main.c 25.5 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>
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/interrupt.h>
50
#include <linux/taskstats_kern.h>
51
#include <linux/delayacct.h>
Linus Torvalds's avatar
Linus Torvalds committed
52 53 54 55
#include <linux/unistd.h>
#include <linux/rmap.h>
#include <linux/mempolicy.h>
#include <linux/key.h>
56
#include <linux/buffer_head.h>
57
#include <linux/page_ext.h>
58
#include <linux/debug_locks.h>
59
#include <linux/debugobjects.h>
Ingo Molnar's avatar
Ingo Molnar committed
60
#include <linux/lockdep.h>
61
#include <linux/kmemleak.h>
62
#include <linux/pid_namespace.h>
63
#include <linux/device.h>
64
#include <linux/kthread.h>
65
#include <linux/sched.h>
66
#include <linux/sched/init.h>
67
#include <linux/signal.h>
68
#include <linux/idr.h>
69
#include <linux/kgdb.h>
70
#include <linux/ftrace.h>
71
#include <linux/async.h>
72
#include <linux/kmemcheck.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 79
#include <linux/blkdev.h>
#include <linux/elevator.h>
80
#include <linux/sched_clock.h>
81
#include <linux/sched/task.h>
82
#include <linux/sched/task_stack.h>
83
#include <linux/context_tracking.h>
84
#include <linux/random.h>
85
#include <linux/list.h>
86
#include <linux/integrity.h>
87
#include <linux/proc_ns.h>
88
#include <linux/io.h>
89
#include <linux/cache.h>
90
#include <linux/rodata_test.h>
Linus Torvalds's avatar
Linus Torvalds committed
91 92 93 94

#include <asm/io.h>
#include <asm/bugs.h>
#include <asm/setup.h>
95
#include <asm/sections.h>
96
#include <asm/cacheflush.h>
Linus Torvalds's avatar
Linus Torvalds committed
97

98
static int kernel_init(void *);
Linus Torvalds's avatar
Linus Torvalds committed
99 100

extern void init_IRQ(void);
101
extern void fork_init(void);
Linus Torvalds's avatar
Linus Torvalds committed
102 103
extern void radix_tree_init(void);

104 105 106 107 108 109 110 111 112
/*
 * 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;

113
enum system_states system_state __read_mostly;
Linus Torvalds's avatar
Linus Torvalds committed
114 115 116 117 118 119 120 121 122 123
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. */
124
void (*__initdata late_time_init)(void);
Linus Torvalds's avatar
Linus Torvalds committed
125

126 127 128 129 130 131
/* 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;
132 133
/* Command line for per-initcall parameter parsing */
static char *initcall_command_line;
Linus Torvalds's avatar
Linus Torvalds committed
134 135

static char *execute_command;
136
static char *ramdisk_execute_command;
Linus Torvalds's avatar
Linus Torvalds committed
137

138 139 140 141
/*
 * Used to generate warnings if static_key manipulation functions are used
 * before jump_label_init is called.
 */
142
bool static_key_initialized __read_mostly;
143 144
EXPORT_SYMBOL_GPL(static_key_initialized);

Jan Beulich's avatar
Jan Beulich committed
145 146 147 148 149 150
/*
 * 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.
151
 * For ex. kdump situation where previous kernel has crashed, BIOS has been
Jan Beulich's avatar
Jan Beulich committed
152 153 154 155
 * skipped and devices will be in unknown state.
 */
unsigned int reset_devices;
EXPORT_SYMBOL(reset_devices);
Linus Torvalds's avatar
Linus Torvalds committed
156

157 158 159 160 161 162 163 164
static int __init set_reset_devices(char *str)
{
	reset_devices = 1;
	return 1;
}

__setup("reset_devices", set_reset_devices);

165 166
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
167 168
static const char *panic_later, *panic_param;

169
extern const struct obs_kernel_param __setup_start[], __setup_end[];
Linus Torvalds's avatar
Linus Torvalds committed
170

171
static bool __init obsolete_checksetup(char *line)
Linus Torvalds's avatar
Linus Torvalds committed
172
{
173
	const struct obs_kernel_param *p;
174
	bool had_early_param = false;
Linus Torvalds's avatar
Linus Torvalds committed
175 176 177 178

	p = __setup_start;
	do {
		int n = strlen(p->str);
179
		if (parameqn(line, p->str, n)) {
Linus Torvalds's avatar
Linus Torvalds committed
180
			if (p->early) {
181 182 183 184
				/* 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
185
				if (line[n] == '\0' || line[n] == '=')
186
					had_early_param = true;
Linus Torvalds's avatar
Linus Torvalds committed
187
			} else if (!p->setup_func) {
188 189
				pr_warn("Parameter %s is obsolete, ignored\n",
					p->str);
190
				return true;
Linus Torvalds's avatar
Linus Torvalds committed
191
			} else if (p->setup_func(line + n))
192
				return true;
Linus Torvalds's avatar
Linus Torvalds committed
193 194 195
		}
		p++;
	} while (p < __setup_end);
196 197

	return had_early_param;
Linus Torvalds's avatar
Linus Torvalds committed
198 199 200 201 202 203 204 205 206 207 208
}

/*
 * 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)
{
209
	console_loglevel = CONSOLE_LOGLEVEL_DEBUG;
210
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
211 212 213 214
}

static int __init quiet_kernel(char *str)
{
215
	console_loglevel = CONSOLE_LOGLEVEL_QUIET;
216
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
217 218
}

219 220
early_param("debug", debug_kernel);
early_param("quiet", quiet_kernel);
Linus Torvalds's avatar
Linus Torvalds committed
221 222 223

static int __init loglevel(char *str)
{
224 225 226 227 228 229 230 231 232 233 234 235 236
	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
237 238
}

239
early_param("loglevel", loglevel);
Linus Torvalds's avatar
Linus Torvalds committed
240

241
/* Change NUL term back to "=", to make "param" the whole string. */
242 243
static int __init repair_env_string(char *param, char *val,
				    const char *unused, void *arg)
Linus Torvalds's avatar
Linus Torvalds committed
244 245 246 247 248 249 250 251 252 253 254 255
{
	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();
	}
256 257 258
	return 0;
}

259
/* Anything after -- gets handed straight to init. */
260 261
static int __init set_init_arg(char *param, char *val,
			       const char *unused, void *arg)
262 263 264 265 266 267
{
	unsigned int i;

	if (panic_later)
		return 0;

268
	repair_env_string(param, val, unused, NULL);
269 270 271 272 273 274 275 276 277 278 279 280

	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;
}

281 282 283 284
/*
 * Unknown boot options get handed to init, unless they look like
 * unused parameters (modprobe will find them in /proc/cmdline).
 */
285 286
static int __init unknown_bootoption(char *param, char *val,
				     const char *unused, void *arg)
287
{
288
	repair_env_string(param, val, unused, NULL);
Linus Torvalds's avatar
Linus Torvalds committed
289 290 291 292 293

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

294 295
	/* Unused module parameter. */
	if (strchr(param, '.') && (!val || strchr(param, '.') < val))
Linus Torvalds's avatar
Linus Torvalds committed
296 297 298 299 300 301 302 303 304 305
		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) {
306
				panic_later = "env";
Linus Torvalds's avatar
Linus Torvalds committed
307 308 309 310 311 312 313 314 315 316 317
				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) {
318
				panic_later = "init";
Linus Torvalds's avatar
Linus Torvalds committed
319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343
				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);

344 345 346 347 348 349 350 351 352 353 354 355
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
356
#ifndef CONFIG_SMP
357
static const unsigned int setup_max_cpus = NR_CPUS;
358
static inline void setup_nr_cpu_ids(void) { }
Linus Torvalds's avatar
Linus Torvalds committed
359 360 361
static inline void smp_prepare_cpus(unsigned int maxcpus) { }
#endif

362 363 364 365 366 367 368 369
/*
 * 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)
{
370 371 372 373 374
	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);
375 376
	strcpy(saved_command_line, boot_command_line);
	strcpy(static_command_line, command_line);
377 378
}

Linus Torvalds's avatar
Linus Torvalds committed
379 380 381 382 383 384 385 386 387
/*
 * 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.
 */

388 389
static __initdata DECLARE_COMPLETION(kthreadd_done);

390
static noinline void __ref rest_init(void)
Linus Torvalds's avatar
Linus Torvalds committed
391
{
392 393
	int pid;

394
	rcu_scheduler_starting();
395
	/*
Peter Zijlstra's avatar
Peter Zijlstra committed
396
	 * We need to spawn init first so that it obtains pid 1, however
397 398 399
	 * the init task will end up wanting to create kthreads, which, if
	 * we schedule it before we create kthreadd, will OOPS.
	 */
400
	kernel_thread(kernel_init, NULL, CLONE_FS);
Linus Torvalds's avatar
Linus Torvalds committed
401
	numa_default_policy();
402
	pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES);
403
	rcu_read_lock();
404
	kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns);
405
	rcu_read_unlock();
406
	complete(&kthreadd_done);
407 408 409

	/*
	 * The boot idle thread must execute schedule()
410
	 * at least once to get things moving:
411
	 */
412
	init_idle_bootup_task(current);
413
	schedule_preempt_disabled();
414
	/* Call into cpu_idle with preempt disabled */
415
	cpu_startup_entry(CPUHP_ONLINE);
416
}
Linus Torvalds's avatar
Linus Torvalds committed
417 418

/* Check for early params. */
419 420
static int __init do_early_param(char *param, char *val,
				 const char *unused, void *arg)
Linus Torvalds's avatar
Linus Torvalds committed
421
{
422
	const struct obs_kernel_param *p;
Linus Torvalds's avatar
Linus Torvalds committed
423 424

	for (p = __setup_start; p < __setup_end; p++) {
425
		if ((p->early && parameq(param, p->str)) ||
426 427 428
		    (strcmp(param, "console") == 0 &&
		     strcmp(p->str, "earlycon") == 0)
		) {
Linus Torvalds's avatar
Linus Torvalds committed
429
			if (p->setup_func(val) != 0)
430
				pr_warn("Malformed early option '%s'\n", param);
Linus Torvalds's avatar
Linus Torvalds committed
431 432 433 434 435 436
		}
	}
	/* We accept everything at this stage. */
	return 0;
}

437 438
void __init parse_early_options(char *cmdline)
{
439 440
	parse_args("early options", cmdline, NULL, 0, 0, 0, NULL,
		   do_early_param);
441 442
}

Linus Torvalds's avatar
Linus Torvalds committed
443 444 445
/* Arch code calls this early on, or if not, just before other parsing. */
void __init parse_early_param(void)
{
446 447
	static int done __initdata;
	static char tmp_cmdline[COMMAND_LINE_SIZE] __initdata;
Linus Torvalds's avatar
Linus Torvalds committed
448 449 450 451 452

	if (done)
		return;

	/* All fall through to do_early_param. */
453
	strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE);
454
	parse_early_options(tmp_cmdline);
Linus Torvalds's avatar
Linus Torvalds committed
455 456 457
	done = 1;
}

458 459
void __init __weak arch_post_acpi_subsys_init(void) { }

460
void __init __weak smp_setup_processor_id(void)
461 462 463
{
}

464
# if THREAD_SIZE >= PAGE_SIZE
465
void __init __weak thread_stack_cache_init(void)
466 467
{
}
468
#endif
469

Pekka Enberg's avatar
Pekka Enberg committed
470 471 472 473 474
/*
 * Set up kernel memory allocators
 */
static void __init mm_init(void)
{
475 476 477 478 479
	/*
	 * page_ext requires contiguous pages,
	 * bigger than MAX_ORDER unless SPARSEMEM.
	 */
	page_ext_init_flatmem();
Pekka Enberg's avatar
Pekka Enberg committed
480 481
	mem_init();
	kmem_cache_init();
482
	percpu_init_late();
483
	pgtable_init();
Pekka Enberg's avatar
Pekka Enberg committed
484
	vmalloc_init();
485
	ioremap_huge_init();
Pekka Enberg's avatar
Pekka Enberg committed
486 487
}

488
asmlinkage __visible void __init start_kernel(void)
Linus Torvalds's avatar
Linus Torvalds committed
489
{
490 491
	char *command_line;
	char *after_dashes;
492

493
	set_task_stack_end_magic(&init_task);
494
	smp_setup_processor_id();
495
	debug_objects_early_init();
496 497

	/*
498
	 * Set up the initial canary ASAP:
499 500 501
	 */
	boot_init_stack_canary();

502
	cgroup_init_early();
Ingo Molnar's avatar
Ingo Molnar committed
503 504

	local_irq_disable();
505
	early_boot_irqs_disabled = true;
Ingo Molnar's avatar
Ingo Molnar committed
506

507 508 509 510
	/*
	 * Interrupts are still disabled. Do necessary setups, then
	 * enable them.
	 */
511
	boot_cpu_init();
Linus Torvalds's avatar
Linus Torvalds committed
512
	page_address_init();
513
	pr_notice("%s", linux_banner);
Linus Torvalds's avatar
Linus Torvalds committed
514
	setup_arch(&command_line);
515
	mm_init_cpumask(&init_mm);
516
	setup_command_line(command_line);
517
	setup_nr_cpu_ids();
518
	setup_per_cpu_areas();
519
	boot_cpu_state_init();
520
	smp_prepare_boot_cpu();	/* arch-specific boot-cpu hooks */
Linus Torvalds's avatar
Linus Torvalds committed
521

522
	build_all_zonelists(NULL, NULL);
523 524
	page_alloc_init();

525
	pr_notice("Kernel command line: %s\n", boot_command_line);
526
	parse_early_param();
527 528 529
	after_dashes = parse_args("Booting kernel",
				  static_command_line, __start___param,
				  __stop___param - __start___param,
530
				  -1, -1, NULL, &unknown_bootoption);
531
	if (!IS_ERR_OR_NULL(after_dashes))
532
		parse_args("Setting init args", after_dashes, NULL, 0, -1, -1,
533
			   NULL, set_init_arg);
534 535 536

	jump_label_init();

537 538 539 540
	/*
	 * These use large bootmem allocations and must precede
	 * kmem_cache_init()
	 */
541
	setup_log_buf(0);
542 543 544 545
	pidhash_init();
	vfs_caches_init_early();
	sort_main_extable();
	trap_init();
Pekka Enberg's avatar
Pekka Enberg committed
546
	mm_init();
547

548 549
	ftrace_init();

550 551 552
	/* trace_printk can be enabled here */
	early_trace_init();

Linus Torvalds's avatar
Linus Torvalds committed
553 554 555 556 557 558 559 560 561 562 563
	/*
	 * 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();
564 565
	if (WARN(!irqs_disabled(),
		 "Interrupts were enabled *very* early, fixing it\n"))
566
		local_irq_disable();
567
	radix_tree_init();
568 569 570 571 572 573 574 575

	/*
	 * 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
576
	rcu_init();
577

578
	/* Trace events are available after this */
579 580
	trace_init();

581
	context_tracking_init();
582 583
	/* init some links before init_ISA_irqs() */
	early_irq_init();
Linus Torvalds's avatar
Linus Torvalds committed
584
	init_IRQ();
585
	tick_init();
586
	rcu_init_nohz();
Linus Torvalds's avatar
Linus Torvalds committed
587
	init_timers();
588
	hrtimers_init();
Linus Torvalds's avatar
Linus Torvalds committed
589
	softirq_init();
590
	timekeeping_init();
591
	time_init();
592
	sched_clock_postinit();
593
	printk_safe_init();
594
	perf_event_init();
595
	profile_init();
596
	call_function_init();
597
	WARN(!irqs_disabled(), "Interrupts were enabled early\n");
598
	early_boot_irqs_disabled = false;
599
	local_irq_enable();
600

601
	kmem_cache_init_late();
Linus Torvalds's avatar
Linus Torvalds committed
602 603 604 605 606 607 608 609

	/*
	 * 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)
610 611
		panic("Too many boot %s vars at `%s'", panic_later,
		      panic_param);
Ingo Molnar's avatar
Ingo Molnar committed
612 613 614

	lockdep_info();

615 616 617 618 619 620 621
	/*
	 * 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();

Linus Torvalds's avatar
Linus Torvalds committed
622 623
#ifdef CONFIG_BLK_DEV_INITRD
	if (initrd_start && !initrd_below_start_ok &&
624
	    page_to_pfn(virt_to_page((void *)initrd_start)) < min_low_pfn) {
625
		pr_crit("initrd overwritten (0x%08lx < 0x%08lx) - disabling it.\n",
626 627
		    page_to_pfn(virt_to_page((void *)initrd_start)),
		    min_low_pfn);
Linus Torvalds's avatar
Linus Torvalds committed
628 629 630
		initrd_start = 0;
	}
#endif
631
	page_ext_init();
632
	debug_objects_mem_init();
633
	kmemleak_init();
634
	setup_per_cpu_pageset();
Linus Torvalds's avatar
Linus Torvalds committed
635 636 637 638 639 640
	numa_policy_init();
	if (late_time_init)
		late_time_init();
	calibrate_delay();
	pidmap_init();
	anon_vma_init();
641
	acpi_early_init();
642
#ifdef CONFIG_X86
643
	if (efi_enabled(EFI_RUNTIME_SERVICES))
644
		efi_enter_virtual_mode();
645
#endif
646
#ifdef CONFIG_X86_ESPFIX64
647 648
	/* Should be run before the first non-init thread is created */
	init_espfix_bsp();
649
#endif
650
	thread_stack_cache_init();
651
	cred_init();
652
	fork_init();
Linus Torvalds's avatar
Linus Torvalds committed
653 654 655 656
	proc_caches_init();
	buffer_init();
	key_init();
	security_init();
657
	dbg_late_init();
658
	vfs_caches_init();
659
	pagecache_init();
Linus Torvalds's avatar
Linus Torvalds committed
660 661
	signals_init();
	proc_root_init();
662
	nsfs_init();
Linus Torvalds's avatar
Linus Torvalds committed
663
	cpuset_init();
664
	cgroup_init();
665
	taskstats_init_early();
666
	delayacct_init();
Linus Torvalds's avatar
Linus Torvalds committed
667 668 669

	check_bugs();

670
	acpi_subsystem_init();
671
	arch_post_acpi_subsys_init();
672
	sfi_init_late();
Linus Torvalds's avatar
Linus Torvalds committed
673

674
	if (efi_enabled(EFI_RUNTIME_SERVICES)) {
675
		efi_free_boot_services();
676
	}
677

Linus Torvalds's avatar
Linus Torvalds committed
678 679 680 681
	/* Do the rest non-__init'ed, we're now alive */
	rest_init();
}

682 683 684 685
/* Call all constructor functions linked into the kernel. */
static void __init do_ctors(void)
{
#ifdef CONFIG_CONSTRUCTORS
686
	ctor_fn_t *fn = (ctor_fn_t *) __ctors_start;
687

688 689
	for (; fn < (ctor_fn_t *) __ctors_end; fn++)
		(*fn)();
690 691 692
#endif
}

693
bool initcall_debug;
694
core_param(initcall_debug, initcall_debug, bool, 0644);
Linus Torvalds's avatar
Linus Torvalds committed
695

696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726
#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;
727
	char fn_name[KSYM_SYMBOL_LEN];
728
	unsigned long addr;
729

730
	if (list_empty(&blacklisted_initcalls))
731 732
		return false;

733 734
	addr = (unsigned long) dereference_function_descriptor(fn);
	sprint_symbol_no_offset(fn_name, addr);
735

736 737 738 739 740 741
	/*
	 * 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');

742
	list_for_each_entry(entry, &blacklisted_initcalls, next) {
743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764
		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);

765
static int __init_or_module do_one_initcall_debug(initcall_t fn)
Linus Torvalds's avatar
Linus Torvalds committed
766
{
767
	ktime_t calltime, delta, rettime;
768 769
	unsigned long long duration;
	int ret;
Linus Torvalds's avatar
Linus Torvalds committed
770

771
	printk(KERN_DEBUG "calling  %pF @ %i\n", fn, task_pid_nr(current));
772
	calltime = ktime_get();
773
	ret = fn();
774 775 776
	rettime = ktime_get();
	delta = ktime_sub(rettime, calltime);
	duration = (unsigned long long) ktime_to_ns(delta) >> 10;
777
	printk(KERN_DEBUG "initcall %pF returned %d after %lld usecs\n",
778
		 fn, ret, duration);
Linus Torvalds's avatar
Linus Torvalds committed
779

780 781 782
	return ret;
}

783
int __init_or_module do_one_initcall(initcall_t fn)
784 785 786
{
	int count = preempt_count();
	int ret;
787
	char msgbuf[64];
788

789 790 791
	if (initcall_blacklisted(fn))
		return -EPERM;

792 793 794 795
	if (initcall_debug)
		ret = do_one_initcall_debug(fn);
	else
		ret = fn();
796

797
	msgbuf[0] = 0;
798

799
	if (preempt_count() != count) {
800
		sprintf(msgbuf, "preemption imbalance ");
801
		preempt_count_set(count);
Linus Torvalds's avatar
Linus Torvalds committed
802
	}
803
	if (irqs_disabled()) {
804
		strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf));
805 806
		local_irq_enable();
	}
807
	WARN(msgbuf[0], "initcall %pF returned with %s\n", fn, msgbuf);
808

809
	add_latent_entropy();
810
	return ret;
811 812 813
}


814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836
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,
};

837
/* Keep these in sync with initcalls in include/linux/init.h */
838
static char *initcall_level_names[] __initdata = {
839 840 841 842 843 844 845 846
	"early",
	"core",
	"postcore",
	"arch",
	"subsys",
	"fs",
	"device",
	"late",
847 848 849
};

static void __init do_initcall_level(int level)
850
{
851
	initcall_t *fn;
852

853
	strcpy(initcall_command_line, saved_command_line);
854
	parse_args(initcall_level_names[level],
855
		   initcall_command_line, __start___param,
856 857
		   __stop___param - __start___param,
		   level, level,
858
		   NULL, &repair_env_string);
859 860

	for (fn = initcall_levels[level]; fn < initcall_levels[level+1]; fn++)
861
		do_one_initcall(*fn);
Linus Torvalds's avatar
Linus Torvalds committed
862 863
}

864 865 866 867
static void __init do_initcalls(void)
{
	int level;

868
	for (level = 0; level < ARRAY_SIZE(initcall_levels) - 1; level++)
869 870 871
		do_initcall_level(level);
}

Linus Torvalds's avatar
Linus Torvalds committed
872 873 874 875 876 877 878 879 880
/*
 * 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)
{
881
	cpuset_init_smp();
882
	shmem_init();
Linus Torvalds's avatar
Linus Torvalds committed
883
	driver_init();
884
	init_irq_proc();
885
	do_ctors();
886
	usermodehelper_enable();
887
	do_initcalls();
Linus Torvalds's avatar
Linus Torvalds committed
888 889
}

890
static void __init do_pre_smp_initcalls(void)
891
{
892
	initcall_t *fn;
893

894
	for (fn = __initcall_start; fn < __initcall0_start; fn++)
895
		do_one_initcall(*fn);
896 897
}

898 899 900 901 902 903 904 905 906 907 908
/*
 * 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();
}

909
static int run_init_process(const char *init_filename)
Linus Torvalds's avatar
Linus Torvalds committed
910 911
{
	argv_init[0] = init_filename;
912
	return do_execve(getname_kernel(init_filename),
913 914
		(const char __user *const __user *)argv_init,
		(const char __user *const __user *)envp_init);
Linus Torvalds's avatar
Linus Torvalds committed
915 916
}

917 918 919 920 921 922 923 924 925 926 927 928 929 930
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;
}

931
static noinline void __init kernel_init_freeable(void);
932

933
#if defined(CONFIG_STRICT_KERNEL_RWX) || defined(CONFIG_STRICT_MODULE_RWX)
934
bool rodata_enabled __ro_after_init = true;
935 936 937 938 939
static int __init set_debug_rodata(char *str)
{
	return strtobool(str, &rodata_enabled);
}
__setup("rodata=", set_debug_rodata);
940
#endif
941

942
#ifdef CONFIG_STRICT_KERNEL_RWX
943 944
static void mark_readonly(void)
{
945
	if (rodata_enabled) {
946
		mark_rodata_ro();
947 948
		rodata_test();
	} else
949 950 951 952 953 954 955 956 957
		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

958
static int __ref kernel_init(void *unused)
959
{
960 961
	int ret;

962
	kernel_init_freeable();
963 964
	/* need to finish all async __init code before freeing the memory */
	async_synchronize_full();
965
	ftrace_free_init_mem();
966
	free_initmem();
967
	mark_readonly();
968 969 970
	system_state = SYSTEM_RUNNING;
	numa_default_policy();

971 972
	rcu_end_inkernel_boot();

973
	if (ramdisk_execute_command) {
974 975
		ret = run_init_process(ramdisk_execute_command);
		if (!ret)
976
			return 0;
977 978
		pr_err("Failed to execute %s (error %d)\n",
		       ramdisk_execute_command, ret);
979 980 981 982 983 984 985 986 987
	}

	/*
	 * 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) {
988 989
		ret = run_init_process(execute_command);
		if (!ret)
990
			return 0;
991 992
		panic("Requested init %s failed (error %d).",
		      execute_command, ret);
993
	}
994 995 996 997
	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"))
998
		return 0;
999

1000
	panic("No working init found.  Try passing init= option to kernel. "
1001
	      "See Linux Documentation/admin-guide/init.rst for guidance.");
1002 1003
}

1004
static noinline void __init kernel_init_freeable(void)
Linus Torvalds's avatar
Linus Torvalds committed
1005
{
1006 1007 1008 1009
	/*
	 * Wait until kthreadd is all set-up.
	 */
	wait_for_completion(&kthreadd_done);
1010 1011 1012 1013

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

1014 1015 1016
	/*
	 * init can allocate pages on any node
	 */
1017
	set_mems_allowed(node_states[N_MEMORY]);
Linus Torvalds's avatar
Linus Torvalds committed
1018 1019 1020
	/*
	 * init can run on any cpu.
	 */
1021
	set_cpus_allowed_ptr(current, cpu_all_mask);
Linus Torvalds's avatar
Linus Torvalds committed
1022

1023 1024
	cad_pid = task_pid(current);

1025
	smp_prepare_cpus(setup_max_cpus);
Linus Torvalds's avatar
Linus Torvalds committed
1026

1027 1028
	workqueue_init();

1029 1030
	init_mm_internals();

Linus Torvalds's avatar
Linus Torvalds committed
1031
	do_pre_smp_initcalls();
1032
	lockup_detector_init();
Linus Torvalds's avatar
Linus Torvalds committed
1033 1034 1035 1036

	smp_init();
	sched_init_smp();

1037 1038
	page_alloc_init_late();

Linus Torvalds's avatar
Linus Torvalds committed
1039 1040
	do_basic_setup();

1041 1042
	/* Open the /dev/console on the rootfs, this should never fail */
	if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0)
1043
		pr_err("Warning: unable to open an initial console.\n");
1044 1045 1046

	(void) sys_dup(0);
	(void) sys_dup(0);
Linus Torvalds's avatar
Linus Torvalds committed
1047 1048 1049 1050
	/*
	 * check if there is an early userspace init.  If yes, let it do all
	 * the work
	 */
1051 1052 1053 1054 1055 1056

	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
1057
		prepare_namespace();
1058
	}
Linus Torvalds's avatar
Linus Torvalds committed
1059 1060 1061 1062 1063

	/*
	 * 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..
1064 1065 1066
	 *
	 * rootfs is available now, try loading the public keys
	 * and default modules
Linus Torvalds's avatar
Linus Torvalds committed
1067
	 */
1068

1069
	integrity_load_keys();
1070
	load_default_modules();
Linus Torvalds's avatar
Linus Torvalds committed
1071
}