init/main.c

   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
   9  *  Simplified starting of init:  Michael A. Griffith <grif@acm.org>
  10  */
  11
  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
  60 #include <asm/io.h>
  61 #include <asm/bugs.h>
  62 #include <asm/setup.h>
  63 #include <asm/sections.h>
  64 #include <asm/cacheflush.h>
  65
  66 #ifdef CONFIG_X86_LOCAL_APIC
  67 #include <asm/smp.h>
  68 #endif
  69
  70 /*
  71  * This is one of the first .c files built. Error out early if we have compiler
  72  * trouble.
  73  *
  74  * Versions of gcc older than that listed below may actually compile and link
  75  * okay, but the end product can have subtle run time bugs.  To avoid associated
  76  * bogus bug reports, we flatly refuse to compile with a gcc that is known to be
  77  * too old from the very beginning.
  78  */
  79 #if (__GNUC__ < 3) || (__GNUC__ == 3 && __GNUC_MINOR__ < 2)
  80 #error Sorry, your GCC is too old. It builds incorrect kernels.
  81 #endif
  82
  83 #if __GNUC__ == 4 && __GNUC_MINOR__ == 1 && __GNUC_PATCHLEVEL__ == 0
  84 #warning gcc-4.1.0 is known to miscompile the kernel.  A different compiler version is recommended.
  85 #endif
  86
  87 static int kernel_init(void *);
  88
  89 extern void init_IRQ(void);
  90 extern void fork_init(unsigned long);
  91 extern void mca_init(void);
  92 extern void sbus_init(void);
  93 extern void signals_init(void);
  94 extern void pidhash_init(void);
  95 extern void pidmap_init(void);
  96 extern void prio_tree_init(void);
  97 extern void radix_tree_init(void);
  98 extern void free_initmem(void);
  99 #ifdef  CONFIG_ACPI
 100 extern void acpi_early_init(void);
 101 #else
 102 static inline void acpi_early_init(void) { }
 103 #endif
 104 #ifndef CONFIG_DEBUG_RODATA
 105 static inline void mark_rodata_ro(void) { }
 106 #endif
 107
 108 #ifdef CONFIG_TC
 109 extern void tc_init(void);
 110 #endif
 111
 112 enum system_states system_state;
 113 EXPORT_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
 121 extern void time_init(void);
 122 /* Default late time init is NULL. archs can override this later. */
 123 void (*late_time_init)(void);
 124 extern void softirq_init(void);
 125
 126 /* Untouched command line saved by arch-specific code. */
 127 char __initdata boot_command_line[COMMAND_LINE_SIZE];
 128 /* Untouched saved command line (eg. for /proc) */
 129 char *saved_command_line;
 130 /* Command line for parameter parsing */
 131 static char *static_command_line;
 132
 133 static char *execute_command;
 134 static char *ramdisk_execute_command;
 135
 136 #ifdef CONFIG_SMP
 137 /* Setup configured maximum number of CPUs to activate */
 138 static unsigned int __initdata max_cpus = NR_CPUS;
 139
 140 /*
 141  * Setup routine for controlling SMP activation
 142  *
 143  * Command-line option of "nosmp" or "maxcpus=0" will disable SMP
 144  * activation entirely (the MPS table probe still happens, though).
 145  *
 146  * Command-line option of "maxcpus=<NUM>", where <NUM> is an integer
 147  * greater than 0, limits the maximum number of CPUs activated in
 148  * SMP mode to <NUM>.
 149  */
 150 #ifndef CONFIG_X86_IO_APIC
 151 static inline void disable_ioapic_setup(void) {};
 152 #endif
 153
 154 static int __init nosmp(char *str)
 155 {
 156         max_cpus = 0;
 157         disable_ioapic_setup();
 158         return 0;
 159 }
 160
 161 early_param("nosmp", nosmp);
 162
 163 static int __init maxcpus(char *str)
 164 {
 165         get_option(&str, &max_cpus);
 166         if (max_cpus == 0)
 167                 disable_ioapic_setup();
 168
 169         return 0;
 170 }
 171
 172 early_param("maxcpus", maxcpus);
 173 #else
 174 #define max_cpus NR_CPUS
 175 #endif
 176
 177 /*
 178  * If set, this is an indication to the drivers that reset the underlying
 179  * device before going ahead with the initialization otherwise driver might
 180  * rely on the BIOS and skip the reset operation.
 181  *
 182  * This is useful if kernel is booting in an unreliable environment.
 183  * For ex. kdump situaiton where previous kernel has crashed, BIOS has been
 184  * skipped and devices will be in unknown state.
 185  */
 186 unsigned int reset_devices;
 187 EXPORT_SYMBOL(reset_devices);
 188
 189 static int __init set_reset_devices(char *str)
 190 {
 191         reset_devices = 1;
 192         return 1;
 193 }
 194
 195 __setup("reset_devices", set_reset_devices);
 196
 197 static char * argv_init[MAX_INIT_ARGS+2] = { "init", NULL, };
 198 char * envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, };
 199 static const char *panic_later, *panic_param;
 200
 201 extern struct obs_kernel_param __setup_start[], __setup_end[];
 202
 203 static int __init obsolete_checksetup(char *line)
 204 {
 205         struct obs_kernel_param *p;
 206         int had_early_param = 0;
 207
 208         p = __setup_start;
 209         do {
 210                 int n = strlen(p->str);
 211                 if (!strncmp(line, p->str, n)) {
 212                         if (p->early) {
 213                                 /* Already done in parse_early_param?
 214                                  * (Needs exact match on param part).
 215                                  * Keep iterating, as we can have early
 216                                  * params and __setups of same names 8( */
 217                                 if (line[n] == '\0' || line[n] == '=')
 218                                         had_early_param = 1;
 219                         } else if (!p->setup_func) {
 220                                 printk(KERN_WARNING "Parameter %s is obsolete,"
 221                                        " ignored\n", p->str);
 222                                 return 1;
 223                         } else if (p->setup_func(line + n))
 224                                 return 1;
 225                 }
 226                 p++;
 227         } while (p < __setup_end);
 228
 229         return had_early_param;
 230 }
 231
 232 /*
 233  * This should be approx 2 Bo*oMips to start (note initial shift), and will
 234  * still work even if initially too large, it will just take slightly longer
 235  */
 236 unsigned long loops_per_jiffy = (1<<12);
 237
 238 EXPORT_SYMBOL(loops_per_jiffy);
 239
 240 static int __init debug_kernel(char *str)
 241 {
 242         if (*str)
 243                 return 0;
 244         console_loglevel = 10;
 245         return 1;
 246 }
 247
 248 static int __init quiet_kernel(char *str)
 249 {
 250         if (*str)
 251                 return 0;
 252         console_loglevel = 4;
 253         return 1;
 254 }
 255
 256 __setup("debug", debug_kernel);
 257 __setup("quiet", quiet_kernel);
 258
 259 static int __init loglevel(char *str)
 260 {
 261         get_option(&str, &console_loglevel);
 262         return 1;
 263 }
 264
 265 __setup("loglevel=", loglevel);
 266
 267 /*
 268  * Unknown boot options get handed to init, unless they look like
 269  * failed parameters
 270  */
 271 static int __init unknown_bootoption(char *param, char *val)
 272 {
 273         /* Change NUL term back to "=", to make "param" the whole string. */
 274         if (val) {
 275                 /* param=val or param="val"? */
 276                 if (val == param+strlen(param)+1)
 277                         val[-1] = '=';
 278                 else if (val == param+strlen(param)+2) {
 279                         val[-2] = '=';
 280                         memmove(val-1, val, strlen(val)+1);
 281                         val--;
 282                 } else
 283                         BUG();
 284         }
 285
 286         /* Handle obsolete-style parameters */
 287         if (obsolete_checksetup(param))
 288                 return 0;
 289
 290         /*
 291          * Preemptive maintenance for "why didn't my mispelled command
 292          * line work?"
 293          */
 294         if (strchr(param, '.') && (!val || strchr(param, '.') < val)) {
 295                 printk(KERN_ERR "Unknown boot option `%s': ignoring\n", param);
 296                 return 0;
 297         }
 298
 299         if (panic_later)
 300                 return 0;
 301
 302         if (val) {
 303                 /* Environment option */
 304                 unsigned int i;
 305                 for (i = 0; envp_init[i]; i++) {
 306                         if (i == MAX_INIT_ENVS) {
 307                                 panic_later = "Too many boot env vars at `%s'";
 308                                 panic_param = param;
 309                         }
 310                         if (!strncmp(param, envp_init[i], val - param))
 311                                 break;
 312                 }
 313                 envp_init[i] = param;
 314         } else {
 315                 /* Command line option */
 316                 unsigned int i;
 317                 for (i = 0; argv_init[i]; i++) {
 318                         if (i == MAX_INIT_ARGS) {
 319                                 panic_later = "Too many boot init vars at `%s'";
 320                                 panic_param = param;
 321                         }
 322                 }
 323                 argv_init[i] = param;
 324         }
 325         return 0;
 326 }
 327
 328 static int __init init_setup(char *str)
 329 {
 330         unsigned int i;
 331
 332         execute_command = str;
 333         /*
 334          * In case LILO is going to boot us with default command line,
 335          * it prepends "auto" before the whole cmdline which makes
 336          * the shell think it should execute a script with such name.
 337          * So we ignore all arguments entered _before_ init=... [MJ]
 338          */
 339         for (i = 1; i < MAX_INIT_ARGS; i++)
 340                 argv_init[i] = NULL;
 341         return 1;
 342 }
 343 __setup("init=", init_setup);
 344
 345 static int __init rdinit_setup(char *str)
 346 {
 347         unsigned int i;
 348
 349         ramdisk_execute_command = str;
 350         /* See "auto" comment in init_setup */
 351         for (i = 1; i < MAX_INIT_ARGS; i++)
 352                 argv_init[i] = NULL;
 353         return 1;
 354 }
 355 __setup("rdinit=", rdinit_setup);
 356
 357 #ifndef CONFIG_SMP
 358
 359 #ifdef CONFIG_X86_LOCAL_APIC
 360 static void __init smp_init(void)
 361 {
 362         APIC_init_uniprocessor();
 363 }
 364 #else
 365 #define smp_init()      do { } while (0)
 366 #endif
 367
 368 static inline void setup_per_cpu_areas(void) { }
 369 static inline void smp_prepare_cpus(unsigned int maxcpus) { }
 370
 371 #else
 372
 373 #ifdef __GENERIC_PER_CPU
 374 unsigned long __per_cpu_offset[NR_CPUS] __read_mostly;
 375
 376 EXPORT_SYMBOL(__per_cpu_offset);
 377
 378 static void __init setup_per_cpu_areas(void)
 379 {
 380         unsigned long size, i;
 381         char *ptr;
 382         unsigned long nr_possible_cpus = num_possible_cpus();
 383
 384         /* Copy section for each CPU (we discard the original) */
 385         size = ALIGN(PERCPU_ENOUGH_ROOM, PAGE_SIZE);
 386         ptr = alloc_bootmem_pages(size * nr_possible_cpus);
 387
 388         for_each_possible_cpu(i) {
 389                 __per_cpu_offset[i] = ptr - __per_cpu_start;
 390                 memcpy(ptr, __per_cpu_start, __per_cpu_end - __per_cpu_start);
 391                 ptr += size;
 392         }
 393 }
 394 #endif /* !__GENERIC_PER_CPU */
 395
 396 /* Called by boot processor to activate the rest. */
 397 static void __init smp_init(void)
 398 {
 399         unsigned int cpu;
 400
 401         /* FIXME: This should be done in userspace --RR */
 402         for_each_present_cpu(cpu) {
 403                 if (num_online_cpus() >= max_cpus)
 404                         break;
 405                 if (!cpu_online(cpu))
 406                         cpu_up(cpu);
 407         }
 408
 409         /* Any cleanup work */
 410         printk(KERN_INFO "Brought up %ld CPUs\n", (long)num_online_cpus());
 411         smp_cpus_done(max_cpus);
 412 }
 413
 414 #endif
 415
 416 /*
 417  * We need to store the untouched command line for future reference.
 418  * We also need to store the touched command line since the parameter
 419  * parsing is performed in place, and we should allow a component to
 420  * store reference of name/value for future reference.
 421  */
 422 static void __init setup_command_line(char *command_line)
 423 {
 424         saved_command_line = alloc_bootmem(strlen (boot_command_line)+1);
 425         static_command_line = alloc_bootmem(strlen (command_line)+1);
 426         strcpy (saved_command_line, boot_command_line);
 427         strcpy (static_command_line, command_line);
 428 }
 429
 430 /*
 431  * We need to finalize in a non-__init function or else race conditions
 432  * between the root thread and the init thread may cause start_kernel to
 433  * be reaped by free_initmem before the root thread has proceeded to
 434  * cpu_idle.
 435  *
 436  * gcc-3.4 accidentally inlines this function, so use noinline.
 437  */
 438
 439 static void noinline __init_refok rest_init(void)
 440         __releases(kernel_lock)
 441 {
 442         int pid;
 443
 444         kernel_thread(kernel_init, NULL, CLONE_FS | CLONE_SIGHAND);
 445         numa_default_policy();
 446         pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES);
 447         kthreadd_task = find_task_by_pid(pid);
 448         unlock_kernel();
 449
 450         /*
 451          * The boot idle thread must execute schedule()
 452          * at least once to get things moving:
 453          */
 454         init_idle_bootup_task(current);
 455         preempt_enable_no_resched();
 456         schedule();
 457         preempt_disable();
 458
 459         /* Call into cpu_idle with preempt disabled */
 460         cpu_idle();
 461 }
 462
 463 /* Check for early params. */
 464 static int __init do_early_param(char *param, char *val)
 465 {
 466         struct obs_kernel_param *p;
 467
 468         for (p = __setup_start; p < __setup_end; p++) {
 469                 if ((p->early && strcmp(param, p->str) == 0) ||
 470                     (strcmp(param, "console") == 0 &&
 471                      strcmp(p->str, "earlycon") == 0)
 472                 ) {
 473                         if (p->setup_func(val) != 0)
 474                                 printk(KERN_WARNING
 475                                        "Malformed early option '%s'\n", param);
 476                 }
 477         }
 478         /* We accept everything at this stage. */
 479         return 0;
 480 }
 481
 482 /* Arch code calls this early on, or if not, just before other parsing. */
 483 void __init parse_early_param(void)
 484 {
 485         static __initdata int done = 0;
 486         static __initdata char tmp_cmdline[COMMAND_LINE_SIZE];
 487
 488         if (done)
 489                 return;
 490
 491         /* All fall through to do_early_param. */
 492         strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE);
 493         parse_args("early options", tmp_cmdline, NULL, 0, do_early_param);
 494         done = 1;
 495 }
 496
 497 /*
 498  *      Activate the first processor.
 499  */
 500
 501 static void __init boot_cpu_init(void)
 502 {
 503         int cpu = smp_processor_id();
 504         /* Mark the boot cpu "present", "online" etc for SMP and UP case */
 505         cpu_set(cpu, cpu_online_map);
 506         cpu_set(cpu, cpu_present_map);
 507         cpu_set(cpu, cpu_possible_map);
 508 }
 509
 510 void __init __attribute__((weak)) smp_setup_processor_id(void)
 511 {
 512 }
 513
 514 asmlinkage void __init start_kernel(void)
 515 {
 516         char * command_line;
 517         extern struct kernel_param __start___param[], __stop___param[];
 518
 519         smp_setup_processor_id();
 520
 521         /*
 522          * Need to run as early as possible, to initialize the
 523          * lockdep hash:
 524          */
 525         unwind_init();
 526         lockdep_init();
 527         cgroup_init_early();
 528
 529         local_irq_disable();
 530         early_boot_irqs_off();
 531         early_init_irq_lock_class();
 532
 533 /*
 534  * Interrupts are still disabled. Do necessary setups, then
 535  * enable them
 536  */
 537         lock_kernel();
 538         tick_init();
 539         boot_cpu_init();
 540         page_address_init();
 541         printk(KERN_NOTICE);
 542         printk(linux_banner);
 543         setup_arch(&command_line);
 544         setup_command_line(command_line);
 545         unwind_setup();
 546         setup_per_cpu_areas();
 547         smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */
 548
 549         /*
 550          * Set up the scheduler prior starting any interrupts (such as the
 551          * timer interrupt). Full topology setup happens at smp_init()
 552          * time - but meanwhile we still have a functioning scheduler.
 553          */
 554         sched_init();
 555         /*
 556          * Disable preemption - early bootup scheduling is extremely
 557          * fragile until we cpu_idle() for the first time.
 558          */
 559         preempt_disable();
 560         build_all_zonelists();
 561         page_alloc_init();
 562         printk(KERN_NOTICE "Kernel command line: %s\n", boot_command_line);
 563         parse_early_param();
 564         parse_args("Booting kernel", static_command_line, __start___param,
 565                    __stop___param - __start___param,
 566                    &unknown_bootoption);
 567         if (!irqs_disabled()) {
 568                 printk(KERN_WARNING "start_kernel(): bug: interrupts were "
 569                                 "enabled *very* early, fixing it\n");
 570                 local_irq_disable();
 571         }
 572         sort_main_extable();
 573         trap_init();
 574         rcu_init();
 575         init_IRQ();
 576         pidhash_init();
 577         init_timers();
 578         hrtimers_init();
 579         softirq_init();
 580         timekeeping_init();
 581         time_init();
 582         profile_init();
 583         if (!irqs_disabled())
 584                 printk("start_kernel(): bug: interrupts were enabled early\n");
 585         early_boot_irqs_on();
 586         local_irq_enable();
 587
 588         /*
 589          * HACK ALERT! This is early. We're enabling the console before
 590          * we've done PCI setups etc, and console_init() must be aware of
 591          * this. But we do want output early, in case something goes wrong.
 592          */
 593         console_init();
 594         if (panic_later)
 595                 panic(panic_later, panic_param);
 596
 597         lockdep_info();
 598
 599         /*
 600          * Need to run this when irqs are enabled, because it wants
 601          * to self-test [hard/soft]-irqs on/off lock inversion bugs
 602          * too:
 603          */
 604         locking_selftest();
 605
 606 #ifdef CONFIG_BLK_DEV_INITRD
 607         if (initrd_start && !initrd_below_start_ok &&
 608                         initrd_start < min_low_pfn << PAGE_SHIFT) {
 609                 printk(KERN_CRIT "initrd overwritten (0x%08lx < 0x%08lx) - "
 610                     "disabling it.\n",initrd_start,min_low_pfn << PAGE_SHIFT);
 611                 initrd_start = 0;
 612         }
 613 #endif
 614         vfs_caches_init_early();
 615         cpuset_init_early();
 616         mem_init();
 617         kmem_cache_init();
 618         setup_per_cpu_pageset();
 619         numa_policy_init();
 620         if (late_time_init)
 621                 late_time_init();
 622         calibrate_delay();
 623         pidmap_init();
 624         pgtable_cache_init();
 625         prio_tree_init();
 626         anon_vma_init();
 627 #ifdef CONFIG_X86
 628         if (efi_enabled)
 629                 efi_enter_virtual_mode();
 630 #endif
 631         fork_init(num_physpages);
 632         proc_caches_init();
 633         buffer_init();
 634         unnamed_dev_init();
 635         key_init();
 636         security_init();
 637         vfs_caches_init(num_physpages);
 638         radix_tree_init();
 639         signals_init();
 640         /* rootfs populating might need page-writeback */
 641         page_writeback_init();
 642 #ifdef CONFIG_PROC_FS
 643         proc_root_init();
 644 #endif
 645         cgroup_init();
 646         cpuset_init();
 647         taskstats_init_early();
 648         delayacct_init();
 649
 650         check_bugs();
 651
 652         acpi_early_init(); /* before LAPIC and SMP init */
 653
 654         /* Do the rest non-__init'ed, we're now alive */
 655         rest_init();
 656 }
 657
 658 static int __initdata initcall_debug;
 659
 660 static int __init initcall_debug_setup(char *str)
 661 {
 662         initcall_debug = 1;
 663         return 1;
 664 }
 665 __setup("initcall_debug", initcall_debug_setup);
 666
 667 extern initcall_t __initcall_start[], __initcall_end[];
 668
 669 static void __init do_initcalls(void)
 670 {
 671         initcall_t *call;
 672         int count = preempt_count();
 673
 674         for (call = __initcall_start; call < __initcall_end; call++) {
 675                 ktime_t t0, t1, delta;
 676                 char *msg = NULL;
 677                 char msgbuf[40];
 678                 int result;
 679
 680                 if (initcall_debug) {
 681                         printk("Calling initcall 0x%p", *call);
 682                         print_fn_descriptor_symbol(": %s()",
 683                                         (unsigned long) *call);
 684                         printk("\n");
 685                         t0 = ktime_get();
 686                 }
 687
 688                 result = (*call)();
 689
 690                 if (initcall_debug) {
 691                         t1 = ktime_get();
 692                         delta = ktime_sub(t1, t0);
 693
 694                         printk("initcall 0x%p", *call);
 695                         print_fn_descriptor_symbol(": %s()",
 696                                         (unsigned long) *call);
 697                         printk(" returned %d.\n", result);
 698
 699                         printk("initcall 0x%p ran for %Ld msecs: ",
 700                                 *call, (unsigned long long)delta.tv64 >> 20);
 701                         print_fn_descriptor_symbol("%s()\n",
 702                                 (unsigned long) *call);
 703                 }
 704
 705                 if (result && result != -ENODEV && initcall_debug) {
 706                         sprintf(msgbuf, "error code %d", result);
 707                         msg = msgbuf;
 708                 }
 709                 if (preempt_count() != count) {
 710                         msg = "preemption imbalance";
 711                         preempt_count() = count;
 712                 }
 713                 if (irqs_disabled()) {
 714                         msg = "disabled interrupts";
 715                         local_irq_enable();
 716                 }
 717                 if (msg) {
 718                         printk(KERN_WARNING "initcall at 0x%p", *call);
 719                         print_fn_descriptor_symbol(": %s()",
 720                                         (unsigned long) *call);
 721                         printk(": returned with %s\n", msg);
 722                 }
 723         }
 724
 725         /* Make sure there is no pending stuff from the initcall sequence */
 726         flush_scheduled_work();
 727 }
 728
 729 /*
 730  * Ok, the machine is now initialized. None of the devices
 731  * have been touched yet, but the CPU subsystem is up and
 732  * running, and memory and process management works.
 733  *
 734  * Now we can finally start doing some real work..
 735  */
 736 static void __init do_basic_setup(void)
 737 {
 738         /* drivers will send hotplug events */
 739         init_workqueues();
 740         usermodehelper_init();
 741         driver_init();
 742         init_irq_proc();
 743         do_initcalls();
 744 }
 745
 746 static int __initdata nosoftlockup;
 747
 748 static int __init nosoftlockup_setup(char *str)
 749 {
 750         nosoftlockup = 1;
 751         return 1;
 752 }
 753 __setup("nosoftlockup", nosoftlockup_setup);
 754
 755 static void __init do_pre_smp_initcalls(void)
 756 {
 757         extern int spawn_ksoftirqd(void);
 758 #ifdef CONFIG_SMP
 759         extern int migration_init(void);
 760
 761         migration_init();
 762 #endif
 763         spawn_ksoftirqd();
 764         if (!nosoftlockup)
 765                 spawn_softlockup_task();
 766 }
 767
 768 static void run_init_process(char *init_filename)
 769 {
 770         argv_init[0] = init_filename;
 771         kernel_execve(init_filename, argv_init, envp_init);
 772 }
 773
 774 /* This is a non __init function. Force it to be noinline otherwise gcc
 775  * makes it inline to init() and it becomes part of init.text section
 776  */
 777 static int noinline init_post(void)
 778 {
 779         free_initmem();
 780         unlock_kernel();
 781         mark_rodata_ro();
 782         system_state = SYSTEM_RUNNING;
 783         numa_default_policy();
 784
 785         if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0)
 786                 printk(KERN_WARNING "Warning: unable to open an initial console.\n");
 787
 788         (void) sys_dup(0);
 789         (void) sys_dup(0);
 790
 791         if (ramdisk_execute_command) {
 792                 run_init_process(ramdisk_execute_command);
 793                 printk(KERN_WARNING "Failed to execute %s\n",
 794                                 ramdisk_execute_command);
 795         }
 796
 797         /*
 798          * We try each of these until one succeeds.
 799          *
 800          * The Bourne shell can be used instead of init if we are
 801          * trying to recover a really broken machine.
 802          */
 803         if (execute_command) {
 804                 run_init_process(execute_command);
 805                 printk(KERN_WARNING "Failed to execute %s.  Attempting "
 806                                         "defaults...\n", execute_command);
 807         }
 808         run_init_process("/sbin/init");
 809         run_init_process("/etc/init");
 810         run_init_process("/bin/init");
 811         run_init_process("/bin/sh");
 812
 813         panic("No init found.  Try passing init= option to kernel.");
 814 }
 815
 816 static int __init kernel_init(void * unused)
 817 {
 818         lock_kernel();
 819         /*
 820          * init can run on any cpu.
 821          */
 822         set_cpus_allowed(current, CPU_MASK_ALL);
 823         /*
 824          * Tell the world that we're going to be the grim
 825          * reaper of innocent orphaned children.
 826          *
 827          * We don't want people to have to make incorrect
 828          * assumptions about where in the task array this
 829          * can be found.
 830          */
 831         init_pid_ns.child_reaper = current;
 832
 833         __set_special_pids(1, 1);
 834         cad_pid = task_pid(current);
 835
 836         smp_prepare_cpus(max_cpus);
 837
 838         do_pre_smp_initcalls();
 839
 840         smp_init();
 841         sched_init_smp();
 842
 843         cpuset_init_smp();
 844
 845         do_basic_setup();
 846
 847         /*
 848          * check if there is an early userspace init.  If yes, let it do all
 849          * the work
 850          */
 851
 852         if (!ramdisk_execute_command)
 853                 ramdisk_execute_command = "/init";
 854
 855         if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) {
 856                 ramdisk_execute_command = NULL;
 857                 prepare_namespace();
 858         }
 859
 860         /*
 861          * Ok, we have completed the initial bootup, and
 862          * we're essentially up and running. Get rid of the
 863          * initmem segments and start the user-mode stuff..
 864          */
 865         init_post();
 866         return 0;
 867 }