main.c 24.7 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11
/*
 *  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
 *  Simplified starting of init:  Michael A. Griffith <grif@acm.org> 
 */

A
Andrew Morton 已提交
12 13
#define DEBUG		/* Enable initcall_debug */

L
Linus Torvalds 已提交
14 15 16 17 18
#include <linux/types.h>
#include <linux/module.h>
#include <linux/proc_fs.h>
#include <linux/kernel.h>
#include <linux/syscalls.h>
I
Ingo Molnar 已提交
19
#include <linux/stackprotector.h>
L
Linus Torvalds 已提交
20 21 22 23 24 25 26
#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>
27
#include <linux/acpi.h>
L
Linus Torvalds 已提交
28 29 30
#include <linux/tty.h>
#include <linux/percpu.h>
#include <linux/kmod.h>
N
Nick Piggin 已提交
31
#include <linux/vmalloc.h>
L
Linus Torvalds 已提交
32
#include <linux/kernel_stat.h>
33
#include <linux/start_kernel.h>
L
Linus Torvalds 已提交
34
#include <linux/security.h>
35
#include <linux/smp.h>
L
Linus Torvalds 已提交
36 37 38 39 40 41 42
#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>
43
#include <linux/cgroup.h>
L
Linus Torvalds 已提交
44
#include <linux/efi.h>
45
#include <linux/tick.h>
46
#include <linux/interrupt.h>
47
#include <linux/taskstats_kern.h>
48
#include <linux/delayacct.h>
L
Linus Torvalds 已提交
49 50 51 52
#include <linux/unistd.h>
#include <linux/rmap.h>
#include <linux/mempolicy.h>
#include <linux/key.h>
A
Adrian Bunk 已提交
53
#include <linux/buffer_head.h>
54
#include <linux/page_cgroup.h>
55
#include <linux/debug_locks.h>
56
#include <linux/debugobjects.h>
I
Ingo Molnar 已提交
57
#include <linux/lockdep.h>
C
Catalin Marinas 已提交
58
#include <linux/kmemleak.h>
59
#include <linux/pid_namespace.h>
60
#include <linux/device.h>
61
#include <linux/kthread.h>
62
#include <linux/sched.h>
63
#include <linux/signal.h>
64
#include <linux/idr.h>
65
#include <linux/kgdb.h>
66
#include <linux/ftrace.h>
67
#include <linux/async.h>
68
#include <linux/kmemcheck.h>
69
#include <linux/sfi.h>
70
#include <linux/shmem_fs.h>
71
#include <linux/slab.h>
72
#include <linux/perf_event.h>
A
Al Viro 已提交
73
#include <linux/file.h>
74
#include <linux/ptrace.h>
75 76
#include <linux/blkdev.h>
#include <linux/elevator.h>
77
#include <linux/sched_clock.h>
78
#include <linux/context_tracking.h>
79
#include <linux/random.h>
80
#include <linux/list.h>
L
Linus Torvalds 已提交
81 82 83 84

#include <asm/io.h>
#include <asm/bugs.h>
#include <asm/setup.h>
85
#include <asm/sections.h>
86
#include <asm/cacheflush.h>
L
Linus Torvalds 已提交
87 88 89 90 91

#ifdef CONFIG_X86_LOCAL_APIC
#include <asm/smp.h>
#endif

92
static int kernel_init(void *);
L
Linus Torvalds 已提交
93 94 95 96

extern void init_IRQ(void);
extern void fork_init(unsigned long);
extern void radix_tree_init(void);
97 98 99
#ifndef CONFIG_DEBUG_RODATA
static inline void mark_rodata_ro(void) { }
#endif
L
Linus Torvalds 已提交
100

101 102 103 104 105 106 107 108 109
/*
 * 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;

110
enum system_states system_state __read_mostly;
L
Linus Torvalds 已提交
111 112 113 114 115 116 117 118 119 120
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. */
121
void (*__initdata late_time_init)(void);
L
Linus Torvalds 已提交
122

123 124 125 126 127 128
/* 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;
129 130
/* Command line for per-initcall parameter parsing */
static char *initcall_command_line;
L
Linus Torvalds 已提交
131 132

static char *execute_command;
133
static char *ramdisk_execute_command;
L
Linus Torvalds 已提交
134

135 136 137 138 139 140 141
/*
 * Used to generate warnings if static_key manipulation functions are used
 * before jump_label_init is called.
 */
bool static_key_initialized __read_mostly = false;
EXPORT_SYMBOL_GPL(static_key_initialized);

J
Jan Beulich 已提交
142 143 144 145 146 147 148 149 150 151 152
/*
 * 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.
 * For ex. kdump situaiton where previous kernel has crashed, BIOS has been
 * skipped and devices will be in unknown state.
 */
unsigned int reset_devices;
EXPORT_SYMBOL(reset_devices);
L
Linus Torvalds 已提交
153

154 155 156 157 158 159 160 161
static int __init set_reset_devices(char *str)
{
	reset_devices = 1;
	return 1;
}

__setup("reset_devices", set_reset_devices);

162 163
static const char * argv_init[MAX_INIT_ARGS+2] = { "init", NULL, };
const char * envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, };
L
Linus Torvalds 已提交
164 165
static const char *panic_later, *panic_param;

166
extern const struct obs_kernel_param __setup_start[], __setup_end[];
L
Linus Torvalds 已提交
167 168 169

static int __init obsolete_checksetup(char *line)
{
170
	const struct obs_kernel_param *p;
171
	int had_early_param = 0;
L
Linus Torvalds 已提交
172 173 174 175

	p = __setup_start;
	do {
		int n = strlen(p->str);
176
		if (parameqn(line, p->str, n)) {
L
Linus Torvalds 已提交
177
			if (p->early) {
178 179 180 181
				/* 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( */
L
Linus Torvalds 已提交
182
				if (line[n] == '\0' || line[n] == '=')
183
					had_early_param = 1;
L
Linus Torvalds 已提交
184
			} else if (!p->setup_func) {
A
Andrew Morton 已提交
185 186
				pr_warn("Parameter %s is obsolete, ignored\n",
					p->str);
L
Linus Torvalds 已提交
187 188 189 190 191 192
				return 1;
			} else if (p->setup_func(line + n))
				return 1;
		}
		p++;
	} while (p < __setup_end);
193 194

	return had_early_param;
L
Linus Torvalds 已提交
195 196 197 198 199 200 201 202 203 204 205 206
}

/*
 * 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)
{
207
	console_loglevel = CONSOLE_LOGLEVEL_DEBUG;
208
	return 0;
L
Linus Torvalds 已提交
209 210 211 212
}

static int __init quiet_kernel(char *str)
{
213
	console_loglevel = CONSOLE_LOGLEVEL_QUIET;
214
	return 0;
L
Linus Torvalds 已提交
215 216
}

217 218
early_param("debug", debug_kernel);
early_param("quiet", quiet_kernel);
L
Linus Torvalds 已提交
219 220 221

static int __init loglevel(char *str)
{
222 223 224 225 226 227 228 229 230 231 232 233 234
	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;
L
Linus Torvalds 已提交
235 236
}

237
early_param("loglevel", loglevel);
L
Linus Torvalds 已提交
238

239
/* Change NUL term back to "=", to make "param" the whole string. */
240
static int __init repair_env_string(char *param, char *val, const char *unused)
L
Linus Torvalds 已提交
241 242 243 244 245 246 247 248 249 250 251 252
{
	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();
	}
253 254 255
	return 0;
}

256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276
/* Anything after -- gets handed straight to init. */
static int __init set_init_arg(char *param, char *val, const char *unused)
{
	unsigned int i;

	if (panic_later)
		return 0;

	repair_env_string(param, val, unused);

	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;
}

277 278 279 280
/*
 * Unknown boot options get handed to init, unless they look like
 * unused parameters (modprobe will find them in /proc/cmdline).
 */
281
static int __init unknown_bootoption(char *param, char *val, const char *unused)
282
{
283
	repair_env_string(param, val, unused);
L
Linus Torvalds 已提交
284 285 286 287 288

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

289 290
	/* Unused module parameter. */
	if (strchr(param, '.') && (!val || strchr(param, '.') < val))
L
Linus Torvalds 已提交
291 292 293 294 295 296 297 298 299 300
		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) {
301
				panic_later = "env";
L
Linus Torvalds 已提交
302 303 304 305 306 307 308 309 310 311 312
				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) {
313
				panic_later = "init";
L
Linus Torvalds 已提交
314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338
				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);

339 340 341 342 343 344 345 346 347 348 349 350
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);

L
Linus Torvalds 已提交
351
#ifndef CONFIG_SMP
352
static const unsigned int setup_max_cpus = NR_CPUS;
L
Linus Torvalds 已提交
353 354 355 356 357 358 359 360 361
#ifdef CONFIG_X86_LOCAL_APIC
static void __init smp_init(void)
{
	APIC_init_uniprocessor();
}
#else
#define smp_init()	do { } while (0)
#endif

362
static inline void setup_nr_cpu_ids(void) { }
L
Linus Torvalds 已提交
363 364 365
static inline void smp_prepare_cpus(unsigned int maxcpus) { }
#endif

366 367 368 369 370 371 372 373
/*
 * 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)
{
374 375 376 377 378
	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);
379 380 381 382
	strcpy (saved_command_line, boot_command_line);
	strcpy (static_command_line, command_line);
}

L
Linus Torvalds 已提交
383 384 385 386 387 388 389 390 391
/*
 * 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.
 */

392 393
static __initdata DECLARE_COMPLETION(kthreadd_done);

394
static noinline void __init_refok rest_init(void)
L
Linus Torvalds 已提交
395
{
396 397
	int pid;

398
	rcu_scheduler_starting();
399
	/*
P
Peter Zijlstra 已提交
400
	 * We need to spawn init first so that it obtains pid 1, however
401 402 403
	 * the init task will end up wanting to create kthreads, which, if
	 * we schedule it before we create kthreadd, will OOPS.
	 */
404
	kernel_thread(kernel_init, NULL, CLONE_FS);
L
Linus Torvalds 已提交
405
	numa_default_policy();
406
	pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES);
407
	rcu_read_lock();
P
Pavel Emelyanov 已提交
408
	kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns);
409
	rcu_read_unlock();
410
	complete(&kthreadd_done);
411 412 413

	/*
	 * The boot idle thread must execute schedule()
I
Ingo Molnar 已提交
414
	 * at least once to get things moving:
415
	 */
I
Ingo Molnar 已提交
416
	init_idle_bootup_task(current);
417
	schedule_preempt_disabled();
418
	/* Call into cpu_idle with preempt disabled */
419
	cpu_startup_entry(CPUHP_ONLINE);
I
Ingo Molnar 已提交
420
}
L
Linus Torvalds 已提交
421 422

/* Check for early params. */
423
static int __init do_early_param(char *param, char *val, const char *unused)
L
Linus Torvalds 已提交
424
{
425
	const struct obs_kernel_param *p;
L
Linus Torvalds 已提交
426 427

	for (p = __setup_start; p < __setup_end; p++) {
428
		if ((p->early && parameq(param, p->str)) ||
429 430 431
		    (strcmp(param, "console") == 0 &&
		     strcmp(p->str, "earlycon") == 0)
		) {
L
Linus Torvalds 已提交
432
			if (p->setup_func(val) != 0)
A
Andrew Morton 已提交
433
				pr_warn("Malformed early option '%s'\n", param);
L
Linus Torvalds 已提交
434 435 436 437 438 439
		}
	}
	/* We accept everything at this stage. */
	return 0;
}

M
Magnus Damm 已提交
440 441
void __init parse_early_options(char *cmdline)
{
442
	parse_args("early options", cmdline, NULL, 0, 0, 0, do_early_param);
M
Magnus Damm 已提交
443 444
}

L
Linus Torvalds 已提交
445 446 447 448 449 450 451 452 453 454
/* Arch code calls this early on, or if not, just before other parsing. */
void __init parse_early_param(void)
{
	static __initdata int done = 0;
	static __initdata char tmp_cmdline[COMMAND_LINE_SIZE];

	if (done)
		return;

	/* All fall through to do_early_param. */
455
	strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE);
M
Magnus Damm 已提交
456
	parse_early_options(tmp_cmdline);
L
Linus Torvalds 已提交
457 458 459 460 461 462 463
	done = 1;
}

/*
 *	Activate the first processor.
 */

464 465 466 467
static void __init boot_cpu_init(void)
{
	int cpu = smp_processor_id();
	/* Mark the boot cpu "present", "online" etc for SMP and UP case */
468
	set_cpu_online(cpu, true);
469
	set_cpu_active(cpu, true);
470 471
	set_cpu_present(cpu, true);
	set_cpu_possible(cpu, true);
472 473
}

474
void __init __weak smp_setup_processor_id(void)
475 476 477
{
}

478
# if THREAD_SIZE >= PAGE_SIZE
479 480 481
void __init __weak thread_info_cache_init(void)
{
}
482
#endif
483

P
Pekka Enberg 已提交
484 485 486 487 488
/*
 * Set up kernel memory allocators
 */
static void __init mm_init(void)
{
489
	/*
490 491
	 * page_cgroup requires contiguous pages,
	 * bigger than MAX_ORDER unless SPARSEMEM.
492 493
	 */
	page_cgroup_init_flatmem();
P
Pekka Enberg 已提交
494 495
	mem_init();
	kmem_cache_init();
496
	percpu_init_late();
497
	pgtable_init();
P
Pekka Enberg 已提交
498 499 500
	vmalloc_init();
}

501
asmlinkage __visible void __init start_kernel(void)
L
Linus Torvalds 已提交
502
{
503
	char * command_line, *after_dashes;
504
	extern const struct kernel_param __start___param[], __stop___param[];
505

I
Ingo Molnar 已提交
506 507 508 509 510
	/*
	 * Need to run as early as possible, to initialize the
	 * lockdep hash:
	 */
	lockdep_init();
511
	smp_setup_processor_id();
512
	debug_objects_early_init();
513 514 515 516 517 518

	/*
	 * Set up the the initial canary ASAP:
	 */
	boot_init_stack_canary();

519
	cgroup_init_early();
I
Ingo Molnar 已提交
520 521

	local_irq_disable();
522
	early_boot_irqs_disabled = true;
I
Ingo Molnar 已提交
523

L
Linus Torvalds 已提交
524 525 526 527
/*
 * Interrupts are still disabled. Do necessary setups, then
 * enable them
 */
528
	boot_cpu_init();
L
Linus Torvalds 已提交
529
	page_address_init();
A
Andrew Morton 已提交
530
	pr_notice("%s", linux_banner);
L
Linus Torvalds 已提交
531
	setup_arch(&command_line);
532
	mm_init_cpumask(&init_mm);
533
	setup_command_line(command_line);
534
	setup_nr_cpu_ids();
535
	setup_per_cpu_areas();
536
	smp_prepare_boot_cpu();	/* arch-specific boot-cpu hooks */
L
Linus Torvalds 已提交
537

538
	build_all_zonelists(NULL, NULL);
539 540
	page_alloc_init();

A
Andrew Morton 已提交
541
	pr_notice("Kernel command line: %s\n", boot_command_line);
542
	parse_early_param();
543 544 545 546 547 548 549
	after_dashes = parse_args("Booting kernel",
				  static_command_line, __start___param,
				  __stop___param - __start___param,
				  -1, -1, &unknown_bootoption);
	if (after_dashes)
		parse_args("Setting init args", after_dashes, NULL, 0, -1, -1,
			   set_init_arg);
550 551 552

	jump_label_init();

553 554 555 556
	/*
	 * These use large bootmem allocations and must precede
	 * kmem_cache_init()
	 */
557
	setup_log_buf(0);
558 559 560 561
	pidhash_init();
	vfs_caches_init_early();
	sort_main_extable();
	trap_init();
P
Pekka Enberg 已提交
562
	mm_init();
563

L
Linus Torvalds 已提交
564 565 566 567 568 569 570 571 572 573 574
	/*
	 * 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();
575
	if (WARN(!irqs_disabled(), "Interrupts were enabled *very* early, fixing it\n"))
576
		local_irq_disable();
P
Peter Zijlstra 已提交
577
	idr_init_cache();
L
Linus Torvalds 已提交
578
	rcu_init();
579
	tick_nohz_init();
580
	context_tracking_init();
Y
Yinghai Lu 已提交
581
	radix_tree_init();
582 583
	/* init some links before init_ISA_irqs() */
	early_irq_init();
L
Linus Torvalds 已提交
584
	init_IRQ();
T
Thomas Gleixner 已提交
585
	tick_init();
L
Linus Torvalds 已提交
586
	init_timers();
587
	hrtimers_init();
L
Linus Torvalds 已提交
588
	softirq_init();
589
	timekeeping_init();
J
john stultz 已提交
590
	time_init();
591
	sched_clock_postinit();
592
	perf_event_init();
593
	profile_init();
594
	call_function_init();
595
	WARN(!irqs_disabled(), "Interrupts were enabled early\n");
596
	early_boot_irqs_disabled = false;
597
	local_irq_enable();
598

599
	kmem_cache_init_late();
L
Linus Torvalds 已提交
600 601 602 603 604 605 606 607

	/*
	 * 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)
608 609
		panic("Too many boot %s vars at `%s'", panic_later,
		      panic_param);
I
Ingo Molnar 已提交
610 611 612

	lockdep_info();

613 614 615 616 617 618 619
	/*
	 * 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();

L
Linus Torvalds 已提交
620 621
#ifdef CONFIG_BLK_DEV_INITRD
	if (initrd_start && !initrd_below_start_ok &&
622
	    page_to_pfn(virt_to_page((void *)initrd_start)) < min_low_pfn) {
A
Andrew Morton 已提交
623
		pr_crit("initrd overwritten (0x%08lx < 0x%08lx) - disabling it.\n",
624 625
		    page_to_pfn(virt_to_page((void *)initrd_start)),
		    min_low_pfn);
L
Linus Torvalds 已提交
626 627 628
		initrd_start = 0;
	}
#endif
629
	page_cgroup_init();
630
	debug_objects_mem_init();
631
	kmemleak_init();
632
	setup_per_cpu_pageset();
L
Linus Torvalds 已提交
633 634 635
	numa_policy_init();
	if (late_time_init)
		late_time_init();
636
	sched_clock_init();
L
Linus Torvalds 已提交
637 638 639
	calibrate_delay();
	pidmap_init();
	anon_vma_init();
640
	acpi_early_init();
641
#ifdef CONFIG_X86
642
	if (efi_enabled(EFI_RUNTIME_SERVICES))
643
		efi_enter_virtual_mode();
644
#endif
645
#ifdef CONFIG_X86_ESPFIX64
646 647
	/* Should be run before the first non-init thread is created */
	init_espfix_bsp();
648
#endif
649
	thread_info_cache_init();
D
David Howells 已提交
650
	cred_init();
651
	fork_init(totalram_pages);
L
Linus Torvalds 已提交
652 653 654 655
	proc_caches_init();
	buffer_init();
	key_init();
	security_init();
656
	dbg_late_init();
657
	vfs_caches_init(totalram_pages);
L
Linus Torvalds 已提交
658 659 660 661
	signals_init();
	/* rootfs populating might need page-writeback */
	page_writeback_init();
	proc_root_init();
662
	cgroup_init();
L
Linus Torvalds 已提交
663
	cpuset_init();
664
	taskstats_init_early();
665
	delayacct_init();
L
Linus Torvalds 已提交
666 667 668

	check_bugs();

669
	sfi_init_late();
L
Linus Torvalds 已提交
670

671
	if (efi_enabled(EFI_RUNTIME_SERVICES)) {
672
		efi_late_init();
673
		efi_free_boot_services();
674
	}
675

676 677
	ftrace_init();

L
Linus Torvalds 已提交
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;
R
Rusty Russell 已提交
694
core_param(initcall_debug, initcall_debug, bool, 0644);
L
Linus Torvalds 已提交
695

696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759
#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 list_head *tmp;
	struct blacklist_entry *entry;
	char *fn_name;

	fn_name = kasprintf(GFP_KERNEL, "%pf", fn);
	if (!fn_name)
		return false;

	list_for_each(tmp, &blacklisted_initcalls) {
		entry = list_entry(tmp, struct blacklist_entry, next);
		if (!strcmp(fn_name, entry->buf)) {
			pr_debug("initcall %s blacklisted\n", fn_name);
			kfree(fn_name);
			return true;
		}
	}

	kfree(fn_name);
	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);

760
static int __init_or_module do_one_initcall_debug(initcall_t fn)
L
Linus Torvalds 已提交
761
{
762
	ktime_t calltime, delta, rettime;
A
Américo Wang 已提交
763 764
	unsigned long long duration;
	int ret;
L
Linus Torvalds 已提交
765

766
	printk(KERN_DEBUG "calling  %pF @ %i\n", fn, task_pid_nr(current));
767
	calltime = ktime_get();
A
Américo Wang 已提交
768
	ret = fn();
769 770 771
	rettime = ktime_get();
	delta = ktime_sub(rettime, calltime);
	duration = (unsigned long long) ktime_to_ns(delta) >> 10;
772
	printk(KERN_DEBUG "initcall %pF returned %d after %lld usecs\n",
A
Andrew Morton 已提交
773
		 fn, ret, duration);
L
Linus Torvalds 已提交
774

775 776 777
	return ret;
}

778
int __init_or_module do_one_initcall(initcall_t fn)
779 780 781
{
	int count = preempt_count();
	int ret;
782
	char msgbuf[64];
783

784 785 786
	if (initcall_blacklisted(fn))
		return -EPERM;

787 788 789 790
	if (initcall_debug)
		ret = do_one_initcall_debug(fn);
	else
		ret = fn();
791

792
	msgbuf[0] = 0;
793

794
	if (preempt_count() != count) {
795
		sprintf(msgbuf, "preemption imbalance ");
796
		preempt_count_set(count);
L
Linus Torvalds 已提交
797
	}
798
	if (irqs_disabled()) {
799
		strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf));
800 801
		local_irq_enable();
	}
802
	WARN(msgbuf[0], "initcall %pF returned with %s\n", fn, msgbuf);
803

A
Américo Wang 已提交
804
	return ret;
805 806 807
}


808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830
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,
};

831
/* Keep these in sync with initcalls in include/linux/init.h */
832
static char *initcall_level_names[] __initdata = {
833 834 835 836 837 838 839 840
	"early",
	"core",
	"postcore",
	"arch",
	"subsys",
	"fs",
	"device",
	"late",
841 842 843
};

static void __init do_initcall_level(int level)
844
{
845
	extern const struct kernel_param __start___param[], __stop___param[];
846
	initcall_t *fn;
847

848
	strcpy(initcall_command_line, saved_command_line);
849
	parse_args(initcall_level_names[level],
850
		   initcall_command_line, __start___param,
851 852
		   __stop___param - __start___param,
		   level, level,
853
		   &repair_env_string);
854 855

	for (fn = initcall_levels[level]; fn < initcall_levels[level+1]; fn++)
856
		do_one_initcall(*fn);
L
Linus Torvalds 已提交
857 858
}

859 860 861 862
static void __init do_initcalls(void)
{
	int level;

863
	for (level = 0; level < ARRAY_SIZE(initcall_levels) - 1; level++)
864 865 866
		do_initcall_level(level);
}

L
Linus Torvalds 已提交
867 868 869 870 871 872 873 874 875
/*
 * 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)
{
876
	cpuset_init_smp();
L
Linus Torvalds 已提交
877
	usermodehelper_init();
878
	shmem_init();
L
Linus Torvalds 已提交
879
	driver_init();
880
	init_irq_proc();
881
	do_ctors();
882
	usermodehelper_enable();
883
	do_initcalls();
884
	random_int_secret_init();
L
Linus Torvalds 已提交
885 886
}

887
static void __init do_pre_smp_initcalls(void)
888
{
889
	initcall_t *fn;
890

891
	for (fn = __initcall_start; fn < __initcall0_start; fn++)
892
		do_one_initcall(*fn);
893 894
}

895 896 897 898 899 900 901 902 903 904 905
/*
 * 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();
}

906
static int run_init_process(const char *init_filename)
L
Linus Torvalds 已提交
907 908
{
	argv_init[0] = init_filename;
909
	return do_execve(getname_kernel(init_filename),
910 911
		(const char __user *const __user *)argv_init,
		(const char __user *const __user *)envp_init);
L
Linus Torvalds 已提交
912 913
}

914 915 916 917 918 919 920 921 922 923 924 925 926 927
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;
}

928
static noinline void __init kernel_init_freeable(void);
929 930

static int __ref kernel_init(void *unused)
931
{
932 933
	int ret;

934
	kernel_init_freeable();
935 936
	/* need to finish all async __init code before freeing the memory */
	async_synchronize_full();
937 938 939 940 941
	free_initmem();
	mark_rodata_ro();
	system_state = SYSTEM_RUNNING;
	numa_default_policy();

A
Al Viro 已提交
942
	flush_delayed_fput();
943

944
	if (ramdisk_execute_command) {
945 946
		ret = run_init_process(ramdisk_execute_command);
		if (!ret)
947
			return 0;
948 949
		pr_err("Failed to execute %s (error %d)\n",
		       ramdisk_execute_command, ret);
950 951 952 953 954 955 956 957 958
	}

	/*
	 * 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) {
959 960
		ret = run_init_process(execute_command);
		if (!ret)
961
			return 0;
962 963
		pr_err("Failed to execute %s (error %d).  Attempting defaults...\n",
			execute_command, ret);
964
	}
965 966 967 968
	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"))
969
		return 0;
970

971
	panic("No working init found.  Try passing init= option to kernel. "
972
	      "See Linux Documentation/init.txt for guidance.");
973 974
}

975
static noinline void __init kernel_init_freeable(void)
L
Linus Torvalds 已提交
976
{
977 978 979 980
	/*
	 * Wait until kthreadd is all set-up.
	 */
	wait_for_completion(&kthreadd_done);
981 982 983 984

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

985 986 987
	/*
	 * init can allocate pages on any node
	 */
988
	set_mems_allowed(node_states[N_MEMORY]);
L
Linus Torvalds 已提交
989 990 991
	/*
	 * init can run on any cpu.
	 */
992
	set_cpus_allowed_ptr(current, cpu_all_mask);
L
Linus Torvalds 已提交
993

994 995
	cad_pid = task_pid(current);

996
	smp_prepare_cpus(setup_max_cpus);
L
Linus Torvalds 已提交
997 998

	do_pre_smp_initcalls();
999
	lockup_detector_init();
L
Linus Torvalds 已提交
1000 1001 1002 1003 1004 1005

	smp_init();
	sched_init_smp();

	do_basic_setup();

1006 1007
	/* Open the /dev/console on the rootfs, this should never fail */
	if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0)
A
Andrew Morton 已提交
1008
		pr_err("Warning: unable to open an initial console.\n");
1009 1010 1011

	(void) sys_dup(0);
	(void) sys_dup(0);
L
Linus Torvalds 已提交
1012 1013 1014 1015
	/*
	 * check if there is an early userspace init.  If yes, let it do all
	 * the work
	 */
1016 1017 1018 1019 1020 1021

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

	if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) {
		ramdisk_execute_command = NULL;
L
Linus Torvalds 已提交
1022
		prepare_namespace();
1023
	}
L
Linus Torvalds 已提交
1024 1025 1026 1027 1028 1029

	/*
	 * 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..
	 */
1030 1031 1032

	/* rootfs is available now, try loading default modules */
	load_default_modules();
L
Linus Torvalds 已提交
1033
}