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 *
29 * Copyright (c) 1982, 1986, 1989, 1991, 1992, 1993
30 * The Regents of the University of California. All rights reserved.
31 * (c) UNIX System Laboratories, Inc.
32 * All or some portions of this file are derived from material licensed
33 * to the University of California by American Telephone and Telegraph
34 * Co. or Unix System Laboratories, Inc. and are reproduced herein with
35 * the permission of UNIX System Laboratories, Inc.
36 *
37 * Redistribution and use in source and binary forms, with or without
38 * modification, are permitted provided that the following conditions
39 * are met:
40 * 1. Redistributions of source code must retain the above copyright
41 * notice, this list of conditions and the following disclaimer.
42 * 2. Redistributions in binary form must reproduce the above copyright
43 * notice, this list of conditions and the following disclaimer in the
44 * documentation and/or other materials provided with the distribution.
45 * 3. All advertising materials mentioning features or use of this software
46 * must display the following acknowledgement:
47 * This product includes software developed by the University of
48 * California, Berkeley and its contributors.
49 * 4. Neither the name of the University nor the names of its contributors
50 * may be used to endorse or promote products derived from this software
51 * without specific prior written permission.
52 *
53 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
54 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
55 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
56 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
57 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
58 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
59 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
60 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
61 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
62 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
63 * SUCH DAMAGE.
64 *
65 * @(#)init_main.c 8.16 (Berkeley) 5/14/95
66 */
67
68 /*
69 *
70 * Mach Operating System
71 * Copyright (c) 1987 Carnegie-Mellon University
72 * All rights reserved. The CMU software License Agreement specifies
73 * the terms and conditions for use and redistribution.
74 */
75 /*
76 * NOTICE: This file was modified by McAfee Research in 2004 to introduce
77 * support for mandatory and extensible security protections. This notice
78 * is included in support of clause 2.2 (b) of the Apple Public License,
79 * Version 2.0.
80 */
81
82 #include <sys/param.h>
83 #include <sys/filedesc.h>
84 #include <sys/kernel.h>
85 #include <sys/mount_internal.h>
86 #include <sys/proc_internal.h>
87 #include <sys/kauth.h>
88 #include <sys/systm.h>
89 #include <sys/vnode_internal.h>
90 #include <sys/conf.h>
91 #include <sys/buf_internal.h>
92 #include <sys/user.h>
93 #include <sys/time.h>
94 #include <sys/systm.h>
95 #include <sys/mman.h>
96 #include <sys/kasl.h>
97
98 #include <security/audit/audit.h>
99
100 #include <sys/malloc.h>
101 #include <sys/dkstat.h>
102 #include <sys/codesign.h>
103
104 #include <kern/startup.h>
105 #include <kern/thread.h>
106 #include <kern/task.h>
107 #include <kern/ast.h>
108 #include <kern/zalloc.h>
109 #include <kern/ux_handler.h> /* for ux_handler_setup() */
110 #include <kern/sched_hygiene.h>
111
112 #include <mach/vm_param.h>
113
114 #include <vm/vm_map.h>
115 #include <vm/vm_kern.h>
116
117 #include <sys/reboot.h>
118 #include <dev/busvar.h> /* for pseudo_inits */
119 #include <sys/kdebug.h>
120 #include <sys/monotonic.h>
121
122 #include <mach/mach_types.h>
123 #include <mach/vm_prot.h>
124 #include <mach/semaphore.h>
125 #include <mach/sync_policy.h>
126 #include <kern/clock.h>
127 #include <sys/csr.h>
128 #include <mach/kern_return.h>
129 #include <mach/thread_act.h> /* for thread_resume() */
130 #include <sys/mcache.h> /* for mcache_init() */
131 #include <sys/mbuf.h> /* for mbinit() */
132 #include <sys/event.h> /* for knote_init() */
133 #include <sys/eventhandler.h> /* for eventhandler_init() */
134 #include <sys/kern_memorystatus.h> /* for memorystatus_init() */
135 #include <sys/kern_memorystatus_freeze.h> /* for memorystatus_freeze_init() */
136 #include <sys/aio_kern.h> /* for aio_init() */
137 #include <sys/semaphore.h> /* for psem_cache_init() */
138 #include <net/dlil.h> /* for dlil_init() */
139 #include <net/iptap.h> /* for iptap_init() */
140 #include <sys/socketvar.h> /* for socketinit() */
141 #include <sys/protosw.h> /* for domaininit() */
142 #include <kern/sched_prim.h> /* for thread_wakeup() */
143 #include <net/if_ether.h> /* for ether_family_init() */
144 #include <net/if_gif.h> /* for gif_init() */
145 #include <miscfs/devfs/devfsdefs.h> /* for devfs_kernel_mount() */
146 #include <vm/vm_kern.h> /* for kmem_suballoc() */
147 #include <sys/proc_uuid_policy.h> /* proc_uuid_policy_init() */
148 #include <netinet/flow_divert.h> /* flow_divert_init() */
149 #include <net/content_filter.h> /* for cfil_init() */
150 #include <net/necp.h> /* for necp_init() */
151 #include <net/network_agent.h> /* for netagent_init() */
152 #include <net/packet_mangler.h> /* for pkt_mnglr_init() */
153 #include <net/if_utun.h> /* for utun_register_control() */
154 #include <net/if_ipsec.h> /* for ipsec_register_control() */
155 #include <net/netsrc.h> /* for netsrc_init() */
156 #include <net/ntstat.h> /* for nstat_init() */
157 #include <netinet/tcp_cc.h> /* for tcp_cc_init() */
158 #include <netinet/mptcp_var.h> /* for mptcp_control_register() */
159 #include <net/nwk_wq.h> /* for nwk_wq_init */
160 #include <net/restricted_in_port.h> /* for restricted_in_port_init() */
161 #include <kern/assert.h> /* for assert() */
162 #include <sys/kern_overrides.h> /* for init_system_override() */
163 #include <sys/lockf.h> /* for lf_init() */
164 #include <sys/fsctl.h>
165
166 #include <net/init.h>
167
168 #if CONFIG_MACF
169 #include <security/mac_framework.h>
170 #include <security/mac_internal.h> /* mac_init_bsd() */
171 #include <security/mac_mach_internal.h> /* mac_update_task_label() */
172 #endif
173
174 #include <machine/exec.h>
175
176 #if CONFIG_NETBOOT
177 #include <sys/netboot.h>
178 #endif
179
180 #if CONFIG_IMAGEBOOT
181 #include <sys/imageboot.h>
182 #endif
183
184 #if PFLOG
185 #include <net/if_pflog.h>
186 #endif
187
188 #if SKYWALK
189 #include <skywalk/os_skywalk_private.h>
190 #endif /* SKYWALK */
191
192 #include <pexpert/pexpert.h>
193 #include <machine/pal_routines.h>
194 #include <console/video_console.h>
195
196 #if CONFIG_XNUPOST
197 #include <tests/xnupost.h>
198 #endif
199
200 void * get_user_regs(thread_t); /* XXX kludge for <machine/thread.h> */
201 void IOKitInitializeTime(void); /* XXX */
202 void IOSleep(unsigned int); /* XXX */
203 void IOSetImageBoot(void); /* XXX */
204 void loopattach(void); /* XXX */
205
206 const char *const copyright =
207 "Copyright (c) 1982, 1986, 1989, 1991, 1993\n\t"
208 "The Regents of the University of California. "
209 "All rights reserved.\n\n";
210
211 /* Components of the first process -- never freed. */
212 SECURITY_READ_ONLY_LATE(struct vfs_context) vfs_context0;
213
214 struct proc proc0 = {
215 .p_comm = "kernel_task",
216 .p_name = "kernel_task",
217 .p_pptr = &proc0,
218 .p_stat = SRUN,
219 #if defined(__LP64__)
220 .p_flag = P_SYSTEM | P_LP64,
221 #else
222 .p_flag = P_SYSTEM,
223 #endif
224 .p_nice = NZERO,
225 .p_uthlist = TAILQ_HEAD_INITIALIZER(proc0.p_uthlist),
226 };
227 static struct plimit limit0;
228 static struct pstats pstats0;
229 SECURITY_READ_ONLY_LATE(proc_t) kernproc = &proc0;
230 proc_t XNU_PTRAUTH_SIGNED_PTR("initproc") initproc;
231
232 long tk_cancc;
233 long tk_nin;
234 long tk_nout;
235 long tk_rawcc;
236
237 int lock_trace = 0;
238 /* Global variables to make pstat happy. We do swapping differently */
239 int nswdev, nswap;
240 int nswapmap;
241 void *swapmap;
242 struct swdevt swdevt[1];
243
244 static LCK_GRP_DECLARE(hostname_lck_grp, "hostname");
245 LCK_MTX_DECLARE(hostname_lock, &hostname_lck_grp);
246 LCK_MTX_DECLARE(domainname_lock, &hostname_lck_grp);
247
248 dev_t rootdev; /* device of the root */
249 dev_t dumpdev; /* device to take dumps on */
250 long dumplo; /* offset into dumpdev */
251 long hostid;
252 char hostname[MAXHOSTNAMELEN];
253 char domainname[MAXDOMNAMELEN];
254 char rootdevice[DEVMAXNAMESIZE];
255
256 struct vnode *rootvp;
257 bool rootvp_is_ssd = false;
258 SECURITY_READ_ONLY_LATE(int) boothowto;
259 /*
260 * -minimalboot indicates that we want userspace to be bootstrapped to a
261 * minimal environment. What constitutes minimal is up to the bootstrap
262 * process.
263 */
264 TUNABLE(int, minimalboot, "-minimalboot", 0);
265 #if CONFIG_DARKBOOT
266 int darkboot = 0;
267 #endif
268
269 extern kern_return_t IOFindBSDRoot(char *, unsigned int, dev_t *, u_int32_t *);
270 extern void IOSecureBSDRoot(const char * rootName);
271 extern kern_return_t IOKitBSDInit(void );
272 extern boolean_t IOSetRecoveryBoot(bsd_bootfail_mode_t, uuid_t, boolean_t);
273 extern void kminit(void);
274 extern void bsd_bufferinit(void);
275 extern void throttle_init(void);
276
277 vm_map_t bsd_pageable_map;
278 vm_map_t mb_map;
279
280 static int bsd_simul_execs;
281 static int bsd_pageable_map_size;
282 __private_extern__ int execargs_cache_size = 0;
283 __private_extern__ int execargs_free_count = 0;
284 __private_extern__ vm_offset_t * execargs_cache = NULL;
285
286 void bsd_exec_setup(int);
287
288 __private_extern__ int bootarg_execfailurereports = 0;
289
290 #if __x86_64__
291 __private_extern__ TUNABLE(int, bootarg_no32exec, "no32exec", 1);
292 #endif
293
294 #if DEVELOPMENT || DEBUG
295 /* Prevent kernel-based ASLR from being used. */
296 __private_extern__ TUNABLE(bool, bootarg_disable_aslr, "-disable_aslr", 0);
297 #endif
298
299 /*
300 * Allow an alternate dyld to be used for testing.
301 */
302
303 #if DEVELOPMENT || DEBUG
304 char dyld_alt_path[MAXPATHLEN];
305 int use_alt_dyld = 0;
306 #endif
307
308 int cmask = CMASK;
309 extern int customnbuf;
310
311 kern_return_t bsd_autoconf(void);
312 void bsd_utaskbootstrap(void);
313
314 #if CONFIG_DEV_KMEM
315 extern void dev_kmem_init(void);
316 #endif
317 extern void select_waitq_init(void);
318 static void process_name(const char *, proc_t);
319
320 static void setconf(void);
321
322 #if CONFIG_BASESYSTEMROOT
323 static int bsd_find_basesystem_dmg(char *bsdmgpath_out, bool *rooted_dmg, bool *skip_signature_check);
324 static boolean_t bsdmgroot_bootable(void);
325 #endif // CONFIG_BASESYSTEMROOT
326
327 static bool bsd_rooted_ramdisk(void);
328
329 #if SYSV_SHM
330 extern void sysv_shm_lock_init(void);
331 #endif
332 #if SYSV_SEM
333 extern void sysv_sem_lock_init(void);
334 #endif
335 #if SYSV_MSG
336 extern void sysv_msg_lock_init(void);
337 #endif
338
339 #if CONFIG_MACF
340 #if defined (__i386__) || defined (__x86_64__)
341 /* MACF policy_check configuration flags; see policy_check.c for details */
342 extern int check_policy_init(int);
343 #endif
344 #endif /* CONFIG_MACF */
345
346 /* If we are using CONFIG_DTRACE */
347 #if CONFIG_DTRACE
348 extern void dtrace_postinit(void);
349 #endif
350
351 /*
352 * Initialization code.
353 * Called from cold start routine as
354 * soon as a stack and segmentation
355 * have been established.
356 * Functions:
357 * turn on clock
358 * hand craft 0th process
359 * call all initialization routines
360 * hand craft 1st user process
361 */
362
363 /*
364 * Sets the name for the given task.
365 */
366 static void
process_name(const char * s,proc_t p)367 process_name(const char *s, proc_t p)
368 {
369 strlcpy(p->p_comm, s, sizeof(p->p_comm));
370 strlcpy(p->p_name, s, sizeof(p->p_name));
371 }
372
373 /* To allow these values to be patched, they're globals here */
374 #include <machine/vmparam.h>
375 struct rlimit vm_initial_limit_stack = { .rlim_cur = DFLSSIZ, .rlim_max = MAXSSIZ - PAGE_MAX_SIZE };
376 struct rlimit vm_initial_limit_data = { .rlim_cur = DFLDSIZ, .rlim_max = MAXDSIZ };
377 struct rlimit vm_initial_limit_core = { .rlim_cur = DFLCSIZ, .rlim_max = MAXCSIZ };
378
379 extern struct os_refgrp rlimit_refgrp;
380
381 extern thread_t cloneproc(task_t, coalition_t, proc_t, int, int);
382 extern int (*mountroot)(void);
383
384 LCK_ATTR_DECLARE(proc_lck_attr, 0, 0);
385 LCK_GRP_DECLARE(proc_lck_grp, "proc");
386 LCK_GRP_DECLARE(proc_slock_grp, "proc-slock");
387 LCK_GRP_DECLARE(proc_fdmlock_grp, "proc-fdmlock");
388 LCK_GRP_DECLARE(proc_mlock_grp, "proc-mlock");
389 LCK_GRP_DECLARE(proc_ucred_mlock_grp, "proc-ucred-mlock");
390 LCK_GRP_DECLARE(proc_dirslock_grp, "proc-dirslock");
391 LCK_GRP_DECLARE(proc_kqhashlock_grp, "proc-kqhashlock");
392 LCK_GRP_DECLARE(proc_knhashlock_grp, "proc-knhashlock");
393
394
395 LCK_MTX_DECLARE_ATTR(proc_list_mlock, &proc_mlock_grp, &proc_lck_attr);
396
397 #if XNU_TARGET_OS_OSX
398 /* hook called after root is mounted XXX temporary hack */
399 void (*mountroot_post_hook)(void);
400 void (*unmountroot_pre_hook)(void);
401 #endif
402 void set_rootvnode(vnode_t);
403
404 extern lck_rw_t rootvnode_rw_lock;
405
406 /* called with an iocount and usecount on new_rootvnode */
407 void
set_rootvnode(vnode_t new_rootvnode)408 set_rootvnode(vnode_t new_rootvnode)
409 {
410 mount_t new_mount = (new_rootvnode != NULL) ? new_rootvnode->v_mount : NULL;
411 vnode_t new_devvp = (new_mount != NULL) ? new_mount->mnt_devvp : NULL;
412 vnode_t old_rootvnode = rootvnode;
413
414 new_rootvnode->v_flag |= VROOT;
415 rootvp = new_devvp;
416 rootvnode = new_rootvnode;
417 kernproc->p_fd.fd_cdir = new_rootvnode;
418 if (new_devvp != NULL) {
419 rootdev = vnode_specrdev(new_devvp);
420 } else if (new_mount != NULL) {
421 rootdev = vfs_statfs(new_mount)->f_fsid.val[0]; /* like ATTR_CMN_DEVID */
422 } else {
423 rootdev = NODEV;
424 }
425
426 if (old_rootvnode) {
427 vnode_rele(old_rootvnode);
428 }
429 }
430
431 #define RAMDEV "md0"
432
433 bool
bsd_rooted_ramdisk(void)434 bsd_rooted_ramdisk(void)
435 {
436 bool is_ramdisk = false;
437 char *dev_path = zalloc(ZV_NAMEI);
438 if (dev_path == NULL) {
439 panic("failed to allocate devpath string!");
440 }
441
442 if (PE_parse_boot_argn("rd", dev_path, MAXPATHLEN)) {
443 if (strncmp(dev_path, RAMDEV, strlen(RAMDEV)) == 0) {
444 is_ramdisk = true;
445 }
446 }
447
448 zfree(ZV_NAMEI, dev_path);
449 return is_ramdisk;
450 }
451
452 /*
453 * This function is called very early on in the Mach startup, from the
454 * function start_kernel_threads() in osfmk/kern/startup.c. It's called
455 * in the context of the current (startup) task using a call to the
456 * function kernel_thread_create() to jump into start_kernel_threads().
457 * Internally, kernel_thread_create() calls thread_create_internal(),
458 * which calls uthread_init(). The function of uthread_init() is
459 * normally to init a uthread structure, and fill out the uu_sigmask,
460 * tro_ucred/tro_proc fields. It skips filling these out in the case of the "task"
461 * being "kernel_task", because the order of operation is inverted. To
462 * account for that, we need to manually fill in at least the contents
463 * of the tro_ucred field so that the uthread structure can be
464 * used like any other.
465 */
466
467 void
bsd_init(void)468 bsd_init(void)
469 {
470 struct uthread *ut;
471 kern_return_t ret;
472 vnode_t init_rootvnode = NULLVP;
473 struct proc_ro_data kernproc_ro_data = {
474 .p_csflags = CS_VALID,
475 };
476 struct task_ro_data kerntask_ro_data = { };
477 #if CONFIG_NETBOOT || CONFIG_IMAGEBOOT
478 boolean_t netboot = FALSE;
479 #endif
480
481 #define DEBUG_BSDINIT 0
482
483 #if DEBUG_BSDINIT
484 #define bsd_init_kprintf(x, ...) kprintf("bsd_init: " x, ## __VA_ARGS__)
485 #else
486 #define bsd_init_kprintf(x, ...)
487 #endif
488
489 throttle_init();
490
491 printf(copyright);
492
493 #if CONFIG_DEV_KMEM
494 bsd_init_kprintf("calling dev_kmem_init\n");
495 dev_kmem_init();
496 #endif
497
498 /* Initialize kauth subsystem before instancing the first credential */
499 bsd_init_kprintf("calling kauth_init\n");
500 kauth_init();
501
502 /* kernel_task->proc = kernproc; */
503 set_bsdtask_info(kernel_task, (void *)kernproc);
504
505 /* set the cred */
506 kauth_cred_set(&kernproc_ro_data.p_ucred, vfs_context0.vc_ucred);
507 kernproc->p_proc_ro = proc_ro_alloc(kernproc, &kernproc_ro_data,
508 kernel_task, &kerntask_ro_data);
509
510 /* give kernproc a name */
511 bsd_init_kprintf("calling process_name\n");
512 process_name("kernel_task", kernproc);
513
514 /* Allocate proc lock attribute */
515
516 lck_mtx_init(&kernproc->p_mlock, &proc_mlock_grp, &proc_lck_attr);
517 lck_mtx_init(&kernproc->p_ucred_mlock, &proc_ucred_mlock_grp, &proc_lck_attr);
518 lck_spin_init(&kernproc->p_slock, &proc_slock_grp, &proc_lck_attr);
519
520 /* Init the file descriptor table. */
521 fdt_init(kernproc);
522 kernproc->p_fd.fd_cmask = (mode_t)cmask;
523
524 assert(bsd_simul_execs != 0);
525 execargs_cache_size = bsd_simul_execs;
526 execargs_free_count = bsd_simul_execs;
527 execargs_cache = zalloc_permanent(bsd_simul_execs * sizeof(vm_offset_t),
528 ZALIGN(vm_offset_t));
529
530 if (current_task() != kernel_task) {
531 printf("bsd_init: We have a problem, "
532 "current task is not kernel task\n");
533 }
534
535 bsd_init_kprintf("calling get_bsdthread_info\n");
536 ut = current_uthread();
537
538 #if CONFIG_MACF
539 /*
540 * Initialize the MAC Framework
541 */
542 mac_policy_initbsd();
543
544 #if defined (__i386__) || defined (__x86_64__)
545 /*
546 * We currently only support this on i386/x86_64, as that is the
547 * only lock code we have instrumented so far.
548 */
549 int policy_check_flags;
550 PE_parse_boot_argn("policy_check", &policy_check_flags, sizeof(policy_check_flags));
551 check_policy_init(policy_check_flags);
552 #endif
553 #endif /* MAC */
554
555 /*
556 * Make a session and group
557 *
558 * No need to hold the pgrp lock,
559 * there are no other BSD threads yet.
560 */
561 struct session *session0 = session_alloc(kernproc);
562 struct pgrp *pgrp0 = pgrp_alloc(0, PGRP_REF_NONE);
563 session0->s_ttypgrpid = 0;
564 pgrp0->pg_session = session0;
565
566 /*
567 * Create process 0.
568 */
569 proc_list_lock();
570 os_ref_init_mask(&kernproc->p_refcount, P_REF_BITS, &p_refgrp, P_REF_NONE);
571 os_ref_init_raw(&kernproc->p_waitref, &p_refgrp);
572
573 /*
574 * Make a group and session, then simulate pinsertchild(),
575 * adjusted for the kernel.
576 */
577 pghash_insert_locked(0, pgrp0);
578
579 LIST_INSERT_HEAD(&pgrp0->pg_members, kernproc, p_pglist);
580 hazard_ptr_init(&kernproc->p_pgrp, pgrp0);
581 LIST_INSERT_HEAD(&allproc, kernproc, p_list);
582
583 LIST_INSERT_HEAD(SESSHASH(0), session0, s_hash);
584 proc_list_unlock();
585
586 kernproc->task = kernel_task;
587
588 #if DEVELOPMENT || DEBUG
589 if (bootarg_disable_aslr) {
590 kernproc->p_flag |= P_DISABLE_ASLR;
591 }
592 #endif
593
594 TAILQ_INSERT_TAIL(&kernproc->p_uthlist, ut, uu_list);
595
596 bsd_init_kprintf("calling kauth_cred_create\n");
597 /*
598 * Officially associate the kernel with vfs_context0.vc_ucred.
599 */
600 #if CONFIG_MACF
601 mac_cred_label_associate_kernel(vfs_context0.vc_ucred);
602 #endif
603 proc_update_creds_onproc(kernproc);
604
605 TAILQ_INIT(&kernproc->p_aio_activeq);
606 TAILQ_INIT(&kernproc->p_aio_doneq);
607 kernproc->p_aio_total_count = 0;
608
609 /* Create the limits structures. */
610 for (uint32_t i = 0; i < ARRAY_COUNT(limit0.pl_rlimit); i++) {
611 limit0.pl_rlimit[i].rlim_cur =
612 limit0.pl_rlimit[i].rlim_max = RLIM_INFINITY;
613 }
614 limit0.pl_rlimit[RLIMIT_NOFILE].rlim_cur = NOFILE;
615 limit0.pl_rlimit[RLIMIT_NPROC].rlim_cur = maxprocperuid;
616 limit0.pl_rlimit[RLIMIT_NPROC].rlim_max = maxproc;
617 limit0.pl_rlimit[RLIMIT_STACK] = vm_initial_limit_stack;
618 limit0.pl_rlimit[RLIMIT_DATA] = vm_initial_limit_data;
619 limit0.pl_rlimit[RLIMIT_CORE] = vm_initial_limit_core;
620 os_ref_init_count(&limit0.pl_refcnt, &rlimit_refgrp, 1);
621
622 hazard_ptr_init(&kernproc->p_limit, &limit0);
623 kernproc->p_stats = &pstats0;
624 proc_sigacts_copy(kernproc, NULL);
625 kernproc->p_subsystem_root_path = NULL;
626
627 /*
628 * Charge root for one process: launchd.
629 */
630 bsd_init_kprintf("calling chgproccnt\n");
631 (void)chgproccnt(0, 1);
632
633 /*
634 * Allocate a kernel submap for pageable memory
635 * for temporary copying (execve()).
636 */
637 {
638 vm_offset_t minimum;
639
640 bsd_init_kprintf("calling kmem_suballoc\n");
641 assert(bsd_pageable_map_size != 0);
642 ret = kmem_suballoc(kernel_map,
643 &minimum,
644 (vm_size_t)bsd_pageable_map_size,
645 TRUE,
646 VM_FLAGS_ANYWHERE,
647 VM_MAP_KERNEL_FLAGS_NONE,
648 VM_KERN_MEMORY_BSD,
649 &bsd_pageable_map);
650 if (ret != KERN_SUCCESS) {
651 panic("bsd_init: Failed to allocate bsd pageable map");
652 }
653 }
654
655 /*
656 * Initialize buffers and hash links for buffers
657 *
658 * SIDE EFFECT: Starts a thread for bcleanbuf_thread(), so must
659 * happen after a credential has been associated with
660 * the kernel task.
661 */
662 bsd_init_kprintf("calling bsd_bufferinit\n");
663 bsd_bufferinit();
664
665 /*
666 * Initialize the calendar.
667 */
668 bsd_init_kprintf("calling IOKitInitializeTime\n");
669 IOKitInitializeTime();
670
671 /* Initialize the file systems. */
672 bsd_init_kprintf("calling vfsinit\n");
673 vfsinit();
674
675 #if CONFIG_PROC_UUID_POLICY
676 /* Initial proc_uuid_policy subsystem */
677 bsd_init_kprintf("calling proc_uuid_policy_init()\n");
678 proc_uuid_policy_init();
679 #endif
680
681 #if SOCKETS
682 /* Initialize per-CPU cache allocator */
683 mcache_init();
684
685 /* Initialize mbuf's. */
686 bsd_init_kprintf("calling mbinit\n");
687 mbinit();
688 restricted_in_port_init();
689 #endif /* SOCKETS */
690
691 /*
692 * Initializes security event auditing.
693 * XXX: Should/could this occur later?
694 */
695 #if CONFIG_AUDIT
696 bsd_init_kprintf("calling audit_init\n");
697 audit_init();
698 #endif
699
700 /* Initialize kqueues */
701 bsd_init_kprintf("calling knote_init\n");
702 knote_init();
703
704 /* Initialize event handler */
705 bsd_init_kprintf("calling eventhandler_init\n");
706 eventhandler_init();
707
708 /* Initialize for async IO */
709 bsd_init_kprintf("calling aio_init\n");
710 aio_init();
711
712 pthread_init();
713 /* POSIX Shm and Sem */
714 bsd_init_kprintf("calling pshm_cache_init\n");
715 pshm_cache_init();
716 bsd_init_kprintf("calling psem_cache_init\n");
717 psem_cache_init();
718 bsd_init_kprintf("calling select_waitq_init\n");
719 select_waitq_init();
720
721 /*
722 * Initialize protocols. Block reception of incoming packets
723 * until everything is ready.
724 */
725 #if NETWORKING
726 bsd_init_kprintf("calling nwk_wq_init\n");
727 nwk_wq_init();
728 bsd_init_kprintf("calling dlil_init\n");
729 dlil_init();
730 #endif /* NETWORKING */
731 #if SOCKETS
732 bsd_init_kprintf("calling socketinit\n");
733 socketinit();
734 bsd_init_kprintf("calling domaininit\n");
735 domaininit();
736 iptap_init();
737 #if FLOW_DIVERT
738 flow_divert_init();
739 #endif /* FLOW_DIVERT */
740 #endif /* SOCKETS */
741 #if SKYWALK
742 bsd_init_kprintf("calling skywalk_init\n");
743 (void) skywalk_init();
744 #endif /* SKYWALK */
745 #if NETWORKING
746 #if NECP
747 /* Initialize Network Extension Control Policies */
748 necp_init();
749 #endif
750 netagent_init();
751 #endif /* NETWORKING */
752
753 #if defined (__x86_64__)
754 hvg_bsd_init();
755 #endif /* DEBUG || DEVELOPMENT */
756
757 #if CONFIG_FREEZE
758 #ifndef CONFIG_MEMORYSTATUS
759 #error "CONFIG_FREEZE defined without matching CONFIG_MEMORYSTATUS"
760 #endif
761 /* Initialise background freezing */
762 bsd_init_kprintf("calling memorystatus_freeze_init\n");
763 memorystatus_freeze_init();
764 #endif
765
766 #if CONFIG_MEMORYSTATUS
767 /* Initialize kernel memory status notifications */
768 bsd_init_kprintf("calling memorystatus_init\n");
769 memorystatus_init();
770 #endif /* CONFIG_MEMORYSTATUS */
771
772 bsd_init_kprintf("calling sysctl_mib_init\n");
773 sysctl_mib_init();
774
775 bsd_init_kprintf("calling bsd_autoconf\n");
776 bsd_autoconf();
777
778 #if CONFIG_DTRACE
779 dtrace_postinit();
780 #endif
781
782 /*
783 * We attach the loopback interface *way* down here to ensure
784 * it happens after autoconf(), otherwise it becomes the
785 * "primary" interface.
786 */
787 #include <loop.h>
788 #if NLOOP > 0
789 bsd_init_kprintf("calling loopattach\n");
790 loopattach(); /* XXX */
791 #endif
792 #if NGIF
793 /* Initialize gif interface (after lo0) */
794 gif_init();
795 #endif
796
797 #if PFLOG
798 /* Initialize packet filter log interface */
799 pfloginit();
800 #endif /* PFLOG */
801
802 #if NETHER > 0
803 /* Register the built-in dlil ethernet interface family */
804 bsd_init_kprintf("calling ether_family_init\n");
805 ether_family_init();
806 #endif /* ETHER */
807
808 #if NETWORKING
809 #if CONTENT_FILTER
810 cfil_init();
811 #endif
812
813 #if PACKET_MANGLER
814 pkt_mnglr_init();
815 #endif
816
817 /*
818 * Register subsystems with kernel control handlers
819 */
820 utun_register_control();
821 #if IPSEC
822 ipsec_register_control();
823 #endif /* IPSEC */
824 netsrc_init();
825 nstat_init();
826 tcp_cc_init();
827 #if MPTCP
828 mptcp_control_register();
829 #endif /* MPTCP */
830
831 /*
832 * The the networking stack is now initialized so it is a good time to call
833 * the clients that are waiting for the networking stack to be usable.
834 */
835 bsd_init_kprintf("calling net_init_run\n");
836 net_init_run();
837 #endif /* NETWORKING */
838
839 bsd_init_kprintf("calling inittodr\n");
840 inittodr(0);
841
842 /* Mount the root file system. */
843 while (TRUE) {
844 int err;
845
846 bsd_init_kprintf("calling setconf\n");
847 setconf();
848 #if CONFIG_NETBOOT
849 netboot = (mountroot == netboot_mountroot);
850 #endif
851
852 bsd_init_kprintf("vfs_mountroot\n");
853 if (0 == (err = vfs_mountroot())) {
854 break;
855 }
856 rootdevice[0] = '\0';
857 #if CONFIG_NETBOOT
858 if (netboot) {
859 PE_display_icon( 0, "noroot"); /* XXX a netboot-specific icon would be nicer */
860 vc_progress_set(FALSE, 0);
861 for (uint32_t i = 1; 1; i *= 2) {
862 printf("bsd_init: failed to mount network root, error %d, %s\n",
863 err, PE_boot_args());
864 printf("We are hanging here...\n");
865 IOSleep(i * 60 * 1000);
866 }
867 /*NOTREACHED*/
868 }
869 #endif
870 printf("cannot mount root, errno = %d\n", err);
871 }
872
873 IOSecureBSDRoot(rootdevice);
874
875 mountlist.tqh_first->mnt_flag |= MNT_ROOTFS;
876
877 bsd_init_kprintf("calling VFS_ROOT\n");
878 /* Get the vnode for '/'. Set fdp->fd_fd.fd_cdir to reference it. */
879 if (VFS_ROOT(mountlist.tqh_first, &init_rootvnode, vfs_context_kernel())) {
880 panic("bsd_init: cannot find root vnode: %s", PE_boot_args());
881 }
882 (void)vnode_ref(init_rootvnode);
883 (void)vnode_put(init_rootvnode);
884
885 lck_rw_lock_exclusive(&rootvnode_rw_lock);
886 set_rootvnode(init_rootvnode);
887 lck_rw_unlock_exclusive(&rootvnode_rw_lock);
888 init_rootvnode = NULLVP; /* use rootvnode after this point */
889
890
891 if (!bsd_rooted_ramdisk()) {
892 boolean_t require_rootauth = FALSE;
893
894 #if XNU_TARGET_OS_OSX && defined(__arm64__)
895 #if CONFIG_IMAGEBOOT
896 /* Apple Silicon MacOS */
897 require_rootauth = !imageboot_desired();
898 #endif // CONFIG_IMAGEBOOT
899 #elif !XNU_TARGET_OS_OSX
900 /* Non MacOS */
901 require_rootauth = TRUE;
902 #endif // XNU_TARGET_OS_OSX && defined(__arm64__)
903
904 if (require_rootauth) {
905 /* enforce sealedness */
906 int autherr = VNOP_IOCTL(rootvnode, FSIOC_KERNEL_ROOTAUTH, NULL, 0, vfs_context_kernel());
907 if (autherr) {
908 panic("rootvp not authenticated after mounting");
909 }
910 }
911 }
912
913
914 #if CONFIG_NETBOOT
915 if (netboot) {
916 int err;
917
918 netboot = TRUE;
919 /* post mount setup */
920 if ((err = netboot_setup()) != 0) {
921 PE_display_icon( 0, "noroot"); /* XXX a netboot-specific icon would be nicer */
922 vc_progress_set(FALSE, 0);
923 for (uint32_t i = 1; 1; i *= 2) {
924 printf("bsd_init: NetBoot could not find root, error %d: %s\n",
925 err, PE_boot_args());
926 printf("We are hanging here...\n");
927 IOSleep(i * 60 * 1000);
928 }
929 /*NOTREACHED*/
930 }
931 }
932 #endif
933
934
935 #if CONFIG_IMAGEBOOT
936 /*
937 * See if a system disk image is present. If so, mount it and
938 * switch the root vnode to point to it
939 */
940 imageboot_type_t imageboot_type = imageboot_needed();
941 if (netboot == FALSE && imageboot_type) {
942 /*
943 * An image was found. No turning back: we're booted
944 * with a kernel from the disk image.
945 */
946 bsd_init_kprintf("doing image boot: type = %d\n", imageboot_type);
947 imageboot_setup(imageboot_type);
948 IOSetImageBoot();
949 }
950
951 #endif /* CONFIG_IMAGEBOOT */
952
953 /* set initial time; all other resource data is already zero'ed */
954 microtime_with_abstime(&kernproc->p_start, &kernproc->p_stats->ps_start);
955
956 #if DEVFS
957 {
958 char mounthere[] = "/dev"; /* !const because of internal casting */
959
960 bsd_init_kprintf("calling devfs_kernel_mount\n");
961 devfs_kernel_mount(mounthere);
962 }
963 #endif /* DEVFS */
964
965 #if CONFIG_BASESYSTEMROOT
966 #if CONFIG_IMAGEBOOT
967 if (bsdmgroot_bootable()) {
968 int error;
969 bool rooted_dmg = false;
970 bool skip_signature_check = false;
971
972 printf("trying to find and mount BaseSystem dmg as root volume\n");
973 #if DEVELOPMENT || DEBUG
974 printf("(set boot-arg -nobsdmgroot to avoid this)\n");
975 #endif // DEVELOPMENT || DEBUG
976
977 char *dmgpath = NULL;
978 dmgpath = zalloc_flags(ZV_NAMEI, Z_ZERO | Z_WAITOK | Z_NOFAIL);
979
980 error = bsd_find_basesystem_dmg(dmgpath, &rooted_dmg, &skip_signature_check);
981 if (error) {
982 bsd_init_kprintf("failed to to find BaseSystem dmg: error = %d\n", error);
983 } else {
984 PE_parse_boot_argn("bsdmgpath", dmgpath, sizeof(dmgpath));
985
986 bsd_init_kprintf("found BaseSystem dmg at: %s\n", dmgpath);
987
988 error = imageboot_pivot_image(dmgpath, IMAGEBOOT_DMG, "/System/Volumes/BaseSystem", "System/Volumes/macOS", rooted_dmg, skip_signature_check);
989 if (error) {
990 bsd_init_kprintf("couldn't mount BaseSystem dmg: error = %d", error);
991 } else {
992 IOSetImageBoot();
993 }
994 }
995 zfree(ZV_NAMEI, dmgpath);
996 }
997 #else /* CONFIG_IMAGEBOOT */
998 #error CONFIG_BASESYSTEMROOT requires CONFIG_IMAGEBOOT
999 #endif /* CONFIG_IMAGEBOOT */
1000 #endif /* CONFIG_BASESYSTEMROOT */
1001
1002 /* Initialize signal state for process 0. */
1003 bsd_init_kprintf("calling siginit\n");
1004 siginit(kernproc);
1005
1006 bsd_init_kprintf("calling bsd_utaskbootstrap\n");
1007 bsd_utaskbootstrap();
1008
1009 pal_kernel_announce();
1010
1011 bsd_init_kprintf("calling mountroot_post_hook\n");
1012
1013 #if XNU_TARGET_OS_OSX
1014 /* invoke post-root-mount hook */
1015 if (mountroot_post_hook != NULL) {
1016 mountroot_post_hook();
1017 }
1018 #endif
1019
1020 #if 0 /* not yet */
1021 consider_zone_gc(FALSE);
1022 #endif
1023
1024 /*
1025 * At this point, we consider the kernel "booted" enough to apply
1026 * stricter timeouts.
1027 */
1028 machine_timeout_bsd_init();
1029
1030 bsd_init_kprintf("done\n");
1031 }
1032
1033 void
bsdinit_task(void)1034 bsdinit_task(void)
1035 {
1036 proc_t p = current_proc();
1037
1038 process_name("init", p);
1039
1040 /* Set up exception-to-signal reflection */
1041 ux_handler_setup();
1042
1043 #if CONFIG_MACF
1044 mac_cred_label_associate_user(proc_ucred(p));
1045 #endif
1046
1047 vm_init_before_launchd();
1048
1049 #if CONFIG_XNUPOST
1050 int result = bsd_list_tests();
1051 result = bsd_do_post();
1052 if (result != 0) {
1053 panic("bsd_do_post: Tests failed with result = 0x%08x", result);
1054 }
1055 #endif
1056
1057 bsd_init_kprintf("bsd_do_post - done");
1058
1059 load_init_program(p);
1060 lock_trace = 1;
1061 }
1062
1063 kern_return_t
bsd_autoconf(void)1064 bsd_autoconf(void)
1065 {
1066 kprintf("bsd_autoconf: calling kminit\n");
1067 kminit();
1068
1069 /*
1070 * Early startup for bsd pseudodevices.
1071 */
1072 {
1073 struct pseudo_init *pi;
1074
1075 for (pi = pseudo_inits; pi->ps_func; pi++) {
1076 (*pi->ps_func)(pi->ps_count);
1077 }
1078 }
1079
1080 return IOKitBSDInit();
1081 }
1082
1083
1084 #include <sys/disklabel.h> /* for MAXPARTITIONS */
1085
1086 static void
setconf(void)1087 setconf(void)
1088 {
1089 u_int32_t flags;
1090 kern_return_t err;
1091
1092 err = IOFindBSDRoot(rootdevice, sizeof(rootdevice), &rootdev, &flags);
1093 if (err) {
1094 printf("setconf: IOFindBSDRoot returned an error (%d);"
1095 "setting rootdevice to 'sd0a'.\n", err); /* XXX DEBUG TEMP */
1096 rootdev = makedev( 6, 0 );
1097 strlcpy(rootdevice, "sd0a", sizeof(rootdevice));
1098 flags = 0;
1099 }
1100
1101 #if CONFIG_NETBOOT
1102 if (flags & 1) {
1103 /* network device */
1104 mountroot = netboot_mountroot;
1105 } else {
1106 #endif
1107 /* otherwise have vfs determine root filesystem */
1108 mountroot = NULL;
1109 #if CONFIG_NETBOOT
1110 }
1111 #endif
1112 }
1113
1114 /*
1115 * Boot into the flavor of Recovery dictated by `mode`.
1116 */
1117 boolean_t
bsd_boot_to_recovery(bsd_bootfail_mode_t mode,uuid_t volume_uuid,boolean_t reboot)1118 bsd_boot_to_recovery(bsd_bootfail_mode_t mode, uuid_t volume_uuid, boolean_t reboot)
1119 {
1120 return IOSetRecoveryBoot(mode, volume_uuid, reboot);
1121 }
1122
1123 void
bsd_utaskbootstrap(void)1124 bsd_utaskbootstrap(void)
1125 {
1126 thread_t thread;
1127 struct uthread *ut;
1128
1129 /*
1130 * Clone the bootstrap process from the kernel process, without
1131 * inheriting either task characteristics or memory from the kernel;
1132 */
1133 thread = cloneproc(TASK_NULL, COALITION_NULL, kernproc, FALSE, TRUE);
1134
1135 /* Hold the reference as it will be dropped during shutdown */
1136 initproc = proc_find(1);
1137 #if __PROC_INTERNAL_DEBUG
1138 if (initproc == PROC_NULL) {
1139 panic("bsd_utaskbootstrap: initproc not set");
1140 }
1141 #endif
1142
1143 zalloc_first_proc_made();
1144
1145 /*
1146 * Since we aren't going back out the normal way to our parent,
1147 * we have to drop the transition locks explicitly.
1148 */
1149 proc_signalend(initproc, 0);
1150 proc_transend(initproc, 0);
1151
1152 ut = (struct uthread *)get_bsdthread_info(thread);
1153 ut->uu_sigmask = 0;
1154 act_set_astbsd(thread);
1155 task_clear_return_wait(get_threadtask(thread), TCRW_CLEAR_ALL_WAIT);
1156 }
1157
1158 static void
parse_bsd_args(void)1159 parse_bsd_args(void)
1160 {
1161 char namep[48];
1162
1163 if (PE_parse_boot_argn("-s", namep, sizeof(namep))) {
1164 boothowto |= RB_SINGLE;
1165 }
1166
1167 if (PE_parse_boot_argn("-x", namep, sizeof(namep))) { /* safe boot */
1168 boothowto |= RB_SAFEBOOT;
1169 }
1170
1171 if (PE_parse_boot_argn("nbuf", &max_nbuf_headers,
1172 sizeof(max_nbuf_headers))) {
1173 customnbuf = 1;
1174 }
1175
1176 #if CONFIG_DARKBOOT
1177 /*
1178 * The darkboot flag is specified by the bootloader and is stored in
1179 * boot_args->bootFlags. This flag is available starting revision 2.
1180 */
1181 boot_args *args = (boot_args *) PE_state.bootArgs;
1182 if ((args != NULL) && (args->Revision >= kBootArgsRevision2)) {
1183 darkboot = (args->bootFlags & kBootFlagsDarkBoot) ? 1 : 0;
1184 } else {
1185 darkboot = 0;
1186 }
1187 #endif
1188
1189 #if DEVELOPMENT || DEBUG
1190 if (PE_parse_boot_argn("alt-dyld", dyld_alt_path, sizeof(dyld_alt_path))) {
1191 if (strlen(dyld_alt_path) > 0) {
1192 use_alt_dyld = 1;
1193 }
1194 }
1195 #endif /* DEVELOPMENT || DEBUG */
1196 }
1197 STARTUP(TUNABLES, STARTUP_RANK_MIDDLE, parse_bsd_args);
1198
1199 #if CONFIG_BASESYSTEMROOT
1200
1201 extern bool IOGetBootUUID(char *);
1202 extern bool IOGetApfsPrebootUUID(char *);
1203
1204
1205 // Get the UUID of the Preboot (and Recovery) folder associated with the
1206 // current boot volume, if applicable. The meaning of the UUID can be
1207 // filesystem-dependent and not all kinds of boots will have a UUID.
1208 // If available, the string will be returned. It does not need to be
1209 // deallocate. (Future: if we need to return the string as a copy that the
1210 // caller must free, we'll introduce a new functcion for that.)
1211 // NULL will be returned if the current boot has no applicable Preboot UUID.
1212 static bool
get_preboot_uuid(uuid_string_t maybe_uuid_string)1213 get_preboot_uuid(uuid_string_t maybe_uuid_string)
1214 {
1215 // try IOGetApfsPrebootUUID
1216 if (IOGetApfsPrebootUUID(maybe_uuid_string)) {
1217 uuid_t maybe_uuid;
1218 int error = uuid_parse(maybe_uuid_string, maybe_uuid);
1219 if (error == 0) {
1220 return true;
1221 }
1222 }
1223
1224 // try IOGetBootUUID
1225 if (IOGetBootUUID(maybe_uuid_string)) {
1226 uuid_t maybe_uuid;
1227 int error = uuid_parse(maybe_uuid_string, maybe_uuid);
1228 if (error == 0) {
1229 return true;
1230 }
1231 }
1232
1233 // didn't find it
1234 return false;
1235 }
1236
1237 #if defined(__arm64__)
1238 extern bool IOGetBootObjectsPath(char *);
1239 #endif
1240
1241 // Find the BaseSystem.dmg to be used as the initial root volume during certain
1242 // kinds of boots.
1243 // This may mount volumes and lookup vnodes.
1244 // The DEVELOPMENT kernel will look for BaseSystem.rooted.dmg first.
1245 // If it returns 0 (no error), then it also writes the absolute path to the
1246 // BaseSystem.dmg into its argument (which must be a char[MAXPATHLEN]).
1247 static
1248 int
bsd_find_basesystem_dmg(char * bsdmgpath_out,bool * rooted_dmg,bool * skip_signature_check)1249 bsd_find_basesystem_dmg(char *bsdmgpath_out, bool *rooted_dmg, bool *skip_signature_check)
1250 {
1251 int error;
1252 size_t len;
1253 char *dmgbasepath;
1254 char *dmgpath;
1255 bool allow_rooted_dmg = false;
1256
1257 dmgbasepath = zalloc_flags(ZV_NAMEI, Z_ZERO | Z_WAITOK);
1258 dmgpath = zalloc_flags(ZV_NAMEI, Z_ZERO | Z_WAITOK);
1259 vnode_t imagevp = NULLVP;
1260
1261 #if DEVELOPMENT || DEBUG
1262 allow_rooted_dmg = true;
1263 #endif
1264
1265 //must provide output bool
1266 if (rooted_dmg && skip_signature_check) {
1267 *rooted_dmg = false;
1268 *skip_signature_check = false;
1269 } else {
1270 error = EINVAL;
1271 goto done;
1272 }
1273
1274 error = vfs_mount_recovery();
1275 if (error) {
1276 goto done;
1277 }
1278
1279 len = strlcpy(dmgbasepath, "/System/Volumes/Recovery/", MAXPATHLEN);
1280 if (len > MAXPATHLEN) {
1281 error = ENAMETOOLONG;
1282 goto done;
1283 }
1284
1285 if (csr_check(CSR_ALLOW_ANY_RECOVERY_OS) == 0) {
1286 *skip_signature_check = true;
1287 allow_rooted_dmg = true;
1288 }
1289
1290 #if defined(__arm64__)
1291 char boot_obj_path[MAXPATHLEN] = "";
1292
1293 if (IOGetBootObjectsPath(boot_obj_path)) {
1294 if (boot_obj_path[0] == '/') {
1295 dmgbasepath[len - 1] = '\0';
1296 }
1297
1298 len = strlcat(dmgbasepath, boot_obj_path, MAXPATHLEN);
1299 if (len > MAXPATHLEN) {
1300 error = ENAMETOOLONG;
1301 goto done;
1302 }
1303
1304 len = strlcat(dmgbasepath, "/usr/standalone/firmware/", MAXPATHLEN);
1305 if (len > MAXPATHLEN) {
1306 error = ENAMETOOLONG;
1307 goto done;
1308 }
1309
1310 if (allow_rooted_dmg) {
1311 len = strlcpy(dmgpath, dmgbasepath, MAXPATHLEN);
1312 if (len > MAXPATHLEN) {
1313 error = ENAMETOOLONG;
1314 goto done;
1315 }
1316
1317 len = strlcat(dmgpath, "arm64eBaseSystem.rooted.dmg", MAXPATHLEN);
1318 if (len > MAXPATHLEN) {
1319 error = ENAMETOOLONG;
1320 goto done;
1321 }
1322
1323 error = vnode_lookup(dmgpath, 0, &imagevp, vfs_context_kernel());
1324 if (error == 0) {
1325 *rooted_dmg = true;
1326 *skip_signature_check = true;
1327 goto done;
1328 }
1329 memset(dmgpath, 0, MAXPATHLEN);
1330 }
1331
1332 len = strlcpy(dmgpath, dmgbasepath, MAXPATHLEN);
1333 if (len > MAXPATHLEN) {
1334 error = ENAMETOOLONG;
1335 goto done;
1336 }
1337
1338 len = strlcat(dmgpath, "arm64eBaseSystem.dmg", MAXPATHLEN);
1339 if (len > MAXPATHLEN) {
1340 error = ENAMETOOLONG;
1341 goto done;
1342 }
1343
1344 error = vnode_lookup(dmgpath, 0, &imagevp, vfs_context_kernel());
1345 if (error == 0) {
1346 goto done;
1347 }
1348 memset(dmgpath, 0, MAXPATHLEN);
1349 dmgbasepath[strlen("/System/Volumes/Recovery/")] = '\0';
1350 }
1351 #endif // __arm64__
1352
1353 uuid_string_t preboot_uuid;
1354 if (!get_preboot_uuid(preboot_uuid)) {
1355 // no preboot? bail out
1356 return EINVAL;
1357 }
1358
1359 len = strlcat(dmgbasepath, preboot_uuid, MAXPATHLEN);
1360 if (len > MAXPATHLEN) {
1361 error = ENAMETOOLONG;
1362 goto done;
1363 }
1364
1365 if (allow_rooted_dmg) {
1366 // Try BaseSystem.rooted.dmg
1367 len = strlcpy(dmgpath, dmgbasepath, MAXPATHLEN);
1368 if (len > MAXPATHLEN) {
1369 error = ENAMETOOLONG;
1370 goto done;
1371 }
1372
1373 len = strlcat(dmgpath, "/BaseSystem.rooted.dmg", MAXPATHLEN);
1374 if (len > MAXPATHLEN) {
1375 error = ENAMETOOLONG;
1376 goto done;
1377 }
1378
1379 error = vnode_lookup(dmgpath, 0, &imagevp, vfs_context_kernel());
1380 if (error == 0) {
1381 // we found it! success!
1382 *rooted_dmg = true;
1383 *skip_signature_check = true;
1384 goto done;
1385 }
1386 }
1387
1388 // Try BaseSystem.dmg
1389 len = strlcpy(dmgpath, dmgbasepath, MAXPATHLEN);
1390 if (len > MAXPATHLEN) {
1391 error = ENAMETOOLONG;
1392 goto done;
1393 }
1394
1395 len = strlcat(dmgpath, "/BaseSystem.dmg", MAXPATHLEN);
1396 if (len > MAXPATHLEN) {
1397 error = ENAMETOOLONG;
1398 goto done;
1399 }
1400
1401 error = vnode_lookup(dmgpath, 0, &imagevp, vfs_context_kernel());
1402 if (error == 0) {
1403 // success!
1404 goto done;
1405 }
1406
1407 done:
1408 if (error == 0) {
1409 strlcpy(bsdmgpath_out, dmgpath, MAXPATHLEN);
1410 } else {
1411 bsd_init_kprintf("%s: error %d\n", __func__, error);
1412 }
1413 if (imagevp != NULLVP) {
1414 vnode_put(imagevp);
1415 }
1416 zfree(ZV_NAMEI, dmgpath);
1417 zfree(ZV_NAMEI, dmgbasepath);
1418 return error;
1419 }
1420
1421 static boolean_t
bsdmgroot_bootable(void)1422 bsdmgroot_bootable(void)
1423 {
1424 #if defined(__arm64__)
1425 #define BSDMGROOT_DEFAULT true
1426 #else
1427 #define BSDMGROOT_DEFAULT false
1428 #endif
1429
1430 boolean_t resolved = BSDMGROOT_DEFAULT;
1431
1432 boolean_t boot_arg_bsdmgroot = false;
1433 boolean_t boot_arg_nobsdmgroot = false;
1434 int error;
1435 mount_t mp;
1436 boolean_t root_part_of_volume_group = false;
1437 struct vfs_attr vfsattr;
1438
1439 mp = rootvnode->v_mount;
1440 VFSATTR_INIT(&vfsattr);
1441 VFSATTR_WANTED(&vfsattr, f_capabilities);
1442
1443 boot_arg_bsdmgroot = PE_parse_boot_argn("-bsdmgroot", NULL, 0);
1444 boot_arg_nobsdmgroot = PE_parse_boot_argn("-nobsdmgroot", NULL, 0);
1445
1446 error = vfs_getattr(mp, &vfsattr, vfs_context_kernel());
1447 if (!error && VFSATTR_IS_SUPPORTED(&vfsattr, f_capabilities)) {
1448 if ((vfsattr.f_capabilities.capabilities[VOL_CAPABILITIES_FORMAT] & VOL_CAP_FMT_VOL_GROUPS) &&
1449 (vfsattr.f_capabilities.valid[VOL_CAPABILITIES_FORMAT] & VOL_CAP_FMT_VOL_GROUPS)) {
1450 root_part_of_volume_group = true;
1451 }
1452 }
1453
1454 boolean_t singleuser = (boothowto & RB_SINGLE) != 0;
1455
1456 // Start with the #defined default above.
1457 // If booting to single-user mode, default to false, because single-
1458 // user mode inside the BaseSystem is probably not what's wanted.
1459 // If the 'yes' boot-arg is set, we'll allow that even in single-user
1460 // mode, we'll assume you know what you're doing.
1461 // The 'no' boot-arg overpowers the 'yes' boot-arg.
1462 // In any case, we will not attempt to root from BaseSystem if the
1463 // original (booter-chosen) root volume isn't in a volume group.
1464 // This is just out of an abundance of caution: if the boot environment
1465 // seems to be "something other than a standard install",
1466 // we'll be conservative in messing with the root volume.
1467
1468 if (singleuser) {
1469 resolved = false;
1470 }
1471
1472 if (boot_arg_bsdmgroot) {
1473 resolved = true;
1474 }
1475
1476 if (boot_arg_nobsdmgroot) {
1477 resolved = false;
1478 }
1479
1480 if (!root_part_of_volume_group) {
1481 resolved = false;
1482 }
1483
1484 return resolved;
1485 }
1486 #endif // CONFIG_BASESYSTEMROOT
1487
1488 void
bsd_exec_setup(int scale)1489 bsd_exec_setup(int scale)
1490 {
1491 switch (scale) {
1492 case 0:
1493 case 1:
1494 bsd_simul_execs = BSD_SIMUL_EXECS;
1495 break;
1496 case 2:
1497 case 3:
1498 bsd_simul_execs = 65;
1499 break;
1500 case 4:
1501 case 5:
1502 bsd_simul_execs = 129;
1503 break;
1504 case 6:
1505 case 7:
1506 bsd_simul_execs = 257;
1507 break;
1508 default:
1509 bsd_simul_execs = 513;
1510 break;
1511 }
1512 bsd_pageable_map_size = (bsd_simul_execs * BSD_PAGEABLE_SIZE_PER_EXEC);
1513 }
1514
1515 #if !CONFIG_NETBOOT
1516 int
1517 netboot_root(void);
1518
1519 int
netboot_root(void)1520 netboot_root(void)
1521 {
1522 return 0;
1523 }
1524 #endif
1525