main.c 25.3 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 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>
Ingo Molnar's avatar
Ingo Molnar committed
19
#include <linux/stackprotector.h>
Linus Torvalds's avatar
Linus Torvalds committed
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>
Linus Torvalds's avatar
Linus Torvalds committed
28 29 30
#include <linux/tty.h>
#include <linux/percpu.h>
#include <linux/kmod.h>
Nick Piggin's avatar
Nick Piggin committed
31
#include <linux/vmalloc.h>
Linus Torvalds's avatar
Linus Torvalds committed
32
#include <linux/kernel_stat.h>
33
#include <linux/start_kernel.h>
Linus Torvalds's avatar
Linus Torvalds committed
34
#include <linux/security.h>
35
#include <linux/smp.h>
Linus Torvalds's avatar
Linus Torvalds committed
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>
Linus Torvalds's avatar
Linus Torvalds committed
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>
Linus Torvalds's avatar
Linus Torvalds committed
49 50 51 52
#include <linux/unistd.h>
#include <linux/rmap.h>
#include <linux/mempolicy.h>
#include <linux/key.h>
Adrian Bunk's avatar
Adrian Bunk committed
53
#include <linux/buffer_head.h>
54
#include <linux/page_ext.h>
55
#include <linux/debug_locks.h>
56
#include <linux/debugobjects.h>
Ingo Molnar's avatar
Ingo Molnar committed
57
#include <linux/lockdep.h>
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>
Al Viro's avatar
Al Viro committed
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>
81
#include <linux/integrity.h>
82
#include <linux/proc_ns.h>
83
#include <linux/io.h>
84
#include <linux/cache.h>
Linus Torvalds's avatar
Linus Torvalds committed
85 86 87 88

#include <asm/io.h>
#include <asm/bugs.h>
#include <asm/setup.h>
89
#include <asm/sections.h>
90
#include <asm/cacheflush.h>
Linus Torvalds's avatar
Linus Torvalds committed
91

92
static int kernel_init(void *);
Linus Torvalds's avatar
Linus Torvalds committed
93 94

extern void init_IRQ(void);
95
extern void fork_init(void);
Linus Torvalds's avatar
Linus Torvalds committed
96 97
extern void radix_tree_init(void);

98 99 100 101 102 103 104 105 106
/*
 * 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;

107
enum system_states system_state __read_mostly;
Linus Torvalds's avatar
Linus Torvalds committed
108 109 110 111 112 113 114 115 116 117
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. */
118
void (*__initdata late_time_init)(void);
Linus Torvalds's avatar
Linus Torvalds committed
119

120 121 122 123 124 125
/* 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;
126 127
/* Command line for per-initcall parameter parsing */
static char *initcall_command_line;
Linus Torvalds's avatar
Linus Torvalds committed
128 129

static char *execute_command;
130
static char *ramdisk_execute_command;
Linus Torvalds's avatar
Linus Torvalds committed
131

132 133 134 135
/*
 * Used to generate warnings if static_key manipulation functions are used
 * before jump_label_init is called.
 */
136
bool static_key_initialized __read_mostly;
137 138
EXPORT_SYMBOL_GPL(static_key_initialized);

Jan Beulich's avatar
Jan Beulich committed
139 140 141 142 143 144
/*
 * 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.
145
 * For ex. kdump situation where previous kernel has crashed, BIOS has been
Jan Beulich's avatar
Jan Beulich committed
146 147 148 149
 * skipped and devices will be in unknown state.
 */
unsigned int reset_devices;
EXPORT_SYMBOL(reset_devices);
Linus Torvalds's avatar
Linus Torvalds committed
150

151 152 153 154 155 156 157 158
static int __init set_reset_devices(char *str)
{
	reset_devices = 1;
	return 1;
}

__setup("reset_devices", set_reset_devices);

159 160
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
161 162
static const char *panic_later, *panic_param;

163
extern const struct obs_kernel_param __setup_start[], __setup_end[];
Linus Torvalds's avatar
Linus Torvalds committed
164

165
static bool __init obsolete_checksetup(char *line)
Linus Torvalds's avatar
Linus Torvalds committed
166
{
167
	const struct obs_kernel_param *p;
168
	bool had_early_param = false;
Linus Torvalds's avatar
Linus Torvalds committed
169 170 171 172

	p = __setup_start;
	do {
		int n = strlen(p->str);
173
		if (parameqn(line, p->str, n)) {
Linus Torvalds's avatar
Linus Torvalds committed
174
			if (p->early) {
175 176 177 178
				/* 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
179
				if (line[n] == '\0' || line[n] == '=')
180
					had_early_param = true;
Linus Torvalds's avatar
Linus Torvalds committed
181
			} else if (!p->setup_func) {
182 183
				pr_warn("Parameter %s is obsolete, ignored\n",
					p->str);
184
				return true;
Linus Torvalds's avatar
Linus Torvalds committed
185
			} else if (p->setup_func(line + n))
186
				return true;
Linus Torvalds's avatar
Linus Torvalds committed
187 188 189
		}
		p++;
	} while (p < __setup_end);
190 191

	return had_early_param;
Linus Torvalds's avatar
Linus Torvalds committed
192 193 194 195 196 197 198 199 200 201 202
}

/*
 * 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)
{
203
	console_loglevel = CONSOLE_LOGLEVEL_DEBUG;
204
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
205 206 207 208
}

static int __init quiet_kernel(char *str)
{
209
	console_loglevel = CONSOLE_LOGLEVEL_QUIET;
210
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
211 212
}

213 214
early_param("debug", debug_kernel);
early_param("quiet", quiet_kernel);
Linus Torvalds's avatar
Linus Torvalds committed
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;
Linus Torvalds's avatar
Linus Torvalds committed
231 232
}

233
early_param("loglevel", loglevel);
Linus Torvalds's avatar
Linus Torvalds committed
234

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

253
/* Anything after -- gets handed straight to init. */
254 255
static int __init set_init_arg(char *param, char *val,
			       const char *unused, void *arg)
256 257 258 259 260 261
{
	unsigned int i;

	if (panic_later)
		return 0;

262
	repair_env_string(param, val, unused, NULL);
263 264 265 266 267 268 269 270 271 272 273 274

	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;
}

275 276 277 278
/*
 * Unknown boot options get handed to init, unless they look like
 * unused parameters (modprobe will find them in /proc/cmdline).
 */
279 280
static int __init unknown_bootoption(char *param, char *val,
				     const char *unused, void *arg)
281
{
282
	repair_env_string(param, val, unused, NULL);
Linus Torvalds's avatar
Linus Torvalds committed
283 284 285 286 287

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

288 289
	/* Unused module parameter. */
	if (strchr(param, '.') && (!val || strchr(param, '.') < val))
Linus Torvalds's avatar
Linus Torvalds committed
290 291 292 293 294 295 296 297 298 299
		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) {
300
				panic_later = "env";
Linus Torvalds's avatar
Linus Torvalds committed
301 302 303 304 305 306 307 308 309 310 311
				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) {
312
				panic_later = "init";
Linus Torvalds's avatar
Linus Torvalds committed
313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337
				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);

338 339 340 341 342 343 344 345 346 347 348 349
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
350
#ifndef CONFIG_SMP
351
static const unsigned int setup_max_cpus = NR_CPUS;
352
static inline void setup_nr_cpu_ids(void) { }
Linus Torvalds's avatar
Linus Torvalds committed
353 354 355
static inline void smp_prepare_cpus(unsigned int maxcpus) { }
#endif

356 357 358 359 360 361 362 363
/*
 * 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)
{
364 365 366 367 368
	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);
369 370
	strcpy(saved_command_line, boot_command_line);
	strcpy(static_command_line, command_line);
371 372
}

Linus Torvalds's avatar
Linus Torvalds committed
373 374 375 376 377 378 379 380 381
/*
 * 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.
 */

382 383
static __initdata DECLARE_COMPLETION(kthreadd_done);

384
static noinline void __ref rest_init(void)
Linus Torvalds's avatar
Linus Torvalds committed
385
{
386 387
	int pid;

388
	rcu_scheduler_starting();
389
	/*
Peter Zijlstra's avatar
Peter Zijlstra committed
390
	 * We need to spawn init first so that it obtains pid 1, however
391 392 393
	 * the init task will end up wanting to create kthreads, which, if
	 * we schedule it before we create kthreadd, will OOPS.
	 */
394
	kernel_thread(kernel_init, NULL, CLONE_FS);
Linus Torvalds's avatar
Linus Torvalds committed
395
	numa_default_policy();
396
	pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES);
397
	rcu_read_lock();
398
	kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns);
399
	rcu_read_unlock();
400
	complete(&kthreadd_done);
401 402 403

	/*
	 * The boot idle thread must execute schedule()
404
	 * at least once to get things moving:
405
	 */
406
	init_idle_bootup_task(current);
407
	schedule_preempt_disabled();
408
	/* Call into cpu_idle with preempt disabled */
409
	cpu_startup_entry(CPUHP_ONLINE);
410
}
Linus Torvalds's avatar
Linus Torvalds committed
411 412

/* Check for early params. */
413 414
static int __init do_early_param(char *param, char *val,
				 const char *unused, void *arg)
Linus Torvalds's avatar
Linus Torvalds committed
415
{
416
	const struct obs_kernel_param *p;
Linus Torvalds's avatar
Linus Torvalds committed
417 418

	for (p = __setup_start; p < __setup_end; p++) {
419
		if ((p->early && parameq(param, p->str)) ||
420 421 422
		    (strcmp(param, "console") == 0 &&
		     strcmp(p->str, "earlycon") == 0)
		) {
Linus Torvalds's avatar
Linus Torvalds committed
423
			if (p->setup_func(val) != 0)
424
				pr_warn("Malformed early option '%s'\n", param);
Linus Torvalds's avatar
Linus Torvalds committed
425 426 427 428 429 430
		}
	}
	/* We accept everything at this stage. */
	return 0;
}

431 432
void __init parse_early_options(char *cmdline)
{
433 434
	parse_args("early options", cmdline, NULL, 0, 0, 0, NULL,
		   do_early_param);
435 436
}

Linus Torvalds's avatar
Linus Torvalds committed
437 438 439
/* Arch code calls this early on, or if not, just before other parsing. */
void __init parse_early_param(void)
{
440 441
	static int done __initdata;
	static char tmp_cmdline[COMMAND_LINE_SIZE] __initdata;
Linus Torvalds's avatar
Linus Torvalds committed
442 443 444 445 446

	if (done)
		return;

	/* All fall through to do_early_param. */
447
	strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE);
448
	parse_early_options(tmp_cmdline);
Linus Torvalds's avatar
Linus Torvalds committed
449 450 451
	done = 1;
}

452 453
void __init __weak arch_post_acpi_subsys_init(void) { }

454
void __init __weak smp_setup_processor_id(void)
455 456 457
{
}

458
# if THREAD_SIZE >= PAGE_SIZE
459
void __init __weak thread_stack_cache_init(void)
460 461
{
}
462
#endif
463

Pekka Enberg's avatar
Pekka Enberg committed
464 465 466 467 468
/*
 * Set up kernel memory allocators
 */
static void __init mm_init(void)
{
469 470 471 472 473
	/*
	 * page_ext requires contiguous pages,
	 * bigger than MAX_ORDER unless SPARSEMEM.
	 */
	page_ext_init_flatmem();
Pekka Enberg's avatar
Pekka Enberg committed
474 475
	mem_init();
	kmem_cache_init();
476
	percpu_init_late();
477
	pgtable_init();
Pekka Enberg's avatar
Pekka Enberg committed
478
	vmalloc_init();
479
	ioremap_huge_init();
Pekka Enberg's avatar
Pekka Enberg committed
480 481
}

482
asmlinkage __visible void __init start_kernel(void)
Linus Torvalds's avatar
Linus Torvalds committed
483
{
484 485
	char *command_line;
	char *after_dashes;
486

487
	set_task_stack_end_magic(&init_task);
488
	smp_setup_processor_id();
489
	debug_objects_early_init();
490 491 492 493 494 495

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

496
	cgroup_init_early();
Ingo Molnar's avatar
Ingo Molnar committed
497 498

	local_irq_disable();
499
	early_boot_irqs_disabled = true;
Ingo Molnar's avatar
Ingo Molnar committed
500

Linus Torvalds's avatar
Linus Torvalds committed
501 502 503 504
/*
 * Interrupts are still disabled. Do necessary setups, then
 * enable them
 */
505
	boot_cpu_init();
Linus Torvalds's avatar
Linus Torvalds committed
506
	page_address_init();
507
	pr_notice("%s", linux_banner);
Linus Torvalds's avatar
Linus Torvalds committed
508
	setup_arch(&command_line);
509
	mm_init_cpumask(&init_mm);
510
	setup_command_line(command_line);
511
	setup_nr_cpu_ids();
512
	setup_per_cpu_areas();
513
	boot_cpu_state_init();
514
	smp_prepare_boot_cpu();	/* arch-specific boot-cpu hooks */
Linus Torvalds's avatar
Linus Torvalds committed
515

516
	build_all_zonelists(NULL, NULL);
517 518
	page_alloc_init();

519
	pr_notice("Kernel command line: %s\n", boot_command_line);
520
	parse_early_param();
521 522 523
	after_dashes = parse_args("Booting kernel",
				  static_command_line, __start___param,
				  __stop___param - __start___param,
524
				  -1, -1, NULL, &unknown_bootoption);
525
	if (!IS_ERR_OR_NULL(after_dashes))
526
		parse_args("Setting init args", after_dashes, NULL, 0, -1, -1,
527
			   NULL, set_init_arg);
528 529 530

	jump_label_init();

531 532 533 534
	/*
	 * These use large bootmem allocations and must precede
	 * kmem_cache_init()
	 */
535
	setup_log_buf(0);
536 537 538 539
	pidhash_init();
	vfs_caches_init_early();
	sort_main_extable();
	trap_init();
Pekka Enberg's avatar
Pekka Enberg committed
540
	mm_init();
541

Linus Torvalds's avatar
Linus Torvalds committed
542 543 544 545 546 547 548 549 550 551 552
	/*
	 * 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();
553 554
	if (WARN(!irqs_disabled(),
		 "Interrupts were enabled *very* early, fixing it\n"))
555
		local_irq_disable();
556
	idr_init_cache();
557 558 559 560 561 562 563 564

	/*
	 * 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
565
	rcu_init();
566 567 568 569

	/* trace_printk() and trace points may be used after this */
	trace_init();

570
	context_tracking_init();
571
	radix_tree_init();
572 573
	/* init some links before init_ISA_irqs() */
	early_irq_init();
Linus Torvalds's avatar
Linus Torvalds committed
574
	init_IRQ();
Thomas Gleixner's avatar
Thomas Gleixner committed
575
	tick_init();
576
	rcu_init_nohz();
Linus Torvalds's avatar
Linus Torvalds committed
577
	init_timers();
578
	hrtimers_init();
Linus Torvalds's avatar
Linus Torvalds committed
579
	softirq_init();
580
	timekeeping_init();
581
	time_init();
582
	sched_clock_postinit();
583
	printk_nmi_init();
584
	perf_event_init();
585
	profile_init();
586
	call_function_init();
587
	WARN(!irqs_disabled(), "Interrupts were enabled early\n");
588
	early_boot_irqs_disabled = false;
589
	local_irq_enable();
590

591
	kmem_cache_init_late();
Linus Torvalds's avatar
Linus Torvalds committed
592 593 594 595 596 597 598 599

	/*
	 * 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)
600 601
		panic("Too many boot %s vars at `%s'", panic_later,
		      panic_param);
Ingo Molnar's avatar
Ingo Molnar committed
602 603 604

	lockdep_info();

605 606 607 608 609 610 611
	/*
	 * 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
612 613
#ifdef CONFIG_BLK_DEV_INITRD
	if (initrd_start && !initrd_below_start_ok &&
614
	    page_to_pfn(virt_to_page((void *)initrd_start)) < min_low_pfn) {
615
		pr_crit("initrd overwritten (0x%08lx < 0x%08lx) - disabling it.\n",
616 617
		    page_to_pfn(virt_to_page((void *)initrd_start)),
		    min_low_pfn);
Linus Torvalds's avatar
Linus Torvalds committed
618 619 620
		initrd_start = 0;
	}
#endif
621
	page_ext_init();
622
	debug_objects_mem_init();
623
	kmemleak_init();
624
	setup_per_cpu_pageset();
Linus Torvalds's avatar
Linus Torvalds committed
625 626 627 628 629 630
	numa_policy_init();
	if (late_time_init)
		late_time_init();
	calibrate_delay();
	pidmap_init();
	anon_vma_init();
631
	acpi_early_init();
632
#ifdef CONFIG_X86
633
	if (efi_enabled(EFI_RUNTIME_SERVICES))
634
		efi_enter_virtual_mode();
635
#endif
636
#ifdef CONFIG_X86_ESPFIX64
637 638
	/* Should be run before the first non-init thread is created */
	init_espfix_bsp();
639
#endif
640
	thread_stack_cache_init();
641
	cred_init();
642
	fork_init();
Linus Torvalds's avatar
Linus Torvalds committed
643 644 645 646
	proc_caches_init();
	buffer_init();
	key_init();
	security_init();
647
	dbg_late_init();
648
	vfs_caches_init();
649
	pagecache_init();
Linus Torvalds's avatar
Linus Torvalds committed
650 651
	signals_init();
	proc_root_init();
652
	nsfs_init();
Linus Torvalds's avatar
Linus Torvalds committed
653
	cpuset_init();
654
	cgroup_init();
655
	taskstats_init_early();
656
	delayacct_init();
Linus Torvalds's avatar
Linus Torvalds committed
657 658 659

	check_bugs();

660
	acpi_subsystem_init();
661
	arch_post_acpi_subsys_init();
662
	sfi_init_late();
Linus Torvalds's avatar
Linus Torvalds committed
663

664
	if (efi_enabled(EFI_RUNTIME_SERVICES)) {
665
		efi_late_init();
666
		efi_free_boot_services();
667
	}
668

669 670
	ftrace_init();

Linus Torvalds's avatar
Linus Torvalds committed
671 672 673 674
	/* Do the rest non-__init'ed, we're now alive */
	rest_init();
}

675 676 677 678
/* Call all constructor functions linked into the kernel. */
static void __init do_ctors(void)
{
#ifdef CONFIG_CONSTRUCTORS
679
	ctor_fn_t *fn = (ctor_fn_t *) __ctors_start;
680

681 682
	for (; fn < (ctor_fn_t *) __ctors_end; fn++)
		(*fn)();
683 684 685
#endif
}

686
bool initcall_debug;
687
core_param(initcall_debug, initcall_debug, bool, 0644);
Linus Torvalds's avatar
Linus Torvalds committed
688

689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719
#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;
720
	char fn_name[KSYM_SYMBOL_LEN];
721
	unsigned long addr;
722

723
	if (list_empty(&blacklisted_initcalls))
724 725
		return false;

726 727
	addr = (unsigned long) dereference_function_descriptor(fn);
	sprint_symbol_no_offset(fn_name, addr);
728

729 730 731 732 733 734
	/*
	 * 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');

735
	list_for_each_entry(entry, &blacklisted_initcalls, next) {
736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757
		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);

758
static int __init_or_module do_one_initcall_debug(initcall_t fn)
Linus Torvalds's avatar
Linus Torvalds committed
759
{
760
	ktime_t calltime, delta, rettime;
Américo Wang's avatar
Américo Wang committed
761 762
	unsigned long long duration;
	int ret;
Linus Torvalds's avatar
Linus Torvalds committed
763

764
	printk(KERN_DEBUG "calling  %pF @ %i\n", fn, task_pid_nr(current));
765
	calltime = ktime_get();
Américo Wang's avatar
Américo Wang committed
766
	ret = fn();
767 768 769
	rettime = ktime_get();
	delta = ktime_sub(rettime, calltime);
	duration = (unsigned long long) ktime_to_ns(delta) >> 10;
770
	printk(KERN_DEBUG "initcall %pF returned %d after %lld usecs\n",
771
		 fn, ret, duration);
Linus Torvalds's avatar
Linus Torvalds committed
772

773 774 775
	return ret;
}

776
int __init_or_module do_one_initcall(initcall_t fn)
777 778 779
{
	int count = preempt_count();
	int ret;
780
	char msgbuf[64];
781

782 783 784
	if (initcall_blacklisted(fn))
		return -EPERM;

785 786 787 788
	if (initcall_debug)
		ret = do_one_initcall_debug(fn);
	else
		ret = fn();
789

790
	msgbuf[0] = 0;
791

792
	if (preempt_count() != count) {
793
		sprintf(msgbuf, "preemption imbalance ");
794
		preempt_count_set(count);
Linus Torvalds's avatar
Linus Torvalds committed
795
	}
796
	if (irqs_disabled()) {
797
		strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf));
798 799
		local_irq_enable();
	}
800
	WARN(msgbuf[0], "initcall %pF returned with %s\n", fn, msgbuf);
801

802
	add_latent_entropy();
Américo Wang's avatar
Américo Wang committed
803
	return ret;
804 805 806
}


807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829
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,
};

830
/* Keep these in sync with initcalls in include/linux/init.h */
831
static char *initcall_level_names[] __initdata = {
832 833 834 835 836 837 838 839
	"early",
	"core",
	"postcore",
	"arch",
	"subsys",
	"fs",
	"device",
	"late",
840 841 842
};

static void __init do_initcall_level(int level)
843
{
844
	initcall_t *fn;
845

846
	strcpy(initcall_command_line, saved_command_line);
847
	parse_args(initcall_level_names[level],
848
		   initcall_command_line, __start___param,
849 850
		   __stop___param - __start___param,
		   level, level,
851
		   NULL, &repair_env_string);
852 853

	for (fn = initcall_levels[level]; fn < initcall_levels[level+1]; fn++)
854
		do_one_initcall(*fn);
Linus Torvalds's avatar
Linus Torvalds committed
855 856
}

857 858 859 860
static void __init do_initcalls(void)
{
	int level;

861
	for (level = 0; level < ARRAY_SIZE(initcall_levels) - 1; level++)
862 863 864
		do_initcall_level(level);
}

Linus Torvalds's avatar
Linus Torvalds committed
865 866 867 868 869 870 871 872 873
/*
 * 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)
{
874
	cpuset_init_smp();
875
	shmem_init();
Linus Torvalds's avatar
Linus Torvalds committed
876
	driver_init();
877
	init_irq_proc();
878
	do_ctors();
879
	usermodehelper_enable();
880
	do_initcalls();
881
	random_int_secret_init();
Linus Torvalds's avatar
Linus Torvalds committed
882 883
}

884
static void __init do_pre_smp_initcalls(void)
885
{
886
	initcall_t *fn;
887

888
	for (fn = __initcall_start; fn < __initcall0_start; fn++)
889
		do_one_initcall(*fn);
890 891
}

892 893 894 895 896 897 898 899 900 901 902
/*
 * 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();
}

903
static int run_init_process(const char *init_filename)
Linus Torvalds's avatar
Linus Torvalds committed
904 905
{
	argv_init[0] = init_filename;
906
	return do_execve(getname_kernel(init_filename),
907 908
		(const char __user *const __user *)argv_init,
		(const char __user *const __user *)envp_init);
Linus Torvalds's avatar
Linus Torvalds committed
909 910
}

911 912 913 914 915 916 917 918 919 920 921 922 923 924
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;
}

925
static noinline void __init kernel_init_freeable(void);
926

927
#if defined(CONFIG_DEBUG_RODATA) || defined(CONFIG_DEBUG_SET_MODULE_RONX)
928
bool rodata_enabled __ro_after_init = true;
929 930 931 932 933
static int __init set_debug_rodata(char *str)
{
	return strtobool(str, &rodata_enabled);
}
__setup("rodata=", set_debug_rodata);
934
#endif
935

936
#ifdef CONFIG_DEBUG_RODATA
937 938 939 940 941 942 943 944 945 946 947 948 949 950
static void mark_readonly(void)
{
	if (rodata_enabled)
		mark_rodata_ro();
	else
		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

951
static int __ref kernel_init(void *unused)
952
{
953 954
	int ret;

955
	kernel_init_freeable();
956 957
	/* need to finish all async __init code before freeing the memory */
	async_synchronize_full();
958
	free_initmem();
959
	mark_readonly();
960 961 962
	system_state = SYSTEM_RUNNING;
	numa_default_policy();

Al Viro's avatar
Al Viro committed
963
	flush_delayed_fput();
964

965 966
	rcu_end_inkernel_boot();

967
	if (ramdisk_execute_command) {
968 969
		ret = run_init_process(ramdisk_execute_command);
		if (!ret)
970
			return 0;
971 972
		pr_err("Failed to execute %s (error %d)\n",
		       ramdisk_execute_command, ret);
973 974 975 976 977 978 979 980 981
	}

	/*
	 * 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) {
982 983
		ret = run_init_process(execute_command);
		if (!ret)
984
			return 0;
985 986
		panic("Requested init %s failed (error %d).",
		      execute_command, ret);
987
	}
988 989 990 991
	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"))
992
		return 0;
993

994
	panic("No working init found.  Try passing init= option to kernel. "
995
	      "See Linux Documentation/admin-guide/init.rst for guidance.");
996 997
}

998
static noinline void __init kernel_init_freeable(void)
Linus Torvalds's avatar
Linus Torvalds committed
999
{
1000 1001 1002 1003
	/*
	 * Wait until kthreadd is all set-up.
	 */
	wait_for_completion(&kthreadd_done);
1004 1005 1006 1007

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

1008 1009 1010
	/*
	 * init can allocate pages on any node
	 */
1011
	set_mems_allowed(node_states[N_MEMORY]);
Linus Torvalds's avatar
Linus Torvalds committed
1012 1013 1014
	/*
	 * init can run on any cpu.
	 */
1015
	set_cpus_allowed_ptr(current, cpu_all_mask);
Linus Torvalds's avatar
Linus Torvalds committed
1016

1017 1018
	cad_pid = task_pid(current);

1019
	smp_prepare_cpus(setup_max_cpus);
Linus Torvalds's avatar
Linus Torvalds committed
1020

1021 1022
	workqueue_init();

Linus Torvalds's avatar
Linus Torvalds committed
1023
	do_pre_smp_initcalls();
1024
	lockup_detector_init();
Linus Torvalds's avatar
Linus Torvalds committed
1025 1026 1027 1028

	smp_init();
	sched_init_smp();

1029 1030
	page_alloc_init_late();

Linus Torvalds's avatar
Linus Torvalds committed
1031 1032
	do_basic_setup();

1033 1034
	/* Open the /dev/console on the rootfs, this should never fail */
	if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0)
1035
		pr_err("Warning: unable to open an initial console.\n");
1036 1037 1038

	(void) sys_dup(0);
	(void) sys_dup(0);
Linus Torvalds's avatar
Linus Torvalds committed
1039 1040 1041 1042
	/*
	 * check if there is an early userspace init.  If yes, let it do all
	 * the work
	 */
1043 1044 1045 1046 1047 1048

	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
1049
		prepare_namespace();
1050
	}
Linus Torvalds's avatar
Linus Torvalds committed
1051 1052 1053 1054 1055

	/*
	 * 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..
1056 1057 1058
	 *
	 * rootfs is available now, try loading the public keys
	 * and default modules
Linus Torvalds's avatar
Linus Torvalds committed
1059
	 */
1060

1061
	integrity_load_keys();
1062
	load_default_modules();
Linus Torvalds's avatar
Linus Torvalds committed
1063
}