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 #define DEBUG /* Enable initcall_debug */ 13 14 #include <linux/types.h> 15 #include <linux/module.h> 16 #include <linux/proc_fs.h> 17 #include <linux/kernel.h> 18 #include <linux/syscalls.h> 19 #include <linux/stackprotector.h> 20 #include <linux/string.h> 21 #include <linux/ctype.h> 22 #include <linux/delay.h> 23 #include <linux/ioport.h> 24 #include <linux/init.h> 25 #include <linux/initrd.h> 26 #include <linux/bootmem.h> 27 #include <linux/acpi.h> 28 #include <linux/tty.h> 29 #include <linux/percpu.h> 30 #include <linux/kmod.h> 31 #include <linux/vmalloc.h> 32 #include <linux/kernel_stat.h> 33 #include <linux/start_kernel.h> 34 #include <linux/security.h> 35 #include <linux/smp.h> 36 #include <linux/profile.h> 37 #include <linux/rcupdate.h> 38 #include <linux/moduleparam.h> 39 #include <linux/kallsyms.h> 40 #include <linux/writeback.h> 41 #include <linux/cpu.h> 42 #include <linux/cpuset.h> 43 #include <linux/cgroup.h> 44 #include <linux/efi.h> 45 #include <linux/tick.h> 46 #include <linux/interrupt.h> 47 #include <linux/taskstats_kern.h> 48 #include <linux/delayacct.h> 49 #include <linux/unistd.h> 50 #include <linux/rmap.h> 51 #include <linux/mempolicy.h> 52 #include <linux/key.h> 53 #include <linux/buffer_head.h> 54 #include <linux/page_cgroup.h> 55 #include <linux/debug_locks.h> 56 #include <linux/debugobjects.h> 57 #include <linux/lockdep.h> 58 #include <linux/kmemleak.h> 59 #include <linux/pid_namespace.h> 60 #include <linux/device.h> 61 #include <linux/kthread.h> 62 #include <linux/sched.h> 63 #include <linux/signal.h> 64 #include <linux/idr.h> 65 #include <linux/kgdb.h> 66 #include <linux/ftrace.h> 67 #include <linux/async.h> 68 #include <linux/kmemcheck.h> 69 #include <linux/sfi.h> 70 #include <linux/shmem_fs.h> 71 #include <linux/slab.h> 72 #include <linux/perf_event.h> 73 #include <linux/file.h> 74 #include <linux/ptrace.h> 75 #include <linux/blkdev.h> 76 #include <linux/elevator.h> 77 #include <linux/sched_clock.h> 78 #include <linux/context_tracking.h> 79 #include <linux/random.h> 80 #include <linux/list.h> 81 82 #include <asm/io.h> 83 #include <asm/bugs.h> 84 #include <asm/setup.h> 85 #include <asm/sections.h> 86 #include <asm/cacheflush.h> 87 88 #ifdef CONFIG_X86_LOCAL_APIC 89 #include <asm/smp.h> 90 #endif 91 92 static int kernel_init(void *); 93 94 extern void init_IRQ(void); 95 extern void fork_init(unsigned long); 96 extern void radix_tree_init(void); 97 #ifndef CONFIG_DEBUG_RODATA 98 static inline void mark_rodata_ro(void) { } 99 #endif 100 101 /* 102 * Debug helper: via this flag we know that we are in 'early bootup code' 103 * where only the boot processor is running with IRQ disabled. This means 104 * two things - IRQ must not be enabled before the flag is cleared and some 105 * operations which are not allowed with IRQ disabled are allowed while the 106 * flag is set. 107 */ 108 bool early_boot_irqs_disabled __read_mostly; 109 110 enum system_states system_state __read_mostly; 111 EXPORT_SYMBOL(system_state); 112 113 /* 114 * Boot command-line arguments 115 */ 116 #define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT 117 #define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT 118 119 extern void time_init(void); 120 /* Default late time init is NULL. archs can override this later. */ 121 void (*__initdata late_time_init)(void); 122 123 /* Untouched command line saved by arch-specific code. */ 124 char __initdata boot_command_line[COMMAND_LINE_SIZE]; 125 /* Untouched saved command line (eg. for /proc) */ 126 char *saved_command_line; 127 /* Command line for parameter parsing */ 128 static char *static_command_line; 129 /* Command line for per-initcall parameter parsing */ 130 static char *initcall_command_line; 131 132 static char *execute_command; 133 static char *ramdisk_execute_command; 134 135 /* 136 * Used to generate warnings if static_key manipulation functions are used 137 * before jump_label_init is called. 138 */ 139 bool static_key_initialized __read_mostly = false; 140 EXPORT_SYMBOL_GPL(static_key_initialized); 141 142 /* 143 * If set, this is an indication to the drivers that reset the underlying 144 * device before going ahead with the initialization otherwise driver might 145 * rely on the BIOS and skip the reset operation. 146 * 147 * This is useful if kernel is booting in an unreliable environment. 148 * For ex. kdump situaiton where previous kernel has crashed, BIOS has been 149 * skipped and devices will be in unknown state. 150 */ 151 unsigned int reset_devices; 152 EXPORT_SYMBOL(reset_devices); 153 154 static int __init set_reset_devices(char *str) 155 { 156 reset_devices = 1; 157 return 1; 158 } 159 160 __setup("reset_devices", set_reset_devices); 161 162 static const char * argv_init[MAX_INIT_ARGS+2] = { "init", NULL, }; 163 const char * envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, }; 164 static const char *panic_later, *panic_param; 165 166 extern const struct obs_kernel_param __setup_start[], __setup_end[]; 167 168 static int __init obsolete_checksetup(char *line) 169 { 170 const struct obs_kernel_param *p; 171 int had_early_param = 0; 172 173 p = __setup_start; 174 do { 175 int n = strlen(p->str); 176 if (parameqn(line, p->str, n)) { 177 if (p->early) { 178 /* Already done in parse_early_param? 179 * (Needs exact match on param part). 180 * Keep iterating, as we can have early 181 * params and __setups of same names 8( */ 182 if (line[n] == '\0' || line[n] == '=') 183 had_early_param = 1; 184 } else if (!p->setup_func) { 185 pr_warn("Parameter %s is obsolete, ignored\n", 186 p->str); 187 return 1; 188 } else if (p->setup_func(line + n)) 189 return 1; 190 } 191 p++; 192 } while (p < __setup_end); 193 194 return had_early_param; 195 } 196 197 /* 198 * This should be approx 2 Bo*oMips to start (note initial shift), and will 199 * still work even if initially too large, it will just take slightly longer 200 */ 201 unsigned long loops_per_jiffy = (1<<12); 202 203 EXPORT_SYMBOL(loops_per_jiffy); 204 205 static int __init debug_kernel(char *str) 206 { 207 console_loglevel = CONSOLE_LOGLEVEL_DEBUG; 208 return 0; 209 } 210 211 static int __init quiet_kernel(char *str) 212 { 213 console_loglevel = CONSOLE_LOGLEVEL_QUIET; 214 return 0; 215 } 216 217 early_param("debug", debug_kernel); 218 early_param("quiet", quiet_kernel); 219 220 static int __init loglevel(char *str) 221 { 222 int newlevel; 223 224 /* 225 * Only update loglevel value when a correct setting was passed, 226 * to prevent blind crashes (when loglevel being set to 0) that 227 * are quite hard to debug 228 */ 229 if (get_option(&str, &newlevel)) { 230 console_loglevel = newlevel; 231 return 0; 232 } 233 234 return -EINVAL; 235 } 236 237 early_param("loglevel", loglevel); 238 239 /* Change NUL term back to "=", to make "param" the whole string. */ 240 static int __init repair_env_string(char *param, char *val, const char *unused) 241 { 242 if (val) { 243 /* param=val or param="val"? */ 244 if (val == param+strlen(param)+1) 245 val[-1] = '='; 246 else if (val == param+strlen(param)+2) { 247 val[-2] = '='; 248 memmove(val-1, val, strlen(val)+1); 249 val--; 250 } else 251 BUG(); 252 } 253 return 0; 254 } 255 256 /* Anything after -- gets handed straight to init. */ 257 static int __init set_init_arg(char *param, char *val, const char *unused) 258 { 259 unsigned int i; 260 261 if (panic_later) 262 return 0; 263 264 repair_env_string(param, val, unused); 265 266 for (i = 0; argv_init[i]; i++) { 267 if (i == MAX_INIT_ARGS) { 268 panic_later = "init"; 269 panic_param = param; 270 return 0; 271 } 272 } 273 argv_init[i] = param; 274 return 0; 275 } 276 277 /* 278 * Unknown boot options get handed to init, unless they look like 279 * unused parameters (modprobe will find them in /proc/cmdline). 280 */ 281 static int __init unknown_bootoption(char *param, char *val, const char *unused) 282 { 283 repair_env_string(param, val, unused); 284 285 /* Handle obsolete-style parameters */ 286 if (obsolete_checksetup(param)) 287 return 0; 288 289 /* Unused module parameter. */ 290 if (strchr(param, '.') && (!val || strchr(param, '.') < val)) 291 return 0; 292 293 if (panic_later) 294 return 0; 295 296 if (val) { 297 /* Environment option */ 298 unsigned int i; 299 for (i = 0; envp_init[i]; i++) { 300 if (i == MAX_INIT_ENVS) { 301 panic_later = "env"; 302 panic_param = param; 303 } 304 if (!strncmp(param, envp_init[i], val - param)) 305 break; 306 } 307 envp_init[i] = param; 308 } else { 309 /* Command line option */ 310 unsigned int i; 311 for (i = 0; argv_init[i]; i++) { 312 if (i == MAX_INIT_ARGS) { 313 panic_later = "init"; 314 panic_param = param; 315 } 316 } 317 argv_init[i] = param; 318 } 319 return 0; 320 } 321 322 static int __init init_setup(char *str) 323 { 324 unsigned int i; 325 326 execute_command = str; 327 /* 328 * In case LILO is going to boot us with default command line, 329 * it prepends "auto" before the whole cmdline which makes 330 * the shell think it should execute a script with such name. 331 * So we ignore all arguments entered _before_ init=... [MJ] 332 */ 333 for (i = 1; i < MAX_INIT_ARGS; i++) 334 argv_init[i] = NULL; 335 return 1; 336 } 337 __setup("init=", init_setup); 338 339 static int __init rdinit_setup(char *str) 340 { 341 unsigned int i; 342 343 ramdisk_execute_command = str; 344 /* See "auto" comment in init_setup */ 345 for (i = 1; i < MAX_INIT_ARGS; i++) 346 argv_init[i] = NULL; 347 return 1; 348 } 349 __setup("rdinit=", rdinit_setup); 350 351 #ifndef CONFIG_SMP 352 static const unsigned int setup_max_cpus = NR_CPUS; 353 #ifdef CONFIG_X86_LOCAL_APIC 354 static void __init smp_init(void) 355 { 356 APIC_init_uniprocessor(); 357 } 358 #else 359 #define smp_init() do { } while (0) 360 #endif 361 362 static inline void setup_nr_cpu_ids(void) { } 363 static inline void smp_prepare_cpus(unsigned int maxcpus) { } 364 #endif 365 366 /* 367 * We need to store the untouched command line for future reference. 368 * We also need to store the touched command line since the parameter 369 * parsing is performed in place, and we should allow a component to 370 * store reference of name/value for future reference. 371 */ 372 static void __init setup_command_line(char *command_line) 373 { 374 saved_command_line = 375 memblock_virt_alloc(strlen(boot_command_line) + 1, 0); 376 initcall_command_line = 377 memblock_virt_alloc(strlen(boot_command_line) + 1, 0); 378 static_command_line = memblock_virt_alloc(strlen(command_line) + 1, 0); 379 strcpy (saved_command_line, boot_command_line); 380 strcpy (static_command_line, command_line); 381 } 382 383 /* 384 * We need to finalize in a non-__init function or else race conditions 385 * between the root thread and the init thread may cause start_kernel to 386 * be reaped by free_initmem before the root thread has proceeded to 387 * cpu_idle. 388 * 389 * gcc-3.4 accidentally inlines this function, so use noinline. 390 */ 391 392 static __initdata DECLARE_COMPLETION(kthreadd_done); 393 394 static noinline void __init_refok rest_init(void) 395 { 396 int pid; 397 398 rcu_scheduler_starting(); 399 /* 400 * We need to spawn init first so that it obtains pid 1, however 401 * the init task will end up wanting to create kthreads, which, if 402 * we schedule it before we create kthreadd, will OOPS. 403 */ 404 kernel_thread(kernel_init, NULL, CLONE_FS); 405 numa_default_policy(); 406 pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES); 407 rcu_read_lock(); 408 kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns); 409 rcu_read_unlock(); 410 complete(&kthreadd_done); 411 412 /* 413 * The boot idle thread must execute schedule() 414 * at least once to get things moving: 415 */ 416 init_idle_bootup_task(current); 417 schedule_preempt_disabled(); 418 /* Call into cpu_idle with preempt disabled */ 419 cpu_startup_entry(CPUHP_ONLINE); 420 } 421 422 /* Check for early params. */ 423 static int __init do_early_param(char *param, char *val, const char *unused) 424 { 425 const struct obs_kernel_param *p; 426 427 for (p = __setup_start; p < __setup_end; p++) { 428 if ((p->early && parameq(param, p->str)) || 429 (strcmp(param, "console") == 0 && 430 strcmp(p->str, "earlycon") == 0) 431 ) { 432 if (p->setup_func(val) != 0) 433 pr_warn("Malformed early option '%s'\n", param); 434 } 435 } 436 /* We accept everything at this stage. */ 437 return 0; 438 } 439 440 void __init parse_early_options(char *cmdline) 441 { 442 parse_args("early options", cmdline, NULL, 0, 0, 0, do_early_param); 443 } 444 445 /* Arch code calls this early on, or if not, just before other parsing. */ 446 void __init parse_early_param(void) 447 { 448 static __initdata int done = 0; 449 static __initdata char tmp_cmdline[COMMAND_LINE_SIZE]; 450 451 if (done) 452 return; 453 454 /* All fall through to do_early_param. */ 455 strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE); 456 parse_early_options(tmp_cmdline); 457 done = 1; 458 } 459 460 /* 461 * Activate the first processor. 462 */ 463 464 static void __init boot_cpu_init(void) 465 { 466 int cpu = smp_processor_id(); 467 /* Mark the boot cpu "present", "online" etc for SMP and UP case */ 468 set_cpu_online(cpu, true); 469 set_cpu_active(cpu, true); 470 set_cpu_present(cpu, true); 471 set_cpu_possible(cpu, true); 472 } 473 474 void __init __weak smp_setup_processor_id(void) 475 { 476 } 477 478 # if THREAD_SIZE >= PAGE_SIZE 479 void __init __weak thread_info_cache_init(void) 480 { 481 } 482 #endif 483 484 /* 485 * Set up kernel memory allocators 486 */ 487 static void __init mm_init(void) 488 { 489 /* 490 * page_cgroup requires contiguous pages, 491 * bigger than MAX_ORDER unless SPARSEMEM. 492 */ 493 page_cgroup_init_flatmem(); 494 mem_init(); 495 kmem_cache_init(); 496 percpu_init_late(); 497 pgtable_init(); 498 vmalloc_init(); 499 } 500 501 asmlinkage __visible void __init start_kernel(void) 502 { 503 char * command_line, *after_dashes; 504 extern const struct kernel_param __start___param[], __stop___param[]; 505 506 /* 507 * Need to run as early as possible, to initialize the 508 * lockdep hash: 509 */ 510 lockdep_init(); 511 smp_setup_processor_id(); 512 debug_objects_early_init(); 513 514 /* 515 * Set up the the initial canary ASAP: 516 */ 517 boot_init_stack_canary(); 518 519 cgroup_init_early(); 520 521 local_irq_disable(); 522 early_boot_irqs_disabled = true; 523 524 /* 525 * Interrupts are still disabled. Do necessary setups, then 526 * enable them 527 */ 528 boot_cpu_init(); 529 page_address_init(); 530 pr_notice("%s", linux_banner); 531 setup_arch(&command_line); 532 mm_init_cpumask(&init_mm); 533 setup_command_line(command_line); 534 setup_nr_cpu_ids(); 535 setup_per_cpu_areas(); 536 smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */ 537 538 build_all_zonelists(NULL, NULL); 539 page_alloc_init(); 540 541 pr_notice("Kernel command line: %s\n", boot_command_line); 542 parse_early_param(); 543 after_dashes = parse_args("Booting kernel", 544 static_command_line, __start___param, 545 __stop___param - __start___param, 546 -1, -1, &unknown_bootoption); 547 if (after_dashes) 548 parse_args("Setting init args", after_dashes, NULL, 0, -1, -1, 549 set_init_arg); 550 551 jump_label_init(); 552 553 /* 554 * These use large bootmem allocations and must precede 555 * kmem_cache_init() 556 */ 557 setup_log_buf(0); 558 pidhash_init(); 559 vfs_caches_init_early(); 560 sort_main_extable(); 561 trap_init(); 562 mm_init(); 563 564 /* 565 * Set up the scheduler prior starting any interrupts (such as the 566 * timer interrupt). Full topology setup happens at smp_init() 567 * time - but meanwhile we still have a functioning scheduler. 568 */ 569 sched_init(); 570 /* 571 * Disable preemption - early bootup scheduling is extremely 572 * fragile until we cpu_idle() for the first time. 573 */ 574 preempt_disable(); 575 if (WARN(!irqs_disabled(), "Interrupts were enabled *very* early, fixing it\n")) 576 local_irq_disable(); 577 idr_init_cache(); 578 rcu_init(); 579 tick_nohz_init(); 580 context_tracking_init(); 581 radix_tree_init(); 582 /* init some links before init_ISA_irqs() */ 583 early_irq_init(); 584 init_IRQ(); 585 tick_init(); 586 init_timers(); 587 hrtimers_init(); 588 softirq_init(); 589 timekeeping_init(); 590 time_init(); 591 sched_clock_postinit(); 592 perf_event_init(); 593 profile_init(); 594 call_function_init(); 595 WARN(!irqs_disabled(), "Interrupts were enabled early\n"); 596 early_boot_irqs_disabled = false; 597 local_irq_enable(); 598 599 kmem_cache_init_late(); 600 601 /* 602 * HACK ALERT! This is early. We're enabling the console before 603 * we've done PCI setups etc, and console_init() must be aware of 604 * this. But we do want output early, in case something goes wrong. 605 */ 606 console_init(); 607 if (panic_later) 608 panic("Too many boot %s vars at `%s'", panic_later, 609 panic_param); 610 611 lockdep_info(); 612 613 /* 614 * Need to run this when irqs are enabled, because it wants 615 * to self-test [hard/soft]-irqs on/off lock inversion bugs 616 * too: 617 */ 618 locking_selftest(); 619 620 #ifdef CONFIG_BLK_DEV_INITRD 621 if (initrd_start && !initrd_below_start_ok && 622 page_to_pfn(virt_to_page((void *)initrd_start)) < min_low_pfn) { 623 pr_crit("initrd overwritten (0x%08lx < 0x%08lx) - disabling it.\n", 624 page_to_pfn(virt_to_page((void *)initrd_start)), 625 min_low_pfn); 626 initrd_start = 0; 627 } 628 #endif 629 page_cgroup_init(); 630 debug_objects_mem_init(); 631 kmemleak_init(); 632 setup_per_cpu_pageset(); 633 numa_policy_init(); 634 if (late_time_init) 635 late_time_init(); 636 sched_clock_init(); 637 calibrate_delay(); 638 pidmap_init(); 639 anon_vma_init(); 640 acpi_early_init(); 641 #ifdef CONFIG_X86 642 if (efi_enabled(EFI_RUNTIME_SERVICES)) 643 efi_enter_virtual_mode(); 644 #endif 645 #ifdef CONFIG_X86_ESPFIX64 646 /* Should be run before the first non-init thread is created */ 647 init_espfix_bsp(); 648 #endif 649 thread_info_cache_init(); 650 cred_init(); 651 fork_init(totalram_pages); 652 proc_caches_init(); 653 buffer_init(); 654 key_init(); 655 security_init(); 656 dbg_late_init(); 657 vfs_caches_init(totalram_pages); 658 signals_init(); 659 /* rootfs populating might need page-writeback */ 660 page_writeback_init(); 661 proc_root_init(); 662 cgroup_init(); 663 cpuset_init(); 664 taskstats_init_early(); 665 delayacct_init(); 666 667 check_bugs(); 668 669 sfi_init_late(); 670 671 if (efi_enabled(EFI_RUNTIME_SERVICES)) { 672 efi_late_init(); 673 efi_free_boot_services(); 674 } 675 676 ftrace_init(); 677 678 /* Do the rest non-__init'ed, we're now alive */ 679 rest_init(); 680 } 681 682 /* Call all constructor functions linked into the kernel. */ 683 static void __init do_ctors(void) 684 { 685 #ifdef CONFIG_CONSTRUCTORS 686 ctor_fn_t *fn = (ctor_fn_t *) __ctors_start; 687 688 for (; fn < (ctor_fn_t *) __ctors_end; fn++) 689 (*fn)(); 690 #endif 691 } 692 693 bool initcall_debug; 694 core_param(initcall_debug, initcall_debug, bool, 0644); 695 696 #ifdef CONFIG_KALLSYMS 697 struct blacklist_entry { 698 struct list_head next; 699 char *buf; 700 }; 701 702 static __initdata_or_module LIST_HEAD(blacklisted_initcalls); 703 704 static int __init initcall_blacklist(char *str) 705 { 706 char *str_entry; 707 struct blacklist_entry *entry; 708 709 /* str argument is a comma-separated list of functions */ 710 do { 711 str_entry = strsep(&str, ","); 712 if (str_entry) { 713 pr_debug("blacklisting initcall %s\n", str_entry); 714 entry = alloc_bootmem(sizeof(*entry)); 715 entry->buf = alloc_bootmem(strlen(str_entry) + 1); 716 strcpy(entry->buf, str_entry); 717 list_add(&entry->next, &blacklisted_initcalls); 718 } 719 } while (str_entry); 720 721 return 0; 722 } 723 724 static bool __init_or_module initcall_blacklisted(initcall_t fn) 725 { 726 struct list_head *tmp; 727 struct blacklist_entry *entry; 728 char *fn_name; 729 730 fn_name = kasprintf(GFP_KERNEL, "%pf", fn); 731 if (!fn_name) 732 return false; 733 734 list_for_each(tmp, &blacklisted_initcalls) { 735 entry = list_entry(tmp, struct blacklist_entry, next); 736 if (!strcmp(fn_name, entry->buf)) { 737 pr_debug("initcall %s blacklisted\n", fn_name); 738 kfree(fn_name); 739 return true; 740 } 741 } 742 743 kfree(fn_name); 744 return false; 745 } 746 #else 747 static int __init initcall_blacklist(char *str) 748 { 749 pr_warn("initcall_blacklist requires CONFIG_KALLSYMS\n"); 750 return 0; 751 } 752 753 static bool __init_or_module initcall_blacklisted(initcall_t fn) 754 { 755 return false; 756 } 757 #endif 758 __setup("initcall_blacklist=", initcall_blacklist); 759 760 static int __init_or_module do_one_initcall_debug(initcall_t fn) 761 { 762 ktime_t calltime, delta, rettime; 763 unsigned long long duration; 764 int ret; 765 766 printk(KERN_DEBUG "calling %pF @ %i\n", fn, task_pid_nr(current)); 767 calltime = ktime_get(); 768 ret = fn(); 769 rettime = ktime_get(); 770 delta = ktime_sub(rettime, calltime); 771 duration = (unsigned long long) ktime_to_ns(delta) >> 10; 772 printk(KERN_DEBUG "initcall %pF returned %d after %lld usecs\n", 773 fn, ret, duration); 774 775 return ret; 776 } 777 778 int __init_or_module do_one_initcall(initcall_t fn) 779 { 780 int count = preempt_count(); 781 int ret; 782 char msgbuf[64]; 783 784 if (initcall_blacklisted(fn)) 785 return -EPERM; 786 787 if (initcall_debug) 788 ret = do_one_initcall_debug(fn); 789 else 790 ret = fn(); 791 792 msgbuf[0] = 0; 793 794 if (preempt_count() != count) { 795 sprintf(msgbuf, "preemption imbalance "); 796 preempt_count_set(count); 797 } 798 if (irqs_disabled()) { 799 strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf)); 800 local_irq_enable(); 801 } 802 WARN(msgbuf[0], "initcall %pF returned with %s\n", fn, msgbuf); 803 804 return ret; 805 } 806 807 808 extern initcall_t __initcall_start[]; 809 extern initcall_t __initcall0_start[]; 810 extern initcall_t __initcall1_start[]; 811 extern initcall_t __initcall2_start[]; 812 extern initcall_t __initcall3_start[]; 813 extern initcall_t __initcall4_start[]; 814 extern initcall_t __initcall5_start[]; 815 extern initcall_t __initcall6_start[]; 816 extern initcall_t __initcall7_start[]; 817 extern initcall_t __initcall_end[]; 818 819 static initcall_t *initcall_levels[] __initdata = { 820 __initcall0_start, 821 __initcall1_start, 822 __initcall2_start, 823 __initcall3_start, 824 __initcall4_start, 825 __initcall5_start, 826 __initcall6_start, 827 __initcall7_start, 828 __initcall_end, 829 }; 830 831 /* Keep these in sync with initcalls in include/linux/init.h */ 832 static char *initcall_level_names[] __initdata = { 833 "early", 834 "core", 835 "postcore", 836 "arch", 837 "subsys", 838 "fs", 839 "device", 840 "late", 841 }; 842 843 static void __init do_initcall_level(int level) 844 { 845 extern const struct kernel_param __start___param[], __stop___param[]; 846 initcall_t *fn; 847 848 strcpy(initcall_command_line, saved_command_line); 849 parse_args(initcall_level_names[level], 850 initcall_command_line, __start___param, 851 __stop___param - __start___param, 852 level, level, 853 &repair_env_string); 854 855 for (fn = initcall_levels[level]; fn < initcall_levels[level+1]; fn++) 856 do_one_initcall(*fn); 857 } 858 859 static void __init do_initcalls(void) 860 { 861 int level; 862 863 for (level = 0; level < ARRAY_SIZE(initcall_levels) - 1; level++) 864 do_initcall_level(level); 865 } 866 867 /* 868 * Ok, the machine is now initialized. None of the devices 869 * have been touched yet, but the CPU subsystem is up and 870 * running, and memory and process management works. 871 * 872 * Now we can finally start doing some real work.. 873 */ 874 static void __init do_basic_setup(void) 875 { 876 cpuset_init_smp(); 877 usermodehelper_init(); 878 shmem_init(); 879 driver_init(); 880 init_irq_proc(); 881 do_ctors(); 882 usermodehelper_enable(); 883 do_initcalls(); 884 random_int_secret_init(); 885 } 886 887 static void __init do_pre_smp_initcalls(void) 888 { 889 initcall_t *fn; 890 891 for (fn = __initcall_start; fn < __initcall0_start; fn++) 892 do_one_initcall(*fn); 893 } 894 895 /* 896 * This function requests modules which should be loaded by default and is 897 * called twice right after initrd is mounted and right before init is 898 * exec'd. If such modules are on either initrd or rootfs, they will be 899 * loaded before control is passed to userland. 900 */ 901 void __init load_default_modules(void) 902 { 903 load_default_elevator_module(); 904 } 905 906 static int run_init_process(const char *init_filename) 907 { 908 argv_init[0] = init_filename; 909 return do_execve(getname_kernel(init_filename), 910 (const char __user *const __user *)argv_init, 911 (const char __user *const __user *)envp_init); 912 } 913 914 static int try_to_run_init_process(const char *init_filename) 915 { 916 int ret; 917 918 ret = run_init_process(init_filename); 919 920 if (ret && ret != -ENOENT) { 921 pr_err("Starting init: %s exists but couldn't execute it (error %d)\n", 922 init_filename, ret); 923 } 924 925 return ret; 926 } 927 928 static noinline void __init kernel_init_freeable(void); 929 930 static int __ref kernel_init(void *unused) 931 { 932 int ret; 933 934 kernel_init_freeable(); 935 /* need to finish all async __init code before freeing the memory */ 936 async_synchronize_full(); 937 free_initmem(); 938 mark_rodata_ro(); 939 system_state = SYSTEM_RUNNING; 940 numa_default_policy(); 941 942 flush_delayed_fput(); 943 944 if (ramdisk_execute_command) { 945 ret = run_init_process(ramdisk_execute_command); 946 if (!ret) 947 return 0; 948 pr_err("Failed to execute %s (error %d)\n", 949 ramdisk_execute_command, ret); 950 } 951 952 /* 953 * We try each of these until one succeeds. 954 * 955 * The Bourne shell can be used instead of init if we are 956 * trying to recover a really broken machine. 957 */ 958 if (execute_command) { 959 ret = run_init_process(execute_command); 960 if (!ret) 961 return 0; 962 pr_err("Failed to execute %s (error %d). Attempting defaults...\n", 963 execute_command, ret); 964 } 965 if (!try_to_run_init_process("/sbin/init") || 966 !try_to_run_init_process("/etc/init") || 967 !try_to_run_init_process("/bin/init") || 968 !try_to_run_init_process("/bin/sh")) 969 return 0; 970 971 panic("No working init found. Try passing init= option to kernel. " 972 "See Linux Documentation/init.txt for guidance."); 973 } 974 975 static noinline void __init kernel_init_freeable(void) 976 { 977 /* 978 * Wait until kthreadd is all set-up. 979 */ 980 wait_for_completion(&kthreadd_done); 981 982 /* Now the scheduler is fully set up and can do blocking allocations */ 983 gfp_allowed_mask = __GFP_BITS_MASK; 984 985 /* 986 * init can allocate pages on any node 987 */ 988 set_mems_allowed(node_states[N_MEMORY]); 989 /* 990 * init can run on any cpu. 991 */ 992 set_cpus_allowed_ptr(current, cpu_all_mask); 993 994 cad_pid = task_pid(current); 995 996 smp_prepare_cpus(setup_max_cpus); 997 998 do_pre_smp_initcalls(); 999 lockup_detector_init(); 1000 1001 smp_init(); 1002 sched_init_smp(); 1003 1004 do_basic_setup(); 1005 1006 /* Open the /dev/console on the rootfs, this should never fail */ 1007 if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0) 1008 pr_err("Warning: unable to open an initial console.\n"); 1009 1010 (void) sys_dup(0); 1011 (void) sys_dup(0); 1012 /* 1013 * check if there is an early userspace init. If yes, let it do all 1014 * the work 1015 */ 1016 1017 if (!ramdisk_execute_command) 1018 ramdisk_execute_command = "/init"; 1019 1020 if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) { 1021 ramdisk_execute_command = NULL; 1022 prepare_namespace(); 1023 } 1024 1025 /* 1026 * Ok, we have completed the initial bootup, and 1027 * we're essentially up and running. Get rid of the 1028 * initmem segments and start the user-mode stuff.. 1029 */ 1030 1031 /* rootfs is available now, try loading default modules */ 1032 load_default_modules(); 1033 } 1034