t, struct user32_extern_proc *__restrict);
247	STATIC void
248	fill_user64_eproc(proc_t, struct user64_eproc *__restrict);
249	STATIC void
250	fill_user64_proc(proc_t, struct user64_kinfo_proc *__restrict);
251	STATIC void
252	fill_user64_externproc(proc_t, struct user64_extern_proc *__restrict);
253	STATIC void
254	fill_user32_proc(proc_t, struct user32_kinfo_proc *__restrict);
255
256	#if CONFIG_NETBOOT
257	extern int
258	netboot_root(void);
259	#endif
260	int
261	sysctl_procargs(int *name, u_int namelen, user_addr_t where,
262	size_t *sizep, proc_t cur_proc);
263	STATIC int
264	sysctl_procargsx(int name, u_int namelen, user_addr_t where, size_t sizep,
265	proc_t cur_proc, int argc_yes);
266	int
267	sysctl_struct(user_addr_t oldp, size_t *oldlenp, user_addr_t newp,
268	size_t newlen, void sp, int* len);
269
270	STATIC int sysdoproc_filt_KERN_PROC_PID(proc_t p, void * arg);
271	STATIC int sysdoproc_filt_KERN_PROC_PGRP(proc_t p, void * arg);
272	STATIC int sysdoproc_filt_KERN_PROC_TTY(proc_t p, void * arg);
273	STATIC int sysdoproc_filt_KERN_PROC_UID(proc_t p, void * arg);
274	STATIC int sysdoproc_filt_KERN_PROC_RUID(proc_t p, void * arg);
275	int sysdoproc_callback(proc_t p, void *arg);
276
277	#if CONFIG_THREAD_GROUPS && (DEVELOPMENT \|\| DEBUG)
278	STATIC int sysctl_get_thread_group_id SYSCTL_HANDLER_ARGS;
279	#endif
280
281	/ forward declarations for non-static STATIC /
282	STATIC void fill_loadavg64(struct loadavg la, struct* user64_loadavg *la64);
283	STATIC void fill_loadavg32(struct loadavg la, struct* user32_loadavg *la32);
284	STATIC int sysctl_handle_kern_threadname(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
285	STATIC int sysctl_sched_stats(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
286	STATIC int sysctl_sched_stats_enable(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
287	#if COUNT_SYSCALLS
288	STATIC int sysctl_docountsyscalls SYSCTL_HANDLER_ARGS;
289	#endif /* COUNT_SYSCALLS */
290	#if defined(XNU_TARGET_OS_OSX)
291	STATIC int sysctl_doprocargs SYSCTL_HANDLER_ARGS;
292	#endif /* defined(XNU_TARGET_OS_OSX) */
293	STATIC int sysctl_doprocargs2 SYSCTL_HANDLER_ARGS;
294	STATIC int sysctl_prochandle SYSCTL_HANDLER_ARGS;
295	STATIC int sysctl_aiomax(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
296	STATIC int sysctl_aioprocmax(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
297	STATIC int sysctl_aiothreads(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
298	STATIC int sysctl_maxproc(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
299	STATIC int sysctl_osversion(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
300	STATIC int sysctl_sysctl_bootargs(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
301	STATIC int sysctl_maxvnodes(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
302	STATIC int sysctl_securelvl(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
303	STATIC int sysctl_domainname(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
304	STATIC int sysctl_hostname(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
305	STATIC int sysctl_procname(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
306	STATIC int sysctl_boottime(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
307	STATIC int sysctl_bootuuid(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
308	STATIC int sysctl_symfile(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
309	#if CONFIG_NETBOOT
310	STATIC int sysctl_netboot(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
311	#endif
312	#ifdef CONFIG_IMGSRC_ACCESS
313	STATIC int sysctl_imgsrcdev(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
314	#endif
315	STATIC int sysctl_usrstack(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
316	STATIC int sysctl_usrstack64(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
317	#if CONFIG_COREDUMP
318	STATIC int sysctl_coredump(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
319	STATIC int sysctl_suid_coredump(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
320	#endif
321	STATIC int sysctl_delayterm(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
322	STATIC int sysctl_rage_vnode(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
323	STATIC int sysctl_kern_check_openevt(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
324	#if DEVELOPMENT \|\| DEBUG
325	STATIC int sysctl_nx(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
326	#endif
327	STATIC int sysctl_loadavg(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
328	STATIC int sysctl_vm_toggle_address_reuse(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
329	STATIC int sysctl_swapusage(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
330	STATIC int fetch_process_cputype( proc_t cur_proc, int name, u_int namelen, cpu_type_t cputype);
331	STATIC int sysctl_sysctl_native(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
332	STATIC int sysctl_sysctl_cputype(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
333	STATIC int sysctl_safeboot(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
334	STATIC int sysctl_singleuser(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
335	STATIC int sysctl_minimalboot(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
336	STATIC int sysctl_slide(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
337
338	#ifdef CONFIG_XNUPOST
339	#include <tests/xnupost.h>
340
341	STATIC int sysctl_debug_test_oslog_ctl(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
342	STATIC int sysctl_debug_test_stackshot_mutex_owner(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
343	STATIC int sysctl_debug_test_stackshot_rwlck_owner(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req);
344	#endif
345
346	extern void IORegistrySetOSBuildVersion(char * build_version);
347	extern int IOParseWorkloadConfig(workload_config_ctx_t ctx, const* char * buffer, size_t size);
348	extern int IOUnparseWorkloadConfig(char buffer, size_t size);
349
350	STATIC void
351	fill_loadavg64(struct loadavg la, struct* user64_loadavg *la64)
352	{
353	la64->ldavg[`0`] = la->ldavg[`0`];
354	la64->ldavg[`1`] = la->ldavg[`1`];
355	la64->ldavg[`2`] = la->ldavg[`2`];
356	la64->fscale = (user64_long_t)la->fscale;
357	}
358
359	STATIC void
360	fill_loadavg32(struct loadavg la, struct* user32_loadavg *la32)
361	{
362	la32->ldavg[`0`] = la->ldavg[`0`];
363	la32->ldavg[`1`] = la->ldavg[`1`];
364	la32->ldavg[`2`] = la->ldavg[`2`];
365	la32->fscale = (user32_long_t)la->fscale;
366	}
367
368	#if COUNT_SYSCALLS
369	extern int do_count_syscalls;
370	#endif
371
372	#ifdef INSECURE
373	int securelevel = -`1`;
374	#else
375	int securelevel;
376	#endif
377
378	STATIC int
379	sysctl_handle_kern_threadname( __unused struct sysctl_oid oidp, __unused void* *arg1,
380	__unused int arg2, struct sysctl_req *req)
381	{
382	int error;
383	struct uthread *ut = current_uthread();
384	user_addr_t oldp = `0`, newp = `0`;
385	size_t *oldlenp = NULL;
386	size_t newlen = `0`;
387
388	oldp = req->oldptr;
389	oldlenp = &(req->oldlen);
390	newp = req->newptr;
391	newlen = req->newlen;
392
393	/ We want the current length, and maybe the string itself /
394	if (oldlenp) {
395	/ if we have no thread name yet tell'em we want MAXTHREADNAMESIZE - 1 /
396	size_t currlen = MAXTHREADNAMESIZE - `1`;
397
398	if (ut->pth_name) {
399	/ use length of current thread name /
400	currlen = strlen(s: ut->pth_name);
401	}
402	if (oldp) {
403	if (*oldlenp < currlen) {
404	return ENOMEM;
405	}
406	/ NOTE - we do not copy the NULL terminator /
407	if (ut->pth_name) {
408	error = copyout(ut->pth_name, oldp, currlen);
409	if (error) {
410	return error;
411	}
412	}
413	}
414	/ return length of thread name minus NULL terminator (just like strlen) /
415	req->oldidx = currlen;
416	}
417
418	/ We want to set the name to something /
419	if (newp) {
420	if (newlen > (MAXTHREADNAMESIZE - `1`)) {
421	return ENAMETOOLONG;
422	}
423	if (!ut->pth_name) {
424	char tmp_pth_name = (char* *)kalloc_data(MAXTHREADNAMESIZE,
425	Z_WAITOK \| Z_ZERO);
426	if (!tmp_pth_name) {
427	return ENOMEM;
428	}
429	if (!OSCompareAndSwapPtr(NULL, tmp_pth_name, &ut->pth_name)) {
430	kfree_data(tmp_pth_name, MAXTHREADNAMESIZE);
431	return EBUSY;
432	}
433	} else {
434	kernel_debug_string_simple(TRACE_STRING_THREADNAME_PREV, str: ut->pth_name);
435	bzero(s: ut->pth_name, MAXTHREADNAMESIZE);
436	}
437	error = copyin(newp, ut->pth_name, newlen);
438	if (error) {
439	return error;
440	}
441
442	kernel_debug_string_simple(TRACE_STRING_THREADNAME, str: ut->pth_name);
443	}
444
445	return `0`;
446	}
447
448	SYSCTL_PROC(_kern, KERN_THREADNAME, threadname, CTLFLAG_ANYBODY \| CTLTYPE_STRING \| CTLFLAG_RW \| CTLFLAG_LOCKED, `0`, `0`, sysctl_handle_kern_threadname, "A", "");
449
450	#define WORKLOAD_CONFIG_MAX_SIZE (128 * 1024 * 1024)
451
452	/ Called locked - sysctl defined without CTLFLAG_LOCKED. /
453	static int
454	sysctl_workload_config SYSCTL_HANDLER_ARGS
455	{
456	#pragma unused(arg1, arg2)
457
458	char *plist_blob = NULL;
459	kern_return_t ret = KERN_FAILURE;
460	int error = -`1`;
461
462	/ Only allow reading of workload config on non-RELEASE kernels. /
463	#if DEVELOPMENT \|\| DEBUG
464
465	const size_t buf_size = req->oldlen;
466
467	if (!req->oldptr) {
468	/ Just looking for the size to allocate. /
469	size_t size = `0`;
470	ret = IOUnparseWorkloadConfig(NULL, &size);
471	if (ret != KERN_SUCCESS) {
472	return ENOMEM;
473	}
474
475	error = SYSCTL_OUT(req, NULL, size);
476	if (error) {
477	return error;
478	}
479	} else {
480	if (buf_size > (WORKLOAD_CONFIG_MAX_SIZE - `1`) \|\|
481	buf_size == `0`) {
482	return EINVAL;
483	}
484
485	plist_blob = kalloc_data(buf_size, Z_WAITOK \| Z_ZERO);
486	if (!plist_blob) {
487	return ENOMEM;
488	}
489
490	size_t size = buf_size;
491	ret = IOUnparseWorkloadConfig(plist_blob, &size);
492	if (ret != KERN_SUCCESS) {
493	kfree_data(plist_blob, buf_size);
494	return ENOMEM;
495	}
496
497	error = SYSCTL_OUT(req, plist_blob, MIN(buf_size, size));
498
499	/ If the buffer was too small to fit the entire config. /
500	if (buf_size < size) {
501	error = ENOMEM;
502	}
503
504	kfree_data(plist_blob, buf_size);
505	if (error) {
506	return error;
507	}
508	}
509	#endif /* DEVELOPMENT \|\| DEBUG */
510
511	if (req->newptr) {
512	size_t newlen = req->newlen;
513	if (newlen > (WORKLOAD_CONFIG_MAX_SIZE - `1`)) {
514	return EINVAL;
515	}
516
517
518	workload_config_ctx_t *ctx = NULL;
519	/*
520	* Only allow workload_config_boot to be loaded once at boot by launchd.
521	*/
522	if (current_proc() == initproc &&
523	!workload_config_initialized(ctx: &workload_config_boot)) {
524	ctx = &workload_config_boot;
525	} else {
526	#if DEVELOPMENT \|\| DEBUG
527	/*
528	* Use the devel config context otherwise. If a devel config has been
529	* initialized it will be used for lookups in place of the boot config.
530	*/
531	ctx = &workload_config_devel;
532	if (workload_config_initialized(ctx)) {
533	workload_config_free(ctx);
534	}
535
536	/ The devel context can be explicitly cleared by an empty string. /
537	if (newlen == `1`) {
538	return `0`;
539	}
540	#else
541	return EINVAL;
542	#endif
543	}
544
545	plist_blob = kalloc_data(newlen + `1`, Z_WAITOK \| Z_ZERO);
546	if (!plist_blob) {
547	return ENOMEM;
548	}
549	error = copyin(req->newptr, plist_blob, newlen);
550	if (error) {
551	kfree_data(plist_blob, newlen + `1`);
552	return error;
553	}
554	plist_blob[newlen] = `'\0'`;
555	ret = IOParseWorkloadConfig(ctx, buffer: plist_blob, size: newlen + `1`);
556
557	kfree_data(plist_blob, newlen + `1`);
558	return ret == KERN_SUCCESS ? `0` : EINVAL;
559	}
560
561	return `0`;
562	}
563
564	SYSCTL_PROC(_kern, OID_AUTO, workload_config, CTLTYPE_STRING \| CTLFLAG_RW \| CTLFLAG_MASKED,
565	`0`, `0`, sysctl_workload_config, "A", "global workgroup configuration plist load/unload");
566
567	#define BSD_HOST 1
568	STATIC int
569	sysctl_sched_stats(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
570	{
571	host_basic_info_data_t hinfo;
572	kern_return_t kret;
573	uint32_t size;
574	uint32_t buf_size = `0`;
575	int changed;
576	mach_msg_type_number_t count = HOST_BASIC_INFO_COUNT;
577	struct _processor_statistics_np *buf;
578	int error;
579
580	kret = host_info(host: (host_t)BSD_HOST, HOST_BASIC_INFO, host_info_out: (host_info_t)&hinfo, host_info_outCnt: &count);
581	if (kret != KERN_SUCCESS) {
582	return EINVAL;
583	}
584
585	size = sizeof(struct _processor_statistics_np) * (hinfo.logical_cpu_max + `2`); / One for RT Queue, One for Fair Share Queue /
586
587	if (req->oldlen < size) {
588	return EINVAL;
589	}
590
591	buf_size = size;
592	buf = (struct _processor_statistics_np *)kalloc_data(buf_size, Z_ZERO \| Z_WAITOK);
593
594	kret = get_sched_statistics(out: buf, count: &size);
595	if (kret != KERN_SUCCESS) {
596	error = EINVAL;
597	goto out;
598	}
599
600	error = sysctl_io_opaque(req, pValue: buf, valueSize: size, changed: &changed);
601	if (error) {
602	goto out;
603	}
604
605	if (changed) {
606	panic("Sched info changed?!");
607	}
608	out:
609	kfree_data(buf, buf_size);
610	return error;
611	}
612
613	SYSCTL_PROC(_kern, OID_AUTO, sched_stats, CTLFLAG_LOCKED, `0`, `0`, sysctl_sched_stats, "-", "");
614
615	STATIC int
616	sysctl_sched_stats_enable(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, __unused struct sysctl_req *req)
617	{
618	boolean_t active;
619	int res;
620
621	if (req->newlen != sizeof(active)) {
622	return EINVAL;
623	}
624
625	res = copyin(req->newptr, &active, sizeof(active));
626	if (res != `0`) {
627	return res;
628	}
629
630	return set_sched_stats_active(active);
631	}
632
633	SYSCTL_PROC(_kern, OID_AUTO, sched_stats_enable, CTLFLAG_LOCKED \| CTLFLAG_WR, `0`, `0`, sysctl_sched_stats_enable, "-", "");
634
635	extern uint32_t sched_debug_flags;
636	SYSCTL_INT(_debug, OID_AUTO, sched, CTLFLAG_RW \| CTLFLAG_LOCKED, &sched_debug_flags, `0`, "scheduler debug");
637
638	#if (DEBUG \|\| DEVELOPMENT)
639	extern boolean_t doprnt_hide_pointers;
640	SYSCTL_INT(_debug, OID_AUTO, hide_kernel_pointers, CTLFLAG_RW \| CTLFLAG_LOCKED, &doprnt_hide_pointers, `0`, "hide kernel pointers from log");
641	#endif
642
643
644	extern int get_kernel_symfile(proc_t, char **);
645
646	#if COUNT_SYSCALLS
647	#define KERN_COUNT_SYSCALLS (KERN_OSTYPE + 1000)
648
649	extern const unsigned int nsysent;
650	extern int syscalls_log[];
651	extern const char *syscallnames[];
652
653	STATIC int
654	sysctl_docountsyscalls SYSCTL_HANDLER_ARGS
655	{
656	__unused int cmd = oidp->oid_arg2; / subcommand/
657	__unused int name = arg1; /* oid element argument vector /
658	__unused int namelen = arg2; / number of oid element arguments /
659	int error, changed;
660
661	int tmp;
662
663	/ valid values passed in:*
664	* = 0 means don't keep called counts for each bsd syscall
665	* > 0 means keep called counts for each bsd syscall
666	* = 2 means dump current counts to the system log
667	* = 3 means reset all counts
668	* for example, to dump current counts:
669	* sysctl -w kern.count_calls=2
670	*/
671	error = sysctl_io_number(req, do_count_syscalls,
672	sizeof(do_count_syscalls), &tmp, &changed);
673
674	if (error != `0` \|\| !changed) {
675	return error;
676	}
677
678	if (tmp == `1`) {
679	do_count_syscalls = `1`;
680	} else if (tmp == `0` \|\| tmp == `2` \|\| tmp == `3`) {
681	for (int i = `0`; i < nsysent; i++) {
682	if (syscalls_log[i] != `0`) {
683	if (tmp == `2`) {
684	printf("%d calls - name %s \n", syscalls_log[i], syscallnames[i]);
685	} else {
686	syscalls_log[i] = `0`;
687	}
688	}
689	}
690	do_count_syscalls = (tmp != `0`);
691	}
692
693	return error;
694	}
695	SYSCTL_PROC(_kern, KERN_COUNT_SYSCALLS, count_syscalls, CTLTYPE_NODE \| CTLFLAG_RD \| CTLFLAG_LOCKED,
696	`0`, / Pointer argument (arg1) /
697	`0`, / Integer argument (arg2) /
698	sysctl_docountsyscalls, / Handler function /
699	NULL, / Data pointer /
700	"");
701	#endif /* COUNT_SYSCALLS */
702
703	/*
704	* The following sysctl_* functions should not be used
705	* any more, as they can only cope with callers in
706	* user mode: Use new-style
707	* sysctl_io_number()
708	* sysctl_io_string()
709	* sysctl_io_opaque()
710	* instead.
711	*/
712
713	STATIC int
714	sysdoproc_filt_KERN_PROC_PID(proc_t p, void * arg)
715	{
716	if (proc_getpid(p) != (pid_t)(int**)arg) {
717	return `0`;
718	} else {
719	return `1`;
720	}
721	}
722
723	STATIC int
724	sysdoproc_filt_KERN_PROC_PGRP(proc_t p, void * arg)
725	{
726	if (p->p_pgrpid != (pid_t)(int**)arg) {
727	return `0`;
728	} else {
729	return `1`;
730	}
731	}
732
733	STATIC int
734	sysdoproc_filt_KERN_PROC_TTY(proc_t p, void * arg)
735	{
736	struct pgrp *pg;
737	dev_t dev = NODEV;
738
739	if ((p->p_flag & P_CONTROLT) && (pg = proc_pgrp(p, NULL)) != PGRP_NULL) {
740	dev = os_atomic_load(&pg->pg_session->s_ttydev, relaxed);
741	pgrp_rele(pgrp: pg);
742	}
743
744	return dev != NODEV && dev == (dev_t)(int* *)arg;
745	}
746
747	STATIC int
748	sysdoproc_filt_KERN_PROC_UID(proc_t p, void * arg)
749	{
750	uid_t uid;
751
752	smr_proc_task_enter();
753	uid = kauth_cred_getuid(cred: proc_ucred_smr(p));
754	smr_proc_task_leave();
755
756	if (uid != (uid_t)(int**)arg) {
757	return `0`;
758	} else {
759	return `1`;
760	}
761	}
762
763
764	STATIC int
765	sysdoproc_filt_KERN_PROC_RUID(proc_t p, void * arg)
766	{
767	uid_t ruid;
768
769	smr_proc_task_enter();
770	ruid = kauth_cred_getruid(cred: proc_ucred_smr(p));
771	smr_proc_task_leave();
772
773	if (ruid != (uid_t)(int**)arg) {
774	return `0`;
775	} else {
776	return `1`;
777	}
778	}
779
780	/*
781	* try over estimating by 5 procs
782	*/
783	#define KERN_PROCSLOP (5 * sizeof(struct kinfo_proc))
784	struct sysdoproc_args {
785	size_t buflen;
786	void *kprocp;
787	boolean_t is_64_bit;
788	user_addr_t dp;
789	size_t needed;
790	unsigned int sizeof_kproc;
791	int *errorp;
792	int uidcheck;
793	int ruidcheck;
794	int ttycheck;
795	int uidval;
796	};
797
798	int
799	sysdoproc_callback(proc_t p, void *arg)
800	{
801	struct sysdoproc_args *args = arg;
802
803	if (args->buflen >= args->sizeof_kproc) {
804	if ((args->ruidcheck != `0`) && (sysdoproc_filt_KERN_PROC_RUID(p, arg: &args->uidval) == `0`)) {
805	return PROC_RETURNED;
806	}
807	if ((args->uidcheck != `0`) && (sysdoproc_filt_KERN_PROC_UID(p, arg: &args->uidval) == `0`)) {
808	return PROC_RETURNED;
809	}
810	if ((args->ttycheck != `0`) && (sysdoproc_filt_KERN_PROC_TTY(p, arg: &args->uidval) == `0`)) {
811	return PROC_RETURNED;
812	}
813
814	bzero(s: args->kprocp, n: args->sizeof_kproc);
815	if (args->is_64_bit) {
816	fill_user64_proc(p, args->kprocp);
817	} else {
818	fill_user32_proc(p, args->kprocp);
819	}
820	int error = copyout(args->kprocp, args->dp, args->sizeof_kproc);
821	if (error) {
822	*args->errorp = error;
823	return PROC_RETURNED_DONE;
824	}
825	args->dp += args->sizeof_kproc;
826	args->buflen -= args->sizeof_kproc;
827	}
828	args->needed += args->sizeof_kproc;
829	return PROC_RETURNED;
830	}
831
832	SYSCTL_NODE(_kern, KERN_PROC, proc, CTLFLAG_RD \| CTLFLAG_LOCKED, `0`, "");
833	STATIC int
834	sysctl_prochandle SYSCTL_HANDLER_ARGS
835	{
836	int cmd = oidp->oid_arg2; / subcommand for multiple nodes /
837	int name = arg1; /* oid element argument vector /
838	int namelen = arg2; / number of oid element arguments /
839	user_addr_t where = req->oldptr;/ user buffer copy out address /
840
841	user_addr_t dp = where;
842	size_t needed = `0`;
843	size_t buflen = where != USER_ADDR_NULL ? req->oldlen : `0`;
844	int error = `0`;
845	boolean_t is_64_bit = proc_is64bit(current_proc());
846	struct user32_kinfo_proc user32_kproc;
847	struct user64_kinfo_proc user_kproc;
848	int sizeof_kproc;
849	void *kprocp;
850	int (filterfn)(proc_t, void* *) = `0`;
851	struct sysdoproc_args args;
852	int uidcheck = `0`;
853	int ruidcheck = `0`;
854	int ttycheck = `0`;
855
856	if (namelen != `1` && !(namelen == `0` && cmd == KERN_PROC_ALL)) {
857	return EINVAL;
858	}
859
860	if (is_64_bit) {
861	sizeof_kproc = sizeof(user_kproc);
862	kprocp = &user_kproc;
863	} else {
864	sizeof_kproc = sizeof(user32_kproc);
865	kprocp = &user32_kproc;
866	}
867
868	switch (cmd) {
869	case KERN_PROC_PID:
870	filterfn = sysdoproc_filt_KERN_PROC_PID;
871	break;
872
873	case KERN_PROC_PGRP:
874	filterfn = sysdoproc_filt_KERN_PROC_PGRP;
875	break;
876
877	case KERN_PROC_TTY:
878	ttycheck = `1`;
879	break;
880
881	case KERN_PROC_UID:
882	uidcheck = `1`;
883	break;
884
885	case KERN_PROC_RUID:
886	ruidcheck = `1`;
887	break;
888
889	case KERN_PROC_ALL:
890	break;
891
892	default:
893	/ must be kern.proc.<unknown> /
894	return ENOTSUP;
895	}
896
897	error = `0`;
898	args.buflen = buflen;
899	args.kprocp = kprocp;
900	args.is_64_bit = is_64_bit;
901	args.dp = dp;
902	args.needed = needed;
903	args.errorp = &error;
904	args.uidcheck = uidcheck;
905	args.ruidcheck = ruidcheck;
906	args.ttycheck = ttycheck;
907	args.sizeof_kproc = sizeof_kproc;
908	if (namelen) {
909	args.uidval = name[`0`];
910	}
911
912	proc_iterate(flags: (PROC_ALLPROCLIST \| PROC_ZOMBPROCLIST),
913	callout: sysdoproc_callback, arg: &args, filterfn, filterarg: name);
914
915	if (error) {
916	return error;
917	}
918
919	dp = args.dp;
920	needed = args.needed;
921
922	if (where != USER_ADDR_NULL) {
923	req->oldlen = dp - where;
924	if (needed > req->oldlen) {
925	return ENOMEM;
926	}
927	} else {
928	needed += KERN_PROCSLOP;
929	req->oldlen = needed;
930	}
931	/ adjust index so we return the right required/consumed amount /
932	req->oldidx += req->oldlen;
933	return `0`;
934	}
935
936
937	/*
938	* We specify the subcommand code for multiple nodes as the 'req->arg2' value
939	* in the sysctl declaration itself, which comes into the handler function
940	* as 'oidp->oid_arg2'.
941	*
942	* For these particular sysctls, since they have well known OIDs, we could
943	* have just obtained it from the '((int *)arg1)[0]' parameter, but that would
944	* not demonstrate how to handle multiple sysctls that used OID_AUTO instead
945	* of a well known value with a common handler function. This is desirable,
946	* because we want well known values to "go away" at some future date.
947	*
948	* It should be noted that the value of '((int *)arg1)[1]' is used for many
949	* an integer parameter to the subcommand for many of these sysctls; we'd
950	* rather have used '((int *)arg1)[0]' for that, or even better, an element
951	* in a structure passed in as the the 'newp' argument to sysctlbyname(3),
952	* and then use leaf-node permissions enforcement, but that would have
953	* necessitated modifying user space code to correspond to the interface
954	* change, and we are striving for binary backward compatibility here; even
955	* though these are SPI, and not intended for use by user space applications
956	* which are not themselves system tools or libraries, some applications
957	* have erroneously used them.
958	*/
959	SYSCTL_PROC(_kern_proc, KERN_PROC_ALL, all, CTLTYPE_NODE \| CTLFLAG_RD \| CTLFLAG_LOCKED,
960	`0`, / Pointer argument (arg1) /
961	KERN_PROC_ALL, / Integer argument (arg2) /
962	sysctl_prochandle, / Handler function /
963	NULL, / Data is size variant on ILP32/LP64 /
964	"");
965	SYSCTL_PROC(_kern_proc, KERN_PROC_PID, pid, CTLTYPE_NODE \| CTLFLAG_RD \| CTLFLAG_LOCKED,
966	`0`, / Pointer argument (arg1) /
967	KERN_PROC_PID, / Integer argument (arg2) /
968	sysctl_prochandle, / Handler function /
969	NULL, / Data is size variant on ILP32/LP64 /
970	"");
971	SYSCTL_PROC(_kern_proc, KERN_PROC_TTY, tty, CTLTYPE_NODE \| CTLFLAG_RD \| CTLFLAG_LOCKED,
972	`0`, / Pointer argument (arg1) /
973	KERN_PROC_TTY, / Integer argument (arg2) /
974	sysctl_prochandle, / Handler function /
975	NULL, / Data is size variant on ILP32/LP64 /
976	"");
977	SYSCTL_PROC(_kern_proc, KERN_PROC_PGRP, pgrp, CTLTYPE_NODE \| CTLFLAG_RD \| CTLFLAG_LOCKED,
978	`0`, / Pointer argument (arg1) /
979	KERN_PROC_PGRP, / Integer argument (arg2) /
980	sysctl_prochandle, / Handler function /
981	NULL, / Data is size variant on ILP32/LP64 /
982	"");
983	SYSCTL_PROC(_kern_proc, KERN_PROC_UID, uid, CTLTYPE_NODE \| CTLFLAG_RD \| CTLFLAG_LOCKED,
984	`0`, / Pointer argument (arg1) /
985	KERN_PROC_UID, / Integer argument (arg2) /
986	sysctl_prochandle, / Handler function /
987	NULL, / Data is size variant on ILP32/LP64 /
988	"");
989	SYSCTL_PROC(_kern_proc, KERN_PROC_RUID, ruid, CTLTYPE_NODE \| CTLFLAG_RD \| CTLFLAG_LOCKED,
990	`0`, / Pointer argument (arg1) /
991	KERN_PROC_RUID, / Integer argument (arg2) /
992	sysctl_prochandle, / Handler function /
993	NULL, / Data is size variant on ILP32/LP64 /
994	"");
995	SYSCTL_PROC(_kern_proc, KERN_PROC_LCID, lcid, CTLTYPE_NODE \| CTLFLAG_RD \| CTLFLAG_LOCKED,
996	`0`, / Pointer argument (arg1) /
997	KERN_PROC_LCID, / Integer argument (arg2) /
998	sysctl_prochandle, / Handler function /
999	NULL, / Data is size variant on ILP32/LP64 /
1000	"");
1001
1002
1003	/*
1004	* Fill in non-zero fields of an eproc structure for the specified process.
1005	*/
1006	STATIC void
1007	fill_user32_eproc(proc_t p, struct user32_eproc *__restrict ep)
1008	{
1009	struct pgrp *pg;
1010	struct session *sessp;
1011	kauth_cred_t my_cred;
1012
1013	pg = proc_pgrp(p, &sessp);
1014
1015	if (pg != PGRP_NULL) {
1016	ep->e_pgid = p->p_pgrpid;
1017	ep->e_jobc = pg->pg_jobc;
1018	if (sessp->s_ttyvp) {
1019	ep->e_flag = EPROC_CTTY;
1020	}
1021	}
1022
1023	ep->e_ppid = p->p_ppid;
1024
1025	smr_proc_task_enter();
1026	my_cred = proc_ucred_smr(p);
1027
1028	/ A fake historical pcred /
1029	ep->e_pcred.p_ruid = kauth_cred_getruid(cred: my_cred);
1030	ep->e_pcred.p_svuid = kauth_cred_getsvuid(cred: my_cred);
1031	ep->e_pcred.p_rgid = kauth_cred_getrgid(cred: my_cred);
1032	ep->e_pcred.p_svgid = kauth_cred_getsvgid(cred: my_cred);
1033
1034	/ A fake historical kauth_cred_t /*
1035	unsigned long refcnt = os_atomic_load(&my_cred->cr_ref, relaxed);
1036	ep->e_ucred.cr_ref = (uint32_t)MIN(refcnt, UINT32_MAX);
1037	ep->e_ucred.cr_uid = kauth_cred_getuid(cred: my_cred);
1038	ep->e_ucred.cr_ngroups = (short)posix_cred_get(cred: my_cred)->cr_ngroups;
1039	bcopy(src: posix_cred_get(cred: my_cred)->cr_groups,
1040	dst: ep->e_ucred.cr_groups, NGROUPS * sizeof(gid_t));
1041
1042	my_cred = NOCRED;
1043	smr_proc_task_leave();
1044
1045	ep->e_tdev = NODEV;
1046	if (pg != PGRP_NULL) {
1047	if (p->p_flag & P_CONTROLT) {
1048	session_lock(sess: sessp);
1049	ep->e_tdev = os_atomic_load(&sessp->s_ttydev, relaxed);
1050	ep->e_tpgid = sessp->s_ttypgrpid;
1051	session_unlock(sess: sessp);
1052	}
1053	if (SESS_LEADER(p, sessp)) {
1054	ep->e_flag \|= EPROC_SLEADER;
1055	}
1056	pgrp_rele(pgrp: pg);
1057	}
1058	}
1059
1060	/*
1061	* Fill in non-zero fields of an LP64 eproc structure for the specified process.
1062	*/
1063	STATIC void
1064	fill_user64_eproc(proc_t p, struct user64_eproc *__restrict ep)
1065	{
1066	struct pgrp *pg;
1067	struct session *sessp;
1068	kauth_cred_t my_cred;
1069
1070	pg = proc_pgrp(p, &sessp);
1071
1072	if (pg != PGRP_NULL) {
1073	ep->e_pgid = p->p_pgrpid;
1074	ep->e_jobc = pg->pg_jobc;
1075	if (sessp->s_ttyvp) {
1076	ep->e_flag = EPROC_CTTY;
1077	}
1078	}
1079
1080	ep->e_ppid = p->p_ppid;
1081
1082	smr_proc_task_enter();
1083	my_cred = proc_ucred_smr(p);
1084
1085	/ A fake historical pcred /
1086	ep->e_pcred.p_ruid = kauth_cred_getruid(cred: my_cred);
1087	ep->e_pcred.p_svuid = kauth_cred_getsvuid(cred: my_cred);
1088	ep->e_pcred.p_rgid = kauth_cred_getrgid(cred: my_cred);
1089	ep->e_pcred.p_svgid = kauth_cred_getsvgid(cred: my_cred);
1090
1091	/ A fake historical kauth_cred_t /*
1092	unsigned long refcnt = os_atomic_load(&my_cred->cr_ref, relaxed);
1093	ep->e_ucred.cr_ref = (uint32_t)MIN(refcnt, UINT32_MAX);
1094	ep->e_ucred.cr_uid = kauth_cred_getuid(cred: my_cred);
1095	ep->e_ucred.cr_ngroups = (short)posix_cred_get(cred: my_cred)->cr_ngroups;
1096	bcopy(src: posix_cred_get(cred: my_cred)->cr_groups,
1097	dst: ep->e_ucred.cr_groups, NGROUPS * sizeof(gid_t));
1098
1099	my_cred = NOCRED;
1100	smr_proc_task_leave();
1101
1102	ep->e_tdev = NODEV;
1103	if (pg != PGRP_NULL) {
1104	if (p->p_flag & P_CONTROLT) {
1105	session_lock(sess: sessp);
1106	ep->e_tdev = os_atomic_load(&sessp->s_ttydev, relaxed);
1107	ep->e_tpgid = sessp->s_ttypgrpid;
1108	session_unlock(sess: sessp);
1109	}
1110	if (SESS_LEADER(p, sessp)) {
1111	ep->e_flag \|= EPROC_SLEADER;
1112	}
1113	pgrp_rele(pgrp: pg);
1114	}
1115	}
1116
1117	/*
1118	* Fill in an eproc structure for the specified process.
1119	* bzeroed by our caller, so only set non-zero fields.
1120	*/
1121	STATIC void
1122	fill_user32_externproc(proc_t p, struct user32_extern_proc *__restrict exp)
1123	{
1124	exp->p_starttime.tv_sec = (user32_time_t)p->p_start.tv_sec;
1125	exp->p_starttime.tv_usec = p->p_start.tv_usec;
1126	exp->p_flag = p->p_flag;
1127	#if DEVELOPMENT \|\| DEBUG
1128	if (p->p_lflag & P_LTRACED && !bootarg_hide_process_traced) {
1129	#else
1130	if (p->p_lflag & P_LTRACED) {
1131	#endif
1132	exp->p_flag \|= P_TRACED;
1133	}
1134	if (p->p_lflag & P_LPPWAIT) {
1135	exp->p_flag \|= P_PPWAIT;
1136	}
1137	if (p->p_lflag & P_LEXIT) {
1138	exp->p_flag \|= P_WEXIT;
1139	}
1140	exp->p_stat = p->p_stat;
1141	exp->p_pid = proc_getpid(p);
1142	#if DEVELOPMENT \|\| DEBUG
1143	if (bootarg_hide_process_traced) {
1144	exp->p_oppid = `0`;
1145	} else
1146	#endif
1147	{
1148	exp->p_oppid = p->p_oppid;
1149	}
1150	/ Mach related /
1151	exp->p_debugger = p->p_debugger;
1152	exp->sigwait = p->sigwait;
1153	/ scheduling /
1154	#ifdef _PROC_HAS_SCHEDINFO_
1155	exp->p_estcpu = p->p_estcpu;
1156	exp->p_pctcpu = p->p_pctcpu;
1157	exp->p_slptime = p->p_slptime;
1158	#endif
1159	exp->p_realtimer.it_interval.tv_sec =
1160	(user32_time_t)p->p_realtimer.it_interval.tv_sec;
1161	exp->p_realtimer.it_interval.tv_usec =
1162	(__int32_t)p->p_realtimer.it_interval.tv_usec;
1163
1164	exp->p_realtimer.it_value.tv_sec =
1165	(user32_time_t)p->p_realtimer.it_value.tv_sec;
1166	exp->p_realtimer.it_value.tv_usec =
1167	(__int32_t)p->p_realtimer.it_value.tv_usec;
1168
1169	exp->p_rtime.tv_sec = (user32_time_t)p->p_rtime.tv_sec;
1170	exp->p_rtime.tv_usec = (__int32_t)p->p_rtime.tv_usec;
1171
1172	exp->p_sigignore = p->p_sigignore;
1173	exp->p_sigcatch = p->p_sigcatch;
1174	exp->p_priority = p->p_priority;
1175	exp->p_nice = p->p_nice;
1176	bcopy(src: &p->p_comm, dst: &exp->p_comm, MAXCOMLEN);
1177	exp->p_xstat = (u_short)MIN(p->p_xstat, USHRT_MAX);
1178	exp->p_acflag = p->p_acflag;
1179	}
1180
1181	/*
1182	* Fill in an LP64 version of extern_proc structure for the specified process.
1183	*/
1184	STATIC void
1185	fill_user64_externproc(proc_t p, struct user64_extern_proc *__restrict exp)
1186	{
1187	exp->p_starttime.tv_sec = p->p_start.tv_sec;
1188	exp->p_starttime.tv_usec = p->p_start.tv_usec;
1189	exp->p_flag = p->p_flag;
1190	#if DEVELOPMENT \|\| DEBUG
1191	if (p->p_lflag & P_LTRACED && !bootarg_hide_process_traced) {
1192	#else
1193	if (p->p_lflag & P_LTRACED) {
1194	#endif
1195	exp->p_flag \|= P_TRACED;
1196	}
1197	if (p->p_lflag & P_LPPWAIT) {
1198	exp->p_flag \|= P_PPWAIT;
1199	}
1200	if (p->p_lflag & P_LEXIT) {
1201	exp->p_flag \|= P_WEXIT;
1202	}
1203	exp->p_stat = p->p_stat;
1204	exp->p_pid = proc_getpid(p);
1205	#if DEVELOPMENT \|\| DEBUG
1206	if (bootarg_hide_process_traced) {
1207	exp->p_oppid = `0`;
1208	} else
1209	#endif
1210	{
1211	exp->p_oppid = p->p_oppid;
1212	}
1213	/ Mach related /
1214	exp->p_debugger = p->p_debugger;
1215	exp->sigwait = p->sigwait;
1216	/ scheduling /
1217	#ifdef _PROC_HAS_SCHEDINFO_
1218	exp->p_estcpu = p->p_estcpu;
1219	exp->p_pctcpu = p->p_pctcpu;
1220	exp->p_slptime = p->p_slptime;
1221	#endif
1222	exp->p_realtimer.it_interval.tv_sec = p->p_realtimer.it_interval.tv_sec;
1223	exp->p_realtimer.it_interval.tv_usec = p->p_realtimer.it_interval.tv_usec;
1224
1225	exp->p_realtimer.it_value.tv_sec = p->p_realtimer.it_value.tv_sec;
1226	exp->p_realtimer.it_value.tv_usec = p->p_realtimer.it_value.tv_usec;
1227
1228	exp->p_rtime.tv_sec = p->p_rtime.tv_sec;
1229	exp->p_rtime.tv_usec = p->p_rtime.tv_usec;
1230
1231	exp->p_sigignore = p->p_sigignore;
1232	exp->p_sigcatch = p->p_sigcatch;
1233	exp->p_priority = p->p_priority;
1234	exp->p_nice = p->p_nice;
1235	bcopy(src: &p->p_comm, dst: &exp->p_comm, MAXCOMLEN);
1236	exp->p_xstat = (u_short)MIN(p->p_xstat, USHRT_MAX);
1237	exp->p_acflag = p->p_acflag;
1238	}
1239
1240	STATIC void
1241	fill_user32_proc(proc_t p, struct user32_kinfo_proc *__restrict kp)
1242	{
1243	/ on a 64 bit kernel, 32 bit users get some truncated information /
1244	fill_user32_externproc(p, exp: &kp->kp_proc);
1245	fill_user32_eproc(p, ep: &kp->kp_eproc);
1246	}
1247
1248	STATIC void
1249	fill_user64_proc(proc_t p, struct user64_kinfo_proc *__restrict kp)
1250	{
1251	fill_user64_externproc(p, exp: &kp->kp_proc);
1252	fill_user64_eproc(p, ep: &kp->kp_eproc);
1253	}
1254
1255	#if defined(XNU_TARGET_OS_OSX)
1256	/*
1257	* Return the top *sizep bytes of the user stack, or the entire area of the
1258	* user stack down through the saved exec_path, whichever is smaller.
1259	*/
1260	STATIC int
1261	sysctl_doprocargs SYSCTL_HANDLER_ARGS
1262	{
1263	__unused int cmd = oidp->oid_arg2; / subcommand/
1264	int name = arg1; /* oid element argument vector /
1265	int namelen = arg2; / number of oid element arguments /
1266	user_addr_t oldp = req->oldptr; / user buffer copy out address /
1267	size_t oldlenp = &req->oldlen; /* user buffer copy out size /
1268	// user_addr_t newp = req->newptr; / user buffer copy in address /
1269	// size_t newlen = req->newlen; / user buffer copy in size /
1270	int error;
1271
1272	error = sysctl_procargsx( name, namelen, where: oldp, sizep: oldlenp, cur_proc: current_proc(), argc_yes: `0`);
1273
1274	/ adjust index so we return the right required/consumed amount /
1275	if (!error) {
1276	req->oldidx += req->oldlen;
1277	}
1278
1279	return error;
1280	}
1281	SYSCTL_PROC(_kern, KERN_PROCARGS, procargs, CTLTYPE_NODE \| CTLFLAG_RD \| CTLFLAG_LOCKED,
1282	`0`, / Pointer argument (arg1) /
1283	`0`, / Integer argument (arg2) /
1284	sysctl_doprocargs, / Handler function /
1285	NULL, / Data pointer /
1286	"");
1287	#endif /* defined(XNU_TARGET_OS_OSX) */
1288
1289	STATIC int
1290	sysctl_doprocargs2 SYSCTL_HANDLER_ARGS
1291	{
1292	__unused int cmd = oidp->oid_arg2; / subcommand/
1293	int name = arg1; /* oid element argument vector /
1294	int namelen = arg2; / number of oid element arguments /
1295	user_addr_t oldp = req->oldptr; / user buffer copy out address /
1296	size_t oldlenp = &req->oldlen; /* user buffer copy out size /
1297	// user_addr_t newp = req->newptr; / user buffer copy in address /
1298	// size_t newlen = req->newlen; / user buffer copy in size /
1299	int error;
1300
1301	error = sysctl_procargsx( name, namelen, where: oldp, sizep: oldlenp, cur_proc: current_proc(), argc_yes: `1`);
1302
1303	/ adjust index so we return the right required/consumed amount /
1304	if (!error) {
1305	req->oldidx += req->oldlen;
1306	}
1307
1308	return error;
1309	}
1310	SYSCTL_PROC(_kern, KERN_PROCARGS2, procargs2, CTLTYPE_NODE \| CTLFLAG_RD \| CTLFLAG_LOCKED,
1311	`0`, / Pointer argument (arg1) /
1312	`0`, / Integer argument (arg2) /
1313	sysctl_doprocargs2, / Handler function /
1314	NULL, / Data pointer /
1315	"");
1316
1317	#define SYSCTL_PROCARGS_READ_ENVVARS_ENTITLEMENT "com.apple.private.read-environment-variables"
1318	STATIC int
1319	sysctl_procargsx(int *name, u_int namelen, user_addr_t where,
1320	size_t sizep, proc_t cur_proc, int* argc_yes)
1321	{
1322	assert(sizep != NULL);
1323	proc_t p = NULL;
1324	size_t buflen = where != USER_ADDR_NULL ? *sizep : `0`;
1325	int error = `0`;
1326	struct _vm_map *proc_map = NULL;
1327	struct task * task;
1328	vm_map_copy_t tmp = NULL;
1329	user_addr_t arg_addr;
1330	size_t arg_size;
1331	caddr_t data;
1332	size_t argslen = `0`;
1333	size_t size = `0`;
1334	vm_offset_t copy_start = `0`, copy_end;
1335	vm_offset_t smallbuffer_start;
1336	kern_return_t ret;
1337	int pid;
1338	uid_t uid;
1339	int argc = -`1`;
1340	size_t argvsize;
1341	size_t remaining;
1342	size_t current_arg_index;
1343	size_t current_arg_len;
1344	const char * current_arg;
1345	bool omit_env_vars = true;
1346	user_addr_t user_stack;
1347	vm_map_offset_t effective_page_mask;
1348
1349	if (namelen < `1`) {
1350	error = EINVAL;
1351	goto finish;
1352	}
1353
1354	if (argc_yes) {
1355	buflen -= sizeof(int); / reserve first word to return argc /
1356	}
1357	/ we only care about buflen when where (oldp from sysctl) is not NULL. /
1358	/ when where (oldp from sysctl) is NULL and sizep (oldlenp from sysctl /
1359	/ is not NULL then the caller wants us to return the length needed to /
1360	/ hold the data we would return /
1361	if (where != USER_ADDR_NULL && (buflen <= `0` \|\| buflen > ARG_MAX)) {
1362	error = EINVAL;
1363	goto finish;
1364	}
1365
1366	/*
1367	* Lookup process by pid
1368	*/
1369	pid = name[`0`];
1370	p = proc_find(pid);
1371	if (p == NULL) {
1372	error = EINVAL;
1373	goto finish;
1374	}
1375
1376	/ Allow reading environment variables if any of the following are true:*
1377	* - kernel is DEVELOPMENT \|\| DEBUG
1378	* - target process is same as current_proc()
1379	* - target process is not cs_restricted
1380	* - SIP is off
1381	* - caller has an entitlement
1382	*/
1383
1384	#if DEVELOPMENT \|\| DEBUG
1385	omit_env_vars = false;
1386	#endif
1387	if (p == current_proc() \|\|
1388	!cs_restricted(p) \|\|
1389	#if CONFIG_CSR
1390	csr_check(CSR_ALLOW_UNRESTRICTED_DTRACE) == `0` \|\|
1391	#endif
1392	IOCurrentTaskHasEntitlement(SYSCTL_PROCARGS_READ_ENVVARS_ENTITLEMENT)
1393	) {
1394	omit_env_vars = false;
1395	}
1396
1397	/*
1398	* Copy the top N bytes of the stack.
1399	* On all machines we have so far, the stack grows
1400	* downwards.
1401	*
1402	* If the user expects no more than N bytes of
1403	* argument list, use that as a guess for the
1404	* size.
1405	*/
1406
1407	if (!p->user_stack) {
1408	error = EINVAL;
1409	goto finish;
1410	}
1411
1412	/ save off argc, argslen, user_stack before releasing the proc /
1413	argc = p->p_argc;
1414	argslen = p->p_argslen;
1415	user_stack = p->user_stack;
1416
1417	/*
1418	* When these sysctls were introduced, the first string in the strings
1419	* section was just the bare path of the executable. However, for security
1420	* reasons we now prefix this string with executable_path= so it can be
1421	* parsed getenv style. To avoid binary compatability issues with exising
1422	* callers of this sysctl, we strip it off here.
1423	* (rdar://problem/13746466)
1424	*/
1425	#define EXECUTABLE_KEY "executable_path="
1426	argslen -= strlen(EXECUTABLE_KEY);
1427
1428	if (where == USER_ADDR_NULL && !omit_env_vars) {
1429	/ caller only wants to know length of proc args data.*
1430	* If we don't need to omit environment variables, we can skip
1431	* copying the target process stack */
1432	goto calculate_size;
1433	}
1434
1435	smr_proc_task_enter();
1436	uid = kauth_cred_getuid(cred: proc_ucred_smr(p));
1437	smr_proc_task_leave();
1438
1439	if ((uid != kauth_cred_getuid(cred: kauth_cred_get()))
1440	&& suser(cred: kauth_cred_get(), acflag: &cur_proc->p_acflag)) {
1441	error = EINVAL;
1442	goto finish;
1443	}
1444
1445	/*
1446	* Before we can block (any VM code), make another
1447	* reference to the map to keep it alive. We do
1448	* that by getting a reference on the task itself.
1449	*/
1450	task = proc_task(p);
1451	if (task == NULL) {
1452	error = EINVAL;
1453	goto finish;
1454	}
1455
1456	/*
1457	* Once we have a task reference we can convert that into a
1458	* map reference, which we will use in the calls below. The
1459	* task/process may change its map after we take this reference
1460	* (see execve), but the worst that will happen then is a return
1461	* of stale info (which is always a possibility).
1462	*/
1463	task_reference(task);
1464	proc_rele(p);
1465	p = NULL;
1466	proc_map = get_task_map_reference(task);
1467	task_deallocate(task);
1468
1469	if (proc_map == NULL) {
1470	error = EINVAL;
1471	goto finish;
1472	}
1473
1474	effective_page_mask = vm_map_page_mask(map: proc_map);
1475
1476	arg_size = vm_map_round_page(argslen, effective_page_mask);
1477
1478	arg_addr = user_stack - arg_size;
1479
1480	ret = kmem_alloc(map: kernel_map, addrp: &copy_start, size: arg_size,
1481	flags: KMA_DATA \| KMA_ZERO, VM_KERN_MEMORY_BSD);
1482	if (ret != KERN_SUCCESS) {
1483	error = ENOMEM;
1484	goto finish;
1485	}
1486
1487	copy_end = copy_start + arg_size;
1488
1489	if (vm_map_copyin(src_map: proc_map, src_addr: (vm_map_address_t)arg_addr,
1490	len: (vm_map_size_t)arg_size, FALSE, copy_result: &tmp) != KERN_SUCCESS) {
1491	error = EIO;
1492	goto finish;
1493	}
1494
1495	/*
1496	* Now that we've done the copyin from the process'
1497	* map, we can release the reference to it.
1498	*/
1499	vm_map_deallocate(map: proc_map);
1500	proc_map = NULL;
1501
1502	if (vm_map_copy_overwrite(dst_map: kernel_map,
1503	dst_addr: (vm_map_address_t)copy_start,
1504	copy: tmp, copy_size: (vm_map_size_t) arg_size, FALSE) != KERN_SUCCESS) {
1505	error = EIO;
1506	goto finish;
1507	}
1508	/ tmp was consumed /
1509	tmp = NULL;
1510
1511	if (omit_env_vars) {
1512	argvsize = `0`;
1513
1514	/ Iterate over everything in argv, plus one for the bare executable path /
1515	for (current_arg_index = `0`; current_arg_index < argc + `1` && argvsize < argslen; ++current_arg_index) {
1516	current_arg = (const char *)(copy_end - argslen) + argvsize;
1517	remaining = argslen - argvsize;
1518	current_arg_len = strnlen(s: current_arg, n: remaining);
1519	if (current_arg_len < remaining) {
1520	/ We have space for the null terminator /
1521	current_arg_len += `1`;
1522
1523	if (current_arg_index == `0`) {
1524	/ The bare executable path may have multiple null bytes after it for alignment /
1525	while (current_arg_len < remaining && current_arg[current_arg_len] == `0`) {
1526	current_arg_len += `1`;
1527	}
1528	}
1529	}
1530	argvsize += current_arg_len;
1531	}
1532	assert(argvsize <= argslen);
1533
1534	/ Adjust argslen and copy_end to make the copyout range extend to the end of argv /
1535	copy_end = copy_end - argslen + argvsize;
1536	argslen = argvsize;
1537	}
1538
1539	if (where == USER_ADDR_NULL) {
1540	/ Skip copyout /
1541	goto calculate_size;
1542	}
1543
1544	if (buflen >= argslen) {
1545	data = (caddr_t) (copy_end - argslen);
1546	size = argslen;
1547	} else {
1548	/*
1549	* Before rdar://25397314, this function contained incorrect logic when buflen is less
1550	* than argslen. The problem was that it copied in `buflen` bytes from the end of the target
1551	* process user stack into the beginning of a buffer of size round_page(buflen), and then
1552	* copied out `buflen` bytes from the end of this buffer. The effect of this was that
1553	* the caller of this sysctl would get zeros at the end of their buffer.
1554	*
1555	* To preserve this behavior, bzero everything from copy_end-round_page(buflen)+buflen to the
1556	* end of the buffer. This emulates copying in only `buflen` bytes.
1557	*
1558	*
1559	* In the old code:
1560	*
1561	* copy_start .... size: round_page(buflen) .... copy_end
1562	* [---copied in data (size: buflen)---\|--- zeros ----------]
1563	* ^
1564	* data = copy_end - buflen
1565	*
1566	*
1567	* In the new code:
1568	* copy_start .... size: round_page(p->argslen) .... full copy_end
1569	* ^ ....................... p->argslen ...............................^
1570	* ^ ^ truncated copy_end ^
1571	* ^ ^ ^ ^
1572	* ^ ................ argslen ........................ ^
1573	* ^ ^ ^ ^
1574	* [-------copied in data (size: round_page(p->argslen))-------:----env vars---]
1575	* ^ ^
1576	* ^ data = copy_end - buflen
1577	* smallbuffer_start = max(copy_end - round_page(buflen), copy_start)
1578	*
1579	*
1580	* Full copy_end: copy_end calculated from copy_start + round_page(p->argslen)
1581	* Truncated copy_end: copy_end after truncation to remove environment variables.
1582	*
1583	* If environment variables were omitted, then we use the truncated copy_end, otherwise
1584	* we use full copy_end.
1585	*
1586	* smallbuffer_start: represents where copy_start would be in the old code.
1587	* data: The beginning of the region we copyout
1588	*/
1589	smallbuffer_start = copy_end - vm_map_round_page(buflen, effective_page_mask);
1590	if (smallbuffer_start < copy_start) {
1591	smallbuffer_start = copy_start;
1592	}
1593	bzero(s: (void *)(smallbuffer_start + buflen), n: copy_end - (smallbuffer_start + buflen));
1594	data = (caddr_t) (copy_end - buflen);
1595	size = buflen;
1596	}
1597
1598	if (argc_yes) {
1599	/ Put processes argc as the first word in the copyout buffer /
1600	suword(addr: where, word: argc);
1601	error = copyout(data, (where + sizeof(int)), size);
1602	size += sizeof(int);
1603	} else {
1604	error = copyout(data, where, size);
1605
1606	/*
1607	* Make the old PROCARGS work to return the executable's path
1608	* But, only if there is enough space in the provided buffer
1609	*
1610	* on entry: data [possibily] points to the beginning of the path
1611	*
1612	* Note: we keep all pointers&sizes aligned to word boundries
1613	*/
1614	if ((!error) && (buflen > `0` && (u_int)buflen > size)) {
1615	int binPath_sz, alignedBinPath_sz = `0`;
1616	int extraSpaceNeeded, addThis;
1617	user_addr_t placeHere;
1618	char * str = (char *) data;
1619	size_t max_len = size;
1620
1621	/ Some apps are really bad about messing up their stacks*
1622	* So, we have to be extra careful about getting the length
1623	* of the executing binary. If we encounter an error, we bail.
1624	*/
1625
1626	/ Limit ourselves to PATH_MAX paths /
1627	if (max_len > PATH_MAX) {
1628	max_len = PATH_MAX;
1629	}
1630
1631	binPath_sz = `0`;
1632
1633	while ((binPath_sz < max_len - `1`) && (*str++ != `0`)) {
1634	binPath_sz++;
1635	}
1636
1637	/ If we have a NUL terminator, copy it, too /
1638	if (binPath_sz < max_len - `1`) {
1639	binPath_sz += `1`;
1640	}
1641
1642	/ Pre-Flight the space requiremnts /
1643
1644	/ Account for the padding that fills out binPath to the next word /
1645	alignedBinPath_sz += (binPath_sz & (sizeof(int) - `1`)) ? (sizeof(int) - (binPath_sz & (sizeof(int) - `1`))) : `0`;
1646
1647	placeHere = where + size;
1648
1649	/ Account for the bytes needed to keep placeHere word aligned /
1650	addThis = (placeHere & (sizeof(int) - `1`)) ? (sizeof(int) - (placeHere & (sizeof(int) - `1`))) : `0`;
1651
1652	/ Add up all the space that is needed /
1653	extraSpaceNeeded = alignedBinPath_sz + addThis + binPath_sz + (`4` * sizeof(int));
1654
1655	/ is there is room to tack on argv[0]? /
1656	if ((buflen & ~(sizeof(int) - `1`)) >= (size + extraSpaceNeeded)) {
1657	placeHere += addThis;
1658	suword(addr: placeHere, word: `0`);
1659	placeHere += sizeof(int);
1660	suword(addr: placeHere, word: `0xBFFF0000`);
1661	placeHere += sizeof(int);
1662	suword(addr: placeHere, word: `0`);
1663	placeHere += sizeof(int);
1664	error = copyout(data, placeHere, binPath_sz);
1665	if (!error) {
1666	placeHere += binPath_sz;
1667	suword(addr: placeHere, word: `0`);
1668	size += extraSpaceNeeded;
1669	}
1670	}
1671	}
1672	}
1673
1674	calculate_size:
1675	/ Size has already been calculated for the where != NULL case /
1676	if (where == USER_ADDR_NULL) {
1677	size = argslen;
1678	if (argc_yes) {
1679	size += sizeof(int);
1680	} else {
1681	/*
1682	* old PROCARGS will return the executable's path and plus some
1683	* extra space for work alignment and data tags
1684	*/
1685	size += PATH_MAX + (`6` * sizeof(int));
1686	}
1687	size += (size & (sizeof(int) - `1`)) ? (sizeof(int) - (size & (sizeof(int) - `1`))) : `0`;
1688	}
1689
1690	*sizep = size;
1691
1692	finish:
1693	if (p != NULL) {
1694	proc_rele(p);
1695	}
1696	if (tmp != NULL) {
1697	vm_map_copy_discard(copy: tmp);
1698	}
1699	if (proc_map != NULL) {
1700	vm_map_deallocate(map: proc_map);
1701	}
1702	if (copy_start != (vm_offset_t) `0`) {
1703	kmem_free(map: kernel_map, addr: copy_start, size: arg_size);
1704	}
1705	return error;
1706	}
1707
1708
1709	/*
1710	* Max number of concurrent aio requests
1711	*/
1712	STATIC int
1713	sysctl_aiomax
1714	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
1715	{
1716	int new_value, changed;
1717	int error = sysctl_io_number(req, bigValue: aio_max_requests, valueSize: sizeof(int), pValue: &new_value, changed: &changed);
1718	if (changed) {
1719	/ make sure the system-wide limit is greater than the per process limit /
1720	if (new_value >= aio_max_requests_per_process && new_value <= AIO_MAX_REQUESTS) {
1721	aio_max_requests = new_value;
1722	} else {
1723	error = EINVAL;
1724	}
1725	}
1726	return error;
1727	}
1728
1729
1730	/*
1731	* Max number of concurrent aio requests per process
1732	*/
1733	STATIC int
1734	sysctl_aioprocmax
1735	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
1736	{
1737	int new_value, changed;
1738	int error = sysctl_io_number(req, bigValue: aio_max_requests_per_process, valueSize: sizeof(int), pValue: &new_value, changed: &changed);
1739	if (changed) {
1740	/ make sure per process limit is less than the system-wide limit /
1741	if (new_value <= aio_max_requests && new_value >= AIO_LISTIO_MAX) {
1742	aio_max_requests_per_process = new_value;
1743	} else {
1744	error = EINVAL;
1745	}
1746	}
1747	return error;
1748	}
1749
1750
1751	/*
1752	* Max number of async IO worker threads
1753	*/
1754	STATIC int
1755	sysctl_aiothreads
1756	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
1757	{
1758	int new_value, changed;
1759	int error = sysctl_io_number(req, bigValue: aio_worker_threads, valueSize: sizeof(int), pValue: &new_value, changed: &changed);
1760	if (changed) {
1761	/ we only allow an increase in the number of worker threads /
1762	if (new_value > aio_worker_threads) {
1763	_aio_create_worker_threads(num: (new_value - aio_worker_threads));
1764	aio_worker_threads = new_value;
1765	} else {
1766	error = EINVAL;
1767	}
1768	}
1769	return error;
1770	}
1771
1772
1773	/*
1774	* System-wide limit on the max number of processes
1775	*/
1776	STATIC int
1777	sysctl_maxproc
1778	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
1779	{
1780	int new_value, changed;
1781	int error = sysctl_io_number(req, bigValue: maxproc, valueSize: sizeof(int), pValue: &new_value, changed: &changed);
1782	if (changed) {
1783	AUDIT_ARG(value32, new_value);
1784	/ make sure the system-wide limit is less than the configured hard*
1785	* limit set at kernel compilation */
1786	if (new_value <= hard_maxproc && new_value > `0`) {
1787	maxproc = new_value;
1788	} else {
1789	error = EINVAL;
1790	}
1791	}
1792	return error;
1793	}
1794
1795	extern int sched_enable_smt;
1796	STATIC int
1797	sysctl_sched_enable_smt
1798	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
1799	{
1800	int new_value, changed;
1801	int error = sysctl_io_number(req, bigValue: sched_enable_smt, valueSize: sizeof(int), pValue: &new_value, changed: &changed);
1802	if (error) {
1803	return error;
1804	}
1805	kern_return_t kret = KERN_SUCCESS;
1806	if (changed) {
1807	AUDIT_ARG(value32, new_value);
1808	if (new_value == `0`) {
1809	sched_enable_smt = `0`;
1810	kret = enable_smt_processors(false);
1811	} else {
1812	sched_enable_smt = `1`;
1813	kret = enable_smt_processors(true);
1814	}
1815	}
1816	switch (kret) {
1817	case KERN_SUCCESS:
1818	error = `0`;
1819	break;
1820	case KERN_INVALID_ARGUMENT:
1821	error = EINVAL;
1822	break;
1823	case KERN_FAILURE:
1824	error = EBUSY;
1825	break;
1826	default:
1827	error = ENOENT;
1828	break;
1829	}
1830
1831	return error;
1832	}
1833
1834	SYSCTL_STRING(_kern, KERN_OSTYPE, ostype,
1835	CTLFLAG_RD \| CTLFLAG_KERN \| CTLFLAG_LOCKED,
1836	ostype, `0`, "");
1837	SYSCTL_STRING(_kern, KERN_OSRELEASE, osrelease,
1838	CTLFLAG_RD \| CTLFLAG_KERN \| CTLFLAG_LOCKED,
1839	osrelease, `0`, "");
1840	SYSCTL_INT(_kern, KERN_OSREV, osrevision,
1841	CTLFLAG_RD \| CTLFLAG_KERN \| CTLFLAG_LOCKED,
1842	(int *)NULL, BSD, "");
1843	SYSCTL_STRING(_kern, KERN_VERSION, version,
1844	CTLFLAG_RD \| CTLFLAG_KERN \| CTLFLAG_LOCKED,
1845	version, `0`, "");
1846	SYSCTL_STRING(_kern, OID_AUTO, uuid,
1847	CTLFLAG_RD \| CTLFLAG_KERN \| CTLFLAG_LOCKED,
1848	&kernel_uuid_string[`0`], `0`, "");
1849
1850	SYSCTL_STRING(_kern, OID_AUTO, osbuildconfig,
1851	CTLFLAG_RD \| CTLFLAG_KERN \| CTLFLAG_LOCKED \| CTLFLAG_MASKED,
1852	&osbuild_config[`0`], `0`, "");
1853
1854	#if DEBUG
1855	#ifndef DKPR
1856	#define DKPR 1
1857	#endif
1858	#endif
1859
1860	#if DKPR
1861	int debug_kprint_syscall = `0`;
1862	char debug_kprint_syscall_process[MAXCOMLEN + `1`];
1863
1864	/ Thread safe: bits and string value are not used to reclaim state /
1865	SYSCTL_INT(_debug, OID_AUTO, kprint_syscall,
1866	CTLFLAG_RW \| CTLFLAG_LOCKED, &debug_kprint_syscall, `0`, "kprintf syscall tracing");
1867	SYSCTL_STRING(_debug, OID_AUTO, kprint_syscall_process,
1868	CTLFLAG_RW \| CTLFLAG_LOCKED, debug_kprint_syscall_process, sizeof(debug_kprint_syscall_process),
1869	"name of process for kprintf syscall tracing");
1870
1871	int
1872	debug_kprint_current_process(const char **namep)
1873	{
1874	struct proc *p = current_proc();
1875
1876	if (p == NULL) {
1877	return `0`;
1878	}
1879
1880	if (debug_kprint_syscall_process[`0`]) {
1881	/ user asked to scope tracing to a particular process name /
1882	if (`0` == strncmp(debug_kprint_syscall_process,
1883	p->p_comm, sizeof(debug_kprint_syscall_process))) {
1884	/ no value in telling the user that we traced what they asked /
1885	if (namep) {
1886	*namep = NULL;
1887	}
1888
1889	return `1`;
1890	} else {
1891	return `0`;
1892	}
1893	}
1894
1895	/ trace all processes. Tell user what we traced /
1896	if (namep) {
1897	*namep = p->p_comm;
1898	}
1899
1900	return `1`;
1901	}
1902	#endif
1903
1904	/ PR-5293665: need to use a callback function for kern.osversion to set*
1905	* osversion in IORegistry */
1906
1907	STATIC int
1908	sysctl_osversion(__unused struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req)
1909	{
1910	int rval = `0`;
1911
1912	rval = sysctl_handle_string(oidp, arg1, arg2, req);
1913
1914	if (req->newptr) {
1915	IORegistrySetOSBuildVersion(build_version: (char *)arg1);
1916	}
1917
1918	return rval;
1919	}
1920
1921	SYSCTL_PROC(_kern, KERN_OSVERSION, osversion,
1922	CTLFLAG_RW \| CTLFLAG_KERN \| CTLTYPE_STRING \| CTLFLAG_LOCKED,
1923	osversion, `256` / OSVERSIZE/,
1924	sysctl_osversion, "A", "");
1925
1926	static bool
1927	_already_set_or_not_launchd(struct sysctl_req req, char* *val)
1928	{
1929	if (req->newptr != `0`) {
1930	/*
1931	* Can only ever be set by launchd, and only once at boot.
1932	*/
1933	if (proc_getpid(req->p) != `1` \|\| val[`0`] != `'\0'`) {
1934	return true;
1935	}
1936	}
1937	return false;
1938	}
1939
1940	#define kRootsInstalledReadWriteEntitlement "com.apple.private.roots-installed-read-write"
1941	#define kRootsInstalledReadOnlyEntitlement "com.apple.private.roots-installed-read-only"
1942	uint64_t roots_installed = `0`;
1943
1944	static int
1945	sysctl_roots_installed
1946	(__unused struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req)
1947	{
1948	int error = `0`;
1949
1950	if (req->newptr != `0`) {
1951	/ a ReadWrite entitlement is required for updating this syscl*
1952	* meanwhile, only allow write once
1953	*/
1954	if (!IOCurrentTaskHasEntitlement(kRootsInstalledReadWriteEntitlement) \|\| (roots_installed != `0`)) {
1955	return EPERM;
1956	}
1957	} else {
1958	/ for reader of this sysctl, need either ReadWrite or ReadOnly entitlement /
1959	if (!IOCurrentTaskHasEntitlement(kRootsInstalledReadWriteEntitlement) &&
1960	!IOCurrentTaskHasEntitlement(kRootsInstalledReadOnlyEntitlement)) {
1961	return EPERM;
1962	}
1963	}
1964
1965	error = sysctl_handle_quad(oidp, arg1, arg2, req);
1966
1967	return error;
1968	}
1969
1970	SYSCTL_PROC(_kern, OID_AUTO, roots_installed,
1971	CTLTYPE_QUAD \| CTLFLAG_RW \| CTLFLAG_ANYBODY \| CTLFLAG_LOCKED,
1972	&roots_installed, sizeof(roots_installed),
1973	sysctl_roots_installed, "Q", "");
1974
1975	#if XNU_TARGET_OS_OSX
1976	static int
1977	sysctl_system_version_compat
1978	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
1979	{
1980	int oldval = (task_has_system_version_compat_enabled(task: current_task()));
1981	int new_value = `0`, changed = `0`;
1982
1983	int error = sysctl_io_number(req, bigValue: oldval, valueSize: sizeof(int), pValue: &new_value, changed: &changed);
1984	if (changed) {
1985	task_set_system_version_compat_enabled(task: current_task(), enable_system_version_compat: (new_value));
1986	}
1987	return error;
1988	}
1989
1990	SYSCTL_PROC(_kern, OID_AUTO, system_version_compat,
1991	CTLTYPE_INT \| CTLFLAG_RW \| CTLFLAG_ANYBODY \| CTLFLAG_LOCKED,
1992	`0`, `0`, sysctl_system_version_compat, "A", "");
1993
1994	char osproductversioncompat[`48`] = { `'\0'` };
1995
1996	static int
1997	sysctl_osproductversioncompat(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req)
1998	{
1999	if (_already_set_or_not_launchd(req, val: osproductversioncompat)) {
2000	return EPERM;
2001	}
2002	return sysctl_handle_string(oidp, arg1, arg2, req);
2003	}
2004
2005
2006	SYSCTL_PROC(_kern, OID_AUTO, osproductversioncompat,
2007	CTLFLAG_RW \| CTLFLAG_KERN \| CTLTYPE_STRING \| CTLFLAG_LOCKED,
2008	osproductversioncompat, sizeof(osproductversioncompat),
2009	sysctl_osproductversioncompat, "A", "The ProductVersion from SystemVersionCompat.plist");
2010	#endif
2011
2012	char osproductversion[`48`] = { `'\0'` };
2013
2014	static char iossupportversion_string[`48`] = { `'\0'` };
2015
2016	static int
2017	sysctl_osproductversion(__unused struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req)
2018	{
2019	if (_already_set_or_not_launchd(req, val: osproductversion)) {
2020	return EPERM;
2021	}
2022
2023	#if XNU_TARGET_OS_OSX
2024	if (task_has_system_version_compat_enabled(task: current_task()) && (osproductversioncompat[`0`] != `'\0'`)) {
2025	return sysctl_handle_string(oidp, arg1: osproductversioncompat, arg2, req);
2026	} else {
2027	return sysctl_handle_string(oidp, arg1, arg2, req);
2028	}
2029	#else
2030	return sysctl_handle_string(oidp, arg1, arg2, req);
2031	#endif
2032	}
2033
2034	#if XNU_TARGET_OS_OSX
2035	static_assert(sizeof(osproductversioncompat) == sizeof(osproductversion),
2036	"osproductversion size matches osproductversioncompat size");
2037	#endif
2038
2039	SYSCTL_PROC(_kern, OID_AUTO, osproductversion,
2040	CTLFLAG_RW \| CTLFLAG_KERN \| CTLTYPE_STRING \| CTLFLAG_LOCKED,
2041	osproductversion, sizeof(osproductversion),
2042	sysctl_osproductversion, "A", "The ProductVersion from SystemVersion.plist");
2043
2044	char osreleasetype[`48`] = { `'\0'` };
2045
2046	STATIC int
2047	sysctl_osreleasetype(__unused struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req)
2048	{
2049	if (_already_set_or_not_launchd(req, val: osreleasetype)) {
2050	return EPERM;
2051	}
2052	return sysctl_handle_string(oidp, arg1, arg2, req);
2053	}
2054
2055	void reset_osreleasetype(void);
2056
2057	void
2058	reset_osreleasetype(void)
2059	{
2060	memset(s: osreleasetype, c: `0`, n: sizeof(osreleasetype));
2061	}
2062
2063	SYSCTL_PROC(_kern, OID_AUTO, osreleasetype,
2064	CTLFLAG_RW \| CTLFLAG_KERN \| CTLTYPE_STRING \| CTLFLAG_LOCKED,
2065	osreleasetype, sizeof(osreleasetype),
2066	sysctl_osreleasetype, "A", "The ReleaseType from SystemVersion.plist");
2067
2068	STATIC int
2069	sysctl_iossupportversion(__unused struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req)
2070	{
2071	if (_already_set_or_not_launchd(req, val: iossupportversion_string)) {
2072	return EPERM;
2073	}
2074
2075	return sysctl_handle_string(oidp, arg1, arg2, req);
2076	}
2077
2078	SYSCTL_PROC(_kern, OID_AUTO, iossupportversion,
2079	CTLFLAG_RW \| CTLFLAG_KERN \| CTLTYPE_STRING \| CTLFLAG_LOCKED,
2080	iossupportversion_string, sizeof(iossupportversion_string),
2081	sysctl_iossupportversion, "A", "The iOSSupportVersion from SystemVersion.plist");
2082
2083	static uint64_t osvariant_status = `0`;
2084
2085	STATIC int
2086	sysctl_osvariant_status(__unused struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req)
2087	{
2088	if (req->newptr != `0`) {
2089	/*
2090	* Can only ever be set by launchd, and only once.
2091	* Reset by usrctl() -> reset_osvariant_status() during
2092	* userspace reboot, since userspace could reboot into
2093	* a different variant.
2094	*/
2095	if (proc_getpid(req->p) != `1` \|\| osvariant_status != `0`) {
2096	return EPERM;
2097	}
2098	}
2099
2100	int err = sysctl_handle_quad(oidp, arg1, arg2, req);
2101
2102	reset_debug_syscall_rejection_mode();
2103
2104	return err;
2105	}
2106
2107	SYSCTL_PROC(_kern, OID_AUTO, osvariant_status,
2108	CTLFLAG_RW \| CTLTYPE_QUAD \| CTLFLAG_LOCKED \| CTLFLAG_MASKED,
2109	&osvariant_status, sizeof(osvariant_status),
2110	sysctl_osvariant_status, "Q", "Opaque flags used to cache OS variant information");
2111
2112	static bool
2113	_os_variant_check_disabled(enum os_variant_property property)
2114	{
2115	return (osvariant_status >> (`32` + property)) & `0x1`;
2116	}
2117
2118	static bool
2119	_os_variant_has(enum os_variant_status_flags_positions p)
2120	{
2121	return ((osvariant_status >> (p * OS_VARIANT_STATUS_BIT_WIDTH)) & OS_VARIANT_STATUS_MASK) == OS_VARIANT_S_YES;
2122	}
2123
2124	bool
2125	os_variant_has_internal_diagnostics(__unused const char *subsystem)
2126	{
2127	if (_os_variant_check_disabled(property: OS_VARIANT_PROPERTY_DIAGNOSTICS)) {
2128	return false;
2129	}
2130	#if XNU_TARGET_OS_OSX
2131	return _os_variant_has(p: OS_VARIANT_SFP_INTERNAL_CONTENT) \|\| _os_variant_has(p: OS_VARIANT_SFP_INTERNAL_DIAGS_PROFILE);
2132	#else
2133	return _os_variant_has(OS_VARIANT_SFP_INTERNAL_RELEASE_TYPE);
2134	#endif /* XNU_TARGET_OS_OSX */
2135	}
2136
2137	void reset_osvariant_status(void);
2138
2139	void
2140	reset_osvariant_status(void)
2141	{
2142	osvariant_status = `0`;
2143	reset_debug_syscall_rejection_mode();
2144	}
2145
2146	extern void commpage_update_dyld_flags(uint64_t);
2147	TUNABLE_WRITEABLE(uint64_t, dyld_flags, "dyld_flags", `0`);
2148
2149	STATIC int
2150	sysctl_dyld_flags(__unused struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req)
2151	{
2152	/*
2153	* Can only ever be set by launchd, possibly several times
2154	* as dyld may change its mind after a userspace reboot.
2155	*/
2156	if (req->newptr != `0` && proc_getpid(req->p) != `1`) {
2157	return EPERM;
2158	}
2159
2160	int res = sysctl_handle_quad(oidp, arg1, arg2, req);
2161	if (req->newptr && res == `0`) {
2162	commpage_update_dyld_flags(dyld_flags);
2163	}
2164	return res;
2165	}
2166
2167	SYSCTL_PROC(_kern, OID_AUTO, dyld_flags,
2168	CTLFLAG_RW \| CTLTYPE_QUAD \| CTLFLAG_LOCKED \| CTLFLAG_MASKED,
2169	&dyld_flags, sizeof(dyld_flags),
2170	sysctl_dyld_flags, "Q", "Opaque flags used to cache dyld system-wide configuration");
2171
2172	#if defined(XNU_TARGET_OS_BRIDGE)
2173	char macosproductversion[MACOS_VERS_LEN] = { `'\0'` };
2174
2175	SYSCTL_STRING(_kern, OID_AUTO, macosproductversion,
2176	CTLFLAG_RW \| CTLFLAG_KERN \| CTLFLAG_LOCKED,
2177	&macosproductversion[`0`], MACOS_VERS_LEN, "The currently running macOS ProductVersion (from SystemVersion.plist on macOS)");
2178
2179	char macosversion[MACOS_VERS_LEN] = { `'\0'` };
2180
2181	SYSCTL_STRING(_kern, OID_AUTO, macosversion,
2182	CTLFLAG_RW \| CTLFLAG_KERN \| CTLFLAG_LOCKED,
2183	&macosversion[`0`], MACOS_VERS_LEN, "The currently running macOS build version");
2184	#endif
2185
2186	STATIC int
2187	sysctl_sysctl_bootargs
2188	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
2189	{
2190	int error;
2191	char buf[BOOT_LINE_LENGTH];
2192
2193	strlcpy(dst: buf, src: PE_boot_args(), BOOT_LINE_LENGTH);
2194	error = sysctl_io_string(req, pValue: buf, BOOT_LINE_LENGTH, trunc: `0`, NULL);
2195	return error;
2196	}
2197
2198	SYSCTL_PROC(_kern, OID_AUTO, bootargs,
2199	CTLFLAG_LOCKED \| CTLFLAG_RD \| CTLFLAG_KERN \| CTLTYPE_STRING,
2200	NULL, `0`,
2201	sysctl_sysctl_bootargs, "A", "bootargs");
2202
2203	STATIC int
2204	sysctl_kernelcacheuuid(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req)
2205	{
2206	int rval = ENOENT;
2207	if (kernelcache_uuid_valid) {
2208	rval = sysctl_handle_string(oidp, arg1, arg2, req);
2209	}
2210	return rval;
2211	}
2212
2213	SYSCTL_PROC(_kern, OID_AUTO, kernelcacheuuid,
2214	CTLFLAG_RD \| CTLFLAG_KERN \| CTLTYPE_STRING \| CTLFLAG_LOCKED,
2215	kernelcache_uuid_string, sizeof(kernelcache_uuid_string),
2216	sysctl_kernelcacheuuid, "A", "");
2217
2218	STATIC int
2219	sysctl_systemfilesetuuid(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req)
2220	{
2221	int rval = ENOENT;
2222	if (pageablekc_uuid_valid) {
2223	rval = sysctl_handle_string(oidp, arg1, arg2, req);
2224	}
2225	return rval;
2226	}
2227
2228	SYSCTL_PROC(_kern, OID_AUTO, systemfilesetuuid,
2229	CTLFLAG_RD \| CTLFLAG_KERN \| CTLTYPE_STRING \| CTLFLAG_LOCKED,
2230	pageablekc_uuid_string, sizeof(pageablekc_uuid_string),
2231	sysctl_systemfilesetuuid, "A", "");
2232
2233	STATIC int
2234	sysctl_auxiliaryfilesetuuid(struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req)
2235	{
2236	int rval = ENOENT;
2237	if (auxkc_uuid_valid) {
2238	rval = sysctl_handle_string(oidp, arg1, arg2, req);
2239	}
2240	return rval;
2241	}
2242
2243	SYSCTL_PROC(_kern, OID_AUTO, auxiliaryfilesetuuid,
2244	CTLFLAG_RD \| CTLFLAG_KERN \| CTLTYPE_STRING \| CTLFLAG_LOCKED,
2245	auxkc_uuid_string, sizeof(auxkc_uuid_string),
2246	sysctl_auxiliaryfilesetuuid, "A", "");
2247
2248	STATIC int
2249	sysctl_filesetuuid(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
2250	{
2251	int rval = ENOENT;
2252	kc_format_t kcformat;
2253	kernel_mach_header_t *mh;
2254	void *uuid = NULL;
2255	unsigned long uuidlen = `0`;
2256	uuid_string_t uuid_str;
2257
2258	if (!PE_get_primary_kc_format(type: &kcformat) \|\| kcformat != KCFormatFileset) {
2259	return rval;
2260	}
2261
2262	mh = (kernel_mach_header_t *)PE_get_kc_header(type: KCKindPrimary);
2263	uuid = getuuidfromheader(mh, &uuidlen);
2264
2265	if ((uuid != NULL) && (uuidlen == sizeof(uuid_t))) {
2266	uuid_unparse_upper(uu: (uuid_t )uuid, out: uuid_str);
2267	rval = sysctl_io_string(req, pValue: (char )uuid_str, valueSize: sizeof*(uuid_str), trunc: `0`, NULL);
2268	}
2269
2270	return rval;
2271	}
2272
2273	SYSCTL_PROC(_kern, OID_AUTO, filesetuuid,
2274	CTLFLAG_RD \| CTLFLAG_KERN \| CTLTYPE_STRING \| CTLFLAG_LOCKED,
2275	NULL, `0`,
2276	sysctl_filesetuuid, "A", "");
2277
2278
2279	SYSCTL_INT(_kern, KERN_MAXFILES, maxfiles,
2280	CTLFLAG_RW \| CTLFLAG_KERN \| CTLFLAG_LOCKED,
2281	&maxfiles, `0`, "");
2282	SYSCTL_INT(_kern, KERN_ARGMAX, argmax,
2283	CTLFLAG_RD \| CTLFLAG_KERN \| CTLFLAG_LOCKED,
2284	(int *)NULL, ARG_MAX, "");
2285	SYSCTL_INT(_kern, KERN_POSIX1, posix1version,
2286	CTLFLAG_RD \| CTLFLAG_KERN \| CTLFLAG_LOCKED,
2287	(int *)NULL, _POSIX_VERSION, "");
2288	SYSCTL_INT(_kern, KERN_NGROUPS, ngroups,
2289	CTLFLAG_RD \| CTLFLAG_KERN \| CTLFLAG_LOCKED,
2290	(int *)NULL, NGROUPS_MAX, "");
2291	SYSCTL_INT(_kern, KERN_JOB_CONTROL, job_control,
2292	CTLFLAG_RD \| CTLFLAG_KERN \| CTLFLAG_LOCKED,
2293	(int *)NULL, `1`, "");
2294	#if 1 /* _POSIX_SAVED_IDS from <unistd.h> */
2295	SYSCTL_INT(_kern, KERN_SAVED_IDS, saved_ids,
2296	CTLFLAG_RD \| CTLFLAG_KERN \| CTLFLAG_LOCKED,
2297	(int *)NULL, `1`, "");
2298	#else
2299	SYSCTL_INT(_kern, KERN_SAVED_IDS, saved_ids,
2300	CTLFLAG_RD \| CTLFLAG_KERN \| CTLFLAG_LOCKED,
2301	NULL, `0`, "");
2302	#endif
2303	SYSCTL_INT(_kern, OID_AUTO, num_files,
2304	CTLFLAG_RD \| CTLFLAG_LOCKED,
2305	&nfiles, `0`, "");
2306	SYSCTL_COMPAT_INT(_kern, OID_AUTO, num_vnodes,
2307	CTLFLAG_RD \| CTLFLAG_LOCKED,
2308	&numvnodes, `0`, "");
2309	SYSCTL_INT(_kern, OID_AUTO, num_tasks,
2310	CTLFLAG_RD \| CTLFLAG_LOCKED,
2311	&task_max, `0`, "");
2312	SYSCTL_INT(_kern, OID_AUTO, num_threads,
2313	CTLFLAG_RD \| CTLFLAG_LOCKED,
2314	&thread_max, `0`, "");
2315	SYSCTL_INT(_kern, OID_AUTO, num_taskthreads,
2316	CTLFLAG_RD \| CTLFLAG_LOCKED,
2317	&task_threadmax, `0`, "");
2318	SYSCTL_LONG(_kern, OID_AUTO, num_recycledvnodes,
2319	CTLFLAG_RD \| CTLFLAG_LOCKED,
2320	&num_recycledvnodes, "");
2321	SYSCTL_COMPAT_INT(_kern, OID_AUTO, free_vnodes,
2322	CTLFLAG_RD \| CTLFLAG_LOCKED,
2323	&freevnodes, `0`, "");
2324
2325	STATIC int
2326	sysctl_maxvnodes(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
2327	{
2328	int oldval = desiredvnodes;
2329	int error = sysctl_io_number(req, bigValue: desiredvnodes, valueSize: sizeof(int), pValue: &desiredvnodes, NULL);
2330
2331	if (oldval != desiredvnodes) {
2332	resize_namecache(newsize: desiredvnodes);
2333	}
2334
2335	return error;
2336	}
2337
2338	SYSCTL_INT(_kern, OID_AUTO, namecache_disabled,
2339	CTLFLAG_RW \| CTLFLAG_LOCKED,
2340	&nc_disabled, `0`, "");
2341
2342	SYSCTL_PROC(_kern, KERN_MAXVNODES, maxvnodes,
2343	CTLTYPE_INT \| CTLFLAG_RW \| CTLFLAG_LOCKED,
2344	`0`, `0`, sysctl_maxvnodes, "I", "");
2345
2346	SYSCTL_PROC(_kern, KERN_MAXPROC, maxproc,
2347	CTLTYPE_INT \| CTLFLAG_RW \| CTLFLAG_LOCKED,
2348	`0`, `0`, sysctl_maxproc, "I", "");
2349
2350	SYSCTL_PROC(_kern, KERN_AIOMAX, aiomax,
2351	CTLTYPE_INT \| CTLFLAG_RW \| CTLFLAG_LOCKED,
2352	`0`, `0`, sysctl_aiomax, "I", "");
2353
2354	SYSCTL_PROC(_kern, KERN_AIOPROCMAX, aioprocmax,
2355	CTLTYPE_INT \| CTLFLAG_RW \| CTLFLAG_LOCKED,
2356	`0`, `0`, sysctl_aioprocmax, "I", "");
2357
2358	SYSCTL_PROC(_kern, KERN_AIOTHREADS, aiothreads,
2359	CTLTYPE_INT \| CTLFLAG_RW \| CTLFLAG_LOCKED,
2360	`0`, `0`, sysctl_aiothreads, "I", "");
2361
2362	SYSCTL_PROC(_kern, OID_AUTO, sched_enable_smt,
2363	CTLTYPE_INT \| CTLFLAG_RW \| CTLFLAG_KERN,
2364	`0`, `0`, sysctl_sched_enable_smt, "I", "");
2365
2366	extern int sched_allow_NO_SMT_threads;
2367	SYSCTL_INT(_kern, OID_AUTO, sched_allow_NO_SMT_threads,
2368	CTLFLAG_KERN \| CTLFLAG_RW \| CTLFLAG_LOCKED,
2369	&sched_allow_NO_SMT_threads, `0`, "");
2370
2371	extern int sched_avoid_cpu0;
2372	SYSCTL_INT(_kern, OID_AUTO, sched_rt_avoid_cpu0,
2373	CTLFLAG_KERN \| CTLFLAG_RW \| CTLFLAG_LOCKED,
2374	&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");
2375
2376	#if (DEVELOPMENT \|\| DEBUG)
2377
2378	static int
2379	sysctl_kern_max_unsafe_rt_quanta(__unused struct sysctl_oid *oidp,
2380	__unused void arg1, __unused int* arg2, struct sysctl_req *req)
2381	{
2382	extern void sched_set_max_unsafe_rt_quanta(int);
2383	extern int max_unsafe_rt_quanta;
2384
2385	int new_value, changed;
2386	int old_value = max_unsafe_rt_quanta;
2387	int error = sysctl_io_number(req, old_value, sizeof(int), &new_value,
2388	&changed);
2389	if (changed) {
2390	sched_set_max_unsafe_rt_quanta(new_value);
2391	}
2392
2393	return error;
2394	}
2395
2396	SYSCTL_PROC(_kern, OID_AUTO, max_unsafe_rt_quanta,
2397	CTLTYPE_INT \| CTLFLAG_RW \| CTLFLAG_LOCKED,
2398	`0`, `0`, sysctl_kern_max_unsafe_rt_quanta, "I",
2399	"Number of quanta to allow a realtime "
2400	"thread to run before being penalized");
2401
2402	static int
2403	sysctl_kern_max_unsafe_fixed_quanta(__unused struct sysctl_oid *oidp,
2404	__unused void arg1, __unused int* arg2, struct sysctl_req *req)
2405	{
2406	extern void sched_set_max_unsafe_fixed_quanta(int);
2407	extern int max_unsafe_fixed_quanta;
2408
2409	int new_value, changed;
2410	int old_value = max_unsafe_fixed_quanta;
2411	int error = sysctl_io_number(req, old_value, sizeof(int), &new_value,
2412	&changed);
2413	if (changed) {
2414	sched_set_max_unsafe_fixed_quanta(new_value);
2415	}
2416
2417	return error;
2418	}
2419
2420	SYSCTL_PROC(_kern, OID_AUTO, max_unsafe_fixed_quanta,
2421	CTLTYPE_INT \| CTLFLAG_RW \| CTLFLAG_LOCKED,
2422	`0`, `0`, sysctl_kern_max_unsafe_fixed_quanta, "I",
2423	"Number of quanta to allow a fixed sched mode "
2424	"thread to run before being penalized");
2425
2426	static int
2427	sysctl_kern_quantum_us(__unused struct sysctl_oid oidp, __unused void* *arg1,
2428	__unused int arg2, struct sysctl_req *req)
2429	{
2430	extern uint64_t sysctl_get_quantum_us(void);
2431	const uint64_t quantum_us = sysctl_get_quantum_us();
2432
2433	return sysctl_io_number(req, quantum_us, sizeof(quantum_us), NULL, NULL);
2434	}
2435
2436	SYSCTL_PROC(_kern, OID_AUTO, quantum_us,
2437	CTLTYPE_QUAD \| CTLFLAG_RD \| CTLFLAG_LOCKED,
2438	`0`, `0`, sysctl_kern_quantum_us, "Q",
2439	"Length of scheduling quantum in microseconds");
2440
2441	extern int smt_sched_bonus_16ths;
2442	SYSCTL_INT(_kern, OID_AUTO, smt_sched_bonus_16ths,
2443	CTLFLAG_KERN \| CTLFLAG_RW \| CTLFLAG_LOCKED,
2444	&smt_sched_bonus_16ths, `0`, "");
2445
2446	extern int smt_timeshare_enabled;
2447	SYSCTL_INT(_kern, OID_AUTO, sched_smt_timeshare_enable,
2448	CTLFLAG_KERN \| CTLFLAG_RW \| CTLFLAG_LOCKED,
2449	&smt_timeshare_enabled, `0`, "");
2450
2451	extern int sched_smt_balance;
2452	SYSCTL_INT(_kern, OID_AUTO, sched_smt_balance,
2453	CTLFLAG_KERN \| CTLFLAG_RW \| CTLFLAG_LOCKED,
2454	&sched_smt_balance, `0`, "");
2455	extern int sched_allow_rt_smt;
2456	SYSCTL_INT(_kern, OID_AUTO, sched_allow_rt_smt,
2457	CTLFLAG_KERN \| CTLFLAG_RW \| CTLFLAG_LOCKED,
2458	&sched_allow_rt_smt, `0`, "");
2459	extern int sched_allow_rt_steal;
2460	SYSCTL_INT(_kern, OID_AUTO, sched_allow_rt_steal,
2461	CTLFLAG_KERN \| CTLFLAG_RW \| CTLFLAG_LOCKED,
2462	&sched_allow_rt_steal, `0`, "");
2463	extern int sched_backup_cpu_timeout_count;
2464	SYSCTL_INT(_kern, OID_AUTO, sched_backup_cpu_timeout_count,
2465	CTLFLAG_KERN \| CTLFLAG_RW \| CTLFLAG_LOCKED,
2466	&sched_backup_cpu_timeout_count, `0`, "The maximum number of 10us delays before allowing a backup cpu to select a thread");
2467	#if __arm64__
2468	/ Scheduler perfcontrol callouts sysctls /
2469	SYSCTL_DECL(_kern_perfcontrol_callout);
2470	SYSCTL_NODE(_kern, OID_AUTO, perfcontrol_callout, CTLFLAG_RW \| CTLFLAG_LOCKED, `0`,
2471	"scheduler perfcontrol callouts");
2472
2473	extern int perfcontrol_callout_stats_enabled;
2474	SYSCTL_INT(_kern_perfcontrol_callout, OID_AUTO, stats_enabled,
2475	CTLFLAG_KERN \| CTLFLAG_RW \| CTLFLAG_LOCKED,
2476	&perfcontrol_callout_stats_enabled, `0`, "");
2477
2478	extern uint64_t perfcontrol_callout_stat_avg(perfcontrol_callout_type_t type,
2479	perfcontrol_callout_stat_t stat);
2480
2481	/ On-Core Callout /
2482	STATIC int
2483	sysctl_perfcontrol_callout_stat
2484	(__unused struct sysctl_oid oidp, void* arg1, int* arg2, struct sysctl_req *req)
2485	{
2486	perfcontrol_callout_stat_t stat = (perfcontrol_callout_stat_t)arg1;
2487	perfcontrol_callout_type_t type = (perfcontrol_callout_type_t)arg2;
2488	return sysctl_io_number(req, (int)perfcontrol_callout_stat_avg(type, stat),
2489	sizeof(int), NULL, NULL);
2490	}
2491
2492	SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, oncore_instr,
2493	CTLTYPE_INT \| CTLFLAG_RD \| CTLFLAG_LOCKED,
2494	(void *)PERFCONTROL_STAT_INSTRS, PERFCONTROL_CALLOUT_ON_CORE,
2495	sysctl_perfcontrol_callout_stat, "I", "");
2496	SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, oncore_cycles,
2497	CTLTYPE_INT \| CTLFLAG_RD \| CTLFLAG_LOCKED,
2498	(void *)PERFCONTROL_STAT_CYCLES, PERFCONTROL_CALLOUT_ON_CORE,
2499	sysctl_perfcontrol_callout_stat, "I", "");
2500	SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, offcore_instr,
2501	CTLTYPE_INT \| CTLFLAG_RD \| CTLFLAG_LOCKED,
2502	(void *)PERFCONTROL_STAT_INSTRS, PERFCONTROL_CALLOUT_OFF_CORE,
2503	sysctl_perfcontrol_callout_stat, "I", "");
2504	SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, offcore_cycles,
2505	CTLTYPE_INT \| CTLFLAG_RD \| CTLFLAG_LOCKED,
2506	(void *)PERFCONTROL_STAT_CYCLES, PERFCONTROL_CALLOUT_OFF_CORE,
2507	sysctl_perfcontrol_callout_stat, "I", "");
2508	SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, context_instr,
2509	CTLTYPE_INT \| CTLFLAG_RD \| CTLFLAG_LOCKED,
2510	(void *)PERFCONTROL_STAT_INSTRS, PERFCONTROL_CALLOUT_CONTEXT,
2511	sysctl_perfcontrol_callout_stat, "I", "");
2512	SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, context_cycles,
2513	CTLTYPE_INT \| CTLFLAG_RD \| CTLFLAG_LOCKED,
2514	(void *)PERFCONTROL_STAT_CYCLES, PERFCONTROL_CALLOUT_CONTEXT,
2515	sysctl_perfcontrol_callout_stat, "I", "");
2516	SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, update_instr,
2517	CTLTYPE_INT \| CTLFLAG_RD \| CTLFLAG_LOCKED,
2518	(void *)PERFCONTROL_STAT_INSTRS, PERFCONTROL_CALLOUT_STATE_UPDATE,
2519	sysctl_perfcontrol_callout_stat, "I", "");
2520	SYSCTL_PROC(_kern_perfcontrol_callout, OID_AUTO, update_cycles,
2521	CTLTYPE_INT \| CTLFLAG_RD \| CTLFLAG_LOCKED,
2522	(void *)PERFCONTROL_STAT_CYCLES, PERFCONTROL_CALLOUT_STATE_UPDATE,
2523	sysctl_perfcontrol_callout_stat, "I", "");
2524
2525	#if __AMP__
2526	extern int sched_amp_idle_steal;
2527	SYSCTL_INT(_kern, OID_AUTO, sched_amp_idle_steal,
2528	CTLFLAG_KERN \| CTLFLAG_RW \| CTLFLAG_LOCKED,
2529	&sched_amp_idle_steal, `0`, "");
2530	extern int sched_amp_spill_steal;
2531	SYSCTL_INT(_kern, OID_AUTO, sched_amp_spill_steal,
2532	CTLFLAG_KERN \| CTLFLAG_RW \| CTLFLAG_LOCKED,
2533	&sched_amp_spill_steal, `0`, "");
2534	extern int sched_amp_spill_count;
2535	SYSCTL_INT(_kern, OID_AUTO, sched_amp_spill_count,
2536	CTLFLAG_KERN \| CTLFLAG_RW \| CTLFLAG_LOCKED,
2537	&sched_amp_spill_count, `0`, "");
2538	extern int sched_amp_spill_deferred_ipi;
2539	SYSCTL_INT(_kern, OID_AUTO, sched_amp_spill_deferred_ipi,
2540	CTLFLAG_KERN \| CTLFLAG_RW \| CTLFLAG_LOCKED,
2541	&sched_amp_spill_deferred_ipi, `0`, "");
2542	extern int sched_amp_pcores_preempt_immediate_ipi;
2543	SYSCTL_INT(_kern, OID_AUTO, sched_amp_pcores_preempt_immediate_ipi,
2544	CTLFLAG_KERN \| CTLFLAG_RW \| CTLFLAG_LOCKED,
2545	&sched_amp_pcores_preempt_immediate_ipi, `0`, "");
2546	#endif /* __AMP__ */
2547	#endif /* __arm64__ */
2548
2549	#if __arm64__
2550	extern int legacy_footprint_entitlement_mode;
2551	SYSCTL_INT(_kern, OID_AUTO, legacy_footprint_entitlement_mode,
2552	CTLFLAG_KERN \| CTLFLAG_RD \| CTLFLAG_LOCKED,
2553	&legacy_footprint_entitlement_mode, `0`, "");
2554	#endif /* __arm64__ */
2555
2556	/*
2557	* Realtime threads are ordered by highest priority first then,
2558	* for threads of the same priority, by earliest deadline first.
2559	* But if sched_rt_runq_strict_priority is false (the default),
2560	* a lower priority thread with an earlier deadline will be preferred
2561	* over a higher priority thread with a later deadline, as long as
2562	* both threads' computations will fit before the later deadline.
2563	*/
2564	extern int sched_rt_runq_strict_priority;
2565	SYSCTL_INT(_kern, OID_AUTO, sched_rt_runq_strict_priority,
2566	CTLFLAG_KERN \| CTLFLAG_RW \| CTLFLAG_LOCKED,
2567	&sched_rt_runq_strict_priority, `0`, "");
2568
2569	static int
2570	sysctl_kern_sched_rt_n_backup_processors(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
2571	{
2572	int new_value, changed;
2573	int old_value = sched_get_rt_n_backup_processors();
2574	int error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
2575	if (changed) {
2576	sched_set_rt_n_backup_processors(new_value);
2577	}
2578
2579	return error;
2580	}
2581
2582	SYSCTL_PROC(_kern, OID_AUTO, sched_rt_n_backup_processors,
2583	CTLTYPE_INT \| CTLFLAG_RW \| CTLFLAG_LOCKED,
2584	`0`, `0`, sysctl_kern_sched_rt_n_backup_processors, "I", "");
2585
2586	static int
2587	sysctl_kern_sched_rt_deadline_epsilon_us(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
2588	{
2589	int new_value, changed;
2590	int old_value = sched_get_rt_deadline_epsilon();
2591	int error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
2592	if (changed) {
2593	sched_set_rt_deadline_epsilon(new_value);
2594	}
2595
2596	return error;
2597	}
2598
2599	SYSCTL_PROC(_kern, OID_AUTO, sched_rt_deadline_epsilon_us,
2600	CTLTYPE_INT \| CTLFLAG_RW \| CTLFLAG_LOCKED,
2601	`0`, `0`, sysctl_kern_sched_rt_deadline_epsilon_us, "I", "");
2602
2603	extern int sched_idle_delay_cpuid;
2604	SYSCTL_INT(_kern, OID_AUTO, sched_idle_delay_cpuid,
2605	CTLFLAG_KERN \| CTLFLAG_RW \| CTLFLAG_LOCKED,
2606	&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");
2607
2608	static int
2609	sysctl_kern_sched_powered_cores(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
2610	{
2611	int new_value, changed;
2612	int old_value = sched_get_powered_cores();
2613	int error = sysctl_io_number(req, old_value, sizeof(int), &new_value, &changed);
2614	if (changed) {
2615	sched_set_powered_cores(new_value);
2616	}
2617
2618	return error;
2619	}
2620
2621	SYSCTL_PROC(_kern, OID_AUTO, sched_powered_cores,
2622	CTLTYPE_INT \| CTLFLAG_RW \| CTLFLAG_LOCKED,
2623	`0`, `0`, sysctl_kern_sched_powered_cores, "I", "");
2624
2625	#endif /* (DEVELOPMENT \|\| DEBUG) */
2626
2627	extern uint64_t perfcontrol_requested_recommended_cores;
2628	SYSCTL_QUAD(_kern, OID_AUTO, sched_recommended_cores,
2629	CTLFLAG_KERN \| CTLFLAG_RD \| CTLFLAG_LOCKED,
2630	&perfcontrol_requested_recommended_cores, "");
2631
2632	static int
2633	sysctl_kern_suspend_cluster_powerdown(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
2634	{
2635	int new_value, changed;
2636	int old_value = get_cluster_powerdown_user_suspended();
2637	int error = sysctl_io_number(req, bigValue: old_value, valueSize: sizeof(int), pValue: &new_value, changed: &changed);
2638	if (!error && changed) {
2639	if (new_value > `0`) {
2640	error = suspend_cluster_powerdown_from_user();
2641	} else {
2642	error = resume_cluster_powerdown_from_user();
2643	}
2644	if (error) {
2645	error = EALREADY;
2646	}
2647	}
2648
2649	return error;
2650	}
2651
2652	SYSCTL_PROC(_kern, OID_AUTO, suspend_cluster_powerdown,
2653	CTLTYPE_INT \| CTLFLAG_RW \| CTLFLAG_LOCKED,
2654	`0`, `0`, sysctl_kern_suspend_cluster_powerdown, "I", "");
2655
2656
2657	STATIC int
2658	sysctl_securelvl
2659	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
2660	{
2661	int new_value, changed;
2662	int error = sysctl_io_number(req, bigValue: securelevel, valueSize: sizeof(int), pValue: &new_value, changed: &changed);
2663	if (changed) {
2664	if (!(new_value < securelevel && proc_getpid(req->p) != `1`)) {
2665	proc_list_lock();
2666	securelevel = new_value;
2667	proc_list_unlock();
2668	} else {
2669	error = EPERM;
2670	}
2671	}
2672	return error;
2673	}
2674
2675	SYSCTL_PROC(_kern, KERN_SECURELVL, securelevel,
2676	CTLTYPE_INT \| CTLFLAG_RW \| CTLFLAG_LOCKED,
2677	`0`, `0`, sysctl_securelvl, "I", "");
2678
2679
2680	STATIC int
2681	sysctl_domainname
2682	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
2683	{
2684	int error, changed;
2685	char tmpname[MAXHOSTNAMELEN] = {};
2686
2687	lck_mtx_lock(lck: &domainname_lock);
2688	strlcpy(dst: tmpname, src: domainname, n: sizeof(tmpname));
2689	lck_mtx_unlock(lck: &domainname_lock);
2690
2691	error = sysctl_io_string(req, pValue: tmpname, valueSize: sizeof(tmpname), trunc: `0`, changed: &changed);
2692	if (!error && changed) {
2693	lck_mtx_lock(lck: &domainname_lock);
2694	strlcpy(dst: domainname, src: tmpname, n: sizeof(domainname));
2695	lck_mtx_unlock(lck: &domainname_lock);
2696	}
2697	return error;
2698	}
2699
2700	SYSCTL_PROC(_kern, KERN_DOMAINNAME, nisdomainname,
2701	CTLTYPE_STRING \| CTLFLAG_RW \| CTLFLAG_LOCKED,
2702	`0`, `0`, sysctl_domainname, "A", "");
2703
2704	SYSCTL_COMPAT_INT(_kern, KERN_HOSTID, hostid,
2705	CTLFLAG_RW \| CTLFLAG_KERN \| CTLFLAG_LOCKED,
2706	&hostid, `0`, "");
2707
2708	STATIC int
2709	sysctl_hostname
2710	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
2711	{
2712	int error, changed;
2713	char tmpname[MAXHOSTNAMELEN] = {};
2714	const char * name;
2715
2716	#if XNU_TARGET_OS_OSX
2717	name = hostname;
2718	#else /* XNU_TARGET_OS_OSX */
2719	#define ENTITLEMENT_USER_ASSIGNED_DEVICE_NAME \
2720	"com.apple.developer.device-information.user-assigned-device-name"
2721	if (csproc_get_platform_binary(current_proc()) \|\|
2722	IOCurrentTaskHasEntitlement(ENTITLEMENT_USER_ASSIGNED_DEVICE_NAME)) {
2723	name = hostname;
2724	} else {
2725	/ Deny writes if we don't pass entitlement check /
2726	if (req->newptr) {
2727	return EPERM;
2728	}
2729
2730	name = "localhost";
2731	}
2732	#endif /* ! XNU_TARGET_OS_OSX */
2733
2734	lck_mtx_lock(lck: &hostname_lock);
2735	strlcpy(dst: tmpname, src: name, n: sizeof(tmpname));
2736	lck_mtx_unlock(lck: &hostname_lock);
2737
2738	error = sysctl_io_string(req, pValue: tmpname, valueSize: sizeof(tmpname), trunc: `1`, changed: &changed);
2739	if (!error && changed) {
2740	lck_mtx_lock(lck: &hostname_lock);
2741	strlcpy(dst: hostname, src: tmpname, n: sizeof(hostname));
2742	lck_mtx_unlock(lck: &hostname_lock);
2743	}
2744	return error;
2745	}
2746
2747	SYSCTL_PROC(_kern, KERN_HOSTNAME, hostname,
2748	CTLTYPE_STRING \| CTLFLAG_RW \| CTLFLAG_LOCKED,
2749	`0`, `0`, sysctl_hostname, "A", "");
2750
2751	STATIC int
2752	sysctl_procname
2753	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
2754	{
2755	/ Original code allowed writing, I'm copying this, although this all makes*
2756	* no sense to me. Besides, this sysctl is never used. */
2757	return sysctl_io_string(req, pValue: &req->p->p_name[`0`], valueSize: (`2` * MAXCOMLEN + `1`), trunc: `1`, NULL);
2758	}
2759
2760	SYSCTL_PROC(_kern, KERN_PROCNAME, procname,
2761	CTLTYPE_STRING \| CTLFLAG_RW \| CTLFLAG_ANYBODY \| CTLFLAG_LOCKED,
2762	`0`, `0`, sysctl_procname, "A", "");
2763
2764	SYSCTL_INT(_kern, KERN_SPECULATIVE_READS, speculative_reads_disabled,
2765	CTLFLAG_RW \| CTLFLAG_KERN \| CTLFLAG_LOCKED,
2766	&speculative_reads_disabled, `0`, "");
2767
2768	SYSCTL_UINT(_kern, OID_AUTO, preheat_max_bytes,
2769	CTLFLAG_RW \| CTLFLAG_KERN \| CTLFLAG_LOCKED,
2770	&preheat_max_bytes, `0`, "");
2771
2772	SYSCTL_UINT(_kern, OID_AUTO, preheat_min_bytes,
2773	CTLFLAG_RW \| CTLFLAG_KERN \| CTLFLAG_LOCKED,
2774	&preheat_min_bytes, `0`, "");
2775
2776	SYSCTL_UINT(_kern, OID_AUTO, speculative_prefetch_max,
2777	CTLFLAG_RW \| CTLFLAG_KERN \| CTLFLAG_LOCKED,
2778	&speculative_prefetch_max, `0`, "");
2779
2780	SYSCTL_UINT(_kern, OID_AUTO, speculative_prefetch_max_iosize,
2781	CTLFLAG_RW \| CTLFLAG_KERN \| CTLFLAG_LOCKED,
2782	&speculative_prefetch_max_iosize, `0`, "");
2783
2784	SYSCTL_UINT(_kern, OID_AUTO, vm_page_free_target,
2785	CTLFLAG_RW \| CTLFLAG_KERN \| CTLFLAG_LOCKED,
2786	&vm_page_free_target, `0`, "");
2787
2788	SYSCTL_UINT(_kern, OID_AUTO, vm_page_free_min,
2789	CTLFLAG_RW \| CTLFLAG_KERN \| CTLFLAG_LOCKED,
2790	&vm_page_free_min, `0`, "");
2791
2792	SYSCTL_UINT(_kern, OID_AUTO, vm_page_free_reserved,
2793	CTLFLAG_RW \| CTLFLAG_KERN \| CTLFLAG_LOCKED,
2794	&vm_page_free_reserved, `0`, "");
2795
2796	SYSCTL_UINT(_kern, OID_AUTO, vm_page_speculative_percentage,
2797	CTLFLAG_RW \| CTLFLAG_KERN \| CTLFLAG_LOCKED,
2798	&vm_pageout_state.vm_page_speculative_percentage, `0`, "");
2799
2800	SYSCTL_UINT(_kern, OID_AUTO, vm_page_speculative_q_age_ms,
2801	CTLFLAG_RW \| CTLFLAG_KERN \| CTLFLAG_LOCKED,
2802	&vm_pageout_state.vm_page_speculative_q_age_ms, `0`, "");
2803
2804	SYSCTL_UINT(_kern, OID_AUTO, vm_max_delayed_work_limit,
2805	CTLFLAG_RW \| CTLFLAG_KERN \| CTLFLAG_LOCKED,
2806	&vm_max_delayed_work_limit, `0`, "");
2807
2808	SYSCTL_UINT(_kern, OID_AUTO, vm_max_batch,
2809	CTLFLAG_RW \| CTLFLAG_KERN \| CTLFLAG_LOCKED,
2810	&vm_max_batch, `0`, "");
2811
2812	SYSCTL_STRING(_kern, OID_AUTO, bootsessionuuid,
2813	CTLFLAG_RD \| CTLFLAG_LOCKED,
2814	&bootsessionuuid_string, sizeof(bootsessionuuid_string), "");
2815
2816
2817	STATIC int
2818	sysctl_boottime
2819	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
2820	{
2821	struct timeval tv;
2822	boottime_timeval(tv: &tv);
2823	struct proc *p = req->p;
2824
2825	if (proc_is64bit(p)) {
2826	struct user64_timeval t = {};
2827	t.tv_sec = tv.tv_sec;
2828	t.tv_usec = tv.tv_usec;
2829	return sysctl_io_opaque(req, pValue: &t, valueSize: sizeof(t), NULL);
2830	} else {
2831	struct user32_timeval t = {};
2832	t.tv_sec = (user32_time_t)tv.tv_sec;
2833	t.tv_usec = tv.tv_usec;
2834	return sysctl_io_opaque(req, pValue: &t, valueSize: sizeof(t), NULL);
2835	}
2836	}
2837
2838	SYSCTL_PROC(_kern, KERN_BOOTTIME, boottime,
2839	CTLTYPE_STRUCT \| CTLFLAG_KERN \| CTLFLAG_RD \| CTLFLAG_LOCKED,
2840	`0`, `0`, sysctl_boottime, "S,timeval", "");
2841
2842	extern bool IOGetBootUUID(char *);
2843
2844	/ non-static: written by imageboot.c /
2845	uuid_string_t fake_bootuuid;
2846
2847	STATIC int
2848	sysctl_bootuuid
2849	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
2850	{
2851	int error = ENOENT;
2852
2853	/ check the first byte to see if the string has been*
2854	* populated. this is a uuid_STRING_t, this check would
2855	* not work with a uuid_t.
2856	*/
2857	if (fake_bootuuid[`0`] != `'\0'`) {
2858	error = sysctl_io_string(req, pValue: fake_bootuuid, valueSize: `0`, trunc: `0`, NULL);
2859	goto out;
2860	}
2861
2862	uuid_string_t uuid_string;
2863	if (IOGetBootUUID(uuid_string)) {
2864	uuid_t boot_uuid;
2865	error = uuid_parse(in: uuid_string, uu: boot_uuid);
2866	if (!error) {
2867	error = sysctl_io_string(req, __DECONST(char *, uuid_string), valueSize: `0`, trunc: `0`, NULL);
2868	}
2869	}
2870
2871	out:
2872	return error;
2873	}
2874
2875	SYSCTL_PROC(_kern, OID_AUTO, bootuuid,
2876	CTLTYPE_STRING \| CTLFLAG_KERN \| CTLFLAG_RD \| CTLFLAG_LOCKED,
2877	`0`, `0`, sysctl_bootuuid, "A", "");
2878
2879
2880	extern bool IOGetApfsPrebootUUID(char *);
2881	extern bool IOGetAssociatedApfsVolgroupUUID(char *);
2882
2883	STATIC int
2884	sysctl_apfsprebootuuid
2885	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
2886	{
2887	int error = ENOENT;
2888
2889	uuid_string_t uuid_string;
2890	if (IOGetApfsPrebootUUID(uuid_string)) {
2891	uuid_t apfs_preboot_uuid;
2892	error = uuid_parse(in: uuid_string, uu: apfs_preboot_uuid);
2893	if (!error) {
2894	error = sysctl_io_string(req, __DECONST(char *, uuid_string), valueSize: `0`, trunc: `0`, NULL);
2895	}
2896	}
2897
2898	return error;
2899	}
2900
2901	SYSCTL_PROC(_kern, OID_AUTO, apfsprebootuuid,
2902	CTLTYPE_STRING \| CTLFLAG_KERN \| CTLFLAG_RD \| CTLFLAG_LOCKED,
2903	`0`, `0`, sysctl_apfsprebootuuid, "A", "");
2904
2905	STATIC int
2906	sysctl_targetsystemvolgroupuuid
2907	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
2908	{
2909	int error = ENOENT;
2910
2911	uuid_string_t uuid_string;
2912	if (IOGetApfsPrebootUUID(uuid_string)) {
2913	uuid_t apfs_preboot_uuid;
2914	error = uuid_parse(in: uuid_string, uu: apfs_preboot_uuid);
2915	if (!error) {
2916	error = sysctl_io_string(req, __DECONST(char *, uuid_string), valueSize: `0`, trunc: `0`, NULL);
2917	}
2918	} else {
2919	/*
2920	* In special boot modes, such as kcgen-mode, the
2921	* apfs-preboot-uuid property will not be set. Instead, a
2922	* different property, associated-volume-group, will be set
2923	* which indicates the UUID of the VolumeGroup containing the
2924	* system volume into which you will boot.
2925	*/
2926	if (IOGetAssociatedApfsVolgroupUUID(uuid_string)) {
2927	uuid_t apfs_preboot_uuid;
2928	error = uuid_parse(in: uuid_string, uu: apfs_preboot_uuid);
2929	if (!error) {
2930	error = sysctl_io_string(req, __DECONST(char *, uuid_string), valueSize: `0`, trunc: `0`, NULL);
2931	}
2932	}
2933	}
2934
2935	return error;
2936	}
2937
2938	SYSCTL_PROC(_kern, OID_AUTO, targetsystemvolgroupuuid,
2939	CTLTYPE_STRING \| CTLFLAG_KERN \| CTLFLAG_RD \| CTLFLAG_MASKED \| CTLFLAG_LOCKED,
2940	`0`, `0`, sysctl_targetsystemvolgroupuuid, "A", "");
2941
2942
2943	extern bool IOGetBootManifestHash(char , size_t );
2944	extern bool IOGetBootObjectsPath(char *);
2945
2946	STATIC int
2947	sysctl_bootobjectspath
2948	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
2949	{
2950	int error = ENOENT;
2951
2952	#if defined(__x86_64__)
2953	/ auth-root-dmg is used for the Intel BaseSystem in some flows,*
2954	* e.g. createinstallmedia and as part of upgrading from 10.15 or earlier
2955	* under these scenarios, set_fake_bootuuid will be called when pivoting to
2956	* the new root filesystem. need honor the fake bootuuid.
2957	*/
2958	if (fake_bootuuid[`0`] != `'\0'`) {
2959	error = sysctl_io_string(req, fake_bootuuid, `0`, `0`, NULL);
2960	} else {
2961	/ for intel mac, boot objects reside in [preboot volume]/[bootuuid]*
2962	* bootuuid and apfsprebootuuid are populated by efiboot and they are alias.
2963	*/
2964	uuid_string_t uuid_string;
2965	if (IOGetBootUUID(uuid_string)) {
2966	uuid_t boot_uuid;
2967	error = uuid_parse(uuid_string, boot_uuid);
2968	if (!error) {
2969	error = sysctl_io_string(req, (char *)uuid_string, `0`, `0`, NULL);
2970	}
2971	}
2972	}
2973	#else
2974	char boot_obj_path[MAXPATHLEN] = { "\0" };
2975	static const char kAsciiHexChars[] = "0123456789ABCDEF";
2976	unsigned int i, j;
2977
2978	/ Hashed with SHA2-384 or SHA1, boot manifest hash is 48 bytes or 20 bytes*
2979	* hence, need a 97 bytes char array for the string.
2980	*/
2981	size_t hash_data_size = CCSHA384_OUTPUT_SIZE;
2982	char hash_data[CCSHA384_OUTPUT_SIZE] = { "\0" };
2983	char boot_manifest_hash[CCSHA384_OUTPUT_SIZE * `2` + `1`] = { "\0" };;
2984
2985	/ for Apple Silicon Macs, there is a boot-objects-path under IODeviceTree:/chosen*
2986	* and boot objects reside in [preboot volume]/[boot-objects-path]
2987	* for embedded platforms, there would be a boot-manifest-hash under IODeviceTree:/chosen
2988	* and boot objects reside in [preboot volume]/[boot-manifest-hash]
2989	*/
2990	if (IOGetBootObjectsPath(boot_obj_path)) {
2991	error = sysctl_io_string(req, pValue: (char *)boot_obj_path, valueSize: `0`, trunc: `0`, NULL);
2992	} else if (IOGetBootManifestHash(hash_data, &hash_data_size)) {
2993	j = `0`;
2994	for (i = `0`; i < hash_data_size; ++i) {
2995	char octet = hash_data[i];
2996	boot_manifest_hash[j++] = kAsciiHexChars[((octet & `0xF0`) >> `4`)];
2997	boot_manifest_hash[j++] = kAsciiHexChars[(octet & `0x0F`)];
2998	}
2999	/ make sure string has null termination /
3000	boot_manifest_hash[j] = `'\0'`;
3001	error = sysctl_io_string(req, pValue: (char *)boot_manifest_hash, valueSize: `0`, trunc: `0`, NULL);
3002	}
3003	#endif
3004	return error;
3005	}
3006
3007	SYSCTL_PROC(_kern, OID_AUTO, bootobjectspath,
3008	CTLTYPE_STRING \| CTLFLAG_KERN \| CTLFLAG_RD \| CTLFLAG_LOCKED,
3009	`0`, `0`, sysctl_bootobjectspath, "A", "");
3010
3011
3012	STATIC int
3013	sysctl_symfile
3014	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
3015	{
3016	char *str;
3017	int error = get_kernel_symfile(req->p, &str);
3018	if (error) {
3019	return error;
3020	}
3021	return sysctl_io_string(req, pValue: str, valueSize: `0`, trunc: `0`, NULL);
3022	}
3023
3024
3025	SYSCTL_PROC(_kern, KERN_SYMFILE, symfile,
3026	CTLTYPE_STRING \| CTLFLAG_RD \| CTLFLAG_LOCKED,
3027	`0`, `0`, sysctl_symfile, "A", "");
3028
3029	#if CONFIG_NETBOOT
3030	STATIC int
3031	sysctl_netboot
3032	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
3033	{
3034	return sysctl_io_number(req, bigValue: netboot_root(), valueSize: sizeof(int), NULL, NULL);
3035	}
3036
3037	SYSCTL_PROC(_kern, KERN_NETBOOT, netboot,
3038	CTLTYPE_INT \| CTLFLAG_RD \| CTLFLAG_LOCKED,
3039	`0`, `0`, sysctl_netboot, "I", "");
3040	#endif
3041
3042	#ifdef CONFIG_IMGSRC_ACCESS
3043	/*
3044	* Legacy--act as if only one layer of nesting is possible.
3045	*/
3046	STATIC int
3047	sysctl_imgsrcdev
3048	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
3049	{
3050	vfs_context_t ctx = vfs_context_current();
3051	vnode_t devvp;
3052	int result;
3053
3054	if (!vfs_context_issuser(ctx)) {
3055	return EPERM;
3056	}
3057
3058	if (imgsrc_rootvnodes[`0`] == NULL) {
3059	return ENOENT;
3060	}
3061
3062	result = vnode_getwithref(vp: imgsrc_rootvnodes[`0`]);
3063	if (result != `0`) {
3064	return result;
3065	}
3066
3067	devvp = vnode_mount(vp: imgsrc_rootvnodes[`0`])->mnt_devvp;
3068	result = vnode_getwithref(vp: devvp);
3069	if (result != `0`) {
3070	goto out;
3071	}
3072
3073	result = sysctl_io_number(req, bigValue: vnode_specrdev(vp: devvp), valueSize: sizeof(dev_t), NULL, NULL);
3074
3075	vnode_put(vp: devvp);
3076	out:
3077	vnode_put(vp: imgsrc_rootvnodes[`0`]);
3078	return result;
3079	}
3080
3081	SYSCTL_PROC(_kern, OID_AUTO, imgsrcdev,
3082	CTLTYPE_INT \| CTLFLAG_RD \| CTLFLAG_LOCKED,
3083	`0`, `0`, sysctl_imgsrcdev, "I", "");
3084
3085	STATIC int
3086	sysctl_imgsrcinfo
3087	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
3088	{
3089	int error;
3090	struct imgsrc_info info[MAX_IMAGEBOOT_NESTING] = {}; / 2 for now, no problem /
3091	uint32_t i;
3092	vnode_t rvp, devvp;
3093
3094	if (imgsrc_rootvnodes[`0`] == NULLVP) {
3095	return ENXIO;
3096	}
3097
3098	for (i = `0`; i < MAX_IMAGEBOOT_NESTING; i++) {
3099	/*
3100	* Go get the root vnode.
3101	*/
3102	rvp = imgsrc_rootvnodes[i];
3103	if (rvp == NULLVP) {
3104	break;
3105	}
3106
3107	error = vnode_get(rvp);
3108	if (error != `0`) {
3109	return error;
3110	}
3111
3112	/*
3113	* For now, no getting at a non-local volume.
3114	*/
3115	devvp = vnode_mount(vp: rvp)->mnt_devvp;
3116	if (devvp == NULL) {
3117	vnode_put(vp: rvp);
3118	return EINVAL;
3119	}
3120
3121	error = vnode_getwithref(vp: devvp);
3122	if (error != `0`) {
3123	vnode_put(vp: rvp);
3124	return error;
3125	}
3126
3127	/*
3128	* Fill in info.
3129	*/
3130	info[i].ii_dev = vnode_specrdev(vp: devvp);
3131	info[i].ii_flags = `0`;
3132	info[i].ii_height = i;
3133	bzero(s: info[i].ii_reserved, n: sizeof(info[i].ii_reserved));
3134
3135	vnode_put(vp: devvp);
3136	vnode_put(vp: rvp);
3137	}
3138
3139	return sysctl_io_opaque(req, pValue: info, valueSize: i * sizeof(info[`0`]), NULL);
3140	}
3141
3142	SYSCTL_PROC(_kern, OID_AUTO, imgsrcinfo,
3143	CTLTYPE_OPAQUE \| CTLFLAG_RD \| CTLFLAG_LOCKED,
3144	`0`, `0`, sysctl_imgsrcinfo, "I", "");
3145
3146	#endif /* CONFIG_IMGSRC_ACCESS */
3147
3148
3149	SYSCTL_DECL(_kern_timer);
3150	SYSCTL_NODE(_kern, OID_AUTO, timer, CTLFLAG_RW \| CTLFLAG_LOCKED, `0`, "timer");
3151
3152
3153	SYSCTL_INT(_kern_timer, OID_AUTO, coalescing_enabled,
3154	CTLFLAG_KERN \| CTLFLAG_RW \| CTLFLAG_LOCKED,
3155	&mach_timer_coalescing_enabled, `0`, "");
3156
3157	SYSCTL_QUAD(_kern_timer, OID_AUTO, deadline_tracking_bin_1,
3158	CTLFLAG_RW \| CTLFLAG_LOCKED,
3159	&timer_deadline_tracking_bin_1, "");
3160	SYSCTL_QUAD(_kern_timer, OID_AUTO, deadline_tracking_bin_2,
3161	CTLFLAG_RW \| CTLFLAG_LOCKED,
3162	&timer_deadline_tracking_bin_2, "");
3163
3164	SYSCTL_DECL(_kern_timer_longterm);
3165	SYSCTL_NODE(_kern_timer, OID_AUTO, longterm, CTLFLAG_RW \| CTLFLAG_LOCKED, `0`, "longterm");
3166
3167
3168	/ Must match definition in osfmk/kern/timer_call.c /
3169	enum {
3170	THRESHOLD, QCOUNT,
3171	ENQUEUES, DEQUEUES, ESCALATES, SCANS, PREEMPTS,
3172	LATENCY, LATENCY_MIN, LATENCY_MAX, LONG_TERM_SCAN_LIMIT,
3173	LONG_TERM_SCAN_INTERVAL, LONG_TERM_SCAN_PAUSES,
3174	SCAN_LIMIT, SCAN_INTERVAL, SCAN_PAUSES, SCAN_POSTPONES,
3175	};
3176	extern uint64_t timer_sysctl_get(int);
3177	extern int timer_sysctl_set(int, uint64_t);
3178
3179	STATIC int
3180	sysctl_timer
3181	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
3182	{
3183	int oid = (int)arg1;
3184	uint64_t value = timer_sysctl_get(oid);
3185	uint64_t new_value;
3186	int error;
3187	int changed;
3188
3189	error = sysctl_io_number(req, bigValue: value, valueSize: sizeof(value), pValue: &new_value, changed: &changed);
3190	if (changed) {
3191	error = timer_sysctl_set(oid, new_value);
3192	}
3193
3194	return error;
3195	}
3196
3197	SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, threshold,
3198	CTLTYPE_QUAD \| CTLFLAG_RW \| CTLFLAG_LOCKED,
3199	(void *) THRESHOLD, `0`, sysctl_timer, "Q", "");
3200	SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, scan_limit,
3201	CTLTYPE_QUAD \| CTLFLAG_RW \| CTLFLAG_LOCKED,
3202	(void *) LONG_TERM_SCAN_LIMIT, `0`, sysctl_timer, "Q", "");
3203	SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, scan_interval,
3204	CTLTYPE_QUAD \| CTLFLAG_RW \| CTLFLAG_LOCKED,
3205	(void *) LONG_TERM_SCAN_INTERVAL, `0`, sysctl_timer, "Q", "");
3206
3207	SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, qlen,
3208	CTLTYPE_QUAD \| CTLFLAG_RD \| CTLFLAG_LOCKED,
3209	(void *) QCOUNT, `0`, sysctl_timer, "Q", "");
3210	SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, scan_pauses,
3211	CTLTYPE_QUAD \| CTLFLAG_RD \| CTLFLAG_LOCKED,
3212	(void *) LONG_TERM_SCAN_PAUSES, `0`, sysctl_timer, "Q", "");
3213
3214	#if DEBUG
3215	SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, enqueues,
3216	CTLTYPE_QUAD \| CTLFLAG_RD \| CTLFLAG_LOCKED,
3217	(void *) ENQUEUES, `0`, sysctl_timer, "Q", "");
3218	SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, dequeues,
3219	CTLTYPE_QUAD \| CTLFLAG_RD \| CTLFLAG_LOCKED,
3220	(void *) DEQUEUES, `0`, sysctl_timer, "Q", "");
3221	SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, escalates,
3222	CTLTYPE_QUAD \| CTLFLAG_RD \| CTLFLAG_LOCKED,
3223	(void *) ESCALATES, `0`, sysctl_timer, "Q", "");
3224	SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, scans,
3225	CTLTYPE_QUAD \| CTLFLAG_RD \| CTLFLAG_LOCKED,
3226	(void *) SCANS, `0`, sysctl_timer, "Q", "");
3227	SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, preempts,
3228	CTLTYPE_QUAD \| CTLFLAG_RD \| CTLFLAG_LOCKED,
3229	(void *) PREEMPTS, `0`, sysctl_timer, "Q", "");
3230	SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, latency,
3231	CTLTYPE_QUAD \| CTLFLAG_RD \| CTLFLAG_LOCKED,
3232	(void *) LATENCY, `0`, sysctl_timer, "Q", "");
3233	SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, latency_min,
3234	CTLTYPE_QUAD \| CTLFLAG_RD \| CTLFLAG_LOCKED,
3235	(void *) LATENCY_MIN, `0`, sysctl_timer, "Q", "");
3236	SYSCTL_PROC(_kern_timer_longterm, OID_AUTO, latency_max,
3237	CTLTYPE_QUAD \| CTLFLAG_RD \| CTLFLAG_LOCKED,
3238	(void *) LATENCY_MAX, `0`, sysctl_timer, "Q", "");
3239	#endif /* DEBUG */
3240
3241	SYSCTL_PROC(_kern_timer, OID_AUTO, scan_limit,
3242	CTLTYPE_QUAD \| CTLFLAG_RW \| CTLFLAG_LOCKED,
3243	(void *) SCAN_LIMIT, `0`, sysctl_timer, "Q", "");
3244	SYSCTL_PROC(_kern_timer, OID_AUTO, scan_interval,
3245	CTLTYPE_QUAD \| CTLFLAG_RW \| CTLFLAG_LOCKED,
3246	(void *) SCAN_INTERVAL, `0`, sysctl_timer, "Q", "");
3247	SYSCTL_PROC(_kern_timer, OID_AUTO, scan_pauses,
3248	CTLTYPE_QUAD \| CTLFLAG_RD \| CTLFLAG_LOCKED,
3249	(void *) SCAN_PAUSES, `0`, sysctl_timer, "Q", "");
3250	SYSCTL_PROC(_kern_timer, OID_AUTO, scan_postpones,
3251	CTLTYPE_QUAD \| CTLFLAG_RD \| CTLFLAG_LOCKED,
3252	(void *) SCAN_POSTPONES, `0`, sysctl_timer, "Q", "");
3253
3254	STATIC int
3255	sysctl_usrstack
3256	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
3257	{
3258	return sysctl_io_number(req, bigValue: (int)req->p->user_stack, valueSize: sizeof(int), NULL, NULL);
3259	}
3260
3261	SYSCTL_PROC(_kern, KERN_USRSTACK32, usrstack,
3262	CTLTYPE_INT \| CTLFLAG_RD \| CTLFLAG_LOCKED,
3263	`0`, `0`, sysctl_usrstack, "I", "");
3264
3265	STATIC int
3266	sysctl_usrstack64
3267	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
3268	{
3269	return sysctl_io_number(req, bigValue: req->p->user_stack, valueSize: sizeof(req->p->user_stack), NULL, NULL);
3270	}
3271
3272	SYSCTL_PROC(_kern, KERN_USRSTACK64, usrstack64,
3273	CTLTYPE_QUAD \| CTLFLAG_RD \| CTLFLAG_LOCKED,
3274	`0`, `0`, sysctl_usrstack64, "Q", "");
3275
3276
3277	#if EXCLAVES_COREDUMP
3278
3279	/ secure kernel coredump support. /
3280	extern unsigned int sc_dump_mode;
3281	SYSCTL_UINT(_kern, OID_AUTO, secure_coredump, CTLFLAG_RD, &sc_dump_mode, `0`, "secure_coredump");
3282
3283	#endif /* EXCLAVES_COREDUMP */
3284
3285
3286	#if CONFIG_COREDUMP
3287
3288	SYSCTL_STRING(_kern, KERN_COREFILE, corefile,
3289	CTLFLAG_RW \| CTLFLAG_KERN \| CTLFLAG_LOCKED,
3290	corefilename, sizeof(corefilename), "");
3291
3292	SYSCTL_STRING(_kern, OID_AUTO, drivercorefile,
3293	CTLFLAG_RW \| CTLFLAG_KERN \| CTLFLAG_LOCKED,
3294	drivercorefilename, sizeof(drivercorefilename), "");
3295
3296	STATIC int
3297	sysctl_coredump
3298	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
3299	{
3300	#ifdef SECURE_KERNEL
3301	(void)req;
3302	return ENOTSUP;
3303	#else
3304	int new_value, changed;
3305	int error = sysctl_io_number(req, bigValue: do_coredump, valueSize: sizeof(int), pValue: &new_value, changed: &changed);
3306	if (changed) {
3307	if ((new_value == `0`) \|\| (new_value == `1`)) {
3308	do_coredump = new_value;
3309	} else {
3310	error = EINVAL;
3311	}
3312	}
3313	return error;
3314	#endif
3315	}
3316
3317	SYSCTL_PROC(_kern, KERN_COREDUMP, coredump,
3318	CTLTYPE_INT \| CTLFLAG_RW \| CTLFLAG_LOCKED,
3319	`0`, `0`, sysctl_coredump, "I", "");
3320
3321	STATIC int
3322	sysctl_suid_coredump
3323	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
3324	{
3325	#ifdef SECURE_KERNEL
3326	(void)req;
3327	return ENOTSUP;
3328	#else
3329	int new_value, changed;
3330	int error = sysctl_io_number(req, bigValue: sugid_coredump, valueSize: sizeof(int), pValue: &new_value, changed: &changed);
3331	if (changed) {
3332	if ((new_value == `0`) \|\| (new_value == `1`)) {
3333	sugid_coredump = new_value;
3334	} else {
3335	error = EINVAL;
3336	}
3337	}
3338	return error;
3339	#endif
3340	}
3341
3342	SYSCTL_PROC(_kern, KERN_SUGID_COREDUMP, sugid_coredump,
3343	CTLTYPE_INT \| CTLFLAG_RW \| CTLFLAG_LOCKED,
3344	`0`, `0`, sysctl_suid_coredump, "I", "");
3345
3346	#endif /* CONFIG_COREDUMP */
3347
3348	STATIC int
3349	sysctl_delayterm
3350	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
3351	{
3352	struct proc *p = req->p;
3353	int new_value, changed;
3354	int error = sysctl_io_number(req, bigValue: (req->p->p_lflag & P_LDELAYTERM)? `1`: `0`, valueSize: sizeof(int), pValue: &new_value, changed: &changed);
3355	if (changed) {
3356	proc_lock(p);
3357	if (new_value) {
3358	req->p->p_lflag \|= P_LDELAYTERM;
3359	} else {
3360	req->p->p_lflag &= ~P_LDELAYTERM;
3361	}
3362	proc_unlock(p);
3363	}
3364	return error;
3365	}
3366
3367	SYSCTL_PROC(_kern, KERN_PROCDELAYTERM, delayterm,
3368	CTLTYPE_INT \| CTLFLAG_RW \| CTLFLAG_LOCKED,
3369	`0`, `0`, sysctl_delayterm, "I", "");
3370
3371
3372	STATIC int
3373	sysctl_rage_vnode
3374	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
3375	{
3376	struct proc *p = req->p;
3377	struct uthread *ut;
3378	int new_value, old_value, changed;
3379	int error;
3380
3381	ut = current_uthread();
3382
3383	if (ut->uu_flag & UT_RAGE_VNODES) {
3384	old_value = KERN_RAGE_THREAD;
3385	} else if (p->p_lflag & P_LRAGE_VNODES) {
3386	old_value = KERN_RAGE_PROC;
3387	} else {
3388	old_value = `0`;
3389	}
3390
3391	error = sysctl_io_number(req, bigValue: old_value, valueSize: sizeof(int), pValue: &new_value, changed: &changed);
3392
3393	if ((error == `0`) && (changed != `0`)) {
3394	switch (new_value) {
3395	case KERN_RAGE_PROC:
3396	proc_lock(p);
3397	p->p_lflag \|= P_LRAGE_VNODES;
3398	proc_unlock(p);
3399	break;
3400	case KERN_UNRAGE_PROC:
3401	proc_lock(p);
3402	p->p_lflag &= ~P_LRAGE_VNODES;
3403	proc_unlock(p);
3404	break;
3405
3406	case KERN_RAGE_THREAD:
3407	ut->uu_flag \|= UT_RAGE_VNODES;
3408	break;
3409	case KERN_UNRAGE_THREAD:
3410	ut = current_uthread();
3411	ut->uu_flag &= ~UT_RAGE_VNODES;
3412	break;
3413	}
3414	}
3415	return error;
3416	}
3417
3418	SYSCTL_PROC(_kern, KERN_RAGEVNODE, rage_vnode,
3419	CTLTYPE_INT \| CTLFLAG_RW \| CTLFLAG_ANYBODY \| CTLFLAG_LOCKED,
3420	`0`, `0`, sysctl_rage_vnode, "I", "");
3421
3422	/ XXX until filecoordinationd fixes a bit of inverted logic. /
3423	STATIC int
3424	sysctl_vfsnspace
3425	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
3426	{
3427	int old_value = `0`, new_value, changed;
3428
3429	return sysctl_io_number(req, bigValue: old_value, valueSize: sizeof(int), pValue: &new_value,
3430	changed: &changed);
3431	}
3432
3433	SYSCTL_PROC(_kern, OID_AUTO, vfsnspace,
3434	CTLTYPE_INT \| CTLFLAG_RW \| CTLFLAG_ANYBODY \| CTLFLAG_LOCKED,
3435	`0`, `0`, sysctl_vfsnspace, "I", "");
3436
3437	/ XXX move this interface into libproc and remove this sysctl /
3438	STATIC int
3439	sysctl_setthread_cpupercent
3440	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
3441	{
3442	int new_value, old_value;
3443	int error = `0`;
3444	kern_return_t kret = KERN_SUCCESS;
3445	uint8_t percent = `0`;
3446	int ms_refill = `0`;
3447
3448	if (!req->newptr) {
3449	return `0`;
3450	}
3451
3452	old_value = `0`;
3453
3454	if ((error = sysctl_io_number(req, bigValue: old_value, valueSize: sizeof(old_value), pValue: &new_value, NULL)) != `0`) {
3455	return error;
3456	}
3457
3458	percent = new_value & `0xff`; / low 8 bytes for perent /
3459	ms_refill = (new_value >> `8`) & `0xffffff`; / upper 24bytes represent ms refill value /
3460	if (percent > `100`) {
3461	return EINVAL;
3462	}
3463
3464	/*
3465	* If the caller is specifying a percentage of 0, this will unset the CPU limit, if present.
3466	*/
3467	kret = percent == `0` ?
3468	thread_set_cpulimit(THREAD_CPULIMIT_DISABLE, percentage: `0`, interval_ns: `0`) :
3469	thread_set_cpulimit(THREAD_CPULIMIT_BLOCK, percentage: percent, interval_ns: ms_refill * (int)NSEC_PER_MSEC);
3470
3471	if (kret != `0`) {
3472	return EIO;
3473	}
3474
3475	return `0`;
3476	}
3477
3478	SYSCTL_PROC(_kern, OID_AUTO, setthread_cpupercent,
3479	CTLTYPE_INT \| CTLFLAG_WR \| CTLFLAG_ANYBODY,
3480	`0`, `0`, sysctl_setthread_cpupercent, "I", "set thread cpu percentage limit");
3481
3482
3483	STATIC int
3484	sysctl_kern_check_openevt
3485	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
3486	{
3487	struct proc *p = req->p;
3488	int new_value, old_value, changed;
3489	int error;
3490
3491	if (p->p_flag & P_CHECKOPENEVT) {
3492	old_value = KERN_OPENEVT_PROC;
3493	} else {
3494	old_value = `0`;
3495	}
3496
3497	error = sysctl_io_number(req, bigValue: old_value, valueSize: sizeof(int), pValue: &new_value, changed: &changed);
3498
3499	if ((error == `0`) && (changed != `0`)) {
3500	switch (new_value) {
3501	case KERN_OPENEVT_PROC:
3502	OSBitOrAtomic(P_CHECKOPENEVT, &p->p_flag);
3503	break;
3504
3505	case KERN_UNOPENEVT_PROC:
3506	OSBitAndAtomic(~((uint32_t)P_CHECKOPENEVT), &p->p_flag);
3507	break;
3508
3509	default:
3510	error = EINVAL;
3511	}
3512	}
3513	return error;
3514	}
3515
3516	SYSCTL_PROC(_kern, KERN_CHECKOPENEVT, check_openevt, CTLTYPE_INT \| CTLFLAG_RW \| CTLFLAG_ANYBODY \| CTLFLAG_LOCKED,
3517	`0`, `0`, sysctl_kern_check_openevt, "I", "set the per-process check-open-evt flag");
3518
3519
3520	#if DEVELOPMENT \|\| DEBUG
3521	STATIC int
3522	sysctl_nx
3523	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
3524	{
3525	#ifdef SECURE_KERNEL
3526	(void)req;
3527	return ENOTSUP;
3528	#else
3529	int new_value, changed;
3530	int error;
3531
3532	error = sysctl_io_number(req, nx_enabled, sizeof(nx_enabled), &new_value, &changed);
3533	if (error) {
3534	return error;
3535	}
3536
3537	if (changed) {
3538	#if defined(__x86_64__)
3539	/*
3540	* Only allow setting if NX is supported on the chip
3541	*/
3542	if (!(cpuid_extfeatures() & CPUID_EXTFEATURE_XD)) {
3543	return ENOTSUP;
3544	}
3545	#endif
3546	nx_enabled = new_value;
3547	}
3548	return error;
3549	#endif /* SECURE_KERNEL */
3550	}
3551	#endif
3552
3553	#if DEVELOPMENT \|\| DEBUG
3554	SYSCTL_PROC(_kern, KERN_NX_PROTECTION, nx,
3555	CTLTYPE_INT \| CTLFLAG_RW \| CTLFLAG_KERN \| CTLFLAG_LOCKED,
3556	`0`, `0`, sysctl_nx, "I", "");
3557	#endif
3558
3559	STATIC int
3560	sysctl_loadavg
3561	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
3562	{
3563	if (proc_is64bit(req->p)) {
3564	struct user64_loadavg loadinfo64 = {};
3565	fill_loadavg64(la: &averunnable, la64: &loadinfo64);
3566	return sysctl_io_opaque(req, pValue: &loadinfo64, valueSize: sizeof(loadinfo64), NULL);
3567	} else {
3568	struct user32_loadavg loadinfo32 = {};
3569	fill_loadavg32(la: &averunnable, la32: &loadinfo32);
3570	return sysctl_io_opaque(req, pValue: &loadinfo32, valueSize: sizeof(loadinfo32), NULL);
3571	}
3572	}
3573
3574	SYSCTL_PROC(_vm, VM_LOADAVG, loadavg,
3575	CTLTYPE_STRUCT \| CTLFLAG_RD \| CTLFLAG_LOCKED,
3576	`0`, `0`, sysctl_loadavg, "S,loadavg", "");
3577
3578	/*
3579	* Note: Thread safe; vm_map_lock protects in vm_toggle_entry_reuse()
3580	*/
3581	STATIC int
3582	sysctl_vm_toggle_address_reuse(__unused struct sysctl_oid oidp, __unused void* *arg1,
3583	__unused int arg2, struct sysctl_req *req)
3584	{
3585	int old_value = `0`, new_value = `0`, error = `0`;
3586
3587	if (vm_toggle_entry_reuse( VM_TOGGLE_GETVALUE, &old_value )) {
3588	return error;
3589	}
3590	error = sysctl_io_number(req, bigValue: old_value, valueSize: sizeof(int), pValue: &new_value, NULL);
3591	if (!error) {
3592	return vm_toggle_entry_reuse(new_value, NULL);
3593	}
3594	return error;
3595	}
3596
3597	SYSCTL_PROC(_debug, OID_AUTO, toggle_address_reuse, CTLFLAG_ANYBODY \| CTLTYPE_INT \| CTLFLAG_RW \| CTLFLAG_LOCKED, `0`, `0`, sysctl_vm_toggle_address_reuse, "I", "");
3598
3599	#ifdef CONFIG_XNUPOST
3600
3601	extern uint32_t xnupost_get_estimated_testdata_size(void);
3602	extern int xnupost_reset_all_tests(void);
3603
3604	STATIC int
3605	sysctl_handle_xnupost_get_tests SYSCTL_HANDLER_ARGS
3606	{
3607	/ fixup unused arguments warnings /
3608	__unused int _oa2 = arg2;
3609	__unused void * _oa1 = arg1;
3610	__unused struct sysctl_oid * _oidp = oidp;
3611
3612	int error = `0`;
3613	user_addr_t oldp = `0`;
3614	user_addr_t newp = `0`;
3615	uint32_t usedbytes = `0`;
3616
3617	oldp = req->oldptr;
3618	newp = req->newptr;
3619
3620	if (newp) {
3621	return ENOTSUP;
3622	}
3623
3624	if ((void *)oldp == NULL) {
3625	/ return estimated size for second call where info can be placed /
3626	req->oldidx = xnupost_get_estimated_testdata_size();
3627	} else {
3628	error = xnupost_export_testdata((void *)oldp, req->oldlen, &usedbytes);
3629	req->oldidx = usedbytes;
3630	}
3631
3632	return error;
3633	}
3634
3635	SYSCTL_PROC(_debug,
3636	OID_AUTO,
3637	xnupost_get_tests,
3638	CTLFLAG_MASKED \| CTLFLAG_ANYBODY \| CTLTYPE_OPAQUE \| CTLFLAG_RD \| CTLFLAG_LOCKED,
3639	`0`,
3640	`0`,
3641	sysctl_handle_xnupost_get_tests,
3642	"-",
3643	"read xnupost test data in kernel");
3644
3645	#if CONFIG_EXT_PANICLOG
3646	/*
3647	* Extensible panic log test hooks
3648	*/
3649	static int
3650	sysctl_debug_ext_paniclog_test_hook SYSCTL_HANDLER_ARGS
3651	{
3652	#pragma unused(arg1, arg2)
3653	int rval = `0`;
3654	uint32_t test_option = `0`;
3655
3656	rval = sysctl_handle_int(oidp, &test_option, `0`, req);
3657
3658	if (rval == `0` && req->newptr) {
3659	rval = ext_paniclog_test_hook(test_option);
3660	}
3661
3662	return rval;
3663	}
3664
3665	SYSCTL_PROC(_debug, OID_AUTO, ext_paniclog_test_hook,
3666	CTLTYPE_INT \| CTLFLAG_RW,
3667	`0`, `0`,
3668	sysctl_debug_ext_paniclog_test_hook, "A", "ext paniclog test hook");
3669
3670	#endif
3671
3672	STATIC int
3673	sysctl_debug_xnupost_ctl SYSCTL_HANDLER_ARGS
3674	{
3675	/ fixup unused arguments warnings /
3676	__unused int _oa2 = arg2;
3677	__unused void * _oa1 = arg1;
3678	__unused struct sysctl_oid * _oidp = oidp;
3679
3680	#define ARRCOUNT 4
3681	/*
3682	* INPUT: ACTION, PARAM1, PARAM2, PARAM3
3683	* OUTPUT: RESULTCODE, ADDITIONAL DATA
3684	*/
3685	int32_t outval[ARRCOUNT] = {`0`};
3686	int32_t input[ARRCOUNT] = {`0`};
3687	int32_t out_size = sizeof(outval);
3688	int32_t in_size = sizeof(input);
3689	int error = `0`;
3690
3691	/ if this is NULL call to find out size, send out size info /
3692	if (!req->newptr) {
3693	goto out;
3694	}
3695
3696	/ pull in provided value from userspace /
3697	error = SYSCTL_IN(req, &input[`0`], in_size);
3698	if (error) {
3699	return error;
3700	}
3701
3702	if (input[`0`] == XTCTL_RESET_TESTDATA) {
3703	outval[`0`] = xnupost_reset_all_tests();
3704	goto out;
3705	}
3706
3707	out:
3708	error = SYSCTL_OUT(req, &outval[`0`], out_size);
3709	return error;
3710	}
3711
3712	SYSCTL_PROC(_debug,
3713	OID_AUTO,
3714	xnupost_testctl,
3715	CTLFLAG_MASKED \| CTLFLAG_ANYBODY \| CTLTYPE_OPAQUE \| CTLFLAG_RW \| CTLFLAG_LOCKED,
3716	`0`,
3717	`0`,
3718	sysctl_debug_xnupost_ctl,
3719	"I",
3720	"xnupost control for kernel testing");
3721
3722	extern void test_oslog_handleOSLogCtl(int32_t * in, int32_t * out, int32_t arraycount);
3723
3724	STATIC int
3725	sysctl_debug_test_oslog_ctl(__unused struct sysctl_oid * oidp, __unused void * arg1, __unused int arg2, struct sysctl_req * req)
3726	{
3727	#define ARRCOUNT 4
3728	int32_t outval[ARRCOUNT] = {`0`};
3729	int32_t input[ARRCOUNT] = {`0`};
3730	int32_t size_outval = sizeof(outval);
3731	int32_t size_inval = sizeof(input);
3732	int32_t error;
3733
3734	/ if this is NULL call to find out size, send out size info /
3735	if (!req->newptr) {
3736	error = SYSCTL_OUT(req, &outval[`0`], size_outval);
3737	return error;
3738	}
3739
3740	/ pull in provided value from userspace /
3741	error = SYSCTL_IN(req, &input[`0`], size_inval);
3742	if (error) {
3743	return error;
3744	}
3745
3746	test_oslog_handleOSLogCtl(input, outval, ARRCOUNT);
3747
3748	error = SYSCTL_OUT(req, &outval[`0`], size_outval);
3749
3750	return error;
3751	}
3752
3753	SYSCTL_PROC(_debug,
3754	OID_AUTO,
3755	test_OSLogCtl,
3756	CTLFLAG_MASKED \| CTLFLAG_ANYBODY \| CTLTYPE_OPAQUE \| CTLFLAG_RW \| CTLFLAG_LOCKED,
3757	`0`,
3758	`0`,
3759	sysctl_debug_test_oslog_ctl,
3760	"I",
3761	"testing oslog in kernel");
3762
3763	#include <mach/task.h>
3764	#include <mach/semaphore.h>
3765
3766	static LCK_GRP_DECLARE(sysctl_debug_test_stackshot_owner_grp, "test-stackshot-owner-grp");
3767	static LCK_MTX_DECLARE(sysctl_debug_test_stackshot_owner_init_mtx,
3768	&sysctl_debug_test_stackshot_owner_grp);
3769
3770	/ This is a sysctl for testing collection of owner info on a lock in kernel space. A multi-threaded*
3771	* test from userland sets this sysctl in such a way that a thread blocks in kernel mode, and a
3772	* stackshot is taken to see if the owner of the lock can be identified.
3773	*
3774	* We can't return to userland with a kernel lock held, so be sure to unlock before we leave.
3775	* the semaphores allow us to artificially create cases where the lock is being held and the
3776	* thread is hanging / taking a long time to do something. */
3777
3778	volatile char sysctl_debug_test_stackshot_mtx_inited = `0`;
3779	semaphore_t sysctl_debug_test_stackshot_mutex_sem;
3780	lck_mtx_t sysctl_debug_test_stackshot_owner_lck;
3781
3782	#define SYSCTL_DEBUG_MTX_ACQUIRE_WAIT 1
3783	#define SYSCTL_DEBUG_MTX_ACQUIRE_NOWAIT 2
3784	#define SYSCTL_DEBUG_MTX_SIGNAL 3
3785	#define SYSCTL_DEBUG_MTX_TEARDOWN 4
3786
3787	STATIC int
3788	sysctl_debug_test_stackshot_mutex_owner(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
3789	{
3790	long long option = -`1`;
3791	/ if the user tries to read the sysctl, we tell them what the address of the lock is (to test against stackshot's output) /
3792	long long mtx_unslid_addr = (long long)VM_KERNEL_UNSLIDE_OR_PERM(&sysctl_debug_test_stackshot_owner_lck);
3793	int error = sysctl_io_number(req, mtx_unslid_addr, sizeof(long long), (void*)&option, NULL);
3794
3795	lck_mtx_lock(&sysctl_debug_test_stackshot_owner_init_mtx);
3796	if (!sysctl_debug_test_stackshot_mtx_inited) {
3797	lck_mtx_init(&sysctl_debug_test_stackshot_owner_lck,
3798	&sysctl_debug_test_stackshot_owner_grp,
3799	LCK_ATTR_NULL);
3800	semaphore_create(kernel_task,
3801	&sysctl_debug_test_stackshot_mutex_sem,
3802	SYNC_POLICY_FIFO, `0`);
3803	sysctl_debug_test_stackshot_mtx_inited = `1`;
3804	}
3805	lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_init_mtx);
3806
3807	if (!error) {
3808	switch (option) {
3809	case SYSCTL_DEBUG_MTX_ACQUIRE_NOWAIT:
3810	lck_mtx_lock(&sysctl_debug_test_stackshot_owner_lck);
3811	lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_lck);
3812	break;
3813	case SYSCTL_DEBUG_MTX_ACQUIRE_WAIT:
3814	lck_mtx_lock(&sysctl_debug_test_stackshot_owner_lck);
3815	semaphore_wait(sysctl_debug_test_stackshot_mutex_sem);
3816	lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_lck);
3817	break;
3818	case SYSCTL_DEBUG_MTX_SIGNAL:
3819	semaphore_signal(sysctl_debug_test_stackshot_mutex_sem);
3820	break;
3821	case SYSCTL_DEBUG_MTX_TEARDOWN:
3822	lck_mtx_lock(&sysctl_debug_test_stackshot_owner_init_mtx);
3823
3824	lck_mtx_destroy(&sysctl_debug_test_stackshot_owner_lck,
3825	&sysctl_debug_test_stackshot_owner_grp);
3826	semaphore_destroy(kernel_task,
3827	sysctl_debug_test_stackshot_mutex_sem);
3828	sysctl_debug_test_stackshot_mtx_inited = `0`;
3829
3830	lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_init_mtx);
3831	break;
3832	case -`1`: / user just wanted to read the value, so do nothing /
3833	break;
3834	default:
3835	error = EINVAL;
3836	break;
3837	}
3838	}
3839	return error;
3840	}
3841
3842	/ we can't return to userland with a kernel rwlock held, so be sure to unlock before we leave.*
3843	* the semaphores allow us to artificially create cases where the lock is being held and the
3844	* thread is hanging / taking a long time to do something. */
3845
3846	SYSCTL_PROC(_debug,
3847	OID_AUTO,
3848	test_MutexOwnerCtl,
3849	CTLFLAG_MASKED \| CTLFLAG_ANYBODY \| CTLTYPE_QUAD \| CTLFLAG_RW \| CTLFLAG_LOCKED,
3850	`0`,
3851	`0`,
3852	sysctl_debug_test_stackshot_mutex_owner,
3853	"-",
3854	"Testing mutex owner in kernel");
3855
3856	volatile char sysctl_debug_test_stackshot_rwlck_inited = `0`;
3857	lck_rw_t sysctl_debug_test_stackshot_owner_rwlck;
3858	semaphore_t sysctl_debug_test_stackshot_rwlck_sem;
3859
3860	#define SYSCTL_DEBUG_KRWLCK_RACQUIRE_NOWAIT 1
3861	#define SYSCTL_DEBUG_KRWLCK_RACQUIRE_WAIT 2
3862	#define SYSCTL_DEBUG_KRWLCK_WACQUIRE_NOWAIT 3
3863	#define SYSCTL_DEBUG_KRWLCK_WACQUIRE_WAIT 4
3864	#define SYSCTL_DEBUG_KRWLCK_SIGNAL 5
3865	#define SYSCTL_DEBUG_KRWLCK_TEARDOWN 6
3866
3867	STATIC int
3868	sysctl_debug_test_stackshot_rwlck_owner(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
3869	{
3870	long long option = -`1`;
3871	/ if the user tries to read the sysctl, we tell them what the address of the lock is*
3872	* (to test against stackshot's output) */
3873	long long rwlck_unslid_addr = (long long)VM_KERNEL_UNSLIDE_OR_PERM(&sysctl_debug_test_stackshot_owner_rwlck);
3874	int error = sysctl_io_number(req, rwlck_unslid_addr, sizeof(long long), (void*)&option, NULL);
3875
3876	lck_mtx_lock(&sysctl_debug_test_stackshot_owner_init_mtx);
3877	if (!sysctl_debug_test_stackshot_rwlck_inited) {
3878	lck_rw_init(&sysctl_debug_test_stackshot_owner_rwlck,
3879	&sysctl_debug_test_stackshot_owner_grp,
3880	LCK_ATTR_NULL);
3881	semaphore_create(kernel_task,
3882	&sysctl_debug_test_stackshot_rwlck_sem,
3883	SYNC_POLICY_FIFO,
3884	`0`);
3885	sysctl_debug_test_stackshot_rwlck_inited = `1`;
3886	}
3887	lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_init_mtx);
3888
3889	if (!error) {
3890	switch (option) {
3891	case SYSCTL_DEBUG_KRWLCK_RACQUIRE_NOWAIT:
3892	lck_rw_lock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_SHARED);
3893	lck_rw_unlock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_SHARED);
3894	break;
3895	case SYSCTL_DEBUG_KRWLCK_RACQUIRE_WAIT:
3896	lck_rw_lock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_SHARED);
3897	semaphore_wait(sysctl_debug_test_stackshot_rwlck_sem);
3898	lck_rw_unlock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_SHARED);
3899	break;
3900	case SYSCTL_DEBUG_KRWLCK_WACQUIRE_NOWAIT:
3901	lck_rw_lock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_EXCLUSIVE);
3902	lck_rw_unlock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_EXCLUSIVE);
3903	break;
3904	case SYSCTL_DEBUG_KRWLCK_WACQUIRE_WAIT:
3905	lck_rw_lock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_EXCLUSIVE);
3906	semaphore_wait(sysctl_debug_test_stackshot_rwlck_sem);
3907	lck_rw_unlock(&sysctl_debug_test_stackshot_owner_rwlck, LCK_RW_TYPE_EXCLUSIVE);
3908	break;
3909	case SYSCTL_DEBUG_KRWLCK_SIGNAL:
3910	semaphore_signal(sysctl_debug_test_stackshot_rwlck_sem);
3911	break;
3912	case SYSCTL_DEBUG_KRWLCK_TEARDOWN:
3913	lck_mtx_lock(&sysctl_debug_test_stackshot_owner_init_mtx);
3914
3915	lck_rw_destroy(&sysctl_debug_test_stackshot_owner_rwlck,
3916	&sysctl_debug_test_stackshot_owner_grp);
3917	semaphore_destroy(kernel_task,
3918	sysctl_debug_test_stackshot_rwlck_sem);
3919	sysctl_debug_test_stackshot_rwlck_inited = `0`;
3920
3921	lck_mtx_unlock(&sysctl_debug_test_stackshot_owner_init_mtx);
3922	break;
3923	case -`1`: / user just wanted to read the value, so do nothing /
3924	break;
3925	default:
3926	error = EINVAL;
3927	break;
3928	}
3929	}
3930	return error;
3931	}
3932
3933
3934	SYSCTL_PROC(_debug,
3935	OID_AUTO,
3936	test_RWLockOwnerCtl,
3937	CTLFLAG_MASKED \| CTLFLAG_ANYBODY \| CTLTYPE_QUAD \| CTLFLAG_RW \| CTLFLAG_LOCKED,
3938	`0`,
3939	`0`,
3940	sysctl_debug_test_stackshot_rwlck_owner,
3941	"-",
3942	"Testing rwlock owner in kernel");
3943	#endif /* !CONFIG_XNUPOST */
3944
3945	STATIC int
3946	sysctl_swapusage
3947	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
3948	{
3949	int error;
3950	uint64_t swap_total;
3951	uint64_t swap_avail;
3952	vm_size_t swap_pagesize;
3953	boolean_t swap_encrypted;
3954	struct xsw_usage xsu = {};
3955
3956	error = macx_swapinfo(total_p: &swap_total,
3957	avail_p: &swap_avail,
3958	pagesize_p: &swap_pagesize,
3959	encrypted_p: &swap_encrypted);
3960	if (error) {
3961	return error;
3962	}
3963
3964	xsu.xsu_total = swap_total;
3965	xsu.xsu_avail = swap_avail;
3966	xsu.xsu_used = swap_total - swap_avail;
3967	xsu.xsu_pagesize = (u_int32_t)MIN(swap_pagesize, UINT32_MAX);
3968	xsu.xsu_encrypted = swap_encrypted;
3969	return sysctl_io_opaque(req, pValue: &xsu, valueSize: sizeof(xsu), NULL);
3970	}
3971
3972
3973
3974	SYSCTL_PROC(_vm, VM_SWAPUSAGE, swapusage,
3975	CTLTYPE_STRUCT \| CTLFLAG_RD \| CTLFLAG_LOCKED,
3976	`0`, `0`, sysctl_swapusage, "S,xsw_usage", "");
3977
3978	extern int vm_swap_enabled;
3979	SYSCTL_INT(_vm, OID_AUTO, swap_enabled, CTLFLAG_RD \| CTLFLAG_LOCKED, &vm_swap_enabled, `0`, "");
3980
3981	#if DEVELOPMENT \|\| DEBUG
3982	extern int vm_num_swap_files_config;
3983	extern int vm_num_swap_files;
3984	extern lck_mtx_t vm_swap_data_lock;
3985	#define VM_MAX_SWAP_FILE_NUM 100
3986
3987	static int
3988	sysctl_vm_config_num_swap_files SYSCTL_HANDLER_ARGS
3989	{
3990	#pragma unused(arg1, arg2)
3991	int error = `0`, val = vm_num_swap_files_config;
3992
3993	error = sysctl_handle_int(oidp, &val, `0`, req);
3994	if (error \|\| !req->newptr) {
3995	goto out;
3996	}
3997
3998	if (!VM_CONFIG_SWAP_IS_ACTIVE && !VM_CONFIG_FREEZER_SWAP_IS_ACTIVE) {
3999	printf("Swap is disabled\n");
4000	error = EINVAL;
4001	goto out;
4002	}
4003
4004	lck_mtx_lock(&vm_swap_data_lock);
4005
4006	if (val < vm_num_swap_files) {
4007	printf("Cannot configure fewer swap files than already exist.\n");
4008	error = EINVAL;
4009	lck_mtx_unlock(&vm_swap_data_lock);
4010	goto out;
4011	}
4012
4013	if (val > VM_MAX_SWAP_FILE_NUM) {
4014	printf("Capping number of swap files to upper bound.\n");
4015	val = VM_MAX_SWAP_FILE_NUM;
4016	}
4017
4018	vm_num_swap_files_config = val;
4019	lck_mtx_unlock(&vm_swap_data_lock);
4020	out:
4021
4022	return `0`;
4023	}
4024
4025	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", "");
4026	#endif /* DEVELOPMENT \|\| DEBUG */
4027
4028	/ this kernel does NOT implement shared_region_make_private_np() /
4029	SYSCTL_INT(_kern, KERN_SHREG_PRIVATIZABLE, shreg_private,
4030	CTLFLAG_RD \| CTLFLAG_LOCKED,
4031	(int *)NULL, `0`, "");
4032
4033	STATIC int
4034	fetch_process_cputype(
4035	proc_t cur_proc,
4036	int *name,
4037	u_int namelen,
4038	cpu_type_t *cputype)
4039	{
4040	proc_t p = PROC_NULL;
4041	int refheld = `0`;
4042	cpu_type_t ret = `0`;
4043	int error = `0`;
4044
4045	if (namelen == `0`) {
4046	p = cur_proc;
4047	} else if (namelen == `1`) {
4048	p = proc_find(pid: name[`0`]);
4049	if (p == NULL) {
4050	return EINVAL;
4051	}
4052	refheld = `1`;
4053	} else {
4054	error = EINVAL;
4055	goto out;
4056	}
4057
4058	ret = cpu_type() & ~CPU_ARCH_MASK;
4059	if (IS_64BIT_PROCESS(p)) {
4060	ret \|= CPU_ARCH_ABI64;
4061	}
4062
4063	*cputype = ret;
4064
4065	if (refheld != `0`) {
4066	proc_rele(p);
4067	}
4068	out:
4069	return error;
4070	}
4071
4072
4073	STATIC int
4074	sysctl_sysctl_native(__unused struct sysctl_oid oidp, void* arg1, int* arg2,
4075	struct sysctl_req *req)
4076	{
4077	int error;
4078	cpu_type_t proc_cputype = `0`;
4079	if ((error = fetch_process_cputype(cur_proc: req->p, name: (int *)arg1, namelen: arg2, cputype: &proc_cputype)) != `0`) {
4080	return error;
4081	}
4082	int res = `1`;
4083	if ((proc_cputype & ~CPU_ARCH_MASK) != (cpu_type() & ~CPU_ARCH_MASK)) {
4084	res = `0`;
4085	}
4086	return SYSCTL_OUT(req, &res, sizeof(res));
4087	}
4088	SYSCTL_PROC(_sysctl, OID_AUTO, proc_native, CTLTYPE_INT \| CTLFLAG_RD \| CTLFLAG_LOCKED, `0`, `0`, sysctl_sysctl_native, "I", "proc_native");
4089
4090	STATIC int
4091	sysctl_sysctl_cputype(__unused struct sysctl_oid oidp, void* arg1, int* arg2,
4092	struct sysctl_req *req)
4093	{
4094	int error;
4095	cpu_type_t proc_cputype = `0`;
4096	if ((error = fetch_process_cputype(cur_proc: req->p, name: (int *)arg1, namelen: arg2, cputype: &proc_cputype)) != `0`) {
4097	return error;
4098	}
4099	return SYSCTL_OUT(req, &proc_cputype, sizeof(proc_cputype));
4100	}
4101	SYSCTL_PROC(_sysctl, OID_AUTO, proc_cputype, CTLTYPE_NODE \| CTLFLAG_RD \| CTLFLAG_LOCKED, `0`, `0`, sysctl_sysctl_cputype, "I", "proc_cputype");
4102
4103	STATIC int
4104	sysctl_safeboot
4105	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
4106	{
4107	return sysctl_io_number(req, bigValue: boothowto & RB_SAFEBOOT ? `1` : `0`, valueSize: sizeof(int), NULL, NULL);
4108	}
4109
4110	SYSCTL_PROC(_kern, KERN_SAFEBOOT, safeboot,
4111	CTLTYPE_INT \| CTLFLAG_RD \| CTLFLAG_LOCKED,
4112	`0`, `0`, sysctl_safeboot, "I", "");
4113
4114	STATIC int
4115	sysctl_singleuser
4116	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
4117	{
4118	return sysctl_io_number(req, bigValue: boothowto & RB_SINGLE ? `1` : `0`, valueSize: sizeof(int), NULL, NULL);
4119	}
4120
4121	SYSCTL_PROC(_kern, OID_AUTO, singleuser,
4122	CTLTYPE_INT \| CTLFLAG_RD \| CTLFLAG_LOCKED,
4123	`0`, `0`, sysctl_singleuser, "I", "");
4124
4125	STATIC int
4126	sysctl_minimalboot
4127	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
4128	{
4129	return sysctl_io_number(req, bigValue: minimalboot, valueSize: sizeof(int), NULL, NULL);
4130	}
4131
4132	SYSCTL_PROC(_kern, OID_AUTO, minimalboot,
4133	CTLTYPE_INT \| CTLFLAG_RD \| CTLFLAG_LOCKED,
4134	`0`, `0`, sysctl_minimalboot, "I", "");
4135
4136	/*
4137	* Controls for debugging affinity sets - see osfmk/kern/affinity.c
4138	*/
4139	extern boolean_t affinity_sets_enabled;
4140	extern int affinity_sets_mapping;
4141
4142	SYSCTL_INT(_kern, OID_AUTO, affinity_sets_enabled,
4143	CTLFLAG_RW \| CTLFLAG_LOCKED, (int *) &affinity_sets_enabled, `0`, "hinting enabled");
4144	SYSCTL_INT(_kern, OID_AUTO, affinity_sets_mapping,
4145	CTLFLAG_RW \| CTLFLAG_LOCKED, &affinity_sets_mapping, `0`, "mapping policy");
4146
4147	/*
4148	* Boolean indicating if KASLR is active.
4149	*/
4150	STATIC int
4151	sysctl_slide
4152	(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
4153	{
4154	uint32_t slide;
4155
4156	slide = vm_kernel_slide ? `1` : `0`;
4157
4158	return sysctl_io_number( req, bigValue: slide, valueSize: sizeof(int), NULL, NULL);
4159	}
4160
4161	SYSCTL_PROC(_kern, OID_AUTO, slide,
4162	CTLTYPE_INT \| CTLFLAG_RD \| CTLFLAG_LOCKED,
4163	`0`, `0`, sysctl_slide, "I", "");
4164
4165	#if DEBUG \|\| DEVELOPMENT
4166	#if defined(__arm64__)
4167	extern vm_offset_t segTEXTEXECB;
4168
4169	static int
4170	sysctl_kernel_text_exec_base_slide SYSCTL_HANDLER_ARGS
4171	{
4172	#pragma unused(arg1, arg2, oidp)
4173	unsigned long slide = `0`;
4174	kc_format_t kc_format;
4175
4176	PE_get_primary_kc_format(&kc_format);
4177
4178	if (kc_format == KCFormatFileset) {
4179	void *kch = PE_get_kc_header(KCKindPrimary);
4180	slide = (unsigned long)segTEXTEXECB - (unsigned long)kch + vm_kernel_slide;
4181	}
4182	return SYSCTL_OUT(req, &slide, sizeof(slide));
4183	}
4184
4185	SYSCTL_QUAD(_kern, OID_AUTO, kernel_slide, CTLFLAG_RD \| CTLFLAG_KERN \| CTLFLAG_LOCKED, &vm_kernel_slide, "");
4186	SYSCTL_QUAD(_kern, OID_AUTO, kernel_text_exec_base, CTLFLAG_RD \| CTLFLAG_KERN \| CTLFLAG_LOCKED, &segTEXTEXECB, "");
4187	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", "");
4188	#endif /* defined(__arm64__) */
4189
4190	/ User address of the PFZ /
4191	extern user32_addr_t commpage_text32_location;
4192	extern user64_addr_t commpage_text64_location;
4193
4194	STATIC int
4195	sysctl_pfz_start SYSCTL_HANDLER_ARGS
4196	{
4197	#pragma unused(oidp, arg1, arg2)
4198
4199	#ifdef __LP64__
4200	return sysctl_io_number(req, commpage_text64_location, sizeof(user64_addr_t), NULL, NULL);
4201	#else
4202	return sysctl_io_number(req, commpage_text32_location, sizeof(user32_addr_t), NULL, NULL);
4203	#endif
4204	}
4205
4206	SYSCTL_PROC(_kern, OID_AUTO, pfz,
4207	CTLTYPE_QUAD \| CTLFLAG_RD \| CTLFLAG_LOCKED \| CTLFLAG_MASKED,
4208	`0`, `0`, sysctl_pfz_start, "I", "");
4209	#endif
4210
4211
4212	/*
4213	* Limit on total memory users can wire.
4214	*
4215	* vm_global_user_wire_limit - system wide limit on wired memory from all processes combined.
4216	*
4217	* vm_per_task_user_wire_limit - per address space limit on wired memory. This puts a cap on the process's rlimit value.
4218	*
4219	* These values are initialized to reasonable defaults at boot time based on the available physical memory in
4220	* kmem_init().
4221	*
4222	* All values are in bytes.
4223	*/
4224
4225	vm_map_size_t vm_global_user_wire_limit;
4226	vm_map_size_t vm_per_task_user_wire_limit;
4227	extern uint64_t max_mem_actual, max_mem;
4228
4229	uint64_t vm_add_wire_count_over_global_limit;
4230	uint64_t vm_add_wire_count_over_user_limit;
4231	/*
4232	* We used to have a global in the kernel called vm_global_no_user_wire_limit which was the inverse
4233	* of vm_global_user_wire_limit. But maintaining both of those is silly, and vm_global_user_wire_limit is the
4234	* real limit.
4235	* This function is for backwards compatibility with userspace
4236	* since we exposed the old global via a sysctl.
4237	*/
4238	STATIC int
4239	sysctl_global_no_user_wire_amount(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
4240	{
4241	vm_map_size_t old_value;
4242	vm_map_size_t new_value;
4243	int changed;
4244	int error;
4245	uint64_t config_memsize = max_mem;
4246	#if defined(XNU_TARGET_OS_OSX)
4247	config_memsize = max_mem_actual;
4248	#endif /* defined(XNU_TARGET_OS_OSX) */
4249
4250	old_value = (vm_map_size_t)(config_memsize - vm_global_user_wire_limit);
4251	error = sysctl_io_number(req, bigValue: old_value, valueSize: sizeof(vm_map_size_t), pValue: &new_value, changed: &changed);
4252	if (changed) {
4253	if ((uint64_t)new_value > config_memsize) {
4254	error = EINVAL;
4255	} else {
4256	vm_global_user_wire_limit = (vm_map_size_t)(config_memsize - new_value);
4257	}
4258	}
4259	return error;
4260	}
4261	/*
4262	* There needs to be a more automatic/elegant way to do this
4263	*/
4264	SYSCTL_QUAD(_vm, OID_AUTO, global_user_wire_limit, CTLFLAG_RW \| CTLFLAG_LOCKED, &vm_global_user_wire_limit, "");
4265	SYSCTL_QUAD(_vm, OID_AUTO, user_wire_limit, CTLFLAG_RW \| CTLFLAG_LOCKED, &vm_per_task_user_wire_limit, "");
4266	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", "");
4267
4268	/*
4269	* Relaxed atomic RW of a 64bit value via sysctl.
4270	*/
4271	STATIC int
4272	sysctl_r_64bit_atomic(uint64_t ptr, struct* sysctl_req *req)
4273	{
4274	uint64_t old_value;
4275	uint64_t new_value;
4276	int error;
4277
4278	old_value = os_atomic_load_wide(ptr, relaxed);
4279	error = sysctl_io_number(req, bigValue: old_value, valueSize: sizeof(vm_map_size_t), pValue: &new_value, NULL);
4280	return error;
4281	}
4282	STATIC int
4283	sysctl_add_wire_count_over_global_limit(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
4284	{
4285	return sysctl_r_64bit_atomic(ptr: &vm_add_wire_count_over_global_limit, req);
4286	}
4287	STATIC int
4288	sysctl_add_wire_count_over_user_limit(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
4289	{
4290	return sysctl_r_64bit_atomic(ptr: &vm_add_wire_count_over_user_limit, req);
4291	}
4292
4293	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", "");
4294	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", "");
4295
4296	#if DEVELOPMENT \|\| DEBUG
4297	/ These sysctls are used to test the wired limit. /
4298	extern unsigned int vm_page_wire_count;
4299	extern uint32_t vm_lopage_free_count;
4300	SYSCTL_INT(_vm, OID_AUTO, page_wire_count, CTLFLAG_RD \| CTLFLAG_LOCKED, &vm_page_wire_count, `0`, "");
4301	SYSCTL_INT(_vm, OID_AUTO, lopage_free_count, CTLFLAG_RD \| CTLFLAG_LOCKED, &vm_lopage_free_count, `0`, "");
4302
4303	/*
4304	* Setting the per task variable exclude_physfootprint_ledger to 1 will allow the calling task to exclude memory entries that are
4305	* tagged by VM_LEDGER_TAG_DEFAULT and flagged by VM_LEDGER_FLAG_EXCLUDE_FOOTPRINT_DEBUG from its phys_footprint ledger.
4306	*/
4307
4308	STATIC int
4309	sysctl_rw_task_no_footprint_for_debug(struct sysctl_oid oidp __unused, void* arg1 __unused, int* arg2 __unused, struct sysctl_req *req)
4310	{
4311	int error;
4312	int value;
4313	proc_t p = current_proc();
4314
4315	if (req->newptr) {
4316	// Write request
4317	error = SYSCTL_IN(req, &value, sizeof(value));
4318	if (!error) {
4319	if (value == `1`) {
4320	task_set_no_footprint_for_debug(proc_task(p), TRUE);
4321	} else if (value == `0`) {
4322	task_set_no_footprint_for_debug(proc_task(p), FALSE);
4323	} else {
4324	error = EINVAL;
4325	}
4326	}
4327	} else {
4328	// Read request
4329	value = task_get_no_footprint_for_debug(proc_task(p));
4330	error = SYSCTL_OUT(req, &value, sizeof(value));
4331	}
4332	return error;
4333	}
4334
4335	SYSCTL_PROC(_vm, OID_AUTO, task_no_footprint_for_debug,
4336	CTLTYPE_INT \| CTLFLAG_RW \| CTLFLAG_LOCKED \| CTLFLAG_ANYBODY,
4337	`0`, `0`, &sysctl_rw_task_no_footprint_for_debug, "I", "Allow debug memory to be excluded from this task's memory footprint (debug only)");
4338
4339	#endif /* DEVELOPMENT \|\| DEBUG */
4340
4341
4342	extern int vm_map_copy_overwrite_aligned_src_not_internal;
4343	extern int vm_map_copy_overwrite_aligned_src_not_symmetric;
4344	extern int vm_map_copy_overwrite_aligned_src_large;
4345	SYSCTL_INT(_vm, OID_AUTO, vm_copy_src_not_internal, CTLFLAG_RD \| CTLFLAG_LOCKED, &vm_map_copy_overwrite_aligned_src_not_internal, `0`, "");
4346	SYSCTL_INT(_vm, OID_AUTO, vm_copy_src_not_symmetric, CTLFLAG_RD \| CTLFLAG_LOCKED, &vm_map_copy_overwrite_aligned_src_not_symmetric, `0`, "");
4347	SYSCTL_INT(_vm, OID_AUTO, vm_copy_src_large, CTLFLAG_RD \| CTLFLAG_LOCKED, &vm_map_copy_overwrite_aligned_src_large, `0`, "");
4348
4349
4350	extern uint32_t vm_page_external_count;
4351
4352	SYSCTL_INT(_vm, OID_AUTO, vm_page_external_count, CTLFLAG_RD \| CTLFLAG_LOCKED, &vm_page_external_count, `0`, "");
4353
4354	SYSCTL_INT(_vm, OID_AUTO, vm_page_filecache_min, CTLFLAG_RD \| CTLFLAG_LOCKED, &vm_pageout_state.vm_page_filecache_min, `0`, "");
4355	SYSCTL_INT(_vm, OID_AUTO, vm_page_xpmapped_min, CTLFLAG_RD \| CTLFLAG_LOCKED, &vm_pageout_state.vm_page_xpmapped_min, `0`, "");
4356
4357	#if DEVELOPMENT \|\| DEBUG
4358	SYSCTL_INT(_vm, OID_AUTO, vm_page_filecache_min_divisor, CTLFLAG_RW \| CTLFLAG_LOCKED, &vm_pageout_state.vm_page_filecache_min_divisor, `0`, "");
4359	SYSCTL_INT(_vm, OID_AUTO, vm_page_xpmapped_min_divisor, CTLFLAG_RW \| CTLFLAG_LOCKED, &vm_pageout_state.vm_page_xpmapped_min_divisor, `0`, "");
4360	extern boolean_t vps_yield_for_pgqlockwaiters;
4361	SYSCTL_INT(_vm, OID_AUTO, vm_pageoutscan_yields_for_pageQlockwaiters, CTLFLAG_RW \| CTLFLAG_LOCKED, &vps_yield_for_pgqlockwaiters, `0`, "");
4362	#endif
4363
4364	extern int vm_compressor_mode;
4365	extern int vm_compressor_is_active;
4366	extern int vm_compressor_available;
4367	extern uint32_t c_seg_bufsize;
4368	extern uint64_t compressor_pool_size;
4369	extern uint32_t vm_ripe_target_age;
4370	extern uint32_t swapout_target_age;
4371	extern int64_t compressor_bytes_used;
4372	extern int64_t c_segment_input_bytes;
4373	extern int64_t c_segment_compressed_bytes;
4374	extern uint32_t compressor_eval_period_in_msecs;
4375	extern uint32_t compressor_sample_min_in_msecs;
4376	extern uint32_t compressor_sample_max_in_msecs;
4377	extern uint32_t compressor_thrashing_threshold_per_10msecs;
4378	extern uint32_t compressor_thrashing_min_per_10msecs;
4379	extern uint32_t vm_compressor_time_thread;
4380
4381	#if DEVELOPMENT \|\| DEBUG
4382	extern uint32_t vm_compressor_minorcompact_threshold_divisor;
4383	extern uint32_t vm_compressor_majorcompact_threshold_divisor;
4384	extern uint32_t vm_compressor_unthrottle_threshold_divisor;
4385	extern uint32_t vm_compressor_catchup_threshold_divisor;
4386
4387	extern uint32_t vm_compressor_minorcompact_threshold_divisor_overridden;
4388	extern uint32_t vm_compressor_majorcompact_threshold_divisor_overridden;
4389	extern uint32_t vm_compressor_unthrottle_threshold_divisor_overridden;
4390	extern uint32_t vm_compressor_catchup_threshold_divisor_overridden;
4391
4392	extern vmct_stats_t vmct_stats;
4393
4394
4395	STATIC int
4396	sysctl_minorcompact_threshold_divisor(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
4397	{
4398	int new_value, changed;
4399	int error = sysctl_io_number(req, vm_compressor_minorcompact_threshold_divisor, sizeof(int), &new_value, &changed);
4400
4401	if (changed) {
4402	vm_compressor_minorcompact_threshold_divisor = new_value;
4403	vm_compressor_minorcompact_threshold_divisor_overridden = `1`;
4404	}
4405	return error;
4406	}
4407
4408	SYSCTL_PROC(_vm, OID_AUTO, compressor_minorcompact_threshold_divisor,
4409	CTLTYPE_INT \| CTLFLAG_LOCKED \| CTLFLAG_RW,
4410	`0`, `0`, sysctl_minorcompact_threshold_divisor, "I", "");
4411
4412
4413	STATIC int
4414	sysctl_majorcompact_threshold_divisor(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
4415	{
4416	int new_value, changed;
4417	int error = sysctl_io_number(req, vm_compressor_majorcompact_threshold_divisor, sizeof(int), &new_value, &changed);
4418
4419	if (changed) {
4420	vm_compressor_majorcompact_threshold_divisor = new_value;
4421	vm_compressor_majorcompact_threshold_divisor_overridden = `1`;
4422	}
4423	return error;
4424	}
4425
4426	SYSCTL_PROC(_vm, OID_AUTO, compressor_majorcompact_threshold_divisor,
4427	CTLTYPE_INT \| CTLFLAG_LOCKED \| CTLFLAG_RW,
4428	`0`, `0`, sysctl_majorcompact_threshold_divisor, "I", "");
4429
4430
4431	STATIC int
4432	sysctl_unthrottle_threshold_divisor(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
4433	{
4434	int new_value, changed;
4435	int error = sysctl_io_number(req, vm_compressor_unthrottle_threshold_divisor, sizeof(int), &new_value, &changed);
4436
4437	if (changed) {
4438	vm_compressor_unthrottle_threshold_divisor = new_value;
4439	vm_compressor_unthrottle_threshold_divisor_overridden = `1`;
4440	}
4441	return error;
4442	}
4443
4444	SYSCTL_PROC(_vm, OID_AUTO, compressor_unthrottle_threshold_divisor,
4445	CTLTYPE_INT \| CTLFLAG_LOCKED \| CTLFLAG_RW,
4446	`0`, `0`, sysctl_unthrottle_threshold_divisor, "I", "");
4447
4448
4449	STATIC int
4450	sysctl_catchup_threshold_divisor(__unused struct sysctl_oid oidp, __unused void* arg1, __unused int* arg2, struct sysctl_req *req)
4451	{
4452	int new_value, changed;
4453	int error = sysctl_io_number(req, vm_compressor_catchup_threshold_divisor, sizeof(int), &new_value, &changed);
4454
4455	if (changed) {
4456	vm_compressor_catchup_threshold_divisor = new_value;
4457	vm_compressor_catchup_threshold_divisor_overridden = `1`;
4458	}
4459	return error;
4460	}
4461
4462	SYSCTL_PROC(_vm, OID_AUTO, compressor_catchup_threshold_divisor,
4463	CTLTYPE_INT \| CTLFLAG_LOCKED \| CTLFLAG_RW,
4464	`0`, `0`, sysctl_catchup_threshold_divisor, "I", "");
4465	#endif
4466
4467
4468	SYSCTL_QUAD(_vm, OID_AUTO, compressor_input_bytes, CTLFLAG_RD \| CTLFLAG_LOCKED, &c_segment_input_bytes, "");
4469	SYSCTL_QUAD(_vm, OID_AUTO, compressor_compressed_bytes, CTLFLAG_RD \| CTLFLAG_LOCKED, &c_segment_compressed_bytes, "");
4470	SYSCTL_QUAD(_vm, OID_AUTO, compressor_bytes_used, CTLFLAG_RD \| CTLFLAG_LOCKED, &compressor_bytes_used, "");
4471
4472	SYSCTL_INT(_vm, OID_AUTO, compressor_mode, CTLFLAG_RD \| CTLFLAG_LOCKED, &vm_compressor_mode, `0`, "");
4473	SYSCTL_INT(_vm, OID_AUTO, compressor_is_active, CTLFLAG_RD \| CTLFLAG_LOCKED, &vm_compressor_is_active, `0`, "");
4474

Browse the source code of xnu/bsd/kern/kern_sysctl.c