]>
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> |
57c8a661 | 28 | #include <linux/memblock.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 | 53 | #include <linux/unistd.h> |
3ea056c5 | 54 | #include <linux/utsname.h> |
1da177e4 LT |
55 | #include <linux/rmap.h> |
56 | #include <linux/mempolicy.h> | |
57 | #include <linux/key.h> | |
b6cd0b77 | 58 | #include <linux/buffer_head.h> |
eefa864b | 59 | #include <linux/page_ext.h> |
9a11b49a | 60 | #include <linux/debug_locks.h> |
3ac7fe5a | 61 | #include <linux/debugobjects.h> |
fbb9ce95 | 62 | #include <linux/lockdep.h> |
3c7b4e6b | 63 | #include <linux/kmemleak.h> |
84d73786 | 64 | #include <linux/pid_namespace.h> |
1f21782e | 65 | #include <linux/device.h> |
73c27992 | 66 | #include <linux/kthread.h> |
e6fe6649 | 67 | #include <linux/sched.h> |
1777e463 | 68 | #include <linux/sched/init.h> |
a1c9eea9 | 69 | #include <linux/signal.h> |
199f0ca5 | 70 | #include <linux/idr.h> |
0b4b3827 | 71 | #include <linux/kgdb.h> |
68bf21aa | 72 | #include <linux/ftrace.h> |
22a9d645 | 73 | #include <linux/async.h> |
6ae6996a | 74 | #include <linux/sfi.h> |
2b2af54a | 75 | #include <linux/shmem_fs.h> |
5a0e3ad6 | 76 | #include <linux/slab.h> |
24a24bb6 | 77 | #include <linux/perf_event.h> |
a74fb73c | 78 | #include <linux/ptrace.h> |
aa8c6248 | 79 | #include <linux/pti.h> |
bb813f4c TH |
80 | #include <linux/blkdev.h> |
81 | #include <linux/elevator.h> | |
5d2a4e91 | 82 | #include <linux/sched/clock.h> |
29930025 | 83 | #include <linux/sched/task.h> |
68db0cf1 | 84 | #include <linux/sched/task_stack.h> |
65f382fd | 85 | #include <linux/context_tracking.h> |
47d06e53 | 86 | #include <linux/random.h> |
7b0b73d7 | 87 | #include <linux/list.h> |
c9cd2ce2 | 88 | #include <linux/integrity.h> |
e149ed2b | 89 | #include <linux/proc_ns.h> |
0ddab1d2 | 90 | #include <linux/io.h> |
39290b38 | 91 | #include <linux/cache.h> |
2959a5f7 | 92 | #include <linux/rodata_test.h> |
33352244 | 93 | #include <linux/jump_label.h> |
ae67d58d | 94 | #include <linux/mem_encrypt.h> |
1da177e4 LT |
95 | |
96 | #include <asm/io.h> | |
97 | #include <asm/bugs.h> | |
98 | #include <asm/setup.h> | |
a940199f | 99 | #include <asm/sections.h> |
37b73c82 | 100 | #include <asm/cacheflush.h> |
1da177e4 | 101 | |
4ee7c60d SRV |
102 | #define CREATE_TRACE_POINTS |
103 | #include <trace/events/initcall.h> | |
104 | ||
aae5f662 | 105 | static int kernel_init(void *); |
1da177e4 LT |
106 | |
107 | extern void init_IRQ(void); | |
ff691f6e | 108 | extern void fork_init(void); |
1da177e4 | 109 | extern void radix_tree_init(void); |
1da177e4 | 110 | |
2ce802f6 TH |
111 | /* |
112 | * Debug helper: via this flag we know that we are in 'early bootup code' | |
113 | * where only the boot processor is running with IRQ disabled. This means | |
114 | * two things - IRQ must not be enabled before the flag is cleared and some | |
115 | * operations which are not allowed with IRQ disabled are allowed while the | |
116 | * flag is set. | |
117 | */ | |
118 | bool early_boot_irqs_disabled __read_mostly; | |
119 | ||
a6826048 | 120 | enum system_states system_state __read_mostly; |
1da177e4 LT |
121 | EXPORT_SYMBOL(system_state); |
122 | ||
123 | /* | |
124 | * Boot command-line arguments | |
125 | */ | |
126 | #define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT | |
127 | #define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT | |
128 | ||
129 | extern void time_init(void); | |
130 | /* Default late time init is NULL. archs can override this later. */ | |
d2e3192b | 131 | void (*__initdata late_time_init)(void); |
1da177e4 | 132 | |
30d7e0d4 ABL |
133 | /* Untouched command line saved by arch-specific code. */ |
134 | char __initdata boot_command_line[COMMAND_LINE_SIZE]; | |
135 | /* Untouched saved command line (eg. for /proc) */ | |
136 | char *saved_command_line; | |
137 | /* Command line for parameter parsing */ | |
138 | static char *static_command_line; | |
08746a65 KM |
139 | /* Command line for per-initcall parameter parsing */ |
140 | static char *initcall_command_line; | |
1da177e4 LT |
141 | |
142 | static char *execute_command; | |
ffdfc409 | 143 | static char *ramdisk_execute_command; |
1da177e4 | 144 | |
c4b2c0c5 HFS |
145 | /* |
146 | * Used to generate warnings if static_key manipulation functions are used | |
147 | * before jump_label_init is called. | |
148 | */ | |
dd4d9fec | 149 | bool static_key_initialized __read_mostly; |
c4b2c0c5 HFS |
150 | EXPORT_SYMBOL_GPL(static_key_initialized); |
151 | ||
8b3b2955 JB |
152 | /* |
153 | * If set, this is an indication to the drivers that reset the underlying | |
154 | * device before going ahead with the initialization otherwise driver might | |
155 | * rely on the BIOS and skip the reset operation. | |
156 | * | |
157 | * This is useful if kernel is booting in an unreliable environment. | |
fc454fdc | 158 | * For ex. kdump situation where previous kernel has crashed, BIOS has been |
8b3b2955 JB |
159 | * skipped and devices will be in unknown state. |
160 | */ | |
161 | unsigned int reset_devices; | |
162 | EXPORT_SYMBOL(reset_devices); | |
1da177e4 | 163 | |
7e96287d VG |
164 | static int __init set_reset_devices(char *str) |
165 | { | |
166 | reset_devices = 1; | |
167 | return 1; | |
168 | } | |
169 | ||
170 | __setup("reset_devices", set_reset_devices); | |
171 | ||
dd4d9fec FF |
172 | static const char *argv_init[MAX_INIT_ARGS+2] = { "init", NULL, }; |
173 | const char *envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, }; | |
1da177e4 LT |
174 | static const char *panic_later, *panic_param; |
175 | ||
914dcaa8 | 176 | extern const struct obs_kernel_param __setup_start[], __setup_end[]; |
1da177e4 | 177 | |
31c025b5 | 178 | static bool __init obsolete_checksetup(char *line) |
1da177e4 | 179 | { |
914dcaa8 | 180 | const struct obs_kernel_param *p; |
31c025b5 | 181 | bool had_early_param = false; |
1da177e4 LT |
182 | |
183 | p = __setup_start; | |
184 | do { | |
185 | int n = strlen(p->str); | |
b1e4d20c | 186 | if (parameqn(line, p->str, n)) { |
1da177e4 | 187 | if (p->early) { |
33df0d19 RR |
188 | /* Already done in parse_early_param? |
189 | * (Needs exact match on param part). | |
190 | * Keep iterating, as we can have early | |
191 | * params and __setups of same names 8( */ | |
1da177e4 | 192 | if (line[n] == '\0' || line[n] == '=') |
31c025b5 | 193 | had_early_param = true; |
1da177e4 | 194 | } else if (!p->setup_func) { |
ea676e84 AM |
195 | pr_warn("Parameter %s is obsolete, ignored\n", |
196 | p->str); | |
31c025b5 | 197 | return true; |
1da177e4 | 198 | } else if (p->setup_func(line + n)) |
31c025b5 | 199 | return true; |
1da177e4 LT |
200 | } |
201 | p++; | |
202 | } while (p < __setup_end); | |
33df0d19 RR |
203 | |
204 | return had_early_param; | |
1da177e4 LT |
205 | } |
206 | ||
207 | /* | |
208 | * This should be approx 2 Bo*oMips to start (note initial shift), and will | |
209 | * still work even if initially too large, it will just take slightly longer | |
210 | */ | |
211 | unsigned long loops_per_jiffy = (1<<12); | |
1da177e4 LT |
212 | EXPORT_SYMBOL(loops_per_jiffy); |
213 | ||
214 | static int __init debug_kernel(char *str) | |
215 | { | |
a8fe19eb | 216 | console_loglevel = CONSOLE_LOGLEVEL_DEBUG; |
f6f21c81 | 217 | return 0; |
1da177e4 LT |
218 | } |
219 | ||
220 | static int __init quiet_kernel(char *str) | |
221 | { | |
a8fe19eb | 222 | console_loglevel = CONSOLE_LOGLEVEL_QUIET; |
f6f21c81 | 223 | return 0; |
1da177e4 LT |
224 | } |
225 | ||
f6f21c81 YL |
226 | early_param("debug", debug_kernel); |
227 | early_param("quiet", quiet_kernel); | |
1da177e4 LT |
228 | |
229 | static int __init loglevel(char *str) | |
230 | { | |
808bf29b AS |
231 | int newlevel; |
232 | ||
233 | /* | |
234 | * Only update loglevel value when a correct setting was passed, | |
235 | * to prevent blind crashes (when loglevel being set to 0) that | |
236 | * are quite hard to debug | |
237 | */ | |
238 | if (get_option(&str, &newlevel)) { | |
239 | console_loglevel = newlevel; | |
240 | return 0; | |
241 | } | |
242 | ||
243 | return -EINVAL; | |
1da177e4 LT |
244 | } |
245 | ||
f6f21c81 | 246 | early_param("loglevel", loglevel); |
1da177e4 | 247 | |
a99cd112 | 248 | /* Change NUL term back to "=", to make "param" the whole string. */ |
ecc86170 LR |
249 | static int __init repair_env_string(char *param, char *val, |
250 | const char *unused, void *arg) | |
1da177e4 | 251 | { |
1da177e4 LT |
252 | if (val) { |
253 | /* param=val or param="val"? */ | |
254 | if (val == param+strlen(param)+1) | |
255 | val[-1] = '='; | |
256 | else if (val == param+strlen(param)+2) { | |
257 | val[-2] = '='; | |
258 | memmove(val-1, val, strlen(val)+1); | |
259 | val--; | |
260 | } else | |
261 | BUG(); | |
262 | } | |
a99cd112 CM |
263 | return 0; |
264 | } | |
265 | ||
51e158c1 | 266 | /* Anything after -- gets handed straight to init. */ |
ecc86170 LR |
267 | static int __init set_init_arg(char *param, char *val, |
268 | const char *unused, void *arg) | |
51e158c1 RR |
269 | { |
270 | unsigned int i; | |
271 | ||
272 | if (panic_later) | |
273 | return 0; | |
274 | ||
ecc86170 | 275 | repair_env_string(param, val, unused, NULL); |
51e158c1 RR |
276 | |
277 | for (i = 0; argv_init[i]; i++) { | |
278 | if (i == MAX_INIT_ARGS) { | |
279 | panic_later = "init"; | |
280 | panic_param = param; | |
281 | return 0; | |
282 | } | |
283 | } | |
284 | argv_init[i] = param; | |
285 | return 0; | |
286 | } | |
287 | ||
a99cd112 CM |
288 | /* |
289 | * Unknown boot options get handed to init, unless they look like | |
290 | * unused parameters (modprobe will find them in /proc/cmdline). | |
291 | */ | |
ecc86170 LR |
292 | static int __init unknown_bootoption(char *param, char *val, |
293 | const char *unused, void *arg) | |
a99cd112 | 294 | { |
ecc86170 | 295 | repair_env_string(param, val, unused, NULL); |
1da177e4 LT |
296 | |
297 | /* Handle obsolete-style parameters */ | |
298 | if (obsolete_checksetup(param)) | |
299 | return 0; | |
300 | ||
f066a4f6 RR |
301 | /* Unused module parameter. */ |
302 | if (strchr(param, '.') && (!val || strchr(param, '.') < val)) | |
1da177e4 | 303 | return 0; |
1da177e4 LT |
304 | |
305 | if (panic_later) | |
306 | return 0; | |
307 | ||
308 | if (val) { | |
309 | /* Environment option */ | |
310 | unsigned int i; | |
311 | for (i = 0; envp_init[i]; i++) { | |
312 | if (i == MAX_INIT_ENVS) { | |
499a4584 | 313 | panic_later = "env"; |
1da177e4 LT |
314 | panic_param = param; |
315 | } | |
316 | if (!strncmp(param, envp_init[i], val - param)) | |
317 | break; | |
318 | } | |
319 | envp_init[i] = param; | |
320 | } else { | |
321 | /* Command line option */ | |
322 | unsigned int i; | |
323 | for (i = 0; argv_init[i]; i++) { | |
324 | if (i == MAX_INIT_ARGS) { | |
499a4584 | 325 | panic_later = "init"; |
1da177e4 LT |
326 | panic_param = param; |
327 | } | |
328 | } | |
329 | argv_init[i] = param; | |
330 | } | |
331 | return 0; | |
332 | } | |
333 | ||
334 | static int __init init_setup(char *str) | |
335 | { | |
336 | unsigned int i; | |
337 | ||
338 | execute_command = str; | |
339 | /* | |
340 | * In case LILO is going to boot us with default command line, | |
341 | * it prepends "auto" before the whole cmdline which makes | |
342 | * the shell think it should execute a script with such name. | |
343 | * So we ignore all arguments entered _before_ init=... [MJ] | |
344 | */ | |
345 | for (i = 1; i < MAX_INIT_ARGS; i++) | |
346 | argv_init[i] = NULL; | |
347 | return 1; | |
348 | } | |
349 | __setup("init=", init_setup); | |
350 | ||
ffdfc409 OJ |
351 | static int __init rdinit_setup(char *str) |
352 | { | |
353 | unsigned int i; | |
354 | ||
355 | ramdisk_execute_command = str; | |
356 | /* See "auto" comment in init_setup */ | |
357 | for (i = 1; i < MAX_INIT_ARGS; i++) | |
358 | argv_init[i] = NULL; | |
359 | return 1; | |
360 | } | |
361 | __setup("rdinit=", rdinit_setup); | |
362 | ||
1da177e4 | 363 | #ifndef CONFIG_SMP |
34db18a0 | 364 | static const unsigned int setup_max_cpus = NR_CPUS; |
e0982e90 | 365 | static inline void setup_nr_cpu_ids(void) { } |
1da177e4 | 366 | static inline void smp_prepare_cpus(unsigned int maxcpus) { } |
1da177e4 LT |
367 | #endif |
368 | ||
30d7e0d4 ABL |
369 | /* |
370 | * We need to store the untouched command line for future reference. | |
371 | * We also need to store the touched command line since the parameter | |
372 | * parsing is performed in place, and we should allow a component to | |
373 | * store reference of name/value for future reference. | |
374 | */ | |
375 | static void __init setup_command_line(char *command_line) | |
376 | { | |
098b081b | 377 | saved_command_line = |
7e1c4e27 | 378 | memblock_alloc(strlen(boot_command_line) + 1, SMP_CACHE_BYTES); |
098b081b | 379 | initcall_command_line = |
7e1c4e27 MR |
380 | memblock_alloc(strlen(boot_command_line) + 1, SMP_CACHE_BYTES); |
381 | static_command_line = memblock_alloc(strlen(command_line) + 1, | |
382 | SMP_CACHE_BYTES); | |
dd4d9fec FF |
383 | strcpy(saved_command_line, boot_command_line); |
384 | strcpy(static_command_line, command_line); | |
30d7e0d4 ABL |
385 | } |
386 | ||
1da177e4 LT |
387 | /* |
388 | * We need to finalize in a non-__init function or else race conditions | |
389 | * between the root thread and the init thread may cause start_kernel to | |
390 | * be reaped by free_initmem before the root thread has proceeded to | |
391 | * cpu_idle. | |
392 | * | |
393 | * gcc-3.4 accidentally inlines this function, so use noinline. | |
394 | */ | |
395 | ||
b433c3d4 PZ |
396 | static __initdata DECLARE_COMPLETION(kthreadd_done); |
397 | ||
53c99bd6 | 398 | noinline void __ref rest_init(void) |
1da177e4 | 399 | { |
8fb12156 | 400 | struct task_struct *tsk; |
73c27992 EB |
401 | int pid; |
402 | ||
7db905e6 | 403 | rcu_scheduler_starting(); |
b433c3d4 | 404 | /* |
97158569 | 405 | * We need to spawn init first so that it obtains pid 1, however |
b433c3d4 PZ |
406 | * the init task will end up wanting to create kthreads, which, if |
407 | * we schedule it before we create kthreadd, will OOPS. | |
408 | */ | |
8fb12156 TG |
409 | pid = kernel_thread(kernel_init, NULL, CLONE_FS); |
410 | /* | |
411 | * Pin init on the boot CPU. Task migration is not properly working | |
412 | * until sched_init_smp() has been run. It will set the allowed | |
413 | * CPUs for init to the non isolated CPUs. | |
414 | */ | |
415 | rcu_read_lock(); | |
416 | tsk = find_task_by_pid_ns(pid, &init_pid_ns); | |
417 | set_cpus_allowed_ptr(tsk, cpumask_of(smp_processor_id())); | |
418 | rcu_read_unlock(); | |
419 | ||
1da177e4 | 420 | numa_default_policy(); |
73c27992 | 421 | pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES); |
d11c563d | 422 | rcu_read_lock(); |
5cd20455 | 423 | kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns); |
d11c563d | 424 | rcu_read_unlock(); |
1c3c5eab TG |
425 | |
426 | /* | |
427 | * Enable might_sleep() and smp_processor_id() checks. | |
f142f08b | 428 | * They cannot be enabled earlier because with CONFIG_PREEMPT=y |
1c3c5eab TG |
429 | * kernel_thread() would trigger might_sleep() splats. With |
430 | * CONFIG_PREEMPT_VOLUNTARY=y the init task might have scheduled | |
431 | * already, but it's stuck on the kthreadd_done completion. | |
432 | */ | |
433 | system_state = SYSTEM_SCHEDULING; | |
434 | ||
b433c3d4 | 435 | complete(&kthreadd_done); |
f340c0d1 IM |
436 | |
437 | /* | |
438 | * The boot idle thread must execute schedule() | |
1df21055 | 439 | * at least once to get things moving: |
f340c0d1 | 440 | */ |
bd2f5536 | 441 | schedule_preempt_disabled(); |
5bfb5d69 | 442 | /* Call into cpu_idle with preempt disabled */ |
a1a04ec3 | 443 | cpu_startup_entry(CPUHP_ONLINE); |
1df21055 | 444 | } |
1da177e4 LT |
445 | |
446 | /* Check for early params. */ | |
ecc86170 LR |
447 | static int __init do_early_param(char *param, char *val, |
448 | const char *unused, void *arg) | |
1da177e4 | 449 | { |
914dcaa8 | 450 | const struct obs_kernel_param *p; |
1da177e4 LT |
451 | |
452 | for (p = __setup_start; p < __setup_end; p++) { | |
b1e4d20c | 453 | if ((p->early && parameq(param, p->str)) || |
18a8bd94 YL |
454 | (strcmp(param, "console") == 0 && |
455 | strcmp(p->str, "earlycon") == 0) | |
456 | ) { | |
1da177e4 | 457 | if (p->setup_func(val) != 0) |
ea676e84 | 458 | pr_warn("Malformed early option '%s'\n", param); |
1da177e4 LT |
459 | } |
460 | } | |
461 | /* We accept everything at this stage. */ | |
462 | return 0; | |
463 | } | |
464 | ||
13977091 MD |
465 | void __init parse_early_options(char *cmdline) |
466 | { | |
ecc86170 LR |
467 | parse_args("early options", cmdline, NULL, 0, 0, 0, NULL, |
468 | do_early_param); | |
13977091 MD |
469 | } |
470 | ||
1da177e4 LT |
471 | /* Arch code calls this early on, or if not, just before other parsing. */ |
472 | void __init parse_early_param(void) | |
473 | { | |
dd4d9fec FF |
474 | static int done __initdata; |
475 | static char tmp_cmdline[COMMAND_LINE_SIZE] __initdata; | |
1da177e4 LT |
476 | |
477 | if (done) | |
478 | return; | |
479 | ||
480 | /* All fall through to do_early_param. */ | |
30d7e0d4 | 481 | strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE); |
13977091 | 482 | parse_early_options(tmp_cmdline); |
1da177e4 LT |
483 | done = 1; |
484 | } | |
485 | ||
e7ff3a47 TG |
486 | void __init __weak arch_post_acpi_subsys_init(void) { } |
487 | ||
839ad62e | 488 | void __init __weak smp_setup_processor_id(void) |
033ab7f8 AM |
489 | { |
490 | } | |
491 | ||
eded09cc | 492 | # if THREAD_SIZE >= PAGE_SIZE |
b235beea | 493 | void __init __weak thread_stack_cache_init(void) |
8c9843e5 BH |
494 | { |
495 | } | |
eded09cc | 496 | #endif |
8c9843e5 | 497 | |
c7753208 TL |
498 | void __init __weak mem_encrypt_init(void) { } |
499 | ||
4e37958d SRV |
500 | bool initcall_debug; |
501 | core_param(initcall_debug, initcall_debug, bool, 0644); | |
b0dc52f1 SRV |
502 | |
503 | #ifdef TRACEPOINTS_ENABLED | |
4e37958d | 504 | static void __init initcall_debug_enable(void); |
b0dc52f1 SRV |
505 | #else |
506 | static inline void initcall_debug_enable(void) | |
507 | { | |
508 | } | |
509 | #endif | |
4e37958d | 510 | |
444f478f PE |
511 | /* |
512 | * Set up kernel memory allocators | |
513 | */ | |
514 | static void __init mm_init(void) | |
515 | { | |
eefa864b JK |
516 | /* |
517 | * page_ext requires contiguous pages, | |
518 | * bigger than MAX_ORDER unless SPARSEMEM. | |
519 | */ | |
520 | page_ext_init_flatmem(); | |
444f478f PE |
521 | mem_init(); |
522 | kmem_cache_init(); | |
b35f1819 | 523 | pgtable_init(); |
444f478f | 524 | vmalloc_init(); |
0ddab1d2 | 525 | ioremap_huge_init(); |
613e396b TG |
526 | /* Should be run before the first non-init thread is created */ |
527 | init_espfix_bsp(); | |
aa8c6248 TG |
528 | /* Should be run after espfix64 is set up. */ |
529 | pti_init(); | |
444f478f PE |
530 | } |
531 | ||
53c99bd6 MS |
532 | void __init __weak arch_call_rest_init(void) |
533 | { | |
534 | rest_init(); | |
535 | } | |
536 | ||
722a9f92 | 537 | asmlinkage __visible void __init start_kernel(void) |
1da177e4 | 538 | { |
dd4d9fec FF |
539 | char *command_line; |
540 | char *after_dashes; | |
033ab7f8 | 541 | |
d4311ff1 | 542 | set_task_stack_end_magic(&init_task); |
73839c5b | 543 | smp_setup_processor_id(); |
3ac7fe5a | 544 | debug_objects_early_init(); |
42059429 | 545 | |
ddbcc7e8 | 546 | cgroup_init_early(); |
fbb9ce95 IM |
547 | |
548 | local_irq_disable(); | |
2ce802f6 | 549 | early_boot_irqs_disabled = true; |
fbb9ce95 | 550 | |
1b3b3b49 VK |
551 | /* |
552 | * Interrupts are still disabled. Do necessary setups, then | |
553 | * enable them. | |
554 | */ | |
44fd2299 | 555 | boot_cpu_init(); |
1da177e4 | 556 | page_address_init(); |
ea676e84 | 557 | pr_notice("%s", linux_banner); |
1da177e4 | 558 | setup_arch(&command_line); |
121388a3 LA |
559 | /* |
560 | * Set up the the initial canary and entropy after arch | |
33d72f38 | 561 | * and after adding latent and command line entropy. |
121388a3 LA |
562 | */ |
563 | add_latent_entropy(); | |
33d72f38 | 564 | add_device_randomness(command_line, strlen(command_line)); |
121388a3 | 565 | boot_init_stack_canary(); |
6345d24d | 566 | mm_init_cpumask(&init_mm); |
30d7e0d4 | 567 | setup_command_line(command_line); |
e0982e90 | 568 | setup_nr_cpu_ids(); |
d6647bdf | 569 | setup_per_cpu_areas(); |
44fd2299 | 570 | smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */ |
b5b1404d | 571 | boot_cpu_hotplug_init(); |
1da177e4 | 572 | |
72675e13 | 573 | build_all_zonelists(NULL); |
83b519e8 PE |
574 | page_alloc_init(); |
575 | ||
ea676e84 | 576 | pr_notice("Kernel command line: %s\n", boot_command_line); |
83b519e8 | 577 | parse_early_param(); |
51e158c1 RR |
578 | after_dashes = parse_args("Booting kernel", |
579 | static_command_line, __start___param, | |
580 | __stop___param - __start___param, | |
ecc86170 | 581 | -1, -1, NULL, &unknown_bootoption); |
3438cf54 | 582 | if (!IS_ERR_OR_NULL(after_dashes)) |
51e158c1 | 583 | parse_args("Setting init args", after_dashes, NULL, 0, -1, -1, |
ecc86170 | 584 | NULL, set_init_arg); |
97ce2c88 JF |
585 | |
586 | jump_label_init(); | |
587 | ||
83b519e8 PE |
588 | /* |
589 | * These use large bootmem allocations and must precede | |
590 | * kmem_cache_init() | |
591 | */ | |
162a7e75 | 592 | setup_log_buf(0); |
83b519e8 PE |
593 | vfs_caches_init_early(); |
594 | sort_main_extable(); | |
595 | trap_init(); | |
444f478f | 596 | mm_init(); |
de03c72c | 597 | |
f631718d SRV |
598 | ftrace_init(); |
599 | ||
e725c731 SRV |
600 | /* trace_printk can be enabled here */ |
601 | early_trace_init(); | |
602 | ||
1da177e4 LT |
603 | /* |
604 | * Set up the scheduler prior starting any interrupts (such as the | |
605 | * timer interrupt). Full topology setup happens at smp_init() | |
606 | * time - but meanwhile we still have a functioning scheduler. | |
607 | */ | |
608 | sched_init(); | |
609 | /* | |
610 | * Disable preemption - early bootup scheduling is extremely | |
611 | * fragile until we cpu_idle() for the first time. | |
612 | */ | |
613 | preempt_disable(); | |
dd4d9fec FF |
614 | if (WARN(!irqs_disabled(), |
615 | "Interrupts were enabled *very* early, fixing it\n")) | |
c4a68306 | 616 | local_irq_disable(); |
0a835c4f | 617 | radix_tree_init(); |
3347fa09 | 618 | |
7d229c66 TS |
619 | /* |
620 | * Set up housekeeping before setting up workqueues to allow the unbound | |
621 | * workqueue to take non-housekeeping into account. | |
622 | */ | |
623 | housekeeping_init(); | |
624 | ||
3347fa09 TH |
625 | /* |
626 | * Allow workqueue creation and work item queueing/cancelling | |
627 | * early. Work item execution depends on kthreads and starts after | |
628 | * workqueue_init(). | |
629 | */ | |
630 | workqueue_init_early(); | |
631 | ||
1da177e4 | 632 | rcu_init(); |
5f893b26 | 633 | |
e725c731 | 634 | /* Trace events are available after this */ |
5f893b26 SRRH |
635 | trace_init(); |
636 | ||
4e37958d SRV |
637 | if (initcall_debug) |
638 | initcall_debug_enable(); | |
639 | ||
65f382fd | 640 | context_tracking_init(); |
0b8f1efa YL |
641 | /* init some links before init_ISA_irqs() */ |
642 | early_irq_init(); | |
1da177e4 | 643 | init_IRQ(); |
ad2b1353 | 644 | tick_init(); |
d6dd50e0 | 645 | rcu_init_nohz(); |
1da177e4 | 646 | init_timers(); |
c0a31329 | 647 | hrtimers_init(); |
1da177e4 | 648 | softirq_init(); |
ad596171 | 649 | timekeeping_init(); |
88fecaa2 | 650 | time_init(); |
f92bac3b | 651 | printk_safe_init(); |
9e630205 | 652 | perf_event_init(); |
93e02814 | 653 | profile_init(); |
d8ad7d11 | 654 | call_function_init(); |
f91eb62f | 655 | WARN(!irqs_disabled(), "Interrupts were enabled early\n"); |
c3bc8fd6 | 656 | |
2ce802f6 | 657 | early_boot_irqs_disabled = false; |
93e02814 | 658 | local_irq_enable(); |
dcce284a | 659 | |
7e85ee0c | 660 | kmem_cache_init_late(); |
1da177e4 LT |
661 | |
662 | /* | |
663 | * HACK ALERT! This is early. We're enabling the console before | |
664 | * we've done PCI setups etc, and console_init() must be aware of | |
665 | * this. But we do want output early, in case something goes wrong. | |
666 | */ | |
667 | console_init(); | |
668 | if (panic_later) | |
499a4584 TH |
669 | panic("Too many boot %s vars at `%s'", panic_later, |
670 | panic_param); | |
fbb9ce95 | 671 | |
c3bc8fd6 | 672 | lockdep_init(); |
fbb9ce95 | 673 | |
9a11b49a IM |
674 | /* |
675 | * Need to run this when irqs are enabled, because it wants | |
676 | * to self-test [hard/soft]-irqs on/off lock inversion bugs | |
677 | * too: | |
678 | */ | |
679 | locking_selftest(); | |
680 | ||
c7753208 TL |
681 | /* |
682 | * This needs to be called before any devices perform DMA | |
683 | * operations that might use the SWIOTLB bounce buffers. It will | |
684 | * mark the bounce buffers as decrypted so that their usage will | |
685 | * not cause "plain-text" data to be decrypted when accessed. | |
686 | */ | |
687 | mem_encrypt_init(); | |
688 | ||
1da177e4 LT |
689 | #ifdef CONFIG_BLK_DEV_INITRD |
690 | if (initrd_start && !initrd_below_start_ok && | |
bd673c7c | 691 | page_to_pfn(virt_to_page((void *)initrd_start)) < min_low_pfn) { |
ea676e84 | 692 | pr_crit("initrd overwritten (0x%08lx < 0x%08lx) - disabling it.\n", |
bd673c7c GU |
693 | page_to_pfn(virt_to_page((void *)initrd_start)), |
694 | min_low_pfn); | |
1da177e4 LT |
695 | initrd_start = 0; |
696 | } | |
697 | #endif | |
fe53ca54 | 698 | page_ext_init(); |
9b090f2d | 699 | kmemleak_init(); |
caba4cbb | 700 | debug_objects_mem_init(); |
e7c8d5c9 | 701 | setup_per_cpu_pageset(); |
1da177e4 | 702 | numa_policy_init(); |
9c71206d | 703 | acpi_early_init(); |
1da177e4 LT |
704 | if (late_time_init) |
705 | late_time_init(); | |
857baa87 | 706 | sched_clock_init(); |
1da177e4 | 707 | calibrate_delay(); |
95846ecf | 708 | pid_idr_init(); |
1da177e4 | 709 | anon_vma_init(); |
11520e5e | 710 | #ifdef CONFIG_X86 |
83e68189 | 711 | if (efi_enabled(EFI_RUNTIME_SERVICES)) |
11520e5e LT |
712 | efi_enter_virtual_mode(); |
713 | #endif | |
b235beea | 714 | thread_stack_cache_init(); |
d84f4f99 | 715 | cred_init(); |
ff691f6e | 716 | fork_init(); |
1da177e4 | 717 | proc_caches_init(); |
3ea056c5 | 718 | uts_ns_init(); |
1da177e4 | 719 | buffer_init(); |
1da177e4 LT |
720 | key_init(); |
721 | security_init(); | |
0b4b3827 | 722 | dbg_late_init(); |
4248b0da | 723 | vfs_caches_init(); |
62906027 | 724 | pagecache_init(); |
1da177e4 | 725 | signals_init(); |
09652320 | 726 | seq_file_init(); |
1da177e4 | 727 | proc_root_init(); |
e149ed2b | 728 | nsfs_init(); |
1da177e4 | 729 | cpuset_init(); |
695df213 | 730 | cgroup_init(); |
c757249a | 731 | taskstats_init_early(); |
ca74e92b | 732 | delayacct_init(); |
1da177e4 LT |
733 | |
734 | check_bugs(); | |
735 | ||
b064a8fa | 736 | acpi_subsystem_init(); |
e7ff3a47 | 737 | arch_post_acpi_subsys_init(); |
6ae6996a | 738 | sfi_init_late(); |
1da177e4 LT |
739 | |
740 | /* Do the rest non-__init'ed, we're now alive */ | |
53c99bd6 | 741 | arch_call_rest_init(); |
1da177e4 LT |
742 | } |
743 | ||
b99b87f7 PO |
744 | /* Call all constructor functions linked into the kernel. */ |
745 | static void __init do_ctors(void) | |
746 | { | |
747 | #ifdef CONFIG_CONSTRUCTORS | |
196a15b4 | 748 | ctor_fn_t *fn = (ctor_fn_t *) __ctors_start; |
b99b87f7 | 749 | |
196a15b4 HS |
750 | for (; fn < (ctor_fn_t *) __ctors_end; fn++) |
751 | (*fn)(); | |
b99b87f7 PO |
752 | #endif |
753 | } | |
754 | ||
7b0b73d7 PB |
755 | #ifdef CONFIG_KALLSYMS |
756 | struct blacklist_entry { | |
757 | struct list_head next; | |
758 | char *buf; | |
759 | }; | |
760 | ||
761 | static __initdata_or_module LIST_HEAD(blacklisted_initcalls); | |
762 | ||
763 | static int __init initcall_blacklist(char *str) | |
764 | { | |
765 | char *str_entry; | |
766 | struct blacklist_entry *entry; | |
767 | ||
768 | /* str argument is a comma-separated list of functions */ | |
769 | do { | |
770 | str_entry = strsep(&str, ","); | |
771 | if (str_entry) { | |
772 | pr_debug("blacklisting initcall %s\n", str_entry); | |
7e1c4e27 MR |
773 | entry = memblock_alloc(sizeof(*entry), |
774 | SMP_CACHE_BYTES); | |
775 | entry->buf = memblock_alloc(strlen(str_entry) + 1, | |
776 | SMP_CACHE_BYTES); | |
7b0b73d7 PB |
777 | strcpy(entry->buf, str_entry); |
778 | list_add(&entry->next, &blacklisted_initcalls); | |
779 | } | |
780 | } while (str_entry); | |
781 | ||
782 | return 0; | |
783 | } | |
784 | ||
785 | static bool __init_or_module initcall_blacklisted(initcall_t fn) | |
786 | { | |
7b0b73d7 | 787 | struct blacklist_entry *entry; |
c8cdd2be | 788 | char fn_name[KSYM_SYMBOL_LEN]; |
0fd5ed8d | 789 | unsigned long addr; |
7b0b73d7 | 790 | |
c8cdd2be | 791 | if (list_empty(&blacklisted_initcalls)) |
7b0b73d7 PB |
792 | return false; |
793 | ||
0fd5ed8d RV |
794 | addr = (unsigned long) dereference_function_descriptor(fn); |
795 | sprint_symbol_no_offset(fn_name, addr); | |
c8cdd2be | 796 | |
841c06d7 PB |
797 | /* |
798 | * fn will be "function_name [module_name]" where [module_name] is not | |
799 | * displayed for built-in init functions. Strip off the [module_name]. | |
800 | */ | |
801 | strreplace(fn_name, ' ', '\0'); | |
802 | ||
e6fd1fb3 | 803 | list_for_each_entry(entry, &blacklisted_initcalls, next) { |
7b0b73d7 PB |
804 | if (!strcmp(fn_name, entry->buf)) { |
805 | pr_debug("initcall %s blacklisted\n", fn_name); | |
7b0b73d7 PB |
806 | return true; |
807 | } | |
808 | } | |
809 | ||
7b0b73d7 PB |
810 | return false; |
811 | } | |
812 | #else | |
813 | static int __init initcall_blacklist(char *str) | |
814 | { | |
815 | pr_warn("initcall_blacklist requires CONFIG_KALLSYMS\n"); | |
816 | return 0; | |
817 | } | |
818 | ||
819 | static bool __init_or_module initcall_blacklisted(initcall_t fn) | |
820 | { | |
821 | return false; | |
822 | } | |
823 | #endif | |
824 | __setup("initcall_blacklist=", initcall_blacklist); | |
825 | ||
4e37958d SRV |
826 | static __init_or_module void |
827 | trace_initcall_start_cb(void *data, initcall_t fn) | |
1da177e4 | 828 | { |
4e37958d | 829 | ktime_t *calltime = (ktime_t *)data; |
1da177e4 | 830 | |
d62cf815 | 831 | printk(KERN_DEBUG "calling %pF @ %i\n", fn, task_pid_nr(current)); |
4e37958d SRV |
832 | *calltime = ktime_get(); |
833 | } | |
834 | ||
835 | static __init_or_module void | |
836 | trace_initcall_finish_cb(void *data, initcall_t fn, int ret) | |
837 | { | |
838 | ktime_t *calltime = (ktime_t *)data; | |
839 | ktime_t delta, rettime; | |
840 | unsigned long long duration; | |
841 | ||
22c5c03b | 842 | rettime = ktime_get(); |
4e37958d | 843 | delta = ktime_sub(rettime, *calltime); |
22c5c03b | 844 | duration = (unsigned long long) ktime_to_ns(delta) >> 10; |
d62cf815 | 845 | printk(KERN_DEBUG "initcall %pF returned %d after %lld usecs\n", |
ea676e84 | 846 | fn, ret, duration); |
4e37958d | 847 | } |
1da177e4 | 848 | |
4e37958d SRV |
849 | static ktime_t initcall_calltime; |
850 | ||
b0dc52f1 | 851 | #ifdef TRACEPOINTS_ENABLED |
4e37958d SRV |
852 | static void __init initcall_debug_enable(void) |
853 | { | |
854 | int ret; | |
855 | ||
856 | ret = register_trace_initcall_start(trace_initcall_start_cb, | |
857 | &initcall_calltime); | |
858 | ret |= register_trace_initcall_finish(trace_initcall_finish_cb, | |
859 | &initcall_calltime); | |
860 | WARN(ret, "Failed to register initcall tracepoints\n"); | |
22c5c03b | 861 | } |
b0dc52f1 SRV |
862 | # define do_trace_initcall_start trace_initcall_start |
863 | # define do_trace_initcall_finish trace_initcall_finish | |
864 | #else | |
865 | static inline void do_trace_initcall_start(initcall_t fn) | |
866 | { | |
867 | if (!initcall_debug) | |
868 | return; | |
869 | trace_initcall_start_cb(&initcall_calltime, fn); | |
870 | } | |
871 | static inline void do_trace_initcall_finish(initcall_t fn, int ret) | |
872 | { | |
873 | if (!initcall_debug) | |
874 | return; | |
875 | trace_initcall_finish_cb(&initcall_calltime, fn, ret); | |
876 | } | |
877 | #endif /* !TRACEPOINTS_ENABLED */ | |
22c5c03b | 878 | |
e4461271 | 879 | int __init_or_module do_one_initcall(initcall_t fn) |
22c5c03b KW |
880 | { |
881 | int count = preempt_count(); | |
ff1c8fac | 882 | char msgbuf[64]; |
4e37958d | 883 | int ret; |
22c5c03b | 884 | |
7b0b73d7 PB |
885 | if (initcall_blacklisted(fn)) |
886 | return -EPERM; | |
887 | ||
b0dc52f1 | 888 | do_trace_initcall_start(fn); |
4e37958d | 889 | ret = fn(); |
b0dc52f1 | 890 | do_trace_initcall_finish(fn, ret); |
8f0c45cd | 891 | |
e0df154f | 892 | msgbuf[0] = 0; |
e662e1cf | 893 | |
e0df154f | 894 | if (preempt_count() != count) { |
bf5d770b | 895 | sprintf(msgbuf, "preemption imbalance "); |
4a2b4b22 | 896 | preempt_count_set(count); |
1da177e4 | 897 | } |
e0df154f | 898 | if (irqs_disabled()) { |
a76bfd0d | 899 | strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf)); |
e0df154f LT |
900 | local_irq_enable(); |
901 | } | |
f91eb62f | 902 | WARN(msgbuf[0], "initcall %pF returned with %s\n", fn, msgbuf); |
59f9415f | 903 | |
38addce8 | 904 | add_latent_entropy(); |
30dbb20e | 905 | return ret; |
e0df154f LT |
906 | } |
907 | ||
908 | ||
1b1eeca7 AB |
909 | extern initcall_entry_t __initcall_start[]; |
910 | extern initcall_entry_t __initcall0_start[]; | |
911 | extern initcall_entry_t __initcall1_start[]; | |
912 | extern initcall_entry_t __initcall2_start[]; | |
913 | extern initcall_entry_t __initcall3_start[]; | |
914 | extern initcall_entry_t __initcall4_start[]; | |
915 | extern initcall_entry_t __initcall5_start[]; | |
916 | extern initcall_entry_t __initcall6_start[]; | |
917 | extern initcall_entry_t __initcall7_start[]; | |
918 | extern initcall_entry_t __initcall_end[]; | |
919 | ||
920 | static initcall_entry_t *initcall_levels[] __initdata = { | |
026cee00 PM |
921 | __initcall0_start, |
922 | __initcall1_start, | |
923 | __initcall2_start, | |
924 | __initcall3_start, | |
925 | __initcall4_start, | |
926 | __initcall5_start, | |
927 | __initcall6_start, | |
928 | __initcall7_start, | |
929 | __initcall_end, | |
930 | }; | |
931 | ||
96263d28 | 932 | /* Keep these in sync with initcalls in include/linux/init.h */ |
026cee00 | 933 | static char *initcall_level_names[] __initdata = { |
a6fb6012 | 934 | "pure", |
9fb48c74 JC |
935 | "core", |
936 | "postcore", | |
937 | "arch", | |
938 | "subsys", | |
939 | "fs", | |
940 | "device", | |
941 | "late", | |
026cee00 PM |
942 | }; |
943 | ||
026cee00 | 944 | static void __init do_initcall_level(int level) |
e0df154f | 945 | { |
1b1eeca7 | 946 | initcall_entry_t *fn; |
e0df154f | 947 | |
08746a65 | 948 | strcpy(initcall_command_line, saved_command_line); |
026cee00 | 949 | parse_args(initcall_level_names[level], |
08746a65 | 950 | initcall_command_line, __start___param, |
026cee00 PM |
951 | __stop___param - __start___param, |
952 | level, level, | |
ecc86170 | 953 | NULL, &repair_env_string); |
026cee00 | 954 | |
4ee7c60d | 955 | trace_initcall_level(initcall_level_names[level]); |
026cee00 | 956 | for (fn = initcall_levels[level]; fn < initcall_levels[level+1]; fn++) |
1b1eeca7 | 957 | do_one_initcall(initcall_from_entry(fn)); |
1da177e4 LT |
958 | } |
959 | ||
026cee00 PM |
960 | static void __init do_initcalls(void) |
961 | { | |
962 | int level; | |
963 | ||
19efb72f | 964 | for (level = 0; level < ARRAY_SIZE(initcall_levels) - 1; level++) |
026cee00 PM |
965 | do_initcall_level(level); |
966 | } | |
967 | ||
1da177e4 LT |
968 | /* |
969 | * Ok, the machine is now initialized. None of the devices | |
970 | * have been touched yet, but the CPU subsystem is up and | |
971 | * running, and memory and process management works. | |
972 | * | |
973 | * Now we can finally start doing some real work.. | |
974 | */ | |
975 | static void __init do_basic_setup(void) | |
976 | { | |
759ee091 | 977 | cpuset_init_smp(); |
41ffe5d5 | 978 | shmem_init(); |
1da177e4 | 979 | driver_init(); |
b04c3afb | 980 | init_irq_proc(); |
b99b87f7 | 981 | do_ctors(); |
d5767c53 | 982 | usermodehelper_enable(); |
b0f84374 | 983 | do_initcalls(); |
1da177e4 LT |
984 | } |
985 | ||
7babe8db | 986 | static void __init do_pre_smp_initcalls(void) |
c2147a50 | 987 | { |
1b1eeca7 | 988 | initcall_entry_t *fn; |
c2147a50 | 989 | |
4ee7c60d | 990 | trace_initcall_level("early"); |
026cee00 | 991 | for (fn = __initcall_start; fn < __initcall0_start; fn++) |
1b1eeca7 | 992 | do_one_initcall(initcall_from_entry(fn)); |
c2147a50 EGM |
993 | } |
994 | ||
a74fb73c | 995 | static int run_init_process(const char *init_filename) |
1da177e4 LT |
996 | { |
997 | argv_init[0] = init_filename; | |
3f5c15d8 | 998 | pr_info("Run %s as init process\n", init_filename); |
c4ad8f98 | 999 | return do_execve(getname_kernel(init_filename), |
ae903caa AV |
1000 | (const char __user *const __user *)argv_init, |
1001 | (const char __user *const __user *)envp_init); | |
1da177e4 LT |
1002 | } |
1003 | ||
ba24762b MO |
1004 | static int try_to_run_init_process(const char *init_filename) |
1005 | { | |
1006 | int ret; | |
1007 | ||
1008 | ret = run_init_process(init_filename); | |
1009 | ||
1010 | if (ret && ret != -ENOENT) { | |
1011 | pr_err("Starting init: %s exists but couldn't execute it (error %d)\n", | |
1012 | init_filename, ret); | |
1013 | } | |
1014 | ||
1015 | return ret; | |
1016 | } | |
1017 | ||
f80b0c90 | 1018 | static noinline void __init kernel_init_freeable(void); |
d6b21238 | 1019 | |
0f5bf6d0 | 1020 | #if defined(CONFIG_STRICT_KERNEL_RWX) || defined(CONFIG_STRICT_MODULE_RWX) |
39290b38 | 1021 | bool rodata_enabled __ro_after_init = true; |
d2aa1aca KC |
1022 | static int __init set_debug_rodata(char *str) |
1023 | { | |
1024 | return strtobool(str, &rodata_enabled); | |
1025 | } | |
1026 | __setup("rodata=", set_debug_rodata); | |
39290b38 | 1027 | #endif |
d2aa1aca | 1028 | |
0f5bf6d0 | 1029 | #ifdef CONFIG_STRICT_KERNEL_RWX |
d2aa1aca KC |
1030 | static void mark_readonly(void) |
1031 | { | |
2959a5f7 | 1032 | if (rodata_enabled) { |
ae646f0b | 1033 | /* |
ba180314 PM |
1034 | * load_module() results in W+X mappings, which are cleaned |
1035 | * up with call_rcu(). Let's make sure that queued work is | |
ae646f0b JH |
1036 | * flushed so that we don't hit false positives looking for |
1037 | * insecure pages which are W+X. | |
1038 | */ | |
ba180314 | 1039 | rcu_barrier(); |
d2aa1aca | 1040 | mark_rodata_ro(); |
2959a5f7 JP |
1041 | rodata_test(); |
1042 | } else | |
d2aa1aca KC |
1043 | pr_info("Kernel memory protection disabled.\n"); |
1044 | } | |
1045 | #else | |
1046 | static inline void mark_readonly(void) | |
1047 | { | |
1048 | pr_warn("This architecture does not have kernel memory protection.\n"); | |
1049 | } | |
1050 | #endif | |
1051 | ||
d6b21238 | 1052 | static int __ref kernel_init(void *unused) |
ee5bfa64 | 1053 | { |
ba24762b MO |
1054 | int ret; |
1055 | ||
d6b21238 | 1056 | kernel_init_freeable(); |
22a9d645 AV |
1057 | /* need to finish all async __init code before freeing the memory */ |
1058 | async_synchronize_full(); | |
b80f0f6c | 1059 | ftrace_free_init_mem(); |
ee5bfa64 | 1060 | free_initmem(); |
d2aa1aca | 1061 | mark_readonly(); |
b976690f JR |
1062 | |
1063 | /* | |
1064 | * Kernel mappings are now finalized - update the userspace page-table | |
1065 | * to finalize PTI. | |
1066 | */ | |
1067 | pti_finalize(); | |
1068 | ||
ee5bfa64 VG |
1069 | system_state = SYSTEM_RUNNING; |
1070 | numa_default_policy(); | |
1071 | ||
967dcb8f PM |
1072 | rcu_end_inkernel_boot(); |
1073 | ||
ee5bfa64 | 1074 | if (ramdisk_execute_command) { |
ba24762b MO |
1075 | ret = run_init_process(ramdisk_execute_command); |
1076 | if (!ret) | |
a74fb73c | 1077 | return 0; |
ba24762b MO |
1078 | pr_err("Failed to execute %s (error %d)\n", |
1079 | ramdisk_execute_command, ret); | |
ee5bfa64 VG |
1080 | } |
1081 | ||
1082 | /* | |
1083 | * We try each of these until one succeeds. | |
1084 | * | |
1085 | * The Bourne shell can be used instead of init if we are | |
1086 | * trying to recover a really broken machine. | |
1087 | */ | |
1088 | if (execute_command) { | |
ba24762b MO |
1089 | ret = run_init_process(execute_command); |
1090 | if (!ret) | |
a74fb73c | 1091 | return 0; |
6ef4536e AL |
1092 | panic("Requested init %s failed (error %d).", |
1093 | execute_command, ret); | |
ee5bfa64 | 1094 | } |
ba24762b MO |
1095 | if (!try_to_run_init_process("/sbin/init") || |
1096 | !try_to_run_init_process("/etc/init") || | |
1097 | !try_to_run_init_process("/bin/init") || | |
1098 | !try_to_run_init_process("/bin/sh")) | |
a74fb73c | 1099 | return 0; |
ee5bfa64 | 1100 | |
ba24762b | 1101 | panic("No working init found. Try passing init= option to kernel. " |
8c27ceff | 1102 | "See Linux Documentation/admin-guide/init.rst for guidance."); |
ee5bfa64 VG |
1103 | } |
1104 | ||
f80b0c90 | 1105 | static noinline void __init kernel_init_freeable(void) |
1da177e4 | 1106 | { |
b433c3d4 PZ |
1107 | /* |
1108 | * Wait until kthreadd is all set-up. | |
1109 | */ | |
1110 | wait_for_completion(&kthreadd_done); | |
31a67102 LT |
1111 | |
1112 | /* Now the scheduler is fully set up and can do blocking allocations */ | |
1113 | gfp_allowed_mask = __GFP_BITS_MASK; | |
1114 | ||
58568d2a MX |
1115 | /* |
1116 | * init can allocate pages on any node | |
1117 | */ | |
3c466d46 | 1118 | set_mems_allowed(node_states[N_MEMORY]); |
1da177e4 | 1119 | |
9ec52099 CLG |
1120 | cad_pid = task_pid(current); |
1121 | ||
ca74a6f8 | 1122 | smp_prepare_cpus(setup_max_cpus); |
1da177e4 | 1123 | |
3347fa09 TH |
1124 | workqueue_init(); |
1125 | ||
597b7305 MH |
1126 | init_mm_internals(); |
1127 | ||
1da177e4 | 1128 | do_pre_smp_initcalls(); |
004417a6 | 1129 | lockup_detector_init(); |
1da177e4 | 1130 | |
1da177e4 LT |
1131 | smp_init(); |
1132 | sched_init_smp(); | |
1133 | ||
0e1cc95b MG |
1134 | page_alloc_init_late(); |
1135 | ||
1da177e4 LT |
1136 | do_basic_setup(); |
1137 | ||
2bd3a997 | 1138 | /* Open the /dev/console on the rootfs, this should never fail */ |
bae217ea | 1139 | if (ksys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0) |
ea676e84 | 1140 | pr_err("Warning: unable to open an initial console.\n"); |
2bd3a997 | 1141 | |
c7248321 DB |
1142 | (void) ksys_dup(0); |
1143 | (void) ksys_dup(0); | |
1da177e4 LT |
1144 | /* |
1145 | * check if there is an early userspace init. If yes, let it do all | |
1146 | * the work | |
1147 | */ | |
ffdfc409 OJ |
1148 | |
1149 | if (!ramdisk_execute_command) | |
1150 | ramdisk_execute_command = "/init"; | |
1151 | ||
cbfe20f5 DB |
1152 | if (ksys_access((const char __user *) |
1153 | ramdisk_execute_command, 0) != 0) { | |
ffdfc409 | 1154 | ramdisk_execute_command = NULL; |
1da177e4 | 1155 | prepare_namespace(); |
ffdfc409 | 1156 | } |
1da177e4 LT |
1157 | |
1158 | /* | |
1159 | * Ok, we have completed the initial bootup, and | |
1160 | * we're essentially up and running. Get rid of the | |
1161 | * initmem segments and start the user-mode stuff.. | |
c9cd2ce2 DK |
1162 | * |
1163 | * rootfs is available now, try loading the public keys | |
1164 | * and default modules | |
1da177e4 | 1165 | */ |
bb813f4c | 1166 | |
c9cd2ce2 | 1167 | integrity_load_keys(); |
1da177e4 | 1168 | } |