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_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#include <linux/cache.h>
85
86#include <asm/io.h>
87#include <asm/bugs.h>
88#include <asm/setup.h>
89#include <asm/sections.h>
90#include <asm/cacheflush.h>
91
92static int kernel_init(void *);
93
94extern void init_IRQ(void);
95extern void fork_init(void);
96extern void radix_tree_init(void);
97
98/*
99 * Debug helper: via this flag we know that we are in 'early bootup code'
100 * where only the boot processor is running with IRQ disabled. This means
101 * two things - IRQ must not be enabled before the flag is cleared and some
102 * operations which are not allowed with IRQ disabled are allowed while the
103 * flag is set.
104 */
105bool early_boot_irqs_disabled __read_mostly;
106
107enum system_states system_state __read_mostly;
108EXPORT_SYMBOL(system_state);
109
110/*
111 * Boot command-line arguments
112 */
113#define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT
114#define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT
115
116extern void time_init(void);
117/* Default late time init is NULL. archs can override this later. */
118void (*__initdata late_time_init)(void);
119
120/* Untouched command line saved by arch-specific code. */
121char __initdata boot_command_line[COMMAND_LINE_SIZE];
122/* Untouched saved command line (eg. for /proc) */
123char *saved_command_line;
124/* Command line for parameter parsing */
125static char *static_command_line;
126/* Command line for per-initcall parameter parsing */
127static char *initcall_command_line;
128
129static char *execute_command;
130static char *ramdisk_execute_command;
131
132/*
133 * Used to generate warnings if static_key manipulation functions are used
134 * before jump_label_init is called.
135 */
136bool static_key_initialized __read_mostly;
137EXPORT_SYMBOL_GPL(static_key_initialized);
138
139/*
140 * If set, this is an indication to the drivers that reset the underlying
141 * device before going ahead with the initialization otherwise driver might
142 * rely on the BIOS and skip the reset operation.
143 *
144 * This is useful if kernel is booting in an unreliable environment.
145 * For ex. kdump situation where previous kernel has crashed, BIOS has been
146 * skipped and devices will be in unknown state.
147 */
148unsigned int reset_devices;
149EXPORT_SYMBOL(reset_devices);
150
151static int __init set_reset_devices(char *str)
152{
153 reset_devices = 1;
154 return 1;
155}
156
157__setup("reset_devices", set_reset_devices);
158
159static const char *argv_init[MAX_INIT_ARGS+2] = { "init", NULL, };
160const char *envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, };
161static const char *panic_later, *panic_param;
162
163extern const struct obs_kernel_param __setup_start[], __setup_end[];
164
165static bool __init obsolete_checksetup(char *line)
166{
167 const struct obs_kernel_param *p;
168 bool had_early_param = false;
169
170 p = __setup_start;
171 do {
172 int n = strlen(p->str);
173 if (parameqn(line, p->str, n)) {
174 if (p->early) {
175 /* Already done in parse_early_param?
176 * (Needs exact match on param part).
177 * Keep iterating, as we can have early
178 * params and __setups of same names 8( */
179 if (line[n] == '\0' || line[n] == '=')
180 had_early_param = true;
181 } else if (!p->setup_func) {
182 pr_warn("Parameter %s is obsolete, ignored\n",
183 p->str);
184 return true;
185 } else if (p->setup_func(line + n))
186 return true;
187 }
188 p++;
189 } while (p < __setup_end);
190
191 return had_early_param;
192}
193
194/*
195 * This should be approx 2 Bo*oMips to start (note initial shift), and will
196 * still work even if initially too large, it will just take slightly longer
197 */
198unsigned long loops_per_jiffy = (1<<12);
199EXPORT_SYMBOL(loops_per_jiffy);
200
201static int __init debug_kernel(char *str)
202{
203 console_loglevel = CONSOLE_LOGLEVEL_DEBUG;
204 return 0;
205}
206
207static int __init quiet_kernel(char *str)
208{
209 console_loglevel = CONSOLE_LOGLEVEL_QUIET;
210 return 0;
211}
212
213early_param("debug", debug_kernel);
214early_param("quiet", quiet_kernel);
215
216static int __init loglevel(char *str)
217{
218 int newlevel;
219
220 /*
221 * Only update loglevel value when a correct setting was passed,
222 * to prevent blind crashes (when loglevel being set to 0) that
223 * are quite hard to debug
224 */
225 if (get_option(&str, &newlevel)) {
226 console_loglevel = newlevel;
227 return 0;
228 }
229
230 return -EINVAL;
231}
232
233early_param("loglevel", loglevel);
234
235/* Change NUL term back to "=", to make "param" the whole string. */
236static int __init repair_env_string(char *param, char *val,
237 const char *unused, void *arg)
238{
239 if (val) {
240 /* param=val or param="val"? */
241 if (val == param+strlen(param)+1)
242 val[-1] = '=';
243 else if (val == param+strlen(param)+2) {
244 val[-2] = '=';
245 memmove(val-1, val, strlen(val)+1);
246 val--;
247 } else
248 BUG();
249 }
250 return 0;
251}
252
253/* Anything after -- gets handed straight to init. */
254static int __init set_init_arg(char *param, char *val,
255 const char *unused, void *arg)
256{
257 unsigned int i;
258
259 if (panic_later)
260 return 0;
261
262 repair_env_string(param, val, unused, NULL);
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,
280 const char *unused, void *arg)
281{
282 repair_env_string(param, val, unused, NULL);
283
284 /* Handle obsolete-style parameters */
285 if (obsolete_checksetup(param))
286 return 0;
287
288 /* Unused module parameter. */
289 if (strchr(param, '.') && (!val || strchr(param, '.') < val))
290 return 0;
291
292 if (panic_later)
293 return 0;
294
295 if (val) {
296 /* Environment option */
297 unsigned int i;
298 for (i = 0; envp_init[i]; i++) {
299 if (i == MAX_INIT_ENVS) {
300 panic_later = "env";
301 panic_param = param;
302 }
303 if (!strncmp(param, envp_init[i], val - param))
304 break;
305 }
306 envp_init[i] = param;
307 } else {
308 /* Command line option */
309 unsigned int i;
310 for (i = 0; argv_init[i]; i++) {
311 if (i == MAX_INIT_ARGS) {
312 panic_later = "init";
313 panic_param = param;
314 }
315 }
316 argv_init[i] = param;
317 }
318 return 0;
319}
320
321static int __init init_setup(char *str)
322{
323 unsigned int i;
324
325 execute_command = str;
326 /*
327 * In case LILO is going to boot us with default command line,
328 * it prepends "auto" before the whole cmdline which makes
329 * the shell think it should execute a script with such name.
330 * So we ignore all arguments entered _before_ init=... [MJ]
331 */
332 for (i = 1; i < MAX_INIT_ARGS; i++)
333 argv_init[i] = NULL;
334 return 1;
335}
336__setup("init=", init_setup);
337
338static int __init rdinit_setup(char *str)
339{
340 unsigned int i;
341
342 ramdisk_execute_command = str;
343 /* See "auto" comment in init_setup */
344 for (i = 1; i < MAX_INIT_ARGS; i++)
345 argv_init[i] = NULL;
346 return 1;
347}
348__setup("rdinit=", rdinit_setup);
349
350#ifndef CONFIG_SMP
351static const unsigned int setup_max_cpus = NR_CPUS;
352static inline void setup_nr_cpu_ids(void) { }
353static inline void smp_prepare_cpus(unsigned int maxcpus) { }
354#endif
355
356/*
357 * We need to store the untouched command line for future reference.
358 * We also need to store the touched command line since the parameter
359 * parsing is performed in place, and we should allow a component to
360 * store reference of name/value for future reference.
361 */
362static void __init setup_command_line(char *command_line)
363{
364 saved_command_line =
365 memblock_virt_alloc(strlen(boot_command_line) + 1, 0);
366 initcall_command_line =
367 memblock_virt_alloc(strlen(boot_command_line) + 1, 0);
368 static_command_line = memblock_virt_alloc(strlen(command_line) + 1, 0);
369 strcpy(saved_command_line, boot_command_line);
370 strcpy(static_command_line, command_line);
371}
372
373/*
374 * We need to finalize in a non-__init function or else race conditions
375 * between the root thread and the init thread may cause start_kernel to
376 * be reaped by free_initmem before the root thread has proceeded to
377 * cpu_idle.
378 *
379 * gcc-3.4 accidentally inlines this function, so use noinline.
380 */
381
382static __initdata DECLARE_COMPLETION(kthreadd_done);
383
384static noinline void __ref rest_init(void)
385{
386 int pid;
387
388 rcu_scheduler_starting();
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,
414 const char *unused, void *arg)
415{
416 const struct obs_kernel_param *p;
417
418 for (p = __setup_start; p < __setup_end; p++) {
419 if ((p->early && parameq(param, p->str)) ||
420 (strcmp(param, "console") == 0 &&
421 strcmp(p->str, "earlycon") == 0)
422 ) {
423 if (p->setup_func(val) != 0)
424 pr_warn("Malformed early option '%s'\n", param);
425 }
426 }
427 /* We accept everything at this stage. */
428 return 0;
429}
430
431void __init parse_early_options(char *cmdline)
432{
433 parse_args("early options", cmdline, NULL, 0, 0, 0, NULL,
434 do_early_param);
435}
436
437/* Arch code calls this early on, or if not, just before other parsing. */
438void __init parse_early_param(void)
439{
440 static int done __initdata;
441 static char tmp_cmdline[COMMAND_LINE_SIZE] __initdata;
442
443 if (done)
444 return;
445
446 /* All fall through to do_early_param. */
447 strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE);
448 parse_early_options(tmp_cmdline);
449 done = 1;
450}
451
452void __init __weak arch_post_acpi_subsys_init(void) { }
453
454void __init __weak smp_setup_processor_id(void)
455{
456}
457
458# if THREAD_SIZE >= PAGE_SIZE
459void __init __weak thread_stack_cache_init(void)
460{
461}
462#endif
463
464/*
465 * Set up kernel memory allocators
466 */
467static void __init mm_init(void)
468{
469 /*
470 * page_ext requires contiguous pages,
471 * bigger than MAX_ORDER unless SPARSEMEM.
472 */
473 page_ext_init_flatmem();
474 mem_init();
475 kmem_cache_init();
476 percpu_init_late();
477 pgtable_init();
478 vmalloc_init();
479 ioremap_huge_init();
480}
481
482asmlinkage __visible void __init start_kernel(void)
483{
484 char *command_line;
485 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 * Set up the the initial canary ASAP:
493 */
494 boot_init_stack_canary();
495
496 cgroup_init_early();
497
498 local_irq_disable();
499 early_boot_irqs_disabled = true;
500
501/*
502 * Interrupts are still disabled. Do necessary setups, then
503 * enable them
504 */
505 boot_cpu_init();
506 page_address_init();
507 pr_notice("%s", linux_banner);
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 */
515
516 build_all_zonelists(NULL, NULL);
517 page_alloc_init();
518
519 pr_notice("Kernel command line: %s\n", boot_command_line);
520 parse_early_param();
521 after_dashes = parse_args("Booting kernel",
522 static_command_line, __start___param,
523 __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 jump_label_init();
530
531 /*
532 * These use large bootmem allocations and must precede
533 * kmem_cache_init()
534 */
535 setup_log_buf(0);
536 pidhash_init();
537 vfs_caches_init_early();
538 sort_main_extable();
539 trap_init();
540 mm_init();
541
542 /*
543 * Set up the scheduler prior starting any interrupts (such as the
544 * timer interrupt). Full topology setup happens at smp_init()
545 * time - but meanwhile we still have a functioning scheduler.
546 */
547 sched_init();
548 /*
549 * Disable preemption - early bootup scheduling is extremely
550 * fragile until we cpu_idle() for the first time.
551 */
552 preempt_disable();
553 if (WARN(!irqs_disabled(),
554 "Interrupts were enabled *very* early, fixing it\n"))
555 local_irq_disable();
556 idr_init_cache();
557
558 /*
559 * Allow workqueue creation and work item queueing/cancelling
560 * early. Work item execution depends on kthreads and starts after
561 * workqueue_init().
562 */
563 workqueue_init_early();
564
565 rcu_init();
566
567 /* trace_printk() and trace points may be used after this */
568 trace_init();
569
570 context_tracking_init();
571 radix_tree_init();
572 /* init some links before init_ISA_irqs() */
573 early_irq_init();
574 init_IRQ();
575 tick_init();
576 rcu_init_nohz();
577 init_timers();
578 hrtimers_init();
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();
592
593 /*
594 * HACK ALERT! This is early. We're enabling the console before
595 * we've done PCI setups etc, and console_init() must be aware of
596 * this. But we do want output early, in case something goes wrong.
597 */
598 console_init();
599 if (panic_later)
600 panic("Too many boot %s vars at `%s'", panic_later,
601 panic_param);
602
603 lockdep_info();
604
605 /*
606 * Need to run this when irqs are enabled, because it wants
607 * to self-test [hard/soft]-irqs on/off lock inversion bugs
608 * too:
609 */
610 locking_selftest();
611
612#ifdef CONFIG_BLK_DEV_INITRD
613 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 page_to_pfn(virt_to_page((void *)initrd_start)),
617 min_low_pfn);
618 initrd_start = 0;
619 }
620#endif
621 page_ext_init();
622 debug_objects_mem_init();
623 kmemleak_init();
624 setup_per_cpu_pageset();
625 numa_policy_init();
626 if (late_time_init)
627 late_time_init();
628 sched_clock_init();
629 calibrate_delay();
630 pidmap_init();
631 anon_vma_init();
632 acpi_early_init();
633#ifdef CONFIG_X86
634 if (efi_enabled(EFI_RUNTIME_SERVICES))
635 efi_enter_virtual_mode();
636#endif
637#ifdef CONFIG_X86_ESPFIX64
638 /* Should be run before the first non-init thread is created */
639 init_espfix_bsp();
640#endif
641 thread_stack_cache_init();
642 cred_init();
643 fork_init();
644 proc_caches_init();
645 buffer_init();
646 key_init();
647 security_init();
648 dbg_late_init();
649 vfs_caches_init();
650 pagecache_init();
651 signals_init();
652 proc_root_init();
653 nsfs_init();
654 cpuset_init();
655 cgroup_init();
656 taskstats_init_early();
657 delayacct_init();
658
659 check_bugs();
660
661 acpi_subsystem_init();
662 arch_post_acpi_subsys_init();
663 sfi_init_late();
664
665 if (efi_enabled(EFI_RUNTIME_SERVICES)) {
666 efi_late_init();
667 efi_free_boot_services();
668 }
669
670 ftrace_init();
671
672 /* Do the rest non-__init'ed, we're now alive */
673 rest_init();
674}
675
676/* Call all constructor functions linked into the kernel. */
677static void __init do_ctors(void)
678{
679#ifdef CONFIG_CONSTRUCTORS
680 ctor_fn_t *fn = (ctor_fn_t *) __ctors_start;
681
682 for (; fn < (ctor_fn_t *) __ctors_end; fn++)
683 (*fn)();
684#endif
685}
686
687bool initcall_debug;
688core_param(initcall_debug, initcall_debug, bool, 0644);
689
690#ifdef CONFIG_KALLSYMS
691struct blacklist_entry {
692 struct list_head next;
693 char *buf;
694};
695
696static __initdata_or_module LIST_HEAD(blacklisted_initcalls);
697
698static int __init initcall_blacklist(char *str)
699{
700 char *str_entry;
701 struct blacklist_entry *entry;
702
703 /* str argument is a comma-separated list of functions */
704 do {
705 str_entry = strsep(&str, ",");
706 if (str_entry) {
707 pr_debug("blacklisting initcall %s\n", str_entry);
708 entry = alloc_bootmem(sizeof(*entry));
709 entry->buf = alloc_bootmem(strlen(str_entry) + 1);
710 strcpy(entry->buf, str_entry);
711 list_add(&entry->next, &blacklisted_initcalls);
712 }
713 } while (str_entry);
714
715 return 0;
716}
717
718static bool __init_or_module initcall_blacklisted(initcall_t fn)
719{
720 struct blacklist_entry *entry;
721 char fn_name[KSYM_SYMBOL_LEN];
722 unsigned long addr;
723
724 if (list_empty(&blacklisted_initcalls))
725 return false;
726
727 addr = (unsigned long) dereference_function_descriptor(fn);
728 sprint_symbol_no_offset(fn_name, addr);
729
730 /*
731 * fn will be "function_name [module_name]" where [module_name] is not
732 * displayed for built-in init functions. Strip off the [module_name].
733 */
734 strreplace(fn_name, ' ', '\0');
735
736 list_for_each_entry(entry, &blacklisted_initcalls, next) {
737 if (!strcmp(fn_name, entry->buf)) {
738 pr_debug("initcall %s blacklisted\n", fn_name);
739 return true;
740 }
741 }
742
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 add_latent_entropy();
804 return ret;
805}
806
807
808extern initcall_t __initcall_start[];
809extern initcall_t __initcall0_start[];
810extern initcall_t __initcall1_start[];
811extern initcall_t __initcall2_start[];
812extern initcall_t __initcall3_start[];
813extern initcall_t __initcall4_start[];
814extern initcall_t __initcall5_start[];
815extern initcall_t __initcall6_start[];
816extern initcall_t __initcall7_start[];
817extern initcall_t __initcall_end[];
818
819static initcall_t *initcall_levels[] __initdata = {
820 __initcall0_start,
821 __initcall1_start,
822 __initcall2_start,
823 __initcall3_start,
824 __initcall4_start,
825 __initcall5_start,
826 __initcall6_start,
827 __initcall7_start,
828 __initcall_end,
829};
830
831/* Keep these in sync with initcalls in include/linux/init.h */
832static char *initcall_level_names[] __initdata = {
833 "early",
834 "core",
835 "postcore",
836 "arch",
837 "subsys",
838 "fs",
839 "device",
840 "late",
841};
842
843static void __init do_initcall_level(int level)
844{
845 initcall_t *fn;
846
847 strcpy(initcall_command_line, saved_command_line);
848 parse_args(initcall_level_names[level],
849 initcall_command_line, __start___param,
850 __stop___param - __start___param,
851 level, level,
852 NULL, &repair_env_string);
853
854 for (fn = initcall_levels[level]; fn < initcall_levels[level+1]; fn++)
855 do_one_initcall(*fn);
856}
857
858static void __init do_initcalls(void)
859{
860 int level;
861
862 for (level = 0; level < ARRAY_SIZE(initcall_levels) - 1; level++)
863 do_initcall_level(level);
864}
865
866/*
867 * Ok, the machine is now initialized. None of the devices
868 * have been touched yet, but the CPU subsystem is up and
869 * running, and memory and process management works.
870 *
871 * Now we can finally start doing some real work..
872 */
873static void __init do_basic_setup(void)
874{
875 cpuset_init_smp();
876 shmem_init();
877 driver_init();
878 init_irq_proc();
879 do_ctors();
880 usermodehelper_enable();
881 do_initcalls();
882}
883
884static 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 * This function requests modules which should be loaded by default and is
894 * called twice right after initrd is mounted and right before init is
895 * exec'd. If such modules are on either initrd or rootfs, they will be
896 * loaded before control is passed to userland.
897 */
898void __init load_default_modules(void)
899{
900 load_default_elevator_module();
901}
902
903static int run_init_process(const char *init_filename)
904{
905 argv_init[0] = init_filename;
906 return do_execve(getname_kernel(init_filename),
907 (const char __user *const __user *)argv_init,
908 (const char __user *const __user *)envp_init);
909}
910
911static int try_to_run_init_process(const char *init_filename)
912{
913 int ret;
914
915 ret = run_init_process(init_filename);
916
917 if (ret && ret != -ENOENT) {
918 pr_err("Starting init: %s exists but couldn't execute it (error %d)\n",
919 init_filename, ret);
920 }
921
922 return ret;
923}
924
925static noinline void __init kernel_init_freeable(void);
926
927#if defined(CONFIG_DEBUG_RODATA) || defined(CONFIG_DEBUG_SET_MODULE_RONX)
928bool rodata_enabled __ro_after_init = true;
929static int __init set_debug_rodata(char *str)
930{
931 return strtobool(str, &rodata_enabled);
932}
933__setup("rodata=", set_debug_rodata);
934#endif
935
936#ifdef CONFIG_DEBUG_RODATA
937static void mark_readonly(void)
938{
939 if (rodata_enabled)
940 mark_rodata_ro();
941 else
942 pr_info("Kernel memory protection disabled.\n");
943}
944#else
945static inline void mark_readonly(void)
946{
947 pr_warn("This architecture does not have kernel memory protection.\n");
948}
949#endif
950
951static int __ref kernel_init(void *unused)
952{
953 int ret;
954
955 kernel_init_freeable();
956 /* need to finish all async __init code before freeing the memory */
957 async_synchronize_full();
958 free_initmem();
959 mark_readonly();
960 system_state = SYSTEM_RUNNING;
961 numa_default_policy();
962
963 flush_delayed_fput();
964
965 rcu_end_inkernel_boot();
966
967 if (ramdisk_execute_command) {
968 ret = run_init_process(ramdisk_execute_command);
969 if (!ret)
970 return 0;
971 pr_err("Failed to execute %s (error %d)\n",
972 ramdisk_execute_command, ret);
973 }
974
975 /*
976 * We try each of these until one succeeds.
977 *
978 * The Bourne shell can be used instead of init if we are
979 * trying to recover a really broken machine.
980 */
981 if (execute_command) {
982 ret = run_init_process(execute_command);
983 if (!ret)
984 return 0;
985 panic("Requested init %s failed (error %d).",
986 execute_command, ret);
987 }
988 if (!try_to_run_init_process("/sbin/init") ||
989 !try_to_run_init_process("/etc/init") ||
990 !try_to_run_init_process("/bin/init") ||
991 !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
998static noinline void __init kernel_init_freeable(void)
999{
1000 /*
1001 * Wait until kthreadd is all set-up.
1002 */
1003 wait_for_completion(&kthreadd_done);
1004
1005 /* Now the scheduler is fully set up and can do blocking allocations */
1006 gfp_allowed_mask = __GFP_BITS_MASK;
1007
1008 /*
1009 * init can allocate pages on any node
1010 */
1011 set_mems_allowed(node_states[N_MEMORY]);
1012 /*
1013 * init can run on any cpu.
1014 */
1015 set_cpus_allowed_ptr(current, cpu_all_mask);
1016
1017 cad_pid = task_pid(current);
1018
1019 smp_prepare_cpus(setup_max_cpus);
1020
1021 workqueue_init();
1022
1023 do_pre_smp_initcalls();
1024 lockup_detector_init();
1025
1026 smp_init();
1027 sched_init_smp();
1028
1029 page_alloc_init_late();
1030
1031 do_basic_setup();
1032
1033 /* Open the /dev/console on the rootfs, this should never fail */
1034 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 (void) sys_dup(0);
1038 (void) sys_dup(0);
1039 /*
1040 * check if there is an early userspace init. If yes, let it do all
1041 * the work
1042 */
1043
1044 if (!ramdisk_execute_command)
1045 ramdisk_execute_command = "/init";
1046
1047 if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) {
1048 ramdisk_execute_command = NULL;
1049 prepare_namespace();
1050 }
1051
1052 /*
1053 * Ok, we have completed the initial bootup, and
1054 * we're essentially up and running. Get rid of the
1055 * initmem segments and start the user-mode stuff..
1056 *
1057 * rootfs is available now, try loading the public keys
1058 * and default modules
1059 */
1060
1061 integrity_load_keys();
1062 load_default_modules();
1063}