| lh | 9ed821d | 2023-04-07 01:36:19 -0700 | [diff] [blame] | 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 |  | 
 | 12 | #include <linux/types.h> | 
 | 13 | #include <linux/module.h> | 
 | 14 | #include <linux/proc_fs.h> | 
 | 15 | #include <linux/kernel.h> | 
 | 16 | #include <linux/syscalls.h> | 
 | 17 | #include <linux/stackprotector.h> | 
 | 18 | #include <linux/string.h> | 
 | 19 | #include <linux/ctype.h> | 
 | 20 | #include <linux/delay.h> | 
 | 21 | #include <linux/ioport.h> | 
 | 22 | #include <linux/init.h> | 
 | 23 | #include <linux/initrd.h> | 
 | 24 | #include <linux/bootmem.h> | 
 | 25 | #include <linux/acpi.h> | 
 | 26 | #include <linux/tty.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/profile.h> | 
 | 35 | #include <linux/rcupdate.h> | 
 | 36 | #include <linux/moduleparam.h> | 
 | 37 | #include <linux/kallsyms.h> | 
 | 38 | #include <linux/writeback.h> | 
 | 39 | #include <linux/cpu.h> | 
 | 40 | #include <linux/cpuset.h> | 
 | 41 | #include <linux/cgroup.h> | 
 | 42 | #include <linux/efi.h> | 
 | 43 | #include <linux/tick.h> | 
 | 44 | #include <linux/interrupt.h> | 
 | 45 | #include <linux/taskstats_kern.h> | 
 | 46 | #include <linux/delayacct.h> | 
 | 47 | #include <linux/unistd.h> | 
 | 48 | #include <linux/rmap.h> | 
 | 49 | #include <linux/mempolicy.h> | 
 | 50 | #include <linux/key.h> | 
 | 51 | #include <linux/buffer_head.h> | 
 | 52 | #include <linux/page_cgroup.h> | 
 | 53 | #include <linux/debug_locks.h> | 
 | 54 | #include <linux/debugobjects.h> | 
 | 55 | #include <linux/lockdep.h> | 
 | 56 | #include <linux/kmemleak.h> | 
 | 57 | #include <linux/pid_namespace.h> | 
 | 58 | #include <linux/device.h> | 
 | 59 | #include <linux/kthread.h> | 
 | 60 | #include <linux/sched.h> | 
 | 61 | #include <linux/signal.h> | 
 | 62 | #include <linux/idr.h> | 
 | 63 | #include <linux/kgdb.h> | 
 | 64 | #include <linux/ftrace.h> | 
 | 65 | #include <linux/async.h> | 
 | 66 | #include <linux/kmemcheck.h> | 
 | 67 | #include <linux/sfi.h> | 
 | 68 | #include <linux/shmem_fs.h> | 
 | 69 | #include <linux/slab.h> | 
 | 70 | #include <linux/perf_event.h> | 
 | 71 | #include <linux/posix-timers.h> | 
 | 72 | #include <linux/random.h> | 
 | 73 |  | 
 | 74 | #include <asm/io.h> | 
 | 75 | #include <asm/bugs.h> | 
 | 76 | #include <asm/setup.h> | 
 | 77 | #include <asm/sections.h> | 
 | 78 | #include <asm/cacheflush.h> | 
 | 79 |  | 
 | 80 | #ifdef CONFIG_X86_LOCAL_APIC | 
 | 81 | #include <asm/smp.h> | 
 | 82 | #endif | 
 | 83 |  | 
 | 84 | static int kernel_init(void *); | 
 | 85 |  | 
 | 86 | extern void init_IRQ(void); | 
 | 87 | extern void fork_init(unsigned long); | 
 | 88 | extern void mca_init(void); | 
 | 89 | extern void sbus_init(void); | 
 | 90 | extern void prio_tree_init(void); | 
 | 91 | extern void radix_tree_init(void); | 
 | 92 | #ifndef CONFIG_DEBUG_RODATA | 
 | 93 | static inline void mark_rodata_ro(void) { } | 
 | 94 | #endif | 
 | 95 |  | 
 | 96 | #ifdef CONFIG_TC | 
 | 97 | extern void tc_init(void); | 
 | 98 | #endif | 
 | 99 |  | 
 | 100 | #ifdef CONFIG_SLOB | 
 | 101 | extern void slob_sizes_init(void); | 
 | 102 | #endif | 
 | 103 |  | 
 | 104 | /* | 
 | 105 |  * Debug helper: via this flag we know that we are in 'early bootup code' | 
 | 106 |  * where only the boot processor is running with IRQ disabled.  This means | 
 | 107 |  * two things - IRQ must not be enabled before the flag is cleared and some | 
 | 108 |  * operations which are not allowed with IRQ disabled are allowed while the | 
 | 109 |  * flag is set. | 
 | 110 |  */ | 
 | 111 | bool early_boot_irqs_disabled __read_mostly; | 
 | 112 |  | 
 | 113 | enum system_states system_state __read_mostly; | 
 | 114 | EXPORT_SYMBOL(system_state); | 
 | 115 | int KERNEL_START_END=0; | 
 | 116 | EXPORT_SYMBOL(KERNEL_START_END); | 
 | 117 |  | 
 | 118 | /* | 
 | 119 |  * Boot command-line arguments | 
 | 120 |  */ | 
 | 121 | #define MAX_INIT_ARGS CONFIG_INIT_ENV_ARG_LIMIT | 
 | 122 | #define MAX_INIT_ENVS CONFIG_INIT_ENV_ARG_LIMIT | 
 | 123 |  | 
 | 124 | extern void time_init(void); | 
 | 125 | /* Default late time init is NULL. archs can override this later. */ | 
 | 126 | void (*__initdata late_time_init)(void); | 
 | 127 | extern void softirq_init(void); | 
 | 128 |  | 
 | 129 | /* Untouched command line saved by arch-specific code. */ | 
 | 130 | char __initdata boot_command_line[COMMAND_LINE_SIZE]; | 
 | 131 | /* Untouched saved command line (eg. for /proc) */ | 
 | 132 | char *saved_command_line; | 
 | 133 | /* Command line for parameter parsing */ | 
 | 134 | static char *static_command_line; | 
 | 135 |  | 
 | 136 | static char *execute_command; | 
 | 137 | static char *ramdisk_execute_command; | 
 | 138 |  | 
 | 139 | /* | 
 | 140 |  * If set, this is an indication to the drivers that reset the underlying | 
 | 141 |  * device before going ahead with the initialization otherwise driver might | 
 | 142 |  * rely on the BIOS and skip the reset operation. | 
 | 143 |  * | 
 | 144 |  * This is useful if kernel is booting in an unreliable environment. | 
 | 145 |  * For ex. kdump situaiton where previous kernel has crashed, BIOS has been | 
 | 146 |  * skipped and devices will be in unknown state. | 
 | 147 |  */ | 
 | 148 | unsigned int reset_devices; | 
 | 149 | EXPORT_SYMBOL(reset_devices); | 
 | 150 |  | 
 | 151 | static int __init set_reset_devices(char *str) | 
 | 152 | { | 
 | 153 | 	reset_devices = 1; | 
 | 154 | 	return 1; | 
 | 155 | } | 
 | 156 |  | 
 | 157 | __setup("reset_devices", set_reset_devices); | 
 | 158 |  | 
 | 159 | static const char * argv_init[MAX_INIT_ARGS+2] = { "init", NULL, }; | 
 | 160 | const char * envp_init[MAX_INIT_ENVS+2] = { "HOME=/", "TERM=linux", NULL, }; | 
 | 161 | static const char *panic_later, *panic_param; | 
 | 162 |  | 
 | 163 | extern const struct obs_kernel_param __setup_start[], __setup_end[]; | 
 | 164 |  | 
 | 165 | static int __init obsolete_checksetup(char *line) | 
 | 166 | { | 
 | 167 | 	const struct obs_kernel_param *p; | 
 | 168 | 	int had_early_param = 0; | 
 | 169 |  | 
 | 170 | 	p = __setup_start; | 
 | 171 | 	do { | 
 | 172 | 		int n = strlen(p->str); | 
 | 173 | 		if (parameqn(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 |  | 
 | 191 | 	return had_early_param; | 
 | 192 | } | 
 | 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 |  | 
 | 200 | EXPORT_SYMBOL(loops_per_jiffy); | 
 | 201 |  | 
 | 202 | static int __init debug_kernel(char *str) | 
 | 203 | { | 
 | 204 | 	console_loglevel = 10; | 
 | 205 | 	return 0; | 
 | 206 | } | 
 | 207 |  | 
 | 208 | static int __init quiet_kernel(char *str) | 
 | 209 | { | 
 | 210 | 	console_loglevel = 4; | 
 | 211 | 	return 0; | 
 | 212 | } | 
 | 213 |  | 
 | 214 | early_param("debug", debug_kernel); | 
 | 215 | early_param("quiet", quiet_kernel); | 
 | 216 |  | 
 | 217 | static int __init loglevel(char *str) | 
 | 218 | { | 
 | 219 | 	int newlevel; | 
 | 220 |  | 
 | 221 | 	/* | 
 | 222 | 	 * Only update loglevel value when a correct setting was passed, | 
 | 223 | 	 * to prevent blind crashes (when loglevel being set to 0) that | 
 | 224 | 	 * are quite hard to debug | 
 | 225 | 	 */ | 
 | 226 | 	if (get_option(&str, &newlevel)) { | 
 | 227 | 		console_loglevel = newlevel; | 
 | 228 | 		return 0; | 
 | 229 | 	} | 
 | 230 |  | 
 | 231 | 	return -EINVAL; | 
 | 232 | } | 
 | 233 |  | 
 | 234 | early_param("loglevel", loglevel); | 
 | 235 |  | 
 | 236 | /* Change NUL term back to "=", to make "param" the whole string. */ | 
 | 237 | static int __init repair_env_string(char *param, char *val) | 
 | 238 | { | 
 | 239 | 	if (val) { | 
 | 240 | 		/* param=val or param="val"? */ | 
 | 241 | 		if (val == param+strlen(param)+1) | 
 | 242 | 			val[-1] = '='; | 
 | 243 | 		else if (val == param+strlen(param)+2) { | 
 | 244 | 			val[-2] = '='; | 
 | 245 | 			memmove(val-1, val, strlen(val)+1); | 
 | 246 | 			val--; | 
 | 247 | 		} else | 
 | 248 | 			BUG(); | 
 | 249 | 	} | 
 | 250 | 	return 0; | 
 | 251 | } | 
 | 252 |  | 
 | 253 | /* | 
 | 254 |  * Unknown boot options get handed to init, unless they look like | 
 | 255 |  * unused parameters (modprobe will find them in /proc/cmdline). | 
 | 256 |  */ | 
 | 257 | static int __init unknown_bootoption(char *param, char *val) | 
 | 258 | { | 
 | 259 | 	repair_env_string(param, val); | 
 | 260 |  | 
 | 261 | 	/* Handle obsolete-style parameters */ | 
 | 262 | 	if (obsolete_checksetup(param)) | 
 | 263 | 		return 0; | 
 | 264 |  | 
 | 265 | 	/* Unused module parameter. */ | 
 | 266 | 	if (strchr(param, '.') && (!val || strchr(param, '.') < val)) | 
 | 267 | 		return 0; | 
 | 268 |  | 
 | 269 | 	if (panic_later) | 
 | 270 | 		return 0; | 
 | 271 |  | 
 | 272 | 	if (val) { | 
 | 273 | 		/* Environment option */ | 
 | 274 | 		unsigned int i; | 
 | 275 | 		for (i = 0; envp_init[i]; i++) { | 
 | 276 | 			if (i == MAX_INIT_ENVS) { | 
 | 277 | 				panic_later = "Too many boot env vars at `%s'"; | 
 | 278 | 				panic_param = param; | 
 | 279 | 			} | 
 | 280 | 			if (!strncmp(param, envp_init[i], val - param)) | 
 | 281 | 				break; | 
 | 282 | 		} | 
 | 283 | 		envp_init[i] = param; | 
 | 284 | 	} else { | 
 | 285 | 		/* Command line option */ | 
 | 286 | 		unsigned int i; | 
 | 287 | 		for (i = 0; argv_init[i]; i++) { | 
 | 288 | 			if (i == MAX_INIT_ARGS) { | 
 | 289 | 				panic_later = "Too many boot init vars at `%s'"; | 
 | 290 | 				panic_param = param; | 
 | 291 | 			} | 
 | 292 | 		} | 
 | 293 | 		argv_init[i] = param; | 
 | 294 | 	} | 
 | 295 | 	return 0; | 
 | 296 | } | 
 | 297 |  | 
 | 298 | static int __init init_setup(char *str) | 
 | 299 | { | 
 | 300 | 	unsigned int i; | 
 | 301 |  | 
 | 302 | 	execute_command = str; | 
 | 303 | 	/* | 
 | 304 | 	 * In case LILO is going to boot us with default command line, | 
 | 305 | 	 * it prepends "auto" before the whole cmdline which makes | 
 | 306 | 	 * the shell think it should execute a script with such name. | 
 | 307 | 	 * So we ignore all arguments entered _before_ init=... [MJ] | 
 | 308 | 	 */ | 
 | 309 | 	for (i = 1; i < MAX_INIT_ARGS; i++) | 
 | 310 | 		argv_init[i] = NULL; | 
 | 311 | 	return 1; | 
 | 312 | } | 
 | 313 | __setup("init=", init_setup); | 
 | 314 |  | 
 | 315 | static int __init rdinit_setup(char *str) | 
 | 316 | { | 
 | 317 | 	unsigned int i; | 
 | 318 |  | 
 | 319 | 	ramdisk_execute_command = str; | 
 | 320 | 	/* See "auto" comment in init_setup */ | 
 | 321 | 	for (i = 1; i < MAX_INIT_ARGS; i++) | 
 | 322 | 		argv_init[i] = NULL; | 
 | 323 | 	return 1; | 
 | 324 | } | 
 | 325 | __setup("rdinit=", rdinit_setup); | 
 | 326 |  | 
 | 327 | #ifndef CONFIG_SMP | 
 | 328 | static const unsigned int setup_max_cpus = NR_CPUS; | 
 | 329 | #ifdef CONFIG_X86_LOCAL_APIC | 
 | 330 | static void __init smp_init(void) | 
 | 331 | { | 
 | 332 | 	APIC_init_uniprocessor(); | 
 | 333 | } | 
 | 334 | #else | 
 | 335 | #define smp_init()	do { } while (0) | 
 | 336 | #endif | 
 | 337 |  | 
 | 338 | static inline void setup_nr_cpu_ids(void) { } | 
 | 339 | static inline void smp_prepare_cpus(unsigned int maxcpus) { } | 
 | 340 | #endif | 
 | 341 |  | 
 | 342 | /* | 
 | 343 |  * We need to store the untouched command line for future reference. | 
 | 344 |  * We also need to store the touched command line since the parameter | 
 | 345 |  * parsing is performed in place, and we should allow a component to | 
 | 346 |  * store reference of name/value for future reference. | 
 | 347 |  */ | 
 | 348 | static void __init setup_command_line(char *command_line) | 
 | 349 | { | 
 | 350 | 	saved_command_line = alloc_bootmem(strlen (boot_command_line)+1); | 
 | 351 | 	static_command_line = alloc_bootmem(strlen (command_line)+1); | 
 | 352 | 	strcpy (saved_command_line, boot_command_line); | 
 | 353 | 	strcpy (static_command_line, command_line); | 
 | 354 | } | 
 | 355 |  | 
 | 356 | /* | 
 | 357 |  * We need to finalize in a non-__init function or else race conditions | 
 | 358 |  * between the root thread and the init thread may cause start_kernel to | 
 | 359 |  * be reaped by free_initmem before the root thread has proceeded to | 
 | 360 |  * cpu_idle. | 
 | 361 |  * | 
 | 362 |  * gcc-3.4 accidentally inlines this function, so use noinline. | 
 | 363 |  */ | 
 | 364 |  | 
 | 365 | static __initdata DECLARE_COMPLETION(kthreadd_done); | 
 | 366 |  | 
 | 367 | static noinline void __init_refok rest_init(void) | 
 | 368 | { | 
 | 369 | 	int pid; | 
 | 370 |  | 
 | 371 | 	rcu_scheduler_starting(); | 
 | 372 | 	/* | 
 | 373 | 	 * We need to spawn init first so that it obtains pid 1, however | 
 | 374 | 	 * the init task will end up wanting to create kthreads, which, if | 
 | 375 | 	 * we schedule it before we create kthreadd, will OOPS. | 
 | 376 | 	 */ | 
 | 377 | 	kernel_thread(kernel_init, NULL, CLONE_FS | CLONE_SIGHAND); | 
 | 378 | 	numa_default_policy(); | 
 | 379 | 	pid = kernel_thread(kthreadd, NULL, CLONE_FS | CLONE_FILES); | 
 | 380 | 	rcu_read_lock(); | 
 | 381 | 	kthreadd_task = find_task_by_pid_ns(pid, &init_pid_ns); | 
 | 382 | 	rcu_read_unlock(); | 
 | 383 | 	complete(&kthreadd_done); | 
 | 384 |  | 
 | 385 | 	/* | 
 | 386 | 	 * The boot idle thread must execute schedule() | 
 | 387 | 	 * at least once to get things moving: | 
 | 388 | 	 */ | 
 | 389 | 	init_idle_bootup_task(current); | 
 | 390 | 	schedule_preempt_disabled(); | 
 | 391 | 	/* Call into cpu_idle with preempt disabled */ | 
 | 392 | 	cpu_idle(); | 
 | 393 | } | 
 | 394 |  | 
 | 395 | /* Check for early params. */ | 
 | 396 | static int __init do_early_param(char *param, char *val) | 
 | 397 | { | 
 | 398 | 	const struct obs_kernel_param *p; | 
 | 399 |  | 
 | 400 | 	for (p = __setup_start; p < __setup_end; p++) { | 
 | 401 | 		if ((p->early && parameq(param, p->str)) || | 
 | 402 | 		    (strcmp(param, "console") == 0 && | 
 | 403 | 		     strcmp(p->str, "earlycon") == 0) | 
 | 404 | 		) { | 
 | 405 | 			if (p->setup_func(val) != 0) | 
 | 406 | 				printk(KERN_WARNING | 
 | 407 | 				       "Malformed early option '%s'\n", param); | 
 | 408 | 		} | 
 | 409 | 	} | 
 | 410 | 	/* We accept everything at this stage. */ | 
 | 411 | 	return 0; | 
 | 412 | } | 
 | 413 |  | 
 | 414 | void __init parse_early_options(char *cmdline) | 
 | 415 | { | 
 | 416 | 	parse_args("early options", cmdline, NULL, 0, 0, 0, do_early_param); | 
 | 417 | } | 
 | 418 |  | 
 | 419 | /* Arch code calls this early on, or if not, just before other parsing. */ | 
 | 420 | void __init parse_early_param(void) | 
 | 421 | { | 
 | 422 | 	static __initdata int done = 0; | 
 | 423 | 	static __initdata char tmp_cmdline[COMMAND_LINE_SIZE]; | 
 | 424 |  | 
 | 425 | 	if (done) | 
 | 426 | 		return; | 
 | 427 |  | 
 | 428 | 	/* All fall through to do_early_param. */ | 
 | 429 | 	strlcpy(tmp_cmdline, boot_command_line, COMMAND_LINE_SIZE); | 
 | 430 | 	parse_early_options(tmp_cmdline); | 
 | 431 | 	done = 1; | 
 | 432 | } | 
 | 433 |  | 
 | 434 | /* | 
 | 435 |  *	Activate the first processor. | 
 | 436 |  */ | 
 | 437 |  | 
 | 438 | static void __init boot_cpu_init(void) | 
 | 439 | { | 
 | 440 | 	int cpu = smp_processor_id(); | 
 | 441 | 	/* Mark the boot cpu "present", "online" etc for SMP and UP case */ | 
 | 442 | 	set_cpu_online(cpu, true); | 
 | 443 | 	set_cpu_active(cpu, true); | 
 | 444 | 	set_cpu_present(cpu, true); | 
 | 445 | 	set_cpu_possible(cpu, true); | 
 | 446 | } | 
 | 447 |  | 
 | 448 | void __init __weak smp_setup_processor_id(void) | 
 | 449 | { | 
 | 450 | } | 
 | 451 |  | 
 | 452 | void __init __weak thread_info_cache_init(void) | 
 | 453 | { | 
 | 454 | } | 
 | 455 |  | 
 | 456 | /* | 
 | 457 |  * Set up kernel memory allocators | 
 | 458 |  */ | 
 | 459 | static void __init mm_init(void) | 
 | 460 | { | 
 | 461 | 	/* | 
 | 462 | 	 * page_cgroup requires contiguous pages, | 
 | 463 | 	 * bigger than MAX_ORDER unless SPARSEMEM. | 
 | 464 | 	 */ | 
 | 465 | 	page_cgroup_init_flatmem(); | 
 | 466 | 	mem_init(); | 
 | 467 | 	kmem_cache_init(); | 
 | 468 | 	percpu_init_late(); | 
 | 469 | 	pgtable_cache_init(); | 
 | 470 | 	vmalloc_init(); | 
 | 471 | } | 
 | 472 |  | 
 | 473 | static void current_kernel_init(void) | 
 | 474 | { | 
 | 475 | #ifdef CONFIG_STACK_SIZE | 
 | 476 | 	current_kernel_thread = &init_thread_info; | 
 | 477 | #endif | 
 | 478 | } | 
 | 479 |  | 
 | 480 | asmlinkage void __init start_kernel(void) | 
 | 481 | { | 
 | 482 |  | 
 | 483 | 	char * command_line; | 
 | 484 | 	extern const struct kernel_param __start___param[], __stop___param[]; | 
 | 485 |  | 
 | 486 | #ifdef CONFIG_SLOB | 
 | 487 | 	slob_sizes_init(); | 
 | 488 | #endif | 
 | 489 |  | 
 | 490 | 	/* | 
 | 491 | 	 * Need to run as early as possible, to initialize the | 
 | 492 | 	 * lockdep hash: | 
 | 493 | 	 */ | 
 | 494 | 	current_kernel_init(); | 
 | 495 |  | 
 | 496 | 	lockdep_init(); | 
 | 497 | 	smp_setup_processor_id(); | 
 | 498 | 	debug_objects_early_init(); | 
 | 499 |  | 
 | 500 | 	/* | 
 | 501 | 	 * Set up the the initial canary ASAP: | 
 | 502 | 	 */ | 
 | 503 | 	boot_init_stack_canary(); | 
 | 504 |  | 
 | 505 | 	cgroup_init_early(); | 
 | 506 |  | 
 | 507 | 	local_irq_disable(); | 
 | 508 | 	early_boot_irqs_disabled = true; | 
 | 509 |  | 
 | 510 |  | 
 | 511 |  | 
 | 512 | /* | 
 | 513 |  * Interrupts are still disabled. Do necessary setups, then | 
 | 514 |  * enable them | 
 | 515 |  */ | 
 | 516 | 	tick_init(); | 
 | 517 | 	boot_cpu_init(); | 
 | 518 | 	page_address_init(); | 
 | 519 | 	/* printk(KERN_NOTICE "%s", linux_banner); */ | 
 | 520 | 	setup_arch(&command_line); | 
 | 521 | 	mm_init_owner(&init_mm, &init_task); | 
 | 522 | 	mm_init_cpumask(&init_mm); | 
 | 523 | 	setup_command_line(command_line); | 
 | 524 | 	setup_nr_cpu_ids(); | 
 | 525 | 	setup_per_cpu_areas(); | 
 | 526 | 	softirq_early_init(); | 
 | 527 | 	smp_prepare_boot_cpu();	/* arch-specific boot-cpu hooks */ | 
 | 528 |  | 
 | 529 | 	build_all_zonelists(NULL); | 
 | 530 | 	page_alloc_init(); | 
 | 531 |  | 
 | 532 | 	/* printk(KERN_NOTICE "Kernel command line: %s\n", boot_command_line); */ | 
 | 533 | 	parse_early_param(); | 
 | 534 | 	parse_args("Booting kernel", static_command_line, __start___param, | 
 | 535 | 		   __stop___param - __start___param, | 
 | 536 | 		   -1, -1, &unknown_bootoption); | 
 | 537 |  | 
 | 538 | 	jump_label_init(); | 
 | 539 |  | 
 | 540 | 	/* | 
 | 541 | 	 * These use large bootmem allocations and must precede | 
 | 542 | 	 * kmem_cache_init() | 
 | 543 | 	 */ | 
 | 544 | 	setup_log_buf(0); | 
 | 545 | 	pidhash_init(); | 
 | 546 | 	vfs_caches_init_early(); | 
 | 547 | 	sort_main_extable(); | 
 | 548 | 	trap_init(); | 
 | 549 | 	mm_init(); | 
 | 550 |  | 
 | 551 | 	/* | 
 | 552 | 	 * Set up the scheduler prior starting any interrupts (such as the | 
 | 553 | 	 * timer interrupt). Full topology setup happens at smp_init() | 
 | 554 | 	 * time - but meanwhile we still have a functioning scheduler. | 
 | 555 | 	 */ | 
 | 556 | 	sched_init(); | 
 | 557 | 	/* | 
 | 558 | 	 * Disable preemption - early bootup scheduling is extremely | 
 | 559 | 	 * fragile until we cpu_idle() for the first time. | 
 | 560 | 	 */ | 
 | 561 | 	preempt_disable(); | 
 | 562 | 	if (!irqs_disabled()) { | 
 | 563 | 		printk(KERN_WARNING "start_kernel(): bug: interrupts were " | 
 | 564 | 				"enabled *very* early, fixing it\n"); | 
 | 565 | 		local_irq_disable(); | 
 | 566 | 	} | 
 | 567 | 	idr_init_cache(); | 
 | 568 | 	perf_event_init(); | 
 | 569 | 	rcu_init(); | 
 | 570 | 	radix_tree_init(); | 
 | 571 | 	/* init some links before init_ISA_irqs() */ | 
 | 572 | 	early_irq_init(); | 
 | 573 | 	init_IRQ(); | 
 | 574 | 	prio_tree_init(); | 
 | 575 | 	init_timers(); | 
 | 576 | 	hrtimers_init(); | 
 | 577 | 	softirq_init(); | 
 | 578 | 	timekeeping_init(); | 
 | 579 | 	time_init(); | 
 | 580 | 	profile_init(); | 
 | 581 | 	call_function_init(); | 
 | 582 | 	if (!irqs_disabled()) | 
 | 583 | 		printk(KERN_CRIT "start_kernel(): bug: interrupts were " | 
 | 584 | 				 "enabled early\n"); | 
 | 585 | 	early_boot_irqs_disabled = false; | 
 | 586 | 	local_irq_enable(); | 
 | 587 |  | 
 | 588 | 	kmem_cache_init_late(); | 
 | 589 |  | 
 | 590 | 	/* | 
 | 591 | 	 * HACK ALERT! This is early. We're enabling the console before | 
 | 592 | 	 * we've done PCI setups etc, and console_init() must be aware of | 
 | 593 | 	 * this. But we do want output early, in case something goes wrong. | 
 | 594 | 	 */ | 
 | 595 | 	console_init(); | 
 | 596 | 	if (panic_later) | 
 | 597 | 		panic(panic_later, panic_param); | 
 | 598 |  | 
 | 599 | 	lockdep_info(); | 
 | 600 |  | 
 | 601 | 	/* | 
 | 602 | 	 * Need to run this when irqs are enabled, because it wants | 
 | 603 | 	 * to self-test [hard/soft]-irqs on/off lock inversion bugs | 
 | 604 | 	 * too: | 
 | 605 | 	 */ | 
 | 606 | 	locking_selftest(); | 
 | 607 |  | 
 | 608 | #ifdef CONFIG_BLK_DEV_INITRD | 
 | 609 | 	if (initrd_start && !initrd_below_start_ok && | 
 | 610 | 	    page_to_pfn(virt_to_page((void *)initrd_start)) < min_low_pfn) { | 
 | 611 | 		printk(KERN_CRIT "initrd overwritten (0x%08lx < 0x%08lx) - " | 
 | 612 | 		    "disabling it.\n", | 
 | 613 | 		    page_to_pfn(virt_to_page((void *)initrd_start)), | 
 | 614 | 		    min_low_pfn); | 
 | 615 | 		initrd_start = 0; | 
 | 616 | 	} | 
 | 617 | #endif | 
 | 618 | 	page_cgroup_init(); | 
 | 619 | 	debug_objects_mem_init(); | 
 | 620 | 	kmemleak_init(); | 
 | 621 | 	setup_per_cpu_pageset(); | 
 | 622 | 	numa_policy_init(); | 
 | 623 | 	if (late_time_init) | 
 | 624 | 		late_time_init(); | 
 | 625 | 	sched_clock_init(); | 
 | 626 | 	calibrate_delay(); | 
 | 627 | 	pidmap_init(); | 
 | 628 | 	anon_vma_init(); | 
 | 629 | #ifdef CONFIG_X86 | 
 | 630 | 	if (efi_enabled(EFI_RUNTIME_SERVICES)) | 
 | 631 | 		efi_enter_virtual_mode(); | 
 | 632 | #endif | 
 | 633 | #ifdef CONFIG_X86_ESPFIX64 | 
 | 634 | 	/* Should be run before the first non-init thread is created */ | 
 | 635 | 	init_espfix_bsp(); | 
 | 636 | #endif | 
 | 637 | 	thread_info_cache_init(); | 
 | 638 | 	cred_init(); | 
 | 639 | 	fork_init(totalram_pages); | 
 | 640 | 	proc_caches_init(); | 
 | 641 | 	buffer_init(); | 
 | 642 | 	key_init(); | 
 | 643 | 	security_init(); | 
 | 644 | 	dbg_late_init(); | 
 | 645 | 	vfs_caches_init(totalram_pages); | 
 | 646 | 	signals_init(); | 
 | 647 | 	/* rootfs populating might need page-writeback */ | 
 | 648 | 	page_writeback_init(); | 
 | 649 | #ifdef CONFIG_PROC_FS | 
 | 650 | 	proc_root_init(); | 
 | 651 | #endif | 
 | 652 | 	cgroup_init(); | 
 | 653 | 	cpuset_init(); | 
 | 654 | 	taskstats_init_early(); | 
 | 655 | 	delayacct_init(); | 
 | 656 |  | 
 | 657 | 	check_bugs(); | 
 | 658 |  | 
 | 659 | 	acpi_early_init(); /* before LAPIC and SMP init */ | 
 | 660 | 	sfi_init_late(); | 
 | 661 |  | 
 | 662 | 	if (efi_enabled(EFI_RUNTIME_SERVICES)) | 
 | 663 | 		efi_free_boot_services(); | 
 | 664 |  | 
 | 665 | 	ftrace_init(); | 
 | 666 |  | 
 | 667 | 	/* Do the rest non-__init'ed, we're now alive */ | 
 | 668 | 	rest_init(); | 
 | 669 | } | 
 | 670 |  | 
 | 671 | /* Call all constructor functions linked into the kernel. */ | 
 | 672 | static void __init do_ctors(void) | 
 | 673 | { | 
 | 674 | #ifdef CONFIG_CONSTRUCTORS | 
 | 675 | 	ctor_fn_t *fn = (ctor_fn_t *) __ctors_start; | 
 | 676 |  | 
 | 677 | 	for (; fn < (ctor_fn_t *) __ctors_end; fn++) | 
 | 678 | 		(*fn)(); | 
 | 679 | #endif | 
 | 680 | } | 
 | 681 |  | 
 | 682 | bool initcall_debug; | 
 | 683 | core_param(initcall_debug, initcall_debug, bool, 0644); | 
 | 684 |  | 
 | 685 | static char msgbuf[64]; | 
 | 686 |  | 
 | 687 | static int __init_or_module do_one_initcall_debug(initcall_t fn) | 
 | 688 | { | 
 | 689 | 	ktime_t calltime, delta, rettime; | 
 | 690 | 	unsigned long long duration; | 
 | 691 | 	int ret; | 
 | 692 |  | 
 | 693 | 	printk(KERN_DEBUG "calling  %pF @ %i\n", fn, task_pid_nr(current)); | 
 | 694 | 	calltime = ktime_get(); | 
 | 695 | 	ret = fn(); | 
 | 696 | 	rettime = ktime_get(); | 
 | 697 | 	delta = ktime_sub(rettime, calltime); | 
 | 698 | 	duration = (unsigned long long) ktime_to_ns(delta) >> 10; | 
 | 699 | 	printk(KERN_DEBUG "initcall %pF returned %d after %lld usecs\n", fn, | 
 | 700 | 		ret, duration); | 
 | 701 |  | 
 | 702 | 	return ret; | 
 | 703 | } | 
 | 704 |  | 
 | 705 | int __init_or_module do_one_initcall(initcall_t fn) | 
 | 706 | { | 
 | 707 | 	int count = preempt_count(); | 
 | 708 | 	int ret; | 
 | 709 |  | 
 | 710 | 	if (initcall_debug) | 
 | 711 | 		ret = do_one_initcall_debug(fn); | 
 | 712 | 	else | 
 | 713 | 		ret = fn(); | 
 | 714 |  | 
 | 715 | 	msgbuf[0] = 0; | 
 | 716 |  | 
 | 717 | 	if (ret && ret != -ENODEV && initcall_debug) | 
 | 718 | 		sprintf(msgbuf, "error code %d ", ret); | 
 | 719 |  | 
 | 720 | 	if (preempt_count() != count) { | 
 | 721 | 		strlcat(msgbuf, "preemption imbalance ", sizeof(msgbuf)); | 
 | 722 | 		preempt_count() = count; | 
 | 723 | 	} | 
 | 724 | 	if (irqs_disabled()) { | 
 | 725 | 		strlcat(msgbuf, "disabled interrupts ", sizeof(msgbuf)); | 
 | 726 | 		local_irq_enable(); | 
 | 727 | 	} | 
 | 728 | 	if (msgbuf[0]) { | 
 | 729 | 		printk("initcall %pF returned with %s\n", fn, msgbuf); | 
 | 730 | 	} | 
 | 731 |  | 
 | 732 | 	return ret; | 
 | 733 | } | 
 | 734 |  | 
 | 735 |  | 
 | 736 | extern initcall_t __initcall_start[]; | 
 | 737 | extern initcall_t __initcall0_start[]; | 
 | 738 | extern initcall_t __initcall1_start[]; | 
 | 739 | extern initcall_t __initcall2_start[]; | 
 | 740 | extern initcall_t __initcall3_start[]; | 
 | 741 | extern initcall_t __initcall4_start[]; | 
 | 742 | extern initcall_t __initcall5_start[]; | 
 | 743 | extern initcall_t __initcall6_start[]; | 
 | 744 | extern initcall_t __initcall7_start[]; | 
 | 745 | extern initcall_t __initcall_end[]; | 
 | 746 |  | 
 | 747 | static initcall_t *initcall_levels[] __initdata = { | 
 | 748 | 	__initcall0_start, | 
 | 749 | 	__initcall1_start, | 
 | 750 | 	__initcall2_start, | 
 | 751 | 	__initcall3_start, | 
 | 752 | 	__initcall4_start, | 
 | 753 | 	__initcall5_start, | 
 | 754 | 	__initcall6_start, | 
 | 755 | 	__initcall7_start, | 
 | 756 | 	__initcall_end, | 
 | 757 | }; | 
 | 758 |  | 
 | 759 | static char *initcall_level_names[] __initdata = { | 
 | 760 | 	"early parameters", | 
 | 761 | 	"core parameters", | 
 | 762 | 	"postcore parameters", | 
 | 763 | 	"arch parameters", | 
 | 764 | 	"subsys parameters", | 
 | 765 | 	"fs parameters", | 
 | 766 | 	"device parameters", | 
 | 767 | 	"late parameters", | 
 | 768 | }; | 
 | 769 |  | 
 | 770 | static void __init do_initcall_level(int level) | 
 | 771 | { | 
 | 772 | 	extern const struct kernel_param __start___param[], __stop___param[]; | 
 | 773 | 	initcall_t *fn; | 
 | 774 |  | 
 | 775 | 	strcpy(static_command_line, saved_command_line); | 
 | 776 | 	parse_args(initcall_level_names[level], | 
 | 777 | 		   static_command_line, __start___param, | 
 | 778 | 		   __stop___param - __start___param, | 
 | 779 | 		   level, level, | 
 | 780 | 		   repair_env_string); | 
 | 781 |  | 
 | 782 | 	for (fn = initcall_levels[level]; fn < initcall_levels[level+1]; fn++) | 
 | 783 | 		do_one_initcall(*fn); | 
 | 784 | } | 
 | 785 |  | 
 | 786 | static void __init do_initcalls(void) | 
 | 787 | { | 
 | 788 | 	int level; | 
 | 789 |  | 
 | 790 | 	for (level = 0; level < ARRAY_SIZE(initcall_levels) - 1; level++) | 
 | 791 | 		do_initcall_level(level); | 
 | 792 | } | 
 | 793 |  | 
 | 794 | /* | 
 | 795 |  * Ok, the machine is now initialized. None of the devices | 
 | 796 |  * have been touched yet, but the CPU subsystem is up and | 
 | 797 |  * running, and memory and process management works. | 
 | 798 |  * | 
 | 799 |  * Now we can finally start doing some real work.. | 
 | 800 |  */ | 
 | 801 | static void __init do_basic_setup(void) | 
 | 802 | { | 
 | 803 | 	cpuset_init_smp(); | 
 | 804 | 	usermodehelper_init(); | 
 | 805 | 	shmem_init(); | 
 | 806 | 	driver_init(); | 
 | 807 | 	init_irq_proc(); | 
 | 808 | 	do_ctors(); | 
 | 809 | 	usermodehelper_enable(); | 
 | 810 | 	do_initcalls(); | 
 | 811 | 	random_int_secret_init(); | 
 | 812 | } | 
 | 813 |  | 
 | 814 | static void __init do_pre_smp_initcalls(void) | 
 | 815 | { | 
 | 816 | 	initcall_t *fn; | 
 | 817 |  | 
 | 818 | 	for (fn = __initcall_start; fn < __initcall0_start; fn++) | 
 | 819 | 		do_one_initcall(*fn); | 
 | 820 | } | 
 | 821 |  | 
 | 822 | static void run_init_process(const char *init_filename) | 
 | 823 | { | 
 | 824 | 	argv_init[0] = init_filename; | 
 | 825 | 	kernel_execve(init_filename, argv_init, envp_init); | 
 | 826 | } | 
 | 827 |  | 
 | 828 | /* This is a non __init function. Force it to be noinline otherwise gcc | 
 | 829 |  * makes it inline to init() and it becomes part of init.text section | 
 | 830 |  */ | 
 | 831 | static noinline int init_post(void) | 
 | 832 | { | 
 | 833 | 	/* need to finish all async __init code before freeing the memory */ | 
 | 834 | 	async_synchronize_full(); | 
 | 835 | 	free_initmem(); | 
 | 836 | 	mark_rodata_ro(); | 
 | 837 | 	system_state = SYSTEM_RUNNING; | 
 | 838 | 	numa_default_policy(); | 
 | 839 |  | 
 | 840 |          KERNEL_START_END=1; | 
 | 841 |  | 
 | 842 |  | 
 | 843 | 	current->signal->flags |= SIGNAL_UNKILLABLE; | 
 | 844 |  | 
 | 845 | 	if (ramdisk_execute_command) { | 
 | 846 | 		run_init_process(ramdisk_execute_command); | 
 | 847 | 		printk(KERN_WARNING "Failed to execute %s\n", | 
 | 848 | 				ramdisk_execute_command); | 
 | 849 | 	} | 
 | 850 |  | 
 | 851 | 	/* | 
 | 852 | 	 * We try each of these until one succeeds. | 
 | 853 | 	 * | 
 | 854 | 	 * The Bourne shell can be used instead of init if we are | 
 | 855 | 	 * trying to recover a really broken machine. | 
 | 856 | 	 */ | 
 | 857 | 	if (execute_command) { | 
 | 858 | 		run_init_process(execute_command); | 
 | 859 | 		printk(KERN_WARNING "Failed to execute %s.  Attempting " | 
 | 860 | 					"defaults...\n", execute_command); | 
 | 861 | 	} | 
 | 862 | #ifdef CONFIG_SYSTEM_RECOVERY | 
 | 863 | 	run_init_process("/recovery/sbin/init"); | 
 | 864 | #else | 
 | 865 | 	run_init_process("/sbin/init"); | 
 | 866 | 	run_init_process("/etc/init"); | 
 | 867 | 	run_init_process("/bin/init"); | 
 | 868 | 	run_init_process("/bin/sh"); | 
 | 869 | #endif | 
 | 870 | 	panic("No init found.  Try passing init= option to kernel. " | 
 | 871 | 	      "See Linux Documentation/init.txt for guidance."); | 
 | 872 | } | 
 | 873 |  | 
 | 874 |  | 
 | 875 | /*/PSBUFµÄÐéÄâµØÖ·ÓëÎïÀíµØÖ·µÄÓ³Éä¹ØÏµ | 
 | 876 | unsigned long virtaddr_to_phys(unsigned long virt) | 
 | 877 | { | 
 | 878 | 	if(virt >= (unsigned long)vir_addr_ddrnet && virt <= ((unsigned long)vir_addr_ddrnet+10*1024*1024))  | 
 | 879 | 	{ | 
 | 880 | 		return 0x27400000 + (virt - (unsigned long)vir_addr_ddrnet); | 
 | 881 | 	} | 
 | 882 | 	return  __pa(virt); | 
 | 883 | } | 
 | 884 | unsigned long physaddr_to_virt(unsigned long phys) | 
 | 885 | { | 
 | 886 | 	if(phys >= 0x27400000 && phys <= (0x27400000 + 10*1024*1024))  | 
 | 887 | 		return (unsigned long)vir_addr_ddrnet + (phys - 0x27400000); | 
 | 888 | 	return __va(phys); | 
 | 889 | } | 
 | 890 | */ | 
 | 891 | static int __init kernel_init(void * unused) | 
 | 892 | { | 
 | 893 | 	/*  | 
 | 894 | 	 * cp ps_buf mmap, for temp test. | 
 | 895 | 	 * add by jkz,2016-01-07. | 
 | 896 | 	 */ | 
 | 897 | 	//int *vir_addr = NULL; | 
 | 898 |  | 
 | 899 | 	/* | 
 | 900 | 	 * Wait until kthreadd is all set-up. | 
 | 901 | 	 */ | 
 | 902 | 	wait_for_completion(&kthreadd_done); | 
 | 903 |  | 
 | 904 | 	/* Now the scheduler is fully set up and can do blocking allocations */ | 
 | 905 | 	gfp_allowed_mask = __GFP_BITS_MASK; | 
 | 906 |  | 
 | 907 | 	/* | 
 | 908 | 	 * init can allocate pages on any node | 
 | 909 | 	 */ | 
 | 910 | 	set_mems_allowed(node_states[N_HIGH_MEMORY]); | 
 | 911 | 	/* | 
 | 912 | 	 * init can run on any cpu. | 
 | 913 | 	 */ | 
 | 914 | 	set_cpus_allowed_ptr(current, cpu_all_mask); | 
 | 915 |  | 
 | 916 | 	cad_pid = task_pid(current); | 
 | 917 |  | 
 | 918 | 	smp_prepare_cpus(setup_max_cpus); | 
 | 919 |  | 
 | 920 | 	do_pre_smp_initcalls(); | 
 | 921 | 	lockup_detector_init(); | 
 | 922 |  | 
 | 923 | 	smp_init(); | 
 | 924 | 	sched_init_smp(); | 
 | 925 |  | 
 | 926 | 	do_basic_setup(); | 
 | 927 |  | 
 | 928 | 	/*  | 
 | 929 | 	 * cp ps_buf mmap, for temp test. | 
 | 930 | 	 * add by jkz,2016-01-07. | 
 | 931 | 	 */ | 
 | 932 |  | 
 | 933 | 	/* Open the /dev/console on the rootfs, this should never fail */ | 
 | 934 | 	if (sys_open((const char __user *) "/dev/console", O_RDWR, 0) < 0) | 
 | 935 | 		printk(KERN_WARNING "Warning: unable to open an initial console.\n"); | 
 | 936 |  | 
 | 937 | 	(void) sys_dup(0); | 
 | 938 | 	(void) sys_dup(0); | 
 | 939 | 	/* | 
 | 940 | 	 * check if there is an early userspace init.  If yes, let it do all | 
 | 941 | 	 * the work | 
 | 942 | 	 */ | 
 | 943 |  | 
 | 944 | 	if (!ramdisk_execute_command) | 
 | 945 | 		ramdisk_execute_command = "/init"; | 
 | 946 |  | 
 | 947 | 	if (sys_access((const char __user *) ramdisk_execute_command, 0) != 0) { | 
 | 948 | 		ramdisk_execute_command = NULL; | 
 | 949 | 		prepare_namespace(); | 
 | 950 | 	} | 
 | 951 |  | 
 | 952 | 	/* | 
 | 953 | 	 * Ok, we have completed the initial bootup, and | 
 | 954 | 	 * we're essentially up and running. Get rid of the | 
 | 955 | 	 * initmem segments and start the user-mode stuff.. | 
 | 956 | 	 */ | 
 | 957 |  | 
 | 958 | 	init_post(); | 
 | 959 | 	return 0; | 
 | 960 | } |