]> bbs.cooldavid.org Git - net-next-2.6.git/blame - init/main.c
kmemcheck: add the kmemcheck documentation
[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>
3c7b4e6b 59#include <linux/kmemleak.h>
84d73786 60#include <linux/pid_namespace.h>
1f21782e 61#include <linux/device.h>
73c27992 62#include <linux/kthread.h>
e6fe6649 63#include <linux/sched.h>
a1c9eea9 64#include <linux/signal.h>
199f0ca5 65#include <linux/idr.h>
68bf21aa 66#include <linux/ftrace.h>
22a9d645 67#include <linux/async.h>
02af61bb 68#include <linux/kmemtrace.h>
3f5ec136 69#include <trace/boot.h>
1da177e4
LT
70
71#include <asm/io.h>
72#include <asm/bugs.h>
73#include <asm/setup.h>
a940199f 74#include <asm/sections.h>
37b73c82 75#include <asm/cacheflush.h>
1da177e4 76
1da177e4
LT
77#ifdef CONFIG_X86_LOCAL_APIC
78#include <asm/smp.h>
79#endif
80
aae5f662 81static int kernel_init(void *);
1da177e4
LT
82
83extern void init_IRQ(void);
1da177e4
LT
84extern void fork_init(unsigned long);
85extern void mca_init(void);
86extern void sbus_init(void);
1da177e4
LT
87extern void prio_tree_init(void);
88extern void radix_tree_init(void);
89extern void free_initmem(void);
1da177e4
LT
90#ifdef CONFIG_ACPI
91extern void acpi_early_init(void);
92#else
93static inline void acpi_early_init(void) { }
94#endif
37b73c82
AV
95#ifndef CONFIG_DEBUG_RODATA
96static inline void mark_rodata_ro(void) { }
97#endif
1da177e4
LT
98
99#ifdef CONFIG_TC
100extern void tc_init(void);
101#endif
102
a6826048 103enum system_states system_state __read_mostly;
1da177e4
LT
104EXPORT_SYMBOL(system_state);
105
106/*
107 * Boot command-line arguments
108 */
109#define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT
110#define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT
111
112extern void time_init(void);
113/* Default late time init is NULL. archs can override this later. */
d2e3192b 114void (*__initdata late_time_init)(void);
1da177e4
LT
115extern void softirq_init(void);
116
30d7e0d4
ABL
117/* Untouched command line saved by arch-specific code. */
118char __initdata boot_command_line[COMMAND_LINE_SIZE];
119/* Untouched saved command line (eg. for /proc) */
120char *saved_command_line;
121/* Command line for parameter parsing */
122static char *static_command_line;
1da177e4
LT
123
124static char *execute_command;
ffdfc409 125static char *ramdisk_execute_command;
1da177e4 126
8b3b2955 127#ifdef CONFIG_SMP
1da177e4 128/* Setup configured maximum number of CPUs to activate */
ca74a6f8 129unsigned int __initdata setup_max_cpus = NR_CPUS;
7e96287d 130
1da177e4
LT
131/*
132 * Setup routine for controlling SMP activation
133 *
134 * Command-line option of "nosmp" or "maxcpus=0" will disable SMP
135 * activation entirely (the MPS table probe still happens, though).
136 *
137 * Command-line option of "maxcpus=<NUM>", where <NUM> is an integer
138 * greater than 0, limits the maximum number of CPUs activated in
139 * SMP mode to <NUM>.
140 */
65a4e574
IM
141
142void __weak arch_disable_smp_support(void) { }
61ec7567 143
1da177e4
LT
144static int __init nosmp(char *str)
145{
ca74a6f8 146 setup_max_cpus = 0;
65a4e574
IM
147 arch_disable_smp_support();
148
8b3b2955 149 return 0;
1da177e4
LT
150}
151
8b3b2955 152early_param("nosmp", nosmp);
1da177e4
LT
153
154static int __init maxcpus(char *str)
155{
ca74a6f8
AK
156 get_option(&str, &setup_max_cpus);
157 if (setup_max_cpus == 0)
65a4e574 158 arch_disable_smp_support();
61ec7567
LB
159
160 return 0;
1da177e4
LT
161}
162
81340977 163early_param("maxcpus", maxcpus);
8b3b2955 164#else
65a4e574 165const unsigned int setup_max_cpus = NR_CPUS;
8b3b2955
JB
166#endif
167
168/*
169 * If set, this is an indication to the drivers that reset the underlying
170 * device before going ahead with the initialization otherwise driver might
171 * rely on the BIOS and skip the reset operation.
172 *
173 * This is useful if kernel is booting in an unreliable environment.
174 * For ex. kdump situaiton where previous kernel has crashed, BIOS has been
175 * skipped and devices will be in unknown state.
176 */
177unsigned int reset_devices;
178EXPORT_SYMBOL(reset_devices);
1da177e4 179
7e96287d
VG
180static int __init set_reset_devices(char *str)
181{
182 reset_devices = 1;
183 return 1;
184}
185
186__setup("reset_devices", set_reset_devices);
187
1da177e4
LT
188static char * argv_init[MAX_INIT_ARGS+2] = { "init", NULL, };
189char * envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, };
190static const char *panic_later, *panic_param;
191
192extern struct obs_kernel_param __setup_start[], __setup_end[];
193
194static int __init obsolete_checksetup(char *line)
195{
196 struct obs_kernel_param *p;
33df0d19 197 int had_early_param = 0;
1da177e4
LT
198
199 p = __setup_start;
200 do {
201 int n = strlen(p->str);
202 if (!strncmp(line, p->str, n)) {
203 if (p->early) {
33df0d19
RR
204 /* Already done in parse_early_param?
205 * (Needs exact match on param part).
206 * Keep iterating, as we can have early
207 * params and __setups of same names 8( */
1da177e4 208 if (line[n] == '\0' || line[n] == '=')
33df0d19 209 had_early_param = 1;
1da177e4
LT
210 } else if (!p->setup_func) {
211 printk(KERN_WARNING "Parameter %s is obsolete,"
212 " ignored\n", p->str);
213 return 1;
214 } else if (p->setup_func(line + n))
215 return 1;
216 }
217 p++;
218 } while (p < __setup_end);
33df0d19
RR
219
220 return had_early_param;
1da177e4
LT
221}
222
223/*
224 * This should be approx 2 Bo*oMips to start (note initial shift), and will
225 * still work even if initially too large, it will just take slightly longer
226 */
227unsigned long loops_per_jiffy = (1<<12);
228
229EXPORT_SYMBOL(loops_per_jiffy);
230
231static int __init debug_kernel(char *str)
232{
1da177e4 233 console_loglevel = 10;
f6f21c81 234 return 0;
1da177e4
LT
235}
236
237static int __init quiet_kernel(char *str)
238{
1da177e4 239 console_loglevel = 4;
f6f21c81 240 return 0;
1da177e4
LT
241}
242
f6f21c81
YL
243early_param("debug", debug_kernel);
244early_param("quiet", quiet_kernel);
1da177e4
LT
245
246static int __init loglevel(char *str)
247{
248 get_option(&str, &console_loglevel);
d9d4fcfe 249 return 0;
1da177e4
LT
250}
251
f6f21c81 252early_param("loglevel", loglevel);
1da177e4
LT
253
254/*
255 * Unknown boot options get handed to init, unless they look like
256 * failed parameters
257 */
258static int __init unknown_bootoption(char *param, char *val)
259{
260 /* Change NUL term back to "=", to make "param" the whole string. */
261 if (val) {
262 /* param=val or param="val"? */
263 if (val == param+strlen(param)+1)
264 val[-1] = '=';
265 else if (val == param+strlen(param)+2) {
266 val[-2] = '=';
267 memmove(val-1, val, strlen(val)+1);
268 val--;
269 } else
270 BUG();
271 }
272
273 /* Handle obsolete-style parameters */
274 if (obsolete_checksetup(param))
275 return 0;
276
277 /*
211fee8a 278 * Preemptive maintenance for "why didn't my misspelled command
1da177e4
LT
279 * line work?"
280 */
281 if (strchr(param, '.') && (!val || strchr(param, '.') < val)) {
282 printk(KERN_ERR "Unknown boot option `%s': ignoring\n", param);
283 return 0;
284 }
285
286 if (panic_later)
287 return 0;
288
289 if (val) {
290 /* Environment option */
291 unsigned int i;
292 for (i = 0; envp_init[i]; i++) {
293 if (i == MAX_INIT_ENVS) {
294 panic_later = "Too many boot env vars at `%s'";
295 panic_param = param;
296 }
297 if (!strncmp(param, envp_init[i], val - param))
298 break;
299 }
300 envp_init[i] = param;
301 } else {
302 /* Command line option */
303 unsigned int i;
304 for (i = 0; argv_init[i]; i++) {
305 if (i == MAX_INIT_ARGS) {
306 panic_later = "Too many boot init vars at `%s'";
307 panic_param = param;
308 }
309 }
310 argv_init[i] = param;
311 }
312 return 0;
313}
314
12d6f21e
IM
315#ifdef CONFIG_DEBUG_PAGEALLOC
316int __read_mostly debug_pagealloc_enabled = 0;
317#endif
318
1da177e4
LT
319static int __init init_setup(char *str)
320{
321 unsigned int i;
322
323 execute_command = str;
324 /*
325 * In case LILO is going to boot us with default command line,
326 * it prepends "auto" before the whole cmdline which makes
327 * the shell think it should execute a script with such name.
328 * So we ignore all arguments entered _before_ init=... [MJ]
329 */
330 for (i = 1; i < MAX_INIT_ARGS; i++)
331 argv_init[i] = NULL;
332 return 1;
333}
334__setup("init=", init_setup);
335
ffdfc409
OJ
336static int __init rdinit_setup(char *str)
337{
338 unsigned int i;
339
340 ramdisk_execute_command = str;
341 /* See "auto" comment in init_setup */
342 for (i = 1; i < MAX_INIT_ARGS; i++)
343 argv_init[i] = NULL;
344 return 1;
345}
346__setup("rdinit=", rdinit_setup);
347
1da177e4
LT
348#ifndef CONFIG_SMP
349
350#ifdef CONFIG_X86_LOCAL_APIC
351static void __init smp_init(void)
352{
353 APIC_init_uniprocessor();
354}
355#else
356#define smp_init() do { } while (0)
357#endif
358
359static inline void setup_per_cpu_areas(void) { }
e0982e90 360static inline void setup_nr_cpu_ids(void) { }
1da177e4
LT
361static inline void smp_prepare_cpus(unsigned int maxcpus) { }
362
363#else
364
321a8e9d
MT
365#if NR_CPUS > BITS_PER_LONG
366cpumask_t cpu_mask_all __read_mostly = CPU_MASK_ALL;
367EXPORT_SYMBOL(cpu_mask_all);
368#endif
369
e0982e90
MT
370/* Setup number of possible processor ids */
371int nr_cpu_ids __read_mostly = NR_CPUS;
372EXPORT_SYMBOL(nr_cpu_ids);
373
374/* An arch may set nr_cpu_ids earlier if needed, so this would be redundant */
375static void __init setup_nr_cpu_ids(void)
376{
e0c0ba73 377 nr_cpu_ids = find_last_bit(cpumask_bits(cpu_possible_mask),NR_CPUS) + 1;
e0982e90
MT
378}
379
dd5af90a 380#ifndef CONFIG_HAVE_SETUP_PER_CPU_AREA
b73b459f 381unsigned long __per_cpu_offset[NR_CPUS] __read_mostly;
1da177e4
LT
382
383EXPORT_SYMBOL(__per_cpu_offset);
384
385static void __init setup_per_cpu_areas(void)
386{
387 unsigned long size, i;
388 char *ptr;
63872f87 389 unsigned long nr_possible_cpus = num_possible_cpus();
1da177e4
LT
390
391 /* Copy section for each CPU (we discard the original) */
b6e3590f
JF
392 size = ALIGN(PERCPU_ENOUGH_ROOM, PAGE_SIZE);
393 ptr = alloc_bootmem_pages(size * nr_possible_cpus);
1da177e4 394
0a945022 395 for_each_possible_cpu(i) {
1da177e4
LT
396 __per_cpu_offset[i] = ptr - __per_cpu_start;
397 memcpy(ptr, __per_cpu_start, __per_cpu_end - __per_cpu_start);
63872f87 398 ptr += size;
1da177e4
LT
399 }
400}
dd5af90a 401#endif /* CONFIG_HAVE_SETUP_PER_CPU_AREA */
1da177e4
LT
402
403/* Called by boot processor to activate the rest. */
404static void __init smp_init(void)
405{
53b8a315 406 unsigned int cpu;
1da177e4 407
e761b772
MK
408 /*
409 * Set up the current CPU as possible to migrate to.
410 * The other ones will be done by cpu_up/cpu_down()
411 */
2b17fa50 412 set_cpu_active(smp_processor_id(), true);
e761b772 413
1da177e4 414 /* FIXME: This should be done in userspace --RR */
53b8a315 415 for_each_present_cpu(cpu) {
ca74a6f8 416 if (num_online_cpus() >= setup_max_cpus)
1da177e4 417 break;
53b8a315
CL
418 if (!cpu_online(cpu))
419 cpu_up(cpu);
1da177e4
LT
420 }
421
422 /* Any cleanup work */
423 printk(KERN_INFO "Brought up %ld CPUs\n", (long)num_online_cpus());
ca74a6f8 424 smp_cpus_done(setup_max_cpus);
1da177e4
LT
425}
426
427#endif
428
30d7e0d4
ABL
429/*
430 * We need to store the untouched command line for future reference.
431 * We also need to store the touched command line since the parameter
432 * parsing is performed in place, and we should allow a component to
433 * store reference of name/value for future reference.
434 */
435static void __init setup_command_line(char *command_line)
436{
437 saved_command_line = alloc_bootmem(strlen (boot_command_line)+1);
438 static_command_line = alloc_bootmem(strlen (command_line)+1);
439 strcpy (saved_command_line, boot_command_line);
440 strcpy (static_command_line, command_line);
441}
442
1da177e4
LT
443/*
444 * We need to finalize in a non-__init function or else race conditions
445 * between the root thread and the init thread may cause start_kernel to
446 * be reaped by free_initmem before the root thread has proceeded to
447 * cpu_idle.
448 *
449 * gcc-3.4 accidentally inlines this function, so use noinline.
450 */
451
f99ebf0a 452static noinline void __init_refok rest_init(void)
1da177e4
LT
453 __releases(kernel_lock)
454{
73c27992
EB
455 int pid;
456
aae5f662 457 kernel_thread(kernel_init, NULL, CLONE_FS | CLONE_SIGHAND);
1da177e4 458 numa_default_policy();
73c27992 459 pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES);
5cd20455 460 kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns);
1da177e4 461 unlock_kernel();
f340c0d1
IM
462
463 /*
464 * The boot idle thread must execute schedule()
1df21055 465 * at least once to get things moving:
f340c0d1 466 */
1df21055 467 init_idle_bootup_task(current);
a6826048 468 rcu_scheduler_starting();
5bfb5d69 469 preempt_enable_no_resched();
f340c0d1 470 schedule();
5bfb5d69 471 preempt_disable();
f340c0d1 472
5bfb5d69 473 /* Call into cpu_idle with preempt disabled */
1da177e4 474 cpu_idle();
1df21055 475}
1da177e4
LT
476
477/* Check for early params. */
478static int __init do_early_param(char *param, char *val)
479{
480 struct obs_kernel_param *p;
481
482 for (p = __setup_start; p < __setup_end; p++) {
18a8bd94
YL
483 if ((p->early && strcmp(param, p->str) == 0) ||
484 (strcmp(param, "console") == 0 &&
485 strcmp(p->str, "earlycon") == 0)
486 ) {
1da177e4
LT
487 if (p->setup_func(val) != 0)
488 printk(KERN_WARNING
489 "Malformed early option '%s'\n", param);
490 }
491 }
492 /* We accept everything at this stage. */
493 return 0;
494}
495
13977091
MD
496void __init parse_early_options(char *cmdline)
497{
498 parse_args("early options", cmdline, NULL, 0, do_early_param);
499}
500
1da177e4
LT
501/* Arch code calls this early on, or if not, just before other parsing. */
502void __init parse_early_param(void)
503{
504 static __initdata int done = 0;
505 static __initdata char tmp_cmdline[COMMAND_LINE_SIZE];
506
507 if (done)
508 return;
509
510 /* All fall through to do_early_param. */
30d7e0d4 511 strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE);
13977091 512 parse_early_options(tmp_cmdline);
1da177e4
LT
513 done = 1;
514}
515
516/*
517 * Activate the first processor.
518 */
519
44fd2299
SS
520static void __init boot_cpu_init(void)
521{
522 int cpu = smp_processor_id();
523 /* Mark the boot cpu "present", "online" etc for SMP and UP case */
915441b6
RR
524 set_cpu_online(cpu, true);
525 set_cpu_present(cpu, true);
526 set_cpu_possible(cpu, true);
44fd2299
SS
527}
528
839ad62e 529void __init __weak smp_setup_processor_id(void)
033ab7f8
AM
530{
531}
532
8c9843e5
BH
533void __init __weak thread_info_cache_init(void)
534{
535}
536
444f478f
PE
537/*
538 * Set up kernel memory allocators
539 */
540static void __init mm_init(void)
541{
542 mem_init();
543 kmem_cache_init();
544 vmalloc_init();
545}
546
1da177e4
LT
547asmlinkage void __init start_kernel(void)
548{
549 char * command_line;
550 extern struct kernel_param __start___param[], __stop___param[];
033ab7f8
AM
551
552 smp_setup_processor_id();
553
fbb9ce95
IM
554 /*
555 * Need to run as early as possible, to initialize the
556 * lockdep hash:
557 */
558 lockdep_init();
3ac7fe5a 559 debug_objects_early_init();
42059429
IM
560
561 /*
562 * Set up the the initial canary ASAP:
563 */
564 boot_init_stack_canary();
565
ddbcc7e8 566 cgroup_init_early();
fbb9ce95
IM
567
568 local_irq_disable();
569 early_boot_irqs_off();
243c7621 570 early_init_irq_lock_class();
fbb9ce95 571
1da177e4
LT
572/*
573 * Interrupts are still disabled. Do necessary setups, then
574 * enable them
575 */
576 lock_kernel();
906568c9 577 tick_init();
44fd2299 578 boot_cpu_init();
1da177e4 579 page_address_init();
657cafa6 580 printk(KERN_NOTICE "%s", linux_banner);
1da177e4 581 setup_arch(&command_line);
cf475ad2 582 mm_init_owner(&init_mm, &init_task);
30d7e0d4 583 setup_command_line(command_line);
1da177e4 584 setup_per_cpu_areas();
e0982e90 585 setup_nr_cpu_ids();
44fd2299 586 smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */
1da177e4 587
83b519e8
PE
588 build_all_zonelists();
589 page_alloc_init();
590
591 printk(KERN_NOTICE "Kernel command line: %s\n", boot_command_line);
592 parse_early_param();
593 parse_args("Booting kernel", static_command_line, __start___param,
594 __stop___param - __start___param,
595 &unknown_bootoption);
596 /*
597 * These use large bootmem allocations and must precede
598 * kmem_cache_init()
599 */
600 pidhash_init();
83b519e8
PE
601 vfs_caches_init_early();
602 sort_main_extable();
603 trap_init();
444f478f 604 mm_init();
1da177e4
LT
605 /*
606 * Set up the scheduler prior starting any interrupts (such as the
607 * timer interrupt). Full topology setup happens at smp_init()
608 * time - but meanwhile we still have a functioning scheduler.
609 */
610 sched_init();
611 /*
612 * Disable preemption - early bootup scheduling is extremely
613 * fragile until we cpu_idle() for the first time.
614 */
615 preempt_disable();
c4a68306
AB
616 if (!irqs_disabled()) {
617 printk(KERN_WARNING "start_kernel(): bug: interrupts were "
618 "enabled *very* early, fixing it\n");
619 local_irq_disable();
620 }
1da177e4 621 rcu_init();
0b8f1efa
YL
622 /* init some links before init_ISA_irqs() */
623 early_irq_init();
1da177e4 624 init_IRQ();
3c7b4e6b 625 prio_tree_init();
1da177e4 626 init_timers();
c0a31329 627 hrtimers_init();
1da177e4 628 softirq_init();
ad596171 629 timekeeping_init();
88fecaa2 630 time_init();
3e51f33f 631 sched_clock_init();
93e02814
HC
632 profile_init();
633 if (!irqs_disabled())
24d431d0
RL
634 printk(KERN_CRIT "start_kernel(): bug: interrupts were "
635 "enabled early\n");
fbb9ce95 636 early_boot_irqs_on();
93e02814 637 local_irq_enable();
1da177e4
LT
638
639 /*
640 * HACK ALERT! This is early. We're enabling the console before
641 * we've done PCI setups etc, and console_init() must be aware of
642 * this. But we do want output early, in case something goes wrong.
643 */
644 console_init();
645 if (panic_later)
646 panic(panic_later, panic_param);
fbb9ce95
IM
647
648 lockdep_info();
649
9a11b49a
IM
650 /*
651 * Need to run this when irqs are enabled, because it wants
652 * to self-test [hard/soft]-irqs on/off lock inversion bugs
653 * too:
654 */
655 locking_selftest();
656
1da177e4
LT
657#ifdef CONFIG_BLK_DEV_INITRD
658 if (initrd_start && !initrd_below_start_ok &&
bd673c7c 659 page_to_pfn(virt_to_page((void *)initrd_start)) < min_low_pfn) {
1da177e4 660 printk(KERN_CRIT "initrd overwritten (0x%08lx < 0x%08lx) - "
fb6624eb 661 "disabling it.\n",
bd673c7c
GU
662 page_to_pfn(virt_to_page((void *)initrd_start)),
663 min_low_pfn);
1da177e4
LT
664 initrd_start = 0;
665 }
666#endif
c417f024 667 cpuset_init_early();
94b6da5a 668 page_cgroup_init();
a03c2a48 669 enable_debug_pagealloc();
d221938c 670 cpu_hotplug_init();
b9ce08c0 671 kmemtrace_init();
3c7b4e6b 672 kmemleak_init();
3ac7fe5a 673 debug_objects_mem_init();
199f0ca5 674 idr_init_cache();
e7c8d5c9 675 setup_per_cpu_pageset();
1da177e4
LT
676 numa_policy_init();
677 if (late_time_init)
678 late_time_init();
679 calibrate_delay();
680 pidmap_init();
681 pgtable_cache_init();
1da177e4
LT
682 anon_vma_init();
683#ifdef CONFIG_X86
684 if (efi_enabled)
685 efi_enter_virtual_mode();
686#endif
8c9843e5 687 thread_info_cache_init();
d84f4f99 688 cred_init();
1da177e4
LT
689 fork_init(num_physpages);
690 proc_caches_init();
691 buffer_init();
1da177e4
LT
692 key_init();
693 security_init();
694 vfs_caches_init(num_physpages);
695 radix_tree_init();
696 signals_init();
697 /* rootfs populating might need page-writeback */
698 page_writeback_init();
699#ifdef CONFIG_PROC_FS
700 proc_root_init();
701#endif
ddbcc7e8 702 cgroup_init();
1da177e4 703 cpuset_init();
c757249a 704 taskstats_init_early();
ca74e92b 705 delayacct_init();
1da177e4
LT
706
707 check_bugs();
708
709 acpi_early_init(); /* before LAPIC and SMP init */
710
68bf21aa
SR
711 ftrace_init();
712
1da177e4
LT
713 /* Do the rest non-__init'ed, we're now alive */
714 rest_init();
715}
716
22a9d645 717int initcall_debug;
d0ea3d7d 718core_param(initcall_debug, initcall_debug, bool, 0644);
1da177e4 719
59f9415f 720int do_one_initcall(initcall_t fn)
1da177e4 721{
1da177e4 722 int count = preempt_count();
74239072 723 ktime_t calltime, delta, rettime;
a76bfd0d 724 char msgbuf[64];
74239072
FW
725 struct boot_trace_call call;
726 struct boot_trace_ret ret;
1da177e4 727
e0df154f 728 if (initcall_debug) {
74239072
FW
729 call.caller = task_pid_nr(current);
730 printk("calling %pF @ %i\n", fn, call.caller);
731 calltime = ktime_get();
732 trace_boot_call(&call, fn);
71566a0d 733 enable_boot_trace();
e0df154f 734 }
1da177e4 735
74239072 736 ret.result = fn();
1da177e4 737
e0df154f 738 if (initcall_debug) {
71566a0d 739 disable_boot_trace();
74239072
FW
740 rettime = ktime_get();
741 delta = ktime_sub(rettime, calltime);
1d926f27 742 ret.duration = (unsigned long long) ktime_to_ns(delta) >> 10;
74239072 743 trace_boot_ret(&ret, fn);
ca538f6b 744 printk("initcall %pF returned %d after %Ld usecs\n", fn,
74239072 745 ret.result, ret.duration);
e0df154f 746 }
8f0c45cd 747
e0df154f 748 msgbuf[0] = 0;
e662e1cf 749
74239072
FW
750 if (ret.result && ret.result != -ENODEV && initcall_debug)
751 sprintf(msgbuf, "error code %d ", ret.result);
e662e1cf 752
e0df154f 753 if (preempt_count() != count) {
a76bfd0d 754 strlcat(msgbuf, "preemption imbalance ", sizeof(msgbuf));
e0df154f 755 preempt_count() = count;
1da177e4 756 }
e0df154f 757 if (irqs_disabled()) {
a76bfd0d 758 strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf));
e0df154f
LT
759 local_irq_enable();
760 }
761 if (msgbuf[0]) {
96d746c6 762 printk("initcall %pF returned with %s\n", fn, msgbuf);
e0df154f 763 }
59f9415f 764
74239072 765 return ret.result;
e0df154f
LT
766}
767
768
c2147a50 769extern initcall_t __initcall_start[], __initcall_end[], __early_initcall_end[];
e0df154f
LT
770
771static void __init do_initcalls(void)
772{
773 initcall_t *call;
774
c2147a50 775 for (call = __early_initcall_end; call < __initcall_end; call++)
e0df154f 776 do_one_initcall(*call);
1da177e4
LT
777
778 /* Make sure there is no pending stuff from the initcall sequence */
779 flush_scheduled_work();
780}
781
782/*
783 * Ok, the machine is now initialized. None of the devices
784 * have been touched yet, but the CPU subsystem is up and
785 * running, and memory and process management works.
786 *
787 * Now we can finally start doing some real work..
788 */
789static void __init do_basic_setup(void)
790{
4446a36f 791 rcu_init_sched(); /* needed by module_init stage. */
4403b406 792 init_workqueues();
759ee091 793 cpuset_init_smp();
1da177e4
LT
794 usermodehelper_init();
795 driver_init();
b04c3afb 796 init_irq_proc();
1da177e4
LT
797 do_initcalls();
798}
799
7babe8db 800static void __init do_pre_smp_initcalls(void)
c2147a50
EGM
801{
802 initcall_t *call;
803
804 for (call = __initcall_start; call < __early_initcall_end; call++)
805 do_one_initcall(*call);
806}
807
1da177e4
LT
808static void run_init_process(char *init_filename)
809{
810 argv_init[0] = init_filename;
67608567 811 kernel_execve(init_filename, argv_init, envp_init);
1da177e4
LT
812}
813
ee5bfa64
VG
814/* This is a non __init function. Force it to be noinline otherwise gcc
815 * makes it inline to init() and it becomes part of init.text section
816 */
f99ebf0a 817static noinline int init_post(void)
acdd052a 818 __releases(kernel_lock)
ee5bfa64 819{
22a9d645
AV
820 /* need to finish all async __init code before freeing the memory */
821 async_synchronize_full();
ee5bfa64
VG
822 free_initmem();
823 unlock_kernel();
824 mark_rodata_ro();
825 system_state = SYSTEM_RUNNING;
826 numa_default_policy();
827
828 if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0)
829 printk(KERN_WARNING "Warning: unable to open an initial console.\n");
830
831 (void) sys_dup(0);
832 (void) sys_dup(0);
833
fae5fa44
ON
834 current->signal->flags |= SIGNAL_UNKILLABLE;
835
ee5bfa64
VG
836 if (ramdisk_execute_command) {
837 run_init_process(ramdisk_execute_command);
838 printk(KERN_WARNING "Failed to execute %s\n",
839 ramdisk_execute_command);
840 }
841
842 /*
843 * We try each of these until one succeeds.
844 *
845 * The Bourne shell can be used instead of init if we are
846 * trying to recover a really broken machine.
847 */
848 if (execute_command) {
849 run_init_process(execute_command);
850 printk(KERN_WARNING "Failed to execute %s. Attempting "
851 "defaults...\n", execute_command);
852 }
853 run_init_process("/sbin/init");
854 run_init_process("/etc/init");
855 run_init_process("/bin/init");
856 run_init_process("/bin/sh");
857
858 panic("No init found. Try passing init= option to kernel.");
859}
860
aae5f662 861static int __init kernel_init(void * unused)
1da177e4
LT
862{
863 lock_kernel();
864 /*
865 * init can run on any cpu.
866 */
1a2142af 867 set_cpus_allowed_ptr(current, cpu_all_mask);
1da177e4
LT
868 /*
869 * Tell the world that we're going to be the grim
870 * reaper of innocent orphaned children.
871 *
872 * We don't want people to have to make incorrect
873 * assumptions about where in the task array this
874 * can be found.
875 */
84d73786 876 init_pid_ns.child_reaper = current;
1da177e4 877
9ec52099
CLG
878 cad_pid = task_pid(current);
879
ca74a6f8 880 smp_prepare_cpus(setup_max_cpus);
1da177e4
LT
881
882 do_pre_smp_initcalls();
3bf77af6 883 start_boot_trace();
1da177e4 884
1da177e4
LT
885 smp_init();
886 sched_init_smp();
887
1da177e4
LT
888 do_basic_setup();
889
890 /*
891 * check if there is an early userspace init. If yes, let it do all
892 * the work
893 */
ffdfc409
OJ
894
895 if (!ramdisk_execute_command)
896 ramdisk_execute_command = "/init";
897
898 if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) {
899 ramdisk_execute_command = NULL;
1da177e4 900 prepare_namespace();
ffdfc409 901 }
1da177e4
LT
902
903 /*
904 * Ok, we have completed the initial bootup, and
905 * we're essentially up and running. Get rid of the
906 * initmem segments and start the user-mode stuff..
907 */
71566a0d 908
ee5bfa64
VG
909 init_post();
910 return 0;
1da177e4 911}