main.c 21 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892893894895896897
  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/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/ftrace.h>
  65. #include <linux/async.h>
  66. #include <trace/boot.h>
  67. #include <asm/io.h>
  68. #include <asm/bugs.h>
  69. #include <asm/setup.h>
  70. #include <asm/sections.h>
  71. #include <asm/cacheflush.h>
  72. #include <trace/kmemtrace.h>
  73. #ifdef CONFIG_X86_LOCAL_APIC
  74. #include <asm/smp.h>
  75. #endif
  76. static int kernel_init(void *);
  77. extern void init_IRQ(void);
  78. extern void fork_init(unsigned long);
  79. extern void mca_init(void);
  80. extern void sbus_init(void);
  81. extern void prio_tree_init(void);
  82. extern void radix_tree_init(void);
  83. extern void free_initmem(void);
  84. #ifdef CONFIG_ACPI
  85. extern void acpi_early_init(void);
  86. #else
  87. static inline void acpi_early_init(void) { }
  88. #endif
  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 __initdata setup_max_cpus = NR_CPUS;
  117. /*
  118. * Setup routine for controlling SMP activation
  119. *
  120. * Command-line option of "nosmp" or "maxcpus=0" will disable SMP
  121. * activation entirely (the MPS table probe still happens, though).
  122. *
  123. * Command-line option of "maxcpus=<NUM>", where <NUM> is an integer
  124. * greater than 0, limits the maximum number of CPUs activated in
  125. * SMP mode to <NUM>.
  126. */
  127. void __weak arch_disable_smp_support(void) { }
  128. static int __init nosmp(char *str)
  129. {
  130. setup_max_cpus = 0;
  131. arch_disable_smp_support();
  132. return 0;
  133. }
  134. early_param("nosmp", nosmp);
  135. static int __init maxcpus(char *str)
  136. {
  137. get_option(&str, &setup_max_cpus);
  138. if (setup_max_cpus == 0)
  139. arch_disable_smp_support();
  140. return 0;
  141. }
  142. early_param("maxcpus", maxcpus);
  143. #else
  144. const unsigned int setup_max_cpus = NR_CPUS;
  145. #endif
  146. /*
  147. * If set, this is an indication to the drivers that reset the underlying
  148. * device before going ahead with the initialization otherwise driver might
  149. * rely on the BIOS and skip the reset operation.
  150. *
  151. * This is useful if kernel is booting in an unreliable environment.
  152. * For ex. kdump situaiton where previous kernel has crashed, BIOS has been
  153. * skipped and devices will be in unknown state.
  154. */
  155. unsigned int reset_devices;
  156. EXPORT_SYMBOL(reset_devices);
  157. static int __init set_reset_devices(char *str)
  158. {
  159. reset_devices = 1;
  160. return 1;
  161. }
  162. __setup("reset_devices", set_reset_devices);
  163. static char * argv_init[MAX_INIT_ARGS+2] = { "init", NULL, };
  164. char * envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, };
  165. static const char *panic_later, *panic_param;
  166. extern struct obs_kernel_param __setup_start[], __setup_end[];
  167. static int __init obsolete_checksetup(char *line)
  168. {
  169. struct obs_kernel_param *p;
  170. int had_early_param = 0;
  171. p = __setup_start;
  172. do {
  173. int n = strlen(p->str);
  174. if (!strncmp(line, p->str, n)) {
  175. if (p->early) {
  176. /* Already done in parse_early_param?
  177. * (Needs exact match on param part).
  178. * Keep iterating, as we can have early
  179. * params and __setups of same names 8( */
  180. if (line[n] == '\0' || line[n] == '=')
  181. had_early_param = 1;
  182. } else if (!p->setup_func) {
  183. printk(KERN_WARNING "Parameter %s is obsolete,"
  184. " ignored\n", p->str);
  185. return 1;
  186. } else if (p->setup_func(line + n))
  187. return 1;
  188. }
  189. p++;
  190. } while (p < __setup_end);
  191. return had_early_param;
  192. }
  193. /*
  194. * This should be approx 2 Bo*oMips to start (note initial shift), and will
  195. * still work even if initially too large, it will just take slightly longer
  196. */
  197. unsigned long loops_per_jiffy = (1<<12);
  198. EXPORT_SYMBOL(loops_per_jiffy);
  199. static int __init debug_kernel(char *str)
  200. {
  201. console_loglevel = 10;
  202. return 0;
  203. }
  204. static int __init quiet_kernel(char *str)
  205. {
  206. console_loglevel = 4;
  207. return 0;
  208. }
  209. early_param("debug", debug_kernel);
  210. early_param("quiet", quiet_kernel);
  211. static int __init loglevel(char *str)
  212. {
  213. get_option(&str, &console_loglevel);
  214. return 0;
  215. }
  216. early_param("loglevel", loglevel);
  217. /*
  218. * Unknown boot options get handed to init, unless they look like
  219. * failed parameters
  220. */
  221. static int __init unknown_bootoption(char *param, char *val)
  222. {
  223. /* Change NUL term back to "=", to make "param" the whole string. */
  224. if (val) {
  225. /* param=val or param="val"? */
  226. if (val == param+strlen(param)+1)
  227. val[-1] = '=';
  228. else if (val == param+strlen(param)+2) {
  229. val[-2] = '=';
  230. memmove(val-1, val, strlen(val)+1);
  231. val--;
  232. } else
  233. BUG();
  234. }
  235. /* Handle obsolete-style parameters */
  236. if (obsolete_checksetup(param))
  237. return 0;
  238. /*
  239. * Preemptive maintenance for "why didn't my misspelled command
  240. * line work?"
  241. */
  242. if (strchr(param, '.') && (!val || strchr(param, '.') < val)) {
  243. printk(KERN_ERR "Unknown boot option `%s': ignoring\n", param);
  244. return 0;
  245. }
  246. if (panic_later)
  247. return 0;
  248. if (val) {
  249. /* Environment option */
  250. unsigned int i;
  251. for (i = 0; envp_init[i]; i++) {
  252. if (i == MAX_INIT_ENVS) {
  253. panic_later = "Too many boot env vars at `%s'";
  254. panic_param = param;
  255. }
  256. if (!strncmp(param, envp_init[i], val - param))
  257. break;
  258. }
  259. envp_init[i] = param;
  260. } else {
  261. /* Command line option */
  262. unsigned int i;
  263. for (i = 0; argv_init[i]; i++) {
  264. if (i == MAX_INIT_ARGS) {
  265. panic_later = "Too many boot init vars at `%s'";
  266. panic_param = param;
  267. }
  268. }
  269. argv_init[i] = param;
  270. }
  271. return 0;
  272. }
  273. #ifdef CONFIG_DEBUG_PAGEALLOC
  274. int __read_mostly debug_pagealloc_enabled = 0;
  275. #endif
  276. static int __init init_setup(char *str)
  277. {
  278. unsigned int i;
  279. execute_command = str;
  280. /*
  281. * In case LILO is going to boot us with default command line,
  282. * it prepends "auto" before the whole cmdline which makes
  283. * the shell think it should execute a script with such name.
  284. * So we ignore all arguments entered _before_ init=... [MJ]
  285. */
  286. for (i = 1; i < MAX_INIT_ARGS; i++)
  287. argv_init[i] = NULL;
  288. return 1;
  289. }
  290. __setup("init=", init_setup);
  291. static int __init rdinit_setup(char *str)
  292. {
  293. unsigned int i;
  294. ramdisk_execute_command = str;
  295. /* See "auto" comment in init_setup */
  296. for (i = 1; i < MAX_INIT_ARGS; i++)
  297. argv_init[i] = NULL;
  298. return 1;
  299. }
  300. __setup("rdinit=", rdinit_setup);
  301. #ifndef CONFIG_SMP
  302. #ifdef CONFIG_X86_LOCAL_APIC
  303. static void __init smp_init(void)
  304. {
  305. APIC_init_uniprocessor();
  306. }
  307. #else
  308. #define smp_init() do { } while (0)
  309. #endif
  310. static inline void setup_per_cpu_areas(void) { }
  311. static inline void setup_nr_cpu_ids(void) { }
  312. static inline void smp_prepare_cpus(unsigned int maxcpus) { }
  313. #else
  314. #if NR_CPUS > BITS_PER_LONG
  315. cpumask_t cpu_mask_all __read_mostly = CPU_MASK_ALL;
  316. EXPORT_SYMBOL(cpu_mask_all);
  317. #endif
  318. /* Setup number of possible processor ids */
  319. int nr_cpu_ids __read_mostly = NR_CPUS;
  320. EXPORT_SYMBOL(nr_cpu_ids);
  321. /* An arch may set nr_cpu_ids earlier if needed, so this would be redundant */
  322. static void __init setup_nr_cpu_ids(void)
  323. {
  324. nr_cpu_ids = find_last_bit(cpumask_bits(cpu_possible_mask),NR_CPUS) + 1;
  325. }
  326. #ifndef CONFIG_HAVE_SETUP_PER_CPU_AREA
  327. unsigned long __per_cpu_offset[NR_CPUS] __read_mostly;
  328. EXPORT_SYMBOL(__per_cpu_offset);
  329. static void __init setup_per_cpu_areas(void)
  330. {
  331. unsigned long size, i;
  332. char *ptr;
  333. unsigned long nr_possible_cpus = num_possible_cpus();
  334. /* Copy section for each CPU (we discard the original) */
  335. size = ALIGN(PERCPU_ENOUGH_ROOM, PAGE_SIZE);
  336. ptr = alloc_bootmem_pages(size * nr_possible_cpus);
  337. for_each_possible_cpu(i) {
  338. __per_cpu_offset[i] = ptr - __per_cpu_start;
  339. memcpy(ptr, __per_cpu_start, __per_cpu_end - __per_cpu_start);
  340. ptr += size;
  341. }
  342. }
  343. #endif /* CONFIG_HAVE_SETUP_PER_CPU_AREA */
  344. /* Called by boot processor to activate the rest. */
  345. static void __init smp_init(void)
  346. {
  347. unsigned int cpu;
  348. /*
  349. * Set up the current CPU as possible to migrate to.
  350. * The other ones will be done by cpu_up/cpu_down()
  351. */
  352. set_cpu_active(smp_processor_id(), true);
  353. /* FIXME: This should be done in userspace --RR */
  354. for_each_present_cpu(cpu) {
  355. if (num_online_cpus() >= setup_max_cpus)
  356. break;
  357. if (!cpu_online(cpu))
  358. cpu_up(cpu);
  359. }
  360. /* Any cleanup work */
  361. printk(KERN_INFO "Brought up %ld CPUs\n", (long)num_online_cpus());
  362. smp_cpus_done(setup_max_cpus);
  363. }
  364. #endif
  365. /*
  366. * We need to store the untouched command line for future reference.
  367. * We also need to store the touched command line since the parameter
  368. * parsing is performed in place, and we should allow a component to
  369. * store reference of name/value for future reference.
  370. */
  371. static void __init setup_command_line(char *command_line)
  372. {
  373. saved_command_line = alloc_bootmem(strlen (boot_command_line)+1);
  374. static_command_line = alloc_bootmem(strlen (command_line)+1);
  375. strcpy (saved_command_line, boot_command_line);
  376. strcpy (static_command_line, command_line);
  377. }
  378. /*
  379. * We need to finalize in a non-__init function or else race conditions
  380. * between the root thread and the init thread may cause start_kernel to
  381. * be reaped by free_initmem before the root thread has proceeded to
  382. * cpu_idle.
  383. *
  384. * gcc-3.4 accidentally inlines this function, so use noinline.
  385. */
  386. static noinline void __init_refok rest_init(void)
  387. __releases(kernel_lock)
  388. {
  389. int pid;
  390. kernel_thread(kernel_init, NULL, CLONE_FS | CLONE_SIGHAND);
  391. numa_default_policy();
  392. pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES);
  393. kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns);
  394. unlock_kernel();
  395. /*
  396. * The boot idle thread must execute schedule()
  397. * at least once to get things moving:
  398. */
  399. init_idle_bootup_task(current);
  400. rcu_scheduler_starting();
  401. preempt_enable_no_resched();
  402. schedule();
  403. preempt_disable();
  404. /* Call into cpu_idle with preempt disabled */
  405. cpu_idle();
  406. }
  407. /* Check for early params. */
  408. static int __init do_early_param(char *param, char *val)
  409. {
  410. struct obs_kernel_param *p;
  411. for (p = __setup_start; p < __setup_end; p++) {
  412. if ((p->early && strcmp(param, p->str) == 0) ||
  413. (strcmp(param, "console") == 0 &&
  414. strcmp(p->str, "earlycon") == 0)
  415. ) {
  416. if (p->setup_func(val) != 0)
  417. printk(KERN_WARNING
  418. "Malformed early option '%s'\n", param);
  419. }
  420. }
  421. /* We accept everything at this stage. */
  422. return 0;
  423. }
  424. void __init parse_early_options(char *cmdline)
  425. {
  426. parse_args("early options", cmdline, NULL, 0, do_early_param);
  427. }
  428. /* Arch code calls this early on, or if not, just before other parsing. */
  429. void __init parse_early_param(void)
  430. {
  431. static __initdata int done = 0;
  432. static __initdata char tmp_cmdline[COMMAND_LINE_SIZE];
  433. if (done)
  434. return;
  435. /* All fall through to do_early_param. */
  436. strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE);
  437. parse_early_options(tmp_cmdline);
  438. done = 1;
  439. }
  440. /*
  441. * Activate the first processor.
  442. */
  443. static void __init boot_cpu_init(void)
  444. {
  445. int cpu = smp_processor_id();
  446. /* Mark the boot cpu "present", "online" etc for SMP and UP case */
  447. set_cpu_online(cpu, true);
  448. set_cpu_present(cpu, true);
  449. set_cpu_possible(cpu, true);
  450. }
  451. void __init __weak smp_setup_processor_id(void)
  452. {
  453. }
  454. void __init __weak thread_info_cache_init(void)
  455. {
  456. }
  457. asmlinkage void __init start_kernel(void)
  458. {
  459. char * command_line;
  460. extern 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. early_init_irq_lock_class();
  476. /*
  477. * Interrupts are still disabled. Do necessary setups, then
  478. * enable them
  479. */
  480. lock_kernel();
  481. tick_init();
  482. boot_cpu_init();
  483. page_address_init();
  484. printk(KERN_NOTICE);
  485. printk(linux_banner);
  486. setup_arch(&command_line);
  487. mm_init_owner(&init_mm, &init_task);
  488. setup_command_line(command_line);
  489. setup_per_cpu_areas();
  490. setup_nr_cpu_ids();
  491. smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */
  492. /*
  493. * Set up the scheduler prior starting any interrupts (such as the
  494. * timer interrupt). Full topology setup happens at smp_init()
  495. * time - but meanwhile we still have a functioning scheduler.
  496. */
  497. sched_init();
  498. /*
  499. * Disable preemption - early bootup scheduling is extremely
  500. * fragile until we cpu_idle() for the first time.
  501. */
  502. preempt_disable();
  503. build_all_zonelists();
  504. page_alloc_init();
  505. printk(KERN_NOTICE "Kernel command line: %s\n", boot_command_line);
  506. parse_early_param();
  507. parse_args("Booting kernel", static_command_line, __start___param,
  508. __stop___param - __start___param,
  509. &unknown_bootoption);
  510. if (!irqs_disabled()) {
  511. printk(KERN_WARNING "start_kernel(): bug: interrupts were "
  512. "enabled *very* early, fixing it\n");
  513. local_irq_disable();
  514. }
  515. sort_main_extable();
  516. trap_init();
  517. rcu_init();
  518. /* init some links before init_ISA_irqs() */
  519. early_irq_init();
  520. init_IRQ();
  521. pidhash_init();
  522. init_timers();
  523. hrtimers_init();
  524. softirq_init();
  525. timekeeping_init();
  526. time_init();
  527. sched_clock_init();
  528. profile_init();
  529. if (!irqs_disabled())
  530. printk(KERN_CRIT "start_kernel(): bug: interrupts were "
  531. "enabled early\n");
  532. early_boot_irqs_on();
  533. local_irq_enable();
  534. /*
  535. * HACK ALERT! This is early. We're enabling the console before
  536. * we've done PCI setups etc, and console_init() must be aware of
  537. * this. But we do want output early, in case something goes wrong.
  538. */
  539. console_init();
  540. if (panic_later)
  541. panic(panic_later, panic_param);
  542. lockdep_info();
  543. /*
  544. * Need to run this when irqs are enabled, because it wants
  545. * to self-test [hard/soft]-irqs on/off lock inversion bugs
  546. * too:
  547. */
  548. locking_selftest();
  549. #ifdef CONFIG_BLK_DEV_INITRD
  550. if (initrd_start && !initrd_below_start_ok &&
  551. page_to_pfn(virt_to_page((void *)initrd_start)) < min_low_pfn) {
  552. printk(KERN_CRIT "initrd overwritten (0x%08lx < 0x%08lx) - "
  553. "disabling it.\n",
  554. page_to_pfn(virt_to_page((void *)initrd_start)),
  555. min_low_pfn);
  556. initrd_start = 0;
  557. }
  558. #endif
  559. vmalloc_init();
  560. vfs_caches_init_early();
  561. cpuset_init_early();
  562. page_cgroup_init();
  563. mem_init();
  564. enable_debug_pagealloc();
  565. cpu_hotplug_init();
  566. kmem_cache_init();
  567. kmemtrace_init();
  568. debug_objects_mem_init();
  569. idr_init_cache();
  570. setup_per_cpu_pageset();
  571. numa_policy_init();
  572. if (late_time_init)
  573. late_time_init();
  574. calibrate_delay();
  575. pidmap_init();
  576. pgtable_cache_init();
  577. prio_tree_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(num_physpages);
  586. proc_caches_init();
  587. buffer_init();
  588. key_init();
  589. security_init();
  590. vfs_caches_init(num_physpages);
  591. radix_tree_init();
  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. ftrace_init();
  605. /* Do the rest non-__init'ed, we're now alive */
  606. rest_init();
  607. }
  608. int initcall_debug;
  609. core_param(initcall_debug, initcall_debug, bool, 0644);
  610. int do_one_initcall(initcall_t fn)
  611. {
  612. int count = preempt_count();
  613. ktime_t calltime, delta, rettime;
  614. char msgbuf[64];
  615. struct boot_trace_call call;
  616. struct boot_trace_ret ret;
  617. if (initcall_debug) {
  618. call.caller = task_pid_nr(current);
  619. printk("calling %pF @ %i\n", fn, call.caller);
  620. calltime = ktime_get();
  621. trace_boot_call(&call, fn);
  622. enable_boot_trace();
  623. }
  624. ret.result = fn();
  625. if (initcall_debug) {
  626. disable_boot_trace();
  627. rettime = ktime_get();
  628. delta = ktime_sub(rettime, calltime);
  629. ret.duration = (unsigned long long) ktime_to_ns(delta) >> 10;
  630. trace_boot_ret(&ret, fn);
  631. printk("initcall %pF returned %d after %Ld usecs\n", fn,
  632. ret.result, ret.duration);
  633. }
  634. msgbuf[0] = 0;
  635. if (ret.result && ret.result != -ENODEV && initcall_debug)
  636. sprintf(msgbuf, "error code %d ", ret.result);
  637. if (preempt_count() != count) {
  638. strlcat(msgbuf, "preemption imbalance ", sizeof(msgbuf));
  639. preempt_count() = count;
  640. }
  641. if (irqs_disabled()) {
  642. strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf));
  643. local_irq_enable();
  644. }
  645. if (msgbuf[0]) {
  646. printk("initcall %pF returned with %s\n", fn, msgbuf);
  647. }
  648. return ret.result;
  649. }
  650. extern initcall_t __initcall_start[], __initcall_end[], __early_initcall_end[];
  651. static void __init do_initcalls(void)
  652. {
  653. initcall_t *call;
  654. for (call = __early_initcall_end; call < __initcall_end; call++)
  655. do_one_initcall(*call);
  656. /* Make sure there is no pending stuff from the initcall sequence */
  657. flush_scheduled_work();
  658. }
  659. /*
  660. * Ok, the machine is now initialized. None of the devices
  661. * have been touched yet, but the CPU subsystem is up and
  662. * running, and memory and process management works.
  663. *
  664. * Now we can finally start doing some real work..
  665. */
  666. static void __init do_basic_setup(void)
  667. {
  668. rcu_init_sched(); /* needed by module_init stage. */
  669. init_workqueues();
  670. cpuset_init_smp();
  671. usermodehelper_init();
  672. driver_init();
  673. init_irq_proc();
  674. do_initcalls();
  675. }
  676. static void __init do_pre_smp_initcalls(void)
  677. {
  678. initcall_t *call;
  679. for (call = __initcall_start; call < __early_initcall_end; call++)
  680. do_one_initcall(*call);
  681. }
  682. static void run_init_process(char *init_filename)
  683. {
  684. argv_init[0] = init_filename;
  685. kernel_execve(init_filename, argv_init, envp_init);
  686. }
  687. /* This is a non __init function. Force it to be noinline otherwise gcc
  688. * makes it inline to init() and it becomes part of init.text section
  689. */
  690. static noinline int init_post(void)
  691. __releases(kernel_lock)
  692. {
  693. /* need to finish all async __init code before freeing the memory */
  694. async_synchronize_full();
  695. free_initmem();
  696. unlock_kernel();
  697. mark_rodata_ro();
  698. system_state = SYSTEM_RUNNING;
  699. numa_default_policy();
  700. if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0)
  701. printk(KERN_WARNING "Warning: unable to open an initial console.\n");
  702. (void) sys_dup(0);
  703. (void) sys_dup(0);
  704. current->signal->flags |= SIGNAL_UNKILLABLE;
  705. if (ramdisk_execute_command) {
  706. run_init_process(ramdisk_execute_command);
  707. printk(KERN_WARNING "Failed to execute %s\n",
  708. ramdisk_execute_command);
  709. }
  710. /*
  711. * We try each of these until one succeeds.
  712. *
  713. * The Bourne shell can be used instead of init if we are
  714. * trying to recover a really broken machine.
  715. */
  716. if (execute_command) {
  717. run_init_process(execute_command);
  718. printk(KERN_WARNING "Failed to execute %s. Attempting "
  719. "defaults...\n", execute_command);
  720. }
  721. run_init_process("/sbin/init");
  722. run_init_process("/etc/init");
  723. run_init_process("/bin/init");
  724. run_init_process("/bin/sh");
  725. panic("No init found. Try passing init= option to kernel.");
  726. }
  727. static int __init kernel_init(void * unused)
  728. {
  729. lock_kernel();
  730. /*
  731. * init can run on any cpu.
  732. */
  733. set_cpus_allowed_ptr(current, cpu_all_mask);
  734. /*
  735. * Tell the world that we're going to be the grim
  736. * reaper of innocent orphaned children.
  737. *
  738. * We don't want people to have to make incorrect
  739. * assumptions about where in the task array this
  740. * can be found.
  741. */
  742. init_pid_ns.child_reaper = current;
  743. cad_pid = task_pid(current);
  744. smp_prepare_cpus(setup_max_cpus);
  745. do_pre_smp_initcalls();
  746. start_boot_trace();
  747. smp_init();
  748. sched_init_smp();
  749. do_basic_setup();
  750. /*
  751. * check if there is an early userspace init. If yes, let it do all
  752. * the work
  753. */
  754. if (!ramdisk_execute_command)
  755. ramdisk_execute_command = "/init";
  756. if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) {
  757. ramdisk_execute_command = NULL;
  758. prepare_namespace();
  759. }
  760. /*
  761. * Ok, we have completed the initial bootup, and
  762. * we're essentially up and running. Get rid of the
  763. * initmem segments and start the user-mode stuff..
  764. */
  765. init_post();
  766. return 0;
  767. }