main.c 22 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921
  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/ioport.h>
  21. #include <linux/init.h>
  22. #include <linux/smp_lock.h>
  23. #include <linux/initrd.h>
  24. #include <linux/bootmem.h>
  25. #include <linux/acpi.h>
  26. #include <linux/tty.h>
  27. #include <linux/percpu.h>
  28. #include <linux/kmod.h>
  29. #include <linux/vmalloc.h>
  30. #include <linux/kernel_stat.h>
  31. #include <linux/start_kernel.h>
  32. #include <linux/security.h>
  33. #include <linux/smp.h>
  34. #include <linux/profile.h>
  35. #include <linux/rcupdate.h>
  36. #include <linux/moduleparam.h>
  37. #include <linux/kallsyms.h>
  38. #include <linux/writeback.h>
  39. #include <linux/cpu.h>
  40. #include <linux/cpuset.h>
  41. #include <linux/cgroup.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/buffer_head.h>
  52. #include <linux/page_cgroup.h>
  53. #include <linux/debug_locks.h>
  54. #include <linux/debugobjects.h>
  55. #include <linux/lockdep.h>
  56. #include <linux/kmemleak.h>
  57. #include <linux/pid_namespace.h>
  58. #include <linux/device.h>
  59. #include <linux/kthread.h>
  60. #include <linux/sched.h>
  61. #include <linux/signal.h>
  62. #include <linux/idr.h>
  63. #include <linux/kgdb.h>
  64. #include <linux/ftrace.h>
  65. #include <linux/async.h>
  66. #include <linux/kmemcheck.h>
  67. #include <linux/sfi.h>
  68. #include <linux/shmem_fs.h>
  69. #include <linux/slab.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. #ifndef CONFIG_DEBUG_RODATA
  87. static inline void mark_rodata_ro(void) { }
  88. #endif
  89. #ifdef CONFIG_TC
  90. extern void tc_init(void);
  91. #endif
  92. enum system_states system_state __read_mostly;
  93. EXPORT_SYMBOL(system_state);
  94. /*
  95. * Boot command-line arguments
  96. */
  97. #define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT
  98. #define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT
  99. extern void time_init(void);
  100. /* Default late time init is NULL. archs can override this later. */
  101. void (*__initdata late_time_init)(void);
  102. extern void softirq_init(void);
  103. /* Untouched command line saved by arch-specific code. */
  104. char __initdata boot_command_line[COMMAND_LINE_SIZE];
  105. /* Untouched saved command line (eg. for /proc) */
  106. char *saved_command_line;
  107. /* Command line for parameter parsing */
  108. static char *static_command_line;
  109. static char *execute_command;
  110. static char *ramdisk_execute_command;
  111. #ifdef CONFIG_SMP
  112. /* Setup configured maximum number of CPUs to activate */
  113. unsigned int setup_max_cpus = NR_CPUS;
  114. EXPORT_SYMBOL(setup_max_cpus);
  115. /*
  116. * Setup routine for controlling SMP activation
  117. *
  118. * Command-line option of "nosmp" or "maxcpus=0" will disable SMP
  119. * activation entirely (the MPS table probe still happens, though).
  120. *
  121. * Command-line option of "maxcpus=<NUM>", where <NUM> is an integer
  122. * greater than 0, limits the maximum number of CPUs activated in
  123. * SMP mode to <NUM>.
  124. */
  125. void __weak arch_disable_smp_support(void) { }
  126. static int __init nosmp(char *str)
  127. {
  128. setup_max_cpus = 0;
  129. arch_disable_smp_support();
  130. return 0;
  131. }
  132. early_param("nosmp", nosmp);
  133. /* this is hard limit */
  134. static int __init nrcpus(char *str)
  135. {
  136. int nr_cpus;
  137. get_option(&str, &nr_cpus);
  138. if (nr_cpus > 0 && nr_cpus < nr_cpu_ids)
  139. nr_cpu_ids = nr_cpus;
  140. return 0;
  141. }
  142. early_param("nr_cpus", nrcpus);
  143. static int __init maxcpus(char *str)
  144. {
  145. get_option(&str, &setup_max_cpus);
  146. if (setup_max_cpus == 0)
  147. arch_disable_smp_support();
  148. return 0;
  149. }
  150. early_param("maxcpus", maxcpus);
  151. #else
  152. static const unsigned int setup_max_cpus = NR_CPUS;
  153. #endif
  154. /*
  155. * If set, this is an indication to the drivers that reset the underlying
  156. * device before going ahead with the initialization otherwise driver might
  157. * rely on the BIOS and skip the reset operation.
  158. *
  159. * This is useful if kernel is booting in an unreliable environment.
  160. * For ex. kdump situaiton where previous kernel has crashed, BIOS has been
  161. * skipped and devices will be in unknown state.
  162. */
  163. unsigned int reset_devices;
  164. EXPORT_SYMBOL(reset_devices);
  165. static int __init set_reset_devices(char *str)
  166. {
  167. reset_devices = 1;
  168. return 1;
  169. }
  170. __setup("reset_devices", set_reset_devices);
  171. static const char * argv_init[MAX_INIT_ARGS+2] = { "init", NULL, };
  172. const char * envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, };
  173. static const char *panic_later, *panic_param;
  174. extern const struct obs_kernel_param __setup_start[], __setup_end[];
  175. static int __init obsolete_checksetup(char *line)
  176. {
  177. const struct obs_kernel_param *p;
  178. int had_early_param = 0;
  179. p = __setup_start;
  180. do {
  181. int n = strlen(p->str);
  182. if (!strncmp(line, p->str, n)) {
  183. if (p->early) {
  184. /* Already done in parse_early_param?
  185. * (Needs exact match on param part).
  186. * Keep iterating, as we can have early
  187. * params and __setups of same names 8( */
  188. if (line[n] == '\0' || line[n] == '=')
  189. had_early_param = 1;
  190. } else if (!p->setup_func) {
  191. printk(KERN_WARNING "Parameter %s is obsolete,"
  192. " ignored\n", p->str);
  193. return 1;
  194. } else if (p->setup_func(line + n))
  195. return 1;
  196. }
  197. p++;
  198. } while (p < __setup_end);
  199. return had_early_param;
  200. }
  201. /*
  202. * This should be approx 2 Bo*oMips to start (note initial shift), and will
  203. * still work even if initially too large, it will just take slightly longer
  204. */
  205. unsigned long loops_per_jiffy = (1<<12);
  206. EXPORT_SYMBOL(loops_per_jiffy);
  207. static int __init debug_kernel(char *str)
  208. {
  209. console_loglevel = 10;
  210. return 0;
  211. }
  212. static int __init quiet_kernel(char *str)
  213. {
  214. console_loglevel = 4;
  215. return 0;
  216. }
  217. early_param("debug", debug_kernel);
  218. early_param("quiet", quiet_kernel);
  219. static int __init loglevel(char *str)
  220. {
  221. get_option(&str, &console_loglevel);
  222. return 0;
  223. }
  224. early_param("loglevel", loglevel);
  225. /*
  226. * Unknown boot options get handed to init, unless they look like
  227. * unused parameters (modprobe will find them in /proc/cmdline).
  228. */
  229. static int __init unknown_bootoption(char *param, char *val)
  230. {
  231. /* Change NUL term back to "=", to make "param" the whole string. */
  232. if (val) {
  233. /* param=val or param="val"? */
  234. if (val == param+strlen(param)+1)
  235. val[-1] = '=';
  236. else if (val == param+strlen(param)+2) {
  237. val[-2] = '=';
  238. memmove(val-1, val, strlen(val)+1);
  239. val--;
  240. } else
  241. BUG();
  242. }
  243. /* Handle obsolete-style parameters */
  244. if (obsolete_checksetup(param))
  245. return 0;
  246. /* Unused module parameter. */
  247. if (strchr(param, '.') && (!val || strchr(param, '.') < val))
  248. return 0;
  249. if (panic_later)
  250. return 0;
  251. if (val) {
  252. /* Environment option */
  253. unsigned int i;
  254. for (i = 0; envp_init[i]; i++) {
  255. if (i == MAX_INIT_ENVS) {
  256. panic_later = "Too many boot env vars at `%s'";
  257. panic_param = param;
  258. }
  259. if (!strncmp(param, envp_init[i], val - param))
  260. break;
  261. }
  262. envp_init[i] = param;
  263. } else {
  264. /* Command line option */
  265. unsigned int i;
  266. for (i = 0; argv_init[i]; i++) {
  267. if (i == MAX_INIT_ARGS) {
  268. panic_later = "Too many boot init vars at `%s'";
  269. panic_param = param;
  270. }
  271. }
  272. argv_init[i] = param;
  273. }
  274. return 0;
  275. }
  276. #ifdef CONFIG_DEBUG_PAGEALLOC
  277. int __read_mostly debug_pagealloc_enabled = 0;
  278. #endif
  279. static int __init init_setup(char *str)
  280. {
  281. unsigned int i;
  282. execute_command = str;
  283. /*
  284. * In case LILO is going to boot us with default command line,
  285. * it prepends "auto" before the whole cmdline which makes
  286. * the shell think it should execute a script with such name.
  287. * So we ignore all arguments entered _before_ init=... [MJ]
  288. */
  289. for (i = 1; i < MAX_INIT_ARGS; i++)
  290. argv_init[i] = NULL;
  291. return 1;
  292. }
  293. __setup("init=", init_setup);
  294. static int __init rdinit_setup(char *str)
  295. {
  296. unsigned int i;
  297. ramdisk_execute_command = str;
  298. /* See "auto" comment in init_setup */
  299. for (i = 1; i < MAX_INIT_ARGS; i++)
  300. argv_init[i] = NULL;
  301. return 1;
  302. }
  303. __setup("rdinit=", rdinit_setup);
  304. #ifndef CONFIG_SMP
  305. #ifdef CONFIG_X86_LOCAL_APIC
  306. static void __init smp_init(void)
  307. {
  308. APIC_init_uniprocessor();
  309. }
  310. #else
  311. #define smp_init() do { } while (0)
  312. #endif
  313. static inline void setup_nr_cpu_ids(void) { }
  314. static inline void smp_prepare_cpus(unsigned int maxcpus) { }
  315. #else
  316. /* Setup number of possible processor ids */
  317. int nr_cpu_ids __read_mostly = NR_CPUS;
  318. EXPORT_SYMBOL(nr_cpu_ids);
  319. /* An arch may set nr_cpu_ids earlier if needed, so this would be redundant */
  320. static void __init setup_nr_cpu_ids(void)
  321. {
  322. nr_cpu_ids = find_last_bit(cpumask_bits(cpu_possible_mask),NR_CPUS) + 1;
  323. }
  324. /* Called by boot processor to activate the rest. */
  325. static void __init smp_init(void)
  326. {
  327. unsigned int cpu;
  328. /* FIXME: This should be done in userspace --RR */
  329. for_each_present_cpu(cpu) {
  330. if (num_online_cpus() >= setup_max_cpus)
  331. break;
  332. if (!cpu_online(cpu))
  333. cpu_up(cpu);
  334. }
  335. /* Any cleanup work */
  336. printk(KERN_INFO "Brought up %ld CPUs\n", (long)num_online_cpus());
  337. smp_cpus_done(setup_max_cpus);
  338. }
  339. #endif
  340. /*
  341. * We need to store the untouched command line for future reference.
  342. * We also need to store the touched command line since the parameter
  343. * parsing is performed in place, and we should allow a component to
  344. * store reference of name/value for future reference.
  345. */
  346. static void __init setup_command_line(char *command_line)
  347. {
  348. saved_command_line = alloc_bootmem(strlen (boot_command_line)+1);
  349. static_command_line = alloc_bootmem(strlen (command_line)+1);
  350. strcpy (saved_command_line, boot_command_line);
  351. strcpy (static_command_line, command_line);
  352. }
  353. /*
  354. * We need to finalize in a non-__init function or else race conditions
  355. * between the root thread and the init thread may cause start_kernel to
  356. * be reaped by free_initmem before the root thread has proceeded to
  357. * cpu_idle.
  358. *
  359. * gcc-3.4 accidentally inlines this function, so use noinline.
  360. */
  361. static __initdata DECLARE_COMPLETION(kthreadd_done);
  362. static noinline void __init_refok rest_init(void)
  363. __releases(kernel_lock)
  364. {
  365. int pid;
  366. rcu_scheduler_starting();
  367. /*
  368. * We need to spawn init first so that it obtains pid 1, however
  369. * the init task will end up wanting to create kthreads, which, if
  370. * we schedule it before we create kthreadd, will OOPS.
  371. */
  372. kernel_thread(kernel_init, NULL, CLONE_FS | CLONE_SIGHAND);
  373. numa_default_policy();
  374. pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES);
  375. rcu_read_lock();
  376. kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns);
  377. rcu_read_unlock();
  378. complete(&kthreadd_done);
  379. /*
  380. * The boot idle thread must execute schedule()
  381. * at least once to get things moving:
  382. */
  383. init_idle_bootup_task(current);
  384. preempt_enable_no_resched();
  385. schedule();
  386. preempt_disable();
  387. /* Call into cpu_idle with preempt disabled */
  388. cpu_idle();
  389. }
  390. /* Check for early params. */
  391. static int __init do_early_param(char *param, char *val)
  392. {
  393. const struct obs_kernel_param *p;
  394. for (p = __setup_start; p < __setup_end; p++) {
  395. if ((p->early && strcmp(param, p->str) == 0) ||
  396. (strcmp(param, "console") == 0 &&
  397. strcmp(p->str, "earlycon") == 0)
  398. ) {
  399. if (p->setup_func(val) != 0)
  400. printk(KERN_WARNING
  401. "Malformed early option '%s'\n", param);
  402. }
  403. }
  404. /* We accept everything at this stage. */
  405. return 0;
  406. }
  407. void __init parse_early_options(char *cmdline)
  408. {
  409. parse_args("early options", cmdline, NULL, 0, do_early_param);
  410. }
  411. /* Arch code calls this early on, or if not, just before other parsing. */
  412. void __init parse_early_param(void)
  413. {
  414. static __initdata int done = 0;
  415. static __initdata char tmp_cmdline[COMMAND_LINE_SIZE];
  416. if (done)
  417. return;
  418. /* All fall through to do_early_param. */
  419. strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE);
  420. parse_early_options(tmp_cmdline);
  421. done = 1;
  422. }
  423. /*
  424. * Activate the first processor.
  425. */
  426. static void __init boot_cpu_init(void)
  427. {
  428. int cpu = smp_processor_id();
  429. /* Mark the boot cpu "present", "online" etc for SMP and UP case */
  430. set_cpu_online(cpu, true);
  431. set_cpu_active(cpu, true);
  432. set_cpu_present(cpu, true);
  433. set_cpu_possible(cpu, true);
  434. }
  435. void __init __weak smp_setup_processor_id(void)
  436. {
  437. }
  438. void __init __weak thread_info_cache_init(void)
  439. {
  440. }
  441. /*
  442. * Set up kernel memory allocators
  443. */
  444. static void __init mm_init(void)
  445. {
  446. /*
  447. * page_cgroup requires countinous pages as memmap
  448. * and it's bigger than MAX_ORDER unless SPARSEMEM.
  449. */
  450. page_cgroup_init_flatmem();
  451. mem_init();
  452. kmem_cache_init();
  453. percpu_init_late();
  454. pgtable_cache_init();
  455. vmalloc_init();
  456. }
  457. asmlinkage void __init start_kernel(void)
  458. {
  459. char * command_line;
  460. extern const struct kernel_param __start___param[], __stop___param[];
  461. smp_setup_processor_id();
  462. /*
  463. * Need to run as early as possible, to initialize the
  464. * lockdep hash:
  465. */
  466. lockdep_init();
  467. debug_objects_early_init();
  468. /*
  469. * Set up the the initial canary ASAP:
  470. */
  471. boot_init_stack_canary();
  472. cgroup_init_early();
  473. local_irq_disable();
  474. early_boot_irqs_off();
  475. /*
  476. * Interrupts are still disabled. Do necessary setups, then
  477. * enable them
  478. */
  479. tick_init();
  480. boot_cpu_init();
  481. page_address_init();
  482. printk(KERN_NOTICE "%s", linux_banner);
  483. setup_arch(&command_line);
  484. mm_init_owner(&init_mm, &init_task);
  485. setup_command_line(command_line);
  486. setup_nr_cpu_ids();
  487. setup_per_cpu_areas();
  488. smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */
  489. build_all_zonelists(NULL);
  490. page_alloc_init();
  491. printk(KERN_NOTICE "Kernel command line: %s\n", boot_command_line);
  492. parse_early_param();
  493. parse_args("Booting kernel", static_command_line, __start___param,
  494. __stop___param - __start___param,
  495. &unknown_bootoption);
  496. /*
  497. * These use large bootmem allocations and must precede
  498. * kmem_cache_init()
  499. */
  500. pidhash_init();
  501. vfs_caches_init_early();
  502. sort_main_extable();
  503. trap_init();
  504. mm_init();
  505. /*
  506. * Set up the scheduler prior starting any interrupts (such as the
  507. * timer interrupt). Full topology setup happens at smp_init()
  508. * time - but meanwhile we still have a functioning scheduler.
  509. */
  510. sched_init();
  511. /*
  512. * Disable preemption - early bootup scheduling is extremely
  513. * fragile until we cpu_idle() for the first time.
  514. */
  515. preempt_disable();
  516. if (!irqs_disabled()) {
  517. printk(KERN_WARNING "start_kernel(): bug: interrupts were "
  518. "enabled *very* early, fixing it\n");
  519. local_irq_disable();
  520. }
  521. rcu_init();
  522. radix_tree_init();
  523. /* init some links before init_ISA_irqs() */
  524. early_irq_init();
  525. init_IRQ();
  526. prio_tree_init();
  527. init_timers();
  528. hrtimers_init();
  529. softirq_init();
  530. timekeeping_init();
  531. time_init();
  532. profile_init();
  533. if (!irqs_disabled())
  534. printk(KERN_CRIT "start_kernel(): bug: interrupts were "
  535. "enabled early\n");
  536. early_boot_irqs_on();
  537. local_irq_enable();
  538. /* Interrupts are enabled now so all GFP allocations are safe. */
  539. gfp_allowed_mask = __GFP_BITS_MASK;
  540. kmem_cache_init_late();
  541. /*
  542. * HACK ALERT! This is early. We're enabling the console before
  543. * we've done PCI setups etc, and console_init() must be aware of
  544. * this. But we do want output early, in case something goes wrong.
  545. */
  546. console_init();
  547. if (panic_later)
  548. panic(panic_later, panic_param);
  549. lockdep_info();
  550. /*
  551. * Need to run this when irqs are enabled, because it wants
  552. * to self-test [hard/soft]-irqs on/off lock inversion bugs
  553. * too:
  554. */
  555. locking_selftest();
  556. #ifdef CONFIG_BLK_DEV_INITRD
  557. if (initrd_start && !initrd_below_start_ok &&
  558. page_to_pfn(virt_to_page((void *)initrd_start)) < min_low_pfn) {
  559. printk(KERN_CRIT "initrd overwritten (0x%08lx < 0x%08lx) - "
  560. "disabling it.\n",
  561. page_to_pfn(virt_to_page((void *)initrd_start)),
  562. min_low_pfn);
  563. initrd_start = 0;
  564. }
  565. #endif
  566. page_cgroup_init();
  567. enable_debug_pagealloc();
  568. kmemleak_init();
  569. debug_objects_mem_init();
  570. idr_init_cache();
  571. setup_per_cpu_pageset();
  572. numa_policy_init();
  573. if (late_time_init)
  574. late_time_init();
  575. sched_clock_init();
  576. calibrate_delay();
  577. pidmap_init();
  578. anon_vma_init();
  579. #ifdef CONFIG_X86
  580. if (efi_enabled)
  581. efi_enter_virtual_mode();
  582. #endif
  583. thread_info_cache_init();
  584. cred_init();
  585. fork_init(totalram_pages);
  586. proc_caches_init();
  587. buffer_init();
  588. key_init();
  589. security_init();
  590. dbg_late_init();
  591. vfs_caches_init(totalram_pages);
  592. signals_init();
  593. /* rootfs populating might need page-writeback */
  594. page_writeback_init();
  595. #ifdef CONFIG_PROC_FS
  596. proc_root_init();
  597. #endif
  598. cgroup_init();
  599. cpuset_init();
  600. taskstats_init_early();
  601. delayacct_init();
  602. check_bugs();
  603. acpi_early_init(); /* before LAPIC and SMP init */
  604. sfi_init_late();
  605. ftrace_init();
  606. /* Do the rest non-__init'ed, we're now alive */
  607. rest_init();
  608. }
  609. /* Call all constructor functions linked into the kernel. */
  610. static void __init do_ctors(void)
  611. {
  612. #ifdef CONFIG_CONSTRUCTORS
  613. ctor_fn_t *fn = (ctor_fn_t *) __ctors_start;
  614. for (; fn < (ctor_fn_t *) __ctors_end; fn++)
  615. (*fn)();
  616. #endif
  617. }
  618. int initcall_debug;
  619. core_param(initcall_debug, initcall_debug, bool, 0644);
  620. static char msgbuf[64];
  621. static int __init_or_module do_one_initcall_debug(initcall_t fn)
  622. {
  623. ktime_t calltime, delta, rettime;
  624. unsigned long long duration;
  625. int ret;
  626. printk(KERN_DEBUG "calling %pF @ %i\n", fn, task_pid_nr(current));
  627. calltime = ktime_get();
  628. ret = fn();
  629. rettime = ktime_get();
  630. delta = ktime_sub(rettime, calltime);
  631. duration = (unsigned long long) ktime_to_ns(delta) >> 10;
  632. printk(KERN_DEBUG "initcall %pF returned %d after %lld usecs\n", fn,
  633. ret, duration);
  634. return ret;
  635. }
  636. int __init_or_module do_one_initcall(initcall_t fn)
  637. {
  638. int count = preempt_count();
  639. int ret;
  640. if (initcall_debug)
  641. ret = do_one_initcall_debug(fn);
  642. else
  643. ret = fn();
  644. msgbuf[0] = 0;
  645. if (ret && ret != -ENODEV && initcall_debug)
  646. sprintf(msgbuf, "error code %d ", ret);
  647. if (preempt_count() != count) {
  648. strlcat(msgbuf, "preemption imbalance ", sizeof(msgbuf));
  649. preempt_count() = count;
  650. }
  651. if (irqs_disabled()) {
  652. strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf));
  653. local_irq_enable();
  654. }
  655. if (msgbuf[0]) {
  656. printk("initcall %pF returned with %s\n", fn, msgbuf);
  657. }
  658. return ret;
  659. }
  660. extern initcall_t __initcall_start[], __initcall_end[], __early_initcall_end[];
  661. static void __init do_initcalls(void)
  662. {
  663. initcall_t *fn;
  664. for (fn = __early_initcall_end; fn < __initcall_end; fn++)
  665. do_one_initcall(*fn);
  666. /* Make sure there is no pending stuff from the initcall sequence */
  667. flush_scheduled_work();
  668. }
  669. /*
  670. * Ok, the machine is now initialized. None of the devices
  671. * have been touched yet, but the CPU subsystem is up and
  672. * running, and memory and process management works.
  673. *
  674. * Now we can finally start doing some real work..
  675. */
  676. static void __init do_basic_setup(void)
  677. {
  678. cpuset_init_smp();
  679. usermodehelper_init();
  680. init_tmpfs();
  681. driver_init();
  682. init_irq_proc();
  683. do_ctors();
  684. do_initcalls();
  685. }
  686. static void __init do_pre_smp_initcalls(void)
  687. {
  688. initcall_t *fn;
  689. for (fn = __initcall_start; fn < __early_initcall_end; fn++)
  690. do_one_initcall(*fn);
  691. }
  692. static void run_init_process(const char *init_filename)
  693. {
  694. argv_init[0] = init_filename;
  695. kernel_execve(init_filename, argv_init, envp_init);
  696. }
  697. /* This is a non __init function. Force it to be noinline otherwise gcc
  698. * makes it inline to init() and it becomes part of init.text section
  699. */
  700. static noinline int init_post(void)
  701. __releases(kernel_lock)
  702. {
  703. /* need to finish all async __init code before freeing the memory */
  704. async_synchronize_full();
  705. free_initmem();
  706. mark_rodata_ro();
  707. system_state = SYSTEM_RUNNING;
  708. numa_default_policy();
  709. current->signal->flags |= SIGNAL_UNKILLABLE;
  710. if (ramdisk_execute_command) {
  711. run_init_process(ramdisk_execute_command);
  712. printk(KERN_WARNING "Failed to execute %s\n",
  713. ramdisk_execute_command);
  714. }
  715. /*
  716. * We try each of these until one succeeds.
  717. *
  718. * The Bourne shell can be used instead of init if we are
  719. * trying to recover a really broken machine.
  720. */
  721. if (execute_command) {
  722. run_init_process(execute_command);
  723. printk(KERN_WARNING "Failed to execute %s. Attempting "
  724. "defaults...\n", execute_command);
  725. }
  726. run_init_process("/sbin/init");
  727. run_init_process("/etc/init");
  728. run_init_process("/bin/init");
  729. run_init_process("/bin/sh");
  730. panic("No init found. Try passing init= option to kernel. "
  731. "See Linux Documentation/init.txt for guidance.");
  732. }
  733. static int __init kernel_init(void * unused)
  734. {
  735. /*
  736. * Wait until kthreadd is all set-up.
  737. */
  738. wait_for_completion(&kthreadd_done);
  739. /*
  740. * init can allocate pages on any node
  741. */
  742. set_mems_allowed(node_states[N_HIGH_MEMORY]);
  743. /*
  744. * init can run on any cpu.
  745. */
  746. set_cpus_allowed_ptr(current, cpu_all_mask);
  747. /*
  748. * Tell the world that we're going to be the grim
  749. * reaper of innocent orphaned children.
  750. *
  751. * We don't want people to have to make incorrect
  752. * assumptions about where in the task array this
  753. * can be found.
  754. */
  755. init_pid_ns.child_reaper = current;
  756. cad_pid = task_pid(current);
  757. smp_prepare_cpus(setup_max_cpus);
  758. do_pre_smp_initcalls();
  759. smp_init();
  760. sched_init_smp();
  761. do_basic_setup();
  762. /* Open the /dev/console on the rootfs, this should never fail */
  763. if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0)
  764. printk(KERN_WARNING "Warning: unable to open an initial console.\n");
  765. (void) sys_dup(0);
  766. (void) sys_dup(0);
  767. /*
  768. * check if there is an early userspace init. If yes, let it do all
  769. * the work
  770. */
  771. if (!ramdisk_execute_command)
  772. ramdisk_execute_command = "/init";
  773. if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) {
  774. ramdisk_execute_command = NULL;
  775. prepare_namespace();
  776. }
  777. /*
  778. * Ok, we have completed the initial bootup, and
  779. * we're essentially up and running. Get rid of the
  780. * initmem segments and start the user-mode stuff..
  781. */
  782. init_post();
  783. return 0;
  784. }