]>
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 | |
dd4d9fec | 9 | * Simplified starting of init: Michael A. Griffith <grif@acm.org> |
1da177e4 LT |
10 | */ |
11 | ||
ea676e84 AM |
12 | #define DEBUG /* Enable initcall_debug */ |
13 | ||
1da177e4 | 14 | #include <linux/types.h> |
8a293be0 | 15 | #include <linux/extable.h> |
1da177e4 LT |
16 | #include <linux/module.h> |
17 | #include <linux/proc_fs.h> | |
5c2c5c55 | 18 | #include <linux/binfmts.h> |
1da177e4 LT |
19 | #include <linux/kernel.h> |
20 | #include <linux/syscalls.h> | |
9b5609fd | 21 | #include <linux/stackprotector.h> |
1da177e4 LT |
22 | #include <linux/string.h> |
23 | #include <linux/ctype.h> | |
24 | #include <linux/delay.h> | |
1da177e4 LT |
25 | #include <linux/ioport.h> |
26 | #include <linux/init.h> | |
1da177e4 | 27 | #include <linux/initrd.h> |
1da177e4 | 28 | #include <linux/bootmem.h> |
4a7a16dc | 29 | #include <linux/acpi.h> |
0c688614 | 30 | #include <linux/console.h> |
38b8d208 | 31 | #include <linux/nmi.h> |
1da177e4 LT |
32 | #include <linux/percpu.h> |
33 | #include <linux/kmod.h> | |
db64fe02 | 34 | #include <linux/vmalloc.h> |
1da177e4 | 35 | #include <linux/kernel_stat.h> |
d7cd5611 | 36 | #include <linux/start_kernel.h> |
1da177e4 | 37 | #include <linux/security.h> |
3d442233 | 38 | #include <linux/smp.h> |
1da177e4 LT |
39 | #include <linux/profile.h> |
40 | #include <linux/rcupdate.h> | |
41 | #include <linux/moduleparam.h> | |
42 | #include <linux/kallsyms.h> | |
43 | #include <linux/writeback.h> | |
44 | #include <linux/cpu.h> | |
45 | #include <linux/cpuset.h> | |
ddbcc7e8 | 46 | #include <linux/cgroup.h> |
1da177e4 | 47 | #include <linux/efi.h> |
906568c9 | 48 | #include <linux/tick.h> |
78634061 | 49 | #include <linux/sched/isolation.h> |
6168a702 | 50 | #include <linux/interrupt.h> |
c757249a | 51 | #include <linux/taskstats_kern.h> |
ca74e92b | 52 | #include <linux/delayacct.h> |
1da177e4 LT |
53 | #include <linux/unistd.h> |
54 | #include <linux/rmap.h> | |
55 | #include <linux/mempolicy.h> | |
56 | #include <linux/key.h> | |
b6cd0b77 | 57 | #include <linux/buffer_head.h> |
eefa864b | 58 | #include <linux/page_ext.h> |
9a11b49a | 59 | #include <linux/debug_locks.h> |
3ac7fe5a | 60 | #include <linux/debugobjects.h> |
fbb9ce95 | 61 | #include <linux/lockdep.h> |
3c7b4e6b | 62 | #include <linux/kmemleak.h> |
84d73786 | 63 | #include <linux/pid_namespace.h> |
1f21782e | 64 | #include <linux/device.h> |
73c27992 | 65 | #include <linux/kthread.h> |
e6fe6649 | 66 | #include <linux/sched.h> |
1777e463 | 67 | #include <linux/sched/init.h> |
a1c9eea9 | 68 | #include <linux/signal.h> |
199f0ca5 | 69 | #include <linux/idr.h> |
0b4b3827 | 70 | #include <linux/kgdb.h> |
68bf21aa | 71 | #include <linux/ftrace.h> |
22a9d645 | 72 | #include <linux/async.h> |
6ae6996a | 73 | #include <linux/sfi.h> |
2b2af54a | 74 | #include <linux/shmem_fs.h> |
5a0e3ad6 | 75 | #include <linux/slab.h> |
24a24bb6 | 76 | #include <linux/perf_event.h> |
a74fb73c | 77 | #include <linux/ptrace.h> |
aa8c6248 | 78 | #include <linux/pti.h> |
bb813f4c TH |
79 | #include <linux/blkdev.h> |
80 | #include <linux/elevator.h> | |
38ff87f7 | 81 | #include <linux/sched_clock.h> |
29930025 | 82 | #include <linux/sched/task.h> |
68db0cf1 | 83 | #include <linux/sched/task_stack.h> |
65f382fd | 84 | #include <linux/context_tracking.h> |
47d06e53 | 85 | #include <linux/random.h> |
7b0b73d7 | 86 | #include <linux/list.h> |
c9cd2ce2 | 87 | #include <linux/integrity.h> |
e149ed2b | 88 | #include <linux/proc_ns.h> |
0ddab1d2 | 89 | #include <linux/io.h> |
39290b38 | 90 | #include <linux/cache.h> |
2959a5f7 | 91 | #include <linux/rodata_test.h> |
1da177e4 LT |
92 | |
93 | #include <asm/io.h> | |
94 | #include <asm/bugs.h> | |
95 | #include <asm/setup.h> | |
a940199f | 96 | #include <asm/sections.h> |
37b73c82 | 97 | #include <asm/cacheflush.h> |
1da177e4 | 98 | |
aae5f662 | 99 | static int kernel_init(void *); |
1da177e4 LT |
100 | |
101 | extern void init_IRQ(void); | |
1da177e4 | 102 | extern void radix_tree_init(void); |
1da177e4 | 103 | |
2ce802f6 TH |
104 | /* |
105 | * Debug helper: via this flag we know that we are in 'early bootup code' | |
106 | * where only the boot processor is running with IRQ disabled. This means | |
107 | * two things - IRQ must not be enabled before the flag is cleared and some | |
108 | * operations which are not allowed with IRQ disabled are allowed while the | |
109 | * flag is set. | |
110 | */ | |
111 | bool early_boot_irqs_disabled __read_mostly; | |
112 | ||
a6826048 | 113 | enum system_states system_state __read_mostly; |
1da177e4 LT |
114 | EXPORT_SYMBOL(system_state); |
115 | ||
116 | /* | |
117 | * Boot command-line arguments | |
118 | */ | |
119 | #define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT | |
120 | #define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT | |
121 | ||
122 | extern void time_init(void); | |
123 | /* Default late time init is NULL. archs can override this later. */ | |
d2e3192b | 124 | void (*__initdata late_time_init)(void); |
1da177e4 | 125 | |
30d7e0d4 ABL |
126 | /* Untouched command line saved by arch-specific code. */ |
127 | char __initdata boot_command_line[COMMAND_LINE_SIZE]; | |
128 | /* Untouched saved command line (eg. for /proc) */ | |
129 | char *saved_command_line; | |
130 | /* Command line for parameter parsing */ | |
131 | static char *static_command_line; | |
08746a65 KM |
132 | /* Command line for per-initcall parameter parsing */ |
133 | static char *initcall_command_line; | |
1da177e4 LT |
134 | |
135 | static char *execute_command; | |
ffdfc409 | 136 | static char *ramdisk_execute_command; |
1da177e4 | 137 | |
c4b2c0c5 HFS |
138 | /* |
139 | * Used to generate warnings if static_key manipulation functions are used | |
140 | * before jump_label_init is called. | |
141 | */ | |
dd4d9fec | 142 | bool static_key_initialized __read_mostly; |
c4b2c0c5 HFS |
143 | EXPORT_SYMBOL_GPL(static_key_initialized); |
144 | ||
8b3b2955 JB |
145 | /* |
146 | * If set, this is an indication to the drivers that reset the underlying | |
147 | * device before going ahead with the initialization otherwise driver might | |
148 | * rely on the BIOS and skip the reset operation. | |
149 | * | |
150 | * This is useful if kernel is booting in an unreliable environment. | |
fc454fdc | 151 | * For ex. kdump situation where previous kernel has crashed, BIOS has been |
8b3b2955 JB |
152 | * skipped and devices will be in unknown state. |
153 | */ | |
154 | unsigned int reset_devices; | |
155 | EXPORT_SYMBOL(reset_devices); | |
1da177e4 | 156 | |
7e96287d VG |
157 | static int __init set_reset_devices(char *str) |
158 | { | |
159 | reset_devices = 1; | |
160 | return 1; | |
161 | } | |
162 | ||
163 | __setup("reset_devices", set_reset_devices); | |
164 | ||
dd4d9fec FF |
165 | static const char *argv_init[MAX_INIT_ARGS+2] = { "init", NULL, }; |
166 | const char *envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, }; | |
1da177e4 LT |
167 | static const char *panic_later, *panic_param; |
168 | ||
914dcaa8 | 169 | extern const struct obs_kernel_param __setup_start[], __setup_end[]; |
1da177e4 | 170 | |
31c025b5 | 171 | static bool __init obsolete_checksetup(char *line) |
1da177e4 | 172 | { |
914dcaa8 | 173 | const struct obs_kernel_param *p; |
31c025b5 | 174 | bool had_early_param = false; |
1da177e4 LT |
175 | |
176 | p = __setup_start; | |
177 | do { | |
178 | int n = strlen(p->str); | |
b1e4d20c | 179 | if (parameqn(line, p->str, n)) { |
1da177e4 | 180 | if (p->early) { |
33df0d19 RR |
181 | /* Already done in parse_early_param? |
182 | * (Needs exact match on param part). | |
183 | * Keep iterating, as we can have early | |
184 | * params and __setups of same names 8( */ | |
1da177e4 | 185 | if (line[n] == '\0' || line[n] == '=') |
31c025b5 | 186 | had_early_param = true; |
1da177e4 | 187 | } else if (!p->setup_func) { |
ea676e84 AM |
188 | pr_warn("Parameter %s is obsolete, ignored\n", |
189 | p->str); | |
31c025b5 | 190 | return true; |
1da177e4 | 191 | } else if (p->setup_func(line + n)) |
31c025b5 | 192 | return true; |
1da177e4 LT |
193 | } |
194 | p++; | |
195 | } while (p < __setup_end); | |
33df0d19 RR |
196 | |
197 | return had_early_param; | |
1da177e4 LT |
198 | } |
199 | ||
200 | /* | |
201 | * This should be approx 2 Bo*oMips to start (note initial shift), and will | |
202 | * still work even if initially too large, it will just take slightly longer | |
203 | */ | |
204 | unsigned long loops_per_jiffy = (1<<12); | |
1da177e4 LT |
205 | EXPORT_SYMBOL(loops_per_jiffy); |
206 | ||
207 | static int __init debug_kernel(char *str) | |
208 | { | |
a8fe19eb | 209 | console_loglevel = CONSOLE_LOGLEVEL_DEBUG; |
f6f21c81 | 210 | return 0; |
1da177e4 LT |
211 | } |
212 | ||
213 | static int __init quiet_kernel(char *str) | |
214 | { | |
a8fe19eb | 215 | console_loglevel = CONSOLE_LOGLEVEL_QUIET; |
f6f21c81 | 216 | return 0; |
1da177e4 LT |
217 | } |
218 | ||
f6f21c81 YL |
219 | early_param("debug", debug_kernel); |
220 | early_param("quiet", quiet_kernel); | |
1da177e4 LT |
221 | |
222 | static int __init loglevel(char *str) | |
223 | { | |
808bf29b AS |
224 | int newlevel; |
225 | ||
226 | /* | |
227 | * Only update loglevel value when a correct setting was passed, | |
228 | * to prevent blind crashes (when loglevel being set to 0) that | |
229 | * are quite hard to debug | |
230 | */ | |
231 | if (get_option(&str, &newlevel)) { | |
232 | console_loglevel = newlevel; | |
233 | return 0; | |
234 | } | |
235 | ||
236 | return -EINVAL; | |
1da177e4 LT |
237 | } |
238 | ||
f6f21c81 | 239 | early_param("loglevel", loglevel); |
1da177e4 | 240 | |
a99cd112 | 241 | /* Change NUL term back to "=", to make "param" the whole string. */ |
ecc86170 LR |
242 | static int __init repair_env_string(char *param, char *val, |
243 | const char *unused, void *arg) | |
1da177e4 | 244 | { |
1da177e4 LT |
245 | if (val) { |
246 | /* param=val or param="val"? */ | |
247 | if (val == param+strlen(param)+1) | |
248 | val[-1] = '='; | |
249 | else if (val == param+strlen(param)+2) { | |
250 | val[-2] = '='; | |
251 | memmove(val-1, val, strlen(val)+1); | |
252 | val--; | |
253 | } else | |
254 | BUG(); | |
255 | } | |
a99cd112 CM |
256 | return 0; |
257 | } | |
258 | ||
51e158c1 | 259 | /* Anything after -- gets handed straight to init. */ |
ecc86170 LR |
260 | static int __init set_init_arg(char *param, char *val, |
261 | const char *unused, void *arg) | |
51e158c1 RR |
262 | { |
263 | unsigned int i; | |
264 | ||
265 | if (panic_later) | |
266 | return 0; | |
267 | ||
ecc86170 | 268 | repair_env_string(param, val, unused, NULL); |
51e158c1 RR |
269 | |
270 | for (i = 0; argv_init[i]; i++) { | |
271 | if (i == MAX_INIT_ARGS) { | |
272 | panic_later = "init"; | |
273 | panic_param = param; | |
274 | return 0; | |
275 | } | |
276 | } | |
277 | argv_init[i] = param; | |
278 | return 0; | |
279 | } | |
280 | ||
a99cd112 CM |
281 | /* |
282 | * Unknown boot options get handed to init, unless they look like | |
283 | * unused parameters (modprobe will find them in /proc/cmdline). | |
284 | */ | |
ecc86170 LR |
285 | static int __init unknown_bootoption(char *param, char *val, |
286 | const char *unused, void *arg) | |
a99cd112 | 287 | { |
ecc86170 | 288 | repair_env_string(param, val, unused, NULL); |
1da177e4 LT |
289 | |
290 | /* Handle obsolete-style parameters */ | |
291 | if (obsolete_checksetup(param)) | |
292 | return 0; | |
293 | ||
f066a4f6 RR |
294 | /* Unused module parameter. */ |
295 | if (strchr(param, '.') && (!val || strchr(param, '.') < val)) | |
1da177e4 | 296 | return 0; |
1da177e4 LT |
297 | |
298 | if (panic_later) | |
299 | return 0; | |
300 | ||
301 | if (val) { | |
302 | /* Environment option */ | |
303 | unsigned int i; | |
304 | for (i = 0; envp_init[i]; i++) { | |
305 | if (i == MAX_INIT_ENVS) { | |
499a4584 | 306 | panic_later = "env"; |
1da177e4 LT |
307 | panic_param = param; |
308 | } | |
309 | if (!strncmp(param, envp_init[i], val - param)) | |
310 | break; | |
311 | } | |
312 | envp_init[i] = param; | |
313 | } else { | |
314 | /* Command line option */ | |
315 | unsigned int i; | |
316 | for (i = 0; argv_init[i]; i++) { | |
317 | if (i == MAX_INIT_ARGS) { | |
499a4584 | 318 | panic_later = "init"; |
1da177e4 LT |
319 | panic_param = param; |
320 | } | |
321 | } | |
322 | argv_init[i] = param; | |
323 | } | |
324 | return 0; | |
325 | } | |
326 | ||
327 | static int __init init_setup(char *str) | |
328 | { | |
329 | unsigned int i; | |
330 | ||
331 | execute_command = str; | |
332 | /* | |
333 | * In case LILO is going to boot us with default command line, | |
334 | * it prepends "auto" before the whole cmdline which makes | |
335 | * the shell think it should execute a script with such name. | |
336 | * So we ignore all arguments entered _before_ init=... [MJ] | |
337 | */ | |
338 | for (i = 1; i < MAX_INIT_ARGS; i++) | |
339 | argv_init[i] = NULL; | |
340 | return 1; | |
341 | } | |
342 | __setup("init=", init_setup); | |
343 | ||
ffdfc409 OJ |
344 | static int __init rdinit_setup(char *str) |
345 | { | |
346 | unsigned int i; | |
347 | ||
348 | ramdisk_execute_command = str; | |
349 | /* See "auto" comment in init_setup */ | |
350 | for (i = 1; i < MAX_INIT_ARGS; i++) | |
351 | argv_init[i] = NULL; | |
352 | return 1; | |
353 | } | |
354 | __setup("rdinit=", rdinit_setup); | |
355 | ||
1da177e4 | 356 | #ifndef CONFIG_SMP |
34db18a0 | 357 | static const unsigned int setup_max_cpus = NR_CPUS; |
e0982e90 | 358 | static inline void setup_nr_cpu_ids(void) { } |
1da177e4 | 359 | static inline void smp_prepare_cpus(unsigned int maxcpus) { } |
1da177e4 LT |
360 | #endif |
361 | ||
30d7e0d4 ABL |
362 | /* |
363 | * We need to store the untouched command line for future reference. | |
364 | * We also need to store the touched command line since the parameter | |
365 | * parsing is performed in place, and we should allow a component to | |
366 | * store reference of name/value for future reference. | |
367 | */ | |
368 | static void __init setup_command_line(char *command_line) | |
369 | { | |
098b081b SS |
370 | saved_command_line = |
371 | memblock_virt_alloc(strlen(boot_command_line) + 1, 0); | |
372 | initcall_command_line = | |
373 | memblock_virt_alloc(strlen(boot_command_line) + 1, 0); | |
374 | static_command_line = memblock_virt_alloc(strlen(command_line) + 1, 0); | |
dd4d9fec FF |
375 | strcpy(saved_command_line, boot_command_line); |
376 | strcpy(static_command_line, command_line); | |
30d7e0d4 ABL |
377 | } |
378 | ||
1da177e4 LT |
379 | /* |
380 | * We need to finalize in a non-__init function or else race conditions | |
381 | * between the root thread and the init thread may cause start_kernel to | |
382 | * be reaped by free_initmem before the root thread has proceeded to | |
383 | * cpu_idle. | |
384 | * | |
385 | * gcc-3.4 accidentally inlines this function, so use noinline. | |
386 | */ | |
387 | ||
b433c3d4 PZ |
388 | static __initdata DECLARE_COMPLETION(kthreadd_done); |
389 | ||
bd721ea7 | 390 | static noinline void __ref rest_init(void) |
1da177e4 | 391 | { |
8fb12156 | 392 | struct task_struct *tsk; |
73c27992 EB |
393 | int pid; |
394 | ||
7db905e6 | 395 | rcu_scheduler_starting(); |
b433c3d4 | 396 | /* |
97158569 | 397 | * We need to spawn init first so that it obtains pid 1, however |
b433c3d4 PZ |
398 | * the init task will end up wanting to create kthreads, which, if |
399 | * we schedule it before we create kthreadd, will OOPS. | |
400 | */ | |
8fb12156 TG |
401 | pid = kernel_thread(kernel_init, NULL, CLONE_FS); |
402 | /* | |
403 | * Pin init on the boot CPU. Task migration is not properly working | |
404 | * until sched_init_smp() has been run. It will set the allowed | |
405 | * CPUs for init to the non isolated CPUs. | |
406 | */ | |
407 | rcu_read_lock(); | |
408 | tsk = find_task_by_pid_ns(pid, &init_pid_ns); | |
409 | set_cpus_allowed_ptr(tsk, cpumask_of(smp_processor_id())); | |
410 | rcu_read_unlock(); | |
411 | ||
1da177e4 | 412 | numa_default_policy(); |
73c27992 | 413 | pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES); |
d11c563d | 414 | rcu_read_lock(); |
5cd20455 | 415 | kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns); |
d11c563d | 416 | rcu_read_unlock(); |
1c3c5eab TG |
417 | |
418 | /* | |
419 | * Enable might_sleep() and smp_processor_id() checks. | |
420 | * They cannot be enabled earlier because with CONFIG_PRREMPT=y | |
421 | * kernel_thread() would trigger might_sleep() splats. With | |
422 | * CONFIG_PREEMPT_VOLUNTARY=y the init task might have scheduled | |
423 | * already, but it's stuck on the kthreadd_done completion. | |
424 | */ | |
425 | system_state = SYSTEM_SCHEDULING; | |
426 | ||
b433c3d4 | 427 | complete(&kthreadd_done); |
f340c0d1 IM |
428 | |
429 | /* | |
430 | * The boot idle thread must execute schedule() | |
1df21055 | 431 | * at least once to get things moving: |
f340c0d1 | 432 | */ |
bd2f5536 | 433 | schedule_preempt_disabled(); |
5bfb5d69 | 434 | /* Call into cpu_idle with preempt disabled */ |
a1a04ec3 | 435 | cpu_startup_entry(CPUHP_ONLINE); |
1df21055 | 436 | } |
1da177e4 LT |
437 | |
438 | /* Check for early params. */ | |
ecc86170 LR |
439 | static int __init do_early_param(char *param, char *val, |
440 | const char *unused, void *arg) | |
1da177e4 | 441 | { |
914dcaa8 | 442 | const struct obs_kernel_param *p; |
1da177e4 LT |
443 | |
444 | for (p = __setup_start; p < __setup_end; p++) { | |
b1e4d20c | 445 | if ((p->early && parameq(param, p->str)) || |
18a8bd94 YL |
446 | (strcmp(param, "console") == 0 && |
447 | strcmp(p->str, "earlycon") == 0) | |
448 | ) { | |
1da177e4 | 449 | if (p->setup_func(val) != 0) |
ea676e84 | 450 | pr_warn("Malformed early option '%s'\n", param); |
1da177e4 LT |
451 | } |
452 | } | |
453 | /* We accept everything at this stage. */ | |
454 | return 0; | |
455 | } | |
456 | ||
13977091 MD |
457 | void __init parse_early_options(char *cmdline) |
458 | { | |
ecc86170 LR |
459 | parse_args("early options", cmdline, NULL, 0, 0, 0, NULL, |
460 | do_early_param); | |
13977091 MD |
461 | } |
462 | ||
1da177e4 LT |
463 | /* Arch code calls this early on, or if not, just before other parsing. */ |
464 | void __init parse_early_param(void) | |
465 | { | |
dd4d9fec FF |
466 | static int done __initdata; |
467 | static char tmp_cmdline[COMMAND_LINE_SIZE] __initdata; | |
1da177e4 LT |
468 | |
469 | if (done) | |
470 | return; | |
471 | ||
472 | /* All fall through to do_early_param. */ | |
30d7e0d4 | 473 | strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE); |
13977091 | 474 | parse_early_options(tmp_cmdline); |
1da177e4 LT |
475 | done = 1; |
476 | } | |
477 | ||
e7ff3a47 TG |
478 | void __init __weak arch_post_acpi_subsys_init(void) { } |
479 | ||
839ad62e | 480 | void __init __weak smp_setup_processor_id(void) |
033ab7f8 AM |
481 | { |
482 | } | |
483 | ||
eded09cc | 484 | # if THREAD_SIZE >= PAGE_SIZE |
b235beea | 485 | void __init __weak thread_stack_cache_init(void) |
8c9843e5 BH |
486 | { |
487 | } | |
eded09cc | 488 | #endif |
8c9843e5 | 489 | |
c7753208 TL |
490 | void __init __weak mem_encrypt_init(void) { } |
491 | ||
444f478f PE |
492 | /* |
493 | * Set up kernel memory allocators | |
494 | */ | |
495 | static void __init mm_init(void) | |
496 | { | |
eefa864b JK |
497 | /* |
498 | * page_ext requires contiguous pages, | |
499 | * bigger than MAX_ORDER unless SPARSEMEM. | |
500 | */ | |
501 | page_ext_init_flatmem(); | |
444f478f PE |
502 | mem_init(); |
503 | kmem_cache_init(); | |
b35f1819 | 504 | pgtable_init(); |
444f478f | 505 | vmalloc_init(); |
0ddab1d2 | 506 | ioremap_huge_init(); |
613e396b TG |
507 | /* Should be run before the first non-init thread is created */ |
508 | init_espfix_bsp(); | |
aa8c6248 TG |
509 | /* Should be run after espfix64 is set up. */ |
510 | pti_init(); | |
444f478f PE |
511 | } |
512 | ||
722a9f92 | 513 | asmlinkage __visible void __init start_kernel(void) |
1da177e4 | 514 | { |
dd4d9fec FF |
515 | char *command_line; |
516 | char *after_dashes; | |
033ab7f8 | 517 | |
d4311ff1 | 518 | set_task_stack_end_magic(&init_task); |
73839c5b | 519 | smp_setup_processor_id(); |
3ac7fe5a | 520 | debug_objects_early_init(); |
42059429 | 521 | |
ddbcc7e8 | 522 | cgroup_init_early(); |
fbb9ce95 IM |
523 | |
524 | local_irq_disable(); | |
2ce802f6 | 525 | early_boot_irqs_disabled = true; |
fbb9ce95 | 526 | |
1b3b3b49 VK |
527 | /* |
528 | * Interrupts are still disabled. Do necessary setups, then | |
529 | * enable them. | |
530 | */ | |
44fd2299 | 531 | boot_cpu_init(); |
1da177e4 | 532 | page_address_init(); |
ea676e84 | 533 | pr_notice("%s", linux_banner); |
1da177e4 | 534 | setup_arch(&command_line); |
121388a3 LA |
535 | /* |
536 | * Set up the the initial canary and entropy after arch | |
33d72f38 | 537 | * and after adding latent and command line entropy. |
121388a3 LA |
538 | */ |
539 | add_latent_entropy(); | |
33d72f38 | 540 | add_device_randomness(command_line, strlen(command_line)); |
121388a3 | 541 | boot_init_stack_canary(); |
6345d24d | 542 | mm_init_cpumask(&init_mm); |
30d7e0d4 | 543 | setup_command_line(command_line); |
e0982e90 | 544 | setup_nr_cpu_ids(); |
d6647bdf | 545 | setup_per_cpu_areas(); |
44fd2299 | 546 | smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */ |
d403c57a | 547 | boot_cpu_hotplug_init(); |
1da177e4 | 548 | |
72675e13 | 549 | build_all_zonelists(NULL); |
83b519e8 PE |
550 | page_alloc_init(); |
551 | ||
ea676e84 | 552 | pr_notice("Kernel command line: %s\n", boot_command_line); |
58b46af0 DW |
553 | /* parameters may set static keys */ |
554 | jump_label_init(); | |
83b519e8 | 555 | parse_early_param(); |
51e158c1 RR |
556 | after_dashes = parse_args("Booting kernel", |
557 | static_command_line, __start___param, | |
558 | __stop___param - __start___param, | |
ecc86170 | 559 | -1, -1, NULL, &unknown_bootoption); |
3438cf54 | 560 | if (!IS_ERR_OR_NULL(after_dashes)) |
51e158c1 | 561 | parse_args("Setting init args", after_dashes, NULL, 0, -1, -1, |
ecc86170 | 562 | NULL, set_init_arg); |
97ce2c88 | 563 | |
83b519e8 PE |
564 | /* |
565 | * These use large bootmem allocations and must precede | |
566 | * kmem_cache_init() | |
567 | */ | |
162a7e75 | 568 | setup_log_buf(0); |
83b519e8 PE |
569 | vfs_caches_init_early(); |
570 | sort_main_extable(); | |
571 | trap_init(); | |
444f478f | 572 | mm_init(); |
de03c72c | 573 | |
f631718d SRV |
574 | ftrace_init(); |
575 | ||
e725c731 SRV |
576 | /* trace_printk can be enabled here */ |
577 | early_trace_init(); | |
578 | ||
1da177e4 LT |
579 | /* |
580 | * Set up the scheduler prior starting any interrupts (such as the | |
581 | * timer interrupt). Full topology setup happens at smp_init() | |
582 | * time - but meanwhile we still have a functioning scheduler. | |
583 | */ | |
584 | sched_init(); | |
585 | /* | |
586 | * Disable preemption - early bootup scheduling is extremely | |
587 | * fragile until we cpu_idle() for the first time. | |
588 | */ | |
589 | preempt_disable(); | |
dd4d9fec FF |
590 | if (WARN(!irqs_disabled(), |
591 | "Interrupts were enabled *very* early, fixing it\n")) | |
c4a68306 | 592 | local_irq_disable(); |
0a835c4f | 593 | radix_tree_init(); |
3347fa09 | 594 | |
7d229c66 TS |
595 | /* |
596 | * Set up housekeeping before setting up workqueues to allow the unbound | |
597 | * workqueue to take non-housekeeping into account. | |
598 | */ | |
599 | housekeeping_init(); | |
600 | ||
3347fa09 TH |
601 | /* |
602 | * Allow workqueue creation and work item queueing/cancelling | |
603 | * early. Work item execution depends on kthreads and starts after | |
604 | * workqueue_init(). | |
605 | */ | |
606 | workqueue_init_early(); | |
607 | ||
1da177e4 | 608 | rcu_init(); |
5f893b26 | 609 | |
e725c731 | 610 | /* Trace events are available after this */ |
5f893b26 SRRH |
611 | trace_init(); |
612 | ||
65f382fd | 613 | context_tracking_init(); |
0b8f1efa YL |
614 | /* init some links before init_ISA_irqs() */ |
615 | early_irq_init(); | |
1da177e4 | 616 | init_IRQ(); |
ad2b1353 | 617 | tick_init(); |
d6dd50e0 | 618 | rcu_init_nohz(); |
1da177e4 | 619 | init_timers(); |
c0a31329 | 620 | hrtimers_init(); |
1da177e4 | 621 | softirq_init(); |
ad596171 | 622 | timekeeping_init(); |
88fecaa2 | 623 | time_init(); |
38ff87f7 | 624 | sched_clock_postinit(); |
f92bac3b | 625 | printk_safe_init(); |
9e630205 | 626 | perf_event_init(); |
93e02814 | 627 | profile_init(); |
d8ad7d11 | 628 | call_function_init(); |
f91eb62f | 629 | WARN(!irqs_disabled(), "Interrupts were enabled early\n"); |
2ce802f6 | 630 | early_boot_irqs_disabled = false; |
93e02814 | 631 | local_irq_enable(); |
dcce284a | 632 | |
7e85ee0c | 633 | kmem_cache_init_late(); |
1da177e4 LT |
634 | |
635 | /* | |
636 | * HACK ALERT! This is early. We're enabling the console before | |
637 | * we've done PCI setups etc, and console_init() must be aware of | |
638 | * this. But we do want output early, in case something goes wrong. | |
639 | */ | |
640 | console_init(); | |
641 | if (panic_later) | |
499a4584 TH |
642 | panic("Too many boot %s vars at `%s'", panic_later, |
643 | panic_param); | |
fbb9ce95 IM |
644 | |
645 | lockdep_info(); | |
646 | ||
9a11b49a IM |
647 | /* |
648 | * Need to run this when irqs are enabled, because it wants | |
649 | * to self-test [hard/soft]-irqs on/off lock inversion bugs | |
650 | * too: | |
651 | */ | |
652 | locking_selftest(); | |
653 | ||
c7753208 TL |
654 | /* |
655 | * This needs to be called before any devices perform DMA | |
656 | * operations that might use the SWIOTLB bounce buffers. It will | |
657 | * mark the bounce buffers as decrypted so that their usage will | |
658 | * not cause "plain-text" data to be decrypted when accessed. | |
659 | */ | |
660 | mem_encrypt_init(); | |
661 | ||
1da177e4 LT |
662 | #ifdef CONFIG_BLK_DEV_INITRD |
663 | if (initrd_start && !initrd_below_start_ok && | |
bd673c7c | 664 | page_to_pfn(virt_to_page((void *)initrd_start)) < min_low_pfn) { |
ea676e84 | 665 | pr_crit("initrd overwritten (0x%08lx < 0x%08lx) - disabling it.\n", |
bd673c7c GU |
666 | page_to_pfn(virt_to_page((void *)initrd_start)), |
667 | min_low_pfn); | |
1da177e4 LT |
668 | initrd_start = 0; |
669 | } | |
670 | #endif | |
9b090f2d | 671 | kmemleak_init(); |
caba4cbb | 672 | debug_objects_mem_init(); |
e7c8d5c9 | 673 | setup_per_cpu_pageset(); |
1da177e4 | 674 | numa_policy_init(); |
9c71206d | 675 | acpi_early_init(); |
1da177e4 LT |
676 | if (late_time_init) |
677 | late_time_init(); | |
678 | calibrate_delay(); | |
95846ecf | 679 | pid_idr_init(); |
1da177e4 | 680 | anon_vma_init(); |
11520e5e | 681 | #ifdef CONFIG_X86 |
83e68189 | 682 | if (efi_enabled(EFI_RUNTIME_SERVICES)) |
11520e5e LT |
683 | efi_enter_virtual_mode(); |
684 | #endif | |
b235beea | 685 | thread_stack_cache_init(); |
d84f4f99 | 686 | cred_init(); |
ff691f6e | 687 | fork_init(); |
1da177e4 LT |
688 | proc_caches_init(); |
689 | buffer_init(); | |
1da177e4 LT |
690 | key_init(); |
691 | security_init(); | |
0b4b3827 | 692 | dbg_late_init(); |
4248b0da | 693 | vfs_caches_init(); |
62906027 | 694 | pagecache_init(); |
1da177e4 | 695 | signals_init(); |
1da177e4 | 696 | proc_root_init(); |
e149ed2b | 697 | nsfs_init(); |
1da177e4 | 698 | cpuset_init(); |
695df213 | 699 | cgroup_init(); |
c757249a | 700 | taskstats_init_early(); |
ca74e92b | 701 | delayacct_init(); |
1da177e4 LT |
702 | |
703 | check_bugs(); | |
704 | ||
b064a8fa | 705 | acpi_subsystem_init(); |
e7ff3a47 | 706 | arch_post_acpi_subsys_init(); |
6ae6996a | 707 | sfi_init_late(); |
1da177e4 | 708 | |
83e68189 | 709 | if (efi_enabled(EFI_RUNTIME_SERVICES)) { |
78510792 | 710 | efi_free_boot_services(); |
2223af38 | 711 | } |
78510792 | 712 | |
1da177e4 LT |
713 | /* Do the rest non-__init'ed, we're now alive */ |
714 | rest_init(); | |
715 | } | |
716 | ||
b99b87f7 PO |
717 | /* Call all constructor functions linked into the kernel. */ |
718 | static void __init do_ctors(void) | |
719 | { | |
720 | #ifdef CONFIG_CONSTRUCTORS | |
196a15b4 | 721 | ctor_fn_t *fn = (ctor_fn_t *) __ctors_start; |
b99b87f7 | 722 | |
196a15b4 HS |
723 | for (; fn < (ctor_fn_t *) __ctors_end; fn++) |
724 | (*fn)(); | |
b99b87f7 PO |
725 | #endif |
726 | } | |
727 | ||
2329abfa | 728 | bool initcall_debug; |
d0ea3d7d | 729 | core_param(initcall_debug, initcall_debug, bool, 0644); |
1da177e4 | 730 | |
7b0b73d7 PB |
731 | #ifdef CONFIG_KALLSYMS |
732 | struct blacklist_entry { | |
733 | struct list_head next; | |
734 | char *buf; | |
735 | }; | |
736 | ||
737 | static __initdata_or_module LIST_HEAD(blacklisted_initcalls); | |
738 | ||
739 | static int __init initcall_blacklist(char *str) | |
740 | { | |
741 | char *str_entry; | |
742 | struct blacklist_entry *entry; | |
743 | ||
744 | /* str argument is a comma-separated list of functions */ | |
745 | do { | |
746 | str_entry = strsep(&str, ","); | |
747 | if (str_entry) { | |
748 | pr_debug("blacklisting initcall %s\n", str_entry); | |
749 | entry = alloc_bootmem(sizeof(*entry)); | |
750 | entry->buf = alloc_bootmem(strlen(str_entry) + 1); | |
751 | strcpy(entry->buf, str_entry); | |
752 | list_add(&entry->next, &blacklisted_initcalls); | |
753 | } | |
754 | } while (str_entry); | |
755 | ||
756 | return 0; | |
757 | } | |
758 | ||
759 | static bool __init_or_module initcall_blacklisted(initcall_t fn) | |
760 | { | |
7b0b73d7 | 761 | struct blacklist_entry *entry; |
c8cdd2be | 762 | char fn_name[KSYM_SYMBOL_LEN]; |
0fd5ed8d | 763 | unsigned long addr; |
7b0b73d7 | 764 | |
c8cdd2be | 765 | if (list_empty(&blacklisted_initcalls)) |
7b0b73d7 PB |
766 | return false; |
767 | ||
0fd5ed8d RV |
768 | addr = (unsigned long) dereference_function_descriptor(fn); |
769 | sprint_symbol_no_offset(fn_name, addr); | |
c8cdd2be | 770 | |
841c06d7 PB |
771 | /* |
772 | * fn will be "function_name [module_name]" where [module_name] is not | |
773 | * displayed for built-in init functions. Strip off the [module_name]. | |
774 | */ | |
775 | strreplace(fn_name, ' ', '\0'); | |
776 | ||
e6fd1fb3 | 777 | list_for_each_entry(entry, &blacklisted_initcalls, next) { |
7b0b73d7 PB |
778 | if (!strcmp(fn_name, entry->buf)) { |
779 | pr_debug("initcall %s blacklisted\n", fn_name); | |
7b0b73d7 PB |
780 | return true; |
781 | } | |
782 | } | |
783 | ||
7b0b73d7 PB |
784 | return false; |
785 | } | |
786 | #else | |
787 | static int __init initcall_blacklist(char *str) | |
788 | { | |
789 | pr_warn("initcall_blacklist requires CONFIG_KALLSYMS\n"); | |
790 | return 0; | |
791 | } | |
792 | ||
793 | static bool __init_or_module initcall_blacklisted(initcall_t fn) | |
794 | { | |
795 | return false; | |
796 | } | |
797 | #endif | |
798 | __setup("initcall_blacklist=", initcall_blacklist); | |
799 | ||
e4461271 | 800 | static int __init_or_module do_one_initcall_debug(initcall_t fn) |
1da177e4 | 801 | { |
41e3a9b4 | 802 | unsigned long long calltime, delta, rettime; |
30dbb20e AW |
803 | unsigned long long duration; |
804 | int ret; | |
1da177e4 | 805 | |
d62cf815 | 806 | printk(KERN_DEBUG "calling %pF @ %i\n", fn, task_pid_nr(current)); |
41e3a9b4 | 807 | calltime = local_clock(); |
30dbb20e | 808 | ret = fn(); |
41e3a9b4 AV |
809 | rettime = local_clock(); |
810 | delta = rettime - calltime; | |
811 | duration = delta >> 10; | |
d62cf815 | 812 | printk(KERN_DEBUG "initcall %pF returned %d after %lld usecs\n", |
ea676e84 | 813 | fn, ret, duration); |
1da177e4 | 814 | |
22c5c03b KW |
815 | return ret; |
816 | } | |
817 | ||
e4461271 | 818 | int __init_or_module do_one_initcall(initcall_t fn) |
22c5c03b KW |
819 | { |
820 | int count = preempt_count(); | |
821 | int ret; | |
ff1c8fac | 822 | char msgbuf[64]; |
22c5c03b | 823 | |
7b0b73d7 PB |
824 | if (initcall_blacklisted(fn)) |
825 | return -EPERM; | |
826 | ||
22c5c03b KW |
827 | if (initcall_debug) |
828 | ret = do_one_initcall_debug(fn); | |
829 | else | |
830 | ret = fn(); | |
8f0c45cd | 831 | |
e0df154f | 832 | msgbuf[0] = 0; |
e662e1cf | 833 | |
e0df154f | 834 | if (preempt_count() != count) { |
bf5d770b | 835 | sprintf(msgbuf, "preemption imbalance "); |
4a2b4b22 | 836 | preempt_count_set(count); |
1da177e4 | 837 | } |
e0df154f | 838 | if (irqs_disabled()) { |
a76bfd0d | 839 | strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf)); |
e0df154f LT |
840 | local_irq_enable(); |
841 | } | |
f91eb62f | 842 | WARN(msgbuf[0], "initcall %pF returned with %s\n", fn, msgbuf); |
59f9415f | 843 | |
38addce8 | 844 | add_latent_entropy(); |
30dbb20e | 845 | return ret; |
e0df154f LT |
846 | } |
847 | ||
848 | ||
026cee00 PM |
849 | extern initcall_t __initcall_start[]; |
850 | extern initcall_t __initcall0_start[]; | |
851 | extern initcall_t __initcall1_start[]; | |
852 | extern initcall_t __initcall2_start[]; | |
853 | extern initcall_t __initcall3_start[]; | |
854 | extern initcall_t __initcall4_start[]; | |
855 | extern initcall_t __initcall5_start[]; | |
856 | extern initcall_t __initcall6_start[]; | |
857 | extern initcall_t __initcall7_start[]; | |
858 | extern initcall_t __initcall_end[]; | |
859 | ||
860 | static initcall_t *initcall_levels[] __initdata = { | |
861 | __initcall0_start, | |
862 | __initcall1_start, | |
863 | __initcall2_start, | |
864 | __initcall3_start, | |
865 | __initcall4_start, | |
866 | __initcall5_start, | |
867 | __initcall6_start, | |
868 | __initcall7_start, | |
869 | __initcall_end, | |
870 | }; | |
871 | ||
96263d28 | 872 | /* Keep these in sync with initcalls in include/linux/init.h */ |
026cee00 | 873 | static char *initcall_level_names[] __initdata = { |
9fb48c74 JC |
874 | "early", |
875 | "core", | |
876 | "postcore", | |
877 | "arch", | |
878 | "subsys", | |
879 | "fs", | |
880 | "device", | |
881 | "late", | |
026cee00 PM |
882 | }; |
883 | ||
026cee00 | 884 | static void __init do_initcall_level(int level) |
e0df154f | 885 | { |
196a15b4 | 886 | initcall_t *fn; |
e0df154f | 887 | |
08746a65 | 888 | strcpy(initcall_command_line, saved_command_line); |
026cee00 | 889 | parse_args(initcall_level_names[level], |
08746a65 | 890 | initcall_command_line, __start___param, |
026cee00 PM |
891 | __stop___param - __start___param, |
892 | level, level, | |
ecc86170 | 893 | NULL, &repair_env_string); |
026cee00 PM |
894 | |
895 | for (fn = initcall_levels[level]; fn < initcall_levels[level+1]; fn++) | |
196a15b4 | 896 | do_one_initcall(*fn); |
1da177e4 LT |
897 | } |
898 | ||
026cee00 PM |
899 | static void __init do_initcalls(void) |
900 | { | |
901 | int level; | |
902 | ||
19efb72f | 903 | for (level = 0; level < ARRAY_SIZE(initcall_levels) - 1; level++) |
026cee00 PM |
904 | do_initcall_level(level); |
905 | } | |
906 | ||
1da177e4 LT |
907 | /* |
908 | * Ok, the machine is now initialized. None of the devices | |
909 | * have been touched yet, but the CPU subsystem is up and | |
910 | * running, and memory and process management works. | |
911 | * | |
912 | * Now we can finally start doing some real work.. | |
913 | */ | |
914 | static void __init do_basic_setup(void) | |
915 | { | |
759ee091 | 916 | cpuset_init_smp(); |
41ffe5d5 | 917 | shmem_init(); |
1da177e4 | 918 | driver_init(); |
b04c3afb | 919 | init_irq_proc(); |
b99b87f7 | 920 | do_ctors(); |
d5767c53 | 921 | usermodehelper_enable(); |
b0f84374 | 922 | do_initcalls(); |
1da177e4 LT |
923 | } |
924 | ||
7babe8db | 925 | static void __init do_pre_smp_initcalls(void) |
c2147a50 | 926 | { |
196a15b4 | 927 | initcall_t *fn; |
c2147a50 | 928 | |
026cee00 | 929 | for (fn = __initcall_start; fn < __initcall0_start; fn++) |
196a15b4 | 930 | do_one_initcall(*fn); |
c2147a50 EGM |
931 | } |
932 | ||
bb813f4c TH |
933 | /* |
934 | * This function requests modules which should be loaded by default and is | |
935 | * called twice right after initrd is mounted and right before init is | |
936 | * exec'd. If such modules are on either initrd or rootfs, they will be | |
937 | * loaded before control is passed to userland. | |
938 | */ | |
939 | void __init load_default_modules(void) | |
940 | { | |
941 | load_default_elevator_module(); | |
942 | } | |
943 | ||
a74fb73c | 944 | static int run_init_process(const char *init_filename) |
1da177e4 LT |
945 | { |
946 | argv_init[0] = init_filename; | |
c4ad8f98 | 947 | return do_execve(getname_kernel(init_filename), |
ae903caa AV |
948 | (const char __user *const __user *)argv_init, |
949 | (const char __user *const __user *)envp_init); | |
1da177e4 LT |
950 | } |
951 | ||
ba24762b MO |
952 | static int try_to_run_init_process(const char *init_filename) |
953 | { | |
954 | int ret; | |
955 | ||
956 | ret = run_init_process(init_filename); | |
957 | ||
958 | if (ret && ret != -ENOENT) { | |
959 | pr_err("Starting init: %s exists but couldn't execute it (error %d)\n", | |
960 | init_filename, ret); | |
961 | } | |
962 | ||
963 | return ret; | |
964 | } | |
965 | ||
f80b0c90 | 966 | static noinline void __init kernel_init_freeable(void); |
d6b21238 | 967 | |
0f5bf6d0 | 968 | #if defined(CONFIG_STRICT_KERNEL_RWX) || defined(CONFIG_STRICT_MODULE_RWX) |
39290b38 | 969 | bool rodata_enabled __ro_after_init = true; |
d2aa1aca KC |
970 | static int __init set_debug_rodata(char *str) |
971 | { | |
972 | return strtobool(str, &rodata_enabled); | |
973 | } | |
974 | __setup("rodata=", set_debug_rodata); | |
39290b38 | 975 | #endif |
d2aa1aca | 976 | |
0f5bf6d0 | 977 | #ifdef CONFIG_STRICT_KERNEL_RWX |
d2aa1aca KC |
978 | static void mark_readonly(void) |
979 | { | |
2959a5f7 | 980 | if (rodata_enabled) { |
9662dfde JH |
981 | /* |
982 | * load_module() results in W+X mappings, which are cleaned up | |
983 | * with call_rcu_sched(). Let's make sure that queued work is | |
984 | * flushed so that we don't hit false positives looking for | |
985 | * insecure pages which are W+X. | |
986 | */ | |
987 | rcu_barrier_sched(); | |
d2aa1aca | 988 | mark_rodata_ro(); |
2959a5f7 JP |
989 | rodata_test(); |
990 | } else | |
d2aa1aca KC |
991 | pr_info("Kernel memory protection disabled.\n"); |
992 | } | |
993 | #else | |
994 | static inline void mark_readonly(void) | |
995 | { | |
996 | pr_warn("This architecture does not have kernel memory protection.\n"); | |
997 | } | |
998 | #endif | |
999 | ||
d6b21238 | 1000 | static int __ref kernel_init(void *unused) |
ee5bfa64 | 1001 | { |
ba24762b MO |
1002 | int ret; |
1003 | ||
d6b21238 | 1004 | kernel_init_freeable(); |
22a9d645 AV |
1005 | /* need to finish all async __init code before freeing the memory */ |
1006 | async_synchronize_full(); | |
b80f0f6c | 1007 | ftrace_free_init_mem(); |
ee5bfa64 | 1008 | free_initmem(); |
d2aa1aca | 1009 | mark_readonly(); |
4cec8b0c JR |
1010 | |
1011 | /* | |
1012 | * Kernel mappings are now finalized - update the userspace page-table | |
1013 | * to finalize PTI. | |
1014 | */ | |
1015 | pti_finalize(); | |
1016 | ||
ee5bfa64 VG |
1017 | system_state = SYSTEM_RUNNING; |
1018 | numa_default_policy(); | |
1019 | ||
967dcb8f PM |
1020 | rcu_end_inkernel_boot(); |
1021 | ||
ee5bfa64 | 1022 | if (ramdisk_execute_command) { |
ba24762b MO |
1023 | ret = run_init_process(ramdisk_execute_command); |
1024 | if (!ret) | |
a74fb73c | 1025 | return 0; |
ba24762b MO |
1026 | pr_err("Failed to execute %s (error %d)\n", |
1027 | ramdisk_execute_command, ret); | |
ee5bfa64 VG |
1028 | } |
1029 | ||
1030 | /* | |
1031 | * We try each of these until one succeeds. | |
1032 | * | |
1033 | * The Bourne shell can be used instead of init if we are | |
1034 | * trying to recover a really broken machine. | |
1035 | */ | |
1036 | if (execute_command) { | |
ba24762b MO |
1037 | ret = run_init_process(execute_command); |
1038 | if (!ret) | |
a74fb73c | 1039 | return 0; |
6ef4536e AL |
1040 | panic("Requested init %s failed (error %d).", |
1041 | execute_command, ret); | |
ee5bfa64 | 1042 | } |
ba24762b MO |
1043 | if (!try_to_run_init_process("/sbin/init") || |
1044 | !try_to_run_init_process("/etc/init") || | |
1045 | !try_to_run_init_process("/bin/init") || | |
1046 | !try_to_run_init_process("/bin/sh")) | |
a74fb73c | 1047 | return 0; |
ee5bfa64 | 1048 | |
ba24762b | 1049 | panic("No working init found. Try passing init= option to kernel. " |
8c27ceff | 1050 | "See Linux Documentation/admin-guide/init.rst for guidance."); |
ee5bfa64 VG |
1051 | } |
1052 | ||
f80b0c90 | 1053 | static noinline void __init kernel_init_freeable(void) |
1da177e4 | 1054 | { |
b433c3d4 PZ |
1055 | /* |
1056 | * Wait until kthreadd is all set-up. | |
1057 | */ | |
1058 | wait_for_completion(&kthreadd_done); | |
31a67102 LT |
1059 | |
1060 | /* Now the scheduler is fully set up and can do blocking allocations */ | |
1061 | gfp_allowed_mask = __GFP_BITS_MASK; | |
1062 | ||
58568d2a MX |
1063 | /* |
1064 | * init can allocate pages on any node | |
1065 | */ | |
3c466d46 | 1066 | set_mems_allowed(node_states[N_MEMORY]); |
1da177e4 | 1067 | |
9ec52099 CLG |
1068 | cad_pid = task_pid(current); |
1069 | ||
ca74a6f8 | 1070 | smp_prepare_cpus(setup_max_cpus); |
1da177e4 | 1071 | |
3347fa09 TH |
1072 | workqueue_init(); |
1073 | ||
597b7305 MH |
1074 | init_mm_internals(); |
1075 | ||
1da177e4 | 1076 | do_pre_smp_initcalls(); |
004417a6 | 1077 | lockup_detector_init(); |
1da177e4 | 1078 | |
1da177e4 LT |
1079 | smp_init(); |
1080 | sched_init_smp(); | |
1081 | ||
0e1cc95b | 1082 | page_alloc_init_late(); |
89b54b65 QC |
1083 | /* Initialize page ext after all struct pages are initialized. */ |
1084 | page_ext_init(); | |
0e1cc95b | 1085 | |
1da177e4 LT |
1086 | do_basic_setup(); |
1087 | ||
2bd3a997 EB |
1088 | /* Open the /dev/console on the rootfs, this should never fail */ |
1089 | if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0) | |
ea676e84 | 1090 | pr_err("Warning: unable to open an initial console.\n"); |
2bd3a997 EB |
1091 | |
1092 | (void) sys_dup(0); | |
1093 | (void) sys_dup(0); | |
1da177e4 LT |
1094 | /* |
1095 | * check if there is an early userspace init. If yes, let it do all | |
1096 | * the work | |
1097 | */ | |
ffdfc409 OJ |
1098 | |
1099 | if (!ramdisk_execute_command) | |
1100 | ramdisk_execute_command = "/init"; | |
1101 | ||
1102 | if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) { | |
1103 | ramdisk_execute_command = NULL; | |
1da177e4 | 1104 | prepare_namespace(); |
ffdfc409 | 1105 | } |
1da177e4 LT |
1106 | |
1107 | /* | |
1108 | * Ok, we have completed the initial bootup, and | |
1109 | * we're essentially up and running. Get rid of the | |
1110 | * initmem segments and start the user-mode stuff.. | |
c9cd2ce2 DK |
1111 | * |
1112 | * rootfs is available now, try loading the public keys | |
1113 | * and default modules | |
1da177e4 | 1114 | */ |
bb813f4c | 1115 | |
c9cd2ce2 | 1116 | integrity_load_keys(); |
bb813f4c | 1117 | load_default_modules(); |
1da177e4 | 1118 | } |