]> bbs.cooldavid.org Git - net-next-2.6.git/blame - init/main.c
drivers/misc: add driver for Texas Instruments DAC7512
[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>
1da177e4
LT
21#include <linux/ioport.h>
22#include <linux/init.h>
23#include <linux/smp_lock.h>
24#include <linux/initrd.h>
1da177e4 25#include <linux/bootmem.h>
4a7a16dc 26#include <linux/acpi.h>
1da177e4
LT
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>
dfec072e 68#include <linux/kmemcheck.h>
02af61bb 69#include <linux/kmemtrace.h>
6ae6996a 70#include <linux/sfi.h>
2b2af54a 71#include <linux/shmem_fs.h>
3f5ec136 72#include <trace/boot.h>
1da177e4
LT
73
74#include <asm/io.h>
75#include <asm/bugs.h>
76#include <asm/setup.h>
a940199f 77#include <asm/sections.h>
37b73c82 78#include <asm/cacheflush.h>
1da177e4 79
1da177e4
LT
80#ifdef CONFIG_X86_LOCAL_APIC
81#include <asm/smp.h>
82#endif
83
aae5f662 84static int kernel_init(void *);
1da177e4
LT
85
86extern void init_IRQ(void);
1da177e4
LT
87extern void fork_init(unsigned long);
88extern void mca_init(void);
89extern void sbus_init(void);
1da177e4
LT
90extern void prio_tree_init(void);
91extern void radix_tree_init(void);
92extern void free_initmem(void);
37b73c82
AV
93#ifndef CONFIG_DEBUG_RODATA
94static inline void mark_rodata_ro(void) { }
95#endif
1da177e4
LT
96
97#ifdef CONFIG_TC
98extern void tc_init(void);
99#endif
100
a6826048 101enum system_states system_state __read_mostly;
1da177e4
LT
102EXPORT_SYMBOL(system_state);
103
104/*
105 * Boot command-line arguments
106 */
107#define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT
108#define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT
109
110extern void time_init(void);
111/* Default late time init is NULL. archs can override this later. */
d2e3192b 112void (*__initdata late_time_init)(void);
1da177e4
LT
113extern void softirq_init(void);
114
30d7e0d4
ABL
115/* Untouched command line saved by arch-specific code. */
116char __initdata boot_command_line[COMMAND_LINE_SIZE];
117/* Untouched saved command line (eg. for /proc) */
118char *saved_command_line;
119/* Command line for parameter parsing */
120static char *static_command_line;
1da177e4
LT
121
122static char *execute_command;
ffdfc409 123static char *ramdisk_execute_command;
1da177e4 124
8b3b2955 125#ifdef CONFIG_SMP
1da177e4 126/* Setup configured maximum number of CPUs to activate */
ca74a6f8 127unsigned int __initdata setup_max_cpus = NR_CPUS;
7e96287d 128
1da177e4
LT
129/*
130 * Setup routine for controlling SMP activation
131 *
132 * Command-line option of "nosmp" or "maxcpus=0" will disable SMP
133 * activation entirely (the MPS table probe still happens, though).
134 *
135 * Command-line option of "maxcpus=<NUM>", where <NUM> is an integer
136 * greater than 0, limits the maximum number of CPUs activated in
137 * SMP mode to <NUM>.
138 */
65a4e574
IM
139
140void __weak arch_disable_smp_support(void) { }
61ec7567 141
1da177e4
LT
142static int __init nosmp(char *str)
143{
ca74a6f8 144 setup_max_cpus = 0;
65a4e574
IM
145 arch_disable_smp_support();
146
8b3b2955 147 return 0;
1da177e4
LT
148}
149
8b3b2955 150early_param("nosmp", nosmp);
1da177e4
LT
151
152static int __init maxcpus(char *str)
153{
ca74a6f8
AK
154 get_option(&str, &setup_max_cpus);
155 if (setup_max_cpus == 0)
65a4e574 156 arch_disable_smp_support();
61ec7567
LB
157
158 return 0;
1da177e4
LT
159}
160
81340977 161early_param("maxcpus", maxcpus);
8b3b2955 162#else
65a4e574 163const unsigned int setup_max_cpus = NR_CPUS;
8b3b2955
JB
164#endif
165
166/*
167 * If set, this is an indication to the drivers that reset the underlying
168 * device before going ahead with the initialization otherwise driver might
169 * rely on the BIOS and skip the reset operation.
170 *
171 * This is useful if kernel is booting in an unreliable environment.
172 * For ex. kdump situaiton where previous kernel has crashed, BIOS has been
173 * skipped and devices will be in unknown state.
174 */
175unsigned int reset_devices;
176EXPORT_SYMBOL(reset_devices);
1da177e4 177
7e96287d
VG
178static int __init set_reset_devices(char *str)
179{
180 reset_devices = 1;
181 return 1;
182}
183
184__setup("reset_devices", set_reset_devices);
185
1da177e4
LT
186static char * argv_init[MAX_INIT_ARGS+2] = { "init", NULL, };
187char * envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, };
188static const char *panic_later, *panic_param;
189
190extern struct obs_kernel_param __setup_start[], __setup_end[];
191
192static int __init obsolete_checksetup(char *line)
193{
194 struct obs_kernel_param *p;
33df0d19 195 int had_early_param = 0;
1da177e4
LT
196
197 p = __setup_start;
198 do {
199 int n = strlen(p->str);
200 if (!strncmp(line, p->str, n)) {
201 if (p->early) {
33df0d19
RR
202 /* Already done in parse_early_param?
203 * (Needs exact match on param part).
204 * Keep iterating, as we can have early
205 * params and __setups of same names 8( */
1da177e4 206 if (line[n] == '\0' || line[n] == '=')
33df0d19 207 had_early_param = 1;
1da177e4
LT
208 } else if (!p->setup_func) {
209 printk(KERN_WARNING "Parameter %s is obsolete,"
210 " ignored\n", p->str);
211 return 1;
212 } else if (p->setup_func(line + n))
213 return 1;
214 }
215 p++;
216 } while (p < __setup_end);
33df0d19
RR
217
218 return had_early_param;
1da177e4
LT
219}
220
221/*
222 * This should be approx 2 Bo*oMips to start (note initial shift), and will
223 * still work even if initially too large, it will just take slightly longer
224 */
225unsigned long loops_per_jiffy = (1<<12);
226
227EXPORT_SYMBOL(loops_per_jiffy);
228
229static int __init debug_kernel(char *str)
230{
1da177e4 231 console_loglevel = 10;
f6f21c81 232 return 0;
1da177e4
LT
233}
234
235static int __init quiet_kernel(char *str)
236{
1da177e4 237 console_loglevel = 4;
f6f21c81 238 return 0;
1da177e4
LT
239}
240
f6f21c81
YL
241early_param("debug", debug_kernel);
242early_param("quiet", quiet_kernel);
1da177e4
LT
243
244static int __init loglevel(char *str)
245{
246 get_option(&str, &console_loglevel);
d9d4fcfe 247 return 0;
1da177e4
LT
248}
249
f6f21c81 250early_param("loglevel", loglevel);
1da177e4
LT
251
252/*
253 * Unknown boot options get handed to init, unless they look like
f066a4f6 254 * unused parameters (modprobe will find them in /proc/cmdline).
1da177e4
LT
255 */
256static int __init unknown_bootoption(char *param, char *val)
257{
258 /* Change NUL term back to "=", to make "param" the whole string. */
259 if (val) {
260 /* param=val or param="val"? */
261 if (val == param+strlen(param)+1)
262 val[-1] = '=';
263 else if (val == param+strlen(param)+2) {
264 val[-2] = '=';
265 memmove(val-1, val, strlen(val)+1);
266 val--;
267 } else
268 BUG();
269 }
270
271 /* Handle obsolete-style parameters */
272 if (obsolete_checksetup(param))
273 return 0;
274
f066a4f6
RR
275 /* Unused module parameter. */
276 if (strchr(param, '.') && (!val || strchr(param, '.') < val))
1da177e4 277 return 0;
1da177e4
LT
278
279 if (panic_later)
280 return 0;
281
282 if (val) {
283 /* Environment option */
284 unsigned int i;
285 for (i = 0; envp_init[i]; i++) {
286 if (i == MAX_INIT_ENVS) {
287 panic_later = "Too many boot env vars at `%s'";
288 panic_param = param;
289 }
290 if (!strncmp(param, envp_init[i], val - param))
291 break;
292 }
293 envp_init[i] = param;
294 } else {
295 /* Command line option */
296 unsigned int i;
297 for (i = 0; argv_init[i]; i++) {
298 if (i == MAX_INIT_ARGS) {
299 panic_later = "Too many boot init vars at `%s'";
300 panic_param = param;
301 }
302 }
303 argv_init[i] = param;
304 }
305 return 0;
306}
307
12d6f21e
IM
308#ifdef CONFIG_DEBUG_PAGEALLOC
309int __read_mostly debug_pagealloc_enabled = 0;
310#endif
311
1da177e4
LT
312static int __init init_setup(char *str)
313{
314 unsigned int i;
315
316 execute_command = str;
317 /*
318 * In case LILO is going to boot us with default command line,
319 * it prepends "auto" before the whole cmdline which makes
320 * the shell think it should execute a script with such name.
321 * So we ignore all arguments entered _before_ init=... [MJ]
322 */
323 for (i = 1; i < MAX_INIT_ARGS; i++)
324 argv_init[i] = NULL;
325 return 1;
326}
327__setup("init=", init_setup);
328
ffdfc409
OJ
329static int __init rdinit_setup(char *str)
330{
331 unsigned int i;
332
333 ramdisk_execute_command = str;
334 /* See "auto" comment in init_setup */
335 for (i = 1; i < MAX_INIT_ARGS; i++)
336 argv_init[i] = NULL;
337 return 1;
338}
339__setup("rdinit=", rdinit_setup);
340
1da177e4
LT
341#ifndef CONFIG_SMP
342
343#ifdef CONFIG_X86_LOCAL_APIC
344static void __init smp_init(void)
345{
346 APIC_init_uniprocessor();
347}
348#else
349#define smp_init() do { } while (0)
350#endif
351
e0982e90 352static inline void setup_nr_cpu_ids(void) { }
1da177e4
LT
353static inline void smp_prepare_cpus(unsigned int maxcpus) { }
354
355#else
356
e0982e90
MT
357/* Setup number of possible processor ids */
358int nr_cpu_ids __read_mostly = NR_CPUS;
359EXPORT_SYMBOL(nr_cpu_ids);
360
361/* An arch may set nr_cpu_ids earlier if needed, so this would be redundant */
362static void __init setup_nr_cpu_ids(void)
363{
e0c0ba73 364 nr_cpu_ids = find_last_bit(cpumask_bits(cpu_possible_mask),NR_CPUS) + 1;
e0982e90
MT
365}
366
1da177e4
LT
367/* Called by boot processor to activate the rest. */
368static void __init smp_init(void)
369{
53b8a315 370 unsigned int cpu;
1da177e4 371
e761b772
MK
372 /*
373 * Set up the current CPU as possible to migrate to.
374 * The other ones will be done by cpu_up/cpu_down()
375 */
2b17fa50 376 set_cpu_active(smp_processor_id(), true);
e761b772 377
1da177e4 378 /* FIXME: This should be done in userspace --RR */
53b8a315 379 for_each_present_cpu(cpu) {
ca74a6f8 380 if (num_online_cpus() >= setup_max_cpus)
1da177e4 381 break;
53b8a315
CL
382 if (!cpu_online(cpu))
383 cpu_up(cpu);
1da177e4
LT
384 }
385
386 /* Any cleanup work */
387 printk(KERN_INFO "Brought up %ld CPUs\n", (long)num_online_cpus());
ca74a6f8 388 smp_cpus_done(setup_max_cpus);
1da177e4
LT
389}
390
391#endif
392
30d7e0d4
ABL
393/*
394 * We need to store the untouched command line for future reference.
395 * We also need to store the touched command line since the parameter
396 * parsing is performed in place, and we should allow a component to
397 * store reference of name/value for future reference.
398 */
399static void __init setup_command_line(char *command_line)
400{
401 saved_command_line = alloc_bootmem(strlen (boot_command_line)+1);
402 static_command_line = alloc_bootmem(strlen (command_line)+1);
403 strcpy (saved_command_line, boot_command_line);
404 strcpy (static_command_line, command_line);
405}
406
1da177e4
LT
407/*
408 * We need to finalize in a non-__init function or else race conditions
409 * between the root thread and the init thread may cause start_kernel to
410 * be reaped by free_initmem before the root thread has proceeded to
411 * cpu_idle.
412 *
413 * gcc-3.4 accidentally inlines this function, so use noinline.
414 */
415
f99ebf0a 416static noinline void __init_refok rest_init(void)
1da177e4
LT
417 __releases(kernel_lock)
418{
73c27992
EB
419 int pid;
420
7db905e6 421 rcu_scheduler_starting();
aae5f662 422 kernel_thread(kernel_init, NULL, CLONE_FS | CLONE_SIGHAND);
1da177e4 423 numa_default_policy();
73c27992 424 pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES);
5cd20455 425 kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns);
1da177e4 426 unlock_kernel();
f340c0d1
IM
427
428 /*
429 * The boot idle thread must execute schedule()
1df21055 430 * at least once to get things moving:
f340c0d1 431 */
1df21055 432 init_idle_bootup_task(current);
5bfb5d69 433 preempt_enable_no_resched();
f340c0d1 434 schedule();
5bfb5d69 435 preempt_disable();
f340c0d1 436
5bfb5d69 437 /* Call into cpu_idle with preempt disabled */
1da177e4 438 cpu_idle();
1df21055 439}
1da177e4
LT
440
441/* Check for early params. */
442static int __init do_early_param(char *param, char *val)
443{
444 struct obs_kernel_param *p;
445
446 for (p = __setup_start; p < __setup_end; p++) {
18a8bd94
YL
447 if ((p->early && strcmp(param, p->str) == 0) ||
448 (strcmp(param, "console") == 0 &&
449 strcmp(p->str, "earlycon") == 0)
450 ) {
1da177e4
LT
451 if (p->setup_func(val) != 0)
452 printk(KERN_WARNING
453 "Malformed early option '%s'\n", param);
454 }
455 }
456 /* We accept everything at this stage. */
457 return 0;
458}
459
13977091
MD
460void __init parse_early_options(char *cmdline)
461{
462 parse_args("early options", cmdline, NULL, 0, do_early_param);
463}
464
1da177e4
LT
465/* Arch code calls this early on, or if not, just before other parsing. */
466void __init parse_early_param(void)
467{
468 static __initdata int done = 0;
469 static __initdata char tmp_cmdline[COMMAND_LINE_SIZE];
470
471 if (done)
472 return;
473
474 /* All fall through to do_early_param. */
30d7e0d4 475 strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE);
13977091 476 parse_early_options(tmp_cmdline);
1da177e4
LT
477 done = 1;
478}
479
480/*
481 * Activate the first processor.
482 */
483
44fd2299
SS
484static void __init boot_cpu_init(void)
485{
486 int cpu = smp_processor_id();
487 /* Mark the boot cpu "present", "online" etc for SMP and UP case */
915441b6
RR
488 set_cpu_online(cpu, true);
489 set_cpu_present(cpu, true);
490 set_cpu_possible(cpu, true);
44fd2299
SS
491}
492
839ad62e 493void __init __weak smp_setup_processor_id(void)
033ab7f8
AM
494{
495}
496
8c9843e5
BH
497void __init __weak thread_info_cache_init(void)
498{
499}
500
444f478f
PE
501/*
502 * Set up kernel memory allocators
503 */
504static void __init mm_init(void)
505{
ca371c0d
KH
506 /*
507 * page_cgroup requires countinous pages as memmap
508 * and it's bigger than MAX_ORDER unless SPARSEMEM.
509 */
510 page_cgroup_init_flatmem();
444f478f
PE
511 mem_init();
512 kmem_cache_init();
c868d550 513 pgtable_cache_init();
444f478f
PE
514 vmalloc_init();
515}
516
1da177e4
LT
517asmlinkage void __init start_kernel(void)
518{
519 char * command_line;
520 extern struct kernel_param __start___param[], __stop___param[];
033ab7f8
AM
521
522 smp_setup_processor_id();
523
fbb9ce95
IM
524 /*
525 * Need to run as early as possible, to initialize the
526 * lockdep hash:
527 */
528 lockdep_init();
3ac7fe5a 529 debug_objects_early_init();
42059429
IM
530
531 /*
532 * Set up the the initial canary ASAP:
533 */
534 boot_init_stack_canary();
535
ddbcc7e8 536 cgroup_init_early();
fbb9ce95
IM
537
538 local_irq_disable();
539 early_boot_irqs_off();
243c7621 540 early_init_irq_lock_class();
fbb9ce95 541
1da177e4
LT
542/*
543 * Interrupts are still disabled. Do necessary setups, then
544 * enable them
545 */
546 lock_kernel();
906568c9 547 tick_init();
44fd2299 548 boot_cpu_init();
1da177e4 549 page_address_init();
657cafa6 550 printk(KERN_NOTICE "%s", linux_banner);
1da177e4 551 setup_arch(&command_line);
cf475ad2 552 mm_init_owner(&init_mm, &init_task);
30d7e0d4 553 setup_command_line(command_line);
e0982e90 554 setup_nr_cpu_ids();
d6647bdf 555 setup_per_cpu_areas();
44fd2299 556 smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */
1da177e4 557
83b519e8
PE
558 build_all_zonelists();
559 page_alloc_init();
560
561 printk(KERN_NOTICE "Kernel command line: %s\n", boot_command_line);
562 parse_early_param();
563 parse_args("Booting kernel", static_command_line, __start___param,
564 __stop___param - __start___param,
565 &unknown_bootoption);
566 /*
567 * These use large bootmem allocations and must precede
568 * kmem_cache_init()
569 */
570 pidhash_init();
83b519e8
PE
571 vfs_caches_init_early();
572 sort_main_extable();
573 trap_init();
444f478f 574 mm_init();
1da177e4
LT
575 /*
576 * Set up the scheduler prior starting any interrupts (such as the
577 * timer interrupt). Full topology setup happens at smp_init()
578 * time - but meanwhile we still have a functioning scheduler.
579 */
580 sched_init();
581 /*
582 * Disable preemption - early bootup scheduling is extremely
583 * fragile until we cpu_idle() for the first time.
584 */
585 preempt_disable();
c4a68306
AB
586 if (!irqs_disabled()) {
587 printk(KERN_WARNING "start_kernel(): bug: interrupts were "
588 "enabled *very* early, fixing it\n");
589 local_irq_disable();
590 }
1da177e4 591 rcu_init();
0b8f1efa
YL
592 /* init some links before init_ISA_irqs() */
593 early_irq_init();
1da177e4 594 init_IRQ();
3c7b4e6b 595 prio_tree_init();
1da177e4 596 init_timers();
c0a31329 597 hrtimers_init();
1da177e4 598 softirq_init();
ad596171 599 timekeeping_init();
88fecaa2 600 time_init();
93e02814
HC
601 profile_init();
602 if (!irqs_disabled())
24d431d0
RL
603 printk(KERN_CRIT "start_kernel(): bug: interrupts were "
604 "enabled early\n");
fbb9ce95 605 early_boot_irqs_on();
93e02814 606 local_irq_enable();
dcce284a
BH
607
608 /* Interrupts are enabled now so all GFP allocations are safe. */
609 set_gfp_allowed_mask(__GFP_BITS_MASK);
610
7e85ee0c 611 kmem_cache_init_late();
1da177e4
LT
612
613 /*
614 * HACK ALERT! This is early. We're enabling the console before
615 * we've done PCI setups etc, and console_init() must be aware of
616 * this. But we do want output early, in case something goes wrong.
617 */
618 console_init();
619 if (panic_later)
620 panic(panic_later, panic_param);
fbb9ce95
IM
621
622 lockdep_info();
623
9a11b49a
IM
624 /*
625 * Need to run this when irqs are enabled, because it wants
626 * to self-test [hard/soft]-irqs on/off lock inversion bugs
627 * too:
628 */
629 locking_selftest();
630
1da177e4
LT
631#ifdef CONFIG_BLK_DEV_INITRD
632 if (initrd_start && !initrd_below_start_ok &&
bd673c7c 633 page_to_pfn(virt_to_page((void *)initrd_start)) < min_low_pfn) {
1da177e4 634 printk(KERN_CRIT "initrd overwritten (0x%08lx < 0x%08lx) - "
fb6624eb 635 "disabling it.\n",
bd673c7c
GU
636 page_to_pfn(virt_to_page((void *)initrd_start)),
637 min_low_pfn);
1da177e4
LT
638 initrd_start = 0;
639 }
640#endif
94b6da5a 641 page_cgroup_init();
a03c2a48 642 enable_debug_pagealloc();
b9ce08c0 643 kmemtrace_init();
3c7b4e6b 644 kmemleak_init();
3ac7fe5a 645 debug_objects_mem_init();
199f0ca5 646 idr_init_cache();
e7c8d5c9 647 setup_per_cpu_pageset();
1da177e4
LT
648 numa_policy_init();
649 if (late_time_init)
650 late_time_init();
fa84e9ee 651 sched_clock_init();
1da177e4
LT
652 calibrate_delay();
653 pidmap_init();
1da177e4
LT
654 anon_vma_init();
655#ifdef CONFIG_X86
656 if (efi_enabled)
657 efi_enter_virtual_mode();
658#endif
8c9843e5 659 thread_info_cache_init();
d84f4f99 660 cred_init();
4481374c 661 fork_init(totalram_pages);
1da177e4
LT
662 proc_caches_init();
663 buffer_init();
1da177e4
LT
664 key_init();
665 security_init();
4481374c 666 vfs_caches_init(totalram_pages);
1da177e4
LT
667 radix_tree_init();
668 signals_init();
669 /* rootfs populating might need page-writeback */
670 page_writeback_init();
671#ifdef CONFIG_PROC_FS
672 proc_root_init();
673#endif
ddbcc7e8 674 cgroup_init();
1da177e4 675 cpuset_init();
c757249a 676 taskstats_init_early();
ca74e92b 677 delayacct_init();
1da177e4
LT
678
679 check_bugs();
680
681 acpi_early_init(); /* before LAPIC and SMP init */
6ae6996a 682 sfi_init_late();
1da177e4 683
68bf21aa
SR
684 ftrace_init();
685
1da177e4
LT
686 /* Do the rest non-__init'ed, we're now alive */
687 rest_init();
688}
689
b99b87f7
PO
690/* Call all constructor functions linked into the kernel. */
691static void __init do_ctors(void)
692{
693#ifdef CONFIG_CONSTRUCTORS
694 ctor_fn_t *call = (ctor_fn_t *) __ctors_start;
695
696 for (; call < (ctor_fn_t *) __ctors_end; call++)
697 (*call)();
698#endif
699}
700
22a9d645 701int initcall_debug;
d0ea3d7d 702core_param(initcall_debug, initcall_debug, bool, 0644);
1da177e4 703
4a683bf9
IM
704static char msgbuf[64];
705static struct boot_trace_call call;
706static struct boot_trace_ret ret;
707
59f9415f 708int do_one_initcall(initcall_t fn)
1da177e4 709{
1da177e4 710 int count = preempt_count();
74239072 711 ktime_t calltime, delta, rettime;
1da177e4 712
e0df154f 713 if (initcall_debug) {
74239072
FW
714 call.caller = task_pid_nr(current);
715 printk("calling %pF @ %i\n", fn, call.caller);
716 calltime = ktime_get();
717 trace_boot_call(&call, fn);
71566a0d 718 enable_boot_trace();
e0df154f 719 }
1da177e4 720
74239072 721 ret.result = fn();
1da177e4 722
e0df154f 723 if (initcall_debug) {
71566a0d 724 disable_boot_trace();
74239072
FW
725 rettime = ktime_get();
726 delta = ktime_sub(rettime, calltime);
1d926f27 727 ret.duration = (unsigned long long) ktime_to_ns(delta) >> 10;
74239072 728 trace_boot_ret(&ret, fn);
ca538f6b 729 printk("initcall %pF returned %d after %Ld usecs\n", fn,
74239072 730 ret.result, ret.duration);
e0df154f 731 }
8f0c45cd 732
e0df154f 733 msgbuf[0] = 0;
e662e1cf 734
74239072
FW
735 if (ret.result && ret.result != -ENODEV && initcall_debug)
736 sprintf(msgbuf, "error code %d ", ret.result);
e662e1cf 737
e0df154f 738 if (preempt_count() != count) {
a76bfd0d 739 strlcat(msgbuf, "preemption imbalance ", sizeof(msgbuf));
e0df154f 740 preempt_count() = count;
1da177e4 741 }
e0df154f 742 if (irqs_disabled()) {
a76bfd0d 743 strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf));
e0df154f
LT
744 local_irq_enable();
745 }
746 if (msgbuf[0]) {
96d746c6 747 printk("initcall %pF returned with %s\n", fn, msgbuf);
e0df154f 748 }
59f9415f 749
74239072 750 return ret.result;
e0df154f
LT
751}
752
753
c2147a50 754extern initcall_t __initcall_start[], __initcall_end[], __early_initcall_end[];
e0df154f
LT
755
756static void __init do_initcalls(void)
757{
758 initcall_t *call;
759
c2147a50 760 for (call = __early_initcall_end; call < __initcall_end; call++)
e0df154f 761 do_one_initcall(*call);
1da177e4
LT
762
763 /* Make sure there is no pending stuff from the initcall sequence */
764 flush_scheduled_work();
765}
766
767/*
768 * Ok, the machine is now initialized. None of the devices
769 * have been touched yet, but the CPU subsystem is up and
770 * running, and memory and process management works.
771 *
772 * Now we can finally start doing some real work..
773 */
774static void __init do_basic_setup(void)
775{
4403b406 776 init_workqueues();
759ee091 777 cpuset_init_smp();
1da177e4 778 usermodehelper_init();
2b2af54a 779 init_tmpfs();
1da177e4 780 driver_init();
b04c3afb 781 init_irq_proc();
b99b87f7 782 do_ctors();
1da177e4
LT
783 do_initcalls();
784}
785
7babe8db 786static void __init do_pre_smp_initcalls(void)
c2147a50
EGM
787{
788 initcall_t *call;
789
790 for (call = __initcall_start; call < __early_initcall_end; call++)
791 do_one_initcall(*call);
792}
793
1da177e4
LT
794static void run_init_process(char *init_filename)
795{
796 argv_init[0] = init_filename;
67608567 797 kernel_execve(init_filename, argv_init, envp_init);
1da177e4
LT
798}
799
ee5bfa64
VG
800/* This is a non __init function. Force it to be noinline otherwise gcc
801 * makes it inline to init() and it becomes part of init.text section
802 */
f99ebf0a 803static noinline int init_post(void)
acdd052a 804 __releases(kernel_lock)
ee5bfa64 805{
22a9d645
AV
806 /* need to finish all async __init code before freeing the memory */
807 async_synchronize_full();
ee5bfa64
VG
808 free_initmem();
809 unlock_kernel();
810 mark_rodata_ro();
811 system_state = SYSTEM_RUNNING;
812 numa_default_policy();
813
814 if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0)
815 printk(KERN_WARNING "Warning: unable to open an initial console.\n");
816
817 (void) sys_dup(0);
818 (void) sys_dup(0);
819
fae5fa44
ON
820 current->signal->flags |= SIGNAL_UNKILLABLE;
821
ee5bfa64
VG
822 if (ramdisk_execute_command) {
823 run_init_process(ramdisk_execute_command);
824 printk(KERN_WARNING "Failed to execute %s\n",
825 ramdisk_execute_command);
826 }
827
828 /*
829 * We try each of these until one succeeds.
830 *
831 * The Bourne shell can be used instead of init if we are
832 * trying to recover a really broken machine.
833 */
834 if (execute_command) {
835 run_init_process(execute_command);
836 printk(KERN_WARNING "Failed to execute %s. Attempting "
837 "defaults...\n", execute_command);
838 }
839 run_init_process("/sbin/init");
840 run_init_process("/etc/init");
841 run_init_process("/bin/init");
842 run_init_process("/bin/sh");
843
844 panic("No init found. Try passing init= option to kernel.");
845}
846
aae5f662 847static int __init kernel_init(void * unused)
1da177e4
LT
848{
849 lock_kernel();
58568d2a
MX
850
851 /*
852 * init can allocate pages on any node
853 */
854 set_mems_allowed(node_possible_map);
1da177e4
LT
855 /*
856 * init can run on any cpu.
857 */
1a2142af 858 set_cpus_allowed_ptr(current, cpu_all_mask);
1da177e4
LT
859 /*
860 * Tell the world that we're going to be the grim
861 * reaper of innocent orphaned children.
862 *
863 * We don't want people to have to make incorrect
864 * assumptions about where in the task array this
865 * can be found.
866 */
84d73786 867 init_pid_ns.child_reaper = current;
1da177e4 868
9ec52099
CLG
869 cad_pid = task_pid(current);
870
ca74a6f8 871 smp_prepare_cpus(setup_max_cpus);
1da177e4
LT
872
873 do_pre_smp_initcalls();
3bf77af6 874 start_boot_trace();
1da177e4 875
1da177e4
LT
876 smp_init();
877 sched_init_smp();
878
1da177e4
LT
879 do_basic_setup();
880
881 /*
882 * check if there is an early userspace init. If yes, let it do all
883 * the work
884 */
ffdfc409
OJ
885
886 if (!ramdisk_execute_command)
887 ramdisk_execute_command = "/init";
888
889 if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) {
890 ramdisk_execute_command = NULL;
1da177e4 891 prepare_namespace();
ffdfc409 892 }
1da177e4
LT
893
894 /*
895 * Ok, we have completed the initial bootup, and
896 * we're essentially up and running. Get rid of the
897 * initmem segments and start the user-mode stuff..
898 */
71566a0d 899
ee5bfa64
VG
900 init_post();
901 return 0;
1da177e4 902}