]> bbs.cooldavid.org Git - net-next-2.6.git/blame - init/main.c
xps: Transmit Packet Steering
[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 27#include <linux/tty.h>
1da177e4
LT
28#include <linux/percpu.h>
29#include <linux/kmod.h>
db64fe02 30#include <linux/vmalloc.h>
1da177e4 31#include <linux/kernel_stat.h>
d7cd5611 32#include <linux/start_kernel.h>
1da177e4 33#include <linux/security.h>
3d442233 34#include <linux/smp.h>
1da177e4
LT
35#include <linux/profile.h>
36#include <linux/rcupdate.h>
37#include <linux/moduleparam.h>
38#include <linux/kallsyms.h>
39#include <linux/writeback.h>
40#include <linux/cpu.h>
41#include <linux/cpuset.h>
ddbcc7e8 42#include <linux/cgroup.h>
1da177e4 43#include <linux/efi.h>
906568c9 44#include <linux/tick.h>
6168a702 45#include <linux/interrupt.h>
c757249a 46#include <linux/taskstats_kern.h>
ca74e92b 47#include <linux/delayacct.h>
1da177e4
LT
48#include <linux/unistd.h>
49#include <linux/rmap.h>
50#include <linux/mempolicy.h>
51#include <linux/key.h>
b6cd0b77 52#include <linux/buffer_head.h>
94b6da5a 53#include <linux/page_cgroup.h>
9a11b49a 54#include <linux/debug_locks.h>
3ac7fe5a 55#include <linux/debugobjects.h>
fbb9ce95 56#include <linux/lockdep.h>
3c7b4e6b 57#include <linux/kmemleak.h>
84d73786 58#include <linux/pid_namespace.h>
1f21782e 59#include <linux/device.h>
73c27992 60#include <linux/kthread.h>
e6fe6649 61#include <linux/sched.h>
a1c9eea9 62#include <linux/signal.h>
199f0ca5 63#include <linux/idr.h>
0b4b3827 64#include <linux/kgdb.h>
68bf21aa 65#include <linux/ftrace.h>
22a9d645 66#include <linux/async.h>
dfec072e 67#include <linux/kmemcheck.h>
6ae6996a 68#include <linux/sfi.h>
2b2af54a 69#include <linux/shmem_fs.h>
5a0e3ad6 70#include <linux/slab.h>
1da177e4
LT
71
72#include <asm/io.h>
73#include <asm/bugs.h>
74#include <asm/setup.h>
a940199f 75#include <asm/sections.h>
37b73c82 76#include <asm/cacheflush.h>
1da177e4 77
1da177e4
LT
78#ifdef CONFIG_X86_LOCAL_APIC
79#include <asm/smp.h>
80#endif
81
aae5f662 82static int kernel_init(void *);
1da177e4
LT
83
84extern void init_IRQ(void);
1da177e4
LT
85extern void fork_init(unsigned long);
86extern void mca_init(void);
87extern void sbus_init(void);
1da177e4
LT
88extern void prio_tree_init(void);
89extern void radix_tree_init(void);
90extern void free_initmem(void);
37b73c82
AV
91#ifndef CONFIG_DEBUG_RODATA
92static inline void mark_rodata_ro(void) { }
93#endif
1da177e4
LT
94
95#ifdef CONFIG_TC
96extern void tc_init(void);
97#endif
98
a6826048 99enum system_states system_state __read_mostly;
1da177e4
LT
100EXPORT_SYMBOL(system_state);
101
102/*
103 * Boot command-line arguments
104 */
105#define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT
106#define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT
107
108extern void time_init(void);
109/* Default late time init is NULL. archs can override this later. */
d2e3192b 110void (*__initdata late_time_init)(void);
1da177e4
LT
111extern void softirq_init(void);
112
30d7e0d4
ABL
113/* Untouched command line saved by arch-specific code. */
114char __initdata boot_command_line[COMMAND_LINE_SIZE];
115/* Untouched saved command line (eg. for /proc) */
116char *saved_command_line;
117/* Command line for parameter parsing */
118static char *static_command_line;
1da177e4
LT
119
120static char *execute_command;
ffdfc409 121static char *ramdisk_execute_command;
1da177e4 122
8b3b2955 123#ifdef CONFIG_SMP
1da177e4 124/* Setup configured maximum number of CPUs to activate */
75cbfb97
TR
125unsigned int setup_max_cpus = NR_CPUS;
126EXPORT_SYMBOL(setup_max_cpus);
127
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 151
2b633e3f
YL
152/* this is hard limit */
153static int __init nrcpus(char *str)
154{
155 int nr_cpus;
156
157 get_option(&str, &nr_cpus);
158 if (nr_cpus > 0 && nr_cpus < nr_cpu_ids)
159 nr_cpu_ids = nr_cpus;
160
161 return 0;
162}
163
164early_param("nr_cpus", nrcpus);
165
1da177e4
LT
166static int __init maxcpus(char *str)
167{
ca74a6f8
AK
168 get_option(&str, &setup_max_cpus);
169 if (setup_max_cpus == 0)
65a4e574 170 arch_disable_smp_support();
61ec7567
LB
171
172 return 0;
1da177e4
LT
173}
174
81340977 175early_param("maxcpus", maxcpus);
8b3b2955 176#else
7463e633 177static const unsigned int setup_max_cpus = NR_CPUS;
8b3b2955
JB
178#endif
179
180/*
181 * If set, this is an indication to the drivers that reset the underlying
182 * device before going ahead with the initialization otherwise driver might
183 * rely on the BIOS and skip the reset operation.
184 *
185 * This is useful if kernel is booting in an unreliable environment.
186 * For ex. kdump situaiton where previous kernel has crashed, BIOS has been
187 * skipped and devices will be in unknown state.
188 */
189unsigned int reset_devices;
190EXPORT_SYMBOL(reset_devices);
1da177e4 191
7e96287d
VG
192static int __init set_reset_devices(char *str)
193{
194 reset_devices = 1;
195 return 1;
196}
197
198__setup("reset_devices", set_reset_devices);
199
d7627467
DH
200static const char * argv_init[MAX_INIT_ARGS+2] = { "init", NULL, };
201const char * envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, };
1da177e4
LT
202static const char *panic_later, *panic_param;
203
914dcaa8 204extern const struct obs_kernel_param __setup_start[], __setup_end[];
1da177e4
LT
205
206static int __init obsolete_checksetup(char *line)
207{
914dcaa8 208 const struct obs_kernel_param *p;
33df0d19 209 int had_early_param = 0;
1da177e4
LT
210
211 p = __setup_start;
212 do {
213 int n = strlen(p->str);
214 if (!strncmp(line, p->str, n)) {
215 if (p->early) {
33df0d19
RR
216 /* Already done in parse_early_param?
217 * (Needs exact match on param part).
218 * Keep iterating, as we can have early
219 * params and __setups of same names 8( */
1da177e4 220 if (line[n] == '\0' || line[n] == '=')
33df0d19 221 had_early_param = 1;
1da177e4
LT
222 } else if (!p->setup_func) {
223 printk(KERN_WARNING "Parameter %s is obsolete,"
224 " ignored\n", p->str);
225 return 1;
226 } else if (p->setup_func(line + n))
227 return 1;
228 }
229 p++;
230 } while (p < __setup_end);
33df0d19
RR
231
232 return had_early_param;
1da177e4
LT
233}
234
235/*
236 * This should be approx 2 Bo*oMips to start (note initial shift), and will
237 * still work even if initially too large, it will just take slightly longer
238 */
239unsigned long loops_per_jiffy = (1<<12);
240
241EXPORT_SYMBOL(loops_per_jiffy);
242
243static int __init debug_kernel(char *str)
244{
1da177e4 245 console_loglevel = 10;
f6f21c81 246 return 0;
1da177e4
LT
247}
248
249static int __init quiet_kernel(char *str)
250{
1da177e4 251 console_loglevel = 4;
f6f21c81 252 return 0;
1da177e4
LT
253}
254
f6f21c81
YL
255early_param("debug", debug_kernel);
256early_param("quiet", quiet_kernel);
1da177e4
LT
257
258static int __init loglevel(char *str)
259{
260 get_option(&str, &console_loglevel);
d9d4fcfe 261 return 0;
1da177e4
LT
262}
263
f6f21c81 264early_param("loglevel", loglevel);
1da177e4
LT
265
266/*
267 * Unknown boot options get handed to init, unless they look like
f066a4f6 268 * unused parameters (modprobe will find them in /proc/cmdline).
1da177e4
LT
269 */
270static int __init unknown_bootoption(char *param, char *val)
271{
272 /* Change NUL term back to "=", to make "param" the whole string. */
273 if (val) {
274 /* param=val or param="val"? */
275 if (val == param+strlen(param)+1)
276 val[-1] = '=';
277 else if (val == param+strlen(param)+2) {
278 val[-2] = '=';
279 memmove(val-1, val, strlen(val)+1);
280 val--;
281 } else
282 BUG();
283 }
284
285 /* Handle obsolete-style parameters */
286 if (obsolete_checksetup(param))
287 return 0;
288
f066a4f6
RR
289 /* Unused module parameter. */
290 if (strchr(param, '.') && (!val || strchr(param, '.') < val))
1da177e4 291 return 0;
1da177e4
LT
292
293 if (panic_later)
294 return 0;
295
296 if (val) {
297 /* Environment option */
298 unsigned int i;
299 for (i = 0; envp_init[i]; i++) {
300 if (i == MAX_INIT_ENVS) {
301 panic_later = "Too many boot env vars at `%s'";
302 panic_param = param;
303 }
304 if (!strncmp(param, envp_init[i], val - param))
305 break;
306 }
307 envp_init[i] = param;
308 } else {
309 /* Command line option */
310 unsigned int i;
311 for (i = 0; argv_init[i]; i++) {
312 if (i == MAX_INIT_ARGS) {
313 panic_later = "Too many boot init vars at `%s'";
314 panic_param = param;
315 }
316 }
317 argv_init[i] = param;
318 }
319 return 0;
320}
321
12d6f21e
IM
322#ifdef CONFIG_DEBUG_PAGEALLOC
323int __read_mostly debug_pagealloc_enabled = 0;
324#endif
325
1da177e4
LT
326static int __init init_setup(char *str)
327{
328 unsigned int i;
329
330 execute_command = str;
331 /*
332 * In case LILO is going to boot us with default command line,
333 * it prepends "auto" before the whole cmdline which makes
334 * the shell think it should execute a script with such name.
335 * So we ignore all arguments entered _before_ init=... [MJ]
336 */
337 for (i = 1; i < MAX_INIT_ARGS; i++)
338 argv_init[i] = NULL;
339 return 1;
340}
341__setup("init=", init_setup);
342
ffdfc409
OJ
343static int __init rdinit_setup(char *str)
344{
345 unsigned int i;
346
347 ramdisk_execute_command = str;
348 /* See "auto" comment in init_setup */
349 for (i = 1; i < MAX_INIT_ARGS; i++)
350 argv_init[i] = NULL;
351 return 1;
352}
353__setup("rdinit=", rdinit_setup);
354
1da177e4
LT
355#ifndef CONFIG_SMP
356
357#ifdef CONFIG_X86_LOCAL_APIC
358static void __init smp_init(void)
359{
360 APIC_init_uniprocessor();
361}
362#else
363#define smp_init() do { } while (0)
364#endif
365
e0982e90 366static inline void setup_nr_cpu_ids(void) { }
1da177e4
LT
367static inline void smp_prepare_cpus(unsigned int maxcpus) { }
368
369#else
370
e0982e90
MT
371/* Setup number of possible processor ids */
372int nr_cpu_ids __read_mostly = NR_CPUS;
373EXPORT_SYMBOL(nr_cpu_ids);
374
375/* An arch may set nr_cpu_ids earlier if needed, so this would be redundant */
376static void __init setup_nr_cpu_ids(void)
377{
e0c0ba73 378 nr_cpu_ids = find_last_bit(cpumask_bits(cpu_possible_mask),NR_CPUS) + 1;
e0982e90
MT
379}
380
1da177e4
LT
381/* Called by boot processor to activate the rest. */
382static void __init smp_init(void)
383{
53b8a315 384 unsigned int cpu;
1da177e4
LT
385
386 /* FIXME: This should be done in userspace --RR */
53b8a315 387 for_each_present_cpu(cpu) {
ca74a6f8 388 if (num_online_cpus() >= setup_max_cpus)
1da177e4 389 break;
53b8a315
CL
390 if (!cpu_online(cpu))
391 cpu_up(cpu);
1da177e4
LT
392 }
393
394 /* Any cleanup work */
395 printk(KERN_INFO "Brought up %ld CPUs\n", (long)num_online_cpus());
ca74a6f8 396 smp_cpus_done(setup_max_cpus);
1da177e4
LT
397}
398
399#endif
400
30d7e0d4
ABL
401/*
402 * We need to store the untouched command line for future reference.
403 * We also need to store the touched command line since the parameter
404 * parsing is performed in place, and we should allow a component to
405 * store reference of name/value for future reference.
406 */
407static void __init setup_command_line(char *command_line)
408{
409 saved_command_line = alloc_bootmem(strlen (boot_command_line)+1);
410 static_command_line = alloc_bootmem(strlen (command_line)+1);
411 strcpy (saved_command_line, boot_command_line);
412 strcpy (static_command_line, command_line);
413}
414
1da177e4
LT
415/*
416 * We need to finalize in a non-__init function or else race conditions
417 * between the root thread and the init thread may cause start_kernel to
418 * be reaped by free_initmem before the root thread has proceeded to
419 * cpu_idle.
420 *
421 * gcc-3.4 accidentally inlines this function, so use noinline.
422 */
423
b433c3d4
PZ
424static __initdata DECLARE_COMPLETION(kthreadd_done);
425
f99ebf0a 426static noinline void __init_refok rest_init(void)
1da177e4 427{
73c27992
EB
428 int pid;
429
7db905e6 430 rcu_scheduler_starting();
b433c3d4 431 /*
97158569 432 * We need to spawn init first so that it obtains pid 1, however
b433c3d4
PZ
433 * the init task will end up wanting to create kthreads, which, if
434 * we schedule it before we create kthreadd, will OOPS.
435 */
aae5f662 436 kernel_thread(kernel_init, NULL, CLONE_FS | CLONE_SIGHAND);
1da177e4 437 numa_default_policy();
73c27992 438 pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES);
d11c563d 439 rcu_read_lock();
5cd20455 440 kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns);
d11c563d 441 rcu_read_unlock();
b433c3d4 442 complete(&kthreadd_done);
f340c0d1
IM
443
444 /*
445 * The boot idle thread must execute schedule()
1df21055 446 * at least once to get things moving:
f340c0d1 447 */
1df21055 448 init_idle_bootup_task(current);
5bfb5d69 449 preempt_enable_no_resched();
f340c0d1 450 schedule();
5bfb5d69 451 preempt_disable();
f340c0d1 452
5bfb5d69 453 /* Call into cpu_idle with preempt disabled */
1da177e4 454 cpu_idle();
1df21055 455}
1da177e4
LT
456
457/* Check for early params. */
458static int __init do_early_param(char *param, char *val)
459{
914dcaa8 460 const struct obs_kernel_param *p;
1da177e4
LT
461
462 for (p = __setup_start; p < __setup_end; p++) {
18a8bd94
YL
463 if ((p->early && strcmp(param, p->str) == 0) ||
464 (strcmp(param, "console") == 0 &&
465 strcmp(p->str, "earlycon") == 0)
466 ) {
1da177e4
LT
467 if (p->setup_func(val) != 0)
468 printk(KERN_WARNING
469 "Malformed early option '%s'\n", param);
470 }
471 }
472 /* We accept everything at this stage. */
473 return 0;
474}
475
13977091
MD
476void __init parse_early_options(char *cmdline)
477{
478 parse_args("early options", cmdline, NULL, 0, do_early_param);
479}
480
1da177e4
LT
481/* Arch code calls this early on, or if not, just before other parsing. */
482void __init parse_early_param(void)
483{
484 static __initdata int done = 0;
485 static __initdata char tmp_cmdline[COMMAND_LINE_SIZE];
486
487 if (done)
488 return;
489
490 /* All fall through to do_early_param. */
30d7e0d4 491 strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE);
13977091 492 parse_early_options(tmp_cmdline);
1da177e4
LT
493 done = 1;
494}
495
496/*
497 * Activate the first processor.
498 */
499
44fd2299
SS
500static void __init boot_cpu_init(void)
501{
502 int cpu = smp_processor_id();
503 /* Mark the boot cpu "present", "online" etc for SMP and UP case */
915441b6 504 set_cpu_online(cpu, true);
933b0618 505 set_cpu_active(cpu, true);
915441b6
RR
506 set_cpu_present(cpu, true);
507 set_cpu_possible(cpu, true);
44fd2299
SS
508}
509
839ad62e 510void __init __weak smp_setup_processor_id(void)
033ab7f8
AM
511{
512}
513
8c9843e5
BH
514void __init __weak thread_info_cache_init(void)
515{
516}
517
444f478f
PE
518/*
519 * Set up kernel memory allocators
520 */
521static void __init mm_init(void)
522{
ca371c0d
KH
523 /*
524 * page_cgroup requires countinous pages as memmap
525 * and it's bigger than MAX_ORDER unless SPARSEMEM.
526 */
527 page_cgroup_init_flatmem();
444f478f
PE
528 mem_init();
529 kmem_cache_init();
099a19d9 530 percpu_init_late();
c868d550 531 pgtable_cache_init();
444f478f
PE
532 vmalloc_init();
533}
534
1da177e4
LT
535asmlinkage void __init start_kernel(void)
536{
537 char * command_line;
914dcaa8 538 extern const struct kernel_param __start___param[], __stop___param[];
033ab7f8
AM
539
540 smp_setup_processor_id();
541
fbb9ce95
IM
542 /*
543 * Need to run as early as possible, to initialize the
544 * lockdep hash:
545 */
546 lockdep_init();
3ac7fe5a 547 debug_objects_early_init();
42059429
IM
548
549 /*
550 * Set up the the initial canary ASAP:
551 */
552 boot_init_stack_canary();
553
ddbcc7e8 554 cgroup_init_early();
fbb9ce95
IM
555
556 local_irq_disable();
557 early_boot_irqs_off();
558
1da177e4
LT
559/*
560 * Interrupts are still disabled. Do necessary setups, then
561 * enable them
562 */
906568c9 563 tick_init();
44fd2299 564 boot_cpu_init();
1da177e4 565 page_address_init();
657cafa6 566 printk(KERN_NOTICE "%s", linux_banner);
1da177e4 567 setup_arch(&command_line);
cf475ad2 568 mm_init_owner(&init_mm, &init_task);
30d7e0d4 569 setup_command_line(command_line);
e0982e90 570 setup_nr_cpu_ids();
d6647bdf 571 setup_per_cpu_areas();
44fd2299 572 smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */
1da177e4 573
1f522509 574 build_all_zonelists(NULL);
83b519e8
PE
575 page_alloc_init();
576
577 printk(KERN_NOTICE "Kernel command line: %s\n", boot_command_line);
578 parse_early_param();
579 parse_args("Booting kernel", static_command_line, __start___param,
580 __stop___param - __start___param,
581 &unknown_bootoption);
582 /*
583 * These use large bootmem allocations and must precede
584 * kmem_cache_init()
585 */
586 pidhash_init();
83b519e8
PE
587 vfs_caches_init_early();
588 sort_main_extable();
589 trap_init();
444f478f 590 mm_init();
1da177e4
LT
591 /*
592 * Set up the scheduler prior starting any interrupts (such as the
593 * timer interrupt). Full topology setup happens at smp_init()
594 * time - but meanwhile we still have a functioning scheduler.
595 */
596 sched_init();
597 /*
598 * Disable preemption - early bootup scheduling is extremely
599 * fragile until we cpu_idle() for the first time.
600 */
601 preempt_disable();
c4a68306
AB
602 if (!irqs_disabled()) {
603 printk(KERN_WARNING "start_kernel(): bug: interrupts were "
604 "enabled *very* early, fixing it\n");
605 local_irq_disable();
606 }
1da177e4 607 rcu_init();
773e3eb7 608 radix_tree_init();
0b8f1efa
YL
609 /* init some links before init_ISA_irqs() */
610 early_irq_init();
1da177e4 611 init_IRQ();
3c7b4e6b 612 prio_tree_init();
1da177e4 613 init_timers();
c0a31329 614 hrtimers_init();
1da177e4 615 softirq_init();
ad596171 616 timekeeping_init();
88fecaa2 617 time_init();
93e02814
HC
618 profile_init();
619 if (!irqs_disabled())
24d431d0
RL
620 printk(KERN_CRIT "start_kernel(): bug: interrupts were "
621 "enabled early\n");
fbb9ce95 622 early_boot_irqs_on();
93e02814 623 local_irq_enable();
dcce284a
BH
624
625 /* Interrupts are enabled now so all GFP allocations are safe. */
452aa699 626 gfp_allowed_mask = __GFP_BITS_MASK;
dcce284a 627
7e85ee0c 628 kmem_cache_init_late();
1da177e4
LT
629
630 /*
631 * HACK ALERT! This is early. We're enabling the console before
632 * we've done PCI setups etc, and console_init() must be aware of
633 * this. But we do want output early, in case something goes wrong.
634 */
635 console_init();
636 if (panic_later)
637 panic(panic_later, panic_param);
fbb9ce95
IM
638
639 lockdep_info();
640
9a11b49a
IM
641 /*
642 * Need to run this when irqs are enabled, because it wants
643 * to self-test [hard/soft]-irqs on/off lock inversion bugs
644 * too:
645 */
646 locking_selftest();
647
1da177e4
LT
648#ifdef CONFIG_BLK_DEV_INITRD
649 if (initrd_start && !initrd_below_start_ok &&
bd673c7c 650 page_to_pfn(virt_to_page((void *)initrd_start)) < min_low_pfn) {
1da177e4 651 printk(KERN_CRIT "initrd overwritten (0x%08lx < 0x%08lx) - "
fb6624eb 652 "disabling it.\n",
bd673c7c
GU
653 page_to_pfn(virt_to_page((void *)initrd_start)),
654 min_low_pfn);
1da177e4
LT
655 initrd_start = 0;
656 }
657#endif
94b6da5a 658 page_cgroup_init();
a03c2a48 659 enable_debug_pagealloc();
3c7b4e6b 660 kmemleak_init();
3ac7fe5a 661 debug_objects_mem_init();
199f0ca5 662 idr_init_cache();
e7c8d5c9 663 setup_per_cpu_pageset();
1da177e4
LT
664 numa_policy_init();
665 if (late_time_init)
666 late_time_init();
fa84e9ee 667 sched_clock_init();
1da177e4
LT
668 calibrate_delay();
669 pidmap_init();
1da177e4
LT
670 anon_vma_init();
671#ifdef CONFIG_X86
672 if (efi_enabled)
673 efi_enter_virtual_mode();
674#endif
8c9843e5 675 thread_info_cache_init();
d84f4f99 676 cred_init();
4481374c 677 fork_init(totalram_pages);
1da177e4
LT
678 proc_caches_init();
679 buffer_init();
1da177e4
LT
680 key_init();
681 security_init();
0b4b3827 682 dbg_late_init();
4481374c 683 vfs_caches_init(totalram_pages);
1da177e4
LT
684 signals_init();
685 /* rootfs populating might need page-writeback */
686 page_writeback_init();
687#ifdef CONFIG_PROC_FS
688 proc_root_init();
689#endif
ddbcc7e8 690 cgroup_init();
1da177e4 691 cpuset_init();
c757249a 692 taskstats_init_early();
ca74e92b 693 delayacct_init();
1da177e4
LT
694
695 check_bugs();
696
697 acpi_early_init(); /* before LAPIC and SMP init */
6ae6996a 698 sfi_init_late();
1da177e4 699
68bf21aa
SR
700 ftrace_init();
701
1da177e4
LT
702 /* Do the rest non-__init'ed, we're now alive */
703 rest_init();
704}
705
b99b87f7
PO
706/* Call all constructor functions linked into the kernel. */
707static void __init do_ctors(void)
708{
709#ifdef CONFIG_CONSTRUCTORS
196a15b4 710 ctor_fn_t *fn = (ctor_fn_t *) __ctors_start;
b99b87f7 711
196a15b4
HS
712 for (; fn < (ctor_fn_t *) __ctors_end; fn++)
713 (*fn)();
b99b87f7
PO
714#endif
715}
716
22a9d645 717int initcall_debug;
d0ea3d7d 718core_param(initcall_debug, initcall_debug, bool, 0644);
1da177e4 719
4a683bf9 720static char msgbuf[64];
4a683bf9 721
e4461271 722static int __init_or_module do_one_initcall_debug(initcall_t fn)
1da177e4 723{
74239072 724 ktime_t calltime, delta, rettime;
30dbb20e
AW
725 unsigned long long duration;
726 int ret;
1da177e4 727
22c5c03b
KW
728 printk(KERN_DEBUG "calling %pF @ %i\n", fn, task_pid_nr(current));
729 calltime = ktime_get();
30dbb20e 730 ret = fn();
22c5c03b
KW
731 rettime = ktime_get();
732 delta = ktime_sub(rettime, calltime);
733 duration = (unsigned long long) ktime_to_ns(delta) >> 10;
734 printk(KERN_DEBUG "initcall %pF returned %d after %lld usecs\n", fn,
735 ret, duration);
1da177e4 736
22c5c03b
KW
737 return ret;
738}
739
e4461271 740int __init_or_module do_one_initcall(initcall_t fn)
22c5c03b
KW
741{
742 int count = preempt_count();
743 int ret;
744
745 if (initcall_debug)
746 ret = do_one_initcall_debug(fn);
747 else
748 ret = fn();
8f0c45cd 749
e0df154f 750 msgbuf[0] = 0;
e662e1cf 751
30dbb20e
AW
752 if (ret && ret != -ENODEV && initcall_debug)
753 sprintf(msgbuf, "error code %d ", ret);
e662e1cf 754
e0df154f 755 if (preempt_count() != count) {
a76bfd0d 756 strlcat(msgbuf, "preemption imbalance ", sizeof(msgbuf));
e0df154f 757 preempt_count() = count;
1da177e4 758 }
e0df154f 759 if (irqs_disabled()) {
a76bfd0d 760 strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf));
e0df154f
LT
761 local_irq_enable();
762 }
763 if (msgbuf[0]) {
96d746c6 764 printk("initcall %pF returned with %s\n", fn, msgbuf);
e0df154f 765 }
59f9415f 766
30dbb20e 767 return ret;
e0df154f
LT
768}
769
770
c2147a50 771extern initcall_t __initcall_start[], __initcall_end[], __early_initcall_end[];
e0df154f
LT
772
773static void __init do_initcalls(void)
774{
196a15b4 775 initcall_t *fn;
e0df154f 776
196a15b4
HS
777 for (fn = __early_initcall_end; fn < __initcall_end; fn++)
778 do_one_initcall(*fn);
1da177e4
LT
779
780 /* Make sure there is no pending stuff from the initcall sequence */
781 flush_scheduled_work();
782}
783
784/*
785 * Ok, the machine is now initialized. None of the devices
786 * have been touched yet, but the CPU subsystem is up and
787 * running, and memory and process management works.
788 *
789 * Now we can finally start doing some real work..
790 */
791static void __init do_basic_setup(void)
792{
759ee091 793 cpuset_init_smp();
1da177e4 794 usermodehelper_init();
2b2af54a 795 init_tmpfs();
1da177e4 796 driver_init();
b04c3afb 797 init_irq_proc();
b99b87f7 798 do_ctors();
1da177e4
LT
799 do_initcalls();
800}
801
7babe8db 802static void __init do_pre_smp_initcalls(void)
c2147a50 803{
196a15b4 804 initcall_t *fn;
c2147a50 805
196a15b4
HS
806 for (fn = __initcall_start; fn < __early_initcall_end; fn++)
807 do_one_initcall(*fn);
c2147a50
EGM
808}
809
d7627467 810static void run_init_process(const char *init_filename)
1da177e4
LT
811{
812 argv_init[0] = init_filename;
67608567 813 kernel_execve(init_filename, argv_init, envp_init);
1da177e4
LT
814}
815
ee5bfa64
VG
816/* This is a non __init function. Force it to be noinline otherwise gcc
817 * makes it inline to init() and it becomes part of init.text section
818 */
f99ebf0a 819static noinline int init_post(void)
ee5bfa64 820{
22a9d645
AV
821 /* need to finish all async __init code before freeing the memory */
822 async_synchronize_full();
ee5bfa64 823 free_initmem();
ee5bfa64
VG
824 mark_rodata_ro();
825 system_state = SYSTEM_RUNNING;
826 numa_default_policy();
827
ee5bfa64 828
fae5fa44
ON
829 current->signal->flags |= SIGNAL_UNKILLABLE;
830
ee5bfa64
VG
831 if (ramdisk_execute_command) {
832 run_init_process(ramdisk_execute_command);
833 printk(KERN_WARNING "Failed to execute %s\n",
834 ramdisk_execute_command);
835 }
836
837 /*
838 * We try each of these until one succeeds.
839 *
840 * The Bourne shell can be used instead of init if we are
841 * trying to recover a really broken machine.
842 */
843 if (execute_command) {
844 run_init_process(execute_command);
845 printk(KERN_WARNING "Failed to execute %s. Attempting "
846 "defaults...\n", execute_command);
847 }
848 run_init_process("/sbin/init");
849 run_init_process("/etc/init");
850 run_init_process("/bin/init");
851 run_init_process("/bin/sh");
852
9a85b8d6
AM
853 panic("No init found. Try passing init= option to kernel. "
854 "See Linux Documentation/init.txt for guidance.");
ee5bfa64
VG
855}
856
aae5f662 857static int __init kernel_init(void * unused)
1da177e4 858{
b433c3d4
PZ
859 /*
860 * Wait until kthreadd is all set-up.
861 */
862 wait_for_completion(&kthreadd_done);
58568d2a
MX
863 /*
864 * init can allocate pages on any node
865 */
5ab116c9 866 set_mems_allowed(node_states[N_HIGH_MEMORY]);
1da177e4
LT
867 /*
868 * init can run on any cpu.
869 */
1a2142af 870 set_cpus_allowed_ptr(current, cpu_all_mask);
1da177e4
LT
871 /*
872 * Tell the world that we're going to be the grim
873 * reaper of innocent orphaned children.
874 *
875 * We don't want people to have to make incorrect
876 * assumptions about where in the task array this
877 * can be found.
878 */
84d73786 879 init_pid_ns.child_reaper = current;
1da177e4 880
9ec52099
CLG
881 cad_pid = task_pid(current);
882
ca74a6f8 883 smp_prepare_cpus(setup_max_cpus);
1da177e4
LT
884
885 do_pre_smp_initcalls();
886
1da177e4
LT
887 smp_init();
888 sched_init_smp();
889
1da177e4
LT
890 do_basic_setup();
891
2bd3a997
EB
892 /* Open the /dev/console on the rootfs, this should never fail */
893 if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0)
894 printk(KERN_WARNING "Warning: unable to open an initial console.\n");
895
896 (void) sys_dup(0);
897 (void) sys_dup(0);
1da177e4
LT
898 /*
899 * check if there is an early userspace init. If yes, let it do all
900 * the work
901 */
ffdfc409
OJ
902
903 if (!ramdisk_execute_command)
904 ramdisk_execute_command = "/init";
905
906 if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) {
907 ramdisk_execute_command = NULL;
1da177e4 908 prepare_namespace();
ffdfc409 909 }
1da177e4
LT
910
911 /*
912 * Ok, we have completed the initial bootup, and
913 * we're essentially up and running. Get rid of the
914 * initmem segments and start the user-mode stuff..
915 */
71566a0d 916
ee5bfa64
VG
917 init_post();
918 return 0;
1da177e4 919}