[TCP]: Seperate DSACK from SACK fast path
[linux-2.6.22.y-op.git] / init / main.c
blob8b4a7d76916235df139b98bd41019a228615a2dd
1 /*
2 * linux/init/main.c
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/efi.h>
43 #include <linux/taskstats_kern.h>
44 #include <linux/delayacct.h>
45 #include <linux/unistd.h>
46 #include <linux/rmap.h>
47 #include <linux/mempolicy.h>
48 #include <linux/key.h>
49 #include <linux/unwind.h>
50 #include <linux/buffer_head.h>
51 #include <linux/debug_locks.h>
52 #include <linux/lockdep.h>
53 #include <linux/pid_namespace.h>
54 #include <linux/device.h>
56 #include <asm/io.h>
57 #include <asm/bugs.h>
58 #include <asm/setup.h>
59 #include <asm/sections.h>
60 #include <asm/cacheflush.h>
62 #ifdef CONFIG_X86_LOCAL_APIC
63 #include <asm/smp.h>
64 #endif
67 * This is one of the first .c files built. Error out early if we have compiler
68 * trouble.
70 * Versions of gcc older than that listed below may actually compile and link
71 * okay, but the end product can have subtle run time bugs. To avoid associated
72 * bogus bug reports, we flatly refuse to compile with a gcc that is known to be
73 * too old from the very beginning.
75 #if (__GNUC__ < 3) || (__GNUC__ == 3 && __GNUC_MINOR__ < 2)
76 #error Sorry, your GCC is too old. It builds incorrect kernels.
77 #endif
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.
81 #endif
83 static int 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 sysctl_init(void);
90 extern void signals_init(void);
91 extern void pidhash_init(void);
92 extern void pidmap_init(void);
93 extern void prio_tree_init(void);
94 extern void radix_tree_init(void);
95 extern void free_initmem(void);
96 extern void prepare_namespace(void);
97 #ifdef CONFIG_ACPI
98 extern void acpi_early_init(void);
99 #else
100 static inline void acpi_early_init(void) { }
101 #endif
102 #ifndef CONFIG_DEBUG_RODATA
103 static inline void mark_rodata_ro(void) { }
104 #endif
106 #ifdef CONFIG_TC
107 extern void tc_init(void);
108 #endif
110 enum system_states system_state;
111 EXPORT_SYMBOL(system_state);
114 * Boot command-line arguments
116 #define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT
117 #define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT
119 extern void time_init(void);
120 /* Default late time init is NULL. archs can override this later. */
121 void (*late_time_init)(void);
122 extern void softirq_init(void);
124 /* Untouched command line (eg. for /proc) saved by arch-specific code. */
125 char saved_command_line[COMMAND_LINE_SIZE];
127 static char *execute_command;
128 static char *ramdisk_execute_command;
130 /* Setup configured maximum number of CPUs to activate */
131 static unsigned int max_cpus = NR_CPUS;
134 * If set, this is an indication to the drivers that reset the underlying
135 * device before going ahead with the initialization otherwise driver might
136 * rely on the BIOS and skip the reset operation.
138 * This is useful if kernel is booting in an unreliable environment.
139 * For ex. kdump situaiton where previous kernel has crashed, BIOS has been
140 * skipped and devices will be in unknown state.
142 unsigned int reset_devices;
143 EXPORT_SYMBOL(reset_devices);
146 * Setup routine for controlling SMP activation
148 * Command-line option of "nosmp" or "maxcpus=0" will disable SMP
149 * activation entirely (the MPS table probe still happens, though).
151 * Command-line option of "maxcpus=<NUM>", where <NUM> is an integer
152 * greater than 0, limits the maximum number of CPUs activated in
153 * SMP mode to <NUM>.
155 static int __init nosmp(char *str)
157 max_cpus = 0;
158 return 1;
161 __setup("nosmp", nosmp);
163 static int __init maxcpus(char *str)
165 get_option(&str, &max_cpus);
166 return 1;
169 __setup("maxcpus=", maxcpus);
171 static int __init set_reset_devices(char *str)
173 reset_devices = 1;
174 return 1;
177 __setup("reset_devices", set_reset_devices);
179 static char * argv_init[MAX_INIT_ARGS+2] = { "init", NULL, };
180 char * envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, };
181 static const char *panic_later, *panic_param;
183 extern struct obs_kernel_param __setup_start[], __setup_end[];
185 static int __init obsolete_checksetup(char *line)
187 struct obs_kernel_param *p;
188 int had_early_param = 0;
190 p = __setup_start;
191 do {
192 int n = strlen(p->str);
193 if (!strncmp(line, p->str, n)) {
194 if (p->early) {
195 /* Already done in parse_early_param?
196 * (Needs exact match on param part).
197 * Keep iterating, as we can have early
198 * params and __setups of same names 8( */
199 if (line[n] == '\0' || line[n] == '=')
200 had_early_param = 1;
201 } else if (!p->setup_func) {
202 printk(KERN_WARNING "Parameter %s is obsolete,"
203 " ignored\n", p->str);
204 return 1;
205 } else if (p->setup_func(line + n))
206 return 1;
208 p++;
209 } while (p < __setup_end);
211 return had_early_param;
215 * This should be approx 2 Bo*oMips to start (note initial shift), and will
216 * still work even if initially too large, it will just take slightly longer
218 unsigned long loops_per_jiffy = (1<<12);
220 EXPORT_SYMBOL(loops_per_jiffy);
222 static int __init debug_kernel(char *str)
224 if (*str)
225 return 0;
226 console_loglevel = 10;
227 return 1;
230 static int __init quiet_kernel(char *str)
232 if (*str)
233 return 0;
234 console_loglevel = 4;
235 return 1;
238 __setup("debug", debug_kernel);
239 __setup("quiet", quiet_kernel);
241 static int __init loglevel(char *str)
243 get_option(&str, &console_loglevel);
244 return 1;
247 __setup("loglevel=", loglevel);
250 * Unknown boot options get handed to init, unless they look like
251 * failed parameters
253 static int __init unknown_bootoption(char *param, char *val)
255 /* Change NUL term back to "=", to make "param" the whole string. */
256 if (val) {
257 /* param=val or param="val"? */
258 if (val == param+strlen(param)+1)
259 val[-1] = '=';
260 else if (val == param+strlen(param)+2) {
261 val[-2] = '=';
262 memmove(val-1, val, strlen(val)+1);
263 val--;
264 } else
265 BUG();
268 /* Handle obsolete-style parameters */
269 if (obsolete_checksetup(param))
270 return 0;
273 * Preemptive maintenance for "why didn't my mispelled command
274 * line work?"
276 if (strchr(param, '.') && (!val || strchr(param, '.') < val)) {
277 printk(KERN_ERR "Unknown boot option `%s': ignoring\n", param);
278 return 0;
281 if (panic_later)
282 return 0;
284 if (val) {
285 /* Environment option */
286 unsigned int i;
287 for (i = 0; envp_init[i]; i++) {
288 if (i == MAX_INIT_ENVS) {
289 panic_later = "Too many boot env vars at `%s'";
290 panic_param = param;
292 if (!strncmp(param, envp_init[i], val - param))
293 break;
295 envp_init[i] = param;
296 } else {
297 /* Command line option */
298 unsigned int i;
299 for (i = 0; argv_init[i]; i++) {
300 if (i == MAX_INIT_ARGS) {
301 panic_later = "Too many boot init vars at `%s'";
302 panic_param = param;
305 argv_init[i] = param;
307 return 0;
310 static int __init init_setup(char *str)
312 unsigned int i;
314 execute_command = str;
316 * In case LILO is going to boot us with default command line,
317 * it prepends "auto" before the whole cmdline which makes
318 * the shell think it should execute a script with such name.
319 * So we ignore all arguments entered _before_ init=... [MJ]
321 for (i = 1; i < MAX_INIT_ARGS; i++)
322 argv_init[i] = NULL;
323 return 1;
325 __setup("init=", init_setup);
327 static int __init rdinit_setup(char *str)
329 unsigned int i;
331 ramdisk_execute_command = str;
332 /* See "auto" comment in init_setup */
333 for (i = 1; i < MAX_INIT_ARGS; i++)
334 argv_init[i] = NULL;
335 return 1;
337 __setup("rdinit=", rdinit_setup);
339 #ifndef CONFIG_SMP
341 #ifdef CONFIG_X86_LOCAL_APIC
342 static void __init smp_init(void)
344 APIC_init_uniprocessor();
346 #else
347 #define smp_init() do { } while (0)
348 #endif
350 static inline void setup_per_cpu_areas(void) { }
351 static inline void smp_prepare_cpus(unsigned int maxcpus) { }
353 #else
355 #ifdef __GENERIC_PER_CPU
356 unsigned long __per_cpu_offset[NR_CPUS] __read_mostly;
358 EXPORT_SYMBOL(__per_cpu_offset);
360 static void __init setup_per_cpu_areas(void)
362 unsigned long size, i;
363 char *ptr;
364 unsigned long nr_possible_cpus = num_possible_cpus();
366 /* Copy section for each CPU (we discard the original) */
367 size = ALIGN(__per_cpu_end - __per_cpu_start, SMP_CACHE_BYTES);
368 #ifdef CONFIG_MODULES
369 if (size < PERCPU_ENOUGH_ROOM)
370 size = PERCPU_ENOUGH_ROOM;
371 #endif
372 ptr = alloc_bootmem(size * nr_possible_cpus);
374 for_each_possible_cpu(i) {
375 __per_cpu_offset[i] = ptr - __per_cpu_start;
376 memcpy(ptr, __per_cpu_start, __per_cpu_end - __per_cpu_start);
377 ptr += size;
380 #endif /* !__GENERIC_PER_CPU */
382 /* Called by boot processor to activate the rest. */
383 static void __init smp_init(void)
385 unsigned int i;
387 /* FIXME: This should be done in userspace --RR */
388 for_each_present_cpu(i) {
389 if (num_online_cpus() >= max_cpus)
390 break;
391 if (!cpu_online(i))
392 cpu_up(i);
395 /* Any cleanup work */
396 printk(KERN_INFO "Brought up %ld CPUs\n", (long)num_online_cpus());
397 smp_cpus_done(max_cpus);
398 #if 0
399 /* Get other processors into their bootup holding patterns. */
401 smp_commence();
402 #endif
405 #endif
408 * We need to finalize in a non-__init function or else race conditions
409 * between the root thread and the init thread may cause start_kernel to
410 * be reaped by free_initmem before the root thread has proceeded to
411 * cpu_idle.
413 * gcc-3.4 accidentally inlines this function, so use noinline.
416 static void noinline rest_init(void)
417 __releases(kernel_lock)
419 kernel_thread(init, NULL, CLONE_FS | CLONE_SIGHAND);
420 numa_default_policy();
421 unlock_kernel();
424 * The boot idle thread must execute schedule()
425 * at least one to get things moving:
427 preempt_enable_no_resched();
428 schedule();
429 preempt_disable();
431 /* Call into cpu_idle with preempt disabled */
432 cpu_idle();
435 /* Check for early params. */
436 static int __init do_early_param(char *param, char *val)
438 struct obs_kernel_param *p;
440 for (p = __setup_start; p < __setup_end; p++) {
441 if (p->early && strcmp(param, p->str) == 0) {
442 if (p->setup_func(val) != 0)
443 printk(KERN_WARNING
444 "Malformed early option '%s'\n", param);
447 /* We accept everything at this stage. */
448 return 0;
451 /* Arch code calls this early on, or if not, just before other parsing. */
452 void __init parse_early_param(void)
454 static __initdata int done = 0;
455 static __initdata char tmp_cmdline[COMMAND_LINE_SIZE];
457 if (done)
458 return;
460 /* All fall through to do_early_param. */
461 strlcpy(tmp_cmdline, saved_command_line, COMMAND_LINE_SIZE);
462 parse_args("early options", tmp_cmdline, NULL, 0, do_early_param);
463 done = 1;
467 * Activate the first processor.
470 static void __init boot_cpu_init(void)
472 int cpu = smp_processor_id();
473 /* Mark the boot cpu "present", "online" etc for SMP and UP case */
474 cpu_set(cpu, cpu_online_map);
475 cpu_set(cpu, cpu_present_map);
476 cpu_set(cpu, cpu_possible_map);
479 void __init __attribute__((weak)) smp_setup_processor_id(void)
483 asmlinkage void __init start_kernel(void)
485 char * command_line;
486 extern struct kernel_param __start___param[], __stop___param[];
488 smp_setup_processor_id();
491 * Need to run as early as possible, to initialize the
492 * lockdep hash:
494 unwind_init();
495 lockdep_init();
497 local_irq_disable();
498 early_boot_irqs_off();
499 early_init_irq_lock_class();
502 * Interrupts are still disabled. Do necessary setups, then
503 * enable them
505 lock_kernel();
506 boot_cpu_init();
507 page_address_init();
508 printk(KERN_NOTICE);
509 printk(linux_banner);
510 setup_arch(&command_line);
511 unwind_setup();
512 setup_per_cpu_areas();
513 smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */
516 * Set up the scheduler prior starting any interrupts (such as the
517 * timer interrupt). Full topology setup happens at smp_init()
518 * time - but meanwhile we still have a functioning scheduler.
520 sched_init();
522 * Disable preemption - early bootup scheduling is extremely
523 * fragile until we cpu_idle() for the first time.
525 preempt_disable();
526 build_all_zonelists();
527 page_alloc_init();
528 printk(KERN_NOTICE "Kernel command line: %s\n", saved_command_line);
529 parse_early_param();
530 parse_args("Booting kernel", command_line, __start___param,
531 __stop___param - __start___param,
532 &unknown_bootoption);
533 if (!irqs_disabled()) {
534 printk(KERN_WARNING "start_kernel(): bug: interrupts were "
535 "enabled *very* early, fixing it\n");
536 local_irq_disable();
538 sort_main_extable();
539 trap_init();
540 rcu_init();
541 init_IRQ();
542 pidhash_init();
543 init_timers();
544 hrtimers_init();
545 softirq_init();
546 timekeeping_init();
547 time_init();
548 profile_init();
549 if (!irqs_disabled())
550 printk("start_kernel(): bug: interrupts were enabled early\n");
551 early_boot_irqs_on();
552 local_irq_enable();
555 * HACK ALERT! This is early. We're enabling the console before
556 * we've done PCI setups etc, and console_init() must be aware of
557 * this. But we do want output early, in case something goes wrong.
559 console_init();
560 if (panic_later)
561 panic(panic_later, panic_param);
563 lockdep_info();
566 * Need to run this when irqs are enabled, because it wants
567 * to self-test [hard/soft]-irqs on/off lock inversion bugs
568 * too:
570 locking_selftest();
572 #ifdef CONFIG_BLK_DEV_INITRD
573 if (initrd_start && !initrd_below_start_ok &&
574 initrd_start < min_low_pfn << PAGE_SHIFT) {
575 printk(KERN_CRIT "initrd overwritten (0x%08lx < 0x%08lx) - "
576 "disabling it.\n",initrd_start,min_low_pfn << PAGE_SHIFT);
577 initrd_start = 0;
579 #endif
580 vfs_caches_init_early();
581 cpuset_init_early();
582 mem_init();
583 kmem_cache_init();
584 setup_per_cpu_pageset();
585 numa_policy_init();
586 if (late_time_init)
587 late_time_init();
588 calibrate_delay();
589 pidmap_init();
590 pgtable_cache_init();
591 prio_tree_init();
592 anon_vma_init();
593 #ifdef CONFIG_X86
594 if (efi_enabled)
595 efi_enter_virtual_mode();
596 #endif
597 fork_init(num_physpages);
598 proc_caches_init();
599 buffer_init();
600 unnamed_dev_init();
601 key_init();
602 security_init();
603 vfs_caches_init(num_physpages);
604 radix_tree_init();
605 signals_init();
606 /* rootfs populating might need page-writeback */
607 page_writeback_init();
608 #ifdef CONFIG_PROC_FS
609 proc_root_init();
610 #endif
611 cpuset_init();
612 taskstats_init_early();
613 delayacct_init();
615 check_bugs();
617 acpi_early_init(); /* before LAPIC and SMP init */
619 /* Do the rest non-__init'ed, we're now alive */
620 rest_init();
623 static int __initdata initcall_debug;
625 static int __init initcall_debug_setup(char *str)
627 initcall_debug = 1;
628 return 1;
630 __setup("initcall_debug", initcall_debug_setup);
632 extern initcall_t __initcall_start[], __initcall_end[];
634 static void __init do_initcalls(void)
636 initcall_t *call;
637 int count = preempt_count();
639 for (call = __initcall_start; call < __initcall_end; call++) {
640 char *msg = NULL;
641 char msgbuf[40];
642 int result;
644 if (initcall_debug) {
645 printk("Calling initcall 0x%p", *call);
646 print_fn_descriptor_symbol(": %s()",
647 (unsigned long) *call);
648 printk("\n");
651 result = (*call)();
653 if (result && result != -ENODEV && initcall_debug) {
654 sprintf(msgbuf, "error code %d", result);
655 msg = msgbuf;
657 if (preempt_count() != count) {
658 msg = "preemption imbalance";
659 preempt_count() = count;
661 if (irqs_disabled()) {
662 msg = "disabled interrupts";
663 local_irq_enable();
665 if (msg) {
666 printk(KERN_WARNING "initcall at 0x%p", *call);
667 print_fn_descriptor_symbol(": %s()",
668 (unsigned long) *call);
669 printk(": returned with %s\n", msg);
673 /* Make sure there is no pending stuff from the initcall sequence */
674 flush_scheduled_work();
678 * Ok, the machine is now initialized. None of the devices
679 * have been touched yet, but the CPU subsystem is up and
680 * running, and memory and process management works.
682 * Now we can finally start doing some real work..
684 static void __init do_basic_setup(void)
686 /* drivers will send hotplug events */
687 init_workqueues();
688 usermodehelper_init();
689 driver_init();
691 #ifdef CONFIG_SYSCTL
692 sysctl_init();
693 #endif
695 do_initcalls();
698 static void __init do_pre_smp_initcalls(void)
700 extern int spawn_ksoftirqd(void);
701 #ifdef CONFIG_SMP
702 extern int migration_init(void);
704 migration_init();
705 #endif
706 spawn_ksoftirqd();
707 spawn_softlockup_task();
710 static void run_init_process(char *init_filename)
712 argv_init[0] = init_filename;
713 kernel_execve(init_filename, argv_init, envp_init);
716 static int init(void * unused)
718 lock_kernel();
720 * init can run on any cpu.
722 set_cpus_allowed(current, CPU_MASK_ALL);
724 * Tell the world that we're going to be the grim
725 * reaper of innocent orphaned children.
727 * We don't want people to have to make incorrect
728 * assumptions about where in the task array this
729 * can be found.
731 init_pid_ns.child_reaper = current;
733 cad_pid = task_pid(current);
735 smp_prepare_cpus(max_cpus);
737 do_pre_smp_initcalls();
739 smp_init();
740 sched_init_smp();
742 cpuset_init_smp();
744 do_basic_setup();
747 * check if there is an early userspace init. If yes, let it do all
748 * the work
751 if (!ramdisk_execute_command)
752 ramdisk_execute_command = "/init";
754 if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) {
755 ramdisk_execute_command = NULL;
756 prepare_namespace();
760 * Ok, we have completed the initial bootup, and
761 * we're essentially up and running. Get rid of the
762 * initmem segments and start the user-mode stuff..
764 free_initmem();
765 unlock_kernel();
766 mark_rodata_ro();
767 system_state = SYSTEM_RUNNING;
768 numa_default_policy();
770 if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0)
771 printk(KERN_WARNING "Warning: unable to open an initial console.\n");
773 (void) sys_dup(0);
774 (void) sys_dup(0);
776 if (ramdisk_execute_command) {
777 run_init_process(ramdisk_execute_command);
778 printk(KERN_WARNING "Failed to execute %s\n",
779 ramdisk_execute_command);
783 * We try each of these until one succeeds.
785 * The Bourne shell can be used instead of init if we are
786 * trying to recover a really broken machine.
788 if (execute_command) {
789 run_init_process(execute_command);
790 printk(KERN_WARNING "Failed to execute %s. Attempting "
791 "defaults...\n", execute_command);
793 run_init_process("/sbin/init");
794 run_init_process("/etc/init");
795 run_init_process("/bin/init");
796 run_init_process("/bin/sh");
798 panic("No init found. Try passing init= option to kernel.");