xref: /xnu-8020.121.3/bsd/kern/bsd_init.c (revision fdd8201d7b966f0c3ea610489d29bd841d358941)
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 <net/remote_vif.h>             /* for rvi_init() */
162 #include <kern/assert.h>                /* for assert() */
163 #include <sys/kern_overrides.h>         /* for init_system_override() */
164 #include <sys/lockf.h>                  /* for lf_init() */
165 #include <sys/fsctl.h>
166 
167 #include <net/init.h>
168 
169 #if CONFIG_MACF
170 #include <security/mac_framework.h>
171 #include <security/mac_internal.h>      /* mac_init_bsd() */
172 #include <security/mac_mach_internal.h> /* mac_update_task_label() */
173 #endif
174 
175 #include <machine/exec.h>
176 
177 #if CONFIG_NETBOOT
178 #include <sys/netboot.h>
179 #endif
180 
181 #if CONFIG_IMAGEBOOT
182 #include <sys/imageboot.h>
183 #endif
184 
185 #if PFLOG
186 #include <net/if_pflog.h>
187 #endif
188 
189 #if SKYWALK
190 #include <skywalk/os_skywalk_private.h>
191 #endif /* SKYWALK */
192 
193 #include <pexpert/pexpert.h>
194 #include <machine/pal_routines.h>
195 #include <console/video_console.h>
196 
197 #if CONFIG_XNUPOST
198 #include <tests/xnupost.h>
199 #endif
200 
201 void * get_user_regs(thread_t);         /* XXX kludge for <machine/thread.h> */
202 void IOKitInitializeTime(void);         /* XXX */
203 void IOSleep(unsigned int);             /* XXX */
204 void IOSetImageBoot(void);              /* XXX */
205 void loopattach(void);                  /* XXX */
206 
207 const char *const copyright =
208     "Copyright (c) 1982, 1986, 1989, 1991, 1993\n\t"
209     "The Regents of the University of California. "
210     "All rights reserved.\n\n";
211 
212 /* Components of the first process -- never freed. */
213 SECURITY_READ_ONLY_LATE(struct vfs_context) vfs_context0;
214 
215 struct proc proc0 = {
216 	.p_comm    = "kernel_task",
217 	.p_name    = "kernel_task",
218 	.p_pptr    = &proc0,
219 	.p_stat    = SRUN,
220 #if defined(__LP64__)
221 	.p_flag    = P_SYSTEM | P_LP64,
222 #else
223 	.p_flag    = P_SYSTEM,
224 #endif
225 	.p_nice    = NZERO,
226 	.p_uthlist = TAILQ_HEAD_INITIALIZER(proc0.p_uthlist),
227 };
228 static struct plimit limit0;
229 static struct pstats pstats0;
230 SECURITY_READ_ONLY_LATE(proc_t) kernproc = &proc0;
231 proc_t XNU_PTRAUTH_SIGNED_PTR("initproc") initproc;
232 
233 long tk_cancc;
234 long tk_nin;
235 long tk_nout;
236 long tk_rawcc;
237 
238 int lock_trace = 0;
239 /* Global variables to make pstat happy. We do swapping differently */
240 int nswdev, nswap;
241 int nswapmap;
242 void *swapmap;
243 struct swdevt swdevt[1];
244 
245 static LCK_GRP_DECLARE(hostname_lck_grp, "hostname");
246 LCK_MTX_DECLARE(hostname_lock, &hostname_lck_grp);
247 LCK_MTX_DECLARE(domainname_lock, &hostname_lck_grp);
248 
249 dev_t   rootdev;                /* device of the root */
250 dev_t   dumpdev;                /* device to take dumps on */
251 long    dumplo;                 /* offset into dumpdev */
252 long    hostid;
253 char    hostname[MAXHOSTNAMELEN];
254 char    domainname[MAXDOMNAMELEN];
255 char    rootdevice[DEVMAXNAMESIZE];
256 
257 struct  vnode *rootvp;
258 bool rootvp_is_ssd = false;
259 SECURITY_READ_ONLY_LATE(int) boothowto;
260 /*
261  * -minimalboot indicates that we want userspace to be bootstrapped to a
262  * minimal environment.  What constitutes minimal is up to the bootstrap
263  * process.
264  */
265 TUNABLE(int, minimalboot, "-minimalboot", 0);
266 #if CONFIG_DARKBOOT
267 int darkboot = 0;
268 #endif
269 
270 extern kern_return_t IOFindBSDRoot(char *, unsigned int, dev_t *, u_int32_t *);
271 extern void IOSecureBSDRoot(const char * rootName);
272 extern kern_return_t IOKitBSDInit(void );
273 extern boolean_t IOSetRecoveryBoot(bsd_bootfail_mode_t, uuid_t, boolean_t);
274 extern void kminit(void);
275 extern void bsd_bufferinit(void);
276 extern void throttle_init(void);
277 
278 vm_map_t        bsd_pageable_map;
279 vm_map_t        mb_map;
280 
281 static  int bsd_simul_execs;
282 static int bsd_pageable_map_size;
283 __private_extern__ int execargs_cache_size = 0;
284 __private_extern__ int execargs_free_count = 0;
285 __private_extern__ vm_offset_t * execargs_cache = NULL;
286 
287 void bsd_exec_setup(int);
288 
289 __private_extern__ int bootarg_execfailurereports = 0;
290 
291 #if __x86_64__
292 __private_extern__ TUNABLE(int, bootarg_no32exec, "no32exec", 1);
293 #endif
294 
295 #if DEVELOPMENT || DEBUG
296 /* Prevent kernel-based ASLR from being used. */
297 __private_extern__ TUNABLE(bool, bootarg_disable_aslr, "-disable_aslr", 0);
298 #endif
299 
300 /*
301  * Allow an alternate dyld to be used for testing.
302  */
303 
304 #if DEVELOPMENT || DEBUG
305 char dyld_alt_path[MAXPATHLEN];
306 int use_alt_dyld = 0;
307 #endif
308 
309 int     cmask = CMASK;
310 extern int customnbuf;
311 
312 kern_return_t bsd_autoconf(void);
313 void bsd_utaskbootstrap(void);
314 
315 #if CONFIG_DEV_KMEM
316 extern void dev_kmem_init(void);
317 #endif
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 SECURITY_READ_ONLY_LATE(struct kmem_range) bsd_pageable_range = {};
407 KMEM_RANGE_REGISTER_DYNAMIC(bsd_pageable, &bsd_pageable_range, ^() {
408 	assert(bsd_pageable_map_size != 0);
409 	return (vm_map_size_t) bsd_pageable_map_size;
410 });
411 
412 /* called with an iocount and usecount on new_rootvnode */
413 void
set_rootvnode(vnode_t new_rootvnode)414 set_rootvnode(vnode_t new_rootvnode)
415 {
416 	mount_t new_mount = (new_rootvnode != NULL) ? new_rootvnode->v_mount : NULL;
417 	vnode_t new_devvp = (new_mount != NULL) ? new_mount->mnt_devvp : NULL;
418 	vnode_t old_rootvnode = rootvnode;
419 
420 	new_rootvnode->v_flag |= VROOT;
421 	rootvp = new_devvp;
422 	rootvnode = new_rootvnode;
423 	kernproc->p_fd.fd_cdir = new_rootvnode;
424 	if (new_devvp != NULL) {
425 		rootdev = vnode_specrdev(new_devvp);
426 	} else if (new_mount != NULL) {
427 		rootdev = vfs_statfs(new_mount)->f_fsid.val[0];  /* like ATTR_CMN_DEVID */
428 	} else {
429 		rootdev = NODEV;
430 	}
431 
432 	if (old_rootvnode) {
433 		vnode_rele(old_rootvnode);
434 	}
435 }
436 
437 #define RAMDEV "md0"
438 
439 bool
bsd_rooted_ramdisk(void)440 bsd_rooted_ramdisk(void)
441 {
442 	bool is_ramdisk = false;
443 	char *dev_path = zalloc(ZV_NAMEI);
444 	if (dev_path == NULL) {
445 		panic("failed to allocate devpath string!");
446 	}
447 
448 	if (PE_parse_boot_argn("rd", dev_path, MAXPATHLEN)) {
449 		if (strncmp(dev_path, RAMDEV, strlen(RAMDEV)) == 0) {
450 			is_ramdisk = true;
451 		}
452 	}
453 
454 	zfree(ZV_NAMEI, dev_path);
455 	return is_ramdisk;
456 }
457 
458 /*
459  * This function is called very early on in the Mach startup, from the
460  * function start_kernel_threads() in osfmk/kern/startup.c.  It's called
461  * in the context of the current (startup) task using a call to the
462  * function kernel_thread_create() to jump into start_kernel_threads().
463  * Internally, kernel_thread_create() calls thread_create_internal(),
464  * which calls uthread_init().  The function of uthread_init() is
465  * normally to init a uthread structure, and fill out the uu_sigmask,
466  * tro_ucred/tro_proc fields.  It skips filling these out in the case of the "task"
467  * being "kernel_task", because the order of operation is inverted.  To
468  * account for that, we need to manually fill in at least the contents
469  * of the tro_ucred field so that the uthread structure can be
470  * used like any other.
471  */
472 void
bsd_init(void)473 bsd_init(void)
474 {
475 	struct uthread *ut;
476 	vnode_t init_rootvnode = NULLVP;
477 	struct proc_ro_data kernproc_ro_data = {
478 		.p_csflags = CS_VALID,
479 	};
480 	struct task_ro_data kerntask_ro_data = { };
481 #if CONFIG_NETBOOT || CONFIG_IMAGEBOOT
482 	boolean_t       netboot = FALSE;
483 #endif
484 
485 #define DEBUG_BSDINIT 0
486 
487 #if DEBUG_BSDINIT
488 #define bsd_init_kprintf(x, ...) kprintf("bsd_init: " x, ## __VA_ARGS__)
489 #else
490 #define bsd_init_kprintf(x, ...)
491 #endif
492 
493 	throttle_init();
494 
495 	printf(copyright);
496 
497 #if CONFIG_DEV_KMEM
498 	bsd_init_kprintf("calling dev_kmem_init\n");
499 	dev_kmem_init();
500 #endif
501 
502 	/* Initialize kauth subsystem before instancing the first credential */
503 	bsd_init_kprintf("calling kauth_init\n");
504 	kauth_init();
505 
506 	/* kernel_task->proc = kernproc; */
507 	set_bsdtask_info(kernel_task, (void *)kernproc);
508 
509 	/* set the cred */
510 	kauth_cred_set(&kernproc_ro_data.p_ucred, vfs_context0.vc_ucred);
511 	kernproc->p_proc_ro = proc_ro_alloc(kernproc, &kernproc_ro_data,
512 	    kernel_task, &kerntask_ro_data);
513 
514 	/* give kernproc a name */
515 	bsd_init_kprintf("calling process_name\n");
516 	process_name("kernel_task", kernproc);
517 
518 	/* Allocate proc lock attribute */
519 
520 	lck_mtx_init(&kernproc->p_mlock, &proc_mlock_grp, &proc_lck_attr);
521 	lck_mtx_init(&kernproc->p_ucred_mlock, &proc_ucred_mlock_grp, &proc_lck_attr);
522 	lck_spin_init(&kernproc->p_slock, &proc_slock_grp, &proc_lck_attr);
523 
524 	/* Init the file descriptor table. */
525 	fdt_init(kernproc);
526 	kernproc->p_fd.fd_cmask = (mode_t)cmask;
527 
528 	assert(bsd_simul_execs != 0);
529 	execargs_cache_size = bsd_simul_execs;
530 	execargs_free_count = bsd_simul_execs;
531 	execargs_cache = zalloc_permanent(bsd_simul_execs * sizeof(vm_offset_t),
532 	    ZALIGN(vm_offset_t));
533 
534 	if (current_task() != kernel_task) {
535 		printf("bsd_init: We have a problem, "
536 		    "current task is not kernel task\n");
537 	}
538 
539 	bsd_init_kprintf("calling get_bsdthread_info\n");
540 	ut = current_uthread();
541 
542 #if CONFIG_MACF
543 	/*
544 	 * Initialize the MAC Framework
545 	 */
546 	mac_policy_initbsd();
547 
548 #if defined (__i386__) || defined (__x86_64__)
549 	/*
550 	 * We currently only support this on i386/x86_64, as that is the
551 	 * only lock code we have instrumented so far.
552 	 */
553 	int policy_check_flags;
554 	PE_parse_boot_argn("policy_check", &policy_check_flags, sizeof(policy_check_flags));
555 	check_policy_init(policy_check_flags);
556 #endif
557 #endif /* MAC */
558 
559 	/*
560 	 * Make a session and group
561 	 *
562 	 * No need to hold the pgrp lock,
563 	 * there are no other BSD threads yet.
564 	 */
565 	struct session *session0 = session_alloc(kernproc);
566 	struct pgrp *pgrp0 = pgrp_alloc(0, PGRP_REF_NONE);
567 	session0->s_ttypgrpid = 0;
568 	pgrp0->pg_session = session0;
569 
570 	/*
571 	 * Create process 0.
572 	 */
573 	proc_list_lock();
574 	os_ref_init_mask(&kernproc->p_refcount, P_REF_BITS, &p_refgrp, P_REF_NONE);
575 	os_ref_init_raw(&kernproc->p_waitref, &p_refgrp);
576 
577 	/*
578 	 * Make a group and session, then simulate pinsertchild(),
579 	 * adjusted for the kernel.
580 	 */
581 	pghash_insert_locked(0, pgrp0);
582 
583 	LIST_INSERT_HEAD(&pgrp0->pg_members, kernproc, p_pglist);
584 	hazard_ptr_init(&kernproc->p_pgrp, pgrp0);
585 	LIST_INSERT_HEAD(&allproc, kernproc, p_list);
586 
587 	LIST_INSERT_HEAD(SESSHASH(0), session0, s_hash);
588 	proc_list_unlock();
589 
590 	kernproc->task = kernel_task;
591 
592 #if DEVELOPMENT || DEBUG
593 	if (bootarg_disable_aslr) {
594 		kernproc->p_flag |= P_DISABLE_ASLR;
595 	}
596 #endif
597 
598 	TAILQ_INSERT_TAIL(&kernproc->p_uthlist, ut, uu_list);
599 
600 	bsd_init_kprintf("calling kauth_cred_create\n");
601 	/*
602 	 * Officially associate the kernel with vfs_context0.vc_ucred.
603 	 */
604 #if CONFIG_MACF
605 	mac_cred_label_associate_kernel(vfs_context0.vc_ucred);
606 #endif
607 	proc_update_creds_onproc(kernproc);
608 
609 	TAILQ_INIT(&kernproc->p_aio_activeq);
610 	TAILQ_INIT(&kernproc->p_aio_doneq);
611 	kernproc->p_aio_total_count = 0;
612 
613 	/* Create the limits structures. */
614 	for (uint32_t i = 0; i < ARRAY_COUNT(limit0.pl_rlimit); i++) {
615 		limit0.pl_rlimit[i].rlim_cur =
616 		    limit0.pl_rlimit[i].rlim_max = RLIM_INFINITY;
617 	}
618 	limit0.pl_rlimit[RLIMIT_NOFILE].rlim_cur = NOFILE;
619 	limit0.pl_rlimit[RLIMIT_NPROC].rlim_cur = maxprocperuid;
620 	limit0.pl_rlimit[RLIMIT_NPROC].rlim_max = maxproc;
621 	limit0.pl_rlimit[RLIMIT_STACK] = vm_initial_limit_stack;
622 	limit0.pl_rlimit[RLIMIT_DATA] = vm_initial_limit_data;
623 	limit0.pl_rlimit[RLIMIT_CORE] = vm_initial_limit_core;
624 	os_ref_init_count(&limit0.pl_refcnt, &rlimit_refgrp, 1);
625 
626 	hazard_ptr_init(&kernproc->p_limit, &limit0);
627 	kernproc->p_stats = &pstats0;
628 	proc_sigacts_copy(kernproc, NULL);
629 	kernproc->p_subsystem_root_path = NULL;
630 
631 	/*
632 	 * Charge root for one process: launchd.
633 	 */
634 	bsd_init_kprintf("calling chgproccnt\n");
635 	(void)chgproccnt(0, 1);
636 
637 	/*
638 	 *	Allocate a kernel submap for pageable memory
639 	 *	for temporary copying (execve()).
640 	 */
641 	bsd_init_kprintf("calling kmem_suballoc\n");
642 	bsd_pageable_map = kmem_suballoc(kernel_map,
643 	    &bsd_pageable_range.min_address,
644 	    (vm_size_t)bsd_pageable_map_size,
645 	    VM_MAP_CREATE_PAGEABLE,
646 	    VM_FLAGS_FIXED_RANGE_SUBALLOC,
647 	    KMS_PERMANENT | KMS_NOFAIL,
648 	    VM_KERN_MEMORY_BSD).kmr_submap;
649 
650 	/*
651 	 * Initialize buffers and hash links for buffers
652 	 *
653 	 * SIDE EFFECT: Starts a thread for bcleanbuf_thread(), so must
654 	 *		happen after a credential has been associated with
655 	 *		the kernel task.
656 	 */
657 	bsd_init_kprintf("calling bsd_bufferinit\n");
658 	bsd_bufferinit();
659 
660 	/*
661 	 * Initialize the calendar.
662 	 */
663 	bsd_init_kprintf("calling IOKitInitializeTime\n");
664 	IOKitInitializeTime();
665 
666 	/* Initialize the file systems. */
667 	bsd_init_kprintf("calling vfsinit\n");
668 	vfsinit();
669 
670 #if CONFIG_PROC_UUID_POLICY
671 	/* Initial proc_uuid_policy subsystem */
672 	bsd_init_kprintf("calling proc_uuid_policy_init()\n");
673 	proc_uuid_policy_init();
674 #endif
675 
676 #if SOCKETS
677 	/* Initialize per-CPU cache allocator */
678 	mcache_init();
679 
680 	/* Initialize mbuf's. */
681 	bsd_init_kprintf("calling mbinit\n");
682 	mbinit();
683 	restricted_in_port_init();
684 #endif /* SOCKETS */
685 
686 	/*
687 	 * Initializes security event auditing.
688 	 * XXX: Should/could this occur later?
689 	 */
690 #if CONFIG_AUDIT
691 	bsd_init_kprintf("calling audit_init\n");
692 	audit_init();
693 #endif
694 
695 	/* Initialize kqueues */
696 	bsd_init_kprintf("calling knote_init\n");
697 	knote_init();
698 
699 	/* Initialize event handler */
700 	bsd_init_kprintf("calling eventhandler_init\n");
701 	eventhandler_init();
702 
703 	/* Initialize for async IO */
704 	bsd_init_kprintf("calling aio_init\n");
705 	aio_init();
706 
707 	pthread_init();
708 	/* POSIX Shm and Sem */
709 	bsd_init_kprintf("calling pshm_cache_init\n");
710 	pshm_cache_init();
711 	bsd_init_kprintf("calling psem_cache_init\n");
712 	psem_cache_init();
713 
714 	/*
715 	 * Initialize protocols.  Block reception of incoming packets
716 	 * until everything is ready.
717 	 */
718 #if NETWORKING
719 	bsd_init_kprintf("calling nwk_wq_init\n");
720 	nwk_wq_init();
721 	bsd_init_kprintf("calling dlil_init\n");
722 	dlil_init();
723 #endif /* NETWORKING */
724 #if SOCKETS
725 	bsd_init_kprintf("calling socketinit\n");
726 	socketinit();
727 	bsd_init_kprintf("calling domaininit\n");
728 	domaininit();
729 	iptap_init();
730 #if FLOW_DIVERT
731 	flow_divert_init();
732 #endif  /* FLOW_DIVERT */
733 #endif /* SOCKETS */
734 #if SKYWALK
735 	bsd_init_kprintf("calling skywalk_init\n");
736 	(void) skywalk_init();
737 #endif /* SKYWALK */
738 #if NETWORKING
739 #if NECP
740 	/* Initialize Network Extension Control Policies */
741 	necp_init();
742 #endif
743 	netagent_init();
744 #endif /* NETWORKING */
745 
746 #if defined (__x86_64__)
747 	hvg_bsd_init();
748 #endif /* DEBUG || DEVELOPMENT */
749 
750 #if CONFIG_FREEZE
751 #ifndef CONFIG_MEMORYSTATUS
752     #error "CONFIG_FREEZE defined without matching CONFIG_MEMORYSTATUS"
753 #endif
754 	/* Initialise background freezing */
755 	bsd_init_kprintf("calling memorystatus_freeze_init\n");
756 	memorystatus_freeze_init();
757 #endif
758 
759 #if CONFIG_MEMORYSTATUS
760 	/* Initialize kernel memory status notifications */
761 	bsd_init_kprintf("calling memorystatus_init\n");
762 	memorystatus_init();
763 #endif /* CONFIG_MEMORYSTATUS */
764 
765 	bsd_init_kprintf("calling sysctl_mib_init\n");
766 	sysctl_mib_init();
767 
768 	bsd_init_kprintf("calling bsd_autoconf\n");
769 	bsd_autoconf();
770 
771 #if CONFIG_DTRACE
772 	dtrace_postinit();
773 #endif
774 
775 	/*
776 	 * We attach the loopback interface *way* down here to ensure
777 	 * it happens after autoconf(), otherwise it becomes the
778 	 * "primary" interface.
779 	 */
780 #include <loop.h>
781 #if NLOOP > 0
782 	bsd_init_kprintf("calling loopattach\n");
783 	loopattach();                   /* XXX */
784 #endif
785 #if NGIF
786 	/* Initialize gif interface (after lo0) */
787 	gif_init();
788 #endif
789 
790 #if PFLOG
791 	/* Initialize packet filter log interface */
792 	pfloginit();
793 #endif /* PFLOG */
794 
795 #if NETHER > 0
796 	/* Register the built-in dlil ethernet interface family */
797 	bsd_init_kprintf("calling ether_family_init\n");
798 	ether_family_init();
799 #endif /* ETHER */
800 
801 #if NETWORKING
802 #if CONTENT_FILTER
803 	cfil_init();
804 #endif
805 
806 #if PACKET_MANGLER
807 	pkt_mnglr_init();
808 #endif
809 
810 	/*
811 	 * Register subsystems with kernel control handlers
812 	 */
813 	utun_register_control();
814 #if IPSEC
815 	ipsec_register_control();
816 #endif /* IPSEC */
817 	netsrc_init();
818 	nstat_init();
819 	tcp_cc_init();
820 #if MPTCP
821 	mptcp_control_register();
822 #endif /* MPTCP */
823 
824 #if REMOTE_VIF
825 	rvi_init();
826 #endif /* REMOTE_VIF */
827 
828 	/*
829 	 * The the networking stack is now initialized so it is a good time to call
830 	 * the clients that are waiting for the networking stack to be usable.
831 	 */
832 	bsd_init_kprintf("calling net_init_run\n");
833 	net_init_run();
834 #endif /* NETWORKING */
835 
836 	bsd_init_kprintf("calling inittodr\n");
837 	inittodr(0);
838 
839 	/* Mount the root file system. */
840 	while (TRUE) {
841 		int err;
842 
843 		bsd_init_kprintf("calling setconf\n");
844 		setconf();
845 #if CONFIG_NETBOOT
846 		netboot = (mountroot == netboot_mountroot);
847 #endif
848 
849 		bsd_init_kprintf("vfs_mountroot\n");
850 		if (0 == (err = vfs_mountroot())) {
851 			break;
852 		}
853 		rootdevice[0] = '\0';
854 #if CONFIG_NETBOOT
855 		if (netboot) {
856 			PE_display_icon( 0, "noroot");  /* XXX a netboot-specific icon would be nicer */
857 			vc_progress_set(FALSE, 0);
858 			for (uint32_t i = 1; 1; i *= 2) {
859 				printf("bsd_init: failed to mount network root, error %d, %s\n",
860 				    err, PE_boot_args());
861 				printf("We are hanging here...\n");
862 				IOSleep(i * 60 * 1000);
863 			}
864 			/*NOTREACHED*/
865 		}
866 #endif
867 		printf("cannot mount root, errno = %d\n", err);
868 	}
869 
870 	IOSecureBSDRoot(rootdevice);
871 
872 	mountlist.tqh_first->mnt_flag |= MNT_ROOTFS;
873 
874 	bsd_init_kprintf("calling VFS_ROOT\n");
875 	/* Get the vnode for '/'.  Set fdp->fd_fd.fd_cdir to reference it. */
876 	if (VFS_ROOT(mountlist.tqh_first, &init_rootvnode, vfs_context_kernel())) {
877 		panic("bsd_init: cannot find root vnode: %s", PE_boot_args());
878 	}
879 	(void)vnode_ref(init_rootvnode);
880 	(void)vnode_put(init_rootvnode);
881 
882 	lck_rw_lock_exclusive(&rootvnode_rw_lock);
883 	set_rootvnode(init_rootvnode);
884 	lck_rw_unlock_exclusive(&rootvnode_rw_lock);
885 	init_rootvnode = NULLVP;  /* use rootvnode after this point */
886 
887 
888 	if (!bsd_rooted_ramdisk()) {
889 		boolean_t require_rootauth = FALSE;
890 
891 #if XNU_TARGET_OS_OSX && defined(__arm64__)
892 #if CONFIG_IMAGEBOOT
893 		/* Apple Silicon MacOS */
894 		require_rootauth = !imageboot_desired();
895 #endif // CONFIG_IMAGEBOOT
896 #elif !XNU_TARGET_OS_OSX
897 		/* Non MacOS */
898 		require_rootauth = TRUE;
899 #endif // XNU_TARGET_OS_OSX && defined(__arm64__)
900 
901 		if (require_rootauth) {
902 			/* enforce sealedness */
903 			int autherr = VNOP_IOCTL(rootvnode, FSIOC_KERNEL_ROOTAUTH, NULL, 0, vfs_context_kernel());
904 			if (autherr) {
905 				panic("rootvp not authenticated after mounting");
906 			}
907 		}
908 	}
909 
910 
911 #if CONFIG_NETBOOT
912 	if (netboot) {
913 		int err;
914 
915 		netboot = TRUE;
916 		/* post mount setup */
917 		if ((err = netboot_setup()) != 0) {
918 			PE_display_icon( 0, "noroot");  /* XXX a netboot-specific icon would be nicer */
919 			vc_progress_set(FALSE, 0);
920 			for (uint32_t i = 1; 1; i *= 2) {
921 				printf("bsd_init: NetBoot could not find root, error %d: %s\n",
922 				    err, PE_boot_args());
923 				printf("We are hanging here...\n");
924 				IOSleep(i * 60 * 1000);
925 			}
926 			/*NOTREACHED*/
927 		}
928 	}
929 #endif
930 
931 
932 #if CONFIG_IMAGEBOOT
933 	/*
934 	 * See if a system disk image is present. If so, mount it and
935 	 * switch the root vnode to point to it
936 	 */
937 	imageboot_type_t imageboot_type = imageboot_needed();
938 	if (netboot == FALSE && imageboot_type) {
939 		/*
940 		 * An image was found.  No turning back: we're booted
941 		 * with a kernel from the disk image.
942 		 */
943 		bsd_init_kprintf("doing image boot: type = %d\n", imageboot_type);
944 		imageboot_setup(imageboot_type);
945 		IOSetImageBoot();
946 	}
947 
948 #endif /* CONFIG_IMAGEBOOT */
949 
950 	/* set initial time; all other resource data is  already zero'ed */
951 	microtime_with_abstime(&kernproc->p_start, &kernproc->p_stats->ps_start);
952 
953 #if DEVFS
954 	{
955 		char mounthere[] = "/dev"; /* !const because of internal casting */
956 
957 		bsd_init_kprintf("calling devfs_kernel_mount\n");
958 		devfs_kernel_mount(mounthere);
959 	}
960 #endif /* DEVFS */
961 
962 #if CONFIG_BASESYSTEMROOT
963 #if CONFIG_IMAGEBOOT
964 	if (bsdmgroot_bootable()) {
965 		int error;
966 		bool rooted_dmg = false;
967 		bool skip_signature_check = false;
968 
969 		printf("trying to find and mount BaseSystem dmg as root volume\n");
970 #if DEVELOPMENT || DEBUG
971 		printf("(set boot-arg -nobsdmgroot to avoid this)\n");
972 #endif // DEVELOPMENT || DEBUG
973 
974 		char *dmgpath = NULL;
975 		dmgpath = zalloc_flags(ZV_NAMEI, Z_ZERO | Z_WAITOK | Z_NOFAIL);
976 
977 		error = bsd_find_basesystem_dmg(dmgpath, &rooted_dmg, &skip_signature_check);
978 		if (error) {
979 			bsd_init_kprintf("failed to to find BaseSystem dmg: error = %d\n", error);
980 		} else {
981 			PE_parse_boot_argn("bsdmgpath", dmgpath, sizeof(dmgpath));
982 
983 			bsd_init_kprintf("found BaseSystem dmg at: %s\n", dmgpath);
984 
985 			error = imageboot_pivot_image(dmgpath, IMAGEBOOT_DMG, "/System/Volumes/BaseSystem", "System/Volumes/macOS", rooted_dmg, skip_signature_check);
986 			if (error) {
987 				bsd_init_kprintf("couldn't mount BaseSystem dmg: error = %d", error);
988 			} else {
989 				IOSetImageBoot();
990 			}
991 		}
992 		zfree(ZV_NAMEI, dmgpath);
993 	}
994 #else /* CONFIG_IMAGEBOOT */
995 #error CONFIG_BASESYSTEMROOT requires CONFIG_IMAGEBOOT
996 #endif /* CONFIG_IMAGEBOOT */
997 #endif /* CONFIG_BASESYSTEMROOT */
998 
999 	/* Initialize signal state for process 0. */
1000 	bsd_init_kprintf("calling siginit\n");
1001 	siginit(kernproc);
1002 
1003 	bsd_init_kprintf("calling bsd_utaskbootstrap\n");
1004 	bsd_utaskbootstrap();
1005 
1006 	pal_kernel_announce();
1007 
1008 	bsd_init_kprintf("calling mountroot_post_hook\n");
1009 
1010 #if XNU_TARGET_OS_OSX
1011 	/* invoke post-root-mount hook */
1012 	if (mountroot_post_hook != NULL) {
1013 		mountroot_post_hook();
1014 	}
1015 #endif
1016 
1017 #if 0 /* not yet */
1018 	consider_zone_gc(FALSE);
1019 #endif
1020 
1021 	/*
1022 	 * At this point, we consider the kernel "booted" enough to apply
1023 	 * stricter timeouts.
1024 	 */
1025 	machine_timeout_bsd_init();
1026 
1027 	bsd_init_kprintf("done\n");
1028 }
1029 
1030 void
bsdinit_task(void)1031 bsdinit_task(void)
1032 {
1033 	proc_t p = current_proc();
1034 
1035 	process_name("init", p);
1036 
1037 	/* Set up exception-to-signal reflection */
1038 	ux_handler_setup();
1039 
1040 #if CONFIG_MACF
1041 	mac_cred_label_associate_user(proc_ucred(p));
1042 #endif
1043 
1044 	vm_init_before_launchd();
1045 
1046 #if CONFIG_XNUPOST
1047 	int result = bsd_list_tests();
1048 	result = bsd_do_post();
1049 	if (result != 0) {
1050 		panic("bsd_do_post: Tests failed with result = 0x%08x", result);
1051 	}
1052 #endif
1053 
1054 	bsd_init_kprintf("bsd_do_post - done");
1055 
1056 	load_init_program(p);
1057 	lock_trace = 1;
1058 }
1059 
1060 kern_return_t
bsd_autoconf(void)1061 bsd_autoconf(void)
1062 {
1063 	kprintf("bsd_autoconf: calling kminit\n");
1064 	kminit();
1065 
1066 	/*
1067 	 * Early startup for bsd pseudodevices.
1068 	 */
1069 	{
1070 		struct pseudo_init *pi;
1071 
1072 		for (pi = pseudo_inits; pi->ps_func; pi++) {
1073 			(*pi->ps_func)(pi->ps_count);
1074 		}
1075 	}
1076 
1077 	return IOKitBSDInit();
1078 }
1079 
1080 
1081 #include <sys/disklabel.h>  /* for MAXPARTITIONS */
1082 
1083 static void
setconf(void)1084 setconf(void)
1085 {
1086 	u_int32_t       flags;
1087 	kern_return_t   err;
1088 
1089 	err = IOFindBSDRoot(rootdevice, sizeof(rootdevice), &rootdev, &flags);
1090 	if (err) {
1091 		printf("setconf: IOFindBSDRoot returned an error (%d);"
1092 		    "setting rootdevice to 'sd0a'.\n", err);     /* XXX DEBUG TEMP */
1093 		rootdev = makedev( 6, 0 );
1094 		strlcpy(rootdevice, "sd0a", sizeof(rootdevice));
1095 		flags = 0;
1096 	}
1097 
1098 #if CONFIG_NETBOOT
1099 	if (flags & 1) {
1100 		/* network device */
1101 		mountroot = netboot_mountroot;
1102 	} else {
1103 #endif
1104 	/* otherwise have vfs determine root filesystem */
1105 	mountroot = NULL;
1106 #if CONFIG_NETBOOT
1107 }
1108 #endif
1109 }
1110 
1111 /*
1112  * Boot into the flavor of Recovery dictated by `mode`.
1113  */
1114 boolean_t
bsd_boot_to_recovery(bsd_bootfail_mode_t mode,uuid_t volume_uuid,boolean_t reboot)1115 bsd_boot_to_recovery(bsd_bootfail_mode_t mode, uuid_t volume_uuid, boolean_t reboot)
1116 {
1117 	return IOSetRecoveryBoot(mode, volume_uuid, reboot);
1118 }
1119 
1120 void
bsd_utaskbootstrap(void)1121 bsd_utaskbootstrap(void)
1122 {
1123 	thread_t thread;
1124 	struct uthread *ut;
1125 
1126 	/*
1127 	 * Clone the bootstrap process from the kernel process, without
1128 	 * inheriting either task characteristics or memory from the kernel;
1129 	 */
1130 	thread = cloneproc(TASK_NULL, COALITION_NULL, kernproc, FALSE, TRUE);
1131 
1132 	/* Hold the reference as it will be dropped during shutdown */
1133 	initproc = proc_find(1);
1134 #if __PROC_INTERNAL_DEBUG
1135 	if (initproc == PROC_NULL) {
1136 		panic("bsd_utaskbootstrap: initproc not set");
1137 	}
1138 #endif
1139 
1140 	zalloc_first_proc_made();
1141 
1142 	/*
1143 	 * Since we aren't going back out the normal way to our parent,
1144 	 * we have to drop the transition locks explicitly.
1145 	 */
1146 	proc_signalend(initproc, 0);
1147 	proc_transend(initproc, 0);
1148 
1149 	ut = (struct uthread *)get_bsdthread_info(thread);
1150 	ut->uu_sigmask = 0;
1151 	act_set_astbsd(thread);
1152 	task_clear_return_wait(get_threadtask(thread), TCRW_CLEAR_ALL_WAIT);
1153 }
1154 
1155 static void
parse_bsd_args(void)1156 parse_bsd_args(void)
1157 {
1158 	char namep[48];
1159 
1160 	if (PE_parse_boot_argn("-s", namep, sizeof(namep))) {
1161 		boothowto |= RB_SINGLE;
1162 	}
1163 
1164 	if (PE_parse_boot_argn("-x", namep, sizeof(namep))) { /* safe boot */
1165 		boothowto |= RB_SAFEBOOT;
1166 	}
1167 
1168 	if (PE_parse_boot_argn("nbuf", &max_nbuf_headers,
1169 	    sizeof(max_nbuf_headers))) {
1170 		customnbuf = 1;
1171 	}
1172 
1173 #if CONFIG_DARKBOOT
1174 	/*
1175 	 * The darkboot flag is specified by the bootloader and is stored in
1176 	 * boot_args->bootFlags. This flag is available starting revision 2.
1177 	 */
1178 	boot_args *args = (boot_args *) PE_state.bootArgs;
1179 	if ((args != NULL) && (args->Revision >= kBootArgsRevision2)) {
1180 		darkboot = (args->bootFlags & kBootFlagsDarkBoot) ? 1 : 0;
1181 	} else {
1182 		darkboot = 0;
1183 	}
1184 #endif
1185 
1186 #if DEVELOPMENT || DEBUG
1187 	if (PE_parse_boot_argn("alt-dyld", dyld_alt_path, sizeof(dyld_alt_path))) {
1188 		if (strlen(dyld_alt_path) > 0) {
1189 			use_alt_dyld = 1;
1190 		}
1191 	}
1192 #endif /* DEVELOPMENT || DEBUG */
1193 }
1194 STARTUP(TUNABLES, STARTUP_RANK_MIDDLE, parse_bsd_args);
1195 
1196 #if CONFIG_BASESYSTEMROOT
1197 
1198 extern bool IOGetBootUUID(char *);
1199 extern bool IOGetApfsPrebootUUID(char *);
1200 
1201 
1202 // This function returns the UUID of the Preboot (and Recovery) folder associated with the
1203 // current boot volume, if applicable. The meaning of the UUID can be
1204 // filesystem-dependent and not all kinds of boots will have a UUID.
1205 // On success, the UUID is copied into the past-in parameter and TRUE is returned.
1206 // In case the current boot has no applicable Preboot UUID, FALSE is returned.
1207 static bool
get_preboot_uuid(uuid_string_t maybe_uuid_string)1208 get_preboot_uuid(uuid_string_t maybe_uuid_string)
1209 {
1210 	// try IOGetApfsPrebootUUID
1211 	if (IOGetApfsPrebootUUID(maybe_uuid_string)) {
1212 		uuid_t maybe_uuid;
1213 		int error = uuid_parse(maybe_uuid_string, maybe_uuid);
1214 		if (error == 0) {
1215 			return true;
1216 		}
1217 	}
1218 
1219 	// try IOGetBootUUID
1220 	if (IOGetBootUUID(maybe_uuid_string)) {
1221 		uuid_t maybe_uuid;
1222 		int error = uuid_parse(maybe_uuid_string, maybe_uuid);
1223 		if (error == 0) {
1224 			return true;
1225 		}
1226 	}
1227 
1228 	// didn't find it
1229 	return false;
1230 }
1231 
1232 #if defined(__arm64__)
1233 extern bool IOGetBootObjectsPath(char *);
1234 #endif
1235 
1236 // Find the BaseSystem.dmg to be used as the initial root volume during certain
1237 // kinds of boots.
1238 // This may mount volumes and lookup vnodes.
1239 // The DEVELOPMENT kernel will look for BaseSystem.rooted.dmg first.
1240 // If it returns 0 (no error), then it also writes the absolute path to the
1241 // BaseSystem.dmg into its argument (which must be a char[MAXPATHLEN]).
1242 static
1243 int
bsd_find_basesystem_dmg(char * bsdmgpath_out,bool * rooted_dmg,bool * skip_signature_check)1244 bsd_find_basesystem_dmg(char *bsdmgpath_out, bool *rooted_dmg, bool *skip_signature_check)
1245 {
1246 	int error;
1247 	size_t len;
1248 	char *dmgbasepath;
1249 	char *dmgpath;
1250 	bool allow_rooted_dmg = false;
1251 
1252 	dmgbasepath = zalloc_flags(ZV_NAMEI, Z_ZERO | Z_WAITOK);
1253 	dmgpath = zalloc_flags(ZV_NAMEI, Z_ZERO | Z_WAITOK);
1254 	vnode_t imagevp = NULLVP;
1255 
1256 #if DEVELOPMENT || DEBUG
1257 	allow_rooted_dmg = true;
1258 #endif
1259 
1260 	//must provide output bool
1261 	if (rooted_dmg && skip_signature_check) {
1262 		*rooted_dmg = false;
1263 		*skip_signature_check = false;
1264 	} else {
1265 		error = EINVAL;
1266 		goto done;
1267 	}
1268 
1269 	error = vfs_mount_recovery();
1270 	if (error) {
1271 		goto done;
1272 	}
1273 
1274 	len = strlcpy(dmgbasepath, "/System/Volumes/Recovery/", MAXPATHLEN);
1275 	if (len > MAXPATHLEN) {
1276 		error = ENAMETOOLONG;
1277 		goto done;
1278 	}
1279 
1280 	if (csr_check(CSR_ALLOW_ANY_RECOVERY_OS) == 0) {
1281 		*skip_signature_check = true;
1282 		allow_rooted_dmg = true;
1283 	}
1284 
1285 #if defined(__arm64__)
1286 	char boot_obj_path[MAXPATHLEN] = "";
1287 
1288 	if (IOGetBootObjectsPath(boot_obj_path)) {
1289 		if (boot_obj_path[0] == '/') {
1290 			dmgbasepath[len - 1] = '\0';
1291 		}
1292 
1293 		len = strlcat(dmgbasepath, boot_obj_path, MAXPATHLEN);
1294 		if (len > MAXPATHLEN) {
1295 			error = ENAMETOOLONG;
1296 			goto done;
1297 		}
1298 
1299 		len = strlcat(dmgbasepath, "/usr/standalone/firmware/", MAXPATHLEN);
1300 		if (len > MAXPATHLEN) {
1301 			error = ENAMETOOLONG;
1302 			goto done;
1303 		}
1304 
1305 		if (allow_rooted_dmg) {
1306 			len = strlcpy(dmgpath, dmgbasepath, MAXPATHLEN);
1307 			if (len > MAXPATHLEN) {
1308 				error = ENAMETOOLONG;
1309 				goto done;
1310 			}
1311 
1312 			len = strlcat(dmgpath, "arm64eBaseSystem.rooted.dmg", MAXPATHLEN);
1313 			if (len > MAXPATHLEN) {
1314 				error = ENAMETOOLONG;
1315 				goto done;
1316 			}
1317 
1318 			error = vnode_lookup(dmgpath, 0, &imagevp, vfs_context_kernel());
1319 			if (error == 0) {
1320 				*rooted_dmg = true;
1321 				*skip_signature_check = true;
1322 				goto done;
1323 			}
1324 			memset(dmgpath, 0, MAXPATHLEN);
1325 		}
1326 
1327 		len = strlcpy(dmgpath, dmgbasepath, MAXPATHLEN);
1328 		if (len > MAXPATHLEN) {
1329 			error = ENAMETOOLONG;
1330 			goto done;
1331 		}
1332 
1333 		len = strlcat(dmgpath, "arm64eBaseSystem.dmg", MAXPATHLEN);
1334 		if (len > MAXPATHLEN) {
1335 			error = ENAMETOOLONG;
1336 			goto done;
1337 		}
1338 
1339 		error = vnode_lookup(dmgpath, 0, &imagevp, vfs_context_kernel());
1340 		if (error == 0) {
1341 			goto done;
1342 		}
1343 		memset(dmgpath, 0, MAXPATHLEN);
1344 		dmgbasepath[strlen("/System/Volumes/Recovery/")] = '\0';
1345 	}
1346 #endif // __arm64__
1347 
1348 	uuid_string_t preboot_uuid;
1349 	if (!get_preboot_uuid(preboot_uuid)) {
1350 		// no preboot? bail out
1351 		return EINVAL;
1352 	}
1353 
1354 	len = strlcat(dmgbasepath, preboot_uuid, MAXPATHLEN);
1355 	if (len > MAXPATHLEN) {
1356 		error = ENAMETOOLONG;
1357 		goto done;
1358 	}
1359 
1360 	if (allow_rooted_dmg) {
1361 		// Try BaseSystem.rooted.dmg
1362 		len = strlcpy(dmgpath, dmgbasepath, MAXPATHLEN);
1363 		if (len > MAXPATHLEN) {
1364 			error = ENAMETOOLONG;
1365 			goto done;
1366 		}
1367 
1368 		len = strlcat(dmgpath, "/BaseSystem.rooted.dmg", MAXPATHLEN);
1369 		if (len > MAXPATHLEN) {
1370 			error = ENAMETOOLONG;
1371 			goto done;
1372 		}
1373 
1374 		error = vnode_lookup(dmgpath, 0, &imagevp, vfs_context_kernel());
1375 		if (error == 0) {
1376 			// we found it! success!
1377 			*rooted_dmg = true;
1378 			*skip_signature_check = true;
1379 			goto done;
1380 		}
1381 	}
1382 
1383 	// Try BaseSystem.dmg
1384 	len = strlcpy(dmgpath, dmgbasepath, MAXPATHLEN);
1385 	if (len > MAXPATHLEN) {
1386 		error = ENAMETOOLONG;
1387 		goto done;
1388 	}
1389 
1390 	len = strlcat(dmgpath, "/BaseSystem.dmg", MAXPATHLEN);
1391 	if (len > MAXPATHLEN) {
1392 		error = ENAMETOOLONG;
1393 		goto done;
1394 	}
1395 
1396 	error = vnode_lookup(dmgpath, 0, &imagevp, vfs_context_kernel());
1397 	if (error == 0) {
1398 		// success!
1399 		goto done;
1400 	}
1401 
1402 done:
1403 	if (error == 0) {
1404 		strlcpy(bsdmgpath_out, dmgpath, MAXPATHLEN);
1405 	} else {
1406 		bsd_init_kprintf("%s: error %d\n", __func__, error);
1407 	}
1408 	if (imagevp != NULLVP) {
1409 		vnode_put(imagevp);
1410 	}
1411 	zfree(ZV_NAMEI, dmgpath);
1412 	zfree(ZV_NAMEI, dmgbasepath);
1413 	return error;
1414 }
1415 
1416 static boolean_t
bsdmgroot_bootable(void)1417 bsdmgroot_bootable(void)
1418 {
1419 #if defined(__arm64__)
1420 #define BSDMGROOT_DEFAULT true
1421 #else
1422 #define BSDMGROOT_DEFAULT false
1423 #endif
1424 
1425 	boolean_t resolved = BSDMGROOT_DEFAULT;
1426 
1427 	boolean_t boot_arg_bsdmgroot = false;
1428 	boolean_t boot_arg_nobsdmgroot = false;
1429 	int error;
1430 	mount_t mp;
1431 	boolean_t root_part_of_volume_group = false;
1432 	struct vfs_attr vfsattr;
1433 
1434 	mp = rootvnode->v_mount;
1435 	VFSATTR_INIT(&vfsattr);
1436 	VFSATTR_WANTED(&vfsattr, f_capabilities);
1437 
1438 	boot_arg_bsdmgroot = PE_parse_boot_argn("-bsdmgroot", NULL, 0);
1439 	boot_arg_nobsdmgroot = PE_parse_boot_argn("-nobsdmgroot", NULL, 0);
1440 
1441 	error = vfs_getattr(mp, &vfsattr, vfs_context_kernel());
1442 	if (!error && VFSATTR_IS_SUPPORTED(&vfsattr, f_capabilities)) {
1443 		if ((vfsattr.f_capabilities.capabilities[VOL_CAPABILITIES_FORMAT] & VOL_CAP_FMT_VOL_GROUPS) &&
1444 		    (vfsattr.f_capabilities.valid[VOL_CAPABILITIES_FORMAT] & VOL_CAP_FMT_VOL_GROUPS)) {
1445 			root_part_of_volume_group = true;
1446 		}
1447 	}
1448 
1449 	boolean_t singleuser = (boothowto & RB_SINGLE) != 0;
1450 
1451 	// Start with the #defined default above.
1452 	// If booting to single-user mode, default to false, because single-
1453 	// user mode inside the BaseSystem is probably not what's wanted.
1454 	// If the 'yes' boot-arg is set, we'll allow that even in single-user
1455 	// mode, we'll assume you know what you're doing.
1456 	// The 'no' boot-arg overpowers the 'yes' boot-arg.
1457 	// In any case, we will not attempt to root from BaseSystem if the
1458 	// original (booter-chosen) root volume isn't in a volume group.
1459 	// This is just out of an abundance of caution: if the boot environment
1460 	// seems to be "something other than a standard install",
1461 	// we'll be conservative in messing with the root volume.
1462 
1463 	if (singleuser) {
1464 		resolved = false;
1465 	}
1466 
1467 	if (boot_arg_bsdmgroot) {
1468 		resolved = true;
1469 	}
1470 
1471 	if (boot_arg_nobsdmgroot) {
1472 		resolved = false;
1473 	}
1474 
1475 	if (!root_part_of_volume_group) {
1476 		resolved = false;
1477 	}
1478 
1479 	return resolved;
1480 }
1481 #endif // CONFIG_BASESYSTEMROOT
1482 
1483 void
bsd_exec_setup(int scale)1484 bsd_exec_setup(int scale)
1485 {
1486 	switch (scale) {
1487 	case 0:
1488 	case 1:
1489 		bsd_simul_execs = BSD_SIMUL_EXECS;
1490 		break;
1491 	case 2:
1492 	case 3:
1493 		bsd_simul_execs = 65;
1494 		break;
1495 	case 4:
1496 	case 5:
1497 		bsd_simul_execs = 129;
1498 		break;
1499 	case 6:
1500 	case 7:
1501 		bsd_simul_execs = 257;
1502 		break;
1503 	default:
1504 		bsd_simul_execs = 513;
1505 		break;
1506 	}
1507 	bsd_pageable_map_size = (bsd_simul_execs * BSD_PAGEABLE_SIZE_PER_EXEC);
1508 }
1509 
1510 #if !CONFIG_NETBOOT
1511 int
1512 netboot_root(void);
1513 
1514 int
netboot_root(void)1515 netboot_root(void)
1516 {
1517 	return 0;
1518 }
1519 #endif
1520