main.c 21.9 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>
L
Linus Torvalds 已提交
79 80 81 82

#include <asm/io.h>
#include <asm/bugs.h>
#include <asm/setup.h>
83
#include <asm/sections.h>
84
#include <asm/cacheflush.h>
L
Linus Torvalds 已提交
85 86 87 88 89

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

90
static int kernel_init(void *);
L
Linus Torvalds 已提交
91 92 93 94 95 96

extern void init_IRQ(void);
extern void fork_init(unsigned long);
extern void mca_init(void);
extern void sbus_init(void);
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

#ifdef CONFIG_TC
extern void tc_init(void);
#endif

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;
L
Linus Torvalds 已提交
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);
L
Linus Torvalds 已提交
126 127
extern void softirq_init(void);

128 129 130 131 132 133
/* 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;
L
Linus Torvalds 已提交
134 135

static char *execute_command;
136
static char *ramdisk_execute_command;
L
Linus Torvalds 已提交
137

J
Jan Beulich 已提交
138 139 140 141 142 143 144 145 146 147 148
/*
 * 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
149

150 151 152 153 154 155 156 157
static int __init set_reset_devices(char *str)
{
	reset_devices = 1;
	return 1;
}

__setup("reset_devices", set_reset_devices);

158 159
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
160 161
static const char *panic_later, *panic_param;

162
extern const struct obs_kernel_param __setup_start[], __setup_end[];
L
Linus Torvalds 已提交
163 164 165

static int __init obsolete_checksetup(char *line)
{
166
	const struct obs_kernel_param *p;
167
	int had_early_param = 0;
L
Linus Torvalds 已提交
168 169 170 171

	p = __setup_start;
	do {
		int n = strlen(p->str);
172
		if (parameqn(line, p->str, n)) {
L
Linus Torvalds 已提交
173
			if (p->early) {
174 175 176 177
				/* 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
178
				if (line[n] == '\0' || line[n] == '=')
179
					had_early_param = 1;
L
Linus Torvalds 已提交
180
			} else if (!p->setup_func) {
A
Andrew Morton 已提交
181 182
				pr_warn("Parameter %s is obsolete, ignored\n",
					p->str);
L
Linus Torvalds 已提交
183 184 185 186 187 188
				return 1;
			} else if (p->setup_func(line + n))
				return 1;
		}
		p++;
	} while (p < __setup_end);
189 190

	return had_early_param;
L
Linus Torvalds 已提交
191 192 193 194 195 196 197 198 199 200 201 202 203
}

/*
 * 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)
{
	console_loglevel = 10;
204
	return 0;
L
Linus Torvalds 已提交
205 206 207 208 209
}

static int __init quiet_kernel(char *str)
{
	console_loglevel = 4;
210
	return 0;
L
Linus Torvalds 已提交
211 212
}

213 214
early_param("debug", debug_kernel);
early_param("quiet", quiet_kernel);
L
Linus Torvalds 已提交
215 216 217

static int __init loglevel(char *str)
{
218 219 220 221 222 223 224 225 226 227 228 229 230
	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
231 232
}

233
early_param("loglevel", loglevel);
L
Linus Torvalds 已提交
234

235
/* Change NUL term back to "=", to make "param" the whole string. */
236
static int __init repair_env_string(char *param, char *val, const char *unused)
L
Linus Torvalds 已提交
237 238 239 240 241 242 243 244 245 246 247 248
{
	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();
	}
249 250 251 252 253 254 255
	return 0;
}

/*
 * Unknown boot options get handed to init, unless they look like
 * unused parameters (modprobe will find them in /proc/cmdline).
 */
256
static int __init unknown_bootoption(char *param, char *val, const char *unused)
257
{
258
	repair_env_string(param, val, unused);
L
Linus Torvalds 已提交
259 260 261 262 263

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

264 265
	/* Unused module parameter. */
	if (strchr(param, '.') && (!val || strchr(param, '.') < val))
L
Linus Torvalds 已提交
266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313
		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) {
				panic_later = "Too many boot env vars at `%s'";
				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) {
				panic_later = "Too many boot init vars at `%s'";
				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);

314 315 316 317 318 319 320 321 322 323 324 325
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
326
#ifndef CONFIG_SMP
327
static const unsigned int setup_max_cpus = NR_CPUS;
L
Linus Torvalds 已提交
328 329 330 331 332 333 334 335 336
#ifdef CONFIG_X86_LOCAL_APIC
static void __init smp_init(void)
{
	APIC_init_uniprocessor();
}
#else
#define smp_init()	do { } while (0)
#endif

337
static inline void setup_nr_cpu_ids(void) { }
L
Linus Torvalds 已提交
338 339 340
static inline void smp_prepare_cpus(unsigned int maxcpus) { }
#endif

341 342 343 344 345 346 347 348 349 350 351 352 353 354
/*
 * 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)
{
	saved_command_line = alloc_bootmem(strlen (boot_command_line)+1);
	static_command_line = alloc_bootmem(strlen (command_line)+1);
	strcpy (saved_command_line, boot_command_line);
	strcpy (static_command_line, command_line);
}

L
Linus Torvalds 已提交
355 356 357 358 359 360 361 362 363
/*
 * 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.
 */

364 365
static __initdata DECLARE_COMPLETION(kthreadd_done);

366
static noinline void __init_refok rest_init(void)
L
Linus Torvalds 已提交
367
{
368 369
	int pid;

370
	rcu_scheduler_starting();
371
	/*
P
Peter Zijlstra 已提交
372
	 * We need to spawn init first so that it obtains pid 1, however
373 374 375
	 * the init task will end up wanting to create kthreads, which, if
	 * we schedule it before we create kthreadd, will OOPS.
	 */
376
	kernel_thread(kernel_init, NULL, CLONE_FS | CLONE_SIGHAND);
L
Linus Torvalds 已提交
377
	numa_default_policy();
378
	pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES);
379
	rcu_read_lock();
P
Pavel Emelyanov 已提交
380
	kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns);
381
	rcu_read_unlock();
382
	complete(&kthreadd_done);
383 384 385

	/*
	 * The boot idle thread must execute schedule()
I
Ingo Molnar 已提交
386
	 * at least once to get things moving:
387
	 */
I
Ingo Molnar 已提交
388
	init_idle_bootup_task(current);
389
	schedule_preempt_disabled();
390
	/* Call into cpu_idle with preempt disabled */
391
	cpu_startup_entry(CPUHP_ONLINE);
I
Ingo Molnar 已提交
392
}
L
Linus Torvalds 已提交
393 394

/* Check for early params. */
395
static int __init do_early_param(char *param, char *val, const char *unused)
L
Linus Torvalds 已提交
396
{
397
	const struct obs_kernel_param *p;
L
Linus Torvalds 已提交
398 399

	for (p = __setup_start; p < __setup_end; p++) {
400
		if ((p->early && parameq(param, p->str)) ||
401 402 403
		    (strcmp(param, "console") == 0 &&
		     strcmp(p->str, "earlycon") == 0)
		) {
L
Linus Torvalds 已提交
404
			if (p->setup_func(val) != 0)
A
Andrew Morton 已提交
405
				pr_warn("Malformed early option '%s'\n", param);
L
Linus Torvalds 已提交
406 407 408 409 410 411
		}
	}
	/* We accept everything at this stage. */
	return 0;
}

M
Magnus Damm 已提交
412 413
void __init parse_early_options(char *cmdline)
{
414
	parse_args("early options", cmdline, NULL, 0, 0, 0, do_early_param);
M
Magnus Damm 已提交
415 416
}

L
Linus Torvalds 已提交
417 418 419 420 421 422 423 424 425 426
/* 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. */
427
	strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE);
M
Magnus Damm 已提交
428
	parse_early_options(tmp_cmdline);
L
Linus Torvalds 已提交
429 430 431 432 433 434 435
	done = 1;
}

/*
 *	Activate the first processor.
 */

436 437 438 439
static void __init boot_cpu_init(void)
{
	int cpu = smp_processor_id();
	/* Mark the boot cpu "present", "online" etc for SMP and UP case */
440
	set_cpu_online(cpu, true);
441
	set_cpu_active(cpu, true);
442 443
	set_cpu_present(cpu, true);
	set_cpu_possible(cpu, true);
444 445
}

446
void __init __weak smp_setup_processor_id(void)
447 448 449
{
}

450
# if THREAD_SIZE >= PAGE_SIZE
451 452 453
void __init __weak thread_info_cache_init(void)
{
}
454
#endif
455

P
Pekka Enberg 已提交
456 457 458 459 460
/*
 * Set up kernel memory allocators
 */
static void __init mm_init(void)
{
461
	/*
462 463
	 * page_cgroup requires contiguous pages,
	 * bigger than MAX_ORDER unless SPARSEMEM.
464 465
	 */
	page_cgroup_init_flatmem();
P
Pekka Enberg 已提交
466 467
	mem_init();
	kmem_cache_init();
468
	percpu_init_late();
469
	pgtable_cache_init();
P
Pekka Enberg 已提交
470 471 472
	vmalloc_init();
}

L
Linus Torvalds 已提交
473 474 475
asmlinkage void __init start_kernel(void)
{
	char * command_line;
476
	extern const struct kernel_param __start___param[], __stop___param[];
477

I
Ingo Molnar 已提交
478 479 480 481 482
	/*
	 * Need to run as early as possible, to initialize the
	 * lockdep hash:
	 */
	lockdep_init();
483
	smp_setup_processor_id();
484
	debug_objects_early_init();
485 486 487 488 489 490

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

491
	cgroup_init_early();
I
Ingo Molnar 已提交
492 493

	local_irq_disable();
494
	early_boot_irqs_disabled = true;
I
Ingo Molnar 已提交
495

L
Linus Torvalds 已提交
496 497 498 499
/*
 * Interrupts are still disabled. Do necessary setups, then
 * enable them
 */
500
	boot_cpu_init();
L
Linus Torvalds 已提交
501
	page_address_init();
A
Andrew Morton 已提交
502
	pr_notice("%s", linux_banner);
L
Linus Torvalds 已提交
503
	setup_arch(&command_line);
504
	mm_init_owner(&init_mm, &init_task);
505
	mm_init_cpumask(&init_mm);
506
	setup_command_line(command_line);
507
	setup_nr_cpu_ids();
508
	setup_per_cpu_areas();
509
	smp_prepare_boot_cpu();	/* arch-specific boot-cpu hooks */
L
Linus Torvalds 已提交
510

511
	build_all_zonelists(NULL, NULL);
512 513
	page_alloc_init();

A
Andrew Morton 已提交
514
	pr_notice("Kernel command line: %s\n", boot_command_line);
515 516 517
	parse_early_param();
	parse_args("Booting kernel", static_command_line, __start___param,
		   __stop___param - __start___param,
518
		   -1, -1, &unknown_bootoption);
519 520 521

	jump_label_init();

522 523 524 525
	/*
	 * These use large bootmem allocations and must precede
	 * kmem_cache_init()
	 */
526
	setup_log_buf(0);
527 528 529 530
	pidhash_init();
	vfs_caches_init_early();
	sort_main_extable();
	trap_init();
P
Pekka Enberg 已提交
531
	mm_init();
532

L
Linus Torvalds 已提交
533 534 535 536 537 538 539 540 541 542 543
	/*
	 * 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();
544
	if (WARN(!irqs_disabled(), "Interrupts were enabled *very* early, fixing it\n"))
545
		local_irq_disable();
P
Peter Zijlstra 已提交
546
	idr_init_cache();
L
Linus Torvalds 已提交
547
	rcu_init();
548
	tick_nohz_init();
549
	context_tracking_init();
Y
Yinghai Lu 已提交
550
	radix_tree_init();
551 552
	/* init some links before init_ISA_irqs() */
	early_irq_init();
L
Linus Torvalds 已提交
553
	init_IRQ();
T
Thomas Gleixner 已提交
554
	tick_init();
L
Linus Torvalds 已提交
555
	init_timers();
556
	hrtimers_init();
L
Linus Torvalds 已提交
557
	softirq_init();
558
	timekeeping_init();
J
john stultz 已提交
559
	time_init();
560
	sched_clock_postinit();
561
	perf_event_init();
562
	profile_init();
563
	call_function_init();
564
	WARN(!irqs_disabled(), "Interrupts were enabled early\n");
565
	early_boot_irqs_disabled = false;
566
	local_irq_enable();
567

568
	kmem_cache_init_late();
L
Linus Torvalds 已提交
569 570 571 572 573 574 575 576 577

	/*
	 * 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)
		panic(panic_later, panic_param);
I
Ingo Molnar 已提交
578 579 580

	lockdep_info();

581 582 583 584 585 586 587
	/*
	 * 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
588 589
#ifdef CONFIG_BLK_DEV_INITRD
	if (initrd_start && !initrd_below_start_ok &&
590
	    page_to_pfn(virt_to_page((void *)initrd_start)) < min_low_pfn) {
A
Andrew Morton 已提交
591
		pr_crit("initrd overwritten (0x%08lx < 0x%08lx) - disabling it.\n",
592 593
		    page_to_pfn(virt_to_page((void *)initrd_start)),
		    min_low_pfn);
L
Linus Torvalds 已提交
594 595 596
		initrd_start = 0;
	}
#endif
597
	page_cgroup_init();
598
	debug_objects_mem_init();
599
	kmemleak_init();
600
	setup_per_cpu_pageset();
L
Linus Torvalds 已提交
601 602 603
	numa_policy_init();
	if (late_time_init)
		late_time_init();
604
	sched_clock_init();
L
Linus Torvalds 已提交
605 606 607
	calibrate_delay();
	pidmap_init();
	anon_vma_init();
608
#ifdef CONFIG_X86
609
	if (efi_enabled(EFI_RUNTIME_SERVICES))
610 611
		efi_enter_virtual_mode();
#endif
612
	thread_info_cache_init();
D
David Howells 已提交
613
	cred_init();
614
	fork_init(totalram_pages);
L
Linus Torvalds 已提交
615 616 617 618
	proc_caches_init();
	buffer_init();
	key_init();
	security_init();
619
	dbg_late_init();
620
	vfs_caches_init(totalram_pages);
L
Linus Torvalds 已提交
621 622 623 624 625 626
	signals_init();
	/* rootfs populating might need page-writeback */
	page_writeback_init();
#ifdef CONFIG_PROC_FS
	proc_root_init();
#endif
627
	cgroup_init();
L
Linus Torvalds 已提交
628
	cpuset_init();
629
	taskstats_init_early();
630
	delayacct_init();
L
Linus Torvalds 已提交
631 632 633 634

	check_bugs();

	acpi_early_init(); /* before LAPIC and SMP init */
635
	sfi_init_late();
L
Linus Torvalds 已提交
636

637
	if (efi_enabled(EFI_RUNTIME_SERVICES)) {
638
		efi_late_init();
639
		efi_free_boot_services();
640
	}
641

642 643
	ftrace_init();

L
Linus Torvalds 已提交
644 645 646 647
	/* Do the rest non-__init'ed, we're now alive */
	rest_init();
}

648 649 650 651
/* Call all constructor functions linked into the kernel. */
static void __init do_ctors(void)
{
#ifdef CONFIG_CONSTRUCTORS
652
	ctor_fn_t *fn = (ctor_fn_t *) __ctors_start;
653

654 655
	for (; fn < (ctor_fn_t *) __ctors_end; fn++)
		(*fn)();
656 657 658
#endif
}

659
bool initcall_debug;
R
Rusty Russell 已提交
660
core_param(initcall_debug, initcall_debug, bool, 0644);
L
Linus Torvalds 已提交
661

662
static int __init_or_module do_one_initcall_debug(initcall_t fn)
L
Linus Torvalds 已提交
663
{
664
	ktime_t calltime, delta, rettime;
A
Américo Wang 已提交
665 666
	unsigned long long duration;
	int ret;
L
Linus Torvalds 已提交
667

A
Andrew Morton 已提交
668
	pr_debug("calling  %pF @ %i\n", fn, task_pid_nr(current));
669
	calltime = ktime_get();
A
Américo Wang 已提交
670
	ret = fn();
671 672 673
	rettime = ktime_get();
	delta = ktime_sub(rettime, calltime);
	duration = (unsigned long long) ktime_to_ns(delta) >> 10;
A
Andrew Morton 已提交
674 675
	pr_debug("initcall %pF returned %d after %lld usecs\n",
		 fn, ret, duration);
L
Linus Torvalds 已提交
676

677 678 679
	return ret;
}

680
int __init_or_module do_one_initcall(initcall_t fn)
681 682 683
{
	int count = preempt_count();
	int ret;
684
	char msgbuf[64];
685 686 687 688 689

	if (initcall_debug)
		ret = do_one_initcall_debug(fn);
	else
		ret = fn();
690

691
	msgbuf[0] = 0;
692

693
	if (preempt_count() != count) {
694
		sprintf(msgbuf, "preemption imbalance ");
695
		preempt_count() = count;
L
Linus Torvalds 已提交
696
	}
697
	if (irqs_disabled()) {
698
		strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf));
699 700
		local_irq_enable();
	}
701
	WARN(msgbuf[0], "initcall %pF returned with %s\n", fn, msgbuf);
702

A
Américo Wang 已提交
703
	return ret;
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
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,
};

730
/* Keep these in sync with initcalls in include/linux/init.h */
731
static char *initcall_level_names[] __initdata = {
732 733 734 735 736 737 738 739
	"early",
	"core",
	"postcore",
	"arch",
	"subsys",
	"fs",
	"device",
	"late",
740 741 742
};

static void __init do_initcall_level(int level)
743
{
744
	extern const struct kernel_param __start___param[], __stop___param[];
745
	initcall_t *fn;
746

747 748 749 750 751
	strcpy(static_command_line, saved_command_line);
	parse_args(initcall_level_names[level],
		   static_command_line, __start___param,
		   __stop___param - __start___param,
		   level, level,
752
		   &repair_env_string);
753 754

	for (fn = initcall_levels[level]; fn < initcall_levels[level+1]; fn++)
755
		do_one_initcall(*fn);
L
Linus Torvalds 已提交
756 757
}

758 759 760 761
static void __init do_initcalls(void)
{
	int level;

762
	for (level = 0; level < ARRAY_SIZE(initcall_levels) - 1; level++)
763 764 765
		do_initcall_level(level);
}

L
Linus Torvalds 已提交
766 767 768 769 770 771 772 773 774
/*
 * 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)
{
775
	cpuset_init_smp();
L
Linus Torvalds 已提交
776
	usermodehelper_init();
777
	shmem_init();
L
Linus Torvalds 已提交
778
	driver_init();
779
	init_irq_proc();
780
	do_ctors();
781
	usermodehelper_enable();
782
	do_initcalls();
L
Linus Torvalds 已提交
783 784
}

785
static void __init do_pre_smp_initcalls(void)
786
{
787
	initcall_t *fn;
788

789
	for (fn = __initcall_start; fn < __initcall0_start; fn++)
790
		do_one_initcall(*fn);
791 792
}

793 794 795 796 797 798 799 800 801 802 803
/*
 * 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();
}

804
static int run_init_process(const char *init_filename)
L
Linus Torvalds 已提交
805 806
{
	argv_init[0] = init_filename;
807 808 809
	return do_execve(init_filename,
		(const char __user *const __user *)argv_init,
		(const char __user *const __user *)envp_init);
L
Linus Torvalds 已提交
810 811
}

812
static noinline void __init kernel_init_freeable(void);
813 814

static int __ref kernel_init(void *unused)
815
{
816
	kernel_init_freeable();
817 818
	/* need to finish all async __init code before freeing the memory */
	async_synchronize_full();
819 820 821 822 823
	free_initmem();
	mark_rodata_ro();
	system_state = SYSTEM_RUNNING;
	numa_default_policy();

A
Al Viro 已提交
824
	flush_delayed_fput();
825

826
	if (ramdisk_execute_command) {
827 828
		if (!run_init_process(ramdisk_execute_command))
			return 0;
A
Andrew Morton 已提交
829
		pr_err("Failed to execute %s\n", ramdisk_execute_command);
830 831 832 833 834 835 836 837 838
	}

	/*
	 * 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) {
839 840
		if (!run_init_process(execute_command))
			return 0;
A
Andrew Morton 已提交
841 842
		pr_err("Failed to execute %s.  Attempting defaults...\n",
			execute_command);
843
	}
844 845 846 847 848
	if (!run_init_process("/sbin/init") ||
	    !run_init_process("/etc/init") ||
	    !run_init_process("/bin/init") ||
	    !run_init_process("/bin/sh"))
		return 0;
849

850 851
	panic("No init found.  Try passing init= option to kernel. "
	      "See Linux Documentation/init.txt for guidance.");
852 853
}

854
static noinline void __init kernel_init_freeable(void)
L
Linus Torvalds 已提交
855
{
856 857 858 859
	/*
	 * Wait until kthreadd is all set-up.
	 */
	wait_for_completion(&kthreadd_done);
860 861 862 863

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

864 865 866
	/*
	 * init can allocate pages on any node
	 */
867
	set_mems_allowed(node_states[N_MEMORY]);
L
Linus Torvalds 已提交
868 869 870
	/*
	 * init can run on any cpu.
	 */
871
	set_cpus_allowed_ptr(current, cpu_all_mask);
L
Linus Torvalds 已提交
872

873 874
	cad_pid = task_pid(current);

875
	smp_prepare_cpus(setup_max_cpus);
L
Linus Torvalds 已提交
876 877

	do_pre_smp_initcalls();
878
	lockup_detector_init();
L
Linus Torvalds 已提交
879 880 881 882 883 884

	smp_init();
	sched_init_smp();

	do_basic_setup();

885 886
	/* 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
887
		pr_err("Warning: unable to open an initial console.\n");
888 889 890

	(void) sys_dup(0);
	(void) sys_dup(0);
L
Linus Torvalds 已提交
891 892 893 894
	/*
	 * check if there is an early userspace init.  If yes, let it do all
	 * the work
	 */
895 896 897 898 899 900

	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
901
		prepare_namespace();
902
	}
L
Linus Torvalds 已提交
903 904 905 906 907 908

	/*
	 * 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..
	 */
909 910 911

	/* rootfs is available now, try loading default modules */
	load_default_modules();
L
Linus Torvalds 已提交
912
}