fork.c 55.5 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
/*
 *  linux/kernel/fork.c
 *
 *  Copyright (C) 1991, 1992  Linus Torvalds
 */

/*
 *  'fork.c' contains the help-routines for the 'fork' system call
 * (see also entry.S and others).
 * Fork is rather simple, once you get the hang of it, but the memory
 * management can be a bitch. See 'mm/memory.c': 'copy_page_range()'
 */

#include <linux/slab.h>
#include <linux/init.h>
#include <linux/unistd.h>
#include <linux/module.h>
#include <linux/vmalloc.h>
#include <linux/completion.h>
#include <linux/personality.h>
#include <linux/mempolicy.h>
#include <linux/sem.h>
#include <linux/file.h>
A
Al Viro 已提交
24
#include <linux/fdtable.h>
25
#include <linux/iocontext.h>
L
Linus Torvalds 已提交
26 27 28
#include <linux/key.h>
#include <linux/binfmts.h>
#include <linux/mman.h>
A
Andrea Arcangeli 已提交
29
#include <linux/mmu_notifier.h>
L
Linus Torvalds 已提交
30
#include <linux/fs.h>
D
Davidlohr Bueso 已提交
31 32
#include <linux/mm.h>
#include <linux/vmacache.h>
S
Serge E. Hallyn 已提交
33
#include <linux/nsproxy.h>
34
#include <linux/capability.h>
L
Linus Torvalds 已提交
35
#include <linux/cpu.h>
36
#include <linux/cgroup.h>
L
Linus Torvalds 已提交
37
#include <linux/security.h>
38
#include <linux/hugetlb.h>
39
#include <linux/seccomp.h>
L
Linus Torvalds 已提交
40 41 42 43
#include <linux/swap.h>
#include <linux/syscalls.h>
#include <linux/jiffies.h>
#include <linux/futex.h>
44
#include <linux/compat.h>
45
#include <linux/kthread.h>
46
#include <linux/task_io_accounting_ops.h>
47
#include <linux/rcupdate.h>
L
Linus Torvalds 已提交
48 49 50
#include <linux/ptrace.h>
#include <linux/mount.h>
#include <linux/audit.h>
51
#include <linux/memcontrol.h>
52
#include <linux/ftrace.h>
53
#include <linux/proc_fs.h>
L
Linus Torvalds 已提交
54 55
#include <linux/profile.h>
#include <linux/rmap.h>
H
Hugh Dickins 已提交
56
#include <linux/ksm.h>
L
Linus Torvalds 已提交
57
#include <linux/acct.h>
58
#include <linux/tsacct_kern.h>
M
Matt Helsley 已提交
59
#include <linux/cn_proc.h>
R
Rafael J. Wysocki 已提交
60
#include <linux/freezer.h>
61
#include <linux/delayacct.h>
62
#include <linux/taskstats_kern.h>
63
#include <linux/random.h>
M
Miloslav Trmac 已提交
64
#include <linux/tty.h>
65
#include <linux/blkdev.h>
66
#include <linux/fs_struct.h>
67
#include <linux/magic.h>
68
#include <linux/perf_event.h>
69
#include <linux/posix-timers.h>
70
#include <linux/user-return-notifier.h>
Y
Ying Han 已提交
71
#include <linux/oom.h>
A
Andrea Arcangeli 已提交
72
#include <linux/khugepaged.h>
73
#include <linux/signalfd.h>
74
#include <linux/uprobes.h>
75
#include <linux/aio.h>
76
#include <linux/compiler.h>
77
#include <linux/sysctl.h>
D
Dmitry Vyukov 已提交
78
#include <linux/kcov.h>
L
Linus Torvalds 已提交
79 80 81 82 83 84 85 86

#include <asm/pgtable.h>
#include <asm/pgalloc.h>
#include <asm/uaccess.h>
#include <asm/mmu_context.h>
#include <asm/cacheflush.h>
#include <asm/tlbflush.h>

87 88
#include <trace/events/sched.h>

89 90 91
#define CREATE_TRACE_POINTS
#include <trace/events/task.h>

92 93 94 95 96 97 98 99 100 101
/*
 * Minimum number of threads to boot the kernel
 */
#define MIN_THREADS 20

/*
 * Maximum number of threads
 */
#define MAX_THREADS FUTEX_TID_MASK

L
Linus Torvalds 已提交
102 103 104 105
/*
 * Protected counters by write_lock_irq(&tasklist_lock)
 */
unsigned long total_forks;	/* Handle normal Linux uptimes. */
106
int nr_threads;			/* The idle threads do not count.. */
L
Linus Torvalds 已提交
107 108 109 110 111

int max_threads;		/* tunable limit on nr_threads */

DEFINE_PER_CPU(unsigned long, process_counts) = 0;

112
__cacheline_aligned DEFINE_RWLOCK(tasklist_lock);  /* outer */
113 114 115 116 117 118 119 120

#ifdef CONFIG_PROVE_RCU
int lockdep_tasklist_lock_is_held(void)
{
	return lockdep_is_held(&tasklist_lock);
}
EXPORT_SYMBOL_GPL(lockdep_tasklist_lock_is_held);
#endif /* #ifdef CONFIG_PROVE_RCU */
L
Linus Torvalds 已提交
121 122 123 124 125 126

int nr_processes(void)
{
	int cpu;
	int total = 0;

127
	for_each_possible_cpu(cpu)
L
Linus Torvalds 已提交
128 129 130 131 132
		total += per_cpu(process_counts, cpu);

	return total;
}

133 134 135 136
void __weak arch_release_task_struct(struct task_struct *tsk)
{
}

137
#ifndef CONFIG_ARCH_TASK_STRUCT_ALLOCATOR
138
static struct kmem_cache *task_struct_cachep;
139 140 141 142 143 144 145 146 147 148

static inline struct task_struct *alloc_task_struct_node(int node)
{
	return kmem_cache_alloc_node(task_struct_cachep, GFP_KERNEL, node);
}

static inline void free_task_struct(struct task_struct *tsk)
{
	kmem_cache_free(task_struct_cachep, tsk);
}
L
Linus Torvalds 已提交
149 150
#endif

151
void __weak arch_release_thread_stack(unsigned long *stack)
152 153 154
{
}

155
#ifndef CONFIG_ARCH_THREAD_STACK_ALLOCATOR
156

157 158 159 160
/*
 * Allocate pages if THREAD_SIZE is >= PAGE_SIZE, otherwise use a
 * kmemcache based allocator.
 */
161
# if THREAD_SIZE >= PAGE_SIZE || defined(CONFIG_VMAP_STACK)
162 163 164 165 166 167 168 169 170 171

#ifdef CONFIG_VMAP_STACK
/*
 * vmalloc() is a bit slow, and calling vfree() enough times will force a TLB
 * flush.  Try to minimize the number of calls by caching stacks.
 */
#define NR_CACHED_STACKS 2
static DEFINE_PER_CPU(struct vm_struct *, cached_stacks[NR_CACHED_STACKS]);
#endif

172
static unsigned long *alloc_thread_stack_node(struct task_struct *tsk, int node)
173
{
174
#ifdef CONFIG_VMAP_STACK
175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196
	void *stack;
	int i;

	local_irq_disable();
	for (i = 0; i < NR_CACHED_STACKS; i++) {
		struct vm_struct *s = this_cpu_read(cached_stacks[i]);

		if (!s)
			continue;
		this_cpu_write(cached_stacks[i], NULL);

		tsk->stack_vm_area = s;
		local_irq_enable();
		return s->addr;
	}
	local_irq_enable();

	stack = __vmalloc_node_range(THREAD_SIZE, THREAD_SIZE,
				     VMALLOC_START, VMALLOC_END,
				     THREADINFO_GFP | __GFP_HIGHMEM,
				     PAGE_KERNEL,
				     0, node, __builtin_return_address(0));
197 198 199 200 201 202 203 204 205 206

	/*
	 * We can't call find_vm_area() in interrupt context, and
	 * free_thread_stack() can be called in interrupt context,
	 * so cache the vm_struct.
	 */
	if (stack)
		tsk->stack_vm_area = find_vm_area(stack);
	return stack;
#else
207 208
	struct page *page = alloc_pages_node(node, THREADINFO_GFP,
					     THREAD_SIZE_ORDER);
209 210

	return page ? page_address(page) : NULL;
211
#endif
212 213
}

214
static inline void free_thread_stack(struct task_struct *tsk)
215
{
216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231
#ifdef CONFIG_VMAP_STACK
	if (task_stack_vm_area(tsk)) {
		unsigned long flags;
		int i;

		local_irq_save(flags);
		for (i = 0; i < NR_CACHED_STACKS; i++) {
			if (this_cpu_read(cached_stacks[i]))
				continue;

			this_cpu_write(cached_stacks[i], tsk->stack_vm_area);
			local_irq_restore(flags);
			return;
		}
		local_irq_restore(flags);

232
		vfree(tsk->stack);
233 234 235 236 237
		return;
	}
#endif

	__free_pages(virt_to_page(tsk->stack), THREAD_SIZE_ORDER);
238
}
239
# else
240
static struct kmem_cache *thread_stack_cache;
241

242
static unsigned long *alloc_thread_stack_node(struct task_struct *tsk,
243 244
						  int node)
{
245
	return kmem_cache_alloc_node(thread_stack_cache, THREADINFO_GFP, node);
246 247
}

248
static void free_thread_stack(struct task_struct *tsk)
249
{
250
	kmem_cache_free(thread_stack_cache, tsk->stack);
251 252
}

253
void thread_stack_cache_init(void)
254
{
255
	thread_stack_cache = kmem_cache_create("thread_stack", THREAD_SIZE,
256
					      THREAD_SIZE, 0, NULL);
257
	BUG_ON(thread_stack_cache == NULL);
258 259
}
# endif
260 261
#endif

L
Linus Torvalds 已提交
262
/* SLAB cache for signal_struct structures (tsk->signal) */
263
static struct kmem_cache *signal_cachep;
L
Linus Torvalds 已提交
264 265

/* SLAB cache for sighand_struct structures (tsk->sighand) */
266
struct kmem_cache *sighand_cachep;
L
Linus Torvalds 已提交
267 268

/* SLAB cache for files_struct structures (tsk->files) */
269
struct kmem_cache *files_cachep;
L
Linus Torvalds 已提交
270 271

/* SLAB cache for fs_struct structures (tsk->fs) */
272
struct kmem_cache *fs_cachep;
L
Linus Torvalds 已提交
273 274

/* SLAB cache for vm_area_struct structures */
275
struct kmem_cache *vm_area_cachep;
L
Linus Torvalds 已提交
276 277

/* SLAB cache for mm_struct structures (tsk->mm) */
278
static struct kmem_cache *mm_cachep;
L
Linus Torvalds 已提交
279

280
static void account_kernel_stack(struct task_struct *tsk, int account)
281
{
282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313
	void *stack = task_stack_page(tsk);
	struct vm_struct *vm = task_stack_vm_area(tsk);

	BUILD_BUG_ON(IS_ENABLED(CONFIG_VMAP_STACK) && PAGE_SIZE % 1024 != 0);

	if (vm) {
		int i;

		BUG_ON(vm->nr_pages != THREAD_SIZE / PAGE_SIZE);

		for (i = 0; i < THREAD_SIZE / PAGE_SIZE; i++) {
			mod_zone_page_state(page_zone(vm->pages[i]),
					    NR_KERNEL_STACK_KB,
					    PAGE_SIZE / 1024 * account);
		}

		/* All stack pages belong to the same memcg. */
		memcg_kmem_update_page_stat(vm->pages[0], MEMCG_KERNEL_STACK_KB,
					    account * (THREAD_SIZE / 1024));
	} else {
		/*
		 * All stack pages are in the same zone and belong to the
		 * same memcg.
		 */
		struct page *first_page = virt_to_page(stack);

		mod_zone_page_state(page_zone(first_page), NR_KERNEL_STACK_KB,
				    THREAD_SIZE / 1024 * account);

		memcg_kmem_update_page_stat(first_page, MEMCG_KERNEL_STACK_KB,
					    account * (THREAD_SIZE / 1024));
	}
314 315
}

316
static void release_task_stack(struct task_struct *tsk)
L
Linus Torvalds 已提交
317
{
318
	account_kernel_stack(tsk, -1);
319
	arch_release_thread_stack(tsk->stack);
320
	free_thread_stack(tsk);
321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349
	tsk->stack = NULL;
#ifdef CONFIG_VMAP_STACK
	tsk->stack_vm_area = NULL;
#endif
}

#ifdef CONFIG_THREAD_INFO_IN_TASK
void put_task_stack(struct task_struct *tsk)
{
	if (atomic_dec_and_test(&tsk->stack_refcount))
		release_task_stack(tsk);
}
#endif

void free_task(struct task_struct *tsk)
{
#ifndef CONFIG_THREAD_INFO_IN_TASK
	/*
	 * The task is finally done with both the stack and thread_info,
	 * so free both.
	 */
	release_task_stack(tsk);
#else
	/*
	 * If the task had a separate stack allocation, it should be gone
	 * by now.
	 */
	WARN_ON_ONCE(atomic_read(&tsk->stack_refcount) != 0);
#endif
I
Ingo Molnar 已提交
350
	rt_mutex_debug_task_free(tsk);
351
	ftrace_graph_exit_task(tsk);
352
	put_seccomp_filter(tsk);
353
	arch_release_task_struct(tsk);
L
Linus Torvalds 已提交
354 355 356 357
	free_task_struct(tsk);
}
EXPORT_SYMBOL(free_task);

358 359
static inline void free_signal_struct(struct signal_struct *sig)
{
360
	taskstats_tgid_free(sig);
361
	sched_autogroup_exit(sig);
362 363 364 365
	/*
	 * __mmdrop is not safe to call from softirq context on x86 due to
	 * pgd_dtor so postpone it to the async context
	 */
366
	if (sig->oom_mm)
367
		mmdrop_async(sig->oom_mm);
368 369 370 371 372
	kmem_cache_free(signal_cachep, sig);
}

static inline void put_signal_struct(struct signal_struct *sig)
{
373
	if (atomic_dec_and_test(&sig->sigcnt))
374 375 376
		free_signal_struct(sig);
}

377
void __put_task_struct(struct task_struct *tsk)
L
Linus Torvalds 已提交
378
{
E
Eugene Teo 已提交
379
	WARN_ON(!tsk->exit_state);
L
Linus Torvalds 已提交
380 381 382
	WARN_ON(atomic_read(&tsk->usage));
	WARN_ON(tsk == current);

383
	cgroup_free(tsk);
384
	task_numa_free(tsk);
385
	security_task_free(tsk);
386
	exit_creds(tsk);
387
	delayacct_tsk_free(tsk);
388
	put_signal_struct(tsk->signal);
L
Linus Torvalds 已提交
389 390 391 392

	if (!profile_handoff_task(tsk))
		free_task(tsk);
}
393
EXPORT_SYMBOL_GPL(__put_task_struct);
L
Linus Torvalds 已提交
394

T
Thomas Gleixner 已提交
395
void __init __weak arch_task_cache_init(void) { }
396

397 398 399
/*
 * set_max_threads
 */
400
static void set_max_threads(unsigned int max_threads_suggested)
401
{
402
	u64 threads;
403 404

	/*
405 406
	 * The number of threads shall be limited such that the thread
	 * structures may only consume a small part of the available memory.
407
	 */
408 409 410 411 412 413
	if (fls64(totalram_pages) + fls64(PAGE_SIZE) > 64)
		threads = MAX_THREADS;
	else
		threads = div64_u64((u64) totalram_pages * (u64) PAGE_SIZE,
				    (u64) THREAD_SIZE * 8UL);

414 415 416
	if (threads > max_threads_suggested)
		threads = max_threads_suggested;

417
	max_threads = clamp_t(u64, threads, MIN_THREADS, MAX_THREADS);
418 419
}

420 421 422 423
#ifdef CONFIG_ARCH_WANTS_DYNAMIC_TASK_STRUCT
/* Initialized by the architecture: */
int arch_task_struct_size __read_mostly;
#endif
424

425
void __init fork_init(void)
L
Linus Torvalds 已提交
426
{
427
	int i;
428
#ifndef CONFIG_ARCH_TASK_STRUCT_ALLOCATOR
L
Linus Torvalds 已提交
429 430 431 432
#ifndef ARCH_MIN_TASKALIGN
#define ARCH_MIN_TASKALIGN	L1_CACHE_BYTES
#endif
	/* create a slab on which task_structs can be allocated */
433 434 435
	task_struct_cachep = kmem_cache_create("task_struct",
			arch_task_struct_size, ARCH_MIN_TASKALIGN,
			SLAB_PANIC|SLAB_NOTRACK|SLAB_ACCOUNT, NULL);
L
Linus Torvalds 已提交
436 437
#endif

438 439 440
	/* do the arch specific task caches init */
	arch_task_cache_init();

441
	set_max_threads(MAX_THREADS);
L
Linus Torvalds 已提交
442 443 444 445 446

	init_task.signal->rlim[RLIMIT_NPROC].rlim_cur = max_threads/2;
	init_task.signal->rlim[RLIMIT_NPROC].rlim_max = max_threads/2;
	init_task.signal->rlim[RLIMIT_SIGPENDING] =
		init_task.signal->rlim[RLIMIT_NPROC];
447

448 449 450
	for (i = 0; i < UCOUNT_COUNTS; i++) {
		init_user_ns.ucount_max[i] = max_threads/2;
	}
L
Linus Torvalds 已提交
451 452
}

453
int __weak arch_dup_task_struct(struct task_struct *dst,
454 455 456 457 458 459
					       struct task_struct *src)
{
	*dst = *src;
	return 0;
}

460 461 462 463 464 465 466 467
void set_task_stack_end_magic(struct task_struct *tsk)
{
	unsigned long *stackend;

	stackend = end_of_stack(tsk);
	*stackend = STACK_END_MAGIC;	/* for overflow detection */
}

468
static struct task_struct *dup_task_struct(struct task_struct *orig, int node)
L
Linus Torvalds 已提交
469 470
{
	struct task_struct *tsk;
471
	unsigned long *stack;
472
	struct vm_struct *stack_vm_area;
P
Peter Zijlstra 已提交
473
	int err;
L
Linus Torvalds 已提交
474

475 476
	if (node == NUMA_NO_NODE)
		node = tsk_fork_get_node(orig);
477
	tsk = alloc_task_struct_node(node);
L
Linus Torvalds 已提交
478 479 480
	if (!tsk)
		return NULL;

481 482
	stack = alloc_thread_stack_node(tsk, node);
	if (!stack)
483
		goto free_tsk;
L
Linus Torvalds 已提交
484

485 486
	stack_vm_area = task_stack_vm_area(tsk);

487
	err = arch_dup_task_struct(tsk, orig);
488 489 490 491 492 493 494 495 496 497

	/*
	 * arch_dup_task_struct() clobbers the stack-related fields.  Make
	 * sure they're properly initialized before using any stack-related
	 * functions again.
	 */
	tsk->stack = stack;
#ifdef CONFIG_VMAP_STACK
	tsk->stack_vm_area = stack_vm_area;
#endif
498 499 500
#ifdef CONFIG_THREAD_INFO_IN_TASK
	atomic_set(&tsk->stack_refcount, 1);
#endif
501

502
	if (err)
503
		goto free_stack;
504

K
Kees Cook 已提交
505 506 507 508 509 510 511 512 513
#ifdef CONFIG_SECCOMP
	/*
	 * We must handle setting up seccomp filters once we're under
	 * the sighand lock in case orig has changed between now and
	 * then. Until then, filter must be NULL to avoid messing up
	 * the usage counts on the error path calling free_task.
	 */
	tsk->seccomp.filter = NULL;
#endif
514 515

	setup_thread_stack(tsk, orig);
516
	clear_user_return_notifier(tsk);
517
	clear_tsk_need_resched(tsk);
518
	set_task_stack_end_magic(tsk);
L
Linus Torvalds 已提交
519

520 521 522 523
#ifdef CONFIG_CC_STACKPROTECTOR
	tsk->stack_canary = get_random_int();
#endif

524 525 526 527 528
	/*
	 * One for us, one for whoever does the "release_task()" (usually
	 * parent)
	 */
	atomic_set(&tsk->usage, 2);
529
#ifdef CONFIG_BLK_DEV_IO_TRACE
530
	tsk->btrace_seq = 0;
531
#endif
532
	tsk->splice_pipe = NULL;
533
	tsk->task_frag.page = NULL;
534
	tsk->wake_q.next = NULL;
535

536
	account_kernel_stack(tsk, 1);
537

D
Dmitry Vyukov 已提交
538 539
	kcov_task_init(tsk);

L
Linus Torvalds 已提交
540
	return tsk;
541

542
free_stack:
543
	free_thread_stack(tsk);
544
free_tsk:
545 546
	free_task_struct(tsk);
	return NULL;
L
Linus Torvalds 已提交
547 548 549
}

#ifdef CONFIG_MMU
550 551
static __latent_entropy int dup_mmap(struct mm_struct *mm,
					struct mm_struct *oldmm)
L
Linus Torvalds 已提交
552
{
553
	struct vm_area_struct *mpnt, *tmp, *prev, **pprev;
L
Linus Torvalds 已提交
554 555 556 557
	struct rb_node **rb_link, *rb_parent;
	int retval;
	unsigned long charge;

558
	uprobe_start_dup_mmap();
559 560 561 562
	if (down_write_killable(&oldmm->mmap_sem)) {
		retval = -EINTR;
		goto fail_uprobe_end;
	}
563
	flush_cache_dup_mm(oldmm);
564
	uprobe_dup_mmap(oldmm, mm);
565 566 567 568
	/*
	 * Not linked in yet - no deadlock potential:
	 */
	down_write_nested(&mm->mmap_sem, SINGLE_DEPTH_NESTING);
569

570 571 572
	/* No ordering required: file already has been exposed. */
	RCU_INIT_POINTER(mm->exe_file, get_mm_exe_file(oldmm));

573
	mm->total_vm = oldmm->total_vm;
574
	mm->data_vm = oldmm->data_vm;
575 576 577
	mm->exec_vm = oldmm->exec_vm;
	mm->stack_vm = oldmm->stack_vm;

L
Linus Torvalds 已提交
578 579 580
	rb_link = &mm->mm_rb.rb_node;
	rb_parent = NULL;
	pprev = &mm->mmap;
H
Hugh Dickins 已提交
581
	retval = ksm_fork(mm, oldmm);
A
Andrea Arcangeli 已提交
582 583 584
	if (retval)
		goto out;
	retval = khugepaged_fork(mm, oldmm);
H
Hugh Dickins 已提交
585 586
	if (retval)
		goto out;
L
Linus Torvalds 已提交
587

588
	prev = NULL;
H
Hugh Dickins 已提交
589
	for (mpnt = oldmm->mmap; mpnt; mpnt = mpnt->vm_next) {
L
Linus Torvalds 已提交
590 591 592
		struct file *file;

		if (mpnt->vm_flags & VM_DONTCOPY) {
593
			vm_stat_account(mm, mpnt->vm_flags, -vma_pages(mpnt));
L
Linus Torvalds 已提交
594 595 596 597
			continue;
		}
		charge = 0;
		if (mpnt->vm_flags & VM_ACCOUNT) {
598 599
			unsigned long len = vma_pages(mpnt);

600
			if (security_vm_enough_memory_mm(oldmm, len)) /* sic */
L
Linus Torvalds 已提交
601 602 603
				goto fail_nomem;
			charge = len;
		}
604
		tmp = kmem_cache_alloc(vm_area_cachep, GFP_KERNEL);
L
Linus Torvalds 已提交
605 606 607
		if (!tmp)
			goto fail_nomem;
		*tmp = *mpnt;
608
		INIT_LIST_HEAD(&tmp->anon_vma_chain);
609 610
		retval = vma_dup_policy(mpnt, tmp);
		if (retval)
L
Linus Torvalds 已提交
611
			goto fail_nomem_policy;
A
Andrea Arcangeli 已提交
612
		tmp->vm_mm = mm;
613 614
		if (anon_vma_fork(tmp, mpnt))
			goto fail_nomem_anon_vma_fork;
E
Eric B Munson 已提交
615 616
		tmp->vm_flags &=
			~(VM_LOCKED|VM_LOCKONFAULT|VM_UFFD_MISSING|VM_UFFD_WP);
617
		tmp->vm_next = tmp->vm_prev = NULL;
618
		tmp->vm_userfaultfd_ctx = NULL_VM_UFFD_CTX;
L
Linus Torvalds 已提交
619 620
		file = tmp->vm_file;
		if (file) {
A
Al Viro 已提交
621
			struct inode *inode = file_inode(file);
H
Hugh Dickins 已提交
622 623
			struct address_space *mapping = file->f_mapping;

L
Linus Torvalds 已提交
624 625 626
			get_file(file);
			if (tmp->vm_flags & VM_DENYWRITE)
				atomic_dec(&inode->i_writecount);
627
			i_mmap_lock_write(mapping);
H
Hugh Dickins 已提交
628
			if (tmp->vm_flags & VM_SHARED)
629
				atomic_inc(&mapping->i_mmap_writable);
H
Hugh Dickins 已提交
630 631
			flush_dcache_mmap_lock(mapping);
			/* insert tmp into the share list, just after mpnt */
632 633
			vma_interval_tree_insert_after(tmp, mpnt,
					&mapping->i_mmap);
H
Hugh Dickins 已提交
634
			flush_dcache_mmap_unlock(mapping);
635
			i_mmap_unlock_write(mapping);
L
Linus Torvalds 已提交
636 637
		}

638 639 640 641 642 643 644 645
		/*
		 * Clear hugetlb-related page reserves for children. This only
		 * affects MAP_PRIVATE mappings. Faults generated by the child
		 * are not guaranteed to succeed, even if read-only
		 */
		if (is_vm_hugetlb_page(tmp))
			reset_vma_resv_huge_pages(tmp);

L
Linus Torvalds 已提交
646
		/*
647
		 * Link in the new vma and copy the page table entries.
L
Linus Torvalds 已提交
648 649 650
		 */
		*pprev = tmp;
		pprev = &tmp->vm_next;
651 652
		tmp->vm_prev = prev;
		prev = tmp;
L
Linus Torvalds 已提交
653 654 655 656 657 658

		__vma_link_rb(mm, tmp, rb_link, rb_parent);
		rb_link = &tmp->vm_rb.rb_right;
		rb_parent = &tmp->vm_rb;

		mm->map_count++;
659
		retval = copy_page_range(mm, oldmm, mpnt);
L
Linus Torvalds 已提交
660 661 662 663 664 665 666

		if (tmp->vm_ops && tmp->vm_ops->open)
			tmp->vm_ops->open(tmp);

		if (retval)
			goto out;
	}
667 668
	/* a new mm has just been created */
	arch_dup_mmap(oldmm, mm);
L
Linus Torvalds 已提交
669 670
	retval = 0;
out:
671
	up_write(&mm->mmap_sem);
H
Hugh Dickins 已提交
672
	flush_tlb_mm(oldmm);
L
Linus Torvalds 已提交
673
	up_write(&oldmm->mmap_sem);
674
fail_uprobe_end:
675
	uprobe_end_dup_mmap();
L
Linus Torvalds 已提交
676
	return retval;
677
fail_nomem_anon_vma_fork:
678
	mpol_put(vma_policy(tmp));
L
Linus Torvalds 已提交
679 680 681 682 683 684 685 686
fail_nomem_policy:
	kmem_cache_free(vm_area_cachep, tmp);
fail_nomem:
	retval = -ENOMEM;
	vm_unacct_memory(charge);
	goto out;
}

687
static inline int mm_alloc_pgd(struct mm_struct *mm)
L
Linus Torvalds 已提交
688 689 690 691 692 693 694
{
	mm->pgd = pgd_alloc(mm);
	if (unlikely(!mm->pgd))
		return -ENOMEM;
	return 0;
}

695
static inline void mm_free_pgd(struct mm_struct *mm)
L
Linus Torvalds 已提交
696
{
697
	pgd_free(mm, mm->pgd);
L
Linus Torvalds 已提交
698 699
}
#else
700 701 702 703 704 705 706
static int dup_mmap(struct mm_struct *mm, struct mm_struct *oldmm)
{
	down_write(&oldmm->mmap_sem);
	RCU_INIT_POINTER(mm->exe_file, get_mm_exe_file(oldmm));
	up_write(&oldmm->mmap_sem);
	return 0;
}
L
Linus Torvalds 已提交
707 708 709 710
#define mm_alloc_pgd(mm)	(0)
#define mm_free_pgd(mm)
#endif /* CONFIG_MMU */

D
Daniel Walker 已提交
711
__cacheline_aligned_in_smp DEFINE_SPINLOCK(mmlist_lock);
L
Linus Torvalds 已提交
712

713
#define allocate_mm()	(kmem_cache_alloc(mm_cachep, GFP_KERNEL))
L
Linus Torvalds 已提交
714 715
#define free_mm(mm)	(kmem_cache_free(mm_cachep, (mm)))

716 717 718 719 720 721 722 723 724 725 726 727
static unsigned long default_dump_filter = MMF_DUMP_FILTER_DEFAULT;

static int __init coredump_filter_setup(char *s)
{
	default_dump_filter =
		(simple_strtoul(s, NULL, 0) << MMF_DUMP_FILTER_SHIFT) &
		MMF_DUMP_FILTER_MASK;
	return 1;
}

__setup("coredump_filter=", coredump_filter_setup);

L
Linus Torvalds 已提交
728 729
#include <linux/init_task.h>

A
Alexey Dobriyan 已提交
730 731 732 733
static void mm_init_aio(struct mm_struct *mm)
{
#ifdef CONFIG_AIO
	spin_lock_init(&mm->ioctx_lock);
734
	mm->ioctx_table = NULL;
A
Alexey Dobriyan 已提交
735 736 737
#endif
}

738 739 740 741 742 743 744
static void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
{
#ifdef CONFIG_MEMCG
	mm->owner = p;
#endif
}

745
static struct mm_struct *mm_init(struct mm_struct *mm, struct task_struct *p)
L
Linus Torvalds 已提交
746
{
747 748 749
	mm->mmap = NULL;
	mm->mm_rb = RB_ROOT;
	mm->vmacache_seqnum = 0;
L
Linus Torvalds 已提交
750 751 752 753
	atomic_set(&mm->mm_users, 1);
	atomic_set(&mm->mm_count, 1);
	init_rwsem(&mm->mmap_sem);
	INIT_LIST_HEAD(&mm->mmlist);
754
	mm->core_state = NULL;
755
	atomic_long_set(&mm->nr_ptes, 0);
756
	mm_nr_pmds_init(mm);
757 758
	mm->map_count = 0;
	mm->locked_vm = 0;
V
Vladimir Davydov 已提交
759
	mm->pinned_vm = 0;
K
KAMEZAWA Hiroyuki 已提交
760
	memset(&mm->rss_stat, 0, sizeof(mm->rss_stat));
L
Linus Torvalds 已提交
761
	spin_lock_init(&mm->page_table_lock);
762
	mm_init_cpumask(mm);
A
Alexey Dobriyan 已提交
763
	mm_init_aio(mm);
764
	mm_init_owner(mm, p);
765
	mmu_notifier_mm_init(mm);
766
	clear_tlb_flush_pending(mm);
767 768 769
#if defined(CONFIG_TRANSPARENT_HUGEPAGE) && !USE_SPLIT_PMD_PTLOCKS
	mm->pmd_huge_pte = NULL;
#endif
L
Linus Torvalds 已提交
770

771 772 773 774 775
	if (current->mm) {
		mm->flags = current->mm->flags & MMF_INIT_MASK;
		mm->def_flags = current->mm->def_flags & VM_INIT_DEF_MASK;
	} else {
		mm->flags = default_dump_filter;
L
Linus Torvalds 已提交
776
		mm->def_flags = 0;
777 778
	}

779 780 781 782 783
	if (mm_alloc_pgd(mm))
		goto fail_nopgd;

	if (init_new_context(p, mm))
		goto fail_nocontext;
784

785 786 787 788 789
	return mm;

fail_nocontext:
	mm_free_pgd(mm);
fail_nopgd:
L
Linus Torvalds 已提交
790 791 792 793
	free_mm(mm);
	return NULL;
}

794 795 796 797 798 799 800 801 802 803 804
static void check_mm(struct mm_struct *mm)
{
	int i;

	for (i = 0; i < NR_MM_COUNTERS; i++) {
		long x = atomic_long_read(&mm->rss_stat.count[i]);

		if (unlikely(x))
			printk(KERN_ALERT "BUG: Bad rss-counter state "
					  "mm:%p idx:%d val:%ld\n", mm, i, x);
	}
805 806 807 808 809 810 811 812

	if (atomic_long_read(&mm->nr_ptes))
		pr_alert("BUG: non-zero nr_ptes on freeing mm: %ld\n",
				atomic_long_read(&mm->nr_ptes));
	if (mm_nr_pmds(mm))
		pr_alert("BUG: non-zero nr_pmds on freeing mm: %ld\n",
				mm_nr_pmds(mm));

813
#if defined(CONFIG_TRANSPARENT_HUGEPAGE) && !USE_SPLIT_PMD_PTLOCKS
S
Sasha Levin 已提交
814
	VM_BUG_ON_MM(mm->pmd_huge_pte, mm);
815 816 817
#endif
}

L
Linus Torvalds 已提交
818 819 820
/*
 * Allocate and initialize an mm_struct.
 */
821
struct mm_struct *mm_alloc(void)
L
Linus Torvalds 已提交
822
{
823
	struct mm_struct *mm;
L
Linus Torvalds 已提交
824 825

	mm = allocate_mm();
826 827 828 829
	if (!mm)
		return NULL;

	memset(mm, 0, sizeof(*mm));
830
	return mm_init(mm, current);
L
Linus Torvalds 已提交
831 832 833 834 835 836 837
}

/*
 * Called when the last reference to the mm
 * is dropped: either by a lazy thread or by
 * mmput. Free the page directory and the mm.
 */
838
void __mmdrop(struct mm_struct *mm)
L
Linus Torvalds 已提交
839 840 841 842
{
	BUG_ON(mm == &init_mm);
	mm_free_pgd(mm);
	destroy_context(mm);
A
Andrea Arcangeli 已提交
843
	mmu_notifier_mm_destroy(mm);
844
	check_mm(mm);
L
Linus Torvalds 已提交
845 846
	free_mm(mm);
}
847
EXPORT_SYMBOL_GPL(__mmdrop);
L
Linus Torvalds 已提交
848

849 850 851 852 853 854 855 856 857
static inline void __mmput(struct mm_struct *mm)
{
	VM_BUG_ON(atomic_read(&mm->mm_users));

	uprobe_clear_state(mm);
	exit_aio(mm);
	ksm_exit(mm);
	khugepaged_exit(mm); /* must run before exit_mmap */
	exit_mmap(mm);
858
	mm_put_huge_zero_page(mm);
859 860 861 862 863 864 865 866
	set_mm_exe_file(mm, NULL);
	if (!list_empty(&mm->mmlist)) {
		spin_lock(&mmlist_lock);
		list_del(&mm->mmlist);
		spin_unlock(&mmlist_lock);
	}
	if (mm->binfmt)
		module_put(mm->binfmt->module);
867
	set_bit(MMF_OOM_SKIP, &mm->flags);
868 869 870
	mmdrop(mm);
}

L
Linus Torvalds 已提交
871 872 873 874 875
/*
 * Decrement the use count and release all resources for an mm.
 */
void mmput(struct mm_struct *mm)
{
A
Andrew Morton 已提交
876 877
	might_sleep();

878 879 880 881 882
	if (atomic_dec_and_test(&mm->mm_users))
		__mmput(mm);
}
EXPORT_SYMBOL_GPL(mmput);

M
Michal Hocko 已提交
883
#ifdef CONFIG_MMU
884 885 886 887 888 889 890 891
static void mmput_async_fn(struct work_struct *work)
{
	struct mm_struct *mm = container_of(work, struct mm_struct, async_put_work);
	__mmput(mm);
}

void mmput_async(struct mm_struct *mm)
{
L
Linus Torvalds 已提交
892
	if (atomic_dec_and_test(&mm->mm_users)) {
893 894
		INIT_WORK(&mm->async_put_work, mmput_async_fn);
		schedule_work(&mm->async_put_work);
L
Linus Torvalds 已提交
895 896
	}
}
M
Michal Hocko 已提交
897
#endif
L
Linus Torvalds 已提交
898

899 900 901 902 903
/**
 * set_mm_exe_file - change a reference to the mm's executable file
 *
 * This changes mm's executable file (shown as symlink /proc/[pid]/exe).
 *
904 905 906 907 908
 * Main users are mmput() and sys_execve(). Callers prevent concurrent
 * invocations: in mmput() nobody alive left, in execve task is single
 * threaded. sys_prctl(PR_SET_MM_MAP/EXE_FILE) also needs to set the
 * mm->exe_file, but does so without using set_mm_exe_file() in order
 * to do avoid the need for any locks.
909
 */
910 911
void set_mm_exe_file(struct mm_struct *mm, struct file *new_exe_file)
{
912 913 914 915 916 917 918 919
	struct file *old_exe_file;

	/*
	 * It is safe to dereference the exe_file without RCU as
	 * this function is only called if nobody else can access
	 * this mm -- see comment above for justification.
	 */
	old_exe_file = rcu_dereference_raw(mm->exe_file);
920

921 922
	if (new_exe_file)
		get_file(new_exe_file);
923 924 925
	rcu_assign_pointer(mm->exe_file, new_exe_file);
	if (old_exe_file)
		fput(old_exe_file);
926 927
}

928 929 930 931 932 933
/**
 * get_mm_exe_file - acquire a reference to the mm's executable file
 *
 * Returns %NULL if mm has no associated executable file.
 * User must release file via fput().
 */
934 935 936 937
struct file *get_mm_exe_file(struct mm_struct *mm)
{
	struct file *exe_file;

938 939 940 941 942
	rcu_read_lock();
	exe_file = rcu_dereference(mm->exe_file);
	if (exe_file && !get_file_rcu(exe_file))
		exe_file = NULL;
	rcu_read_unlock();
943 944
	return exe_file;
}
945
EXPORT_SYMBOL(get_mm_exe_file);
946

M
Mateusz Guzik 已提交
947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968
/**
 * get_task_exe_file - acquire a reference to the task's executable file
 *
 * Returns %NULL if task's mm (if any) has no associated executable file or
 * this is a kernel thread with borrowed mm (see the comment above get_task_mm).
 * User must release file via fput().
 */
struct file *get_task_exe_file(struct task_struct *task)
{
	struct file *exe_file = NULL;
	struct mm_struct *mm;

	task_lock(task);
	mm = task->mm;
	if (mm) {
		if (!(task->flags & PF_KTHREAD))
			exe_file = get_mm_exe_file(mm);
	}
	task_unlock(task);
	return exe_file;
}
EXPORT_SYMBOL(get_task_exe_file);
969

L
Linus Torvalds 已提交
970 971 972
/**
 * get_task_mm - acquire a reference to the task's mm
 *
973
 * Returns %NULL if the task has no mm.  Checks PF_KTHREAD (meaning
L
Linus Torvalds 已提交
974 975 976 977 978 979 980 981 982 983 984 985
 * this kernel workthread has transiently adopted a user mm with use_mm,
 * to do its AIO) is not set and if so returns a reference to it, after
 * bumping up the use count.  User must release the mm via mmput()
 * after use.  Typically used by /proc and ptrace.
 */
struct mm_struct *get_task_mm(struct task_struct *task)
{
	struct mm_struct *mm;

	task_lock(task);
	mm = task->mm;
	if (mm) {
986
		if (task->flags & PF_KTHREAD)
L
Linus Torvalds 已提交
987 988 989 990 991 992 993 994 995
			mm = NULL;
		else
			atomic_inc(&mm->mm_users);
	}
	task_unlock(task);
	return mm;
}
EXPORT_SYMBOL_GPL(get_task_mm);

996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015
struct mm_struct *mm_access(struct task_struct *task, unsigned int mode)
{
	struct mm_struct *mm;
	int err;

	err =  mutex_lock_killable(&task->signal->cred_guard_mutex);
	if (err)
		return ERR_PTR(err);

	mm = get_task_mm(task);
	if (mm && mm != current->mm &&
			!ptrace_may_access(task, mode)) {
		mmput(mm);
		mm = ERR_PTR(-EACCES);
	}
	mutex_unlock(&task->signal->cred_guard_mutex);

	return mm;
}

1016
static void complete_vfork_done(struct task_struct *tsk)
1017
{
O
Oleg Nesterov 已提交
1018
	struct completion *vfork;
1019

O
Oleg Nesterov 已提交
1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045
	task_lock(tsk);
	vfork = tsk->vfork_done;
	if (likely(vfork)) {
		tsk->vfork_done = NULL;
		complete(vfork);
	}
	task_unlock(tsk);
}

static int wait_for_vfork_done(struct task_struct *child,
				struct completion *vfork)
{
	int killed;

	freezer_do_not_count();
	killed = wait_for_completion_killable(vfork);
	freezer_count();

	if (killed) {
		task_lock(child);
		child->vfork_done = NULL;
		task_unlock(child);
	}

	put_task_struct(child);
	return killed;
1046 1047
}

L
Linus Torvalds 已提交
1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062
/* Please note the differences between mmput and mm_release.
 * mmput is called whenever we stop holding onto a mm_struct,
 * error success whatever.
 *
 * mm_release is called after a mm_struct has been removed
 * from the current process.
 *
 * This difference is important for error handling, when we
 * only half set up a mm_struct for a new process and need to restore
 * the old one.  Because we mmput the new mm_struct before
 * restoring the old one. . .
 * Eric Biederman 10 January 1998
 */
void mm_release(struct task_struct *tsk, struct mm_struct *mm)
{
1063 1064
	/* Get rid of any futexes when releasing the mm */
#ifdef CONFIG_FUTEX
1065
	if (unlikely(tsk->robust_list)) {
1066
		exit_robust_list(tsk);
1067 1068
		tsk->robust_list = NULL;
	}
1069
#ifdef CONFIG_COMPAT
1070
	if (unlikely(tsk->compat_robust_list)) {
1071
		compat_exit_robust_list(tsk);
1072 1073
		tsk->compat_robust_list = NULL;
	}
1074
#endif
1075 1076
	if (unlikely(!list_empty(&tsk->pi_state_list)))
		exit_pi_state_list(tsk);
1077 1078
#endif

1079 1080
	uprobe_free_utask(tsk);

L
Linus Torvalds 已提交
1081 1082 1083
	/* Get rid of any cached register state */
	deactivate_mm(tsk, mm);

1084
	/*
1085 1086 1087
	 * Signal userspace if we're not exiting with a core dump
	 * because we want to leave the value intact for debugging
	 * purposes.
1088
	 */
1089
	if (tsk->clear_child_tid) {
1090
		if (!(tsk->signal->flags & SIGNAL_GROUP_COREDUMP) &&
1091 1092 1093 1094 1095 1096 1097 1098 1099
		    atomic_read(&mm->mm_users) > 1) {
			/*
			 * We don't check the error code - if userspace has
			 * not set up a proper pointer then tough luck.
			 */
			put_user(0, tsk->clear_child_tid);
			sys_futex(tsk->clear_child_tid, FUTEX_WAKE,
					1, NULL, NULL, 0);
		}
L
Linus Torvalds 已提交
1100 1101
		tsk->clear_child_tid = NULL;
	}
1102 1103 1104 1105 1106 1107 1108

	/*
	 * All done, finally we can wake up parent and return this mm to him.
	 * Also kthread_stop() uses this completion for synchronization.
	 */
	if (tsk->vfork_done)
		complete_vfork_done(tsk);
L
Linus Torvalds 已提交
1109 1110
}

1111 1112 1113 1114
/*
 * Allocate a new mm structure and copy contents from the
 * mm structure of the passed in task structure.
 */
1115
static struct mm_struct *dup_mm(struct task_struct *tsk)
1116 1117 1118 1119 1120 1121 1122 1123 1124 1125
{
	struct mm_struct *mm, *oldmm = current->mm;
	int err;

	mm = allocate_mm();
	if (!mm)
		goto fail_nomem;

	memcpy(mm, oldmm, sizeof(*mm));

1126
	if (!mm_init(mm, tsk))
1127 1128 1129 1130 1131 1132 1133 1134 1135
		goto fail_nomem;

	err = dup_mmap(mm, oldmm);
	if (err)
		goto free_pt;

	mm->hiwater_rss = get_mm_rss(mm);
	mm->hiwater_vm = mm->total_vm;

1136 1137 1138
	if (mm->binfmt && !try_module_get(mm->binfmt->module))
		goto free_pt;

1139 1140 1141
	return mm;

free_pt:
1142 1143
	/* don't put binfmt in mmput, we haven't got module yet */
	mm->binfmt = NULL;
1144 1145 1146 1147 1148 1149
	mmput(mm);

fail_nomem:
	return NULL;
}

1150
static int copy_mm(unsigned long clone_flags, struct task_struct *tsk)
L
Linus Torvalds 已提交
1151
{
1152
	struct mm_struct *mm, *oldmm;
L
Linus Torvalds 已提交
1153 1154 1155 1156
	int retval;

	tsk->min_flt = tsk->maj_flt = 0;
	tsk->nvcsw = tsk->nivcsw = 0;
1157 1158 1159
#ifdef CONFIG_DETECT_HUNG_TASK
	tsk->last_switch_count = tsk->nvcsw + tsk->nivcsw;
#endif
L
Linus Torvalds 已提交
1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172

	tsk->mm = NULL;
	tsk->active_mm = NULL;

	/*
	 * Are we cloning a kernel thread?
	 *
	 * We need to steal a active VM for that..
	 */
	oldmm = current->mm;
	if (!oldmm)
		return 0;

D
Davidlohr Bueso 已提交
1173 1174 1175
	/* initialize the new vmacache entries */
	vmacache_flush(tsk);

L
Linus Torvalds 已提交
1176 1177 1178 1179 1180 1181 1182
	if (clone_flags & CLONE_VM) {
		atomic_inc(&oldmm->mm_users);
		mm = oldmm;
		goto good_mm;
	}

	retval = -ENOMEM;
1183
	mm = dup_mm(tsk);
L
Linus Torvalds 已提交
1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195
	if (!mm)
		goto fail_nomem;

good_mm:
	tsk->mm = mm;
	tsk->active_mm = mm;
	return 0;

fail_nomem:
	return retval;
}

A
Alexey Dobriyan 已提交
1196
static int copy_fs(unsigned long clone_flags, struct task_struct *tsk)
L
Linus Torvalds 已提交
1197
{
A
Al Viro 已提交
1198
	struct fs_struct *fs = current->fs;
L
Linus Torvalds 已提交
1199
	if (clone_flags & CLONE_FS) {
A
Al Viro 已提交
1200
		/* tsk->fs is already what we want */
N
Nick Piggin 已提交
1201
		spin_lock(&fs->lock);
A
Al Viro 已提交
1202
		if (fs->in_exec) {
N
Nick Piggin 已提交
1203
			spin_unlock(&fs->lock);
A
Al Viro 已提交
1204 1205 1206
			return -EAGAIN;
		}
		fs->users++;
N
Nick Piggin 已提交
1207
		spin_unlock(&fs->lock);
L
Linus Torvalds 已提交
1208 1209
		return 0;
	}
A
Al Viro 已提交
1210
	tsk->fs = copy_fs_struct(fs);
L
Linus Torvalds 已提交
1211 1212 1213 1214 1215
	if (!tsk->fs)
		return -ENOMEM;
	return 0;
}

1216
static int copy_files(unsigned long clone_flags, struct task_struct *tsk)
1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242
{
	struct files_struct *oldf, *newf;
	int error = 0;

	/*
	 * A background process may not have any files ...
	 */
	oldf = current->files;
	if (!oldf)
		goto out;

	if (clone_flags & CLONE_FILES) {
		atomic_inc(&oldf->count);
		goto out;
	}

	newf = dup_fd(oldf, &error);
	if (!newf)
		goto out;

	tsk->files = newf;
	error = 0;
out:
	return error;
}

1243
static int copy_io(unsigned long clone_flags, struct task_struct *tsk)
1244 1245 1246
{
#ifdef CONFIG_BLOCK
	struct io_context *ioc = current->io_context;
1247
	struct io_context *new_ioc;
1248 1249 1250

	if (!ioc)
		return 0;
1251 1252 1253 1254
	/*
	 * Share io context with parent, if CLONE_IO is set
	 */
	if (clone_flags & CLONE_IO) {
T
Tejun Heo 已提交
1255 1256
		ioc_task_link(ioc);
		tsk->io_context = ioc;
1257
	} else if (ioprio_valid(ioc->ioprio)) {
1258 1259
		new_ioc = get_task_io_context(tsk, GFP_KERNEL, NUMA_NO_NODE);
		if (unlikely(!new_ioc))
1260 1261
			return -ENOMEM;

1262
		new_ioc->ioprio = ioc->ioprio;
1263
		put_io_context(new_ioc);
1264 1265 1266 1267 1268
	}
#endif
	return 0;
}

A
Alexey Dobriyan 已提交
1269
static int copy_sighand(unsigned long clone_flags, struct task_struct *tsk)
L
Linus Torvalds 已提交
1270 1271 1272
{
	struct sighand_struct *sig;

Z
Zhaolei 已提交
1273
	if (clone_flags & CLONE_SIGHAND) {
L
Linus Torvalds 已提交
1274 1275 1276 1277
		atomic_inc(&current->sighand->count);
		return 0;
	}
	sig = kmem_cache_alloc(sighand_cachep, GFP_KERNEL);
I
Ingo Molnar 已提交
1278
	rcu_assign_pointer(tsk->sighand, sig);
L
Linus Torvalds 已提交
1279 1280
	if (!sig)
		return -ENOMEM;
1281

L
Linus Torvalds 已提交
1282 1283 1284 1285 1286
	atomic_set(&sig->count, 1);
	memcpy(sig->action, current->sighand->action, sizeof(sig->action));
	return 0;
}

1287
void __cleanup_sighand(struct sighand_struct *sighand)
1288
{
1289 1290
	if (atomic_dec_and_test(&sighand->count)) {
		signalfd_cleanup(sighand);
1291 1292 1293 1294
		/*
		 * sighand_cachep is SLAB_DESTROY_BY_RCU so we can free it
		 * without an RCU grace period, see __lock_task_sighand().
		 */
1295
		kmem_cache_free(sighand_cachep, sighand);
1296
	}
1297 1298
}

1299 1300 1301 1302 1303
/*
 * Initialize POSIX timer handling for a thread group.
 */
static void posix_cpu_timers_init_group(struct signal_struct *sig)
{
1304 1305
	unsigned long cpu_limit;

1306
	cpu_limit = READ_ONCE(sig->rlim[RLIMIT_CPU].rlim_cur);
1307 1308
	if (cpu_limit != RLIM_INFINITY) {
		sig->cputime_expires.prof_exp = secs_to_cputime(cpu_limit);
1309
		sig->cputimer.running = true;
1310 1311
	}

1312 1313 1314 1315 1316 1317
	/* The timer lists. */
	INIT_LIST_HEAD(&sig->cpu_timers[0]);
	INIT_LIST_HEAD(&sig->cpu_timers[1]);
	INIT_LIST_HEAD(&sig->cpu_timers[2]);
}

A
Alexey Dobriyan 已提交
1318
static int copy_signal(unsigned long clone_flags, struct task_struct *tsk)
L
Linus Torvalds 已提交
1319 1320 1321
{
	struct signal_struct *sig;

1322
	if (clone_flags & CLONE_THREAD)
1323 1324
		return 0;

1325
	sig = kmem_cache_zalloc(signal_cachep, GFP_KERNEL);
L
Linus Torvalds 已提交
1326 1327 1328 1329
	tsk->signal = sig;
	if (!sig)
		return -ENOMEM;

1330
	sig->nr_threads = 1;
L
Linus Torvalds 已提交
1331
	atomic_set(&sig->live, 1);
1332
	atomic_set(&sig->sigcnt, 1);
1333 1334 1335 1336 1337

	/* list_add(thread_node, thread_head) without INIT_LIST_HEAD() */
	sig->thread_head = (struct list_head)LIST_HEAD_INIT(tsk->thread_node);
	tsk->thread_node = (struct list_head)LIST_HEAD_INIT(sig->thread_head);

L
Linus Torvalds 已提交
1338
	init_waitqueue_head(&sig->wait_chldexit);
1339
	sig->curr_target = tsk;
L
Linus Torvalds 已提交
1340 1341
	init_sigpending(&sig->shared_pending);
	INIT_LIST_HEAD(&sig->posix_timers);
1342
	seqlock_init(&sig->stats_lock);
1343
	prev_cputime_init(&sig->prev_cputime);
L
Linus Torvalds 已提交
1344

1345
	hrtimer_init(&sig->real_timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
L
Linus Torvalds 已提交
1346 1347 1348 1349 1350 1351
	sig->real_timer.function = it_real_fn;

	task_lock(current->group_leader);
	memcpy(sig->rlim, current->signal->rlim, sizeof sig->rlim);
	task_unlock(current->group_leader);

1352 1353
	posix_cpu_timers_init_group(sig);

M
Miloslav Trmac 已提交
1354
	tty_audit_fork(sig);
1355
	sched_autogroup_fork(sig);
M
Miloslav Trmac 已提交
1356

D
David Rientjes 已提交
1357
	sig->oom_score_adj = current->signal->oom_score_adj;
1358
	sig->oom_score_adj_min = current->signal->oom_score_adj_min;
1359

1360 1361 1362
	sig->has_child_subreaper = current->signal->has_child_subreaper ||
				   current->signal->is_child_subreaper;

1363 1364
	mutex_init(&sig->cred_guard_mutex);

L
Linus Torvalds 已提交
1365 1366 1367
	return 0;
}

K
Kees Cook 已提交
1368 1369 1370 1371 1372 1373 1374 1375 1376
static void copy_seccomp(struct task_struct *p)
{
#ifdef CONFIG_SECCOMP
	/*
	 * Must be called with sighand->lock held, which is common to
	 * all threads in the group. Holding cred_guard_mutex is not
	 * needed because this new task is not yet running and cannot
	 * be racing exec.
	 */
1377
	assert_spin_locked(&current->sighand->siglock);
K
Kees Cook 已提交
1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400

	/* Ref-count the new filter user, and assign it. */
	get_seccomp_filter(current);
	p->seccomp = current->seccomp;

	/*
	 * Explicitly enable no_new_privs here in case it got set
	 * between the task_struct being duplicated and holding the
	 * sighand lock. The seccomp state and nnp must be in sync.
	 */
	if (task_no_new_privs(current))
		task_set_no_new_privs(p);

	/*
	 * If the parent gained a seccomp mode after copying thread
	 * flags and between before we held the sighand lock, we have
	 * to manually enable the seccomp thread flag here.
	 */
	if (p->seccomp.mode != SECCOMP_MODE_DISABLED)
		set_tsk_thread_flag(p, TIF_SECCOMP);
#endif
}

1401
SYSCALL_DEFINE1(set_tid_address, int __user *, tidptr)
L
Linus Torvalds 已提交
1402 1403 1404
{
	current->clear_child_tid = tidptr;

1405
	return task_pid_vnr(current);
L
Linus Torvalds 已提交
1406 1407
}

A
Alexey Dobriyan 已提交
1408
static void rt_mutex_init_task(struct task_struct *p)
I
Ingo Molnar 已提交
1409
{
1410
	raw_spin_lock_init(&p->pi_lock);
1411
#ifdef CONFIG_RT_MUTEXES
1412 1413
	p->pi_waiters = RB_ROOT;
	p->pi_waiters_leftmost = NULL;
I
Ingo Molnar 已提交
1414 1415 1416 1417
	p->pi_blocked_on = NULL;
#endif
}

1418 1419 1420 1421 1422
/*
 * Initialize POSIX timer handling for a single task.
 */
static void posix_cpu_timers_init(struct task_struct *tsk)
{
1423 1424
	tsk->cputime_expires.prof_exp = 0;
	tsk->cputime_expires.virt_exp = 0;
1425 1426 1427 1428 1429 1430
	tsk->cputime_expires.sched_exp = 0;
	INIT_LIST_HEAD(&tsk->cpu_timers[0]);
	INIT_LIST_HEAD(&tsk->cpu_timers[1]);
	INIT_LIST_HEAD(&tsk->cpu_timers[2]);
}

1431 1432 1433 1434 1435 1436
static inline void
init_task_pid(struct task_struct *task, enum pid_type type, struct pid *pid)
{
	 task->pids[type].pid = pid;
}

L
Linus Torvalds 已提交
1437 1438 1439 1440 1441 1442 1443 1444
/*
 * This creates a new process as a copy of the old one,
 * but does not actually start it yet.
 *
 * It copies the registers, and all the appropriate
 * parts of the process environment (as per the clone
 * flags). The actual kick-off is left to the caller.
 */
1445 1446
static __latent_entropy struct task_struct *copy_process(
					unsigned long clone_flags,
1447 1448 1449
					unsigned long stack_start,
					unsigned long stack_size,
					int __user *child_tidptr,
R
Roland McGrath 已提交
1450
					struct pid *pid,
1451
					int trace,
1452 1453
					unsigned long tls,
					int node)
L
Linus Torvalds 已提交
1454 1455
{
	int retval;
1456
	struct task_struct *p;
L
Linus Torvalds 已提交
1457 1458 1459 1460

	if ((clone_flags & (CLONE_NEWNS|CLONE_FS)) == (CLONE_NEWNS|CLONE_FS))
		return ERR_PTR(-EINVAL);

1461 1462 1463
	if ((clone_flags & (CLONE_NEWUSER|CLONE_FS)) == (CLONE_NEWUSER|CLONE_FS))
		return ERR_PTR(-EINVAL);

L
Linus Torvalds 已提交
1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478
	/*
	 * Thread groups must share signals as well, and detached threads
	 * can only be started up within the thread group.
	 */
	if ((clone_flags & CLONE_THREAD) && !(clone_flags & CLONE_SIGHAND))
		return ERR_PTR(-EINVAL);

	/*
	 * Shared signal handlers imply shared VM. By way of the above,
	 * thread groups also imply shared VM. Blocking this case allows
	 * for various simplifications in other code.
	 */
	if ((clone_flags & CLONE_SIGHAND) && !(clone_flags & CLONE_VM))
		return ERR_PTR(-EINVAL);

1479 1480 1481 1482 1483 1484 1485 1486 1487 1488
	/*
	 * Siblings of global init remain as zombies on exit since they are
	 * not reaped by their parent (swapper). To solve this and to avoid
	 * multi-rooted process trees, prevent global and container-inits
	 * from creating siblings.
	 */
	if ((clone_flags & CLONE_PARENT) &&
				current->signal->flags & SIGNAL_UNKILLABLE)
		return ERR_PTR(-EINVAL);

1489
	/*
1490
	 * If the new process will be in a different pid or user namespace
1491
	 * do not allow it to share a thread group with the forking task.
1492
	 */
1493
	if (clone_flags & CLONE_THREAD) {
1494 1495 1496 1497 1498
		if ((clone_flags & (CLONE_NEWUSER | CLONE_NEWPID)) ||
		    (task_active_pid_ns(current) !=
				current->nsproxy->pid_ns_for_children))
			return ERR_PTR(-EINVAL);
	}
1499

L
Linus Torvalds 已提交
1500 1501 1502 1503 1504
	retval = security_task_create(clone_flags);
	if (retval)
		goto fork_out;

	retval = -ENOMEM;
1505
	p = dup_task_struct(current, node);
L
Linus Torvalds 已提交
1506 1507 1508
	if (!p)
		goto fork_out;

1509 1510
	ftrace_graph_init_task(p);

1511 1512
	rt_mutex_init_task(p);

I
Ingo Molnar 已提交
1513
#ifdef CONFIG_PROVE_LOCKING
1514 1515 1516
	DEBUG_LOCKS_WARN_ON(!p->hardirqs_enabled);
	DEBUG_LOCKS_WARN_ON(!p->softirqs_enabled);
#endif
L
Linus Torvalds 已提交
1517
	retval = -EAGAIN;
1518
	if (atomic_read(&p->real_cred->user->processes) >=
1519
			task_rlimit(p, RLIMIT_NPROC)) {
1520 1521
		if (p->real_cred->user != INIT_USER &&
		    !capable(CAP_SYS_RESOURCE) && !capable(CAP_SYS_ADMIN))
L
Linus Torvalds 已提交
1522 1523
			goto bad_fork_free;
	}
1524
	current->flags &= ~PF_NPROC_EXCEEDED;
L
Linus Torvalds 已提交
1525

1526 1527 1528
	retval = copy_creds(p, clone_flags);
	if (retval < 0)
		goto bad_fork_free;
L
Linus Torvalds 已提交
1529 1530 1531 1532 1533 1534

	/*
	 * If multiple threads are within copy_process(), then this check
	 * triggers too late. This doesn't hurt, the check is only there
	 * to stop root fork bombs.
	 */
1535
	retval = -EAGAIN;
L
Linus Torvalds 已提交
1536 1537 1538
	if (nr_threads >= max_threads)
		goto bad_fork_cleanup_count;

1539
	delayacct_tsk_init(p);	/* Must remain after dup_task_struct() */
1540 1541
	p->flags &= ~(PF_SUPERPRIV | PF_WQ_WORKER);
	p->flags |= PF_FORKNOEXEC;
L
Linus Torvalds 已提交
1542 1543
	INIT_LIST_HEAD(&p->children);
	INIT_LIST_HEAD(&p->sibling);
1544
	rcu_copy_process(p);
L
Linus Torvalds 已提交
1545 1546 1547 1548 1549
	p->vfork_done = NULL;
	spin_lock_init(&p->alloc_lock);

	init_sigpending(&p->pending);

1550 1551
	p->utime = p->stime = p->gtime = 0;
	p->utimescaled = p->stimescaled = 0;
1552 1553
	prev_cputime_init(&p->prev_cputime);

1554
#ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1555
	seqcount_init(&p->vtime_seqcount);
1556
	p->vtime_snap = 0;
1557
	p->vtime_snap_whence = VTIME_INACTIVE;
1558 1559
#endif

1560 1561 1562
#if defined(SPLIT_RSS_COUNTING)
	memset(&p->rss_stat, 0, sizeof(p->rss_stat));
#endif
1563

1564 1565
	p->default_timer_slack_ns = current->timer_slack_ns;

1566
	task_io_accounting_init(&p->ioac);
L
Linus Torvalds 已提交
1567 1568
	acct_clear_integrals(p);

1569
	posix_cpu_timers_init(p);
L
Linus Torvalds 已提交
1570

1571
	p->start_time = ktime_get_ns();
1572
	p->real_start_time = ktime_get_boot_ns();
L
Linus Torvalds 已提交
1573 1574
	p->io_context = NULL;
	p->audit_context = NULL;
1575
	cgroup_fork(p);
L
Linus Torvalds 已提交
1576
#ifdef CONFIG_NUMA
1577
	p->mempolicy = mpol_dup(p->mempolicy);
1578 1579 1580
	if (IS_ERR(p->mempolicy)) {
		retval = PTR_ERR(p->mempolicy);
		p->mempolicy = NULL;
1581
		goto bad_fork_cleanup_threadgroup_lock;
1582
	}
L
Linus Torvalds 已提交
1583
#endif
1584 1585 1586
#ifdef CONFIG_CPUSETS
	p->cpuset_mem_spread_rotor = NUMA_NO_NODE;
	p->cpuset_slab_spread_rotor = NUMA_NO_NODE;
1587
	seqcount_init(&p->mems_allowed_seq);
1588
#endif
1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603
#ifdef CONFIG_TRACE_IRQFLAGS
	p->irq_events = 0;
	p->hardirqs_enabled = 0;
	p->hardirq_enable_ip = 0;
	p->hardirq_enable_event = 0;
	p->hardirq_disable_ip = _THIS_IP_;
	p->hardirq_disable_event = 0;
	p->softirqs_enabled = 1;
	p->softirq_enable_ip = _THIS_IP_;
	p->softirq_enable_event = 0;
	p->softirq_disable_ip = 0;
	p->softirq_disable_event = 0;
	p->hardirq_context = 0;
	p->softirq_context = 0;
#endif
1604 1605 1606

	p->pagefault_disabled = 0;

I
Ingo Molnar 已提交
1607 1608 1609 1610 1611
#ifdef CONFIG_LOCKDEP
	p->lockdep_depth = 0; /* no locks held yet */
	p->curr_chain_key = 0;
	p->lockdep_recursion = 0;
#endif
L
Linus Torvalds 已提交
1612

1613 1614 1615
#ifdef CONFIG_DEBUG_MUTEXES
	p->blocked_on = NULL; /* not blocked yet */
#endif
K
Kent Overstreet 已提交
1616 1617 1618 1619
#ifdef CONFIG_BCACHE
	p->sequential_io	= 0;
	p->sequential_io_avg	= 0;
#endif
1620

1621
	/* Perform scheduler related setup. Assign this task to a CPU. */
1622 1623 1624
	retval = sched_fork(clone_flags, p);
	if (retval)
		goto bad_fork_cleanup_policy;
1625

1626
	retval = perf_event_init_task(p);
1627 1628
	if (retval)
		goto bad_fork_cleanup_policy;
1629 1630
	retval = audit_alloc(p);
	if (retval)
P
Peter Zijlstra 已提交
1631
		goto bad_fork_cleanup_perf;
L
Linus Torvalds 已提交
1632
	/* copy all the process information */
1633
	shm_init_task(p);
1634 1635
	retval = copy_semundo(clone_flags, p);
	if (retval)
L
Linus Torvalds 已提交
1636
		goto bad_fork_cleanup_audit;
1637 1638
	retval = copy_files(clone_flags, p);
	if (retval)
L
Linus Torvalds 已提交
1639
		goto bad_fork_cleanup_semundo;
1640 1641
	retval = copy_fs(clone_flags, p);
	if (retval)
L
Linus Torvalds 已提交
1642
		goto bad_fork_cleanup_files;
1643 1644
	retval = copy_sighand(clone_flags, p);
	if (retval)
L
Linus Torvalds 已提交
1645
		goto bad_fork_cleanup_fs;
1646 1647
	retval = copy_signal(clone_flags, p);
	if (retval)
L
Linus Torvalds 已提交
1648
		goto bad_fork_cleanup_sighand;
1649 1650
	retval = copy_mm(clone_flags, p);
	if (retval)
L
Linus Torvalds 已提交
1651
		goto bad_fork_cleanup_signal;
1652 1653
	retval = copy_namespaces(clone_flags, p);
	if (retval)
D
David Howells 已提交
1654
		goto bad_fork_cleanup_mm;
1655 1656
	retval = copy_io(clone_flags, p);
	if (retval)
1657
		goto bad_fork_cleanup_namespaces;
1658
	retval = copy_thread_tls(clone_flags, stack_start, stack_size, p, tls);
L
Linus Torvalds 已提交
1659
	if (retval)
1660
		goto bad_fork_cleanup_io;
L
Linus Torvalds 已提交
1661

1662
	if (pid != &init_struct_pid) {
1663
		pid = alloc_pid(p->nsproxy->pid_ns_for_children);
1664 1665
		if (IS_ERR(pid)) {
			retval = PTR_ERR(pid);
1666
			goto bad_fork_cleanup_thread;
1667
		}
1668 1669
	}

L
Linus Torvalds 已提交
1670 1671 1672 1673
	p->set_child_tid = (clone_flags & CLONE_CHILD_SETTID) ? child_tidptr : NULL;
	/*
	 * Clear TID on mm_release()?
	 */
1674
	p->clear_child_tid = (clone_flags & CLONE_CHILD_CLEARTID) ? child_tidptr : NULL;
1675 1676 1677
#ifdef CONFIG_BLOCK
	p->plug = NULL;
#endif
1678
#ifdef CONFIG_FUTEX
1679 1680 1681 1682
	p->robust_list = NULL;
#ifdef CONFIG_COMPAT
	p->compat_robust_list = NULL;
#endif
1683 1684
	INIT_LIST_HEAD(&p->pi_state_list);
	p->pi_state_cache = NULL;
1685
#endif
1686 1687 1688 1689
	/*
	 * sigaltstack should be cleared when sharing the same VM
	 */
	if ((clone_flags & (CLONE_VM|CLONE_VFORK)) == CLONE_VM)
1690
		sas_ss_reset(p);
1691

L
Linus Torvalds 已提交
1692
	/*
1693 1694
	 * Syscall tracing and stepping should be turned off in the
	 * child regardless of CLONE_PTRACE.
L
Linus Torvalds 已提交
1695
	 */
1696
	user_disable_single_step(p);
L
Linus Torvalds 已提交
1697
	clear_tsk_thread_flag(p, TIF_SYSCALL_TRACE);
1698 1699 1700
#ifdef TIF_SYSCALL_EMU
	clear_tsk_thread_flag(p, TIF_SYSCALL_EMU);
#endif
A
Arjan van de Ven 已提交
1701
	clear_all_latency_tracing(p);
L
Linus Torvalds 已提交
1702 1703

	/* ok, now we should be set up.. */
1704 1705
	p->pid = pid_nr(pid);
	if (clone_flags & CLONE_THREAD) {
1706
		p->exit_signal = -1;
1707 1708 1709 1710 1711 1712 1713 1714 1715 1716
		p->group_leader = current->group_leader;
		p->tgid = current->tgid;
	} else {
		if (clone_flags & CLONE_PARENT)
			p->exit_signal = current->group_leader->exit_signal;
		else
			p->exit_signal = (clone_flags & CSIGNAL);
		p->group_leader = p;
		p->tgid = p->pid;
	}
1717

1718 1719
	p->nr_dirtied = 0;
	p->nr_dirtied_pause = 128 >> (PAGE_SHIFT - 10);
1720
	p->dirty_paused_when = 0;
1721

1722
	p->pdeath_signal = 0;
O
Oleg Nesterov 已提交
1723
	INIT_LIST_HEAD(&p->thread_group);
A
Al Viro 已提交
1724
	p->task_works = NULL;
L
Linus Torvalds 已提交
1725

1726
	threadgroup_change_begin(current);
1727 1728 1729 1730 1731 1732
	/*
	 * Ensure that the cgroup subsystem policies allow the new process to be
	 * forked. It should be noted the the new process's css_set can be changed
	 * between here and cgroup_post_fork() if an organisation operation is in
	 * progress.
	 */
1733
	retval = cgroup_can_fork(p);
1734 1735 1736
	if (retval)
		goto bad_fork_free_pid;

1737 1738 1739 1740
	/*
	 * Make it visible to the rest of the system, but dont wake it up yet.
	 * Need tasklist lock for parent etc handling!
	 */
L
Linus Torvalds 已提交
1741 1742 1743
	write_lock_irq(&tasklist_lock);

	/* CLONE_PARENT re-uses the old parent */
1744
	if (clone_flags & (CLONE_PARENT|CLONE_THREAD)) {
L
Linus Torvalds 已提交
1745
		p->real_parent = current->real_parent;
1746 1747
		p->parent_exec_id = current->parent_exec_id;
	} else {
L
Linus Torvalds 已提交
1748
		p->real_parent = current;
1749 1750
		p->parent_exec_id = current->self_exec_id;
	}
L
Linus Torvalds 已提交
1751

1752
	spin_lock(&current->sighand->siglock);
1753

K
Kees Cook 已提交
1754 1755 1756 1757 1758 1759
	/*
	 * Copy seccomp details explicitly here, in case they were changed
	 * before holding sighand lock.
	 */
	copy_seccomp(p);

1760 1761 1762 1763 1764 1765 1766
	/*
	 * Process group and session signals need to be delivered to just the
	 * parent before the fork or both the parent and the child after the
	 * fork. Restart if a signal comes in before we add the new process to
	 * it's process group.
	 * A fatal signal pending means that current will exit, so the new
	 * thread can't slip out of an OOM kill (or normal SIGKILL).
1767
	*/
D
Daniel Walker 已提交
1768
	recalc_sigpending();
1769 1770 1771 1772
	if (signal_pending(current)) {
		spin_unlock(&current->sighand->siglock);
		write_unlock_irq(&tasklist_lock);
		retval = -ERESTARTNOINTR;
1773
		goto bad_fork_cancel_cgroup;
1774 1775
	}

1776
	if (likely(p->pid)) {
T
Tejun Heo 已提交
1777
		ptrace_init_task(p, (clone_flags & CLONE_PTRACE) || trace);
1778

1779
		init_task_pid(p, PIDTYPE_PID, pid);
1780
		if (thread_group_leader(p)) {
1781 1782 1783
			init_task_pid(p, PIDTYPE_PGID, task_pgrp(current));
			init_task_pid(p, PIDTYPE_SID, task_session(current));

1784
			if (is_child_reaper(pid)) {
1785
				ns_of_pid(pid)->child_reaper = p;
1786 1787
				p->signal->flags |= SIGNAL_UNKILLABLE;
			}
1788

1789
			p->signal->leader_pid = pid;
A
Alan Cox 已提交
1790
			p->signal->tty = tty_kref_get(current->signal->tty);
1791
			list_add_tail(&p->sibling, &p->real_parent->children);
1792
			list_add_tail_rcu(&p->tasks, &init_task.tasks);
1793 1794
			attach_pid(p, PIDTYPE_PGID);
			attach_pid(p, PIDTYPE_SID);
1795
			__this_cpu_inc(process_counts);
1796 1797 1798 1799 1800 1801
		} else {
			current->signal->nr_threads++;
			atomic_inc(&current->signal->live);
			atomic_inc(&current->signal->sigcnt);
			list_add_tail_rcu(&p->thread_group,
					  &p->group_leader->thread_group);
1802 1803
			list_add_tail_rcu(&p->thread_node,
					  &p->signal->thread_head);
1804
		}
1805
		attach_pid(p, PIDTYPE_PID);
1806
		nr_threads++;
L
Linus Torvalds 已提交
1807 1808 1809
	}

	total_forks++;
1810
	spin_unlock(&current->sighand->siglock);
1811
	syscall_tracepoint_update(p);
L
Linus Torvalds 已提交
1812
	write_unlock_irq(&tasklist_lock);
1813

1814
	proc_fork_connector(p);
1815
	cgroup_post_fork(p);
1816
	threadgroup_change_end(current);
1817
	perf_event_fork(p);
1818 1819

	trace_task_newtask(p, clone_flags);
1820
	uprobe_copy_process(p, clone_flags);
1821

L
Linus Torvalds 已提交
1822 1823
	return p;

1824
bad_fork_cancel_cgroup:
1825
	cgroup_cancel_fork(p);
1826
bad_fork_free_pid:
1827
	threadgroup_change_end(current);
1828 1829
	if (pid != &init_struct_pid)
		free_pid(pid);
1830 1831
bad_fork_cleanup_thread:
	exit_thread(p);
1832
bad_fork_cleanup_io:
1833 1834
	if (p->io_context)
		exit_io_context(p);
S
Serge E. Hallyn 已提交
1835
bad_fork_cleanup_namespaces:
1836
	exit_task_namespaces(p);
L
Linus Torvalds 已提交
1837
bad_fork_cleanup_mm:
D
David Rientjes 已提交
1838
	if (p->mm)
L
Linus Torvalds 已提交
1839 1840
		mmput(p->mm);
bad_fork_cleanup_signal:
1841
	if (!(clone_flags & CLONE_THREAD))
1842
		free_signal_struct(p->signal);
L
Linus Torvalds 已提交
1843
bad_fork_cleanup_sighand:
1844
	__cleanup_sighand(p->sighand);
L
Linus Torvalds 已提交
1845 1846 1847 1848 1849 1850 1851 1852
bad_fork_cleanup_fs:
	exit_fs(p); /* blocking */
bad_fork_cleanup_files:
	exit_files(p); /* blocking */
bad_fork_cleanup_semundo:
	exit_sem(p);
bad_fork_cleanup_audit:
	audit_free(p);
P
Peter Zijlstra 已提交
1853
bad_fork_cleanup_perf:
1854
	perf_event_free_task(p);
P
Peter Zijlstra 已提交
1855
bad_fork_cleanup_policy:
L
Linus Torvalds 已提交
1856
#ifdef CONFIG_NUMA
1857
	mpol_put(p->mempolicy);
1858
bad_fork_cleanup_threadgroup_lock:
L
Linus Torvalds 已提交
1859
#endif
1860
	delayacct_tsk_free(p);
L
Linus Torvalds 已提交
1861
bad_fork_cleanup_count:
D
David Howells 已提交
1862
	atomic_dec(&p->cred->user->processes);
1863
	exit_creds(p);
L
Linus Torvalds 已提交
1864
bad_fork_free:
1865
	put_task_stack(p);
L
Linus Torvalds 已提交
1866
	free_task(p);
1867 1868
fork_out:
	return ERR_PTR(retval);
L
Linus Torvalds 已提交
1869 1870
}

1871 1872 1873 1874 1875 1876 1877 1878 1879 1880
static inline void init_idle_pids(struct pid_link *links)
{
	enum pid_type type;

	for (type = PIDTYPE_PID; type < PIDTYPE_MAX; ++type) {
		INIT_HLIST_NODE(&links[type].node); /* not really needed */
		links[type].pid = &init_struct_pid;
	}
}

1881
struct task_struct *fork_idle(int cpu)
L
Linus Torvalds 已提交
1882
{
1883
	struct task_struct *task;
1884 1885
	task = copy_process(CLONE_VM, 0, 0, NULL, &init_struct_pid, 0, 0,
			    cpu_to_node(cpu));
1886 1887
	if (!IS_ERR(task)) {
		init_idle_pids(task->pids);
1888
		init_idle(task, cpu);
1889
	}
1890

L
Linus Torvalds 已提交
1891 1892 1893 1894 1895 1896 1897 1898 1899
	return task;
}

/*
 *  Ok, this is the main fork-routine.
 *
 * It copies the process, and if successful kick-starts
 * it and waits for it to finish using the VM if required.
 */
1900
long _do_fork(unsigned long clone_flags,
L
Linus Torvalds 已提交
1901 1902 1903
	      unsigned long stack_start,
	      unsigned long stack_size,
	      int __user *parent_tidptr,
1904 1905
	      int __user *child_tidptr,
	      unsigned long tls)
L
Linus Torvalds 已提交
1906 1907 1908
{
	struct task_struct *p;
	int trace = 0;
1909
	long nr;
L
Linus Torvalds 已提交
1910

R
Roland McGrath 已提交
1911
	/*
T
Tejun Heo 已提交
1912 1913 1914 1915
	 * Determine whether and which event to report to ptracer.  When
	 * called from kernel_thread or CLONE_UNTRACED is explicitly
	 * requested, no event is reported; otherwise, report if the event
	 * for the type of forking is enabled.
R
Roland McGrath 已提交
1916
	 */
1917
	if (!(clone_flags & CLONE_UNTRACED)) {
T
Tejun Heo 已提交
1918 1919 1920 1921 1922 1923 1924 1925 1926 1927
		if (clone_flags & CLONE_VFORK)
			trace = PTRACE_EVENT_VFORK;
		else if ((clone_flags & CSIGNAL) != SIGCHLD)
			trace = PTRACE_EVENT_CLONE;
		else
			trace = PTRACE_EVENT_FORK;

		if (likely(!ptrace_event_enabled(current, trace)))
			trace = 0;
	}
L
Linus Torvalds 已提交
1928

A
Al Viro 已提交
1929
	p = copy_process(clone_flags, stack_start, stack_size,
1930
			 child_tidptr, NULL, trace, tls, NUMA_NO_NODE);
1931
	add_latent_entropy();
L
Linus Torvalds 已提交
1932 1933 1934 1935 1936 1937
	/*
	 * Do this prior waking up the new thread - the thread pointer
	 * might get invalid after that point, if the thread exits quickly.
	 */
	if (!IS_ERR(p)) {
		struct completion vfork;
1938
		struct pid *pid;
L
Linus Torvalds 已提交
1939

1940 1941
		trace_sched_process_fork(current, p);

1942 1943
		pid = get_task_pid(p, PIDTYPE_PID);
		nr = pid_vnr(pid);
1944 1945 1946

		if (clone_flags & CLONE_PARENT_SETTID)
			put_user(nr, parent_tidptr);
1947

L
Linus Torvalds 已提交
1948 1949 1950
		if (clone_flags & CLONE_VFORK) {
			p->vfork_done = &vfork;
			init_completion(&vfork);
O
Oleg Nesterov 已提交
1951
			get_task_struct(p);
L
Linus Torvalds 已提交
1952 1953
		}

1954
		wake_up_new_task(p);
L
Linus Torvalds 已提交
1955

T
Tejun Heo 已提交
1956 1957
		/* forking complete and child started to run, tell ptracer */
		if (unlikely(trace))
1958
			ptrace_event_pid(trace, pid);
R
Roland McGrath 已提交
1959

L
Linus Torvalds 已提交
1960
		if (clone_flags & CLONE_VFORK) {
O
Oleg Nesterov 已提交
1961
			if (!wait_for_vfork_done(p, &vfork))
1962
				ptrace_event_pid(PTRACE_EVENT_VFORK_DONE, pid);
L
Linus Torvalds 已提交
1963
		}
1964 1965

		put_pid(pid);
L
Linus Torvalds 已提交
1966
	} else {
1967
		nr = PTR_ERR(p);
L
Linus Torvalds 已提交
1968
	}
1969
	return nr;
L
Linus Torvalds 已提交
1970 1971
}

1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985
#ifndef CONFIG_HAVE_COPY_THREAD_TLS
/* For compatibility with architectures that call do_fork directly rather than
 * using the syscall entry points below. */
long do_fork(unsigned long clone_flags,
	      unsigned long stack_start,
	      unsigned long stack_size,
	      int __user *parent_tidptr,
	      int __user *child_tidptr)
{
	return _do_fork(clone_flags, stack_start, stack_size,
			parent_tidptr, child_tidptr, 0);
}
#endif

1986 1987 1988 1989 1990
/*
 * Create a kernel thread.
 */
pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags)
{
1991 1992
	return _do_fork(flags|CLONE_VM|CLONE_UNTRACED, (unsigned long)fn,
		(unsigned long)arg, NULL, NULL, 0);
1993 1994
}

A
Al Viro 已提交
1995 1996 1997 1998
#ifdef __ARCH_WANT_SYS_FORK
SYSCALL_DEFINE0(fork)
{
#ifdef CONFIG_MMU
1999
	return _do_fork(SIGCHLD, 0, 0, NULL, NULL, 0);
A
Al Viro 已提交
2000 2001
#else
	/* can not support in nommu mode */
2002
	return -EINVAL;
A
Al Viro 已提交
2003 2004 2005 2006 2007 2008 2009
#endif
}
#endif

#ifdef __ARCH_WANT_SYS_VFORK
SYSCALL_DEFINE0(vfork)
{
2010 2011
	return _do_fork(CLONE_VFORK | CLONE_VM | SIGCHLD, 0,
			0, NULL, NULL, 0);
A
Al Viro 已提交
2012 2013 2014 2015 2016 2017 2018
}
#endif

#ifdef __ARCH_WANT_SYS_CLONE
#ifdef CONFIG_CLONE_BACKWARDS
SYSCALL_DEFINE5(clone, unsigned long, clone_flags, unsigned long, newsp,
		 int __user *, parent_tidptr,
2019
		 unsigned long, tls,
A
Al Viro 已提交
2020 2021 2022 2023 2024
		 int __user *, child_tidptr)
#elif defined(CONFIG_CLONE_BACKWARDS2)
SYSCALL_DEFINE5(clone, unsigned long, newsp, unsigned long, clone_flags,
		 int __user *, parent_tidptr,
		 int __user *, child_tidptr,
2025
		 unsigned long, tls)
M
Michal Simek 已提交
2026 2027 2028 2029 2030
#elif defined(CONFIG_CLONE_BACKWARDS3)
SYSCALL_DEFINE6(clone, unsigned long, clone_flags, unsigned long, newsp,
		int, stack_size,
		int __user *, parent_tidptr,
		int __user *, child_tidptr,
2031
		unsigned long, tls)
A
Al Viro 已提交
2032 2033 2034 2035
#else
SYSCALL_DEFINE5(clone, unsigned long, clone_flags, unsigned long, newsp,
		 int __user *, parent_tidptr,
		 int __user *, child_tidptr,
2036
		 unsigned long, tls)
A
Al Viro 已提交
2037 2038
#endif
{
2039
	return _do_fork(clone_flags, newsp, 0, parent_tidptr, child_tidptr, tls);
A
Al Viro 已提交
2040 2041 2042
}
#endif

2043 2044 2045 2046
#ifndef ARCH_MIN_MMSTRUCT_ALIGN
#define ARCH_MIN_MMSTRUCT_ALIGN 0
#endif

2047
static void sighand_ctor(void *data)
2048 2049 2050
{
	struct sighand_struct *sighand = data;

C
Christoph Lameter 已提交
2051
	spin_lock_init(&sighand->siglock);
D
Davide Libenzi 已提交
2052
	init_waitqueue_head(&sighand->signalfd_wqh);
2053 2054
}

L
Linus Torvalds 已提交
2055 2056 2057 2058
void __init proc_caches_init(void)
{
	sighand_cachep = kmem_cache_create("sighand_cache",
			sizeof(struct sighand_struct), 0,
V
Vegard Nossum 已提交
2059
			SLAB_HWCACHE_ALIGN|SLAB_PANIC|SLAB_DESTROY_BY_RCU|
2060
			SLAB_NOTRACK|SLAB_ACCOUNT, sighand_ctor);
L
Linus Torvalds 已提交
2061 2062
	signal_cachep = kmem_cache_create("signal_cache",
			sizeof(struct signal_struct), 0,
2063 2064
			SLAB_HWCACHE_ALIGN|SLAB_PANIC|SLAB_NOTRACK|SLAB_ACCOUNT,
			NULL);
2065
	files_cachep = kmem_cache_create("files_cache",
L
Linus Torvalds 已提交
2066
			sizeof(struct files_struct), 0,
2067 2068
			SLAB_HWCACHE_ALIGN|SLAB_PANIC|SLAB_NOTRACK|SLAB_ACCOUNT,
			NULL);
2069
	fs_cachep = kmem_cache_create("fs_cache",
L
Linus Torvalds 已提交
2070
			sizeof(struct fs_struct), 0,
2071 2072
			SLAB_HWCACHE_ALIGN|SLAB_PANIC|SLAB_NOTRACK|SLAB_ACCOUNT,
			NULL);
2073 2074 2075 2076 2077 2078 2079
	/*
	 * FIXME! The "sizeof(struct mm_struct)" currently includes the
	 * whole struct cpumask for the OFFSTACK case. We could change
	 * this to *only* allocate as much of it as required by the
	 * maximum number of CPU's we can ever have.  The cpumask_allocation
	 * is at the end of the structure, exactly for that reason.
	 */
L
Linus Torvalds 已提交
2080
	mm_cachep = kmem_cache_create("mm_struct",
2081
			sizeof(struct mm_struct), ARCH_MIN_MMSTRUCT_ALIGN,
2082 2083 2084
			SLAB_HWCACHE_ALIGN|SLAB_PANIC|SLAB_NOTRACK|SLAB_ACCOUNT,
			NULL);
	vm_area_cachep = KMEM_CACHE(vm_area_struct, SLAB_PANIC|SLAB_ACCOUNT);
2085
	mmap_init();
2086
	nsproxy_cache_init();
L
Linus Torvalds 已提交
2087
}
2088 2089

/*
2090
 * Check constraints on flags passed to the unshare system call.
2091
 */
2092
static int check_unshare_flags(unsigned long unshare_flags)
2093
{
2094 2095
	if (unshare_flags & ~(CLONE_THREAD|CLONE_FS|CLONE_NEWNS|CLONE_SIGHAND|
				CLONE_VM|CLONE_FILES|CLONE_SYSVSEM|
2096
				CLONE_NEWUTS|CLONE_NEWIPC|CLONE_NEWNET|
2097
				CLONE_NEWUSER|CLONE_NEWPID|CLONE_NEWCGROUP))
2098
		return -EINVAL;
2099
	/*
2100 2101 2102 2103
	 * Not implemented, but pretend it works if there is nothing
	 * to unshare.  Note that unsharing the address space or the
	 * signal handlers also need to unshare the signal queues (aka
	 * CLONE_THREAD).
2104
	 */
2105
	if (unshare_flags & (CLONE_THREAD | CLONE_SIGHAND | CLONE_VM)) {
2106 2107 2108 2109 2110 2111 2112 2113 2114
		if (!thread_group_empty(current))
			return -EINVAL;
	}
	if (unshare_flags & (CLONE_SIGHAND | CLONE_VM)) {
		if (atomic_read(&current->sighand->count) > 1)
			return -EINVAL;
	}
	if (unshare_flags & CLONE_VM) {
		if (!current_is_single_threaded())
2115 2116
			return -EINVAL;
	}
2117 2118 2119 2120 2121

	return 0;
}

/*
2122
 * Unshare the filesystem structure if it is being shared
2123 2124 2125 2126 2127
 */
static int unshare_fs(unsigned long unshare_flags, struct fs_struct **new_fsp)
{
	struct fs_struct *fs = current->fs;

A
Al Viro 已提交
2128 2129 2130 2131 2132 2133 2134 2135 2136 2137
	if (!(unshare_flags & CLONE_FS) || !fs)
		return 0;

	/* don't need lock here; in the worst case we'll do useless copy */
	if (fs->users == 1)
		return 0;

	*new_fsp = copy_fs_struct(fs);
	if (!*new_fsp)
		return -ENOMEM;
2138 2139 2140 2141 2142

	return 0;
}

/*
2143
 * Unshare file descriptor table if it is being shared
2144 2145 2146 2147
 */
static int unshare_fd(unsigned long unshare_flags, struct files_struct **new_fdp)
{
	struct files_struct *fd = current->files;
2148
	int error = 0;
2149 2150

	if ((unshare_flags & CLONE_FILES) &&
2151 2152 2153 2154 2155
	    (fd && atomic_read(&fd->count) > 1)) {
		*new_fdp = dup_fd(fd, &error);
		if (!*new_fdp)
			return error;
	}
2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167

	return 0;
}

/*
 * unshare allows a process to 'unshare' part of the process
 * context which was originally shared using clone.  copy_*
 * functions used by do_fork() cannot be used here directly
 * because they modify an inactive task_struct that is being
 * constructed. Here we are modifying the current, active,
 * task_struct.
 */
2168
SYSCALL_DEFINE1(unshare, unsigned long, unshare_flags)
2169 2170 2171
{
	struct fs_struct *fs, *new_fs = NULL;
	struct files_struct *fd, *new_fd = NULL;
2172
	struct cred *new_cred = NULL;
2173
	struct nsproxy *new_nsproxy = NULL;
2174
	int do_sysvsem = 0;
2175
	int err;
2176

2177
	/*
2178 2179
	 * If unsharing a user namespace must also unshare the thread group
	 * and unshare the filesystem root and working directories.
2180 2181
	 */
	if (unshare_flags & CLONE_NEWUSER)
2182
		unshare_flags |= CLONE_THREAD | CLONE_FS;
2183 2184 2185 2186 2187
	/*
	 * If unsharing vm, must also unshare signal handlers.
	 */
	if (unshare_flags & CLONE_VM)
		unshare_flags |= CLONE_SIGHAND;
2188 2189 2190 2191 2192
	/*
	 * If unsharing a signal handlers, must also unshare the signal queues.
	 */
	if (unshare_flags & CLONE_SIGHAND)
		unshare_flags |= CLONE_THREAD;
2193 2194 2195 2196 2197
	/*
	 * If unsharing namespace, must also unshare filesystem information.
	 */
	if (unshare_flags & CLONE_NEWNS)
		unshare_flags |= CLONE_FS;
2198 2199 2200 2201

	err = check_unshare_flags(unshare_flags);
	if (err)
		goto bad_unshare_out;
2202 2203 2204 2205 2206 2207
	/*
	 * CLONE_NEWIPC must also detach from the undolist: after switching
	 * to a new ipc namespace, the semaphore arrays from the old
	 * namespace are unreachable.
	 */
	if (unshare_flags & (CLONE_NEWIPC|CLONE_SYSVSEM))
2208
		do_sysvsem = 1;
2209 2210
	err = unshare_fs(unshare_flags, &new_fs);
	if (err)
2211
		goto bad_unshare_out;
2212 2213
	err = unshare_fd(unshare_flags, &new_fd);
	if (err)
2214
		goto bad_unshare_cleanup_fs;
2215
	err = unshare_userns(unshare_flags, &new_cred);
2216
	if (err)
2217
		goto bad_unshare_cleanup_fd;
2218 2219 2220 2221
	err = unshare_nsproxy_namespaces(unshare_flags, &new_nsproxy,
					 new_cred, new_fs);
	if (err)
		goto bad_unshare_cleanup_cred;
2222

2223
	if (new_fs || new_fd || do_sysvsem || new_cred || new_nsproxy) {
2224 2225 2226 2227 2228 2229
		if (do_sysvsem) {
			/*
			 * CLONE_SYSVSEM is equivalent to sys_exit().
			 */
			exit_sem(current);
		}
2230 2231 2232 2233 2234
		if (unshare_flags & CLONE_NEWIPC) {
			/* Orphan segments in old ns (see sem above). */
			exit_shm(current);
			shm_init_task(current);
		}
S
Serge E. Hallyn 已提交
2235

A
Alan Cox 已提交
2236
		if (new_nsproxy)
2237
			switch_task_namespaces(current, new_nsproxy);
2238

2239 2240
		task_lock(current);

2241 2242
		if (new_fs) {
			fs = current->fs;
N
Nick Piggin 已提交
2243
			spin_lock(&fs->lock);
2244
			current->fs = new_fs;
A
Al Viro 已提交
2245 2246 2247 2248
			if (--fs->users)
				new_fs = NULL;
			else
				new_fs = fs;
N
Nick Piggin 已提交
2249
			spin_unlock(&fs->lock);
2250 2251 2252 2253 2254 2255 2256 2257 2258
		}

		if (new_fd) {
			fd = current->files;
			current->files = new_fd;
			new_fd = fd;
		}

		task_unlock(current);
2259 2260 2261 2262 2263 2264

		if (new_cred) {
			/* Install the new user namespace */
			commit_creds(new_cred);
			new_cred = NULL;
		}
2265 2266
	}

2267 2268 2269
bad_unshare_cleanup_cred:
	if (new_cred)
		put_cred(new_cred);
2270 2271 2272 2273 2274 2275
bad_unshare_cleanup_fd:
	if (new_fd)
		put_files_struct(new_fd);

bad_unshare_cleanup_fs:
	if (new_fs)
A
Al Viro 已提交
2276
		free_fs_struct(new_fs);
2277 2278 2279 2280

bad_unshare_out:
	return err;
}
2281 2282 2283 2284 2285 2286 2287 2288 2289 2290

/*
 *	Helper to unshare the files of the current task.
 *	We don't want to expose copy_files internals to
 *	the exec layer of the kernel.
 */

int unshare_files(struct files_struct **displaced)
{
	struct task_struct *task = current;
2291
	struct files_struct *copy = NULL;
2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304
	int error;

	error = unshare_fd(CLONE_FILES, &copy);
	if (error || !copy) {
		*displaced = NULL;
		return error;
	}
	*displaced = task->files;
	task_lock(task);
	task->files = copy;
	task_unlock(task);
	return 0;
}
2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327

int sysctl_max_threads(struct ctl_table *table, int write,
		       void __user *buffer, size_t *lenp, loff_t *ppos)
{
	struct ctl_table t;
	int ret;
	int threads = max_threads;
	int min = MIN_THREADS;
	int max = MAX_THREADS;

	t = *table;
	t.data = &threads;
	t.extra1 = &min;
	t.extra2 = &max;

	ret = proc_dointvec_minmax(&t, write, buffer, lenp, ppos);
	if (ret || !write)
		return ret;

	set_max_threads(threads);

	return 0;
}