Commit | Line | Data |
---|---|---|
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 LT |
33 | #include <linux/security.h> |
34 | #include <linux/workqueue.h> | |
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> | |
42 | #include <linux/efi.h> | |
c757249a | 43 | #include <linux/taskstats_kern.h> |
ca74e92b | 44 | #include <linux/delayacct.h> |
1da177e4 LT |
45 | #include <linux/unistd.h> |
46 | #include <linux/rmap.h> | |
47 | #include <linux/mempolicy.h> | |
48 | #include <linux/key.h> | |
4552d5dc | 49 | #include <linux/unwind.h> |
b6cd0b77 | 50 | #include <linux/buffer_head.h> |
9a11b49a | 51 | #include <linux/debug_locks.h> |
fbb9ce95 | 52 | #include <linux/lockdep.h> |
a2ee8649 | 53 | #include <linux/utsrelease.h> |
84d73786 | 54 | #include <linux/pid_namespace.h> |
a2ee8649 | 55 | #include <linux/compile.h> |
1da177e4 LT |
56 | |
57 | #include <asm/io.h> | |
58 | #include <asm/bugs.h> | |
59 | #include <asm/setup.h> | |
a940199f | 60 | #include <asm/sections.h> |
37b73c82 | 61 | #include <asm/cacheflush.h> |
1da177e4 | 62 | |
1da177e4 LT |
63 | #ifdef CONFIG_X86_LOCAL_APIC |
64 | #include <asm/smp.h> | |
65 | #endif | |
66 | ||
67 | /* | |
69c99ac1 CQH |
68 | * This is one of the first .c files built. Error out early if we have compiler |
69 | * trouble. | |
70 | * | |
71 | * Versions of gcc older than that listed below may actually compile and link | |
72 | * okay, but the end product can have subtle run time bugs. To avoid associated | |
73 | * bogus bug reports, we flatly refuse to compile with a gcc that is known to be | |
74 | * too old from the very beginning. | |
1da177e4 | 75 | */ |
fd285bb5 | 76 | #if (__GNUC__ < 3) || (__GNUC__ == 3 && __GNUC_MINOR__ < 2) |
1da177e4 LT |
77 | #error Sorry, your GCC is too old. It builds incorrect kernels. |
78 | #endif | |
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 | ||
1da177e4 LT |
84 | static int init(void *); |
85 | ||
86 | extern void init_IRQ(void); | |
1da177e4 LT |
87 | extern void fork_init(unsigned long); |
88 | extern void mca_init(void); | |
89 | extern void sbus_init(void); | |
90 | extern void sysctl_init(void); | |
91 | extern void signals_init(void); | |
1da177e4 LT |
92 | extern void pidhash_init(void); |
93 | extern void pidmap_init(void); | |
94 | extern void prio_tree_init(void); | |
95 | extern void radix_tree_init(void); | |
96 | extern void free_initmem(void); | |
97 | extern void populate_rootfs(void); | |
98 | extern void driver_init(void); | |
99 | extern void prepare_namespace(void); | |
100 | #ifdef CONFIG_ACPI | |
101 | extern void acpi_early_init(void); | |
102 | #else | |
103 | static inline void acpi_early_init(void) { } | |
104 | #endif | |
37b73c82 AV |
105 | #ifndef CONFIG_DEBUG_RODATA |
106 | static inline void mark_rodata_ro(void) { } | |
107 | #endif | |
1da177e4 LT |
108 | |
109 | #ifdef CONFIG_TC | |
110 | extern void tc_init(void); | |
111 | #endif | |
112 | ||
113 | enum system_states system_state; | |
114 | EXPORT_SYMBOL(system_state); | |
115 | ||
116 | /* | |
117 | * Boot command-line arguments | |
118 | */ | |
119 | #define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT | |
120 | #define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT | |
121 | ||
122 | extern void time_init(void); | |
123 | /* Default late time init is NULL. archs can override this later. */ | |
124 | void (*late_time_init)(void); | |
125 | extern void softirq_init(void); | |
126 | ||
127 | /* Untouched command line (eg. for /proc) saved by arch-specific code. */ | |
128 | char saved_command_line[COMMAND_LINE_SIZE]; | |
129 | ||
130 | static char *execute_command; | |
ffdfc409 | 131 | static char *ramdisk_execute_command; |
1da177e4 LT |
132 | |
133 | /* Setup configured maximum number of CPUs to activate */ | |
134 | static unsigned int max_cpus = NR_CPUS; | |
135 | ||
7e96287d VG |
136 | /* |
137 | * If set, this is an indication to the drivers that reset the underlying | |
138 | * device before going ahead with the initialization otherwise driver might | |
139 | * rely on the BIOS and skip the reset operation. | |
140 | * | |
141 | * This is useful if kernel is booting in an unreliable environment. | |
142 | * For ex. kdump situaiton where previous kernel has crashed, BIOS has been | |
143 | * skipped and devices will be in unknown state. | |
144 | */ | |
145 | unsigned int reset_devices; | |
146 | EXPORT_SYMBOL(reset_devices); | |
147 | ||
1da177e4 LT |
148 | /* |
149 | * Setup routine for controlling SMP activation | |
150 | * | |
151 | * Command-line option of "nosmp" or "maxcpus=0" will disable SMP | |
152 | * activation entirely (the MPS table probe still happens, though). | |
153 | * | |
154 | * Command-line option of "maxcpus=<NUM>", where <NUM> is an integer | |
155 | * greater than 0, limits the maximum number of CPUs activated in | |
156 | * SMP mode to <NUM>. | |
157 | */ | |
158 | static int __init nosmp(char *str) | |
159 | { | |
160 | max_cpus = 0; | |
161 | return 1; | |
162 | } | |
163 | ||
164 | __setup("nosmp", nosmp); | |
165 | ||
166 | static int __init maxcpus(char *str) | |
167 | { | |
168 | get_option(&str, &max_cpus); | |
169 | return 1; | |
170 | } | |
171 | ||
172 | __setup("maxcpus=", maxcpus); | |
173 | ||
7e96287d VG |
174 | static int __init set_reset_devices(char *str) |
175 | { | |
176 | reset_devices = 1; | |
177 | return 1; | |
178 | } | |
179 | ||
180 | __setup("reset_devices", set_reset_devices); | |
181 | ||
1da177e4 LT |
182 | static char * argv_init[MAX_INIT_ARGS+2] = { "init", NULL, }; |
183 | char * envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, }; | |
184 | static const char *panic_later, *panic_param; | |
185 | ||
186 | extern struct obs_kernel_param __setup_start[], __setup_end[]; | |
187 | ||
188 | static int __init obsolete_checksetup(char *line) | |
189 | { | |
190 | struct obs_kernel_param *p; | |
33df0d19 | 191 | int had_early_param = 0; |
1da177e4 LT |
192 | |
193 | p = __setup_start; | |
194 | do { | |
195 | int n = strlen(p->str); | |
196 | if (!strncmp(line, p->str, n)) { | |
197 | if (p->early) { | |
33df0d19 RR |
198 | /* Already done in parse_early_param? |
199 | * (Needs exact match on param part). | |
200 | * Keep iterating, as we can have early | |
201 | * params and __setups of same names 8( */ | |
1da177e4 | 202 | if (line[n] == '\0' || line[n] == '=') |
33df0d19 | 203 | had_early_param = 1; |
1da177e4 LT |
204 | } else if (!p->setup_func) { |
205 | printk(KERN_WARNING "Parameter %s is obsolete," | |
206 | " ignored\n", p->str); | |
207 | return 1; | |
208 | } else if (p->setup_func(line + n)) | |
209 | return 1; | |
210 | } | |
211 | p++; | |
212 | } while (p < __setup_end); | |
33df0d19 RR |
213 | |
214 | return had_early_param; | |
1da177e4 LT |
215 | } |
216 | ||
217 | /* | |
218 | * This should be approx 2 Bo*oMips to start (note initial shift), and will | |
219 | * still work even if initially too large, it will just take slightly longer | |
220 | */ | |
221 | unsigned long loops_per_jiffy = (1<<12); | |
222 | ||
223 | EXPORT_SYMBOL(loops_per_jiffy); | |
224 | ||
225 | static int __init debug_kernel(char *str) | |
226 | { | |
227 | if (*str) | |
228 | return 0; | |
229 | console_loglevel = 10; | |
230 | return 1; | |
231 | } | |
232 | ||
233 | static int __init quiet_kernel(char *str) | |
234 | { | |
235 | if (*str) | |
236 | return 0; | |
237 | console_loglevel = 4; | |
238 | return 1; | |
239 | } | |
240 | ||
241 | __setup("debug", debug_kernel); | |
242 | __setup("quiet", quiet_kernel); | |
243 | ||
244 | static int __init loglevel(char *str) | |
245 | { | |
246 | get_option(&str, &console_loglevel); | |
247 | return 1; | |
248 | } | |
249 | ||
250 | __setup("loglevel=", loglevel); | |
251 | ||
252 | /* | |
253 | * Unknown boot options get handed to init, unless they look like | |
254 | * failed parameters | |
255 | */ | |
256 | static 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 | ||
275 | /* | |
276 | * Preemptive maintenance for "why didn't my mispelled command | |
277 | * line work?" | |
278 | */ | |
279 | if (strchr(param, '.') && (!val || strchr(param, '.') < val)) { | |
280 | printk(KERN_ERR "Unknown boot option `%s': ignoring\n", param); | |
281 | return 0; | |
282 | } | |
283 | ||
284 | if (panic_later) | |
285 | return 0; | |
286 | ||
287 | if (val) { | |
288 | /* Environment option */ | |
289 | unsigned int i; | |
290 | for (i = 0; envp_init[i]; i++) { | |
291 | if (i == MAX_INIT_ENVS) { | |
292 | panic_later = "Too many boot env vars at `%s'"; | |
293 | panic_param = param; | |
294 | } | |
295 | if (!strncmp(param, envp_init[i], val - param)) | |
296 | break; | |
297 | } | |
298 | envp_init[i] = param; | |
299 | } else { | |
300 | /* Command line option */ | |
301 | unsigned int i; | |
302 | for (i = 0; argv_init[i]; i++) { | |
303 | if (i == MAX_INIT_ARGS) { | |
304 | panic_later = "Too many boot init vars at `%s'"; | |
305 | panic_param = param; | |
306 | } | |
307 | } | |
308 | argv_init[i] = param; | |
309 | } | |
310 | return 0; | |
311 | } | |
312 | ||
313 | static int __init init_setup(char *str) | |
314 | { | |
315 | unsigned int i; | |
316 | ||
317 | execute_command = str; | |
318 | /* | |
319 | * In case LILO is going to boot us with default command line, | |
320 | * it prepends "auto" before the whole cmdline which makes | |
321 | * the shell think it should execute a script with such name. | |
322 | * So we ignore all arguments entered _before_ init=... [MJ] | |
323 | */ | |
324 | for (i = 1; i < MAX_INIT_ARGS; i++) | |
325 | argv_init[i] = NULL; | |
326 | return 1; | |
327 | } | |
328 | __setup("init=", init_setup); | |
329 | ||
ffdfc409 OJ |
330 | static int __init rdinit_setup(char *str) |
331 | { | |
332 | unsigned int i; | |
333 | ||
334 | ramdisk_execute_command = str; | |
335 | /* See "auto" comment in init_setup */ | |
336 | for (i = 1; i < MAX_INIT_ARGS; i++) | |
337 | argv_init[i] = NULL; | |
338 | return 1; | |
339 | } | |
340 | __setup("rdinit=", rdinit_setup); | |
341 | ||
1da177e4 LT |
342 | #ifndef CONFIG_SMP |
343 | ||
344 | #ifdef CONFIG_X86_LOCAL_APIC | |
345 | static void __init smp_init(void) | |
346 | { | |
347 | APIC_init_uniprocessor(); | |
348 | } | |
349 | #else | |
350 | #define smp_init() do { } while (0) | |
351 | #endif | |
352 | ||
353 | static inline void setup_per_cpu_areas(void) { } | |
354 | static inline void smp_prepare_cpus(unsigned int maxcpus) { } | |
355 | ||
356 | #else | |
357 | ||
358 | #ifdef __GENERIC_PER_CPU | |
b73b459f | 359 | unsigned long __per_cpu_offset[NR_CPUS] __read_mostly; |
1da177e4 LT |
360 | |
361 | EXPORT_SYMBOL(__per_cpu_offset); | |
362 | ||
363 | static void __init setup_per_cpu_areas(void) | |
364 | { | |
365 | unsigned long size, i; | |
366 | char *ptr; | |
63872f87 | 367 | unsigned long nr_possible_cpus = num_possible_cpus(); |
1da177e4 LT |
368 | |
369 | /* Copy section for each CPU (we discard the original) */ | |
370 | size = ALIGN(__per_cpu_end - __per_cpu_start, SMP_CACHE_BYTES); | |
371 | #ifdef CONFIG_MODULES | |
372 | if (size < PERCPU_ENOUGH_ROOM) | |
373 | size = PERCPU_ENOUGH_ROOM; | |
374 | #endif | |
63872f87 | 375 | ptr = alloc_bootmem(size * nr_possible_cpus); |
1da177e4 | 376 | |
0a945022 | 377 | for_each_possible_cpu(i) { |
1da177e4 LT |
378 | __per_cpu_offset[i] = ptr - __per_cpu_start; |
379 | memcpy(ptr, __per_cpu_start, __per_cpu_end - __per_cpu_start); | |
63872f87 | 380 | ptr += size; |
1da177e4 LT |
381 | } |
382 | } | |
383 | #endif /* !__GENERIC_PER_CPU */ | |
384 | ||
385 | /* Called by boot processor to activate the rest. */ | |
386 | static void __init smp_init(void) | |
387 | { | |
388 | unsigned int i; | |
389 | ||
390 | /* FIXME: This should be done in userspace --RR */ | |
391 | for_each_present_cpu(i) { | |
392 | if (num_online_cpus() >= max_cpus) | |
393 | break; | |
394 | if (!cpu_online(i)) | |
395 | cpu_up(i); | |
396 | } | |
397 | ||
398 | /* Any cleanup work */ | |
399 | printk(KERN_INFO "Brought up %ld CPUs\n", (long)num_online_cpus()); | |
400 | smp_cpus_done(max_cpus); | |
401 | #if 0 | |
402 | /* Get other processors into their bootup holding patterns. */ | |
403 | ||
404 | smp_commence(); | |
405 | #endif | |
406 | } | |
407 | ||
408 | #endif | |
409 | ||
410 | /* | |
411 | * We need to finalize in a non-__init function or else race conditions | |
412 | * between the root thread and the init thread may cause start_kernel to | |
413 | * be reaped by free_initmem before the root thread has proceeded to | |
414 | * cpu_idle. | |
415 | * | |
416 | * gcc-3.4 accidentally inlines this function, so use noinline. | |
417 | */ | |
418 | ||
419 | static void noinline rest_init(void) | |
420 | __releases(kernel_lock) | |
421 | { | |
422 | kernel_thread(init, NULL, CLONE_FS | CLONE_SIGHAND); | |
423 | numa_default_policy(); | |
424 | unlock_kernel(); | |
f340c0d1 IM |
425 | |
426 | /* | |
427 | * The boot idle thread must execute schedule() | |
428 | * at least one to get things moving: | |
429 | */ | |
5bfb5d69 | 430 | preempt_enable_no_resched(); |
f340c0d1 | 431 | schedule(); |
5bfb5d69 | 432 | preempt_disable(); |
f340c0d1 | 433 | |
5bfb5d69 | 434 | /* Call into cpu_idle with preempt disabled */ |
1da177e4 LT |
435 | cpu_idle(); |
436 | } | |
437 | ||
438 | /* Check for early params. */ | |
439 | static int __init do_early_param(char *param, char *val) | |
440 | { | |
441 | struct obs_kernel_param *p; | |
442 | ||
443 | for (p = __setup_start; p < __setup_end; p++) { | |
444 | if (p->early && strcmp(param, p->str) == 0) { | |
445 | if (p->setup_func(val) != 0) | |
446 | printk(KERN_WARNING | |
447 | "Malformed early option '%s'\n", param); | |
448 | } | |
449 | } | |
450 | /* We accept everything at this stage. */ | |
451 | return 0; | |
452 | } | |
453 | ||
454 | /* Arch code calls this early on, or if not, just before other parsing. */ | |
455 | void __init parse_early_param(void) | |
456 | { | |
457 | static __initdata int done = 0; | |
458 | static __initdata char tmp_cmdline[COMMAND_LINE_SIZE]; | |
459 | ||
460 | if (done) | |
461 | return; | |
462 | ||
463 | /* All fall through to do_early_param. */ | |
464 | strlcpy(tmp_cmdline, saved_command_line, COMMAND_LINE_SIZE); | |
465 | parse_args("early options", tmp_cmdline, NULL, 0, do_early_param); | |
466 | done = 1; | |
467 | } | |
468 | ||
469 | /* | |
470 | * Activate the first processor. | |
471 | */ | |
472 | ||
44fd2299 SS |
473 | static void __init boot_cpu_init(void) |
474 | { | |
475 | int cpu = smp_processor_id(); | |
476 | /* Mark the boot cpu "present", "online" etc for SMP and UP case */ | |
477 | cpu_set(cpu, cpu_online_map); | |
478 | cpu_set(cpu, cpu_present_map); | |
479 | cpu_set(cpu, cpu_possible_map); | |
480 | } | |
481 | ||
033ab7f8 AM |
482 | void __init __attribute__((weak)) smp_setup_processor_id(void) |
483 | { | |
484 | } | |
485 | ||
1da177e4 LT |
486 | asmlinkage void __init start_kernel(void) |
487 | { | |
488 | char * command_line; | |
489 | extern struct kernel_param __start___param[], __stop___param[]; | |
033ab7f8 AM |
490 | |
491 | smp_setup_processor_id(); | |
492 | ||
fbb9ce95 IM |
493 | /* |
494 | * Need to run as early as possible, to initialize the | |
495 | * lockdep hash: | |
496 | */ | |
c9538ed4 | 497 | unwind_init(); |
fbb9ce95 IM |
498 | lockdep_init(); |
499 | ||
500 | local_irq_disable(); | |
501 | early_boot_irqs_off(); | |
243c7621 | 502 | early_init_irq_lock_class(); |
fbb9ce95 | 503 | |
1da177e4 LT |
504 | /* |
505 | * Interrupts are still disabled. Do necessary setups, then | |
506 | * enable them | |
507 | */ | |
508 | lock_kernel(); | |
44fd2299 | 509 | boot_cpu_init(); |
1da177e4 LT |
510 | page_address_init(); |
511 | printk(KERN_NOTICE); | |
a2ee8649 | 512 | printk(linux_banner, UTS_RELEASE, UTS_VERSION); |
1da177e4 | 513 | setup_arch(&command_line); |
690a973f | 514 | unwind_setup(); |
1da177e4 | 515 | setup_per_cpu_areas(); |
44fd2299 | 516 | smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */ |
1da177e4 LT |
517 | |
518 | /* | |
519 | * Set up the scheduler prior starting any interrupts (such as the | |
520 | * timer interrupt). Full topology setup happens at smp_init() | |
521 | * time - but meanwhile we still have a functioning scheduler. | |
522 | */ | |
523 | sched_init(); | |
524 | /* | |
525 | * Disable preemption - early bootup scheduling is extremely | |
526 | * fragile until we cpu_idle() for the first time. | |
527 | */ | |
528 | preempt_disable(); | |
529 | build_all_zonelists(); | |
530 | page_alloc_init(); | |
531 | printk(KERN_NOTICE "Kernel command line: %s\n", saved_command_line); | |
532 | parse_early_param(); | |
533 | parse_args("Booting kernel", command_line, __start___param, | |
534 | __stop___param - __start___param, | |
535 | &unknown_bootoption); | |
536 | sort_main_extable(); | |
537 | trap_init(); | |
538 | rcu_init(); | |
539 | init_IRQ(); | |
540 | pidhash_init(); | |
541 | init_timers(); | |
c0a31329 | 542 | hrtimers_init(); |
1da177e4 | 543 | softirq_init(); |
ad596171 | 544 | timekeeping_init(); |
88fecaa2 | 545 | time_init(); |
93e02814 HC |
546 | profile_init(); |
547 | if (!irqs_disabled()) | |
548 | printk("start_kernel(): bug: interrupts were enabled early\n"); | |
fbb9ce95 | 549 | early_boot_irqs_on(); |
93e02814 | 550 | local_irq_enable(); |
1da177e4 LT |
551 | |
552 | /* | |
553 | * HACK ALERT! This is early. We're enabling the console before | |
554 | * we've done PCI setups etc, and console_init() must be aware of | |
555 | * this. But we do want output early, in case something goes wrong. | |
556 | */ | |
557 | console_init(); | |
558 | if (panic_later) | |
559 | panic(panic_later, panic_param); | |
fbb9ce95 IM |
560 | |
561 | lockdep_info(); | |
562 | ||
9a11b49a IM |
563 | /* |
564 | * Need to run this when irqs are enabled, because it wants | |
565 | * to self-test [hard/soft]-irqs on/off lock inversion bugs | |
566 | * too: | |
567 | */ | |
568 | locking_selftest(); | |
569 | ||
1da177e4 LT |
570 | #ifdef CONFIG_BLK_DEV_INITRD |
571 | if (initrd_start && !initrd_below_start_ok && | |
572 | initrd_start < min_low_pfn << PAGE_SHIFT) { | |
573 | printk(KERN_CRIT "initrd overwritten (0x%08lx < 0x%08lx) - " | |
574 | "disabling it.\n",initrd_start,min_low_pfn << PAGE_SHIFT); | |
575 | initrd_start = 0; | |
576 | } | |
577 | #endif | |
578 | vfs_caches_init_early(); | |
c417f024 | 579 | cpuset_init_early(); |
1da177e4 LT |
580 | mem_init(); |
581 | kmem_cache_init(); | |
e7c8d5c9 | 582 | setup_per_cpu_pageset(); |
1da177e4 LT |
583 | numa_policy_init(); |
584 | if (late_time_init) | |
585 | late_time_init(); | |
586 | calibrate_delay(); | |
587 | pidmap_init(); | |
588 | pgtable_cache_init(); | |
589 | prio_tree_init(); | |
590 | anon_vma_init(); | |
591 | #ifdef CONFIG_X86 | |
592 | if (efi_enabled) | |
593 | efi_enter_virtual_mode(); | |
594 | #endif | |
595 | fork_init(num_physpages); | |
596 | proc_caches_init(); | |
597 | buffer_init(); | |
598 | unnamed_dev_init(); | |
599 | key_init(); | |
600 | security_init(); | |
601 | vfs_caches_init(num_physpages); | |
602 | radix_tree_init(); | |
603 | signals_init(); | |
604 | /* rootfs populating might need page-writeback */ | |
605 | page_writeback_init(); | |
606 | #ifdef CONFIG_PROC_FS | |
607 | proc_root_init(); | |
608 | #endif | |
609 | cpuset_init(); | |
c757249a | 610 | taskstats_init_early(); |
ca74e92b | 611 | delayacct_init(); |
1da177e4 LT |
612 | |
613 | check_bugs(); | |
614 | ||
615 | acpi_early_init(); /* before LAPIC and SMP init */ | |
616 | ||
617 | /* Do the rest non-__init'ed, we're now alive */ | |
618 | rest_init(); | |
619 | } | |
620 | ||
621 | static int __initdata initcall_debug; | |
622 | ||
623 | static int __init initcall_debug_setup(char *str) | |
624 | { | |
625 | initcall_debug = 1; | |
626 | return 1; | |
627 | } | |
628 | __setup("initcall_debug", initcall_debug_setup); | |
629 | ||
1da177e4 LT |
630 | extern initcall_t __initcall_start[], __initcall_end[]; |
631 | ||
632 | static void __init do_initcalls(void) | |
633 | { | |
634 | initcall_t *call; | |
635 | int count = preempt_count(); | |
636 | ||
637 | for (call = __initcall_start; call < __initcall_end; call++) { | |
c1cda48a AM |
638 | char *msg = NULL; |
639 | char msgbuf[40]; | |
640 | int result; | |
1da177e4 LT |
641 | |
642 | if (initcall_debug) { | |
c1cda48a AM |
643 | printk("Calling initcall 0x%p", *call); |
644 | print_fn_descriptor_symbol(": %s()", | |
645 | (unsigned long) *call); | |
1da177e4 LT |
646 | printk("\n"); |
647 | } | |
648 | ||
c1cda48a | 649 | result = (*call)(); |
1da177e4 | 650 | |
f3a19cb4 | 651 | if (result && result != -ENODEV && initcall_debug) { |
c1cda48a AM |
652 | sprintf(msgbuf, "error code %d", result); |
653 | msg = msgbuf; | |
654 | } | |
1da177e4 LT |
655 | if (preempt_count() != count) { |
656 | msg = "preemption imbalance"; | |
657 | preempt_count() = count; | |
658 | } | |
659 | if (irqs_disabled()) { | |
660 | msg = "disabled interrupts"; | |
661 | local_irq_enable(); | |
662 | } | |
663 | if (msg) { | |
c1cda48a AM |
664 | printk(KERN_WARNING "initcall at 0x%p", *call); |
665 | print_fn_descriptor_symbol(": %s()", | |
666 | (unsigned long) *call); | |
667 | printk(": returned with %s\n", msg); | |
1da177e4 LT |
668 | } |
669 | } | |
670 | ||
671 | /* Make sure there is no pending stuff from the initcall sequence */ | |
672 | flush_scheduled_work(); | |
673 | } | |
674 | ||
675 | /* | |
676 | * Ok, the machine is now initialized. None of the devices | |
677 | * have been touched yet, but the CPU subsystem is up and | |
678 | * running, and memory and process management works. | |
679 | * | |
680 | * Now we can finally start doing some real work.. | |
681 | */ | |
682 | static void __init do_basic_setup(void) | |
683 | { | |
684 | /* drivers will send hotplug events */ | |
685 | init_workqueues(); | |
686 | usermodehelper_init(); | |
687 | driver_init(); | |
688 | ||
689 | #ifdef CONFIG_SYSCTL | |
690 | sysctl_init(); | |
691 | #endif | |
692 | ||
1da177e4 LT |
693 | do_initcalls(); |
694 | } | |
695 | ||
696 | static void do_pre_smp_initcalls(void) | |
697 | { | |
698 | extern int spawn_ksoftirqd(void); | |
699 | #ifdef CONFIG_SMP | |
700 | extern int migration_init(void); | |
701 | ||
702 | migration_init(); | |
703 | #endif | |
704 | spawn_ksoftirqd(); | |
8446f1d3 | 705 | spawn_softlockup_task(); |
1da177e4 LT |
706 | } |
707 | ||
708 | static void run_init_process(char *init_filename) | |
709 | { | |
710 | argv_init[0] = init_filename; | |
67608567 | 711 | kernel_execve(init_filename, argv_init, envp_init); |
1da177e4 LT |
712 | } |
713 | ||
1da177e4 LT |
714 | static int init(void * unused) |
715 | { | |
716 | lock_kernel(); | |
717 | /* | |
718 | * init can run on any cpu. | |
719 | */ | |
720 | set_cpus_allowed(current, CPU_MASK_ALL); | |
721 | /* | |
722 | * Tell the world that we're going to be the grim | |
723 | * reaper of innocent orphaned children. | |
724 | * | |
725 | * We don't want people to have to make incorrect | |
726 | * assumptions about where in the task array this | |
727 | * can be found. | |
728 | */ | |
84d73786 | 729 | init_pid_ns.child_reaper = current; |
1da177e4 | 730 | |
9ec52099 CLG |
731 | cad_pid = task_pid(current); |
732 | ||
1da177e4 LT |
733 | smp_prepare_cpus(max_cpus); |
734 | ||
735 | do_pre_smp_initcalls(); | |
736 | ||
1da177e4 LT |
737 | smp_init(); |
738 | sched_init_smp(); | |
739 | ||
740 | cpuset_init_smp(); | |
741 | ||
742 | /* | |
743 | * Do this before initcalls, because some drivers want to access | |
744 | * firmware files. | |
745 | */ | |
746 | populate_rootfs(); | |
747 | ||
748 | do_basic_setup(); | |
749 | ||
750 | /* | |
751 | * check if there is an early userspace init. If yes, let it do all | |
752 | * the work | |
753 | */ | |
ffdfc409 OJ |
754 | |
755 | if (!ramdisk_execute_command) | |
756 | ramdisk_execute_command = "/init"; | |
757 | ||
758 | if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) { | |
759 | ramdisk_execute_command = NULL; | |
1da177e4 | 760 | prepare_namespace(); |
ffdfc409 | 761 | } |
1da177e4 LT |
762 | |
763 | /* | |
764 | * Ok, we have completed the initial bootup, and | |
765 | * we're essentially up and running. Get rid of the | |
766 | * initmem segments and start the user-mode stuff.. | |
767 | */ | |
768 | free_initmem(); | |
769 | unlock_kernel(); | |
37b73c82 | 770 | mark_rodata_ro(); |
1da177e4 LT |
771 | system_state = SYSTEM_RUNNING; |
772 | numa_default_policy(); | |
773 | ||
774 | if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0) | |
775 | printk(KERN_WARNING "Warning: unable to open an initial console.\n"); | |
776 | ||
777 | (void) sys_dup(0); | |
778 | (void) sys_dup(0); | |
ffdfc409 OJ |
779 | |
780 | if (ramdisk_execute_command) { | |
781 | run_init_process(ramdisk_execute_command); | |
782 | printk(KERN_WARNING "Failed to execute %s\n", | |
783 | ramdisk_execute_command); | |
784 | } | |
785 | ||
1da177e4 LT |
786 | /* |
787 | * We try each of these until one succeeds. | |
788 | * | |
789 | * The Bourne shell can be used instead of init if we are | |
790 | * trying to recover a really broken machine. | |
791 | */ | |
c1d7ef70 | 792 | if (execute_command) { |
1da177e4 | 793 | run_init_process(execute_command); |
c1d7ef70 AB |
794 | printk(KERN_WARNING "Failed to execute %s. Attempting " |
795 | "defaults...\n", execute_command); | |
796 | } | |
1da177e4 LT |
797 | run_init_process("/sbin/init"); |
798 | run_init_process("/etc/init"); | |
799 | run_init_process("/bin/init"); | |
800 | run_init_process("/bin/sh"); | |
801 | ||
802 | panic("No init found. Try passing init= option to kernel."); | |
803 | } |