]>
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> | |
ddbcc7e8 | 42 | #include <linux/cgroup.h> |
1da177e4 | 43 | #include <linux/efi.h> |
906568c9 | 44 | #include <linux/tick.h> |
6168a702 | 45 | #include <linux/interrupt.h> |
c757249a | 46 | #include <linux/taskstats_kern.h> |
ca74e92b | 47 | #include <linux/delayacct.h> |
1da177e4 LT |
48 | #include <linux/unistd.h> |
49 | #include <linux/rmap.h> | |
50 | #include <linux/mempolicy.h> | |
51 | #include <linux/key.h> | |
4552d5dc | 52 | #include <linux/unwind.h> |
b6cd0b77 | 53 | #include <linux/buffer_head.h> |
9a11b49a | 54 | #include <linux/debug_locks.h> |
fbb9ce95 | 55 | #include <linux/lockdep.h> |
84d73786 | 56 | #include <linux/pid_namespace.h> |
1f21782e | 57 | #include <linux/device.h> |
73c27992 | 58 | #include <linux/kthread.h> |
e6fe6649 | 59 | #include <linux/sched.h> |
a1c9eea9 | 60 | #include <linux/signal.h> |
199f0ca5 | 61 | #include <linux/idr.h> |
1da177e4 LT |
62 | |
63 | #include <asm/io.h> | |
64 | #include <asm/bugs.h> | |
65 | #include <asm/setup.h> | |
a940199f | 66 | #include <asm/sections.h> |
37b73c82 | 67 | #include <asm/cacheflush.h> |
1da177e4 | 68 | |
1da177e4 LT |
69 | #ifdef CONFIG_X86_LOCAL_APIC |
70 | #include <asm/smp.h> | |
71 | #endif | |
72 | ||
73 | /* | |
69c99ac1 CQH |
74 | * This is one of the first .c files built. Error out early if we have compiler |
75 | * trouble. | |
1da177e4 | 76 | */ |
1da177e4 | 77 | |
f1a60dbf AM |
78 | #if __GNUC__ == 4 && __GNUC_MINOR__ == 1 && __GNUC_PATCHLEVEL__ == 0 |
79 | #warning gcc-4.1.0 is known to miscompile the kernel. A different compiler version is recommended. | |
80 | #endif | |
81 | ||
aae5f662 | 82 | static int kernel_init(void *); |
1da177e4 LT |
83 | |
84 | extern void init_IRQ(void); | |
1da177e4 LT |
85 | extern void fork_init(unsigned long); |
86 | extern void mca_init(void); | |
87 | extern void sbus_init(void); | |
1da177e4 LT |
88 | extern void pidhash_init(void); |
89 | extern void pidmap_init(void); | |
90 | extern void prio_tree_init(void); | |
91 | extern void radix_tree_init(void); | |
92 | extern void free_initmem(void); | |
1da177e4 LT |
93 | #ifdef CONFIG_ACPI |
94 | extern void acpi_early_init(void); | |
95 | #else | |
96 | static inline void acpi_early_init(void) { } | |
97 | #endif | |
37b73c82 AV |
98 | #ifndef CONFIG_DEBUG_RODATA |
99 | static inline void mark_rodata_ro(void) { } | |
100 | #endif | |
1da177e4 LT |
101 | |
102 | #ifdef CONFIG_TC | |
103 | extern void tc_init(void); | |
104 | #endif | |
105 | ||
106 | enum system_states system_state; | |
107 | EXPORT_SYMBOL(system_state); | |
108 | ||
109 | /* | |
110 | * Boot command-line arguments | |
111 | */ | |
112 | #define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT | |
113 | #define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT | |
114 | ||
115 | extern void time_init(void); | |
116 | /* Default late time init is NULL. archs can override this later. */ | |
117 | void (*late_time_init)(void); | |
118 | extern void softirq_init(void); | |
119 | ||
30d7e0d4 ABL |
120 | /* Untouched command line saved by arch-specific code. */ |
121 | char __initdata boot_command_line[COMMAND_LINE_SIZE]; | |
122 | /* Untouched saved command line (eg. for /proc) */ | |
123 | char *saved_command_line; | |
124 | /* Command line for parameter parsing */ | |
125 | static char *static_command_line; | |
1da177e4 LT |
126 | |
127 | static char *execute_command; | |
ffdfc409 | 128 | static char *ramdisk_execute_command; |
1da177e4 | 129 | |
8b3b2955 | 130 | #ifdef CONFIG_SMP |
1da177e4 | 131 | /* Setup configured maximum number of CPUs to activate */ |
ca74a6f8 | 132 | unsigned int __initdata setup_max_cpus = NR_CPUS; |
7e96287d | 133 | |
1da177e4 LT |
134 | /* |
135 | * Setup routine for controlling SMP activation | |
136 | * | |
137 | * Command-line option of "nosmp" or "maxcpus=0" will disable SMP | |
138 | * activation entirely (the MPS table probe still happens, though). | |
139 | * | |
140 | * Command-line option of "maxcpus=<NUM>", where <NUM> is an integer | |
141 | * greater than 0, limits the maximum number of CPUs activated in | |
142 | * SMP mode to <NUM>. | |
143 | */ | |
61ec7567 LB |
144 | #ifndef CONFIG_X86_IO_APIC |
145 | static inline void disable_ioapic_setup(void) {}; | |
146 | #endif | |
147 | ||
1da177e4 LT |
148 | static int __init nosmp(char *str) |
149 | { | |
ca74a6f8 | 150 | setup_max_cpus = 0; |
61ec7567 | 151 | disable_ioapic_setup(); |
8b3b2955 | 152 | return 0; |
1da177e4 LT |
153 | } |
154 | ||
8b3b2955 | 155 | early_param("nosmp", nosmp); |
1da177e4 LT |
156 | |
157 | static int __init maxcpus(char *str) | |
158 | { | |
ca74a6f8 AK |
159 | get_option(&str, &setup_max_cpus); |
160 | if (setup_max_cpus == 0) | |
61ec7567 LB |
161 | disable_ioapic_setup(); |
162 | ||
163 | return 0; | |
1da177e4 LT |
164 | } |
165 | ||
81340977 | 166 | early_param("maxcpus", maxcpus); |
8b3b2955 | 167 | #else |
ca74a6f8 | 168 | #define setup_max_cpus NR_CPUS |
8b3b2955 JB |
169 | #endif |
170 | ||
171 | /* | |
172 | * If set, this is an indication to the drivers that reset the underlying | |
173 | * device before going ahead with the initialization otherwise driver might | |
174 | * rely on the BIOS and skip the reset operation. | |
175 | * | |
176 | * This is useful if kernel is booting in an unreliable environment. | |
177 | * For ex. kdump situaiton where previous kernel has crashed, BIOS has been | |
178 | * skipped and devices will be in unknown state. | |
179 | */ | |
180 | unsigned int reset_devices; | |
181 | EXPORT_SYMBOL(reset_devices); | |
1da177e4 | 182 | |
7e96287d VG |
183 | static int __init set_reset_devices(char *str) |
184 | { | |
185 | reset_devices = 1; | |
186 | return 1; | |
187 | } | |
188 | ||
189 | __setup("reset_devices", set_reset_devices); | |
190 | ||
1da177e4 LT |
191 | static char * argv_init[MAX_INIT_ARGS+2] = { "init", NULL, }; |
192 | char * envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, }; | |
193 | static const char *panic_later, *panic_param; | |
194 | ||
195 | extern struct obs_kernel_param __setup_start[], __setup_end[]; | |
196 | ||
197 | static int __init obsolete_checksetup(char *line) | |
198 | { | |
199 | struct obs_kernel_param *p; | |
33df0d19 | 200 | int had_early_param = 0; |
1da177e4 LT |
201 | |
202 | p = __setup_start; | |
203 | do { | |
204 | int n = strlen(p->str); | |
205 | if (!strncmp(line, p->str, n)) { | |
206 | if (p->early) { | |
33df0d19 RR |
207 | /* Already done in parse_early_param? |
208 | * (Needs exact match on param part). | |
209 | * Keep iterating, as we can have early | |
210 | * params and __setups of same names 8( */ | |
1da177e4 | 211 | if (line[n] == '\0' || line[n] == '=') |
33df0d19 | 212 | had_early_param = 1; |
1da177e4 LT |
213 | } else if (!p->setup_func) { |
214 | printk(KERN_WARNING "Parameter %s is obsolete," | |
215 | " ignored\n", p->str); | |
216 | return 1; | |
217 | } else if (p->setup_func(line + n)) | |
218 | return 1; | |
219 | } | |
220 | p++; | |
221 | } while (p < __setup_end); | |
33df0d19 RR |
222 | |
223 | return had_early_param; | |
1da177e4 LT |
224 | } |
225 | ||
226 | /* | |
227 | * This should be approx 2 Bo*oMips to start (note initial shift), and will | |
228 | * still work even if initially too large, it will just take slightly longer | |
229 | */ | |
230 | unsigned long loops_per_jiffy = (1<<12); | |
231 | ||
232 | EXPORT_SYMBOL(loops_per_jiffy); | |
233 | ||
234 | static int __init debug_kernel(char *str) | |
235 | { | |
1da177e4 | 236 | console_loglevel = 10; |
f6f21c81 | 237 | return 0; |
1da177e4 LT |
238 | } |
239 | ||
240 | static int __init quiet_kernel(char *str) | |
241 | { | |
1da177e4 | 242 | console_loglevel = 4; |
f6f21c81 | 243 | return 0; |
1da177e4 LT |
244 | } |
245 | ||
f6f21c81 YL |
246 | early_param("debug", debug_kernel); |
247 | early_param("quiet", quiet_kernel); | |
1da177e4 LT |
248 | |
249 | static int __init loglevel(char *str) | |
250 | { | |
251 | get_option(&str, &console_loglevel); | |
d9d4fcfe | 252 | return 0; |
1da177e4 LT |
253 | } |
254 | ||
f6f21c81 | 255 | early_param("loglevel", loglevel); |
1da177e4 LT |
256 | |
257 | /* | |
258 | * Unknown boot options get handed to init, unless they look like | |
259 | * failed parameters | |
260 | */ | |
261 | static int __init unknown_bootoption(char *param, char *val) | |
262 | { | |
263 | /* Change NUL term back to "=", to make "param" the whole string. */ | |
264 | if (val) { | |
265 | /* param=val or param="val"? */ | |
266 | if (val == param+strlen(param)+1) | |
267 | val[-1] = '='; | |
268 | else if (val == param+strlen(param)+2) { | |
269 | val[-2] = '='; | |
270 | memmove(val-1, val, strlen(val)+1); | |
271 | val--; | |
272 | } else | |
273 | BUG(); | |
274 | } | |
275 | ||
276 | /* Handle obsolete-style parameters */ | |
277 | if (obsolete_checksetup(param)) | |
278 | return 0; | |
279 | ||
280 | /* | |
211fee8a | 281 | * Preemptive maintenance for "why didn't my misspelled command |
1da177e4 LT |
282 | * line work?" |
283 | */ | |
284 | if (strchr(param, '.') && (!val || strchr(param, '.') < val)) { | |
285 | printk(KERN_ERR "Unknown boot option `%s': ignoring\n", param); | |
286 | return 0; | |
287 | } | |
288 | ||
289 | if (panic_later) | |
290 | return 0; | |
291 | ||
292 | if (val) { | |
293 | /* Environment option */ | |
294 | unsigned int i; | |
295 | for (i = 0; envp_init[i]; i++) { | |
296 | if (i == MAX_INIT_ENVS) { | |
297 | panic_later = "Too many boot env vars at `%s'"; | |
298 | panic_param = param; | |
299 | } | |
300 | if (!strncmp(param, envp_init[i], val - param)) | |
301 | break; | |
302 | } | |
303 | envp_init[i] = param; | |
304 | } else { | |
305 | /* Command line option */ | |
306 | unsigned int i; | |
307 | for (i = 0; argv_init[i]; i++) { | |
308 | if (i == MAX_INIT_ARGS) { | |
309 | panic_later = "Too many boot init vars at `%s'"; | |
310 | panic_param = param; | |
311 | } | |
312 | } | |
313 | argv_init[i] = param; | |
314 | } | |
315 | return 0; | |
316 | } | |
317 | ||
12d6f21e IM |
318 | #ifdef CONFIG_DEBUG_PAGEALLOC |
319 | int __read_mostly debug_pagealloc_enabled = 0; | |
320 | #endif | |
321 | ||
1da177e4 LT |
322 | static int __init init_setup(char *str) |
323 | { | |
324 | unsigned int i; | |
325 | ||
326 | execute_command = str; | |
327 | /* | |
328 | * In case LILO is going to boot us with default command line, | |
329 | * it prepends "auto" before the whole cmdline which makes | |
330 | * the shell think it should execute a script with such name. | |
331 | * So we ignore all arguments entered _before_ init=... [MJ] | |
332 | */ | |
333 | for (i = 1; i < MAX_INIT_ARGS; i++) | |
334 | argv_init[i] = NULL; | |
335 | return 1; | |
336 | } | |
337 | __setup("init=", init_setup); | |
338 | ||
ffdfc409 OJ |
339 | static int __init rdinit_setup(char *str) |
340 | { | |
341 | unsigned int i; | |
342 | ||
343 | ramdisk_execute_command = str; | |
344 | /* See "auto" comment in init_setup */ | |
345 | for (i = 1; i < MAX_INIT_ARGS; i++) | |
346 | argv_init[i] = NULL; | |
347 | return 1; | |
348 | } | |
349 | __setup("rdinit=", rdinit_setup); | |
350 | ||
1da177e4 LT |
351 | #ifndef CONFIG_SMP |
352 | ||
353 | #ifdef CONFIG_X86_LOCAL_APIC | |
354 | static void __init smp_init(void) | |
355 | { | |
356 | APIC_init_uniprocessor(); | |
357 | } | |
358 | #else | |
359 | #define smp_init() do { } while (0) | |
360 | #endif | |
361 | ||
362 | static inline void setup_per_cpu_areas(void) { } | |
e0982e90 | 363 | static inline void setup_nr_cpu_ids(void) { } |
1da177e4 LT |
364 | static inline void smp_prepare_cpus(unsigned int maxcpus) { } |
365 | ||
366 | #else | |
367 | ||
321a8e9d MT |
368 | #if NR_CPUS > BITS_PER_LONG |
369 | cpumask_t cpu_mask_all __read_mostly = CPU_MASK_ALL; | |
370 | EXPORT_SYMBOL(cpu_mask_all); | |
371 | #endif | |
372 | ||
e0982e90 MT |
373 | /* Setup number of possible processor ids */ |
374 | int nr_cpu_ids __read_mostly = NR_CPUS; | |
375 | EXPORT_SYMBOL(nr_cpu_ids); | |
376 | ||
377 | /* An arch may set nr_cpu_ids earlier if needed, so this would be redundant */ | |
378 | static void __init setup_nr_cpu_ids(void) | |
379 | { | |
380 | int cpu, highest_cpu = 0; | |
381 | ||
382 | for_each_possible_cpu(cpu) | |
383 | highest_cpu = cpu; | |
384 | ||
385 | nr_cpu_ids = highest_cpu + 1; | |
386 | } | |
387 | ||
dd5af90a | 388 | #ifndef CONFIG_HAVE_SETUP_PER_CPU_AREA |
b73b459f | 389 | unsigned long __per_cpu_offset[NR_CPUS] __read_mostly; |
1da177e4 LT |
390 | |
391 | EXPORT_SYMBOL(__per_cpu_offset); | |
392 | ||
393 | static void __init setup_per_cpu_areas(void) | |
394 | { | |
395 | unsigned long size, i; | |
396 | char *ptr; | |
63872f87 | 397 | unsigned long nr_possible_cpus = num_possible_cpus(); |
1da177e4 LT |
398 | |
399 | /* Copy section for each CPU (we discard the original) */ | |
b6e3590f JF |
400 | size = ALIGN(PERCPU_ENOUGH_ROOM, PAGE_SIZE); |
401 | ptr = alloc_bootmem_pages(size * nr_possible_cpus); | |
1da177e4 | 402 | |
0a945022 | 403 | for_each_possible_cpu(i) { |
1da177e4 LT |
404 | __per_cpu_offset[i] = ptr - __per_cpu_start; |
405 | memcpy(ptr, __per_cpu_start, __per_cpu_end - __per_cpu_start); | |
63872f87 | 406 | ptr += size; |
1da177e4 LT |
407 | } |
408 | } | |
dd5af90a | 409 | #endif /* CONFIG_HAVE_SETUP_PER_CPU_AREA */ |
1da177e4 LT |
410 | |
411 | /* Called by boot processor to activate the rest. */ | |
412 | static void __init smp_init(void) | |
413 | { | |
53b8a315 | 414 | unsigned int cpu; |
1da177e4 LT |
415 | |
416 | /* FIXME: This should be done in userspace --RR */ | |
53b8a315 | 417 | for_each_present_cpu(cpu) { |
ca74a6f8 | 418 | if (num_online_cpus() >= setup_max_cpus) |
1da177e4 | 419 | break; |
53b8a315 CL |
420 | if (!cpu_online(cpu)) |
421 | cpu_up(cpu); | |
1da177e4 LT |
422 | } |
423 | ||
424 | /* Any cleanup work */ | |
425 | printk(KERN_INFO "Brought up %ld CPUs\n", (long)num_online_cpus()); | |
ca74a6f8 | 426 | smp_cpus_done(setup_max_cpus); |
1da177e4 LT |
427 | } |
428 | ||
429 | #endif | |
430 | ||
30d7e0d4 ABL |
431 | /* |
432 | * We need to store the untouched command line for future reference. | |
433 | * We also need to store the touched command line since the parameter | |
434 | * parsing is performed in place, and we should allow a component to | |
435 | * store reference of name/value for future reference. | |
436 | */ | |
437 | static void __init setup_command_line(char *command_line) | |
438 | { | |
439 | saved_command_line = alloc_bootmem(strlen (boot_command_line)+1); | |
440 | static_command_line = alloc_bootmem(strlen (command_line)+1); | |
441 | strcpy (saved_command_line, boot_command_line); | |
442 | strcpy (static_command_line, command_line); | |
443 | } | |
444 | ||
1da177e4 LT |
445 | /* |
446 | * We need to finalize in a non-__init function or else race conditions | |
447 | * between the root thread and the init thread may cause start_kernel to | |
448 | * be reaped by free_initmem before the root thread has proceeded to | |
449 | * cpu_idle. | |
450 | * | |
451 | * gcc-3.4 accidentally inlines this function, so use noinline. | |
452 | */ | |
453 | ||
92080309 | 454 | static void noinline __init_refok rest_init(void) |
1da177e4 LT |
455 | __releases(kernel_lock) |
456 | { | |
73c27992 EB |
457 | int pid; |
458 | ||
aae5f662 | 459 | kernel_thread(kernel_init, NULL, CLONE_FS | CLONE_SIGHAND); |
1da177e4 | 460 | numa_default_policy(); |
73c27992 | 461 | pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES); |
5cd20455 | 462 | kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns); |
1da177e4 | 463 | unlock_kernel(); |
f340c0d1 IM |
464 | |
465 | /* | |
466 | * The boot idle thread must execute schedule() | |
1df21055 | 467 | * at least once to get things moving: |
f340c0d1 | 468 | */ |
1df21055 | 469 | init_idle_bootup_task(current); |
5bfb5d69 | 470 | preempt_enable_no_resched(); |
f340c0d1 | 471 | schedule(); |
5bfb5d69 | 472 | preempt_disable(); |
f340c0d1 | 473 | |
5bfb5d69 | 474 | /* Call into cpu_idle with preempt disabled */ |
1da177e4 | 475 | cpu_idle(); |
1df21055 | 476 | } |
1da177e4 LT |
477 | |
478 | /* Check for early params. */ | |
479 | static int __init do_early_param(char *param, char *val) | |
480 | { | |
481 | struct obs_kernel_param *p; | |
482 | ||
483 | for (p = __setup_start; p < __setup_end; p++) { | |
18a8bd94 YL |
484 | if ((p->early && strcmp(param, p->str) == 0) || |
485 | (strcmp(param, "console") == 0 && | |
486 | strcmp(p->str, "earlycon") == 0) | |
487 | ) { | |
1da177e4 LT |
488 | if (p->setup_func(val) != 0) |
489 | printk(KERN_WARNING | |
490 | "Malformed early option '%s'\n", param); | |
491 | } | |
492 | } | |
493 | /* We accept everything at this stage. */ | |
494 | return 0; | |
495 | } | |
496 | ||
497 | /* Arch code calls this early on, or if not, just before other parsing. */ | |
498 | void __init parse_early_param(void) | |
499 | { | |
500 | static __initdata int done = 0; | |
501 | static __initdata char tmp_cmdline[COMMAND_LINE_SIZE]; | |
502 | ||
503 | if (done) | |
504 | return; | |
505 | ||
506 | /* All fall through to do_early_param. */ | |
30d7e0d4 | 507 | strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE); |
1da177e4 LT |
508 | parse_args("early options", tmp_cmdline, NULL, 0, do_early_param); |
509 | done = 1; | |
510 | } | |
511 | ||
512 | /* | |
513 | * Activate the first processor. | |
514 | */ | |
515 | ||
44fd2299 SS |
516 | static void __init boot_cpu_init(void) |
517 | { | |
518 | int cpu = smp_processor_id(); | |
519 | /* Mark the boot cpu "present", "online" etc for SMP and UP case */ | |
520 | cpu_set(cpu, cpu_online_map); | |
521 | cpu_set(cpu, cpu_present_map); | |
522 | cpu_set(cpu, cpu_possible_map); | |
523 | } | |
524 | ||
839ad62e | 525 | void __init __weak smp_setup_processor_id(void) |
033ab7f8 AM |
526 | { |
527 | } | |
528 | ||
8c9843e5 BH |
529 | void __init __weak thread_info_cache_init(void) |
530 | { | |
531 | } | |
532 | ||
1da177e4 LT |
533 | asmlinkage void __init start_kernel(void) |
534 | { | |
535 | char * command_line; | |
536 | extern struct kernel_param __start___param[], __stop___param[]; | |
033ab7f8 AM |
537 | |
538 | smp_setup_processor_id(); | |
539 | ||
fbb9ce95 IM |
540 | /* |
541 | * Need to run as early as possible, to initialize the | |
542 | * lockdep hash: | |
543 | */ | |
c9538ed4 | 544 | unwind_init(); |
fbb9ce95 | 545 | lockdep_init(); |
ddbcc7e8 | 546 | cgroup_init_early(); |
fbb9ce95 IM |
547 | |
548 | local_irq_disable(); | |
549 | early_boot_irqs_off(); | |
243c7621 | 550 | early_init_irq_lock_class(); |
fbb9ce95 | 551 | |
1da177e4 LT |
552 | /* |
553 | * Interrupts are still disabled. Do necessary setups, then | |
554 | * enable them | |
555 | */ | |
556 | lock_kernel(); | |
906568c9 | 557 | tick_init(); |
44fd2299 | 558 | boot_cpu_init(); |
1da177e4 LT |
559 | page_address_init(); |
560 | printk(KERN_NOTICE); | |
8993780a | 561 | printk(linux_banner); |
1da177e4 | 562 | setup_arch(&command_line); |
cf475ad2 | 563 | mm_init_owner(&init_mm, &init_task); |
30d7e0d4 | 564 | setup_command_line(command_line); |
690a973f | 565 | unwind_setup(); |
1da177e4 | 566 | setup_per_cpu_areas(); |
e0982e90 | 567 | setup_nr_cpu_ids(); |
44fd2299 | 568 | smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */ |
1da177e4 LT |
569 | |
570 | /* | |
571 | * Set up the scheduler prior starting any interrupts (such as the | |
572 | * timer interrupt). Full topology setup happens at smp_init() | |
573 | * time - but meanwhile we still have a functioning scheduler. | |
574 | */ | |
575 | sched_init(); | |
576 | /* | |
577 | * Disable preemption - early bootup scheduling is extremely | |
578 | * fragile until we cpu_idle() for the first time. | |
579 | */ | |
580 | preempt_disable(); | |
581 | build_all_zonelists(); | |
582 | page_alloc_init(); | |
30d7e0d4 | 583 | printk(KERN_NOTICE "Kernel command line: %s\n", boot_command_line); |
1da177e4 | 584 | parse_early_param(); |
30d7e0d4 | 585 | parse_args("Booting kernel", static_command_line, __start___param, |
1da177e4 LT |
586 | __stop___param - __start___param, |
587 | &unknown_bootoption); | |
c4a68306 AB |
588 | if (!irqs_disabled()) { |
589 | printk(KERN_WARNING "start_kernel(): bug: interrupts were " | |
590 | "enabled *very* early, fixing it\n"); | |
591 | local_irq_disable(); | |
592 | } | |
1da177e4 LT |
593 | sort_main_extable(); |
594 | trap_init(); | |
595 | rcu_init(); | |
596 | init_IRQ(); | |
597 | pidhash_init(); | |
598 | init_timers(); | |
c0a31329 | 599 | hrtimers_init(); |
1da177e4 | 600 | softirq_init(); |
ad596171 | 601 | timekeeping_init(); |
88fecaa2 | 602 | time_init(); |
93e02814 HC |
603 | profile_init(); |
604 | if (!irqs_disabled()) | |
605 | printk("start_kernel(): bug: interrupts were enabled early\n"); | |
fbb9ce95 | 606 | early_boot_irqs_on(); |
93e02814 | 607 | local_irq_enable(); |
1da177e4 LT |
608 | |
609 | /* | |
610 | * HACK ALERT! This is early. We're enabling the console before | |
611 | * we've done PCI setups etc, and console_init() must be aware of | |
612 | * this. But we do want output early, in case something goes wrong. | |
613 | */ | |
614 | console_init(); | |
615 | if (panic_later) | |
616 | panic(panic_later, panic_param); | |
fbb9ce95 IM |
617 | |
618 | lockdep_info(); | |
619 | ||
9a11b49a IM |
620 | /* |
621 | * Need to run this when irqs are enabled, because it wants | |
622 | * to self-test [hard/soft]-irqs on/off lock inversion bugs | |
623 | * too: | |
624 | */ | |
625 | locking_selftest(); | |
626 | ||
1da177e4 LT |
627 | #ifdef CONFIG_BLK_DEV_INITRD |
628 | if (initrd_start && !initrd_below_start_ok && | |
629 | initrd_start < min_low_pfn << PAGE_SHIFT) { | |
630 | printk(KERN_CRIT "initrd overwritten (0x%08lx < 0x%08lx) - " | |
631 | "disabling it.\n",initrd_start,min_low_pfn << PAGE_SHIFT); | |
632 | initrd_start = 0; | |
633 | } | |
634 | #endif | |
635 | vfs_caches_init_early(); | |
c417f024 | 636 | cpuset_init_early(); |
1da177e4 | 637 | mem_init(); |
a03c2a48 | 638 | enable_debug_pagealloc(); |
d221938c | 639 | cpu_hotplug_init(); |
1da177e4 | 640 | kmem_cache_init(); |
199f0ca5 | 641 | idr_init_cache(); |
e7c8d5c9 | 642 | setup_per_cpu_pageset(); |
1da177e4 LT |
643 | numa_policy_init(); |
644 | if (late_time_init) | |
645 | late_time_init(); | |
646 | calibrate_delay(); | |
647 | pidmap_init(); | |
648 | pgtable_cache_init(); | |
649 | prio_tree_init(); | |
650 | anon_vma_init(); | |
651 | #ifdef CONFIG_X86 | |
652 | if (efi_enabled) | |
653 | efi_enter_virtual_mode(); | |
654 | #endif | |
8c9843e5 | 655 | thread_info_cache_init(); |
1da177e4 LT |
656 | fork_init(num_physpages); |
657 | proc_caches_init(); | |
658 | buffer_init(); | |
659 | unnamed_dev_init(); | |
660 | key_init(); | |
661 | security_init(); | |
662 | vfs_caches_init(num_physpages); | |
663 | radix_tree_init(); | |
664 | signals_init(); | |
665 | /* rootfs populating might need page-writeback */ | |
666 | page_writeback_init(); | |
667 | #ifdef CONFIG_PROC_FS | |
668 | proc_root_init(); | |
669 | #endif | |
ddbcc7e8 | 670 | cgroup_init(); |
1da177e4 | 671 | cpuset_init(); |
c757249a | 672 | taskstats_init_early(); |
ca74e92b | 673 | delayacct_init(); |
1da177e4 LT |
674 | |
675 | check_bugs(); | |
676 | ||
677 | acpi_early_init(); /* before LAPIC and SMP init */ | |
678 | ||
679 | /* Do the rest non-__init'ed, we're now alive */ | |
680 | rest_init(); | |
681 | } | |
682 | ||
683 | static int __initdata initcall_debug; | |
684 | ||
685 | static int __init initcall_debug_setup(char *str) | |
686 | { | |
687 | initcall_debug = 1; | |
688 | return 1; | |
689 | } | |
690 | __setup("initcall_debug", initcall_debug_setup); | |
691 | ||
1da177e4 LT |
692 | extern initcall_t __initcall_start[], __initcall_end[]; |
693 | ||
694 | static void __init do_initcalls(void) | |
695 | { | |
696 | initcall_t *call; | |
697 | int count = preempt_count(); | |
698 | ||
699 | for (call = __initcall_start; call < __initcall_end; call++) { | |
8f0c45cd | 700 | ktime_t t0, t1, delta; |
c1cda48a AM |
701 | char *msg = NULL; |
702 | char msgbuf[40]; | |
703 | int result; | |
1da177e4 LT |
704 | |
705 | if (initcall_debug) { | |
626adeb6 | 706 | print_fn_descriptor_symbol("calling %s()\n", |
c1cda48a | 707 | (unsigned long) *call); |
8f0c45cd | 708 | t0 = ktime_get(); |
1da177e4 LT |
709 | } |
710 | ||
c1cda48a | 711 | result = (*call)(); |
1da177e4 | 712 | |
8f0c45cd IM |
713 | if (initcall_debug) { |
714 | t1 = ktime_get(); | |
715 | delta = ktime_sub(t1, t0); | |
716 | ||
626adeb6 | 717 | print_fn_descriptor_symbol("initcall %s()", |
8f0c45cd | 718 | (unsigned long) *call); |
626adeb6 BH |
719 | printk(" returned %d after %Ld msecs\n", result, |
720 | (unsigned long long) delta.tv64 >> 20); | |
8f0c45cd IM |
721 | } |
722 | ||
f3a19cb4 | 723 | if (result && result != -ENODEV && initcall_debug) { |
c1cda48a AM |
724 | sprintf(msgbuf, "error code %d", result); |
725 | msg = msgbuf; | |
726 | } | |
1da177e4 LT |
727 | if (preempt_count() != count) { |
728 | msg = "preemption imbalance"; | |
729 | preempt_count() = count; | |
730 | } | |
731 | if (irqs_disabled()) { | |
732 | msg = "disabled interrupts"; | |
733 | local_irq_enable(); | |
734 | } | |
735 | if (msg) { | |
626adeb6 | 736 | print_fn_descriptor_symbol(KERN_WARNING "initcall %s()", |
c1cda48a | 737 | (unsigned long) *call); |
626adeb6 | 738 | printk(" returned with %s\n", msg); |
1da177e4 LT |
739 | } |
740 | } | |
741 | ||
742 | /* Make sure there is no pending stuff from the initcall sequence */ | |
743 | flush_scheduled_work(); | |
744 | } | |
745 | ||
746 | /* | |
747 | * Ok, the machine is now initialized. None of the devices | |
748 | * have been touched yet, but the CPU subsystem is up and | |
749 | * running, and memory and process management works. | |
750 | * | |
751 | * Now we can finally start doing some real work.. | |
752 | */ | |
753 | static void __init do_basic_setup(void) | |
754 | { | |
755 | /* drivers will send hotplug events */ | |
756 | init_workqueues(); | |
757 | usermodehelper_init(); | |
758 | driver_init(); | |
b04c3afb | 759 | init_irq_proc(); |
1da177e4 LT |
760 | do_initcalls(); |
761 | } | |
762 | ||
97842216 DJ |
763 | static int __initdata nosoftlockup; |
764 | ||
765 | static int __init nosoftlockup_setup(char *str) | |
766 | { | |
767 | nosoftlockup = 1; | |
768 | return 1; | |
769 | } | |
770 | __setup("nosoftlockup", nosoftlockup_setup); | |
771 | ||
88d20328 | 772 | static void __init do_pre_smp_initcalls(void) |
1da177e4 LT |
773 | { |
774 | extern int spawn_ksoftirqd(void); | |
1da177e4 LT |
775 | |
776 | migration_init(); | |
1da177e4 | 777 | spawn_ksoftirqd(); |
97842216 DJ |
778 | if (!nosoftlockup) |
779 | spawn_softlockup_task(); | |
1da177e4 LT |
780 | } |
781 | ||
782 | static void run_init_process(char *init_filename) | |
783 | { | |
784 | argv_init[0] = init_filename; | |
67608567 | 785 | kernel_execve(init_filename, argv_init, envp_init); |
1da177e4 LT |
786 | } |
787 | ||
ee5bfa64 VG |
788 | /* This is a non __init function. Force it to be noinline otherwise gcc |
789 | * makes it inline to init() and it becomes part of init.text section | |
790 | */ | |
791 | static int noinline init_post(void) | |
792 | { | |
793 | free_initmem(); | |
794 | unlock_kernel(); | |
795 | mark_rodata_ro(); | |
796 | system_state = SYSTEM_RUNNING; | |
797 | numa_default_policy(); | |
798 | ||
799 | if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0) | |
800 | printk(KERN_WARNING "Warning: unable to open an initial console.\n"); | |
801 | ||
802 | (void) sys_dup(0); | |
803 | (void) sys_dup(0); | |
804 | ||
fae5fa44 ON |
805 | current->signal->flags |= SIGNAL_UNKILLABLE; |
806 | ||
ee5bfa64 VG |
807 | if (ramdisk_execute_command) { |
808 | run_init_process(ramdisk_execute_command); | |
809 | printk(KERN_WARNING "Failed to execute %s\n", | |
810 | ramdisk_execute_command); | |
811 | } | |
812 | ||
813 | /* | |
814 | * We try each of these until one succeeds. | |
815 | * | |
816 | * The Bourne shell can be used instead of init if we are | |
817 | * trying to recover a really broken machine. | |
818 | */ | |
819 | if (execute_command) { | |
820 | run_init_process(execute_command); | |
821 | printk(KERN_WARNING "Failed to execute %s. Attempting " | |
822 | "defaults...\n", execute_command); | |
823 | } | |
824 | run_init_process("/sbin/init"); | |
825 | run_init_process("/etc/init"); | |
826 | run_init_process("/bin/init"); | |
827 | run_init_process("/bin/sh"); | |
828 | ||
829 | panic("No init found. Try passing init= option to kernel."); | |
830 | } | |
831 | ||
aae5f662 | 832 | static int __init kernel_init(void * unused) |
1da177e4 LT |
833 | { |
834 | lock_kernel(); | |
835 | /* | |
836 | * init can run on any cpu. | |
837 | */ | |
321a8e9d | 838 | set_cpus_allowed_ptr(current, CPU_MASK_ALL_PTR); |
1da177e4 LT |
839 | /* |
840 | * Tell the world that we're going to be the grim | |
841 | * reaper of innocent orphaned children. | |
842 | * | |
843 | * We don't want people to have to make incorrect | |
844 | * assumptions about where in the task array this | |
845 | * can be found. | |
846 | */ | |
84d73786 | 847 | init_pid_ns.child_reaper = current; |
1da177e4 | 848 | |
9ec52099 CLG |
849 | cad_pid = task_pid(current); |
850 | ||
ca74a6f8 | 851 | smp_prepare_cpus(setup_max_cpus); |
1da177e4 LT |
852 | |
853 | do_pre_smp_initcalls(); | |
854 | ||
1da177e4 LT |
855 | smp_init(); |
856 | sched_init_smp(); | |
857 | ||
858 | cpuset_init_smp(); | |
859 | ||
1da177e4 LT |
860 | do_basic_setup(); |
861 | ||
862 | /* | |
863 | * check if there is an early userspace init. If yes, let it do all | |
864 | * the work | |
865 | */ | |
ffdfc409 OJ |
866 | |
867 | if (!ramdisk_execute_command) | |
868 | ramdisk_execute_command = "/init"; | |
869 | ||
870 | if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) { | |
871 | ramdisk_execute_command = NULL; | |
1da177e4 | 872 | prepare_namespace(); |
ffdfc409 | 873 | } |
1da177e4 LT |
874 | |
875 | /* | |
876 | * Ok, we have completed the initial bootup, and | |
877 | * we're essentially up and running. Get rid of the | |
878 | * initmem segments and start the user-mode stuff.. | |
879 | */ | |
ee5bfa64 VG |
880 | init_post(); |
881 | return 0; | |
1da177e4 | 882 | } |