4 * Copyright (C) 1991, 1992 Linus Torvalds
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>
12 #include <linux/types.h>
13 #include <linux/module.h>
14 #include <linux/proc_fs.h>
15 #include <linux/kernel.h>
16 #include <linux/syscalls.h>
17 #include <linux/string.h>
18 #include <linux/ctype.h>
19 #include <linux/delay.h>
20 #include <linux/utsname.h>
21 #include <linux/ioport.h>
22 #include <linux/init.h>
23 #include <linux/smp_lock.h>
24 #include <linux/initrd.h>
25 #include <linux/hdreg.h>
26 #include <linux/bootmem.h>
27 #include <linux/tty.h>
28 #include <linux/gfp.h>
29 #include <linux/percpu.h>
30 #include <linux/kmod.h>
31 #include <linux/kernel_stat.h>
32 #include <linux/start_kernel.h>
33 #include <linux/security.h>
34 #include <linux/workqueue.h>
35 #include <linux/profile.h>
36 #include <linux/rcupdate.h>
37 #include <linux/moduleparam.h>
38 #include <linux/kallsyms.h>
39 #include <linux/writeback.h>
40 #include <linux/cpu.h>
41 #include <linux/cpuset.h>
42 #include <linux/cgroup.h>
43 #include <linux/efi.h>
44 #include <linux/tick.h>
45 #include <linux/interrupt.h>
46 #include <linux/taskstats_kern.h>
47 #include <linux/delayacct.h>
48 #include <linux/unistd.h>
49 #include <linux/rmap.h>
50 #include <linux/mempolicy.h>
51 #include <linux/key.h>
52 #include <linux/unwind.h>
53 #include <linux/buffer_head.h>
54 #include <linux/debug_locks.h>
55 #include <linux/lockdep.h>
56 #include <linux/pid_namespace.h>
57 #include <linux/device.h>
58 #include <linux/kthread.h>
59 #include <linux/sched.h>
60 #include <linux/signal.h>
64 #include <asm/setup.h>
65 #include <asm/sections.h>
66 #include <asm/cacheflush.h>
68 #ifdef CONFIG_X86_LOCAL_APIC
73 * This is one of the first .c files built. Error out early if we have compiler
77 #if __GNUC__ == 4 && __GNUC_MINOR__ == 1 && __GNUC_PATCHLEVEL__ == 0
78 #warning gcc-4.1.0 is known to miscompile the kernel. A different compiler version is recommended.
81 static int kernel_init(void *);
83 extern void init_IRQ(void);
84 extern void fork_init(unsigned long);
85 extern void mca_init(void);
86 extern void sbus_init(void);
87 extern void pidhash_init(void);
88 extern void pidmap_init(void);
89 extern void prio_tree_init(void);
90 extern void radix_tree_init(void);
91 extern void free_initmem(void);
93 extern void acpi_early_init(void);
95 static inline void acpi_early_init(void) { }
97 #ifndef CONFIG_DEBUG_RODATA
98 static inline void mark_rodata_ro(void) { }
102 extern void tc_init(void);
105 #ifdef CONFIG_ACPI_CUSTOM_DSDT_INITRD
106 extern int populate_rootfs(void);
108 static inline void populate_rootfs(void) {}
111 enum system_states system_state;
112 EXPORT_SYMBOL(system_state);
115 * Boot command-line arguments
117 #define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT
118 #define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT
120 extern void time_init(void);
121 /* Default late time init is NULL. archs can override this later. */
122 void (*late_time_init)(void);
123 extern void softirq_init(void);
125 /* Untouched command line saved by arch-specific code. */
126 char __initdata boot_command_line[COMMAND_LINE_SIZE];
127 /* Untouched saved command line (eg. for /proc) */
128 char *saved_command_line;
129 /* Command line for parameter parsing */
130 static char *static_command_line;
132 static char *execute_command;
133 static char *ramdisk_execute_command;
136 /* Setup configured maximum number of CPUs to activate */
137 unsigned int __initdata setup_max_cpus = NR_CPUS;
140 * Setup routine for controlling SMP activation
142 * Command-line option of "nosmp" or "maxcpus=0" will disable SMP
143 * activation entirely (the MPS table probe still happens, though).
145 * Command-line option of "maxcpus=<NUM>", where <NUM> is an integer
146 * greater than 0, limits the maximum number of CPUs activated in
149 #ifndef CONFIG_X86_IO_APIC
150 static inline void disable_ioapic_setup(void) {};
153 static int __init nosmp(char *str)
156 disable_ioapic_setup();
160 early_param("nosmp", nosmp);
162 static int __init maxcpus(char *str)
164 get_option(&str, &setup_max_cpus);
165 if (setup_max_cpus == 0)
166 disable_ioapic_setup();
171 early_param("maxcpus", maxcpus);
173 #define setup_max_cpus NR_CPUS
177 * If set, this is an indication to the drivers that reset the underlying
178 * device before going ahead with the initialization otherwise driver might
179 * rely on the BIOS and skip the reset operation.
181 * This is useful if kernel is booting in an unreliable environment.
182 * For ex. kdump situaiton where previous kernel has crashed, BIOS has been
183 * skipped and devices will be in unknown state.
185 unsigned int reset_devices;
186 EXPORT_SYMBOL(reset_devices);
188 static int __init set_reset_devices(char *str)
194 __setup("reset_devices", set_reset_devices);
196 static char * argv_init[MAX_INIT_ARGS+2] = { "init", NULL, };
197 char * envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, };
198 static const char *panic_later, *panic_param;
200 extern struct obs_kernel_param __setup_start[], __setup_end[];
202 static int __init obsolete_checksetup(char *line)
204 struct obs_kernel_param *p;
205 int had_early_param = 0;
209 int n = strlen(p->str);
210 if (!strncmp(line, p->str, n)) {
212 /* Already done in parse_early_param?
213 * (Needs exact match on param part).
214 * Keep iterating, as we can have early
215 * params and __setups of same names 8( */
216 if (line[n] == '\0' || line[n] == '=')
218 } else if (!p->setup_func) {
219 printk(KERN_WARNING "Parameter %s is obsolete,"
220 " ignored\n", p->str);
222 } else if (p->setup_func(line + n))
226 } while (p < __setup_end);
228 return had_early_param;
232 * This should be approx 2 Bo*oMips to start (note initial shift), and will
233 * still work even if initially too large, it will just take slightly longer
235 unsigned long loops_per_jiffy = (1<<12);
237 EXPORT_SYMBOL(loops_per_jiffy);
239 static int __init debug_kernel(char *str)
243 console_loglevel = 10;
247 static int __init quiet_kernel(char *str)
251 console_loglevel = 4;
255 __setup("debug", debug_kernel);
256 __setup("quiet", quiet_kernel);
258 static int __init loglevel(char *str)
260 get_option(&str, &console_loglevel);
264 __setup("loglevel=", loglevel);
267 * Unknown boot options get handed to init, unless they look like
270 static int __init unknown_bootoption(char *param, char *val)
272 /* Change NUL term back to "=", to make "param" the whole string. */
274 /* param=val or param="val"? */
275 if (val == param+strlen(param)+1)
277 else if (val == param+strlen(param)+2) {
279 memmove(val-1, val, strlen(val)+1);
285 /* Handle obsolete-style parameters */
286 if (obsolete_checksetup(param))
290 * Preemptive maintenance for "why didn't my misspelled command
293 if (strchr(param, '.') && (!val || strchr(param, '.') < val)) {
294 printk(KERN_ERR "Unknown boot option `%s': ignoring\n", param);
302 /* Environment option */
304 for (i = 0; envp_init[i]; i++) {
305 if (i == MAX_INIT_ENVS) {
306 panic_later = "Too many boot env vars at `%s'";
309 if (!strncmp(param, envp_init[i], val - param))
312 envp_init[i] = param;
314 /* Command line option */
316 for (i = 0; argv_init[i]; i++) {
317 if (i == MAX_INIT_ARGS) {
318 panic_later = "Too many boot init vars at `%s'";
322 argv_init[i] = param;
327 #ifdef CONFIG_DEBUG_PAGEALLOC
328 int __read_mostly debug_pagealloc_enabled = 0;
331 static int __init init_setup(char *str)
335 execute_command = str;
337 * In case LILO is going to boot us with default command line,
338 * it prepends "auto" before the whole cmdline which makes
339 * the shell think it should execute a script with such name.
340 * So we ignore all arguments entered _before_ init=... [MJ]
342 for (i = 1; i < MAX_INIT_ARGS; i++)
346 __setup("init=", init_setup);
348 static int __init rdinit_setup(char *str)
352 ramdisk_execute_command = str;
353 /* See "auto" comment in init_setup */
354 for (i = 1; i < MAX_INIT_ARGS; i++)
358 __setup("rdinit=", rdinit_setup);
362 #ifdef CONFIG_X86_LOCAL_APIC
363 static void __init smp_init(void)
365 APIC_init_uniprocessor();
368 #define smp_init() do { } while (0)
371 static inline void setup_per_cpu_areas(void) { }
372 static inline void smp_prepare_cpus(unsigned int maxcpus) { }
376 #ifndef CONFIG_HAVE_SETUP_PER_CPU_AREA
377 unsigned long __per_cpu_offset[NR_CPUS] __read_mostly;
379 EXPORT_SYMBOL(__per_cpu_offset);
381 static void __init setup_per_cpu_areas(void)
383 unsigned long size, i;
385 unsigned long nr_possible_cpus = num_possible_cpus();
387 /* Copy section for each CPU (we discard the original) */
388 size = ALIGN(PERCPU_ENOUGH_ROOM, PAGE_SIZE);
389 ptr = alloc_bootmem_pages(size * nr_possible_cpus);
391 for_each_possible_cpu(i) {
392 __per_cpu_offset[i] = ptr - __per_cpu_start;
393 memcpy(ptr, __per_cpu_start, __per_cpu_end - __per_cpu_start);
397 #endif /* CONFIG_HAVE_SETUP_PER_CPU_AREA */
399 /* Called by boot processor to activate the rest. */
400 static void __init smp_init(void)
404 /* FIXME: This should be done in userspace --RR */
405 for_each_present_cpu(cpu) {
406 if (num_online_cpus() >= setup_max_cpus)
408 if (!cpu_online(cpu))
412 /* Any cleanup work */
413 printk(KERN_INFO "Brought up %ld CPUs\n", (long)num_online_cpus());
414 smp_cpus_done(setup_max_cpus);
420 * We need to store the untouched command line for future reference.
421 * We also need to store the touched command line since the parameter
422 * parsing is performed in place, and we should allow a component to
423 * store reference of name/value for future reference.
425 static void __init setup_command_line(char *command_line)
427 saved_command_line = alloc_bootmem(strlen (boot_command_line)+1);
428 static_command_line = alloc_bootmem(strlen (command_line)+1);
429 strcpy (saved_command_line, boot_command_line);
430 strcpy (static_command_line, command_line);
434 * We need to finalize in a non-__init function or else race conditions
435 * between the root thread and the init thread may cause start_kernel to
436 * be reaped by free_initmem before the root thread has proceeded to
439 * gcc-3.4 accidentally inlines this function, so use noinline.
442 static void noinline __init_refok rest_init(void)
443 __releases(kernel_lock)
447 kernel_thread(kernel_init, NULL, CLONE_FS | CLONE_SIGHAND);
448 numa_default_policy();
449 pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES);
450 kthreadd_task = find_task_by_pid(pid);
454 * The boot idle thread must execute schedule()
455 * at least once to get things moving:
457 init_idle_bootup_task(current);
458 preempt_enable_no_resched();
462 /* Call into cpu_idle with preempt disabled */
466 /* Check for early params. */
467 static int __init do_early_param(char *param, char *val)
469 struct obs_kernel_param *p;
471 for (p = __setup_start; p < __setup_end; p++) {
472 if ((p->early && strcmp(param, p->str) == 0) ||
473 (strcmp(param, "console") == 0 &&
474 strcmp(p->str, "earlycon") == 0)
476 if (p->setup_func(val) != 0)
478 "Malformed early option '%s'\n", param);
481 /* We accept everything at this stage. */
485 /* Arch code calls this early on, or if not, just before other parsing. */
486 void __init parse_early_param(void)
488 static __initdata int done = 0;
489 static __initdata char tmp_cmdline[COMMAND_LINE_SIZE];
494 /* All fall through to do_early_param. */
495 strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE);
496 parse_args("early options", tmp_cmdline, NULL, 0, do_early_param);
501 * Activate the first processor.
504 static void __init boot_cpu_init(void)
506 int cpu = smp_processor_id();
507 /* Mark the boot cpu "present", "online" etc for SMP and UP case */
508 cpu_set(cpu, cpu_online_map);
509 cpu_set(cpu, cpu_present_map);
510 cpu_set(cpu, cpu_possible_map);
513 void __init __attribute__((weak)) smp_setup_processor_id(void)
517 asmlinkage void __init start_kernel(void)
520 extern struct kernel_param __start___param[], __stop___param[];
522 smp_setup_processor_id();
525 * Need to run as early as possible, to initialize the
533 early_boot_irqs_off();
534 early_init_irq_lock_class();
537 * Interrupts are still disabled. Do necessary setups, then
545 printk(linux_banner);
546 setup_arch(&command_line);
547 setup_command_line(command_line);
549 setup_per_cpu_areas();
550 smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */
553 * Set up the scheduler prior starting any interrupts (such as the
554 * timer interrupt). Full topology setup happens at smp_init()
555 * time - but meanwhile we still have a functioning scheduler.
559 * Disable preemption - early bootup scheduling is extremely
560 * fragile until we cpu_idle() for the first time.
563 build_all_zonelists();
565 enable_debug_pagealloc();
566 printk(KERN_NOTICE "Kernel command line: %s\n", boot_command_line);
568 parse_args("Booting kernel", static_command_line, __start___param,
569 __stop___param - __start___param,
570 &unknown_bootoption);
571 if (!irqs_disabled()) {
572 printk(KERN_WARNING "start_kernel(): bug: interrupts were "
573 "enabled *very* early, fixing it\n");
587 if (!irqs_disabled())
588 printk("start_kernel(): bug: interrupts were enabled early\n");
589 early_boot_irqs_on();
593 * HACK ALERT! This is early. We're enabling the console before
594 * we've done PCI setups etc, and console_init() must be aware of
595 * this. But we do want output early, in case something goes wrong.
599 panic(panic_later, panic_param);
604 * Need to run this when irqs are enabled, because it wants
605 * to self-test [hard/soft]-irqs on/off lock inversion bugs
610 #ifdef CONFIG_BLK_DEV_INITRD
611 if (initrd_start && !initrd_below_start_ok &&
612 initrd_start < min_low_pfn << PAGE_SHIFT) {
613 printk(KERN_CRIT "initrd overwritten (0x%08lx < 0x%08lx) - "
614 "disabling it.\n",initrd_start,min_low_pfn << PAGE_SHIFT);
618 vfs_caches_init_early();
623 setup_per_cpu_pageset();
629 pgtable_cache_init();
634 efi_enter_virtual_mode();
636 fork_init(num_physpages);
642 vfs_caches_init(num_physpages);
645 /* rootfs populating might need page-writeback */
646 page_writeback_init();
647 #ifdef CONFIG_PROC_FS
652 taskstats_init_early();
657 populate_rootfs(); /* For DSDT override from initramfs */
658 acpi_early_init(); /* before LAPIC and SMP init */
660 /* Do the rest non-__init'ed, we're now alive */
664 static int __initdata initcall_debug;
666 static int __init initcall_debug_setup(char *str)
671 __setup("initcall_debug", initcall_debug_setup);
673 extern initcall_t __initcall_start[], __initcall_end[];
675 static void __init do_initcalls(void)
678 int count = preempt_count();
680 for (call = __initcall_start; call < __initcall_end; call++) {
681 ktime_t t0, t1, delta;
686 if (initcall_debug) {
687 printk("Calling initcall 0x%p", *call);
688 print_fn_descriptor_symbol(": %s()",
689 (unsigned long) *call);
696 if (initcall_debug) {
698 delta = ktime_sub(t1, t0);
700 printk("initcall 0x%p", *call);
701 print_fn_descriptor_symbol(": %s()",
702 (unsigned long) *call);
703 printk(" returned %d.\n", result);
705 printk("initcall 0x%p ran for %Ld msecs: ",
706 *call, (unsigned long long)delta.tv64 >> 20);
707 print_fn_descriptor_symbol("%s()\n",
708 (unsigned long) *call);
711 if (result && result != -ENODEV && initcall_debug) {
712 sprintf(msgbuf, "error code %d", result);
715 if (preempt_count() != count) {
716 msg = "preemption imbalance";
717 preempt_count() = count;
719 if (irqs_disabled()) {
720 msg = "disabled interrupts";
724 printk(KERN_WARNING "initcall at 0x%p", *call);
725 print_fn_descriptor_symbol(": %s()",
726 (unsigned long) *call);
727 printk(": returned with %s\n", msg);
731 /* Make sure there is no pending stuff from the initcall sequence */
732 flush_scheduled_work();
736 * Ok, the machine is now initialized. None of the devices
737 * have been touched yet, but the CPU subsystem is up and
738 * running, and memory and process management works.
740 * Now we can finally start doing some real work..
742 static void __init do_basic_setup(void)
744 /* drivers will send hotplug events */
746 usermodehelper_init();
752 static int __initdata nosoftlockup;
754 static int __init nosoftlockup_setup(char *str)
759 __setup("nosoftlockup", nosoftlockup_setup);
761 static void __init do_pre_smp_initcalls(void)
763 extern int spawn_ksoftirqd(void);
768 spawn_softlockup_task();
771 static void run_init_process(char *init_filename)
773 argv_init[0] = init_filename;
774 kernel_execve(init_filename, argv_init, envp_init);
777 /* This is a non __init function. Force it to be noinline otherwise gcc
778 * makes it inline to init() and it becomes part of init.text section
780 static int noinline init_post(void)
785 system_state = SYSTEM_RUNNING;
786 numa_default_policy();
788 if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0)
789 printk(KERN_WARNING "Warning: unable to open an initial console.\n");
794 if (ramdisk_execute_command) {
795 run_init_process(ramdisk_execute_command);
796 printk(KERN_WARNING "Failed to execute %s\n",
797 ramdisk_execute_command);
801 * We try each of these until one succeeds.
803 * The Bourne shell can be used instead of init if we are
804 * trying to recover a really broken machine.
806 if (execute_command) {
807 run_init_process(execute_command);
808 printk(KERN_WARNING "Failed to execute %s. Attempting "
809 "defaults...\n", execute_command);
811 run_init_process("/sbin/init");
812 run_init_process("/etc/init");
813 run_init_process("/bin/init");
814 run_init_process("/bin/sh");
816 panic("No init found. Try passing init= option to kernel.");
819 static int __init kernel_init(void * unused)
823 * init can run on any cpu.
825 set_cpus_allowed(current, CPU_MASK_ALL);
827 * Tell the world that we're going to be the grim
828 * reaper of innocent orphaned children.
830 * We don't want people to have to make incorrect
831 * assumptions about where in the task array this
834 init_pid_ns.child_reaper = current;
836 __set_special_pids(1, 1);
837 cad_pid = task_pid(current);
839 smp_prepare_cpus(setup_max_cpus);
841 do_pre_smp_initcalls();
851 * check if there is an early userspace init. If yes, let it do all
855 if (!ramdisk_execute_command)
856 ramdisk_execute_command = "/init";
858 if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) {
859 ramdisk_execute_command = NULL;
864 * Ok, we have completed the initial bootup, and
865 * we're essentially up and running. Get rid of the
866 * initmem segments and start the user-mode stuff..