]>
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> | |
1da177e4 LT |
25 | #include <linux/bootmem.h> |
26 | #include <linux/tty.h> | |
27 | #include <linux/gfp.h> | |
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/workqueue.h> |
36 | #include <linux/profile.h> | |
37 | #include <linux/rcupdate.h> | |
38 | #include <linux/moduleparam.h> | |
39 | #include <linux/kallsyms.h> | |
40 | #include <linux/writeback.h> | |
41 | #include <linux/cpu.h> | |
42 | #include <linux/cpuset.h> | |
ddbcc7e8 | 43 | #include <linux/cgroup.h> |
1da177e4 | 44 | #include <linux/efi.h> |
906568c9 | 45 | #include <linux/tick.h> |
6168a702 | 46 | #include <linux/interrupt.h> |
c757249a | 47 | #include <linux/taskstats_kern.h> |
ca74e92b | 48 | #include <linux/delayacct.h> |
1da177e4 LT |
49 | #include <linux/unistd.h> |
50 | #include <linux/rmap.h> | |
51 | #include <linux/mempolicy.h> | |
52 | #include <linux/key.h> | |
4552d5dc | 53 | #include <linux/unwind.h> |
b6cd0b77 | 54 | #include <linux/buffer_head.h> |
94b6da5a | 55 | #include <linux/page_cgroup.h> |
9a11b49a | 56 | #include <linux/debug_locks.h> |
3ac7fe5a | 57 | #include <linux/debugobjects.h> |
fbb9ce95 | 58 | #include <linux/lockdep.h> |
84d73786 | 59 | #include <linux/pid_namespace.h> |
1f21782e | 60 | #include <linux/device.h> |
73c27992 | 61 | #include <linux/kthread.h> |
e6fe6649 | 62 | #include <linux/sched.h> |
a1c9eea9 | 63 | #include <linux/signal.h> |
199f0ca5 | 64 | #include <linux/idr.h> |
68bf21aa | 65 | #include <linux/ftrace.h> |
1da177e4 LT |
66 | |
67 | #include <asm/io.h> | |
68 | #include <asm/bugs.h> | |
69 | #include <asm/setup.h> | |
a940199f | 70 | #include <asm/sections.h> |
37b73c82 | 71 | #include <asm/cacheflush.h> |
1da177e4 | 72 | |
1da177e4 LT |
73 | #ifdef CONFIG_X86_LOCAL_APIC |
74 | #include <asm/smp.h> | |
75 | #endif | |
76 | ||
77 | /* | |
69c99ac1 CQH |
78 | * This is one of the first .c files built. Error out early if we have compiler |
79 | * trouble. | |
1da177e4 | 80 | */ |
1da177e4 | 81 | |
f1a60dbf AM |
82 | #if __GNUC__ == 4 && __GNUC_MINOR__ == 1 && __GNUC_PATCHLEVEL__ == 0 |
83 | #warning gcc-4.1.0 is known to miscompile the kernel. A different compiler version is recommended. | |
84 | #endif | |
85 | ||
aae5f662 | 86 | static int kernel_init(void *); |
1da177e4 LT |
87 | |
88 | extern void init_IRQ(void); | |
1da177e4 LT |
89 | extern void fork_init(unsigned long); |
90 | extern void mca_init(void); | |
91 | extern void sbus_init(void); | |
1da177e4 LT |
92 | extern void prio_tree_init(void); |
93 | extern void radix_tree_init(void); | |
94 | extern void free_initmem(void); | |
1da177e4 LT |
95 | #ifdef CONFIG_ACPI |
96 | extern void acpi_early_init(void); | |
97 | #else | |
98 | static inline void acpi_early_init(void) { } | |
99 | #endif | |
37b73c82 AV |
100 | #ifndef CONFIG_DEBUG_RODATA |
101 | static inline void mark_rodata_ro(void) { } | |
102 | #endif | |
1da177e4 LT |
103 | |
104 | #ifdef CONFIG_TC | |
105 | extern void tc_init(void); | |
106 | #endif | |
107 | ||
108 | enum system_states system_state; | |
109 | EXPORT_SYMBOL(system_state); | |
110 | ||
111 | /* | |
112 | * Boot command-line arguments | |
113 | */ | |
114 | #define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT | |
115 | #define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT | |
116 | ||
117 | extern void time_init(void); | |
118 | /* Default late time init is NULL. archs can override this later. */ | |
119 | void (*late_time_init)(void); | |
120 | extern void softirq_init(void); | |
121 | ||
30d7e0d4 ABL |
122 | /* Untouched command line saved by arch-specific code. */ |
123 | char __initdata boot_command_line[COMMAND_LINE_SIZE]; | |
124 | /* Untouched saved command line (eg. for /proc) */ | |
125 | char *saved_command_line; | |
126 | /* Command line for parameter parsing */ | |
127 | static char *static_command_line; | |
1da177e4 LT |
128 | |
129 | static char *execute_command; | |
ffdfc409 | 130 | static char *ramdisk_execute_command; |
1da177e4 | 131 | |
8b3b2955 | 132 | #ifdef CONFIG_SMP |
1da177e4 | 133 | /* Setup configured maximum number of CPUs to activate */ |
ca74a6f8 | 134 | unsigned int __initdata setup_max_cpus = NR_CPUS; |
7e96287d | 135 | |
1da177e4 LT |
136 | /* |
137 | * Setup routine for controlling SMP activation | |
138 | * | |
139 | * Command-line option of "nosmp" or "maxcpus=0" will disable SMP | |
140 | * activation entirely (the MPS table probe still happens, though). | |
141 | * | |
142 | * Command-line option of "maxcpus=<NUM>", where <NUM> is an integer | |
143 | * greater than 0, limits the maximum number of CPUs activated in | |
144 | * SMP mode to <NUM>. | |
145 | */ | |
61ec7567 LB |
146 | #ifndef CONFIG_X86_IO_APIC |
147 | static inline void disable_ioapic_setup(void) {}; | |
148 | #endif | |
149 | ||
1da177e4 LT |
150 | static int __init nosmp(char *str) |
151 | { | |
ca74a6f8 | 152 | setup_max_cpus = 0; |
61ec7567 | 153 | disable_ioapic_setup(); |
8b3b2955 | 154 | return 0; |
1da177e4 LT |
155 | } |
156 | ||
8b3b2955 | 157 | early_param("nosmp", nosmp); |
1da177e4 LT |
158 | |
159 | static int __init maxcpus(char *str) | |
160 | { | |
ca74a6f8 AK |
161 | get_option(&str, &setup_max_cpus); |
162 | if (setup_max_cpus == 0) | |
61ec7567 LB |
163 | disable_ioapic_setup(); |
164 | ||
165 | return 0; | |
1da177e4 LT |
166 | } |
167 | ||
81340977 | 168 | early_param("maxcpus", maxcpus); |
8b3b2955 | 169 | #else |
ca74a6f8 | 170 | #define setup_max_cpus NR_CPUS |
8b3b2955 JB |
171 | #endif |
172 | ||
173 | /* | |
174 | * If set, this is an indication to the drivers that reset the underlying | |
175 | * device before going ahead with the initialization otherwise driver might | |
176 | * rely on the BIOS and skip the reset operation. | |
177 | * | |
178 | * This is useful if kernel is booting in an unreliable environment. | |
179 | * For ex. kdump situaiton where previous kernel has crashed, BIOS has been | |
180 | * skipped and devices will be in unknown state. | |
181 | */ | |
182 | unsigned int reset_devices; | |
183 | EXPORT_SYMBOL(reset_devices); | |
1da177e4 | 184 | |
7e96287d VG |
185 | static int __init set_reset_devices(char *str) |
186 | { | |
187 | reset_devices = 1; | |
188 | return 1; | |
189 | } | |
190 | ||
191 | __setup("reset_devices", set_reset_devices); | |
192 | ||
1da177e4 LT |
193 | static char * argv_init[MAX_INIT_ARGS+2] = { "init", NULL, }; |
194 | char * envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, }; | |
195 | static const char *panic_later, *panic_param; | |
196 | ||
197 | extern struct obs_kernel_param __setup_start[], __setup_end[]; | |
198 | ||
199 | static int __init obsolete_checksetup(char *line) | |
200 | { | |
201 | struct obs_kernel_param *p; | |
33df0d19 | 202 | int had_early_param = 0; |
1da177e4 LT |
203 | |
204 | p = __setup_start; | |
205 | do { | |
206 | int n = strlen(p->str); | |
207 | if (!strncmp(line, p->str, n)) { | |
208 | if (p->early) { | |
33df0d19 RR |
209 | /* Already done in parse_early_param? |
210 | * (Needs exact match on param part). | |
211 | * Keep iterating, as we can have early | |
212 | * params and __setups of same names 8( */ | |
1da177e4 | 213 | if (line[n] == '\0' || line[n] == '=') |
33df0d19 | 214 | had_early_param = 1; |
1da177e4 LT |
215 | } else if (!p->setup_func) { |
216 | printk(KERN_WARNING "Parameter %s is obsolete," | |
217 | " ignored\n", p->str); | |
218 | return 1; | |
219 | } else if (p->setup_func(line + n)) | |
220 | return 1; | |
221 | } | |
222 | p++; | |
223 | } while (p < __setup_end); | |
33df0d19 RR |
224 | |
225 | return had_early_param; | |
1da177e4 LT |
226 | } |
227 | ||
228 | /* | |
229 | * This should be approx 2 Bo*oMips to start (note initial shift), and will | |
230 | * still work even if initially too large, it will just take slightly longer | |
231 | */ | |
232 | unsigned long loops_per_jiffy = (1<<12); | |
233 | ||
234 | EXPORT_SYMBOL(loops_per_jiffy); | |
235 | ||
236 | static int __init debug_kernel(char *str) | |
237 | { | |
1da177e4 | 238 | console_loglevel = 10; |
f6f21c81 | 239 | return 0; |
1da177e4 LT |
240 | } |
241 | ||
242 | static int __init quiet_kernel(char *str) | |
243 | { | |
1da177e4 | 244 | console_loglevel = 4; |
f6f21c81 | 245 | return 0; |
1da177e4 LT |
246 | } |
247 | ||
f6f21c81 YL |
248 | early_param("debug", debug_kernel); |
249 | early_param("quiet", quiet_kernel); | |
1da177e4 LT |
250 | |
251 | static int __init loglevel(char *str) | |
252 | { | |
253 | get_option(&str, &console_loglevel); | |
d9d4fcfe | 254 | return 0; |
1da177e4 LT |
255 | } |
256 | ||
f6f21c81 | 257 | early_param("loglevel", loglevel); |
1da177e4 LT |
258 | |
259 | /* | |
260 | * Unknown boot options get handed to init, unless they look like | |
261 | * failed parameters | |
262 | */ | |
263 | static int __init unknown_bootoption(char *param, char *val) | |
264 | { | |
265 | /* Change NUL term back to "=", to make "param" the whole string. */ | |
266 | if (val) { | |
267 | /* param=val or param="val"? */ | |
268 | if (val == param+strlen(param)+1) | |
269 | val[-1] = '='; | |
270 | else if (val == param+strlen(param)+2) { | |
271 | val[-2] = '='; | |
272 | memmove(val-1, val, strlen(val)+1); | |
273 | val--; | |
274 | } else | |
275 | BUG(); | |
276 | } | |
277 | ||
278 | /* Handle obsolete-style parameters */ | |
279 | if (obsolete_checksetup(param)) | |
280 | return 0; | |
281 | ||
282 | /* | |
211fee8a | 283 | * Preemptive maintenance for "why didn't my misspelled command |
1da177e4 LT |
284 | * line work?" |
285 | */ | |
286 | if (strchr(param, '.') && (!val || strchr(param, '.') < val)) { | |
287 | printk(KERN_ERR "Unknown boot option `%s': ignoring\n", param); | |
288 | return 0; | |
289 | } | |
290 | ||
291 | if (panic_later) | |
292 | return 0; | |
293 | ||
294 | if (val) { | |
295 | /* Environment option */ | |
296 | unsigned int i; | |
297 | for (i = 0; envp_init[i]; i++) { | |
298 | if (i == MAX_INIT_ENVS) { | |
299 | panic_later = "Too many boot env vars at `%s'"; | |
300 | panic_param = param; | |
301 | } | |
302 | if (!strncmp(param, envp_init[i], val - param)) | |
303 | break; | |
304 | } | |
305 | envp_init[i] = param; | |
306 | } else { | |
307 | /* Command line option */ | |
308 | unsigned int i; | |
309 | for (i = 0; argv_init[i]; i++) { | |
310 | if (i == MAX_INIT_ARGS) { | |
311 | panic_later = "Too many boot init vars at `%s'"; | |
312 | panic_param = param; | |
313 | } | |
314 | } | |
315 | argv_init[i] = param; | |
316 | } | |
317 | return 0; | |
318 | } | |
319 | ||
12d6f21e IM |
320 | #ifdef CONFIG_DEBUG_PAGEALLOC |
321 | int __read_mostly debug_pagealloc_enabled = 0; | |
322 | #endif | |
323 | ||
1da177e4 LT |
324 | static int __init init_setup(char *str) |
325 | { | |
326 | unsigned int i; | |
327 | ||
328 | execute_command = str; | |
329 | /* | |
330 | * In case LILO is going to boot us with default command line, | |
331 | * it prepends "auto" before the whole cmdline which makes | |
332 | * the shell think it should execute a script with such name. | |
333 | * So we ignore all arguments entered _before_ init=... [MJ] | |
334 | */ | |
335 | for (i = 1; i < MAX_INIT_ARGS; i++) | |
336 | argv_init[i] = NULL; | |
337 | return 1; | |
338 | } | |
339 | __setup("init=", init_setup); | |
340 | ||
ffdfc409 OJ |
341 | static int __init rdinit_setup(char *str) |
342 | { | |
343 | unsigned int i; | |
344 | ||
345 | ramdisk_execute_command = str; | |
346 | /* See "auto" comment in init_setup */ | |
347 | for (i = 1; i < MAX_INIT_ARGS; i++) | |
348 | argv_init[i] = NULL; | |
349 | return 1; | |
350 | } | |
351 | __setup("rdinit=", rdinit_setup); | |
352 | ||
1da177e4 LT |
353 | #ifndef CONFIG_SMP |
354 | ||
355 | #ifdef CONFIG_X86_LOCAL_APIC | |
356 | static void __init smp_init(void) | |
357 | { | |
358 | APIC_init_uniprocessor(); | |
359 | } | |
360 | #else | |
361 | #define smp_init() do { } while (0) | |
362 | #endif | |
363 | ||
364 | static inline void setup_per_cpu_areas(void) { } | |
e0982e90 | 365 | static inline void setup_nr_cpu_ids(void) { } |
1da177e4 LT |
366 | static inline void smp_prepare_cpus(unsigned int maxcpus) { } |
367 | ||
368 | #else | |
369 | ||
321a8e9d MT |
370 | #if NR_CPUS > BITS_PER_LONG |
371 | cpumask_t cpu_mask_all __read_mostly = CPU_MASK_ALL; | |
372 | EXPORT_SYMBOL(cpu_mask_all); | |
373 | #endif | |
374 | ||
e0982e90 MT |
375 | /* Setup number of possible processor ids */ |
376 | int nr_cpu_ids __read_mostly = NR_CPUS; | |
377 | EXPORT_SYMBOL(nr_cpu_ids); | |
378 | ||
379 | /* An arch may set nr_cpu_ids earlier if needed, so this would be redundant */ | |
380 | static void __init setup_nr_cpu_ids(void) | |
381 | { | |
382 | int cpu, highest_cpu = 0; | |
383 | ||
384 | for_each_possible_cpu(cpu) | |
385 | highest_cpu = cpu; | |
386 | ||
387 | nr_cpu_ids = highest_cpu + 1; | |
388 | } | |
389 | ||
dd5af90a | 390 | #ifndef CONFIG_HAVE_SETUP_PER_CPU_AREA |
b73b459f | 391 | unsigned long __per_cpu_offset[NR_CPUS] __read_mostly; |
1da177e4 LT |
392 | |
393 | EXPORT_SYMBOL(__per_cpu_offset); | |
394 | ||
395 | static void __init setup_per_cpu_areas(void) | |
396 | { | |
397 | unsigned long size, i; | |
398 | char *ptr; | |
63872f87 | 399 | unsigned long nr_possible_cpus = num_possible_cpus(); |
1da177e4 LT |
400 | |
401 | /* Copy section for each CPU (we discard the original) */ | |
b6e3590f JF |
402 | size = ALIGN(PERCPU_ENOUGH_ROOM, PAGE_SIZE); |
403 | ptr = alloc_bootmem_pages(size * nr_possible_cpus); | |
1da177e4 | 404 | |
0a945022 | 405 | for_each_possible_cpu(i) { |
1da177e4 LT |
406 | __per_cpu_offset[i] = ptr - __per_cpu_start; |
407 | memcpy(ptr, __per_cpu_start, __per_cpu_end - __per_cpu_start); | |
63872f87 | 408 | ptr += size; |
1da177e4 LT |
409 | } |
410 | } | |
dd5af90a | 411 | #endif /* CONFIG_HAVE_SETUP_PER_CPU_AREA */ |
1da177e4 LT |
412 | |
413 | /* Called by boot processor to activate the rest. */ | |
414 | static void __init smp_init(void) | |
415 | { | |
53b8a315 | 416 | unsigned int cpu; |
1da177e4 | 417 | |
e761b772 MK |
418 | /* |
419 | * Set up the current CPU as possible to migrate to. | |
420 | * The other ones will be done by cpu_up/cpu_down() | |
421 | */ | |
422 | cpu = smp_processor_id(); | |
423 | cpu_set(cpu, cpu_active_map); | |
424 | ||
1da177e4 | 425 | /* FIXME: This should be done in userspace --RR */ |
53b8a315 | 426 | for_each_present_cpu(cpu) { |
ca74a6f8 | 427 | if (num_online_cpus() >= setup_max_cpus) |
1da177e4 | 428 | break; |
53b8a315 CL |
429 | if (!cpu_online(cpu)) |
430 | cpu_up(cpu); | |
1da177e4 LT |
431 | } |
432 | ||
433 | /* Any cleanup work */ | |
434 | printk(KERN_INFO "Brought up %ld CPUs\n", (long)num_online_cpus()); | |
ca74a6f8 | 435 | smp_cpus_done(setup_max_cpus); |
1da177e4 LT |
436 | } |
437 | ||
438 | #endif | |
439 | ||
30d7e0d4 ABL |
440 | /* |
441 | * We need to store the untouched command line for future reference. | |
442 | * We also need to store the touched command line since the parameter | |
443 | * parsing is performed in place, and we should allow a component to | |
444 | * store reference of name/value for future reference. | |
445 | */ | |
446 | static void __init setup_command_line(char *command_line) | |
447 | { | |
448 | saved_command_line = alloc_bootmem(strlen (boot_command_line)+1); | |
449 | static_command_line = alloc_bootmem(strlen (command_line)+1); | |
450 | strcpy (saved_command_line, boot_command_line); | |
451 | strcpy (static_command_line, command_line); | |
452 | } | |
453 | ||
1da177e4 LT |
454 | /* |
455 | * We need to finalize in a non-__init function or else race conditions | |
456 | * between the root thread and the init thread may cause start_kernel to | |
457 | * be reaped by free_initmem before the root thread has proceeded to | |
458 | * cpu_idle. | |
459 | * | |
460 | * gcc-3.4 accidentally inlines this function, so use noinline. | |
461 | */ | |
462 | ||
92080309 | 463 | static void noinline __init_refok rest_init(void) |
1da177e4 LT |
464 | __releases(kernel_lock) |
465 | { | |
73c27992 EB |
466 | int pid; |
467 | ||
aae5f662 | 468 | kernel_thread(kernel_init, NULL, CLONE_FS | CLONE_SIGHAND); |
1da177e4 | 469 | numa_default_policy(); |
73c27992 | 470 | pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES); |
5cd20455 | 471 | kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns); |
1da177e4 | 472 | unlock_kernel(); |
f340c0d1 IM |
473 | |
474 | /* | |
475 | * The boot idle thread must execute schedule() | |
1df21055 | 476 | * at least once to get things moving: |
f340c0d1 | 477 | */ |
1df21055 | 478 | init_idle_bootup_task(current); |
5bfb5d69 | 479 | preempt_enable_no_resched(); |
f340c0d1 | 480 | schedule(); |
5bfb5d69 | 481 | preempt_disable(); |
f340c0d1 | 482 | |
5bfb5d69 | 483 | /* Call into cpu_idle with preempt disabled */ |
1da177e4 | 484 | cpu_idle(); |
1df21055 | 485 | } |
1da177e4 LT |
486 | |
487 | /* Check for early params. */ | |
488 | static int __init do_early_param(char *param, char *val) | |
489 | { | |
490 | struct obs_kernel_param *p; | |
491 | ||
492 | for (p = __setup_start; p < __setup_end; p++) { | |
18a8bd94 YL |
493 | if ((p->early && strcmp(param, p->str) == 0) || |
494 | (strcmp(param, "console") == 0 && | |
495 | strcmp(p->str, "earlycon") == 0) | |
496 | ) { | |
1da177e4 LT |
497 | if (p->setup_func(val) != 0) |
498 | printk(KERN_WARNING | |
499 | "Malformed early option '%s'\n", param); | |
500 | } | |
501 | } | |
502 | /* We accept everything at this stage. */ | |
503 | return 0; | |
504 | } | |
505 | ||
506 | /* Arch code calls this early on, or if not, just before other parsing. */ | |
507 | void __init parse_early_param(void) | |
508 | { | |
509 | static __initdata int done = 0; | |
510 | static __initdata char tmp_cmdline[COMMAND_LINE_SIZE]; | |
511 | ||
512 | if (done) | |
513 | return; | |
514 | ||
515 | /* All fall through to do_early_param. */ | |
30d7e0d4 | 516 | strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE); |
1da177e4 LT |
517 | parse_args("early options", tmp_cmdline, NULL, 0, do_early_param); |
518 | done = 1; | |
519 | } | |
520 | ||
521 | /* | |
522 | * Activate the first processor. | |
523 | */ | |
524 | ||
44fd2299 SS |
525 | static void __init boot_cpu_init(void) |
526 | { | |
527 | int cpu = smp_processor_id(); | |
528 | /* Mark the boot cpu "present", "online" etc for SMP and UP case */ | |
529 | cpu_set(cpu, cpu_online_map); | |
530 | cpu_set(cpu, cpu_present_map); | |
531 | cpu_set(cpu, cpu_possible_map); | |
532 | } | |
533 | ||
839ad62e | 534 | void __init __weak smp_setup_processor_id(void) |
033ab7f8 AM |
535 | { |
536 | } | |
537 | ||
8c9843e5 BH |
538 | void __init __weak thread_info_cache_init(void) |
539 | { | |
540 | } | |
541 | ||
1da177e4 LT |
542 | asmlinkage void __init start_kernel(void) |
543 | { | |
544 | char * command_line; | |
545 | extern struct kernel_param __start___param[], __stop___param[]; | |
033ab7f8 AM |
546 | |
547 | smp_setup_processor_id(); | |
548 | ||
fbb9ce95 IM |
549 | /* |
550 | * Need to run as early as possible, to initialize the | |
551 | * lockdep hash: | |
552 | */ | |
c9538ed4 | 553 | unwind_init(); |
fbb9ce95 | 554 | lockdep_init(); |
3ac7fe5a | 555 | debug_objects_early_init(); |
ddbcc7e8 | 556 | cgroup_init_early(); |
fbb9ce95 IM |
557 | |
558 | local_irq_disable(); | |
559 | early_boot_irqs_off(); | |
243c7621 | 560 | early_init_irq_lock_class(); |
fbb9ce95 | 561 | |
1da177e4 LT |
562 | /* |
563 | * Interrupts are still disabled. Do necessary setups, then | |
564 | * enable them | |
565 | */ | |
566 | lock_kernel(); | |
906568c9 | 567 | tick_init(); |
44fd2299 | 568 | boot_cpu_init(); |
1da177e4 LT |
569 | page_address_init(); |
570 | printk(KERN_NOTICE); | |
8993780a | 571 | printk(linux_banner); |
1da177e4 | 572 | setup_arch(&command_line); |
cf475ad2 | 573 | mm_init_owner(&init_mm, &init_task); |
30d7e0d4 | 574 | setup_command_line(command_line); |
690a973f | 575 | unwind_setup(); |
1da177e4 | 576 | setup_per_cpu_areas(); |
e0982e90 | 577 | setup_nr_cpu_ids(); |
44fd2299 | 578 | smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */ |
1da177e4 LT |
579 | |
580 | /* | |
581 | * Set up the scheduler prior starting any interrupts (such as the | |
582 | * timer interrupt). Full topology setup happens at smp_init() | |
583 | * time - but meanwhile we still have a functioning scheduler. | |
584 | */ | |
585 | sched_init(); | |
586 | /* | |
587 | * Disable preemption - early bootup scheduling is extremely | |
588 | * fragile until we cpu_idle() for the first time. | |
589 | */ | |
590 | preempt_disable(); | |
591 | build_all_zonelists(); | |
592 | page_alloc_init(); | |
30d7e0d4 | 593 | printk(KERN_NOTICE "Kernel command line: %s\n", boot_command_line); |
1da177e4 | 594 | parse_early_param(); |
30d7e0d4 | 595 | parse_args("Booting kernel", static_command_line, __start___param, |
1da177e4 LT |
596 | __stop___param - __start___param, |
597 | &unknown_bootoption); | |
c4a68306 AB |
598 | if (!irqs_disabled()) { |
599 | printk(KERN_WARNING "start_kernel(): bug: interrupts were " | |
600 | "enabled *very* early, fixing it\n"); | |
601 | local_irq_disable(); | |
602 | } | |
1da177e4 LT |
603 | sort_main_extable(); |
604 | trap_init(); | |
605 | rcu_init(); | |
606 | init_IRQ(); | |
607 | pidhash_init(); | |
608 | init_timers(); | |
c0a31329 | 609 | hrtimers_init(); |
1da177e4 | 610 | softirq_init(); |
ad596171 | 611 | timekeeping_init(); |
88fecaa2 | 612 | time_init(); |
3e51f33f | 613 | sched_clock_init(); |
93e02814 HC |
614 | profile_init(); |
615 | if (!irqs_disabled()) | |
616 | printk("start_kernel(): bug: interrupts were enabled early\n"); | |
fbb9ce95 | 617 | early_boot_irqs_on(); |
93e02814 | 618 | local_irq_enable(); |
1da177e4 LT |
619 | |
620 | /* | |
621 | * HACK ALERT! This is early. We're enabling the console before | |
622 | * we've done PCI setups etc, and console_init() must be aware of | |
623 | * this. But we do want output early, in case something goes wrong. | |
624 | */ | |
625 | console_init(); | |
626 | if (panic_later) | |
627 | panic(panic_later, panic_param); | |
fbb9ce95 IM |
628 | |
629 | lockdep_info(); | |
630 | ||
9a11b49a IM |
631 | /* |
632 | * Need to run this when irqs are enabled, because it wants | |
633 | * to self-test [hard/soft]-irqs on/off lock inversion bugs | |
634 | * too: | |
635 | */ | |
636 | locking_selftest(); | |
637 | ||
1da177e4 LT |
638 | #ifdef CONFIG_BLK_DEV_INITRD |
639 | if (initrd_start && !initrd_below_start_ok && | |
bd673c7c | 640 | page_to_pfn(virt_to_page((void *)initrd_start)) < min_low_pfn) { |
1da177e4 | 641 | printk(KERN_CRIT "initrd overwritten (0x%08lx < 0x%08lx) - " |
fb6624eb | 642 | "disabling it.\n", |
bd673c7c GU |
643 | page_to_pfn(virt_to_page((void *)initrd_start)), |
644 | min_low_pfn); | |
1da177e4 LT |
645 | initrd_start = 0; |
646 | } | |
647 | #endif | |
db64fe02 | 648 | vmalloc_init(); |
1da177e4 | 649 | vfs_caches_init_early(); |
c417f024 | 650 | cpuset_init_early(); |
94b6da5a | 651 | page_cgroup_init(); |
1da177e4 | 652 | mem_init(); |
a03c2a48 | 653 | enable_debug_pagealloc(); |
d221938c | 654 | cpu_hotplug_init(); |
1da177e4 | 655 | kmem_cache_init(); |
3ac7fe5a | 656 | debug_objects_mem_init(); |
199f0ca5 | 657 | idr_init_cache(); |
e7c8d5c9 | 658 | setup_per_cpu_pageset(); |
1da177e4 LT |
659 | numa_policy_init(); |
660 | if (late_time_init) | |
661 | late_time_init(); | |
662 | calibrate_delay(); | |
663 | pidmap_init(); | |
664 | pgtable_cache_init(); | |
665 | prio_tree_init(); | |
666 | anon_vma_init(); | |
667 | #ifdef CONFIG_X86 | |
668 | if (efi_enabled) | |
669 | efi_enter_virtual_mode(); | |
670 | #endif | |
8c9843e5 | 671 | thread_info_cache_init(); |
1da177e4 LT |
672 | fork_init(num_physpages); |
673 | proc_caches_init(); | |
674 | buffer_init(); | |
675 | unnamed_dev_init(); | |
676 | key_init(); | |
677 | security_init(); | |
678 | vfs_caches_init(num_physpages); | |
679 | radix_tree_init(); | |
680 | signals_init(); | |
681 | /* rootfs populating might need page-writeback */ | |
682 | page_writeback_init(); | |
683 | #ifdef CONFIG_PROC_FS | |
684 | proc_root_init(); | |
685 | #endif | |
ddbcc7e8 | 686 | cgroup_init(); |
1da177e4 | 687 | cpuset_init(); |
c757249a | 688 | taskstats_init_early(); |
ca74e92b | 689 | delayacct_init(); |
1da177e4 LT |
690 | |
691 | check_bugs(); | |
692 | ||
693 | acpi_early_init(); /* before LAPIC and SMP init */ | |
694 | ||
68bf21aa SR |
695 | ftrace_init(); |
696 | ||
1da177e4 LT |
697 | /* Do the rest non-__init'ed, we're now alive */ |
698 | rest_init(); | |
699 | } | |
700 | ||
59f9415f | 701 | static int initcall_debug; |
d0ea3d7d | 702 | core_param(initcall_debug, initcall_debug, bool, 0644); |
1da177e4 | 703 | |
59f9415f | 704 | int do_one_initcall(initcall_t fn) |
1da177e4 | 705 | { |
1da177e4 | 706 | int count = preempt_count(); |
cb5ab742 | 707 | ktime_t delta; |
a76bfd0d | 708 | char msgbuf[64]; |
3bf77af6 | 709 | struct boot_trace it; |
1da177e4 | 710 | |
e0df154f | 711 | if (initcall_debug) { |
3bf77af6 | 712 | it.caller = task_pid_nr(current); |
3bf77af6 | 713 | printk("calling %pF @ %i\n", fn, it.caller); |
cb5ab742 | 714 | it.calltime = ktime_get(); |
e0df154f | 715 | } |
1da177e4 | 716 | |
cb5ab742 | 717 | it.result = fn(); |
1da177e4 | 718 | |
e0df154f | 719 | if (initcall_debug) { |
cb5ab742 FW |
720 | it.rettime = ktime_get(); |
721 | delta = ktime_sub(it.rettime, it.calltime); | |
ca538f6b TB |
722 | it.duration = (unsigned long long) delta.tv64 >> 10; |
723 | printk("initcall %pF returned %d after %Ld usecs\n", fn, | |
cb5ab742 | 724 | it.result, it.duration); |
5601020f | 725 | trace_boot(&it, fn); |
e0df154f | 726 | } |
8f0c45cd | 727 | |
e0df154f | 728 | msgbuf[0] = 0; |
e662e1cf | 729 | |
cb5ab742 FW |
730 | if (it.result && it.result != -ENODEV && initcall_debug) |
731 | sprintf(msgbuf, "error code %d ", it.result); | |
e662e1cf | 732 | |
e0df154f | 733 | if (preempt_count() != count) { |
a76bfd0d | 734 | strlcat(msgbuf, "preemption imbalance ", sizeof(msgbuf)); |
e0df154f | 735 | preempt_count() = count; |
1da177e4 | 736 | } |
e0df154f | 737 | if (irqs_disabled()) { |
a76bfd0d | 738 | strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf)); |
e0df154f LT |
739 | local_irq_enable(); |
740 | } | |
741 | if (msgbuf[0]) { | |
96d746c6 | 742 | printk("initcall %pF returned with %s\n", fn, msgbuf); |
e0df154f | 743 | } |
59f9415f | 744 | |
cb5ab742 | 745 | return it.result; |
e0df154f LT |
746 | } |
747 | ||
748 | ||
c2147a50 | 749 | extern initcall_t __initcall_start[], __initcall_end[], __early_initcall_end[]; |
e0df154f LT |
750 | |
751 | static void __init do_initcalls(void) | |
752 | { | |
753 | initcall_t *call; | |
754 | ||
c2147a50 | 755 | for (call = __early_initcall_end; call < __initcall_end; call++) |
e0df154f | 756 | do_one_initcall(*call); |
1da177e4 LT |
757 | |
758 | /* Make sure there is no pending stuff from the initcall sequence */ | |
759 | flush_scheduled_work(); | |
760 | } | |
761 | ||
762 | /* | |
763 | * Ok, the machine is now initialized. None of the devices | |
764 | * have been touched yet, but the CPU subsystem is up and | |
765 | * running, and memory and process management works. | |
766 | * | |
767 | * Now we can finally start doing some real work.. | |
768 | */ | |
769 | static void __init do_basic_setup(void) | |
770 | { | |
4446a36f | 771 | rcu_init_sched(); /* needed by module_init stage. */ |
1da177e4 LT |
772 | usermodehelper_init(); |
773 | driver_init(); | |
b04c3afb | 774 | init_irq_proc(); |
1da177e4 LT |
775 | do_initcalls(); |
776 | } | |
777 | ||
7babe8db | 778 | static void __init do_pre_smp_initcalls(void) |
c2147a50 EGM |
779 | { |
780 | initcall_t *call; | |
781 | ||
782 | for (call = __initcall_start; call < __early_initcall_end; call++) | |
783 | do_one_initcall(*call); | |
784 | } | |
785 | ||
1da177e4 LT |
786 | static void run_init_process(char *init_filename) |
787 | { | |
788 | argv_init[0] = init_filename; | |
67608567 | 789 | kernel_execve(init_filename, argv_init, envp_init); |
1da177e4 LT |
790 | } |
791 | ||
ee5bfa64 VG |
792 | /* This is a non __init function. Force it to be noinline otherwise gcc |
793 | * makes it inline to init() and it becomes part of init.text section | |
794 | */ | |
795 | static int noinline init_post(void) | |
796 | { | |
797 | free_initmem(); | |
798 | unlock_kernel(); | |
799 | mark_rodata_ro(); | |
800 | system_state = SYSTEM_RUNNING; | |
801 | numa_default_policy(); | |
802 | ||
803 | if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0) | |
804 | printk(KERN_WARNING "Warning: unable to open an initial console.\n"); | |
805 | ||
806 | (void) sys_dup(0); | |
807 | (void) sys_dup(0); | |
808 | ||
fae5fa44 ON |
809 | current->signal->flags |= SIGNAL_UNKILLABLE; |
810 | ||
ee5bfa64 VG |
811 | if (ramdisk_execute_command) { |
812 | run_init_process(ramdisk_execute_command); | |
813 | printk(KERN_WARNING "Failed to execute %s\n", | |
814 | ramdisk_execute_command); | |
815 | } | |
816 | ||
817 | /* | |
818 | * We try each of these until one succeeds. | |
819 | * | |
820 | * The Bourne shell can be used instead of init if we are | |
821 | * trying to recover a really broken machine. | |
822 | */ | |
823 | if (execute_command) { | |
824 | run_init_process(execute_command); | |
825 | printk(KERN_WARNING "Failed to execute %s. Attempting " | |
826 | "defaults...\n", execute_command); | |
827 | } | |
828 | run_init_process("/sbin/init"); | |
829 | run_init_process("/etc/init"); | |
830 | run_init_process("/bin/init"); | |
831 | run_init_process("/bin/sh"); | |
832 | ||
833 | panic("No init found. Try passing init= option to kernel."); | |
834 | } | |
835 | ||
aae5f662 | 836 | static int __init kernel_init(void * unused) |
1da177e4 LT |
837 | { |
838 | lock_kernel(); | |
839 | /* | |
840 | * init can run on any cpu. | |
841 | */ | |
321a8e9d | 842 | set_cpus_allowed_ptr(current, CPU_MASK_ALL_PTR); |
1da177e4 LT |
843 | /* |
844 | * Tell the world that we're going to be the grim | |
845 | * reaper of innocent orphaned children. | |
846 | * | |
847 | * We don't want people to have to make incorrect | |
848 | * assumptions about where in the task array this | |
849 | * can be found. | |
850 | */ | |
84d73786 | 851 | init_pid_ns.child_reaper = current; |
1da177e4 | 852 | |
9ec52099 CLG |
853 | cad_pid = task_pid(current); |
854 | ||
a802dd0e HC |
855 | init_workqueues(); |
856 | ||
ca74a6f8 | 857 | smp_prepare_cpus(setup_max_cpus); |
1da177e4 LT |
858 | |
859 | do_pre_smp_initcalls(); | |
3bf77af6 | 860 | start_boot_trace(); |
1da177e4 | 861 | |
1da177e4 LT |
862 | smp_init(); |
863 | sched_init_smp(); | |
864 | ||
865 | cpuset_init_smp(); | |
866 | ||
1da177e4 LT |
867 | do_basic_setup(); |
868 | ||
869 | /* | |
870 | * check if there is an early userspace init. If yes, let it do all | |
871 | * the work | |
872 | */ | |
ffdfc409 OJ |
873 | |
874 | if (!ramdisk_execute_command) | |
875 | ramdisk_execute_command = "/init"; | |
876 | ||
877 | if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) { | |
878 | ramdisk_execute_command = NULL; | |
1da177e4 | 879 | prepare_namespace(); |
ffdfc409 | 880 | } |
1da177e4 LT |
881 | |
882 | /* | |
883 | * Ok, we have completed the initial bootup, and | |
884 | * we're essentially up and running. Get rid of the | |
885 | * initmem segments and start the user-mode stuff.. | |
886 | */ | |
097d036a | 887 | stop_boot_trace(); |
ee5bfa64 VG |
888 | init_post(); |
889 | return 0; | |
1da177e4 | 890 | } |