main.c 26.8 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1 2 3 4 5 6 7 8
/*
 *  linux/init/main.c
 *
 *  Copyright (C) 1991, 1992  Linus Torvalds
 *
 *  GK 2/5/95  -  Changed to support mounting root fs via NFS
 *  Added initrd & change_root: Werner Almesberger & Hans Lermen, Feb '96
 *  Moan early if gcc is old, avoiding bogus kernels - Paul Gortmaker, May '96
9
 *  Simplified starting of init:  Michael A. Griffith <grif@acm.org>
Linus Torvalds's avatar
Linus Torvalds committed
10 11
 */

12 13
#define DEBUG		/* Enable initcall_debug */

Linus Torvalds's avatar
Linus Torvalds committed
14
#include <linux/types.h>
15
#include <linux/extable.h>
Linus Torvalds's avatar
Linus Torvalds committed
16 17
#include <linux/module.h>
#include <linux/proc_fs.h>
18
#include <linux/binfmts.h>
Linus Torvalds's avatar
Linus Torvalds committed
19 20
#include <linux/kernel.h>
#include <linux/syscalls.h>
Ingo Molnar's avatar
Ingo Molnar committed
21
#include <linux/stackprotector.h>
Linus Torvalds's avatar
Linus Torvalds committed
22 23 24 25 26 27 28
#include <linux/string.h>
#include <linux/ctype.h>
#include <linux/delay.h>
#include <linux/ioport.h>
#include <linux/init.h>
#include <linux/initrd.h>
#include <linux/bootmem.h>
29
#include <linux/acpi.h>
30
#include <linux/console.h>
31
#include <linux/nmi.h>
Linus Torvalds's avatar
Linus Torvalds committed
32 33
#include <linux/percpu.h>
#include <linux/kmod.h>
Nick Piggin's avatar
Nick Piggin committed
34
#include <linux/vmalloc.h>
Linus Torvalds's avatar
Linus Torvalds committed
35
#include <linux/kernel_stat.h>
36
#include <linux/start_kernel.h>
Linus Torvalds's avatar
Linus Torvalds committed
37
#include <linux/security.h>
38
#include <linux/smp.h>
Linus Torvalds's avatar
Linus Torvalds committed
39 40 41 42 43 44 45
#include <linux/profile.h>
#include <linux/rcupdate.h>
#include <linux/moduleparam.h>
#include <linux/kallsyms.h>
#include <linux/writeback.h>
#include <linux/cpu.h>
#include <linux/cpuset.h>
46
#include <linux/cgroup.h>
Linus Torvalds's avatar
Linus Torvalds committed
47
#include <linux/efi.h>
48
#include <linux/tick.h>
49
#include <linux/sched/isolation.h>
50
#include <linux/interrupt.h>
51
#include <linux/taskstats_kern.h>
52
#include <linux/delayacct.h>
Linus Torvalds's avatar
Linus Torvalds committed
53 54 55 56
#include <linux/unistd.h>
#include <linux/rmap.h>
#include <linux/mempolicy.h>
#include <linux/key.h>
Adrian Bunk's avatar
Adrian Bunk committed
57
#include <linux/buffer_head.h>
58
#include <linux/page_ext.h>
59
#include <linux/debug_locks.h>
60
#include <linux/debugobjects.h>
Ingo Molnar's avatar
Ingo Molnar committed
61
#include <linux/lockdep.h>
62
#include <linux/kmemleak.h>
63
#include <linux/pid_namespace.h>
64
#include <linux/device.h>
65
#include <linux/kthread.h>
66
#include <linux/sched.h>
67
#include <linux/sched/init.h>
68
#include <linux/signal.h>
69
#include <linux/idr.h>
70
#include <linux/kgdb.h>
71
#include <linux/ftrace.h>
72
#include <linux/async.h>
73
#include <linux/sfi.h>
74
#include <linux/shmem_fs.h>
75
#include <linux/slab.h>
76
#include <linux/perf_event.h>
77
#include <linux/ptrace.h>
78
#include <linux/pti.h>
79 80
#include <linux/blkdev.h>
#include <linux/elevator.h>
81
#include <linux/sched_clock.h>
82
#include <linux/sched/task.h>
83
#include <linux/sched/task_stack.h>
84
#include <linux/context_tracking.h>
85
#include <linux/random.h>
86
#include <linux/list.h>
87
#include <linux/integrity.h>
88
#include <linux/proc_ns.h>
89
#include <linux/io.h>
90
#include <linux/cache.h>
91
#include <linux/rodata_test.h>
Linus Torvalds's avatar
Linus Torvalds committed
92 93 94 95

#include <asm/io.h>
#include <asm/bugs.h>
#include <asm/setup.h>
96
#include <asm/sections.h>
97
#include <asm/cacheflush.h>
Linus Torvalds's avatar
Linus Torvalds committed
98

99
static int kernel_init(void *);
Linus Torvalds's avatar
Linus Torvalds committed
100 101

extern void init_IRQ(void);
102
extern void fork_init(void);
Linus Torvalds's avatar
Linus Torvalds committed
103 104
extern void radix_tree_init(void);

105 106 107 108 109 110 111 112 113
/*
 * Debug helper: via this flag we know that we are in 'early bootup code'
 * where only the boot processor is running with IRQ disabled.  This means
 * two things - IRQ must not be enabled before the flag is cleared and some
 * operations which are not allowed with IRQ disabled are allowed while the
 * flag is set.
 */
bool early_boot_irqs_disabled __read_mostly;

114
enum system_states system_state __read_mostly;
Linus Torvalds's avatar
Linus Torvalds committed
115 116 117 118 119 120 121 122 123 124
EXPORT_SYMBOL(system_state);

/*
 * Boot command-line arguments
 */
#define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT
#define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT

extern void time_init(void);
/* Default late time init is NULL. archs can override this later. */
125
void (*__initdata late_time_init)(void);
Linus Torvalds's avatar
Linus Torvalds committed
126

127 128 129 130 131 132
/* Untouched command line saved by arch-specific code. */
char __initdata boot_command_line[COMMAND_LINE_SIZE];
/* Untouched saved command line (eg. for /proc) */
char *saved_command_line;
/* Command line for parameter parsing */
static char *static_command_line;
133 134
/* Command line for per-initcall parameter parsing */
static char *initcall_command_line;
Linus Torvalds's avatar
Linus Torvalds committed
135 136

static char *execute_command;
137
static char *ramdisk_execute_command;
Linus Torvalds's avatar
Linus Torvalds committed
138

139 140 141 142
/*
 * Used to generate warnings if static_key manipulation functions are used
 * before jump_label_init is called.
 */
143
bool static_key_initialized __read_mostly;
144 145
EXPORT_SYMBOL_GPL(static_key_initialized);

Jan Beulich's avatar
Jan Beulich committed
146 147 148 149 150 151
/*
 * If set, this is an indication to the drivers that reset the underlying
 * device before going ahead with the initialization otherwise driver might
 * rely on the BIOS and skip the reset operation.
 *
 * This is useful if kernel is booting in an unreliable environment.
152
 * For ex. kdump situation where previous kernel has crashed, BIOS has been
Jan Beulich's avatar
Jan Beulich committed
153 154 155 156
 * skipped and devices will be in unknown state.
 */
unsigned int reset_devices;
EXPORT_SYMBOL(reset_devices);
Linus Torvalds's avatar
Linus Torvalds committed
157

158 159 160 161 162 163 164 165
static int __init set_reset_devices(char *str)
{
	reset_devices = 1;
	return 1;
}

__setup("reset_devices", set_reset_devices);

166 167
static const char *argv_init[MAX_INIT_ARGS+2] = { "init", NULL, };
const char *envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, };
Linus Torvalds's avatar
Linus Torvalds committed
168 169
static const char *panic_later, *panic_param;

170
extern const struct obs_kernel_param __setup_start[], __setup_end[];
Linus Torvalds's avatar
Linus Torvalds committed
171

172
static bool __init obsolete_checksetup(char *line)
Linus Torvalds's avatar
Linus Torvalds committed
173
{
174
	const struct obs_kernel_param *p;
175
	bool had_early_param = false;
Linus Torvalds's avatar
Linus Torvalds committed
176 177 178 179

	p = __setup_start;
	do {
		int n = strlen(p->str);
180
		if (parameqn(line, p->str, n)) {
Linus Torvalds's avatar
Linus Torvalds committed
181
			if (p->early) {
182 183 184 185
				/* Already done in parse_early_param?
				 * (Needs exact match on param part).
				 * Keep iterating, as we can have early
				 * params and __setups of same names 8( */
Linus Torvalds's avatar
Linus Torvalds committed
186
				if (line[n] == '\0' || line[n] == '=')
187
					had_early_param = true;
Linus Torvalds's avatar
Linus Torvalds committed
188
			} else if (!p->setup_func) {
189 190
				pr_warn("Parameter %s is obsolete, ignored\n",
					p->str);
191
				return true;
Linus Torvalds's avatar
Linus Torvalds committed
192
			} else if (p->setup_func(line + n))
193
				return true;
Linus Torvalds's avatar
Linus Torvalds committed
194 195 196
		}
		p++;
	} while (p < __setup_end);
197 198

	return had_early_param;
Linus Torvalds's avatar
Linus Torvalds committed
199 200 201 202 203 204 205 206 207 208 209
}

/*
 * This should be approx 2 Bo*oMips to start (note initial shift), and will
 * still work even if initially too large, it will just take slightly longer
 */
unsigned long loops_per_jiffy = (1<<12);
EXPORT_SYMBOL(loops_per_jiffy);

static int __init debug_kernel(char *str)
{
210
	console_loglevel = CONSOLE_LOGLEVEL_DEBUG;
211
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
212 213 214 215
}

static int __init quiet_kernel(char *str)
{
216
	console_loglevel = CONSOLE_LOGLEVEL_QUIET;
217
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
218 219
}

220 221
early_param("debug", debug_kernel);
early_param("quiet", quiet_kernel);
Linus Torvalds's avatar
Linus Torvalds committed
222 223 224

static int __init loglevel(char *str)
{
225 226 227 228 229 230 231 232 233 234 235 236 237
	int newlevel;

	/*
	 * Only update loglevel value when a correct setting was passed,
	 * to prevent blind crashes (when loglevel being set to 0) that
	 * are quite hard to debug
	 */
	if (get_option(&str, &newlevel)) {
		console_loglevel = newlevel;
		return 0;
	}

	return -EINVAL;
Linus Torvalds's avatar
Linus Torvalds committed
238 239
}

240
early_param("loglevel", loglevel);
Linus Torvalds's avatar
Linus Torvalds committed
241

242
/* Change NUL term back to "=", to make "param" the whole string. */
243 244
static int __init repair_env_string(char *param, char *val,
				    const char *unused, void *arg)
Linus Torvalds's avatar
Linus Torvalds committed
245 246 247 248 249 250 251 252 253 254 255 256
{
	if (val) {
		/* param=val or param="val"? */
		if (val == param+strlen(param)+1)
			val[-1] = '=';
		else if (val == param+strlen(param)+2) {
			val[-2] = '=';
			memmove(val-1, val, strlen(val)+1);
			val--;
		} else
			BUG();
	}
257 258 259
	return 0;
}

260
/* Anything after -- gets handed straight to init. */
261 262
static int __init set_init_arg(char *param, char *val,
			       const char *unused, void *arg)
263 264 265 266 267 268
{
	unsigned int i;

	if (panic_later)
		return 0;

269
	repair_env_string(param, val, unused, NULL);
270 271 272 273 274 275 276 277 278 279 280 281

	for (i = 0; argv_init[i]; i++) {
		if (i == MAX_INIT_ARGS) {
			panic_later = "init";
			panic_param = param;
			return 0;
		}
	}
	argv_init[i] = param;
	return 0;
}

282 283 284 285
/*
 * Unknown boot options get handed to init, unless they look like
 * unused parameters (modprobe will find them in /proc/cmdline).
 */
286 287
static int __init unknown_bootoption(char *param, char *val,
				     const char *unused, void *arg)
288
{
289
	repair_env_string(param, val, unused, NULL);
Linus Torvalds's avatar
Linus Torvalds committed
290 291 292 293 294

	/* Handle obsolete-style parameters */
	if (obsolete_checksetup(param))
		return 0;

295 296
	/* Unused module parameter. */
	if (strchr(param, '.') && (!val || strchr(param, '.') < val))
Linus Torvalds's avatar
Linus Torvalds committed
297 298 299 300 301 302 303 304 305 306
		return 0;

	if (panic_later)
		return 0;

	if (val) {
		/* Environment option */
		unsigned int i;
		for (i = 0; envp_init[i]; i++) {
			if (i == MAX_INIT_ENVS) {
307
				panic_later = "env";
Linus Torvalds's avatar
Linus Torvalds committed
308 309 310 311 312 313 314 315 316 317 318
				panic_param = param;
			}
			if (!strncmp(param, envp_init[i], val - param))
				break;
		}
		envp_init[i] = param;
	} else {
		/* Command line option */
		unsigned int i;
		for (i = 0; argv_init[i]; i++) {
			if (i == MAX_INIT_ARGS) {
319
				panic_later = "init";
Linus Torvalds's avatar
Linus Torvalds committed
320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344
				panic_param = param;
			}
		}
		argv_init[i] = param;
	}
	return 0;
}

static int __init init_setup(char *str)
{
	unsigned int i;

	execute_command = str;
	/*
	 * In case LILO is going to boot us with default command line,
	 * it prepends "auto" before the whole cmdline which makes
	 * the shell think it should execute a script with such name.
	 * So we ignore all arguments entered _before_ init=... [MJ]
	 */
	for (i = 1; i < MAX_INIT_ARGS; i++)
		argv_init[i] = NULL;
	return 1;
}
__setup("init=", init_setup);

345 346 347 348 349 350 351 352 353 354 355 356
static int __init rdinit_setup(char *str)
{
	unsigned int i;

	ramdisk_execute_command = str;
	/* See "auto" comment in init_setup */
	for (i = 1; i < MAX_INIT_ARGS; i++)
		argv_init[i] = NULL;
	return 1;
}
__setup("rdinit=", rdinit_setup);

Linus Torvalds's avatar
Linus Torvalds committed
357
#ifndef CONFIG_SMP
358
static const unsigned int setup_max_cpus = NR_CPUS;
359
static inline void setup_nr_cpu_ids(void) { }
Linus Torvalds's avatar
Linus Torvalds committed
360 361 362
static inline void smp_prepare_cpus(unsigned int maxcpus) { }
#endif

363 364 365 366 367 368 369 370
/*
 * We need to store the untouched command line for future reference.
 * We also need to store the touched command line since the parameter
 * parsing is performed in place, and we should allow a component to
 * store reference of name/value for future reference.
 */
static void __init setup_command_line(char *command_line)
{
371 372 373 374 375
	saved_command_line =
		memblock_virt_alloc(strlen(boot_command_line) + 1, 0);
	initcall_command_line =
		memblock_virt_alloc(strlen(boot_command_line) + 1, 0);
	static_command_line = memblock_virt_alloc(strlen(command_line) + 1, 0);
376 377
	strcpy(saved_command_line, boot_command_line);
	strcpy(static_command_line, command_line);
378 379
}

Linus Torvalds's avatar
Linus Torvalds committed
380 381 382 383 384 385 386 387 388
/*
 * We need to finalize in a non-__init function or else race conditions
 * between the root thread and the init thread may cause start_kernel to
 * be reaped by free_initmem before the root thread has proceeded to
 * cpu_idle.
 *
 * gcc-3.4 accidentally inlines this function, so use noinline.
 */

389 390
static __initdata DECLARE_COMPLETION(kthreadd_done);

391
static noinline void __ref rest_init(void)
Linus Torvalds's avatar
Linus Torvalds committed
392
{
393
	struct task_struct *tsk;
394 395
	int pid;

396
	rcu_scheduler_starting();
397
	/*
Peter Zijlstra's avatar
Peter Zijlstra committed
398
	 * We need to spawn init first so that it obtains pid 1, however
399 400 401
	 * the init task will end up wanting to create kthreads, which, if
	 * we schedule it before we create kthreadd, will OOPS.
	 */
402 403 404 405 406 407 408 409 410 411 412
	pid = kernel_thread(kernel_init, NULL, CLONE_FS);
	/*
	 * Pin init on the boot CPU. Task migration is not properly working
	 * until sched_init_smp() has been run. It will set the allowed
	 * CPUs for init to the non isolated CPUs.
	 */
	rcu_read_lock();
	tsk = find_task_by_pid_ns(pid, &init_pid_ns);
	set_cpus_allowed_ptr(tsk, cpumask_of(smp_processor_id()));
	rcu_read_unlock();

Linus Torvalds's avatar
Linus Torvalds committed
413
	numa_default_policy();
414
	pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES);
415
	rcu_read_lock();
416
	kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns);
417
	rcu_read_unlock();
418 419 420 421 422 423 424 425 426 427

	/*
	 * Enable might_sleep() and smp_processor_id() checks.
	 * They cannot be enabled earlier because with CONFIG_PRREMPT=y
	 * kernel_thread() would trigger might_sleep() splats. With
	 * CONFIG_PREEMPT_VOLUNTARY=y the init task might have scheduled
	 * already, but it's stuck on the kthreadd_done completion.
	 */
	system_state = SYSTEM_SCHEDULING;

428
	complete(&kthreadd_done);
429 430 431

	/*
	 * The boot idle thread must execute schedule()
432
	 * at least once to get things moving:
433
	 */
434
	schedule_preempt_disabled();
435
	/* Call into cpu_idle with preempt disabled */
436
	cpu_startup_entry(CPUHP_ONLINE);
437
}
Linus Torvalds's avatar
Linus Torvalds committed
438 439

/* Check for early params. */
440 441
static int __init do_early_param(char *param, char *val,
				 const char *unused, void *arg)
Linus Torvalds's avatar
Linus Torvalds committed
442
{
443
	const struct obs_kernel_param *p;
Linus Torvalds's avatar
Linus Torvalds committed
444 445

	for (p = __setup_start; p < __setup_end; p++) {
446
		if ((p->early && parameq(param, p->str)) ||
447 448 449
		    (strcmp(param, "console") == 0 &&
		     strcmp(p->str, "earlycon") == 0)
		) {
Linus Torvalds's avatar
Linus Torvalds committed
450
			if (p->setup_func(val) != 0)
451
				pr_warn("Malformed early option '%s'\n", param);
Linus Torvalds's avatar
Linus Torvalds committed
452 453 454 455 456 457
		}
	}
	/* We accept everything at this stage. */
	return 0;
}

458 459
void __init parse_early_options(char *cmdline)
{
460 461
	parse_args("early options", cmdline, NULL, 0, 0, 0, NULL,
		   do_early_param);
462 463
}

Linus Torvalds's avatar
Linus Torvalds committed
464 465 466
/* Arch code calls this early on, or if not, just before other parsing. */
void __init parse_early_param(void)
{
467 468
	static int done __initdata;
	static char tmp_cmdline[COMMAND_LINE_SIZE] __initdata;
Linus Torvalds's avatar
Linus Torvalds committed
469 470 471 472 473

	if (done)
		return;

	/* All fall through to do_early_param. */
474
	strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE);
475
	parse_early_options(tmp_cmdline);
Linus Torvalds's avatar
Linus Torvalds committed
476 477 478
	done = 1;
}

479 480
void __init __weak arch_post_acpi_subsys_init(void) { }

481
void __init __weak smp_setup_processor_id(void)
482 483 484
{
}

485
# if THREAD_SIZE >= PAGE_SIZE
486
void __init __weak thread_stack_cache_init(void)
487 488
{
}
489
#endif
490

491 492
void __init __weak mem_encrypt_init(void) { }

Pekka Enberg's avatar
Pekka Enberg committed
493 494 495 496 497
/*
 * Set up kernel memory allocators
 */
static void __init mm_init(void)
{
498 499 500 501 502
	/*
	 * page_ext requires contiguous pages,
	 * bigger than MAX_ORDER unless SPARSEMEM.
	 */
	page_ext_init_flatmem();
Pekka Enberg's avatar
Pekka Enberg committed
503 504
	mem_init();
	kmem_cache_init();
505
	pgtable_init();
Pekka Enberg's avatar
Pekka Enberg committed
506
	vmalloc_init();
507
	ioremap_huge_init();
508 509
	/* Should be run before the first non-init thread is created */
	init_espfix_bsp();
510 511
	/* Should be run after espfix64 is set up. */
	pti_init();
Pekka Enberg's avatar
Pekka Enberg committed
512 513
}

514
asmlinkage __visible void __init start_kernel(void)
Linus Torvalds's avatar
Linus Torvalds committed
515
{
516 517
	char *command_line;
	char *after_dashes;
518

519
	set_task_stack_end_magic(&init_task);
520
	smp_setup_processor_id();
521
	debug_objects_early_init();
522

523
	cgroup_init_early();
Ingo Molnar's avatar
Ingo Molnar committed
524 525

	local_irq_disable();
526
	early_boot_irqs_disabled = true;
Ingo Molnar's avatar
Ingo Molnar committed
527

528 529 530 531
	/*
	 * Interrupts are still disabled. Do necessary setups, then
	 * enable them.
	 */
532
	boot_cpu_init();
Linus Torvalds's avatar
Linus Torvalds committed
533
	page_address_init();
534
	pr_notice("%s", linux_banner);
Linus Torvalds's avatar
Linus Torvalds committed
535
	setup_arch(&command_line);
536 537
	/*
	 * Set up the the initial canary and entropy after arch
538
	 * and after adding latent and command line entropy.
539 540
	 */
	add_latent_entropy();
541
	add_device_randomness(command_line, strlen(command_line));
542
	boot_init_stack_canary();
543
	mm_init_cpumask(&init_mm);
544
	setup_command_line(command_line);
545
	setup_nr_cpu_ids();
546
	setup_per_cpu_areas();
547
	boot_cpu_state_init();
548
	smp_prepare_boot_cpu();	/* arch-specific boot-cpu hooks */
Linus Torvalds's avatar
Linus Torvalds committed
549

550
	build_all_zonelists(NULL);
551 552
	page_alloc_init();

553
	pr_notice("Kernel command line: %s\n", boot_command_line);
554
	parse_early_param();
555 556 557
	after_dashes = parse_args("Booting kernel",
				  static_command_line, __start___param,
				  __stop___param - __start___param,
558
				  -1, -1, NULL, &unknown_bootoption);
559
	if (!IS_ERR_OR_NULL(after_dashes))
560
		parse_args("Setting init args", after_dashes, NULL, 0, -1, -1,
561
			   NULL, set_init_arg);
562 563 564

	jump_label_init();

565 566 567 568
	/*
	 * These use large bootmem allocations and must precede
	 * kmem_cache_init()
	 */
569
	setup_log_buf(0);
570 571 572
	vfs_caches_init_early();
	sort_main_extable();
	trap_init();
Pekka Enberg's avatar
Pekka Enberg committed
573
	mm_init();
574

575 576
	ftrace_init();

577 578 579
	/* trace_printk can be enabled here */
	early_trace_init();

Linus Torvalds's avatar
Linus Torvalds committed
580 581 582 583 584 585 586 587 588 589 590
	/*
	 * Set up the scheduler prior starting any interrupts (such as the
	 * timer interrupt). Full topology setup happens at smp_init()
	 * time - but meanwhile we still have a functioning scheduler.
	 */
	sched_init();
	/*
	 * Disable preemption - early bootup scheduling is extremely
	 * fragile until we cpu_idle() for the first time.
	 */
	preempt_disable();
591 592
	if (WARN(!irqs_disabled(),
		 "Interrupts were enabled *very* early, fixing it\n"))
593
		local_irq_disable();
594
	radix_tree_init();
595

596 597 598 599 600 601
	/*
	 * Set up housekeeping before setting up workqueues to allow the unbound
	 * workqueue to take non-housekeeping into account.
	 */
	housekeeping_init();

602 603 604 605 606 607 608
	/*
	 * Allow workqueue creation and work item queueing/cancelling
	 * early.  Work item execution depends on kthreads and starts after
	 * workqueue_init().
	 */
	workqueue_init_early();

Linus Torvalds's avatar
Linus Torvalds committed
609
	rcu_init();
610

611
	/* Trace events are available after this */
612 613
	trace_init();

614
	context_tracking_init();
615 616
	/* init some links before init_ISA_irqs() */
	early_irq_init();
Linus Torvalds's avatar
Linus Torvalds committed
617
	init_IRQ();
Thomas Gleixner's avatar
Thomas Gleixner committed
618
	tick_init();
619
	rcu_init_nohz();
Linus Torvalds's avatar
Linus Torvalds committed
620
	init_timers();
621
	hrtimers_init();
Linus Torvalds's avatar
Linus Torvalds committed
622
	softirq_init();
623
	timekeeping_init();
624
	time_init();
625
	sched_clock_postinit();
626
	printk_safe_init();
627
	perf_event_init();
628
	profile_init();
629
	call_function_init();
630
	WARN(!irqs_disabled(), "Interrupts were enabled early\n");
631
	early_boot_irqs_disabled = false;
632
	local_irq_enable();
633

634
	kmem_cache_init_late();
Linus Torvalds's avatar
Linus Torvalds committed
635 636 637 638 639 640 641 642

	/*
	 * HACK ALERT! This is early. We're enabling the console before
	 * we've done PCI setups etc, and console_init() must be aware of
	 * this. But we do want output early, in case something goes wrong.
	 */
	console_init();
	if (panic_later)
643 644
		panic("Too many boot %s vars at `%s'", panic_later,
		      panic_param);
Ingo Molnar's avatar
Ingo Molnar committed
645 646 647

	lockdep_info();

648 649 650 651 652 653 654
	/*
	 * Need to run this when irqs are enabled, because it wants
	 * to self-test [hard/soft]-irqs on/off lock inversion bugs
	 * too:
	 */
	locking_selftest();

655 656 657 658 659 660 661 662
	/*
	 * This needs to be called before any devices perform DMA
	 * operations that might use the SWIOTLB bounce buffers. It will
	 * mark the bounce buffers as decrypted so that their usage will
	 * not cause "plain-text" data to be decrypted when accessed.
	 */
	mem_encrypt_init();

Linus Torvalds's avatar
Linus Torvalds committed
663 664
#ifdef CONFIG_BLK_DEV_INITRD
	if (initrd_start && !initrd_below_start_ok &&
665
	    page_to_pfn(virt_to_page((void *)initrd_start)) < min_low_pfn) {
666
		pr_crit("initrd overwritten (0x%08lx < 0x%08lx) - disabling it.\n",
667 668
		    page_to_pfn(virt_to_page((void *)initrd_start)),
		    min_low_pfn);
Linus Torvalds's avatar
Linus Torvalds committed
669 670 671
		initrd_start = 0;
	}
#endif
672
	page_ext_init();
673
	kmemleak_init();
674
	debug_objects_mem_init();
675
	setup_per_cpu_pageset();
Linus Torvalds's avatar
Linus Torvalds committed
676
	numa_policy_init();
677
	acpi_early_init();
Linus Torvalds's avatar
Linus Torvalds committed
678 679 680
	if (late_time_init)
		late_time_init();
	calibrate_delay();
681
	pid_idr_init();
Linus Torvalds's avatar
Linus Torvalds committed
682
	anon_vma_init();
683
#ifdef CONFIG_X86
684
	if (efi_enabled(EFI_RUNTIME_SERVICES))
685 686
		efi_enter_virtual_mode();
#endif
687
	thread_stack_cache_init();
688
	cred_init();
689
	fork_init();
Linus Torvalds's avatar
Linus Torvalds committed
690 691 692 693
	proc_caches_init();
	buffer_init();
	key_init();
	security_init();
694
	dbg_late_init();
695
	vfs_caches_init();
696
	pagecache_init();
Linus Torvalds's avatar
Linus Torvalds committed
697 698
	signals_init();
	proc_root_init();
699
	nsfs_init();
Linus Torvalds's avatar
Linus Torvalds committed
700
	cpuset_init();
701
	cgroup_init();
702
	taskstats_init_early();
703
	delayacct_init();
Linus Torvalds's avatar
Linus Torvalds committed
704 705 706

	check_bugs();

707
	acpi_subsystem_init();
708
	arch_post_acpi_subsys_init();
709
	sfi_init_late();
Linus Torvalds's avatar
Linus Torvalds committed
710

711
	if (efi_enabled(EFI_RUNTIME_SERVICES)) {
712
		efi_free_boot_services();
713
	}
714

Linus Torvalds's avatar
Linus Torvalds committed
715 716 717 718
	/* Do the rest non-__init'ed, we're now alive */
	rest_init();
}

719 720 721 722
/* Call all constructor functions linked into the kernel. */
static void __init do_ctors(void)
{
#ifdef CONFIG_CONSTRUCTORS
723
	ctor_fn_t *fn = (ctor_fn_t *) __ctors_start;
724

725 726
	for (; fn < (ctor_fn_t *) __ctors_end; fn++)
		(*fn)();
727 728 729
#endif
}

730
bool initcall_debug;
731
core_param(initcall_debug, initcall_debug, bool, 0644);
Linus Torvalds's avatar
Linus Torvalds committed
732

733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763
#ifdef CONFIG_KALLSYMS
struct blacklist_entry {
	struct list_head next;
	char *buf;
};

static __initdata_or_module LIST_HEAD(blacklisted_initcalls);

static int __init initcall_blacklist(char *str)
{
	char *str_entry;
	struct blacklist_entry *entry;

	/* str argument is a comma-separated list of functions */
	do {
		str_entry = strsep(&str, ",");
		if (str_entry) {
			pr_debug("blacklisting initcall %s\n", str_entry);
			entry = alloc_bootmem(sizeof(*entry));
			entry->buf = alloc_bootmem(strlen(str_entry) + 1);
			strcpy(entry->buf, str_entry);
			list_add(&entry->next, &blacklisted_initcalls);
		}
	} while (str_entry);

	return 0;
}

static bool __init_or_module initcall_blacklisted(initcall_t fn)
{
	struct blacklist_entry *entry;
764
	char fn_name[KSYM_SYMBOL_LEN];
765
	unsigned long addr;
766

767
	if (list_empty(&blacklisted_initcalls))
768 769
		return false;

770 771
	addr = (unsigned long) dereference_function_descriptor(fn);
	sprint_symbol_no_offset(fn_name, addr);
772

773 774 775 776 777 778
	/*
	 * fn will be "function_name [module_name]" where [module_name] is not
	 * displayed for built-in init functions.  Strip off the [module_name].
	 */
	strreplace(fn_name, ' ', '\0');

779
	list_for_each_entry(entry, &blacklisted_initcalls, next) {
780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801
		if (!strcmp(fn_name, entry->buf)) {
			pr_debug("initcall %s blacklisted\n", fn_name);
			return true;
		}
	}

	return false;
}
#else
static int __init initcall_blacklist(char *str)
{
	pr_warn("initcall_blacklist requires CONFIG_KALLSYMS\n");
	return 0;
}

static bool __init_or_module initcall_blacklisted(initcall_t fn)
{
	return false;
}
#endif
__setup("initcall_blacklist=", initcall_blacklist);

802
static int __init_or_module do_one_initcall_debug(initcall_t fn)
Linus Torvalds's avatar
Linus Torvalds committed
803
{
804
	ktime_t calltime, delta, rettime;
Américo Wang's avatar
Américo Wang committed
805 806
	unsigned long long duration;
	int ret;
Linus Torvalds's avatar
Linus Torvalds committed
807

808
	printk(KERN_DEBUG "calling  %pF @ %i\n", fn, task_pid_nr(current));
809
	calltime = ktime_get();
Américo Wang's avatar
Américo Wang committed
810
	ret = fn();
811 812 813
	rettime = ktime_get();
	delta = ktime_sub(rettime, calltime);
	duration = (unsigned long long) ktime_to_ns(delta) >> 10;
814
	printk(KERN_DEBUG "initcall %pF returned %d after %lld usecs\n",
815
		 fn, ret, duration);
Linus Torvalds's avatar
Linus Torvalds committed
816

817 818 819
	return ret;
}

820
int __init_or_module do_one_initcall(initcall_t fn)
821 822 823
{
	int count = preempt_count();
	int ret;
824
	char msgbuf[64];
825

826 827 828
	if (initcall_blacklisted(fn))
		return -EPERM;

829 830 831 832
	if (initcall_debug)
		ret = do_one_initcall_debug(fn);
	else
		ret = fn();
833

834
	msgbuf[0] = 0;
835

836
	if (preempt_count() != count) {
837
		sprintf(msgbuf, "preemption imbalance ");
838
		preempt_count_set(count);
Linus Torvalds's avatar
Linus Torvalds committed
839
	}
840
	if (irqs_disabled()) {
841
		strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf));
842 843
		local_irq_enable();
	}
844
	WARN(msgbuf[0], "initcall %pF returned with %s\n", fn, msgbuf);
845

846
	add_latent_entropy();
Américo Wang's avatar
Américo Wang committed
847
	return ret;
848 849 850
}


851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873
extern initcall_t __initcall_start[];
extern initcall_t __initcall0_start[];
extern initcall_t __initcall1_start[];
extern initcall_t __initcall2_start[];
extern initcall_t __initcall3_start[];
extern initcall_t __initcall4_start[];
extern initcall_t __initcall5_start[];
extern initcall_t __initcall6_start[];
extern initcall_t __initcall7_start[];
extern initcall_t __initcall_end[];

static initcall_t *initcall_levels[] __initdata = {
	__initcall0_start,
	__initcall1_start,
	__initcall2_start,
	__initcall3_start,
	__initcall4_start,
	__initcall5_start,
	__initcall6_start,
	__initcall7_start,
	__initcall_end,
};

874
/* Keep these in sync with initcalls in include/linux/init.h */
875
static char *initcall_level_names[] __initdata = {
876 877 878 879 880 881 882 883
	"early",
	"core",
	"postcore",
	"arch",
	"subsys",
	"fs",
	"device",
	"late",
884 885 886
};

static void __init do_initcall_level(int level)
887
{
888
	initcall_t *fn;
889

890
	strcpy(initcall_command_line, saved_command_line);
891
	parse_args(initcall_level_names[level],
892
		   initcall_command_line, __start___param,
893 894
		   __stop___param - __start___param,
		   level, level,
895
		   NULL, &repair_env_string);
896 897

	for (fn = initcall_levels[level]; fn < initcall_levels[level+1]; fn++)
898
		do_one_initcall(*fn);
Linus Torvalds's avatar
Linus Torvalds committed
899 900
}

901 902 903 904
static void __init do_initcalls(void)
{
	int level;

905
	for (level = 0; level < ARRAY_SIZE(initcall_levels) - 1; level++)
906 907 908
		do_initcall_level(level);
}

Linus Torvalds's avatar
Linus Torvalds committed
909 910 911 912 913 914 915 916 917
/*
 * Ok, the machine is now initialized. None of the devices
 * have been touched yet, but the CPU subsystem is up and
 * running, and memory and process management works.
 *
 * Now we can finally start doing some real work..
 */
static void __init do_basic_setup(void)
{
918
	cpuset_init_smp();
919
	shmem_init();
Linus Torvalds's avatar
Linus Torvalds committed
920
	driver_init();
921
	init_irq_proc();
922
	do_ctors();
923
	usermodehelper_enable();
924
	do_initcalls();
Linus Torvalds's avatar
Linus Torvalds committed
925 926
}

927
static void __init do_pre_smp_initcalls(void)
928
{
929
	initcall_t *fn;
930

931
	for (fn = __initcall_start; fn < __initcall0_start; fn++)
932
		do_one_initcall(*fn);
933 934
}

935 936 937 938 939 940 941 942 943 944 945
/*
 * This function requests modules which should be loaded by default and is
 * called twice right after initrd is mounted and right before init is
 * exec'd.  If such modules are on either initrd or rootfs, they will be
 * loaded before control is passed to userland.
 */
void __init load_default_modules(void)
{
	load_default_elevator_module();
}

946
static int run_init_process(const char *init_filename)
Linus Torvalds's avatar
Linus Torvalds committed
947 948
{
	argv_init[0] = init_filename;
949
	return do_execve(getname_kernel(init_filename),
950 951
		(const char __user *const __user *)argv_init,
		(const char __user *const __user *)envp_init);
Linus Torvalds's avatar
Linus Torvalds committed
952 953
}

954 955 956 957 958 959 960 961 962 963 964 965 966 967
static int try_to_run_init_process(const char *init_filename)
{
	int ret;

	ret = run_init_process(init_filename);

	if (ret && ret != -ENOENT) {
		pr_err("Starting init: %s exists but couldn't execute it (error %d)\n",
		       init_filename, ret);
	}

	return ret;
}

968
static noinline void __init kernel_init_freeable(void);
969

970
#if defined(CONFIG_STRICT_KERNEL_RWX) || defined(CONFIG_STRICT_MODULE_RWX)
971
bool rodata_enabled __ro_after_init = true;
972 973 974 975 976
static int __init set_debug_rodata(char *str)
{
	return strtobool(str, &rodata_enabled);
}
__setup("rodata=", set_debug_rodata);
977
#endif
978

979
#ifdef CONFIG_STRICT_KERNEL_RWX
980 981
static void mark_readonly(void)
{
982
	if (rodata_enabled) {
983
		mark_rodata_ro();
984 985
		rodata_test();
	} else
986 987 988 989 990 991 992 993 994
		pr_info("Kernel memory protection disabled.\n");
}
#else
static inline void mark_readonly(void)
{
	pr_warn("This architecture does not have kernel memory protection.\n");
}
#endif

995
static int __ref kernel_init(void *unused)
996
{
997 998
	int ret;

999
	kernel_init_freeable();
1000 1001
	/* need to finish all async __init code before freeing the memory */
	async_synchronize_full();
1002
	ftrace_free_init_mem();
1003
	free_initmem();
1004
	mark_readonly();
1005 1006 1007
	system_state = SYSTEM_RUNNING;
	numa_default_policy();

1008 1009
	rcu_end_inkernel_boot();

1010
	if (ramdisk_execute_command) {
1011 1012
		ret = run_init_process(ramdisk_execute_command);
		if (!ret)
1013
			return 0;
1014 1015
		pr_err("Failed to execute %s (error %d)\n",
		       ramdisk_execute_command, ret);
1016 1017 1018 1019 1020 1021 1022 1023 1024
	}

	/*
	 * We try each of these until one succeeds.
	 *
	 * The Bourne shell can be used instead of init if we are
	 * trying to recover a really broken machine.
	 */
	if (execute_command) {
1025 1026
		ret = run_init_process(execute_command);
		if (!ret)
1027
			return 0;
1028 1029
		panic("Requested init %s failed (error %d).",
		      execute_command, ret);
1030
	}
1031 1032 1033 1034
	if (!try_to_run_init_process("/sbin/init") ||
	    !try_to_run_init_process("/etc/init") ||
	    !try_to_run_init_process("/bin/init") ||
	    !try_to_run_init_process("/bin/sh"))
1035
		return 0;
1036

1037
	panic("No working init found.  Try passing init= option to kernel. "
1038
	      "See Linux Documentation/admin-guide/init.rst for guidance.");
1039 1040
}

1041
static noinline void __init kernel_init_freeable(void)
Linus Torvalds's avatar
Linus Torvalds committed
1042
{
1043 1044 1045 1046
	/*
	 * Wait until kthreadd is all set-up.
	 */
	wait_for_completion(&kthreadd_done);
1047 1048 1049 1050

	/* Now the scheduler is fully set up and can do blocking allocations */
	gfp_allowed_mask = __GFP_BITS_MASK;

1051 1052 1053
	/*
	 * init can allocate pages on any node
	 */
1054
	set_mems_allowed(node_states[N_MEMORY]);
Linus Torvalds's avatar
Linus Torvalds committed
1055

1056 1057
	cad_pid = task_pid(current);

1058
	smp_prepare_cpus(setup_max_cpus);
Linus Torvalds's avatar
Linus Torvalds committed
1059

1060 1061
	workqueue_init();

1062 1063
	init_mm_internals();

Linus Torvalds's avatar
Linus Torvalds committed
1064
	do_pre_smp_initcalls();
1065
	lockup_detector_init();
Linus Torvalds's avatar
Linus Torvalds committed
1066 1067 1068 1069

	smp_init();
	sched_init_smp();

1070 1071
	page_alloc_init_late();

Linus Torvalds's avatar
Linus Torvalds committed
1072 1073
	do_basic_setup();

1074 1075
	/* Open the /dev/console on the rootfs, this should never fail */
	if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0)
1076
		pr_err("Warning: unable to open an initial console.\n");
1077 1078 1079

	(void) sys_dup(0);
	(void) sys_dup(0);
Linus Torvalds's avatar
Linus Torvalds committed
1080 1081 1082 1083
	/*
	 * check if there is an early userspace init.  If yes, let it do all
	 * the work
	 */
1084 1085 1086 1087 1088 1089

	if (!ramdisk_execute_command)
		ramdisk_execute_command = "/init";

	if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) {
		ramdisk_execute_command = NULL;
Linus Torvalds's avatar
Linus Torvalds committed
1090
		prepare_namespace();
1091
	}
Linus Torvalds's avatar
Linus Torvalds committed
1092 1093 1094 1095 1096

	/*
	 * Ok, we have completed the initial bootup, and
	 * we're essentially up and running. Get rid of the
	 * initmem segments and start the user-mode stuff..
1097 1098 1099
	 *
	 * rootfs is available now, try loading the public keys
	 * and default modules
Linus Torvalds's avatar
Linus Torvalds committed
1100
	 */
1101

1102
	integrity_load_keys();
1103
	load_default_modules();
Linus Torvalds's avatar
Linus Torvalds committed
1104
}