]> bbs.cooldavid.org Git - net-next-2.6.git/blame - init/main.c
proc: mounts_poll() make consistent to mdstat_poll
[net-next-2.6.git] / init / main.c
CommitLineData
1da177e4
LT
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
1da177e4
LT
12#include <linux/types.h>
13#include <linux/module.h>
14#include <linux/proc_fs.h>
1da177e4
LT
15#include <linux/kernel.h>
16#include <linux/syscalls.h>
9b5609fd 17#include <linux/stackprotector.h>
1da177e4
LT
18#include <linux/string.h>
19#include <linux/ctype.h>
20#include <linux/delay.h>
21#include <linux/utsname.h>
22#include <linux/ioport.h>
23#include <linux/init.h>
24#include <linux/smp_lock.h>
25#include <linux/initrd.h>
1da177e4
LT
26#include <linux/bootmem.h>
27#include <linux/tty.h>
28#include <linux/gfp.h>
29#include <linux/percpu.h>
30#include <linux/kmod.h>
db64fe02 31#include <linux/vmalloc.h>
1da177e4 32#include <linux/kernel_stat.h>
d7cd5611 33#include <linux/start_kernel.h>
1da177e4 34#include <linux/security.h>
3d442233 35#include <linux/smp.h>
1da177e4
LT
36#include <linux/workqueue.h>
37#include <linux/profile.h>
38#include <linux/rcupdate.h>
39#include <linux/moduleparam.h>
40#include <linux/kallsyms.h>
41#include <linux/writeback.h>
42#include <linux/cpu.h>
43#include <linux/cpuset.h>
ddbcc7e8 44#include <linux/cgroup.h>
1da177e4 45#include <linux/efi.h>
906568c9 46#include <linux/tick.h>
6168a702 47#include <linux/interrupt.h>
c757249a 48#include <linux/taskstats_kern.h>
ca74e92b 49#include <linux/delayacct.h>
1da177e4
LT
50#include <linux/unistd.h>
51#include <linux/rmap.h>
52#include <linux/mempolicy.h>
53#include <linux/key.h>
b6cd0b77 54#include <linux/buffer_head.h>
94b6da5a 55#include <linux/page_cgroup.h>
9a11b49a 56#include <linux/debug_locks.h>
3ac7fe5a 57#include <linux/debugobjects.h>
fbb9ce95 58#include <linux/lockdep.h>
84d73786 59#include <linux/pid_namespace.h>
1f21782e 60#include <linux/device.h>
73c27992 61#include <linux/kthread.h>
e6fe6649 62#include <linux/sched.h>
a1c9eea9 63#include <linux/signal.h>
199f0ca5 64#include <linux/idr.h>
68bf21aa 65#include <linux/ftrace.h>
22a9d645 66#include <linux/async.h>
3f5ec136 67#include <trace/boot.h>
1da177e4
LT
68
69#include <asm/io.h>
70#include <asm/bugs.h>
71#include <asm/setup.h>
a940199f 72#include <asm/sections.h>
37b73c82 73#include <asm/cacheflush.h>
36994e58 74#include <trace/kmemtrace.h>
1da177e4 75
1da177e4
LT
76#ifdef CONFIG_X86_LOCAL_APIC
77#include <asm/smp.h>
78#endif
79
aae5f662 80static int kernel_init(void *);
1da177e4
LT
81
82extern void init_IRQ(void);
1da177e4
LT
83extern void fork_init(unsigned long);
84extern void mca_init(void);
85extern void sbus_init(void);
1da177e4
LT
86extern void prio_tree_init(void);
87extern void radix_tree_init(void);
88extern void free_initmem(void);
1da177e4
LT
89#ifdef CONFIG_ACPI
90extern void acpi_early_init(void);
91#else
92static inline void acpi_early_init(void) { }
93#endif
37b73c82
AV
94#ifndef CONFIG_DEBUG_RODATA
95static inline void mark_rodata_ro(void) { }
96#endif
1da177e4
LT
97
98#ifdef CONFIG_TC
99extern void tc_init(void);
100#endif
101
a6826048 102enum system_states system_state __read_mostly;
1da177e4
LT
103EXPORT_SYMBOL(system_state);
104
105/*
106 * Boot command-line arguments
107 */
108#define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT
109#define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT
110
111extern void time_init(void);
112/* Default late time init is NULL. archs can override this later. */
d2e3192b 113void (*__initdata late_time_init)(void);
1da177e4
LT
114extern void softirq_init(void);
115
30d7e0d4
ABL
116/* Untouched command line saved by arch-specific code. */
117char __initdata boot_command_line[COMMAND_LINE_SIZE];
118/* Untouched saved command line (eg. for /proc) */
119char *saved_command_line;
120/* Command line for parameter parsing */
121static char *static_command_line;
1da177e4
LT
122
123static char *execute_command;
ffdfc409 124static char *ramdisk_execute_command;
1da177e4 125
8b3b2955 126#ifdef CONFIG_SMP
1da177e4 127/* Setup configured maximum number of CPUs to activate */
ca74a6f8 128unsigned int __initdata setup_max_cpus = NR_CPUS;
7e96287d 129
1da177e4
LT
130/*
131 * Setup routine for controlling SMP activation
132 *
133 * Command-line option of "nosmp" or "maxcpus=0" will disable SMP
134 * activation entirely (the MPS table probe still happens, though).
135 *
136 * Command-line option of "maxcpus=<NUM>", where <NUM> is an integer
137 * greater than 0, limits the maximum number of CPUs activated in
138 * SMP mode to <NUM>.
139 */
65a4e574
IM
140
141void __weak arch_disable_smp_support(void) { }
61ec7567 142
1da177e4
LT
143static int __init nosmp(char *str)
144{
ca74a6f8 145 setup_max_cpus = 0;
65a4e574
IM
146 arch_disable_smp_support();
147
8b3b2955 148 return 0;
1da177e4
LT
149}
150
8b3b2955 151early_param("nosmp", nosmp);
1da177e4
LT
152
153static int __init maxcpus(char *str)
154{
ca74a6f8
AK
155 get_option(&str, &setup_max_cpus);
156 if (setup_max_cpus == 0)
65a4e574 157 arch_disable_smp_support();
61ec7567
LB
158
159 return 0;
1da177e4
LT
160}
161
81340977 162early_param("maxcpus", maxcpus);
8b3b2955 163#else
65a4e574 164const unsigned int setup_max_cpus = NR_CPUS;
8b3b2955
JB
165#endif
166
167/*
168 * If set, this is an indication to the drivers that reset the underlying
169 * device before going ahead with the initialization otherwise driver might
170 * rely on the BIOS and skip the reset operation.
171 *
172 * This is useful if kernel is booting in an unreliable environment.
173 * For ex. kdump situaiton where previous kernel has crashed, BIOS has been
174 * skipped and devices will be in unknown state.
175 */
176unsigned int reset_devices;
177EXPORT_SYMBOL(reset_devices);
1da177e4 178
7e96287d
VG
179static int __init set_reset_devices(char *str)
180{
181 reset_devices = 1;
182 return 1;
183}
184
185__setup("reset_devices", set_reset_devices);
186
1da177e4
LT
187static char * argv_init[MAX_INIT_ARGS+2] = { "init", NULL, };
188char * envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, };
189static const char *panic_later, *panic_param;
190
191extern struct obs_kernel_param __setup_start[], __setup_end[];
192
193static int __init obsolete_checksetup(char *line)
194{
195 struct obs_kernel_param *p;
33df0d19 196 int had_early_param = 0;
1da177e4
LT
197
198 p = __setup_start;
199 do {
200 int n = strlen(p->str);
201 if (!strncmp(line, p->str, n)) {
202 if (p->early) {
33df0d19
RR
203 /* Already done in parse_early_param?
204 * (Needs exact match on param part).
205 * Keep iterating, as we can have early
206 * params and __setups of same names 8( */
1da177e4 207 if (line[n] == '\0' || line[n] == '=')
33df0d19 208 had_early_param = 1;
1da177e4
LT
209 } else if (!p->setup_func) {
210 printk(KERN_WARNING "Parameter %s is obsolete,"
211 " ignored\n", p->str);
212 return 1;
213 } else if (p->setup_func(line + n))
214 return 1;
215 }
216 p++;
217 } while (p < __setup_end);
33df0d19
RR
218
219 return had_early_param;
1da177e4
LT
220}
221
222/*
223 * This should be approx 2 Bo*oMips to start (note initial shift), and will
224 * still work even if initially too large, it will just take slightly longer
225 */
226unsigned long loops_per_jiffy = (1<<12);
227
228EXPORT_SYMBOL(loops_per_jiffy);
229
230static int __init debug_kernel(char *str)
231{
1da177e4 232 console_loglevel = 10;
f6f21c81 233 return 0;
1da177e4
LT
234}
235
236static int __init quiet_kernel(char *str)
237{
1da177e4 238 console_loglevel = 4;
f6f21c81 239 return 0;
1da177e4
LT
240}
241
f6f21c81
YL
242early_param("debug", debug_kernel);
243early_param("quiet", quiet_kernel);
1da177e4
LT
244
245static int __init loglevel(char *str)
246{
247 get_option(&str, &console_loglevel);
d9d4fcfe 248 return 0;
1da177e4
LT
249}
250
f6f21c81 251early_param("loglevel", loglevel);
1da177e4
LT
252
253/*
254 * Unknown boot options get handed to init, unless they look like
255 * failed parameters
256 */
257static int __init unknown_bootoption(char *param, char *val)
258{
259 /* Change NUL term back to "=", to make "param" the whole string. */
260 if (val) {
261 /* param=val or param="val"? */
262 if (val == param+strlen(param)+1)
263 val[-1] = '=';
264 else if (val == param+strlen(param)+2) {
265 val[-2] = '=';
266 memmove(val-1, val, strlen(val)+1);
267 val--;
268 } else
269 BUG();
270 }
271
272 /* Handle obsolete-style parameters */
273 if (obsolete_checksetup(param))
274 return 0;
275
276 /*
211fee8a 277 * Preemptive maintenance for "why didn't my misspelled command
1da177e4
LT
278 * line work?"
279 */
280 if (strchr(param, '.') && (!val || strchr(param, '.') < val)) {
281 printk(KERN_ERR "Unknown boot option `%s': ignoring\n", param);
282 return 0;
283 }
284
285 if (panic_later)
286 return 0;
287
288 if (val) {
289 /* Environment option */
290 unsigned int i;
291 for (i = 0; envp_init[i]; i++) {
292 if (i == MAX_INIT_ENVS) {
293 panic_later = "Too many boot env vars at `%s'";
294 panic_param = param;
295 }
296 if (!strncmp(param, envp_init[i], val - param))
297 break;
298 }
299 envp_init[i] = param;
300 } else {
301 /* Command line option */
302 unsigned int i;
303 for (i = 0; argv_init[i]; i++) {
304 if (i == MAX_INIT_ARGS) {
305 panic_later = "Too many boot init vars at `%s'";
306 panic_param = param;
307 }
308 }
309 argv_init[i] = param;
310 }
311 return 0;
312}
313
12d6f21e
IM
314#ifdef CONFIG_DEBUG_PAGEALLOC
315int __read_mostly debug_pagealloc_enabled = 0;
316#endif
317
1da177e4
LT
318static int __init init_setup(char *str)
319{
320 unsigned int i;
321
322 execute_command = str;
323 /*
324 * In case LILO is going to boot us with default command line,
325 * it prepends "auto" before the whole cmdline which makes
326 * the shell think it should execute a script with such name.
327 * So we ignore all arguments entered _before_ init=... [MJ]
328 */
329 for (i = 1; i < MAX_INIT_ARGS; i++)
330 argv_init[i] = NULL;
331 return 1;
332}
333__setup("init=", init_setup);
334
ffdfc409
OJ
335static int __init rdinit_setup(char *str)
336{
337 unsigned int i;
338
339 ramdisk_execute_command = str;
340 /* See "auto" comment in init_setup */
341 for (i = 1; i < MAX_INIT_ARGS; i++)
342 argv_init[i] = NULL;
343 return 1;
344}
345__setup("rdinit=", rdinit_setup);
346
1da177e4
LT
347#ifndef CONFIG_SMP
348
349#ifdef CONFIG_X86_LOCAL_APIC
350static void __init smp_init(void)
351{
352 APIC_init_uniprocessor();
353}
354#else
355#define smp_init() do { } while (0)
356#endif
357
358static inline void setup_per_cpu_areas(void) { }
e0982e90 359static inline void setup_nr_cpu_ids(void) { }
1da177e4
LT
360static inline void smp_prepare_cpus(unsigned int maxcpus) { }
361
362#else
363
321a8e9d
MT
364#if NR_CPUS > BITS_PER_LONG
365cpumask_t cpu_mask_all __read_mostly = CPU_MASK_ALL;
366EXPORT_SYMBOL(cpu_mask_all);
367#endif
368
e0982e90
MT
369/* Setup number of possible processor ids */
370int nr_cpu_ids __read_mostly = NR_CPUS;
371EXPORT_SYMBOL(nr_cpu_ids);
372
373/* An arch may set nr_cpu_ids earlier if needed, so this would be redundant */
374static void __init setup_nr_cpu_ids(void)
375{
e0c0ba73 376 nr_cpu_ids = find_last_bit(cpumask_bits(cpu_possible_mask),NR_CPUS) + 1;
e0982e90
MT
377}
378
dd5af90a 379#ifndef CONFIG_HAVE_SETUP_PER_CPU_AREA
b73b459f 380unsigned long __per_cpu_offset[NR_CPUS] __read_mostly;
1da177e4
LT
381
382EXPORT_SYMBOL(__per_cpu_offset);
383
384static void __init setup_per_cpu_areas(void)
385{
386 unsigned long size, i;
387 char *ptr;
63872f87 388 unsigned long nr_possible_cpus = num_possible_cpus();
1da177e4
LT
389
390 /* Copy section for each CPU (we discard the original) */
b6e3590f
JF
391 size = ALIGN(PERCPU_ENOUGH_ROOM, PAGE_SIZE);
392 ptr = alloc_bootmem_pages(size * nr_possible_cpus);
1da177e4 393
0a945022 394 for_each_possible_cpu(i) {
1da177e4
LT
395 __per_cpu_offset[i] = ptr - __per_cpu_start;
396 memcpy(ptr, __per_cpu_start, __per_cpu_end - __per_cpu_start);
63872f87 397 ptr += size;
1da177e4
LT
398 }
399}
dd5af90a 400#endif /* CONFIG_HAVE_SETUP_PER_CPU_AREA */
1da177e4
LT
401
402/* Called by boot processor to activate the rest. */
403static void __init smp_init(void)
404{
53b8a315 405 unsigned int cpu;
1da177e4 406
e761b772
MK
407 /*
408 * Set up the current CPU as possible to migrate to.
409 * The other ones will be done by cpu_up/cpu_down()
410 */
2b17fa50 411 set_cpu_active(smp_processor_id(), true);
e761b772 412
1da177e4 413 /* FIXME: This should be done in userspace --RR */
53b8a315 414 for_each_present_cpu(cpu) {
ca74a6f8 415 if (num_online_cpus() >= setup_max_cpus)
1da177e4 416 break;
53b8a315
CL
417 if (!cpu_online(cpu))
418 cpu_up(cpu);
1da177e4
LT
419 }
420
421 /* Any cleanup work */
422 printk(KERN_INFO "Brought up %ld CPUs\n", (long)num_online_cpus());
ca74a6f8 423 smp_cpus_done(setup_max_cpus);
1da177e4
LT
424}
425
426#endif
427
30d7e0d4
ABL
428/*
429 * We need to store the untouched command line for future reference.
430 * We also need to store the touched command line since the parameter
431 * parsing is performed in place, and we should allow a component to
432 * store reference of name/value for future reference.
433 */
434static void __init setup_command_line(char *command_line)
435{
436 saved_command_line = alloc_bootmem(strlen (boot_command_line)+1);
437 static_command_line = alloc_bootmem(strlen (command_line)+1);
438 strcpy (saved_command_line, boot_command_line);
439 strcpy (static_command_line, command_line);
440}
441
1da177e4
LT
442/*
443 * We need to finalize in a non-__init function or else race conditions
444 * between the root thread and the init thread may cause start_kernel to
445 * be reaped by free_initmem before the root thread has proceeded to
446 * cpu_idle.
447 *
448 * gcc-3.4 accidentally inlines this function, so use noinline.
449 */
450
f99ebf0a 451static noinline void __init_refok rest_init(void)
1da177e4
LT
452 __releases(kernel_lock)
453{
73c27992
EB
454 int pid;
455
aae5f662 456 kernel_thread(kernel_init, NULL, CLONE_FS | CLONE_SIGHAND);
1da177e4 457 numa_default_policy();
73c27992 458 pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES);
5cd20455 459 kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns);
1da177e4 460 unlock_kernel();
f340c0d1
IM
461
462 /*
463 * The boot idle thread must execute schedule()
1df21055 464 * at least once to get things moving:
f340c0d1 465 */
1df21055 466 init_idle_bootup_task(current);
a6826048 467 rcu_scheduler_starting();
5bfb5d69 468 preempt_enable_no_resched();
f340c0d1 469 schedule();
5bfb5d69 470 preempt_disable();
f340c0d1 471
5bfb5d69 472 /* Call into cpu_idle with preempt disabled */
1da177e4 473 cpu_idle();
1df21055 474}
1da177e4
LT
475
476/* Check for early params. */
477static int __init do_early_param(char *param, char *val)
478{
479 struct obs_kernel_param *p;
480
481 for (p = __setup_start; p < __setup_end; p++) {
18a8bd94
YL
482 if ((p->early && strcmp(param, p->str) == 0) ||
483 (strcmp(param, "console") == 0 &&
484 strcmp(p->str, "earlycon") == 0)
485 ) {
1da177e4
LT
486 if (p->setup_func(val) != 0)
487 printk(KERN_WARNING
488 "Malformed early option '%s'\n", param);
489 }
490 }
491 /* We accept everything at this stage. */
492 return 0;
493}
494
495/* Arch code calls this early on, or if not, just before other parsing. */
496void __init parse_early_param(void)
497{
498 static __initdata int done = 0;
499 static __initdata char tmp_cmdline[COMMAND_LINE_SIZE];
500
501 if (done)
502 return;
503
504 /* All fall through to do_early_param. */
30d7e0d4 505 strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE);
1da177e4
LT
506 parse_args("early options", tmp_cmdline, NULL, 0, do_early_param);
507 done = 1;
508}
509
510/*
511 * Activate the first processor.
512 */
513
44fd2299
SS
514static void __init boot_cpu_init(void)
515{
516 int cpu = smp_processor_id();
517 /* Mark the boot cpu "present", "online" etc for SMP and UP case */
915441b6
RR
518 set_cpu_online(cpu, true);
519 set_cpu_present(cpu, true);
520 set_cpu_possible(cpu, true);
44fd2299
SS
521}
522
839ad62e 523void __init __weak smp_setup_processor_id(void)
033ab7f8
AM
524{
525}
526
8c9843e5
BH
527void __init __weak thread_info_cache_init(void)
528{
529}
530
1da177e4
LT
531asmlinkage void __init start_kernel(void)
532{
533 char * command_line;
534 extern struct kernel_param __start___param[], __stop___param[];
033ab7f8
AM
535
536 smp_setup_processor_id();
537
fbb9ce95
IM
538 /*
539 * Need to run as early as possible, to initialize the
540 * lockdep hash:
541 */
542 lockdep_init();
3ac7fe5a 543 debug_objects_early_init();
42059429
IM
544
545 /*
546 * Set up the the initial canary ASAP:
547 */
548 boot_init_stack_canary();
549
ddbcc7e8 550 cgroup_init_early();
fbb9ce95
IM
551
552 local_irq_disable();
553 early_boot_irqs_off();
243c7621 554 early_init_irq_lock_class();
fbb9ce95 555
1da177e4
LT
556/*
557 * Interrupts are still disabled. Do necessary setups, then
558 * enable them
559 */
560 lock_kernel();
906568c9 561 tick_init();
44fd2299 562 boot_cpu_init();
1da177e4
LT
563 page_address_init();
564 printk(KERN_NOTICE);
8993780a 565 printk(linux_banner);
1da177e4 566 setup_arch(&command_line);
cf475ad2 567 mm_init_owner(&init_mm, &init_task);
30d7e0d4 568 setup_command_line(command_line);
1da177e4 569 setup_per_cpu_areas();
e0982e90 570 setup_nr_cpu_ids();
44fd2299 571 smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */
1da177e4
LT
572
573 /*
574 * Set up the scheduler prior starting any interrupts (such as the
575 * timer interrupt). Full topology setup happens at smp_init()
576 * time - but meanwhile we still have a functioning scheduler.
577 */
578 sched_init();
579 /*
580 * Disable preemption - early bootup scheduling is extremely
581 * fragile until we cpu_idle() for the first time.
582 */
583 preempt_disable();
584 build_all_zonelists();
585 page_alloc_init();
30d7e0d4 586 printk(KERN_NOTICE "Kernel command line: %s\n", boot_command_line);
1da177e4 587 parse_early_param();
30d7e0d4 588 parse_args("Booting kernel", static_command_line, __start___param,
1da177e4
LT
589 __stop___param - __start___param,
590 &unknown_bootoption);
c4a68306
AB
591 if (!irqs_disabled()) {
592 printk(KERN_WARNING "start_kernel(): bug: interrupts were "
593 "enabled *very* early, fixing it\n");
594 local_irq_disable();
595 }
1da177e4
LT
596 sort_main_extable();
597 trap_init();
598 rcu_init();
0b8f1efa
YL
599 /* init some links before init_ISA_irqs() */
600 early_irq_init();
1da177e4
LT
601 init_IRQ();
602 pidhash_init();
603 init_timers();
c0a31329 604 hrtimers_init();
1da177e4 605 softirq_init();
ad596171 606 timekeeping_init();
88fecaa2 607 time_init();
3e51f33f 608 sched_clock_init();
93e02814
HC
609 profile_init();
610 if (!irqs_disabled())
24d431d0
RL
611 printk(KERN_CRIT "start_kernel(): bug: interrupts were "
612 "enabled early\n");
fbb9ce95 613 early_boot_irqs_on();
93e02814 614 local_irq_enable();
1da177e4
LT
615
616 /*
617 * HACK ALERT! This is early. We're enabling the console before
618 * we've done PCI setups etc, and console_init() must be aware of
619 * this. But we do want output early, in case something goes wrong.
620 */
621 console_init();
622 if (panic_later)
623 panic(panic_later, panic_param);
fbb9ce95
IM
624
625 lockdep_info();
626
9a11b49a
IM
627 /*
628 * Need to run this when irqs are enabled, because it wants
629 * to self-test [hard/soft]-irqs on/off lock inversion bugs
630 * too:
631 */
632 locking_selftest();
633
1da177e4
LT
634#ifdef CONFIG_BLK_DEV_INITRD
635 if (initrd_start && !initrd_below_start_ok &&
bd673c7c 636 page_to_pfn(virt_to_page((void *)initrd_start)) < min_low_pfn) {
1da177e4 637 printk(KERN_CRIT "initrd overwritten (0x%08lx < 0x%08lx) - "
fb6624eb 638 "disabling it.\n",
bd673c7c
GU
639 page_to_pfn(virt_to_page((void *)initrd_start)),
640 min_low_pfn);
1da177e4
LT
641 initrd_start = 0;
642 }
643#endif
db64fe02 644 vmalloc_init();
1da177e4 645 vfs_caches_init_early();
c417f024 646 cpuset_init_early();
94b6da5a 647 page_cgroup_init();
1da177e4 648 mem_init();
a03c2a48 649 enable_debug_pagealloc();
d221938c 650 cpu_hotplug_init();
1da177e4 651 kmem_cache_init();
b9ce08c0 652 kmemtrace_init();
3ac7fe5a 653 debug_objects_mem_init();
199f0ca5 654 idr_init_cache();
e7c8d5c9 655 setup_per_cpu_pageset();
1da177e4
LT
656 numa_policy_init();
657 if (late_time_init)
658 late_time_init();
659 calibrate_delay();
660 pidmap_init();
661 pgtable_cache_init();
662 prio_tree_init();
663 anon_vma_init();
664#ifdef CONFIG_X86
665 if (efi_enabled)
666 efi_enter_virtual_mode();
667#endif
8c9843e5 668 thread_info_cache_init();
d84f4f99 669 cred_init();
1da177e4
LT
670 fork_init(num_physpages);
671 proc_caches_init();
672 buffer_init();
1da177e4
LT
673 key_init();
674 security_init();
675 vfs_caches_init(num_physpages);
676 radix_tree_init();
677 signals_init();
678 /* rootfs populating might need page-writeback */
679 page_writeback_init();
680#ifdef CONFIG_PROC_FS
681 proc_root_init();
682#endif
ddbcc7e8 683 cgroup_init();
1da177e4 684 cpuset_init();
c757249a 685 taskstats_init_early();
ca74e92b 686 delayacct_init();
1da177e4
LT
687
688 check_bugs();
689
690 acpi_early_init(); /* before LAPIC and SMP init */
691
68bf21aa
SR
692 ftrace_init();
693
1da177e4
LT
694 /* Do the rest non-__init'ed, we're now alive */
695 rest_init();
696}
697
22a9d645 698int initcall_debug;
d0ea3d7d 699core_param(initcall_debug, initcall_debug, bool, 0644);
1da177e4 700
59f9415f 701int do_one_initcall(initcall_t fn)
1da177e4 702{
1da177e4 703 int count = preempt_count();
74239072 704 ktime_t calltime, delta, rettime;
a76bfd0d 705 char msgbuf[64];
74239072
FW
706 struct boot_trace_call call;
707 struct boot_trace_ret ret;
1da177e4 708
e0df154f 709 if (initcall_debug) {
74239072
FW
710 call.caller = task_pid_nr(current);
711 printk("calling %pF @ %i\n", fn, call.caller);
712 calltime = ktime_get();
713 trace_boot_call(&call, fn);
71566a0d 714 enable_boot_trace();
e0df154f 715 }
1da177e4 716
74239072 717 ret.result = fn();
1da177e4 718
e0df154f 719 if (initcall_debug) {
71566a0d 720 disable_boot_trace();
74239072
FW
721 rettime = ktime_get();
722 delta = ktime_sub(rettime, calltime);
1d926f27 723 ret.duration = (unsigned long long) ktime_to_ns(delta) >> 10;
74239072 724 trace_boot_ret(&ret, fn);
ca538f6b 725 printk("initcall %pF returned %d after %Ld usecs\n", fn,
74239072 726 ret.result, ret.duration);
e0df154f 727 }
8f0c45cd 728
e0df154f 729 msgbuf[0] = 0;
e662e1cf 730
74239072
FW
731 if (ret.result && ret.result != -ENODEV && initcall_debug)
732 sprintf(msgbuf, "error code %d ", ret.result);
e662e1cf 733
e0df154f 734 if (preempt_count() != count) {
a76bfd0d 735 strlcat(msgbuf, "preemption imbalance ", sizeof(msgbuf));
e0df154f 736 preempt_count() = count;
1da177e4 737 }
e0df154f 738 if (irqs_disabled()) {
a76bfd0d 739 strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf));
e0df154f
LT
740 local_irq_enable();
741 }
742 if (msgbuf[0]) {
96d746c6 743 printk("initcall %pF returned with %s\n", fn, msgbuf);
e0df154f 744 }
59f9415f 745
74239072 746 return ret.result;
e0df154f
LT
747}
748
749
c2147a50 750extern initcall_t __initcall_start[], __initcall_end[], __early_initcall_end[];
e0df154f
LT
751
752static void __init do_initcalls(void)
753{
754 initcall_t *call;
755
c2147a50 756 for (call = __early_initcall_end; call < __initcall_end; call++)
e0df154f 757 do_one_initcall(*call);
1da177e4
LT
758
759 /* Make sure there is no pending stuff from the initcall sequence */
760 flush_scheduled_work();
761}
762
763/*
764 * Ok, the machine is now initialized. None of the devices
765 * have been touched yet, but the CPU subsystem is up and
766 * running, and memory and process management works.
767 *
768 * Now we can finally start doing some real work..
769 */
770static void __init do_basic_setup(void)
771{
4446a36f 772 rcu_init_sched(); /* needed by module_init stage. */
4403b406 773 init_workqueues();
759ee091 774 cpuset_init_smp();
1da177e4
LT
775 usermodehelper_init();
776 driver_init();
b04c3afb 777 init_irq_proc();
1da177e4
LT
778 do_initcalls();
779}
780
7babe8db 781static void __init do_pre_smp_initcalls(void)
c2147a50
EGM
782{
783 initcall_t *call;
784
785 for (call = __initcall_start; call < __early_initcall_end; call++)
786 do_one_initcall(*call);
787}
788
1da177e4
LT
789static void run_init_process(char *init_filename)
790{
791 argv_init[0] = init_filename;
67608567 792 kernel_execve(init_filename, argv_init, envp_init);
1da177e4
LT
793}
794
ee5bfa64
VG
795/* This is a non __init function. Force it to be noinline otherwise gcc
796 * makes it inline to init() and it becomes part of init.text section
797 */
f99ebf0a 798static noinline int init_post(void)
acdd052a 799 __releases(kernel_lock)
ee5bfa64 800{
22a9d645
AV
801 /* need to finish all async __init code before freeing the memory */
802 async_synchronize_full();
ee5bfa64
VG
803 free_initmem();
804 unlock_kernel();
805 mark_rodata_ro();
806 system_state = SYSTEM_RUNNING;
807 numa_default_policy();
808
809 if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0)
810 printk(KERN_WARNING "Warning: unable to open an initial console.\n");
811
812 (void) sys_dup(0);
813 (void) sys_dup(0);
814
fae5fa44
ON
815 current->signal->flags |= SIGNAL_UNKILLABLE;
816
ee5bfa64
VG
817 if (ramdisk_execute_command) {
818 run_init_process(ramdisk_execute_command);
819 printk(KERN_WARNING "Failed to execute %s\n",
820 ramdisk_execute_command);
821 }
822
823 /*
824 * We try each of these until one succeeds.
825 *
826 * The Bourne shell can be used instead of init if we are
827 * trying to recover a really broken machine.
828 */
829 if (execute_command) {
830 run_init_process(execute_command);
831 printk(KERN_WARNING "Failed to execute %s. Attempting "
832 "defaults...\n", execute_command);
833 }
834 run_init_process("/sbin/init");
835 run_init_process("/etc/init");
836 run_init_process("/bin/init");
837 run_init_process("/bin/sh");
838
839 panic("No init found. Try passing init= option to kernel.");
840}
841
aae5f662 842static int __init kernel_init(void * unused)
1da177e4
LT
843{
844 lock_kernel();
845 /*
846 * init can run on any cpu.
847 */
1a2142af 848 set_cpus_allowed_ptr(current, cpu_all_mask);
1da177e4
LT
849 /*
850 * Tell the world that we're going to be the grim
851 * reaper of innocent orphaned children.
852 *
853 * We don't want people to have to make incorrect
854 * assumptions about where in the task array this
855 * can be found.
856 */
84d73786 857 init_pid_ns.child_reaper = current;
1da177e4 858
9ec52099
CLG
859 cad_pid = task_pid(current);
860
ca74a6f8 861 smp_prepare_cpus(setup_max_cpus);
1da177e4
LT
862
863 do_pre_smp_initcalls();
3bf77af6 864 start_boot_trace();
1da177e4 865
1da177e4
LT
866 smp_init();
867 sched_init_smp();
868
1da177e4
LT
869 do_basic_setup();
870
871 /*
872 * check if there is an early userspace init. If yes, let it do all
873 * the work
874 */
ffdfc409
OJ
875
876 if (!ramdisk_execute_command)
877 ramdisk_execute_command = "/init";
878
879 if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) {
880 ramdisk_execute_command = NULL;
1da177e4 881 prepare_namespace();
ffdfc409 882 }
1da177e4
LT
883
884 /*
885 * Ok, we have completed the initial bootup, and
886 * we're essentially up and running. Get rid of the
887 * initmem segments and start the user-mode stuff..
888 */
71566a0d 889
ee5bfa64
VG
890 init_post();
891 return 0;
1da177e4 892}