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 #if (DEVELOPMENT || DEBUG)
113 #include <kern/debug.h>
114 #endif
115
116 #include <mach/vm_param.h>
117
118 #include <vm/vm_map.h>
119 #include <vm/vm_kern.h>
120
121 #include <sys/reboot.h>
122 #include <dev/busvar.h> /* for pseudo_inits */
123 #include <sys/kdebug.h>
124 #include <sys/monotonic.h>
125
126 #include <mach/mach_types.h>
127 #include <mach/vm_prot.h>
128 #include <mach/semaphore.h>
129 #include <mach/sync_policy.h>
130 #include <kern/clock.h>
131 #include <sys/csr.h>
132 #include <mach/kern_return.h>
133 #include <mach/thread_act.h> /* for thread_resume() */
134 #include <sys/mcache.h> /* for mcache_init() */
135 #include <sys/mbuf.h> /* for mbinit() */
136 #include <sys/event.h> /* for knote_init() */
137 #include <sys/eventhandler.h> /* for eventhandler_init() */
138 #include <sys/kern_memorystatus.h> /* for memorystatus_init() */
139 #include <sys/kern_memorystatus_freeze.h> /* for memorystatus_freeze_init() */
140 #include <sys/aio_kern.h> /* for aio_init() */
141 #include <sys/semaphore.h> /* for psem_cache_init() */
142 #include <net/dlil.h> /* for dlil_init() */
143 #include <net/iptap.h> /* for iptap_init() */
144 #include <sys/socketvar.h> /* for socketinit() */
145 #include <sys/protosw.h> /* for domaininit() */
146 #include <kern/sched_prim.h> /* for thread_wakeup() */
147 #include <net/if_ether.h> /* for ether_family_init() */
148 #include <net/if_gif.h> /* for gif_init() */
149 #include <miscfs/devfs/devfsdefs.h> /* for devfs_kernel_mount() */
150 #include <vm/vm_kern.h> /* for kmem_suballoc() */
151 #include <sys/proc_uuid_policy.h> /* proc_uuid_policy_init() */
152 #include <netinet/flow_divert.h> /* flow_divert_init() */
153 #include <net/content_filter.h> /* for cfil_init() */
154 #include <net/necp.h> /* for necp_init() */
155 #include <net/network_agent.h> /* for netagent_init() */
156 #include <net/packet_mangler.h> /* for pkt_mnglr_init() */
157 #include <net/if_utun.h> /* for utun_register_control() */
158 #include <net/if_ipsec.h> /* for ipsec_register_control() */
159 #include <net/netsrc.h> /* for netsrc_init() */
160 #include <net/ntstat.h> /* for nstat_init() */
161 #include <netinet/tcp_cc.h> /* for tcp_cc_init() */
162 #include <netinet/mptcp_var.h> /* for mptcp_control_register() */
163 #include <net/nwk_wq.h> /* for nwk_wq_init */
164 #include <net/restricted_in_port.h> /* for restricted_in_port_init() */
165 #include <net/remote_vif.h> /* for rvi_init() */
166 #include <kern/assert.h> /* for assert() */
167 #include <sys/kern_overrides.h> /* for init_system_override() */
168 #include <sys/lockf.h> /* for lf_init() */
169 #include <sys/fsctl.h>
170
171 #include <net/init.h>
172
173 #if CONFIG_MACF
174 #include <security/mac_framework.h>
175 #include <security/mac_internal.h> /* mac_init_bsd() */
176 #include <security/mac_mach_internal.h> /* mac_update_task_label() */
177 #endif
178
179 #include <machine/exec.h>
180
181 #if CONFIG_NETBOOT
182 #include <sys/netboot.h>
183 #endif
184
185 #if CONFIG_IMAGEBOOT
186 #include <sys/imageboot.h>
187 #endif
188
189 #if PFLOG
190 #include <net/if_pflog.h>
191 #endif
192
193 #if SKYWALK
194 #include <skywalk/os_skywalk_private.h>
195 #endif /* SKYWALK */
196
197 #include <pexpert/pexpert.h>
198 #include <machine/pal_routines.h>
199 #include <console/video_console.h>
200
201 #if CONFIG_XNUPOST
202 #include <tests/xnupost.h>
203 #endif
204
205 void * get_user_regs(thread_t); /* XXX kludge for <machine/thread.h> */
206 void IOKitInitializeTime(void); /* XXX */
207 void IOSleep(unsigned int); /* XXX */
208 void IOSetImageBoot(void); /* XXX */
209 void loopattach(void); /* XXX */
210
211 void ipc_task_enable(task_t task);
212
213 const char *const copyright =
214 "Copyright (c) 1982, 1986, 1989, 1991, 1993\n\t"
215 "The Regents of the University of California. "
216 "All rights reserved.\n\n";
217
218 /* Components of the first process -- never freed. */
219 SECURITY_READ_ONLY_LATE(struct vfs_context) vfs_context0;
220
221 static struct plimit limit0;
222 static struct pstats pstats0;
223 SECURITY_READ_ONLY_LATE(proc_t) kernproc;
224 proc_t XNU_PTRAUTH_SIGNED_PTR("initproc") initproc;
225
226 long tk_cancc;
227 long tk_nin;
228 long tk_nout;
229 long tk_rawcc;
230
231 int lock_trace = 0;
232 /* Global variables to make pstat happy. We do swapping differently */
233 int nswdev, nswap;
234 int nswapmap;
235 void *swapmap;
236 struct swdevt swdevt[1];
237
238 static LCK_GRP_DECLARE(hostname_lck_grp, "hostname");
239 LCK_MTX_DECLARE(hostname_lock, &hostname_lck_grp);
240 LCK_MTX_DECLARE(domainname_lock, &hostname_lck_grp);
241
242 dev_t rootdev; /* device of the root */
243 dev_t dumpdev; /* device to take dumps on */
244 long dumplo; /* offset into dumpdev */
245 long hostid;
246 char hostname[MAXHOSTNAMELEN];
247 char domainname[MAXDOMNAMELEN];
248 char rootdevice[DEVMAXNAMESIZE];
249
250 struct vnode *rootvp;
251 bool rootvp_is_ssd = false;
252 SECURITY_READ_ONLY_LATE(int) boothowto;
253 /*
254 * -minimalboot indicates that we want userspace to be bootstrapped to a
255 * minimal environment. What constitutes minimal is up to the bootstrap
256 * process.
257 */
258 TUNABLE(int, minimalboot, "-minimalboot", 0);
259 #if CONFIG_DARKBOOT
260 int darkboot = 0;
261 #endif
262
263 extern kern_return_t IOFindBSDRoot(char *, unsigned int, dev_t *, u_int32_t *);
264 extern void IOSecureBSDRoot(const char * rootName);
265 extern kern_return_t IOKitBSDInit(void );
266 extern boolean_t IOSetRecoveryBoot(bsd_bootfail_mode_t, uuid_t, boolean_t);
267 extern void kminit(void);
268 extern void bsd_bufferinit(void);
269 extern void throttle_init(void);
270
271 vm_map_t bsd_pageable_map;
272 vm_map_t mb_map;
273
274 static int bsd_simul_execs;
275 static int bsd_pageable_map_size;
276 __private_extern__ int execargs_cache_size = 0;
277 __private_extern__ int execargs_free_count = 0;
278 __private_extern__ vm_offset_t * execargs_cache = NULL;
279
280 void bsd_exec_setup(int);
281
282 __private_extern__ int bootarg_execfailurereports = 0;
283
284 #if __x86_64__
285 __private_extern__ TUNABLE(int, bootarg_no32exec, "no32exec", 1);
286 #endif
287
288 #if DEVELOPMENT || DEBUG
289 /* Prevent kernel-based ASLR from being used. */
290 __private_extern__ TUNABLE(bool, bootarg_disable_aslr, "-disable_aslr", 0);
291 #endif
292
293 /*
294 * Allow an alternate dyld to be used for testing.
295 */
296
297 #if DEVELOPMENT || DEBUG
298 char dyld_alt_path[MAXPATHLEN];
299 int use_alt_dyld = 0;
300
301 char panic_on_proc_crash[NAME_MAX];
302 int use_panic_on_proc_crash = 0;
303
304 char panic_on_proc_exit[NAME_MAX];
305 int use_panic_on_proc_exit = 0;
306
307 char panic_on_proc_spawn_fail[NAME_MAX];
308 int use_panic_on_proc_spawn_fail = 0;
309
310 char dyld_suffix[NAME_MAX];
311 int use_dyld_suffix = 0;
312 #endif
313
314 int cmask = CMASK;
315 extern int customnbuf;
316
317 kern_return_t bsd_autoconf(void);
318 void bsd_utaskbootstrap(void);
319
320 #if CONFIG_DEV_KMEM
321 extern void dev_kmem_init(void);
322 #endif
323 static void process_name(const char *, proc_t);
324
325 static void setconf(void);
326
327 #if CONFIG_BASESYSTEMROOT
328 static int bsd_find_basesystem_dmg(char *bsdmgpath_out, bool *rooted_dmg, bool *skip_signature_check);
329 static boolean_t bsdmgroot_bootable(void);
330 #endif // CONFIG_BASESYSTEMROOT
331
332 static bool bsd_rooted_ramdisk(void);
333
334 #if SYSV_SHM
335 extern void sysv_shm_lock_init(void);
336 #endif
337 #if SYSV_SEM
338 extern void sysv_sem_lock_init(void);
339 #endif
340 #if SYSV_MSG
341 extern void sysv_msg_lock_init(void);
342 #endif
343
344 #if CONFIG_MACF
345 #if defined (__i386__) || defined (__x86_64__)
346 /* MACF policy_check configuration flags; see policy_check.c for details */
347 extern int check_policy_init(int);
348 #endif
349 #endif /* CONFIG_MACF */
350
351 /* If we are using CONFIG_DTRACE */
352 #if CONFIG_DTRACE
353 extern void dtrace_postinit(void);
354 #endif
355
356 /*
357 * Initialization code.
358 * Called from cold start routine as
359 * soon as a stack and segmentation
360 * have been established.
361 * Functions:
362 * turn on clock
363 * hand craft 0th process
364 * call all initialization routines
365 * hand craft 1st user process
366 */
367
368 /*
369 * Sets the name for the given task.
370 */
371 static void
process_name(const char * s,proc_t p)372 process_name(const char *s, proc_t p)
373 {
374 strlcpy(p->p_comm, s, sizeof(p->p_comm));
375 strlcpy(p->p_name, s, sizeof(p->p_name));
376 }
377
378 /* To allow these values to be patched, they're globals here */
379 #include <machine/vmparam.h>
380 struct rlimit vm_initial_limit_stack = { .rlim_cur = DFLSSIZ, .rlim_max = MAXSSIZ - PAGE_MAX_SIZE };
381 struct rlimit vm_initial_limit_data = { .rlim_cur = DFLDSIZ, .rlim_max = MAXDSIZ };
382 struct rlimit vm_initial_limit_core = { .rlim_cur = DFLCSIZ, .rlim_max = MAXCSIZ };
383
384 extern struct os_refgrp rlimit_refgrp;
385
386 extern int (*mountroot)(void);
387
388 LCK_ATTR_DECLARE(proc_lck_attr, 0, 0);
389 LCK_GRP_DECLARE(proc_lck_grp, "proc");
390 LCK_GRP_DECLARE(proc_slock_grp, "proc-slock");
391 LCK_GRP_DECLARE(proc_fdmlock_grp, "proc-fdmlock");
392 LCK_GRP_DECLARE(proc_mlock_grp, "proc-mlock");
393 LCK_GRP_DECLARE(proc_ucred_mlock_grp, "proc-ucred-mlock");
394 LCK_GRP_DECLARE(proc_dirslock_grp, "proc-dirslock");
395 LCK_GRP_DECLARE(proc_kqhashlock_grp, "proc-kqhashlock");
396 LCK_GRP_DECLARE(proc_knhashlock_grp, "proc-knhashlock");
397
398
399 LCK_MTX_DECLARE_ATTR(proc_list_mlock, &proc_mlock_grp, &proc_lck_attr);
400
401 #if XNU_TARGET_OS_OSX
402 /* hook called after root is mounted XXX temporary hack */
403 void (*mountroot_post_hook)(void);
404 void (*unmountroot_pre_hook)(void);
405 #endif
406 void set_rootvnode(vnode_t);
407
408 extern lck_rw_t rootvnode_rw_lock;
409
410 SECURITY_READ_ONLY_LATE(struct mach_vm_range) bsd_pageable_range = {};
411 KMEM_RANGE_REGISTER_DYNAMIC(bsd_pageable, &bsd_pageable_range, ^() {
412 assert(bsd_pageable_map_size != 0);
413 return (vm_map_size_t) bsd_pageable_map_size;
414 });
415
416 /* called with an iocount and usecount on new_rootvnode */
417 void
set_rootvnode(vnode_t new_rootvnode)418 set_rootvnode(vnode_t new_rootvnode)
419 {
420 mount_t new_mount = (new_rootvnode != NULL) ? new_rootvnode->v_mount : NULL;
421 vnode_t new_devvp = (new_mount != NULL) ? new_mount->mnt_devvp : NULL;
422 vnode_t old_rootvnode = rootvnode;
423
424 new_rootvnode->v_flag |= VROOT;
425 rootvp = new_devvp;
426 rootvnode = new_rootvnode;
427 kernproc->p_fd.fd_cdir = new_rootvnode;
428 if (new_devvp != NULL) {
429 rootdev = vnode_specrdev(new_devvp);
430 } else if (new_mount != NULL) {
431 rootdev = vfs_statfs(new_mount)->f_fsid.val[0]; /* like ATTR_CMN_DEVID */
432 } else {
433 rootdev = NODEV;
434 }
435
436 if (old_rootvnode) {
437 vnode_rele(old_rootvnode);
438 }
439 }
440
441 #define RAMDEV "md0"
442
443 bool
bsd_rooted_ramdisk(void)444 bsd_rooted_ramdisk(void)
445 {
446 bool is_ramdisk = false;
447 char *dev_path = zalloc(ZV_NAMEI);
448 if (dev_path == NULL) {
449 panic("failed to allocate devpath string!");
450 }
451
452 if (PE_parse_boot_argn("rd", dev_path, MAXPATHLEN)) {
453 if (strncmp(dev_path, RAMDEV, strlen(RAMDEV)) == 0) {
454 is_ramdisk = true;
455 }
456 }
457
458 zfree(ZV_NAMEI, dev_path);
459 return is_ramdisk;
460 }
461
462 /*
463 * This function is called very early on in the Mach startup, from the
464 * function start_kernel_threads() in osfmk/kern/startup.c. It's called
465 * in the context of the current (startup) task using a call to the
466 * function kernel_thread_create() to jump into start_kernel_threads().
467 * Internally, kernel_thread_create() calls thread_create_internal(),
468 * which calls uthread_init(). The function of uthread_init() is
469 * normally to init a uthread structure, and fill out the uu_sigmask,
470 * tro_ucred/tro_proc fields. It skips filling these out in the case of the "task"
471 * being "kernel_task", because the order of operation is inverted. To
472 * account for that, we need to manually fill in at least the contents
473 * of the tro_ucred field so that the uthread structure can be
474 * used like any other.
475 */
476 void
bsd_init(void)477 bsd_init(void)
478 {
479 struct uthread *ut;
480 vnode_t init_rootvnode = NULLVP;
481 struct proc_ro_data kernproc_ro_data = {
482 .p_csflags = CS_VALID,
483 };
484 struct task_ro_data kerntask_ro_data = { };
485 #if CONFIG_NETBOOT || CONFIG_IMAGEBOOT
486 boolean_t netboot = FALSE;
487 #endif
488
489 #if (DEVELOPMENT || DEBUG)
490 platform_stall_panic_or_spin(PLATFORM_STALL_XNU_LOCATION_BSD_INIT);
491 #endif
492
493 #define DEBUG_BSDINIT 0
494
495 #if DEBUG_BSDINIT
496 #define bsd_init_kprintf(x, ...) kprintf("bsd_init: " x, ## __VA_ARGS__)
497 #else
498 #define bsd_init_kprintf(x, ...)
499 #endif
500
501 throttle_init();
502
503 printf(copyright);
504
505 #if CONFIG_DEV_KMEM
506 bsd_init_kprintf("calling dev_kmem_init\n");
507 dev_kmem_init();
508 #endif
509
510 /* Initialize kauth subsystem before instancing the first credential */
511 bsd_init_kprintf("calling kauth_init\n");
512 kauth_init();
513
514 /* kernel_task->proc = kernproc; */
515 set_bsdtask_info(kernel_task, (void *)kernproc);
516
517 /* Set the parent of kernproc to itself */
518 kernproc->p_pptr = kernproc;
519
520 /* Set the state to SRUN */
521 kernproc->p_stat = SRUN;
522
523 /* Set the proc flags */
524 #if defined(__LP64__)
525 kernproc->p_flag = P_SYSTEM | P_LP64;
526 #else
527 kernproc->p_flag = P_SYSTEM;
528 #endif
529
530 kernproc->p_nice = NZERO;
531 TAILQ_INIT(&kernproc->p_uthlist);
532
533 /* set the cred */
534 kauth_cred_set(&kernproc_ro_data.p_ucred, vfs_context0.vc_ucred);
535 kernproc->p_proc_ro = proc_ro_alloc(kernproc, &kernproc_ro_data,
536 kernel_task, &kerntask_ro_data);
537
538 /* give kernproc a name */
539 bsd_init_kprintf("calling process_name\n");
540 process_name("kernel_task", kernproc);
541
542 /* Allocate proc lock attribute */
543
544 lck_mtx_init(&kernproc->p_mlock, &proc_mlock_grp, &proc_lck_attr);
545 lck_mtx_init(&kernproc->p_ucred_mlock, &proc_ucred_mlock_grp, &proc_lck_attr);
546 lck_spin_init(&kernproc->p_slock, &proc_slock_grp, &proc_lck_attr);
547
548 /* Init the file descriptor table. */
549 fdt_init(kernproc);
550 kernproc->p_fd.fd_cmask = (mode_t)cmask;
551
552 assert(bsd_simul_execs != 0);
553 execargs_cache_size = bsd_simul_execs;
554 execargs_free_count = bsd_simul_execs;
555 execargs_cache = zalloc_permanent(bsd_simul_execs * sizeof(vm_offset_t),
556 ZALIGN(vm_offset_t));
557
558 if (current_task() != kernel_task) {
559 printf("bsd_init: We have a problem, "
560 "current task is not kernel task\n");
561 }
562
563 bsd_init_kprintf("calling get_bsdthread_info\n");
564 ut = current_uthread();
565
566 #if CONFIG_MACF
567 /*
568 * Initialize the MAC Framework
569 */
570 mac_policy_initbsd();
571
572 #if defined (__i386__) || defined (__x86_64__)
573 /*
574 * We currently only support this on i386/x86_64, as that is the
575 * only lock code we have instrumented so far.
576 */
577 int policy_check_flags;
578 PE_parse_boot_argn("policy_check", &policy_check_flags, sizeof(policy_check_flags));
579 check_policy_init(policy_check_flags);
580 #endif
581 #endif /* MAC */
582
583 /*
584 * Make a session and group
585 *
586 * No need to hold the pgrp lock,
587 * there are no other BSD threads yet.
588 */
589 struct session *session0 = session_alloc(kernproc);
590 struct pgrp *pgrp0 = pgrp_alloc(0, PGRP_REF_NONE);
591 session0->s_ttypgrpid = 0;
592 pgrp0->pg_session = session0;
593
594 /*
595 * Create process 0.
596 */
597 proc_list_lock();
598 os_ref_init_mask(&kernproc->p_refcount, P_REF_BITS, &p_refgrp, P_REF_NONE);
599 os_ref_init_raw(&kernproc->p_waitref, &p_refgrp);
600 proc_ref_hold_proc_task_struct(kernproc);
601
602 /*
603 * Make a group and session, then simulate pinsertchild(),
604 * adjusted for the kernel.
605 */
606 pghash_insert_locked(0, pgrp0);
607
608 LIST_INSERT_HEAD(&pgrp0->pg_members, kernproc, p_pglist);
609 smr_init_store(&kernproc->p_pgrp, pgrp0);
610 LIST_INSERT_HEAD(&allproc, kernproc, p_list);
611
612 LIST_INSERT_HEAD(SESSHASH(0), session0, s_hash);
613 proc_list_unlock();
614
615 proc_set_task(kernproc, kernel_task);
616
617 #if DEVELOPMENT || DEBUG
618 if (bootarg_disable_aslr) {
619 kernproc->p_flag |= P_DISABLE_ASLR;
620 }
621 #endif
622
623 TAILQ_INSERT_TAIL(&kernproc->p_uthlist, ut, uu_list);
624
625 bsd_init_kprintf("calling kauth_cred_create\n");
626 /*
627 * Officially associate the kernel with vfs_context0.vc_ucred.
628 */
629 #if CONFIG_MACF
630 mac_cred_label_associate_kernel(vfs_context0.vc_ucred);
631 #endif
632 proc_update_creds_onproc(kernproc);
633
634 TAILQ_INIT(&kernproc->p_aio_activeq);
635 TAILQ_INIT(&kernproc->p_aio_doneq);
636 kernproc->p_aio_total_count = 0;
637
638 /* Create the limits structures. */
639 for (uint32_t i = 0; i < ARRAY_COUNT(limit0.pl_rlimit); i++) {
640 limit0.pl_rlimit[i].rlim_cur =
641 limit0.pl_rlimit[i].rlim_max = RLIM_INFINITY;
642 }
643 limit0.pl_rlimit[RLIMIT_NOFILE].rlim_cur = NOFILE;
644 limit0.pl_rlimit[RLIMIT_NPROC].rlim_cur = maxprocperuid;
645 limit0.pl_rlimit[RLIMIT_NPROC].rlim_max = maxproc;
646 limit0.pl_rlimit[RLIMIT_STACK] = vm_initial_limit_stack;
647 limit0.pl_rlimit[RLIMIT_DATA] = vm_initial_limit_data;
648 limit0.pl_rlimit[RLIMIT_CORE] = vm_initial_limit_core;
649 os_ref_init_count(&limit0.pl_refcnt, &rlimit_refgrp, 1);
650
651 smr_init_store(&kernproc->p_limit, &limit0);
652 kernproc->p_stats = &pstats0;
653 proc_sigacts_copy(kernproc, NULL);
654 kernproc->p_subsystem_root_path = NULL;
655
656 /*
657 * Charge root for one process: launchd.
658 */
659 bsd_init_kprintf("calling chgproccnt\n");
660 (void)chgproccnt(0, 1);
661
662 /*
663 * Allocate a kernel submap for pageable memory
664 * for temporary copying (execve()).
665 */
666 bsd_init_kprintf("calling kmem_suballoc\n");
667 bsd_pageable_map = kmem_suballoc(kernel_map,
668 &bsd_pageable_range.min_address,
669 (vm_size_t)bsd_pageable_map_size,
670 VM_MAP_CREATE_PAGEABLE,
671 VM_FLAGS_FIXED_RANGE_SUBALLOC,
672 KMS_PERMANENT | KMS_NOFAIL,
673 VM_KERN_MEMORY_BSD).kmr_submap;
674
675 /*
676 * Initialize buffers and hash links for buffers
677 *
678 * SIDE EFFECT: Starts a thread for bcleanbuf_thread(), so must
679 * happen after a credential has been associated with
680 * the kernel task.
681 */
682 bsd_init_kprintf("calling bsd_bufferinit\n");
683 bsd_bufferinit();
684
685 /*
686 * Initialize the calendar.
687 */
688 bsd_init_kprintf("calling IOKitInitializeTime\n");
689 IOKitInitializeTime();
690
691 /* Initialize the file systems. */
692 bsd_init_kprintf("calling vfsinit\n");
693 vfsinit();
694
695 #if CONFIG_PROC_UUID_POLICY
696 /* Initial proc_uuid_policy subsystem */
697 bsd_init_kprintf("calling proc_uuid_policy_init()\n");
698 proc_uuid_policy_init();
699 #endif
700
701 #if SOCKETS
702 /* Initialize per-CPU cache allocator */
703 mcache_init();
704
705 /* Initialize mbuf's. */
706 bsd_init_kprintf("calling mbinit\n");
707 mbinit();
708 restricted_in_port_init();
709 #endif /* SOCKETS */
710
711 /*
712 * Initializes security event auditing.
713 * XXX: Should/could this occur later?
714 */
715 #if CONFIG_AUDIT
716 bsd_init_kprintf("calling audit_init\n");
717 audit_init();
718 #endif
719
720 /* Initialize kqueues */
721 bsd_init_kprintf("calling knote_init\n");
722 knote_init();
723
724 /* Initialize event handler */
725 bsd_init_kprintf("calling eventhandler_init\n");
726 eventhandler_init();
727
728 /* Initialize for async IO */
729 bsd_init_kprintf("calling aio_init\n");
730 aio_init();
731
732 pthread_init();
733 /* POSIX Shm and Sem */
734 bsd_init_kprintf("calling pshm_cache_init\n");
735 pshm_cache_init();
736 bsd_init_kprintf("calling psem_cache_init\n");
737 psem_cache_init();
738
739 /*
740 * Initialize protocols. Block reception of incoming packets
741 * until everything is ready.
742 */
743 #if NETWORKING
744 bsd_init_kprintf("calling nwk_wq_init\n");
745 nwk_wq_init();
746 bsd_init_kprintf("calling dlil_init\n");
747 dlil_init();
748 #endif /* NETWORKING */
749 #if SOCKETS
750 bsd_init_kprintf("calling socketinit\n");
751 socketinit();
752 bsd_init_kprintf("calling domaininit\n");
753 domaininit();
754 iptap_init();
755 #if FLOW_DIVERT
756 flow_divert_init();
757 #endif /* FLOW_DIVERT */
758 #endif /* SOCKETS */
759 #if SKYWALK
760 bsd_init_kprintf("calling skywalk_init\n");
761 (void) skywalk_init();
762 #endif /* SKYWALK */
763 #if NETWORKING
764 #if NECP
765 /* Initialize Network Extension Control Policies */
766 necp_init();
767 #endif
768 netagent_init();
769 #endif /* NETWORKING */
770
771 #if CONFIG_FREEZE
772 #ifndef CONFIG_MEMORYSTATUS
773 #error "CONFIG_FREEZE defined without matching CONFIG_MEMORYSTATUS"
774 #endif
775 /* Initialise background freezing */
776 bsd_init_kprintf("calling memorystatus_freeze_init\n");
777 memorystatus_freeze_init();
778 #endif
779
780 #if CONFIG_MEMORYSTATUS
781 /* Initialize kernel memory status notifications */
782 bsd_init_kprintf("calling memorystatus_init\n");
783 memorystatus_init();
784 #endif /* CONFIG_MEMORYSTATUS */
785
786 bsd_init_kprintf("calling sysctl_mib_init\n");
787 sysctl_mib_init();
788
789 bsd_init_kprintf("calling bsd_autoconf\n");
790 bsd_autoconf();
791
792 #if CONFIG_DTRACE
793 dtrace_postinit();
794 #endif
795
796 /*
797 * We attach the loopback interface *way* down here to ensure
798 * it happens after autoconf(), otherwise it becomes the
799 * "primary" interface.
800 */
801 #include <loop.h>
802 #if NLOOP > 0
803 bsd_init_kprintf("calling loopattach\n");
804 loopattach(); /* XXX */
805 #endif
806 #if NGIF
807 /* Initialize gif interface (after lo0) */
808 gif_init();
809 #endif
810
811 #if PFLOG
812 /* Initialize packet filter log interface */
813 pfloginit();
814 #endif /* PFLOG */
815
816 #if NETHER > 0
817 /* Register the built-in dlil ethernet interface family */
818 bsd_init_kprintf("calling ether_family_init\n");
819 ether_family_init();
820 #endif /* ETHER */
821
822 #if NETWORKING
823 #if CONTENT_FILTER
824 cfil_init();
825 #endif
826
827 #if PACKET_MANGLER
828 pkt_mnglr_init();
829 #endif
830
831 /*
832 * Register subsystems with kernel control handlers
833 */
834 utun_register_control();
835 #if IPSEC
836 ipsec_register_control();
837 #endif /* IPSEC */
838 netsrc_init();
839 nstat_init();
840 tcp_cc_init();
841 #if MPTCP
842 mptcp_control_register();
843 #endif /* MPTCP */
844
845 #if REMOTE_VIF
846 rvi_init();
847 #endif /* REMOTE_VIF */
848
849 /*
850 * The the networking stack is now initialized so it is a good time to call
851 * the clients that are waiting for the networking stack to be usable.
852 */
853 bsd_init_kprintf("calling net_init_run\n");
854 net_init_run();
855 #endif /* NETWORKING */
856
857 bsd_init_kprintf("calling inittodr\n");
858 inittodr(0);
859
860 /* Mount the root file system. */
861 while (TRUE) {
862 int err;
863
864 bsd_init_kprintf("calling setconf\n");
865 setconf();
866 #if CONFIG_NETBOOT
867 netboot = (mountroot == netboot_mountroot);
868 #endif
869
870 bsd_init_kprintf("vfs_mountroot\n");
871 if (0 == (err = vfs_mountroot())) {
872 break;
873 }
874 rootdevice[0] = '\0';
875 #if CONFIG_NETBOOT
876 if (netboot) {
877 PE_display_icon( 0, "noroot"); /* XXX a netboot-specific icon would be nicer */
878 vc_progress_set(FALSE, 0);
879 for (uint32_t i = 1; 1; i *= 2) {
880 printf("bsd_init: failed to mount network root, error %d, %s\n",
881 err, PE_boot_args());
882 printf("We are hanging here...\n");
883 IOSleep(i * 60 * 1000);
884 }
885 /*NOTREACHED*/
886 }
887 #endif
888 printf("cannot mount root, errno = %d\n", err);
889 }
890
891 IOSecureBSDRoot(rootdevice);
892
893 mountlist.tqh_first->mnt_flag |= MNT_ROOTFS;
894
895 bsd_init_kprintf("calling VFS_ROOT\n");
896 /* Get the vnode for '/'. Set fdp->fd_fd.fd_cdir to reference it. */
897 if (VFS_ROOT(mountlist.tqh_first, &init_rootvnode, vfs_context_kernel())) {
898 panic("bsd_init: cannot find root vnode: %s", PE_boot_args());
899 }
900 (void)vnode_ref(init_rootvnode);
901 (void)vnode_put(init_rootvnode);
902
903 lck_rw_lock_exclusive(&rootvnode_rw_lock);
904 set_rootvnode(init_rootvnode);
905 lck_rw_unlock_exclusive(&rootvnode_rw_lock);
906 init_rootvnode = NULLVP; /* use rootvnode after this point */
907
908
909 if (!bsd_rooted_ramdisk()) {
910 boolean_t require_rootauth = FALSE;
911
912 #if XNU_TARGET_OS_OSX && defined(__arm64__)
913 #if CONFIG_IMAGEBOOT
914 /* Apple Silicon MacOS */
915 require_rootauth = !imageboot_desired();
916 #endif // CONFIG_IMAGEBOOT
917 #elif !XNU_TARGET_OS_OSX
918 /* Non MacOS */
919 require_rootauth = TRUE;
920 #endif // XNU_TARGET_OS_OSX && defined(__arm64__)
921
922 if (require_rootauth) {
923 /* enforce sealedness */
924 int autherr = VNOP_IOCTL(rootvnode, FSIOC_KERNEL_ROOTAUTH, NULL, 0, vfs_context_kernel());
925 if (autherr) {
926 panic("rootvp not authenticated after mounting");
927 }
928 }
929 }
930
931
932 #if CONFIG_NETBOOT
933 if (netboot) {
934 int err;
935
936 netboot = TRUE;
937 /* post mount setup */
938 if ((err = netboot_setup()) != 0) {
939 PE_display_icon( 0, "noroot"); /* XXX a netboot-specific icon would be nicer */
940 vc_progress_set(FALSE, 0);
941 for (uint32_t i = 1; 1; i *= 2) {
942 printf("bsd_init: NetBoot could not find root, error %d: %s\n",
943 err, PE_boot_args());
944 printf("We are hanging here...\n");
945 IOSleep(i * 60 * 1000);
946 }
947 /*NOTREACHED*/
948 }
949 }
950 #endif
951
952
953 #if CONFIG_IMAGEBOOT
954 /*
955 * See if a system disk image is present. If so, mount it and
956 * switch the root vnode to point to it
957 */
958 imageboot_type_t imageboot_type = imageboot_needed();
959 if (netboot == FALSE && imageboot_type) {
960 /*
961 * An image was found. No turning back: we're booted
962 * with a kernel from the disk image.
963 */
964 bsd_init_kprintf("doing image boot: type = %d\n", imageboot_type);
965 imageboot_setup(imageboot_type);
966 IOSetImageBoot();
967 }
968
969 #endif /* CONFIG_IMAGEBOOT */
970
971 /* set initial time; all other resource data is already zero'ed */
972 microtime_with_abstime(&kernproc->p_start, &kernproc->p_stats->ps_start);
973
974 #if DEVFS
975 {
976 char mounthere[] = "/dev"; /* !const because of internal casting */
977
978 bsd_init_kprintf("calling devfs_kernel_mount\n");
979 devfs_kernel_mount(mounthere);
980 }
981 #endif /* DEVFS */
982
983 #if CONFIG_BASESYSTEMROOT
984 #if CONFIG_IMAGEBOOT
985 if (bsdmgroot_bootable()) {
986 int error;
987 bool rooted_dmg = false;
988 bool skip_signature_check = false;
989
990 printf("trying to find and mount BaseSystem dmg as root volume\n");
991 #if DEVELOPMENT || DEBUG
992 printf("(set boot-arg -nobsdmgroot to avoid this)\n");
993 #endif // DEVELOPMENT || DEBUG
994
995 char *dmgpath = NULL;
996 dmgpath = zalloc_flags(ZV_NAMEI, Z_ZERO | Z_WAITOK | Z_NOFAIL);
997
998 error = bsd_find_basesystem_dmg(dmgpath, &rooted_dmg, &skip_signature_check);
999 if (error) {
1000 bsd_init_kprintf("failed to to find BaseSystem dmg: error = %d\n", error);
1001 } else {
1002 PE_parse_boot_argn("bsdmgpath", dmgpath, sizeof(dmgpath));
1003
1004 bsd_init_kprintf("found BaseSystem dmg at: %s\n", dmgpath);
1005
1006 error = imageboot_pivot_image(dmgpath, IMAGEBOOT_DMG, "/System/Volumes/BaseSystem", "System/Volumes/macOS", rooted_dmg, skip_signature_check);
1007 if (error) {
1008 bsd_init_kprintf("couldn't mount BaseSystem dmg: error = %d", error);
1009 } else {
1010 IOSetImageBoot();
1011 }
1012 }
1013 zfree(ZV_NAMEI, dmgpath);
1014 }
1015 #else /* CONFIG_IMAGEBOOT */
1016 #error CONFIG_BASESYSTEMROOT requires CONFIG_IMAGEBOOT
1017 #endif /* CONFIG_IMAGEBOOT */
1018 #endif /* CONFIG_BASESYSTEMROOT */
1019
1020 /* Initialize signal state for process 0. */
1021 bsd_init_kprintf("calling siginit\n");
1022 siginit(kernproc);
1023
1024 bsd_init_kprintf("calling bsd_utaskbootstrap\n");
1025 bsd_utaskbootstrap();
1026
1027 pal_kernel_announce();
1028
1029 bsd_init_kprintf("calling mountroot_post_hook\n");
1030
1031 #if XNU_TARGET_OS_OSX
1032 /* invoke post-root-mount hook */
1033 if (mountroot_post_hook != NULL) {
1034 mountroot_post_hook();
1035 }
1036 #endif
1037
1038 #if 0 /* not yet */
1039 consider_zone_gc(FALSE);
1040 #endif
1041
1042 /*
1043 * At this point, we consider the kernel "booted" enough to apply
1044 * stricter timeouts.
1045 */
1046 machine_timeout_bsd_init();
1047
1048 bsd_init_kprintf("done\n");
1049 }
1050
1051 void
bsdinit_task(void)1052 bsdinit_task(void)
1053 {
1054 proc_t p = current_proc();
1055
1056 process_name("init", p);
1057
1058 /* Set up exception-to-signal reflection */
1059 ux_handler_setup();
1060
1061 #if CONFIG_MACF
1062 mac_cred_label_associate_user(proc_ucred(p));
1063 #endif
1064
1065 vm_init_before_launchd();
1066
1067 #if CONFIG_XNUPOST
1068 int result = bsd_list_tests();
1069 result = bsd_do_post();
1070 if (result != 0) {
1071 panic("bsd_do_post: Tests failed with result = 0x%08x", result);
1072 }
1073 #endif
1074
1075 bsd_init_kprintf("bsd_do_post - done");
1076
1077 load_init_program(p);
1078 lock_trace = 1;
1079 }
1080
1081 kern_return_t
bsd_autoconf(void)1082 bsd_autoconf(void)
1083 {
1084 kprintf("bsd_autoconf: calling kminit\n");
1085 kminit();
1086
1087 /*
1088 * Early startup for bsd pseudodevices.
1089 */
1090 {
1091 struct pseudo_init *pi;
1092
1093 for (pi = pseudo_inits; pi->ps_func; pi++) {
1094 (*pi->ps_func)(pi->ps_count);
1095 }
1096 }
1097
1098 return IOKitBSDInit();
1099 }
1100
1101
1102 #include <sys/disklabel.h> /* for MAXPARTITIONS */
1103
1104 static void
setconf(void)1105 setconf(void)
1106 {
1107 u_int32_t flags;
1108 kern_return_t err;
1109
1110 err = IOFindBSDRoot(rootdevice, sizeof(rootdevice), &rootdev, &flags);
1111 if (err) {
1112 printf("setconf: IOFindBSDRoot returned an error (%d);"
1113 "setting rootdevice to 'sd0a'.\n", err); /* XXX DEBUG TEMP */
1114 rootdev = makedev( 6, 0 );
1115 strlcpy(rootdevice, "sd0a", sizeof(rootdevice));
1116 flags = 0;
1117 }
1118
1119 #if CONFIG_NETBOOT
1120 if (flags & 1) {
1121 /* network device */
1122 mountroot = netboot_mountroot;
1123 } else {
1124 #endif
1125 /* otherwise have vfs determine root filesystem */
1126 mountroot = NULL;
1127 #if CONFIG_NETBOOT
1128 }
1129 #endif
1130 }
1131
1132 /*
1133 * Boot into the flavor of Recovery dictated by `mode`.
1134 */
1135 boolean_t
bsd_boot_to_recovery(bsd_bootfail_mode_t mode,uuid_t volume_uuid,boolean_t reboot)1136 bsd_boot_to_recovery(bsd_bootfail_mode_t mode, uuid_t volume_uuid, boolean_t reboot)
1137 {
1138 return IOSetRecoveryBoot(mode, volume_uuid, reboot);
1139 }
1140
1141 void
bsd_utaskbootstrap(void)1142 bsd_utaskbootstrap(void)
1143 {
1144 thread_t thread;
1145 struct uthread *ut;
1146
1147 /*
1148 * Clone the bootstrap process from the kernel process, without
1149 * inheriting either task characteristics or memory from the kernel;
1150 */
1151 thread = cloneproc(TASK_NULL, NULL, kernproc, CLONEPROC_FLAGS_MEMSTAT_INTERNAL);
1152
1153 /* Hold the reference as it will be dropped during shutdown */
1154 initproc = proc_find(1);
1155 #if __PROC_INTERNAL_DEBUG
1156 if (initproc == PROC_NULL) {
1157 panic("bsd_utaskbootstrap: initproc not set");
1158 }
1159 #endif
1160
1161 zalloc_first_proc_made();
1162
1163 /*
1164 * Since we aren't going back out the normal way to our parent,
1165 * we have to drop the transition locks explicitly.
1166 */
1167 proc_signalend(initproc, 0);
1168 proc_transend(initproc, 0);
1169
1170 ut = (struct uthread *)get_bsdthread_info(thread);
1171 ut->uu_sigmask = 0;
1172 act_set_astbsd(thread);
1173
1174 ipc_task_enable(get_threadtask(thread));
1175
1176 task_clear_return_wait(get_threadtask(thread), TCRW_CLEAR_ALL_WAIT);
1177 }
1178
1179 static void
parse_bsd_args(void)1180 parse_bsd_args(void)
1181 {
1182 char namep[48];
1183
1184 if (PE_parse_boot_argn("-s", namep, sizeof(namep))) {
1185 boothowto |= RB_SINGLE;
1186 }
1187
1188 if (PE_parse_boot_argn("-x", namep, sizeof(namep))) { /* safe boot */
1189 boothowto |= RB_SAFEBOOT;
1190 }
1191
1192 if (PE_parse_boot_argn("nbuf", &max_nbuf_headers,
1193 sizeof(max_nbuf_headers))) {
1194 customnbuf = 1;
1195 }
1196
1197 #if CONFIG_DARKBOOT
1198 /*
1199 * The darkboot flag is specified by the bootloader and is stored in
1200 * boot_args->bootFlags. This flag is available starting revision 2.
1201 */
1202 boot_args *args = (boot_args *) PE_state.bootArgs;
1203 if ((args != NULL) && (args->Revision >= kBootArgsRevision2)) {
1204 darkboot = (args->bootFlags & kBootFlagsDarkBoot) ? 1 : 0;
1205 } else {
1206 darkboot = 0;
1207 }
1208 #endif
1209
1210 #if DEVELOPMENT || DEBUG
1211 if (PE_parse_boot_argn("dyldsuffix", dyld_suffix, sizeof(dyld_suffix))) {
1212 if (strlen(dyld_suffix) > 0) {
1213 use_dyld_suffix = 1;
1214 }
1215 }
1216
1217 if (PE_parse_boot_argn("alt-dyld", dyld_alt_path, sizeof(dyld_alt_path))) {
1218 if (strlen(dyld_alt_path) > 0) {
1219 use_alt_dyld = 1;
1220 }
1221 }
1222
1223 if (PE_parse_boot_arg_str("panic-on-proc-crash", panic_on_proc_crash, sizeof(panic_on_proc_crash))) {
1224 if (strlen(panic_on_proc_crash) > 0) {
1225 use_panic_on_proc_crash = 1;
1226 }
1227 }
1228
1229 if (PE_parse_boot_arg_str("panic-on-proc-exit", panic_on_proc_exit, sizeof(panic_on_proc_exit))) {
1230 if (strlen(panic_on_proc_exit) > 0) {
1231 use_panic_on_proc_exit = 1;
1232 }
1233 }
1234
1235 if (PE_parse_boot_arg_str("panic-on-proc-spawn-fail", panic_on_proc_spawn_fail, sizeof(panic_on_proc_spawn_fail))) {
1236 if (strlen(panic_on_proc_spawn_fail) > 0) {
1237 use_panic_on_proc_spawn_fail = 1;
1238 }
1239 }
1240 #endif /* DEVELOPMENT || DEBUG */
1241 }
1242 STARTUP(TUNABLES, STARTUP_RANK_MIDDLE, parse_bsd_args);
1243
1244 #if CONFIG_BASESYSTEMROOT
1245
1246 extern bool IOGetBootUUID(char *);
1247 extern bool IOGetApfsPrebootUUID(char *);
1248
1249
1250 // This function returns the UUID of the Preboot (and Recovery) folder associated with the
1251 // current boot volume, if applicable. The meaning of the UUID can be
1252 // filesystem-dependent and not all kinds of boots will have a UUID.
1253 // On success, the UUID is copied into the past-in parameter and TRUE is returned.
1254 // In case the current boot has no applicable Preboot UUID, FALSE is returned.
1255 static bool
get_preboot_uuid(uuid_string_t maybe_uuid_string)1256 get_preboot_uuid(uuid_string_t maybe_uuid_string)
1257 {
1258 // try IOGetApfsPrebootUUID
1259 if (IOGetApfsPrebootUUID(maybe_uuid_string)) {
1260 uuid_t maybe_uuid;
1261 int error = uuid_parse(maybe_uuid_string, maybe_uuid);
1262 if (error == 0) {
1263 return true;
1264 }
1265 }
1266
1267 // try IOGetBootUUID
1268 if (IOGetBootUUID(maybe_uuid_string)) {
1269 uuid_t maybe_uuid;
1270 int error = uuid_parse(maybe_uuid_string, maybe_uuid);
1271 if (error == 0) {
1272 return true;
1273 }
1274 }
1275
1276 // didn't find it
1277 return false;
1278 }
1279
1280 #if defined(__arm64__)
1281 extern bool IOGetBootObjectsPath(char *);
1282 #endif
1283
1284 // Find the BaseSystem.dmg to be used as the initial root volume during certain
1285 // kinds of boots.
1286 // This may mount volumes and lookup vnodes.
1287 // The DEVELOPMENT kernel will look for BaseSystem.rooted.dmg first.
1288 // If it returns 0 (no error), then it also writes the absolute path to the
1289 // BaseSystem.dmg into its argument (which must be a char[MAXPATHLEN]).
1290 static
1291 int
bsd_find_basesystem_dmg(char * bsdmgpath_out,bool * rooted_dmg,bool * skip_signature_check)1292 bsd_find_basesystem_dmg(char *bsdmgpath_out, bool *rooted_dmg, bool *skip_signature_check)
1293 {
1294 int error;
1295 size_t len;
1296 char *dmgbasepath;
1297 char *dmgpath;
1298 bool allow_rooted_dmg = false;
1299
1300 dmgbasepath = zalloc_flags(ZV_NAMEI, Z_ZERO | Z_WAITOK);
1301 dmgpath = zalloc_flags(ZV_NAMEI, Z_ZERO | Z_WAITOK);
1302 vnode_t imagevp = NULLVP;
1303
1304 #if DEVELOPMENT || DEBUG
1305 allow_rooted_dmg = true;
1306 #endif
1307
1308 //must provide output bool
1309 if (rooted_dmg && skip_signature_check) {
1310 *rooted_dmg = false;
1311 *skip_signature_check = false;
1312 } else {
1313 error = EINVAL;
1314 goto done;
1315 }
1316
1317 error = vfs_mount_recovery();
1318 if (error) {
1319 goto done;
1320 }
1321
1322 len = strlcpy(dmgbasepath, "/System/Volumes/Recovery/", MAXPATHLEN);
1323 if (len > MAXPATHLEN) {
1324 error = ENAMETOOLONG;
1325 goto done;
1326 }
1327
1328 if (csr_check(CSR_ALLOW_ANY_RECOVERY_OS) == 0) {
1329 *skip_signature_check = true;
1330 allow_rooted_dmg = true;
1331 }
1332
1333 #if defined(__arm64__)
1334 char boot_obj_path[MAXPATHLEN] = "";
1335
1336 if (IOGetBootObjectsPath(boot_obj_path)) {
1337 if (boot_obj_path[0] == '/') {
1338 dmgbasepath[len - 1] = '\0';
1339 }
1340
1341 len = strlcat(dmgbasepath, boot_obj_path, MAXPATHLEN);
1342 if (len > MAXPATHLEN) {
1343 error = ENAMETOOLONG;
1344 goto done;
1345 }
1346
1347 len = strlcat(dmgbasepath, "/usr/standalone/firmware/", MAXPATHLEN);
1348 if (len > MAXPATHLEN) {
1349 error = ENAMETOOLONG;
1350 goto done;
1351 }
1352
1353 if (allow_rooted_dmg) {
1354 len = strlcpy(dmgpath, dmgbasepath, MAXPATHLEN);
1355 if (len > MAXPATHLEN) {
1356 error = ENAMETOOLONG;
1357 goto done;
1358 }
1359
1360 len = strlcat(dmgpath, "arm64eBaseSystem.rooted.dmg", MAXPATHLEN);
1361 if (len > MAXPATHLEN) {
1362 error = ENAMETOOLONG;
1363 goto done;
1364 }
1365
1366 error = vnode_lookup(dmgpath, 0, &imagevp, vfs_context_kernel());
1367 if (error == 0) {
1368 *rooted_dmg = true;
1369 *skip_signature_check = true;
1370 goto done;
1371 }
1372 memset(dmgpath, 0, MAXPATHLEN);
1373 }
1374
1375 len = strlcpy(dmgpath, dmgbasepath, MAXPATHLEN);
1376 if (len > MAXPATHLEN) {
1377 error = ENAMETOOLONG;
1378 goto done;
1379 }
1380
1381 len = strlcat(dmgpath, "arm64eBaseSystem.dmg", MAXPATHLEN);
1382 if (len > MAXPATHLEN) {
1383 error = ENAMETOOLONG;
1384 goto done;
1385 }
1386
1387 error = vnode_lookup(dmgpath, 0, &imagevp, vfs_context_kernel());
1388 if (error == 0) {
1389 goto done;
1390 }
1391 memset(dmgpath, 0, MAXPATHLEN);
1392 dmgbasepath[strlen("/System/Volumes/Recovery/")] = '\0';
1393 }
1394 #endif // __arm64__
1395
1396 uuid_string_t preboot_uuid;
1397 if (!get_preboot_uuid(preboot_uuid)) {
1398 // no preboot? bail out
1399 return EINVAL;
1400 }
1401
1402 len = strlcat(dmgbasepath, preboot_uuid, MAXPATHLEN);
1403 if (len > MAXPATHLEN) {
1404 error = ENAMETOOLONG;
1405 goto done;
1406 }
1407
1408 if (allow_rooted_dmg) {
1409 // Try BaseSystem.rooted.dmg
1410 len = strlcpy(dmgpath, dmgbasepath, MAXPATHLEN);
1411 if (len > MAXPATHLEN) {
1412 error = ENAMETOOLONG;
1413 goto done;
1414 }
1415
1416 len = strlcat(dmgpath, "/BaseSystem.rooted.dmg", MAXPATHLEN);
1417 if (len > MAXPATHLEN) {
1418 error = ENAMETOOLONG;
1419 goto done;
1420 }
1421
1422 error = vnode_lookup(dmgpath, 0, &imagevp, vfs_context_kernel());
1423 if (error == 0) {
1424 // we found it! success!
1425 *rooted_dmg = true;
1426 *skip_signature_check = true;
1427 goto done;
1428 }
1429 }
1430
1431 // Try BaseSystem.dmg
1432 len = strlcpy(dmgpath, dmgbasepath, MAXPATHLEN);
1433 if (len > MAXPATHLEN) {
1434 error = ENAMETOOLONG;
1435 goto done;
1436 }
1437
1438 len = strlcat(dmgpath, "/BaseSystem.dmg", MAXPATHLEN);
1439 if (len > MAXPATHLEN) {
1440 error = ENAMETOOLONG;
1441 goto done;
1442 }
1443
1444 error = vnode_lookup(dmgpath, 0, &imagevp, vfs_context_kernel());
1445 if (error == 0) {
1446 // success!
1447 goto done;
1448 }
1449
1450 done:
1451 if (error == 0) {
1452 strlcpy(bsdmgpath_out, dmgpath, MAXPATHLEN);
1453 } else {
1454 bsd_init_kprintf("%s: error %d\n", __func__, error);
1455 }
1456 if (imagevp != NULLVP) {
1457 vnode_put(imagevp);
1458 }
1459 zfree(ZV_NAMEI, dmgpath);
1460 zfree(ZV_NAMEI, dmgbasepath);
1461 return error;
1462 }
1463
1464 static boolean_t
bsdmgroot_bootable(void)1465 bsdmgroot_bootable(void)
1466 {
1467 #if defined(__arm64__)
1468 #define BSDMGROOT_DEFAULT true
1469 #else
1470 #define BSDMGROOT_DEFAULT false
1471 #endif
1472
1473 boolean_t resolved = BSDMGROOT_DEFAULT;
1474
1475 boolean_t boot_arg_bsdmgroot = false;
1476 boolean_t boot_arg_nobsdmgroot = false;
1477 int error;
1478 mount_t mp;
1479 boolean_t root_part_of_volume_group = false;
1480 struct vfs_attr vfsattr;
1481
1482 mp = rootvnode->v_mount;
1483 VFSATTR_INIT(&vfsattr);
1484 VFSATTR_WANTED(&vfsattr, f_capabilities);
1485
1486 boot_arg_bsdmgroot = PE_parse_boot_argn("-bsdmgroot", NULL, 0);
1487 boot_arg_nobsdmgroot = PE_parse_boot_argn("-nobsdmgroot", NULL, 0);
1488
1489 error = vfs_getattr(mp, &vfsattr, vfs_context_kernel());
1490 if (!error && VFSATTR_IS_SUPPORTED(&vfsattr, f_capabilities)) {
1491 if ((vfsattr.f_capabilities.capabilities[VOL_CAPABILITIES_FORMAT] & VOL_CAP_FMT_VOL_GROUPS) &&
1492 (vfsattr.f_capabilities.valid[VOL_CAPABILITIES_FORMAT] & VOL_CAP_FMT_VOL_GROUPS)) {
1493 root_part_of_volume_group = true;
1494 }
1495 }
1496
1497 boolean_t singleuser = (boothowto & RB_SINGLE) != 0;
1498
1499 // Start with the #defined default above.
1500 // If booting to single-user mode, default to false, because single-
1501 // user mode inside the BaseSystem is probably not what's wanted.
1502 // If the 'yes' boot-arg is set, we'll allow that even in single-user
1503 // mode, we'll assume you know what you're doing.
1504 // The 'no' boot-arg overpowers the 'yes' boot-arg.
1505 // In any case, we will not attempt to root from BaseSystem if the
1506 // original (booter-chosen) root volume isn't in a volume group.
1507 // This is just out of an abundance of caution: if the boot environment
1508 // seems to be "something other than a standard install",
1509 // we'll be conservative in messing with the root volume.
1510
1511 if (singleuser) {
1512 resolved = false;
1513 }
1514
1515 if (boot_arg_bsdmgroot) {
1516 resolved = true;
1517 }
1518
1519 if (boot_arg_nobsdmgroot) {
1520 resolved = false;
1521 }
1522
1523 if (!root_part_of_volume_group) {
1524 resolved = false;
1525 }
1526
1527 return resolved;
1528 }
1529 #endif // CONFIG_BASESYSTEMROOT
1530
1531 void
bsd_exec_setup(int scale)1532 bsd_exec_setup(int scale)
1533 {
1534 switch (scale) {
1535 case 0:
1536 case 1:
1537 bsd_simul_execs = BSD_SIMUL_EXECS;
1538 break;
1539 case 2:
1540 case 3:
1541 bsd_simul_execs = 65;
1542 break;
1543 case 4:
1544 case 5:
1545 bsd_simul_execs = 129;
1546 break;
1547 case 6:
1548 case 7:
1549 bsd_simul_execs = 257;
1550 break;
1551 default:
1552 bsd_simul_execs = 513;
1553 break;
1554 }
1555 bsd_pageable_map_size = (bsd_simul_execs * BSD_PAGEABLE_SIZE_PER_EXEC);
1556 }
1557
1558 #if !CONFIG_NETBOOT
1559 int
1560 netboot_root(void);
1561
1562 int
netboot_root(void)1563 netboot_root(void)
1564 {
1565 return 0;
1566 }
1567 #endif
1568