main.c 21.8 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>
L
Linus Torvalds 已提交
77 78 79 80

#include <asm/io.h>
#include <asm/bugs.h>
#include <asm/setup.h>
81
#include <asm/sections.h>
82
#include <asm/cacheflush.h>
L
Linus Torvalds 已提交
83 84 85 86 87

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

88
static int kernel_init(void *);
L
Linus Torvalds 已提交
89 90 91 92 93 94

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);
95 96 97
#ifndef CONFIG_DEBUG_RODATA
static inline void mark_rodata_ro(void) { }
#endif
L
Linus Torvalds 已提交
98 99 100 101 102

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

103 104 105 106 107 108 109 110 111
/*
 * 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;

112
enum system_states system_state __read_mostly;
L
Linus Torvalds 已提交
113 114 115 116 117 118 119 120 121 122
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. */
123
void (*__initdata late_time_init)(void);
L
Linus Torvalds 已提交
124 125
extern void softirq_init(void);

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;
L
Linus Torvalds 已提交
132 133

static char *execute_command;
134
static char *ramdisk_execute_command;
L
Linus Torvalds 已提交
135

J
Jan Beulich 已提交
136 137 138 139 140 141 142 143 144 145 146
/*
 * 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
147

148 149 150 151 152 153 154 155
static int __init set_reset_devices(char *str)
{
	reset_devices = 1;
	return 1;
}

__setup("reset_devices", set_reset_devices);

156 157
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
158 159
static const char *panic_later, *panic_param;

160
extern const struct obs_kernel_param __setup_start[], __setup_end[];
L
Linus Torvalds 已提交
161 162 163

static int __init obsolete_checksetup(char *line)
{
164
	const struct obs_kernel_param *p;
165
	int had_early_param = 0;
L
Linus Torvalds 已提交
166 167 168 169

	p = __setup_start;
	do {
		int n = strlen(p->str);
170
		if (parameqn(line, p->str, n)) {
L
Linus Torvalds 已提交
171
			if (p->early) {
172 173 174 175
				/* 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
176
				if (line[n] == '\0' || line[n] == '=')
177
					had_early_param = 1;
L
Linus Torvalds 已提交
178
			} else if (!p->setup_func) {
A
Andrew Morton 已提交
179 180
				pr_warn("Parameter %s is obsolete, ignored\n",
					p->str);
L
Linus Torvalds 已提交
181 182 183 184 185 186
				return 1;
			} else if (p->setup_func(line + n))
				return 1;
		}
		p++;
	} while (p < __setup_end);
187 188

	return had_early_param;
L
Linus Torvalds 已提交
189 190 191 192 193 194 195 196 197 198 199 200 201
}

/*
 * 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;
202
	return 0;
L
Linus Torvalds 已提交
203 204 205 206 207
}

static int __init quiet_kernel(char *str)
{
	console_loglevel = 4;
208
	return 0;
L
Linus Torvalds 已提交
209 210
}

211 212
early_param("debug", debug_kernel);
early_param("quiet", quiet_kernel);
L
Linus Torvalds 已提交
213 214 215

static int __init loglevel(char *str)
{
216 217 218 219 220 221 222 223 224 225 226 227 228
	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
229 230
}

231
early_param("loglevel", loglevel);
L
Linus Torvalds 已提交
232

233
/* Change NUL term back to "=", to make "param" the whole string. */
234
static int __init repair_env_string(char *param, char *val, const char *unused)
L
Linus Torvalds 已提交
235 236 237 238 239 240 241 242 243 244 245 246
{
	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();
	}
247 248 249 250 251 252 253
	return 0;
}

/*
 * Unknown boot options get handed to init, unless they look like
 * unused parameters (modprobe will find them in /proc/cmdline).
 */
254
static int __init unknown_bootoption(char *param, char *val, const char *unused)
255
{
256
	repair_env_string(param, val, unused);
L
Linus Torvalds 已提交
257 258 259 260 261

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

262 263
	/* Unused module parameter. */
	if (strchr(param, '.') && (!val || strchr(param, '.') < val))
L
Linus Torvalds 已提交
264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311
		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);

312 313 314 315 316 317 318 319 320 321 322 323
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
324
#ifndef CONFIG_SMP
325
static const unsigned int setup_max_cpus = NR_CPUS;
L
Linus Torvalds 已提交
326 327 328 329 330 331 332 333 334
#ifdef CONFIG_X86_LOCAL_APIC
static void __init smp_init(void)
{
	APIC_init_uniprocessor();
}
#else
#define smp_init()	do { } while (0)
#endif

335
static inline void setup_nr_cpu_ids(void) { }
L
Linus Torvalds 已提交
336 337 338
static inline void smp_prepare_cpus(unsigned int maxcpus) { }
#endif

339 340 341 342 343 344 345 346 347 348 349 350 351 352
/*
 * 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
353 354 355 356 357 358 359 360 361
/*
 * 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.
 */

362 363
static __initdata DECLARE_COMPLETION(kthreadd_done);

364
static noinline void __init_refok rest_init(void)
L
Linus Torvalds 已提交
365
{
366 367
	int pid;

368
	rcu_scheduler_starting();
369
	/*
P
Peter Zijlstra 已提交
370
	 * We need to spawn init first so that it obtains pid 1, however
371 372 373
	 * the init task will end up wanting to create kthreads, which, if
	 * we schedule it before we create kthreadd, will OOPS.
	 */
374
	kernel_thread(kernel_init, NULL, CLONE_FS | CLONE_SIGHAND);
L
Linus Torvalds 已提交
375
	numa_default_policy();
376
	pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES);
377
	rcu_read_lock();
P
Pavel Emelyanov 已提交
378
	kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns);
379
	rcu_read_unlock();
380
	complete(&kthreadd_done);
381 382 383

	/*
	 * The boot idle thread must execute schedule()
I
Ingo Molnar 已提交
384
	 * at least once to get things moving:
385
	 */
I
Ingo Molnar 已提交
386
	init_idle_bootup_task(current);
387
	schedule_preempt_disabled();
388
	/* Call into cpu_idle with preempt disabled */
389
	cpu_startup_entry(CPUHP_ONLINE);
I
Ingo Molnar 已提交
390
}
L
Linus Torvalds 已提交
391 392

/* Check for early params. */
393
static int __init do_early_param(char *param, char *val, const char *unused)
L
Linus Torvalds 已提交
394
{
395
	const struct obs_kernel_param *p;
L
Linus Torvalds 已提交
396 397

	for (p = __setup_start; p < __setup_end; p++) {
398
		if ((p->early && parameq(param, p->str)) ||
399 400 401
		    (strcmp(param, "console") == 0 &&
		     strcmp(p->str, "earlycon") == 0)
		) {
L
Linus Torvalds 已提交
402
			if (p->setup_func(val) != 0)
A
Andrew Morton 已提交
403
				pr_warn("Malformed early option '%s'\n", param);
L
Linus Torvalds 已提交
404 405 406 407 408 409
		}
	}
	/* We accept everything at this stage. */
	return 0;
}

M
Magnus Damm 已提交
410 411
void __init parse_early_options(char *cmdline)
{
412
	parse_args("early options", cmdline, NULL, 0, 0, 0, do_early_param);
M
Magnus Damm 已提交
413 414
}

L
Linus Torvalds 已提交
415 416 417 418 419 420 421 422 423 424
/* 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. */
425
	strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE);
M
Magnus Damm 已提交
426
	parse_early_options(tmp_cmdline);
L
Linus Torvalds 已提交
427 428 429 430 431 432 433
	done = 1;
}

/*
 *	Activate the first processor.
 */

434 435 436 437
static void __init boot_cpu_init(void)
{
	int cpu = smp_processor_id();
	/* Mark the boot cpu "present", "online" etc for SMP and UP case */
438
	set_cpu_online(cpu, true);
439
	set_cpu_active(cpu, true);
440 441
	set_cpu_present(cpu, true);
	set_cpu_possible(cpu, true);
442 443
}

444
void __init __weak smp_setup_processor_id(void)
445 446 447
{
}

448
# if THREAD_SIZE >= PAGE_SIZE
449 450 451
void __init __weak thread_info_cache_init(void)
{
}
452
#endif
453

P
Pekka Enberg 已提交
454 455 456 457 458
/*
 * Set up kernel memory allocators
 */
static void __init mm_init(void)
{
459
	/*
460 461
	 * page_cgroup requires contiguous pages,
	 * bigger than MAX_ORDER unless SPARSEMEM.
462 463
	 */
	page_cgroup_init_flatmem();
P
Pekka Enberg 已提交
464 465
	mem_init();
	kmem_cache_init();
466
	percpu_init_late();
467
	pgtable_cache_init();
P
Pekka Enberg 已提交
468 469 470
	vmalloc_init();
}

L
Linus Torvalds 已提交
471 472 473
asmlinkage void __init start_kernel(void)
{
	char * command_line;
474
	extern const struct kernel_param __start___param[], __stop___param[];
475

I
Ingo Molnar 已提交
476 477 478 479 480
	/*
	 * Need to run as early as possible, to initialize the
	 * lockdep hash:
	 */
	lockdep_init();
481
	smp_setup_processor_id();
482
	debug_objects_early_init();
483 484 485 486 487 488

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

489
	cgroup_init_early();
I
Ingo Molnar 已提交
490 491

	local_irq_disable();
492
	early_boot_irqs_disabled = true;
I
Ingo Molnar 已提交
493

L
Linus Torvalds 已提交
494 495 496 497
/*
 * Interrupts are still disabled. Do necessary setups, then
 * enable them
 */
498
	boot_cpu_init();
L
Linus Torvalds 已提交
499
	page_address_init();
A
Andrew Morton 已提交
500
	pr_notice("%s", linux_banner);
L
Linus Torvalds 已提交
501
	setup_arch(&command_line);
502
	mm_init_owner(&init_mm, &init_task);
503
	mm_init_cpumask(&init_mm);
504
	setup_command_line(command_line);
505
	setup_nr_cpu_ids();
506
	setup_per_cpu_areas();
507
	smp_prepare_boot_cpu();	/* arch-specific boot-cpu hooks */
L
Linus Torvalds 已提交
508

509
	build_all_zonelists(NULL, NULL);
510 511
	page_alloc_init();

A
Andrew Morton 已提交
512
	pr_notice("Kernel command line: %s\n", boot_command_line);
513 514 515
	parse_early_param();
	parse_args("Booting kernel", static_command_line, __start___param,
		   __stop___param - __start___param,
516
		   -1, -1, &unknown_bootoption);
517 518 519

	jump_label_init();

520 521 522 523
	/*
	 * These use large bootmem allocations and must precede
	 * kmem_cache_init()
	 */
524
	setup_log_buf(0);
525 526 527 528
	pidhash_init();
	vfs_caches_init_early();
	sort_main_extable();
	trap_init();
P
Pekka Enberg 已提交
529
	mm_init();
530

L
Linus Torvalds 已提交
531 532 533 534 535 536 537 538 539 540 541
	/*
	 * 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();
542
	if (WARN(!irqs_disabled(), "Interrupts were enabled *very* early, fixing it\n"))
543
		local_irq_disable();
P
Peter Zijlstra 已提交
544
	idr_init_cache();
L
Linus Torvalds 已提交
545
	rcu_init();
546
	tick_nohz_init();
Y
Yinghai Lu 已提交
547
	radix_tree_init();
548 549
	/* init some links before init_ISA_irqs() */
	early_irq_init();
L
Linus Torvalds 已提交
550
	init_IRQ();
T
Thomas Gleixner 已提交
551
	tick_init();
L
Linus Torvalds 已提交
552
	init_timers();
553
	hrtimers_init();
L
Linus Torvalds 已提交
554
	softirq_init();
555
	timekeeping_init();
J
john stultz 已提交
556
	time_init();
557
	perf_event_init();
558
	profile_init();
559
	call_function_init();
560
	WARN(!irqs_disabled(), "Interrupts were enabled early\n");
561
	early_boot_irqs_disabled = false;
562
	local_irq_enable();
563

564
	kmem_cache_init_late();
L
Linus Torvalds 已提交
565 566 567 568 569 570 571 572 573

	/*
	 * 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
574 575 576

	lockdep_info();

577 578 579 580 581 582 583
	/*
	 * 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
584 585
#ifdef CONFIG_BLK_DEV_INITRD
	if (initrd_start && !initrd_below_start_ok &&
586
	    page_to_pfn(virt_to_page((void *)initrd_start)) < min_low_pfn) {
A
Andrew Morton 已提交
587
		pr_crit("initrd overwritten (0x%08lx < 0x%08lx) - disabling it.\n",
588 589
		    page_to_pfn(virt_to_page((void *)initrd_start)),
		    min_low_pfn);
L
Linus Torvalds 已提交
590 591 592
		initrd_start = 0;
	}
#endif
593
	page_cgroup_init();
594
	debug_objects_mem_init();
595
	kmemleak_init();
596
	setup_per_cpu_pageset();
L
Linus Torvalds 已提交
597 598 599
	numa_policy_init();
	if (late_time_init)
		late_time_init();
600
	sched_clock_init();
L
Linus Torvalds 已提交
601 602 603
	calibrate_delay();
	pidmap_init();
	anon_vma_init();
604
#ifdef CONFIG_X86
605
	if (efi_enabled(EFI_RUNTIME_SERVICES))
606 607
		efi_enter_virtual_mode();
#endif
608
	thread_info_cache_init();
D
David Howells 已提交
609
	cred_init();
610
	fork_init(totalram_pages);
L
Linus Torvalds 已提交
611 612 613 614
	proc_caches_init();
	buffer_init();
	key_init();
	security_init();
615
	dbg_late_init();
616
	vfs_caches_init(totalram_pages);
L
Linus Torvalds 已提交
617 618 619 620 621 622
	signals_init();
	/* rootfs populating might need page-writeback */
	page_writeback_init();
#ifdef CONFIG_PROC_FS
	proc_root_init();
#endif
623
	cgroup_init();
L
Linus Torvalds 已提交
624
	cpuset_init();
625
	taskstats_init_early();
626
	delayacct_init();
L
Linus Torvalds 已提交
627 628 629 630

	check_bugs();

	acpi_early_init(); /* before LAPIC and SMP init */
631
	sfi_init_late();
L
Linus Torvalds 已提交
632

633
	if (efi_enabled(EFI_RUNTIME_SERVICES)) {
634
		efi_late_init();
635
		efi_free_boot_services();
636
	}
637

638 639
	ftrace_init();

L
Linus Torvalds 已提交
640 641 642 643
	/* Do the rest non-__init'ed, we're now alive */
	rest_init();
}

644 645 646 647
/* Call all constructor functions linked into the kernel. */
static void __init do_ctors(void)
{
#ifdef CONFIG_CONSTRUCTORS
648
	ctor_fn_t *fn = (ctor_fn_t *) __ctors_start;
649

650 651
	for (; fn < (ctor_fn_t *) __ctors_end; fn++)
		(*fn)();
652 653 654
#endif
}

655
bool initcall_debug;
R
Rusty Russell 已提交
656
core_param(initcall_debug, initcall_debug, bool, 0644);
L
Linus Torvalds 已提交
657

658 659
static char msgbuf[64];

660
static int __init_or_module do_one_initcall_debug(initcall_t fn)
L
Linus Torvalds 已提交
661
{
662
	ktime_t calltime, delta, rettime;
A
Américo Wang 已提交
663 664
	unsigned long long duration;
	int ret;
L
Linus Torvalds 已提交
665

A
Andrew Morton 已提交
666
	pr_debug("calling  %pF @ %i\n", fn, task_pid_nr(current));
667
	calltime = ktime_get();
A
Américo Wang 已提交
668
	ret = fn();
669 670 671
	rettime = ktime_get();
	delta = ktime_sub(rettime, calltime);
	duration = (unsigned long long) ktime_to_ns(delta) >> 10;
A
Andrew Morton 已提交
672 673
	pr_debug("initcall %pF returned %d after %lld usecs\n",
		 fn, ret, duration);
L
Linus Torvalds 已提交
674

675 676 677
	return ret;
}

678
int __init_or_module do_one_initcall(initcall_t fn)
679 680 681 682 683 684 685 686
{
	int count = preempt_count();
	int ret;

	if (initcall_debug)
		ret = do_one_initcall_debug(fn);
	else
		ret = fn();
687

688
	msgbuf[0] = 0;
689

690
	if (preempt_count() != count) {
691
		sprintf(msgbuf, "preemption imbalance ");
692
		preempt_count() = count;
L
Linus Torvalds 已提交
693
	}
694
	if (irqs_disabled()) {
695
		strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf));
696 697
		local_irq_enable();
	}
698
	WARN(msgbuf[0], "initcall %pF returned with %s\n", fn, msgbuf);
699

A
Américo Wang 已提交
700
	return ret;
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
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,
};

727
/* Keep these in sync with initcalls in include/linux/init.h */
728
static char *initcall_level_names[] __initdata = {
729 730 731 732 733 734 735 736
	"early",
	"core",
	"postcore",
	"arch",
	"subsys",
	"fs",
	"device",
	"late",
737 738 739
};

static void __init do_initcall_level(int level)
740
{
741
	extern const struct kernel_param __start___param[], __stop___param[];
742
	initcall_t *fn;
743

744 745 746 747 748
	strcpy(static_command_line, saved_command_line);
	parse_args(initcall_level_names[level],
		   static_command_line, __start___param,
		   __stop___param - __start___param,
		   level, level,
749
		   &repair_env_string);
750 751

	for (fn = initcall_levels[level]; fn < initcall_levels[level+1]; fn++)
752
		do_one_initcall(*fn);
L
Linus Torvalds 已提交
753 754
}

755 756 757 758
static void __init do_initcalls(void)
{
	int level;

759
	for (level = 0; level < ARRAY_SIZE(initcall_levels) - 1; level++)
760 761 762
		do_initcall_level(level);
}

L
Linus Torvalds 已提交
763 764 765 766 767 768 769 770 771
/*
 * 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)
{
772
	cpuset_init_smp();
L
Linus Torvalds 已提交
773
	usermodehelper_init();
774
	shmem_init();
L
Linus Torvalds 已提交
775
	driver_init();
776
	init_irq_proc();
777
	do_ctors();
778
	usermodehelper_enable();
779
	do_initcalls();
L
Linus Torvalds 已提交
780 781
}

782
static void __init do_pre_smp_initcalls(void)
783
{
784
	initcall_t *fn;
785

786
	for (fn = __initcall_start; fn < __initcall0_start; fn++)
787
		do_one_initcall(*fn);
788 789
}

790 791 792 793 794 795 796 797 798 799 800
/*
 * 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();
}

801
static int run_init_process(const char *init_filename)
L
Linus Torvalds 已提交
802 803
{
	argv_init[0] = init_filename;
804 805 806
	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
807 808
}

809
static noinline void __init kernel_init_freeable(void);
810 811

static int __ref kernel_init(void *unused)
812
{
813
	kernel_init_freeable();
814 815
	/* need to finish all async __init code before freeing the memory */
	async_synchronize_full();
816 817 818 819 820
	free_initmem();
	mark_rodata_ro();
	system_state = SYSTEM_RUNNING;
	numa_default_policy();

A
Al Viro 已提交
821
	flush_delayed_fput();
822

823
	if (ramdisk_execute_command) {
824 825
		if (!run_init_process(ramdisk_execute_command))
			return 0;
A
Andrew Morton 已提交
826
		pr_err("Failed to execute %s\n", ramdisk_execute_command);
827 828 829 830 831 832 833 834 835
	}

	/*
	 * 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) {
836 837
		if (!run_init_process(execute_command))
			return 0;
A
Andrew Morton 已提交
838 839
		pr_err("Failed to execute %s.  Attempting defaults...\n",
			execute_command);
840
	}
841 842 843 844 845
	if (!run_init_process("/sbin/init") ||
	    !run_init_process("/etc/init") ||
	    !run_init_process("/bin/init") ||
	    !run_init_process("/bin/sh"))
		return 0;
846

847 848
	panic("No init found.  Try passing init= option to kernel. "
	      "See Linux Documentation/init.txt for guidance.");
849 850
}

851
static noinline void __init kernel_init_freeable(void)
L
Linus Torvalds 已提交
852
{
853 854 855 856
	/*
	 * Wait until kthreadd is all set-up.
	 */
	wait_for_completion(&kthreadd_done);
857 858 859 860

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

861 862 863
	/*
	 * init can allocate pages on any node
	 */
864
	set_mems_allowed(node_states[N_MEMORY]);
L
Linus Torvalds 已提交
865 866 867
	/*
	 * init can run on any cpu.
	 */
868
	set_cpus_allowed_ptr(current, cpu_all_mask);
L
Linus Torvalds 已提交
869

870 871
	cad_pid = task_pid(current);

872
	smp_prepare_cpus(setup_max_cpus);
L
Linus Torvalds 已提交
873 874

	do_pre_smp_initcalls();
875
	lockup_detector_init();
L
Linus Torvalds 已提交
876 877 878 879 880 881

	smp_init();
	sched_init_smp();

	do_basic_setup();

882 883
	/* 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
884
		pr_err("Warning: unable to open an initial console.\n");
885 886 887

	(void) sys_dup(0);
	(void) sys_dup(0);
L
Linus Torvalds 已提交
888 889 890 891
	/*
	 * check if there is an early userspace init.  If yes, let it do all
	 * the work
	 */
892 893 894 895 896 897

	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
898
		prepare_namespace();
899
	}
L
Linus Torvalds 已提交
900 901 902 903 904 905

	/*
	 * 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..
	 */
906 907 908

	/* rootfs is available now, try loading default modules */
	load_default_modules();
L
Linus Torvalds 已提交
909
}