blob: 603b1ad88cb68b0d676af32d69d2d5bd0b5bafbd [file] [log] [blame]
rjw1f884582022-01-06 17:20:42 +08001/*
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
12#define DEBUG /* Enable initcall_debug */
13
14#include <linux/types.h>
15#include <linux/extable.h>
16#include <linux/module.h>
17#include <linux/proc_fs.h>
18#include <linux/binfmts.h>
19#include <linux/kernel.h>
20#include <linux/syscalls.h>
21#include <linux/stackprotector.h>
22#include <linux/string.h>
23#include <linux/ctype.h>
24#include <linux/delay.h>
25#include <linux/ioport.h>
26#include <linux/init.h>
27#include <linux/initrd.h>
28#include <linux/bootmem.h>
29#include <linux/acpi.h>
30#include <linux/console.h>
31#include <linux/nmi.h>
32#include <linux/percpu.h>
33#include <linux/kmod.h>
34#include <linux/vmalloc.h>
35#include <linux/kernel_stat.h>
36#include <linux/start_kernel.h>
37#include <linux/security.h>
38#include <linux/smp.h>
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>
46#include <linux/cgroup.h>
47#include <linux/efi.h>
48#include <linux/tick.h>
49#include <linux/interrupt.h>
50#include <linux/taskstats_kern.h>
51#include <linux/delayacct.h>
52#include <linux/unistd.h>
53#include <linux/rmap.h>
54#include <linux/mempolicy.h>
55#include <linux/key.h>
56#include <linux/buffer_head.h>
57#include <linux/page_ext.h>
58#include <linux/debug_locks.h>
59#include <linux/debugobjects.h>
60#include <linux/lockdep.h>
61#include <linux/kmemleak.h>
62#include <linux/pid_namespace.h>
63#include <linux/device.h>
64#include <linux/kthread.h>
65#include <linux/sched.h>
66#include <linux/sched/init.h>
67#include <linux/signal.h>
68#include <linux/idr.h>
69#include <linux/kgdb.h>
70#include <linux/ftrace.h>
71#include <linux/async.h>
72#include <linux/sfi.h>
73#include <linux/shmem_fs.h>
74#include <linux/slab.h>
75#include <linux/perf_event.h>
76#include <linux/ptrace.h>
77#include <linux/pti.h>
78#include <linux/blkdev.h>
79#include <linux/elevator.h>
80#include <linux/sched_clock.h>
81#include <linux/sched/task.h>
82#include <linux/sched/task_stack.h>
83#include <linux/context_tracking.h>
84#include <linux/random.h>
85#include <linux/list.h>
86#include <linux/integrity.h>
87#include <linux/proc_ns.h>
88#include <linux/io.h>
89#include <linux/cache.h>
90#include <linux/rodata_test.h>
91
92#include <asm/io.h>
93#include <asm/bugs.h>
94#include <asm/setup.h>
95#include <asm/sections.h>
96#include <asm/cacheflush.h>
97
98static int kernel_init(void *);
99
100extern void init_IRQ(void);
101extern void radix_tree_init(void);
102
103/*
104 * Debug helper: via this flag we know that we are in 'early bootup code'
105 * where only the boot processor is running with IRQ disabled. This means
106 * two things - IRQ must not be enabled before the flag is cleared and some
107 * operations which are not allowed with IRQ disabled are allowed while the
108 * flag is set.
109 */
110bool early_boot_irqs_disabled __read_mostly;
111
112enum system_states system_state __read_mostly;
113EXPORT_SYMBOL(system_state);
114
115/*
116 * Boot command-line arguments
117 */
118#define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT
119#define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT
120
121extern void time_init(void);
122/* Default late time init is NULL. archs can override this later. */
123void (*__initdata late_time_init)(void);
124
125/* Untouched command line saved by arch-specific code. */
126char __initdata boot_command_line[COMMAND_LINE_SIZE];
127/* Untouched saved command line (eg. for /proc) */
128char *saved_command_line;
129/* Command line for parameter parsing */
130static char *static_command_line;
131/* Command line for per-initcall parameter parsing */
132static char *initcall_command_line;
133
134static char *execute_command;
135static char *ramdisk_execute_command;
136
137/*
138 * Used to generate warnings if static_key manipulation functions are used
139 * before jump_label_init is called.
140 */
141bool static_key_initialized __read_mostly;
142EXPORT_SYMBOL_GPL(static_key_initialized);
143
144/*
145 * If set, this is an indication to the drivers that reset the underlying
146 * device before going ahead with the initialization otherwise driver might
147 * rely on the BIOS and skip the reset operation.
148 *
149 * This is useful if kernel is booting in an unreliable environment.
150 * For ex. kdump situation where previous kernel has crashed, BIOS has been
151 * skipped and devices will be in unknown state.
152 */
153unsigned int reset_devices;
154EXPORT_SYMBOL(reset_devices);
155
156static int __init set_reset_devices(char *str)
157{
158 reset_devices = 1;
159 return 1;
160}
161
162__setup("reset_devices", set_reset_devices);
163
164static const char *argv_init[MAX_INIT_ARGS+2] = { "init", NULL, };
165const char *envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, };
166static const char *panic_later, *panic_param;
167
168extern const struct obs_kernel_param __setup_start[], __setup_end[];
169
170static bool __init obsolete_checksetup(char *line)
171{
172 const struct obs_kernel_param *p;
173 bool had_early_param = false;
174
175 p = __setup_start;
176 do {
177 int n = strlen(p->str);
178 if (parameqn(line, p->str, n)) {
179 if (p->early) {
180 /* Already done in parse_early_param?
181 * (Needs exact match on param part).
182 * Keep iterating, as we can have early
183 * params and __setups of same names 8( */
184 if (line[n] == '\0' || line[n] == '=')
185 had_early_param = true;
186 } else if (!p->setup_func) {
187 pr_warn("Parameter %s is obsolete, ignored\n",
188 p->str);
189 return true;
190 } else if (p->setup_func(line + n))
191 return true;
192 }
193 p++;
194 } while (p < __setup_end);
195
196 return had_early_param;
197}
198
199/*
200 * This should be approx 2 Bo*oMips to start (note initial shift), and will
201 * still work even if initially too large, it will just take slightly longer
202 */
203unsigned long loops_per_jiffy = (1<<12);
204EXPORT_SYMBOL(loops_per_jiffy);
205
206static int __init debug_kernel(char *str)
207{
208 console_loglevel = CONSOLE_LOGLEVEL_DEBUG;
209 return 0;
210}
211
212static int __init quiet_kernel(char *str)
213{
214 console_loglevel = CONSOLE_LOGLEVEL_QUIET;
215 return 0;
216}
217
218early_param("debug", debug_kernel);
219early_param("quiet", quiet_kernel);
220
221static int __init loglevel(char *str)
222{
223 int newlevel;
224
225 /*
226 * Only update loglevel value when a correct setting was passed,
227 * to prevent blind crashes (when loglevel being set to 0) that
228 * are quite hard to debug
229 */
230 if (get_option(&str, &newlevel)) {
231 console_loglevel = newlevel;
232 return 0;
233 }
234
235 return -EINVAL;
236}
237
238early_param("loglevel", loglevel);
239
240/* Change NUL term back to "=", to make "param" the whole string. */
241static int __init repair_env_string(char *param, char *val,
242 const char *unused, void *arg)
243{
244 if (val) {
245 /* param=val or param="val"? */
246 if (val == param+strlen(param)+1)
247 val[-1] = '=';
248 else if (val == param+strlen(param)+2) {
249 val[-2] = '=';
250 memmove(val-1, val, strlen(val)+1);
251 val--;
252 } else
253 BUG();
254 }
255 return 0;
256}
257
258/* Anything after -- gets handed straight to init. */
259static int __init set_init_arg(char *param, char *val,
260 const char *unused, void *arg)
261{
262 unsigned int i;
263
264 if (panic_later)
265 return 0;
266
267 repair_env_string(param, val, unused, NULL);
268
269 for (i = 0; argv_init[i]; i++) {
270 if (i == MAX_INIT_ARGS) {
271 panic_later = "init";
272 panic_param = param;
273 return 0;
274 }
275 }
276 argv_init[i] = param;
277 return 0;
278}
279
280/*
281 * Unknown boot options get handed to init, unless they look like
282 * unused parameters (modprobe will find them in /proc/cmdline).
283 */
284static int __init unknown_bootoption(char *param, char *val,
285 const char *unused, void *arg)
286{
287 repair_env_string(param, val, unused, NULL);
288
289 /* Handle obsolete-style parameters */
290 if (obsolete_checksetup(param))
291 return 0;
292
293 /* Unused module parameter. */
294 if (strchr(param, '.') && (!val || strchr(param, '.') < val))
295 return 0;
296
297 if (panic_later)
298 return 0;
299
300 if (val) {
301 /* Environment option */
302 unsigned int i;
303 for (i = 0; envp_init[i]; i++) {
304 if (i == MAX_INIT_ENVS) {
305 panic_later = "env";
306 panic_param = param;
307 }
308 if (!strncmp(param, envp_init[i], val - param))
309 break;
310 }
311 envp_init[i] = param;
312 } else {
313 /* Command line option */
314 unsigned int i;
315 for (i = 0; argv_init[i]; i++) {
316 if (i == MAX_INIT_ARGS) {
317 panic_later = "init";
318 panic_param = param;
319 }
320 }
321 argv_init[i] = param;
322 }
323 return 0;
324}
325
326static int __init init_setup(char *str)
327{
328 unsigned int i;
329
330 execute_command = str;
331 /*
332 * In case LILO is going to boot us with default command line,
333 * it prepends "auto" before the whole cmdline which makes
334 * the shell think it should execute a script with such name.
335 * So we ignore all arguments entered _before_ init=... [MJ]
336 */
337 for (i = 1; i < MAX_INIT_ARGS; i++)
338 argv_init[i] = NULL;
339 return 1;
340}
341__setup("init=", init_setup);
342
343static int __init rdinit_setup(char *str)
344{
345 unsigned int i;
346
347 ramdisk_execute_command = str;
348 /* See "auto" comment in init_setup */
349 for (i = 1; i < MAX_INIT_ARGS; i++)
350 argv_init[i] = NULL;
351 return 1;
352}
353__setup("rdinit=", rdinit_setup);
354
355#ifndef CONFIG_SMP
356static const unsigned int setup_max_cpus = NR_CPUS;
357static inline void setup_nr_cpu_ids(void) { }
358static inline void smp_prepare_cpus(unsigned int maxcpus) { }
359#endif
360
361/*
362 * We need to store the untouched command line for future reference.
363 * We also need to store the touched command line since the parameter
364 * parsing is performed in place, and we should allow a component to
365 * store reference of name/value for future reference.
366 */
367static void __init setup_command_line(char *command_line)
368{
369 saved_command_line =
370 memblock_virt_alloc(strlen(boot_command_line) + 1, 0);
371 initcall_command_line =
372 memblock_virt_alloc(strlen(boot_command_line) + 1, 0);
373 static_command_line = memblock_virt_alloc(strlen(command_line) + 1, 0);
374 strcpy(saved_command_line, boot_command_line);
375 strcpy(static_command_line, command_line);
376}
377
378/*
379 * We need to finalize in a non-__init function or else race conditions
380 * between the root thread and the init thread may cause start_kernel to
381 * be reaped by free_initmem before the root thread has proceeded to
382 * cpu_idle.
383 *
384 * gcc-3.4 accidentally inlines this function, so use noinline.
385 */
386
387static __initdata DECLARE_COMPLETION(kthreadd_done);
388
389static noinline void __ref rest_init(void)
390{
391 struct task_struct *tsk;
392 int pid;
393
394 rcu_scheduler_starting();
395 /*
396 * We need to spawn init first so that it obtains pid 1, however
397 * the init task will end up wanting to create kthreads, which, if
398 * we schedule it before we create kthreadd, will OOPS.
399 */
400 pid = kernel_thread(kernel_init, NULL, CLONE_FS);
401 /*
402 * Pin init on the boot CPU. Task migration is not properly working
403 * until sched_init_smp() has been run. It will set the allowed
404 * CPUs for init to the non isolated CPUs.
405 */
406 rcu_read_lock();
407 tsk = find_task_by_pid_ns(pid, &init_pid_ns);
408 set_cpus_allowed_ptr(tsk, cpumask_of(smp_processor_id()));
409 rcu_read_unlock();
410
411 numa_default_policy();
412 pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES);
413 rcu_read_lock();
414 kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns);
415 rcu_read_unlock();
416
417 /*
418 * Enable might_sleep() and smp_processor_id() checks.
419 * They cannot be enabled earlier because with CONFIG_PRREMPT=y
420 * kernel_thread() would trigger might_sleep() splats. With
421 * CONFIG_PREEMPT_VOLUNTARY=y the init task might have scheduled
422 * already, but it's stuck on the kthreadd_done completion.
423 */
424 system_state = SYSTEM_SCHEDULING;
425
426 complete(&kthreadd_done);
427
428 /*
429 * The boot idle thread must execute schedule()
430 * at least once to get things moving:
431 */
432 schedule_preempt_disabled();
433 /* Call into cpu_idle with preempt disabled */
434 cpu_startup_entry(CPUHP_ONLINE);
435}
436
437/* Check for early params. */
438static int __init do_early_param(char *param, char *val,
439 const char *unused, void *arg)
440{
441 const struct obs_kernel_param *p;
442
443 for (p = __setup_start; p < __setup_end; p++) {
444 if ((p->early && parameq(param, p->str)) ||
445 (strcmp(param, "console") == 0 &&
446 strcmp(p->str, "earlycon") == 0)
447 ) {
448 if (p->setup_func(val) != 0)
449 pr_warn("Malformed early option '%s'\n", param);
450 }
451 }
452 /* We accept everything at this stage. */
453 return 0;
454}
455
456void __init parse_early_options(char *cmdline)
457{
458 parse_args("early options", cmdline, NULL, 0, 0, 0, NULL,
459 do_early_param);
460}
461
462/* Arch code calls this early on, or if not, just before other parsing. */
463void __init parse_early_param(void)
464{
465 static int done __initdata;
466 static char tmp_cmdline[COMMAND_LINE_SIZE] __initdata;
467
468 if (done)
469 return;
470
471 /* All fall through to do_early_param. */
472 strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE);
473 parse_early_options(tmp_cmdline);
474 done = 1;
475}
476
477void __init __weak arch_post_acpi_subsys_init(void) { }
478
479void __init __weak smp_setup_processor_id(void)
480{
481}
482
483# if THREAD_SIZE >= PAGE_SIZE
484void __init __weak thread_stack_cache_init(void)
485{
486}
487#endif
488
489void __init __weak mem_encrypt_init(void) { }
490
491/*
492 * Set up kernel memory allocators
493 */
494static void __init mm_init(void)
495{
496 /*
497 * page_ext requires contiguous pages,
498 * bigger than MAX_ORDER unless SPARSEMEM.
499 */
500 page_ext_init_flatmem();
501 mem_init();
502 kmem_cache_init();
503 pgtable_init();
504 vmalloc_init();
505 ioremap_huge_init();
506 /* Should be run before the first non-init thread is created */
507 init_espfix_bsp();
508 /* Should be run after espfix64 is set up. */
509 pti_init();
510}
511
512asmlinkage __visible void __init start_kernel(void)
513{
514 char *command_line;
515 char *after_dashes;
516
517 set_task_stack_end_magic(&init_task);
518 smp_setup_processor_id();
519 debug_objects_early_init();
520
521 cgroup_init_early();
522
523 local_irq_disable();
524 early_boot_irqs_disabled = true;
525
526 /*
527 * Interrupts are still disabled. Do necessary setups, then
528 * enable them.
529 */
530 boot_cpu_init();
531 page_address_init();
532 pr_notice("%s", linux_banner);
533 setup_arch(&command_line);
534 /*
535 * Set up the the initial canary and entropy after arch
536 * and after adding latent and command line entropy.
537 */
538 add_latent_entropy();
539 add_device_randomness(command_line, strlen(command_line));
540 boot_init_stack_canary();
541 mm_init_cpumask(&init_mm);
542 setup_command_line(command_line);
543 setup_nr_cpu_ids();
544 setup_per_cpu_areas();
545 smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */
546 boot_cpu_hotplug_init();
547
548 build_all_zonelists(NULL);
549 page_alloc_init();
550
551 pr_notice("Kernel command line: %s\n", boot_command_line);
552 /* parameters may set static keys */
553 jump_label_init();
554 parse_early_param();
555 after_dashes = parse_args("Booting kernel",
556 static_command_line, __start___param,
557 __stop___param - __start___param,
558 -1, -1, NULL, &unknown_bootoption);
559 if (!IS_ERR_OR_NULL(after_dashes))
560 parse_args("Setting init args", after_dashes, NULL, 0, -1, -1,
561 NULL, set_init_arg);
562
563 /*
564 * These use large bootmem allocations and must precede
565 * kmem_cache_init()
566 */
567 setup_log_buf(0);
568 pidhash_init();
569 vfs_caches_init_early();
570 sort_main_extable();
571 trap_init();
572 mm_init();
573
574 ftrace_init();
575
576 /* trace_printk can be enabled here */
577 early_trace_init();
578
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();
590 if (WARN(!irqs_disabled(),
591 "Interrupts were enabled *very* early, fixing it\n"))
592 local_irq_disable();
593 radix_tree_init();
594
595 /*
596 * Allow workqueue creation and work item queueing/cancelling
597 * early. Work item execution depends on kthreads and starts after
598 * workqueue_init().
599 */
600 workqueue_init_early();
601
602 rcu_init();
603
604 /* Trace events are available after this */
605 trace_init();
606
607 context_tracking_init();
608 /* init some links before init_ISA_irqs() */
609 early_irq_init();
610 init_IRQ();
611 tick_init();
612 rcu_init_nohz();
613 init_timers();
614 hrtimers_init();
615 softirq_init();
616 timekeeping_init();
617 time_init();
618 sched_clock_postinit();
619 printk_safe_init();
620 perf_event_init();
621 profile_init();
622 call_function_init();
623 WARN(!irqs_disabled(), "Interrupts were enabled early\n");
624 early_boot_irqs_disabled = false;
625 local_irq_enable();
626
627 kmem_cache_init_late();
628
629 /*
630 * HACK ALERT! This is early. We're enabling the console before
631 * we've done PCI setups etc, and console_init() must be aware of
632 * this. But we do want output early, in case something goes wrong.
633 */
634 console_init();
635 if (panic_later)
636 panic("Too many boot %s vars at `%s'", panic_later,
637 panic_param);
638
639 lockdep_info();
640
641 /*
642 * Need to run this when irqs are enabled, because it wants
643 * to self-test [hard/soft]-irqs on/off lock inversion bugs
644 * too:
645 */
646 locking_selftest();
647
648 /*
649 * This needs to be called before any devices perform DMA
650 * operations that might use the SWIOTLB bounce buffers. It will
651 * mark the bounce buffers as decrypted so that their usage will
652 * not cause "plain-text" data to be decrypted when accessed.
653 */
654 mem_encrypt_init();
655
656#ifdef CONFIG_BLK_DEV_INITRD
657 if (initrd_start && !initrd_below_start_ok &&
658 page_to_pfn(virt_to_page((void *)initrd_start)) < min_low_pfn) {
659 pr_crit("initrd overwritten (0x%08lx < 0x%08lx) - disabling it.\n",
660 page_to_pfn(virt_to_page((void *)initrd_start)),
661 min_low_pfn);
662 initrd_start = 0;
663 }
664#endif
665 kmemleak_init();
666 debug_objects_mem_init();
667 setup_per_cpu_pageset();
668 numa_policy_init();
669 if (late_time_init)
670 late_time_init();
671 calibrate_delay();
672 pidmap_init();
673 anon_vma_init();
674 acpi_early_init();
675#ifdef CONFIG_X86
676 if (efi_enabled(EFI_RUNTIME_SERVICES))
677 efi_enter_virtual_mode();
678#endif
679 thread_stack_cache_init();
680 cred_init();
681 fork_init();
682 proc_caches_init();
683 buffer_init();
684 key_init();
685 security_init();
686 dbg_late_init();
687 vfs_caches_init();
688 pagecache_init();
689 signals_init();
690 proc_root_init();
691 nsfs_init();
692 cpuset_init();
693 cgroup_init();
694 taskstats_init_early();
695 delayacct_init();
696
697 check_bugs();
698
699 acpi_subsystem_init();
700 arch_post_acpi_subsys_init();
701 sfi_init_late();
702
703 if (efi_enabled(EFI_RUNTIME_SERVICES)) {
704 efi_free_boot_services();
705 }
706
707 /* Do the rest non-__init'ed, we're now alive */
708 rest_init();
709
710 prevent_tail_call_optimization();
711}
712
713/* Call all constructor functions linked into the kernel. */
714static void __init do_ctors(void)
715{
716#ifdef CONFIG_CONSTRUCTORS
717 ctor_fn_t *fn = (ctor_fn_t *) __ctors_start;
718
719 for (; fn < (ctor_fn_t *) __ctors_end; fn++)
720 (*fn)();
721#endif
722}
723
724bool initcall_debug;
725core_param(initcall_debug, initcall_debug, bool, 0644);
726
727#ifdef CONFIG_KALLSYMS
728struct blacklist_entry {
729 struct list_head next;
730 char *buf;
731};
732
733static __initdata_or_module LIST_HEAD(blacklisted_initcalls);
734
735static int __init initcall_blacklist(char *str)
736{
737 char *str_entry;
738 struct blacklist_entry *entry;
739
740 /* str argument is a comma-separated list of functions */
741 do {
742 str_entry = strsep(&str, ",");
743 if (str_entry) {
744 pr_debug("blacklisting initcall %s\n", str_entry);
745 entry = alloc_bootmem(sizeof(*entry));
746 entry->buf = alloc_bootmem(strlen(str_entry) + 1);
747 strcpy(entry->buf, str_entry);
748 list_add(&entry->next, &blacklisted_initcalls);
749 }
750 } while (str_entry);
751
752 return 0;
753}
754
755static bool __init_or_module initcall_blacklisted(initcall_t fn)
756{
757 struct blacklist_entry *entry;
758 char fn_name[KSYM_SYMBOL_LEN];
759 unsigned long addr;
760
761 if (list_empty(&blacklisted_initcalls))
762 return false;
763
764 addr = (unsigned long) dereference_function_descriptor(fn);
765 sprint_symbol_no_offset(fn_name, addr);
766
767 /*
768 * fn will be "function_name [module_name]" where [module_name] is not
769 * displayed for built-in init functions. Strip off the [module_name].
770 */
771 strreplace(fn_name, ' ', '\0');
772
773 list_for_each_entry(entry, &blacklisted_initcalls, next) {
774 if (!strcmp(fn_name, entry->buf)) {
775 pr_debug("initcall %s blacklisted\n", fn_name);
776 return true;
777 }
778 }
779
780 return false;
781}
782#else
783static int __init initcall_blacklist(char *str)
784{
785 pr_warn("initcall_blacklist requires CONFIG_KALLSYMS\n");
786 return 0;
787}
788
789static bool __init_or_module initcall_blacklisted(initcall_t fn)
790{
791 return false;
792}
793#endif
794__setup("initcall_blacklist=", initcall_blacklist);
795
796static int __init_or_module do_one_initcall_debug(initcall_t fn)
797{
798 ktime_t calltime, delta, rettime;
799 unsigned long long duration;
800 int ret;
801
802 printk(KERN_DEBUG "calling %pF @ %i\n", fn, task_pid_nr(current));
803 calltime = ktime_get();
804 ret = fn();
805 rettime = ktime_get();
806 delta = ktime_sub(rettime, calltime);
807 duration = (unsigned long long) ktime_to_ns(delta) >> 10;
808 printk(KERN_DEBUG "initcall %pF returned %d after %lld usecs\n",
809 fn, ret, duration);
810
811 return ret;
812}
813
814int __init_or_module do_one_initcall(initcall_t fn)
815{
816 int count = preempt_count();
817 int ret;
818 char msgbuf[64];
819
820 if (initcall_blacklisted(fn))
821 return -EPERM;
822
823 if (initcall_debug)
824 ret = do_one_initcall_debug(fn);
825 else
826 ret = fn();
827
828 msgbuf[0] = 0;
829
830 if (preempt_count() != count) {
831 sprintf(msgbuf, "preemption imbalance ");
832 preempt_count_set(count);
833 }
834 if (irqs_disabled()) {
835 strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf));
836 local_irq_enable();
837 }
838 WARN(msgbuf[0], "initcall %pF returned with %s\n", fn, msgbuf);
839
840 add_latent_entropy();
841 return ret;
842}
843
844
845extern initcall_t __initcall_start[];
846extern initcall_t __initcall0_start[];
847extern initcall_t __initcall1_start[];
848extern initcall_t __initcall2_start[];
849extern initcall_t __initcall3_start[];
850extern initcall_t __initcall4_start[];
851extern initcall_t __initcall5_start[];
852extern initcall_t __initcall6_start[];
853extern initcall_t __initcall7_start[];
854extern initcall_t __initcall_end[];
855
856static initcall_t *initcall_levels[] __initdata = {
857 __initcall0_start,
858 __initcall1_start,
859 __initcall2_start,
860 __initcall3_start,
861 __initcall4_start,
862 __initcall5_start,
863 __initcall6_start,
864 __initcall7_start,
865 __initcall_end,
866};
867
868/* Keep these in sync with initcalls in include/linux/init.h */
869static char *initcall_level_names[] __initdata = {
870 "early",
871 "core",
872 "postcore",
873 "arch",
874 "subsys",
875 "fs",
876 "device",
877 "late",
878};
879
880static void __init do_initcall_level(int level)
881{
882 initcall_t *fn;
883
884 strcpy(initcall_command_line, saved_command_line);
885 parse_args(initcall_level_names[level],
886 initcall_command_line, __start___param,
887 __stop___param - __start___param,
888 level, level,
889 NULL, &repair_env_string);
890
891 for (fn = initcall_levels[level]; fn < initcall_levels[level+1]; fn++)
892 do_one_initcall(*fn);
893}
894
895static void __init do_initcalls(void)
896{
897 int level;
898
899 for (level = 0; level < ARRAY_SIZE(initcall_levels) - 1; level++)
900 do_initcall_level(level);
901}
902
903/*
904 * Ok, the machine is now initialized. None of the devices
905 * have been touched yet, but the CPU subsystem is up and
906 * running, and memory and process management works.
907 *
908 * Now we can finally start doing some real work..
909 */
910static void __init do_basic_setup(void)
911{
912 cpuset_init_smp();
913 shmem_init();
914 driver_init();
915 init_irq_proc();
916 do_ctors();
917 usermodehelper_enable();
918 do_initcalls();
919}
920
921static void __init do_pre_smp_initcalls(void)
922{
923 initcall_t *fn;
924
925 for (fn = __initcall_start; fn < __initcall0_start; fn++)
926 do_one_initcall(*fn);
927}
928
929/*
930 * This function requests modules which should be loaded by default and is
931 * called twice right after initrd is mounted and right before init is
932 * exec'd. If such modules are on either initrd or rootfs, they will be
933 * loaded before control is passed to userland.
934 */
935void __init load_default_modules(void)
936{
937 load_default_elevator_module();
938}
939
940static int run_init_process(const char *init_filename)
941{
942 argv_init[0] = init_filename;
943 return do_execve(getname_kernel(init_filename),
944 (const char __user *const __user *)argv_init,
945 (const char __user *const __user *)envp_init);
946}
947
948static int try_to_run_init_process(const char *init_filename)
949{
950 int ret;
951
952 ret = run_init_process(init_filename);
953
954 if (ret && ret != -ENOENT) {
955 pr_err("Starting init: %s exists but couldn't execute it (error %d)\n",
956 init_filename, ret);
957 }
958
959 return ret;
960}
961
962static noinline void __init kernel_init_freeable(void);
963
964#if defined(CONFIG_STRICT_KERNEL_RWX) || defined(CONFIG_STRICT_MODULE_RWX)
965bool rodata_enabled __ro_after_init = true;
966static int __init set_debug_rodata(char *str)
967{
968 return strtobool(str, &rodata_enabled);
969}
970__setup("rodata=", set_debug_rodata);
971#endif
972
973#ifdef CONFIG_STRICT_KERNEL_RWX
974static void mark_readonly(void)
975{
976 if (rodata_enabled) {
977 /*
978 * load_module() results in W+X mappings, which are cleaned up
979 * with call_rcu_sched(). Let's make sure that queued work is
980 * flushed so that we don't hit false positives looking for
981 * insecure pages which are W+X.
982 */
983 rcu_barrier_sched();
984 mark_rodata_ro();
985 rodata_test();
986 } else
987 pr_info("Kernel memory protection disabled.\n");
988}
989#else
990static inline void mark_readonly(void)
991{
992 pr_warn("This architecture does not have kernel memory protection.\n");
993}
994#endif
995
996static int __ref kernel_init(void *unused)
997{
998 int ret;
999
1000 kernel_init_freeable();
1001 /* need to finish all async __init code before freeing the memory */
1002 async_synchronize_full();
1003 ftrace_free_init_mem();
1004 free_initmem();
1005 mark_readonly();
1006 system_state = SYSTEM_RUNNING;
1007 numa_default_policy();
1008
1009 rcu_end_inkernel_boot();
1010
1011 if (ramdisk_execute_command) {
1012 ret = run_init_process(ramdisk_execute_command);
1013 if (!ret)
1014 return 0;
1015 pr_err("Failed to execute %s (error %d)\n",
1016 ramdisk_execute_command, ret);
1017 }
1018
1019 /*
1020 * We try each of these until one succeeds.
1021 *
1022 * The Bourne shell can be used instead of init if we are
1023 * trying to recover a really broken machine.
1024 */
1025 if (execute_command) {
1026 ret = run_init_process(execute_command);
1027 if (!ret)
1028 return 0;
1029 panic("Requested init %s failed (error %d).",
1030 execute_command, ret);
1031 }
1032 if (!try_to_run_init_process("/sbin/init") ||
1033 !try_to_run_init_process("/etc/init") ||
1034 !try_to_run_init_process("/bin/init") ||
1035 !try_to_run_init_process("/bin/sh"))
1036 return 0;
1037
1038 panic("No working init found. Try passing init= option to kernel. "
1039 "See Linux Documentation/admin-guide/init.rst for guidance.");
1040}
1041
1042static noinline void __init kernel_init_freeable(void)
1043{
1044 /*
1045 * Wait until kthreadd is all set-up.
1046 */
1047 wait_for_completion(&kthreadd_done);
1048
1049 /* Now the scheduler is fully set up and can do blocking allocations */
1050 gfp_allowed_mask = __GFP_BITS_MASK;
1051
1052 /*
1053 * init can allocate pages on any node
1054 */
1055 set_mems_allowed(node_states[N_MEMORY]);
1056
1057 cad_pid = task_pid(current);
1058
1059 smp_prepare_cpus(setup_max_cpus);
1060
1061 workqueue_init();
1062
1063 init_mm_internals();
1064
1065 do_pre_smp_initcalls();
1066 lockup_detector_init();
1067
1068 smp_init();
1069 sched_init_smp();
1070
1071 page_alloc_init_late();
1072 /* Initialize page ext after all struct pages are initialized. */
1073 page_ext_init();
1074
1075 do_basic_setup();
1076
1077 /* Open the /dev/console on the rootfs, this should never fail */
1078 if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0)
1079 pr_err("Warning: unable to open an initial console.\n");
1080
1081 (void) sys_dup(0);
1082 (void) sys_dup(0);
1083 /*
1084 * check if there is an early userspace init. If yes, let it do all
1085 * the work
1086 */
1087
1088 if (!ramdisk_execute_command)
1089 ramdisk_execute_command = "/init";
1090
1091 if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) {
1092 ramdisk_execute_command = NULL;
1093 prepare_namespace();
1094 }
1095
1096 /*
1097 * Ok, we have completed the initial bootup, and
1098 * we're essentially up and running. Get rid of the
1099 * initmem segments and start the user-mode stuff..
1100 *
1101 * rootfs is available now, try loading the public keys
1102 * and default modules
1103 */
1104
1105 integrity_load_keys();
1106 load_default_modules();
1107}