exec.c 39.3 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 63

#include <trace/events/task.h>
64
#include "internal.h"
65
#include "coredump.h"
L
Linus Torvalds 已提交
66

67 68
#include <trace/events/sched.h>

A
Alan Cox 已提交
69 70
int suid_dumpable = 0;

A
Alexey Dobriyan 已提交
71
static LIST_HEAD(formats);
L
Linus Torvalds 已提交
72 73
static DEFINE_RWLOCK(binfmt_lock);

A
Al Viro 已提交
74
void __register_binfmt(struct linux_binfmt * fmt, int insert)
L
Linus Torvalds 已提交
75
{
A
Al Viro 已提交
76
	BUG_ON(!fmt);
77 78
	if (WARN_ON(!fmt->load_binary))
		return;
L
Linus Torvalds 已提交
79
	write_lock(&binfmt_lock);
I
Ivan Kokshaysky 已提交
80 81
	insert ? list_add(&fmt->lh, &formats) :
		 list_add_tail(&fmt->lh, &formats);
L
Linus Torvalds 已提交
82 83 84
	write_unlock(&binfmt_lock);
}

I
Ivan Kokshaysky 已提交
85
EXPORT_SYMBOL(__register_binfmt);
L
Linus Torvalds 已提交
86

87
void unregister_binfmt(struct linux_binfmt * fmt)
L
Linus Torvalds 已提交
88 89
{
	write_lock(&binfmt_lock);
A
Alexey Dobriyan 已提交
90
	list_del(&fmt->lh);
L
Linus Torvalds 已提交
91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106
	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.
 */
107
SYSCALL_DEFINE1(uselib, const char __user *, library)
L
Linus Torvalds 已提交
108
{
A
Al Viro 已提交
109
	struct linux_binfmt *fmt;
110
	struct file *file;
111
	struct filename *tmp = getname(library);
112
	int error = PTR_ERR(tmp);
113 114 115
	static const struct open_flags uselib_flags = {
		.open_flag = O_LARGEFILE | O_RDONLY | __FMODE_EXEC,
		.acc_mode = MAY_READ | MAY_EXEC | MAY_OPEN,
116 117
		.intent = LOOKUP_OPEN,
		.lookup_flags = LOOKUP_FOLLOW,
118
	};
119

120 121 122
	if (IS_ERR(tmp))
		goto out;

123
	file = do_filp_open(AT_FDCWD, tmp, &uselib_flags);
124 125 126
	putname(tmp);
	error = PTR_ERR(file);
	if (IS_ERR(file))
L
Linus Torvalds 已提交
127 128 129
		goto out;

	error = -EINVAL;
A
Al Viro 已提交
130
	if (!S_ISREG(file_inode(file)->i_mode))
L
Linus Torvalds 已提交
131 132
		goto exit;

133
	error = -EACCES;
134
	if (file->f_path.mnt->mnt_flags & MNT_NOEXEC)
L
Linus Torvalds 已提交
135 136
		goto exit;

137
	fsnotify_open(file);
138

L
Linus Torvalds 已提交
139 140
	error = -ENOEXEC;

A
Al Viro 已提交
141 142 143 144 145 146
	read_lock(&binfmt_lock);
	list_for_each_entry(fmt, &formats, lh) {
		if (!fmt->load_shlib)
			continue;
		if (!try_module_get(fmt->module))
			continue;
L
Linus Torvalds 已提交
147
		read_unlock(&binfmt_lock);
A
Al Viro 已提交
148 149 150 151 152
		error = fmt->load_shlib(file);
		read_lock(&binfmt_lock);
		put_binfmt(fmt);
		if (error != -ENOEXEC)
			break;
L
Linus Torvalds 已提交
153
	}
A
Al Viro 已提交
154
	read_unlock(&binfmt_lock);
155
exit:
L
Linus Torvalds 已提交
156 157 158 159 160
	fput(file);
out:
  	return error;
}

161
#ifdef CONFIG_MMU
O
Oleg Nesterov 已提交
162 163 164 165 166 167
/*
 * 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).
 */
168
static void acct_arg_size(struct linux_binprm *bprm, unsigned long pages)
169 170 171 172 173 174 175 176 177 178 179
{
	struct mm_struct *mm = current->mm;
	long diff = (long)(pages - bprm->vma_pages);

	if (!mm || !diff)
		return;

	bprm->vma_pages = pages;
	add_mm_counter(mm, MM_ANONPAGES, diff);
}

180
static struct page *get_arg_page(struct linux_binprm *bprm, unsigned long pos,
181 182 183 184 185 186 187
		int write)
{
	struct page *page;
	int ret;

#ifdef CONFIG_STACK_GROWSUP
	if (write) {
188
		ret = expand_downwards(bprm->vma, pos);
189 190 191 192 193 194 195 196 197 198 199
		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;
200 201
		struct rlimit *rlim;

202 203
		acct_arg_size(bprm, size / PAGE_SIZE);

204 205 206 207 208 209
		/*
		 * We've historically supported up to 32 pages (ARG_MAX)
		 * of argument strings even with small stacks
		 */
		if (size <= ARG_MAX)
			return page;
210 211 212 213 214 215 216 217

		/*
		 * 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.
		 */
218
		rlim = current->signal->rlim;
J
Jiri Slaby 已提交
219
		if (size > ACCESS_ONCE(rlim[RLIMIT_STACK].rlim_cur) / 4) {
220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248
			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)
{
249
	int err;
250 251 252 253 254
	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)
255
		return -ENOMEM;
256 257 258 259 260 261 262 263 264 265

	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.
	 */
266
	BUILD_BUG_ON(VM_STACK_FLAGS & VM_STACK_INCOMPLETE_SETUP);
267 268
	vma->vm_end = STACK_TOP_MAX;
	vma->vm_start = vma->vm_end - PAGE_SIZE;
269
	vma->vm_flags = VM_SOFTDIRTY | VM_STACK_FLAGS | VM_STACK_INCOMPLETE_SETUP;
270
	vma->vm_page_prot = vm_get_page_prot(vma->vm_flags);
271
	INIT_LIST_HEAD(&vma->anon_vma_chain);
272

273
	err = insert_vm_struct(mm, vma);
274
	if (err)
275 276 277 278 279 280 281
		goto err;

	mm->stack_vm = mm->total_vm = 1;
	up_write(&mm->mmap_sem);
	bprm->p = vma->vm_end - sizeof(void *);
	return 0;
err:
282 283 284
	up_write(&mm->mmap_sem);
	bprm->vma = NULL;
	kmem_cache_free(vm_area_cachep, vma);
285 286 287 288 289 290 291 292 293 294
	return err;
}

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

#else

295
static inline void acct_arg_size(struct linux_binprm *bprm, unsigned long pages)
296 297 298
{
}

299
static struct page *get_arg_page(struct linux_binprm *bprm, unsigned long pos,
300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358
		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().
 */
359
static int bprm_mm_init(struct linux_binprm *bprm)
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
{
	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;
}

388
struct user_arg_ptr {
389 390 391 392 393 394
#ifdef CONFIG_COMPAT
	bool is_compat;
#endif
	union {
		const char __user *const __user *native;
#ifdef CONFIG_COMPAT
A
Al Viro 已提交
395
		const compat_uptr_t __user *compat;
396 397
#endif
	} ptr;
398 399 400
};

static const char __user *get_user_arg_ptr(struct user_arg_ptr argv, int nr)
401
{
402 403 404 405 406 407 408 409
	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);
410

411 412 413 414 415
		return compat_ptr(compat);
	}
#endif

	if (get_user(native, argv.ptr.native + nr))
416 417
		return ERR_PTR(-EFAULT);

418
	return native;
419 420
}

L
Linus Torvalds 已提交
421 422 423
/*
 * count() counts the number of strings in array ARGV.
 */
424
static int count(struct user_arg_ptr argv, int max)
L
Linus Torvalds 已提交
425 426 427
{
	int i = 0;

428
	if (argv.ptr.native != NULL) {
L
Linus Torvalds 已提交
429
		for (;;) {
430
			const char __user *p = get_user_arg_ptr(argv, i);
L
Linus Torvalds 已提交
431 432 433

			if (!p)
				break;
434 435 436 437

			if (IS_ERR(p))
				return -EFAULT;

438
			if (i >= max)
L
Linus Torvalds 已提交
439
				return -E2BIG;
440
			++i;
441 442 443

			if (fatal_signal_pending(current))
				return -ERESTARTNOHAND;
L
Linus Torvalds 已提交
444 445 446 447 448 449 450
			cond_resched();
		}
	}
	return i;
}

/*
451 452 453
 * '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
454
 */
455
static int copy_strings(int argc, struct user_arg_ptr argv,
A
Adrian Bunk 已提交
456
			struct linux_binprm *bprm)
L
Linus Torvalds 已提交
457 458 459
{
	struct page *kmapped_page = NULL;
	char *kaddr = NULL;
460
	unsigned long kpos = 0;
L
Linus Torvalds 已提交
461 462 463
	int ret;

	while (argc-- > 0) {
464
		const char __user *str;
L
Linus Torvalds 已提交
465 466 467
		int len;
		unsigned long pos;

468 469 470
		ret = -EFAULT;
		str = get_user_arg_ptr(argv, argc);
		if (IS_ERR(str))
L
Linus Torvalds 已提交
471 472
			goto out;

473 474 475 476 477 478
		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
479 480
			goto out;

481
		/* We're going to work our way backwords. */
L
Linus Torvalds 已提交
482
		pos = bprm->p;
483 484
		str += len;
		bprm->p -= len;
L
Linus Torvalds 已提交
485 486 487 488

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

489 490 491 492
			if (fatal_signal_pending(current)) {
				ret = -ERESTARTNOHAND;
				goto out;
			}
493 494
			cond_resched();

L
Linus Torvalds 已提交
495
			offset = pos % PAGE_SIZE;
496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511
			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
512
				if (!page) {
513
					ret = -E2BIG;
L
Linus Torvalds 已提交
514 515 516
					goto out;
				}

517 518
				if (kmapped_page) {
					flush_kernel_dcache_page(kmapped_page);
L
Linus Torvalds 已提交
519
					kunmap(kmapped_page);
520 521
					put_arg_page(kmapped_page);
				}
L
Linus Torvalds 已提交
522 523
				kmapped_page = page;
				kaddr = kmap(kmapped_page);
524 525
				kpos = pos & PAGE_MASK;
				flush_arg_page(bprm, kpos, kmapped_page);
L
Linus Torvalds 已提交
526
			}
527
			if (copy_from_user(kaddr+offset, str, bytes_to_copy)) {
L
Linus Torvalds 已提交
528 529 530 531 532 533 534
				ret = -EFAULT;
				goto out;
			}
		}
	}
	ret = 0;
out:
535 536
	if (kmapped_page) {
		flush_kernel_dcache_page(kmapped_page);
L
Linus Torvalds 已提交
537
		kunmap(kmapped_page);
538 539
		put_arg_page(kmapped_page);
	}
L
Linus Torvalds 已提交
540 541 542 543 544 545
	return ret;
}

/*
 * Like copy_strings, but get argv and its values from kernel memory.
 */
546
int copy_strings_kernel(int argc, const char *const *__argv,
547
			struct linux_binprm *bprm)
L
Linus Torvalds 已提交
548 549 550
{
	int r;
	mm_segment_t oldfs = get_fs();
551
	struct user_arg_ptr argv = {
552
		.ptr.native = (const char __user *const  __user *)__argv,
553 554
	};

L
Linus Torvalds 已提交
555
	set_fs(KERNEL_DS);
556
	r = copy_strings(argc, argv, bprm);
L
Linus Torvalds 已提交
557
	set_fs(oldfs);
558

L
Linus Torvalds 已提交
559 560 561 562 563
	return r;
}
EXPORT_SYMBOL(copy_strings_kernel);

#ifdef CONFIG_MMU
564

L
Linus Torvalds 已提交
565
/*
566 567 568
 * 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
569
 *
570 571 572 573 574 575
 * 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
576
 */
577
static int shift_arg_pages(struct vm_area_struct *vma, unsigned long shift)
L
Linus Torvalds 已提交
578 579
{
	struct mm_struct *mm = vma->vm_mm;
580 581 582 583 584
	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
585
	struct mmu_gather tlb;
L
Linus Torvalds 已提交
586

587
	BUG_ON(new_start > new_end);
L
Linus Torvalds 已提交
588

589 590 591 592 593 594 595 596 597 598
	/*
	 * 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)
	 */
599 600
	if (vma_adjust(vma, new_start, old_end, vma->vm_pgoff, NULL))
		return -ENOMEM;
601 602 603 604 605 606

	/*
	 * 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,
607
				       vma, new_start, length, false))
608 609 610
		return -ENOMEM;

	lru_add_drain();
611
	tlb_gather_mmu(&tlb, mm, old_start, old_end);
612 613 614 615
	if (new_end > old_start) {
		/*
		 * when the old and new regions overlap clear from new_end.
		 */
P
Peter Zijlstra 已提交
616
		free_pgd_range(&tlb, new_end, old_end, new_end,
617
			vma->vm_next ? vma->vm_next->vm_start : USER_PGTABLES_CEILING);
618 619 620 621 622 623 624
	} 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
625
		free_pgd_range(&tlb, old_start, old_end, new_end,
626
			vma->vm_next ? vma->vm_next->vm_start : USER_PGTABLES_CEILING);
L
Linus Torvalds 已提交
627
	}
628
	tlb_finish_mmu(&tlb, old_start, old_end);
629 630

	/*
631
	 * Shrink the vma to just the new range.  Always succeeds.
632 633 634 635
	 */
	vma_adjust(vma, new_start, new_end, vma->vm_pgoff, NULL);

	return 0;
L
Linus Torvalds 已提交
636 637
}

638 639 640 641
/*
 * 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
642 643 644 645
int setup_arg_pages(struct linux_binprm *bprm,
		    unsigned long stack_top,
		    int executable_stack)
{
646 647
	unsigned long ret;
	unsigned long stack_shift;
L
Linus Torvalds 已提交
648
	struct mm_struct *mm = current->mm;
649 650 651 652
	struct vm_area_struct *vma = bprm->vma;
	struct vm_area_struct *prev = NULL;
	unsigned long vm_flags;
	unsigned long stack_base;
653 654 655
	unsigned long stack_size;
	unsigned long stack_expand;
	unsigned long rlim_stack;
L
Linus Torvalds 已提交
656 657 658

#ifdef CONFIG_STACK_GROWSUP
	/* Limit stack size to 1GB */
J
Jiri Slaby 已提交
659
	stack_base = rlimit_max(RLIMIT_STACK);
L
Linus Torvalds 已提交
660 661 662
	if (stack_base > (1 << 30))
		stack_base = 1 << 30;

663 664 665
	/* 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
666

667
	stack_base = PAGE_ALIGN(stack_top - stack_base);
L
Linus Torvalds 已提交
668

669 670 671
	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
672
#else
673 674
	stack_top = arch_align_stack(stack_top);
	stack_top = PAGE_ALIGN(stack_top);
675 676 677 678 679

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

680 681 682
	stack_shift = vma->vm_end - stack_top;

	bprm->p -= stack_shift;
L
Linus Torvalds 已提交
683 684 685 686
	mm->arg_start = bprm->p;
#endif

	if (bprm->loader)
687 688
		bprm->loader -= stack_shift;
	bprm->exec -= stack_shift;
L
Linus Torvalds 已提交
689 690

	down_write(&mm->mmap_sem);
691
	vm_flags = VM_STACK_FLAGS;
692 693 694 695 696 697 698 699 700 701 702

	/*
	 * 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;
703
	vm_flags |= VM_STACK_INCOMPLETE_SETUP;
704 705 706 707 708 709 710 711 712 713

	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);
714 715
		if (ret)
			goto out_unlock;
L
Linus Torvalds 已提交
716 717
	}

718 719 720
	/* mprotect_fixup is overkill to remove the temporary stack flags */
	vma->vm_flags &= ~VM_STACK_INCOMPLETE_SETUP;

721
	stack_expand = 131072UL; /* randomly 32*4k (or 2*64k) pages */
722 723 724 725 726 727
	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;
728
#ifdef CONFIG_STACK_GROWSUP
729 730 731 732
	if (stack_size + stack_expand > rlim_stack)
		stack_base = vma->vm_start + rlim_stack;
	else
		stack_base = vma->vm_end + stack_expand;
733
#else
734 735 736 737
	if (stack_size + stack_expand > rlim_stack)
		stack_base = vma->vm_end - rlim_stack;
	else
		stack_base = vma->vm_start - stack_expand;
738
#endif
739
	current->mm->start_stack = bprm->p;
740 741 742 743 744
	ret = expand_stack(vma, stack_base);
	if (ret)
		ret = -EFAULT;

out_unlock:
L
Linus Torvalds 已提交
745
	up_write(&mm->mmap_sem);
746
	return ret;
L
Linus Torvalds 已提交
747 748 749 750 751 752 753 754
}
EXPORT_SYMBOL(setup_arg_pages);

#endif /* CONFIG_MMU */

struct file *open_exec(const char *name)
{
	struct file *file;
755
	int err;
756
	struct filename tmp = { .name = name };
757 758 759
	static const struct open_flags open_exec_flags = {
		.open_flag = O_LARGEFILE | O_RDONLY | __FMODE_EXEC,
		.acc_mode = MAY_EXEC | MAY_OPEN,
760 761
		.intent = LOOKUP_OPEN,
		.lookup_flags = LOOKUP_FOLLOW,
762
	};
L
Linus Torvalds 已提交
763

764
	file = do_filp_open(AT_FDCWD, &tmp, &open_exec_flags);
765
	if (IS_ERR(file))
766 767 768
		goto out;

	err = -EACCES;
A
Al Viro 已提交
769
	if (!S_ISREG(file_inode(file)->i_mode))
770
		goto exit;
771

772 773
	if (file->f_path.mnt->mnt_flags & MNT_NOEXEC)
		goto exit;
774

775
	fsnotify_open(file);
776

777
	err = deny_write_access(file);
778 779
	if (err)
		goto exit;
L
Linus Torvalds 已提交
780

781
out:
782 783
	return file;

784 785
exit:
	fput(file);
786 787
	return ERR_PTR(err);
}
L
Linus Torvalds 已提交
788 789
EXPORT_SYMBOL(open_exec);

M
Mimi Zohar 已提交
790 791
int kernel_read(struct file *file, loff_t offset,
		char *addr, unsigned long count)
L
Linus Torvalds 已提交
792 793 794 795 796 797 798 799 800 801 802 803 804 805 806
{
	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);

A
Al Viro 已提交
807 808 809 810 811 812 813 814 815
ssize_t read_code(struct file *file, unsigned long addr, loff_t pos, size_t len)
{
	ssize_t res = file->f_op->read(file, (void __user *)addr, len, &pos);
	if (res > 0)
		flush_icache_range(addr, addr + len);
	return res;
}
EXPORT_SYMBOL(read_code);

L
Linus Torvalds 已提交
816 817 818 819 820 821 822 823 824 825 826
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;
	mm_release(tsk, old_mm);

	if (old_mm) {
827
		sync_mm_rss(old_mm);
L
Linus Torvalds 已提交
828 829 830 831
		/*
		 * 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
832
		 * checking core_state and changing tsk->mm.
L
Linus Torvalds 已提交
833 834
		 */
		down_read(&old_mm->mmap_sem);
835
		if (unlikely(old_mm->core_state)) {
L
Linus Torvalds 已提交
836 837 838 839 840 841 842 843 844 845 846 847 848
			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);
	task_unlock(tsk);
	arch_pick_mmap_layout(mm);
	if (old_mm) {
		up_read(&old_mm->mmap_sem);
849
		BUG_ON(active_mm != old_mm);
850
		setmax_mm_hiwater_rss(&tsk->signal->maxrss, old_mm);
851
		mm_update_next_owner(old_mm);
L
Linus Torvalds 已提交
852 853 854 855 856 857 858 859 860 861 862 863 864
		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().)
 */
865
static int de_thread(struct task_struct *tsk)
L
Linus Torvalds 已提交
866 867
{
	struct signal_struct *sig = tsk->signal;
868
	struct sighand_struct *oldsighand = tsk->sighand;
L
Linus Torvalds 已提交
869 870
	spinlock_t *lock = &oldsighand->siglock;

871
	if (thread_group_empty(tsk))
L
Linus Torvalds 已提交
872 873 874 875 876 877
		goto no_thread_group;

	/*
	 * Kill all other threads in the thread group.
	 */
	spin_lock_irq(lock);
878
	if (signal_group_exit(sig)) {
L
Linus Torvalds 已提交
879 880 881 882 883 884 885
		/*
		 * Another group action in progress, just
		 * return so that the signal is processed.
		 */
		spin_unlock_irq(lock);
		return -EAGAIN;
	}
886

887
	sig->group_exit_task = tsk;
888 889 890
	sig->notify_count = zap_other_threads(tsk);
	if (!thread_group_leader(tsk))
		sig->notify_count--;
L
Linus Torvalds 已提交
891

892
	while (sig->notify_count) {
O
Oleg Nesterov 已提交
893
		__set_current_state(TASK_KILLABLE);
L
Linus Torvalds 已提交
894 895
		spin_unlock_irq(lock);
		schedule();
O
Oleg Nesterov 已提交
896 897
		if (unlikely(__fatal_signal_pending(tsk)))
			goto killed;
L
Linus Torvalds 已提交
898 899 900 901 902 903 904 905 906
		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:
	 */
907
	if (!thread_group_leader(tsk)) {
908
		struct task_struct *leader = tsk->group_leader;
909

910
		sig->notify_count = -1;	/* for exit_notify() */
911
		for (;;) {
912
			threadgroup_change_begin(tsk);
913 914 915
			write_lock_irq(&tasklist_lock);
			if (likely(leader->exit_state))
				break;
O
Oleg Nesterov 已提交
916
			__set_current_state(TASK_KILLABLE);
917
			write_unlock_irq(&tasklist_lock);
918
			threadgroup_change_end(tsk);
919
			schedule();
O
Oleg Nesterov 已提交
920 921
			if (unlikely(__fatal_signal_pending(tsk)))
				goto killed;
922
		}
L
Linus Torvalds 已提交
923

924 925 926 927 928 929 930 931 932 933
		/*
		 * 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).
		 */
934
		tsk->start_time = leader->start_time;
935
		tsk->real_start_time = leader->real_start_time;
936

937 938
		BUG_ON(!same_thread_group(leader, tsk));
		BUG_ON(has_group_leader_pid(tsk));
L
Linus Torvalds 已提交
939 940 941 942 943 944
		/*
		 * 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:
		 */
945 946

		/* Become a process group leader with the old leader's pid.
947 948
		 * The old leader becomes a thread of the this thread group.
		 * Note: The old leader also uses this pid until release_task
949 950
		 *       is called.  Odd but simple and correct.
		 */
951
		tsk->pid = leader->pid;
952
		change_pid(tsk, PIDTYPE_PID, task_pid(leader));
953 954
		transfer_pid(leader, tsk, PIDTYPE_PGID);
		transfer_pid(leader, tsk, PIDTYPE_SID);
955

956
		list_replace_rcu(&leader->tasks, &tsk->tasks);
957
		list_replace_init(&leader->sibling, &tsk->sibling);
L
Linus Torvalds 已提交
958

959 960
		tsk->group_leader = tsk;
		leader->group_leader = tsk;
961

962
		tsk->exit_signal = SIGCHLD;
963
		leader->exit_signal = -1;
964 965 966

		BUG_ON(leader->exit_state != EXIT_ZOMBIE);
		leader->exit_state = EXIT_DEAD;
967 968 969 970 971 972 973 974

		/*
		 * We are going to release_task()->ptrace_unlink() silently,
		 * the tracer can sleep in do_wait(). EXIT_DEAD guarantees
		 * the tracer wont't block again waiting for this thread.
		 */
		if (unlikely(leader->ptrace))
			__wake_up_parent(leader, leader->parent);
L
Linus Torvalds 已提交
975
		write_unlock_irq(&tasklist_lock);
976
		threadgroup_change_end(tsk);
977 978

		release_task(leader);
979
	}
L
Linus Torvalds 已提交
980

981 982
	sig->group_exit_task = NULL;
	sig->notify_count = 0;
L
Linus Torvalds 已提交
983 984

no_thread_group:
985 986 987
	/* we have changed execution domain */
	tsk->exit_signal = SIGCHLD;

L
Linus Torvalds 已提交
988
	exit_itimers(sig);
989
	flush_itimer_signals();
990

991 992
	if (atomic_read(&oldsighand->count) != 1) {
		struct sighand_struct *newsighand;
L
Linus Torvalds 已提交
993
		/*
994 995
		 * This ->sighand is shared with the CLONE_SIGHAND
		 * but not CLONE_THREAD task, switch to the new one.
L
Linus Torvalds 已提交
996
		 */
997 998 999 1000
		newsighand = kmem_cache_alloc(sighand_cachep, GFP_KERNEL);
		if (!newsighand)
			return -ENOMEM;

L
Linus Torvalds 已提交
1001 1002 1003 1004 1005 1006
		atomic_set(&newsighand->count, 1);
		memcpy(newsighand->action, oldsighand->action,
		       sizeof(newsighand->action));

		write_lock_irq(&tasklist_lock);
		spin_lock(&oldsighand->siglock);
1007
		rcu_assign_pointer(tsk->sighand, newsighand);
L
Linus Torvalds 已提交
1008 1009 1010
		spin_unlock(&oldsighand->siglock);
		write_unlock_irq(&tasklist_lock);

1011
		__cleanup_sighand(oldsighand);
L
Linus Torvalds 已提交
1012 1013
	}

1014
	BUG_ON(!thread_group_leader(tsk));
L
Linus Torvalds 已提交
1015
	return 0;
O
Oleg Nesterov 已提交
1016 1017 1018 1019 1020 1021 1022 1023

killed:
	/* protects against exit_notify() and __exit_signal() */
	read_lock(&tasklist_lock);
	sig->group_exit_task = NULL;
	sig->notify_count = 0;
	read_unlock(&tasklist_lock);
	return -EAGAIN;
L
Linus Torvalds 已提交
1024
}
O
Oleg Nesterov 已提交
1025

1026
char *get_task_comm(char *buf, struct task_struct *tsk)
L
Linus Torvalds 已提交
1027 1028 1029 1030 1031
{
	/* buf must be at least sizeof(tsk->comm) in size */
	task_lock(tsk);
	strncpy(buf, tsk->comm, sizeof(tsk->comm));
	task_unlock(tsk);
1032
	return buf;
L
Linus Torvalds 已提交
1033
}
1034
EXPORT_SYMBOL_GPL(get_task_comm);
L
Linus Torvalds 已提交
1035

1036 1037 1038 1039 1040
/*
 * These functions flushes out all traces of the currently running executable
 * so that a new one can be started
 */

L
Linus Torvalds 已提交
1041 1042 1043
void set_task_comm(struct task_struct *tsk, char *buf)
{
	task_lock(tsk);
1044
	trace_task_rename(tsk, buf);
L
Linus Torvalds 已提交
1045 1046
	strlcpy(tsk->comm, buf, sizeof(tsk->comm));
	task_unlock(tsk);
1047
	perf_event_comm(tsk);
L
Linus Torvalds 已提交
1048 1049
}

1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064
static void filename_to_taskname(char *tcomm, const char *fn, unsigned int len)
{
	int i, ch;

	/* Copies the binary name from after last slash */
	for (i = 0; (ch = *(fn++)) != '\0';) {
		if (ch == '/')
			i = 0; /* overwrite what we wrote */
		else
			if (i < len - 1)
				tcomm[i++] = ch;
	}
	tcomm[i] = '\0';
}

L
Linus Torvalds 已提交
1065 1066
int flush_old_exec(struct linux_binprm * bprm)
{
1067
	int retval;
L
Linus Torvalds 已提交
1068 1069 1070 1071 1072 1073 1074 1075 1076

	/*
	 * 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
1077 1078
	set_mm_exe_file(bprm->mm, bprm->file);

1079
	filename_to_taskname(bprm->tcomm, bprm->filename, sizeof(bprm->tcomm));
L
Linus Torvalds 已提交
1080 1081 1082
	/*
	 * Release all of the old mmap stuff
	 */
1083
	acct_arg_size(bprm, 0);
L
Linus Torvalds 已提交
1084 1085
	retval = exec_mmap(bprm->mm);
	if (retval)
1086
		goto out;
L
Linus Torvalds 已提交
1087 1088

	bprm->mm = NULL;		/* We're using it now */
1089

1090
	set_fs(USER_DS);
1091 1092
	current->flags &=
		~(PF_RANDOMIZE | PF_FORKNOEXEC | PF_KTHREAD | PF_NOFREEZE);
1093 1094 1095
	flush_thread();
	current->personality &= ~bprm->per_clear;

1096 1097 1098 1099 1100 1101 1102
	return 0;

out:
	return retval;
}
EXPORT_SYMBOL(flush_old_exec);

1103 1104
void would_dump(struct linux_binprm *bprm, struct file *file)
{
A
Al Viro 已提交
1105
	if (inode_permission(file_inode(file), MAY_READ) < 0)
1106 1107 1108 1109
		bprm->interp_flags |= BINPRM_FLAGS_ENFORCE_NONDUMP;
}
EXPORT_SYMBOL(would_dump);

1110 1111 1112
void setup_new_exec(struct linux_binprm * bprm)
{
	arch_pick_mmap_layout(current->mm);
L
Linus Torvalds 已提交
1113 1114 1115 1116

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

1117
	if (uid_eq(current_euid(), current_uid()) && gid_eq(current_egid(), current_gid()))
1118
		set_dumpable(current->mm, SUID_DUMP_USER);
A
Alan Cox 已提交
1119
	else
1120
		set_dumpable(current->mm, suid_dumpable);
A
Alan Cox 已提交
1121

1122
	set_task_comm(current, bprm->tcomm);
L
Linus Torvalds 已提交
1123

1124 1125 1126 1127 1128 1129
	/* 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;

1130
	/* install the new credentials */
1131 1132
	if (!uid_eq(bprm->cred->uid, current_euid()) ||
	    !gid_eq(bprm->cred->gid, current_egid())) {
1133
		current->pdeath_signal = 0;
1134 1135 1136 1137
	} else {
		would_dump(bprm, bprm->file);
		if (bprm->interp_flags & BINPRM_FLAGS_ENFORCE_NONDUMP)
			set_dumpable(current->mm, suid_dumpable);
L
Linus Torvalds 已提交
1138 1139 1140 1141 1142 1143 1144 1145
	}

	/* An exec changes our domain. We are no longer part of the thread
	   group */

	current->self_exec_id++;
			
	flush_signal_handlers(current, 0);
1146
	do_close_on_exec(current->files);
L
Linus Torvalds 已提交
1147
}
1148
EXPORT_SYMBOL(setup_new_exec);
L
Linus Torvalds 已提交
1149

1150 1151 1152 1153 1154 1155 1156 1157
/*
 * 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)
{
1158
	if (mutex_lock_interruptible(&current->signal->cred_guard_mutex))
1159 1160 1161 1162 1163 1164
		return -ERESTARTNOINTR;

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

1165
	mutex_unlock(&current->signal->cred_guard_mutex);
1166 1167 1168 1169 1170 1171 1172
	return -ENOMEM;
}

void free_bprm(struct linux_binprm *bprm)
{
	free_arg_pages(bprm);
	if (bprm->cred) {
1173
		mutex_unlock(&current->signal->cred_guard_mutex);
1174 1175
		abort_creds(bprm->cred);
	}
1176 1177 1178
	/* If a binfmt changed the interp, free it. */
	if (bprm->interp != bprm->filename)
		kfree(bprm->interp);
1179 1180 1181
	kfree(bprm);
}

1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193
int bprm_change_interp(char *interp, struct linux_binprm *bprm)
{
	/* If a binfmt changed the interp, free it first. */
	if (bprm->interp != bprm->filename)
		kfree(bprm->interp);
	bprm->interp = kstrdup(interp, GFP_KERNEL);
	if (!bprm->interp)
		return -ENOMEM;
	return 0;
}
EXPORT_SYMBOL(bprm_change_interp);

1194 1195 1196 1197 1198 1199 1200 1201 1202
/*
 * 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;
1203 1204 1205 1206 1207 1208 1209 1210 1211

	/*
	 * Disable monitoring for regular users
	 * when executing setuid binaries. Must
	 * wait until new credentials are committed
	 * by commit_creds() above
	 */
	if (get_dumpable(current->mm) != SUID_DUMP_USER)
		perf_event_exit_task(current);
1212 1213
	/*
	 * cred_guard_mutex must be held at least to this point to prevent
1214
	 * ptrace_attach() from altering our determination of the task's
1215 1216
	 * credentials; any time after this it may be unlocked.
	 */
1217
	security_bprm_committed_creds(bprm);
1218
	mutex_unlock(&current->signal->cred_guard_mutex);
1219 1220 1221 1222 1223
}
EXPORT_SYMBOL(install_exec_creds);

/*
 * determine how safe it is to execute the proposed program
1224
 * - the caller must hold ->cred_guard_mutex to protect against
1225 1226
 *   PTRACE_ATTACH
 */
A
Al Viro 已提交
1227
static int check_unsafe_exec(struct linux_binprm *bprm)
1228
{
D
David Howells 已提交
1229
	struct task_struct *p = current, *t;
1230
	unsigned n_fs;
A
Al Viro 已提交
1231
	int res = 0;
1232

T
Tejun Heo 已提交
1233 1234 1235 1236 1237 1238
	if (p->ptrace) {
		if (p->ptrace & PT_PTRACE_CAP)
			bprm->unsafe |= LSM_UNSAFE_PTRACE_CAP;
		else
			bprm->unsafe |= LSM_UNSAFE_PTRACE;
	}
1239

1240 1241 1242 1243 1244 1245 1246
	/*
	 * This isn't strictly necessary, but it makes it harder for LSMs to
	 * mess up.
	 */
	if (current->no_new_privs)
		bprm->unsafe |= LSM_UNSAFE_NO_NEW_PRIVS;

D
David Howells 已提交
1247
	n_fs = 1;
N
Nick Piggin 已提交
1248
	spin_lock(&p->fs->lock);
1249
	rcu_read_lock();
D
David Howells 已提交
1250 1251 1252 1253
	for (t = next_thread(p); t != p; t = next_thread(t)) {
		if (t->fs == p->fs)
			n_fs++;
	}
1254
	rcu_read_unlock();
D
David Howells 已提交
1255

1256
	if (p->fs->users > n_fs) {
1257
		bprm->unsafe |= LSM_UNSAFE_SHARE;
A
Al Viro 已提交
1258
	} else {
1259 1260 1261 1262 1263
		res = -EAGAIN;
		if (!p->fs->in_exec) {
			p->fs->in_exec = 1;
			res = 1;
		}
A
Al Viro 已提交
1264
	}
N
Nick Piggin 已提交
1265
	spin_unlock(&p->fs->lock);
A
Al Viro 已提交
1266 1267

	return res;
1268 1269
}

L
Linus Torvalds 已提交
1270 1271 1272
/* 
 * Fill the binprm structure from the inode. 
 * Check permissions, then read the first 128 (BINPRM_BUF_SIZE) bytes
1273 1274
 *
 * This may be called multiple times for binary chains (scripts for example).
L
Linus Torvalds 已提交
1275 1276 1277
 */
int prepare_binprm(struct linux_binprm *bprm)
{
A
Al Viro 已提交
1278 1279
	struct inode *inode = file_inode(bprm->file);
	umode_t mode = inode->i_mode;
L
Linus Torvalds 已提交
1280 1281 1282
	int retval;


1283 1284 1285
	/* 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
1286

1287
	if (!(bprm->file->f_path.mnt->mnt_flags & MNT_NOSUID) &&
1288 1289 1290
	    !current->no_new_privs &&
	    kuid_has_mapping(bprm->cred->user_ns, inode->i_uid) &&
	    kgid_has_mapping(bprm->cred->user_ns, inode->i_gid)) {
L
Linus Torvalds 已提交
1291 1292
		/* Set-uid? */
		if (mode & S_ISUID) {
1293 1294
			bprm->per_clear |= PER_CLEAR_ON_SETID;
			bprm->cred->euid = inode->i_uid;
L
Linus Torvalds 已提交
1295 1296 1297 1298 1299 1300 1301 1302 1303
		}

		/* 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)) {
1304 1305
			bprm->per_clear |= PER_CLEAR_ON_SETID;
			bprm->cred->egid = inode->i_gid;
L
Linus Torvalds 已提交
1306 1307 1308 1309
		}
	}

	/* fill in binprm security blob */
1310
	retval = security_bprm_set_creds(bprm);
L
Linus Torvalds 已提交
1311 1312
	if (retval)
		return retval;
1313
	bprm->cred_prepared = 1;
L
Linus Torvalds 已提交
1314

1315 1316
	memset(bprm->buf, 0, BINPRM_BUF_SIZE);
	return kernel_read(bprm->file, 0, bprm->buf, BINPRM_BUF_SIZE);
L
Linus Torvalds 已提交
1317 1318 1319 1320
}

EXPORT_SYMBOL(prepare_binprm);

N
Nick Piggin 已提交
1321 1322 1323 1324 1325
/*
 * 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.
 */
1326
int remove_arg_zero(struct linux_binprm *bprm)
L
Linus Torvalds 已提交
1327
{
1328 1329 1330 1331
	int ret = 0;
	unsigned long offset;
	char *kaddr;
	struct page *page;
N
Nick Piggin 已提交
1332

1333 1334
	if (!bprm->argc)
		return 0;
L
Linus Torvalds 已提交
1335

1336 1337 1338 1339 1340 1341 1342
	do {
		offset = bprm->p & ~PAGE_MASK;
		page = get_arg_page(bprm, bprm->p, 0);
		if (!page) {
			ret = -EFAULT;
			goto out;
		}
1343
		kaddr = kmap_atomic(page);
N
Nick Piggin 已提交
1344

1345 1346 1347
		for (; offset < PAGE_SIZE && kaddr[offset];
				offset++, bprm->p++)
			;
N
Nick Piggin 已提交
1348

1349
		kunmap_atomic(kaddr);
1350
		put_arg_page(page);
N
Nick Piggin 已提交
1351

1352 1353 1354
		if (offset == PAGE_SIZE)
			free_arg_page(bprm, (bprm->p >> PAGE_SHIFT) - 1);
	} while (offset == PAGE_SIZE);
N
Nick Piggin 已提交
1355

1356 1357 1358
	bprm->p++;
	bprm->argc--;
	ret = 0;
N
Nick Piggin 已提交
1359

1360 1361
out:
	return ret;
L
Linus Torvalds 已提交
1362 1363 1364
}
EXPORT_SYMBOL(remove_arg_zero);

1365
#define printable(c) (((c)=='\t') || ((c)=='\n') || (0x20<=(c) && (c)<=0x7e))
L
Linus Torvalds 已提交
1366 1367 1368
/*
 * cycle the list of binary formats handler, until one recognizes the image
 */
1369
int search_binary_handler(struct linux_binprm *bprm)
L
Linus Torvalds 已提交
1370
{
1371
	bool need_retry = IS_ENABLED(CONFIG_MODULES);
L
Linus Torvalds 已提交
1372
	struct linux_binfmt *fmt;
1373
	int retval;
L
Linus Torvalds 已提交
1374

1375
	/* This allows 4 levels of binfmt rewrites before failing hard. */
1376
	if (bprm->recursion_depth > 5)
1377 1378
		return -ELOOP;

L
Linus Torvalds 已提交
1379 1380 1381 1382 1383
	retval = security_bprm_check(bprm);
	if (retval)
		return retval;

	retval = -ENOENT;
1384 1385 1386 1387 1388 1389 1390 1391 1392
 retry:
	read_lock(&binfmt_lock);
	list_for_each_entry(fmt, &formats, lh) {
		if (!try_module_get(fmt->module))
			continue;
		read_unlock(&binfmt_lock);
		bprm->recursion_depth++;
		retval = fmt->load_binary(bprm);
		bprm->recursion_depth--;
1393 1394
		if (retval >= 0 || retval != -ENOEXEC ||
		    bprm->mm == NULL || bprm->file == NULL) {
L
Linus Torvalds 已提交
1395
			put_binfmt(fmt);
1396
			return retval;
L
Linus Torvalds 已提交
1397
		}
1398 1399
		read_lock(&binfmt_lock);
		put_binfmt(fmt);
L
Linus Torvalds 已提交
1400
	}
1401 1402
	read_unlock(&binfmt_lock);

1403
	if (need_retry && retval == -ENOEXEC) {
1404 1405 1406
		if (printable(bprm->buf[0]) && printable(bprm->buf[1]) &&
		    printable(bprm->buf[2]) && printable(bprm->buf[3]))
			return retval;
1407 1408
		if (request_module("binfmt-%04x", *(ushort *)(bprm->buf + 2)) < 0)
			return retval;
1409 1410 1411 1412
		need_retry = false;
		goto retry;
	}

L
Linus Torvalds 已提交
1413 1414 1415 1416
	return retval;
}
EXPORT_SYMBOL(search_binary_handler);

1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429
static int exec_binprm(struct linux_binprm *bprm)
{
	pid_t old_pid, old_vpid;
	int ret;

	/* Need to fetch pid before load_binary changes it */
	old_pid = current->pid;
	rcu_read_lock();
	old_vpid = task_pid_nr_ns(current, task_active_pid_ns(current->parent));
	rcu_read_unlock();

	ret = search_binary_handler(bprm);
	if (ret >= 0) {
1430
		audit_bprm(bprm);
1431 1432 1433
		trace_sched_process_exec(current, old_pid, bprm);
		ptrace_event(PTRACE_EVENT_EXEC, old_vpid);
		current->did_exec = 1;
1434
		proc_exec_connector(current);
1435 1436 1437 1438 1439 1440

		if (bprm->file) {
			allow_write_access(bprm->file);
			fput(bprm->file);
			bprm->file = NULL; /* to catch use-after-free */
		}
1441 1442 1443 1444 1445
	}

	return ret;
}

L
Linus Torvalds 已提交
1446 1447 1448
/*
 * sys_execve() executes a new program.
 */
1449 1450
static int do_execve_common(const char *filename,
				struct user_arg_ptr argv,
1451
				struct user_arg_ptr envp)
L
Linus Torvalds 已提交
1452 1453 1454
{
	struct linux_binprm *bprm;
	struct file *file;
1455
	struct files_struct *displaced;
1456
	bool clear_in_exec;
L
Linus Torvalds 已提交
1457
	int retval;
1458 1459 1460 1461 1462 1463 1464 1465

	/*
	 * We move the actual failure in case of RLIMIT_NPROC excess from
	 * set*uid() to execve() because too many poorly written programs
	 * don't check setuid() return code.  Here we additionally recheck
	 * whether NPROC limit is still exceeded.
	 */
	if ((current->flags & PF_NPROC_EXCEEDED) &&
1466
	    atomic_read(&current_user()->processes) > rlimit(RLIMIT_NPROC)) {
1467 1468 1469 1470 1471 1472 1473
		retval = -EAGAIN;
		goto out_ret;
	}

	/* We're below the limit (still or again), so we don't want to make
	 * further execve() calls fail. */
	current->flags &= ~PF_NPROC_EXCEEDED;
L
Linus Torvalds 已提交
1474

1475
	retval = unshare_files(&displaced);
1476 1477 1478
	if (retval)
		goto out_ret;

L
Linus Torvalds 已提交
1479
	retval = -ENOMEM;
1480
	bprm = kzalloc(sizeof(*bprm), GFP_KERNEL);
L
Linus Torvalds 已提交
1481
	if (!bprm)
1482
		goto out_files;
L
Linus Torvalds 已提交
1483

1484 1485
	retval = prepare_bprm_creds(bprm);
	if (retval)
1486
		goto out_free;
A
Al Viro 已提交
1487 1488

	retval = check_unsafe_exec(bprm);
1489
	if (retval < 0)
1490
		goto out_free;
1491
	clear_in_exec = retval;
1492
	current->in_execve = 1;
1493

L
Linus Torvalds 已提交
1494 1495 1496
	file = open_exec(filename);
	retval = PTR_ERR(file);
	if (IS_ERR(file))
A
Al Viro 已提交
1497
		goto out_unmark;
L
Linus Torvalds 已提交
1498 1499 1500 1501 1502 1503 1504

	sched_exec();

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

1505 1506 1507
	retval = bprm_mm_init(bprm);
	if (retval)
		goto out_file;
L
Linus Torvalds 已提交
1508

1509
	bprm->argc = count(argv, MAX_ARG_STRINGS);
L
Linus Torvalds 已提交
1510
	if ((retval = bprm->argc) < 0)
1511
		goto out;
L
Linus Torvalds 已提交
1512

1513
	bprm->envc = count(envp, MAX_ARG_STRINGS);
L
Linus Torvalds 已提交
1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533
	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;

1534
	retval = exec_binprm(bprm);
1535 1536
	if (retval < 0)
		goto out;
L
Linus Torvalds 已提交
1537

1538
	/* execve succeeded */
A
Al Viro 已提交
1539
	current->fs->in_exec = 0;
1540
	current->in_execve = 0;
1541
	acct_update_integrals(current);
1542
	task_numa_free(current);
1543 1544 1545 1546
	free_bprm(bprm);
	if (displaced)
		put_files_struct(displaced);
	return retval;
L
Linus Torvalds 已提交
1547

1548
out:
1549 1550 1551 1552
	if (bprm->mm) {
		acct_arg_size(bprm, 0);
		mmput(bprm->mm);
	}
L
Linus Torvalds 已提交
1553 1554 1555 1556 1557 1558

out_file:
	if (bprm->file) {
		allow_write_access(bprm->file);
		fput(bprm->file);
	}
1559

A
Al Viro 已提交
1560
out_unmark:
1561 1562
	if (clear_in_exec)
		current->fs->in_exec = 0;
1563
	current->in_execve = 0;
1564 1565

out_free:
1566
	free_bprm(bprm);
L
Linus Torvalds 已提交
1567

1568
out_files:
1569 1570
	if (displaced)
		reset_files_struct(displaced);
L
Linus Torvalds 已提交
1571 1572 1573 1574
out_ret:
	return retval;
}

1575 1576
int do_execve(const char *filename,
	const char __user *const __user *__argv,
1577
	const char __user *const __user *__envp)
1578
{
1579 1580
	struct user_arg_ptr argv = { .ptr.native = __argv };
	struct user_arg_ptr envp = { .ptr.native = __envp };
1581
	return do_execve_common(filename, argv, envp);
1582 1583 1584
}

#ifdef CONFIG_COMPAT
1585
static int compat_do_execve(const char *filename,
A
Al Viro 已提交
1586
	const compat_uptr_t __user *__argv,
1587
	const compat_uptr_t __user *__envp)
1588 1589 1590 1591 1592 1593 1594 1595 1596
{
	struct user_arg_ptr argv = {
		.is_compat = true,
		.ptr.compat = __argv,
	};
	struct user_arg_ptr envp = {
		.is_compat = true,
		.ptr.compat = __envp,
	};
1597
	return do_execve_common(filename, argv, envp);
1598
}
1599
#endif
1600

1601
void set_binfmt(struct linux_binfmt *new)
L
Linus Torvalds 已提交
1602
{
1603 1604 1605 1606
	struct mm_struct *mm = current->mm;

	if (mm->binfmt)
		module_put(mm->binfmt->module);
L
Linus Torvalds 已提交
1607

1608
	mm->binfmt = new;
1609 1610
	if (new)
		__module_get(new->module);
L
Linus Torvalds 已提交
1611 1612 1613 1614
}

EXPORT_SYMBOL(set_binfmt);

1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637
/*
 * 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) {
1638
	case SUID_DUMP_DISABLE:
1639 1640 1641 1642
		clear_bit(MMF_DUMPABLE, &mm->flags);
		smp_wmb();
		clear_bit(MMF_DUMP_SECURELY, &mm->flags);
		break;
1643
	case SUID_DUMP_USER:
1644 1645 1646 1647
		set_bit(MMF_DUMPABLE, &mm->flags);
		smp_wmb();
		clear_bit(MMF_DUMP_SECURELY, &mm->flags);
		break;
1648
	case SUID_DUMP_ROOT:
1649 1650 1651 1652 1653 1654 1655
		set_bit(MMF_DUMP_SECURELY, &mm->flags);
		smp_wmb();
		set_bit(MMF_DUMPABLE, &mm->flags);
		break;
	}
}

1656
int __get_dumpable(unsigned long mm_flags)
1657 1658 1659
{
	int ret;

1660
	ret = mm_flags & MMF_DUMPABLE_MASK;
1661
	return (ret > SUID_DUMP_USER) ? SUID_DUMP_ROOT : ret;
1662 1663
}

1664 1665 1666 1667 1668 1669
/*
 * This returns the actual value of the suid_dumpable flag. For things
 * that are using this for checking for privilege transitions, it must
 * test against SUID_DUMP_USER rather than treating it as a boolean
 * value.
 */
1670 1671 1672 1673 1674
int get_dumpable(struct mm_struct *mm)
{
	return __get_dumpable(mm->flags);
}

A
Al Viro 已提交
1675 1676 1677 1678 1679
SYSCALL_DEFINE3(execve,
		const char __user *, filename,
		const char __user *const __user *, argv,
		const char __user *const __user *, envp)
{
1680
	struct filename *path = getname(filename);
A
Al Viro 已提交
1681 1682
	int error = PTR_ERR(path);
	if (!IS_ERR(path)) {
1683
		error = do_execve(path->name, argv, envp);
A
Al Viro 已提交
1684 1685 1686 1687 1688 1689 1690 1691 1692
		putname(path);
	}
	return error;
}
#ifdef CONFIG_COMPAT
asmlinkage long compat_sys_execve(const char __user * filename,
	const compat_uptr_t __user * argv,
	const compat_uptr_t __user * envp)
{
1693
	struct filename *path = getname(filename);
A
Al Viro 已提交
1694 1695
	int error = PTR_ERR(path);
	if (!IS_ERR(path)) {
1696
		error = compat_do_execve(path->name, argv, envp);
A
Al Viro 已提交
1697 1698 1699 1700 1701
		putname(path);
	}
	return error;
}
#endif