exec.c 51.3 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 24 25 26
/*
 *  linux/fs/exec.c
 *
 *  Copyright (C) 1991, 1992  Linus Torvalds
 */

/*
 * #!-checking implemented by tytso.
 */
/*
 * Demand-loading implemented 01.12.91 - no need to read anything but
 * the header into memory. The inode of the executable is put into
 * "current->executable", and page faults do the actual loading. Clean.
 *
 * Once more I can proudly say that linux stood up to being changed: it
 * was less than 2 hours work to get demand-loading completely implemented.
 *
 * Demand loading changed July 1993 by Eric Youngdale.   Use mmap instead,
 * current->executable is only used by the procfs.  This allows a dispatch
 * table to check for several different types  of binary formats.  We keep
 * trying until we recognize the file or we run out of supported binary
 * formats. 
 */

#include <linux/slab.h>
#include <linux/file.h>
A
Al Viro 已提交
27
#include <linux/fdtable.h>
H
Hugh Dickins 已提交
28
#include <linux/mm.h>
L
Linus Torvalds 已提交
29 30
#include <linux/stat.h>
#include <linux/fcntl.h>
H
Hugh Dickins 已提交
31
#include <linux/swap.h>
32
#include <linux/string.h>
L
Linus Torvalds 已提交
33
#include <linux/init.h>
34
#include <linux/pagemap.h>
35
#include <linux/perf_event.h>
L
Linus Torvalds 已提交
36 37 38 39 40 41
#include <linux/highmem.h>
#include <linux/spinlock.h>
#include <linux/key.h>
#include <linux/personality.h>
#include <linux/binfmts.h>
#include <linux/utsname.h>
42
#include <linux/pid_namespace.h>
L
Linus Torvalds 已提交
43 44 45 46 47
#include <linux/module.h>
#include <linux/namei.h>
#include <linux/mount.h>
#include <linux/security.h>
#include <linux/syscalls.h>
48
#include <linux/tsacct_kern.h>
M
Matt Helsley 已提交
49
#include <linux/cn_proc.h>
A
Al Viro 已提交
50
#include <linux/audit.h>
R
Roland McGrath 已提交
51
#include <linux/tracehook.h>
J
Johannes Berg 已提交
52
#include <linux/kmod.h>
53
#include <linux/fsnotify.h>
54
#include <linux/fs_struct.h>
55
#include <linux/pipe_fs_i.h>
Y
Ying Han 已提交
56
#include <linux/oom.h>
57
#include <linux/compat.h>
L
Linus Torvalds 已提交
58 59 60

#include <asm/uaccess.h>
#include <asm/mmu_context.h>
61
#include <asm/tlb.h>
62
#include "internal.h"
L
Linus Torvalds 已提交
63 64

int core_uses_pid;
65
char core_pattern[CORENAME_MAX_SIZE] = "core";
66
unsigned int core_pipe_limit;
A
Alan Cox 已提交
67 68
int suid_dumpable = 0;

69 70 71 72 73 74
struct core_name {
	char *corename;
	int used, size;
};
static atomic_t call_count = ATOMIC_INIT(1);

L
Linus Torvalds 已提交
75 76
/* The maximal length of core_pattern is also specified in sysctl.c */

A
Alexey Dobriyan 已提交
77
static LIST_HEAD(formats);
L
Linus Torvalds 已提交
78 79
static DEFINE_RWLOCK(binfmt_lock);

I
Ivan Kokshaysky 已提交
80
int __register_binfmt(struct linux_binfmt * fmt, int insert)
L
Linus Torvalds 已提交
81 82 83 84
{
	if (!fmt)
		return -EINVAL;
	write_lock(&binfmt_lock);
I
Ivan Kokshaysky 已提交
85 86
	insert ? list_add(&fmt->lh, &formats) :
		 list_add_tail(&fmt->lh, &formats);
L
Linus Torvalds 已提交
87 88 89 90
	write_unlock(&binfmt_lock);
	return 0;	
}

I
Ivan Kokshaysky 已提交
91
EXPORT_SYMBOL(__register_binfmt);
L
Linus Torvalds 已提交
92

93
void unregister_binfmt(struct linux_binfmt * fmt)
L
Linus Torvalds 已提交
94 95
{
	write_lock(&binfmt_lock);
A
Alexey Dobriyan 已提交
96
	list_del(&fmt->lh);
L
Linus Torvalds 已提交
97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112
	write_unlock(&binfmt_lock);
}

EXPORT_SYMBOL(unregister_binfmt);

static inline void put_binfmt(struct linux_binfmt * fmt)
{
	module_put(fmt->module);
}

/*
 * Note that a shared library must be both readable and executable due to
 * security reasons.
 *
 * Also note that we take the address to load from from the file itself.
 */
113
SYSCALL_DEFINE1(uselib, const char __user *, library)
L
Linus Torvalds 已提交
114
{
115 116 117
	struct file *file;
	char *tmp = getname(library);
	int error = PTR_ERR(tmp);
118 119 120 121 122
	static const struct open_flags uselib_flags = {
		.open_flag = O_LARGEFILE | O_RDONLY | __FMODE_EXEC,
		.acc_mode = MAY_READ | MAY_EXEC | MAY_OPEN,
		.intent = LOOKUP_OPEN
	};
123

124 125 126
	if (IS_ERR(tmp))
		goto out;

127
	file = do_filp_open(AT_FDCWD, tmp, &uselib_flags, LOOKUP_FOLLOW);
128 129 130
	putname(tmp);
	error = PTR_ERR(file);
	if (IS_ERR(file))
L
Linus Torvalds 已提交
131 132 133
		goto out;

	error = -EINVAL;
134
	if (!S_ISREG(file->f_path.dentry->d_inode->i_mode))
L
Linus Torvalds 已提交
135 136
		goto exit;

137
	error = -EACCES;
138
	if (file->f_path.mnt->mnt_flags & MNT_NOEXEC)
L
Linus Torvalds 已提交
139 140
		goto exit;

141
	fsnotify_open(file);
142

L
Linus Torvalds 已提交
143 144 145 146 147
	error = -ENOEXEC;
	if(file->f_op) {
		struct linux_binfmt * fmt;

		read_lock(&binfmt_lock);
A
Alexey Dobriyan 已提交
148
		list_for_each_entry(fmt, &formats, lh) {
L
Linus Torvalds 已提交
149 150 151 152 153 154 155 156 157 158 159 160 161
			if (!fmt->load_shlib)
				continue;
			if (!try_module_get(fmt->module))
				continue;
			read_unlock(&binfmt_lock);
			error = fmt->load_shlib(file);
			read_lock(&binfmt_lock);
			put_binfmt(fmt);
			if (error != -ENOEXEC)
				break;
		}
		read_unlock(&binfmt_lock);
	}
162
exit:
L
Linus Torvalds 已提交
163 164 165 166 167
	fput(file);
out:
  	return error;
}

168
#ifdef CONFIG_MMU
O
Oleg Nesterov 已提交
169 170 171 172 173 174
/*
 * The nascent bprm->mm is not visible until exec_mmap() but it can
 * use a lot of memory, account these pages in current->mm temporary
 * for oom_badness()->get_mm_rss(). Once exec succeeds or fails, we
 * change the counter back via acct_arg_size(0).
 */
175
static void acct_arg_size(struct linux_binprm *bprm, unsigned long pages)
176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193
{
	struct mm_struct *mm = current->mm;
	long diff = (long)(pages - bprm->vma_pages);

	if (!mm || !diff)
		return;

	bprm->vma_pages = pages;

#ifdef SPLIT_RSS_COUNTING
	add_mm_counter(mm, MM_ANONPAGES, diff);
#else
	spin_lock(&mm->page_table_lock);
	add_mm_counter(mm, MM_ANONPAGES, diff);
	spin_unlock(&mm->page_table_lock);
#endif
}

194
static struct page *get_arg_page(struct linux_binprm *bprm, unsigned long pos,
195 196 197 198 199 200 201
		int write)
{
	struct page *page;
	int ret;

#ifdef CONFIG_STACK_GROWSUP
	if (write) {
202
		ret = expand_downwards(bprm->vma, pos);
203 204 205 206 207 208 209 210 211 212 213
		if (ret < 0)
			return NULL;
	}
#endif
	ret = get_user_pages(current, bprm->mm, pos,
			1, write, 1, &page, NULL);
	if (ret <= 0)
		return NULL;

	if (write) {
		unsigned long size = bprm->vma->vm_end - bprm->vma->vm_start;
214 215
		struct rlimit *rlim;

216 217
		acct_arg_size(bprm, size / PAGE_SIZE);

218 219 220 221 222 223
		/*
		 * We've historically supported up to 32 pages (ARG_MAX)
		 * of argument strings even with small stacks
		 */
		if (size <= ARG_MAX)
			return page;
224 225 226 227 228 229 230 231

		/*
		 * Limit to 1/4-th the stack size for the argv+env strings.
		 * This ensures that:
		 *  - the remaining binfmt code will not run out of stack space,
		 *  - the program will have a reasonable amount of stack left
		 *    to work from.
		 */
232
		rlim = current->signal->rlim;
J
Jiri Slaby 已提交
233
		if (size > ACCESS_ONCE(rlim[RLIMIT_STACK].rlim_cur) / 4) {
234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262
			put_page(page);
			return NULL;
		}
	}

	return page;
}

static void put_arg_page(struct page *page)
{
	put_page(page);
}

static void free_arg_page(struct linux_binprm *bprm, int i)
{
}

static void free_arg_pages(struct linux_binprm *bprm)
{
}

static void flush_arg_page(struct linux_binprm *bprm, unsigned long pos,
		struct page *page)
{
	flush_cache_page(bprm->vma, pos, page_to_pfn(page));
}

static int __bprm_mm_init(struct linux_binprm *bprm)
{
263
	int err;
264 265 266 267 268
	struct vm_area_struct *vma = NULL;
	struct mm_struct *mm = bprm->mm;

	bprm->vma = vma = kmem_cache_zalloc(vm_area_cachep, GFP_KERNEL);
	if (!vma)
269
		return -ENOMEM;
270 271 272 273 274 275 276 277 278 279

	down_write(&mm->mmap_sem);
	vma->vm_mm = mm;

	/*
	 * Place the stack at the largest stack address the architecture
	 * supports. Later, we'll move this to an appropriate place. We don't
	 * use STACK_TOP because that can depend on attributes which aren't
	 * configured yet.
	 */
280
	BUG_ON(VM_STACK_FLAGS & VM_STACK_INCOMPLETE_SETUP);
281 282
	vma->vm_end = STACK_TOP_MAX;
	vma->vm_start = vma->vm_end - PAGE_SIZE;
283
	vma->vm_flags = VM_STACK_FLAGS | VM_STACK_INCOMPLETE_SETUP;
284
	vma->vm_page_prot = vm_get_page_prot(vma->vm_flags);
285
	INIT_LIST_HEAD(&vma->anon_vma_chain);
286 287 288 289 290

	err = security_file_mmap(NULL, 0, 0, 0, vma->vm_start, 1);
	if (err)
		goto err;

291
	err = insert_vm_struct(mm, vma);
292
	if (err)
293 294 295 296 297 298 299
		goto err;

	mm->stack_vm = mm->total_vm = 1;
	up_write(&mm->mmap_sem);
	bprm->p = vma->vm_end - sizeof(void *);
	return 0;
err:
300 301 302
	up_write(&mm->mmap_sem);
	bprm->vma = NULL;
	kmem_cache_free(vm_area_cachep, vma);
303 304 305 306 307 308 309 310 311 312
	return err;
}

static bool valid_arg_len(struct linux_binprm *bprm, long len)
{
	return len <= MAX_ARG_STRLEN;
}

#else

313
static inline void acct_arg_size(struct linux_binprm *bprm, unsigned long pages)
314 315 316
{
}

317
static struct page *get_arg_page(struct linux_binprm *bprm, unsigned long pos,
318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405
		int write)
{
	struct page *page;

	page = bprm->page[pos / PAGE_SIZE];
	if (!page && write) {
		page = alloc_page(GFP_HIGHUSER|__GFP_ZERO);
		if (!page)
			return NULL;
		bprm->page[pos / PAGE_SIZE] = page;
	}

	return page;
}

static void put_arg_page(struct page *page)
{
}

static void free_arg_page(struct linux_binprm *bprm, int i)
{
	if (bprm->page[i]) {
		__free_page(bprm->page[i]);
		bprm->page[i] = NULL;
	}
}

static void free_arg_pages(struct linux_binprm *bprm)
{
	int i;

	for (i = 0; i < MAX_ARG_PAGES; i++)
		free_arg_page(bprm, i);
}

static void flush_arg_page(struct linux_binprm *bprm, unsigned long pos,
		struct page *page)
{
}

static int __bprm_mm_init(struct linux_binprm *bprm)
{
	bprm->p = PAGE_SIZE * MAX_ARG_PAGES - sizeof(void *);
	return 0;
}

static bool valid_arg_len(struct linux_binprm *bprm, long len)
{
	return len <= bprm->p;
}

#endif /* CONFIG_MMU */

/*
 * Create a new mm_struct and populate it with a temporary stack
 * vm_area_struct.  We don't have enough context at this point to set the stack
 * flags, permissions, and offset, so we use temporary values.  We'll update
 * them later in setup_arg_pages().
 */
int bprm_mm_init(struct linux_binprm *bprm)
{
	int err;
	struct mm_struct *mm = NULL;

	bprm->mm = mm = mm_alloc();
	err = -ENOMEM;
	if (!mm)
		goto err;

	err = init_new_context(current, mm);
	if (err)
		goto err;

	err = __bprm_mm_init(bprm);
	if (err)
		goto err;

	return 0;

err:
	if (mm) {
		bprm->mm = NULL;
		mmdrop(mm);
	}

	return err;
}

406
struct user_arg_ptr {
407 408 409 410 411 412 413 414 415
#ifdef CONFIG_COMPAT
	bool is_compat;
#endif
	union {
		const char __user *const __user *native;
#ifdef CONFIG_COMPAT
		compat_uptr_t __user *compat;
#endif
	} ptr;
416 417 418
};

static const char __user *get_user_arg_ptr(struct user_arg_ptr argv, int nr)
419
{
420 421 422 423 424 425 426 427
	const char __user *native;

#ifdef CONFIG_COMPAT
	if (unlikely(argv.is_compat)) {
		compat_uptr_t compat;

		if (get_user(compat, argv.ptr.compat + nr))
			return ERR_PTR(-EFAULT);
428

429 430 431 432 433
		return compat_ptr(compat);
	}
#endif

	if (get_user(native, argv.ptr.native + nr))
434 435
		return ERR_PTR(-EFAULT);

436
	return native;
437 438
}

L
Linus Torvalds 已提交
439 440 441
/*
 * count() counts the number of strings in array ARGV.
 */
442
static int count(struct user_arg_ptr argv, int max)
L
Linus Torvalds 已提交
443 444 445
{
	int i = 0;

446
	if (argv.ptr.native != NULL) {
L
Linus Torvalds 已提交
447
		for (;;) {
448
			const char __user *p = get_user_arg_ptr(argv, i);
L
Linus Torvalds 已提交
449 450 451

			if (!p)
				break;
452 453 454 455

			if (IS_ERR(p))
				return -EFAULT;

456
			if (i++ >= max)
L
Linus Torvalds 已提交
457
				return -E2BIG;
458 459 460

			if (fatal_signal_pending(current))
				return -ERESTARTNOHAND;
L
Linus Torvalds 已提交
461 462 463 464 465 466 467
			cond_resched();
		}
	}
	return i;
}

/*
468 469 470
 * 'copy_strings()' copies argument/environment strings from the old
 * processes's memory to the new process's stack.  The call to get_user_pages()
 * ensures the destination page is created and not swapped out.
L
Linus Torvalds 已提交
471
 */
472
static int copy_strings(int argc, struct user_arg_ptr argv,
A
Adrian Bunk 已提交
473
			struct linux_binprm *bprm)
L
Linus Torvalds 已提交
474 475 476
{
	struct page *kmapped_page = NULL;
	char *kaddr = NULL;
477
	unsigned long kpos = 0;
L
Linus Torvalds 已提交
478 479 480
	int ret;

	while (argc-- > 0) {
481
		const char __user *str;
L
Linus Torvalds 已提交
482 483 484
		int len;
		unsigned long pos;

485 486 487
		ret = -EFAULT;
		str = get_user_arg_ptr(argv, argc);
		if (IS_ERR(str))
L
Linus Torvalds 已提交
488 489
			goto out;

490 491 492 493 494 495
		len = strnlen_user(str, MAX_ARG_STRLEN);
		if (!len)
			goto out;

		ret = -E2BIG;
		if (!valid_arg_len(bprm, len))
L
Linus Torvalds 已提交
496 497
			goto out;

498
		/* We're going to work our way backwords. */
L
Linus Torvalds 已提交
499
		pos = bprm->p;
500 501
		str += len;
		bprm->p -= len;
L
Linus Torvalds 已提交
502 503 504 505

		while (len > 0) {
			int offset, bytes_to_copy;

506 507 508 509
			if (fatal_signal_pending(current)) {
				ret = -ERESTARTNOHAND;
				goto out;
			}
510 511
			cond_resched();

L
Linus Torvalds 已提交
512
			offset = pos % PAGE_SIZE;
513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528
			if (offset == 0)
				offset = PAGE_SIZE;

			bytes_to_copy = offset;
			if (bytes_to_copy > len)
				bytes_to_copy = len;

			offset -= bytes_to_copy;
			pos -= bytes_to_copy;
			str -= bytes_to_copy;
			len -= bytes_to_copy;

			if (!kmapped_page || kpos != (pos & PAGE_MASK)) {
				struct page *page;

				page = get_arg_page(bprm, pos, 1);
L
Linus Torvalds 已提交
529
				if (!page) {
530
					ret = -E2BIG;
L
Linus Torvalds 已提交
531 532 533
					goto out;
				}

534 535
				if (kmapped_page) {
					flush_kernel_dcache_page(kmapped_page);
L
Linus Torvalds 已提交
536
					kunmap(kmapped_page);
537 538
					put_arg_page(kmapped_page);
				}
L
Linus Torvalds 已提交
539 540
				kmapped_page = page;
				kaddr = kmap(kmapped_page);
541 542
				kpos = pos & PAGE_MASK;
				flush_arg_page(bprm, kpos, kmapped_page);
L
Linus Torvalds 已提交
543
			}
544
			if (copy_from_user(kaddr+offset, str, bytes_to_copy)) {
L
Linus Torvalds 已提交
545 546 547 548 549 550 551
				ret = -EFAULT;
				goto out;
			}
		}
	}
	ret = 0;
out:
552 553
	if (kmapped_page) {
		flush_kernel_dcache_page(kmapped_page);
L
Linus Torvalds 已提交
554
		kunmap(kmapped_page);
555 556
		put_arg_page(kmapped_page);
	}
L
Linus Torvalds 已提交
557 558 559 560 561 562
	return ret;
}

/*
 * Like copy_strings, but get argv and its values from kernel memory.
 */
563
int copy_strings_kernel(int argc, const char *const *__argv,
564
			struct linux_binprm *bprm)
L
Linus Torvalds 已提交
565 566 567
{
	int r;
	mm_segment_t oldfs = get_fs();
568
	struct user_arg_ptr argv = {
569
		.ptr.native = (const char __user *const  __user *)__argv,
570 571
	};

L
Linus Torvalds 已提交
572
	set_fs(KERNEL_DS);
573
	r = copy_strings(argc, argv, bprm);
L
Linus Torvalds 已提交
574
	set_fs(oldfs);
575

L
Linus Torvalds 已提交
576 577 578 579 580
	return r;
}
EXPORT_SYMBOL(copy_strings_kernel);

#ifdef CONFIG_MMU
581

L
Linus Torvalds 已提交
582
/*
583 584 585
 * During bprm_mm_init(), we create a temporary stack at STACK_TOP_MAX.  Once
 * the binfmt code determines where the new stack should reside, we shift it to
 * its final location.  The process proceeds as follows:
L
Linus Torvalds 已提交
586
 *
587 588 589 590 591 592
 * 1) Use shift to calculate the new vma endpoints.
 * 2) Extend vma to cover both the old and new ranges.  This ensures the
 *    arguments passed to subsequent functions are consistent.
 * 3) Move vma's page tables to the new range.
 * 4) Free up any cleared pgd range.
 * 5) Shrink the vma to cover only the new range.
L
Linus Torvalds 已提交
593
 */
594
static int shift_arg_pages(struct vm_area_struct *vma, unsigned long shift)
L
Linus Torvalds 已提交
595 596
{
	struct mm_struct *mm = vma->vm_mm;
597 598 599 600 601
	unsigned long old_start = vma->vm_start;
	unsigned long old_end = vma->vm_end;
	unsigned long length = old_end - old_start;
	unsigned long new_start = old_start - shift;
	unsigned long new_end = old_end - shift;
P
Peter Zijlstra 已提交
602
	struct mmu_gather tlb;
L
Linus Torvalds 已提交
603

604
	BUG_ON(new_start > new_end);
L
Linus Torvalds 已提交
605

606 607 608 609 610 611 612 613 614 615
	/*
	 * ensure there are no vmas between where we want to go
	 * and where we are
	 */
	if (vma != find_vma(mm, new_start))
		return -EFAULT;

	/*
	 * cover the whole range: [new_start, old_end)
	 */
616 617
	if (vma_adjust(vma, new_start, old_end, vma->vm_pgoff, NULL))
		return -ENOMEM;
618 619 620 621 622 623 624 625 626 627

	/*
	 * move the page tables downwards, on failure we rely on
	 * process cleanup to remove whatever mess we made.
	 */
	if (length != move_page_tables(vma, old_start,
				       vma, new_start, length))
		return -ENOMEM;

	lru_add_drain();
P
Peter Zijlstra 已提交
628
	tlb_gather_mmu(&tlb, mm, 0);
629 630 631 632
	if (new_end > old_start) {
		/*
		 * when the old and new regions overlap clear from new_end.
		 */
P
Peter Zijlstra 已提交
633
		free_pgd_range(&tlb, new_end, old_end, new_end,
634 635 636 637 638 639 640 641
			vma->vm_next ? vma->vm_next->vm_start : 0);
	} else {
		/*
		 * otherwise, clean from old_start; this is done to not touch
		 * the address space in [new_end, old_start) some architectures
		 * have constraints on va-space that make this illegal (IA64) -
		 * for the others its just a little faster.
		 */
P
Peter Zijlstra 已提交
642
		free_pgd_range(&tlb, old_start, old_end, new_end,
643
			vma->vm_next ? vma->vm_next->vm_start : 0);
L
Linus Torvalds 已提交
644
	}
P
Peter Zijlstra 已提交
645
	tlb_finish_mmu(&tlb, new_end, old_end);
646 647

	/*
648
	 * Shrink the vma to just the new range.  Always succeeds.
649 650 651 652
	 */
	vma_adjust(vma, new_start, new_end, vma->vm_pgoff, NULL);

	return 0;
L
Linus Torvalds 已提交
653 654
}

655 656 657 658
/*
 * Finalizes the stack vm_area_struct. The flags and permissions are updated,
 * the stack is optionally relocated, and some extra space is added.
 */
L
Linus Torvalds 已提交
659 660 661 662
int setup_arg_pages(struct linux_binprm *bprm,
		    unsigned long stack_top,
		    int executable_stack)
{
663 664
	unsigned long ret;
	unsigned long stack_shift;
L
Linus Torvalds 已提交
665
	struct mm_struct *mm = current->mm;
666 667 668 669
	struct vm_area_struct *vma = bprm->vma;
	struct vm_area_struct *prev = NULL;
	unsigned long vm_flags;
	unsigned long stack_base;
670 671 672
	unsigned long stack_size;
	unsigned long stack_expand;
	unsigned long rlim_stack;
L
Linus Torvalds 已提交
673 674 675

#ifdef CONFIG_STACK_GROWSUP
	/* Limit stack size to 1GB */
J
Jiri Slaby 已提交
676
	stack_base = rlimit_max(RLIMIT_STACK);
L
Linus Torvalds 已提交
677 678 679
	if (stack_base > (1 << 30))
		stack_base = 1 << 30;

680 681 682
	/* Make sure we didn't let the argument array grow too large. */
	if (vma->vm_end - vma->vm_start > stack_base)
		return -ENOMEM;
L
Linus Torvalds 已提交
683

684
	stack_base = PAGE_ALIGN(stack_top - stack_base);
L
Linus Torvalds 已提交
685

686 687 688
	stack_shift = vma->vm_start - stack_base;
	mm->arg_start = bprm->p - stack_shift;
	bprm->p = vma->vm_end - stack_shift;
L
Linus Torvalds 已提交
689
#else
690 691
	stack_top = arch_align_stack(stack_top);
	stack_top = PAGE_ALIGN(stack_top);
692 693 694 695 696

	if (unlikely(stack_top < mmap_min_addr) ||
	    unlikely(vma->vm_end - vma->vm_start >= stack_top - mmap_min_addr))
		return -ENOMEM;

697 698 699
	stack_shift = vma->vm_end - stack_top;

	bprm->p -= stack_shift;
L
Linus Torvalds 已提交
700 701 702 703
	mm->arg_start = bprm->p;
#endif

	if (bprm->loader)
704 705
		bprm->loader -= stack_shift;
	bprm->exec -= stack_shift;
L
Linus Torvalds 已提交
706 707

	down_write(&mm->mmap_sem);
708
	vm_flags = VM_STACK_FLAGS;
709 710 711 712 713 714 715 716 717 718 719

	/*
	 * Adjust stack execute permissions; explicitly enable for
	 * EXSTACK_ENABLE_X, disable for EXSTACK_DISABLE_X and leave alone
	 * (arch default) otherwise.
	 */
	if (unlikely(executable_stack == EXSTACK_ENABLE_X))
		vm_flags |= VM_EXEC;
	else if (executable_stack == EXSTACK_DISABLE_X)
		vm_flags &= ~VM_EXEC;
	vm_flags |= mm->def_flags;
720
	vm_flags |= VM_STACK_INCOMPLETE_SETUP;
721 722 723 724 725 726 727 728 729 730

	ret = mprotect_fixup(vma, &prev, vma->vm_start, vma->vm_end,
			vm_flags);
	if (ret)
		goto out_unlock;
	BUG_ON(prev != vma);

	/* Move stack pages down in memory. */
	if (stack_shift) {
		ret = shift_arg_pages(vma, stack_shift);
731 732
		if (ret)
			goto out_unlock;
L
Linus Torvalds 已提交
733 734
	}

735 736 737
	/* mprotect_fixup is overkill to remove the temporary stack flags */
	vma->vm_flags &= ~VM_STACK_INCOMPLETE_SETUP;

738
	stack_expand = 131072UL; /* randomly 32*4k (or 2*64k) pages */
739 740 741 742 743 744
	stack_size = vma->vm_end - vma->vm_start;
	/*
	 * Align this down to a page boundary as expand_stack
	 * will align it up.
	 */
	rlim_stack = rlimit(RLIMIT_STACK) & PAGE_MASK;
745
#ifdef CONFIG_STACK_GROWSUP
746 747 748 749
	if (stack_size + stack_expand > rlim_stack)
		stack_base = vma->vm_start + rlim_stack;
	else
		stack_base = vma->vm_end + stack_expand;
750
#else
751 752 753 754
	if (stack_size + stack_expand > rlim_stack)
		stack_base = vma->vm_end - rlim_stack;
	else
		stack_base = vma->vm_start - stack_expand;
755
#endif
756
	current->mm->start_stack = bprm->p;
757 758 759 760 761
	ret = expand_stack(vma, stack_base);
	if (ret)
		ret = -EFAULT;

out_unlock:
L
Linus Torvalds 已提交
762
	up_write(&mm->mmap_sem);
763
	return ret;
L
Linus Torvalds 已提交
764 765 766 767 768 769 770 771
}
EXPORT_SYMBOL(setup_arg_pages);

#endif /* CONFIG_MMU */

struct file *open_exec(const char *name)
{
	struct file *file;
772
	int err;
773 774 775 776 777
	static const struct open_flags open_exec_flags = {
		.open_flag = O_LARGEFILE | O_RDONLY | __FMODE_EXEC,
		.acc_mode = MAY_EXEC | MAY_OPEN,
		.intent = LOOKUP_OPEN
	};
L
Linus Torvalds 已提交
778

779
	file = do_filp_open(AT_FDCWD, name, &open_exec_flags, LOOKUP_FOLLOW);
780
	if (IS_ERR(file))
781 782 783
		goto out;

	err = -EACCES;
784 785
	if (!S_ISREG(file->f_path.dentry->d_inode->i_mode))
		goto exit;
786

787 788
	if (file->f_path.mnt->mnt_flags & MNT_NOEXEC)
		goto exit;
789

790
	fsnotify_open(file);
791

792
	err = deny_write_access(file);
793 794
	if (err)
		goto exit;
L
Linus Torvalds 已提交
795

796
out:
797 798
	return file;

799 800
exit:
	fput(file);
801 802
	return ERR_PTR(err);
}
L
Linus Torvalds 已提交
803 804
EXPORT_SYMBOL(open_exec);

M
Mimi Zohar 已提交
805 806
int kernel_read(struct file *file, loff_t offset,
		char *addr, unsigned long count)
L
Linus Torvalds 已提交
807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829
{
	mm_segment_t old_fs;
	loff_t pos = offset;
	int result;

	old_fs = get_fs();
	set_fs(get_ds());
	/* The cast to a user pointer is valid due to the set_fs() */
	result = vfs_read(file, (void __user *)addr, count, &pos);
	set_fs(old_fs);
	return result;
}

EXPORT_SYMBOL(kernel_read);

static int exec_mmap(struct mm_struct *mm)
{
	struct task_struct *tsk;
	struct mm_struct * old_mm, *active_mm;

	/* Notify parent that we're no longer interested in the old VM */
	tsk = current;
	old_mm = current->mm;
830
	sync_mm_rss(tsk, old_mm);
L
Linus Torvalds 已提交
831 832 833 834 835 836 837
	mm_release(tsk, old_mm);

	if (old_mm) {
		/*
		 * Make sure that if there is a core dump in progress
		 * for the old mm, we get out and die instead of going
		 * through with the exec.  We must hold mmap_sem around
838
		 * checking core_state and changing tsk->mm.
L
Linus Torvalds 已提交
839 840
		 */
		down_read(&old_mm->mmap_sem);
841
		if (unlikely(old_mm->core_state)) {
L
Linus Torvalds 已提交
842 843 844 845 846 847 848 849 850
			up_read(&old_mm->mmap_sem);
			return -EINTR;
		}
	}
	task_lock(tsk);
	active_mm = tsk->active_mm;
	tsk->mm = mm;
	tsk->active_mm = mm;
	activate_mm(active_mm, mm);
Y
Ying Han 已提交
851 852 853 854
	if (old_mm && tsk->signal->oom_score_adj == OOM_SCORE_ADJ_MIN) {
		atomic_dec(&old_mm->oom_disable_count);
		atomic_inc(&tsk->mm->oom_disable_count);
	}
L
Linus Torvalds 已提交
855 856 857 858
	task_unlock(tsk);
	arch_pick_mmap_layout(mm);
	if (old_mm) {
		up_read(&old_mm->mmap_sem);
859
		BUG_ON(active_mm != old_mm);
860
		mm_update_next_owner(old_mm);
L
Linus Torvalds 已提交
861 862 863 864 865 866 867 868 869 870 871 872 873
		mmput(old_mm);
		return 0;
	}
	mmdrop(active_mm);
	return 0;
}

/*
 * This function makes sure the current process has its own signal table,
 * so that flush_signal_handlers can later reset the handlers without
 * disturbing other processes.  (Other processes might share the signal
 * table via the CLONE_SIGHAND option to clone().)
 */
874
static int de_thread(struct task_struct *tsk)
L
Linus Torvalds 已提交
875 876
{
	struct signal_struct *sig = tsk->signal;
877
	struct sighand_struct *oldsighand = tsk->sighand;
L
Linus Torvalds 已提交
878 879
	spinlock_t *lock = &oldsighand->siglock;

880
	if (thread_group_empty(tsk))
L
Linus Torvalds 已提交
881 882 883 884 885 886
		goto no_thread_group;

	/*
	 * Kill all other threads in the thread group.
	 */
	spin_lock_irq(lock);
887
	if (signal_group_exit(sig)) {
L
Linus Torvalds 已提交
888 889 890 891 892 893 894
		/*
		 * Another group action in progress, just
		 * return so that the signal is processed.
		 */
		spin_unlock_irq(lock);
		return -EAGAIN;
	}
895

896
	sig->group_exit_task = tsk;
897 898 899
	sig->notify_count = zap_other_threads(tsk);
	if (!thread_group_leader(tsk))
		sig->notify_count--;
L
Linus Torvalds 已提交
900

901
	while (sig->notify_count) {
L
Linus Torvalds 已提交
902 903 904 905 906 907 908 909 910 911 912 913
		__set_current_state(TASK_UNINTERRUPTIBLE);
		spin_unlock_irq(lock);
		schedule();
		spin_lock_irq(lock);
	}
	spin_unlock_irq(lock);

	/*
	 * At this point all other threads have exited, all we have to
	 * do is to wait for the thread group leader to become inactive,
	 * and to assume its PID:
	 */
914
	if (!thread_group_leader(tsk)) {
915
		struct task_struct *leader = tsk->group_leader;
916

917
		sig->notify_count = -1;	/* for exit_notify() */
918 919 920 921 922 923 924 925
		for (;;) {
			write_lock_irq(&tasklist_lock);
			if (likely(leader->exit_state))
				break;
			__set_current_state(TASK_UNINTERRUPTIBLE);
			write_unlock_irq(&tasklist_lock);
			schedule();
		}
L
Linus Torvalds 已提交
926

927 928 929 930 931 932 933 934 935 936
		/*
		 * The only record we have of the real-time age of a
		 * process, regardless of execs it's done, is start_time.
		 * All the past CPU time is accumulated in signal_struct
		 * from sister threads now dead.  But in this non-leader
		 * exec, nothing survives from the original leader thread,
		 * whose birth marks the true age of this process now.
		 * When we take on its identity by switching to its PID, we
		 * also take its birthdate (always earlier than our own).
		 */
937
		tsk->start_time = leader->start_time;
938

939 940
		BUG_ON(!same_thread_group(leader, tsk));
		BUG_ON(has_group_leader_pid(tsk));
L
Linus Torvalds 已提交
941 942 943 944 945 946
		/*
		 * An exec() starts a new thread group with the
		 * TGID of the previous thread group. Rehash the
		 * two threads with a switched PID, and release
		 * the former thread group leader:
		 */
947 948

		/* Become a process group leader with the old leader's pid.
949 950
		 * The old leader becomes a thread of the this thread group.
		 * Note: The old leader also uses this pid until release_task
951 952
		 *       is called.  Odd but simple and correct.
		 */
953 954
		detach_pid(tsk, PIDTYPE_PID);
		tsk->pid = leader->pid;
955
		attach_pid(tsk, PIDTYPE_PID,  task_pid(leader));
956 957
		transfer_pid(leader, tsk, PIDTYPE_PGID);
		transfer_pid(leader, tsk, PIDTYPE_SID);
958

959
		list_replace_rcu(&leader->tasks, &tsk->tasks);
960
		list_replace_init(&leader->sibling, &tsk->sibling);
L
Linus Torvalds 已提交
961

962 963
		tsk->group_leader = tsk;
		leader->group_leader = tsk;
964

965
		tsk->exit_signal = SIGCHLD;
966
		leader->exit_signal = -1;
967 968 969

		BUG_ON(leader->exit_state != EXIT_ZOMBIE);
		leader->exit_state = EXIT_DEAD;
L
Linus Torvalds 已提交
970
		write_unlock_irq(&tasklist_lock);
971 972

		release_task(leader);
973
	}
L
Linus Torvalds 已提交
974

975 976
	sig->group_exit_task = NULL;
	sig->notify_count = 0;
L
Linus Torvalds 已提交
977 978

no_thread_group:
J
Jiri Pirko 已提交
979 980 981
	if (current->mm)
		setmax_mm_hiwater_rss(&sig->maxrss, current->mm);

L
Linus Torvalds 已提交
982
	exit_itimers(sig);
983
	flush_itimer_signals();
984

985 986
	if (atomic_read(&oldsighand->count) != 1) {
		struct sighand_struct *newsighand;
L
Linus Torvalds 已提交
987
		/*
988 989
		 * This ->sighand is shared with the CLONE_SIGHAND
		 * but not CLONE_THREAD task, switch to the new one.
L
Linus Torvalds 已提交
990
		 */
991 992 993 994
		newsighand = kmem_cache_alloc(sighand_cachep, GFP_KERNEL);
		if (!newsighand)
			return -ENOMEM;

L
Linus Torvalds 已提交
995 996 997 998 999 1000
		atomic_set(&newsighand->count, 1);
		memcpy(newsighand->action, oldsighand->action,
		       sizeof(newsighand->action));

		write_lock_irq(&tasklist_lock);
		spin_lock(&oldsighand->siglock);
1001
		rcu_assign_pointer(tsk->sighand, newsighand);
L
Linus Torvalds 已提交
1002 1003 1004
		spin_unlock(&oldsighand->siglock);
		write_unlock_irq(&tasklist_lock);

1005
		__cleanup_sighand(oldsighand);
L
Linus Torvalds 已提交
1006 1007
	}

1008
	BUG_ON(!thread_group_leader(tsk));
L
Linus Torvalds 已提交
1009 1010
	return 0;
}
O
Oleg Nesterov 已提交
1011

L
Linus Torvalds 已提交
1012 1013 1014 1015
/*
 * These functions flushes out all traces of the currently running executable
 * so that a new one can be started
 */
1016
static void flush_old_files(struct files_struct * files)
L
Linus Torvalds 已提交
1017 1018
{
	long j = -1;
1019
	struct fdtable *fdt;
L
Linus Torvalds 已提交
1020 1021 1022 1023 1024 1025 1026

	spin_lock(&files->file_lock);
	for (;;) {
		unsigned long set, i;

		j++;
		i = j * __NFDBITS;
1027
		fdt = files_fdtable(files);
1028
		if (i >= fdt->max_fds)
L
Linus Torvalds 已提交
1029
			break;
1030
		set = fdt->close_on_exec->fds_bits[j];
L
Linus Torvalds 已提交
1031 1032
		if (!set)
			continue;
1033
		fdt->close_on_exec->fds_bits[j] = 0;
L
Linus Torvalds 已提交
1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045
		spin_unlock(&files->file_lock);
		for ( ; set ; i++,set >>= 1) {
			if (set & 1) {
				sys_close(i);
			}
		}
		spin_lock(&files->file_lock);

	}
	spin_unlock(&files->file_lock);
}

1046
char *get_task_comm(char *buf, struct task_struct *tsk)
L
Linus Torvalds 已提交
1047 1048 1049 1050 1051
{
	/* buf must be at least sizeof(tsk->comm) in size */
	task_lock(tsk);
	strncpy(buf, tsk->comm, sizeof(tsk->comm));
	task_unlock(tsk);
1052
	return buf;
L
Linus Torvalds 已提交
1053
}
1054
EXPORT_SYMBOL_GPL(get_task_comm);
L
Linus Torvalds 已提交
1055 1056 1057 1058

void set_task_comm(struct task_struct *tsk, char *buf)
{
	task_lock(tsk);
1059 1060 1061 1062 1063 1064 1065 1066 1067

	/*
	 * Threads may access current->comm without holding
	 * the task lock, so write the string carefully.
	 * Readers without a lock may see incomplete new
	 * names but are safe from non-terminating string reads.
	 */
	memset(tsk->comm, 0, TASK_COMM_LEN);
	wmb();
L
Linus Torvalds 已提交
1068 1069
	strlcpy(tsk->comm, buf, sizeof(tsk->comm));
	task_unlock(tsk);
1070
	perf_event_comm(tsk);
L
Linus Torvalds 已提交
1071 1072 1073 1074
}

int flush_old_exec(struct linux_binprm * bprm)
{
1075
	int retval;
L
Linus Torvalds 已提交
1076 1077 1078 1079 1080 1081 1082 1083 1084

	/*
	 * Make sure we have a private signal table and that
	 * we are unassociated from the previous thread group.
	 */
	retval = de_thread(current);
	if (retval)
		goto out;

M
Matt Helsley 已提交
1085 1086
	set_mm_exe_file(bprm->mm, bprm->file);

L
Linus Torvalds 已提交
1087 1088 1089
	/*
	 * Release all of the old mmap stuff
	 */
1090
	acct_arg_size(bprm, 0);
L
Linus Torvalds 已提交
1091 1092
	retval = exec_mmap(bprm->mm);
	if (retval)
1093
		goto out;
L
Linus Torvalds 已提交
1094 1095

	bprm->mm = NULL;		/* We're using it now */
1096

1097
	current->flags &= ~(PF_RANDOMIZE | PF_KTHREAD);
1098 1099 1100
	flush_thread();
	current->personality &= ~bprm->per_clear;

1101 1102 1103 1104 1105 1106 1107 1108 1109 1110
	return 0;

out:
	return retval;
}
EXPORT_SYMBOL(flush_old_exec);

void setup_new_exec(struct linux_binprm * bprm)
{
	int i, ch;
1111
	const char *name;
1112 1113 1114
	char tcomm[sizeof(current->comm)];

	arch_pick_mmap_layout(current->mm);
L
Linus Torvalds 已提交
1115 1116 1117 1118

	/* This is the point of no return */
	current->sas_ss_sp = current->sas_ss_size = 0;

1119
	if (current_euid() == current_uid() && current_egid() == current_gid())
1120
		set_dumpable(current->mm, 1);
A
Alan Cox 已提交
1121
	else
1122
		set_dumpable(current->mm, suid_dumpable);
A
Alan Cox 已提交
1123

L
Linus Torvalds 已提交
1124
	name = bprm->filename;
1125 1126

	/* Copies the binary name from after last slash */
L
Linus Torvalds 已提交
1127 1128
	for (i=0; (ch = *(name++)) != '\0';) {
		if (ch == '/')
1129
			i = 0; /* overwrite what we wrote */
L
Linus Torvalds 已提交
1130 1131 1132 1133 1134 1135 1136
		else
			if (i < (sizeof(tcomm) - 1))
				tcomm[i++] = ch;
	}
	tcomm[i] = '\0';
	set_task_comm(current, tcomm);

1137 1138 1139 1140 1141 1142
	/* Set the new mm task size. We have to do that late because it may
	 * depend on TIF_32BIT which is only updated in flush_thread() on
	 * some architectures like powerpc
	 */
	current->mm->task_size = TASK_SIZE;

1143 1144 1145
	/* install the new credentials */
	if (bprm->cred->uid != current_euid() ||
	    bprm->cred->gid != current_egid()) {
1146 1147
		current->pdeath_signal = 0;
	} else if (file_permission(bprm->file, MAY_READ) ||
1148
		   bprm->interp_flags & BINPRM_FLAGS_ENFORCE_NONDUMP) {
1149
		set_dumpable(current->mm, suid_dumpable);
L
Linus Torvalds 已提交
1150 1151
	}

I
Ingo Molnar 已提交
1152 1153 1154 1155 1156
	/*
	 * Flush performance counters when crossing a
	 * security domain:
	 */
	if (!get_dumpable(current->mm))
1157
		perf_event_exit_task(current);
I
Ingo Molnar 已提交
1158

L
Linus Torvalds 已提交
1159 1160 1161 1162 1163 1164 1165 1166
	/* An exec changes our domain. We are no longer part of the thread
	   group */

	current->self_exec_id++;
			
	flush_signal_handlers(current, 0);
	flush_old_files(current->files);
}
1167
EXPORT_SYMBOL(setup_new_exec);
L
Linus Torvalds 已提交
1168

1169 1170 1171 1172 1173 1174 1175 1176
/*
 * Prepare credentials and lock ->cred_guard_mutex.
 * install_exec_creds() commits the new creds and drops the lock.
 * Or, if exec fails before, free_bprm() should release ->cred and
 * and unlock.
 */
int prepare_bprm_creds(struct linux_binprm *bprm)
{
1177
	if (mutex_lock_interruptible(&current->signal->cred_guard_mutex))
1178 1179 1180 1181 1182 1183
		return -ERESTARTNOINTR;

	bprm->cred = prepare_exec_creds();
	if (likely(bprm->cred))
		return 0;

1184
	mutex_unlock(&current->signal->cred_guard_mutex);
1185 1186 1187 1188 1189 1190 1191
	return -ENOMEM;
}

void free_bprm(struct linux_binprm *bprm)
{
	free_arg_pages(bprm);
	if (bprm->cred) {
1192
		mutex_unlock(&current->signal->cred_guard_mutex);
1193 1194 1195 1196 1197
		abort_creds(bprm->cred);
	}
	kfree(bprm);
}

1198 1199 1200 1201 1202 1203 1204 1205 1206
/*
 * install the new credentials for this executable
 */
void install_exec_creds(struct linux_binprm *bprm)
{
	security_bprm_committing_creds(bprm);

	commit_creds(bprm->cred);
	bprm->cred = NULL;
1207 1208
	/*
	 * cred_guard_mutex must be held at least to this point to prevent
1209
	 * ptrace_attach() from altering our determination of the task's
1210 1211
	 * credentials; any time after this it may be unlocked.
	 */
1212
	security_bprm_committed_creds(bprm);
1213
	mutex_unlock(&current->signal->cred_guard_mutex);
1214 1215 1216 1217 1218
}
EXPORT_SYMBOL(install_exec_creds);

/*
 * determine how safe it is to execute the proposed program
1219
 * - the caller must hold ->cred_guard_mutex to protect against
1220 1221
 *   PTRACE_ATTACH
 */
A
Al Viro 已提交
1222
int check_unsafe_exec(struct linux_binprm *bprm)
1223
{
D
David Howells 已提交
1224
	struct task_struct *p = current, *t;
1225
	unsigned n_fs;
A
Al Viro 已提交
1226
	int res = 0;
1227

T
Tejun Heo 已提交
1228 1229 1230 1231 1232 1233
	if (p->ptrace) {
		if (p->ptrace & PT_PTRACE_CAP)
			bprm->unsafe |= LSM_UNSAFE_PTRACE_CAP;
		else
			bprm->unsafe |= LSM_UNSAFE_PTRACE;
	}
1234

D
David Howells 已提交
1235
	n_fs = 1;
N
Nick Piggin 已提交
1236
	spin_lock(&p->fs->lock);
1237
	rcu_read_lock();
D
David Howells 已提交
1238 1239 1240 1241
	for (t = next_thread(p); t != p; t = next_thread(t)) {
		if (t->fs == p->fs)
			n_fs++;
	}
1242
	rcu_read_unlock();
D
David Howells 已提交
1243

1244
	if (p->fs->users > n_fs) {
1245
		bprm->unsafe |= LSM_UNSAFE_SHARE;
A
Al Viro 已提交
1246
	} else {
1247 1248 1249 1250 1251
		res = -EAGAIN;
		if (!p->fs->in_exec) {
			p->fs->in_exec = 1;
			res = 1;
		}
A
Al Viro 已提交
1252
	}
N
Nick Piggin 已提交
1253
	spin_unlock(&p->fs->lock);
A
Al Viro 已提交
1254 1255

	return res;
1256 1257
}

L
Linus Torvalds 已提交
1258 1259 1260
/* 
 * Fill the binprm structure from the inode. 
 * Check permissions, then read the first 128 (BINPRM_BUF_SIZE) bytes
1261 1262
 *
 * This may be called multiple times for binary chains (scripts for example).
L
Linus Torvalds 已提交
1263 1264 1265
 */
int prepare_binprm(struct linux_binprm *bprm)
{
1266
	umode_t mode;
1267
	struct inode * inode = bprm->file->f_path.dentry->d_inode;
L
Linus Torvalds 已提交
1268 1269 1270 1271 1272 1273
	int retval;

	mode = inode->i_mode;
	if (bprm->file->f_op == NULL)
		return -EACCES;

1274 1275 1276
	/* clear any previous set[ug]id data from a previous binary */
	bprm->cred->euid = current_euid();
	bprm->cred->egid = current_egid();
L
Linus Torvalds 已提交
1277

1278
	if (!(bprm->file->f_path.mnt->mnt_flags & MNT_NOSUID)) {
L
Linus Torvalds 已提交
1279 1280
		/* Set-uid? */
		if (mode & S_ISUID) {
1281 1282
			bprm->per_clear |= PER_CLEAR_ON_SETID;
			bprm->cred->euid = inode->i_uid;
L
Linus Torvalds 已提交
1283 1284 1285 1286 1287 1288 1289 1290 1291
		}

		/* Set-gid? */
		/*
		 * If setgid is set but no group execute bit then this
		 * is a candidate for mandatory locking, not a setgid
		 * executable.
		 */
		if ((mode & (S_ISGID | S_IXGRP)) == (S_ISGID | S_IXGRP)) {
1292 1293
			bprm->per_clear |= PER_CLEAR_ON_SETID;
			bprm->cred->egid = inode->i_gid;
L
Linus Torvalds 已提交
1294 1295 1296 1297
		}
	}

	/* fill in binprm security blob */
1298
	retval = security_bprm_set_creds(bprm);
L
Linus Torvalds 已提交
1299 1300
	if (retval)
		return retval;
1301
	bprm->cred_prepared = 1;
L
Linus Torvalds 已提交
1302

1303 1304
	memset(bprm->buf, 0, BINPRM_BUF_SIZE);
	return kernel_read(bprm->file, 0, bprm->buf, BINPRM_BUF_SIZE);
L
Linus Torvalds 已提交
1305 1306 1307 1308
}

EXPORT_SYMBOL(prepare_binprm);

N
Nick Piggin 已提交
1309 1310 1311 1312 1313
/*
 * Arguments are '\0' separated strings found at the location bprm->p
 * points to; chop off the first by relocating brpm->p to right after
 * the first '\0' encountered.
 */
1314
int remove_arg_zero(struct linux_binprm *bprm)
L
Linus Torvalds 已提交
1315
{
1316 1317 1318 1319
	int ret = 0;
	unsigned long offset;
	char *kaddr;
	struct page *page;
N
Nick Piggin 已提交
1320

1321 1322
	if (!bprm->argc)
		return 0;
L
Linus Torvalds 已提交
1323

1324 1325 1326 1327 1328 1329 1330 1331
	do {
		offset = bprm->p & ~PAGE_MASK;
		page = get_arg_page(bprm, bprm->p, 0);
		if (!page) {
			ret = -EFAULT;
			goto out;
		}
		kaddr = kmap_atomic(page, KM_USER0);
N
Nick Piggin 已提交
1332

1333 1334 1335
		for (; offset < PAGE_SIZE && kaddr[offset];
				offset++, bprm->p++)
			;
N
Nick Piggin 已提交
1336

1337 1338
		kunmap_atomic(kaddr, KM_USER0);
		put_arg_page(page);
N
Nick Piggin 已提交
1339

1340 1341 1342
		if (offset == PAGE_SIZE)
			free_arg_page(bprm, (bprm->p >> PAGE_SHIFT) - 1);
	} while (offset == PAGE_SIZE);
N
Nick Piggin 已提交
1343

1344 1345 1346
	bprm->p++;
	bprm->argc--;
	ret = 0;
N
Nick Piggin 已提交
1347

1348 1349
out:
	return ret;
L
Linus Torvalds 已提交
1350 1351 1352 1353 1354 1355 1356 1357
}
EXPORT_SYMBOL(remove_arg_zero);

/*
 * cycle the list of binary formats handler, until one recognizes the image
 */
int search_binary_handler(struct linux_binprm *bprm,struct pt_regs *regs)
{
1358
	unsigned int depth = bprm->recursion_depth;
L
Linus Torvalds 已提交
1359 1360 1361 1362 1363 1364 1365 1366 1367 1368
	int try,retval;
	struct linux_binfmt *fmt;

	retval = security_bprm_check(bprm);
	if (retval)
		return retval;

	/* kernel module loader fixup */
	/* so we don't try to load run modprobe in kernel space. */
	set_fs(USER_DS);
A
Al Viro 已提交
1369 1370 1371 1372 1373

	retval = audit_bprm(bprm);
	if (retval)
		return retval;

L
Linus Torvalds 已提交
1374 1375 1376
	retval = -ENOENT;
	for (try=0; try<2; try++) {
		read_lock(&binfmt_lock);
A
Alexey Dobriyan 已提交
1377
		list_for_each_entry(fmt, &formats, lh) {
L
Linus Torvalds 已提交
1378 1379 1380 1381 1382 1383 1384
			int (*fn)(struct linux_binprm *, struct pt_regs *) = fmt->load_binary;
			if (!fn)
				continue;
			if (!try_module_get(fmt->module))
				continue;
			read_unlock(&binfmt_lock);
			retval = fn(bprm, regs);
1385 1386 1387 1388 1389 1390
			/*
			 * Restore the depth counter to its starting value
			 * in this call, so we don't have to rely on every
			 * load_binary function to restore it on return.
			 */
			bprm->recursion_depth = depth;
L
Linus Torvalds 已提交
1391
			if (retval >= 0) {
1392
				if (depth == 0)
T
Tejun Heo 已提交
1393
					ptrace_event(PTRACE_EVENT_EXEC, 0);
L
Linus Torvalds 已提交
1394 1395 1396 1397 1398 1399
				put_binfmt(fmt);
				allow_write_access(bprm->file);
				if (bprm->file)
					fput(bprm->file);
				bprm->file = NULL;
				current->did_exec = 1;
M
Matt Helsley 已提交
1400
				proc_exec_connector(current);
L
Linus Torvalds 已提交
1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414
				return retval;
			}
			read_lock(&binfmt_lock);
			put_binfmt(fmt);
			if (retval != -ENOEXEC || bprm->mm == NULL)
				break;
			if (!bprm->file) {
				read_unlock(&binfmt_lock);
				return retval;
			}
		}
		read_unlock(&binfmt_lock);
		if (retval != -ENOEXEC || bprm->mm == NULL) {
			break;
J
Johannes Berg 已提交
1415 1416
#ifdef CONFIG_MODULES
		} else {
L
Linus Torvalds 已提交
1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434
#define printable(c) (((c)=='\t') || ((c)=='\n') || (0x20<=(c) && (c)<=0x7e))
			if (printable(bprm->buf[0]) &&
			    printable(bprm->buf[1]) &&
			    printable(bprm->buf[2]) &&
			    printable(bprm->buf[3]))
				break; /* -ENOEXEC */
			request_module("binfmt-%04x", *(unsigned short *)(&bprm->buf[2]));
#endif
		}
	}
	return retval;
}

EXPORT_SYMBOL(search_binary_handler);

/*
 * sys_execve() executes a new program.
 */
1435 1436 1437 1438
static int do_execve_common(const char *filename,
				struct user_arg_ptr argv,
				struct user_arg_ptr envp,
				struct pt_regs *regs)
L
Linus Torvalds 已提交
1439 1440 1441
{
	struct linux_binprm *bprm;
	struct file *file;
1442
	struct files_struct *displaced;
1443
	bool clear_in_exec;
L
Linus Torvalds 已提交
1444 1445
	int retval;

1446
	retval = unshare_files(&displaced);
1447 1448 1449
	if (retval)
		goto out_ret;

L
Linus Torvalds 已提交
1450
	retval = -ENOMEM;
1451
	bprm = kzalloc(sizeof(*bprm), GFP_KERNEL);
L
Linus Torvalds 已提交
1452
	if (!bprm)
1453
		goto out_files;
L
Linus Torvalds 已提交
1454

1455 1456
	retval = prepare_bprm_creds(bprm);
	if (retval)
1457
		goto out_free;
A
Al Viro 已提交
1458 1459

	retval = check_unsafe_exec(bprm);
1460
	if (retval < 0)
1461
		goto out_free;
1462
	clear_in_exec = retval;
1463
	current->in_execve = 1;
1464

L
Linus Torvalds 已提交
1465 1466 1467
	file = open_exec(filename);
	retval = PTR_ERR(file);
	if (IS_ERR(file))
A
Al Viro 已提交
1468
		goto out_unmark;
L
Linus Torvalds 已提交
1469 1470 1471 1472 1473 1474 1475

	sched_exec();

	bprm->file = file;
	bprm->filename = filename;
	bprm->interp = filename;

1476 1477 1478
	retval = bprm_mm_init(bprm);
	if (retval)
		goto out_file;
L
Linus Torvalds 已提交
1479

1480
	bprm->argc = count(argv, MAX_ARG_STRINGS);
L
Linus Torvalds 已提交
1481
	if ((retval = bprm->argc) < 0)
1482
		goto out;
L
Linus Torvalds 已提交
1483

1484
	bprm->envc = count(envp, MAX_ARG_STRINGS);
L
Linus Torvalds 已提交
1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505
	if ((retval = bprm->envc) < 0)
		goto out;

	retval = prepare_binprm(bprm);
	if (retval < 0)
		goto out;

	retval = copy_strings_kernel(1, &bprm->filename, bprm);
	if (retval < 0)
		goto out;

	bprm->exec = bprm->p;
	retval = copy_strings(bprm->envc, envp, bprm);
	if (retval < 0)
		goto out;

	retval = copy_strings(bprm->argc, argv, bprm);
	if (retval < 0)
		goto out;

	retval = search_binary_handler(bprm,regs);
1506 1507
	if (retval < 0)
		goto out;
L
Linus Torvalds 已提交
1508

1509
	/* execve succeeded */
A
Al Viro 已提交
1510
	current->fs->in_exec = 0;
1511
	current->in_execve = 0;
1512 1513 1514 1515 1516
	acct_update_integrals(current);
	free_bprm(bprm);
	if (displaced)
		put_files_struct(displaced);
	return retval;
L
Linus Torvalds 已提交
1517

1518
out:
1519 1520 1521 1522
	if (bprm->mm) {
		acct_arg_size(bprm, 0);
		mmput(bprm->mm);
	}
L
Linus Torvalds 已提交
1523 1524 1525 1526 1527 1528

out_file:
	if (bprm->file) {
		allow_write_access(bprm->file);
		fput(bprm->file);
	}
1529

A
Al Viro 已提交
1530
out_unmark:
1531 1532
	if (clear_in_exec)
		current->fs->in_exec = 0;
1533
	current->in_execve = 0;
1534 1535

out_free:
1536
	free_bprm(bprm);
L
Linus Torvalds 已提交
1537

1538
out_files:
1539 1540
	if (displaced)
		reset_files_struct(displaced);
L
Linus Torvalds 已提交
1541 1542 1543 1544
out_ret:
	return retval;
}

1545 1546 1547 1548 1549
int do_execve(const char *filename,
	const char __user *const __user *__argv,
	const char __user *const __user *__envp,
	struct pt_regs *regs)
{
1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568
	struct user_arg_ptr argv = { .ptr.native = __argv };
	struct user_arg_ptr envp = { .ptr.native = __envp };
	return do_execve_common(filename, argv, envp, regs);
}

#ifdef CONFIG_COMPAT
int compat_do_execve(char *filename,
	compat_uptr_t __user *__argv,
	compat_uptr_t __user *__envp,
	struct pt_regs *regs)
{
	struct user_arg_ptr argv = {
		.is_compat = true,
		.ptr.compat = __argv,
	};
	struct user_arg_ptr envp = {
		.is_compat = true,
		.ptr.compat = __envp,
	};
1569 1570
	return do_execve_common(filename, argv, envp, regs);
}
1571
#endif
1572

1573
void set_binfmt(struct linux_binfmt *new)
L
Linus Torvalds 已提交
1574
{
1575 1576 1577 1578
	struct mm_struct *mm = current->mm;

	if (mm->binfmt)
		module_put(mm->binfmt->module);
L
Linus Torvalds 已提交
1579

1580
	mm->binfmt = new;
1581 1582
	if (new)
		__module_get(new->module);
L
Linus Torvalds 已提交
1583 1584 1585 1586
}

EXPORT_SYMBOL(set_binfmt);

1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631
static int expand_corename(struct core_name *cn)
{
	char *old_corename = cn->corename;

	cn->size = CORENAME_MAX_SIZE * atomic_inc_return(&call_count);
	cn->corename = krealloc(old_corename, cn->size, GFP_KERNEL);

	if (!cn->corename) {
		kfree(old_corename);
		return -ENOMEM;
	}

	return 0;
}

static int cn_printf(struct core_name *cn, const char *fmt, ...)
{
	char *cur;
	int need;
	int ret;
	va_list arg;

	va_start(arg, fmt);
	need = vsnprintf(NULL, 0, fmt, arg);
	va_end(arg);

	if (likely(need < cn->size - cn->used - 1))
		goto out_printf;

	ret = expand_corename(cn);
	if (ret)
		goto expand_fail;

out_printf:
	cur = cn->corename + cn->used;
	va_start(arg, fmt);
	vsnprintf(cur, need + 1, fmt, arg);
	va_end(arg);
	cn->used += need;
	return 0;

expand_fail:
	return ret;
}

1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666
static int cn_print_exe_file(struct core_name *cn)
{
	struct file *exe_file;
	char *pathbuf, *path, *p;
	int ret;

	exe_file = get_mm_exe_file(current->mm);
	if (!exe_file)
		return cn_printf(cn, "(unknown)");

	pathbuf = kmalloc(PATH_MAX, GFP_TEMPORARY);
	if (!pathbuf) {
		ret = -ENOMEM;
		goto put_exe_file;
	}

	path = d_path(&exe_file->f_path, pathbuf, PATH_MAX);
	if (IS_ERR(path)) {
		ret = PTR_ERR(path);
		goto free_buf;
	}

	for (p = path; *p; p++)
		if (*p == '/')
			*p = '!';

	ret = cn_printf(cn, "%s", path);

free_buf:
	kfree(pathbuf);
put_exe_file:
	fput(exe_file);
	return ret;
}

L
Linus Torvalds 已提交
1667 1668 1669 1670
/* format_corename will inspect the pattern parameter, and output a
 * name into corename, which must have space for at least
 * CORENAME_MAX_SIZE bytes plus one byte for the zero terminator.
 */
1671
static int format_corename(struct core_name *cn, long signr)
L
Linus Torvalds 已提交
1672
{
1673
	const struct cred *cred = current_cred();
1674 1675
	const char *pat_ptr = core_pattern;
	int ispipe = (*pat_ptr == '|');
L
Linus Torvalds 已提交
1676
	int pid_in_pattern = 0;
1677 1678 1679 1680 1681 1682 1683 1684
	int err = 0;

	cn->size = CORENAME_MAX_SIZE * atomic_read(&call_count);
	cn->corename = kmalloc(cn->size, GFP_KERNEL);
	cn->used = 0;

	if (!cn->corename)
		return -ENOMEM;
L
Linus Torvalds 已提交
1685 1686 1687 1688 1689

	/* Repeat as long as we have more pattern to process and more output
	   space */
	while (*pat_ptr) {
		if (*pat_ptr != '%') {
1690
			if (*pat_ptr == 0)
L
Linus Torvalds 已提交
1691
				goto out;
1692
			err = cn_printf(cn, "%c", *pat_ptr++);
L
Linus Torvalds 已提交
1693 1694
		} else {
			switch (*++pat_ptr) {
1695
			/* single % at the end, drop that */
L
Linus Torvalds 已提交
1696 1697 1698 1699
			case 0:
				goto out;
			/* Double percent, output one percent */
			case '%':
1700
				err = cn_printf(cn, "%c", '%');
L
Linus Torvalds 已提交
1701 1702 1703 1704
				break;
			/* pid */
			case 'p':
				pid_in_pattern = 1;
1705 1706
				err = cn_printf(cn, "%d",
					      task_tgid_vnr(current));
L
Linus Torvalds 已提交
1707 1708 1709
				break;
			/* uid */
			case 'u':
1710
				err = cn_printf(cn, "%d", cred->uid);
L
Linus Torvalds 已提交
1711 1712 1713
				break;
			/* gid */
			case 'g':
1714
				err = cn_printf(cn, "%d", cred->gid);
L
Linus Torvalds 已提交
1715 1716 1717
				break;
			/* signal that caused the coredump */
			case 's':
1718
				err = cn_printf(cn, "%ld", signr);
L
Linus Torvalds 已提交
1719 1720 1721 1722 1723
				break;
			/* UNIX time of coredump */
			case 't': {
				struct timeval tv;
				do_gettimeofday(&tv);
1724
				err = cn_printf(cn, "%lu", tv.tv_sec);
L
Linus Torvalds 已提交
1725 1726 1727 1728 1729
				break;
			}
			/* hostname */
			case 'h':
				down_read(&uts_sem);
1730 1731
				err = cn_printf(cn, "%s",
					      utsname()->nodename);
L
Linus Torvalds 已提交
1732 1733 1734 1735
				up_read(&uts_sem);
				break;
			/* executable */
			case 'e':
1736
				err = cn_printf(cn, "%s", current->comm);
L
Linus Torvalds 已提交
1737
				break;
1738 1739 1740
			case 'E':
				err = cn_print_exe_file(cn);
				break;
1741 1742
			/* core limit size */
			case 'c':
1743 1744
				err = cn_printf(cn, "%lu",
					      rlimit(RLIMIT_CORE));
1745
				break;
L
Linus Torvalds 已提交
1746 1747 1748 1749 1750
			default:
				break;
			}
			++pat_ptr;
		}
1751 1752 1753

		if (err)
			return err;
L
Linus Torvalds 已提交
1754
	}
1755

L
Linus Torvalds 已提交
1756 1757 1758 1759
	/* Backward compatibility with core_uses_pid:
	 *
	 * If core_pattern does not include a %p (as is the default)
	 * and core_uses_pid is set, then .%pid will be appended to
1760
	 * the filename. Do not do this for piped commands. */
1761
	if (!ispipe && !pid_in_pattern && core_uses_pid) {
1762 1763 1764
		err = cn_printf(cn, ".%d", task_tgid_vnr(current));
		if (err)
			return err;
L
Linus Torvalds 已提交
1765
	}
1766 1767
out:
	return ispipe;
L
Linus Torvalds 已提交
1768 1769
}

1770
static int zap_process(struct task_struct *start, int exit_code)
1771 1772
{
	struct task_struct *t;
1773
	int nr = 0;
1774

1775
	start->signal->flags = SIGNAL_GROUP_EXIT;
1776
	start->signal->group_exit_code = exit_code;
1777
	start->signal->group_stop_count = 0;
1778 1779 1780

	t = start;
	do {
1781
		task_clear_jobctl_pending(t, JOBCTL_PENDING_MASK);
1782
		if (t != current && t->mm) {
1783 1784
			sigaddset(&t->pending.signal, SIGKILL);
			signal_wake_up(t, 1);
1785
			nr++;
1786
		}
1787
	} while_each_thread(start, t);
1788 1789

	return nr;
1790 1791
}

1792
static inline int zap_threads(struct task_struct *tsk, struct mm_struct *mm,
1793
				struct core_state *core_state, int exit_code)
L
Linus Torvalds 已提交
1794 1795
{
	struct task_struct *g, *p;
1796
	unsigned long flags;
1797
	int nr = -EAGAIN;
1798 1799

	spin_lock_irq(&tsk->sighand->siglock);
1800
	if (!signal_group_exit(tsk->signal)) {
1801
		mm->core_state = core_state;
1802
		nr = zap_process(tsk, exit_code);
L
Linus Torvalds 已提交
1803
	}
1804
	spin_unlock_irq(&tsk->sighand->siglock);
1805 1806
	if (unlikely(nr < 0))
		return nr;
L
Linus Torvalds 已提交
1807

1808
	if (atomic_read(&mm->mm_users) == nr + 1)
1809
		goto done;
1810 1811
	/*
	 * We should find and kill all tasks which use this mm, and we should
1812
	 * count them correctly into ->nr_threads. We don't take tasklist
1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839
	 * lock, but this is safe wrt:
	 *
	 * fork:
	 *	None of sub-threads can fork after zap_process(leader). All
	 *	processes which were created before this point should be
	 *	visible to zap_threads() because copy_process() adds the new
	 *	process to the tail of init_task.tasks list, and lock/unlock
	 *	of ->siglock provides a memory barrier.
	 *
	 * do_exit:
	 *	The caller holds mm->mmap_sem. This means that the task which
	 *	uses this mm can't pass exit_mm(), so it can't exit or clear
	 *	its ->mm.
	 *
	 * de_thread:
	 *	It does list_replace_rcu(&leader->tasks, &current->tasks),
	 *	we must see either old or new leader, this does not matter.
	 *	However, it can change p->sighand, so lock_task_sighand(p)
	 *	must be used. Since p->mm != NULL and we hold ->mmap_sem
	 *	it can't fail.
	 *
	 *	Note also that "g" can be the old leader with ->mm == NULL
	 *	and already unhashed and thus removed from ->thread_group.
	 *	This is OK, __unhash_process()->list_del_rcu() does not
	 *	clear the ->next pointer, we will find the new leader via
	 *	next_thread().
	 */
1840
	rcu_read_lock();
1841
	for_each_process(g) {
1842 1843
		if (g == tsk->group_leader)
			continue;
1844 1845
		if (g->flags & PF_KTHREAD)
			continue;
1846 1847 1848
		p = g;
		do {
			if (p->mm) {
1849
				if (unlikely(p->mm == mm)) {
1850
					lock_task_sighand(p, &flags);
1851
					nr += zap_process(p, exit_code);
1852 1853
					unlock_task_sighand(p, &flags);
				}
1854 1855
				break;
			}
1856
		} while_each_thread(g, p);
1857
	}
1858
	rcu_read_unlock();
1859
done:
1860
	atomic_set(&core_state->nr_threads, nr);
1861
	return nr;
L
Linus Torvalds 已提交
1862 1863
}

1864
static int coredump_wait(int exit_code, struct core_state *core_state)
L
Linus Torvalds 已提交
1865
{
1866 1867 1868
	struct task_struct *tsk = current;
	struct mm_struct *mm = tsk->mm;
	struct completion *vfork_done;
1869
	int core_waiters = -EBUSY;
L
Linus Torvalds 已提交
1870

1871
	init_completion(&core_state->startup);
1872 1873
	core_state->dumper.task = tsk;
	core_state->dumper.next = NULL;
1874 1875 1876 1877

	down_write(&mm->mmap_sem);
	if (!mm->core_state)
		core_waiters = zap_threads(tsk, mm, core_state, exit_code);
O
Oleg Nesterov 已提交
1878 1879
	up_write(&mm->mmap_sem);

1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892
	if (unlikely(core_waiters < 0))
		goto fail;

	/*
	 * Make sure nobody is waiting for us to release the VM,
	 * otherwise we can deadlock when we wait on each other
	 */
	vfork_done = tsk->vfork_done;
	if (vfork_done) {
		tsk->vfork_done = NULL;
		complete(vfork_done);
	}

O
Oleg Nesterov 已提交
1893
	if (core_waiters)
1894
		wait_for_completion(&core_state->startup);
1895 1896
fail:
	return core_waiters;
L
Linus Torvalds 已提交
1897 1898
}

O
Oleg Nesterov 已提交
1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919
static void coredump_finish(struct mm_struct *mm)
{
	struct core_thread *curr, *next;
	struct task_struct *task;

	next = mm->core_state->dumper.next;
	while ((curr = next) != NULL) {
		next = curr->next;
		task = curr->task;
		/*
		 * see exit_mm(), curr->task must not see
		 * ->task == NULL before we read ->next.
		 */
		smp_mb();
		curr->task = NULL;
		wake_up_process(task);
	}

	mm->core_state = NULL;
}

1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960
/*
 * set_dumpable converts traditional three-value dumpable to two flags and
 * stores them into mm->flags.  It modifies lower two bits of mm->flags, but
 * these bits are not changed atomically.  So get_dumpable can observe the
 * intermediate state.  To avoid doing unexpected behavior, get get_dumpable
 * return either old dumpable or new one by paying attention to the order of
 * modifying the bits.
 *
 * dumpable |   mm->flags (binary)
 * old  new | initial interim  final
 * ---------+-----------------------
 *  0    1  |   00      01      01
 *  0    2  |   00      10(*)   11
 *  1    0  |   01      00      00
 *  1    2  |   01      11      11
 *  2    0  |   11      10(*)   00
 *  2    1  |   11      11      01
 *
 * (*) get_dumpable regards interim value of 10 as 11.
 */
void set_dumpable(struct mm_struct *mm, int value)
{
	switch (value) {
	case 0:
		clear_bit(MMF_DUMPABLE, &mm->flags);
		smp_wmb();
		clear_bit(MMF_DUMP_SECURELY, &mm->flags);
		break;
	case 1:
		set_bit(MMF_DUMPABLE, &mm->flags);
		smp_wmb();
		clear_bit(MMF_DUMP_SECURELY, &mm->flags);
		break;
	case 2:
		set_bit(MMF_DUMP_SECURELY, &mm->flags);
		smp_wmb();
		set_bit(MMF_DUMPABLE, &mm->flags);
		break;
	}
}

1961
static int __get_dumpable(unsigned long mm_flags)
1962 1963 1964
{
	int ret;

1965
	ret = mm_flags & MMF_DUMPABLE_MASK;
1966 1967 1968
	return (ret >= 2) ? 2 : ret;
}

1969 1970 1971 1972 1973
int get_dumpable(struct mm_struct *mm)
{
	return __get_dumpable(mm->flags);
}

1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996
static void wait_for_dump_helpers(struct file *file)
{
	struct pipe_inode_info *pipe;

	pipe = file->f_path.dentry->d_inode->i_pipe;

	pipe_lock(pipe);
	pipe->readers++;
	pipe->writers--;

	while ((pipe->readers > 1) && (!signal_pending(current))) {
		wake_up_interruptible_sync(&pipe->wait);
		kill_fasync(&pipe->fasync_readers, SIGIO, POLL_IN);
		pipe_wait(pipe);
	}

	pipe->readers--;
	pipe->writers++;
	pipe_unlock(pipe);

}


1997
/*
H
Holger Hans Peter Freyther 已提交
1998
 * umh_pipe_setup
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040
 * helper function to customize the process used
 * to collect the core in userspace.  Specifically
 * it sets up a pipe and installs it as fd 0 (stdin)
 * for the process.  Returns 0 on success, or
 * PTR_ERR on failure.
 * Note that it also sets the core limit to 1.  This
 * is a special value that we use to trap recursive
 * core dumps
 */
static int umh_pipe_setup(struct subprocess_info *info)
{
	struct file *rp, *wp;
	struct fdtable *fdt;
	struct coredump_params *cp = (struct coredump_params *)info->data;
	struct files_struct *cf = current->files;

	wp = create_write_pipe(0);
	if (IS_ERR(wp))
		return PTR_ERR(wp);

	rp = create_read_pipe(wp, 0);
	if (IS_ERR(rp)) {
		free_write_pipe(wp);
		return PTR_ERR(rp);
	}

	cp->file = wp;

	sys_close(0);
	fd_install(0, rp);
	spin_lock(&cf->file_lock);
	fdt = files_fdtable(cf);
	FD_SET(0, fdt->open_fds);
	FD_CLR(0, fdt->close_on_exec);
	spin_unlock(&cf->file_lock);

	/* and disallow core files too */
	current->signal->rlim[RLIMIT_CORE] = (struct rlimit){1, 1};

	return 0;
}

W
WANG Cong 已提交
2041
void do_coredump(long signr, int exit_code, struct pt_regs *regs)
L
Linus Torvalds 已提交
2042
{
2043
	struct core_state core_state;
2044
	struct core_name cn;
L
Linus Torvalds 已提交
2045 2046
	struct mm_struct *mm = current->mm;
	struct linux_binfmt * binfmt;
D
David Howells 已提交
2047 2048
	const struct cred *old_cred;
	struct cred *cred;
L
Linus Torvalds 已提交
2049
	int retval = 0;
A
Alan Cox 已提交
2050
	int flag = 0;
O
Oleg Nesterov 已提交
2051
	int ispipe;
2052
	static atomic_t core_dump_count = ATOMIC_INIT(0);
2053 2054 2055
	struct coredump_params cprm = {
		.signr = signr,
		.regs = regs,
J
Jiri Slaby 已提交
2056
		.limit = rlimit(RLIMIT_CORE),
2057 2058 2059 2060 2061 2062
		/*
		 * We must use the same mm->flags while dumping core to avoid
		 * inconsistency of bit flags, since this flag is not protected
		 * by any locks.
		 */
		.mm_flags = mm->flags,
2063
	};
L
Linus Torvalds 已提交
2064

S
Steve Grubb 已提交
2065 2066
	audit_core_dumps(signr);

2067
	binfmt = mm->binfmt;
L
Linus Torvalds 已提交
2068 2069
	if (!binfmt || !binfmt->core_dump)
		goto fail;
2070 2071
	if (!__get_dumpable(cprm.mm_flags))
		goto fail;
D
David Howells 已提交
2072 2073

	cred = prepare_creds();
2074
	if (!cred)
D
David Howells 已提交
2075
		goto fail;
A
Alan Cox 已提交
2076 2077 2078 2079 2080
	/*
	 *	We cannot trust fsuid as being the "true" uid of the
	 *	process nor do we know its entire history. We only know it
	 *	was tainted so we dump it as root in mode 2.
	 */
2081 2082
	if (__get_dumpable(cprm.mm_flags) == 2) {
		/* Setuid core dump mode */
A
Alan Cox 已提交
2083
		flag = O_EXCL;		/* Stop rewrite attacks */
D
David Howells 已提交
2084
		cred->fsuid = 0;	/* Dump root private */
A
Alan Cox 已提交
2085
	}
2086

2087
	retval = coredump_wait(exit_code, &core_state);
2088 2089
	if (retval < 0)
		goto fail_creds;
D
David Howells 已提交
2090 2091

	old_cred = override_creds(cred);
L
Linus Torvalds 已提交
2092 2093 2094 2095 2096 2097 2098

	/*
	 * Clear any false indication of pending signals that might
	 * be seen by the filesystem code called to write the core file.
	 */
	clear_thread_flag(TIF_SIGPENDING);

2099 2100 2101 2102 2103 2104 2105
	ispipe = format_corename(&cn, signr);

	if (ispipe == -ENOMEM) {
		printk(KERN_WARNING "format_corename failed\n");
		printk(KERN_WARNING "Aborting core\n");
		goto fail_corename;
	}
2106

2107
 	if (ispipe) {
O
Oleg Nesterov 已提交
2108 2109 2110
		int dump_count;
		char **helper_argv;

2111
		if (cprm.limit == 1) {
2112 2113 2114
			/*
			 * Normally core limits are irrelevant to pipes, since
			 * we're not writing to the file system, but we use
2115 2116
			 * cprm.limit of 1 here as a speacial value. Any
			 * non-1 limit gets set to RLIM_INFINITY below, but
2117 2118
			 * a limit of 0 skips the dump.  This is a consistent
			 * way to catch recursive crashes.  We can still crash
2119
			 * if the core_pattern binary sets RLIM_CORE =  !1
2120 2121 2122 2123 2124 2125 2126
			 * but it runs as root, and can do lots of stupid things
			 * Note that we use task_tgid_vnr here to grab the pid
			 * of the process group leader.  That way we get the
			 * right pid if a thread in a multi-threaded
			 * core_pattern process dies.
			 */
			printk(KERN_WARNING
2127
				"Process %d(%s) has RLIMIT_CORE set to 1\n",
2128 2129 2130 2131
				task_tgid_vnr(current), current->comm);
			printk(KERN_WARNING "Aborting core\n");
			goto fail_unlock;
		}
O
Oleg Nesterov 已提交
2132
		cprm.limit = RLIM_INFINITY;
2133

2134 2135 2136 2137 2138 2139 2140 2141
		dump_count = atomic_inc_return(&core_dump_count);
		if (core_pipe_limit && (core_pipe_limit < dump_count)) {
			printk(KERN_WARNING "Pid %d(%s) over core_pipe_limit\n",
			       task_tgid_vnr(current), current->comm);
			printk(KERN_WARNING "Skipping core dump\n");
			goto fail_dropcount;
		}

2142
		helper_argv = argv_split(GFP_KERNEL, cn.corename+1, NULL);
2143 2144 2145
		if (!helper_argv) {
			printk(KERN_WARNING "%s failed to allocate memory\n",
			       __func__);
2146
			goto fail_dropcount;
2147
		}
2148

O
Oleg Nesterov 已提交
2149 2150 2151 2152 2153
		retval = call_usermodehelper_fns(helper_argv[0], helper_argv,
					NULL, UMH_WAIT_EXEC, umh_pipe_setup,
					NULL, &cprm);
		argv_free(helper_argv);
		if (retval) {
2154
 			printk(KERN_INFO "Core dump to %s pipe failed\n",
2155
			       cn.corename);
O
Oleg Nesterov 已提交
2156
			goto close_fail;
2157
 		}
2158 2159 2160 2161 2162 2163
	} else {
		struct inode *inode;

		if (cprm.limit < binfmt->min_coredump)
			goto fail_unlock;

2164
		cprm.file = filp_open(cn.corename,
2165 2166
				 O_CREAT | 2 | O_NOFOLLOW | O_LARGEFILE | flag,
				 0600);
2167 2168
		if (IS_ERR(cprm.file))
			goto fail_unlock;
L
Linus Torvalds 已提交
2169

2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191
		inode = cprm.file->f_path.dentry->d_inode;
		if (inode->i_nlink > 1)
			goto close_fail;
		if (d_unhashed(cprm.file->f_path.dentry))
			goto close_fail;
		/*
		 * AK: actually i see no reason to not allow this for named
		 * pipes etc, but keep the previous behaviour for now.
		 */
		if (!S_ISREG(inode->i_mode))
			goto close_fail;
		/*
		 * Dont allow local users get cute and trick others to coredump
		 * into their pre-created files.
		 */
		if (inode->i_uid != current_fsuid())
			goto close_fail;
		if (!cprm.file->f_op || !cprm.file->f_op->write)
			goto close_fail;
		if (do_truncate(cprm.file->f_path.dentry, 0, 0, cprm.file))
			goto close_fail;
	}
L
Linus Torvalds 已提交
2192

2193
	retval = binfmt->core_dump(&cprm);
L
Linus Torvalds 已提交
2194 2195
	if (retval)
		current->signal->group_exit_code |= 0x80;
O
Oleg Nesterov 已提交
2196

2197
	if (ispipe && core_pipe_limit)
2198
		wait_for_dump_helpers(cprm.file);
O
Oleg Nesterov 已提交
2199 2200 2201
close_fail:
	if (cprm.file)
		filp_close(cprm.file, NULL);
2202
fail_dropcount:
O
Oleg Nesterov 已提交
2203
	if (ispipe)
2204
		atomic_dec(&core_dump_count);
L
Linus Torvalds 已提交
2205
fail_unlock:
2206 2207
	kfree(cn.corename);
fail_corename:
2208
	coredump_finish(mm);
D
David Howells 已提交
2209
	revert_creds(old_cred);
2210
fail_creds:
D
David Howells 已提交
2211
	put_cred(cred);
L
Linus Torvalds 已提交
2212
fail:
W
WANG Cong 已提交
2213
	return;
L
Linus Torvalds 已提交
2214
}
2215 2216 2217 2218 2219 2220 2221 2222 2223 2224

/*
 * Core dumping helper functions.  These are the only things you should
 * do on a core-file: use only these functions to write out all the
 * necessary info.
 */
int dump_write(struct file *file, const void *addr, int nr)
{
	return access_ok(VERIFY_READ, addr, nr) && file->f_op->write(file, addr, nr, &file->f_pos) == nr;
}
2225
EXPORT_SYMBOL(dump_write);
2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253

int dump_seek(struct file *file, loff_t off)
{
	int ret = 1;

	if (file->f_op->llseek && file->f_op->llseek != no_llseek) {
		if (file->f_op->llseek(file, off, SEEK_CUR) < 0)
			return 0;
	} else {
		char *buf = (char *)get_zeroed_page(GFP_KERNEL);

		if (!buf)
			return 0;
		while (off > 0) {
			unsigned long n = off;

			if (n > PAGE_SIZE)
				n = PAGE_SIZE;
			if (!dump_write(file, buf, n)) {
				ret = 0;
				break;
			}
			off -= n;
		}
		free_page((unsigned long)buf);
	}
	return ret;
}
2254
EXPORT_SYMBOL(dump_seek);