]>
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> | |
906568c9 | 43 | #include <linux/tick.h> |
6168a702 | 44 | #include <linux/interrupt.h> |
c757249a | 45 | #include <linux/taskstats_kern.h> |
ca74e92b | 46 | #include <linux/delayacct.h> |
1da177e4 LT |
47 | #include <linux/unistd.h> |
48 | #include <linux/rmap.h> | |
49 | #include <linux/mempolicy.h> | |
50 | #include <linux/key.h> | |
4552d5dc | 51 | #include <linux/unwind.h> |
b6cd0b77 | 52 | #include <linux/buffer_head.h> |
9a11b49a | 53 | #include <linux/debug_locks.h> |
fbb9ce95 | 54 | #include <linux/lockdep.h> |
84d73786 | 55 | #include <linux/pid_namespace.h> |
1f21782e | 56 | #include <linux/device.h> |
73c27992 | 57 | #include <linux/kthread.h> |
1da177e4 LT |
58 | |
59 | #include <asm/io.h> | |
60 | #include <asm/bugs.h> | |
61 | #include <asm/setup.h> | |
a940199f | 62 | #include <asm/sections.h> |
37b73c82 | 63 | #include <asm/cacheflush.h> |
1da177e4 | 64 | |
1da177e4 LT |
65 | #ifdef CONFIG_X86_LOCAL_APIC |
66 | #include <asm/smp.h> | |
67 | #endif | |
68 | ||
69 | /* | |
69c99ac1 CQH |
70 | * This is one of the first .c files built. Error out early if we have compiler |
71 | * trouble. | |
72 | * | |
73 | * Versions of gcc older than that listed below may actually compile and link | |
74 | * okay, but the end product can have subtle run time bugs. To avoid associated | |
75 | * bogus bug reports, we flatly refuse to compile with a gcc that is known to be | |
76 | * too old from the very beginning. | |
1da177e4 | 77 | */ |
fd285bb5 | 78 | #if (__GNUC__ < 3) || (__GNUC__ == 3 && __GNUC_MINOR__ < 2) |
1da177e4 LT |
79 | #error Sorry, your GCC is too old. It builds incorrect kernels. |
80 | #endif | |
81 | ||
f1a60dbf AM |
82 | #if __GNUC__ == 4 && __GNUC_MINOR__ == 1 && __GNUC_PATCHLEVEL__ == 0 |
83 | #warning gcc-4.1.0 is known to miscompile the kernel. A different compiler version is recommended. | |
84 | #endif | |
85 | ||
aae5f662 | 86 | static int kernel_init(void *); |
1da177e4 LT |
87 | |
88 | extern void init_IRQ(void); | |
1da177e4 LT |
89 | extern void fork_init(unsigned long); |
90 | extern void mca_init(void); | |
91 | extern void sbus_init(void); | |
1da177e4 | 92 | extern void signals_init(void); |
1da177e4 LT |
93 | extern void pidhash_init(void); |
94 | extern void pidmap_init(void); | |
95 | extern void prio_tree_init(void); | |
96 | extern void radix_tree_init(void); | |
97 | extern void free_initmem(void); | |
1da177e4 LT |
98 | #ifdef CONFIG_ACPI |
99 | extern void acpi_early_init(void); | |
100 | #else | |
101 | static inline void acpi_early_init(void) { } | |
102 | #endif | |
37b73c82 AV |
103 | #ifndef CONFIG_DEBUG_RODATA |
104 | static inline void mark_rodata_ro(void) { } | |
105 | #endif | |
1da177e4 LT |
106 | |
107 | #ifdef CONFIG_TC | |
108 | extern void tc_init(void); | |
109 | #endif | |
110 | ||
111 | enum system_states system_state; | |
112 | EXPORT_SYMBOL(system_state); | |
113 | ||
114 | /* | |
115 | * Boot command-line arguments | |
116 | */ | |
117 | #define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT | |
118 | #define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT | |
119 | ||
120 | extern void time_init(void); | |
121 | /* Default late time init is NULL. archs can override this later. */ | |
122 | void (*late_time_init)(void); | |
123 | extern void softirq_init(void); | |
124 | ||
30d7e0d4 ABL |
125 | /* Untouched command line saved by arch-specific code. */ |
126 | char __initdata boot_command_line[COMMAND_LINE_SIZE]; | |
127 | /* Untouched saved command line (eg. for /proc) */ | |
128 | char *saved_command_line; | |
129 | /* Command line for parameter parsing */ | |
130 | static char *static_command_line; | |
1da177e4 LT |
131 | |
132 | static char *execute_command; | |
ffdfc409 | 133 | static char *ramdisk_execute_command; |
1da177e4 LT |
134 | |
135 | /* Setup configured maximum number of CPUs to activate */ | |
136 | static unsigned int max_cpus = NR_CPUS; | |
137 | ||
7e96287d VG |
138 | /* |
139 | * If set, this is an indication to the drivers that reset the underlying | |
140 | * device before going ahead with the initialization otherwise driver might | |
141 | * rely on the BIOS and skip the reset operation. | |
142 | * | |
143 | * This is useful if kernel is booting in an unreliable environment. | |
144 | * For ex. kdump situaiton where previous kernel has crashed, BIOS has been | |
145 | * skipped and devices will be in unknown state. | |
146 | */ | |
147 | unsigned int reset_devices; | |
148 | EXPORT_SYMBOL(reset_devices); | |
149 | ||
1da177e4 LT |
150 | /* |
151 | * Setup routine for controlling SMP activation | |
152 | * | |
153 | * Command-line option of "nosmp" or "maxcpus=0" will disable SMP | |
154 | * activation entirely (the MPS table probe still happens, though). | |
155 | * | |
156 | * Command-line option of "maxcpus=<NUM>", where <NUM> is an integer | |
157 | * greater than 0, limits the maximum number of CPUs activated in | |
158 | * SMP mode to <NUM>. | |
159 | */ | |
160 | static int __init nosmp(char *str) | |
161 | { | |
162 | max_cpus = 0; | |
163 | return 1; | |
164 | } | |
165 | ||
166 | __setup("nosmp", nosmp); | |
167 | ||
168 | static int __init maxcpus(char *str) | |
169 | { | |
170 | get_option(&str, &max_cpus); | |
171 | return 1; | |
172 | } | |
173 | ||
174 | __setup("maxcpus=", maxcpus); | |
175 | ||
7e96287d VG |
176 | static int __init set_reset_devices(char *str) |
177 | { | |
178 | reset_devices = 1; | |
179 | return 1; | |
180 | } | |
181 | ||
182 | __setup("reset_devices", set_reset_devices); | |
183 | ||
1da177e4 LT |
184 | static char * argv_init[MAX_INIT_ARGS+2] = { "init", NULL, }; |
185 | char * envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, }; | |
186 | static const char *panic_later, *panic_param; | |
187 | ||
188 | extern struct obs_kernel_param __setup_start[], __setup_end[]; | |
189 | ||
190 | static int __init obsolete_checksetup(char *line) | |
191 | { | |
192 | struct obs_kernel_param *p; | |
33df0d19 | 193 | int had_early_param = 0; |
1da177e4 LT |
194 | |
195 | p = __setup_start; | |
196 | do { | |
197 | int n = strlen(p->str); | |
198 | if (!strncmp(line, p->str, n)) { | |
199 | if (p->early) { | |
33df0d19 RR |
200 | /* Already done in parse_early_param? |
201 | * (Needs exact match on param part). | |
202 | * Keep iterating, as we can have early | |
203 | * params and __setups of same names 8( */ | |
1da177e4 | 204 | if (line[n] == '\0' || line[n] == '=') |
33df0d19 | 205 | had_early_param = 1; |
1da177e4 LT |
206 | } else if (!p->setup_func) { |
207 | printk(KERN_WARNING "Parameter %s is obsolete," | |
208 | " ignored\n", p->str); | |
209 | return 1; | |
210 | } else if (p->setup_func(line + n)) | |
211 | return 1; | |
212 | } | |
213 | p++; | |
214 | } while (p < __setup_end); | |
33df0d19 RR |
215 | |
216 | return had_early_param; | |
1da177e4 LT |
217 | } |
218 | ||
219 | /* | |
220 | * This should be approx 2 Bo*oMips to start (note initial shift), and will | |
221 | * still work even if initially too large, it will just take slightly longer | |
222 | */ | |
223 | unsigned long loops_per_jiffy = (1<<12); | |
224 | ||
225 | EXPORT_SYMBOL(loops_per_jiffy); | |
226 | ||
227 | static int __init debug_kernel(char *str) | |
228 | { | |
229 | if (*str) | |
230 | return 0; | |
231 | console_loglevel = 10; | |
232 | return 1; | |
233 | } | |
234 | ||
235 | static int __init quiet_kernel(char *str) | |
236 | { | |
237 | if (*str) | |
238 | return 0; | |
239 | console_loglevel = 4; | |
240 | return 1; | |
241 | } | |
242 | ||
243 | __setup("debug", debug_kernel); | |
244 | __setup("quiet", quiet_kernel); | |
245 | ||
246 | static int __init loglevel(char *str) | |
247 | { | |
248 | get_option(&str, &console_loglevel); | |
249 | return 1; | |
250 | } | |
251 | ||
252 | __setup("loglevel=", loglevel); | |
253 | ||
254 | /* | |
255 | * Unknown boot options get handed to init, unless they look like | |
256 | * failed parameters | |
257 | */ | |
258 | static int __init unknown_bootoption(char *param, char *val) | |
259 | { | |
260 | /* Change NUL term back to "=", to make "param" the whole string. */ | |
261 | if (val) { | |
262 | /* param=val or param="val"? */ | |
263 | if (val == param+strlen(param)+1) | |
264 | val[-1] = '='; | |
265 | else if (val == param+strlen(param)+2) { | |
266 | val[-2] = '='; | |
267 | memmove(val-1, val, strlen(val)+1); | |
268 | val--; | |
269 | } else | |
270 | BUG(); | |
271 | } | |
272 | ||
273 | /* Handle obsolete-style parameters */ | |
274 | if (obsolete_checksetup(param)) | |
275 | return 0; | |
276 | ||
277 | /* | |
278 | * Preemptive maintenance for "why didn't my mispelled command | |
279 | * line work?" | |
280 | */ | |
281 | if (strchr(param, '.') && (!val || strchr(param, '.') < val)) { | |
282 | printk(KERN_ERR "Unknown boot option `%s': ignoring\n", param); | |
283 | return 0; | |
284 | } | |
285 | ||
286 | if (panic_later) | |
287 | return 0; | |
288 | ||
289 | if (val) { | |
290 | /* Environment option */ | |
291 | unsigned int i; | |
292 | for (i = 0; envp_init[i]; i++) { | |
293 | if (i == MAX_INIT_ENVS) { | |
294 | panic_later = "Too many boot env vars at `%s'"; | |
295 | panic_param = param; | |
296 | } | |
297 | if (!strncmp(param, envp_init[i], val - param)) | |
298 | break; | |
299 | } | |
300 | envp_init[i] = param; | |
301 | } else { | |
302 | /* Command line option */ | |
303 | unsigned int i; | |
304 | for (i = 0; argv_init[i]; i++) { | |
305 | if (i == MAX_INIT_ARGS) { | |
306 | panic_later = "Too many boot init vars at `%s'"; | |
307 | panic_param = param; | |
308 | } | |
309 | } | |
310 | argv_init[i] = param; | |
311 | } | |
312 | return 0; | |
313 | } | |
314 | ||
315 | static int __init init_setup(char *str) | |
316 | { | |
317 | unsigned int i; | |
318 | ||
319 | execute_command = str; | |
320 | /* | |
321 | * In case LILO is going to boot us with default command line, | |
322 | * it prepends "auto" before the whole cmdline which makes | |
323 | * the shell think it should execute a script with such name. | |
324 | * So we ignore all arguments entered _before_ init=... [MJ] | |
325 | */ | |
326 | for (i = 1; i < MAX_INIT_ARGS; i++) | |
327 | argv_init[i] = NULL; | |
328 | return 1; | |
329 | } | |
330 | __setup("init=", init_setup); | |
331 | ||
ffdfc409 OJ |
332 | static int __init rdinit_setup(char *str) |
333 | { | |
334 | unsigned int i; | |
335 | ||
336 | ramdisk_execute_command = str; | |
337 | /* See "auto" comment in init_setup */ | |
338 | for (i = 1; i < MAX_INIT_ARGS; i++) | |
339 | argv_init[i] = NULL; | |
340 | return 1; | |
341 | } | |
342 | __setup("rdinit=", rdinit_setup); | |
343 | ||
1da177e4 LT |
344 | #ifndef CONFIG_SMP |
345 | ||
346 | #ifdef CONFIG_X86_LOCAL_APIC | |
347 | static void __init smp_init(void) | |
348 | { | |
349 | APIC_init_uniprocessor(); | |
350 | } | |
351 | #else | |
352 | #define smp_init() do { } while (0) | |
353 | #endif | |
354 | ||
355 | static inline void setup_per_cpu_areas(void) { } | |
356 | static inline void smp_prepare_cpus(unsigned int maxcpus) { } | |
357 | ||
358 | #else | |
359 | ||
360 | #ifdef __GENERIC_PER_CPU | |
b73b459f | 361 | unsigned long __per_cpu_offset[NR_CPUS] __read_mostly; |
1da177e4 LT |
362 | |
363 | EXPORT_SYMBOL(__per_cpu_offset); | |
364 | ||
365 | static void __init setup_per_cpu_areas(void) | |
366 | { | |
367 | unsigned long size, i; | |
368 | char *ptr; | |
63872f87 | 369 | unsigned long nr_possible_cpus = num_possible_cpus(); |
1da177e4 LT |
370 | |
371 | /* Copy section for each CPU (we discard the original) */ | |
b6e3590f JF |
372 | size = ALIGN(PERCPU_ENOUGH_ROOM, PAGE_SIZE); |
373 | ptr = alloc_bootmem_pages(size * nr_possible_cpus); | |
1da177e4 | 374 | |
0a945022 | 375 | for_each_possible_cpu(i) { |
1da177e4 LT |
376 | __per_cpu_offset[i] = ptr - __per_cpu_start; |
377 | memcpy(ptr, __per_cpu_start, __per_cpu_end - __per_cpu_start); | |
63872f87 | 378 | ptr += size; |
1da177e4 LT |
379 | } |
380 | } | |
381 | #endif /* !__GENERIC_PER_CPU */ | |
382 | ||
383 | /* Called by boot processor to activate the rest. */ | |
384 | static void __init smp_init(void) | |
385 | { | |
53b8a315 | 386 | unsigned int cpu; |
1da177e4 LT |
387 | |
388 | /* FIXME: This should be done in userspace --RR */ | |
53b8a315 | 389 | for_each_present_cpu(cpu) { |
1da177e4 LT |
390 | if (num_online_cpus() >= max_cpus) |
391 | break; | |
53b8a315 CL |
392 | if (!cpu_online(cpu)) |
393 | cpu_up(cpu); | |
1da177e4 LT |
394 | } |
395 | ||
396 | /* Any cleanup work */ | |
397 | printk(KERN_INFO "Brought up %ld CPUs\n", (long)num_online_cpus()); | |
398 | smp_cpus_done(max_cpus); | |
1da177e4 LT |
399 | } |
400 | ||
401 | #endif | |
402 | ||
30d7e0d4 ABL |
403 | /* |
404 | * We need to store the untouched command line for future reference. | |
405 | * We also need to store the touched command line since the parameter | |
406 | * parsing is performed in place, and we should allow a component to | |
407 | * store reference of name/value for future reference. | |
408 | */ | |
409 | static void __init setup_command_line(char *command_line) | |
410 | { | |
411 | saved_command_line = alloc_bootmem(strlen (boot_command_line)+1); | |
412 | static_command_line = alloc_bootmem(strlen (command_line)+1); | |
413 | strcpy (saved_command_line, boot_command_line); | |
414 | strcpy (static_command_line, command_line); | |
415 | } | |
416 | ||
1da177e4 LT |
417 | /* |
418 | * We need to finalize in a non-__init function or else race conditions | |
419 | * between the root thread and the init thread may cause start_kernel to | |
420 | * be reaped by free_initmem before the root thread has proceeded to | |
421 | * cpu_idle. | |
422 | * | |
423 | * gcc-3.4 accidentally inlines this function, so use noinline. | |
424 | */ | |
425 | ||
426 | static void noinline rest_init(void) | |
427 | __releases(kernel_lock) | |
428 | { | |
73c27992 EB |
429 | int pid; |
430 | ||
aae5f662 | 431 | kernel_thread(kernel_init, NULL, CLONE_FS | CLONE_SIGHAND); |
1da177e4 | 432 | numa_default_policy(); |
73c27992 EB |
433 | pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES); |
434 | kthreadd_task = find_task_by_pid(pid); | |
1da177e4 | 435 | unlock_kernel(); |
f340c0d1 IM |
436 | |
437 | /* | |
438 | * The boot idle thread must execute schedule() | |
439 | * at least one to get things moving: | |
440 | */ | |
5bfb5d69 | 441 | preempt_enable_no_resched(); |
f340c0d1 | 442 | schedule(); |
5bfb5d69 | 443 | preempt_disable(); |
f340c0d1 | 444 | |
5bfb5d69 | 445 | /* Call into cpu_idle with preempt disabled */ |
1da177e4 LT |
446 | cpu_idle(); |
447 | } | |
448 | ||
449 | /* Check for early params. */ | |
450 | static int __init do_early_param(char *param, char *val) | |
451 | { | |
452 | struct obs_kernel_param *p; | |
453 | ||
454 | for (p = __setup_start; p < __setup_end; p++) { | |
455 | if (p->early && strcmp(param, p->str) == 0) { | |
456 | if (p->setup_func(val) != 0) | |
457 | printk(KERN_WARNING | |
458 | "Malformed early option '%s'\n", param); | |
459 | } | |
460 | } | |
461 | /* We accept everything at this stage. */ | |
462 | return 0; | |
463 | } | |
464 | ||
465 | /* Arch code calls this early on, or if not, just before other parsing. */ | |
466 | void __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); |
1da177e4 LT |
476 | parse_args("early options", tmp_cmdline, NULL, 0, do_early_param); |
477 | done = 1; | |
478 | } | |
479 | ||
480 | /* | |
481 | * Activate the first processor. | |
482 | */ | |
483 | ||
44fd2299 SS |
484 | static 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 */ | |
488 | cpu_set(cpu, cpu_online_map); | |
489 | cpu_set(cpu, cpu_present_map); | |
490 | cpu_set(cpu, cpu_possible_map); | |
491 | } | |
492 | ||
033ab7f8 AM |
493 | void __init __attribute__((weak)) smp_setup_processor_id(void) |
494 | { | |
495 | } | |
496 | ||
1da177e4 LT |
497 | asmlinkage void __init start_kernel(void) |
498 | { | |
499 | char * command_line; | |
500 | extern struct kernel_param __start___param[], __stop___param[]; | |
033ab7f8 AM |
501 | |
502 | smp_setup_processor_id(); | |
503 | ||
fbb9ce95 IM |
504 | /* |
505 | * Need to run as early as possible, to initialize the | |
506 | * lockdep hash: | |
507 | */ | |
c9538ed4 | 508 | unwind_init(); |
fbb9ce95 IM |
509 | lockdep_init(); |
510 | ||
511 | local_irq_disable(); | |
512 | early_boot_irqs_off(); | |
243c7621 | 513 | early_init_irq_lock_class(); |
fbb9ce95 | 514 | |
1da177e4 LT |
515 | /* |
516 | * Interrupts are still disabled. Do necessary setups, then | |
517 | * enable them | |
518 | */ | |
519 | lock_kernel(); | |
906568c9 | 520 | tick_init(); |
44fd2299 | 521 | boot_cpu_init(); |
1da177e4 LT |
522 | page_address_init(); |
523 | printk(KERN_NOTICE); | |
8993780a | 524 | printk(linux_banner); |
1da177e4 | 525 | setup_arch(&command_line); |
30d7e0d4 | 526 | setup_command_line(command_line); |
690a973f | 527 | unwind_setup(); |
1da177e4 | 528 | setup_per_cpu_areas(); |
44fd2299 | 529 | smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */ |
1da177e4 LT |
530 | |
531 | /* | |
532 | * Set up the scheduler prior starting any interrupts (such as the | |
533 | * timer interrupt). Full topology setup happens at smp_init() | |
534 | * time - but meanwhile we still have a functioning scheduler. | |
535 | */ | |
536 | sched_init(); | |
537 | /* | |
538 | * Disable preemption - early bootup scheduling is extremely | |
539 | * fragile until we cpu_idle() for the first time. | |
540 | */ | |
541 | preempt_disable(); | |
542 | build_all_zonelists(); | |
543 | page_alloc_init(); | |
30d7e0d4 | 544 | printk(KERN_NOTICE "Kernel command line: %s\n", boot_command_line); |
1da177e4 | 545 | parse_early_param(); |
30d7e0d4 | 546 | parse_args("Booting kernel", static_command_line, __start___param, |
1da177e4 LT |
547 | __stop___param - __start___param, |
548 | &unknown_bootoption); | |
c4a68306 AB |
549 | if (!irqs_disabled()) { |
550 | printk(KERN_WARNING "start_kernel(): bug: interrupts were " | |
551 | "enabled *very* early, fixing it\n"); | |
552 | local_irq_disable(); | |
553 | } | |
1da177e4 LT |
554 | sort_main_extable(); |
555 | trap_init(); | |
556 | rcu_init(); | |
557 | init_IRQ(); | |
558 | pidhash_init(); | |
559 | init_timers(); | |
c0a31329 | 560 | hrtimers_init(); |
1da177e4 | 561 | softirq_init(); |
ad596171 | 562 | timekeeping_init(); |
88fecaa2 | 563 | time_init(); |
93e02814 HC |
564 | profile_init(); |
565 | if (!irqs_disabled()) | |
566 | printk("start_kernel(): bug: interrupts were enabled early\n"); | |
fbb9ce95 | 567 | early_boot_irqs_on(); |
93e02814 | 568 | local_irq_enable(); |
1da177e4 LT |
569 | |
570 | /* | |
571 | * HACK ALERT! This is early. We're enabling the console before | |
572 | * we've done PCI setups etc, and console_init() must be aware of | |
573 | * this. But we do want output early, in case something goes wrong. | |
574 | */ | |
575 | console_init(); | |
576 | if (panic_later) | |
577 | panic(panic_later, panic_param); | |
fbb9ce95 IM |
578 | |
579 | lockdep_info(); | |
580 | ||
9a11b49a IM |
581 | /* |
582 | * Need to run this when irqs are enabled, because it wants | |
583 | * to self-test [hard/soft]-irqs on/off lock inversion bugs | |
584 | * too: | |
585 | */ | |
586 | locking_selftest(); | |
587 | ||
1da177e4 LT |
588 | #ifdef CONFIG_BLK_DEV_INITRD |
589 | if (initrd_start && !initrd_below_start_ok && | |
590 | initrd_start < min_low_pfn << PAGE_SHIFT) { | |
591 | printk(KERN_CRIT "initrd overwritten (0x%08lx < 0x%08lx) - " | |
592 | "disabling it.\n",initrd_start,min_low_pfn << PAGE_SHIFT); | |
593 | initrd_start = 0; | |
594 | } | |
595 | #endif | |
596 | vfs_caches_init_early(); | |
c417f024 | 597 | cpuset_init_early(); |
1da177e4 LT |
598 | mem_init(); |
599 | kmem_cache_init(); | |
e7c8d5c9 | 600 | setup_per_cpu_pageset(); |
1da177e4 LT |
601 | numa_policy_init(); |
602 | if (late_time_init) | |
603 | late_time_init(); | |
604 | calibrate_delay(); | |
605 | pidmap_init(); | |
606 | pgtable_cache_init(); | |
607 | prio_tree_init(); | |
608 | anon_vma_init(); | |
609 | #ifdef CONFIG_X86 | |
610 | if (efi_enabled) | |
611 | efi_enter_virtual_mode(); | |
612 | #endif | |
613 | fork_init(num_physpages); | |
614 | proc_caches_init(); | |
615 | buffer_init(); | |
616 | unnamed_dev_init(); | |
617 | key_init(); | |
618 | security_init(); | |
619 | vfs_caches_init(num_physpages); | |
620 | radix_tree_init(); | |
621 | signals_init(); | |
622 | /* rootfs populating might need page-writeback */ | |
623 | page_writeback_init(); | |
624 | #ifdef CONFIG_PROC_FS | |
625 | proc_root_init(); | |
626 | #endif | |
627 | cpuset_init(); | |
c757249a | 628 | taskstats_init_early(); |
ca74e92b | 629 | delayacct_init(); |
1da177e4 LT |
630 | |
631 | check_bugs(); | |
632 | ||
633 | acpi_early_init(); /* before LAPIC and SMP init */ | |
634 | ||
635 | /* Do the rest non-__init'ed, we're now alive */ | |
636 | rest_init(); | |
637 | } | |
638 | ||
639 | static int __initdata initcall_debug; | |
640 | ||
641 | static int __init initcall_debug_setup(char *str) | |
642 | { | |
643 | initcall_debug = 1; | |
644 | return 1; | |
645 | } | |
646 | __setup("initcall_debug", initcall_debug_setup); | |
647 | ||
1da177e4 LT |
648 | extern initcall_t __initcall_start[], __initcall_end[]; |
649 | ||
650 | static void __init do_initcalls(void) | |
651 | { | |
652 | initcall_t *call; | |
653 | int count = preempt_count(); | |
654 | ||
655 | for (call = __initcall_start; call < __initcall_end; call++) { | |
8f0c45cd | 656 | ktime_t t0, t1, delta; |
c1cda48a AM |
657 | char *msg = NULL; |
658 | char msgbuf[40]; | |
659 | int result; | |
1da177e4 LT |
660 | |
661 | if (initcall_debug) { | |
c1cda48a AM |
662 | printk("Calling initcall 0x%p", *call); |
663 | print_fn_descriptor_symbol(": %s()", | |
664 | (unsigned long) *call); | |
1da177e4 | 665 | printk("\n"); |
8f0c45cd | 666 | t0 = ktime_get(); |
1da177e4 LT |
667 | } |
668 | ||
c1cda48a | 669 | result = (*call)(); |
1da177e4 | 670 | |
8f0c45cd IM |
671 | if (initcall_debug) { |
672 | t1 = ktime_get(); | |
673 | delta = ktime_sub(t1, t0); | |
674 | ||
675 | printk("initcall 0x%p", *call); | |
676 | print_fn_descriptor_symbol(": %s()", | |
677 | (unsigned long) *call); | |
678 | printk(" returned %d.\n", result); | |
679 | ||
680 | printk("initcall 0x%p ran for %Ld msecs: ", | |
681 | *call, (unsigned long long)delta.tv64 >> 20); | |
682 | print_fn_descriptor_symbol("%s()\n", | |
683 | (unsigned long) *call); | |
684 | } | |
685 | ||
f3a19cb4 | 686 | if (result && result != -ENODEV && initcall_debug) { |
c1cda48a AM |
687 | sprintf(msgbuf, "error code %d", result); |
688 | msg = msgbuf; | |
689 | } | |
1da177e4 LT |
690 | if (preempt_count() != count) { |
691 | msg = "preemption imbalance"; | |
692 | preempt_count() = count; | |
693 | } | |
694 | if (irqs_disabled()) { | |
695 | msg = "disabled interrupts"; | |
696 | local_irq_enable(); | |
697 | } | |
698 | if (msg) { | |
c1cda48a AM |
699 | printk(KERN_WARNING "initcall at 0x%p", *call); |
700 | print_fn_descriptor_symbol(": %s()", | |
701 | (unsigned long) *call); | |
702 | printk(": returned with %s\n", msg); | |
1da177e4 LT |
703 | } |
704 | } | |
705 | ||
706 | /* Make sure there is no pending stuff from the initcall sequence */ | |
707 | flush_scheduled_work(); | |
708 | } | |
709 | ||
710 | /* | |
711 | * Ok, the machine is now initialized. None of the devices | |
712 | * have been touched yet, but the CPU subsystem is up and | |
713 | * running, and memory and process management works. | |
714 | * | |
715 | * Now we can finally start doing some real work.. | |
716 | */ | |
717 | static void __init do_basic_setup(void) | |
718 | { | |
719 | /* drivers will send hotplug events */ | |
720 | init_workqueues(); | |
721 | usermodehelper_init(); | |
722 | driver_init(); | |
b04c3afb | 723 | init_irq_proc(); |
1da177e4 LT |
724 | do_initcalls(); |
725 | } | |
726 | ||
88d20328 | 727 | static void __init do_pre_smp_initcalls(void) |
1da177e4 LT |
728 | { |
729 | extern int spawn_ksoftirqd(void); | |
730 | #ifdef CONFIG_SMP | |
731 | extern int migration_init(void); | |
732 | ||
733 | migration_init(); | |
734 | #endif | |
735 | spawn_ksoftirqd(); | |
8446f1d3 | 736 | spawn_softlockup_task(); |
1da177e4 LT |
737 | } |
738 | ||
739 | static void run_init_process(char *init_filename) | |
740 | { | |
741 | argv_init[0] = init_filename; | |
67608567 | 742 | kernel_execve(init_filename, argv_init, envp_init); |
1da177e4 LT |
743 | } |
744 | ||
ee5bfa64 VG |
745 | /* This is a non __init function. Force it to be noinline otherwise gcc |
746 | * makes it inline to init() and it becomes part of init.text section | |
747 | */ | |
748 | static int noinline init_post(void) | |
749 | { | |
750 | free_initmem(); | |
751 | unlock_kernel(); | |
752 | mark_rodata_ro(); | |
753 | system_state = SYSTEM_RUNNING; | |
754 | numa_default_policy(); | |
755 | ||
756 | if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0) | |
757 | printk(KERN_WARNING "Warning: unable to open an initial console.\n"); | |
758 | ||
759 | (void) sys_dup(0); | |
760 | (void) sys_dup(0); | |
761 | ||
762 | if (ramdisk_execute_command) { | |
763 | run_init_process(ramdisk_execute_command); | |
764 | printk(KERN_WARNING "Failed to execute %s\n", | |
765 | ramdisk_execute_command); | |
766 | } | |
767 | ||
768 | /* | |
769 | * We try each of these until one succeeds. | |
770 | * | |
771 | * The Bourne shell can be used instead of init if we are | |
772 | * trying to recover a really broken machine. | |
773 | */ | |
774 | if (execute_command) { | |
775 | run_init_process(execute_command); | |
776 | printk(KERN_WARNING "Failed to execute %s. Attempting " | |
777 | "defaults...\n", execute_command); | |
778 | } | |
779 | run_init_process("/sbin/init"); | |
780 | run_init_process("/etc/init"); | |
781 | run_init_process("/bin/init"); | |
782 | run_init_process("/bin/sh"); | |
783 | ||
784 | panic("No init found. Try passing init= option to kernel."); | |
785 | } | |
786 | ||
aae5f662 | 787 | static int __init kernel_init(void * unused) |
1da177e4 LT |
788 | { |
789 | lock_kernel(); | |
790 | /* | |
791 | * init can run on any cpu. | |
792 | */ | |
793 | set_cpus_allowed(current, CPU_MASK_ALL); | |
794 | /* | |
795 | * Tell the world that we're going to be the grim | |
796 | * reaper of innocent orphaned children. | |
797 | * | |
798 | * We don't want people to have to make incorrect | |
799 | * assumptions about where in the task array this | |
800 | * can be found. | |
801 | */ | |
84d73786 | 802 | init_pid_ns.child_reaper = current; |
1da177e4 | 803 | |
9ec52099 CLG |
804 | cad_pid = task_pid(current); |
805 | ||
1da177e4 LT |
806 | smp_prepare_cpus(max_cpus); |
807 | ||
808 | do_pre_smp_initcalls(); | |
809 | ||
1da177e4 LT |
810 | smp_init(); |
811 | sched_init_smp(); | |
812 | ||
813 | cpuset_init_smp(); | |
814 | ||
1da177e4 LT |
815 | do_basic_setup(); |
816 | ||
817 | /* | |
818 | * check if there is an early userspace init. If yes, let it do all | |
819 | * the work | |
820 | */ | |
ffdfc409 OJ |
821 | |
822 | if (!ramdisk_execute_command) | |
823 | ramdisk_execute_command = "/init"; | |
824 | ||
825 | if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) { | |
826 | ramdisk_execute_command = NULL; | |
1da177e4 | 827 | prepare_namespace(); |
ffdfc409 | 828 | } |
1da177e4 LT |
829 | |
830 | /* | |
831 | * Ok, we have completed the initial bootup, and | |
832 | * we're essentially up and running. Get rid of the | |
833 | * initmem segments and start the user-mode stuff.. | |
834 | */ | |
ee5bfa64 VG |
835 | init_post(); |
836 | return 0; | |
1da177e4 | 837 | } |