main.c 21 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792793794795796797798799800801802803804805806807808809810811812813814815816817818819820821822823824825826827828829830831832833834835836837838839840841842843844845846847848849850851852853854855856857858859860861862863864865866867868869870871872873874875876877878879880881882883884885886887888
  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/string.h>
  17. #include <linux/ctype.h>
  18. #include <linux/delay.h>
  19. #include <linux/utsname.h>
  20. #include <linux/ioport.h>
  21. #include <linux/init.h>
  22. #include <linux/smp_lock.h>
  23. #include <linux/initrd.h>
  24. #include <linux/bootmem.h>
  25. #include <linux/tty.h>
  26. #include <linux/gfp.h>
  27. #include <linux/percpu.h>
  28. #include <linux/kmod.h>
  29. #include <linux/vmalloc.h>
  30. #include <linux/kernel_stat.h>
  31. #include <linux/start_kernel.h>
  32. #include <linux/security.h>
  33. #include <linux/smp.h>
  34. #include <linux/workqueue.h>
  35. #include <linux/profile.h>
  36. #include <linux/rcupdate.h>
  37. #include <linux/moduleparam.h>
  38. #include <linux/kallsyms.h>
  39. #include <linux/writeback.h>
  40. #include <linux/cpu.h>
  41. #include <linux/cpuset.h>
  42. #include <linux/cgroup.h>
  43. #include <linux/efi.h>
  44. #include <linux/tick.h>
  45. #include <linux/interrupt.h>
  46. #include <linux/taskstats_kern.h>
  47. #include <linux/delayacct.h>
  48. #include <linux/unistd.h>
  49. #include <linux/rmap.h>
  50. #include <linux/mempolicy.h>
  51. #include <linux/key.h>
  52. #include <linux/unwind.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 <asm/io.h>
  66. #include <asm/bugs.h>
  67. #include <asm/setup.h>
  68. #include <asm/sections.h>
  69. #include <asm/cacheflush.h>
  70. #ifdef CONFIG_X86_LOCAL_APIC
  71. #include <asm/smp.h>
  72. #endif
  73. /*
  74. * This is one of the first .c files built. Error out early if we have compiler
  75. * trouble.
  76. */
  77. #if __GNUC__ == 4 && __GNUC_MINOR__ == 1 && __GNUC_PATCHLEVEL__ == 0
  78. #warning gcc-4.1.0 is known to miscompile the kernel. A different compiler version is recommended.
  79. #endif
  80. static int kernel_init(void *);
  81. extern void init_IRQ(void);
  82. extern void fork_init(unsigned long);
  83. extern void mca_init(void);
  84. extern void sbus_init(void);
  85. extern void prio_tree_init(void);
  86. extern void radix_tree_init(void);
  87. extern void free_initmem(void);
  88. #ifdef CONFIG_ACPI
  89. extern void acpi_early_init(void);
  90. #else
  91. static inline void acpi_early_init(void) { }
  92. #endif
  93. #ifndef CONFIG_DEBUG_RODATA
  94. static inline void mark_rodata_ro(void) { }
  95. #endif
  96. #ifdef CONFIG_TC
  97. extern void tc_init(void);
  98. #endif
  99. enum system_states system_state;
  100. EXPORT_SYMBOL(system_state);
  101. /*
  102. * Boot command-line arguments
  103. */
  104. #define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT
  105. #define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT
  106. extern void time_init(void);
  107. /* Default late time init is NULL. archs can override this later. */
  108. void (*late_time_init)(void);
  109. extern void softirq_init(void);
  110. /* Untouched command line saved by arch-specific code. */
  111. char __initdata boot_command_line[COMMAND_LINE_SIZE];
  112. /* Untouched saved command line (eg. for /proc) */
  113. char *saved_command_line;
  114. /* Command line for parameter parsing */
  115. static char *static_command_line;
  116. static char *execute_command;
  117. static char *ramdisk_execute_command;
  118. #ifdef CONFIG_SMP
  119. /* Setup configured maximum number of CPUs to activate */
  120. unsigned int __initdata setup_max_cpus = NR_CPUS;
  121. /*
  122. * Setup routine for controlling SMP activation
  123. *
  124. * Command-line option of "nosmp" or "maxcpus=0" will disable SMP
  125. * activation entirely (the MPS table probe still happens, though).
  126. *
  127. * Command-line option of "maxcpus=<NUM>", where <NUM> is an integer
  128. * greater than 0, limits the maximum number of CPUs activated in
  129. * SMP mode to <NUM>.
  130. */
  131. #ifndef CONFIG_X86_IO_APIC
  132. static inline void disable_ioapic_setup(void) {};
  133. #endif
  134. static int __init nosmp(char *str)
  135. {
  136. setup_max_cpus = 0;
  137. disable_ioapic_setup();
  138. return 0;
  139. }
  140. early_param("nosmp", nosmp);
  141. static int __init maxcpus(char *str)
  142. {
  143. get_option(&str, &setup_max_cpus);
  144. if (setup_max_cpus == 0)
  145. disable_ioapic_setup();
  146. return 0;
  147. }
  148. early_param("maxcpus", maxcpus);
  149. #else
  150. #define setup_max_cpus NR_CPUS
  151. #endif
  152. /*
  153. * If set, this is an indication to the drivers that reset the underlying
  154. * device before going ahead with the initialization otherwise driver might
  155. * rely on the BIOS and skip the reset operation.
  156. *
  157. * This is useful if kernel is booting in an unreliable environment.
  158. * For ex. kdump situaiton where previous kernel has crashed, BIOS has been
  159. * skipped and devices will be in unknown state.
  160. */
  161. unsigned int reset_devices;
  162. EXPORT_SYMBOL(reset_devices);
  163. static int __init set_reset_devices(char *str)
  164. {
  165. reset_devices = 1;
  166. return 1;
  167. }
  168. __setup("reset_devices", set_reset_devices);
  169. static char * argv_init[MAX_INIT_ARGS+2] = { "init", NULL, };
  170. char * envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, };
  171. static const char *panic_later, *panic_param;
  172. extern struct obs_kernel_param __setup_start[], __setup_end[];
  173. static int __init obsolete_checksetup(char *line)
  174. {
  175. struct obs_kernel_param *p;
  176. int had_early_param = 0;
  177. p = __setup_start;
  178. do {
  179. int n = strlen(p->str);
  180. if (!strncmp(line, p->str, n)) {
  181. if (p->early) {
  182. /* Already done in parse_early_param?
  183. * (Needs exact match on param part).
  184. * Keep iterating, as we can have early
  185. * params and __setups of same names 8( */
  186. if (line[n] == '\0' || line[n] == '=')
  187. had_early_param = 1;
  188. } else if (!p->setup_func) {
  189. printk(KERN_WARNING "Parameter %s is obsolete,"
  190. " ignored\n", p->str);
  191. return 1;
  192. } else if (p->setup_func(line + n))
  193. return 1;
  194. }
  195. p++;
  196. } while (p < __setup_end);
  197. return had_early_param;
  198. }
  199. /*
  200. * This should be approx 2 Bo*oMips to start (note initial shift), and will
  201. * still work even if initially too large, it will just take slightly longer
  202. */
  203. unsigned long loops_per_jiffy = (1<<12);
  204. EXPORT_SYMBOL(loops_per_jiffy);
  205. static int __init debug_kernel(char *str)
  206. {
  207. console_loglevel = 10;
  208. return 0;
  209. }
  210. static int __init quiet_kernel(char *str)
  211. {
  212. console_loglevel = 4;
  213. return 0;
  214. }
  215. early_param("debug", debug_kernel);
  216. early_param("quiet", quiet_kernel);
  217. static int __init loglevel(char *str)
  218. {
  219. get_option(&str, &console_loglevel);
  220. return 0;
  221. }
  222. early_param("loglevel", loglevel);
  223. /*
  224. * Unknown boot options get handed to init, unless they look like
  225. * failed parameters
  226. */
  227. static int __init unknown_bootoption(char *param, char *val)
  228. {
  229. /* Change NUL term back to "=", to make "param" the whole string. */
  230. if (val) {
  231. /* param=val or param="val"? */
  232. if (val == param+strlen(param)+1)
  233. val[-1] = '=';
  234. else if (val == param+strlen(param)+2) {
  235. val[-2] = '=';
  236. memmove(val-1, val, strlen(val)+1);
  237. val--;
  238. } else
  239. BUG();
  240. }
  241. /* Handle obsolete-style parameters */
  242. if (obsolete_checksetup(param))
  243. return 0;
  244. /*
  245. * Preemptive maintenance for "why didn't my misspelled command
  246. * line work?"
  247. */
  248. if (strchr(param, '.') && (!val || strchr(param, '.') < val)) {
  249. printk(KERN_ERR "Unknown boot option `%s': ignoring\n", param);
  250. return 0;
  251. }
  252. if (panic_later)
  253. return 0;
  254. if (val) {
  255. /* Environment option */
  256. unsigned int i;
  257. for (i = 0; envp_init[i]; i++) {
  258. if (i == MAX_INIT_ENVS) {
  259. panic_later = "Too many boot env vars at `%s'";
  260. panic_param = param;
  261. }
  262. if (!strncmp(param, envp_init[i], val - param))
  263. break;
  264. }
  265. envp_init[i] = param;
  266. } else {
  267. /* Command line option */
  268. unsigned int i;
  269. for (i = 0; argv_init[i]; i++) {
  270. if (i == MAX_INIT_ARGS) {
  271. panic_later = "Too many boot init vars at `%s'";
  272. panic_param = param;
  273. }
  274. }
  275. argv_init[i] = param;
  276. }
  277. return 0;
  278. }
  279. #ifdef CONFIG_DEBUG_PAGEALLOC
  280. int __read_mostly debug_pagealloc_enabled = 0;
  281. #endif
  282. static int __init init_setup(char *str)
  283. {
  284. unsigned int i;
  285. execute_command = str;
  286. /*
  287. * In case LILO is going to boot us with default command line,
  288. * it prepends "auto" before the whole cmdline which makes
  289. * the shell think it should execute a script with such name.
  290. * So we ignore all arguments entered _before_ init=... [MJ]
  291. */
  292. for (i = 1; i < MAX_INIT_ARGS; i++)
  293. argv_init[i] = NULL;
  294. return 1;
  295. }
  296. __setup("init=", init_setup);
  297. static int __init rdinit_setup(char *str)
  298. {
  299. unsigned int i;
  300. ramdisk_execute_command = str;
  301. /* See "auto" comment in init_setup */
  302. for (i = 1; i < MAX_INIT_ARGS; i++)
  303. argv_init[i] = NULL;
  304. return 1;
  305. }
  306. __setup("rdinit=", rdinit_setup);
  307. #ifndef CONFIG_SMP
  308. #ifdef CONFIG_X86_LOCAL_APIC
  309. static void __init smp_init(void)
  310. {
  311. APIC_init_uniprocessor();
  312. }
  313. #else
  314. #define smp_init() do { } while (0)
  315. #endif
  316. static inline void setup_per_cpu_areas(void) { }
  317. static inline void setup_nr_cpu_ids(void) { }
  318. static inline void smp_prepare_cpus(unsigned int maxcpus) { }
  319. #else
  320. #if NR_CPUS > BITS_PER_LONG
  321. cpumask_t cpu_mask_all __read_mostly = CPU_MASK_ALL;
  322. EXPORT_SYMBOL(cpu_mask_all);
  323. #endif
  324. /* Setup number of possible processor ids */
  325. int nr_cpu_ids __read_mostly = NR_CPUS;
  326. EXPORT_SYMBOL(nr_cpu_ids);
  327. /* An arch may set nr_cpu_ids earlier if needed, so this would be redundant */
  328. static void __init setup_nr_cpu_ids(void)
  329. {
  330. int cpu, highest_cpu = 0;
  331. for_each_possible_cpu(cpu)
  332. highest_cpu = cpu;
  333. nr_cpu_ids = highest_cpu + 1;
  334. }
  335. #ifndef CONFIG_HAVE_SETUP_PER_CPU_AREA
  336. unsigned long __per_cpu_offset[NR_CPUS] __read_mostly;
  337. EXPORT_SYMBOL(__per_cpu_offset);
  338. static void __init setup_per_cpu_areas(void)
  339. {
  340. unsigned long size, i;
  341. char *ptr;
  342. unsigned long nr_possible_cpus = num_possible_cpus();
  343. /* Copy section for each CPU (we discard the original) */
  344. size = ALIGN(PERCPU_ENOUGH_ROOM, PAGE_SIZE);
  345. ptr = alloc_bootmem_pages(size * nr_possible_cpus);
  346. for_each_possible_cpu(i) {
  347. __per_cpu_offset[i] = ptr - __per_cpu_start;
  348. memcpy(ptr, __per_cpu_start, __per_cpu_end - __per_cpu_start);
  349. ptr += size;
  350. }
  351. }
  352. #endif /* CONFIG_HAVE_SETUP_PER_CPU_AREA */
  353. /* Called by boot processor to activate the rest. */
  354. static void __init smp_init(void)
  355. {
  356. unsigned int cpu;
  357. /*
  358. * Set up the current CPU as possible to migrate to.
  359. * The other ones will be done by cpu_up/cpu_down()
  360. */
  361. cpu = smp_processor_id();
  362. cpu_set(cpu, cpu_active_map);
  363. /* FIXME: This should be done in userspace --RR */
  364. for_each_present_cpu(cpu) {
  365. if (num_online_cpus() >= setup_max_cpus)
  366. break;
  367. if (!cpu_online(cpu))
  368. cpu_up(cpu);
  369. }
  370. /* Any cleanup work */
  371. printk(KERN_INFO "Brought up %ld CPUs\n", (long)num_online_cpus());
  372. smp_cpus_done(setup_max_cpus);
  373. }
  374. #endif
  375. /*
  376. * We need to store the untouched command line for future reference.
  377. * We also need to store the touched command line since the parameter
  378. * parsing is performed in place, and we should allow a component to
  379. * store reference of name/value for future reference.
  380. */
  381. static void __init setup_command_line(char *command_line)
  382. {
  383. saved_command_line = alloc_bootmem(strlen (boot_command_line)+1);
  384. static_command_line = alloc_bootmem(strlen (command_line)+1);
  385. strcpy (saved_command_line, boot_command_line);
  386. strcpy (static_command_line, command_line);
  387. }
  388. /*
  389. * We need to finalize in a non-__init function or else race conditions
  390. * between the root thread and the init thread may cause start_kernel to
  391. * be reaped by free_initmem before the root thread has proceeded to
  392. * cpu_idle.
  393. *
  394. * gcc-3.4 accidentally inlines this function, so use noinline.
  395. */
  396. static void noinline __init_refok rest_init(void)
  397. __releases(kernel_lock)
  398. {
  399. int pid;
  400. kernel_thread(kernel_init, NULL, CLONE_FS | CLONE_SIGHAND);
  401. numa_default_policy();
  402. pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES);
  403. kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns);
  404. unlock_kernel();
  405. /*
  406. * The boot idle thread must execute schedule()
  407. * at least once to get things moving:
  408. */
  409. init_idle_bootup_task(current);
  410. preempt_enable_no_resched();
  411. schedule();
  412. preempt_disable();
  413. /* Call into cpu_idle with preempt disabled */
  414. cpu_idle();
  415. }
  416. /* Check for early params. */
  417. static int __init do_early_param(char *param, char *val)
  418. {
  419. struct obs_kernel_param *p;
  420. for (p = __setup_start; p < __setup_end; p++) {
  421. if ((p->early && strcmp(param, p->str) == 0) ||
  422. (strcmp(param, "console") == 0 &&
  423. strcmp(p->str, "earlycon") == 0)
  424. ) {
  425. if (p->setup_func(val) != 0)
  426. printk(KERN_WARNING
  427. "Malformed early option '%s'\n", param);
  428. }
  429. }
  430. /* We accept everything at this stage. */
  431. return 0;
  432. }
  433. /* Arch code calls this early on, or if not, just before other parsing. */
  434. void __init parse_early_param(void)
  435. {
  436. static __initdata int done = 0;
  437. static __initdata char tmp_cmdline[COMMAND_LINE_SIZE];
  438. if (done)
  439. return;
  440. /* All fall through to do_early_param. */
  441. strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE);
  442. parse_args("early options", tmp_cmdline, NULL, 0, do_early_param);
  443. done = 1;
  444. }
  445. /*
  446. * Activate the first processor.
  447. */
  448. static void __init boot_cpu_init(void)
  449. {
  450. int cpu = smp_processor_id();
  451. /* Mark the boot cpu "present", "online" etc for SMP and UP case */
  452. cpu_set(cpu, cpu_online_map);
  453. cpu_set(cpu, cpu_present_map);
  454. cpu_set(cpu, cpu_possible_map);
  455. }
  456. void __init __weak smp_setup_processor_id(void)
  457. {
  458. }
  459. void __init __weak thread_info_cache_init(void)
  460. {
  461. }
  462. asmlinkage void __init start_kernel(void)
  463. {
  464. char * command_line;
  465. extern struct kernel_param __start___param[], __stop___param[];
  466. smp_setup_processor_id();
  467. /*
  468. * Need to run as early as possible, to initialize the
  469. * lockdep hash:
  470. */
  471. unwind_init();
  472. lockdep_init();
  473. debug_objects_early_init();
  474. cgroup_init_early();
  475. local_irq_disable();
  476. early_boot_irqs_off();
  477. early_init_irq_lock_class();
  478. /*
  479. * Interrupts are still disabled. Do necessary setups, then
  480. * enable them
  481. */
  482. lock_kernel();
  483. tick_init();
  484. boot_cpu_init();
  485. page_address_init();
  486. printk(KERN_NOTICE);
  487. printk(linux_banner);
  488. setup_arch(&command_line);
  489. mm_init_owner(&init_mm, &init_task);
  490. setup_command_line(command_line);
  491. unwind_setup();
  492. setup_per_cpu_areas();
  493. setup_nr_cpu_ids();
  494. smp_prepare_boot_cpu(); /* arch-specific boot-cpu hooks */
  495. /*
  496. * Set up the scheduler prior starting any interrupts (such as the
  497. * timer interrupt). Full topology setup happens at smp_init()
  498. * time - but meanwhile we still have a functioning scheduler.
  499. */
  500. sched_init();
  501. /*
  502. * Disable preemption - early bootup scheduling is extremely
  503. * fragile until we cpu_idle() for the first time.
  504. */
  505. preempt_disable();
  506. build_all_zonelists();
  507. page_alloc_init();
  508. printk(KERN_NOTICE "Kernel command line: %s\n", boot_command_line);
  509. parse_early_param();
  510. parse_args("Booting kernel", static_command_line, __start___param,
  511. __stop___param - __start___param,
  512. &unknown_bootoption);
  513. if (!irqs_disabled()) {
  514. printk(KERN_WARNING "start_kernel(): bug: interrupts were "
  515. "enabled *very* early, fixing it\n");
  516. local_irq_disable();
  517. }
  518. sort_main_extable();
  519. trap_init();
  520. rcu_init();
  521. init_IRQ();
  522. pidhash_init();
  523. init_timers();
  524. hrtimers_init();
  525. softirq_init();
  526. timekeeping_init();
  527. time_init();
  528. sched_clock_init();
  529. profile_init();
  530. if (!irqs_disabled())
  531. printk("start_kernel(): bug: interrupts were 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. debug_objects_mem_init();
  568. idr_init_cache();
  569. setup_per_cpu_pageset();
  570. numa_policy_init();
  571. if (late_time_init)
  572. late_time_init();
  573. calibrate_delay();
  574. pidmap_init();
  575. pgtable_cache_init();
  576. prio_tree_init();
  577. anon_vma_init();
  578. #ifdef CONFIG_X86
  579. if (efi_enabled)
  580. efi_enter_virtual_mode();
  581. #endif
  582. thread_info_cache_init();
  583. fork_init(num_physpages);
  584. proc_caches_init();
  585. buffer_init();
  586. key_init();
  587. security_init();
  588. vfs_caches_init(num_physpages);
  589. radix_tree_init();
  590. signals_init();
  591. /* rootfs populating might need page-writeback */
  592. page_writeback_init();
  593. #ifdef CONFIG_PROC_FS
  594. proc_root_init();
  595. #endif
  596. cgroup_init();
  597. cpuset_init();
  598. taskstats_init_early();
  599. delayacct_init();
  600. check_bugs();
  601. acpi_early_init(); /* before LAPIC and SMP init */
  602. ftrace_init();
  603. /* Do the rest non-__init'ed, we're now alive */
  604. rest_init();
  605. }
  606. static int initcall_debug;
  607. core_param(initcall_debug, initcall_debug, bool, 0644);
  608. int do_one_initcall(initcall_t fn)
  609. {
  610. int count = preempt_count();
  611. ktime_t delta;
  612. char msgbuf[64];
  613. struct boot_trace it;
  614. if (initcall_debug) {
  615. it.caller = task_pid_nr(current);
  616. printk("calling %pF @ %i\n", fn, it.caller);
  617. it.calltime = ktime_get();
  618. }
  619. it.result = fn();
  620. if (initcall_debug) {
  621. it.rettime = ktime_get();
  622. delta = ktime_sub(it.rettime, it.calltime);
  623. it.duration = (unsigned long long) delta.tv64 >> 10;
  624. printk("initcall %pF returned %d after %Ld usecs\n", fn,
  625. it.result, it.duration);
  626. trace_boot(&it, fn);
  627. }
  628. msgbuf[0] = 0;
  629. if (it.result && it.result != -ENODEV && initcall_debug)
  630. sprintf(msgbuf, "error code %d ", it.result);
  631. if (preempt_count() != count) {
  632. strlcat(msgbuf, "preemption imbalance ", sizeof(msgbuf));
  633. preempt_count() = count;
  634. }
  635. if (irqs_disabled()) {
  636. strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf));
  637. local_irq_enable();
  638. }
  639. if (msgbuf[0]) {
  640. printk("initcall %pF returned with %s\n", fn, msgbuf);
  641. }
  642. return it.result;
  643. }
  644. extern initcall_t __initcall_start[], __initcall_end[], __early_initcall_end[];
  645. static void __init do_initcalls(void)
  646. {
  647. initcall_t *call;
  648. for (call = __early_initcall_end; call < __initcall_end; call++)
  649. do_one_initcall(*call);
  650. /* Make sure there is no pending stuff from the initcall sequence */
  651. flush_scheduled_work();
  652. }
  653. /*
  654. * Ok, the machine is now initialized. None of the devices
  655. * have been touched yet, but the CPU subsystem is up and
  656. * running, and memory and process management works.
  657. *
  658. * Now we can finally start doing some real work..
  659. */
  660. static void __init do_basic_setup(void)
  661. {
  662. rcu_init_sched(); /* needed by module_init stage. */
  663. init_workqueues();
  664. usermodehelper_init();
  665. driver_init();
  666. init_irq_proc();
  667. do_initcalls();
  668. }
  669. static void __init do_pre_smp_initcalls(void)
  670. {
  671. initcall_t *call;
  672. for (call = __initcall_start; call < __early_initcall_end; call++)
  673. do_one_initcall(*call);
  674. }
  675. static void run_init_process(char *init_filename)
  676. {
  677. argv_init[0] = init_filename;
  678. kernel_execve(init_filename, argv_init, envp_init);
  679. }
  680. /* This is a non __init function. Force it to be noinline otherwise gcc
  681. * makes it inline to init() and it becomes part of init.text section
  682. */
  683. static int noinline init_post(void)
  684. {
  685. free_initmem();
  686. unlock_kernel();
  687. mark_rodata_ro();
  688. system_state = SYSTEM_RUNNING;
  689. numa_default_policy();
  690. if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0)
  691. printk(KERN_WARNING "Warning: unable to open an initial console.\n");
  692. (void) sys_dup(0);
  693. (void) sys_dup(0);
  694. current->signal->flags |= SIGNAL_UNKILLABLE;
  695. if (ramdisk_execute_command) {
  696. run_init_process(ramdisk_execute_command);
  697. printk(KERN_WARNING "Failed to execute %s\n",
  698. ramdisk_execute_command);
  699. }
  700. /*
  701. * We try each of these until one succeeds.
  702. *
  703. * The Bourne shell can be used instead of init if we are
  704. * trying to recover a really broken machine.
  705. */
  706. if (execute_command) {
  707. run_init_process(execute_command);
  708. printk(KERN_WARNING "Failed to execute %s. Attempting "
  709. "defaults...\n", execute_command);
  710. }
  711. run_init_process("/sbin/init");
  712. run_init_process("/etc/init");
  713. run_init_process("/bin/init");
  714. run_init_process("/bin/sh");
  715. panic("No init found. Try passing init= option to kernel.");
  716. }
  717. static int __init kernel_init(void * unused)
  718. {
  719. lock_kernel();
  720. /*
  721. * init can run on any cpu.
  722. */
  723. set_cpus_allowed_ptr(current, CPU_MASK_ALL_PTR);
  724. /*
  725. * Tell the world that we're going to be the grim
  726. * reaper of innocent orphaned children.
  727. *
  728. * We don't want people to have to make incorrect
  729. * assumptions about where in the task array this
  730. * can be found.
  731. */
  732. init_pid_ns.child_reaper = current;
  733. cad_pid = task_pid(current);
  734. smp_prepare_cpus(setup_max_cpus);
  735. do_pre_smp_initcalls();
  736. start_boot_trace();
  737. smp_init();
  738. sched_init_smp();
  739. cpuset_init_smp();
  740. do_basic_setup();
  741. /*
  742. * check if there is an early userspace init. If yes, let it do all
  743. * the work
  744. */
  745. if (!ramdisk_execute_command)
  746. ramdisk_execute_command = "/init";
  747. if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) {
  748. ramdisk_execute_command = NULL;
  749. prepare_namespace();
  750. }
  751. /*
  752. * Ok, we have completed the initial bootup, and
  753. * we're essentially up and running. Get rid of the
  754. * initmem segments and start the user-mode stuff..
  755. */
  756. stop_boot_trace();
  757. init_post();
  758. return 0;
  759. }