main.c 21 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890
  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. #ifdef CONFIG_X86_LOCAL_APIC
  73. #include <asm/smp.h>
  74. #endif
  75. static int kernel_init(void *);
  76. extern void init_IRQ(void);
  77. extern void fork_init(unsigned long);
  78. extern void mca_init(void);
  79. extern void sbus_init(void);
  80. extern void prio_tree_init(void);
  81. extern void radix_tree_init(void);
  82. extern void free_initmem(void);
  83. #ifdef CONFIG_ACPI
  84. extern void acpi_early_init(void);
  85. #else
  86. static inline void acpi_early_init(void) { }
  87. #endif
  88. #ifndef CONFIG_DEBUG_RODATA
  89. static inline void mark_rodata_ro(void) { }
  90. #endif
  91. #ifdef CONFIG_TC
  92. extern void tc_init(void);
  93. #endif
  94. enum system_states system_state;
  95. EXPORT_SYMBOL(system_state);
  96. /*
  97. * Boot command-line arguments
  98. */
  99. #define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT
  100. #define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT
  101. extern void time_init(void);
  102. /* Default late time init is NULL. archs can override this later. */
  103. void (*__initdata late_time_init)(void);
  104. extern void softirq_init(void);
  105. /* Untouched command line saved by arch-specific code. */
  106. char __initdata boot_command_line[COMMAND_LINE_SIZE];
  107. /* Untouched saved command line (eg. for /proc) */
  108. char *saved_command_line;
  109. /* Command line for parameter parsing */
  110. static char *static_command_line;
  111. static char *execute_command;
  112. static char *ramdisk_execute_command;
  113. #ifdef CONFIG_SMP
  114. /* Setup configured maximum number of CPUs to activate */
  115. unsigned int __initdata setup_max_cpus = NR_CPUS;
  116. /*
  117. * Setup routine for controlling SMP activation
  118. *
  119. * Command-line option of "nosmp" or "maxcpus=0" will disable SMP
  120. * activation entirely (the MPS table probe still happens, though).
  121. *
  122. * Command-line option of "maxcpus=<NUM>", where <NUM> is an integer
  123. * greater than 0, limits the maximum number of CPUs activated in
  124. * SMP mode to <NUM>.
  125. */
  126. #ifndef CONFIG_X86_IO_APIC
  127. static inline void disable_ioapic_setup(void) {};
  128. #endif
  129. static int __init nosmp(char *str)
  130. {
  131. setup_max_cpus = 0;
  132. disable_ioapic_setup();
  133. return 0;
  134. }
  135. early_param("nosmp", nosmp);
  136. static int __init maxcpus(char *str)
  137. {
  138. get_option(&str, &setup_max_cpus);
  139. if (setup_max_cpus == 0)
  140. disable_ioapic_setup();
  141. return 0;
  142. }
  143. early_param("maxcpus", maxcpus);
  144. #else
  145. #define setup_max_cpus NR_CPUS
  146. #endif
  147. /*
  148. * If set, this is an indication to the drivers that reset the underlying
  149. * device before going ahead with the initialization otherwise driver might
  150. * rely on the BIOS and skip the reset operation.
  151. *
  152. * This is useful if kernel is booting in an unreliable environment.
  153. * For ex. kdump situaiton where previous kernel has crashed, BIOS has been
  154. * skipped and devices will be in unknown state.
  155. */
  156. unsigned int reset_devices;
  157. EXPORT_SYMBOL(reset_devices);
  158. static int __init set_reset_devices(char *str)
  159. {
  160. reset_devices = 1;
  161. return 1;
  162. }
  163. __setup("reset_devices", set_reset_devices);
  164. static char * argv_init[MAX_INIT_ARGS+2] = { "init", NULL, };
  165. char * envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, };
  166. static const char *panic_later, *panic_param;
  167. extern struct obs_kernel_param __setup_start[], __setup_end[];
  168. static int __init obsolete_checksetup(char *line)
  169. {
  170. struct obs_kernel_param *p;
  171. int had_early_param = 0;
  172. p = __setup_start;
  173. do {
  174. int n = strlen(p->str);
  175. if (!strncmp(line, p->str, n)) {
  176. if (p->early) {
  177. /* Already done in parse_early_param?
  178. * (Needs exact match on param part).
  179. * Keep iterating, as we can have early
  180. * params and __setups of same names 8( */
  181. if (line[n] == '\0' || line[n] == '=')
  182. had_early_param = 1;
  183. } else if (!p->setup_func) {
  184. printk(KERN_WARNING "Parameter %s is obsolete,"
  185. " ignored\n", p->str);
  186. return 1;
  187. } else if (p->setup_func(line + n))
  188. return 1;
  189. }
  190. p++;
  191. } while (p < __setup_end);
  192. return had_early_param;
  193. }
  194. /*
  195. * This should be approx 2 Bo*oMips to start (note initial shift), and will
  196. * still work even if initially too large, it will just take slightly longer
  197. */
  198. unsigned long loops_per_jiffy = (1<<12);
  199. EXPORT_SYMBOL(loops_per_jiffy);
  200. static int __init debug_kernel(char *str)
  201. {
  202. console_loglevel = 10;
  203. return 0;
  204. }
  205. static int __init quiet_kernel(char *str)
  206. {
  207. console_loglevel = 4;
  208. return 0;
  209. }
  210. early_param("debug", debug_kernel);
  211. early_param("quiet", quiet_kernel);
  212. static int __init loglevel(char *str)
  213. {
  214. get_option(&str, &console_loglevel);
  215. return 0;
  216. }
  217. early_param("loglevel", loglevel);
  218. /*
  219. * Unknown boot options get handed to init, unless they look like
  220. * failed parameters
  221. */
  222. static int __init unknown_bootoption(char *param, char *val)
  223. {
  224. /* Change NUL term back to "=", to make "param" the whole string. */
  225. if (val) {
  226. /* param=val or param="val"? */
  227. if (val == param+strlen(param)+1)
  228. val[-1] = '=';
  229. else if (val == param+strlen(param)+2) {
  230. val[-2] = '=';
  231. memmove(val-1, val, strlen(val)+1);
  232. val--;
  233. } else
  234. BUG();
  235. }
  236. /* Handle obsolete-style parameters */
  237. if (obsolete_checksetup(param))
  238. return 0;
  239. /*
  240. * Preemptive maintenance for "why didn't my misspelled command
  241. * line work?"
  242. */
  243. if (strchr(param, '.') && (!val || strchr(param, '.') < val)) {
  244. printk(KERN_ERR "Unknown boot option `%s': ignoring\n", param);
  245. return 0;
  246. }
  247. if (panic_later)
  248. return 0;
  249. if (val) {
  250. /* Environment option */
  251. unsigned int i;
  252. for (i = 0; envp_init[i]; i++) {
  253. if (i == MAX_INIT_ENVS) {
  254. panic_later = "Too many boot env vars at `%s'";
  255. panic_param = param;
  256. }
  257. if (!strncmp(param, envp_init[i], val - param))
  258. break;
  259. }
  260. envp_init[i] = param;
  261. } else {
  262. /* Command line option */
  263. unsigned int i;
  264. for (i = 0; argv_init[i]; i++) {
  265. if (i == MAX_INIT_ARGS) {
  266. panic_later = "Too many boot init vars at `%s'";
  267. panic_param = param;
  268. }
  269. }
  270. argv_init[i] = param;
  271. }
  272. return 0;
  273. }
  274. #ifdef CONFIG_DEBUG_PAGEALLOC
  275. int __read_mostly debug_pagealloc_enabled = 0;
  276. #endif
  277. static int __init init_setup(char *str)
  278. {
  279. unsigned int i;
  280. execute_command = str;
  281. /*
  282. * In case LILO is going to boot us with default command line,
  283. * it prepends "auto" before the whole cmdline which makes
  284. * the shell think it should execute a script with such name.
  285. * So we ignore all arguments entered _before_ init=... [MJ]
  286. */
  287. for (i = 1; i < MAX_INIT_ARGS; i++)
  288. argv_init[i] = NULL;
  289. return 1;
  290. }
  291. __setup("init=", init_setup);
  292. static int __init rdinit_setup(char *str)
  293. {
  294. unsigned int i;
  295. ramdisk_execute_command = str;
  296. /* See "auto" comment in init_setup */
  297. for (i = 1; i < MAX_INIT_ARGS; i++)
  298. argv_init[i] = NULL;
  299. return 1;
  300. }
  301. __setup("rdinit=", rdinit_setup);
  302. #ifndef CONFIG_SMP
  303. #ifdef CONFIG_X86_LOCAL_APIC
  304. static void __init smp_init(void)
  305. {
  306. APIC_init_uniprocessor();
  307. }
  308. #else
  309. #define smp_init() do { } while (0)
  310. #endif
  311. static inline void setup_per_cpu_areas(void) { }
  312. static inline void setup_nr_cpu_ids(void) { }
  313. static inline void smp_prepare_cpus(unsigned int maxcpus) { }
  314. #else
  315. #if NR_CPUS > BITS_PER_LONG
  316. cpumask_t cpu_mask_all __read_mostly = CPU_MASK_ALL;
  317. EXPORT_SYMBOL(cpu_mask_all);
  318. #endif
  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. #ifndef CONFIG_HAVE_SETUP_PER_CPU_AREA
  328. unsigned long __per_cpu_offset[NR_CPUS] __read_mostly;
  329. EXPORT_SYMBOL(__per_cpu_offset);
  330. static void __init setup_per_cpu_areas(void)
  331. {
  332. unsigned long size, i;
  333. char *ptr;
  334. unsigned long nr_possible_cpus = num_possible_cpus();
  335. /* Copy section for each CPU (we discard the original) */
  336. size = ALIGN(PERCPU_ENOUGH_ROOM, PAGE_SIZE);
  337. ptr = alloc_bootmem_pages(size * nr_possible_cpus);
  338. for_each_possible_cpu(i) {
  339. __per_cpu_offset[i] = ptr - __per_cpu_start;
  340. memcpy(ptr, __per_cpu_start, __per_cpu_end - __per_cpu_start);
  341. ptr += size;
  342. }
  343. }
  344. #endif /* CONFIG_HAVE_SETUP_PER_CPU_AREA */
  345. /* Called by boot processor to activate the rest. */
  346. static void __init smp_init(void)
  347. {
  348. unsigned int cpu;
  349. /*
  350. * Set up the current CPU as possible to migrate to.
  351. * The other ones will be done by cpu_up/cpu_down()
  352. */
  353. cpu = smp_processor_id();
  354. cpu_set(cpu, cpu_active_map);
  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. preempt_enable_no_resched();
  403. schedule();
  404. preempt_disable();
  405. /* Call into cpu_idle with preempt disabled */
  406. cpu_idle();
  407. }
  408. /* Check for early params. */
  409. static int __init do_early_param(char *param, char *val)
  410. {
  411. struct obs_kernel_param *p;
  412. for (p = __setup_start; p < __setup_end; p++) {
  413. if ((p->early && strcmp(param, p->str) == 0) ||
  414. (strcmp(param, "console") == 0 &&
  415. strcmp(p->str, "earlycon") == 0)
  416. ) {
  417. if (p->setup_func(val) != 0)
  418. printk(KERN_WARNING
  419. "Malformed early option '%s'\n", param);
  420. }
  421. }
  422. /* We accept everything at this stage. */
  423. return 0;
  424. }
  425. /* Arch code calls this early on, or if not, just before other parsing. */
  426. void __init parse_early_param(void)
  427. {
  428. static __initdata int done = 0;
  429. static __initdata char tmp_cmdline[COMMAND_LINE_SIZE];
  430. if (done)
  431. return;
  432. /* All fall through to do_early_param. */
  433. strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE);
  434. parse_args("early options", tmp_cmdline, NULL, 0, do_early_param);
  435. done = 1;
  436. }
  437. /*
  438. * Activate the first processor.
  439. */
  440. static void __init boot_cpu_init(void)
  441. {
  442. int cpu = smp_processor_id();
  443. /* Mark the boot cpu "present", "online" etc for SMP and UP case */
  444. set_cpu_online(cpu, true);
  445. set_cpu_present(cpu, true);
  446. set_cpu_possible(cpu, true);
  447. }
  448. void __init __weak smp_setup_processor_id(void)
  449. {
  450. }
  451. void __init __weak thread_info_cache_init(void)
  452. {
  453. }
  454. asmlinkage void __init start_kernel(void)
  455. {
  456. char * command_line;
  457. extern struct kernel_param __start___param[], __stop___param[];
  458. smp_setup_processor_id();
  459. /*
  460. * Need to run as early as possible, to initialize the
  461. * lockdep hash:
  462. */
  463. lockdep_init();
  464. debug_objects_early_init();
  465. /*
  466. * Set up the the initial canary ASAP:
  467. */
  468. boot_init_stack_canary();
  469. cgroup_init_early();
  470. local_irq_disable();
  471. early_boot_irqs_off();
  472. early_init_irq_lock_class();
  473. /*
  474. * Interrupts are still disabled. Do necessary setups, then
  475. * enable them
  476. */
  477. lock_kernel();
  478. tick_init();
  479. boot_cpu_init();
  480. page_address_init();
  481. printk(KERN_NOTICE);
  482. printk(linux_banner);
  483. setup_arch(&command_line);
  484. mm_init_owner(&init_mm, &init_task);
  485. setup_command_line(command_line);
  486. setup_per_cpu_areas();
  487. setup_nr_cpu_ids();
  488. smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */
  489. /*
  490. * Set up the scheduler prior starting any interrupts (such as the
  491. * timer interrupt). Full topology setup happens at smp_init()
  492. * time - but meanwhile we still have a functioning scheduler.
  493. */
  494. sched_init();
  495. /*
  496. * Disable preemption - early bootup scheduling is extremely
  497. * fragile until we cpu_idle() for the first time.
  498. */
  499. preempt_disable();
  500. build_all_zonelists();
  501. page_alloc_init();
  502. printk(KERN_NOTICE "Kernel command line: %s\n", boot_command_line);
  503. parse_early_param();
  504. parse_args("Booting kernel", static_command_line, __start___param,
  505. __stop___param - __start___param,
  506. &unknown_bootoption);
  507. if (!irqs_disabled()) {
  508. printk(KERN_WARNING "start_kernel(): bug: interrupts were "
  509. "enabled *very* early, fixing it\n");
  510. local_irq_disable();
  511. }
  512. sort_main_extable();
  513. trap_init();
  514. rcu_init();
  515. /* init some links before init_ISA_irqs() */
  516. early_irq_init();
  517. init_IRQ();
  518. pidhash_init();
  519. init_timers();
  520. hrtimers_init();
  521. softirq_init();
  522. timekeeping_init();
  523. time_init();
  524. sched_clock_init();
  525. profile_init();
  526. if (!irqs_disabled())
  527. printk(KERN_CRIT "start_kernel(): bug: interrupts were "
  528. "enabled early\n");
  529. early_boot_irqs_on();
  530. local_irq_enable();
  531. /*
  532. * HACK ALERT! This is early. We're enabling the console before
  533. * we've done PCI setups etc, and console_init() must be aware of
  534. * this. But we do want output early, in case something goes wrong.
  535. */
  536. console_init();
  537. if (panic_later)
  538. panic(panic_later, panic_param);
  539. lockdep_info();
  540. /*
  541. * Need to run this when irqs are enabled, because it wants
  542. * to self-test [hard/soft]-irqs on/off lock inversion bugs
  543. * too:
  544. */
  545. locking_selftest();
  546. #ifdef CONFIG_BLK_DEV_INITRD
  547. if (initrd_start && !initrd_below_start_ok &&
  548. page_to_pfn(virt_to_page((void *)initrd_start)) < min_low_pfn) {
  549. printk(KERN_CRIT "initrd overwritten (0x%08lx < 0x%08lx) - "
  550. "disabling it.\n",
  551. page_to_pfn(virt_to_page((void *)initrd_start)),
  552. min_low_pfn);
  553. initrd_start = 0;
  554. }
  555. #endif
  556. vmalloc_init();
  557. vfs_caches_init_early();
  558. cpuset_init_early();
  559. page_cgroup_init();
  560. mem_init();
  561. enable_debug_pagealloc();
  562. cpu_hotplug_init();
  563. kmem_cache_init();
  564. debug_objects_mem_init();
  565. idr_init_cache();
  566. setup_per_cpu_pageset();
  567. numa_policy_init();
  568. if (late_time_init)
  569. late_time_init();
  570. calibrate_delay();
  571. pidmap_init();
  572. pgtable_cache_init();
  573. prio_tree_init();
  574. anon_vma_init();
  575. #ifdef CONFIG_X86
  576. if (efi_enabled)
  577. efi_enter_virtual_mode();
  578. #endif
  579. thread_info_cache_init();
  580. cred_init();
  581. fork_init(num_physpages);
  582. proc_caches_init();
  583. buffer_init();
  584. key_init();
  585. security_init();
  586. vfs_caches_init(num_physpages);
  587. radix_tree_init();
  588. signals_init();
  589. /* rootfs populating might need page-writeback */
  590. page_writeback_init();
  591. #ifdef CONFIG_PROC_FS
  592. proc_root_init();
  593. #endif
  594. cgroup_init();
  595. cpuset_init();
  596. taskstats_init_early();
  597. delayacct_init();
  598. check_bugs();
  599. acpi_early_init(); /* before LAPIC and SMP init */
  600. ftrace_init();
  601. /* Do the rest non-__init'ed, we're now alive */
  602. rest_init();
  603. }
  604. int initcall_debug;
  605. core_param(initcall_debug, initcall_debug, bool, 0644);
  606. int do_one_initcall(initcall_t fn)
  607. {
  608. int count = preempt_count();
  609. ktime_t calltime, delta, rettime;
  610. char msgbuf[64];
  611. struct boot_trace_call call;
  612. struct boot_trace_ret ret;
  613. if (initcall_debug) {
  614. call.caller = task_pid_nr(current);
  615. printk("calling %pF @ %i\n", fn, call.caller);
  616. calltime = ktime_get();
  617. trace_boot_call(&call, fn);
  618. enable_boot_trace();
  619. }
  620. ret.result = fn();
  621. if (initcall_debug) {
  622. disable_boot_trace();
  623. rettime = ktime_get();
  624. delta = ktime_sub(rettime, calltime);
  625. ret.duration = (unsigned long long) ktime_to_ns(delta) >> 10;
  626. trace_boot_ret(&ret, fn);
  627. printk("initcall %pF returned %d after %Ld usecs\n", fn,
  628. ret.result, ret.duration);
  629. }
  630. msgbuf[0] = 0;
  631. if (ret.result && ret.result != -ENODEV && initcall_debug)
  632. sprintf(msgbuf, "error code %d ", ret.result);
  633. if (preempt_count() != count) {
  634. strlcat(msgbuf, "preemption imbalance ", sizeof(msgbuf));
  635. preempt_count() = count;
  636. }
  637. if (irqs_disabled()) {
  638. strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf));
  639. local_irq_enable();
  640. }
  641. if (msgbuf[0]) {
  642. printk("initcall %pF returned with %s\n", fn, msgbuf);
  643. }
  644. return ret.result;
  645. }
  646. extern initcall_t __initcall_start[], __initcall_end[], __early_initcall_end[];
  647. static void __init do_initcalls(void)
  648. {
  649. initcall_t *call;
  650. for (call = __early_initcall_end; call < __initcall_end; call++)
  651. do_one_initcall(*call);
  652. /* Make sure there is no pending stuff from the initcall sequence */
  653. flush_scheduled_work();
  654. }
  655. /*
  656. * Ok, the machine is now initialized. None of the devices
  657. * have been touched yet, but the CPU subsystem is up and
  658. * running, and memory and process management works.
  659. *
  660. * Now we can finally start doing some real work..
  661. */
  662. static void __init do_basic_setup(void)
  663. {
  664. rcu_init_sched(); /* needed by module_init stage. */
  665. init_workqueues();
  666. usermodehelper_init();
  667. driver_init();
  668. init_irq_proc();
  669. do_initcalls();
  670. }
  671. static void __init do_pre_smp_initcalls(void)
  672. {
  673. initcall_t *call;
  674. for (call = __initcall_start; call < __early_initcall_end; call++)
  675. do_one_initcall(*call);
  676. }
  677. static void run_init_process(char *init_filename)
  678. {
  679. argv_init[0] = init_filename;
  680. kernel_execve(init_filename, argv_init, envp_init);
  681. }
  682. /* This is a non __init function. Force it to be noinline otherwise gcc
  683. * makes it inline to init() and it becomes part of init.text section
  684. */
  685. static noinline int init_post(void)
  686. {
  687. /* need to finish all async __init code before freeing the memory */
  688. async_synchronize_full();
  689. free_initmem();
  690. unlock_kernel();
  691. mark_rodata_ro();
  692. system_state = SYSTEM_RUNNING;
  693. numa_default_policy();
  694. if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0)
  695. printk(KERN_WARNING "Warning: unable to open an initial console.\n");
  696. (void) sys_dup(0);
  697. (void) sys_dup(0);
  698. current->signal->flags |= SIGNAL_UNKILLABLE;
  699. if (ramdisk_execute_command) {
  700. run_init_process(ramdisk_execute_command);
  701. printk(KERN_WARNING "Failed to execute %s\n",
  702. ramdisk_execute_command);
  703. }
  704. /*
  705. * We try each of these until one succeeds.
  706. *
  707. * The Bourne shell can be used instead of init if we are
  708. * trying to recover a really broken machine.
  709. */
  710. if (execute_command) {
  711. run_init_process(execute_command);
  712. printk(KERN_WARNING "Failed to execute %s. Attempting "
  713. "defaults...\n", execute_command);
  714. }
  715. run_init_process("/sbin/init");
  716. run_init_process("/etc/init");
  717. run_init_process("/bin/init");
  718. run_init_process("/bin/sh");
  719. panic("No init found. Try passing init= option to kernel.");
  720. }
  721. static int __init kernel_init(void * unused)
  722. {
  723. lock_kernel();
  724. /*
  725. * init can run on any cpu.
  726. */
  727. set_cpus_allowed_ptr(current, CPU_MASK_ALL_PTR);
  728. /*
  729. * Tell the world that we're going to be the grim
  730. * reaper of innocent orphaned children.
  731. *
  732. * We don't want people to have to make incorrect
  733. * assumptions about where in the task array this
  734. * can be found.
  735. */
  736. init_pid_ns.child_reaper = current;
  737. cad_pid = task_pid(current);
  738. smp_prepare_cpus(setup_max_cpus);
  739. do_pre_smp_initcalls();
  740. start_boot_trace();
  741. smp_init();
  742. sched_init_smp();
  743. cpuset_init_smp();
  744. do_basic_setup();
  745. /*
  746. * check if there is an early userspace init. If yes, let it do all
  747. * the work
  748. */
  749. if (!ramdisk_execute_command)
  750. ramdisk_execute_command = "/init";
  751. if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) {
  752. ramdisk_execute_command = NULL;
  753. prepare_namespace();
  754. }
  755. /*
  756. * Ok, we have completed the initial bootup, and
  757. * we're essentially up and running. Get rid of the
  758. * initmem segments and start the user-mode stuff..
  759. */
  760. init_post();
  761. return 0;
  762. }