main.c 22 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918
  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. #include <linux/types.h>
  12. #include <linux/module.h>
  13. #include <linux/proc_fs.h>
  14. #include <linux/kernel.h>
  15. #include <linux/syscalls.h>
  16. #include <linux/stackprotector.h>
  17. #include <linux/string.h>
  18. #include <linux/ctype.h>
  19. #include <linux/delay.h>
  20. #include <linux/utsname.h>
  21. #include <linux/ioport.h>
  22. #include <linux/init.h>
  23. #include <linux/smp_lock.h>
  24. #include <linux/initrd.h>
  25. #include <linux/bootmem.h>
  26. #include <linux/tty.h>
  27. #include <linux/gfp.h>
  28. #include <linux/percpu.h>
  29. #include <linux/kmod.h>
  30. #include <linux/vmalloc.h>
  31. #include <linux/kernel_stat.h>
  32. #include <linux/start_kernel.h>
  33. #include <linux/security.h>
  34. #include <linux/smp.h>
  35. #include <linux/workqueue.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/ftrace.h>
  66. #include <linux/async.h>
  67. #include <linux/kmemcheck.h>
  68. #include <linux/kmemtrace.h>
  69. #include <trace/boot.h>
  70. #include <asm/io.h>
  71. #include <asm/bugs.h>
  72. #include <asm/setup.h>
  73. #include <asm/sections.h>
  74. #include <asm/cacheflush.h>
  75. #ifdef CONFIG_X86_LOCAL_APIC
  76. #include <asm/smp.h>
  77. #endif
  78. static int kernel_init(void *);
  79. extern void init_IRQ(void);
  80. extern void fork_init(unsigned long);
  81. extern void mca_init(void);
  82. extern void sbus_init(void);
  83. extern void prio_tree_init(void);
  84. extern void radix_tree_init(void);
  85. extern void free_initmem(void);
  86. #ifdef CONFIG_ACPI
  87. extern void acpi_early_init(void);
  88. #else
  89. static inline void acpi_early_init(void) { }
  90. #endif
  91. #ifndef CONFIG_DEBUG_RODATA
  92. static inline void mark_rodata_ro(void) { }
  93. #endif
  94. #ifdef CONFIG_TC
  95. extern void tc_init(void);
  96. #endif
  97. enum system_states system_state __read_mostly;
  98. EXPORT_SYMBOL(system_state);
  99. /*
  100. * Boot command-line arguments
  101. */
  102. #define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT
  103. #define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT
  104. extern void time_init(void);
  105. /* Default late time init is NULL. archs can override this later. */
  106. void (*__initdata late_time_init)(void);
  107. extern void softirq_init(void);
  108. /* Untouched command line saved by arch-specific code. */
  109. char __initdata boot_command_line[COMMAND_LINE_SIZE];
  110. /* Untouched saved command line (eg. for /proc) */
  111. char *saved_command_line;
  112. /* Command line for parameter parsing */
  113. static char *static_command_line;
  114. static char *execute_command;
  115. static char *ramdisk_execute_command;
  116. #ifdef CONFIG_SMP
  117. /* Setup configured maximum number of CPUs to activate */
  118. unsigned int __initdata setup_max_cpus = NR_CPUS;
  119. /*
  120. * Setup routine for controlling SMP activation
  121. *
  122. * Command-line option of "nosmp" or "maxcpus=0" will disable SMP
  123. * activation entirely (the MPS table probe still happens, though).
  124. *
  125. * Command-line option of "maxcpus=<NUM>", where <NUM> is an integer
  126. * greater than 0, limits the maximum number of CPUs activated in
  127. * SMP mode to <NUM>.
  128. */
  129. void __weak arch_disable_smp_support(void) { }
  130. static int __init nosmp(char *str)
  131. {
  132. setup_max_cpus = 0;
  133. arch_disable_smp_support();
  134. return 0;
  135. }
  136. early_param("nosmp", nosmp);
  137. static int __init maxcpus(char *str)
  138. {
  139. get_option(&str, &setup_max_cpus);
  140. if (setup_max_cpus == 0)
  141. arch_disable_smp_support();
  142. return 0;
  143. }
  144. early_param("maxcpus", maxcpus);
  145. #else
  146. const unsigned int setup_max_cpus = NR_CPUS;
  147. #endif
  148. /*
  149. * If set, this is an indication to the drivers that reset the underlying
  150. * device before going ahead with the initialization otherwise driver might
  151. * rely on the BIOS and skip the reset operation.
  152. *
  153. * This is useful if kernel is booting in an unreliable environment.
  154. * For ex. kdump situaiton where previous kernel has crashed, BIOS has been
  155. * skipped and devices will be in unknown state.
  156. */
  157. unsigned int reset_devices;
  158. EXPORT_SYMBOL(reset_devices);
  159. static int __init set_reset_devices(char *str)
  160. {
  161. reset_devices = 1;
  162. return 1;
  163. }
  164. __setup("reset_devices", set_reset_devices);
  165. static char * argv_init[MAX_INIT_ARGS+2] = { "init", NULL, };
  166. char * envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, };
  167. static const char *panic_later, *panic_param;
  168. extern struct obs_kernel_param __setup_start[], __setup_end[];
  169. static int __init obsolete_checksetup(char *line)
  170. {
  171. struct obs_kernel_param *p;
  172. int had_early_param = 0;
  173. p = __setup_start;
  174. do {
  175. int n = strlen(p->str);
  176. if (!strncmp(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. printk(KERN_WARNING "Parameter %s is obsolete,"
  186. " ignored\n", 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. return had_early_param;
  194. }
  195. /*
  196. * This should be approx 2 Bo*oMips to start (note initial shift), and will
  197. * still work even if initially too large, it will just take slightly longer
  198. */
  199. unsigned long loops_per_jiffy = (1<<12);
  200. EXPORT_SYMBOL(loops_per_jiffy);
  201. static int __init debug_kernel(char *str)
  202. {
  203. console_loglevel = 10;
  204. return 0;
  205. }
  206. static int __init quiet_kernel(char *str)
  207. {
  208. console_loglevel = 4;
  209. return 0;
  210. }
  211. early_param("debug", debug_kernel);
  212. early_param("quiet", quiet_kernel);
  213. static int __init loglevel(char *str)
  214. {
  215. get_option(&str, &console_loglevel);
  216. return 0;
  217. }
  218. early_param("loglevel", loglevel);
  219. /*
  220. * Unknown boot options get handed to init, unless they look like
  221. * failed parameters
  222. */
  223. static int __init unknown_bootoption(char *param, char *val)
  224. {
  225. /* Change NUL term back to "=", to make "param" the whole string. */
  226. if (val) {
  227. /* param=val or param="val"? */
  228. if (val == param+strlen(param)+1)
  229. val[-1] = '=';
  230. else if (val == param+strlen(param)+2) {
  231. val[-2] = '=';
  232. memmove(val-1, val, strlen(val)+1);
  233. val--;
  234. } else
  235. BUG();
  236. }
  237. /* Handle obsolete-style parameters */
  238. if (obsolete_checksetup(param))
  239. return 0;
  240. /*
  241. * Preemptive maintenance for "why didn't my misspelled command
  242. * line work?"
  243. */
  244. if (strchr(param, '.') && (!val || strchr(param, '.') < val)) {
  245. printk(KERN_ERR "Unknown boot option `%s': ignoring\n", param);
  246. return 0;
  247. }
  248. if (panic_later)
  249. return 0;
  250. if (val) {
  251. /* Environment option */
  252. unsigned int i;
  253. for (i = 0; envp_init[i]; i++) {
  254. if (i == MAX_INIT_ENVS) {
  255. panic_later = "Too many boot env vars at `%s'";
  256. panic_param = param;
  257. }
  258. if (!strncmp(param, envp_init[i], val - param))
  259. break;
  260. }
  261. envp_init[i] = param;
  262. } else {
  263. /* Command line option */
  264. unsigned int i;
  265. for (i = 0; argv_init[i]; i++) {
  266. if (i == MAX_INIT_ARGS) {
  267. panic_later = "Too many boot init vars at `%s'";
  268. panic_param = param;
  269. }
  270. }
  271. argv_init[i] = param;
  272. }
  273. return 0;
  274. }
  275. #ifdef CONFIG_DEBUG_PAGEALLOC
  276. int __read_mostly debug_pagealloc_enabled = 0;
  277. #endif
  278. static int __init init_setup(char *str)
  279. {
  280. unsigned int i;
  281. execute_command = str;
  282. /*
  283. * In case LILO is going to boot us with default command line,
  284. * it prepends "auto" before the whole cmdline which makes
  285. * the shell think it should execute a script with such name.
  286. * So we ignore all arguments entered _before_ init=... [MJ]
  287. */
  288. for (i = 1; i < MAX_INIT_ARGS; i++)
  289. argv_init[i] = NULL;
  290. return 1;
  291. }
  292. __setup("init=", init_setup);
  293. static int __init rdinit_setup(char *str)
  294. {
  295. unsigned int i;
  296. ramdisk_execute_command = str;
  297. /* See "auto" comment in init_setup */
  298. for (i = 1; i < MAX_INIT_ARGS; i++)
  299. argv_init[i] = NULL;
  300. return 1;
  301. }
  302. __setup("rdinit=", rdinit_setup);
  303. #ifndef CONFIG_SMP
  304. #ifdef CONFIG_X86_LOCAL_APIC
  305. static void __init smp_init(void)
  306. {
  307. APIC_init_uniprocessor();
  308. }
  309. #else
  310. #define smp_init() do { } while (0)
  311. #endif
  312. static inline void setup_per_cpu_areas(void) { }
  313. static inline void setup_nr_cpu_ids(void) { }
  314. static inline void smp_prepare_cpus(unsigned int maxcpus) { }
  315. #else
  316. #if NR_CPUS > BITS_PER_LONG
  317. cpumask_t cpu_mask_all __read_mostly = CPU_MASK_ALL;
  318. EXPORT_SYMBOL(cpu_mask_all);
  319. #endif
  320. /* Setup number of possible processor ids */
  321. int nr_cpu_ids __read_mostly = NR_CPUS;
  322. EXPORT_SYMBOL(nr_cpu_ids);
  323. /* An arch may set nr_cpu_ids earlier if needed, so this would be redundant */
  324. static void __init setup_nr_cpu_ids(void)
  325. {
  326. nr_cpu_ids = find_last_bit(cpumask_bits(cpu_possible_mask),NR_CPUS) + 1;
  327. }
  328. #ifndef CONFIG_HAVE_SETUP_PER_CPU_AREA
  329. unsigned long __per_cpu_offset[NR_CPUS] __read_mostly;
  330. EXPORT_SYMBOL(__per_cpu_offset);
  331. static void __init setup_per_cpu_areas(void)
  332. {
  333. unsigned long size, i;
  334. char *ptr;
  335. unsigned long nr_possible_cpus = num_possible_cpus();
  336. /* Copy section for each CPU (we discard the original) */
  337. size = ALIGN(PERCPU_ENOUGH_ROOM, PAGE_SIZE);
  338. ptr = alloc_bootmem_pages(size * nr_possible_cpus);
  339. for_each_possible_cpu(i) {
  340. __per_cpu_offset[i] = ptr - __per_cpu_start;
  341. memcpy(ptr, __per_cpu_start, __per_cpu_end - __per_cpu_start);
  342. ptr += size;
  343. }
  344. }
  345. #endif /* CONFIG_HAVE_SETUP_PER_CPU_AREA */
  346. /* Called by boot processor to activate the rest. */
  347. static void __init smp_init(void)
  348. {
  349. unsigned int cpu;
  350. /*
  351. * Set up the current CPU as possible to migrate to.
  352. * The other ones will be done by cpu_up/cpu_down()
  353. */
  354. set_cpu_active(smp_processor_id(), true);
  355. /* FIXME: This should be done in userspace --RR */
  356. for_each_present_cpu(cpu) {
  357. if (num_online_cpus() >= setup_max_cpus)
  358. break;
  359. if (!cpu_online(cpu))
  360. cpu_up(cpu);
  361. }
  362. /* Any cleanup work */
  363. printk(KERN_INFO "Brought up %ld CPUs\n", (long)num_online_cpus());
  364. smp_cpus_done(setup_max_cpus);
  365. }
  366. #endif
  367. /*
  368. * We need to store the untouched command line for future reference.
  369. * We also need to store the touched command line since the parameter
  370. * parsing is performed in place, and we should allow a component to
  371. * store reference of name/value for future reference.
  372. */
  373. static void __init setup_command_line(char *command_line)
  374. {
  375. saved_command_line = alloc_bootmem(strlen (boot_command_line)+1);
  376. static_command_line = alloc_bootmem(strlen (command_line)+1);
  377. strcpy (saved_command_line, boot_command_line);
  378. strcpy (static_command_line, command_line);
  379. }
  380. /*
  381. * We need to finalize in a non-__init function or else race conditions
  382. * between the root thread and the init thread may cause start_kernel to
  383. * be reaped by free_initmem before the root thread has proceeded to
  384. * cpu_idle.
  385. *
  386. * gcc-3.4 accidentally inlines this function, so use noinline.
  387. */
  388. static noinline void __init_refok rest_init(void)
  389. __releases(kernel_lock)
  390. {
  391. int pid;
  392. kernel_thread(kernel_init, NULL, CLONE_FS | CLONE_SIGHAND);
  393. numa_default_policy();
  394. pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES);
  395. kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns);
  396. unlock_kernel();
  397. /*
  398. * The boot idle thread must execute schedule()
  399. * at least once to get things moving:
  400. */
  401. init_idle_bootup_task(current);
  402. rcu_scheduler_starting();
  403. preempt_enable_no_resched();
  404. schedule();
  405. preempt_disable();
  406. /* Call into cpu_idle with preempt disabled */
  407. cpu_idle();
  408. }
  409. /* Check for early params. */
  410. static int __init do_early_param(char *param, char *val)
  411. {
  412. struct obs_kernel_param *p;
  413. for (p = __setup_start; p < __setup_end; p++) {
  414. if ((p->early && strcmp(param, p->str) == 0) ||
  415. (strcmp(param, "console") == 0 &&
  416. strcmp(p->str, "earlycon") == 0)
  417. ) {
  418. if (p->setup_func(val) != 0)
  419. printk(KERN_WARNING
  420. "Malformed early option '%s'\n", param);
  421. }
  422. }
  423. /* We accept everything at this stage. */
  424. return 0;
  425. }
  426. void __init parse_early_options(char *cmdline)
  427. {
  428. parse_args("early options", cmdline, NULL, 0, do_early_param);
  429. }
  430. /* Arch code calls this early on, or if not, just before other parsing. */
  431. void __init parse_early_param(void)
  432. {
  433. static __initdata int done = 0;
  434. static __initdata char tmp_cmdline[COMMAND_LINE_SIZE];
  435. if (done)
  436. return;
  437. /* All fall through to do_early_param. */
  438. strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE);
  439. parse_early_options(tmp_cmdline);
  440. done = 1;
  441. }
  442. /*
  443. * Activate the first processor.
  444. */
  445. static void __init boot_cpu_init(void)
  446. {
  447. int cpu = smp_processor_id();
  448. /* Mark the boot cpu "present", "online" etc for SMP and UP case */
  449. set_cpu_online(cpu, true);
  450. set_cpu_present(cpu, true);
  451. set_cpu_possible(cpu, true);
  452. }
  453. void __init __weak smp_setup_processor_id(void)
  454. {
  455. }
  456. void __init __weak thread_info_cache_init(void)
  457. {
  458. }
  459. /*
  460. * Set up kernel memory allocators
  461. */
  462. static void __init mm_init(void)
  463. {
  464. /*
  465. * page_cgroup requires countinous pages as memmap
  466. * and it's bigger than MAX_ORDER unless SPARSEMEM.
  467. */
  468. page_cgroup_init_flatmem();
  469. mem_init();
  470. kmem_cache_init();
  471. vmalloc_init();
  472. }
  473. asmlinkage void __init start_kernel(void)
  474. {
  475. char * command_line;
  476. extern struct kernel_param __start___param[], __stop___param[];
  477. smp_setup_processor_id();
  478. /*
  479. * Need to run as early as possible, to initialize the
  480. * lockdep hash:
  481. */
  482. lockdep_init();
  483. debug_objects_early_init();
  484. /*
  485. * Set up the the initial canary ASAP:
  486. */
  487. boot_init_stack_canary();
  488. cgroup_init_early();
  489. local_irq_disable();
  490. early_boot_irqs_off();
  491. early_init_irq_lock_class();
  492. /*
  493. * Interrupts are still disabled. Do necessary setups, then
  494. * enable them
  495. */
  496. lock_kernel();
  497. tick_init();
  498. boot_cpu_init();
  499. page_address_init();
  500. printk(KERN_NOTICE "%s", linux_banner);
  501. setup_arch(&command_line);
  502. mm_init_owner(&init_mm, &init_task);
  503. setup_command_line(command_line);
  504. setup_per_cpu_areas();
  505. setup_nr_cpu_ids();
  506. smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */
  507. build_all_zonelists();
  508. page_alloc_init();
  509. printk(KERN_NOTICE "Kernel command line: %s\n", boot_command_line);
  510. parse_early_param();
  511. parse_args("Booting kernel", static_command_line, __start___param,
  512. __stop___param - __start___param,
  513. &unknown_bootoption);
  514. /*
  515. * These use large bootmem allocations and must precede
  516. * kmem_cache_init()
  517. */
  518. pidhash_init();
  519. vfs_caches_init_early();
  520. sort_main_extable();
  521. trap_init();
  522. mm_init();
  523. /*
  524. * Set up the scheduler prior starting any interrupts (such as the
  525. * timer interrupt). Full topology setup happens at smp_init()
  526. * time - but meanwhile we still have a functioning scheduler.
  527. */
  528. sched_init();
  529. /*
  530. * Disable preemption - early bootup scheduling is extremely
  531. * fragile until we cpu_idle() for the first time.
  532. */
  533. preempt_disable();
  534. if (!irqs_disabled()) {
  535. printk(KERN_WARNING "start_kernel(): bug: interrupts were "
  536. "enabled *very* early, fixing it\n");
  537. local_irq_disable();
  538. }
  539. rcu_init();
  540. /* init some links before init_ISA_irqs() */
  541. early_irq_init();
  542. init_IRQ();
  543. prio_tree_init();
  544. init_timers();
  545. hrtimers_init();
  546. softirq_init();
  547. timekeeping_init();
  548. time_init();
  549. sched_clock_init();
  550. profile_init();
  551. if (!irqs_disabled())
  552. printk(KERN_CRIT "start_kernel(): bug: interrupts were "
  553. "enabled early\n");
  554. early_boot_irqs_on();
  555. local_irq_enable();
  556. kmem_cache_init_late();
  557. /*
  558. * HACK ALERT! This is early. We're enabling the console before
  559. * we've done PCI setups etc, and console_init() must be aware of
  560. * this. But we do want output early, in case something goes wrong.
  561. */
  562. console_init();
  563. if (panic_later)
  564. panic(panic_later, panic_param);
  565. lockdep_info();
  566. /*
  567. * Need to run this when irqs are enabled, because it wants
  568. * to self-test [hard/soft]-irqs on/off lock inversion bugs
  569. * too:
  570. */
  571. locking_selftest();
  572. #ifdef CONFIG_BLK_DEV_INITRD
  573. if (initrd_start && !initrd_below_start_ok &&
  574. page_to_pfn(virt_to_page((void *)initrd_start)) < min_low_pfn) {
  575. printk(KERN_CRIT "initrd overwritten (0x%08lx < 0x%08lx) - "
  576. "disabling it.\n",
  577. page_to_pfn(virt_to_page((void *)initrd_start)),
  578. min_low_pfn);
  579. initrd_start = 0;
  580. }
  581. #endif
  582. cpuset_init_early();
  583. page_cgroup_init();
  584. enable_debug_pagealloc();
  585. cpu_hotplug_init();
  586. kmemtrace_init();
  587. kmemleak_init();
  588. debug_objects_mem_init();
  589. idr_init_cache();
  590. setup_per_cpu_pageset();
  591. numa_policy_init();
  592. if (late_time_init)
  593. late_time_init();
  594. calibrate_delay();
  595. pidmap_init();
  596. pgtable_cache_init();
  597. anon_vma_init();
  598. #ifdef CONFIG_X86
  599. if (efi_enabled)
  600. efi_enter_virtual_mode();
  601. #endif
  602. thread_info_cache_init();
  603. cred_init();
  604. fork_init(num_physpages);
  605. proc_caches_init();
  606. buffer_init();
  607. key_init();
  608. security_init();
  609. vfs_caches_init(num_physpages);
  610. radix_tree_init();
  611. signals_init();
  612. /* rootfs populating might need page-writeback */
  613. page_writeback_init();
  614. #ifdef CONFIG_PROC_FS
  615. proc_root_init();
  616. #endif
  617. cgroup_init();
  618. cpuset_init();
  619. taskstats_init_early();
  620. delayacct_init();
  621. check_bugs();
  622. acpi_early_init(); /* before LAPIC and SMP init */
  623. ftrace_init();
  624. /* Do the rest non-__init'ed, we're now alive */
  625. rest_init();
  626. }
  627. int initcall_debug;
  628. core_param(initcall_debug, initcall_debug, bool, 0644);
  629. int do_one_initcall(initcall_t fn)
  630. {
  631. int count = preempt_count();
  632. ktime_t calltime, delta, rettime;
  633. char msgbuf[64];
  634. struct boot_trace_call call;
  635. struct boot_trace_ret ret;
  636. if (initcall_debug) {
  637. call.caller = task_pid_nr(current);
  638. printk("calling %pF @ %i\n", fn, call.caller);
  639. calltime = ktime_get();
  640. trace_boot_call(&call, fn);
  641. enable_boot_trace();
  642. }
  643. ret.result = fn();
  644. if (initcall_debug) {
  645. disable_boot_trace();
  646. rettime = ktime_get();
  647. delta = ktime_sub(rettime, calltime);
  648. ret.duration = (unsigned long long) ktime_to_ns(delta) >> 10;
  649. trace_boot_ret(&ret, fn);
  650. printk("initcall %pF returned %d after %Ld usecs\n", fn,
  651. ret.result, ret.duration);
  652. }
  653. msgbuf[0] = 0;
  654. if (ret.result && ret.result != -ENODEV && initcall_debug)
  655. sprintf(msgbuf, "error code %d ", ret.result);
  656. if (preempt_count() != count) {
  657. strlcat(msgbuf, "preemption imbalance ", sizeof(msgbuf));
  658. preempt_count() = count;
  659. }
  660. if (irqs_disabled()) {
  661. strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf));
  662. local_irq_enable();
  663. }
  664. if (msgbuf[0]) {
  665. printk("initcall %pF returned with %s\n", fn, msgbuf);
  666. }
  667. return ret.result;
  668. }
  669. extern initcall_t __initcall_start[], __initcall_end[], __early_initcall_end[];
  670. static void __init do_initcalls(void)
  671. {
  672. initcall_t *call;
  673. for (call = __early_initcall_end; call < __initcall_end; call++)
  674. do_one_initcall(*call);
  675. /* Make sure there is no pending stuff from the initcall sequence */
  676. flush_scheduled_work();
  677. }
  678. /*
  679. * Ok, the machine is now initialized. None of the devices
  680. * have been touched yet, but the CPU subsystem is up and
  681. * running, and memory and process management works.
  682. *
  683. * Now we can finally start doing some real work..
  684. */
  685. static void __init do_basic_setup(void)
  686. {
  687. rcu_init_sched(); /* needed by module_init stage. */
  688. init_workqueues();
  689. cpuset_init_smp();
  690. usermodehelper_init();
  691. driver_init();
  692. init_irq_proc();
  693. do_initcalls();
  694. }
  695. static void __init do_pre_smp_initcalls(void)
  696. {
  697. initcall_t *call;
  698. for (call = __initcall_start; call < __early_initcall_end; call++)
  699. do_one_initcall(*call);
  700. }
  701. static void run_init_process(char *init_filename)
  702. {
  703. argv_init[0] = init_filename;
  704. kernel_execve(init_filename, argv_init, envp_init);
  705. }
  706. /* This is a non __init function. Force it to be noinline otherwise gcc
  707. * makes it inline to init() and it becomes part of init.text section
  708. */
  709. static noinline int init_post(void)
  710. __releases(kernel_lock)
  711. {
  712. /* need to finish all async __init code before freeing the memory */
  713. async_synchronize_full();
  714. free_initmem();
  715. unlock_kernel();
  716. mark_rodata_ro();
  717. system_state = SYSTEM_RUNNING;
  718. numa_default_policy();
  719. if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0)
  720. printk(KERN_WARNING "Warning: unable to open an initial console.\n");
  721. (void) sys_dup(0);
  722. (void) sys_dup(0);
  723. current->signal->flags |= SIGNAL_UNKILLABLE;
  724. if (ramdisk_execute_command) {
  725. run_init_process(ramdisk_execute_command);
  726. printk(KERN_WARNING "Failed to execute %s\n",
  727. ramdisk_execute_command);
  728. }
  729. /*
  730. * We try each of these until one succeeds.
  731. *
  732. * The Bourne shell can be used instead of init if we are
  733. * trying to recover a really broken machine.
  734. */
  735. if (execute_command) {
  736. run_init_process(execute_command);
  737. printk(KERN_WARNING "Failed to execute %s. Attempting "
  738. "defaults...\n", execute_command);
  739. }
  740. run_init_process("/sbin/init");
  741. run_init_process("/etc/init");
  742. run_init_process("/bin/init");
  743. run_init_process("/bin/sh");
  744. panic("No init found. Try passing init= option to kernel.");
  745. }
  746. static int __init kernel_init(void * unused)
  747. {
  748. lock_kernel();
  749. /*
  750. * init can run on any cpu.
  751. */
  752. set_cpus_allowed_ptr(current, cpu_all_mask);
  753. /*
  754. * Tell the world that we're going to be the grim
  755. * reaper of innocent orphaned children.
  756. *
  757. * We don't want people to have to make incorrect
  758. * assumptions about where in the task array this
  759. * can be found.
  760. */
  761. init_pid_ns.child_reaper = current;
  762. cad_pid = task_pid(current);
  763. smp_prepare_cpus(setup_max_cpus);
  764. do_pre_smp_initcalls();
  765. start_boot_trace();
  766. smp_init();
  767. sched_init_smp();
  768. do_basic_setup();
  769. /*
  770. * check if there is an early userspace init. If yes, let it do all
  771. * the work
  772. */
  773. if (!ramdisk_execute_command)
  774. ramdisk_execute_command = "/init";
  775. if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) {
  776. ramdisk_execute_command = NULL;
  777. prepare_namespace();
  778. }
  779. /*
  780. * Ok, we have completed the initial bootup, and
  781. * we're essentially up and running. Get rid of the
  782. * initmem segments and start the user-mode stuff..
  783. */
  784. init_post();
  785. return 0;
  786. }