]>
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 <[email protected]> | |
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> | |
9b5609fd | 17 | #include <linux/stackprotector.h> |
1da177e4 LT |
18 | #include <linux/string.h> |
19 | #include <linux/ctype.h> | |
20 | #include <linux/delay.h> | |
1da177e4 LT |
21 | #include <linux/ioport.h> |
22 | #include <linux/init.h> | |
23 | #include <linux/smp_lock.h> | |
24 | #include <linux/initrd.h> | |
1da177e4 | 25 | #include <linux/bootmem.h> |
4a7a16dc | 26 | #include <linux/acpi.h> |
1da177e4 | 27 | #include <linux/tty.h> |
1da177e4 LT |
28 | #include <linux/percpu.h> |
29 | #include <linux/kmod.h> | |
db64fe02 | 30 | #include <linux/vmalloc.h> |
1da177e4 | 31 | #include <linux/kernel_stat.h> |
d7cd5611 | 32 | #include <linux/start_kernel.h> |
1da177e4 | 33 | #include <linux/security.h> |
3d442233 | 34 | #include <linux/smp.h> |
1da177e4 LT |
35 | #include <linux/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> | |
b6cd0b77 | 52 | #include <linux/buffer_head.h> |
94b6da5a | 53 | #include <linux/page_cgroup.h> |
9a11b49a | 54 | #include <linux/debug_locks.h> |
3ac7fe5a | 55 | #include <linux/debugobjects.h> |
fbb9ce95 | 56 | #include <linux/lockdep.h> |
3c7b4e6b | 57 | #include <linux/kmemleak.h> |
84d73786 | 58 | #include <linux/pid_namespace.h> |
1f21782e | 59 | #include <linux/device.h> |
73c27992 | 60 | #include <linux/kthread.h> |
e6fe6649 | 61 | #include <linux/sched.h> |
a1c9eea9 | 62 | #include <linux/signal.h> |
199f0ca5 | 63 | #include <linux/idr.h> |
0b4b3827 | 64 | #include <linux/kgdb.h> |
68bf21aa | 65 | #include <linux/ftrace.h> |
22a9d645 | 66 | #include <linux/async.h> |
dfec072e | 67 | #include <linux/kmemcheck.h> |
6ae6996a | 68 | #include <linux/sfi.h> |
2b2af54a | 69 | #include <linux/shmem_fs.h> |
5a0e3ad6 | 70 | #include <linux/slab.h> |
1da177e4 LT |
71 | |
72 | #include <asm/io.h> | |
73 | #include <asm/bugs.h> | |
74 | #include <asm/setup.h> | |
a940199f | 75 | #include <asm/sections.h> |
37b73c82 | 76 | #include <asm/cacheflush.h> |
1da177e4 | 77 | |
1da177e4 LT |
78 | #ifdef CONFIG_X86_LOCAL_APIC |
79 | #include <asm/smp.h> | |
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 prio_tree_init(void); |
89 | extern void radix_tree_init(void); | |
90 | extern void free_initmem(void); | |
37b73c82 AV |
91 | #ifndef CONFIG_DEBUG_RODATA |
92 | static inline void mark_rodata_ro(void) { } | |
93 | #endif | |
1da177e4 LT |
94 | |
95 | #ifdef CONFIG_TC | |
96 | extern void tc_init(void); | |
97 | #endif | |
98 | ||
a6826048 | 99 | enum system_states system_state __read_mostly; |
1da177e4 LT |
100 | EXPORT_SYMBOL(system_state); |
101 | ||
102 | /* | |
103 | * Boot command-line arguments | |
104 | */ | |
105 | #define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT | |
106 | #define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT | |
107 | ||
108 | extern void time_init(void); | |
109 | /* Default late time init is NULL. archs can override this later. */ | |
d2e3192b | 110 | void (*__initdata late_time_init)(void); |
1da177e4 LT |
111 | extern void softirq_init(void); |
112 | ||
30d7e0d4 ABL |
113 | /* Untouched command line saved by arch-specific code. */ |
114 | char __initdata boot_command_line[COMMAND_LINE_SIZE]; | |
115 | /* Untouched saved command line (eg. for /proc) */ | |
116 | char *saved_command_line; | |
117 | /* Command line for parameter parsing */ | |
118 | static char *static_command_line; | |
1da177e4 LT |
119 | |
120 | static char *execute_command; | |
ffdfc409 | 121 | static char *ramdisk_execute_command; |
1da177e4 | 122 | |
8b3b2955 | 123 | #ifdef CONFIG_SMP |
1da177e4 | 124 | /* Setup configured maximum number of CPUs to activate */ |
75cbfb97 TR |
125 | unsigned int setup_max_cpus = NR_CPUS; |
126 | EXPORT_SYMBOL(setup_max_cpus); | |
127 | ||
7e96287d | 128 | |
1da177e4 LT |
129 | /* |
130 | * Setup routine for controlling SMP activation | |
131 | * | |
132 | * Command-line option of "nosmp" or "maxcpus=0" will disable SMP | |
133 | * activation entirely (the MPS table probe still happens, though). | |
134 | * | |
135 | * Command-line option of "maxcpus=<NUM>", where <NUM> is an integer | |
136 | * greater than 0, limits the maximum number of CPUs activated in | |
137 | * SMP mode to <NUM>. | |
138 | */ | |
65a4e574 IM |
139 | |
140 | void __weak arch_disable_smp_support(void) { } | |
61ec7567 | 141 | |
1da177e4 LT |
142 | static int __init nosmp(char *str) |
143 | { | |
ca74a6f8 | 144 | setup_max_cpus = 0; |
65a4e574 IM |
145 | arch_disable_smp_support(); |
146 | ||
8b3b2955 | 147 | return 0; |
1da177e4 LT |
148 | } |
149 | ||
8b3b2955 | 150 | early_param("nosmp", nosmp); |
1da177e4 | 151 | |
2b633e3f YL |
152 | /* this is hard limit */ |
153 | static int __init nrcpus(char *str) | |
154 | { | |
155 | int nr_cpus; | |
156 | ||
157 | get_option(&str, &nr_cpus); | |
158 | if (nr_cpus > 0 && nr_cpus < nr_cpu_ids) | |
159 | nr_cpu_ids = nr_cpus; | |
160 | ||
161 | return 0; | |
162 | } | |
163 | ||
164 | early_param("nr_cpus", nrcpus); | |
165 | ||
1da177e4 LT |
166 | static int __init maxcpus(char *str) |
167 | { | |
ca74a6f8 AK |
168 | get_option(&str, &setup_max_cpus); |
169 | if (setup_max_cpus == 0) | |
65a4e574 | 170 | arch_disable_smp_support(); |
61ec7567 LB |
171 | |
172 | return 0; | |
1da177e4 LT |
173 | } |
174 | ||
81340977 | 175 | early_param("maxcpus", maxcpus); |
8b3b2955 | 176 | #else |
7463e633 | 177 | static const unsigned int setup_max_cpus = NR_CPUS; |
8b3b2955 JB |
178 | #endif |
179 | ||
180 | /* | |
181 | * If set, this is an indication to the drivers that reset the underlying | |
182 | * device before going ahead with the initialization otherwise driver might | |
183 | * rely on the BIOS and skip the reset operation. | |
184 | * | |
185 | * This is useful if kernel is booting in an unreliable environment. | |
186 | * For ex. kdump situaiton where previous kernel has crashed, BIOS has been | |
187 | * skipped and devices will be in unknown state. | |
188 | */ | |
189 | unsigned int reset_devices; | |
190 | EXPORT_SYMBOL(reset_devices); | |
1da177e4 | 191 | |
7e96287d VG |
192 | static int __init set_reset_devices(char *str) |
193 | { | |
194 | reset_devices = 1; | |
195 | return 1; | |
196 | } | |
197 | ||
198 | __setup("reset_devices", set_reset_devices); | |
199 | ||
d7627467 DH |
200 | static const char * argv_init[MAX_INIT_ARGS+2] = { "init", NULL, }; |
201 | const char * envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, }; | |
1da177e4 LT |
202 | static const char *panic_later, *panic_param; |
203 | ||
914dcaa8 | 204 | extern const struct obs_kernel_param __setup_start[], __setup_end[]; |
1da177e4 LT |
205 | |
206 | static int __init obsolete_checksetup(char *line) | |
207 | { | |
914dcaa8 | 208 | const struct obs_kernel_param *p; |
33df0d19 | 209 | int had_early_param = 0; |
1da177e4 LT |
210 | |
211 | p = __setup_start; | |
212 | do { | |
213 | int n = strlen(p->str); | |
214 | if (!strncmp(line, p->str, n)) { | |
215 | if (p->early) { | |
33df0d19 RR |
216 | /* Already done in parse_early_param? |
217 | * (Needs exact match on param part). | |
218 | * Keep iterating, as we can have early | |
219 | * params and __setups of same names 8( */ | |
1da177e4 | 220 | if (line[n] == '\0' || line[n] == '=') |
33df0d19 | 221 | had_early_param = 1; |
1da177e4 LT |
222 | } else if (!p->setup_func) { |
223 | printk(KERN_WARNING "Parameter %s is obsolete," | |
224 | " ignored\n", p->str); | |
225 | return 1; | |
226 | } else if (p->setup_func(line + n)) | |
227 | return 1; | |
228 | } | |
229 | p++; | |
230 | } while (p < __setup_end); | |
33df0d19 RR |
231 | |
232 | return had_early_param; | |
1da177e4 LT |
233 | } |
234 | ||
235 | /* | |
236 | * This should be approx 2 Bo*oMips to start (note initial shift), and will | |
237 | * still work even if initially too large, it will just take slightly longer | |
238 | */ | |
239 | unsigned long loops_per_jiffy = (1<<12); | |
240 | ||
241 | EXPORT_SYMBOL(loops_per_jiffy); | |
242 | ||
243 | static int __init debug_kernel(char *str) | |
244 | { | |
1da177e4 | 245 | console_loglevel = 10; |
f6f21c81 | 246 | return 0; |
1da177e4 LT |
247 | } |
248 | ||
249 | static int __init quiet_kernel(char *str) | |
250 | { | |
1da177e4 | 251 | console_loglevel = 4; |
f6f21c81 | 252 | return 0; |
1da177e4 LT |
253 | } |
254 | ||
f6f21c81 YL |
255 | early_param("debug", debug_kernel); |
256 | early_param("quiet", quiet_kernel); | |
1da177e4 LT |
257 | |
258 | static int __init loglevel(char *str) | |
259 | { | |
260 | get_option(&str, &console_loglevel); | |
d9d4fcfe | 261 | return 0; |
1da177e4 LT |
262 | } |
263 | ||
f6f21c81 | 264 | early_param("loglevel", loglevel); |
1da177e4 LT |
265 | |
266 | /* | |
267 | * Unknown boot options get handed to init, unless they look like | |
f066a4f6 | 268 | * unused parameters (modprobe will find them in /proc/cmdline). |
1da177e4 LT |
269 | */ |
270 | static int __init unknown_bootoption(char *param, char *val) | |
271 | { | |
272 | /* Change NUL term back to "=", to make "param" the whole string. */ | |
273 | if (val) { | |
274 | /* param=val or param="val"? */ | |
275 | if (val == param+strlen(param)+1) | |
276 | val[-1] = '='; | |
277 | else if (val == param+strlen(param)+2) { | |
278 | val[-2] = '='; | |
279 | memmove(val-1, val, strlen(val)+1); | |
280 | val--; | |
281 | } else | |
282 | BUG(); | |
283 | } | |
284 | ||
285 | /* Handle obsolete-style parameters */ | |
286 | if (obsolete_checksetup(param)) | |
287 | return 0; | |
288 | ||
f066a4f6 RR |
289 | /* Unused module parameter. */ |
290 | if (strchr(param, '.') && (!val || strchr(param, '.') < val)) | |
1da177e4 | 291 | return 0; |
1da177e4 LT |
292 | |
293 | if (panic_later) | |
294 | return 0; | |
295 | ||
296 | if (val) { | |
297 | /* Environment option */ | |
298 | unsigned int i; | |
299 | for (i = 0; envp_init[i]; i++) { | |
300 | if (i == MAX_INIT_ENVS) { | |
301 | panic_later = "Too many boot env vars at `%s'"; | |
302 | panic_param = param; | |
303 | } | |
304 | if (!strncmp(param, envp_init[i], val - param)) | |
305 | break; | |
306 | } | |
307 | envp_init[i] = param; | |
308 | } else { | |
309 | /* Command line option */ | |
310 | unsigned int i; | |
311 | for (i = 0; argv_init[i]; i++) { | |
312 | if (i == MAX_INIT_ARGS) { | |
313 | panic_later = "Too many boot init vars at `%s'"; | |
314 | panic_param = param; | |
315 | } | |
316 | } | |
317 | argv_init[i] = param; | |
318 | } | |
319 | return 0; | |
320 | } | |
321 | ||
12d6f21e IM |
322 | #ifdef CONFIG_DEBUG_PAGEALLOC |
323 | int __read_mostly debug_pagealloc_enabled = 0; | |
324 | #endif | |
325 | ||
1da177e4 LT |
326 | static int __init init_setup(char *str) |
327 | { | |
328 | unsigned int i; | |
329 | ||
330 | execute_command = str; | |
331 | /* | |
332 | * In case LILO is going to boot us with default command line, | |
333 | * it prepends "auto" before the whole cmdline which makes | |
334 | * the shell think it should execute a script with such name. | |
335 | * So we ignore all arguments entered _before_ init=... [MJ] | |
336 | */ | |
337 | for (i = 1; i < MAX_INIT_ARGS; i++) | |
338 | argv_init[i] = NULL; | |
339 | return 1; | |
340 | } | |
341 | __setup("init=", init_setup); | |
342 | ||
ffdfc409 OJ |
343 | static int __init rdinit_setup(char *str) |
344 | { | |
345 | unsigned int i; | |
346 | ||
347 | ramdisk_execute_command = str; | |
348 | /* See "auto" comment in init_setup */ | |
349 | for (i = 1; i < MAX_INIT_ARGS; i++) | |
350 | argv_init[i] = NULL; | |
351 | return 1; | |
352 | } | |
353 | __setup("rdinit=", rdinit_setup); | |
354 | ||
1da177e4 LT |
355 | #ifndef CONFIG_SMP |
356 | ||
357 | #ifdef CONFIG_X86_LOCAL_APIC | |
358 | static void __init smp_init(void) | |
359 | { | |
360 | APIC_init_uniprocessor(); | |
361 | } | |
362 | #else | |
363 | #define smp_init() do { } while (0) | |
364 | #endif | |
365 | ||
e0982e90 | 366 | static inline void setup_nr_cpu_ids(void) { } |
1da177e4 LT |
367 | static inline void smp_prepare_cpus(unsigned int maxcpus) { } |
368 | ||
369 | #else | |
370 | ||
e0982e90 MT |
371 | /* Setup number of possible processor ids */ |
372 | int nr_cpu_ids __read_mostly = NR_CPUS; | |
373 | EXPORT_SYMBOL(nr_cpu_ids); | |
374 | ||
375 | /* An arch may set nr_cpu_ids earlier if needed, so this would be redundant */ | |
376 | static void __init setup_nr_cpu_ids(void) | |
377 | { | |
e0c0ba73 | 378 | nr_cpu_ids = find_last_bit(cpumask_bits(cpu_possible_mask),NR_CPUS) + 1; |
e0982e90 MT |
379 | } |
380 | ||
1da177e4 LT |
381 | /* Called by boot processor to activate the rest. */ |
382 | static void __init smp_init(void) | |
383 | { | |
53b8a315 | 384 | unsigned int cpu; |
1da177e4 LT |
385 | |
386 | /* FIXME: This should be done in userspace --RR */ | |
53b8a315 | 387 | for_each_present_cpu(cpu) { |
ca74a6f8 | 388 | if (num_online_cpus() >= setup_max_cpus) |
1da177e4 | 389 | break; |
53b8a315 CL |
390 | if (!cpu_online(cpu)) |
391 | cpu_up(cpu); | |
1da177e4 LT |
392 | } |
393 | ||
394 | /* Any cleanup work */ | |
395 | printk(KERN_INFO "Brought up %ld CPUs\n", (long)num_online_cpus()); | |
ca74a6f8 | 396 | smp_cpus_done(setup_max_cpus); |
1da177e4 LT |
397 | } |
398 | ||
399 | #endif | |
400 | ||
30d7e0d4 ABL |
401 | /* |
402 | * We need to store the untouched command line for future reference. | |
403 | * We also need to store the touched command line since the parameter | |
404 | * parsing is performed in place, and we should allow a component to | |
405 | * store reference of name/value for future reference. | |
406 | */ | |
407 | static void __init setup_command_line(char *command_line) | |
408 | { | |
409 | saved_command_line = alloc_bootmem(strlen (boot_command_line)+1); | |
410 | static_command_line = alloc_bootmem(strlen (command_line)+1); | |
411 | strcpy (saved_command_line, boot_command_line); | |
412 | strcpy (static_command_line, command_line); | |
413 | } | |
414 | ||
1da177e4 LT |
415 | /* |
416 | * We need to finalize in a non-__init function or else race conditions | |
417 | * between the root thread and the init thread may cause start_kernel to | |
418 | * be reaped by free_initmem before the root thread has proceeded to | |
419 | * cpu_idle. | |
420 | * | |
421 | * gcc-3.4 accidentally inlines this function, so use noinline. | |
422 | */ | |
423 | ||
b433c3d4 PZ |
424 | static __initdata DECLARE_COMPLETION(kthreadd_done); |
425 | ||
f99ebf0a | 426 | static noinline void __init_refok rest_init(void) |
1da177e4 LT |
427 | __releases(kernel_lock) |
428 | { | |
73c27992 EB |
429 | int pid; |
430 | ||
7db905e6 | 431 | rcu_scheduler_starting(); |
b433c3d4 | 432 | /* |
97158569 | 433 | * We need to spawn init first so that it obtains pid 1, however |
b433c3d4 PZ |
434 | * the init task will end up wanting to create kthreads, which, if |
435 | * we schedule it before we create kthreadd, will OOPS. | |
436 | */ | |
aae5f662 | 437 | kernel_thread(kernel_init, NULL, CLONE_FS | CLONE_SIGHAND); |
1da177e4 | 438 | numa_default_policy(); |
73c27992 | 439 | pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES); |
d11c563d | 440 | rcu_read_lock(); |
5cd20455 | 441 | kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns); |
d11c563d | 442 | rcu_read_unlock(); |
b433c3d4 | 443 | complete(&kthreadd_done); |
f340c0d1 IM |
444 | |
445 | /* | |
446 | * The boot idle thread must execute schedule() | |
1df21055 | 447 | * at least once to get things moving: |
f340c0d1 | 448 | */ |
1df21055 | 449 | init_idle_bootup_task(current); |
5bfb5d69 | 450 | preempt_enable_no_resched(); |
f340c0d1 | 451 | schedule(); |
5bfb5d69 | 452 | preempt_disable(); |
f340c0d1 | 453 | |
5bfb5d69 | 454 | /* Call into cpu_idle with preempt disabled */ |
1da177e4 | 455 | cpu_idle(); |
1df21055 | 456 | } |
1da177e4 LT |
457 | |
458 | /* Check for early params. */ | |
459 | static int __init do_early_param(char *param, char *val) | |
460 | { | |
914dcaa8 | 461 | const struct obs_kernel_param *p; |
1da177e4 LT |
462 | |
463 | for (p = __setup_start; p < __setup_end; p++) { | |
18a8bd94 YL |
464 | if ((p->early && strcmp(param, p->str) == 0) || |
465 | (strcmp(param, "console") == 0 && | |
466 | strcmp(p->str, "earlycon") == 0) | |
467 | ) { | |
1da177e4 LT |
468 | if (p->setup_func(val) != 0) |
469 | printk(KERN_WARNING | |
470 | "Malformed early option '%s'\n", param); | |
471 | } | |
472 | } | |
473 | /* We accept everything at this stage. */ | |
474 | return 0; | |
475 | } | |
476 | ||
13977091 MD |
477 | void __init parse_early_options(char *cmdline) |
478 | { | |
479 | parse_args("early options", cmdline, NULL, 0, do_early_param); | |
480 | } | |
481 | ||
1da177e4 LT |
482 | /* Arch code calls this early on, or if not, just before other parsing. */ |
483 | void __init parse_early_param(void) | |
484 | { | |
485 | static __initdata int done = 0; | |
486 | static __initdata char tmp_cmdline[COMMAND_LINE_SIZE]; | |
487 | ||
488 | if (done) | |
489 | return; | |
490 | ||
491 | /* All fall through to do_early_param. */ | |
30d7e0d4 | 492 | strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE); |
13977091 | 493 | parse_early_options(tmp_cmdline); |
1da177e4 LT |
494 | done = 1; |
495 | } | |
496 | ||
497 | /* | |
498 | * Activate the first processor. | |
499 | */ | |
500 | ||
44fd2299 SS |
501 | static void __init boot_cpu_init(void) |
502 | { | |
503 | int cpu = smp_processor_id(); | |
504 | /* Mark the boot cpu "present", "online" etc for SMP and UP case */ | |
915441b6 | 505 | set_cpu_online(cpu, true); |
933b0618 | 506 | set_cpu_active(cpu, true); |
915441b6 RR |
507 | set_cpu_present(cpu, true); |
508 | set_cpu_possible(cpu, true); | |
44fd2299 SS |
509 | } |
510 | ||
839ad62e | 511 | void __init __weak smp_setup_processor_id(void) |
033ab7f8 AM |
512 | { |
513 | } | |
514 | ||
8c9843e5 BH |
515 | void __init __weak thread_info_cache_init(void) |
516 | { | |
517 | } | |
518 | ||
444f478f PE |
519 | /* |
520 | * Set up kernel memory allocators | |
521 | */ | |
522 | static void __init mm_init(void) | |
523 | { | |
ca371c0d KH |
524 | /* |
525 | * page_cgroup requires countinous pages as memmap | |
526 | * and it's bigger than MAX_ORDER unless SPARSEMEM. | |
527 | */ | |
528 | page_cgroup_init_flatmem(); | |
444f478f PE |
529 | mem_init(); |
530 | kmem_cache_init(); | |
099a19d9 | 531 | percpu_init_late(); |
c868d550 | 532 | pgtable_cache_init(); |
444f478f PE |
533 | vmalloc_init(); |
534 | } | |
535 | ||
1da177e4 LT |
536 | asmlinkage void __init start_kernel(void) |
537 | { | |
538 | char * command_line; | |
914dcaa8 | 539 | extern const struct kernel_param __start___param[], __stop___param[]; |
033ab7f8 AM |
540 | |
541 | smp_setup_processor_id(); | |
542 | ||
fbb9ce95 IM |
543 | /* |
544 | * Need to run as early as possible, to initialize the | |
545 | * lockdep hash: | |
546 | */ | |
547 | lockdep_init(); | |
3ac7fe5a | 548 | debug_objects_early_init(); |
42059429 IM |
549 | |
550 | /* | |
551 | * Set up the the initial canary ASAP: | |
552 | */ | |
553 | boot_init_stack_canary(); | |
554 | ||
ddbcc7e8 | 555 | cgroup_init_early(); |
fbb9ce95 IM |
556 | |
557 | local_irq_disable(); | |
558 | early_boot_irqs_off(); | |
243c7621 | 559 | early_init_irq_lock_class(); |
fbb9ce95 | 560 | |
1da177e4 LT |
561 | /* |
562 | * Interrupts are still disabled. Do necessary setups, then | |
563 | * enable them | |
564 | */ | |
906568c9 | 565 | tick_init(); |
44fd2299 | 566 | boot_cpu_init(); |
1da177e4 | 567 | page_address_init(); |
657cafa6 | 568 | printk(KERN_NOTICE "%s", linux_banner); |
1da177e4 | 569 | setup_arch(&command_line); |
cf475ad2 | 570 | mm_init_owner(&init_mm, &init_task); |
30d7e0d4 | 571 | setup_command_line(command_line); |
e0982e90 | 572 | setup_nr_cpu_ids(); |
d6647bdf | 573 | setup_per_cpu_areas(); |
44fd2299 | 574 | smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */ |
1da177e4 | 575 | |
1f522509 | 576 | build_all_zonelists(NULL); |
83b519e8 PE |
577 | page_alloc_init(); |
578 | ||
579 | printk(KERN_NOTICE "Kernel command line: %s\n", boot_command_line); | |
580 | parse_early_param(); | |
581 | parse_args("Booting kernel", static_command_line, __start___param, | |
582 | __stop___param - __start___param, | |
583 | &unknown_bootoption); | |
584 | /* | |
585 | * These use large bootmem allocations and must precede | |
586 | * kmem_cache_init() | |
587 | */ | |
588 | pidhash_init(); | |
83b519e8 PE |
589 | vfs_caches_init_early(); |
590 | sort_main_extable(); | |
591 | trap_init(); | |
444f478f | 592 | mm_init(); |
1da177e4 LT |
593 | /* |
594 | * Set up the scheduler prior starting any interrupts (such as the | |
595 | * timer interrupt). Full topology setup happens at smp_init() | |
596 | * time - but meanwhile we still have a functioning scheduler. | |
597 | */ | |
598 | sched_init(); | |
599 | /* | |
600 | * Disable preemption - early bootup scheduling is extremely | |
601 | * fragile until we cpu_idle() for the first time. | |
602 | */ | |
603 | preempt_disable(); | |
c4a68306 AB |
604 | if (!irqs_disabled()) { |
605 | printk(KERN_WARNING "start_kernel(): bug: interrupts were " | |
606 | "enabled *very* early, fixing it\n"); | |
607 | local_irq_disable(); | |
608 | } | |
1da177e4 | 609 | rcu_init(); |
773e3eb7 | 610 | radix_tree_init(); |
0b8f1efa YL |
611 | /* init some links before init_ISA_irqs() */ |
612 | early_irq_init(); | |
1da177e4 | 613 | init_IRQ(); |
3c7b4e6b | 614 | prio_tree_init(); |
1da177e4 | 615 | init_timers(); |
c0a31329 | 616 | hrtimers_init(); |
1da177e4 | 617 | softirq_init(); |
ad596171 | 618 | timekeeping_init(); |
88fecaa2 | 619 | time_init(); |
93e02814 HC |
620 | profile_init(); |
621 | if (!irqs_disabled()) | |
24d431d0 RL |
622 | printk(KERN_CRIT "start_kernel(): bug: interrupts were " |
623 | "enabled early\n"); | |
fbb9ce95 | 624 | early_boot_irqs_on(); |
93e02814 | 625 | local_irq_enable(); |
dcce284a BH |
626 | |
627 | /* Interrupts are enabled now so all GFP allocations are safe. */ | |
452aa699 | 628 | gfp_allowed_mask = __GFP_BITS_MASK; |
dcce284a | 629 | |
7e85ee0c | 630 | kmem_cache_init_late(); |
1da177e4 LT |
631 | |
632 | /* | |
633 | * HACK ALERT! This is early. We're enabling the console before | |
634 | * we've done PCI setups etc, and console_init() must be aware of | |
635 | * this. But we do want output early, in case something goes wrong. | |
636 | */ | |
637 | console_init(); | |
638 | if (panic_later) | |
639 | panic(panic_later, panic_param); | |
fbb9ce95 IM |
640 | |
641 | lockdep_info(); | |
642 | ||
9a11b49a IM |
643 | /* |
644 | * Need to run this when irqs are enabled, because it wants | |
645 | * to self-test [hard/soft]-irqs on/off lock inversion bugs | |
646 | * too: | |
647 | */ | |
648 | locking_selftest(); | |
649 | ||
1da177e4 LT |
650 | #ifdef CONFIG_BLK_DEV_INITRD |
651 | if (initrd_start && !initrd_below_start_ok && | |
bd673c7c | 652 | page_to_pfn(virt_to_page((void *)initrd_start)) < min_low_pfn) { |
1da177e4 | 653 | printk(KERN_CRIT "initrd overwritten (0x%08lx < 0x%08lx) - " |
fb6624eb | 654 | "disabling it.\n", |
bd673c7c GU |
655 | page_to_pfn(virt_to_page((void *)initrd_start)), |
656 | min_low_pfn); | |
1da177e4 LT |
657 | initrd_start = 0; |
658 | } | |
659 | #endif | |
94b6da5a | 660 | page_cgroup_init(); |
a03c2a48 | 661 | enable_debug_pagealloc(); |
3c7b4e6b | 662 | kmemleak_init(); |
3ac7fe5a | 663 | debug_objects_mem_init(); |
199f0ca5 | 664 | idr_init_cache(); |
e7c8d5c9 | 665 | setup_per_cpu_pageset(); |
1da177e4 LT |
666 | numa_policy_init(); |
667 | if (late_time_init) | |
668 | late_time_init(); | |
fa84e9ee | 669 | sched_clock_init(); |
1da177e4 LT |
670 | calibrate_delay(); |
671 | pidmap_init(); | |
1da177e4 LT |
672 | anon_vma_init(); |
673 | #ifdef CONFIG_X86 | |
674 | if (efi_enabled) | |
675 | efi_enter_virtual_mode(); | |
676 | #endif | |
8c9843e5 | 677 | thread_info_cache_init(); |
d84f4f99 | 678 | cred_init(); |
4481374c | 679 | fork_init(totalram_pages); |
1da177e4 LT |
680 | proc_caches_init(); |
681 | buffer_init(); | |
1da177e4 LT |
682 | key_init(); |
683 | security_init(); | |
0b4b3827 | 684 | dbg_late_init(); |
4481374c | 685 | vfs_caches_init(totalram_pages); |
1da177e4 LT |
686 | signals_init(); |
687 | /* rootfs populating might need page-writeback */ | |
688 | page_writeback_init(); | |
689 | #ifdef CONFIG_PROC_FS | |
690 | proc_root_init(); | |
691 | #endif | |
ddbcc7e8 | 692 | cgroup_init(); |
1da177e4 | 693 | cpuset_init(); |
c757249a | 694 | taskstats_init_early(); |
ca74e92b | 695 | delayacct_init(); |
1da177e4 LT |
696 | |
697 | check_bugs(); | |
698 | ||
699 | acpi_early_init(); /* before LAPIC and SMP init */ | |
6ae6996a | 700 | sfi_init_late(); |
1da177e4 | 701 | |
68bf21aa SR |
702 | ftrace_init(); |
703 | ||
1da177e4 LT |
704 | /* Do the rest non-__init'ed, we're now alive */ |
705 | rest_init(); | |
706 | } | |
707 | ||
b99b87f7 PO |
708 | /* Call all constructor functions linked into the kernel. */ |
709 | static void __init do_ctors(void) | |
710 | { | |
711 | #ifdef CONFIG_CONSTRUCTORS | |
196a15b4 | 712 | ctor_fn_t *fn = (ctor_fn_t *) __ctors_start; |
b99b87f7 | 713 | |
196a15b4 HS |
714 | for (; fn < (ctor_fn_t *) __ctors_end; fn++) |
715 | (*fn)(); | |
b99b87f7 PO |
716 | #endif |
717 | } | |
718 | ||
22a9d645 | 719 | int initcall_debug; |
d0ea3d7d | 720 | core_param(initcall_debug, initcall_debug, bool, 0644); |
1da177e4 | 721 | |
4a683bf9 | 722 | static char msgbuf[64]; |
4a683bf9 | 723 | |
e4461271 | 724 | static int __init_or_module do_one_initcall_debug(initcall_t fn) |
1da177e4 | 725 | { |
74239072 | 726 | ktime_t calltime, delta, rettime; |
30dbb20e AW |
727 | unsigned long long duration; |
728 | int ret; | |
1da177e4 | 729 | |
22c5c03b KW |
730 | printk(KERN_DEBUG "calling %pF @ %i\n", fn, task_pid_nr(current)); |
731 | calltime = ktime_get(); | |
30dbb20e | 732 | ret = fn(); |
22c5c03b KW |
733 | rettime = ktime_get(); |
734 | delta = ktime_sub(rettime, calltime); | |
735 | duration = (unsigned long long) ktime_to_ns(delta) >> 10; | |
736 | printk(KERN_DEBUG "initcall %pF returned %d after %lld usecs\n", fn, | |
737 | ret, duration); | |
1da177e4 | 738 | |
22c5c03b KW |
739 | return ret; |
740 | } | |
741 | ||
e4461271 | 742 | int __init_or_module do_one_initcall(initcall_t fn) |
22c5c03b KW |
743 | { |
744 | int count = preempt_count(); | |
745 | int ret; | |
746 | ||
747 | if (initcall_debug) | |
748 | ret = do_one_initcall_debug(fn); | |
749 | else | |
750 | ret = fn(); | |
8f0c45cd | 751 | |
e0df154f | 752 | msgbuf[0] = 0; |
e662e1cf | 753 | |
30dbb20e AW |
754 | if (ret && ret != -ENODEV && initcall_debug) |
755 | sprintf(msgbuf, "error code %d ", ret); | |
e662e1cf | 756 | |
e0df154f | 757 | if (preempt_count() != count) { |
a76bfd0d | 758 | strlcat(msgbuf, "preemption imbalance ", sizeof(msgbuf)); |
e0df154f | 759 | preempt_count() = count; |
1da177e4 | 760 | } |
e0df154f | 761 | if (irqs_disabled()) { |
a76bfd0d | 762 | strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf)); |
e0df154f LT |
763 | local_irq_enable(); |
764 | } | |
765 | if (msgbuf[0]) { | |
96d746c6 | 766 | printk("initcall %pF returned with %s\n", fn, msgbuf); |
e0df154f | 767 | } |
59f9415f | 768 | |
30dbb20e | 769 | return ret; |
e0df154f LT |
770 | } |
771 | ||
772 | ||
c2147a50 | 773 | extern initcall_t __initcall_start[], __initcall_end[], __early_initcall_end[]; |
e0df154f LT |
774 | |
775 | static void __init do_initcalls(void) | |
776 | { | |
196a15b4 | 777 | initcall_t *fn; |
e0df154f | 778 | |
196a15b4 HS |
779 | for (fn = __early_initcall_end; fn < __initcall_end; fn++) |
780 | do_one_initcall(*fn); | |
1da177e4 LT |
781 | |
782 | /* Make sure there is no pending stuff from the initcall sequence */ | |
783 | flush_scheduled_work(); | |
784 | } | |
785 | ||
786 | /* | |
787 | * Ok, the machine is now initialized. None of the devices | |
788 | * have been touched yet, but the CPU subsystem is up and | |
789 | * running, and memory and process management works. | |
790 | * | |
791 | * Now we can finally start doing some real work.. | |
792 | */ | |
793 | static void __init do_basic_setup(void) | |
794 | { | |
759ee091 | 795 | cpuset_init_smp(); |
1da177e4 | 796 | usermodehelper_init(); |
2b2af54a | 797 | init_tmpfs(); |
1da177e4 | 798 | driver_init(); |
b04c3afb | 799 | init_irq_proc(); |
b99b87f7 | 800 | do_ctors(); |
1da177e4 LT |
801 | do_initcalls(); |
802 | } | |
803 | ||
7babe8db | 804 | static void __init do_pre_smp_initcalls(void) |
c2147a50 | 805 | { |
196a15b4 | 806 | initcall_t *fn; |
c2147a50 | 807 | |
196a15b4 HS |
808 | for (fn = __initcall_start; fn < __early_initcall_end; fn++) |
809 | do_one_initcall(*fn); | |
c2147a50 EGM |
810 | } |
811 | ||
d7627467 | 812 | static void run_init_process(const char *init_filename) |
1da177e4 LT |
813 | { |
814 | argv_init[0] = init_filename; | |
67608567 | 815 | kernel_execve(init_filename, argv_init, envp_init); |
1da177e4 LT |
816 | } |
817 | ||
ee5bfa64 VG |
818 | /* This is a non __init function. Force it to be noinline otherwise gcc |
819 | * makes it inline to init() and it becomes part of init.text section | |
820 | */ | |
f99ebf0a | 821 | static noinline int init_post(void) |
acdd052a | 822 | __releases(kernel_lock) |
ee5bfa64 | 823 | { |
22a9d645 AV |
824 | /* need to finish all async __init code before freeing the memory */ |
825 | async_synchronize_full(); | |
ee5bfa64 | 826 | free_initmem(); |
ee5bfa64 VG |
827 | mark_rodata_ro(); |
828 | system_state = SYSTEM_RUNNING; | |
829 | numa_default_policy(); | |
830 | ||
ee5bfa64 | 831 | |
fae5fa44 ON |
832 | current->signal->flags |= SIGNAL_UNKILLABLE; |
833 | ||
ee5bfa64 VG |
834 | if (ramdisk_execute_command) { |
835 | run_init_process(ramdisk_execute_command); | |
836 | printk(KERN_WARNING "Failed to execute %s\n", | |
837 | ramdisk_execute_command); | |
838 | } | |
839 | ||
840 | /* | |
841 | * We try each of these until one succeeds. | |
842 | * | |
843 | * The Bourne shell can be used instead of init if we are | |
844 | * trying to recover a really broken machine. | |
845 | */ | |
846 | if (execute_command) { | |
847 | run_init_process(execute_command); | |
848 | printk(KERN_WARNING "Failed to execute %s. Attempting " | |
849 | "defaults...\n", execute_command); | |
850 | } | |
851 | run_init_process("/sbin/init"); | |
852 | run_init_process("/etc/init"); | |
853 | run_init_process("/bin/init"); | |
854 | run_init_process("/bin/sh"); | |
855 | ||
9a85b8d6 AM |
856 | panic("No init found. Try passing init= option to kernel. " |
857 | "See Linux Documentation/init.txt for guidance."); | |
ee5bfa64 VG |
858 | } |
859 | ||
aae5f662 | 860 | static int __init kernel_init(void * unused) |
1da177e4 | 861 | { |
b433c3d4 PZ |
862 | /* |
863 | * Wait until kthreadd is all set-up. | |
864 | */ | |
865 | wait_for_completion(&kthreadd_done); | |
58568d2a MX |
866 | /* |
867 | * init can allocate pages on any node | |
868 | */ | |
5ab116c9 | 869 | set_mems_allowed(node_states[N_HIGH_MEMORY]); |
1da177e4 LT |
870 | /* |
871 | * init can run on any cpu. | |
872 | */ | |
1a2142af | 873 | set_cpus_allowed_ptr(current, cpu_all_mask); |
1da177e4 LT |
874 | /* |
875 | * Tell the world that we're going to be the grim | |
876 | * reaper of innocent orphaned children. | |
877 | * | |
878 | * We don't want people to have to make incorrect | |
879 | * assumptions about where in the task array this | |
880 | * can be found. | |
881 | */ | |
84d73786 | 882 | init_pid_ns.child_reaper = current; |
1da177e4 | 883 | |
9ec52099 CLG |
884 | cad_pid = task_pid(current); |
885 | ||
ca74a6f8 | 886 | smp_prepare_cpus(setup_max_cpus); |
1da177e4 LT |
887 | |
888 | do_pre_smp_initcalls(); | |
889 | ||
1da177e4 LT |
890 | smp_init(); |
891 | sched_init_smp(); | |
892 | ||
1da177e4 LT |
893 | do_basic_setup(); |
894 | ||
2bd3a997 EB |
895 | /* Open the /dev/console on the rootfs, this should never fail */ |
896 | if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0) | |
897 | printk(KERN_WARNING "Warning: unable to open an initial console.\n"); | |
898 | ||
899 | (void) sys_dup(0); | |
900 | (void) sys_dup(0); | |
1da177e4 LT |
901 | /* |
902 | * check if there is an early userspace init. If yes, let it do all | |
903 | * the work | |
904 | */ | |
ffdfc409 OJ |
905 | |
906 | if (!ramdisk_execute_command) | |
907 | ramdisk_execute_command = "/init"; | |
908 | ||
909 | if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) { | |
910 | ramdisk_execute_command = NULL; | |
1da177e4 | 911 | prepare_namespace(); |
ffdfc409 | 912 | } |
1da177e4 LT |
913 | |
914 | /* | |
915 | * Ok, we have completed the initial bootup, and | |
916 | * we're essentially up and running. Get rid of the | |
917 | * initmem segments and start the user-mode stuff.. | |
918 | */ | |
71566a0d | 919 | |
ee5bfa64 VG |
920 | init_post(); |
921 | return 0; | |
1da177e4 | 922 | } |