fork.c 42.7 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>
S
Serge E. Hallyn 已提交
31
#include <linux/nsproxy.h>
32
#include <linux/capability.h>
L
Linus Torvalds 已提交
33
#include <linux/cpu.h>
34
#include <linux/cgroup.h>
L
Linus Torvalds 已提交
35
#include <linux/security.h>
36
#include <linux/hugetlb.h>
L
Linus Torvalds 已提交
37 38 39 40
#include <linux/swap.h>
#include <linux/syscalls.h>
#include <linux/jiffies.h>
#include <linux/futex.h>
41
#include <linux/compat.h>
42
#include <linux/kthread.h>
43
#include <linux/task_io_accounting_ops.h>
44
#include <linux/rcupdate.h>
L
Linus Torvalds 已提交
45 46 47
#include <linux/ptrace.h>
#include <linux/mount.h>
#include <linux/audit.h>
48
#include <linux/memcontrol.h>
49
#include <linux/ftrace.h>
L
Linus Torvalds 已提交
50 51
#include <linux/profile.h>
#include <linux/rmap.h>
H
Hugh Dickins 已提交
52
#include <linux/ksm.h>
L
Linus Torvalds 已提交
53
#include <linux/acct.h>
54
#include <linux/tsacct_kern.h>
M
Matt Helsley 已提交
55
#include <linux/cn_proc.h>
R
Rafael J. Wysocki 已提交
56
#include <linux/freezer.h>
57
#include <linux/delayacct.h>
58
#include <linux/taskstats_kern.h>
59
#include <linux/random.h>
M
Miloslav Trmac 已提交
60
#include <linux/tty.h>
61
#include <linux/blkdev.h>
62
#include <linux/fs_struct.h>
63
#include <linux/magic.h>
64
#include <linux/perf_event.h>
65
#include <linux/posix-timers.h>
66
#include <linux/user-return-notifier.h>
Y
Ying Han 已提交
67
#include <linux/oom.h>
A
Andrea Arcangeli 已提交
68
#include <linux/khugepaged.h>
69
#include <linux/signalfd.h>
L
Linus Torvalds 已提交
70 71 72 73 74 75 76 77

#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>

78 79
#include <trace/events/sched.h>

80 81 82
#define CREATE_TRACE_POINTS
#include <trace/events/task.h>

L
Linus Torvalds 已提交
83 84 85 86
/*
 * Protected counters by write_lock_irq(&tasklist_lock)
 */
unsigned long total_forks;	/* Handle normal Linux uptimes. */
87
int nr_threads;			/* The idle threads do not count.. */
L
Linus Torvalds 已提交
88 89 90 91 92

int max_threads;		/* tunable limit on nr_threads */

DEFINE_PER_CPU(unsigned long, process_counts) = 0;

93
__cacheline_aligned DEFINE_RWLOCK(tasklist_lock);  /* outer */
94 95 96 97 98 99 100 101

#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
102 103 104 105 106 107

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

108
	for_each_possible_cpu(cpu)
L
Linus Torvalds 已提交
109 110 111 112 113 114
		total += per_cpu(process_counts, cpu);

	return total;
}

#ifndef __HAVE_ARCH_TASK_STRUCT_ALLOCATOR
115 116 117 118
# define alloc_task_struct_node(node)		\
		kmem_cache_alloc_node(task_struct_cachep, GFP_KERNEL, node)
# define free_task_struct(tsk)			\
		kmem_cache_free(task_struct_cachep, (tsk))
119
static struct kmem_cache *task_struct_cachep;
L
Linus Torvalds 已提交
120 121
#endif

122
#ifndef __HAVE_ARCH_THREAD_INFO_ALLOCATOR
123 124
static struct thread_info *alloc_thread_info_node(struct task_struct *tsk,
						  int node)
125 126 127 128 129 130
{
#ifdef CONFIG_DEBUG_STACK_USAGE
	gfp_t mask = GFP_KERNEL | __GFP_ZERO;
#else
	gfp_t mask = GFP_KERNEL;
#endif
131 132 133
	struct page *page = alloc_pages_node(node, mask, THREAD_SIZE_ORDER);

	return page ? page_address(page) : NULL;
134 135 136 137 138 139 140 141
}

static inline void free_thread_info(struct thread_info *ti)
{
	free_pages((unsigned long)ti, THREAD_SIZE_ORDER);
}
#endif

L
Linus Torvalds 已提交
142
/* SLAB cache for signal_struct structures (tsk->signal) */
143
static struct kmem_cache *signal_cachep;
L
Linus Torvalds 已提交
144 145

/* SLAB cache for sighand_struct structures (tsk->sighand) */
146
struct kmem_cache *sighand_cachep;
L
Linus Torvalds 已提交
147 148

/* SLAB cache for files_struct structures (tsk->files) */
149
struct kmem_cache *files_cachep;
L
Linus Torvalds 已提交
150 151

/* SLAB cache for fs_struct structures (tsk->fs) */
152
struct kmem_cache *fs_cachep;
L
Linus Torvalds 已提交
153 154

/* SLAB cache for vm_area_struct structures */
155
struct kmem_cache *vm_area_cachep;
L
Linus Torvalds 已提交
156 157

/* SLAB cache for mm_struct structures (tsk->mm) */
158
static struct kmem_cache *mm_cachep;
L
Linus Torvalds 已提交
159

160 161 162 163 164 165 166
static void account_kernel_stack(struct thread_info *ti, int account)
{
	struct zone *zone = page_zone(virt_to_page(ti));

	mod_zone_page_state(zone, NR_KERNEL_STACK, account);
}

L
Linus Torvalds 已提交
167 168
void free_task(struct task_struct *tsk)
{
169
	account_kernel_stack(tsk->stack, -1);
R
Roman Zippel 已提交
170
	free_thread_info(tsk->stack);
I
Ingo Molnar 已提交
171
	rt_mutex_debug_task_free(tsk);
172
	ftrace_graph_exit_task(tsk);
L
Linus Torvalds 已提交
173 174 175 176
	free_task_struct(tsk);
}
EXPORT_SYMBOL(free_task);

177 178
static inline void free_signal_struct(struct signal_struct *sig)
{
179
	taskstats_tgid_free(sig);
180
	sched_autogroup_exit(sig);
181 182 183 184 185
	kmem_cache_free(signal_cachep, sig);
}

static inline void put_signal_struct(struct signal_struct *sig)
{
186
	if (atomic_dec_and_test(&sig->sigcnt))
187 188 189
		free_signal_struct(sig);
}

190
void __put_task_struct(struct task_struct *tsk)
L
Linus Torvalds 已提交
191
{
E
Eugene Teo 已提交
192
	WARN_ON(!tsk->exit_state);
L
Linus Torvalds 已提交
193 194 195
	WARN_ON(atomic_read(&tsk->usage));
	WARN_ON(tsk == current);

196
	exit_creds(tsk);
197
	delayacct_tsk_free(tsk);
198
	put_signal_struct(tsk->signal);
L
Linus Torvalds 已提交
199 200 201 202

	if (!profile_handoff_task(tsk))
		free_task(tsk);
}
203
EXPORT_SYMBOL_GPL(__put_task_struct);
L
Linus Torvalds 已提交
204

S
Suresh Siddha 已提交
205 206 207 208 209 210 211
/*
 * macro override instead of weak attribute alias, to workaround
 * gcc 4.1.0 and 4.1.1 bugs with weak attribute and empty functions.
 */
#ifndef arch_task_cache_init
#define arch_task_cache_init()
#endif
212

L
Linus Torvalds 已提交
213 214 215 216 217 218 219 220 221
void __init fork_init(unsigned long mempages)
{
#ifndef __HAVE_ARCH_TASK_STRUCT_ALLOCATOR
#ifndef ARCH_MIN_TASKALIGN
#define ARCH_MIN_TASKALIGN	L1_CACHE_BYTES
#endif
	/* create a slab on which task_structs can be allocated */
	task_struct_cachep =
		kmem_cache_create("task_struct", sizeof(struct task_struct),
V
Vegard Nossum 已提交
222
			ARCH_MIN_TASKALIGN, SLAB_PANIC | SLAB_NOTRACK, NULL);
L
Linus Torvalds 已提交
223 224
#endif

225 226 227
	/* do the arch specific task caches init */
	arch_task_cache_init();

L
Linus Torvalds 已提交
228 229 230 231 232 233 234 235 236 237
	/*
	 * The default maximum number of threads is set to a safe
	 * value: the thread structures can take up at most half
	 * of memory.
	 */
	max_threads = mempages / (8 * THREAD_SIZE / PAGE_SIZE);

	/*
	 * we need to allow at least 20 threads to boot a system
	 */
238
	if (max_threads < 20)
L
Linus Torvalds 已提交
239 240 241 242 243 244 245 246
		max_threads = 20;

	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];
}

247 248 249 250 251 252 253
int __attribute__((weak)) arch_dup_task_struct(struct task_struct *dst,
					       struct task_struct *src)
{
	*dst = *src;
	return 0;
}

L
Linus Torvalds 已提交
254 255 256 257
static struct task_struct *dup_task_struct(struct task_struct *orig)
{
	struct task_struct *tsk;
	struct thread_info *ti;
258
	unsigned long *stackend;
259
	int node = tsk_fork_get_node(orig);
P
Peter Zijlstra 已提交
260
	int err;
L
Linus Torvalds 已提交
261 262 263

	prepare_to_copy(orig);

264
	tsk = alloc_task_struct_node(node);
L
Linus Torvalds 已提交
265 266 267
	if (!tsk)
		return NULL;

268
	ti = alloc_thread_info_node(tsk, node);
L
Linus Torvalds 已提交
269 270 271 272 273
	if (!ti) {
		free_task_struct(tsk);
		return NULL;
	}

274
	err = arch_dup_task_struct(tsk, orig);
275 276 277
	if (err)
		goto out;

R
Roman Zippel 已提交
278
	tsk->stack = ti;
P
Peter Zijlstra 已提交
279

280
	setup_thread_stack(tsk, orig);
281
	clear_user_return_notifier(tsk);
282
	clear_tsk_need_resched(tsk);
283 284
	stackend = end_of_stack(tsk);
	*stackend = STACK_END_MAGIC;	/* for overflow detection */
L
Linus Torvalds 已提交
285

286 287 288 289
#ifdef CONFIG_CC_STACKPROTECTOR
	tsk->stack_canary = get_random_int();
#endif

290 291 292 293 294
	/*
	 * One for us, one for whoever does the "release_task()" (usually
	 * parent)
	 */
	atomic_set(&tsk->usage, 2);
295
#ifdef CONFIG_BLK_DEV_IO_TRACE
296
	tsk->btrace_seq = 0;
297
#endif
298
	tsk->splice_pipe = NULL;
299 300 301

	account_kernel_stack(ti, 1);

L
Linus Torvalds 已提交
302
	return tsk;
303 304 305 306 307

out:
	free_thread_info(ti);
	free_task_struct(tsk);
	return NULL;
L
Linus Torvalds 已提交
308 309 310
}

#ifdef CONFIG_MMU
A
Alexey Dobriyan 已提交
311
static int dup_mmap(struct mm_struct *mm, struct mm_struct *oldmm)
L
Linus Torvalds 已提交
312
{
313
	struct vm_area_struct *mpnt, *tmp, *prev, **pprev;
L
Linus Torvalds 已提交
314 315 316 317 318 319
	struct rb_node **rb_link, *rb_parent;
	int retval;
	unsigned long charge;
	struct mempolicy *pol;

	down_write(&oldmm->mmap_sem);
320
	flush_cache_dup_mm(oldmm);
321 322 323 324
	/*
	 * Not linked in yet - no deadlock potential:
	 */
	down_write_nested(&mm->mmap_sem, SINGLE_DEPTH_NESTING);
325

L
Linus Torvalds 已提交
326 327 328 329
	mm->locked_vm = 0;
	mm->mmap = NULL;
	mm->mmap_cache = NULL;
	mm->free_area_cache = oldmm->mmap_base;
330
	mm->cached_hole_size = ~0UL;
L
Linus Torvalds 已提交
331
	mm->map_count = 0;
332
	cpumask_clear(mm_cpumask(mm));
L
Linus Torvalds 已提交
333 334 335 336
	mm->mm_rb = RB_ROOT;
	rb_link = &mm->mm_rb.rb_node;
	rb_parent = NULL;
	pprev = &mm->mmap;
H
Hugh Dickins 已提交
337
	retval = ksm_fork(mm, oldmm);
A
Andrea Arcangeli 已提交
338 339 340
	if (retval)
		goto out;
	retval = khugepaged_fork(mm, oldmm);
H
Hugh Dickins 已提交
341 342
	if (retval)
		goto out;
L
Linus Torvalds 已提交
343

344
	prev = NULL;
H
Hugh Dickins 已提交
345
	for (mpnt = oldmm->mmap; mpnt; mpnt = mpnt->vm_next) {
L
Linus Torvalds 已提交
346 347 348
		struct file *file;

		if (mpnt->vm_flags & VM_DONTCOPY) {
349 350
			long pages = vma_pages(mpnt);
			mm->total_vm -= pages;
351
			vm_stat_account(mm, mpnt->vm_flags, mpnt->vm_file,
352
								-pages);
L
Linus Torvalds 已提交
353 354 355 356 357 358 359 360 361
			continue;
		}
		charge = 0;
		if (mpnt->vm_flags & VM_ACCOUNT) {
			unsigned int len = (mpnt->vm_end - mpnt->vm_start) >> PAGE_SHIFT;
			if (security_vm_enough_memory(len))
				goto fail_nomem;
			charge = len;
		}
362
		tmp = kmem_cache_alloc(vm_area_cachep, GFP_KERNEL);
L
Linus Torvalds 已提交
363 364 365
		if (!tmp)
			goto fail_nomem;
		*tmp = *mpnt;
366
		INIT_LIST_HEAD(&tmp->anon_vma_chain);
367
		pol = mpol_dup(vma_policy(mpnt));
L
Linus Torvalds 已提交
368 369 370 371
		retval = PTR_ERR(pol);
		if (IS_ERR(pol))
			goto fail_nomem_policy;
		vma_set_policy(tmp, pol);
A
Andrea Arcangeli 已提交
372
		tmp->vm_mm = mm;
373 374
		if (anon_vma_fork(tmp, mpnt))
			goto fail_nomem_anon_vma_fork;
L
Linus Torvalds 已提交
375
		tmp->vm_flags &= ~VM_LOCKED;
376
		tmp->vm_next = tmp->vm_prev = NULL;
L
Linus Torvalds 已提交
377 378
		file = tmp->vm_file;
		if (file) {
379
			struct inode *inode = file->f_path.dentry->d_inode;
H
Hugh Dickins 已提交
380 381
			struct address_space *mapping = file->f_mapping;

L
Linus Torvalds 已提交
382 383 384
			get_file(file);
			if (tmp->vm_flags & VM_DENYWRITE)
				atomic_dec(&inode->i_writecount);
385
			mutex_lock(&mapping->i_mmap_mutex);
H
Hugh Dickins 已提交
386 387 388 389
			if (tmp->vm_flags & VM_SHARED)
				mapping->i_mmap_writable++;
			flush_dcache_mmap_lock(mapping);
			/* insert tmp into the share list, just after mpnt */
L
Linus Torvalds 已提交
390
			vma_prio_tree_add(tmp, mpnt);
H
Hugh Dickins 已提交
391
			flush_dcache_mmap_unlock(mapping);
392
			mutex_unlock(&mapping->i_mmap_mutex);
L
Linus Torvalds 已提交
393 394
		}

395 396 397 398 399 400 401 402
		/*
		 * 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
403
		/*
404
		 * Link in the new vma and copy the page table entries.
L
Linus Torvalds 已提交
405 406 407
		 */
		*pprev = tmp;
		pprev = &tmp->vm_next;
408 409
		tmp->vm_prev = prev;
		prev = tmp;
L
Linus Torvalds 已提交
410 411 412 413 414 415

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

		mm->map_count++;
416
		retval = copy_page_range(mm, oldmm, mpnt);
L
Linus Torvalds 已提交
417 418 419 420 421 422 423

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

		if (retval)
			goto out;
	}
424 425
	/* a new mm has just been created */
	arch_dup_mmap(oldmm, mm);
L
Linus Torvalds 已提交
426 427
	retval = 0;
out:
428
	up_write(&mm->mmap_sem);
H
Hugh Dickins 已提交
429
	flush_tlb_mm(oldmm);
L
Linus Torvalds 已提交
430 431
	up_write(&oldmm->mmap_sem);
	return retval;
432 433
fail_nomem_anon_vma_fork:
	mpol_put(pol);
L
Linus Torvalds 已提交
434 435 436 437 438 439 440 441
fail_nomem_policy:
	kmem_cache_free(vm_area_cachep, tmp);
fail_nomem:
	retval = -ENOMEM;
	vm_unacct_memory(charge);
	goto out;
}

442
static inline int mm_alloc_pgd(struct mm_struct *mm)
L
Linus Torvalds 已提交
443 444 445 446 447 448 449
{
	mm->pgd = pgd_alloc(mm);
	if (unlikely(!mm->pgd))
		return -ENOMEM;
	return 0;
}

450
static inline void mm_free_pgd(struct mm_struct *mm)
L
Linus Torvalds 已提交
451
{
452
	pgd_free(mm, mm->pgd);
L
Linus Torvalds 已提交
453 454 455 456 457 458 459
}
#else
#define dup_mmap(mm, oldmm)	(0)
#define mm_alloc_pgd(mm)	(0)
#define mm_free_pgd(mm)
#endif /* CONFIG_MMU */

D
Daniel Walker 已提交
460
__cacheline_aligned_in_smp DEFINE_SPINLOCK(mmlist_lock);
L
Linus Torvalds 已提交
461

462
#define allocate_mm()	(kmem_cache_alloc(mm_cachep, GFP_KERNEL))
L
Linus Torvalds 已提交
463 464
#define free_mm(mm)	(kmem_cache_free(mm_cachep, (mm)))

465 466 467 468 469 470 471 472 473 474 475 476
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
477 478
#include <linux/init_task.h>

A
Alexey Dobriyan 已提交
479 480 481 482 483 484 485 486
static void mm_init_aio(struct mm_struct *mm)
{
#ifdef CONFIG_AIO
	spin_lock_init(&mm->ioctx_lock);
	INIT_HLIST_HEAD(&mm->ioctx_list);
#endif
}

487
static struct mm_struct *mm_init(struct mm_struct *mm, struct task_struct *p)
L
Linus Torvalds 已提交
488 489 490 491 492
{
	atomic_set(&mm->mm_users, 1);
	atomic_set(&mm->mm_count, 1);
	init_rwsem(&mm->mmap_sem);
	INIT_LIST_HEAD(&mm->mmlist);
H
Hugh Dickins 已提交
493 494
	mm->flags = (current->mm) ?
		(current->mm->flags & MMF_INIT_MASK) : default_dump_filter;
495
	mm->core_state = NULL;
L
Linus Torvalds 已提交
496
	mm->nr_ptes = 0;
K
KAMEZAWA Hiroyuki 已提交
497
	memset(&mm->rss_stat, 0, sizeof(mm->rss_stat));
L
Linus Torvalds 已提交
498 499
	spin_lock_init(&mm->page_table_lock);
	mm->free_area_cache = TASK_UNMAPPED_BASE;
500
	mm->cached_hole_size = ~0UL;
A
Alexey Dobriyan 已提交
501
	mm_init_aio(mm);
502
	mm_init_owner(mm, p);
L
Linus Torvalds 已提交
503 504 505

	if (likely(!mm_alloc_pgd(mm))) {
		mm->def_flags = 0;
A
Andrea Arcangeli 已提交
506
		mmu_notifier_mm_init(mm);
L
Linus Torvalds 已提交
507 508
		return mm;
	}
509

L
Linus Torvalds 已提交
510 511 512 513 514 515 516
	free_mm(mm);
	return NULL;
}

/*
 * Allocate and initialize an mm_struct.
 */
517
struct mm_struct *mm_alloc(void)
L
Linus Torvalds 已提交
518
{
519
	struct mm_struct *mm;
L
Linus Torvalds 已提交
520 521

	mm = allocate_mm();
522 523 524 525
	if (!mm)
		return NULL;

	memset(mm, 0, sizeof(*mm));
526 527
	mm_init_cpumask(mm);
	return mm_init(mm, current);
L
Linus Torvalds 已提交
528 529 530 531 532 533 534
}

/*
 * 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.
 */
535
void __mmdrop(struct mm_struct *mm)
L
Linus Torvalds 已提交
536 537 538 539
{
	BUG_ON(mm == &init_mm);
	mm_free_pgd(mm);
	destroy_context(mm);
A
Andrea Arcangeli 已提交
540
	mmu_notifier_mm_destroy(mm);
541 542 543
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
	VM_BUG_ON(mm->pmd_huge_pte);
#endif
L
Linus Torvalds 已提交
544 545
	free_mm(mm);
}
546
EXPORT_SYMBOL_GPL(__mmdrop);
L
Linus Torvalds 已提交
547 548 549 550 551 552

/*
 * Decrement the use count and release all resources for an mm.
 */
void mmput(struct mm_struct *mm)
{
A
Andrew Morton 已提交
553 554
	might_sleep();

L
Linus Torvalds 已提交
555 556
	if (atomic_dec_and_test(&mm->mm_users)) {
		exit_aio(mm);
557
		ksm_exit(mm);
A
Andrea Arcangeli 已提交
558
		khugepaged_exit(mm); /* must run before exit_mmap */
L
Linus Torvalds 已提交
559
		exit_mmap(mm);
M
Matt Helsley 已提交
560
		set_mm_exe_file(mm, NULL);
L
Linus Torvalds 已提交
561 562 563 564 565 566
		if (!list_empty(&mm->mmlist)) {
			spin_lock(&mmlist_lock);
			list_del(&mm->mmlist);
			spin_unlock(&mmlist_lock);
		}
		put_swap_token(mm);
567 568
		if (mm->binfmt)
			module_put(mm->binfmt->module);
L
Linus Torvalds 已提交
569 570 571 572 573
		mmdrop(mm);
	}
}
EXPORT_SYMBOL_GPL(mmput);

574 575 576 577 578 579 580 581 582 583 584 585 586
/*
 * We added or removed a vma mapping the executable. The vmas are only mapped
 * during exec and are not mapped with the mmap system call.
 * Callers must hold down_write() on the mm's mmap_sem for these
 */
void added_exe_file_vma(struct mm_struct *mm)
{
	mm->num_exe_file_vmas++;
}

void removed_exe_file_vma(struct mm_struct *mm)
{
	mm->num_exe_file_vmas--;
587
	if ((mm->num_exe_file_vmas == 0) && mm->exe_file) {
588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624
		fput(mm->exe_file);
		mm->exe_file = NULL;
	}

}

void set_mm_exe_file(struct mm_struct *mm, struct file *new_exe_file)
{
	if (new_exe_file)
		get_file(new_exe_file);
	if (mm->exe_file)
		fput(mm->exe_file);
	mm->exe_file = new_exe_file;
	mm->num_exe_file_vmas = 0;
}

struct file *get_mm_exe_file(struct mm_struct *mm)
{
	struct file *exe_file;

	/* We need mmap_sem to protect against races with removal of
	 * VM_EXECUTABLE vmas */
	down_read(&mm->mmap_sem);
	exe_file = mm->exe_file;
	if (exe_file)
		get_file(exe_file);
	up_read(&mm->mmap_sem);
	return exe_file;
}

static void dup_mm_exe_file(struct mm_struct *oldmm, struct mm_struct *newmm)
{
	/* It's safe to write the exe_file pointer without exe_file_lock because
	 * this is called during fork when the task is not yet in /proc */
	newmm->exe_file = get_mm_exe_file(oldmm);
}

L
Linus Torvalds 已提交
625 626 627
/**
 * get_task_mm - acquire a reference to the task's mm
 *
628
 * Returns %NULL if the task has no mm.  Checks PF_KTHREAD (meaning
L
Linus Torvalds 已提交
629 630 631 632 633 634 635 636 637 638 639 640
 * 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) {
641
		if (task->flags & PF_KTHREAD)
L
Linus Torvalds 已提交
642 643 644 645 646 647 648 649 650
			mm = NULL;
		else
			atomic_inc(&mm->mm_users);
	}
	task_unlock(task);
	return mm;
}
EXPORT_SYMBOL_GPL(get_task_mm);

651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670
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;
}

671 672 673 674 675 676 677 678
void complete_vfork_done(struct task_struct *tsk)
{
	struct completion *vfork_done = tsk->vfork_done;

	tsk->vfork_done = NULL;
	complete(vfork_done);
}

L
Linus Torvalds 已提交
679 680 681 682 683 684 685 686 687 688 689 690 691 692 693
/* 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)
{
694 695
	/* Get rid of any futexes when releasing the mm */
#ifdef CONFIG_FUTEX
696
	if (unlikely(tsk->robust_list)) {
697
		exit_robust_list(tsk);
698 699
		tsk->robust_list = NULL;
	}
700
#ifdef CONFIG_COMPAT
701
	if (unlikely(tsk->compat_robust_list)) {
702
		compat_exit_robust_list(tsk);
703 704
		tsk->compat_robust_list = NULL;
	}
705
#endif
706 707
	if (unlikely(!list_empty(&tsk->pi_state_list)))
		exit_pi_state_list(tsk);
708 709
#endif

L
Linus Torvalds 已提交
710 711 712
	/* Get rid of any cached register state */
	deactivate_mm(tsk, mm);

713 714
	if (tsk->vfork_done)
		complete_vfork_done(tsk);
715 716 717 718 719 720 721

	/*
	 * If we're exiting normally, clear a user-space tid field if
	 * requested.  We leave this alone when dying by signal, to leave
	 * the value intact in a core dump, and to save the unnecessary
	 * trouble otherwise.  Userland only wants this done for a sys_exit.
	 */
722 723 724 725 726 727 728 729 730 731 732
	if (tsk->clear_child_tid) {
		if (!(tsk->flags & PF_SIGNALED) &&
		    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
733 734 735 736
		tsk->clear_child_tid = NULL;
	}
}

737 738 739 740
/*
 * Allocate a new mm structure and copy contents from the
 * mm structure of the passed in task structure.
 */
741
struct mm_struct *dup_mm(struct task_struct *tsk)
742 743 744 745 746 747 748 749 750 751 752 753
{
	struct mm_struct *mm, *oldmm = current->mm;
	int err;

	if (!oldmm)
		return NULL;

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

	memcpy(mm, oldmm, sizeof(*mm));
754
	mm_init_cpumask(mm);
755

756 757 758 759
	/* Initializing for Swap token stuff */
	mm->token_priority = 0;
	mm->last_interval = 0;

760 761 762 763
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
	mm->pmd_huge_pte = NULL;
#endif

764
	if (!mm_init(mm, tsk))
765 766 767 768 769
		goto fail_nomem;

	if (init_new_context(tsk, mm))
		goto fail_nocontext;

M
Matt Helsley 已提交
770 771
	dup_mm_exe_file(oldmm, mm);

772 773 774 775 776 777 778
	err = dup_mmap(mm, oldmm);
	if (err)
		goto free_pt;

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

779 780 781
	if (mm->binfmt && !try_module_get(mm->binfmt->module))
		goto free_pt;

782 783 784
	return mm;

free_pt:
785 786
	/* don't put binfmt in mmput, we haven't got module yet */
	mm->binfmt = NULL;
787 788 789 790 791 792 793 794 795 796 797 798 799 800 801
	mmput(mm);

fail_nomem:
	return NULL;

fail_nocontext:
	/*
	 * If init_new_context() failed, we cannot use mmput() to free the mm
	 * because it calls destroy_context()
	 */
	mm_free_pgd(mm);
	free_mm(mm);
	return NULL;
}

802
static int copy_mm(unsigned long clone_flags, struct task_struct *tsk)
L
Linus Torvalds 已提交
803
{
804
	struct mm_struct *mm, *oldmm;
L
Linus Torvalds 已提交
805 806 807 808
	int retval;

	tsk->min_flt = tsk->maj_flt = 0;
	tsk->nvcsw = tsk->nivcsw = 0;
809 810 811
#ifdef CONFIG_DETECT_HUNG_TASK
	tsk->last_switch_count = tsk->nvcsw + tsk->nivcsw;
#endif
L
Linus Torvalds 已提交
812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831

	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;

	if (clone_flags & CLONE_VM) {
		atomic_inc(&oldmm->mm_users);
		mm = oldmm;
		goto good_mm;
	}

	retval = -ENOMEM;
832
	mm = dup_mm(tsk);
L
Linus Torvalds 已提交
833 834 835 836
	if (!mm)
		goto fail_nomem;

good_mm:
837 838 839 840
	/* Initializing for Swap token stuff */
	mm->token_priority = 0;
	mm->last_interval = 0;

L
Linus Torvalds 已提交
841 842 843 844 845 846 847 848
	tsk->mm = mm;
	tsk->active_mm = mm;
	return 0;

fail_nomem:
	return retval;
}

A
Alexey Dobriyan 已提交
849
static int copy_fs(unsigned long clone_flags, struct task_struct *tsk)
L
Linus Torvalds 已提交
850
{
A
Al Viro 已提交
851
	struct fs_struct *fs = current->fs;
L
Linus Torvalds 已提交
852
	if (clone_flags & CLONE_FS) {
A
Al Viro 已提交
853
		/* tsk->fs is already what we want */
N
Nick Piggin 已提交
854
		spin_lock(&fs->lock);
A
Al Viro 已提交
855
		if (fs->in_exec) {
N
Nick Piggin 已提交
856
			spin_unlock(&fs->lock);
A
Al Viro 已提交
857 858 859
			return -EAGAIN;
		}
		fs->users++;
N
Nick Piggin 已提交
860
		spin_unlock(&fs->lock);
L
Linus Torvalds 已提交
861 862
		return 0;
	}
A
Al Viro 已提交
863
	tsk->fs = copy_fs_struct(fs);
L
Linus Torvalds 已提交
864 865 866 867 868
	if (!tsk->fs)
		return -ENOMEM;
	return 0;
}

869
static int copy_files(unsigned long clone_flags, struct task_struct *tsk)
870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895
{
	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;
}

896
static int copy_io(unsigned long clone_flags, struct task_struct *tsk)
897 898 899
{
#ifdef CONFIG_BLOCK
	struct io_context *ioc = current->io_context;
900
	struct io_context *new_ioc;
901 902 903

	if (!ioc)
		return 0;
904 905 906 907 908 909 910 911
	/*
	 * Share io context with parent, if CLONE_IO is set
	 */
	if (clone_flags & CLONE_IO) {
		tsk->io_context = ioc_task_link(ioc);
		if (unlikely(!tsk->io_context))
			return -ENOMEM;
	} else if (ioprio_valid(ioc->ioprio)) {
912 913
		new_ioc = get_task_io_context(tsk, GFP_KERNEL, NUMA_NO_NODE);
		if (unlikely(!new_ioc))
914 915
			return -ENOMEM;

916
		new_ioc->ioprio = ioc->ioprio;
917
		put_io_context(new_ioc);
918 919 920 921 922
	}
#endif
	return 0;
}

A
Alexey Dobriyan 已提交
923
static int copy_sighand(unsigned long clone_flags, struct task_struct *tsk)
L
Linus Torvalds 已提交
924 925 926
{
	struct sighand_struct *sig;

Z
Zhaolei 已提交
927
	if (clone_flags & CLONE_SIGHAND) {
L
Linus Torvalds 已提交
928 929 930 931
		atomic_inc(&current->sighand->count);
		return 0;
	}
	sig = kmem_cache_alloc(sighand_cachep, GFP_KERNEL);
I
Ingo Molnar 已提交
932
	rcu_assign_pointer(tsk->sighand, sig);
L
Linus Torvalds 已提交
933 934 935 936 937 938 939
	if (!sig)
		return -ENOMEM;
	atomic_set(&sig->count, 1);
	memcpy(sig->action, current->sighand->action, sizeof(sig->action));
	return 0;
}

940
void __cleanup_sighand(struct sighand_struct *sighand)
941
{
942 943
	if (atomic_dec_and_test(&sighand->count)) {
		signalfd_cleanup(sighand);
944
		kmem_cache_free(sighand_cachep, sighand);
945
	}
946 947
}

948 949 950 951 952 953

/*
 * Initialize POSIX timer handling for a thread group.
 */
static void posix_cpu_timers_init_group(struct signal_struct *sig)
{
954 955
	unsigned long cpu_limit;

956 957 958
	/* Thread group counters. */
	thread_group_cputime_init(sig);

959 960 961
	cpu_limit = ACCESS_ONCE(sig->rlim[RLIMIT_CPU].rlim_cur);
	if (cpu_limit != RLIM_INFINITY) {
		sig->cputime_expires.prof_exp = secs_to_cputime(cpu_limit);
962 963 964
		sig->cputimer.running = 1;
	}

965 966 967 968 969 970
	/* 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
971
static int copy_signal(unsigned long clone_flags, struct task_struct *tsk)
L
Linus Torvalds 已提交
972 973 974
{
	struct signal_struct *sig;

975
	if (clone_flags & CLONE_THREAD)
976 977
		return 0;

978
	sig = kmem_cache_zalloc(signal_cachep, GFP_KERNEL);
L
Linus Torvalds 已提交
979 980 981 982
	tsk->signal = sig;
	if (!sig)
		return -ENOMEM;

983
	sig->nr_threads = 1;
L
Linus Torvalds 已提交
984
	atomic_set(&sig->live, 1);
985
	atomic_set(&sig->sigcnt, 1);
L
Linus Torvalds 已提交
986
	init_waitqueue_head(&sig->wait_chldexit);
987 988
	if (clone_flags & CLONE_NEWPID)
		sig->flags |= SIGNAL_UNKILLABLE;
989
	sig->curr_target = tsk;
L
Linus Torvalds 已提交
990 991 992
	init_sigpending(&sig->shared_pending);
	INIT_LIST_HEAD(&sig->posix_timers);

993
	hrtimer_init(&sig->real_timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
L
Linus Torvalds 已提交
994 995 996 997 998 999
	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);

1000 1001
	posix_cpu_timers_init_group(sig);

M
Miloslav Trmac 已提交
1002
	tty_audit_fork(sig);
1003
	sched_autogroup_fork(sig);
M
Miloslav Trmac 已提交
1004

1005
#ifdef CONFIG_CGROUPS
1006
	init_rwsem(&sig->group_rwsem);
1007 1008
#endif

1009
	sig->oom_adj = current->signal->oom_adj;
D
David Rientjes 已提交
1010
	sig->oom_score_adj = current->signal->oom_score_adj;
1011
	sig->oom_score_adj_min = current->signal->oom_score_adj_min;
1012

1013 1014
	mutex_init(&sig->cred_guard_mutex);

L
Linus Torvalds 已提交
1015 1016 1017
	return 0;
}

A
Alexey Dobriyan 已提交
1018
static void copy_flags(unsigned long clone_flags, struct task_struct *p)
L
Linus Torvalds 已提交
1019 1020 1021
{
	unsigned long new_flags = p->flags;

T
Tejun Heo 已提交
1022
	new_flags &= ~(PF_SUPERPRIV | PF_WQ_WORKER);
L
Linus Torvalds 已提交
1023
	new_flags |= PF_FORKNOEXEC;
R
Roland McGrath 已提交
1024
	new_flags |= PF_STARTING;
L
Linus Torvalds 已提交
1025 1026 1027
	p->flags = new_flags;
}

1028
SYSCALL_DEFINE1(set_tid_address, int __user *, tidptr)
L
Linus Torvalds 已提交
1029 1030 1031
{
	current->clear_child_tid = tidptr;

1032
	return task_pid_vnr(current);
L
Linus Torvalds 已提交
1033 1034
}

A
Alexey Dobriyan 已提交
1035
static void rt_mutex_init_task(struct task_struct *p)
I
Ingo Molnar 已提交
1036
{
1037
	raw_spin_lock_init(&p->pi_lock);
1038
#ifdef CONFIG_RT_MUTEXES
1039
	plist_head_init(&p->pi_waiters);
I
Ingo Molnar 已提交
1040 1041 1042 1043
	p->pi_blocked_on = NULL;
#endif
}

1044 1045 1046 1047 1048 1049 1050
#ifdef CONFIG_MM_OWNER
void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
{
	mm->owner = p;
}
#endif /* CONFIG_MM_OWNER */

1051 1052 1053 1054 1055
/*
 * Initialize POSIX timer handling for a single task.
 */
static void posix_cpu_timers_init(struct task_struct *tsk)
{
1056 1057
	tsk->cputime_expires.prof_exp = 0;
	tsk->cputime_expires.virt_exp = 0;
1058 1059 1060 1061 1062 1063
	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]);
}

L
Linus Torvalds 已提交
1064 1065 1066 1067 1068 1069 1070 1071
/*
 * 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.
 */
1072 1073 1074 1075 1076
static struct task_struct *copy_process(unsigned long clone_flags,
					unsigned long stack_start,
					struct pt_regs *regs,
					unsigned long stack_size,
					int __user *child_tidptr,
R
Roland McGrath 已提交
1077 1078
					struct pid *pid,
					int trace)
L
Linus Torvalds 已提交
1079 1080
{
	int retval;
1081
	struct task_struct *p;
1082
	int cgroup_callbacks_done = 0;
L
Linus Torvalds 已提交
1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101

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

	/*
	 * 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);

1102 1103 1104 1105 1106 1107 1108 1109 1110 1111
	/*
	 * 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);

L
Linus Torvalds 已提交
1112 1113 1114 1115 1116 1117 1118 1119 1120
	retval = security_task_create(clone_flags);
	if (retval)
		goto fork_out;

	retval = -ENOMEM;
	p = dup_task_struct(current);
	if (!p)
		goto fork_out;

1121 1122
	ftrace_graph_init_task(p);

1123 1124
	rt_mutex_init_task(p);

I
Ingo Molnar 已提交
1125
#ifdef CONFIG_PROVE_LOCKING
1126 1127 1128
	DEBUG_LOCKS_WARN_ON(!p->hardirqs_enabled);
	DEBUG_LOCKS_WARN_ON(!p->softirqs_enabled);
#endif
L
Linus Torvalds 已提交
1129
	retval = -EAGAIN;
1130
	if (atomic_read(&p->real_cred->user->processes) >=
1131
			task_rlimit(p, RLIMIT_NPROC)) {
L
Linus Torvalds 已提交
1132
		if (!capable(CAP_SYS_ADMIN) && !capable(CAP_SYS_RESOURCE) &&
1133
		    p->real_cred->user != INIT_USER)
L
Linus Torvalds 已提交
1134 1135
			goto bad_fork_free;
	}
1136
	current->flags &= ~PF_NPROC_EXCEEDED;
L
Linus Torvalds 已提交
1137

1138 1139 1140
	retval = copy_creds(p, clone_flags);
	if (retval < 0)
		goto bad_fork_free;
L
Linus Torvalds 已提交
1141 1142 1143 1144 1145 1146

	/*
	 * 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.
	 */
1147
	retval = -EAGAIN;
L
Linus Torvalds 已提交
1148 1149 1150
	if (nr_threads >= max_threads)
		goto bad_fork_cleanup_count;

A
Al Viro 已提交
1151
	if (!try_module_get(task_thread_info(p)->exec_domain->module))
L
Linus Torvalds 已提交
1152 1153 1154
		goto bad_fork_cleanup_count;

	p->did_exec = 0;
1155
	delayacct_tsk_init(p);	/* Must remain after dup_task_struct() */
L
Linus Torvalds 已提交
1156 1157 1158
	copy_flags(clone_flags, p);
	INIT_LIST_HEAD(&p->children);
	INIT_LIST_HEAD(&p->sibling);
1159
	rcu_copy_process(p);
L
Linus Torvalds 已提交
1160 1161 1162 1163 1164
	p->vfork_done = NULL;
	spin_lock_init(&p->alloc_lock);

	init_sigpending(&p->pending);

1165 1166
	p->utime = p->stime = p->gtime = 0;
	p->utimescaled = p->stimescaled = 0;
1167
#ifndef CONFIG_VIRT_CPU_ACCOUNTING
1168
	p->prev_utime = p->prev_stime = 0;
1169
#endif
1170 1171 1172
#if defined(SPLIT_RSS_COUNTING)
	memset(&p->rss_stat, 0, sizeof(p->rss_stat));
#endif
1173

1174 1175
	p->default_timer_slack_ns = current->timer_slack_ns;

1176
	task_io_accounting_init(&p->ioac);
L
Linus Torvalds 已提交
1177 1178
	acct_clear_integrals(p);

1179
	posix_cpu_timers_init(p);
L
Linus Torvalds 已提交
1180 1181

	do_posix_clock_monotonic_gettime(&p->start_time);
1182 1183
	p->real_start_time = p->start_time;
	monotonic_to_bootbased(&p->real_start_time);
L
Linus Torvalds 已提交
1184 1185
	p->io_context = NULL;
	p->audit_context = NULL;
1186
	if (clone_flags & CLONE_THREAD)
1187
		threadgroup_change_begin(current);
1188
	cgroup_fork(p);
L
Linus Torvalds 已提交
1189
#ifdef CONFIG_NUMA
1190
	p->mempolicy = mpol_dup(p->mempolicy);
1191 1192 1193 1194 1195
	if (IS_ERR(p->mempolicy)) {
		retval = PTR_ERR(p->mempolicy);
		p->mempolicy = NULL;
		goto bad_fork_cleanup_cgroup;
	}
1196
	mpol_fix_fork_child_flag(p);
L
Linus Torvalds 已提交
1197
#endif
1198 1199 1200 1201
#ifdef CONFIG_CPUSETS
	p->cpuset_mem_spread_rotor = NUMA_NO_NODE;
	p->cpuset_slab_spread_rotor = NUMA_NO_NODE;
#endif
1202 1203
#ifdef CONFIG_TRACE_IRQFLAGS
	p->irq_events = 0;
1204 1205 1206
#ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
	p->hardirqs_enabled = 1;
#else
1207
	p->hardirqs_enabled = 0;
1208
#endif
1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220
	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
I
Ingo Molnar 已提交
1221 1222 1223 1224 1225
#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
1226

1227 1228 1229
#ifdef CONFIG_DEBUG_MUTEXES
	p->blocked_on = NULL; /* not blocked yet */
#endif
1230 1231 1232 1233
#ifdef CONFIG_CGROUP_MEM_RES_CTLR
	p->memcg_batch.do_batch = 0;
	p->memcg_batch.memcg = NULL;
#endif
1234

1235
	/* Perform scheduler related setup. Assign this task to a CPU. */
1236
	sched_fork(p);
1237

1238
	retval = perf_event_init_task(p);
1239 1240
	if (retval)
		goto bad_fork_cleanup_policy;
1241 1242
	retval = audit_alloc(p);
	if (retval)
1243
		goto bad_fork_cleanup_policy;
L
Linus Torvalds 已提交
1244
	/* copy all the process information */
1245 1246
	retval = copy_semundo(clone_flags, p);
	if (retval)
L
Linus Torvalds 已提交
1247
		goto bad_fork_cleanup_audit;
1248 1249
	retval = copy_files(clone_flags, p);
	if (retval)
L
Linus Torvalds 已提交
1250
		goto bad_fork_cleanup_semundo;
1251 1252
	retval = copy_fs(clone_flags, p);
	if (retval)
L
Linus Torvalds 已提交
1253
		goto bad_fork_cleanup_files;
1254 1255
	retval = copy_sighand(clone_flags, p);
	if (retval)
L
Linus Torvalds 已提交
1256
		goto bad_fork_cleanup_fs;
1257 1258
	retval = copy_signal(clone_flags, p);
	if (retval)
L
Linus Torvalds 已提交
1259
		goto bad_fork_cleanup_sighand;
1260 1261
	retval = copy_mm(clone_flags, p);
	if (retval)
L
Linus Torvalds 已提交
1262
		goto bad_fork_cleanup_signal;
1263 1264
	retval = copy_namespaces(clone_flags, p);
	if (retval)
D
David Howells 已提交
1265
		goto bad_fork_cleanup_mm;
1266 1267
	retval = copy_io(clone_flags, p);
	if (retval)
1268
		goto bad_fork_cleanup_namespaces;
A
Alexey Dobriyan 已提交
1269
	retval = copy_thread(clone_flags, stack_start, stack_size, p, regs);
L
Linus Torvalds 已提交
1270
	if (retval)
1271
		goto bad_fork_cleanup_io;
L
Linus Torvalds 已提交
1272

1273 1274
	if (pid != &init_struct_pid) {
		retval = -ENOMEM;
1275
		pid = alloc_pid(p->nsproxy->pid_ns);
1276
		if (!pid)
1277
			goto bad_fork_cleanup_io;
1278 1279 1280 1281 1282 1283 1284
	}

	p->pid = pid_nr(pid);
	p->tgid = p->pid;
	if (clone_flags & CLONE_THREAD)
		p->tgid = current->tgid;

L
Linus Torvalds 已提交
1285 1286 1287 1288
	p->set_child_tid = (clone_flags & CLONE_CHILD_SETTID) ? child_tidptr : NULL;
	/*
	 * Clear TID on mm_release()?
	 */
1289
	p->clear_child_tid = (clone_flags & CLONE_CHILD_CLEARTID) ? child_tidptr : NULL;
1290 1291 1292
#ifdef CONFIG_BLOCK
	p->plug = NULL;
#endif
1293
#ifdef CONFIG_FUTEX
1294 1295 1296 1297
	p->robust_list = NULL;
#ifdef CONFIG_COMPAT
	p->compat_robust_list = NULL;
#endif
1298 1299
	INIT_LIST_HEAD(&p->pi_state_list);
	p->pi_state_cache = NULL;
1300
#endif
1301 1302 1303 1304 1305 1306
	/*
	 * sigaltstack should be cleared when sharing the same VM
	 */
	if ((clone_flags & (CLONE_VM|CLONE_VFORK)) == CLONE_VM)
		p->sas_ss_sp = p->sas_ss_size = 0;

L
Linus Torvalds 已提交
1307
	/*
1308 1309
	 * Syscall tracing and stepping should be turned off in the
	 * child regardless of CLONE_PTRACE.
L
Linus Torvalds 已提交
1310
	 */
1311
	user_disable_single_step(p);
L
Linus Torvalds 已提交
1312
	clear_tsk_thread_flag(p, TIF_SYSCALL_TRACE);
1313 1314 1315
#ifdef TIF_SYSCALL_EMU
	clear_tsk_thread_flag(p, TIF_SYSCALL_EMU);
#endif
A
Arjan van de Ven 已提交
1316
	clear_all_latency_tracing(p);
L
Linus Torvalds 已提交
1317 1318 1319 1320 1321 1322

	/* ok, now we should be set up.. */
	p->exit_signal = (clone_flags & CLONE_THREAD) ? -1 : (clone_flags & CSIGNAL);
	p->pdeath_signal = 0;
	p->exit_state = 0;

1323 1324
	p->nr_dirtied = 0;
	p->nr_dirtied_pause = 128 >> (PAGE_SHIFT - 10);
1325
	p->dirty_paused_when = 0;
1326

L
Linus Torvalds 已提交
1327 1328 1329 1330 1331
	/*
	 * Ok, make it visible to the rest of the system.
	 * We dont wake it up yet.
	 */
	p->group_leader = p;
O
Oleg Nesterov 已提交
1332
	INIT_LIST_HEAD(&p->thread_group);
L
Linus Torvalds 已提交
1333

1334 1335 1336 1337 1338 1339
	/* Now that the task is set up, run cgroup callbacks if
	 * necessary. We need to run them before the task is visible
	 * on the tasklist. */
	cgroup_fork_callbacks(p);
	cgroup_callbacks_done = 1;

L
Linus Torvalds 已提交
1340 1341 1342 1343
	/* Need tasklist lock for parent etc handling! */
	write_lock_irq(&tasklist_lock);

	/* CLONE_PARENT re-uses the old parent */
1344
	if (clone_flags & (CLONE_PARENT|CLONE_THREAD)) {
L
Linus Torvalds 已提交
1345
		p->real_parent = current->real_parent;
1346 1347
		p->parent_exec_id = current->parent_exec_id;
	} else {
L
Linus Torvalds 已提交
1348
		p->real_parent = current;
1349 1350
		p->parent_exec_id = current->self_exec_id;
	}
L
Linus Torvalds 已提交
1351

1352
	spin_lock(&current->sighand->siglock);
1353 1354 1355 1356 1357 1358 1359 1360

	/*
	 * 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).
1361
	*/
D
Daniel Walker 已提交
1362
	recalc_sigpending();
1363 1364 1365 1366
	if (signal_pending(current)) {
		spin_unlock(&current->sighand->siglock);
		write_unlock_irq(&tasklist_lock);
		retval = -ERESTARTNOINTR;
1367
		goto bad_fork_free_pid;
1368 1369
	}

L
Linus Torvalds 已提交
1370
	if (clone_flags & CLONE_THREAD) {
1371
		current->signal->nr_threads++;
1372
		atomic_inc(&current->signal->live);
1373
		atomic_inc(&current->signal->sigcnt);
L
Linus Torvalds 已提交
1374
		p->group_leader = current->group_leader;
O
Oleg Nesterov 已提交
1375
		list_add_tail_rcu(&p->thread_group, &p->group_leader->thread_group);
L
Linus Torvalds 已提交
1376 1377
	}

1378
	if (likely(p->pid)) {
T
Tejun Heo 已提交
1379
		ptrace_init_task(p, (clone_flags & CLONE_PTRACE) || trace);
1380 1381

		if (thread_group_leader(p)) {
1382
			if (is_child_reaper(pid))
1383
				p->nsproxy->pid_ns->child_reaper = p;
1384

1385
			p->signal->leader_pid = pid;
A
Alan Cox 已提交
1386
			p->signal->tty = tty_kref_get(current->signal->tty);
E
Eric W. Biederman 已提交
1387 1388
			attach_pid(p, PIDTYPE_PGID, task_pgrp(current));
			attach_pid(p, PIDTYPE_SID, task_session(current));
1389
			list_add_tail(&p->sibling, &p->real_parent->children);
1390
			list_add_tail_rcu(&p->tasks, &init_task.tasks);
1391
			__this_cpu_inc(process_counts);
1392
		}
1393
		attach_pid(p, PIDTYPE_PID, pid);
1394
		nr_threads++;
L
Linus Torvalds 已提交
1395 1396 1397
	}

	total_forks++;
1398
	spin_unlock(&current->sighand->siglock);
L
Linus Torvalds 已提交
1399
	write_unlock_irq(&tasklist_lock);
1400
	proc_fork_connector(p);
1401
	cgroup_post_fork(p);
1402
	if (clone_flags & CLONE_THREAD)
1403
		threadgroup_change_end(current);
1404
	perf_event_fork(p);
1405 1406 1407

	trace_task_newtask(p, clone_flags);

L
Linus Torvalds 已提交
1408 1409
	return p;

1410 1411 1412
bad_fork_free_pid:
	if (pid != &init_struct_pid)
		free_pid(pid);
1413
bad_fork_cleanup_io:
1414 1415
	if (p->io_context)
		exit_io_context(p);
S
Serge E. Hallyn 已提交
1416
bad_fork_cleanup_namespaces:
1417
	exit_task_namespaces(p);
L
Linus Torvalds 已提交
1418
bad_fork_cleanup_mm:
D
David Rientjes 已提交
1419
	if (p->mm)
L
Linus Torvalds 已提交
1420 1421
		mmput(p->mm);
bad_fork_cleanup_signal:
1422
	if (!(clone_flags & CLONE_THREAD))
1423
		free_signal_struct(p->signal);
L
Linus Torvalds 已提交
1424
bad_fork_cleanup_sighand:
1425
	__cleanup_sighand(p->sighand);
L
Linus Torvalds 已提交
1426 1427 1428 1429 1430 1431 1432 1433 1434
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);
bad_fork_cleanup_policy:
1435
	perf_event_free_task(p);
L
Linus Torvalds 已提交
1436
#ifdef CONFIG_NUMA
1437
	mpol_put(p->mempolicy);
1438
bad_fork_cleanup_cgroup:
L
Linus Torvalds 已提交
1439
#endif
1440
	if (clone_flags & CLONE_THREAD)
1441
		threadgroup_change_end(current);
1442
	cgroup_exit(p, cgroup_callbacks_done);
1443
	delayacct_tsk_free(p);
A
Al Viro 已提交
1444
	module_put(task_thread_info(p)->exec_domain->module);
L
Linus Torvalds 已提交
1445
bad_fork_cleanup_count:
D
David Howells 已提交
1446
	atomic_dec(&p->cred->user->processes);
1447
	exit_creds(p);
L
Linus Torvalds 已提交
1448 1449
bad_fork_free:
	free_task(p);
1450 1451
fork_out:
	return ERR_PTR(retval);
L
Linus Torvalds 已提交
1452 1453
}

A
Adrian Bunk 已提交
1454
noinline struct pt_regs * __cpuinit __attribute__((weak)) idle_regs(struct pt_regs *regs)
L
Linus Torvalds 已提交
1455 1456 1457 1458 1459
{
	memset(regs, 0, sizeof(struct pt_regs));
	return regs;
}

1460 1461 1462 1463 1464 1465 1466 1467 1468 1469
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;
	}
}

1470
struct task_struct * __cpuinit fork_idle(int cpu)
L
Linus Torvalds 已提交
1471
{
1472
	struct task_struct *task;
L
Linus Torvalds 已提交
1473 1474
	struct pt_regs regs;

1475
	task = copy_process(CLONE_VM, 0, idle_regs(&regs), 0, NULL,
R
Roland McGrath 已提交
1476
			    &init_struct_pid, 0);
1477 1478
	if (!IS_ERR(task)) {
		init_idle_pids(task->pids);
1479
		init_idle(task, cpu);
1480
	}
1481

L
Linus Torvalds 已提交
1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499
	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.
 */
long do_fork(unsigned long clone_flags,
	      unsigned long stack_start,
	      struct pt_regs *regs,
	      unsigned long stack_size,
	      int __user *parent_tidptr,
	      int __user *child_tidptr)
{
	struct task_struct *p;
	int trace = 0;
1500
	long nr;
L
Linus Torvalds 已提交
1501

1502 1503 1504 1505 1506 1507 1508 1509 1510 1511
	/*
	 * Do some preliminary argument and permissions checking before we
	 * actually start allocating stuff
	 */
	if (clone_flags & CLONE_NEWUSER) {
		if (clone_flags & CLONE_THREAD)
			return -EINVAL;
		/* hopefully this check will go away when userns support is
		 * complete
		 */
1512 1513
		if (!capable(CAP_SYS_ADMIN) || !capable(CAP_SETUID) ||
				!capable(CAP_SETGID))
1514 1515 1516
			return -EPERM;
	}

R
Roland McGrath 已提交
1517
	/*
T
Tejun Heo 已提交
1518 1519 1520 1521
	 * 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
1522
	 */
T
Tejun Heo 已提交
1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533
	if (likely(user_mode(regs)) && !(clone_flags & CLONE_UNTRACED)) {
		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
1534

1535
	p = copy_process(clone_flags, stack_start, regs, stack_size,
R
Roland McGrath 已提交
1536
			 child_tidptr, NULL, trace);
L
Linus Torvalds 已提交
1537 1538 1539 1540 1541 1542 1543
	/*
	 * 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;

1544 1545
		trace_sched_process_fork(current, p);

1546
		nr = task_pid_vnr(p);
1547 1548 1549

		if (clone_flags & CLONE_PARENT_SETTID)
			put_user(nr, parent_tidptr);
1550

L
Linus Torvalds 已提交
1551 1552 1553 1554 1555
		if (clone_flags & CLONE_VFORK) {
			p->vfork_done = &vfork;
			init_completion(&vfork);
		}

R
Roland McGrath 已提交
1556 1557 1558
		/*
		 * We set PF_STARTING at creation in case tracing wants to
		 * use this to distinguish a fully live task from one that
T
Tejun Heo 已提交
1559 1560
		 * hasn't finished SIGSTOP raising yet.  Now we clear it
		 * and set the child going.
R
Roland McGrath 已提交
1561 1562 1563
		 */
		p->flags &= ~PF_STARTING;

1564
		wake_up_new_task(p);
L
Linus Torvalds 已提交
1565

T
Tejun Heo 已提交
1566 1567 1568
		/* forking complete and child started to run, tell ptracer */
		if (unlikely(trace))
			ptrace_event(trace, nr);
R
Roland McGrath 已提交
1569

L
Linus Torvalds 已提交
1570
		if (clone_flags & CLONE_VFORK) {
R
Rafael J. Wysocki 已提交
1571
			freezer_do_not_count();
L
Linus Torvalds 已提交
1572
			wait_for_completion(&vfork);
R
Rafael J. Wysocki 已提交
1573
			freezer_count();
T
Tejun Heo 已提交
1574
			ptrace_event(PTRACE_EVENT_VFORK_DONE, nr);
L
Linus Torvalds 已提交
1575 1576
		}
	} else {
1577
		nr = PTR_ERR(p);
L
Linus Torvalds 已提交
1578
	}
1579
	return nr;
L
Linus Torvalds 已提交
1580 1581
}

1582 1583 1584 1585
#ifndef ARCH_MIN_MMSTRUCT_ALIGN
#define ARCH_MIN_MMSTRUCT_ALIGN 0
#endif

1586
static void sighand_ctor(void *data)
1587 1588 1589
{
	struct sighand_struct *sighand = data;

C
Christoph Lameter 已提交
1590
	spin_lock_init(&sighand->siglock);
D
Davide Libenzi 已提交
1591
	init_waitqueue_head(&sighand->signalfd_wqh);
1592 1593
}

L
Linus Torvalds 已提交
1594 1595 1596 1597
void __init proc_caches_init(void)
{
	sighand_cachep = kmem_cache_create("sighand_cache",
			sizeof(struct sighand_struct), 0,
V
Vegard Nossum 已提交
1598 1599
			SLAB_HWCACHE_ALIGN|SLAB_PANIC|SLAB_DESTROY_BY_RCU|
			SLAB_NOTRACK, sighand_ctor);
L
Linus Torvalds 已提交
1600 1601
	signal_cachep = kmem_cache_create("signal_cache",
			sizeof(struct signal_struct), 0,
V
Vegard Nossum 已提交
1602
			SLAB_HWCACHE_ALIGN|SLAB_PANIC|SLAB_NOTRACK, NULL);
1603
	files_cachep = kmem_cache_create("files_cache",
L
Linus Torvalds 已提交
1604
			sizeof(struct files_struct), 0,
V
Vegard Nossum 已提交
1605
			SLAB_HWCACHE_ALIGN|SLAB_PANIC|SLAB_NOTRACK, NULL);
1606
	fs_cachep = kmem_cache_create("fs_cache",
L
Linus Torvalds 已提交
1607
			sizeof(struct fs_struct), 0,
V
Vegard Nossum 已提交
1608
			SLAB_HWCACHE_ALIGN|SLAB_PANIC|SLAB_NOTRACK, NULL);
1609 1610 1611 1612 1613 1614 1615
	/*
	 * 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
1616
	mm_cachep = kmem_cache_create("mm_struct",
1617
			sizeof(struct mm_struct), ARCH_MIN_MMSTRUCT_ALIGN,
V
Vegard Nossum 已提交
1618
			SLAB_HWCACHE_ALIGN|SLAB_PANIC|SLAB_NOTRACK, NULL);
1619
	vm_area_cachep = KMEM_CACHE(vm_area_struct, SLAB_PANIC);
1620
	mmap_init();
1621
	nsproxy_cache_init();
L
Linus Torvalds 已提交
1622
}
1623 1624

/*
1625
 * Check constraints on flags passed to the unshare system call.
1626
 */
1627
static int check_unshare_flags(unsigned long unshare_flags)
1628
{
1629 1630 1631 1632
	if (unshare_flags & ~(CLONE_THREAD|CLONE_FS|CLONE_NEWNS|CLONE_SIGHAND|
				CLONE_VM|CLONE_FILES|CLONE_SYSVSEM|
				CLONE_NEWUTS|CLONE_NEWIPC|CLONE_NEWNET))
		return -EINVAL;
1633
	/*
1634 1635 1636
	 * Not implemented, but pretend it works if there is nothing to
	 * unshare. Note that unsharing CLONE_THREAD or CLONE_SIGHAND
	 * needs to unshare vm.
1637
	 */
1638 1639 1640 1641 1642
	if (unshare_flags & (CLONE_THREAD | CLONE_SIGHAND | CLONE_VM)) {
		/* FIXME: get_task_mm() increments ->mm_users */
		if (atomic_read(&current->mm->mm_users) > 1)
			return -EINVAL;
	}
1643 1644 1645 1646 1647

	return 0;
}

/*
1648
 * Unshare the filesystem structure if it is being shared
1649 1650 1651 1652 1653
 */
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
1654 1655 1656 1657 1658 1659 1660 1661 1662 1663
	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;
1664 1665 1666 1667 1668

	return 0;
}

/*
1669
 * Unshare file descriptor table if it is being shared
1670 1671 1672 1673
 */
static int unshare_fd(unsigned long unshare_flags, struct files_struct **new_fdp)
{
	struct files_struct *fd = current->files;
1674
	int error = 0;
1675 1676

	if ((unshare_flags & CLONE_FILES) &&
1677 1678 1679 1680 1681
	    (fd && atomic_read(&fd->count) > 1)) {
		*new_fdp = dup_fd(fd, &error);
		if (!*new_fdp)
			return error;
	}
1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693

	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.
 */
1694
SYSCALL_DEFINE1(unshare, unsigned long, unshare_flags)
1695 1696 1697
{
	struct fs_struct *fs, *new_fs = NULL;
	struct files_struct *fd, *new_fd = NULL;
1698
	struct nsproxy *new_nsproxy = NULL;
1699
	int do_sysvsem = 0;
1700
	int err;
1701

1702 1703
	err = check_unshare_flags(unshare_flags);
	if (err)
1704 1705
		goto bad_unshare_out;

1706 1707 1708 1709 1710
	/*
	 * If unsharing namespace, must also unshare filesystem information.
	 */
	if (unshare_flags & CLONE_NEWNS)
		unshare_flags |= CLONE_FS;
1711 1712 1713 1714 1715 1716
	/*
	 * 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))
1717
		do_sysvsem = 1;
1718 1719
	err = unshare_fs(unshare_flags, &new_fs);
	if (err)
1720
		goto bad_unshare_out;
1721 1722
	err = unshare_fd(unshare_flags, &new_fd);
	if (err)
1723
		goto bad_unshare_cleanup_fs;
1724 1725
	err = unshare_nsproxy_namespaces(unshare_flags, &new_nsproxy, new_fs);
	if (err)
1726
		goto bad_unshare_cleanup_fd;
1727

1728
	if (new_fs || new_fd || do_sysvsem || new_nsproxy) {
1729 1730 1731 1732 1733 1734
		if (do_sysvsem) {
			/*
			 * CLONE_SYSVSEM is equivalent to sys_exit().
			 */
			exit_sem(current);
		}
S
Serge E. Hallyn 已提交
1735

1736
		if (new_nsproxy) {
1737 1738
			switch_task_namespaces(current, new_nsproxy);
			new_nsproxy = NULL;
1739
		}
1740

1741 1742
		task_lock(current);

1743 1744
		if (new_fs) {
			fs = current->fs;
N
Nick Piggin 已提交
1745
			spin_lock(&fs->lock);
1746
			current->fs = new_fs;
A
Al Viro 已提交
1747 1748 1749 1750
			if (--fs->users)
				new_fs = NULL;
			else
				new_fs = fs;
N
Nick Piggin 已提交
1751
			spin_unlock(&fs->lock);
1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762
		}

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

		task_unlock(current);
	}

1763
	if (new_nsproxy)
1764
		put_nsproxy(new_nsproxy);
1765

1766 1767 1768 1769 1770 1771
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
1772
		free_fs_struct(new_fs);
1773 1774 1775 1776

bad_unshare_out:
	return err;
}
1777 1778 1779 1780 1781 1782 1783 1784 1785 1786

/*
 *	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;
1787
	struct files_struct *copy = NULL;
1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800
	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;
}