1/*
2 *  linux/init/main.c
3 *
4 *  Copyright (C) 1991, 1992  Linus Torvalds
5 *
6 *  GK 2/5/95  -  Changed to support mounting root fs via NFS
7 *  Added initrd & change_root: Werner Almesberger & Hans Lermen, Feb '96
8 *  Moan early if gcc is old, avoiding bogus kernels - Paul Gortmaker, May '96
9 *  Simplified starting of init:  Michael A. Griffith <grif@acm.org>
10 */
11
12#define DEBUG		/* Enable initcall_debug */
13
14#include <linux/types.h>
15#include <linux/module.h>
16#include <linux/proc_fs.h>
17#include <linux/kernel.h>
18#include <linux/syscalls.h>
19#include <linux/stackprotector.h>
20#include <linux/string.h>
21#include <linux/ctype.h>
22#include <linux/delay.h>
23#include <linux/ioport.h>
24#include <linux/init.h>
25#include <linux/initrd.h>
26#include <linux/bootmem.h>
27#include <linux/acpi.h>
28#include <linux/tty.h>
29#include <linux/percpu.h>
30#include <linux/kmod.h>
31#include <linux/vmalloc.h>
32#include <linux/kernel_stat.h>
33#include <linux/start_kernel.h>
34#include <linux/security.h>
35#include <linux/smp.h>
36#include <linux/profile.h>
37#include <linux/rcupdate.h>
38#include <linux/moduleparam.h>
39#include <linux/kallsyms.h>
40#include <linux/writeback.h>
41#include <linux/cpu.h>
42#include <linux/cpuset.h>
43#include <linux/cgroup.h>
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>
49#include <linux/unistd.h>
50#include <linux/rmap.h>
51#include <linux/mempolicy.h>
52#include <linux/key.h>
53#include <linux/buffer_head.h>
54#include <linux/page_ext.h>
55#include <linux/debug_locks.h>
56#include <linux/debugobjects.h>
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>
73#include <linux/file.h>
74#include <linux/ptrace.h>
75#include <linux/blkdev.h>
76#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
85#include <asm/io.h>
86#include <asm/bugs.h>
87#include <asm/setup.h>
88#include <asm/sections.h>
89#include <asm/cacheflush.h>
90
91static int kernel_init(void *);
92
93extern void init_IRQ(void);
94extern void fork_init(void);
95extern void radix_tree_init(void);
96#ifndef CONFIG_DEBUG_RODATA
97static inline void mark_rodata_ro(void) { }
98#endif
99
100/*
101 * Debug helper: via this flag we know that we are in 'early bootup code'
102 * where only the boot processor is running with IRQ disabled.  This means
103 * two things - IRQ must not be enabled before the flag is cleared and some
104 * operations which are not allowed with IRQ disabled are allowed while the
105 * flag is set.
106 */
107bool early_boot_irqs_disabled __read_mostly;
108
109enum system_states system_state __read_mostly;
110EXPORT_SYMBOL(system_state);
111
112/*
113 * Boot command-line arguments
114 */
115#define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT
116#define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT
117
118extern void time_init(void);
119/* Default late time init is NULL. archs can override this later. */
120void (*__initdata late_time_init)(void);
121
122/* Untouched command line saved by arch-specific code. */
123char __initdata boot_command_line[COMMAND_LINE_SIZE];
124/* Untouched saved command line (eg. for /proc) */
125char *saved_command_line;
126/* Command line for parameter parsing */
127static char *static_command_line;
128/* Command line for per-initcall parameter parsing */
129static char *initcall_command_line;
130
131static char *execute_command;
132static char *ramdisk_execute_command;
133
134/*
135 * Used to generate warnings if static_key manipulation functions are used
136 * before jump_label_init is called.
137 */
138bool static_key_initialized __read_mostly;
139EXPORT_SYMBOL_GPL(static_key_initialized);
140
141/*
142 * If set, this is an indication to the drivers that reset the underlying
143 * device before going ahead with the initialization otherwise driver might
144 * rely on the BIOS and skip the reset operation.
145 *
146 * This is useful if kernel is booting in an unreliable environment.
147 * For ex. kdump situation where previous kernel has crashed, BIOS has been
148 * skipped and devices will be in unknown state.
149 */
150unsigned int reset_devices;
151EXPORT_SYMBOL(reset_devices);
152
153static int __init set_reset_devices(char *str)
154{
155	reset_devices = 1;
156	return 1;
157}
158
159__setup("reset_devices", set_reset_devices);
160
161static const char *argv_init[MAX_INIT_ARGS+2] = { "init", NULL, };
162const char *envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, };
163static const char *panic_later, *panic_param;
164
165extern const struct obs_kernel_param __setup_start[], __setup_end[];
166
167static int __init obsolete_checksetup(char *line)
168{
169	const struct obs_kernel_param *p;
170	int had_early_param = 0;
171
172	p = __setup_start;
173	do {
174		int n = strlen(p->str);
175		if (parameqn(line, p->str, n)) {
176			if (p->early) {
177				/* Already done in parse_early_param?
178				 * (Needs exact match on param part).
179				 * Keep iterating, as we can have early
180				 * params and __setups of same names 8( */
181				if (line[n] == '\0' || line[n] == '=')
182					had_early_param = 1;
183			} else if (!p->setup_func) {
184				pr_warn("Parameter %s is obsolete, ignored\n",
185					p->str);
186				return 1;
187			} else if (p->setup_func(line + n))
188				return 1;
189		}
190		p++;
191	} while (p < __setup_end);
192
193	return had_early_param;
194}
195
196/*
197 * This should be approx 2 Bo*oMips to start (note initial shift), and will
198 * still work even if initially too large, it will just take slightly longer
199 */
200unsigned long loops_per_jiffy = (1<<12);
201EXPORT_SYMBOL(loops_per_jiffy);
202
203static int __init debug_kernel(char *str)
204{
205	console_loglevel = CONSOLE_LOGLEVEL_DEBUG;
206	return 0;
207}
208
209static int __init quiet_kernel(char *str)
210{
211	console_loglevel = CONSOLE_LOGLEVEL_QUIET;
212	return 0;
213}
214
215early_param("debug", debug_kernel);
216early_param("quiet", quiet_kernel);
217
218static int __init loglevel(char *str)
219{
220	int newlevel;
221
222	/*
223	 * Only update loglevel value when a correct setting was passed,
224	 * to prevent blind crashes (when loglevel being set to 0) that
225	 * are quite hard to debug
226	 */
227	if (get_option(&str, &newlevel)) {
228		console_loglevel = newlevel;
229		return 0;
230	}
231
232	return -EINVAL;
233}
234
235early_param("loglevel", loglevel);
236
237/* Change NUL term back to "=", to make "param" the whole string. */
238static int __init repair_env_string(char *param, char *val, const char *unused)
239{
240	if (val) {
241		/* param=val or param="val"? */
242		if (val == param+strlen(param)+1)
243			val[-1] = '=';
244		else if (val == param+strlen(param)+2) {
245			val[-2] = '=';
246			memmove(val-1, val, strlen(val)+1);
247			val--;
248		} else
249			BUG();
250	}
251	return 0;
252}
253
254/* Anything after -- gets handed straight to init. */
255static int __init set_init_arg(char *param, char *val, const char *unused)
256{
257	unsigned int i;
258
259	if (panic_later)
260		return 0;
261
262	repair_env_string(param, val, unused);
263
264	for (i = 0; argv_init[i]; i++) {
265		if (i == MAX_INIT_ARGS) {
266			panic_later = "init";
267			panic_param = param;
268			return 0;
269		}
270	}
271	argv_init[i] = param;
272	return 0;
273}
274
275/*
276 * Unknown boot options get handed to init, unless they look like
277 * unused parameters (modprobe will find them in /proc/cmdline).
278 */
279static int __init unknown_bootoption(char *param, char *val, const char *unused)
280{
281	repair_env_string(param, val, unused);
282
283	/* Handle obsolete-style parameters */
284	if (obsolete_checksetup(param))
285		return 0;
286
287	/* Unused module parameter. */
288	if (strchr(param, '.') && (!val || strchr(param, '.') < val))
289		return 0;
290
291	if (panic_later)
292		return 0;
293
294	if (val) {
295		/* Environment option */
296		unsigned int i;
297		for (i = 0; envp_init[i]; i++) {
298			if (i == MAX_INIT_ENVS) {
299				panic_later = "env";
300				panic_param = param;
301			}
302			if (!strncmp(param, envp_init[i], val - param))
303				break;
304		}
305		envp_init[i] = param;
306	} else {
307		/* Command line option */
308		unsigned int i;
309		for (i = 0; argv_init[i]; i++) {
310			if (i == MAX_INIT_ARGS) {
311				panic_later = "init";
312				panic_param = param;
313			}
314		}
315		argv_init[i] = param;
316	}
317	return 0;
318}
319
320static int __init init_setup(char *str)
321{
322	unsigned int i;
323
324	execute_command = str;
325	/*
326	 * In case LILO is going to boot us with default command line,
327	 * it prepends "auto" before the whole cmdline which makes
328	 * the shell think it should execute a script with such name.
329	 * So we ignore all arguments entered _before_ init=... [MJ]
330	 */
331	for (i = 1; i < MAX_INIT_ARGS; i++)
332		argv_init[i] = NULL;
333	return 1;
334}
335__setup("init=", init_setup);
336
337static int __init rdinit_setup(char *str)
338{
339	unsigned int i;
340
341	ramdisk_execute_command = str;
342	/* See "auto" comment in init_setup */
343	for (i = 1; i < MAX_INIT_ARGS; i++)
344		argv_init[i] = NULL;
345	return 1;
346}
347__setup("rdinit=", rdinit_setup);
348
349#ifndef CONFIG_SMP
350static const unsigned int setup_max_cpus = NR_CPUS;
351static inline void setup_nr_cpu_ids(void) { }
352static inline void smp_prepare_cpus(unsigned int maxcpus) { }
353#endif
354
355/*
356 * We need to store the untouched command line for future reference.
357 * We also need to store the touched command line since the parameter
358 * parsing is performed in place, and we should allow a component to
359 * store reference of name/value for future reference.
360 */
361static void __init setup_command_line(char *command_line)
362{
363	saved_command_line =
364		memblock_virt_alloc(strlen(boot_command_line) + 1, 0);
365	initcall_command_line =
366		memblock_virt_alloc(strlen(boot_command_line) + 1, 0);
367	static_command_line = memblock_virt_alloc(strlen(command_line) + 1, 0);
368	strcpy(saved_command_line, boot_command_line);
369	strcpy(static_command_line, command_line);
370}
371
372/*
373 * We need to finalize in a non-__init function or else race conditions
374 * between the root thread and the init thread may cause start_kernel to
375 * be reaped by free_initmem before the root thread has proceeded to
376 * cpu_idle.
377 *
378 * gcc-3.4 accidentally inlines this function, so use noinline.
379 */
380
381static __initdata DECLARE_COMPLETION(kthreadd_done);
382
383static noinline void __init_refok rest_init(void)
384{
385	int pid;
386
387	rcu_scheduler_starting();
388	smpboot_thread_init();
389	/*
390	 * We need to spawn init first so that it obtains pid 1, however
391	 * the init task will end up wanting to create kthreads, which, if
392	 * we schedule it before we create kthreadd, will OOPS.
393	 */
394	kernel_thread(kernel_init, NULL, CLONE_FS);
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}
411
412/* Check for early params. */
413static int __init do_early_param(char *param, char *val, const char *unused)
414{
415	const struct obs_kernel_param *p;
416
417	for (p = __setup_start; p < __setup_end; p++) {
418		if ((p->early && parameq(param, p->str)) ||
419		    (strcmp(param, "console") == 0 &&
420		     strcmp(p->str, "earlycon") == 0)
421		) {
422			if (p->setup_func(val) != 0)
423				pr_warn("Malformed early option '%s'\n", param);
424		}
425	}
426	/* We accept everything at this stage. */
427	return 0;
428}
429
430void __init parse_early_options(char *cmdline)
431{
432	parse_args("early options", cmdline, NULL, 0, 0, 0, do_early_param);
433}
434
435/* Arch code calls this early on, or if not, just before other parsing. */
436void __init parse_early_param(void)
437{
438	static int done __initdata;
439	static char tmp_cmdline[COMMAND_LINE_SIZE] __initdata;
440
441	if (done)
442		return;
443
444	/* All fall through to do_early_param. */
445	strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE);
446	parse_early_options(tmp_cmdline);
447	done = 1;
448}
449
450/*
451 *	Activate the first processor.
452 */
453
454static void __init boot_cpu_init(void)
455{
456	int cpu = smp_processor_id();
457	/* Mark the boot cpu "present", "online" etc for SMP and UP case */
458	set_cpu_online(cpu, true);
459	set_cpu_active(cpu, true);
460	set_cpu_present(cpu, true);
461	set_cpu_possible(cpu, true);
462}
463
464void __init __weak smp_setup_processor_id(void)
465{
466}
467
468# if THREAD_SIZE >= PAGE_SIZE
469void __init __weak thread_info_cache_init(void)
470{
471}
472#endif
473
474/*
475 * Set up kernel memory allocators
476 */
477static void __init mm_init(void)
478{
479	/*
480	 * page_ext requires contiguous pages,
481	 * bigger than MAX_ORDER unless SPARSEMEM.
482	 */
483	page_ext_init_flatmem();
484	mem_init();
485	kmem_cache_init();
486	percpu_init_late();
487	pgtable_init();
488	vmalloc_init();
489	ioremap_huge_init();
490}
491
492asmlinkage __visible void __init start_kernel(void)
493{
494	char *command_line;
495	char *after_dashes;
496
497	/*
498	 * Need to run as early as possible, to initialize the
499	 * lockdep hash:
500	 */
501	lockdep_init();
502	set_task_stack_end_magic(&init_task);
503	smp_setup_processor_id();
504	debug_objects_early_init();
505
506	/*
507	 * Set up the the initial canary ASAP:
508	 */
509	boot_init_stack_canary();
510
511	cgroup_init_early();
512
513	local_irq_disable();
514	early_boot_irqs_disabled = true;
515
516/*
517 * Interrupts are still disabled. Do necessary setups, then
518 * enable them
519 */
520	boot_cpu_init();
521	page_address_init();
522	pr_notice("%s", linux_banner);
523	setup_arch(&command_line);
524	mm_init_cpumask(&init_mm);
525	setup_command_line(command_line);
526	setup_nr_cpu_ids();
527	setup_per_cpu_areas();
528	smp_prepare_boot_cpu();	/* arch-specific boot-cpu hooks */
529
530	build_all_zonelists(NULL, NULL);
531	page_alloc_init();
532
533	pr_notice("Kernel command line: %s\n", boot_command_line);
534	parse_early_param();
535	after_dashes = parse_args("Booting kernel",
536				  static_command_line, __start___param,
537				  __stop___param - __start___param,
538				  -1, -1, &unknown_bootoption);
539	if (!IS_ERR_OR_NULL(after_dashes))
540		parse_args("Setting init args", after_dashes, NULL, 0, -1, -1,
541			   set_init_arg);
542
543	jump_label_init();
544
545	/*
546	 * These use large bootmem allocations and must precede
547	 * kmem_cache_init()
548	 */
549	setup_log_buf(0);
550	pidhash_init();
551	vfs_caches_init_early();
552	sort_main_extable();
553	trap_init();
554	mm_init();
555
556	/*
557	 * Set up the scheduler prior starting any interrupts (such as the
558	 * timer interrupt). Full topology setup happens at smp_init()
559	 * time - but meanwhile we still have a functioning scheduler.
560	 */
561	sched_init();
562	/*
563	 * Disable preemption - early bootup scheduling is extremely
564	 * fragile until we cpu_idle() for the first time.
565	 */
566	preempt_disable();
567	if (WARN(!irqs_disabled(),
568		 "Interrupts were enabled *very* early, fixing it\n"))
569		local_irq_disable();
570	idr_init_cache();
571	rcu_init();
572
573	/* trace_printk() and trace points may be used after this */
574	trace_init();
575
576	context_tracking_init();
577	radix_tree_init();
578	/* init some links before init_ISA_irqs() */
579	early_irq_init();
580	init_IRQ();
581	tick_init();
582	rcu_init_nohz();
583	init_timers();
584	hrtimers_init();
585	softirq_init();
586	timekeeping_init();
587	time_init();
588	sched_clock_postinit();
589	perf_event_init();
590	profile_init();
591	call_function_init();
592	WARN(!irqs_disabled(), "Interrupts were enabled early\n");
593	early_boot_irqs_disabled = false;
594	local_irq_enable();
595
596	kmem_cache_init_late();
597
598	/*
599	 * HACK ALERT! This is early. We're enabling the console before
600	 * we've done PCI setups etc, and console_init() must be aware of
601	 * this. But we do want output early, in case something goes wrong.
602	 */
603	console_init();
604	if (panic_later)
605		panic("Too many boot %s vars at `%s'", panic_later,
606		      panic_param);
607
608	lockdep_info();
609
610	/*
611	 * Need to run this when irqs are enabled, because it wants
612	 * to self-test [hard/soft]-irqs on/off lock inversion bugs
613	 * too:
614	 */
615	locking_selftest();
616
617#ifdef CONFIG_BLK_DEV_INITRD
618	if (initrd_start && !initrd_below_start_ok &&
619	    page_to_pfn(virt_to_page((void *)initrd_start)) < min_low_pfn) {
620		pr_crit("initrd overwritten (0x%08lx < 0x%08lx) - disabling it.\n",
621		    page_to_pfn(virt_to_page((void *)initrd_start)),
622		    min_low_pfn);
623		initrd_start = 0;
624	}
625#endif
626	page_ext_init();
627	debug_objects_mem_init();
628	kmemleak_init();
629	setup_per_cpu_pageset();
630	numa_policy_init();
631	if (late_time_init)
632		late_time_init();
633	sched_clock_init();
634	calibrate_delay();
635	pidmap_init();
636	anon_vma_init();
637	acpi_early_init();
638#ifdef CONFIG_X86
639	if (efi_enabled(EFI_RUNTIME_SERVICES))
640		efi_enter_virtual_mode();
641#endif
642#ifdef CONFIG_X86_ESPFIX64
643	/* Should be run before the first non-init thread is created */
644	init_espfix_bsp();
645#endif
646	thread_info_cache_init();
647	cred_init();
648	fork_init();
649	proc_caches_init();
650	buffer_init();
651	key_init();
652	security_init();
653	dbg_late_init();
654	vfs_caches_init(totalram_pages);
655	signals_init();
656	/* rootfs populating might need page-writeback */
657	page_writeback_init();
658	proc_root_init();
659	nsfs_init();
660	cpuset_init();
661	cgroup_init();
662	taskstats_init_early();
663	delayacct_init();
664
665	check_bugs();
666
667	acpi_subsystem_init();
668	sfi_init_late();
669
670	if (efi_enabled(EFI_RUNTIME_SERVICES)) {
671		efi_late_init();
672		efi_free_boot_services();
673	}
674
675	ftrace_init();
676
677	/* Do the rest non-__init'ed, we're now alive */
678	rest_init();
679}
680
681/* Call all constructor functions linked into the kernel. */
682static void __init do_ctors(void)
683{
684#ifdef CONFIG_CONSTRUCTORS
685	ctor_fn_t *fn = (ctor_fn_t *) __ctors_start;
686
687	for (; fn < (ctor_fn_t *) __ctors_end; fn++)
688		(*fn)();
689#endif
690}
691
692bool initcall_debug;
693core_param(initcall_debug, initcall_debug, bool, 0644);
694
695#ifdef CONFIG_KALLSYMS
696struct blacklist_entry {
697	struct list_head next;
698	char *buf;
699};
700
701static __initdata_or_module LIST_HEAD(blacklisted_initcalls);
702
703static int __init initcall_blacklist(char *str)
704{
705	char *str_entry;
706	struct blacklist_entry *entry;
707
708	/* str argument is a comma-separated list of functions */
709	do {
710		str_entry = strsep(&str, ",");
711		if (str_entry) {
712			pr_debug("blacklisting initcall %s\n", str_entry);
713			entry = alloc_bootmem(sizeof(*entry));
714			entry->buf = alloc_bootmem(strlen(str_entry) + 1);
715			strcpy(entry->buf, str_entry);
716			list_add(&entry->next, &blacklisted_initcalls);
717		}
718	} while (str_entry);
719
720	return 0;
721}
722
723static bool __init_or_module initcall_blacklisted(initcall_t fn)
724{
725	struct list_head *tmp;
726	struct blacklist_entry *entry;
727	char *fn_name;
728
729	fn_name = kasprintf(GFP_KERNEL, "%pf", fn);
730	if (!fn_name)
731		return false;
732
733	list_for_each(tmp, &blacklisted_initcalls) {
734		entry = list_entry(tmp, struct blacklist_entry, next);
735		if (!strcmp(fn_name, entry->buf)) {
736			pr_debug("initcall %s blacklisted\n", fn_name);
737			kfree(fn_name);
738			return true;
739		}
740	}
741
742	kfree(fn_name);
743	return false;
744}
745#else
746static int __init initcall_blacklist(char *str)
747{
748	pr_warn("initcall_blacklist requires CONFIG_KALLSYMS\n");
749	return 0;
750}
751
752static bool __init_or_module initcall_blacklisted(initcall_t fn)
753{
754	return false;
755}
756#endif
757__setup("initcall_blacklist=", initcall_blacklist);
758
759static int __init_or_module do_one_initcall_debug(initcall_t fn)
760{
761	ktime_t calltime, delta, rettime;
762	unsigned long long duration;
763	int ret;
764
765	printk(KERN_DEBUG "calling  %pF @ %i\n", fn, task_pid_nr(current));
766	calltime = ktime_get();
767	ret = fn();
768	rettime = ktime_get();
769	delta = ktime_sub(rettime, calltime);
770	duration = (unsigned long long) ktime_to_ns(delta) >> 10;
771	printk(KERN_DEBUG "initcall %pF returned %d after %lld usecs\n",
772		 fn, ret, duration);
773
774	return ret;
775}
776
777int __init_or_module do_one_initcall(initcall_t fn)
778{
779	int count = preempt_count();
780	int ret;
781	char msgbuf[64];
782
783	if (initcall_blacklisted(fn))
784		return -EPERM;
785
786	if (initcall_debug)
787		ret = do_one_initcall_debug(fn);
788	else
789		ret = fn();
790
791	msgbuf[0] = 0;
792
793	if (preempt_count() != count) {
794		sprintf(msgbuf, "preemption imbalance ");
795		preempt_count_set(count);
796	}
797	if (irqs_disabled()) {
798		strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf));
799		local_irq_enable();
800	}
801	WARN(msgbuf[0], "initcall %pF returned with %s\n", fn, msgbuf);
802
803	return ret;
804}
805
806
807extern initcall_t __initcall_start[];
808extern initcall_t __initcall0_start[];
809extern initcall_t __initcall1_start[];
810extern initcall_t __initcall2_start[];
811extern initcall_t __initcall3_start[];
812extern initcall_t __initcall4_start[];
813extern initcall_t __initcall5_start[];
814extern initcall_t __initcall6_start[];
815extern initcall_t __initcall7_start[];
816extern initcall_t __initcall_end[];
817
818static initcall_t *initcall_levels[] __initdata = {
819	__initcall0_start,
820	__initcall1_start,
821	__initcall2_start,
822	__initcall3_start,
823	__initcall4_start,
824	__initcall5_start,
825	__initcall6_start,
826	__initcall7_start,
827	__initcall_end,
828};
829
830/* Keep these in sync with initcalls in include/linux/init.h */
831static char *initcall_level_names[] __initdata = {
832	"early",
833	"core",
834	"postcore",
835	"arch",
836	"subsys",
837	"fs",
838	"device",
839	"late",
840};
841
842static 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		   __stop___param - __start___param,
850		   level, level,
851		   &repair_env_string);
852
853	for (fn = initcall_levels[level]; fn < initcall_levels[level+1]; fn++)
854		do_one_initcall(*fn);
855}
856
857static void __init do_initcalls(void)
858{
859	int level;
860
861	for (level = 0; level < ARRAY_SIZE(initcall_levels) - 1; level++)
862		do_initcall_level(level);
863}
864
865/*
866 * Ok, the machine is now initialized. None of the devices
867 * have been touched yet, but the CPU subsystem is up and
868 * running, and memory and process management works.
869 *
870 * Now we can finally start doing some real work..
871 */
872static void __init do_basic_setup(void)
873{
874	cpuset_init_smp();
875	usermodehelper_init();
876	shmem_init();
877	driver_init();
878	init_irq_proc();
879	do_ctors();
880	usermodehelper_enable();
881	do_initcalls();
882	random_int_secret_init();
883}
884
885static void __init do_pre_smp_initcalls(void)
886{
887	initcall_t *fn;
888
889	for (fn = __initcall_start; fn < __initcall0_start; fn++)
890		do_one_initcall(*fn);
891}
892
893/*
894 * This function requests modules which should be loaded by default and is
895 * called twice right after initrd is mounted and right before init is
896 * exec'd.  If such modules are on either initrd or rootfs, they will be
897 * loaded before control is passed to userland.
898 */
899void __init load_default_modules(void)
900{
901	load_default_elevator_module();
902}
903
904static int run_init_process(const char *init_filename)
905{
906	argv_init[0] = init_filename;
907	return do_execve(getname_kernel(init_filename),
908		(const char __user *const __user *)argv_init,
909		(const char __user *const __user *)envp_init);
910}
911
912static int try_to_run_init_process(const char *init_filename)
913{
914	int ret;
915
916	ret = run_init_process(init_filename);
917
918	if (ret && ret != -ENOENT) {
919		pr_err("Starting init: %s exists but couldn't execute it (error %d)\n",
920		       init_filename, ret);
921	}
922
923	return ret;
924}
925
926static noinline void __init kernel_init_freeable(void);
927
928static int __ref kernel_init(void *unused)
929{
930	int ret;
931
932	kernel_init_freeable();
933	/* need to finish all async __init code before freeing the memory */
934	async_synchronize_full();
935	free_initmem();
936	mark_rodata_ro();
937	system_state = SYSTEM_RUNNING;
938	numa_default_policy();
939
940	flush_delayed_fput();
941
942	if (ramdisk_execute_command) {
943		ret = run_init_process(ramdisk_execute_command);
944		if (!ret)
945			return 0;
946		pr_err("Failed to execute %s (error %d)\n",
947		       ramdisk_execute_command, ret);
948	}
949
950	/*
951	 * We try each of these until one succeeds.
952	 *
953	 * The Bourne shell can be used instead of init if we are
954	 * trying to recover a really broken machine.
955	 */
956	if (execute_command) {
957		ret = run_init_process(execute_command);
958		if (!ret)
959			return 0;
960		panic("Requested init %s failed (error %d).",
961		      execute_command, ret);
962	}
963	if (!try_to_run_init_process("/sbin/init") ||
964	    !try_to_run_init_process("/etc/init") ||
965	    !try_to_run_init_process("/bin/init") ||
966	    !try_to_run_init_process("/bin/sh"))
967		return 0;
968
969	panic("No working init found.  Try passing init= option to kernel. "
970	      "See Linux Documentation/init.txt for guidance.");
971}
972
973static noinline void __init kernel_init_freeable(void)
974{
975	/*
976	 * Wait until kthreadd is all set-up.
977	 */
978	wait_for_completion(&kthreadd_done);
979
980	/* Now the scheduler is fully set up and can do blocking allocations */
981	gfp_allowed_mask = __GFP_BITS_MASK;
982
983	/*
984	 * init can allocate pages on any node
985	 */
986	set_mems_allowed(node_states[N_MEMORY]);
987	/*
988	 * init can run on any cpu.
989	 */
990	set_cpus_allowed_ptr(current, cpu_all_mask);
991
992	cad_pid = task_pid(current);
993
994	smp_prepare_cpus(setup_max_cpus);
995
996	do_pre_smp_initcalls();
997	lockup_detector_init();
998
999	smp_init();
1000	sched_init_smp();
1001
1002	do_basic_setup();
1003
1004	/* Open the /dev/console on the rootfs, this should never fail */
1005	if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0)
1006		pr_err("Warning: unable to open an initial console.\n");
1007
1008	(void) sys_dup(0);
1009	(void) sys_dup(0);
1010	/*
1011	 * check if there is an early userspace init.  If yes, let it do all
1012	 * the work
1013	 */
1014
1015	if (!ramdisk_execute_command)
1016		ramdisk_execute_command = "/init";
1017
1018	if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) {
1019		ramdisk_execute_command = NULL;
1020		prepare_namespace();
1021	}
1022
1023	/*
1024	 * Ok, we have completed the initial bootup, and
1025	 * we're essentially up and running. Get rid of the
1026	 * initmem segments and start the user-mode stuff..
1027	 *
1028	 * rootfs is available now, try loading the public keys
1029	 * and default modules
1030	 */
1031
1032	integrity_load_keys();
1033	load_default_modules();
1034}
1035