Loading...
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#include <linux/types.h>
13#include <linux/module.h>
14#include <linux/proc_fs.h>
15#include <linux/kernel.h>
16#include <linux/syscalls.h>
17#include <linux/stackprotector.h>
18#include <linux/string.h>
19#include <linux/ctype.h>
20#include <linux/delay.h>
21#include <linux/ioport.h>
22#include <linux/init.h>
23#include <linux/initrd.h>
24#include <linux/bootmem.h>
25#include <linux/acpi.h>
26#include <linux/tty.h>
27#include <linux/percpu.h>
28#include <linux/kmod.h>
29#include <linux/vmalloc.h>
30#include <linux/kernel_stat.h>
31#include <linux/start_kernel.h>
32#include <linux/security.h>
33#include <linux/smp.h>
34#include <linux/profile.h>
35#include <linux/rcupdate.h>
36#include <linux/moduleparam.h>
37#include <linux/kallsyms.h>
38#include <linux/writeback.h>
39#include <linux/cpu.h>
40#include <linux/cpuset.h>
41#include <linux/cgroup.h>
42#include <linux/efi.h>
43#include <linux/tick.h>
44#include <linux/interrupt.h>
45#include <linux/taskstats_kern.h>
46#include <linux/delayacct.h>
47#include <linux/unistd.h>
48#include <linux/rmap.h>
49#include <linux/mempolicy.h>
50#include <linux/key.h>
51#include <linux/buffer_head.h>
52#include <linux/page_cgroup.h>
53#include <linux/debug_locks.h>
54#include <linux/debugobjects.h>
55#include <linux/lockdep.h>
56#include <linux/kmemleak.h>
57#include <linux/pid_namespace.h>
58#include <linux/device.h>
59#include <linux/kthread.h>
60#include <linux/sched.h>
61#include <linux/signal.h>
62#include <linux/idr.h>
63#include <linux/kgdb.h>
64#include <linux/ftrace.h>
65#include <linux/async.h>
66#include <linux/kmemcheck.h>
67#include <linux/sfi.h>
68#include <linux/shmem_fs.h>
69#include <linux/slab.h>
70#include <linux/perf_event.h>
71
72#include <asm/io.h>
73#include <asm/bugs.h>
74#include <asm/setup.h>
75#include <asm/sections.h>
76#include <asm/cacheflush.h>
77
78#ifdef CONFIG_X86_LOCAL_APIC
79#include <asm/smp.h>
80#endif
81
82static int kernel_init(void *);
83
84extern void init_IRQ(void);
85extern void fork_init(unsigned long);
86extern void mca_init(void);
87extern void sbus_init(void);
88extern void prio_tree_init(void);
89extern void radix_tree_init(void);
90extern void free_initmem(void);
91#ifndef CONFIG_DEBUG_RODATA
92static inline void mark_rodata_ro(void) { }
93#endif
94
95#ifdef CONFIG_TC
96extern void tc_init(void);
97#endif
98
99/*
100 * Debug helper: via this flag we know that we are in 'early bootup code'
101 * where only the boot processor is running with IRQ disabled. This means
102 * two things - IRQ must not be enabled before the flag is cleared and some
103 * operations which are not allowed with IRQ disabled are allowed while the
104 * flag is set.
105 */
106bool early_boot_irqs_disabled __read_mostly;
107
108enum system_states system_state __read_mostly;
109EXPORT_SYMBOL(system_state);
110
111/*
112 * Boot command-line arguments
113 */
114#define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT
115#define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT
116
117extern void time_init(void);
118/* Default late time init is NULL. archs can override this later. */
119void (*__initdata late_time_init)(void);
120extern void softirq_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
129static char *execute_command;
130static char *ramdisk_execute_command;
131
132/*
133 * If set, this is an indication to the drivers that reset the underlying
134 * device before going ahead with the initialization otherwise driver might
135 * rely on the BIOS and skip the reset operation.
136 *
137 * This is useful if kernel is booting in an unreliable environment.
138 * For ex. kdump situaiton where previous kernel has crashed, BIOS has been
139 * skipped and devices will be in unknown state.
140 */
141unsigned int reset_devices;
142EXPORT_SYMBOL(reset_devices);
143
144static int __init set_reset_devices(char *str)
145{
146 reset_devices = 1;
147 return 1;
148}
149
150__setup("reset_devices", set_reset_devices);
151
152static const char * argv_init[MAX_INIT_ARGS+2] = { "init", NULL, };
153const char * envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, };
154static const char *panic_later, *panic_param;
155
156extern const struct obs_kernel_param __setup_start[], __setup_end[];
157
158static int __init obsolete_checksetup(char *line)
159{
160 const struct obs_kernel_param *p;
161 int had_early_param = 0;
162
163 p = __setup_start;
164 do {
165 int n = strlen(p->str);
166 if (!strncmp(line, p->str, n)) {
167 if (p->early) {
168 /* Already done in parse_early_param?
169 * (Needs exact match on param part).
170 * Keep iterating, as we can have early
171 * params and __setups of same names 8( */
172 if (line[n] == '\0' || line[n] == '=')
173 had_early_param = 1;
174 } else if (!p->setup_func) {
175 printk(KERN_WARNING "Parameter %s is obsolete,"
176 " ignored\n", p->str);
177 return 1;
178 } else if (p->setup_func(line + n))
179 return 1;
180 }
181 p++;
182 } while (p < __setup_end);
183
184 return had_early_param;
185}
186
187/*
188 * This should be approx 2 Bo*oMips to start (note initial shift), and will
189 * still work even if initially too large, it will just take slightly longer
190 */
191unsigned long loops_per_jiffy = (1<<12);
192
193EXPORT_SYMBOL(loops_per_jiffy);
194
195static int __init debug_kernel(char *str)
196{
197 console_loglevel = 10;
198 return 0;
199}
200
201static int __init quiet_kernel(char *str)
202{
203 console_loglevel = 4;
204 return 0;
205}
206
207early_param("debug", debug_kernel);
208early_param("quiet", quiet_kernel);
209
210static int __init loglevel(char *str)
211{
212 int newlevel;
213
214 /*
215 * Only update loglevel value when a correct setting was passed,
216 * to prevent blind crashes (when loglevel being set to 0) that
217 * are quite hard to debug
218 */
219 if (get_option(&str, &newlevel)) {
220 console_loglevel = newlevel;
221 return 0;
222 }
223
224 return -EINVAL;
225}
226
227early_param("loglevel", loglevel);
228
229/*
230 * Unknown boot options get handed to init, unless they look like
231 * unused parameters (modprobe will find them in /proc/cmdline).
232 */
233static int __init unknown_bootoption(char *param, char *val)
234{
235 /* Change NUL term back to "=", to make "param" the whole string. */
236 if (val) {
237 /* param=val or param="val"? */
238 if (val == param+strlen(param)+1)
239 val[-1] = '=';
240 else if (val == param+strlen(param)+2) {
241 val[-2] = '=';
242 memmove(val-1, val, strlen(val)+1);
243 val--;
244 } else
245 BUG();
246 }
247
248 /* Handle obsolete-style parameters */
249 if (obsolete_checksetup(param))
250 return 0;
251
252 /* Unused module parameter. */
253 if (strchr(param, '.') && (!val || strchr(param, '.') < val))
254 return 0;
255
256 if (panic_later)
257 return 0;
258
259 if (val) {
260 /* Environment option */
261 unsigned int i;
262 for (i = 0; envp_init[i]; i++) {
263 if (i == MAX_INIT_ENVS) {
264 panic_later = "Too many boot env vars at `%s'";
265 panic_param = param;
266 }
267 if (!strncmp(param, envp_init[i], val - param))
268 break;
269 }
270 envp_init[i] = param;
271 } else {
272 /* Command line option */
273 unsigned int i;
274 for (i = 0; argv_init[i]; i++) {
275 if (i == MAX_INIT_ARGS) {
276 panic_later = "Too many boot init vars at `%s'";
277 panic_param = param;
278 }
279 }
280 argv_init[i] = param;
281 }
282 return 0;
283}
284
285#ifdef CONFIG_DEBUG_PAGEALLOC
286int __read_mostly debug_pagealloc_enabled = 0;
287#endif
288
289static int __init init_setup(char *str)
290{
291 unsigned int i;
292
293 execute_command = str;
294 /*
295 * In case LILO is going to boot us with default command line,
296 * it prepends "auto" before the whole cmdline which makes
297 * the shell think it should execute a script with such name.
298 * So we ignore all arguments entered _before_ init=... [MJ]
299 */
300 for (i = 1; i < MAX_INIT_ARGS; i++)
301 argv_init[i] = NULL;
302 return 1;
303}
304__setup("init=", init_setup);
305
306static int __init rdinit_setup(char *str)
307{
308 unsigned int i;
309
310 ramdisk_execute_command = str;
311 /* See "auto" comment in init_setup */
312 for (i = 1; i < MAX_INIT_ARGS; i++)
313 argv_init[i] = NULL;
314 return 1;
315}
316__setup("rdinit=", rdinit_setup);
317
318#ifndef CONFIG_SMP
319static const unsigned int setup_max_cpus = NR_CPUS;
320#ifdef CONFIG_X86_LOCAL_APIC
321static void __init smp_init(void)
322{
323 APIC_init_uniprocessor();
324}
325#else
326#define smp_init() do { } while (0)
327#endif
328
329static inline void setup_nr_cpu_ids(void) { }
330static inline void smp_prepare_cpus(unsigned int maxcpus) { }
331#endif
332
333/*
334 * We need to store the untouched command line for future reference.
335 * We also need to store the touched command line since the parameter
336 * parsing is performed in place, and we should allow a component to
337 * store reference of name/value for future reference.
338 */
339static void __init setup_command_line(char *command_line)
340{
341 saved_command_line = alloc_bootmem(strlen (boot_command_line)+1);
342 static_command_line = alloc_bootmem(strlen (command_line)+1);
343 strcpy (saved_command_line, boot_command_line);
344 strcpy (static_command_line, command_line);
345}
346
347/*
348 * We need to finalize in a non-__init function or else race conditions
349 * between the root thread and the init thread may cause start_kernel to
350 * be reaped by free_initmem before the root thread has proceeded to
351 * cpu_idle.
352 *
353 * gcc-3.4 accidentally inlines this function, so use noinline.
354 */
355
356static __initdata DECLARE_COMPLETION(kthreadd_done);
357
358static noinline void __init_refok rest_init(void)
359{
360 int pid;
361
362 rcu_scheduler_starting();
363 /*
364 * We need to spawn init first so that it obtains pid 1, however
365 * the init task will end up wanting to create kthreads, which, if
366 * we schedule it before we create kthreadd, will OOPS.
367 */
368 kernel_thread(kernel_init, NULL, CLONE_FS | CLONE_SIGHAND);
369 numa_default_policy();
370 pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES);
371 rcu_read_lock();
372 kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns);
373 rcu_read_unlock();
374 complete(&kthreadd_done);
375
376 /*
377 * The boot idle thread must execute schedule()
378 * at least once to get things moving:
379 */
380 init_idle_bootup_task(current);
381 preempt_enable_no_resched();
382 schedule();
383
384 /* Call into cpu_idle with preempt disabled */
385 preempt_disable();
386 cpu_idle();
387}
388
389/* Check for early params. */
390static int __init do_early_param(char *param, char *val)
391{
392 const struct obs_kernel_param *p;
393
394 for (p = __setup_start; p < __setup_end; p++) {
395 if ((p->early && strcmp(param, p->str) == 0) ||
396 (strcmp(param, "console") == 0 &&
397 strcmp(p->str, "earlycon") == 0)
398 ) {
399 if (p->setup_func(val) != 0)
400 printk(KERN_WARNING
401 "Malformed early option '%s'\n", param);
402 }
403 }
404 /* We accept everything at this stage. */
405 return 0;
406}
407
408void __init parse_early_options(char *cmdline)
409{
410 parse_args("early options", cmdline, NULL, 0, do_early_param);
411}
412
413/* Arch code calls this early on, or if not, just before other parsing. */
414void __init parse_early_param(void)
415{
416 static __initdata int done = 0;
417 static __initdata char tmp_cmdline[COMMAND_LINE_SIZE];
418
419 if (done)
420 return;
421
422 /* All fall through to do_early_param. */
423 strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE);
424 parse_early_options(tmp_cmdline);
425 done = 1;
426}
427
428/*
429 * Activate the first processor.
430 */
431
432static void __init boot_cpu_init(void)
433{
434 int cpu = smp_processor_id();
435 /* Mark the boot cpu "present", "online" etc for SMP and UP case */
436 set_cpu_online(cpu, true);
437 set_cpu_active(cpu, true);
438 set_cpu_present(cpu, true);
439 set_cpu_possible(cpu, true);
440}
441
442void __init __weak smp_setup_processor_id(void)
443{
444}
445
446void __init __weak thread_info_cache_init(void)
447{
448}
449
450/*
451 * Set up kernel memory allocators
452 */
453static void __init mm_init(void)
454{
455 /*
456 * page_cgroup requires countinous pages as memmap
457 * and it's bigger than MAX_ORDER unless SPARSEMEM.
458 */
459 page_cgroup_init_flatmem();
460 mem_init();
461 kmem_cache_init();
462 percpu_init_late();
463 pgtable_cache_init();
464 vmalloc_init();
465}
466
467asmlinkage void __init start_kernel(void)
468{
469 char * command_line;
470 extern const struct kernel_param __start___param[], __stop___param[];
471
472 smp_setup_processor_id();
473
474 /*
475 * Need to run as early as possible, to initialize the
476 * lockdep hash:
477 */
478 lockdep_init();
479 debug_objects_early_init();
480
481 /*
482 * Set up the the initial canary ASAP:
483 */
484 boot_init_stack_canary();
485
486 cgroup_init_early();
487
488 local_irq_disable();
489 early_boot_irqs_disabled = true;
490
491/*
492 * Interrupts are still disabled. Do necessary setups, then
493 * enable them
494 */
495 tick_init();
496 boot_cpu_init();
497 page_address_init();
498 printk(KERN_NOTICE "%s", linux_banner);
499 setup_arch(&command_line);
500 mm_init_owner(&init_mm, &init_task);
501 mm_init_cpumask(&init_mm);
502 setup_command_line(command_line);
503 setup_nr_cpu_ids();
504 setup_per_cpu_areas();
505 smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */
506
507 build_all_zonelists(NULL);
508 page_alloc_init();
509
510 printk(KERN_NOTICE "Kernel command line: %s\n", boot_command_line);
511 parse_early_param();
512 parse_args("Booting kernel", static_command_line, __start___param,
513 __stop___param - __start___param,
514 &unknown_bootoption);
515 /*
516 * These use large bootmem allocations and must precede
517 * kmem_cache_init()
518 */
519 setup_log_buf(0);
520 pidhash_init();
521 vfs_caches_init_early();
522 sort_main_extable();
523 trap_init();
524 mm_init();
525
526 /*
527 * Set up the scheduler prior starting any interrupts (such as the
528 * timer interrupt). Full topology setup happens at smp_init()
529 * time - but meanwhile we still have a functioning scheduler.
530 */
531 sched_init();
532 /*
533 * Disable preemption - early bootup scheduling is extremely
534 * fragile until we cpu_idle() for the first time.
535 */
536 preempt_disable();
537 if (!irqs_disabled()) {
538 printk(KERN_WARNING "start_kernel(): bug: interrupts were "
539 "enabled *very* early, fixing it\n");
540 local_irq_disable();
541 }
542 idr_init_cache();
543 perf_event_init();
544 rcu_init();
545 radix_tree_init();
546 /* init some links before init_ISA_irqs() */
547 early_irq_init();
548 init_IRQ();
549 prio_tree_init();
550 init_timers();
551 hrtimers_init();
552 softirq_init();
553 timekeeping_init();
554 time_init();
555 profile_init();
556 call_function_init();
557 if (!irqs_disabled())
558 printk(KERN_CRIT "start_kernel(): bug: interrupts were "
559 "enabled early\n");
560 early_boot_irqs_disabled = false;
561 local_irq_enable();
562
563 /* Interrupts are enabled now so all GFP allocations are safe. */
564 gfp_allowed_mask = __GFP_BITS_MASK;
565
566 kmem_cache_init_late();
567
568 /*
569 * HACK ALERT! This is early. We're enabling the console before
570 * we've done PCI setups etc, and console_init() must be aware of
571 * this. But we do want output early, in case something goes wrong.
572 */
573 console_init();
574 if (panic_later)
575 panic(panic_later, panic_param);
576
577 lockdep_info();
578
579 /*
580 * Need to run this when irqs are enabled, because it wants
581 * to self-test [hard/soft]-irqs on/off lock inversion bugs
582 * too:
583 */
584 locking_selftest();
585
586#ifdef CONFIG_BLK_DEV_INITRD
587 if (initrd_start && !initrd_below_start_ok &&
588 page_to_pfn(virt_to_page((void *)initrd_start)) < min_low_pfn) {
589 printk(KERN_CRIT "initrd overwritten (0x%08lx < 0x%08lx) - "
590 "disabling it.\n",
591 page_to_pfn(virt_to_page((void *)initrd_start)),
592 min_low_pfn);
593 initrd_start = 0;
594 }
595#endif
596 page_cgroup_init();
597 enable_debug_pagealloc();
598 debug_objects_mem_init();
599 kmemleak_init();
600 setup_per_cpu_pageset();
601 numa_policy_init();
602 if (late_time_init)
603 late_time_init();
604 sched_clock_init();
605 calibrate_delay();
606 pidmap_init();
607 anon_vma_init();
608#ifdef CONFIG_X86
609 if (efi_enabled)
610 efi_enter_virtual_mode();
611#endif
612 thread_info_cache_init();
613 cred_init();
614 fork_init(totalram_pages);
615 proc_caches_init();
616 buffer_init();
617 key_init();
618 security_init();
619 dbg_late_init();
620 vfs_caches_init(totalram_pages);
621 signals_init();
622 /* rootfs populating might need page-writeback */
623 page_writeback_init();
624#ifdef CONFIG_PROC_FS
625 proc_root_init();
626#endif
627 cgroup_init();
628 cpuset_init();
629 taskstats_init_early();
630 delayacct_init();
631
632 check_bugs();
633
634 acpi_early_init(); /* before LAPIC and SMP init */
635 sfi_init_late();
636
637 ftrace_init();
638
639 /* Do the rest non-__init'ed, we're now alive */
640 rest_init();
641}
642
643/* Call all constructor functions linked into the kernel. */
644static void __init do_ctors(void)
645{
646#ifdef CONFIG_CONSTRUCTORS
647 ctor_fn_t *fn = (ctor_fn_t *) __ctors_start;
648
649 for (; fn < (ctor_fn_t *) __ctors_end; fn++)
650 (*fn)();
651#endif
652}
653
654int initcall_debug;
655core_param(initcall_debug, initcall_debug, bool, 0644);
656
657static char msgbuf[64];
658
659static int __init_or_module do_one_initcall_debug(initcall_t fn)
660{
661 ktime_t calltime, delta, rettime;
662 unsigned long long duration;
663 int ret;
664
665 printk(KERN_DEBUG "calling %pF @ %i\n", fn, task_pid_nr(current));
666 calltime = ktime_get();
667 ret = fn();
668 rettime = ktime_get();
669 delta = ktime_sub(rettime, calltime);
670 duration = (unsigned long long) ktime_to_ns(delta) >> 10;
671 printk(KERN_DEBUG "initcall %pF returned %d after %lld usecs\n", fn,
672 ret, duration);
673
674 return ret;
675}
676
677int __init_or_module do_one_initcall(initcall_t fn)
678{
679 int count = preempt_count();
680 int ret;
681
682 if (initcall_debug)
683 ret = do_one_initcall_debug(fn);
684 else
685 ret = fn();
686
687 msgbuf[0] = 0;
688
689 if (ret && ret != -ENODEV && initcall_debug)
690 sprintf(msgbuf, "error code %d ", ret);
691
692 if (preempt_count() != count) {
693 strlcat(msgbuf, "preemption imbalance ", sizeof(msgbuf));
694 preempt_count() = count;
695 }
696 if (irqs_disabled()) {
697 strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf));
698 local_irq_enable();
699 }
700 if (msgbuf[0]) {
701 printk("initcall %pF returned with %s\n", fn, msgbuf);
702 }
703
704 return ret;
705}
706
707
708extern initcall_t __initcall_start[], __initcall_end[], __early_initcall_end[];
709
710static void __init do_initcalls(void)
711{
712 initcall_t *fn;
713
714 for (fn = __early_initcall_end; fn < __initcall_end; fn++)
715 do_one_initcall(*fn);
716}
717
718/*
719 * Ok, the machine is now initialized. None of the devices
720 * have been touched yet, but the CPU subsystem is up and
721 * running, and memory and process management works.
722 *
723 * Now we can finally start doing some real work..
724 */
725static void __init do_basic_setup(void)
726{
727 cpuset_init_smp();
728 usermodehelper_init();
729 shmem_init();
730 driver_init();
731 init_irq_proc();
732 do_ctors();
733 usermodehelper_enable();
734 do_initcalls();
735}
736
737static void __init do_pre_smp_initcalls(void)
738{
739 initcall_t *fn;
740
741 for (fn = __initcall_start; fn < __early_initcall_end; fn++)
742 do_one_initcall(*fn);
743}
744
745static void run_init_process(const char *init_filename)
746{
747 argv_init[0] = init_filename;
748 kernel_execve(init_filename, argv_init, envp_init);
749}
750
751/* This is a non __init function. Force it to be noinline otherwise gcc
752 * makes it inline to init() and it becomes part of init.text section
753 */
754static noinline int init_post(void)
755{
756 /* need to finish all async __init code before freeing the memory */
757 async_synchronize_full();
758 free_initmem();
759 mark_rodata_ro();
760 system_state = SYSTEM_RUNNING;
761 numa_default_policy();
762
763
764 current->signal->flags |= SIGNAL_UNKILLABLE;
765
766 if (ramdisk_execute_command) {
767 run_init_process(ramdisk_execute_command);
768 printk(KERN_WARNING "Failed to execute %s\n",
769 ramdisk_execute_command);
770 }
771
772 /*
773 * We try each of these until one succeeds.
774 *
775 * The Bourne shell can be used instead of init if we are
776 * trying to recover a really broken machine.
777 */
778 if (execute_command) {
779 run_init_process(execute_command);
780 printk(KERN_WARNING "Failed to execute %s. Attempting "
781 "defaults...\n", execute_command);
782 }
783 run_init_process("/sbin/init");
784 run_init_process("/etc/init");
785 run_init_process("/bin/init");
786 run_init_process("/bin/sh");
787
788 panic("No init found. Try passing init= option to kernel. "
789 "See Linux Documentation/init.txt for guidance.");
790}
791
792static int __init kernel_init(void * unused)
793{
794 /*
795 * Wait until kthreadd is all set-up.
796 */
797 wait_for_completion(&kthreadd_done);
798 /*
799 * init can allocate pages on any node
800 */
801 set_mems_allowed(node_states[N_HIGH_MEMORY]);
802 /*
803 * init can run on any cpu.
804 */
805 set_cpus_allowed_ptr(current, cpu_all_mask);
806
807 cad_pid = task_pid(current);
808
809 smp_prepare_cpus(setup_max_cpus);
810
811 do_pre_smp_initcalls();
812 lockup_detector_init();
813
814 smp_init();
815 sched_init_smp();
816
817 do_basic_setup();
818
819 /* Open the /dev/console on the rootfs, this should never fail */
820 if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0)
821 printk(KERN_WARNING "Warning: unable to open an initial console.\n");
822
823 (void) sys_dup(0);
824 (void) sys_dup(0);
825 /*
826 * check if there is an early userspace init. If yes, let it do all
827 * the work
828 */
829
830 if (!ramdisk_execute_command)
831 ramdisk_execute_command = "/init";
832
833 if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) {
834 ramdisk_execute_command = NULL;
835 prepare_namespace();
836 }
837
838 /*
839 * Ok, we have completed the initial bootup, and
840 * we're essentially up and running. Get rid of the
841 * initmem segments and start the user-mode stuff..
842 */
843
844 init_post();
845 return 0;
846}
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_cgroup.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
81#include <asm/io.h>
82#include <asm/bugs.h>
83#include <asm/setup.h>
84#include <asm/sections.h>
85#include <asm/cacheflush.h>
86
87#ifdef CONFIG_X86_LOCAL_APIC
88#include <asm/smp.h>
89#endif
90
91static int kernel_init(void *);
92
93extern void init_IRQ(void);
94extern void fork_init(unsigned long);
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 = false;
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 situaiton 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);
201
202EXPORT_SYMBOL(loops_per_jiffy);
203
204static int __init debug_kernel(char *str)
205{
206 console_loglevel = 10;
207 return 0;
208}
209
210static int __init quiet_kernel(char *str)
211{
212 console_loglevel = 4;
213 return 0;
214}
215
216early_param("debug", debug_kernel);
217early_param("quiet", quiet_kernel);
218
219static int __init loglevel(char *str)
220{
221 int newlevel;
222
223 /*
224 * Only update loglevel value when a correct setting was passed,
225 * to prevent blind crashes (when loglevel being set to 0) that
226 * are quite hard to debug
227 */
228 if (get_option(&str, &newlevel)) {
229 console_loglevel = newlevel;
230 return 0;
231 }
232
233 return -EINVAL;
234}
235
236early_param("loglevel", loglevel);
237
238/* Change NUL term back to "=", to make "param" the whole string. */
239static int __init repair_env_string(char *param, char *val, const char *unused)
240{
241 if (val) {
242 /* param=val or param="val"? */
243 if (val == param+strlen(param)+1)
244 val[-1] = '=';
245 else if (val == param+strlen(param)+2) {
246 val[-2] = '=';
247 memmove(val-1, val, strlen(val)+1);
248 val--;
249 } else
250 BUG();
251 }
252 return 0;
253}
254
255/*
256 * Unknown boot options get handed to init, unless they look like
257 * unused parameters (modprobe will find them in /proc/cmdline).
258 */
259static int __init unknown_bootoption(char *param, char *val, const char *unused)
260{
261 repair_env_string(param, val, unused);
262
263 /* Handle obsolete-style parameters */
264 if (obsolete_checksetup(param))
265 return 0;
266
267 /* Unused module parameter. */
268 if (strchr(param, '.') && (!val || strchr(param, '.') < val))
269 return 0;
270
271 if (panic_later)
272 return 0;
273
274 if (val) {
275 /* Environment option */
276 unsigned int i;
277 for (i = 0; envp_init[i]; i++) {
278 if (i == MAX_INIT_ENVS) {
279 panic_later = "env";
280 panic_param = param;
281 }
282 if (!strncmp(param, envp_init[i], val - param))
283 break;
284 }
285 envp_init[i] = param;
286 } else {
287 /* Command line option */
288 unsigned int i;
289 for (i = 0; argv_init[i]; i++) {
290 if (i == MAX_INIT_ARGS) {
291 panic_later = "init";
292 panic_param = param;
293 }
294 }
295 argv_init[i] = param;
296 }
297 return 0;
298}
299
300static int __init init_setup(char *str)
301{
302 unsigned int i;
303
304 execute_command = str;
305 /*
306 * In case LILO is going to boot us with default command line,
307 * it prepends "auto" before the whole cmdline which makes
308 * the shell think it should execute a script with such name.
309 * So we ignore all arguments entered _before_ init=... [MJ]
310 */
311 for (i = 1; i < MAX_INIT_ARGS; i++)
312 argv_init[i] = NULL;
313 return 1;
314}
315__setup("init=", init_setup);
316
317static int __init rdinit_setup(char *str)
318{
319 unsigned int i;
320
321 ramdisk_execute_command = str;
322 /* See "auto" comment in init_setup */
323 for (i = 1; i < MAX_INIT_ARGS; i++)
324 argv_init[i] = NULL;
325 return 1;
326}
327__setup("rdinit=", rdinit_setup);
328
329#ifndef CONFIG_SMP
330static const unsigned int setup_max_cpus = NR_CPUS;
331#ifdef CONFIG_X86_LOCAL_APIC
332static void __init smp_init(void)
333{
334 APIC_init_uniprocessor();
335}
336#else
337#define smp_init() do { } while (0)
338#endif
339
340static inline void setup_nr_cpu_ids(void) { }
341static inline void smp_prepare_cpus(unsigned int maxcpus) { }
342#endif
343
344/*
345 * We need to store the untouched command line for future reference.
346 * We also need to store the touched command line since the parameter
347 * parsing is performed in place, and we should allow a component to
348 * store reference of name/value for future reference.
349 */
350static void __init setup_command_line(char *command_line)
351{
352 saved_command_line =
353 memblock_virt_alloc(strlen(boot_command_line) + 1, 0);
354 initcall_command_line =
355 memblock_virt_alloc(strlen(boot_command_line) + 1, 0);
356 static_command_line = memblock_virt_alloc(strlen(command_line) + 1, 0);
357 strcpy (saved_command_line, boot_command_line);
358 strcpy (static_command_line, command_line);
359}
360
361/*
362 * We need to finalize in a non-__init function or else race conditions
363 * between the root thread and the init thread may cause start_kernel to
364 * be reaped by free_initmem before the root thread has proceeded to
365 * cpu_idle.
366 *
367 * gcc-3.4 accidentally inlines this function, so use noinline.
368 */
369
370static __initdata DECLARE_COMPLETION(kthreadd_done);
371
372static noinline void __init_refok rest_init(void)
373{
374 int pid;
375
376 rcu_scheduler_starting();
377 /*
378 * We need to spawn init first so that it obtains pid 1, however
379 * the init task will end up wanting to create kthreads, which, if
380 * we schedule it before we create kthreadd, will OOPS.
381 */
382 kernel_thread(kernel_init, NULL, CLONE_FS | CLONE_SIGHAND);
383 numa_default_policy();
384 pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES);
385 rcu_read_lock();
386 kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns);
387 rcu_read_unlock();
388 complete(&kthreadd_done);
389
390 /*
391 * The boot idle thread must execute schedule()
392 * at least once to get things moving:
393 */
394 init_idle_bootup_task(current);
395 schedule_preempt_disabled();
396 /* Call into cpu_idle with preempt disabled */
397 cpu_startup_entry(CPUHP_ONLINE);
398}
399
400/* Check for early params. */
401static int __init do_early_param(char *param, char *val, const char *unused)
402{
403 const struct obs_kernel_param *p;
404
405 for (p = __setup_start; p < __setup_end; p++) {
406 if ((p->early && parameq(param, p->str)) ||
407 (strcmp(param, "console") == 0 &&
408 strcmp(p->str, "earlycon") == 0)
409 ) {
410 if (p->setup_func(val) != 0)
411 pr_warn("Malformed early option '%s'\n", param);
412 }
413 }
414 /* We accept everything at this stage. */
415 return 0;
416}
417
418void __init parse_early_options(char *cmdline)
419{
420 parse_args("early options", cmdline, NULL, 0, 0, 0, do_early_param);
421}
422
423/* Arch code calls this early on, or if not, just before other parsing. */
424void __init parse_early_param(void)
425{
426 static __initdata int done = 0;
427 static __initdata char tmp_cmdline[COMMAND_LINE_SIZE];
428
429 if (done)
430 return;
431
432 /* All fall through to do_early_param. */
433 strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE);
434 parse_early_options(tmp_cmdline);
435 done = 1;
436}
437
438/*
439 * Activate the first processor.
440 */
441
442static void __init boot_cpu_init(void)
443{
444 int cpu = smp_processor_id();
445 /* Mark the boot cpu "present", "online" etc for SMP and UP case */
446 set_cpu_online(cpu, true);
447 set_cpu_active(cpu, true);
448 set_cpu_present(cpu, true);
449 set_cpu_possible(cpu, true);
450}
451
452void __init __weak smp_setup_processor_id(void)
453{
454}
455
456# if THREAD_SIZE >= PAGE_SIZE
457void __init __weak thread_info_cache_init(void)
458{
459}
460#endif
461
462/*
463 * Set up kernel memory allocators
464 */
465static void __init mm_init(void)
466{
467 /*
468 * page_cgroup requires contiguous pages,
469 * bigger than MAX_ORDER unless SPARSEMEM.
470 */
471 page_cgroup_init_flatmem();
472 mem_init();
473 kmem_cache_init();
474 percpu_init_late();
475 pgtable_init();
476 vmalloc_init();
477}
478
479asmlinkage __visible void __init start_kernel(void)
480{
481 char * command_line;
482 extern const struct kernel_param __start___param[], __stop___param[];
483
484 /*
485 * Need to run as early as possible, to initialize the
486 * lockdep hash:
487 */
488 lockdep_init();
489 smp_setup_processor_id();
490 debug_objects_early_init();
491
492 /*
493 * Set up the the initial canary ASAP:
494 */
495 boot_init_stack_canary();
496
497 cgroup_init_early();
498
499 local_irq_disable();
500 early_boot_irqs_disabled = true;
501
502/*
503 * Interrupts are still disabled. Do necessary setups, then
504 * enable them
505 */
506 boot_cpu_init();
507 page_address_init();
508 pr_notice("%s", linux_banner);
509 setup_arch(&command_line);
510 mm_init_owner(&init_mm, &init_task);
511 mm_init_cpumask(&init_mm);
512 setup_command_line(command_line);
513 setup_nr_cpu_ids();
514 setup_per_cpu_areas();
515 smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */
516
517 build_all_zonelists(NULL, NULL);
518 page_alloc_init();
519
520 pr_notice("Kernel command line: %s\n", boot_command_line);
521 parse_early_param();
522 parse_args("Booting kernel", static_command_line, __start___param,
523 __stop___param - __start___param,
524 -1, -1, &unknown_bootoption);
525
526 jump_label_init();
527
528 /*
529 * These use large bootmem allocations and must precede
530 * kmem_cache_init()
531 */
532 setup_log_buf(0);
533 pidhash_init();
534 vfs_caches_init_early();
535 sort_main_extable();
536 trap_init();
537 mm_init();
538
539 /*
540 * Set up the scheduler prior starting any interrupts (such as the
541 * timer interrupt). Full topology setup happens at smp_init()
542 * time - but meanwhile we still have a functioning scheduler.
543 */
544 sched_init();
545 /*
546 * Disable preemption - early bootup scheduling is extremely
547 * fragile until we cpu_idle() for the first time.
548 */
549 preempt_disable();
550 if (WARN(!irqs_disabled(), "Interrupts were enabled *very* early, fixing it\n"))
551 local_irq_disable();
552 idr_init_cache();
553 rcu_init();
554 tick_nohz_init();
555 context_tracking_init();
556 radix_tree_init();
557 /* init some links before init_ISA_irqs() */
558 early_irq_init();
559 init_IRQ();
560 tick_init();
561 init_timers();
562 hrtimers_init();
563 softirq_init();
564 timekeeping_init();
565 time_init();
566 sched_clock_postinit();
567 perf_event_init();
568 profile_init();
569 call_function_init();
570 WARN(!irqs_disabled(), "Interrupts were enabled early\n");
571 early_boot_irqs_disabled = false;
572 local_irq_enable();
573
574 kmem_cache_init_late();
575
576 /*
577 * HACK ALERT! This is early. We're enabling the console before
578 * we've done PCI setups etc, and console_init() must be aware of
579 * this. But we do want output early, in case something goes wrong.
580 */
581 console_init();
582 if (panic_later)
583 panic("Too many boot %s vars at `%s'", panic_later,
584 panic_param);
585
586 lockdep_info();
587
588 /*
589 * Need to run this when irqs are enabled, because it wants
590 * to self-test [hard/soft]-irqs on/off lock inversion bugs
591 * too:
592 */
593 locking_selftest();
594
595#ifdef CONFIG_BLK_DEV_INITRD
596 if (initrd_start && !initrd_below_start_ok &&
597 page_to_pfn(virt_to_page((void *)initrd_start)) < min_low_pfn) {
598 pr_crit("initrd overwritten (0x%08lx < 0x%08lx) - disabling it.\n",
599 page_to_pfn(virt_to_page((void *)initrd_start)),
600 min_low_pfn);
601 initrd_start = 0;
602 }
603#endif
604 page_cgroup_init();
605 debug_objects_mem_init();
606 kmemleak_init();
607 setup_per_cpu_pageset();
608 numa_policy_init();
609 if (late_time_init)
610 late_time_init();
611 sched_clock_init();
612 calibrate_delay();
613 pidmap_init();
614 anon_vma_init();
615 acpi_early_init();
616#ifdef CONFIG_X86
617 if (efi_enabled(EFI_RUNTIME_SERVICES))
618 efi_enter_virtual_mode();
619#endif
620 thread_info_cache_init();
621 cred_init();
622 fork_init(totalram_pages);
623 proc_caches_init();
624 buffer_init();
625 key_init();
626 security_init();
627 dbg_late_init();
628 vfs_caches_init(totalram_pages);
629 signals_init();
630 /* rootfs populating might need page-writeback */
631 page_writeback_init();
632#ifdef CONFIG_PROC_FS
633 proc_root_init();
634#endif
635 cgroup_init();
636 cpuset_init();
637 taskstats_init_early();
638 delayacct_init();
639
640 check_bugs();
641
642 sfi_init_late();
643
644 if (efi_enabled(EFI_RUNTIME_SERVICES)) {
645 efi_late_init();
646 efi_free_boot_services();
647 }
648
649 ftrace_init();
650
651 /* Do the rest non-__init'ed, we're now alive */
652 rest_init();
653}
654
655/* Call all constructor functions linked into the kernel. */
656static void __init do_ctors(void)
657{
658#ifdef CONFIG_CONSTRUCTORS
659 ctor_fn_t *fn = (ctor_fn_t *) __ctors_start;
660
661 for (; fn < (ctor_fn_t *) __ctors_end; fn++)
662 (*fn)();
663#endif
664}
665
666bool initcall_debug;
667core_param(initcall_debug, initcall_debug, bool, 0644);
668
669static int __init_or_module do_one_initcall_debug(initcall_t fn)
670{
671 ktime_t calltime, delta, rettime;
672 unsigned long long duration;
673 int ret;
674
675 pr_debug("calling %pF @ %i\n", fn, task_pid_nr(current));
676 calltime = ktime_get();
677 ret = fn();
678 rettime = ktime_get();
679 delta = ktime_sub(rettime, calltime);
680 duration = (unsigned long long) ktime_to_ns(delta) >> 10;
681 pr_debug("initcall %pF returned %d after %lld usecs\n",
682 fn, ret, duration);
683
684 return ret;
685}
686
687int __init_or_module do_one_initcall(initcall_t fn)
688{
689 int count = preempt_count();
690 int ret;
691 char msgbuf[64];
692
693 if (initcall_debug)
694 ret = do_one_initcall_debug(fn);
695 else
696 ret = fn();
697
698 msgbuf[0] = 0;
699
700 if (preempt_count() != count) {
701 sprintf(msgbuf, "preemption imbalance ");
702 preempt_count_set(count);
703 }
704 if (irqs_disabled()) {
705 strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf));
706 local_irq_enable();
707 }
708 WARN(msgbuf[0], "initcall %pF returned with %s\n", fn, msgbuf);
709
710 return ret;
711}
712
713
714extern initcall_t __initcall_start[];
715extern initcall_t __initcall0_start[];
716extern initcall_t __initcall1_start[];
717extern initcall_t __initcall2_start[];
718extern initcall_t __initcall3_start[];
719extern initcall_t __initcall4_start[];
720extern initcall_t __initcall5_start[];
721extern initcall_t __initcall6_start[];
722extern initcall_t __initcall7_start[];
723extern initcall_t __initcall_end[];
724
725static initcall_t *initcall_levels[] __initdata = {
726 __initcall0_start,
727 __initcall1_start,
728 __initcall2_start,
729 __initcall3_start,
730 __initcall4_start,
731 __initcall5_start,
732 __initcall6_start,
733 __initcall7_start,
734 __initcall_end,
735};
736
737/* Keep these in sync with initcalls in include/linux/init.h */
738static char *initcall_level_names[] __initdata = {
739 "early",
740 "core",
741 "postcore",
742 "arch",
743 "subsys",
744 "fs",
745 "device",
746 "late",
747};
748
749static void __init do_initcall_level(int level)
750{
751 extern const struct kernel_param __start___param[], __stop___param[];
752 initcall_t *fn;
753
754 strcpy(initcall_command_line, saved_command_line);
755 parse_args(initcall_level_names[level],
756 initcall_command_line, __start___param,
757 __stop___param - __start___param,
758 level, level,
759 &repair_env_string);
760
761 for (fn = initcall_levels[level]; fn < initcall_levels[level+1]; fn++)
762 do_one_initcall(*fn);
763}
764
765static void __init do_initcalls(void)
766{
767 int level;
768
769 for (level = 0; level < ARRAY_SIZE(initcall_levels) - 1; level++)
770 do_initcall_level(level);
771}
772
773/*
774 * Ok, the machine is now initialized. None of the devices
775 * have been touched yet, but the CPU subsystem is up and
776 * running, and memory and process management works.
777 *
778 * Now we can finally start doing some real work..
779 */
780static void __init do_basic_setup(void)
781{
782 cpuset_init_smp();
783 usermodehelper_init();
784 shmem_init();
785 driver_init();
786 init_irq_proc();
787 do_ctors();
788 usermodehelper_enable();
789 do_initcalls();
790 random_int_secret_init();
791}
792
793static void __init do_pre_smp_initcalls(void)
794{
795 initcall_t *fn;
796
797 for (fn = __initcall_start; fn < __initcall0_start; fn++)
798 do_one_initcall(*fn);
799}
800
801/*
802 * This function requests modules which should be loaded by default and is
803 * called twice right after initrd is mounted and right before init is
804 * exec'd. If such modules are on either initrd or rootfs, they will be
805 * loaded before control is passed to userland.
806 */
807void __init load_default_modules(void)
808{
809 load_default_elevator_module();
810}
811
812static int run_init_process(const char *init_filename)
813{
814 argv_init[0] = init_filename;
815 return do_execve(getname_kernel(init_filename),
816 (const char __user *const __user *)argv_init,
817 (const char __user *const __user *)envp_init);
818}
819
820static int try_to_run_init_process(const char *init_filename)
821{
822 int ret;
823
824 ret = run_init_process(init_filename);
825
826 if (ret && ret != -ENOENT) {
827 pr_err("Starting init: %s exists but couldn't execute it (error %d)\n",
828 init_filename, ret);
829 }
830
831 return ret;
832}
833
834static noinline void __init kernel_init_freeable(void);
835
836static int __ref kernel_init(void *unused)
837{
838 int ret;
839
840 kernel_init_freeable();
841 /* need to finish all async __init code before freeing the memory */
842 async_synchronize_full();
843 free_initmem();
844 mark_rodata_ro();
845 system_state = SYSTEM_RUNNING;
846 numa_default_policy();
847
848 flush_delayed_fput();
849
850 if (ramdisk_execute_command) {
851 ret = run_init_process(ramdisk_execute_command);
852 if (!ret)
853 return 0;
854 pr_err("Failed to execute %s (error %d)\n",
855 ramdisk_execute_command, ret);
856 }
857
858 /*
859 * We try each of these until one succeeds.
860 *
861 * The Bourne shell can be used instead of init if we are
862 * trying to recover a really broken machine.
863 */
864 if (execute_command) {
865 ret = run_init_process(execute_command);
866 if (!ret)
867 return 0;
868 pr_err("Failed to execute %s (error %d). Attempting defaults...\n",
869 execute_command, ret);
870 }
871 if (!try_to_run_init_process("/sbin/init") ||
872 !try_to_run_init_process("/etc/init") ||
873 !try_to_run_init_process("/bin/init") ||
874 !try_to_run_init_process("/bin/sh"))
875 return 0;
876
877 panic("No working init found. Try passing init= option to kernel. "
878 "See Linux Documentation/init.txt for guidance.");
879}
880
881static noinline void __init kernel_init_freeable(void)
882{
883 /*
884 * Wait until kthreadd is all set-up.
885 */
886 wait_for_completion(&kthreadd_done);
887
888 /* Now the scheduler is fully set up and can do blocking allocations */
889 gfp_allowed_mask = __GFP_BITS_MASK;
890
891 /*
892 * init can allocate pages on any node
893 */
894 set_mems_allowed(node_states[N_MEMORY]);
895 /*
896 * init can run on any cpu.
897 */
898 set_cpus_allowed_ptr(current, cpu_all_mask);
899
900 cad_pid = task_pid(current);
901
902 smp_prepare_cpus(setup_max_cpus);
903
904 do_pre_smp_initcalls();
905 lockup_detector_init();
906
907 smp_init();
908 sched_init_smp();
909
910 do_basic_setup();
911
912 /* Open the /dev/console on the rootfs, this should never fail */
913 if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0)
914 pr_err("Warning: unable to open an initial console.\n");
915
916 (void) sys_dup(0);
917 (void) sys_dup(0);
918 /*
919 * check if there is an early userspace init. If yes, let it do all
920 * the work
921 */
922
923 if (!ramdisk_execute_command)
924 ramdisk_execute_command = "/init";
925
926 if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) {
927 ramdisk_execute_command = NULL;
928 prepare_namespace();
929 }
930
931 /*
932 * Ok, we have completed the initial bootup, and
933 * we're essentially up and running. Get rid of the
934 * initmem segments and start the user-mode stuff..
935 */
936
937 /* rootfs is available now, try loading default modules */
938 load_default_modules();
939}