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