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/efi.h>
  43 #include <linux/tick.h>
  44 #include <linux/interrupt.h>
  45 #include <linux/taskstats_kern.h>
  46 #include <linux/delayacct.h>
  47 #include <linux/unistd.h>
  48 #include <linux/rmap.h>
  49 #include <linux/mempolicy.h>
  50 #include <linux/key.h>
  51 #include <linux/unwind.h>
  52 #include <linux/buffer_head.h>
  53 #include <linux/debug_locks.h>
  54 #include <linux/lockdep.h>
  55 #include <linux/pid_namespace.h>
  56 #include <linux/device.h>
  57
  58 #include <asm/io.h>
  59 #include <asm/bugs.h>
  60 #include <asm/setup.h>
  61 #include <asm/sections.h>
  62 #include <asm/cacheflush.h>
  63
  64 #ifdef CONFIG_X86_LOCAL_APIC
  65 #include <asm/smp.h>
  66 #endif
  67
  68 /*
  69  * This is one of the first .c files built. Error out early if we have compiler
  70  * trouble.
  71  *
  72  * Versions of gcc older than that listed below may actually compile and link
  73  * okay, but the end product can have subtle run time bugs.  To avoid associated
  74  * bogus bug reports, we flatly refuse to compile with a gcc that is known to be
  75  * too old from the very beginning.
  76  */
  77 #if (__GNUC__ < 3) || (__GNUC__ == 3 && __GNUC_MINOR__ < 2)
  78 #error Sorry, your GCC is too old. It builds incorrect kernels.
  79 #endif
  80
  81 #if __GNUC__ == 4 && __GNUC_MINOR__ == 1 && __GNUC_PATCHLEVEL__ == 0
  82 #warning gcc-4.1.0 is known to miscompile the kernel.  A different compiler version is recommended.
  83 #endif
  84
  85 static int init(void *);
  86
  87 extern void init_IRQ(void);
  88 extern void fork_init(unsigned long);
  89 extern void mca_init(void);
  90 extern void sbus_init(void);
  91 extern void signals_init(void);
  92 extern void pidhash_init(void);
  93 extern void pidmap_init(void);
  94 extern void prio_tree_init(void);
  95 extern void radix_tree_init(void);
  96 extern void free_initmem(void);
  97 extern void prepare_namespace(void);
  98 #ifdef  CONFIG_ACPI
  99 extern void acpi_early_init(void);
 100 #else
 101 static inline void acpi_early_init(void) { }
 102 #endif
 103 #ifndef CONFIG_DEBUG_RODATA
 104 static inline void mark_rodata_ro(void) { }
 105 #endif
 106
 107 #ifdef CONFIG_TC
 108 extern void tc_init(void);
 109 #endif
 110
 111 enum system_states system_state;
 112 EXPORT_SYMBOL(system_state);
 113
 114 /*
 115  * Boot command-line arguments
 116  */
 117 #define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT
 118 #define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT
 119
 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);
 124
 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;
 131
 132 static char *execute_command;
 133 static char *ramdisk_execute_command;
 134
 135 /* Setup configured maximum number of CPUs to activate */
 136 static unsigned int max_cpus = NR_CPUS;
 137
 138 /*
 139  * If set, this is an indication to the drivers that reset the underlying
 140  * device before going ahead with the initialization otherwise driver might
 141  * rely on the BIOS and skip the reset operation.
 142  *
 143  * This is useful if kernel is booting in an unreliable environment.
 144  * For ex. kdump situaiton where previous kernel has crashed, BIOS has been
 145  * skipped and devices will be in unknown state.
 146  */
 147 unsigned int reset_devices;
 148 EXPORT_SYMBOL(reset_devices);
 149
 150 /*
 151  * Setup routine for controlling SMP activation
 152  *
 153  * Command-line option of "nosmp" or "maxcpus=0" will disable SMP
 154  * activation entirely (the MPS table probe still happens, though).
 155  *
 156  * Command-line option of "maxcpus=<NUM>", where <NUM> is an integer
 157  * greater than 0, limits the maximum number of CPUs activated in
 158  * SMP mode to <NUM>.
 159  */
 160 static int __init nosmp(char *str)
 161 {
 162         max_cpus = 0;
 163         return 1;
 164 }
 165
 166 __setup("nosmp", nosmp);
 167
 168 static int __init maxcpus(char *str)
 169 {
 170         get_option(&str, &max_cpus);
 171         return 1;
 172 }
 173
 174 __setup("maxcpus=", maxcpus);
 175
 176 static int __init set_reset_devices(char *str)
 177 {
 178         reset_devices = 1;
 179         return 1;
 180 }
 181
 182 __setup("reset_devices", set_reset_devices);
 183
 184 static char * argv_init[MAX_INIT_ARGS+2] = { "init", NULL, };
 185 char * envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, };
 186 static const char *panic_later, *panic_param;
 187
 188 extern struct obs_kernel_param __setup_start[], __setup_end[];
 189
 190 static int __init obsolete_checksetup(char *line)
 191 {
 192         struct obs_kernel_param *p;
 193         int had_early_param = 0;
 194
 195         p = __setup_start;
 196         do {
 197                 int n = strlen(p->str);
 198                 if (!strncmp(line, p->str, n)) {
 199                         if (p->early) {
 200                                 /* Already done in parse_early_param?
 201                                  * (Needs exact match on param part).
 202                                  * Keep iterating, as we can have early
 203                                  * params and __setups of same names 8( */
 204                                 if (line[n] == '\0' || line[n] == '=')
 205                                         had_early_param = 1;
 206                         } else if (!p->setup_func) {
 207                                 printk(KERN_WARNING "Parameter %s is obsolete,"
 208                                        " ignored\n", p->str);
 209                                 return 1;
 210                         } else if (p->setup_func(line + n))
 211                                 return 1;
 212                 }
 213                 p++;
 214         } while (p < __setup_end);
 215
 216         return had_early_param;
 217 }
 218
 219 /*
 220  * This should be approx 2 Bo*oMips to start (note initial shift), and will
 221  * still work even if initially too large, it will just take slightly longer
 222  */
 223 unsigned long loops_per_jiffy = (1<<12);
 224
 225 EXPORT_SYMBOL(loops_per_jiffy);
 226
 227 static int __init debug_kernel(char *str)
 228 {
 229         if (*str)
 230                 return 0;
 231         console_loglevel = 10;
 232         return 1;
 233 }
 234
 235 static int __init quiet_kernel(char *str)
 236 {
 237         if (*str)
 238                 return 0;
 239         console_loglevel = 4;
 240         return 1;
 241 }
 242
 243 __setup("debug", debug_kernel);
 244 __setup("quiet", quiet_kernel);
 245
 246 static int __init loglevel(char *str)
 247 {
 248         get_option(&str, &console_loglevel);
 249         return 1;
 250 }
 251
 252 __setup("loglevel=", loglevel);
 253
 254 /*
 255  * Unknown boot options get handed to init, unless they look like
 256  * failed parameters
 257  */
 258 static int __init unknown_bootoption(char *param, char *val)
 259 {
 260         /* Change NUL term back to "=", to make "param" the whole string. */
 261         if (val) {
 262                 /* param=val or param="val"? */
 263                 if (val == param+strlen(param)+1)
 264                         val[-1] = '=';
 265                 else if (val == param+strlen(param)+2) {
 266                         val[-2] = '=';
 267                         memmove(val-1, val, strlen(val)+1);
 268                         val--;
 269                 } else
 270                         BUG();
 271         }
 272
 273         /* Handle obsolete-style parameters */
 274         if (obsolete_checksetup(param))
 275                 return 0;
 276
 277         /*
 278          * Preemptive maintenance for "why didn't my mispelled command
 279          * line work?"
 280          */
 281         if (strchr(param, '.') && (!val || strchr(param, '.') < val)) {
 282                 printk(KERN_ERR "Unknown boot option `%s': ignoring\n", param);
 283                 return 0;
 284         }
 285
 286         if (panic_later)
 287                 return 0;
 288
 289         if (val) {
 290                 /* Environment option */
 291                 unsigned int i;
 292                 for (i = 0; envp_init[i]; i++) {
 293                         if (i == MAX_INIT_ENVS) {
 294                                 panic_later = "Too many boot env vars at `%s'";
 295                                 panic_param = param;
 296                         }
 297                         if (!strncmp(param, envp_init[i], val - param))
 298                                 break;
 299                 }
 300                 envp_init[i] = param;
 301         } else {
 302                 /* Command line option */
 303                 unsigned int i;
 304                 for (i = 0; argv_init[i]; i++) {
 305                         if (i == MAX_INIT_ARGS) {
 306                                 panic_later = "Too many boot init vars at `%s'";
 307                                 panic_param = param;
 308                         }
 309                 }
 310                 argv_init[i] = param;
 311         }
 312         return 0;
 313 }
 314
 315 static int __init init_setup(char *str)
 316 {
 317         unsigned int i;
 318
 319         execute_command = str;
 320         /*
 321          * In case LILO is going to boot us with default command line,
 322          * it prepends "auto" before the whole cmdline which makes
 323          * the shell think it should execute a script with such name.
 324          * So we ignore all arguments entered _before_ init=... [MJ]
 325          */
 326         for (i = 1; i < MAX_INIT_ARGS; i++)
 327                 argv_init[i] = NULL;
 328         return 1;
 329 }
 330 __setup("init=", init_setup);
 331
 332 static int __init rdinit_setup(char *str)
 333 {
 334         unsigned int i;
 335
 336         ramdisk_execute_command = str;
 337         /* See "auto" comment in init_setup */
 338         for (i = 1; i < MAX_INIT_ARGS; i++)
 339                 argv_init[i] = NULL;
 340         return 1;
 341 }
 342 __setup("rdinit=", rdinit_setup);
 343
 344 #ifndef CONFIG_SMP
 345
 346 #ifdef CONFIG_X86_LOCAL_APIC
 347 static void __init smp_init(void)
 348 {
 349         APIC_init_uniprocessor();
 350 }
 351 #else
 352 #define smp_init()      do { } while (0)
 353 #endif
 354
 355 static inline void setup_per_cpu_areas(void) { }
 356 static inline void smp_prepare_cpus(unsigned int maxcpus) { }
 357
 358 #else
 359
 360 #ifdef __GENERIC_PER_CPU
 361 unsigned long __per_cpu_offset[NR_CPUS] __read_mostly;
 362
 363 EXPORT_SYMBOL(__per_cpu_offset);
 364
 365 static void __init setup_per_cpu_areas(void)
 366 {
 367         unsigned long size, i;
 368         char *ptr;
 369         unsigned long nr_possible_cpus = num_possible_cpus();
 370
 371         /* Copy section for each CPU (we discard the original) */
 372         size = ALIGN(__per_cpu_end - __per_cpu_start, SMP_CACHE_BYTES);
 373 #ifdef CONFIG_MODULES
 374         if (size < PERCPU_ENOUGH_ROOM)
 375                 size = PERCPU_ENOUGH_ROOM;
 376 #endif
 377         ptr = alloc_bootmem(size * nr_possible_cpus);
 378
 379         for_each_possible_cpu(i) {
 380                 __per_cpu_offset[i] = ptr - __per_cpu_start;
 381                 memcpy(ptr, __per_cpu_start, __per_cpu_end - __per_cpu_start);
 382                 ptr += size;
 383         }
 384 }
 385 #endif /* !__GENERIC_PER_CPU */
 386
 387 /* Called by boot processor to activate the rest. */
 388 static void __init smp_init(void)
 389 {
 390         unsigned int cpu;
 391         unsigned highest = 0;
 392
 393         for_each_cpu_mask(cpu, cpu_possible_map)
 394                 highest = cpu;
 395         nr_cpu_ids = highest + 1;
 396
 397         /* FIXME: This should be done in userspace --RR */
 398         for_each_present_cpu(cpu) {
 399                 if (num_online_cpus() >= max_cpus)
 400                         break;
 401                 if (!cpu_online(cpu))
 402                         cpu_up(cpu);
 403         }
 404
 405         /* Any cleanup work */
 406         printk(KERN_INFO "Brought up %ld CPUs\n", (long)num_online_cpus());
 407         smp_cpus_done(max_cpus);
 408 }
 409
 410 #endif
 411
 412 /*
 413  * We need to store the untouched command line for future reference.
 414  * We also need to store the touched command line since the parameter
 415  * parsing is performed in place, and we should allow a component to
 416  * store reference of name/value for future reference.
 417  */
 418 static void __init setup_command_line(char *command_line)
 419 {
 420         saved_command_line = alloc_bootmem(strlen (boot_command_line)+1);
 421         static_command_line = alloc_bootmem(strlen (command_line)+1);
 422         strcpy (saved_command_line, boot_command_line);
 423         strcpy (static_command_line, command_line);
 424 }
 425
 426 /*
 427  * We need to finalize in a non-__init function or else race conditions
 428  * between the root thread and the init thread may cause start_kernel to
 429  * be reaped by free_initmem before the root thread has proceeded to
 430  * cpu_idle.
 431  *
 432  * gcc-3.4 accidentally inlines this function, so use noinline.
 433  */
 434
 435 static void noinline rest_init(void)
 436         __releases(kernel_lock)
 437 {
 438         kernel_thread(init, NULL, CLONE_FS | CLONE_SIGHAND);
 439         numa_default_policy();
 440         unlock_kernel();
 441
 442         /*
 443          * The boot idle thread must execute schedule()
 444          * at least one to get things moving:
 445          */
 446         preempt_enable_no_resched();
 447         schedule();
 448         preempt_disable();
 449
 450         /* Call into cpu_idle with preempt disabled */
 451         cpu_idle();
 452 }
 453
 454 /* Check for early params. */
 455 static int __init do_early_param(char *param, char *val)
 456 {
 457         struct obs_kernel_param *p;
 458
 459         for (p = __setup_start; p < __setup_end; p++) {
 460                 if (p->early && strcmp(param, p->str) == 0) {
 461                         if (p->setup_func(val) != 0)
 462                                 printk(KERN_WARNING
 463                                        "Malformed early option '%s'\n", param);
 464                 }
 465         }
 466         /* We accept everything at this stage. */
 467         return 0;
 468 }
 469
 470 /* Arch code calls this early on, or if not, just before other parsing. */
 471 void __init parse_early_param(void)
 472 {
 473         static __initdata int done = 0;
 474         static __initdata char tmp_cmdline[COMMAND_LINE_SIZE];
 475
 476         if (done)
 477                 return;
 478
 479         /* All fall through to do_early_param. */
 480         strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE);
 481         parse_args("early options", tmp_cmdline, NULL, 0, do_early_param);
 482         done = 1;
 483 }
 484
 485 /*
 486  *      Activate the first processor.
 487  */
 488
 489 static void __init boot_cpu_init(void)
 490 {
 491         int cpu = smp_processor_id();
 492         /* Mark the boot cpu "present", "online" etc for SMP and UP case */
 493         cpu_set(cpu, cpu_online_map);
 494         cpu_set(cpu, cpu_present_map);
 495         cpu_set(cpu, cpu_possible_map);
 496 }
 497
 498 void __init __attribute__((weak)) smp_setup_processor_id(void)
 499 {
 500 }
 501
 502 asmlinkage void __init start_kernel(void)
 503 {
 504         char * command_line;
 505         extern struct kernel_param __start___param[], __stop___param[];
 506
 507         smp_setup_processor_id();
 508
 509         /*
 510          * Need to run as early as possible, to initialize the
 511          * lockdep hash:
 512          */
 513         unwind_init();
 514         lockdep_init();
 515
 516         local_irq_disable();
 517         early_boot_irqs_off();
 518         early_init_irq_lock_class();
 519
 520 /*
 521  * Interrupts are still disabled. Do necessary setups, then
 522  * enable them
 523  */
 524         lock_kernel();
 525         tick_init();
 526         boot_cpu_init();
 527         page_address_init();
 528         printk(KERN_NOTICE);
 529         printk(linux_banner);
 530         setup_arch(&command_line);
 531         setup_command_line(command_line);
 532         unwind_setup();
 533         setup_per_cpu_areas();
 534         smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */
 535
 536         /*
 537          * Set up the scheduler prior starting any interrupts (such as the
 538          * timer interrupt). Full topology setup happens at smp_init()
 539          * time - but meanwhile we still have a functioning scheduler.
 540          */
 541         sched_init();
 542         /*
 543          * Disable preemption - early bootup scheduling is extremely
 544          * fragile until we cpu_idle() for the first time.
 545          */
 546         preempt_disable();
 547         build_all_zonelists();
 548         page_alloc_init();
 549         printk(KERN_NOTICE "Kernel command line: %s\n", boot_command_line);
 550         parse_early_param();
 551         parse_args("Booting kernel", static_command_line, __start___param,
 552                    __stop___param - __start___param,
 553                    &unknown_bootoption);
 554         if (!irqs_disabled()) {
 555                 printk(KERN_WARNING "start_kernel(): bug: interrupts were "
 556                                 "enabled *very* early, fixing it\n");
 557                 local_irq_disable();
 558         }
 559         sort_main_extable();
 560         trap_init();
 561         rcu_init();
 562         init_IRQ();
 563         pidhash_init();
 564         init_timers();
 565         hrtimers_init();
 566         softirq_init();
 567         timekeeping_init();
 568         time_init();
 569         profile_init();
 570         if (!irqs_disabled())
 571                 printk("start_kernel(): bug: interrupts were enabled early\n");
 572         early_boot_irqs_on();
 573         local_irq_enable();
 574
 575         /*
 576          * HACK ALERT! This is early. We're enabling the console before
 577          * we've done PCI setups etc, and console_init() must be aware of
 578          * this. But we do want output early, in case something goes wrong.
 579          */
 580         console_init();
 581         if (panic_later)
 582                 panic(panic_later, panic_param);
 583
 584         lockdep_info();
 585
 586         /*
 587          * Need to run this when irqs are enabled, because it wants
 588          * to self-test [hard/soft]-irqs on/off lock inversion bugs
 589          * too:
 590          */
 591         locking_selftest();
 592
 593 #ifdef CONFIG_BLK_DEV_INITRD
 594         if (initrd_start && !initrd_below_start_ok &&
 595                         initrd_start < min_low_pfn << PAGE_SHIFT) {
 596                 printk(KERN_CRIT "initrd overwritten (0x%08lx < 0x%08lx) - "
 597                     "disabling it.\n",initrd_start,min_low_pfn << PAGE_SHIFT);
 598                 initrd_start = 0;
 599         }
 600 #endif
 601         vfs_caches_init_early();
 602         cpuset_init_early();
 603         mem_init();
 604         kmem_cache_init();
 605         setup_per_cpu_pageset();
 606         numa_policy_init();
 607         if (late_time_init)
 608                 late_time_init();
 609         calibrate_delay();
 610         pidmap_init();
 611         pgtable_cache_init();
 612         prio_tree_init();
 613         anon_vma_init();
 614 #ifdef CONFIG_X86
 615         if (efi_enabled)
 616                 efi_enter_virtual_mode();
 617 #endif
 618         fork_init(num_physpages);
 619         proc_caches_init();
 620         buffer_init();
 621         unnamed_dev_init();
 622         key_init();
 623         security_init();
 624         vfs_caches_init(num_physpages);
 625         radix_tree_init();
 626         signals_init();
 627         /* rootfs populating might need page-writeback */
 628         page_writeback_init();
 629 #ifdef CONFIG_PROC_FS
 630         proc_root_init();
 631 #endif
 632         cpuset_init();
 633         taskstats_init_early();
 634         delayacct_init();
 635
 636         check_bugs();
 637
 638         acpi_early_init(); /* before LAPIC and SMP init */
 639
 640         /* Do the rest non-__init'ed, we're now alive */
 641         rest_init();
 642 }
 643
 644 static int __initdata initcall_debug;
 645
 646 static int __init initcall_debug_setup(char *str)
 647 {
 648         initcall_debug = 1;
 649         return 1;
 650 }
 651 __setup("initcall_debug", initcall_debug_setup);
 652
 653 extern initcall_t __initcall_start[], __initcall_end[];
 654
 655 static void __init do_initcalls(void)
 656 {
 657         initcall_t *call;
 658         int count = preempt_count();
 659
 660         for (call = __initcall_start; call < __initcall_end; call++) {
 661                 char *msg = NULL;
 662                 char msgbuf[40];
 663                 int result;
 664
 665                 if (initcall_debug) {
 666                         printk("Calling initcall 0x%p", *call);
 667                         print_fn_descriptor_symbol(": %s()",
 668                                         (unsigned long) *call);
 669                         printk("\n");
 670                 }
 671
 672                 result = (*call)();
 673
 674                 if (result && result != -ENODEV && initcall_debug) {
 675                         sprintf(msgbuf, "error code %d", result);
 676                         msg = msgbuf;
 677                 }
 678                 if (preempt_count() != count) {
 679                         msg = "preemption imbalance";
 680                         preempt_count() = count;
 681                 }
 682                 if (irqs_disabled()) {
 683                         msg = "disabled interrupts";
 684                         local_irq_enable();
 685                 }
 686                 if (msg) {
 687                         printk(KERN_WARNING "initcall at 0x%p", *call);
 688                         print_fn_descriptor_symbol(": %s()",
 689                                         (unsigned long) *call);
 690                         printk(": returned with %s\n", msg);
 691                 }
 692         }
 693
 694         /* Make sure there is no pending stuff from the initcall sequence */
 695         flush_scheduled_work();
 696 }
 697
 698 /*
 699  * Ok, the machine is now initialized. None of the devices
 700  * have been touched yet, but the CPU subsystem is up and
 701  * running, and memory and process management works.
 702  *
 703  * Now we can finally start doing some real work..
 704  */
 705 static void __init do_basic_setup(void)
 706 {
 707         /* drivers will send hotplug events */
 708         init_workqueues();
 709         usermodehelper_init();
 710         driver_init();
 711         init_irq_proc();
 712         do_initcalls();
 713 }
 714
 715 static void __init do_pre_smp_initcalls(void)
 716 {
 717         extern int spawn_ksoftirqd(void);
 718 #ifdef CONFIG_SMP
 719         extern int migration_init(void);
 720
 721         migration_init();
 722 #endif
 723         spawn_ksoftirqd();
 724         spawn_softlockup_task();
 725 }
 726
 727 static void run_init_process(char *init_filename)
 728 {
 729         argv_init[0] = init_filename;
 730         kernel_execve(init_filename, argv_init, envp_init);
 731 }
 732
 733 /* This is a non __init function. Force it to be noinline otherwise gcc
 734  * makes it inline to init() and it becomes part of init.text section
 735  */
 736 static int noinline init_post(void)
 737 {
 738         free_initmem();
 739         unlock_kernel();
 740         mark_rodata_ro();
 741         system_state = SYSTEM_RUNNING;
 742         numa_default_policy();
 743
 744         if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0)
 745                 printk(KERN_WARNING "Warning: unable to open an initial console.\n");
 746
 747         (void) sys_dup(0);
 748         (void) sys_dup(0);
 749
 750         if (ramdisk_execute_command) {
 751                 run_init_process(ramdisk_execute_command);
 752                 printk(KERN_WARNING "Failed to execute %s\n",
 753                                 ramdisk_execute_command);
 754         }
 755
 756         /*
 757          * We try each of these until one succeeds.
 758          *
 759          * The Bourne shell can be used instead of init if we are
 760          * trying to recover a really broken machine.
 761          */
 762         if (execute_command) {
 763                 run_init_process(execute_command);
 764                 printk(KERN_WARNING "Failed to execute %s.  Attempting "
 765                                         "defaults...\n", execute_command);
 766         }
 767         run_init_process("/sbin/init");
 768         run_init_process("/etc/init");
 769         run_init_process("/bin/init");
 770         run_init_process("/bin/sh");
 771
 772         panic("No init found.  Try passing init= option to kernel.");
 773 }
 774
 775 static int __init init(void * unused)
 776 {
 777         lock_kernel();
 778         /*
 779          * init can run on any cpu.
 780          */
 781         set_cpus_allowed(current, CPU_MASK_ALL);
 782         /*
 783          * Tell the world that we're going to be the grim
 784          * reaper of innocent orphaned children.
 785          *
 786          * We don't want people to have to make incorrect
 787          * assumptions about where in the task array this
 788          * can be found.
 789          */
 790         init_pid_ns.child_reaper = current;
 791
 792         cad_pid = task_pid(current);
 793
 794         smp_prepare_cpus(max_cpus);
 795
 796         do_pre_smp_initcalls();
 797
 798         smp_init();
 799         sched_init_smp();
 800
 801         cpuset_init_smp();
 802
 803         do_basic_setup();
 804
 805         /*
 806          * check if there is an early userspace init.  If yes, let it do all
 807          * the work
 808          */
 809
 810         if (!ramdisk_execute_command)
 811                 ramdisk_execute_command = "/init";
 812
 813         if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) {
 814                 ramdisk_execute_command = NULL;
 815                 prepare_namespace();
 816         }
 817
 818         /*
 819          * Ok, we have completed the initial bootup, and
 820          * we're essentially up and running. Get rid of the
 821          * initmem segments and start the user-mode stuff..
 822          */
 823         init_post();
 824         return 0;
 825 }