main.c 21 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888889890891892
  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/acpi.h>
  27. #include <linux/tty.h>
  28. #include <linux/gfp.h>
  29. #include <linux/percpu.h>
  30. #include <linux/kmod.h>
  31. #include <linux/vmalloc.h>
  32. #include <linux/kernel_stat.h>
  33. #include <linux/start_kernel.h>
  34. #include <linux/security.h>
  35. #include <linux/smp.h>
  36. #include <linux/workqueue.h>
  37. #include <linux/profile.h>
  38. #include <linux/rcupdate.h>
  39. #include <linux/moduleparam.h>
  40. #include <linux/kallsyms.h>
  41. #include <linux/writeback.h>
  42. #include <linux/cpu.h>
  43. #include <linux/cpuset.h>
  44. #include <linux/cgroup.h>
  45. #include <linux/efi.h>
  46. #include <linux/tick.h>
  47. #include <linux/interrupt.h>
  48. #include <linux/taskstats_kern.h>
  49. #include <linux/delayacct.h>
  50. #include <linux/unistd.h>
  51. #include <linux/rmap.h>
  52. #include <linux/mempolicy.h>
  53. #include <linux/key.h>
  54. #include <linux/buffer_head.h>
  55. #include <linux/page_cgroup.h>
  56. #include <linux/debug_locks.h>
  57. #include <linux/debugobjects.h>
  58. #include <linux/lockdep.h>
  59. #include <linux/pid_namespace.h>
  60. #include <linux/device.h>
  61. #include <linux/kthread.h>
  62. #include <linux/sched.h>
  63. #include <linux/signal.h>
  64. #include <linux/idr.h>
  65. #include <linux/ftrace.h>
  66. #include <linux/async.h>
  67. #include <trace/boot.h>
  68. #include <asm/io.h>
  69. #include <asm/bugs.h>
  70. #include <asm/setup.h>
  71. #include <asm/sections.h>
  72. #include <asm/cacheflush.h>
  73. #include <trace/kmemtrace.h>
  74. #ifdef CONFIG_X86_LOCAL_APIC
  75. #include <asm/smp.h>
  76. #endif
  77. static int kernel_init(void *);
  78. extern void init_IRQ(void);
  79. extern void fork_init(unsigned long);
  80. extern void mca_init(void);
  81. extern void sbus_init(void);
  82. extern void prio_tree_init(void);
  83. extern void radix_tree_init(void);
  84. extern void free_initmem(void);
  85. #ifndef CONFIG_DEBUG_RODATA
  86. static inline void mark_rodata_ro(void) { }
  87. #endif
  88. #ifdef CONFIG_TC
  89. extern void tc_init(void);
  90. #endif
  91. enum system_states system_state __read_mostly;
  92. EXPORT_SYMBOL(system_state);
  93. /*
  94. * Boot command-line arguments
  95. */
  96. #define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT
  97. #define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT
  98. extern void time_init(void);
  99. /* Default late time init is NULL. archs can override this later. */
  100. void (*__initdata late_time_init)(void);
  101. extern void softirq_init(void);
  102. /* Untouched command line saved by arch-specific code. */
  103. char __initdata boot_command_line[COMMAND_LINE_SIZE];
  104. /* Untouched saved command line (eg. for /proc) */
  105. char *saved_command_line;
  106. /* Command line for parameter parsing */
  107. static char *static_command_line;
  108. static char *execute_command;
  109. static char *ramdisk_execute_command;
  110. #ifdef CONFIG_SMP
  111. /* Setup configured maximum number of CPUs to activate */
  112. unsigned int __initdata setup_max_cpus = NR_CPUS;
  113. /*
  114. * Setup routine for controlling SMP activation
  115. *
  116. * Command-line option of "nosmp" or "maxcpus=0" will disable SMP
  117. * activation entirely (the MPS table probe still happens, though).
  118. *
  119. * Command-line option of "maxcpus=<NUM>", where <NUM> is an integer
  120. * greater than 0, limits the maximum number of CPUs activated in
  121. * SMP mode to <NUM>.
  122. */
  123. void __weak arch_disable_smp_support(void) { }
  124. static int __init nosmp(char *str)
  125. {
  126. setup_max_cpus = 0;
  127. arch_disable_smp_support();
  128. return 0;
  129. }
  130. early_param("nosmp", nosmp);
  131. static int __init maxcpus(char *str)
  132. {
  133. get_option(&str, &setup_max_cpus);
  134. if (setup_max_cpus == 0)
  135. arch_disable_smp_support();
  136. return 0;
  137. }
  138. early_param("maxcpus", maxcpus);
  139. #else
  140. const unsigned int setup_max_cpus = NR_CPUS;
  141. #endif
  142. /*
  143. * If set, this is an indication to the drivers that reset the underlying
  144. * device before going ahead with the initialization otherwise driver might
  145. * rely on the BIOS and skip the reset operation.
  146. *
  147. * This is useful if kernel is booting in an unreliable environment.
  148. * For ex. kdump situaiton where previous kernel has crashed, BIOS has been
  149. * skipped and devices will be in unknown state.
  150. */
  151. unsigned int reset_devices;
  152. EXPORT_SYMBOL(reset_devices);
  153. static int __init set_reset_devices(char *str)
  154. {
  155. reset_devices = 1;
  156. return 1;
  157. }
  158. __setup("reset_devices", set_reset_devices);
  159. static char * argv_init[MAX_INIT_ARGS+2] = { "init", NULL, };
  160. char * envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, };
  161. static const char *panic_later, *panic_param;
  162. extern struct obs_kernel_param __setup_start[], __setup_end[];
  163. static int __init obsolete_checksetup(char *line)
  164. {
  165. struct obs_kernel_param *p;
  166. int had_early_param = 0;
  167. p = __setup_start;
  168. do {
  169. int n = strlen(p->str);
  170. if (!strncmp(line, p->str, n)) {
  171. if (p->early) {
  172. /* Already done in parse_early_param?
  173. * (Needs exact match on param part).
  174. * Keep iterating, as we can have early
  175. * params and __setups of same names 8( */
  176. if (line[n] == '\0' || line[n] == '=')
  177. had_early_param = 1;
  178. } else if (!p->setup_func) {
  179. printk(KERN_WARNING "Parameter %s is obsolete,"
  180. " ignored\n", p->str);
  181. return 1;
  182. } else if (p->setup_func(line + n))
  183. return 1;
  184. }
  185. p++;
  186. } while (p < __setup_end);
  187. return had_early_param;
  188. }
  189. /*
  190. * This should be approx 2 Bo*oMips to start (note initial shift), and will
  191. * still work even if initially too large, it will just take slightly longer
  192. */
  193. unsigned long loops_per_jiffy = (1<<12);
  194. EXPORT_SYMBOL(loops_per_jiffy);
  195. static int __init debug_kernel(char *str)
  196. {
  197. console_loglevel = 10;
  198. return 0;
  199. }
  200. static int __init quiet_kernel(char *str)
  201. {
  202. console_loglevel = 4;
  203. return 0;
  204. }
  205. early_param("debug", debug_kernel);
  206. early_param("quiet", quiet_kernel);
  207. static int __init loglevel(char *str)
  208. {
  209. get_option(&str, &console_loglevel);
  210. return 0;
  211. }
  212. early_param("loglevel", loglevel);
  213. /*
  214. * Unknown boot options get handed to init, unless they look like
  215. * failed parameters
  216. */
  217. static int __init unknown_bootoption(char *param, char *val)
  218. {
  219. /* Change NUL term back to "=", to make "param" the whole string. */
  220. if (val) {
  221. /* param=val or param="val"? */
  222. if (val == param+strlen(param)+1)
  223. val[-1] = '=';
  224. else if (val == param+strlen(param)+2) {
  225. val[-2] = '=';
  226. memmove(val-1, val, strlen(val)+1);
  227. val--;
  228. } else
  229. BUG();
  230. }
  231. /* Handle obsolete-style parameters */
  232. if (obsolete_checksetup(param))
  233. return 0;
  234. /*
  235. * Preemptive maintenance for "why didn't my misspelled command
  236. * line work?"
  237. */
  238. if (strchr(param, '.') && (!val || strchr(param, '.') < val)) {
  239. printk(KERN_ERR "Unknown boot option `%s': ignoring\n", param);
  240. return 0;
  241. }
  242. if (panic_later)
  243. return 0;
  244. if (val) {
  245. /* Environment option */
  246. unsigned int i;
  247. for (i = 0; envp_init[i]; i++) {
  248. if (i == MAX_INIT_ENVS) {
  249. panic_later = "Too many boot env vars at `%s'";
  250. panic_param = param;
  251. }
  252. if (!strncmp(param, envp_init[i], val - param))
  253. break;
  254. }
  255. envp_init[i] = param;
  256. } else {
  257. /* Command line option */
  258. unsigned int i;
  259. for (i = 0; argv_init[i]; i++) {
  260. if (i == MAX_INIT_ARGS) {
  261. panic_later = "Too many boot init vars at `%s'";
  262. panic_param = param;
  263. }
  264. }
  265. argv_init[i] = param;
  266. }
  267. return 0;
  268. }
  269. #ifdef CONFIG_DEBUG_PAGEALLOC
  270. int __read_mostly debug_pagealloc_enabled = 0;
  271. #endif
  272. static int __init init_setup(char *str)
  273. {
  274. unsigned int i;
  275. execute_command = str;
  276. /*
  277. * In case LILO is going to boot us with default command line,
  278. * it prepends "auto" before the whole cmdline which makes
  279. * the shell think it should execute a script with such name.
  280. * So we ignore all arguments entered _before_ init=... [MJ]
  281. */
  282. for (i = 1; i < MAX_INIT_ARGS; i++)
  283. argv_init[i] = NULL;
  284. return 1;
  285. }
  286. __setup("init=", init_setup);
  287. static int __init rdinit_setup(char *str)
  288. {
  289. unsigned int i;
  290. ramdisk_execute_command = str;
  291. /* See "auto" comment in init_setup */
  292. for (i = 1; i < MAX_INIT_ARGS; i++)
  293. argv_init[i] = NULL;
  294. return 1;
  295. }
  296. __setup("rdinit=", rdinit_setup);
  297. #ifndef CONFIG_SMP
  298. #ifdef CONFIG_X86_LOCAL_APIC
  299. static void __init smp_init(void)
  300. {
  301. APIC_init_uniprocessor();
  302. }
  303. #else
  304. #define smp_init() do { } while (0)
  305. #endif
  306. static inline void setup_per_cpu_areas(void) { }
  307. static inline void setup_nr_cpu_ids(void) { }
  308. static inline void smp_prepare_cpus(unsigned int maxcpus) { }
  309. #else
  310. #if NR_CPUS > BITS_PER_LONG
  311. cpumask_t cpu_mask_all __read_mostly = CPU_MASK_ALL;
  312. EXPORT_SYMBOL(cpu_mask_all);
  313. #endif
  314. /* Setup number of possible processor ids */
  315. int nr_cpu_ids __read_mostly = NR_CPUS;
  316. EXPORT_SYMBOL(nr_cpu_ids);
  317. /* An arch may set nr_cpu_ids earlier if needed, so this would be redundant */
  318. static void __init setup_nr_cpu_ids(void)
  319. {
  320. nr_cpu_ids = find_last_bit(cpumask_bits(cpu_possible_mask),NR_CPUS) + 1;
  321. }
  322. #ifndef CONFIG_HAVE_SETUP_PER_CPU_AREA
  323. unsigned long __per_cpu_offset[NR_CPUS] __read_mostly;
  324. EXPORT_SYMBOL(__per_cpu_offset);
  325. static void __init setup_per_cpu_areas(void)
  326. {
  327. unsigned long size, i;
  328. char *ptr;
  329. unsigned long nr_possible_cpus = num_possible_cpus();
  330. /* Copy section for each CPU (we discard the original) */
  331. size = ALIGN(PERCPU_ENOUGH_ROOM, PAGE_SIZE);
  332. ptr = alloc_bootmem_pages(size * nr_possible_cpus);
  333. for_each_possible_cpu(i) {
  334. __per_cpu_offset[i] = ptr - __per_cpu_start;
  335. memcpy(ptr, __per_cpu_start, __per_cpu_end - __per_cpu_start);
  336. ptr += size;
  337. }
  338. }
  339. #endif /* CONFIG_HAVE_SETUP_PER_CPU_AREA */
  340. /* Called by boot processor to activate the rest. */
  341. static void __init smp_init(void)
  342. {
  343. unsigned int cpu;
  344. /*
  345. * Set up the current CPU as possible to migrate to.
  346. * The other ones will be done by cpu_up/cpu_down()
  347. */
  348. set_cpu_active(smp_processor_id(), true);
  349. /* FIXME: This should be done in userspace --RR */
  350. for_each_present_cpu(cpu) {
  351. if (num_online_cpus() >= setup_max_cpus)
  352. break;
  353. if (!cpu_online(cpu))
  354. cpu_up(cpu);
  355. }
  356. /* Any cleanup work */
  357. printk(KERN_INFO "Brought up %ld CPUs\n", (long)num_online_cpus());
  358. smp_cpus_done(setup_max_cpus);
  359. }
  360. #endif
  361. /*
  362. * We need to store the untouched command line for future reference.
  363. * We also need to store the touched command line since the parameter
  364. * parsing is performed in place, and we should allow a component to
  365. * store reference of name/value for future reference.
  366. */
  367. static void __init setup_command_line(char *command_line)
  368. {
  369. saved_command_line = alloc_bootmem(strlen (boot_command_line)+1);
  370. static_command_line = alloc_bootmem(strlen (command_line)+1);
  371. strcpy (saved_command_line, boot_command_line);
  372. strcpy (static_command_line, command_line);
  373. }
  374. /*
  375. * We need to finalize in a non-__init function or else race conditions
  376. * between the root thread and the init thread may cause start_kernel to
  377. * be reaped by free_initmem before the root thread has proceeded to
  378. * cpu_idle.
  379. *
  380. * gcc-3.4 accidentally inlines this function, so use noinline.
  381. */
  382. static noinline void __init_refok rest_init(void)
  383. __releases(kernel_lock)
  384. {
  385. int pid;
  386. kernel_thread(kernel_init, NULL, CLONE_FS | CLONE_SIGHAND);
  387. numa_default_policy();
  388. pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES);
  389. kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns);
  390. unlock_kernel();
  391. /*
  392. * The boot idle thread must execute schedule()
  393. * at least once to get things moving:
  394. */
  395. init_idle_bootup_task(current);
  396. rcu_scheduler_starting();
  397. preempt_enable_no_resched();
  398. schedule();
  399. preempt_disable();
  400. /* Call into cpu_idle with preempt disabled */
  401. cpu_idle();
  402. }
  403. /* Check for early params. */
  404. static int __init do_early_param(char *param, char *val)
  405. {
  406. struct obs_kernel_param *p;
  407. for (p = __setup_start; p < __setup_end; p++) {
  408. if ((p->early && strcmp(param, p->str) == 0) ||
  409. (strcmp(param, "console") == 0 &&
  410. strcmp(p->str, "earlycon") == 0)
  411. ) {
  412. if (p->setup_func(val) != 0)
  413. printk(KERN_WARNING
  414. "Malformed early option '%s'\n", param);
  415. }
  416. }
  417. /* We accept everything at this stage. */
  418. return 0;
  419. }
  420. void __init parse_early_options(char *cmdline)
  421. {
  422. parse_args("early options", cmdline, NULL, 0, do_early_param);
  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_early_options(tmp_cmdline);
  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 "%s", linux_banner);
  481. setup_arch(&command_line);
  482. mm_init_owner(&init_mm, &init_task);
  483. setup_command_line(command_line);
  484. setup_per_cpu_areas();
  485. setup_nr_cpu_ids();
  486. smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */
  487. /*
  488. * Set up the scheduler prior starting any interrupts (such as the
  489. * timer interrupt). Full topology setup happens at smp_init()
  490. * time - but meanwhile we still have a functioning scheduler.
  491. */
  492. sched_init();
  493. /*
  494. * Disable preemption - early bootup scheduling is extremely
  495. * fragile until we cpu_idle() for the first time.
  496. */
  497. preempt_disable();
  498. build_all_zonelists();
  499. page_alloc_init();
  500. printk(KERN_NOTICE "Kernel command line: %s\n", boot_command_line);
  501. parse_early_param();
  502. parse_args("Booting kernel", static_command_line, __start___param,
  503. __stop___param - __start___param,
  504. &unknown_bootoption);
  505. if (!irqs_disabled()) {
  506. printk(KERN_WARNING "start_kernel(): bug: interrupts were "
  507. "enabled *very* early, fixing it\n");
  508. local_irq_disable();
  509. }
  510. sort_main_extable();
  511. trap_init();
  512. rcu_init();
  513. /* init some links before init_ISA_irqs() */
  514. early_irq_init();
  515. init_IRQ();
  516. pidhash_init();
  517. init_timers();
  518. hrtimers_init();
  519. softirq_init();
  520. timekeeping_init();
  521. time_init();
  522. sched_clock_init();
  523. profile_init();
  524. if (!irqs_disabled())
  525. printk(KERN_CRIT "start_kernel(): bug: interrupts were "
  526. "enabled early\n");
  527. early_boot_irqs_on();
  528. local_irq_enable();
  529. /*
  530. * HACK ALERT! This is early. We're enabling the console before
  531. * we've done PCI setups etc, and console_init() must be aware of
  532. * this. But we do want output early, in case something goes wrong.
  533. */
  534. console_init();
  535. if (panic_later)
  536. panic(panic_later, panic_param);
  537. lockdep_info();
  538. /*
  539. * Need to run this when irqs are enabled, because it wants
  540. * to self-test [hard/soft]-irqs on/off lock inversion bugs
  541. * too:
  542. */
  543. locking_selftest();
  544. #ifdef CONFIG_BLK_DEV_INITRD
  545. if (initrd_start && !initrd_below_start_ok &&
  546. page_to_pfn(virt_to_page((void *)initrd_start)) < min_low_pfn) {
  547. printk(KERN_CRIT "initrd overwritten (0x%08lx < 0x%08lx) - "
  548. "disabling it.\n",
  549. page_to_pfn(virt_to_page((void *)initrd_start)),
  550. min_low_pfn);
  551. initrd_start = 0;
  552. }
  553. #endif
  554. vmalloc_init();
  555. vfs_caches_init_early();
  556. cpuset_init_early();
  557. page_cgroup_init();
  558. mem_init();
  559. enable_debug_pagealloc();
  560. cpu_hotplug_init();
  561. kmem_cache_init();
  562. kmemtrace_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. cpuset_init_smp();
  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. __releases(kernel_lock)
  687. {
  688. /* need to finish all async __init code before freeing the memory */
  689. async_synchronize_full();
  690. free_initmem();
  691. unlock_kernel();
  692. mark_rodata_ro();
  693. system_state = SYSTEM_RUNNING;
  694. numa_default_policy();
  695. if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0)
  696. printk(KERN_WARNING "Warning: unable to open an initial console.\n");
  697. (void) sys_dup(0);
  698. (void) sys_dup(0);
  699. current->signal->flags |= SIGNAL_UNKILLABLE;
  700. if (ramdisk_execute_command) {
  701. run_init_process(ramdisk_execute_command);
  702. printk(KERN_WARNING "Failed to execute %s\n",
  703. ramdisk_execute_command);
  704. }
  705. /*
  706. * We try each of these until one succeeds.
  707. *
  708. * The Bourne shell can be used instead of init if we are
  709. * trying to recover a really broken machine.
  710. */
  711. if (execute_command) {
  712. run_init_process(execute_command);
  713. printk(KERN_WARNING "Failed to execute %s. Attempting "
  714. "defaults...\n", execute_command);
  715. }
  716. run_init_process("/sbin/init");
  717. run_init_process("/etc/init");
  718. run_init_process("/bin/init");
  719. run_init_process("/bin/sh");
  720. panic("No init found. Try passing init= option to kernel.");
  721. }
  722. static int __init kernel_init(void * unused)
  723. {
  724. lock_kernel();
  725. /*
  726. * init can run on any cpu.
  727. */
  728. set_cpus_allowed_ptr(current, cpu_all_mask);
  729. /*
  730. * Tell the world that we're going to be the grim
  731. * reaper of innocent orphaned children.
  732. *
  733. * We don't want people to have to make incorrect
  734. * assumptions about where in the task array this
  735. * can be found.
  736. */
  737. init_pid_ns.child_reaper = current;
  738. cad_pid = task_pid(current);
  739. smp_prepare_cpus(setup_max_cpus);
  740. do_pre_smp_initcalls();
  741. start_boot_trace();
  742. smp_init();
  743. sched_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. }