main.c 21 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891
  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 __read_mostly;
  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. void __weak arch_disable_smp_support(void) { }
  127. static int __init nosmp(char *str)
  128. {
  129. setup_max_cpus = 0;
  130. arch_disable_smp_support();
  131. return 0;
  132. }
  133. early_param("nosmp", nosmp);
  134. static int __init maxcpus(char *str)
  135. {
  136. get_option(&str, &setup_max_cpus);
  137. if (setup_max_cpus == 0)
  138. arch_disable_smp_support();
  139. return 0;
  140. }
  141. early_param("maxcpus", maxcpus);
  142. #else
  143. const unsigned int setup_max_cpus = NR_CPUS;
  144. #endif
  145. /*
  146. * If set, this is an indication to the drivers that reset the underlying
  147. * device before going ahead with the initialization otherwise driver might
  148. * rely on the BIOS and skip the reset operation.
  149. *
  150. * This is useful if kernel is booting in an unreliable environment.
  151. * For ex. kdump situaiton where previous kernel has crashed, BIOS has been
  152. * skipped and devices will be in unknown state.
  153. */
  154. unsigned int reset_devices;
  155. EXPORT_SYMBOL(reset_devices);
  156. static int __init set_reset_devices(char *str)
  157. {
  158. reset_devices = 1;
  159. return 1;
  160. }
  161. __setup("reset_devices", set_reset_devices);
  162. static char * argv_init[MAX_INIT_ARGS+2] = { "init", NULL, };
  163. char * envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, };
  164. static const char *panic_later, *panic_param;
  165. extern struct obs_kernel_param __setup_start[], __setup_end[];
  166. static int __init obsolete_checksetup(char *line)
  167. {
  168. struct obs_kernel_param *p;
  169. int had_early_param = 0;
  170. p = __setup_start;
  171. do {
  172. int n = strlen(p->str);
  173. if (!strncmp(line, p->str, n)) {
  174. if (p->early) {
  175. /* Already done in parse_early_param?
  176. * (Needs exact match on param part).
  177. * Keep iterating, as we can have early
  178. * params and __setups of same names 8( */
  179. if (line[n] == '\0' || line[n] == '=')
  180. had_early_param = 1;
  181. } else if (!p->setup_func) {
  182. printk(KERN_WARNING "Parameter %s is obsolete,"
  183. " ignored\n", p->str);
  184. return 1;
  185. } else if (p->setup_func(line + n))
  186. return 1;
  187. }
  188. p++;
  189. } while (p < __setup_end);
  190. return had_early_param;
  191. }
  192. /*
  193. * This should be approx 2 Bo*oMips to start (note initial shift), and will
  194. * still work even if initially too large, it will just take slightly longer
  195. */
  196. unsigned long loops_per_jiffy = (1<<12);
  197. EXPORT_SYMBOL(loops_per_jiffy);
  198. static int __init debug_kernel(char *str)
  199. {
  200. console_loglevel = 10;
  201. return 0;
  202. }
  203. static int __init quiet_kernel(char *str)
  204. {
  205. console_loglevel = 4;
  206. return 0;
  207. }
  208. early_param("debug", debug_kernel);
  209. early_param("quiet", quiet_kernel);
  210. static int __init loglevel(char *str)
  211. {
  212. get_option(&str, &console_loglevel);
  213. return 0;
  214. }
  215. early_param("loglevel", loglevel);
  216. /*
  217. * Unknown boot options get handed to init, unless they look like
  218. * failed parameters
  219. */
  220. static int __init unknown_bootoption(char *param, char *val)
  221. {
  222. /* Change NUL term back to "=", to make "param" the whole string. */
  223. if (val) {
  224. /* param=val or param="val"? */
  225. if (val == param+strlen(param)+1)
  226. val[-1] = '=';
  227. else if (val == param+strlen(param)+2) {
  228. val[-2] = '=';
  229. memmove(val-1, val, strlen(val)+1);
  230. val--;
  231. } else
  232. BUG();
  233. }
  234. /* Handle obsolete-style parameters */
  235. if (obsolete_checksetup(param))
  236. return 0;
  237. /*
  238. * Preemptive maintenance for "why didn't my misspelled command
  239. * line work?"
  240. */
  241. if (strchr(param, '.') && (!val || strchr(param, '.') < val)) {
  242. printk(KERN_ERR "Unknown boot option `%s': ignoring\n", param);
  243. return 0;
  244. }
  245. if (panic_later)
  246. return 0;
  247. if (val) {
  248. /* Environment option */
  249. unsigned int i;
  250. for (i = 0; envp_init[i]; i++) {
  251. if (i == MAX_INIT_ENVS) {
  252. panic_later = "Too many boot env vars at `%s'";
  253. panic_param = param;
  254. }
  255. if (!strncmp(param, envp_init[i], val - param))
  256. break;
  257. }
  258. envp_init[i] = param;
  259. } else {
  260. /* Command line option */
  261. unsigned int i;
  262. for (i = 0; argv_init[i]; i++) {
  263. if (i == MAX_INIT_ARGS) {
  264. panic_later = "Too many boot init vars at `%s'";
  265. panic_param = param;
  266. }
  267. }
  268. argv_init[i] = param;
  269. }
  270. return 0;
  271. }
  272. #ifdef CONFIG_DEBUG_PAGEALLOC
  273. int __read_mostly debug_pagealloc_enabled = 0;
  274. #endif
  275. static int __init init_setup(char *str)
  276. {
  277. unsigned int i;
  278. execute_command = str;
  279. /*
  280. * In case LILO is going to boot us with default command line,
  281. * it prepends "auto" before the whole cmdline which makes
  282. * the shell think it should execute a script with such name.
  283. * So we ignore all arguments entered _before_ init=... [MJ]
  284. */
  285. for (i = 1; i < MAX_INIT_ARGS; i++)
  286. argv_init[i] = NULL;
  287. return 1;
  288. }
  289. __setup("init=", init_setup);
  290. static int __init rdinit_setup(char *str)
  291. {
  292. unsigned int i;
  293. ramdisk_execute_command = str;
  294. /* See "auto" comment in init_setup */
  295. for (i = 1; i < MAX_INIT_ARGS; i++)
  296. argv_init[i] = NULL;
  297. return 1;
  298. }
  299. __setup("rdinit=", rdinit_setup);
  300. #ifndef CONFIG_SMP
  301. #ifdef CONFIG_X86_LOCAL_APIC
  302. static void __init smp_init(void)
  303. {
  304. APIC_init_uniprocessor();
  305. }
  306. #else
  307. #define smp_init() do { } while (0)
  308. #endif
  309. static inline void setup_per_cpu_areas(void) { }
  310. static inline void setup_nr_cpu_ids(void) { }
  311. static inline void smp_prepare_cpus(unsigned int maxcpus) { }
  312. #else
  313. #if NR_CPUS > BITS_PER_LONG
  314. cpumask_t cpu_mask_all __read_mostly = CPU_MASK_ALL;
  315. EXPORT_SYMBOL(cpu_mask_all);
  316. #endif
  317. /* Setup number of possible processor ids */
  318. int nr_cpu_ids __read_mostly = NR_CPUS;
  319. EXPORT_SYMBOL(nr_cpu_ids);
  320. /* An arch may set nr_cpu_ids earlier if needed, so this would be redundant */
  321. static void __init setup_nr_cpu_ids(void)
  322. {
  323. nr_cpu_ids = find_last_bit(cpumask_bits(cpu_possible_mask),NR_CPUS) + 1;
  324. }
  325. #ifndef CONFIG_HAVE_SETUP_PER_CPU_AREA
  326. unsigned long __per_cpu_offset[NR_CPUS] __read_mostly;
  327. EXPORT_SYMBOL(__per_cpu_offset);
  328. static void __init setup_per_cpu_areas(void)
  329. {
  330. unsigned long size, i;
  331. char *ptr;
  332. unsigned long nr_possible_cpus = num_possible_cpus();
  333. /* Copy section for each CPU (we discard the original) */
  334. size = ALIGN(PERCPU_ENOUGH_ROOM, PAGE_SIZE);
  335. ptr = alloc_bootmem_pages(size * nr_possible_cpus);
  336. for_each_possible_cpu(i) {
  337. __per_cpu_offset[i] = ptr - __per_cpu_start;
  338. memcpy(ptr, __per_cpu_start, __per_cpu_end - __per_cpu_start);
  339. ptr += size;
  340. }
  341. }
  342. #endif /* CONFIG_HAVE_SETUP_PER_CPU_AREA */
  343. /* Called by boot processor to activate the rest. */
  344. static void __init smp_init(void)
  345. {
  346. unsigned int cpu;
  347. /*
  348. * Set up the current CPU as possible to migrate to.
  349. * The other ones will be done by cpu_up/cpu_down()
  350. */
  351. cpu = smp_processor_id();
  352. cpu_set(cpu, cpu_active_map);
  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. /* Arch code calls this early on, or if not, just before other parsing. */
  425. void __init parse_early_param(void)
  426. {
  427. static __initdata int done = 0;
  428. static __initdata char tmp_cmdline[COMMAND_LINE_SIZE];
  429. if (done)
  430. return;
  431. /* All fall through to do_early_param. */
  432. strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE);
  433. parse_args("early options", tmp_cmdline, NULL, 0, do_early_param);
  434. done = 1;
  435. }
  436. /*
  437. * Activate the first processor.
  438. */
  439. static void __init boot_cpu_init(void)
  440. {
  441. int cpu = smp_processor_id();
  442. /* Mark the boot cpu "present", "online" etc for SMP and UP case */
  443. set_cpu_online(cpu, true);
  444. set_cpu_present(cpu, true);
  445. set_cpu_possible(cpu, true);
  446. }
  447. void __init __weak smp_setup_processor_id(void)
  448. {
  449. }
  450. void __init __weak thread_info_cache_init(void)
  451. {
  452. }
  453. asmlinkage void __init start_kernel(void)
  454. {
  455. char * command_line;
  456. extern struct kernel_param __start___param[], __stop___param[];
  457. smp_setup_processor_id();
  458. /*
  459. * Need to run as early as possible, to initialize the
  460. * lockdep hash:
  461. */
  462. lockdep_init();
  463. debug_objects_early_init();
  464. /*
  465. * Set up the the initial canary ASAP:
  466. */
  467. boot_init_stack_canary();
  468. cgroup_init_early();
  469. local_irq_disable();
  470. early_boot_irqs_off();
  471. early_init_irq_lock_class();
  472. /*
  473. * Interrupts are still disabled. Do necessary setups, then
  474. * enable them
  475. */
  476. lock_kernel();
  477. tick_init();
  478. boot_cpu_init();
  479. page_address_init();
  480. printk(KERN_NOTICE);
  481. printk(linux_banner);
  482. setup_arch(&command_line);
  483. mm_init_owner(&init_mm, &init_task);
  484. setup_command_line(command_line);
  485. setup_per_cpu_areas();
  486. setup_nr_cpu_ids();
  487. smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */
  488. /*
  489. * Set up the scheduler prior starting any interrupts (such as the
  490. * timer interrupt). Full topology setup happens at smp_init()
  491. * time - but meanwhile we still have a functioning scheduler.
  492. */
  493. sched_init();
  494. /*
  495. * Disable preemption - early bootup scheduling is extremely
  496. * fragile until we cpu_idle() for the first time.
  497. */
  498. preempt_disable();
  499. build_all_zonelists();
  500. page_alloc_init();
  501. printk(KERN_NOTICE "Kernel command line: %s\n", boot_command_line);
  502. parse_early_param();
  503. parse_args("Booting kernel", static_command_line, __start___param,
  504. __stop___param - __start___param,
  505. &unknown_bootoption);
  506. if (!irqs_disabled()) {
  507. printk(KERN_WARNING "start_kernel(): bug: interrupts were "
  508. "enabled *very* early, fixing it\n");
  509. local_irq_disable();
  510. }
  511. sort_main_extable();
  512. trap_init();
  513. rcu_init();
  514. /* init some links before init_ISA_irqs() */
  515. early_irq_init();
  516. init_IRQ();
  517. pidhash_init();
  518. init_timers();
  519. hrtimers_init();
  520. softirq_init();
  521. timekeeping_init();
  522. time_init();
  523. sched_clock_init();
  524. profile_init();
  525. if (!irqs_disabled())
  526. printk(KERN_CRIT "start_kernel(): bug: interrupts were "
  527. "enabled early\n");
  528. early_boot_irqs_on();
  529. local_irq_enable();
  530. /*
  531. * HACK ALERT! This is early. We're enabling the console before
  532. * we've done PCI setups etc, and console_init() must be aware of
  533. * this. But we do want output early, in case something goes wrong.
  534. */
  535. console_init();
  536. if (panic_later)
  537. panic(panic_later, panic_param);
  538. lockdep_info();
  539. /*
  540. * Need to run this when irqs are enabled, because it wants
  541. * to self-test [hard/soft]-irqs on/off lock inversion bugs
  542. * too:
  543. */
  544. locking_selftest();
  545. #ifdef CONFIG_BLK_DEV_INITRD
  546. if (initrd_start && !initrd_below_start_ok &&
  547. page_to_pfn(virt_to_page((void *)initrd_start)) < min_low_pfn) {
  548. printk(KERN_CRIT "initrd overwritten (0x%08lx < 0x%08lx) - "
  549. "disabling it.\n",
  550. page_to_pfn(virt_to_page((void *)initrd_start)),
  551. min_low_pfn);
  552. initrd_start = 0;
  553. }
  554. #endif
  555. vmalloc_init();
  556. vfs_caches_init_early();
  557. cpuset_init_early();
  558. page_cgroup_init();
  559. mem_init();
  560. enable_debug_pagealloc();
  561. cpu_hotplug_init();
  562. kmem_cache_init();
  563. debug_objects_mem_init();
  564. idr_init_cache();
  565. setup_per_cpu_pageset();
  566. numa_policy_init();
  567. if (late_time_init)
  568. late_time_init();
  569. calibrate_delay();
  570. pidmap_init();
  571. pgtable_cache_init();
  572. prio_tree_init();
  573. anon_vma_init();
  574. #ifdef CONFIG_X86
  575. if (efi_enabled)
  576. efi_enter_virtual_mode();
  577. #endif
  578. thread_info_cache_init();
  579. cred_init();
  580. fork_init(num_physpages);
  581. proc_caches_init();
  582. buffer_init();
  583. key_init();
  584. security_init();
  585. vfs_caches_init(num_physpages);
  586. radix_tree_init();
  587. signals_init();
  588. /* rootfs populating might need page-writeback */
  589. page_writeback_init();
  590. #ifdef CONFIG_PROC_FS
  591. proc_root_init();
  592. #endif
  593. cgroup_init();
  594. cpuset_init();
  595. taskstats_init_early();
  596. delayacct_init();
  597. check_bugs();
  598. acpi_early_init(); /* before LAPIC and SMP init */
  599. ftrace_init();
  600. /* Do the rest non-__init'ed, we're now alive */
  601. rest_init();
  602. }
  603. int initcall_debug;
  604. core_param(initcall_debug, initcall_debug, bool, 0644);
  605. int do_one_initcall(initcall_t fn)
  606. {
  607. int count = preempt_count();
  608. ktime_t calltime, delta, rettime;
  609. char msgbuf[64];
  610. struct boot_trace_call call;
  611. struct boot_trace_ret ret;
  612. if (initcall_debug) {
  613. call.caller = task_pid_nr(current);
  614. printk("calling %pF @ %i\n", fn, call.caller);
  615. calltime = ktime_get();
  616. trace_boot_call(&call, fn);
  617. enable_boot_trace();
  618. }
  619. ret.result = fn();
  620. if (initcall_debug) {
  621. disable_boot_trace();
  622. rettime = ktime_get();
  623. delta = ktime_sub(rettime, calltime);
  624. ret.duration = (unsigned long long) ktime_to_ns(delta) >> 10;
  625. trace_boot_ret(&ret, fn);
  626. printk("initcall %pF returned %d after %Ld usecs\n", fn,
  627. ret.result, ret.duration);
  628. }
  629. msgbuf[0] = 0;
  630. if (ret.result && ret.result != -ENODEV && initcall_debug)
  631. sprintf(msgbuf, "error code %d ", ret.result);
  632. if (preempt_count() != count) {
  633. strlcat(msgbuf, "preemption imbalance ", sizeof(msgbuf));
  634. preempt_count() = count;
  635. }
  636. if (irqs_disabled()) {
  637. strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf));
  638. local_irq_enable();
  639. }
  640. if (msgbuf[0]) {
  641. printk("initcall %pF returned with %s\n", fn, msgbuf);
  642. }
  643. return ret.result;
  644. }
  645. extern initcall_t __initcall_start[], __initcall_end[], __early_initcall_end[];
  646. static void __init do_initcalls(void)
  647. {
  648. initcall_t *call;
  649. for (call = __early_initcall_end; call < __initcall_end; call++)
  650. do_one_initcall(*call);
  651. /* Make sure there is no pending stuff from the initcall sequence */
  652. flush_scheduled_work();
  653. }
  654. /*
  655. * Ok, the machine is now initialized. None of the devices
  656. * have been touched yet, but the CPU subsystem is up and
  657. * running, and memory and process management works.
  658. *
  659. * Now we can finally start doing some real work..
  660. */
  661. static void __init do_basic_setup(void)
  662. {
  663. rcu_init_sched(); /* needed by module_init stage. */
  664. init_workqueues();
  665. usermodehelper_init();
  666. driver_init();
  667. init_irq_proc();
  668. do_initcalls();
  669. }
  670. static void __init do_pre_smp_initcalls(void)
  671. {
  672. initcall_t *call;
  673. for (call = __initcall_start; call < __early_initcall_end; call++)
  674. do_one_initcall(*call);
  675. }
  676. static void run_init_process(char *init_filename)
  677. {
  678. argv_init[0] = init_filename;
  679. kernel_execve(init_filename, argv_init, envp_init);
  680. }
  681. /* This is a non __init function. Force it to be noinline otherwise gcc
  682. * makes it inline to init() and it becomes part of init.text section
  683. */
  684. static noinline int init_post(void)
  685. {
  686. /* need to finish all async __init code before freeing the memory */
  687. async_synchronize_full();
  688. free_initmem();
  689. unlock_kernel();
  690. mark_rodata_ro();
  691. system_state = SYSTEM_RUNNING;
  692. numa_default_policy();
  693. if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0)
  694. printk(KERN_WARNING "Warning: unable to open an initial console.\n");
  695. (void) sys_dup(0);
  696. (void) sys_dup(0);
  697. current->signal->flags |= SIGNAL_UNKILLABLE;
  698. if (ramdisk_execute_command) {
  699. run_init_process(ramdisk_execute_command);
  700. printk(KERN_WARNING "Failed to execute %s\n",
  701. ramdisk_execute_command);
  702. }
  703. /*
  704. * We try each of these until one succeeds.
  705. *
  706. * The Bourne shell can be used instead of init if we are
  707. * trying to recover a really broken machine.
  708. */
  709. if (execute_command) {
  710. run_init_process(execute_command);
  711. printk(KERN_WARNING "Failed to execute %s. Attempting "
  712. "defaults...\n", execute_command);
  713. }
  714. run_init_process("/sbin/init");
  715. run_init_process("/etc/init");
  716. run_init_process("/bin/init");
  717. run_init_process("/bin/sh");
  718. panic("No init found. Try passing init= option to kernel.");
  719. }
  720. static int __init kernel_init(void * unused)
  721. {
  722. lock_kernel();
  723. /*
  724. * init can run on any cpu.
  725. */
  726. set_cpus_allowed_ptr(current, CPU_MASK_ALL_PTR);
  727. /*
  728. * Tell the world that we're going to be the grim
  729. * reaper of innocent orphaned children.
  730. *
  731. * We don't want people to have to make incorrect
  732. * assumptions about where in the task array this
  733. * can be found.
  734. */
  735. init_pid_ns.child_reaper = current;
  736. cad_pid = task_pid(current);
  737. smp_prepare_cpus(setup_max_cpus);
  738. do_pre_smp_initcalls();
  739. start_boot_trace();
  740. smp_init();
  741. sched_init_smp();
  742. cpuset_init_smp();
  743. do_basic_setup();
  744. /*
  745. * check if there is an early userspace init. If yes, let it do all
  746. * the work
  747. */
  748. if (!ramdisk_execute_command)
  749. ramdisk_execute_command = "/init";
  750. if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) {
  751. ramdisk_execute_command = NULL;
  752. prepare_namespace();
  753. }
  754. /*
  755. * Ok, we have completed the initial bootup, and
  756. * we're essentially up and running. Get rid of the
  757. * initmem segments and start the user-mode stuff..
  758. */
  759. init_post();
  760. return 0;
  761. }