main.c 22 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897898899900901902903904905906907908909910911912913914915916917918919920921922923924925926927928929930931
  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/workqueue.h>
  35. #include <linux/profile.h>
  36. #include <linux/rcupdate.h>
  37. #include <linux/moduleparam.h>
  38. #include <linux/kallsyms.h>
  39. #include <linux/writeback.h>
  40. #include <linux/cpu.h>
  41. #include <linux/cpuset.h>
  42. #include <linux/cgroup.h>
  43. #include <linux/efi.h>
  44. #include <linux/tick.h>
  45. #include <linux/interrupt.h>
  46. #include <linux/taskstats_kern.h>
  47. #include <linux/delayacct.h>
  48. #include <linux/unistd.h>
  49. #include <linux/rmap.h>
  50. #include <linux/mempolicy.h>
  51. #include <linux/key.h>
  52. #include <linux/buffer_head.h>
  53. #include <linux/page_cgroup.h>
  54. #include <linux/debug_locks.h>
  55. #include <linux/debugobjects.h>
  56. #include <linux/lockdep.h>
  57. #include <linux/kmemleak.h>
  58. #include <linux/pid_namespace.h>
  59. #include <linux/device.h>
  60. #include <linux/kthread.h>
  61. #include <linux/sched.h>
  62. #include <linux/signal.h>
  63. #include <linux/idr.h>
  64. #include <linux/kgdb.h>
  65. #include <linux/ftrace.h>
  66. #include <linux/async.h>
  67. #include <linux/kmemcheck.h>
  68. #include <linux/kmemtrace.h>
  69. #include <linux/sfi.h>
  70. #include <linux/shmem_fs.h>
  71. #include <linux/slab.h>
  72. #include <trace/boot.h>
  73. #include <asm/io.h>
  74. #include <asm/bugs.h>
  75. #include <asm/setup.h>
  76. #include <asm/sections.h>
  77. #include <asm/cacheflush.h>
  78. #ifdef CONFIG_X86_LOCAL_APIC
  79. #include <asm/smp.h>
  80. #endif
  81. static int kernel_init(void *);
  82. extern void init_IRQ(void);
  83. extern void fork_init(unsigned long);
  84. extern void mca_init(void);
  85. extern void sbus_init(void);
  86. extern void prio_tree_init(void);
  87. extern void radix_tree_init(void);
  88. extern void free_initmem(void);
  89. #ifndef CONFIG_DEBUG_RODATA
  90. static inline void mark_rodata_ro(void) { }
  91. #endif
  92. #ifdef CONFIG_TC
  93. extern void tc_init(void);
  94. #endif
  95. enum system_states system_state __read_mostly;
  96. EXPORT_SYMBOL(system_state);
  97. /*
  98. * Boot command-line arguments
  99. */
  100. #define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT
  101. #define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT
  102. extern void time_init(void);
  103. /* Default late time init is NULL. archs can override this later. */
  104. void (*__initdata late_time_init)(void);
  105. extern void softirq_init(void);
  106. /* Untouched command line saved by arch-specific code. */
  107. char __initdata boot_command_line[COMMAND_LINE_SIZE];
  108. /* Untouched saved command line (eg. for /proc) */
  109. char *saved_command_line;
  110. /* Command line for parameter parsing */
  111. static char *static_command_line;
  112. static char *execute_command;
  113. static char *ramdisk_execute_command;
  114. #ifdef CONFIG_SMP
  115. /* Setup configured maximum number of CPUs to activate */
  116. unsigned int setup_max_cpus = NR_CPUS;
  117. EXPORT_SYMBOL(setup_max_cpus);
  118. /*
  119. * Setup routine for controlling SMP activation
  120. *
  121. * Command-line option of "nosmp" or "maxcpus=0" will disable SMP
  122. * activation entirely (the MPS table probe still happens, though).
  123. *
  124. * Command-line option of "maxcpus=<NUM>", where <NUM> is an integer
  125. * greater than 0, limits the maximum number of CPUs activated in
  126. * SMP mode to <NUM>.
  127. */
  128. void __weak arch_disable_smp_support(void) { }
  129. static int __init nosmp(char *str)
  130. {
  131. setup_max_cpus = 0;
  132. arch_disable_smp_support();
  133. return 0;
  134. }
  135. early_param("nosmp", nosmp);
  136. /* this is hard limit */
  137. static int __init nrcpus(char *str)
  138. {
  139. int nr_cpus;
  140. get_option(&str, &nr_cpus);
  141. if (nr_cpus > 0 && nr_cpus < nr_cpu_ids)
  142. nr_cpu_ids = nr_cpus;
  143. return 0;
  144. }
  145. early_param("nr_cpus", nrcpus);
  146. static int __init maxcpus(char *str)
  147. {
  148. get_option(&str, &setup_max_cpus);
  149. if (setup_max_cpus == 0)
  150. arch_disable_smp_support();
  151. return 0;
  152. }
  153. early_param("maxcpus", maxcpus);
  154. #else
  155. static const unsigned int setup_max_cpus = NR_CPUS;
  156. #endif
  157. /*
  158. * If set, this is an indication to the drivers that reset the underlying
  159. * device before going ahead with the initialization otherwise driver might
  160. * rely on the BIOS and skip the reset operation.
  161. *
  162. * This is useful if kernel is booting in an unreliable environment.
  163. * For ex. kdump situaiton where previous kernel has crashed, BIOS has been
  164. * skipped and devices will be in unknown state.
  165. */
  166. unsigned int reset_devices;
  167. EXPORT_SYMBOL(reset_devices);
  168. static int __init set_reset_devices(char *str)
  169. {
  170. reset_devices = 1;
  171. return 1;
  172. }
  173. __setup("reset_devices", set_reset_devices);
  174. static char * argv_init[MAX_INIT_ARGS+2] = { "init", NULL, };
  175. char * envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, };
  176. static const char *panic_later, *panic_param;
  177. extern struct obs_kernel_param __setup_start[], __setup_end[];
  178. static int __init obsolete_checksetup(char *line)
  179. {
  180. struct obs_kernel_param *p;
  181. int had_early_param = 0;
  182. p = __setup_start;
  183. do {
  184. int n = strlen(p->str);
  185. if (!strncmp(line, p->str, n)) {
  186. if (p->early) {
  187. /* Already done in parse_early_param?
  188. * (Needs exact match on param part).
  189. * Keep iterating, as we can have early
  190. * params and __setups of same names 8( */
  191. if (line[n] == '\0' || line[n] == '=')
  192. had_early_param = 1;
  193. } else if (!p->setup_func) {
  194. printk(KERN_WARNING "Parameter %s is obsolete,"
  195. " ignored\n", p->str);
  196. return 1;
  197. } else if (p->setup_func(line + n))
  198. return 1;
  199. }
  200. p++;
  201. } while (p < __setup_end);
  202. return had_early_param;
  203. }
  204. /*
  205. * This should be approx 2 Bo*oMips to start (note initial shift), and will
  206. * still work even if initially too large, it will just take slightly longer
  207. */
  208. unsigned long loops_per_jiffy = (1<<12);
  209. EXPORT_SYMBOL(loops_per_jiffy);
  210. static int __init debug_kernel(char *str)
  211. {
  212. console_loglevel = 10;
  213. return 0;
  214. }
  215. static int __init quiet_kernel(char *str)
  216. {
  217. console_loglevel = 4;
  218. return 0;
  219. }
  220. early_param("debug", debug_kernel);
  221. early_param("quiet", quiet_kernel);
  222. static int __init loglevel(char *str)
  223. {
  224. get_option(&str, &console_loglevel);
  225. return 0;
  226. }
  227. early_param("loglevel", loglevel);
  228. /*
  229. * Unknown boot options get handed to init, unless they look like
  230. * unused parameters (modprobe will find them in /proc/cmdline).
  231. */
  232. static int __init unknown_bootoption(char *param, char *val)
  233. {
  234. /* Change NUL term back to "=", to make "param" the whole string. */
  235. if (val) {
  236. /* param=val or param="val"? */
  237. if (val == param+strlen(param)+1)
  238. val[-1] = '=';
  239. else if (val == param+strlen(param)+2) {
  240. val[-2] = '=';
  241. memmove(val-1, val, strlen(val)+1);
  242. val--;
  243. } else
  244. BUG();
  245. }
  246. /* Handle obsolete-style parameters */
  247. if (obsolete_checksetup(param))
  248. return 0;
  249. /* Unused module parameter. */
  250. if (strchr(param, '.') && (!val || strchr(param, '.') < val))
  251. return 0;
  252. if (panic_later)
  253. return 0;
  254. if (val) {
  255. /* Environment option */
  256. unsigned int i;
  257. for (i = 0; envp_init[i]; i++) {
  258. if (i == MAX_INIT_ENVS) {
  259. panic_later = "Too many boot env vars at `%s'";
  260. panic_param = param;
  261. }
  262. if (!strncmp(param, envp_init[i], val - param))
  263. break;
  264. }
  265. envp_init[i] = param;
  266. } else {
  267. /* Command line option */
  268. unsigned int i;
  269. for (i = 0; argv_init[i]; i++) {
  270. if (i == MAX_INIT_ARGS) {
  271. panic_later = "Too many boot init vars at `%s'";
  272. panic_param = param;
  273. }
  274. }
  275. argv_init[i] = param;
  276. }
  277. return 0;
  278. }
  279. #ifdef CONFIG_DEBUG_PAGEALLOC
  280. int __read_mostly debug_pagealloc_enabled = 0;
  281. #endif
  282. static int __init init_setup(char *str)
  283. {
  284. unsigned int i;
  285. execute_command = str;
  286. /*
  287. * In case LILO is going to boot us with default command line,
  288. * it prepends "auto" before the whole cmdline which makes
  289. * the shell think it should execute a script with such name.
  290. * So we ignore all arguments entered _before_ init=... [MJ]
  291. */
  292. for (i = 1; i < MAX_INIT_ARGS; i++)
  293. argv_init[i] = NULL;
  294. return 1;
  295. }
  296. __setup("init=", init_setup);
  297. static int __init rdinit_setup(char *str)
  298. {
  299. unsigned int i;
  300. ramdisk_execute_command = str;
  301. /* See "auto" comment in init_setup */
  302. for (i = 1; i < MAX_INIT_ARGS; i++)
  303. argv_init[i] = NULL;
  304. return 1;
  305. }
  306. __setup("rdinit=", rdinit_setup);
  307. #ifndef CONFIG_SMP
  308. #ifdef CONFIG_X86_LOCAL_APIC
  309. static void __init smp_init(void)
  310. {
  311. APIC_init_uniprocessor();
  312. }
  313. #else
  314. #define smp_init() do { } while (0)
  315. #endif
  316. static inline void setup_nr_cpu_ids(void) { }
  317. static inline void smp_prepare_cpus(unsigned int maxcpus) { }
  318. #else
  319. /* Setup number of possible processor ids */
  320. int nr_cpu_ids __read_mostly = NR_CPUS;
  321. EXPORT_SYMBOL(nr_cpu_ids);
  322. /* An arch may set nr_cpu_ids earlier if needed, so this would be redundant */
  323. static void __init setup_nr_cpu_ids(void)
  324. {
  325. nr_cpu_ids = find_last_bit(cpumask_bits(cpu_possible_mask),NR_CPUS) + 1;
  326. }
  327. /* Called by boot processor to activate the rest. */
  328. static void __init smp_init(void)
  329. {
  330. unsigned int cpu;
  331. /* FIXME: This should be done in userspace --RR */
  332. for_each_present_cpu(cpu) {
  333. if (num_online_cpus() >= setup_max_cpus)
  334. break;
  335. if (!cpu_online(cpu))
  336. cpu_up(cpu);
  337. }
  338. /* Any cleanup work */
  339. printk(KERN_INFO "Brought up %ld CPUs\n", (long)num_online_cpus());
  340. smp_cpus_done(setup_max_cpus);
  341. }
  342. #endif
  343. /*
  344. * We need to store the untouched command line for future reference.
  345. * We also need to store the touched command line since the parameter
  346. * parsing is performed in place, and we should allow a component to
  347. * store reference of name/value for future reference.
  348. */
  349. static void __init setup_command_line(char *command_line)
  350. {
  351. saved_command_line = alloc_bootmem(strlen (boot_command_line)+1);
  352. static_command_line = alloc_bootmem(strlen (command_line)+1);
  353. strcpy (saved_command_line, boot_command_line);
  354. strcpy (static_command_line, command_line);
  355. }
  356. /*
  357. * We need to finalize in a non-__init function or else race conditions
  358. * between the root thread and the init thread may cause start_kernel to
  359. * be reaped by free_initmem before the root thread has proceeded to
  360. * cpu_idle.
  361. *
  362. * gcc-3.4 accidentally inlines this function, so use noinline.
  363. */
  364. static __initdata DECLARE_COMPLETION(kthreadd_done);
  365. static noinline void __init_refok rest_init(void)
  366. __releases(kernel_lock)
  367. {
  368. int pid;
  369. rcu_scheduler_starting();
  370. /*
  371. * We need to spawn init first so that it obtains pid 1, however
  372. * the init task will end up wanting to create kthreads, which, if
  373. * we schedule it before we create kthreadd, will OOPS.
  374. */
  375. kernel_thread(kernel_init, NULL, CLONE_FS | CLONE_SIGHAND);
  376. numa_default_policy();
  377. pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES);
  378. rcu_read_lock();
  379. kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns);
  380. rcu_read_unlock();
  381. complete(&kthreadd_done);
  382. unlock_kernel();
  383. /*
  384. * The boot idle thread must execute schedule()
  385. * at least once to get things moving:
  386. */
  387. init_idle_bootup_task(current);
  388. preempt_enable_no_resched();
  389. schedule();
  390. preempt_disable();
  391. /* Call into cpu_idle with preempt disabled */
  392. cpu_idle();
  393. }
  394. /* Check for early params. */
  395. static int __init do_early_param(char *param, char *val)
  396. {
  397. struct obs_kernel_param *p;
  398. for (p = __setup_start; p < __setup_end; p++) {
  399. if ((p->early && strcmp(param, p->str) == 0) ||
  400. (strcmp(param, "console") == 0 &&
  401. strcmp(p->str, "earlycon") == 0)
  402. ) {
  403. if (p->setup_func(val) != 0)
  404. printk(KERN_WARNING
  405. "Malformed early option '%s'\n", param);
  406. }
  407. }
  408. /* We accept everything at this stage. */
  409. return 0;
  410. }
  411. void __init parse_early_options(char *cmdline)
  412. {
  413. parse_args("early options", cmdline, NULL, 0, do_early_param);
  414. }
  415. /* Arch code calls this early on, or if not, just before other parsing. */
  416. void __init parse_early_param(void)
  417. {
  418. static __initdata int done = 0;
  419. static __initdata char tmp_cmdline[COMMAND_LINE_SIZE];
  420. if (done)
  421. return;
  422. /* All fall through to do_early_param. */
  423. strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE);
  424. parse_early_options(tmp_cmdline);
  425. done = 1;
  426. }
  427. /*
  428. * Activate the first processor.
  429. */
  430. static void __init boot_cpu_init(void)
  431. {
  432. int cpu = smp_processor_id();
  433. /* Mark the boot cpu "present", "online" etc for SMP and UP case */
  434. set_cpu_online(cpu, true);
  435. set_cpu_active(cpu, true);
  436. set_cpu_present(cpu, true);
  437. set_cpu_possible(cpu, true);
  438. }
  439. void __init __weak smp_setup_processor_id(void)
  440. {
  441. }
  442. void __init __weak thread_info_cache_init(void)
  443. {
  444. }
  445. /*
  446. * Set up kernel memory allocators
  447. */
  448. static void __init mm_init(void)
  449. {
  450. /*
  451. * page_cgroup requires countinous pages as memmap
  452. * and it's bigger than MAX_ORDER unless SPARSEMEM.
  453. */
  454. page_cgroup_init_flatmem();
  455. mem_init();
  456. kmem_cache_init();
  457. pgtable_cache_init();
  458. vmalloc_init();
  459. }
  460. asmlinkage void __init start_kernel(void)
  461. {
  462. char * command_line;
  463. extern struct kernel_param __start___param[], __stop___param[];
  464. smp_setup_processor_id();
  465. /*
  466. * Need to run as early as possible, to initialize the
  467. * lockdep hash:
  468. */
  469. lockdep_init();
  470. debug_objects_early_init();
  471. /*
  472. * Set up the the initial canary ASAP:
  473. */
  474. boot_init_stack_canary();
  475. cgroup_init_early();
  476. local_irq_disable();
  477. early_boot_irqs_off();
  478. early_init_irq_lock_class();
  479. /*
  480. * Interrupts are still disabled. Do necessary setups, then
  481. * enable them
  482. */
  483. lock_kernel();
  484. tick_init();
  485. boot_cpu_init();
  486. page_address_init();
  487. printk(KERN_NOTICE "%s", linux_banner);
  488. setup_arch(&command_line);
  489. mm_init_owner(&init_mm, &init_task);
  490. setup_command_line(command_line);
  491. setup_nr_cpu_ids();
  492. setup_per_cpu_areas();
  493. smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */
  494. build_all_zonelists(NULL);
  495. page_alloc_init();
  496. printk(KERN_NOTICE "Kernel command line: %s\n", boot_command_line);
  497. parse_early_param();
  498. parse_args("Booting kernel", static_command_line, __start___param,
  499. __stop___param - __start___param,
  500. &unknown_bootoption);
  501. /*
  502. * These use large bootmem allocations and must precede
  503. * kmem_cache_init()
  504. */
  505. pidhash_init();
  506. vfs_caches_init_early();
  507. sort_main_extable();
  508. trap_init();
  509. mm_init();
  510. /*
  511. * Set up the scheduler prior starting any interrupts (such as the
  512. * timer interrupt). Full topology setup happens at smp_init()
  513. * time - but meanwhile we still have a functioning scheduler.
  514. */
  515. sched_init();
  516. /*
  517. * Disable preemption - early bootup scheduling is extremely
  518. * fragile until we cpu_idle() for the first time.
  519. */
  520. preempt_disable();
  521. if (!irqs_disabled()) {
  522. printk(KERN_WARNING "start_kernel(): bug: interrupts were "
  523. "enabled *very* early, fixing it\n");
  524. local_irq_disable();
  525. }
  526. rcu_init();
  527. radix_tree_init();
  528. /* init some links before init_ISA_irqs() */
  529. early_irq_init();
  530. init_IRQ();
  531. prio_tree_init();
  532. init_timers();
  533. hrtimers_init();
  534. softirq_init();
  535. timekeeping_init();
  536. time_init();
  537. profile_init();
  538. if (!irqs_disabled())
  539. printk(KERN_CRIT "start_kernel(): bug: interrupts were "
  540. "enabled early\n");
  541. early_boot_irqs_on();
  542. local_irq_enable();
  543. /* Interrupts are enabled now so all GFP allocations are safe. */
  544. gfp_allowed_mask = __GFP_BITS_MASK;
  545. kmem_cache_init_late();
  546. /*
  547. * HACK ALERT! This is early. We're enabling the console before
  548. * we've done PCI setups etc, and console_init() must be aware of
  549. * this. But we do want output early, in case something goes wrong.
  550. */
  551. console_init();
  552. if (panic_later)
  553. panic(panic_later, panic_param);
  554. lockdep_info();
  555. /*
  556. * Need to run this when irqs are enabled, because it wants
  557. * to self-test [hard/soft]-irqs on/off lock inversion bugs
  558. * too:
  559. */
  560. locking_selftest();
  561. #ifdef CONFIG_BLK_DEV_INITRD
  562. if (initrd_start && !initrd_below_start_ok &&
  563. page_to_pfn(virt_to_page((void *)initrd_start)) < min_low_pfn) {
  564. printk(KERN_CRIT "initrd overwritten (0x%08lx < 0x%08lx) - "
  565. "disabling it.\n",
  566. page_to_pfn(virt_to_page((void *)initrd_start)),
  567. min_low_pfn);
  568. initrd_start = 0;
  569. }
  570. #endif
  571. page_cgroup_init();
  572. enable_debug_pagealloc();
  573. kmemtrace_init();
  574. kmemleak_init();
  575. debug_objects_mem_init();
  576. idr_init_cache();
  577. setup_per_cpu_pageset();
  578. numa_policy_init();
  579. if (late_time_init)
  580. late_time_init();
  581. sched_clock_init();
  582. calibrate_delay();
  583. pidmap_init();
  584. anon_vma_init();
  585. #ifdef CONFIG_X86
  586. if (efi_enabled)
  587. efi_enter_virtual_mode();
  588. #endif
  589. thread_info_cache_init();
  590. cred_init();
  591. fork_init(totalram_pages);
  592. proc_caches_init();
  593. buffer_init();
  594. key_init();
  595. security_init();
  596. dbg_late_init();
  597. vfs_caches_init(totalram_pages);
  598. signals_init();
  599. /* rootfs populating might need page-writeback */
  600. page_writeback_init();
  601. #ifdef CONFIG_PROC_FS
  602. proc_root_init();
  603. #endif
  604. cgroup_init();
  605. cpuset_init();
  606. taskstats_init_early();
  607. delayacct_init();
  608. check_bugs();
  609. acpi_early_init(); /* before LAPIC and SMP init */
  610. sfi_init_late();
  611. ftrace_init();
  612. /* Do the rest non-__init'ed, we're now alive */
  613. rest_init();
  614. }
  615. /* Call all constructor functions linked into the kernel. */
  616. static void __init do_ctors(void)
  617. {
  618. #ifdef CONFIG_CONSTRUCTORS
  619. ctor_fn_t *fn = (ctor_fn_t *) __ctors_start;
  620. for (; fn < (ctor_fn_t *) __ctors_end; fn++)
  621. (*fn)();
  622. #endif
  623. }
  624. int initcall_debug;
  625. core_param(initcall_debug, initcall_debug, bool, 0644);
  626. static char msgbuf[64];
  627. static struct boot_trace_call call;
  628. static struct boot_trace_ret ret;
  629. int do_one_initcall(initcall_t fn)
  630. {
  631. int count = preempt_count();
  632. ktime_t calltime, delta, rettime;
  633. if (initcall_debug) {
  634. call.caller = task_pid_nr(current);
  635. printk("calling %pF @ %i\n", fn, call.caller);
  636. calltime = ktime_get();
  637. trace_boot_call(&call, fn);
  638. enable_boot_trace();
  639. }
  640. ret.result = fn();
  641. if (initcall_debug) {
  642. disable_boot_trace();
  643. rettime = ktime_get();
  644. delta = ktime_sub(rettime, calltime);
  645. ret.duration = (unsigned long long) ktime_to_ns(delta) >> 10;
  646. trace_boot_ret(&ret, fn);
  647. printk("initcall %pF returned %d after %Ld usecs\n", fn,
  648. ret.result, ret.duration);
  649. }
  650. msgbuf[0] = 0;
  651. if (ret.result && ret.result != -ENODEV && initcall_debug)
  652. sprintf(msgbuf, "error code %d ", ret.result);
  653. if (preempt_count() != count) {
  654. strlcat(msgbuf, "preemption imbalance ", sizeof(msgbuf));
  655. preempt_count() = count;
  656. }
  657. if (irqs_disabled()) {
  658. strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf));
  659. local_irq_enable();
  660. }
  661. if (msgbuf[0]) {
  662. printk("initcall %pF returned with %s\n", fn, msgbuf);
  663. }
  664. return ret.result;
  665. }
  666. extern initcall_t __initcall_start[], __initcall_end[], __early_initcall_end[];
  667. static void __init do_initcalls(void)
  668. {
  669. initcall_t *fn;
  670. for (fn = __early_initcall_end; fn < __initcall_end; fn++)
  671. do_one_initcall(*fn);
  672. /* Make sure there is no pending stuff from the initcall sequence */
  673. flush_scheduled_work();
  674. }
  675. /*
  676. * Ok, the machine is now initialized. None of the devices
  677. * have been touched yet, but the CPU subsystem is up and
  678. * running, and memory and process management works.
  679. *
  680. * Now we can finally start doing some real work..
  681. */
  682. static void __init do_basic_setup(void)
  683. {
  684. init_workqueues();
  685. cpuset_init_smp();
  686. usermodehelper_init();
  687. init_tmpfs();
  688. driver_init();
  689. init_irq_proc();
  690. do_ctors();
  691. do_initcalls();
  692. }
  693. static void __init do_pre_smp_initcalls(void)
  694. {
  695. initcall_t *fn;
  696. for (fn = __initcall_start; fn < __early_initcall_end; fn++)
  697. do_one_initcall(*fn);
  698. }
  699. static void run_init_process(char *init_filename)
  700. {
  701. argv_init[0] = init_filename;
  702. kernel_execve(init_filename, argv_init, envp_init);
  703. }
  704. /* This is a non __init function. Force it to be noinline otherwise gcc
  705. * makes it inline to init() and it becomes part of init.text section
  706. */
  707. static noinline int init_post(void)
  708. __releases(kernel_lock)
  709. {
  710. /* need to finish all async __init code before freeing the memory */
  711. async_synchronize_full();
  712. free_initmem();
  713. unlock_kernel();
  714. mark_rodata_ro();
  715. system_state = SYSTEM_RUNNING;
  716. numa_default_policy();
  717. current->signal->flags |= SIGNAL_UNKILLABLE;
  718. if (ramdisk_execute_command) {
  719. run_init_process(ramdisk_execute_command);
  720. printk(KERN_WARNING "Failed to execute %s\n",
  721. ramdisk_execute_command);
  722. }
  723. /*
  724. * We try each of these until one succeeds.
  725. *
  726. * The Bourne shell can be used instead of init if we are
  727. * trying to recover a really broken machine.
  728. */
  729. if (execute_command) {
  730. run_init_process(execute_command);
  731. printk(KERN_WARNING "Failed to execute %s. Attempting "
  732. "defaults...\n", execute_command);
  733. }
  734. run_init_process("/sbin/init");
  735. run_init_process("/etc/init");
  736. run_init_process("/bin/init");
  737. run_init_process("/bin/sh");
  738. panic("No init found. Try passing init= option to kernel. "
  739. "See Linux Documentation/init.txt for guidance.");
  740. }
  741. static int __init kernel_init(void * unused)
  742. {
  743. /*
  744. * Wait until kthreadd is all set-up.
  745. */
  746. wait_for_completion(&kthreadd_done);
  747. lock_kernel();
  748. /*
  749. * init can allocate pages on any node
  750. */
  751. set_mems_allowed(node_states[N_HIGH_MEMORY]);
  752. /*
  753. * init can run on any cpu.
  754. */
  755. set_cpus_allowed_ptr(current, cpu_all_mask);
  756. /*
  757. * Tell the world that we're going to be the grim
  758. * reaper of innocent orphaned children.
  759. *
  760. * We don't want people to have to make incorrect
  761. * assumptions about where in the task array this
  762. * can be found.
  763. */
  764. init_pid_ns.child_reaper = current;
  765. cad_pid = task_pid(current);
  766. smp_prepare_cpus(setup_max_cpus);
  767. do_pre_smp_initcalls();
  768. start_boot_trace();
  769. smp_init();
  770. sched_init_smp();
  771. do_basic_setup();
  772. /* Open the /dev/console on the rootfs, this should never fail */
  773. if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0)
  774. printk(KERN_WARNING "Warning: unable to open an initial console.\n");
  775. (void) sys_dup(0);
  776. (void) sys_dup(0);
  777. /*
  778. * check if there is an early userspace init. If yes, let it do all
  779. * the work
  780. */
  781. if (!ramdisk_execute_command)
  782. ramdisk_execute_command = "/init";
  783. if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) {
  784. ramdisk_execute_command = NULL;
  785. prepare_namespace();
  786. }
  787. /*
  788. * Ok, we have completed the initial bootup, and
  789. * we're essentially up and running. Get rid of the
  790. * initmem segments and start the user-mode stuff..
  791. */
  792. init_post();
  793. return 0;
  794. }