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/bootmem.h>
26 #include <linux/tty.h>
27 #include <linux/gfp.h>
28 #include <linux/percpu.h>
29 #include <linux/kmod.h>
30 #include <linux/kernel_stat.h>
31 #include <linux/start_kernel.h>
32 #include <linux/security.h>
33 #include <linux/smp.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/debugobjects.h>
56 #include <linux/lockdep.h>
57 #include <linux/pid_namespace.h>
58 #include <linux/device.h>
59 #include <linux/kthread.h>
60 #include <linux/sched.h>
61 #include <linux/signal.h>
62 #include <linux/idr.h>
66 #include <asm/setup.h>
67 #include <asm/sections.h>
68 #include <asm/cacheflush.h>
70 #ifdef CONFIG_X86_LOCAL_APIC
75 * This is one of the first .c files built. Error out early if we have compiler
79 #if __GNUC__ == 4 && __GNUC_MINOR__ == 1 && __GNUC_PATCHLEVEL__ == 0
80 #warning gcc-4.1.0 is known to miscompile the kernel. A different compiler version is recommended.
83 static int kernel_init(void *);
85 extern void init_IRQ(void);
86 extern void fork_init(unsigned long);
87 extern void mca_init(void);
88 extern void sbus_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 enum system_states system_state;
106 EXPORT_SYMBOL(system_state);
109 * Boot command-line arguments
111 #define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT
112 #define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT
114 extern void time_init(void);
115 /* Default late time init is NULL. archs can override this later. */
116 void (*late_time_init)(void);
117 extern void softirq_init(void);
119 /* Untouched command line saved by arch-specific code. */
120 char __initdata boot_command_line[COMMAND_LINE_SIZE];
121 /* Untouched saved command line (eg. for /proc) */
122 char *saved_command_line;
123 /* Command line for parameter parsing */
124 static char *static_command_line;
126 static char *execute_command;
127 static char *ramdisk_execute_command;
130 /* Setup configured maximum number of CPUs to activate */
131 unsigned int __initdata setup_max_cpus = NR_CPUS;
134 * Setup routine for controlling SMP activation
136 * Command-line option of "nosmp" or "maxcpus=0" will disable SMP
137 * activation entirely (the MPS table probe still happens, though).
139 * Command-line option of "maxcpus=<NUM>", where <NUM> is an integer
140 * greater than 0, limits the maximum number of CPUs activated in
143 #ifndef CONFIG_X86_IO_APIC
144 static inline void disable_ioapic_setup(void) {};
147 static int __init nosmp(char *str)
150 disable_ioapic_setup();
154 early_param("nosmp", nosmp);
156 static int __init maxcpus(char *str)
158 get_option(&str, &setup_max_cpus);
159 if (setup_max_cpus == 0)
160 disable_ioapic_setup();
165 early_param("maxcpus", maxcpus);
167 #define setup_max_cpus NR_CPUS
171 * If set, this is an indication to the drivers that reset the underlying
172 * device before going ahead with the initialization otherwise driver might
173 * rely on the BIOS and skip the reset operation.
175 * This is useful if kernel is booting in an unreliable environment.
176 * For ex. kdump situaiton where previous kernel has crashed, BIOS has been
177 * skipped and devices will be in unknown state.
179 unsigned int reset_devices;
180 EXPORT_SYMBOL(reset_devices);
182 static int __init set_reset_devices(char *str)
188 __setup("reset_devices", set_reset_devices);
190 static char * argv_init[MAX_INIT_ARGS+2] = { "init", NULL, };
191 char * envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, };
192 static const char *panic_later, *panic_param;
194 extern struct obs_kernel_param __setup_start[], __setup_end[];
196 static int __init obsolete_checksetup(char *line)
198 struct obs_kernel_param *p;
199 int had_early_param = 0;
203 int n = strlen(p->str);
204 if (!strncmp(line, p->str, n)) {
206 /* Already done in parse_early_param?
207 * (Needs exact match on param part).
208 * Keep iterating, as we can have early
209 * params and __setups of same names 8( */
210 if (line[n] == '\0' || line[n] == '=')
212 } else if (!p->setup_func) {
213 printk(KERN_WARNING "Parameter %s is obsolete,"
214 " ignored\n", p->str);
216 } else if (p->setup_func(line + n))
220 } while (p < __setup_end);
222 return had_early_param;
226 * This should be approx 2 Bo*oMips to start (note initial shift), and will
227 * still work even if initially too large, it will just take slightly longer
229 unsigned long loops_per_jiffy = (1<<12);
231 EXPORT_SYMBOL(loops_per_jiffy);
233 static int __init debug_kernel(char *str)
235 console_loglevel = 10;
239 static int __init quiet_kernel(char *str)
241 console_loglevel = 4;
245 early_param("debug", debug_kernel);
246 early_param("quiet", quiet_kernel);
248 static int __init loglevel(char *str)
250 get_option(&str, &console_loglevel);
254 early_param("loglevel", loglevel);
257 * Unknown boot options get handed to init, unless they look like
260 static int __init unknown_bootoption(char *param, char *val)
262 /* Change NUL term back to "=", to make "param" the whole string. */
264 /* param=val or param="val"? */
265 if (val == param+strlen(param)+1)
267 else if (val == param+strlen(param)+2) {
269 memmove(val-1, val, strlen(val)+1);
275 /* Handle obsolete-style parameters */
276 if (obsolete_checksetup(param))
280 * Preemptive maintenance for "why didn't my misspelled command
283 if (strchr(param, '.') && (!val || strchr(param, '.') < val)) {
284 printk(KERN_ERR "Unknown boot option `%s': ignoring\n", param);
292 /* Environment option */
294 for (i = 0; envp_init[i]; i++) {
295 if (i == MAX_INIT_ENVS) {
296 panic_later = "Too many boot env vars at `%s'";
299 if (!strncmp(param, envp_init[i], val - param))
302 envp_init[i] = param;
304 /* Command line option */
306 for (i = 0; argv_init[i]; i++) {
307 if (i == MAX_INIT_ARGS) {
308 panic_later = "Too many boot init vars at `%s'";
312 argv_init[i] = param;
317 #ifdef CONFIG_DEBUG_PAGEALLOC
318 int __read_mostly debug_pagealloc_enabled = 0;
321 static int __init init_setup(char *str)
325 execute_command = str;
327 * In case LILO is going to boot us with default command line,
328 * it prepends "auto" before the whole cmdline which makes
329 * the shell think it should execute a script with such name.
330 * So we ignore all arguments entered _before_ init=... [MJ]
332 for (i = 1; i < MAX_INIT_ARGS; i++)
336 __setup("init=", init_setup);
338 static int __init rdinit_setup(char *str)
342 ramdisk_execute_command = str;
343 /* See "auto" comment in init_setup */
344 for (i = 1; i < MAX_INIT_ARGS; i++)
348 __setup("rdinit=", rdinit_setup);
352 #ifdef CONFIG_X86_LOCAL_APIC
353 static void __init smp_init(void)
355 APIC_init_uniprocessor();
358 #define smp_init() do { } while (0)
361 static inline void setup_per_cpu_areas(void) { }
362 static inline void setup_nr_cpu_ids(void) { }
363 static inline void smp_prepare_cpus(unsigned int maxcpus) { }
367 #if NR_CPUS > BITS_PER_LONG
368 cpumask_t cpu_mask_all __read_mostly = CPU_MASK_ALL;
369 EXPORT_SYMBOL(cpu_mask_all);
372 /* Setup number of possible processor ids */
373 int nr_cpu_ids __read_mostly = NR_CPUS;
374 EXPORT_SYMBOL(nr_cpu_ids);
376 /* An arch may set nr_cpu_ids earlier if needed, so this would be redundant */
377 static void __init setup_nr_cpu_ids(void)
379 int cpu, highest_cpu = 0;
381 for_each_possible_cpu(cpu)
384 nr_cpu_ids = highest_cpu + 1;
387 #ifndef CONFIG_HAVE_SETUP_PER_CPU_AREA
388 unsigned long __per_cpu_offset[NR_CPUS] __read_mostly;
390 EXPORT_SYMBOL(__per_cpu_offset);
392 static void __init setup_per_cpu_areas(void)
394 unsigned long size, i, old_size;
396 unsigned long nr_possible_cpus = num_possible_cpus();
397 unsigned long align = 1;
400 /* Copy section for each CPU (we discard the original) */
401 old_size = PERCPU_ENOUGH_ROOM;
402 da_size = per_cpu_dyn_array_size(&align);
403 align = max_t(unsigned long, PAGE_SIZE, align);
404 size = ALIGN(old_size + da_size, align);
405 ptr = alloc_bootmem_pages(size * nr_possible_cpus);
407 for_each_possible_cpu(i) {
408 __per_cpu_offset[i] = ptr - __per_cpu_start;
409 memcpy(ptr, __per_cpu_start, __per_cpu_end - __per_cpu_start);
410 per_cpu_alloc_dyn_array(i, ptr + old_size);
414 #endif /* CONFIG_HAVE_SETUP_PER_CPU_AREA */
416 /* Called by boot processor to activate the rest. */
417 static void __init smp_init(void)
422 * Set up the current CPU as possible to migrate to.
423 * The other ones will be done by cpu_up/cpu_down()
425 cpu = smp_processor_id();
426 cpu_set(cpu, cpu_active_map);
428 /* FIXME: This should be done in userspace --RR */
429 for_each_present_cpu(cpu) {
430 if (num_online_cpus() >= setup_max_cpus)
432 if (!cpu_online(cpu))
436 /* Any cleanup work */
437 printk(KERN_INFO "Brought up %ld CPUs\n", (long)num_online_cpus());
438 smp_cpus_done(setup_max_cpus);
444 * We need to store the untouched command line for future reference.
445 * We also need to store the touched command line since the parameter
446 * parsing is performed in place, and we should allow a component to
447 * store reference of name/value for future reference.
449 static void __init setup_command_line(char *command_line)
451 saved_command_line = alloc_bootmem(strlen (boot_command_line)+1);
452 static_command_line = alloc_bootmem(strlen (command_line)+1);
453 strcpy (saved_command_line, boot_command_line);
454 strcpy (static_command_line, command_line);
458 * We need to finalize in a non-__init function or else race conditions
459 * between the root thread and the init thread may cause start_kernel to
460 * be reaped by free_initmem before the root thread has proceeded to
463 * gcc-3.4 accidentally inlines this function, so use noinline.
466 static void noinline __init_refok rest_init(void)
467 __releases(kernel_lock)
471 kernel_thread(kernel_init, NULL, CLONE_FS | CLONE_SIGHAND);
472 numa_default_policy();
473 pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES);
474 kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns);
478 * The boot idle thread must execute schedule()
479 * at least once to get things moving:
481 init_idle_bootup_task(current);
482 preempt_enable_no_resched();
486 /* Call into cpu_idle with preempt disabled */
490 /* Check for early params. */
491 static int __init do_early_param(char *param, char *val)
493 struct obs_kernel_param *p;
495 for (p = __setup_start; p < __setup_end; p++) {
496 if ((p->early && strcmp(param, p->str) == 0) ||
497 (strcmp(param, "console") == 0 &&
498 strcmp(p->str, "earlycon") == 0)
500 if (p->setup_func(val) != 0)
502 "Malformed early option '%s'\n", param);
505 /* We accept everything at this stage. */
509 /* Arch code calls this early on, or if not, just before other parsing. */
510 void __init parse_early_param(void)
512 static __initdata int done = 0;
513 static __initdata char tmp_cmdline[COMMAND_LINE_SIZE];
518 /* All fall through to do_early_param. */
519 strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE);
520 parse_args("early options", tmp_cmdline, NULL, 0, do_early_param);
525 * Activate the first processor.
528 static void __init boot_cpu_init(void)
530 int cpu = smp_processor_id();
531 /* Mark the boot cpu "present", "online" etc for SMP and UP case */
532 cpu_set(cpu, cpu_online_map);
533 cpu_set(cpu, cpu_present_map);
534 cpu_set(cpu, cpu_possible_map);
537 void __init __weak smp_setup_processor_id(void)
541 void __init __weak thread_info_cache_init(void)
545 void pre_alloc_dyn_array(void)
547 #ifdef CONFIG_HAVE_DYN_ARRAY
548 unsigned long total_size = 0, size, phys;
549 unsigned long max_align = 1;
550 struct dyn_array **daa;
553 /* get the total size at first */
554 for (daa = __dyn_array_start ; daa < __dyn_array_end; daa++) {
555 struct dyn_array *da = *daa;
557 size = da->size * (*da->nr);
558 print_fn_descriptor_symbol("dyn_array %s ", da->name);
559 printk(KERN_CONT "size:%#lx nr:%d align:%#lx\n",
560 da->size, *da->nr, da->align);
561 total_size += roundup(size, da->align);
562 if (da->align > max_align)
563 max_align = da->align;
566 printk(KERN_DEBUG "dyn_array total_size: %#lx\n",
571 /* allocate them all together */
572 max_align = max_t(unsigned long, max_align, PAGE_SIZE);
573 ptr = __alloc_bootmem_nopanic(total_size, max_align, 0);
575 panic("Can not alloc dyn_alloc\n");
577 phys = virt_to_phys(ptr);
578 for (daa = __dyn_array_start ; daa < __dyn_array_end; daa++) {
579 struct dyn_array *da = *daa;
581 size = da->size * (*da->nr);
582 print_fn_descriptor_symbol("dyn_array %s ", da->name);
584 phys = roundup(phys, da->align);
585 *da->name = phys_to_virt(phys);
586 printk(KERN_CONT " ==> [%#lx - %#lx]\n", phys, phys + size);
594 #ifdef CONFIF_GENERIC_HARDIRQS
597 for (i = 0; i < NR_IRQS; i++)
603 unsigned long per_cpu_dyn_array_size(unsigned long *align)
605 unsigned long total_size = 0;
606 #ifdef CONFIG_HAVE_DYN_ARRAY
608 struct dyn_array **daa;
609 unsigned max_align = 1;
611 for (daa = __per_cpu_dyn_array_start ; daa < __per_cpu_dyn_array_end; daa++) {
612 struct dyn_array *da = *daa;
614 size = da->size * (*da->nr);
615 print_fn_descriptor_symbol("per_cpu_dyn_array %s ", da->name);
616 printk(KERN_CONT "size:%#lx nr:%d align:%#lx\n",
617 da->size, *da->nr, da->align);
618 total_size += roundup(size, da->align);
619 if (da->align > max_align)
620 max_align = da->align;
623 printk(KERN_DEBUG "per_cpu_dyn_array total_size: %#lx\n",
631 void per_cpu_alloc_dyn_array(int cpu, char *ptr)
633 #ifdef CONFIG_HAVE_DYN_ARRAY
634 unsigned long size, phys;
635 struct dyn_array **daa;
639 phys = virt_to_phys(ptr);
640 for (daa = __per_cpu_dyn_array_start ; daa < __per_cpu_dyn_array_end; daa++) {
641 struct dyn_array *da = *daa;
643 size = da->size * (*da->nr);
644 print_fn_descriptor_symbol("per_cpu_dyn_array %s ", da->name);
646 phys = roundup(phys, da->align);
647 addr = (unsigned long)da->name;
648 addr += per_cpu_offset(cpu);
649 array = (void **)addr;
650 *array = phys_to_virt(phys);
651 *da->name = *array; /* so init_work could use it directly */
652 printk(KERN_CONT " ==> [%#lx - %#lx]\n", phys, phys + size);
663 asmlinkage void __init start_kernel(void)
666 extern struct kernel_param __start___param[], __stop___param[];
668 smp_setup_processor_id();
671 * Need to run as early as possible, to initialize the
676 debug_objects_early_init();
680 early_boot_irqs_off();
681 early_init_irq_lock_class();
684 * Interrupts are still disabled. Do necessary setups, then
692 printk(linux_banner);
693 setup_arch(&command_line);
694 pre_alloc_dyn_array();
695 mm_init_owner(&init_mm, &init_task);
696 setup_command_line(command_line);
698 setup_per_cpu_areas();
700 smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */
703 * Set up the scheduler prior starting any interrupts (such as the
704 * timer interrupt). Full topology setup happens at smp_init()
705 * time - but meanwhile we still have a functioning scheduler.
709 * Disable preemption - early bootup scheduling is extremely
710 * fragile until we cpu_idle() for the first time.
713 build_all_zonelists();
715 printk(KERN_NOTICE "Kernel command line: %s\n", boot_command_line);
717 parse_args("Booting kernel", static_command_line, __start___param,
718 __stop___param - __start___param,
719 &unknown_bootoption);
720 if (!irqs_disabled()) {
721 printk(KERN_WARNING "start_kernel(): bug: interrupts were "
722 "enabled *very* early, fixing it\n");
737 if (!irqs_disabled())
738 printk("start_kernel(): bug: interrupts were enabled early\n");
739 early_boot_irqs_on();
743 * HACK ALERT! This is early. We're enabling the console before
744 * we've done PCI setups etc, and console_init() must be aware of
745 * this. But we do want output early, in case something goes wrong.
749 panic(panic_later, panic_param);
754 * Need to run this when irqs are enabled, because it wants
755 * to self-test [hard/soft]-irqs on/off lock inversion bugs
760 #ifdef CONFIG_BLK_DEV_INITRD
761 if (initrd_start && !initrd_below_start_ok &&
762 page_to_pfn(virt_to_page((void *)initrd_start)) < min_low_pfn) {
763 printk(KERN_CRIT "initrd overwritten (0x%08lx < 0x%08lx) - "
765 page_to_pfn(virt_to_page((void *)initrd_start)),
770 vfs_caches_init_early();
773 enable_debug_pagealloc();
776 debug_objects_mem_init();
778 setup_per_cpu_pageset();
784 pgtable_cache_init();
789 efi_enter_virtual_mode();
791 thread_info_cache_init();
792 fork_init(num_physpages);
798 vfs_caches_init(num_physpages);
801 /* rootfs populating might need page-writeback */
802 page_writeback_init();
803 #ifdef CONFIG_PROC_FS
808 taskstats_init_early();
813 acpi_early_init(); /* before LAPIC and SMP init */
815 /* Do the rest non-__init'ed, we're now alive */
819 static int initcall_debug;
821 static int __init initcall_debug_setup(char *str)
826 __setup("initcall_debug", initcall_debug_setup);
828 int do_one_initcall(initcall_t fn)
830 int count = preempt_count();
831 ktime_t t0, t1, delta;
835 if (initcall_debug) {
836 printk("calling %pF @ %i\n", fn, task_pid_nr(current));
842 if (initcall_debug) {
844 delta = ktime_sub(t1, t0);
846 printk("initcall %pF returned %d after %Ld msecs\n",
848 (unsigned long long) delta.tv64 >> 20);
853 if (result && result != -ENODEV && initcall_debug)
854 sprintf(msgbuf, "error code %d ", result);
856 if (preempt_count() != count) {
857 strlcat(msgbuf, "preemption imbalance ", sizeof(msgbuf));
858 preempt_count() = count;
860 if (irqs_disabled()) {
861 strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf));
865 printk("initcall %pF returned with %s\n", fn, msgbuf);
872 extern initcall_t __initcall_start[], __initcall_end[], __early_initcall_end[];
874 static void __init do_initcalls(void)
878 for (call = __early_initcall_end; call < __initcall_end; call++)
879 do_one_initcall(*call);
881 /* Make sure there is no pending stuff from the initcall sequence */
882 flush_scheduled_work();
886 * Ok, the machine is now initialized. None of the devices
887 * have been touched yet, but the CPU subsystem is up and
888 * running, and memory and process management works.
890 * Now we can finally start doing some real work..
892 static void __init do_basic_setup(void)
894 rcu_init_sched(); /* needed by module_init stage. */
895 /* drivers will send hotplug events */
897 usermodehelper_init();
903 static void __init do_pre_smp_initcalls(void)
907 for (call = __initcall_start; call < __early_initcall_end; call++)
908 do_one_initcall(*call);
911 static void run_init_process(char *init_filename)
913 argv_init[0] = init_filename;
914 kernel_execve(init_filename, argv_init, envp_init);
917 /* This is a non __init function. Force it to be noinline otherwise gcc
918 * makes it inline to init() and it becomes part of init.text section
920 static int noinline init_post(void)
925 system_state = SYSTEM_RUNNING;
926 numa_default_policy();
928 if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0)
929 printk(KERN_WARNING "Warning: unable to open an initial console.\n");
934 current->signal->flags |= SIGNAL_UNKILLABLE;
936 if (ramdisk_execute_command) {
937 run_init_process(ramdisk_execute_command);
938 printk(KERN_WARNING "Failed to execute %s\n",
939 ramdisk_execute_command);
943 * We try each of these until one succeeds.
945 * The Bourne shell can be used instead of init if we are
946 * trying to recover a really broken machine.
948 if (execute_command) {
949 run_init_process(execute_command);
950 printk(KERN_WARNING "Failed to execute %s. Attempting "
951 "defaults...\n", execute_command);
953 run_init_process("/sbin/init");
954 run_init_process("/etc/init");
955 run_init_process("/bin/init");
956 run_init_process("/bin/sh");
958 panic("No init found. Try passing init= option to kernel.");
961 static int __init kernel_init(void * unused)
965 * init can run on any cpu.
967 set_cpus_allowed_ptr(current, CPU_MASK_ALL_PTR);
969 * Tell the world that we're going to be the grim
970 * reaper of innocent orphaned children.
972 * We don't want people to have to make incorrect
973 * assumptions about where in the task array this
976 init_pid_ns.child_reaper = current;
978 cad_pid = task_pid(current);
980 smp_prepare_cpus(setup_max_cpus);
982 do_pre_smp_initcalls();
992 * check if there is an early userspace init. If yes, let it do all
996 if (!ramdisk_execute_command)
997 ramdisk_execute_command = "/init";
999 if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) {
1000 ramdisk_execute_command = NULL;
1001 prepare_namespace();
1005 * Ok, we have completed the initial bootup, and
1006 * we're essentially up and running. Get rid of the
1007 * initmem segments and start the user-mode stuff..