sched: rework of "prioritize non-migratable tasks over migratable ones"
[GitHub/mt8127/android_kernel_alcatel_ttab.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>
17#include <linux/string.h>
18#include <linux/ctype.h>
19#include <linux/delay.h>
20#include <linux/utsname.h>
21#include <linux/ioport.h>
22#include <linux/init.h>
23#include <linux/smp_lock.h>
24#include <linux/initrd.h>
25#include <linux/hdreg.h>
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>
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/workqueue.h>
36#include <linux/profile.h>
37#include <linux/rcupdate.h>
38#include <linux/moduleparam.h>
39#include <linux/kallsyms.h>
40#include <linux/writeback.h>
41#include <linux/cpu.h>
42#include <linux/cpuset.h>
ddbcc7e8 43#include <linux/cgroup.h>
1da177e4 44#include <linux/efi.h>
906568c9 45#include <linux/tick.h>
6168a702 46#include <linux/interrupt.h>
c757249a 47#include <linux/taskstats_kern.h>
ca74e92b 48#include <linux/delayacct.h>
1da177e4
LT
49#include <linux/unistd.h>
50#include <linux/rmap.h>
51#include <linux/mempolicy.h>
52#include <linux/key.h>
4552d5dc 53#include <linux/unwind.h>
b6cd0b77 54#include <linux/buffer_head.h>
9a11b49a 55#include <linux/debug_locks.h>
3ac7fe5a 56#include <linux/debugobjects.h>
fbb9ce95 57#include <linux/lockdep.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>
1da177e4
LT
64
65#include <asm/io.h>
66#include <asm/bugs.h>
67#include <asm/setup.h>
a940199f 68#include <asm/sections.h>
37b73c82 69#include <asm/cacheflush.h>
1da177e4 70
1da177e4
LT
71#ifdef CONFIG_X86_LOCAL_APIC
72#include <asm/smp.h>
73#endif
74
75/*
69c99ac1
CQH
76 * This is one of the first .c files built. Error out early if we have compiler
77 * trouble.
1da177e4 78 */
1da177e4 79
f1a60dbf
AM
80#if __GNUC__ == 4 && __GNUC_MINOR__ == 1 && __GNUC_PATCHLEVEL__ == 0
81#warning gcc-4.1.0 is known to miscompile the kernel. A different compiler version is recommended.
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 pidhash_init(void);
91extern void pidmap_init(void);
92extern void prio_tree_init(void);
93extern void radix_tree_init(void);
94extern void free_initmem(void);
1da177e4
LT
95#ifdef CONFIG_ACPI
96extern void acpi_early_init(void);
97#else
98static inline void acpi_early_init(void) { }
99#endif
37b73c82
AV
100#ifndef CONFIG_DEBUG_RODATA
101static inline void mark_rodata_ro(void) { }
102#endif
1da177e4
LT
103
104#ifdef CONFIG_TC
105extern void tc_init(void);
106#endif
107
108enum system_states system_state;
109EXPORT_SYMBOL(system_state);
110
111/*
112 * Boot command-line arguments
113 */
114#define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT
115#define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT
116
117extern void time_init(void);
118/* Default late time init is NULL. archs can override this later. */
119void (*late_time_init)(void);
120extern void softirq_init(void);
121
30d7e0d4
ABL
122/* Untouched command line saved by arch-specific code. */
123char __initdata boot_command_line[COMMAND_LINE_SIZE];
124/* Untouched saved command line (eg. for /proc) */
125char *saved_command_line;
126/* Command line for parameter parsing */
127static char *static_command_line;
1da177e4
LT
128
129static char *execute_command;
ffdfc409 130static char *ramdisk_execute_command;
1da177e4 131
8b3b2955 132#ifdef CONFIG_SMP
1da177e4 133/* Setup configured maximum number of CPUs to activate */
ca74a6f8 134unsigned int __initdata setup_max_cpus = NR_CPUS;
7e96287d 135
1da177e4
LT
136/*
137 * Setup routine for controlling SMP activation
138 *
139 * Command-line option of "nosmp" or "maxcpus=0" will disable SMP
140 * activation entirely (the MPS table probe still happens, though).
141 *
142 * Command-line option of "maxcpus=<NUM>", where <NUM> is an integer
143 * greater than 0, limits the maximum number of CPUs activated in
144 * SMP mode to <NUM>.
145 */
61ec7567
LB
146#ifndef CONFIG_X86_IO_APIC
147static inline void disable_ioapic_setup(void) {};
148#endif
149
1da177e4
LT
150static int __init nosmp(char *str)
151{
ca74a6f8 152 setup_max_cpus = 0;
61ec7567 153 disable_ioapic_setup();
8b3b2955 154 return 0;
1da177e4
LT
155}
156
8b3b2955 157early_param("nosmp", nosmp);
1da177e4
LT
158
159static int __init maxcpus(char *str)
160{
ca74a6f8
AK
161 get_option(&str, &setup_max_cpus);
162 if (setup_max_cpus == 0)
61ec7567
LB
163 disable_ioapic_setup();
164
165 return 0;
1da177e4
LT
166}
167
81340977 168early_param("maxcpus", maxcpus);
8b3b2955 169#else
ca74a6f8 170#define setup_max_cpus NR_CPUS
8b3b2955
JB
171#endif
172
173/*
174 * If set, this is an indication to the drivers that reset the underlying
175 * device before going ahead with the initialization otherwise driver might
176 * rely on the BIOS and skip the reset operation.
177 *
178 * This is useful if kernel is booting in an unreliable environment.
179 * For ex. kdump situaiton where previous kernel has crashed, BIOS has been
180 * skipped and devices will be in unknown state.
181 */
182unsigned int reset_devices;
183EXPORT_SYMBOL(reset_devices);
1da177e4 184
7e96287d
VG
185static int __init set_reset_devices(char *str)
186{
187 reset_devices = 1;
188 return 1;
189}
190
191__setup("reset_devices", set_reset_devices);
192
1da177e4
LT
193static char * argv_init[MAX_INIT_ARGS+2] = { "init", NULL, };
194char * envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, };
195static const char *panic_later, *panic_param;
196
197extern struct obs_kernel_param __setup_start[], __setup_end[];
198
199static int __init obsolete_checksetup(char *line)
200{
201 struct obs_kernel_param *p;
33df0d19 202 int had_early_param = 0;
1da177e4
LT
203
204 p = __setup_start;
205 do {
206 int n = strlen(p->str);
207 if (!strncmp(line, p->str, n)) {
208 if (p->early) {
33df0d19
RR
209 /* Already done in parse_early_param?
210 * (Needs exact match on param part).
211 * Keep iterating, as we can have early
212 * params and __setups of same names 8( */
1da177e4 213 if (line[n] == '\0' || line[n] == '=')
33df0d19 214 had_early_param = 1;
1da177e4
LT
215 } else if (!p->setup_func) {
216 printk(KERN_WARNING "Parameter %s is obsolete,"
217 " ignored\n", p->str);
218 return 1;
219 } else if (p->setup_func(line + n))
220 return 1;
221 }
222 p++;
223 } while (p < __setup_end);
33df0d19
RR
224
225 return had_early_param;
1da177e4
LT
226}
227
228/*
229 * This should be approx 2 Bo*oMips to start (note initial shift), and will
230 * still work even if initially too large, it will just take slightly longer
231 */
232unsigned long loops_per_jiffy = (1<<12);
233
234EXPORT_SYMBOL(loops_per_jiffy);
235
236static int __init debug_kernel(char *str)
237{
1da177e4 238 console_loglevel = 10;
f6f21c81 239 return 0;
1da177e4
LT
240}
241
242static int __init quiet_kernel(char *str)
243{
1da177e4 244 console_loglevel = 4;
f6f21c81 245 return 0;
1da177e4
LT
246}
247
f6f21c81
YL
248early_param("debug", debug_kernel);
249early_param("quiet", quiet_kernel);
1da177e4
LT
250
251static int __init loglevel(char *str)
252{
253 get_option(&str, &console_loglevel);
d9d4fcfe 254 return 0;
1da177e4
LT
255}
256
f6f21c81 257early_param("loglevel", loglevel);
1da177e4
LT
258
259/*
260 * Unknown boot options get handed to init, unless they look like
261 * failed parameters
262 */
263static int __init unknown_bootoption(char *param, char *val)
264{
265 /* Change NUL term back to "=", to make "param" the whole string. */
266 if (val) {
267 /* param=val or param="val"? */
268 if (val == param+strlen(param)+1)
269 val[-1] = '=';
270 else if (val == param+strlen(param)+2) {
271 val[-2] = '=';
272 memmove(val-1, val, strlen(val)+1);
273 val--;
274 } else
275 BUG();
276 }
277
278 /* Handle obsolete-style parameters */
279 if (obsolete_checksetup(param))
280 return 0;
281
282 /*
211fee8a 283 * Preemptive maintenance for "why didn't my misspelled command
1da177e4
LT
284 * line work?"
285 */
286 if (strchr(param, '.') && (!val || strchr(param, '.') < val)) {
287 printk(KERN_ERR "Unknown boot option `%s': ignoring\n", param);
288 return 0;
289 }
290
291 if (panic_later)
292 return 0;
293
294 if (val) {
295 /* Environment option */
296 unsigned int i;
297 for (i = 0; envp_init[i]; i++) {
298 if (i == MAX_INIT_ENVS) {
299 panic_later = "Too many boot env vars at `%s'";
300 panic_param = param;
301 }
302 if (!strncmp(param, envp_init[i], val - param))
303 break;
304 }
305 envp_init[i] = param;
306 } else {
307 /* Command line option */
308 unsigned int i;
309 for (i = 0; argv_init[i]; i++) {
310 if (i == MAX_INIT_ARGS) {
311 panic_later = "Too many boot init vars at `%s'";
312 panic_param = param;
313 }
314 }
315 argv_init[i] = param;
316 }
317 return 0;
318}
319
12d6f21e
IM
320#ifdef CONFIG_DEBUG_PAGEALLOC
321int __read_mostly debug_pagealloc_enabled = 0;
322#endif
323
1da177e4
LT
324static int __init init_setup(char *str)
325{
326 unsigned int i;
327
328 execute_command = str;
329 /*
330 * In case LILO is going to boot us with default command line,
331 * it prepends "auto" before the whole cmdline which makes
332 * the shell think it should execute a script with such name.
333 * So we ignore all arguments entered _before_ init=... [MJ]
334 */
335 for (i = 1; i < MAX_INIT_ARGS; i++)
336 argv_init[i] = NULL;
337 return 1;
338}
339__setup("init=", init_setup);
340
ffdfc409
OJ
341static int __init rdinit_setup(char *str)
342{
343 unsigned int i;
344
345 ramdisk_execute_command = str;
346 /* See "auto" comment in init_setup */
347 for (i = 1; i < MAX_INIT_ARGS; i++)
348 argv_init[i] = NULL;
349 return 1;
350}
351__setup("rdinit=", rdinit_setup);
352
1da177e4
LT
353#ifndef CONFIG_SMP
354
355#ifdef CONFIG_X86_LOCAL_APIC
356static void __init smp_init(void)
357{
358 APIC_init_uniprocessor();
359}
360#else
361#define smp_init() do { } while (0)
362#endif
363
364static inline void setup_per_cpu_areas(void) { }
e0982e90 365static inline void setup_nr_cpu_ids(void) { }
1da177e4
LT
366static inline void smp_prepare_cpus(unsigned int maxcpus) { }
367
368#else
369
321a8e9d
MT
370#if NR_CPUS > BITS_PER_LONG
371cpumask_t cpu_mask_all __read_mostly = CPU_MASK_ALL;
372EXPORT_SYMBOL(cpu_mask_all);
373#endif
374
e0982e90
MT
375/* Setup number of possible processor ids */
376int nr_cpu_ids __read_mostly = NR_CPUS;
377EXPORT_SYMBOL(nr_cpu_ids);
378
379/* An arch may set nr_cpu_ids earlier if needed, so this would be redundant */
380static void __init setup_nr_cpu_ids(void)
381{
382 int cpu, highest_cpu = 0;
383
384 for_each_possible_cpu(cpu)
385 highest_cpu = cpu;
386
387 nr_cpu_ids = highest_cpu + 1;
388}
389
dd5af90a 390#ifndef CONFIG_HAVE_SETUP_PER_CPU_AREA
b73b459f 391unsigned long __per_cpu_offset[NR_CPUS] __read_mostly;
1da177e4
LT
392
393EXPORT_SYMBOL(__per_cpu_offset);
394
395static void __init setup_per_cpu_areas(void)
396{
397 unsigned long size, i;
398 char *ptr;
63872f87 399 unsigned long nr_possible_cpus = num_possible_cpus();
1da177e4
LT
400
401 /* Copy section for each CPU (we discard the original) */
b6e3590f
JF
402 size = ALIGN(PERCPU_ENOUGH_ROOM, PAGE_SIZE);
403 ptr = alloc_bootmem_pages(size * nr_possible_cpus);
1da177e4 404
0a945022 405 for_each_possible_cpu(i) {
1da177e4
LT
406 __per_cpu_offset[i] = ptr - __per_cpu_start;
407 memcpy(ptr, __per_cpu_start, __per_cpu_end - __per_cpu_start);
63872f87 408 ptr += size;
1da177e4
LT
409 }
410}
dd5af90a 411#endif /* CONFIG_HAVE_SETUP_PER_CPU_AREA */
1da177e4
LT
412
413/* Called by boot processor to activate the rest. */
414static void __init smp_init(void)
415{
53b8a315 416 unsigned int cpu;
1da177e4
LT
417
418 /* FIXME: This should be done in userspace --RR */
53b8a315 419 for_each_present_cpu(cpu) {
ca74a6f8 420 if (num_online_cpus() >= setup_max_cpus)
1da177e4 421 break;
53b8a315
CL
422 if (!cpu_online(cpu))
423 cpu_up(cpu);
1da177e4
LT
424 }
425
426 /* Any cleanup work */
427 printk(KERN_INFO "Brought up %ld CPUs\n", (long)num_online_cpus());
ca74a6f8 428 smp_cpus_done(setup_max_cpus);
1da177e4
LT
429}
430
431#endif
432
30d7e0d4
ABL
433/*
434 * We need to store the untouched command line for future reference.
435 * We also need to store the touched command line since the parameter
436 * parsing is performed in place, and we should allow a component to
437 * store reference of name/value for future reference.
438 */
439static void __init setup_command_line(char *command_line)
440{
441 saved_command_line = alloc_bootmem(strlen (boot_command_line)+1);
442 static_command_line = alloc_bootmem(strlen (command_line)+1);
443 strcpy (saved_command_line, boot_command_line);
444 strcpy (static_command_line, command_line);
445}
446
1da177e4
LT
447/*
448 * We need to finalize in a non-__init function or else race conditions
449 * between the root thread and the init thread may cause start_kernel to
450 * be reaped by free_initmem before the root thread has proceeded to
451 * cpu_idle.
452 *
453 * gcc-3.4 accidentally inlines this function, so use noinline.
454 */
455
92080309 456static void noinline __init_refok rest_init(void)
1da177e4
LT
457 __releases(kernel_lock)
458{
73c27992
EB
459 int pid;
460
aae5f662 461 kernel_thread(kernel_init, NULL, CLONE_FS | CLONE_SIGHAND);
1da177e4 462 numa_default_policy();
73c27992 463 pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES);
5cd20455 464 kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns);
1da177e4 465 unlock_kernel();
f340c0d1
IM
466
467 /*
468 * The boot idle thread must execute schedule()
1df21055 469 * at least once to get things moving:
f340c0d1 470 */
1df21055 471 init_idle_bootup_task(current);
5bfb5d69 472 preempt_enable_no_resched();
f340c0d1 473 schedule();
5bfb5d69 474 preempt_disable();
f340c0d1 475
5bfb5d69 476 /* Call into cpu_idle with preempt disabled */
1da177e4 477 cpu_idle();
1df21055 478}
1da177e4
LT
479
480/* Check for early params. */
481static int __init do_early_param(char *param, char *val)
482{
483 struct obs_kernel_param *p;
484
485 for (p = __setup_start; p < __setup_end; p++) {
18a8bd94
YL
486 if ((p->early && strcmp(param, p->str) == 0) ||
487 (strcmp(param, "console") == 0 &&
488 strcmp(p->str, "earlycon") == 0)
489 ) {
1da177e4
LT
490 if (p->setup_func(val) != 0)
491 printk(KERN_WARNING
492 "Malformed early option '%s'\n", param);
493 }
494 }
495 /* We accept everything at this stage. */
496 return 0;
497}
498
499/* Arch code calls this early on, or if not, just before other parsing. */
500void __init parse_early_param(void)
501{
502 static __initdata int done = 0;
503 static __initdata char tmp_cmdline[COMMAND_LINE_SIZE];
504
505 if (done)
506 return;
507
508 /* All fall through to do_early_param. */
30d7e0d4 509 strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE);
1da177e4
LT
510 parse_args("early options", tmp_cmdline, NULL, 0, do_early_param);
511 done = 1;
512}
513
514/*
515 * Activate the first processor.
516 */
517
44fd2299
SS
518static void __init boot_cpu_init(void)
519{
520 int cpu = smp_processor_id();
521 /* Mark the boot cpu "present", "online" etc for SMP and UP case */
522 cpu_set(cpu, cpu_online_map);
523 cpu_set(cpu, cpu_present_map);
524 cpu_set(cpu, cpu_possible_map);
525}
526
839ad62e 527void __init __weak smp_setup_processor_id(void)
033ab7f8
AM
528{
529}
530
8c9843e5
BH
531void __init __weak thread_info_cache_init(void)
532{
533}
534
1da177e4
LT
535asmlinkage void __init start_kernel(void)
536{
537 char * command_line;
538 extern 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 */
c9538ed4 546 unwind_init();
fbb9ce95 547 lockdep_init();
3ac7fe5a 548 debug_objects_early_init();
ddbcc7e8 549 cgroup_init_early();
fbb9ce95
IM
550
551 local_irq_disable();
552 early_boot_irqs_off();
243c7621 553 early_init_irq_lock_class();
fbb9ce95 554
1da177e4
LT
555/*
556 * Interrupts are still disabled. Do necessary setups, then
557 * enable them
558 */
559 lock_kernel();
906568c9 560 tick_init();
44fd2299 561 boot_cpu_init();
1da177e4
LT
562 page_address_init();
563 printk(KERN_NOTICE);
8993780a 564 printk(linux_banner);
1da177e4 565 setup_arch(&command_line);
cf475ad2 566 mm_init_owner(&init_mm, &init_task);
30d7e0d4 567 setup_command_line(command_line);
690a973f 568 unwind_setup();
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();
599 init_IRQ();
600 pidhash_init();
601 init_timers();
c0a31329 602 hrtimers_init();
1da177e4 603 softirq_init();
ad596171 604 timekeeping_init();
88fecaa2 605 time_init();
3e51f33f 606 sched_clock_init();
93e02814
HC
607 profile_init();
608 if (!irqs_disabled())
609 printk("start_kernel(): bug: interrupts were enabled early\n");
fbb9ce95 610 early_boot_irqs_on();
93e02814 611 local_irq_enable();
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 &&
633 initrd_start < min_low_pfn << PAGE_SHIFT) {
634 printk(KERN_CRIT "initrd overwritten (0x%08lx < 0x%08lx) - "
635 "disabling it.\n",initrd_start,min_low_pfn << PAGE_SHIFT);
636 initrd_start = 0;
637 }
638#endif
639 vfs_caches_init_early();
c417f024 640 cpuset_init_early();
1da177e4 641 mem_init();
a03c2a48 642 enable_debug_pagealloc();
d221938c 643 cpu_hotplug_init();
1da177e4 644 kmem_cache_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();
651 calibrate_delay();
652 pidmap_init();
653 pgtable_cache_init();
654 prio_tree_init();
655 anon_vma_init();
656#ifdef CONFIG_X86
657 if (efi_enabled)
658 efi_enter_virtual_mode();
659#endif
8c9843e5 660 thread_info_cache_init();
1da177e4
LT
661 fork_init(num_physpages);
662 proc_caches_init();
663 buffer_init();
664 unnamed_dev_init();
665 key_init();
666 security_init();
667 vfs_caches_init(num_physpages);
668 radix_tree_init();
669 signals_init();
670 /* rootfs populating might need page-writeback */
671 page_writeback_init();
672#ifdef CONFIG_PROC_FS
673 proc_root_init();
674#endif
ddbcc7e8 675 cgroup_init();
1da177e4 676 cpuset_init();
c757249a 677 taskstats_init_early();
ca74e92b 678 delayacct_init();
1da177e4
LT
679
680 check_bugs();
681
682 acpi_early_init(); /* before LAPIC and SMP init */
683
684 /* Do the rest non-__init'ed, we're now alive */
685 rest_init();
686}
687
688static int __initdata initcall_debug;
689
690static int __init initcall_debug_setup(char *str)
691{
692 initcall_debug = 1;
693 return 1;
694}
695__setup("initcall_debug", initcall_debug_setup);
696
e0df154f 697static void __init do_one_initcall(initcall_t fn)
1da177e4 698{
1da177e4 699 int count = preempt_count();
e0df154f 700 ktime_t t0, t1, delta;
a76bfd0d 701 char msgbuf[64];
e0df154f 702 int result;
1da177e4 703
e0df154f
LT
704 if (initcall_debug) {
705 print_fn_descriptor_symbol("calling %s\n", fn);
706 t0 = ktime_get();
707 }
1da177e4 708
e0df154f 709 result = fn();
1da177e4 710
e0df154f
LT
711 if (initcall_debug) {
712 t1 = ktime_get();
713 delta = ktime_sub(t1, t0);
8f0c45cd 714
e0df154f
LT
715 print_fn_descriptor_symbol("initcall %s", fn);
716 printk(" returned %d after %Ld msecs\n", result,
717 (unsigned long long) delta.tv64 >> 20);
718 }
8f0c45cd 719
e0df154f 720 msgbuf[0] = 0;
e662e1cf 721
e0df154f
LT
722 if (result && result != -ENODEV && initcall_debug)
723 sprintf(msgbuf, "error code %d ", result);
e662e1cf 724
e0df154f 725 if (preempt_count() != count) {
a76bfd0d 726 strlcat(msgbuf, "preemption imbalance ", sizeof(msgbuf));
e0df154f 727 preempt_count() = count;
1da177e4 728 }
e0df154f 729 if (irqs_disabled()) {
a76bfd0d 730 strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf));
e0df154f
LT
731 local_irq_enable();
732 }
733 if (msgbuf[0]) {
734 print_fn_descriptor_symbol(KERN_WARNING "initcall %s", fn);
735 printk(" returned with %s\n", msgbuf);
736 }
737}
738
739
740extern initcall_t __initcall_start[], __initcall_end[];
741
742static void __init do_initcalls(void)
743{
744 initcall_t *call;
745
746 for (call = __initcall_start; call < __initcall_end; call++)
747 do_one_initcall(*call);
1da177e4
LT
748
749 /* Make sure there is no pending stuff from the initcall sequence */
750 flush_scheduled_work();
751}
752
753/*
754 * Ok, the machine is now initialized. None of the devices
755 * have been touched yet, but the CPU subsystem is up and
756 * running, and memory and process management works.
757 *
758 * Now we can finally start doing some real work..
759 */
760static void __init do_basic_setup(void)
761{
4446a36f 762 rcu_init_sched(); /* needed by module_init stage. */
1da177e4
LT
763 /* drivers will send hotplug events */
764 init_workqueues();
765 usermodehelper_init();
766 driver_init();
b04c3afb 767 init_irq_proc();
1da177e4
LT
768 do_initcalls();
769}
770
97842216
DJ
771static int __initdata nosoftlockup;
772
773static int __init nosoftlockup_setup(char *str)
774{
775 nosoftlockup = 1;
776 return 1;
777}
778__setup("nosoftlockup", nosoftlockup_setup);
779
88d20328 780static void __init do_pre_smp_initcalls(void)
1da177e4
LT
781{
782 extern int spawn_ksoftirqd(void);
1da177e4 783
3d442233 784 init_call_single_data();
1da177e4 785 migration_init();
1da177e4 786 spawn_ksoftirqd();
97842216
DJ
787 if (!nosoftlockup)
788 spawn_softlockup_task();
1da177e4
LT
789}
790
791static void run_init_process(char *init_filename)
792{
793 argv_init[0] = init_filename;
67608567 794 kernel_execve(init_filename, argv_init, envp_init);
1da177e4
LT
795}
796
ee5bfa64
VG
797/* This is a non __init function. Force it to be noinline otherwise gcc
798 * makes it inline to init() and it becomes part of init.text section
799 */
800static int noinline init_post(void)
801{
802 free_initmem();
803 unlock_kernel();
804 mark_rodata_ro();
805 system_state = SYSTEM_RUNNING;
806 numa_default_policy();
807
808 if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0)
809 printk(KERN_WARNING "Warning: unable to open an initial console.\n");
810
811 (void) sys_dup(0);
812 (void) sys_dup(0);
813
fae5fa44
ON
814 current->signal->flags |= SIGNAL_UNKILLABLE;
815
ee5bfa64
VG
816 if (ramdisk_execute_command) {
817 run_init_process(ramdisk_execute_command);
818 printk(KERN_WARNING "Failed to execute %s\n",
819 ramdisk_execute_command);
820 }
821
822 /*
823 * We try each of these until one succeeds.
824 *
825 * The Bourne shell can be used instead of init if we are
826 * trying to recover a really broken machine.
827 */
828 if (execute_command) {
829 run_init_process(execute_command);
830 printk(KERN_WARNING "Failed to execute %s. Attempting "
831 "defaults...\n", execute_command);
832 }
833 run_init_process("/sbin/init");
834 run_init_process("/etc/init");
835 run_init_process("/bin/init");
836 run_init_process("/bin/sh");
837
838 panic("No init found. Try passing init= option to kernel.");
839}
840
aae5f662 841static int __init kernel_init(void * unused)
1da177e4
LT
842{
843 lock_kernel();
844 /*
845 * init can run on any cpu.
846 */
321a8e9d 847 set_cpus_allowed_ptr(current, CPU_MASK_ALL_PTR);
1da177e4
LT
848 /*
849 * Tell the world that we're going to be the grim
850 * reaper of innocent orphaned children.
851 *
852 * We don't want people to have to make incorrect
853 * assumptions about where in the task array this
854 * can be found.
855 */
84d73786 856 init_pid_ns.child_reaper = current;
1da177e4 857
9ec52099
CLG
858 cad_pid = task_pid(current);
859
ca74a6f8 860 smp_prepare_cpus(setup_max_cpus);
1da177e4
LT
861
862 do_pre_smp_initcalls();
863
1da177e4
LT
864 smp_init();
865 sched_init_smp();
866
867 cpuset_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 */
ee5bfa64
VG
889 init_post();
890 return 0;
1da177e4 891}