]>
Commit | Line | Data |
---|---|---|
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 | ||
12 | #include <linux/types.h> | |
13 | #include <linux/module.h> | |
14 | #include <linux/proc_fs.h> | |
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> | |
32 | #include <linux/start_kernel.h> | |
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/cgroup.h> | |
43 | #include <linux/efi.h> | |
44 | #include <linux/tick.h> | |
45 | #include <linux/interrupt.h> | |
46 | #include <linux/taskstats_kern.h> | |
47 | #include <linux/delayacct.h> | |
48 | #include <linux/unistd.h> | |
49 | #include <linux/rmap.h> | |
50 | #include <linux/mempolicy.h> | |
51 | #include <linux/key.h> | |
52 | #include <linux/unwind.h> | |
53 | #include <linux/buffer_head.h> | |
54 | #include <linux/debug_locks.h> | |
55 | #include <linux/debugobjects.h> | |
56 | #include <linux/lockdep.h> | |
57 | #include <linux/pid_namespace.h> | |
58 | #include <linux/device.h> | |
59 | #include <linux/kthread.h> | |
60 | #include <linux/sched.h> | |
61 | #include <linux/signal.h> | |
62 | #include <linux/idr.h> | |
63 | ||
64 | #include <asm/io.h> | |
65 | #include <asm/bugs.h> | |
66 | #include <asm/setup.h> | |
67 | #include <asm/sections.h> | |
68 | #include <asm/cacheflush.h> | |
69 | ||
70 | #ifdef CONFIG_X86_LOCAL_APIC | |
71 | #include <asm/smp.h> | |
72 | #endif | |
73 | ||
74 | /* | |
75 | * This is one of the first .c files built. Error out early if we have compiler | |
76 | * trouble. | |
77 | */ | |
78 | ||
79 | #if __GNUC__ == 4 && __GNUC_MINOR__ == 1 && __GNUC_PATCHLEVEL__ == 0 | |
80 | #warning gcc-4.1.0 is known to miscompile the kernel. A different compiler version is recommended. | |
81 | #endif | |
82 | ||
83 | static int kernel_init(void *); | |
84 | ||
85 | extern void init_IRQ(void); | |
86 | extern void fork_init(unsigned long); | |
87 | extern void mca_init(void); | |
88 | extern void sbus_init(void); | |
89 | extern void pidhash_init(void); | |
90 | extern void pidmap_init(void); | |
91 | extern void prio_tree_init(void); | |
92 | extern void radix_tree_init(void); | |
93 | extern void free_initmem(void); | |
94 | #ifdef CONFIG_ACPI | |
95 | extern void acpi_early_init(void); | |
96 | #else | |
97 | static inline void acpi_early_init(void) { } | |
98 | #endif | |
99 | #ifndef CONFIG_DEBUG_RODATA | |
100 | static inline void mark_rodata_ro(void) { } | |
101 | #endif | |
102 | ||
103 | #ifdef CONFIG_TC | |
104 | extern void tc_init(void); | |
105 | #endif | |
106 | ||
107 | enum system_states system_state; | |
108 | EXPORT_SYMBOL(system_state); | |
109 | ||
110 | /* | |
111 | * Boot command-line arguments | |
112 | */ | |
113 | #define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT | |
114 | #define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT | |
115 | ||
116 | extern void time_init(void); | |
117 | /* Default late time init is NULL. archs can override this later. */ | |
118 | void (*late_time_init)(void); | |
119 | extern void softirq_init(void); | |
120 | ||
121 | /* Untouched command line saved by arch-specific code. */ | |
122 | char __initdata boot_command_line[COMMAND_LINE_SIZE]; | |
123 | /* Untouched saved command line (eg. for /proc) */ | |
124 | char *saved_command_line; | |
125 | /* Command line for parameter parsing */ | |
126 | static char *static_command_line; | |
127 | ||
128 | static char *execute_command; | |
129 | static char *ramdisk_execute_command; | |
130 | ||
131 | #ifdef CONFIG_SMP | |
132 | /* Setup configured maximum number of CPUs to activate */ | |
133 | unsigned int __initdata setup_max_cpus = NR_CPUS; | |
134 | ||
135 | /* | |
136 | * Setup routine for controlling SMP activation | |
137 | * | |
138 | * Command-line option of "nosmp" or "maxcpus=0" will disable SMP | |
139 | * activation entirely (the MPS table probe still happens, though). | |
140 | * | |
141 | * Command-line option of "maxcpus=<NUM>", where <NUM> is an integer | |
142 | * greater than 0, limits the maximum number of CPUs activated in | |
143 | * SMP mode to <NUM>. | |
144 | */ | |
145 | #ifndef CONFIG_X86_IO_APIC | |
146 | static inline void disable_ioapic_setup(void) {}; | |
147 | #endif | |
148 | ||
149 | static int __init nosmp(char *str) | |
150 | { | |
151 | setup_max_cpus = 0; | |
152 | disable_ioapic_setup(); | |
153 | return 0; | |
154 | } | |
155 | ||
156 | early_param("nosmp", nosmp); | |
157 | ||
158 | static int __init maxcpus(char *str) | |
159 | { | |
160 | get_option(&str, &setup_max_cpus); | |
161 | if (setup_max_cpus == 0) | |
162 | disable_ioapic_setup(); | |
163 | ||
164 | return 0; | |
165 | } | |
166 | ||
167 | early_param("maxcpus", maxcpus); | |
168 | #else | |
169 | #define setup_max_cpus NR_CPUS | |
170 | #endif | |
171 | ||
172 | /* | |
173 | * If set, this is an indication to the drivers that reset the underlying | |
174 | * device before going ahead with the initialization otherwise driver might | |
175 | * rely on the BIOS and skip the reset operation. | |
176 | * | |
177 | * This is useful if kernel is booting in an unreliable environment. | |
178 | * For ex. kdump situaiton where previous kernel has crashed, BIOS has been | |
179 | * skipped and devices will be in unknown state. | |
180 | */ | |
181 | unsigned int reset_devices; | |
182 | EXPORT_SYMBOL(reset_devices); | |
183 | ||
184 | static int __init set_reset_devices(char *str) | |
185 | { | |
186 | reset_devices = 1; | |
187 | return 1; | |
188 | } | |
189 | ||
190 | __setup("reset_devices", set_reset_devices); | |
191 | ||
192 | static char * argv_init[MAX_INIT_ARGS+2] = { "init", NULL, }; | |
193 | char * envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, }; | |
194 | static const char *panic_later, *panic_param; | |
195 | ||
196 | extern struct obs_kernel_param __setup_start[], __setup_end[]; | |
197 | ||
198 | static int __init obsolete_checksetup(char *line) | |
199 | { | |
200 | struct obs_kernel_param *p; | |
201 | int had_early_param = 0; | |
202 | ||
203 | p = __setup_start; | |
204 | do { | |
205 | int n = strlen(p->str); | |
206 | if (!strncmp(line, p->str, n)) { | |
207 | if (p->early) { | |
208 | /* Already done in parse_early_param? | |
209 | * (Needs exact match on param part). | |
210 | * Keep iterating, as we can have early | |
211 | * params and __setups of same names 8( */ | |
212 | if (line[n] == '\0' || line[n] == '=') | |
213 | had_early_param = 1; | |
214 | } else if (!p->setup_func) { | |
215 | printk(KERN_WARNING "Parameter %s is obsolete," | |
216 | " ignored\n", p->str); | |
217 | return 1; | |
218 | } else if (p->setup_func(line + n)) | |
219 | return 1; | |
220 | } | |
221 | p++; | |
222 | } while (p < __setup_end); | |
223 | ||
224 | return had_early_param; | |
225 | } | |
226 | ||
227 | /* | |
228 | * This should be approx 2 Bo*oMips to start (note initial shift), and will | |
229 | * still work even if initially too large, it will just take slightly longer | |
230 | */ | |
231 | unsigned long loops_per_jiffy = (1<<12); | |
232 | ||
233 | EXPORT_SYMBOL(loops_per_jiffy); | |
234 | ||
235 | static int __init debug_kernel(char *str) | |
236 | { | |
237 | console_loglevel = 10; | |
238 | return 0; | |
239 | } | |
240 | ||
241 | static int __init quiet_kernel(char *str) | |
242 | { | |
243 | console_loglevel = 4; | |
244 | return 0; | |
245 | } | |
246 | ||
247 | early_param("debug", debug_kernel); | |
248 | early_param("quiet", quiet_kernel); | |
249 | ||
250 | static int __init loglevel(char *str) | |
251 | { | |
252 | get_option(&str, &console_loglevel); | |
253 | return 0; | |
254 | } | |
255 | ||
256 | early_param("loglevel", loglevel); | |
257 | ||
258 | /* | |
259 | * Unknown boot options get handed to init, unless they look like | |
260 | * failed parameters | |
261 | */ | |
262 | static int __init unknown_bootoption(char *param, char *val) | |
263 | { | |
264 | /* Change NUL term back to "=", to make "param" the whole string. */ | |
265 | if (val) { | |
266 | /* param=val or param="val"? */ | |
267 | if (val == param+strlen(param)+1) | |
268 | val[-1] = '='; | |
269 | else if (val == param+strlen(param)+2) { | |
270 | val[-2] = '='; | |
271 | memmove(val-1, val, strlen(val)+1); | |
272 | val--; | |
273 | } else | |
274 | BUG(); | |
275 | } | |
276 | ||
277 | /* Handle obsolete-style parameters */ | |
278 | if (obsolete_checksetup(param)) | |
279 | return 0; | |
280 | ||
281 | /* | |
282 | * Preemptive maintenance for "why didn't my misspelled command | |
283 | * line work?" | |
284 | */ | |
285 | if (strchr(param, '.') && (!val || strchr(param, '.') < val)) { | |
286 | printk(KERN_ERR "Unknown boot option `%s': ignoring\n", param); | |
287 | return 0; | |
288 | } | |
289 | ||
290 | if (panic_later) | |
291 | return 0; | |
292 | ||
293 | if (val) { | |
294 | /* Environment option */ | |
295 | unsigned int i; | |
296 | for (i = 0; envp_init[i]; i++) { | |
297 | if (i == MAX_INIT_ENVS) { | |
298 | panic_later = "Too many boot env vars at `%s'"; | |
299 | panic_param = param; | |
300 | } | |
301 | if (!strncmp(param, envp_init[i], val - param)) | |
302 | break; | |
303 | } | |
304 | envp_init[i] = param; | |
305 | } else { | |
306 | /* Command line option */ | |
307 | unsigned int i; | |
308 | for (i = 0; argv_init[i]; i++) { | |
309 | if (i == MAX_INIT_ARGS) { | |
310 | panic_later = "Too many boot init vars at `%s'"; | |
311 | panic_param = param; | |
312 | } | |
313 | } | |
314 | argv_init[i] = param; | |
315 | } | |
316 | return 0; | |
317 | } | |
318 | ||
319 | #ifdef CONFIG_DEBUG_PAGEALLOC | |
320 | int __read_mostly debug_pagealloc_enabled = 0; | |
321 | #endif | |
322 | ||
323 | static int __init init_setup(char *str) | |
324 | { | |
325 | unsigned int i; | |
326 | ||
327 | execute_command = str; | |
328 | /* | |
329 | * In case LILO is going to boot us with default command line, | |
330 | * it prepends "auto" before the whole cmdline which makes | |
331 | * the shell think it should execute a script with such name. | |
332 | * So we ignore all arguments entered _before_ init=... [MJ] | |
333 | */ | |
334 | for (i = 1; i < MAX_INIT_ARGS; i++) | |
335 | argv_init[i] = NULL; | |
336 | return 1; | |
337 | } | |
338 | __setup("init=", init_setup); | |
339 | ||
340 | static int __init rdinit_setup(char *str) | |
341 | { | |
342 | unsigned int i; | |
343 | ||
344 | ramdisk_execute_command = str; | |
345 | /* See "auto" comment in init_setup */ | |
346 | for (i = 1; i < MAX_INIT_ARGS; i++) | |
347 | argv_init[i] = NULL; | |
348 | return 1; | |
349 | } | |
350 | __setup("rdinit=", rdinit_setup); | |
351 | ||
352 | #ifndef CONFIG_SMP | |
353 | ||
354 | #ifdef CONFIG_X86_LOCAL_APIC | |
355 | static void __init smp_init(void) | |
356 | { | |
357 | APIC_init_uniprocessor(); | |
358 | } | |
359 | #else | |
360 | #define smp_init() do { } while (0) | |
361 | #endif | |
362 | ||
363 | static inline void setup_per_cpu_areas(void) { } | |
364 | static inline void setup_nr_cpu_ids(void) { } | |
365 | static inline void smp_prepare_cpus(unsigned int maxcpus) { } | |
366 | ||
367 | #else | |
368 | ||
369 | #if NR_CPUS > BITS_PER_LONG | |
370 | cpumask_t cpu_mask_all __read_mostly = CPU_MASK_ALL; | |
371 | EXPORT_SYMBOL(cpu_mask_all); | |
372 | #endif | |
373 | ||
374 | /* Setup number of possible processor ids */ | |
375 | int nr_cpu_ids __read_mostly = NR_CPUS; | |
376 | EXPORT_SYMBOL(nr_cpu_ids); | |
377 | ||
378 | /* An arch may set nr_cpu_ids earlier if needed, so this would be redundant */ | |
379 | static void __init setup_nr_cpu_ids(void) | |
380 | { | |
381 | int cpu, highest_cpu = 0; | |
382 | ||
383 | for_each_possible_cpu(cpu) | |
384 | highest_cpu = cpu; | |
385 | ||
386 | nr_cpu_ids = highest_cpu + 1; | |
387 | } | |
388 | ||
389 | #ifndef CONFIG_HAVE_SETUP_PER_CPU_AREA | |
390 | unsigned long __per_cpu_offset[NR_CPUS] __read_mostly; | |
391 | ||
392 | EXPORT_SYMBOL(__per_cpu_offset); | |
393 | ||
394 | static void __init setup_per_cpu_areas(void) | |
395 | { | |
396 | unsigned long size, i; | |
397 | char *ptr; | |
398 | unsigned long nr_possible_cpus = num_possible_cpus(); | |
399 | ||
400 | /* Copy section for each CPU (we discard the original) */ | |
401 | size = ALIGN(PERCPU_ENOUGH_ROOM, PAGE_SIZE); | |
402 | ptr = alloc_bootmem_pages(size * nr_possible_cpus); | |
403 | ||
404 | for_each_possible_cpu(i) { | |
405 | __per_cpu_offset[i] = ptr - __per_cpu_start; | |
406 | memcpy(ptr, __per_cpu_start, __per_cpu_end - __per_cpu_start); | |
407 | ptr += size; | |
408 | } | |
409 | } | |
410 | #endif /* CONFIG_HAVE_SETUP_PER_CPU_AREA */ | |
411 | ||
412 | /* Called by boot processor to activate the rest. */ | |
413 | static void __init smp_init(void) | |
414 | { | |
415 | unsigned int cpu; | |
416 | ||
417 | /* FIXME: This should be done in userspace --RR */ | |
418 | for_each_present_cpu(cpu) { | |
419 | if (num_online_cpus() >= setup_max_cpus) | |
420 | break; | |
421 | if (!cpu_online(cpu)) | |
422 | cpu_up(cpu); | |
423 | } | |
424 | ||
425 | /* Any cleanup work */ | |
426 | printk(KERN_INFO "Brought up %ld CPUs\n", (long)num_online_cpus()); | |
427 | smp_cpus_done(setup_max_cpus); | |
428 | } | |
429 | ||
430 | #endif | |
431 | ||
432 | /* | |
433 | * We need to store the untouched command line for future reference. | |
434 | * We also need to store the touched command line since the parameter | |
435 | * parsing is performed in place, and we should allow a component to | |
436 | * store reference of name/value for future reference. | |
437 | */ | |
438 | static void __init setup_command_line(char *command_line) | |
439 | { | |
440 | saved_command_line = alloc_bootmem(strlen (boot_command_line)+1); | |
441 | static_command_line = alloc_bootmem(strlen (command_line)+1); | |
442 | strcpy (saved_command_line, boot_command_line); | |
443 | strcpy (static_command_line, command_line); | |
444 | } | |
445 | ||
446 | /* | |
447 | * We need to finalize in a non-__init function or else race conditions | |
448 | * between the root thread and the init thread may cause start_kernel to | |
449 | * be reaped by free_initmem before the root thread has proceeded to | |
450 | * cpu_idle. | |
451 | * | |
452 | * gcc-3.4 accidentally inlines this function, so use noinline. | |
453 | */ | |
454 | ||
455 | static void noinline __init_refok rest_init(void) | |
456 | __releases(kernel_lock) | |
457 | { | |
458 | int pid; | |
459 | ||
460 | kernel_thread(kernel_init, NULL, CLONE_FS | CLONE_SIGHAND); | |
461 | numa_default_policy(); | |
462 | pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES); | |
463 | kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns); | |
464 | unlock_kernel(); | |
465 | ||
466 | /* | |
467 | * The boot idle thread must execute schedule() | |
468 | * at least once to get things moving: | |
469 | */ | |
470 | init_idle_bootup_task(current); | |
471 | preempt_enable_no_resched(); | |
472 | schedule(); | |
473 | preempt_disable(); | |
474 | ||
475 | /* Call into cpu_idle with preempt disabled */ | |
476 | cpu_idle(); | |
477 | } | |
478 | ||
479 | /* Check for early params. */ | |
480 | static int __init do_early_param(char *param, char *val) | |
481 | { | |
482 | struct obs_kernel_param *p; | |
483 | ||
484 | for (p = __setup_start; p < __setup_end; p++) { | |
485 | if ((p->early && strcmp(param, p->str) == 0) || | |
486 | (strcmp(param, "console") == 0 && | |
487 | strcmp(p->str, "earlycon") == 0) | |
488 | ) { | |
489 | if (p->setup_func(val) != 0) | |
490 | printk(KERN_WARNING | |
491 | "Malformed early option '%s'\n", param); | |
492 | } | |
493 | } | |
494 | /* We accept everything at this stage. */ | |
495 | return 0; | |
496 | } | |
497 | ||
498 | /* Arch code calls this early on, or if not, just before other parsing. */ | |
499 | void __init parse_early_param(void) | |
500 | { | |
501 | static __initdata int done = 0; | |
502 | static __initdata char tmp_cmdline[COMMAND_LINE_SIZE]; | |
503 | ||
504 | if (done) | |
505 | return; | |
506 | ||
507 | /* All fall through to do_early_param. */ | |
508 | strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE); | |
509 | parse_args("early options", tmp_cmdline, NULL, 0, do_early_param); | |
510 | done = 1; | |
511 | } | |
512 | ||
513 | /* | |
514 | * Activate the first processor. | |
515 | */ | |
516 | ||
517 | static void __init boot_cpu_init(void) | |
518 | { | |
519 | int cpu = smp_processor_id(); | |
520 | /* Mark the boot cpu "present", "online" etc for SMP and UP case */ | |
521 | cpu_set(cpu, cpu_online_map); | |
522 | cpu_set(cpu, cpu_present_map); | |
523 | cpu_set(cpu, cpu_possible_map); | |
524 | } | |
525 | ||
526 | void __init __weak smp_setup_processor_id(void) | |
527 | { | |
528 | } | |
529 | ||
530 | void __init __weak thread_info_cache_init(void) | |
531 | { | |
532 | } | |
533 | ||
534 | asmlinkage void __init start_kernel(void) | |
535 | { | |
536 | char * command_line; | |
537 | extern struct kernel_param __start___param[], __stop___param[]; | |
538 | ||
539 | smp_setup_processor_id(); | |
540 | ||
541 | /* | |
542 | * Need to run as early as possible, to initialize the | |
543 | * lockdep hash: | |
544 | */ | |
545 | unwind_init(); | |
546 | lockdep_init(); | |
547 | debug_objects_early_init(); | |
548 | cgroup_init_early(); | |
549 | ||
550 | local_irq_disable(); | |
551 | early_boot_irqs_off(); | |
552 | early_init_irq_lock_class(); | |
553 | ||
554 | /* | |
555 | * Interrupts are still disabled. Do necessary setups, then | |
556 | * enable them | |
557 | */ | |
558 | lock_kernel(); | |
559 | tick_init(); | |
560 | boot_cpu_init(); | |
561 | page_address_init(); | |
562 | printk(KERN_NOTICE); | |
563 | printk(linux_banner); | |
564 | setup_arch(&command_line); | |
565 | mm_init_owner(&init_mm, &init_task); | |
566 | setup_command_line(command_line); | |
567 | unwind_setup(); | |
568 | setup_per_cpu_areas(); | |
569 | setup_nr_cpu_ids(); | |
570 | smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */ | |
571 | ||
572 | /* | |
573 | * Set up the scheduler prior starting any interrupts (such as the | |
574 | * timer interrupt). Full topology setup happens at smp_init() | |
575 | * time - but meanwhile we still have a functioning scheduler. | |
576 | */ | |
577 | sched_init(); | |
578 | /* | |
579 | * Disable preemption - early bootup scheduling is extremely | |
580 | * fragile until we cpu_idle() for the first time. | |
581 | */ | |
582 | preempt_disable(); | |
583 | build_all_zonelists(); | |
584 | page_alloc_init(); | |
585 | printk(KERN_NOTICE "Kernel command line: %s\n", boot_command_line); | |
586 | parse_early_param(); | |
587 | parse_args("Booting kernel", static_command_line, __start___param, | |
588 | __stop___param - __start___param, | |
589 | &unknown_bootoption); | |
590 | if (!irqs_disabled()) { | |
591 | printk(KERN_WARNING "start_kernel(): bug: interrupts were " | |
592 | "enabled *very* early, fixing it\n"); | |
593 | local_irq_disable(); | |
594 | } | |
595 | sort_main_extable(); | |
596 | trap_init(); | |
597 | rcu_init(); | |
598 | init_IRQ(); | |
599 | pidhash_init(); | |
600 | init_timers(); | |
601 | hrtimers_init(); | |
602 | softirq_init(); | |
603 | timekeeping_init(); | |
604 | time_init(); | |
605 | sched_clock_init(); | |
606 | profile_init(); | |
607 | if (!irqs_disabled()) | |
608 | printk("start_kernel(): bug: interrupts were enabled early\n"); | |
609 | early_boot_irqs_on(); | |
610 | local_irq_enable(); | |
611 | ||
612 | /* | |
613 | * HACK ALERT! This is early. We're enabling the console before | |
614 | * we've done PCI setups etc, and console_init() must be aware of | |
615 | * this. But we do want output early, in case something goes wrong. | |
616 | */ | |
617 | console_init(); | |
618 | if (panic_later) | |
619 | panic(panic_later, panic_param); | |
620 | ||
621 | lockdep_info(); | |
622 | ||
623 | /* | |
624 | * Need to run this when irqs are enabled, because it wants | |
625 | * to self-test [hard/soft]-irqs on/off lock inversion bugs | |
626 | * too: | |
627 | */ | |
628 | locking_selftest(); | |
629 | ||
630 | #ifdef CONFIG_BLK_DEV_INITRD | |
631 | if (initrd_start && !initrd_below_start_ok && | |
632 | initrd_start < min_low_pfn << PAGE_SHIFT) { | |
633 | printk(KERN_CRIT "initrd overwritten (0x%08lx < 0x%08lx) - " | |
634 | "disabling it.\n",initrd_start,min_low_pfn << PAGE_SHIFT); | |
635 | initrd_start = 0; | |
636 | } | |
637 | #endif | |
638 | vfs_caches_init_early(); | |
639 | cpuset_init_early(); | |
640 | mem_init(); | |
641 | enable_debug_pagealloc(); | |
642 | cpu_hotplug_init(); | |
643 | kmem_cache_init(); | |
644 | debug_objects_mem_init(); | |
645 | idr_init_cache(); | |
646 | setup_per_cpu_pageset(); | |
647 | numa_policy_init(); | |
648 | if (late_time_init) | |
649 | late_time_init(); | |
650 | calibrate_delay(); | |
651 | pidmap_init(); | |
652 | pgtable_cache_init(); | |
653 | prio_tree_init(); | |
654 | anon_vma_init(); | |
655 | #ifdef CONFIG_X86 | |
656 | if (efi_enabled) | |
657 | efi_enter_virtual_mode(); | |
658 | #endif | |
659 | thread_info_cache_init(); | |
660 | fork_init(num_physpages); | |
661 | proc_caches_init(); | |
662 | buffer_init(); | |
663 | unnamed_dev_init(); | |
664 | key_init(); | |
665 | security_init(); | |
666 | vfs_caches_init(num_physpages); | |
667 | radix_tree_init(); | |
668 | signals_init(); | |
669 | /* rootfs populating might need page-writeback */ | |
670 | page_writeback_init(); | |
671 | #ifdef CONFIG_PROC_FS | |
672 | proc_root_init(); | |
673 | #endif | |
674 | cgroup_init(); | |
675 | cpuset_init(); | |
676 | taskstats_init_early(); | |
677 | delayacct_init(); | |
678 | ||
679 | check_bugs(); | |
680 | ||
681 | acpi_early_init(); /* before LAPIC and SMP init */ | |
682 | ||
683 | /* Do the rest non-__init'ed, we're now alive */ | |
684 | rest_init(); | |
685 | } | |
686 | ||
687 | static int __initdata initcall_debug; | |
688 | ||
689 | static int __init initcall_debug_setup(char *str) | |
690 | { | |
691 | initcall_debug = 1; | |
692 | return 1; | |
693 | } | |
694 | __setup("initcall_debug", initcall_debug_setup); | |
695 | ||
696 | static void __init do_one_initcall(initcall_t fn) | |
697 | { | |
698 | int count = preempt_count(); | |
699 | ktime_t t0, t1, delta; | |
700 | char msgbuf[64]; | |
701 | int result; | |
702 | ||
703 | if (initcall_debug) { | |
704 | print_fn_descriptor_symbol("calling %s\n", fn); | |
705 | t0 = ktime_get(); | |
706 | } | |
707 | ||
708 | result = fn(); | |
709 | ||
710 | if (initcall_debug) { | |
711 | t1 = ktime_get(); | |
712 | delta = ktime_sub(t1, t0); | |
713 | ||
714 | print_fn_descriptor_symbol("initcall %s", fn); | |
715 | printk(" returned %d after %Ld msecs\n", result, | |
716 | (unsigned long long) delta.tv64 >> 20); | |
717 | } | |
718 | ||
719 | msgbuf[0] = 0; | |
720 | ||
721 | if (result && result != -ENODEV && initcall_debug) | |
722 | sprintf(msgbuf, "error code %d ", result); | |
723 | ||
724 | if (preempt_count() != count) { | |
725 | strlcat(msgbuf, "preemption imbalance ", sizeof(msgbuf)); | |
726 | preempt_count() = count; | |
727 | } | |
728 | if (irqs_disabled()) { | |
729 | strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf)); | |
730 | local_irq_enable(); | |
731 | } | |
732 | if (msgbuf[0]) { | |
733 | print_fn_descriptor_symbol(KERN_WARNING "initcall %s", fn); | |
734 | printk(" returned with %s\n", msgbuf); | |
735 | } | |
736 | } | |
737 | ||
738 | ||
739 | extern initcall_t __initcall_start[], __initcall_end[]; | |
740 | ||
741 | static void __init do_initcalls(void) | |
742 | { | |
743 | initcall_t *call; | |
744 | ||
745 | for (call = __initcall_start; call < __initcall_end; call++) | |
746 | do_one_initcall(*call); | |
747 | ||
748 | /* Make sure there is no pending stuff from the initcall sequence */ | |
749 | flush_scheduled_work(); | |
750 | } | |
751 | ||
752 | /* | |
753 | * Ok, the machine is now initialized. None of the devices | |
754 | * have been touched yet, but the CPU subsystem is up and | |
755 | * running, and memory and process management works. | |
756 | * | |
757 | * Now we can finally start doing some real work.. | |
758 | */ | |
759 | static void __init do_basic_setup(void) | |
760 | { | |
761 | /* drivers will send hotplug events */ | |
762 | init_workqueues(); | |
763 | usermodehelper_init(); | |
764 | driver_init(); | |
765 | init_irq_proc(); | |
766 | do_initcalls(); | |
767 | } | |
768 | ||
769 | static int __initdata nosoftlockup; | |
770 | ||
771 | static int __init nosoftlockup_setup(char *str) | |
772 | { | |
773 | nosoftlockup = 1; | |
774 | return 1; | |
775 | } | |
776 | __setup("nosoftlockup", nosoftlockup_setup); | |
777 | ||
778 | static void __init do_pre_smp_initcalls(void) | |
779 | { | |
780 | extern int spawn_ksoftirqd(void); | |
781 | ||
782 | migration_init(); | |
783 | spawn_ksoftirqd(); | |
784 | if (!nosoftlockup) | |
785 | spawn_softlockup_task(); | |
786 | } | |
787 | ||
788 | static void run_init_process(char *init_filename) | |
789 | { | |
790 | argv_init[0] = init_filename; | |
791 | kernel_execve(init_filename, argv_init, envp_init); | |
792 | } | |
793 | ||
794 | /* This is a non __init function. Force it to be noinline otherwise gcc | |
795 | * makes it inline to init() and it becomes part of init.text section | |
796 | */ | |
797 | static int noinline init_post(void) | |
798 | { | |
799 | free_initmem(); | |
800 | unlock_kernel(); | |
801 | mark_rodata_ro(); | |
802 | system_state = SYSTEM_RUNNING; | |
803 | numa_default_policy(); | |
804 | ||
805 | if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0) | |
806 | printk(KERN_WARNING "Warning: unable to open an initial console.\n"); | |
807 | ||
808 | (void) sys_dup(0); | |
809 | (void) sys_dup(0); | |
810 | ||
811 | current->signal->flags |= SIGNAL_UNKILLABLE; | |
812 | ||
813 | if (ramdisk_execute_command) { | |
814 | run_init_process(ramdisk_execute_command); | |
815 | printk(KERN_WARNING "Failed to execute %s\n", | |
816 | ramdisk_execute_command); | |
817 | } | |
818 | ||
819 | /* | |
820 | * We try each of these until one succeeds. | |
821 | * | |
822 | * The Bourne shell can be used instead of init if we are | |
823 | * trying to recover a really broken machine. | |
824 | */ | |
825 | if (execute_command) { | |
826 | run_init_process(execute_command); | |
827 | printk(KERN_WARNING "Failed to execute %s. Attempting " | |
828 | "defaults...\n", execute_command); | |
829 | } | |
830 | run_init_process("/sbin/init"); | |
831 | run_init_process("/etc/init"); | |
832 | run_init_process("/bin/init"); | |
833 | run_init_process("/bin/sh"); | |
834 | ||
835 | panic("No init found. Try passing init= option to kernel."); | |
836 | } | |
837 | ||
838 | static int __init kernel_init(void * unused) | |
839 | { | |
840 | lock_kernel(); | |
841 | /* | |
842 | * init can run on any cpu. | |
843 | */ | |
844 | set_cpus_allowed_ptr(current, CPU_MASK_ALL_PTR); | |
845 | /* | |
846 | * Tell the world that we're going to be the grim | |
847 | * reaper of innocent orphaned children. | |
848 | * | |
849 | * We don't want people to have to make incorrect | |
850 | * assumptions about where in the task array this | |
851 | * can be found. | |
852 | */ | |
853 | init_pid_ns.child_reaper = current; | |
854 | ||
855 | cad_pid = task_pid(current); | |
856 | ||
857 | smp_prepare_cpus(setup_max_cpus); | |
858 | ||
859 | do_pre_smp_initcalls(); | |
860 | ||
861 | smp_init(); | |
862 | sched_init_smp(); | |
863 | ||
864 | cpuset_init_smp(); | |
865 | ||
866 | do_basic_setup(); | |
867 | ||
868 | /* | |
869 | * check if there is an early userspace init. If yes, let it do all | |
870 | * the work | |
871 | */ | |
872 | ||
873 | if (!ramdisk_execute_command) | |
874 | ramdisk_execute_command = "/init"; | |
875 | ||
876 | if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) { | |
877 | ramdisk_execute_command = NULL; | |
878 | prepare_namespace(); | |
879 | } | |
880 | ||
881 | /* | |
882 | * Ok, we have completed the initial bootup, and | |
883 | * we're essentially up and running. Get rid of the | |
884 | * initmem segments and start the user-mode stuff.. | |
885 | */ | |
886 | init_post(); | |
887 | return 0; | |
888 | } |