compat.c 37.0 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10
/*
 *  linux/fs/compat.c
 *
 *  Kernel compatibililty routines for e.g. 32 bit syscall support
 *  on 64 bit kernels.
 *
 *  Copyright (C) 2002       Stephen Rothwell, IBM Corporation
 *  Copyright (C) 1997-2000  Jakub Jelinek  (jakub@redhat.com)
 *  Copyright (C) 1998       Eddie C. Dost  (ecd@skynet.be)
 *  Copyright (C) 2001,2002  Andi Kleen, SuSE Labs 
P
Pavel Machek 已提交
11
 *  Copyright (C) 2003       Pavel Machek (pavel@ucw.cz)
L
Linus Torvalds 已提交
12 13 14 15 16 17
 *
 *  This program is free software; you can redistribute it and/or modify
 *  it under the terms of the GNU General Public License version 2 as
 *  published by the Free Software Foundation.
 */

18
#include <linux/stddef.h>
19
#include <linux/kernel.h>
L
Linus Torvalds 已提交
20 21 22 23 24 25 26 27
#include <linux/linkage.h>
#include <linux/compat.h>
#include <linux/errno.h>
#include <linux/time.h>
#include <linux/fs.h>
#include <linux/fcntl.h>
#include <linux/namei.h>
#include <linux/file.h>
A
Al Viro 已提交
28
#include <linux/fdtable.h>
L
Linus Torvalds 已提交
29 30 31 32
#include <linux/vfs.h>
#include <linux/ioctl.h>
#include <linux/init.h>
#include <linux/ncp_mount.h>
33
#include <linux/nfs4_mount.h>
L
Linus Torvalds 已提交
34 35 36
#include <linux/syscalls.h>
#include <linux/ctype.h>
#include <linux/dirent.h>
R
Robert Love 已提交
37
#include <linux/fsnotify.h>
L
Linus Torvalds 已提交
38 39 40
#include <linux/highuid.h>
#include <linux/personality.h>
#include <linux/rwsem.h>
41
#include <linux/tsacct_kern.h>
42
#include <linux/security.h>
A
Al Viro 已提交
43
#include <linux/highmem.h>
44
#include <linux/signal.h>
A
Al Viro 已提交
45
#include <linux/poll.h>
46
#include <linux/mm.h>
A
Al Viro 已提交
47
#include <linux/fs_struct.h>
48
#include <linux/slab.h>
W
wu zhangjin 已提交
49
#include <linux/pagemap.h>
50
#include <linux/aio.h>
L
Linus Torvalds 已提交
51 52 53 54

#include <asm/uaccess.h>
#include <asm/mmu_context.h>
#include <asm/ioctls.h>
55
#include "internal.h"
56

57 58 59 60 61 62 63 64 65 66 67 68 69 70
int compat_log = 1;

int compat_printk(const char *fmt, ...)
{
	va_list ap;
	int ret;
	if (!compat_log)
		return 0;
	va_start(ap, fmt);
	ret = vprintk(fmt, ap);
	va_end(ap);
	return ret;
}

L
Linus Torvalds 已提交
71 72 73 74
/*
 * Not all architectures have sys_utime, so implement this in terms
 * of sys_utimes.
 */
75 76
COMPAT_SYSCALL_DEFINE2(utime, const char __user *, filename,
		       struct compat_utimbuf __user *, t)
L
Linus Torvalds 已提交
77
{
U
Ulrich Drepper 已提交
78
	struct timespec tv[2];
L
Linus Torvalds 已提交
79 80 81 82 83

	if (t) {
		if (get_user(tv[0].tv_sec, &t->actime) ||
		    get_user(tv[1].tv_sec, &t->modtime))
			return -EFAULT;
U
Ulrich Drepper 已提交
84 85
		tv[0].tv_nsec = 0;
		tv[1].tv_nsec = 0;
L
Linus Torvalds 已提交
86
	}
U
Ulrich Drepper 已提交
87 88 89
	return do_utimes(AT_FDCWD, filename, t ? tv : NULL, 0);
}

90
COMPAT_SYSCALL_DEFINE4(utimensat, unsigned int, dfd, const char __user *, filename, struct compat_timespec __user *, t, int, flags)
U
Ulrich Drepper 已提交
91 92 93 94
{
	struct timespec tv[2];

	if  (t) {
95 96
		if (compat_get_timespec(&tv[0], &t[0]) ||
		    compat_get_timespec(&tv[1], &t[1]))
U
Ulrich Drepper 已提交
97 98 99 100 101 102
			return -EFAULT;

		if (tv[0].tv_nsec == UTIME_OMIT && tv[1].tv_nsec == UTIME_OMIT)
			return 0;
	}
	return do_utimes(dfd, filename, t ? tv : NULL, flags);
L
Linus Torvalds 已提交
103 104
}

105
COMPAT_SYSCALL_DEFINE3(futimesat, unsigned int, dfd, const char __user *, filename, struct compat_timeval __user *, t)
L
Linus Torvalds 已提交
106
{
U
Ulrich Drepper 已提交
107
	struct timespec tv[2];
L
Linus Torvalds 已提交
108

109
	if (t) {
L
Linus Torvalds 已提交
110
		if (get_user(tv[0].tv_sec, &t[0].tv_sec) ||
U
Ulrich Drepper 已提交
111
		    get_user(tv[0].tv_nsec, &t[0].tv_usec) ||
L
Linus Torvalds 已提交
112
		    get_user(tv[1].tv_sec, &t[1].tv_sec) ||
U
Ulrich Drepper 已提交
113
		    get_user(tv[1].tv_nsec, &t[1].tv_usec))
114
			return -EFAULT;
U
Ulrich Drepper 已提交
115 116 117 118 119
		if (tv[0].tv_nsec >= 1000000 || tv[0].tv_nsec < 0 ||
		    tv[1].tv_nsec >= 1000000 || tv[1].tv_nsec < 0)
			return -EINVAL;
		tv[0].tv_nsec *= 1000;
		tv[1].tv_nsec *= 1000;
120
	}
U
Ulrich Drepper 已提交
121
	return do_utimes(dfd, filename, t ? tv : NULL, 0);
122 123
}

124
COMPAT_SYSCALL_DEFINE2(utimes, const char __user *, filename, struct compat_timeval __user *, t)
125 126
{
	return compat_sys_futimesat(AT_FDCWD, filename, t);
L
Linus Torvalds 已提交
127 128
}

129 130
static int cp_compat_stat(struct kstat *stat, struct compat_stat __user *ubuf)
{
131
	struct compat_stat tmp;
132

133 134
	if (!old_valid_dev(stat->dev) || !old_valid_dev(stat->rdev))
		return -EOVERFLOW;
135

136 137 138 139
	memset(&tmp, 0, sizeof(tmp));
	tmp.st_dev = old_encode_dev(stat->dev);
	tmp.st_ino = stat->ino;
	if (sizeof(tmp.st_ino) < sizeof(stat->ino) && tmp.st_ino != stat->ino)
140
		return -EOVERFLOW;
141 142 143
	tmp.st_mode = stat->mode;
	tmp.st_nlink = stat->nlink;
	if (tmp.st_nlink != stat->nlink)
144
		return -EOVERFLOW;
145 146
	SET_UID(tmp.st_uid, from_kuid_munged(current_user_ns(), stat->uid));
	SET_GID(tmp.st_gid, from_kgid_munged(current_user_ns(), stat->gid));
147 148 149 150 151 152 153 154 155 156 157 158 159
	tmp.st_rdev = old_encode_dev(stat->rdev);
	if ((u64) stat->size > MAX_NON_LFS)
		return -EOVERFLOW;
	tmp.st_size = stat->size;
	tmp.st_atime = stat->atime.tv_sec;
	tmp.st_atime_nsec = stat->atime.tv_nsec;
	tmp.st_mtime = stat->mtime.tv_sec;
	tmp.st_mtime_nsec = stat->mtime.tv_nsec;
	tmp.st_ctime = stat->ctime.tv_sec;
	tmp.st_ctime_nsec = stat->ctime.tv_nsec;
	tmp.st_blocks = stat->blocks;
	tmp.st_blksize = stat->blksize;
	return copy_to_user(ubuf, &tmp, sizeof(tmp)) ? -EFAULT : 0;
160 161
}

162 163
COMPAT_SYSCALL_DEFINE2(newstat, const char __user *, filename,
		       struct compat_stat __user *, statbuf)
L
Linus Torvalds 已提交
164 165
{
	struct kstat stat;
166
	int error;
L
Linus Torvalds 已提交
167

168 169 170 171
	error = vfs_stat(filename, &stat);
	if (error)
		return error;
	return cp_compat_stat(&stat, statbuf);
L
Linus Torvalds 已提交
172 173
}

174 175
COMPAT_SYSCALL_DEFINE2(newlstat, const char __user *, filename,
		       struct compat_stat __user *, statbuf)
L
Linus Torvalds 已提交
176 177
{
	struct kstat stat;
178
	int error;
L
Linus Torvalds 已提交
179

180 181 182 183
	error = vfs_lstat(filename, &stat);
	if (error)
		return error;
	return cp_compat_stat(&stat, statbuf);
L
Linus Torvalds 已提交
184 185
}

186
#ifndef __ARCH_WANT_STAT64
187 188 189
COMPAT_SYSCALL_DEFINE4(newfstatat, unsigned int, dfd,
		       const char __user *, filename,
		       struct compat_stat __user *, statbuf, int, flag)
190 191
{
	struct kstat stat;
192
	int error;
193

194 195 196 197
	error = vfs_fstatat(dfd, filename, &stat, flag);
	if (error)
		return error;
	return cp_compat_stat(&stat, statbuf);
198
}
199
#endif
200

201 202
COMPAT_SYSCALL_DEFINE2(newfstat, unsigned int, fd,
		       struct compat_stat __user *, statbuf)
L
Linus Torvalds 已提交
203 204 205 206 207 208 209 210 211 212 213 214 215
{
	struct kstat stat;
	int error = vfs_fstat(fd, &stat);

	if (!error)
		error = cp_compat_stat(&stat, statbuf);
	return error;
}

static int put_compat_statfs(struct compat_statfs __user *ubuf, struct kstatfs *kbuf)
{
	
	if (sizeof ubuf->f_blocks == 4) {
J
Jon Tollefson 已提交
216 217
		if ((kbuf->f_blocks | kbuf->f_bfree | kbuf->f_bavail |
		     kbuf->f_bsize | kbuf->f_frsize) & 0xffffffff00000000ULL)
L
Linus Torvalds 已提交
218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239
			return -EOVERFLOW;
		/* f_files and f_ffree may be -1; it's okay
		 * to stuff that into 32 bits */
		if (kbuf->f_files != 0xffffffffffffffffULL
		 && (kbuf->f_files & 0xffffffff00000000ULL))
			return -EOVERFLOW;
		if (kbuf->f_ffree != 0xffffffffffffffffULL
		 && (kbuf->f_ffree & 0xffffffff00000000ULL))
			return -EOVERFLOW;
	}
	if (!access_ok(VERIFY_WRITE, ubuf, sizeof(*ubuf)) ||
	    __put_user(kbuf->f_type, &ubuf->f_type) ||
	    __put_user(kbuf->f_bsize, &ubuf->f_bsize) ||
	    __put_user(kbuf->f_blocks, &ubuf->f_blocks) ||
	    __put_user(kbuf->f_bfree, &ubuf->f_bfree) ||
	    __put_user(kbuf->f_bavail, &ubuf->f_bavail) ||
	    __put_user(kbuf->f_files, &ubuf->f_files) ||
	    __put_user(kbuf->f_ffree, &ubuf->f_ffree) ||
	    __put_user(kbuf->f_namelen, &ubuf->f_namelen) ||
	    __put_user(kbuf->f_fsid.val[0], &ubuf->f_fsid.val[0]) ||
	    __put_user(kbuf->f_fsid.val[1], &ubuf->f_fsid.val[1]) ||
	    __put_user(kbuf->f_frsize, &ubuf->f_frsize) ||
240 241
	    __put_user(kbuf->f_flags, &ubuf->f_flags) ||
	    __clear_user(ubuf->f_spare, sizeof(ubuf->f_spare)))
L
Linus Torvalds 已提交
242 243 244 245 246
		return -EFAULT;
	return 0;
}

/*
247
 * The following statfs calls are copies of code from fs/statfs.c and
L
Linus Torvalds 已提交
248 249
 * should be checked against those from time to time
 */
250
COMPAT_SYSCALL_DEFINE2(statfs, const char __user *, pathname, struct compat_statfs __user *, buf)
L
Linus Torvalds 已提交
251
{
A
Al Viro 已提交
252 253 254 255
	struct kstatfs tmp;
	int error = user_statfs(pathname, &tmp);
	if (!error)
		error = put_compat_statfs(buf, &tmp);
L
Linus Torvalds 已提交
256 257 258
	return error;
}

259
COMPAT_SYSCALL_DEFINE2(fstatfs, unsigned int, fd, struct compat_statfs __user *, buf)
L
Linus Torvalds 已提交
260 261
{
	struct kstatfs tmp;
A
Al Viro 已提交
262
	int error = fd_statfs(fd, &tmp);
263 264
	if (!error)
		error = put_compat_statfs(buf, &tmp);
L
Linus Torvalds 已提交
265 266 267 268 269 270
	return error;
}

static int put_compat_statfs64(struct compat_statfs64 __user *ubuf, struct kstatfs *kbuf)
{
	if (sizeof ubuf->f_blocks == 4) {
J
Jon Tollefson 已提交
271 272
		if ((kbuf->f_blocks | kbuf->f_bfree | kbuf->f_bavail |
		     kbuf->f_bsize | kbuf->f_frsize) & 0xffffffff00000000ULL)
L
Linus Torvalds 已提交
273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293
			return -EOVERFLOW;
		/* f_files and f_ffree may be -1; it's okay
		 * to stuff that into 32 bits */
		if (kbuf->f_files != 0xffffffffffffffffULL
		 && (kbuf->f_files & 0xffffffff00000000ULL))
			return -EOVERFLOW;
		if (kbuf->f_ffree != 0xffffffffffffffffULL
		 && (kbuf->f_ffree & 0xffffffff00000000ULL))
			return -EOVERFLOW;
	}
	if (!access_ok(VERIFY_WRITE, ubuf, sizeof(*ubuf)) ||
	    __put_user(kbuf->f_type, &ubuf->f_type) ||
	    __put_user(kbuf->f_bsize, &ubuf->f_bsize) ||
	    __put_user(kbuf->f_blocks, &ubuf->f_blocks) ||
	    __put_user(kbuf->f_bfree, &ubuf->f_bfree) ||
	    __put_user(kbuf->f_bavail, &ubuf->f_bavail) ||
	    __put_user(kbuf->f_files, &ubuf->f_files) ||
	    __put_user(kbuf->f_ffree, &ubuf->f_ffree) ||
	    __put_user(kbuf->f_namelen, &ubuf->f_namelen) ||
	    __put_user(kbuf->f_fsid.val[0], &ubuf->f_fsid.val[0]) ||
	    __put_user(kbuf->f_fsid.val[1], &ubuf->f_fsid.val[1]) ||
294 295 296
	    __put_user(kbuf->f_frsize, &ubuf->f_frsize) ||
	    __put_user(kbuf->f_flags, &ubuf->f_flags) ||
	    __clear_user(ubuf->f_spare, sizeof(ubuf->f_spare)))
L
Linus Torvalds 已提交
297 298 299 300
		return -EFAULT;
	return 0;
}

301
COMPAT_SYSCALL_DEFINE3(statfs64, const char __user *, pathname, compat_size_t, sz, struct compat_statfs64 __user *, buf)
L
Linus Torvalds 已提交
302
{
A
Al Viro 已提交
303
	struct kstatfs tmp;
L
Linus Torvalds 已提交
304 305 306 307 308
	int error;

	if (sz != sizeof(*buf))
		return -EINVAL;

A
Al Viro 已提交
309 310 311
	error = user_statfs(pathname, &tmp);
	if (!error)
		error = put_compat_statfs64(buf, &tmp);
L
Linus Torvalds 已提交
312 313 314
	return error;
}

315
COMPAT_SYSCALL_DEFINE3(fstatfs64, unsigned int, fd, compat_size_t, sz, struct compat_statfs64 __user *, buf)
L
Linus Torvalds 已提交
316 317 318 319 320 321 322
{
	struct kstatfs tmp;
	int error;

	if (sz != sizeof(*buf))
		return -EINVAL;

A
Al Viro 已提交
323
	error = fd_statfs(fd, &tmp);
324 325
	if (!error)
		error = put_compat_statfs64(buf, &tmp);
L
Linus Torvalds 已提交
326 327 328
	return error;
}

C
Christoph Hellwig 已提交
329 330 331 332 333
/*
 * This is a copy of sys_ustat, just dealing with a structure layout.
 * Given how simple this syscall is that apporach is more maintainable
 * than the various conversion hacks.
 */
334
COMPAT_SYSCALL_DEFINE2(ustat, unsigned, dev, struct compat_ustat __user *, u)
C
Christoph Hellwig 已提交
335 336 337
{
	struct compat_ustat tmp;
	struct kstatfs sbuf;
A
Al Viro 已提交
338
	int err = vfs_ustat(new_decode_dev(dev), &sbuf);
C
Christoph Hellwig 已提交
339 340 341 342 343 344 345 346 347 348 349
	if (err)
		return err;

	memset(&tmp, 0, sizeof(struct compat_ustat));
	tmp.f_tfree = sbuf.f_bfree;
	tmp.f_tinode = sbuf.f_ffree;
	if (copy_to_user(u, &tmp, sizeof(struct compat_ustat)))
		return -EFAULT;
	return 0;
}

L
Linus Torvalds 已提交
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
static int get_compat_flock(struct flock *kfl, struct compat_flock __user *ufl)
{
	if (!access_ok(VERIFY_READ, ufl, sizeof(*ufl)) ||
	    __get_user(kfl->l_type, &ufl->l_type) ||
	    __get_user(kfl->l_whence, &ufl->l_whence) ||
	    __get_user(kfl->l_start, &ufl->l_start) ||
	    __get_user(kfl->l_len, &ufl->l_len) ||
	    __get_user(kfl->l_pid, &ufl->l_pid))
		return -EFAULT;
	return 0;
}

static int put_compat_flock(struct flock *kfl, struct compat_flock __user *ufl)
{
	if (!access_ok(VERIFY_WRITE, ufl, sizeof(*ufl)) ||
	    __put_user(kfl->l_type, &ufl->l_type) ||
	    __put_user(kfl->l_whence, &ufl->l_whence) ||
	    __put_user(kfl->l_start, &ufl->l_start) ||
	    __put_user(kfl->l_len, &ufl->l_len) ||
	    __put_user(kfl->l_pid, &ufl->l_pid))
		return -EFAULT;
	return 0;
}

#ifndef HAVE_ARCH_GET_COMPAT_FLOCK64
static int get_compat_flock64(struct flock *kfl, struct compat_flock64 __user *ufl)
{
	if (!access_ok(VERIFY_READ, ufl, sizeof(*ufl)) ||
	    __get_user(kfl->l_type, &ufl->l_type) ||
	    __get_user(kfl->l_whence, &ufl->l_whence) ||
	    __get_user(kfl->l_start, &ufl->l_start) ||
	    __get_user(kfl->l_len, &ufl->l_len) ||
	    __get_user(kfl->l_pid, &ufl->l_pid))
		return -EFAULT;
	return 0;
}
#endif

#ifndef HAVE_ARCH_PUT_COMPAT_FLOCK64
static int put_compat_flock64(struct flock *kfl, struct compat_flock64 __user *ufl)
{
	if (!access_ok(VERIFY_WRITE, ufl, sizeof(*ufl)) ||
	    __put_user(kfl->l_type, &ufl->l_type) ||
	    __put_user(kfl->l_whence, &ufl->l_whence) ||
	    __put_user(kfl->l_start, &ufl->l_start) ||
	    __put_user(kfl->l_len, &ufl->l_len) ||
	    __put_user(kfl->l_pid, &ufl->l_pid))
		return -EFAULT;
	return 0;
}
#endif

402 403 404 405 406 407 408 409 410 411 412 413 414 415 416
static unsigned int
convert_fcntl_cmd(unsigned int cmd)
{
	switch (cmd) {
	case F_GETLK64:
		return F_GETLK;
	case F_SETLK64:
		return F_SETLK;
	case F_SETLKW64:
		return F_SETLKW;
	}

	return cmd;
}

417 418
COMPAT_SYSCALL_DEFINE3(fcntl64, unsigned int, fd, unsigned int, cmd,
		       compat_ulong_t, arg)
L
Linus Torvalds 已提交
419 420 421 422
{
	mm_segment_t old_fs;
	struct flock f;
	long ret;
423
	unsigned int conv_cmd;
L
Linus Torvalds 已提交
424 425 426 427 428 429 430 431 432 433 434 435 436

	switch (cmd) {
	case F_GETLK:
	case F_SETLK:
	case F_SETLKW:
		ret = get_compat_flock(&f, compat_ptr(arg));
		if (ret != 0)
			break;
		old_fs = get_fs();
		set_fs(KERNEL_DS);
		ret = sys_fcntl(fd, cmd, (unsigned long)&f);
		set_fs(old_fs);
		if (cmd == F_GETLK && ret == 0) {
437
			/* GETLK was successful and we need to return the data...
438 439 440 441 442 443 444 445 446 447 448
			 * but it needs to fit in the compat structure.
			 * l_start shouldn't be too big, unless the original
			 * start + end is greater than COMPAT_OFF_T_MAX, in which
			 * case the app was asking for trouble, so we return
			 * -EOVERFLOW in that case.
			 * l_len could be too big, in which case we just truncate it,
			 * and only allow the app to see that part of the conflicting
			 * lock that might make sense to it anyway
			 */

			if (f.l_start > COMPAT_OFF_T_MAX)
L
Linus Torvalds 已提交
449
				ret = -EOVERFLOW;
450 451
			if (f.l_len > COMPAT_OFF_T_MAX)
				f.l_len = COMPAT_OFF_T_MAX;
L
Linus Torvalds 已提交
452 453 454 455 456 457 458 459
			if (ret == 0)
				ret = put_compat_flock(&f, compat_ptr(arg));
		}
		break;

	case F_GETLK64:
	case F_SETLK64:
	case F_SETLKW64:
460 461 462
	case F_GETLKP:
	case F_SETLKP:
	case F_SETLKPW:
L
Linus Torvalds 已提交
463 464 465 466 467
		ret = get_compat_flock64(&f, compat_ptr(arg));
		if (ret != 0)
			break;
		old_fs = get_fs();
		set_fs(KERNEL_DS);
468 469
		conv_cmd = convert_fcntl_cmd(cmd);
		ret = sys_fcntl(fd, conv_cmd, (unsigned long)&f);
L
Linus Torvalds 已提交
470
		set_fs(old_fs);
471
		if ((conv_cmd == F_GETLK || conv_cmd == F_GETLKP) && ret == 0) {
472 473
			/* need to return lock information - see above for commentary */
			if (f.l_start > COMPAT_LOFF_T_MAX)
L
Linus Torvalds 已提交
474
				ret = -EOVERFLOW;
475 476
			if (f.l_len > COMPAT_LOFF_T_MAX)
				f.l_len = COMPAT_LOFF_T_MAX;
L
Linus Torvalds 已提交
477 478 479 480 481 482 483 484 485 486 487 488
			if (ret == 0)
				ret = put_compat_flock64(&f, compat_ptr(arg));
		}
		break;

	default:
		ret = sys_fcntl(fd, cmd, arg);
		break;
	}
	return ret;
}

489 490
COMPAT_SYSCALL_DEFINE3(fcntl, unsigned int, fd, unsigned int, cmd,
		       compat_ulong_t, arg)
L
Linus Torvalds 已提交
491
{
492 493 494 495 496 497 498
	switch (cmd) {
	case F_GETLK64:
	case F_SETLK64:
	case F_SETLKW64:
	case F_GETLKP:
	case F_SETLKP:
	case F_SETLKPW:
L
Linus Torvalds 已提交
499
		return -EINVAL;
500
	}
L
Linus Torvalds 已提交
501 502 503
	return compat_sys_fcntl64(fd, cmd, arg);
}

504
COMPAT_SYSCALL_DEFINE2(io_setup, unsigned, nr_reqs, u32 __user *, ctx32p)
L
Linus Torvalds 已提交
505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522
{
	long ret;
	aio_context_t ctx64;

	mm_segment_t oldfs = get_fs();
	if (unlikely(get_user(ctx64, ctx32p)))
		return -EFAULT;

	set_fs(KERNEL_DS);
	/* The __user pointer cast is valid because of the set_fs() */
	ret = sys_io_setup(nr_reqs, (aio_context_t __user *) &ctx64);
	set_fs(oldfs);
	/* truncating is ok because it's a user address */
	if (!ret)
		ret = put_user((u32) ctx64, ctx32p);
	return ret;
}

523 524 525 526 527
COMPAT_SYSCALL_DEFINE5(io_getevents, compat_aio_context_t, ctx_id,
		       compat_long_t, min_nr,
		       compat_long_t, nr,
		       struct io_event __user *, events,
		       struct compat_timespec __user *, timeout)
L
Linus Torvalds 已提交
528 529 530 531 532
{
	struct timespec t;
	struct timespec __user *ut = NULL;

	if (timeout) {
533
		if (compat_get_timespec(&t, timeout))
534
			return -EFAULT;
L
Linus Torvalds 已提交
535 536 537

		ut = compat_alloc_user_space(sizeof(*ut));
		if (copy_to_user(ut, &t, sizeof(t)) )
538
			return -EFAULT;
L
Linus Torvalds 已提交
539
	} 
540
	return sys_io_getevents(ctx_id, min_nr, nr, events, ut);
L
Linus Torvalds 已提交
541 542
}

543 544 545 546 547 548
/* A write operation does a read from user space and vice versa */
#define vrfy_dir(type) ((type) == READ ? VERIFY_WRITE : VERIFY_READ)

ssize_t compat_rw_copy_check_uvector(int type,
		const struct compat_iovec __user *uvector, unsigned long nr_segs,
		unsigned long fast_segs, struct iovec *fast_pointer,
549
		struct iovec **ret_pointer)
550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569
{
	compat_ssize_t tot_len;
	struct iovec *iov = *ret_pointer = fast_pointer;
	ssize_t ret = 0;
	int seg;

	/*
	 * SuS says "The readv() function *may* fail if the iovcnt argument
	 * was less than or equal to 0, or greater than {IOV_MAX}.  Linux has
	 * traditionally returned zero for zero segments, so...
	 */
	if (nr_segs == 0)
		goto out;

	ret = -EINVAL;
	if (nr_segs > UIO_MAXIOV || nr_segs < 0)
		goto out;
	if (nr_segs > fast_segs) {
		ret = -ENOMEM;
		iov = kmalloc(nr_segs*sizeof(struct iovec), GFP_KERNEL);
570
		if (iov == NULL)
571 572 573 574
			goto out;
	}
	*ret_pointer = iov;

575 576 577 578
	ret = -EFAULT;
	if (!access_ok(VERIFY_READ, uvector, nr_segs*sizeof(*uvector)))
		goto out;

579 580 581
	/*
	 * Single unix specification:
	 * We should -EINVAL if an element length is not >= 0 and fitting an
582
	 * ssize_t.
583
	 *
584 585
	 * In Linux, the total length is limited to MAX_RW_COUNT, there is
	 * no overflow possibility.
586 587 588 589 590 591 592 593 594 595 596 597 598 599
	 */
	tot_len = 0;
	ret = -EINVAL;
	for (seg = 0; seg < nr_segs; seg++) {
		compat_uptr_t buf;
		compat_ssize_t len;

		if (__get_user(len, &uvector->iov_len) ||
		   __get_user(buf, &uvector->iov_base)) {
			ret = -EFAULT;
			goto out;
		}
		if (len < 0)	/* size_t not fitting in compat_ssize_t .. */
			goto out;
600
		if (type >= 0 &&
C
Christopher Yeoh 已提交
601
		    !access_ok(vrfy_dir(type), compat_ptr(buf), len)) {
602 603 604
			ret = -EFAULT;
			goto out;
		}
605 606 607
		if (len > MAX_RW_COUNT - tot_len)
			len = MAX_RW_COUNT - tot_len;
		tot_len += len;
608 609 610 611 612 613 614 615 616 617 618
		iov->iov_base = compat_ptr(buf);
		iov->iov_len = (compat_size_t) len;
		uvector++;
		iov++;
	}
	ret = tot_len;

out:
	return ret;
}

L
Linus Torvalds 已提交
619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635
static inline long
copy_iocb(long nr, u32 __user *ptr32, struct iocb __user * __user *ptr64)
{
	compat_uptr_t uptr;
	int i;

	for (i = 0; i < nr; ++i) {
		if (get_user(uptr, ptr32 + i))
			return -EFAULT;
		if (put_user(compat_ptr(uptr), ptr64 + i))
			return -EFAULT;
	}
	return 0;
}

#define MAX_AIO_SUBMITS 	(PAGE_SIZE/sizeof(struct iocb *))

636 637
COMPAT_SYSCALL_DEFINE3(io_submit, compat_aio_context_t, ctx_id,
		       int, nr, u32 __user *, iocb)
L
Linus Torvalds 已提交
638 639 640 641 642 643 644 645 646 647 648 649 650
{
	struct iocb __user * __user *iocb64; 
	long ret;

	if (unlikely(nr < 0))
		return -EINVAL;

	if (nr > MAX_AIO_SUBMITS)
		nr = MAX_AIO_SUBMITS;
	
	iocb64 = compat_alloc_user_space(nr * sizeof(*iocb64));
	ret = copy_iocb(nr, iocb, iocb64);
	if (!ret)
651
		ret = do_io_submit(ctx_id, nr, iocb64, 1);
L
Linus Torvalds 已提交
652 653 654 655 656 657
	return ret;
}

struct compat_ncp_mount_data {
	compat_int_t version;
	compat_uint_t ncp_fd;
658
	__compat_uid_t mounted_uid;
L
Linus Torvalds 已提交
659 660 661 662 663
	compat_pid_t wdog_pid;
	unsigned char mounted_vol[NCP_VOLNAME_LEN + 1];
	compat_uint_t time_out;
	compat_uint_t retry_count;
	compat_uint_t flags;
664 665
	__compat_uid_t uid;
	__compat_gid_t gid;
L
Linus Torvalds 已提交
666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720
	compat_mode_t file_mode;
	compat_mode_t dir_mode;
};

struct compat_ncp_mount_data_v4 {
	compat_int_t version;
	compat_ulong_t flags;
	compat_ulong_t mounted_uid;
	compat_long_t wdog_pid;
	compat_uint_t ncp_fd;
	compat_uint_t time_out;
	compat_uint_t retry_count;
	compat_ulong_t uid;
	compat_ulong_t gid;
	compat_ulong_t file_mode;
	compat_ulong_t dir_mode;
};

static void *do_ncp_super_data_conv(void *raw_data)
{
	int version = *(unsigned int *)raw_data;

	if (version == 3) {
		struct compat_ncp_mount_data *c_n = raw_data;
		struct ncp_mount_data *n = raw_data;

		n->dir_mode = c_n->dir_mode;
		n->file_mode = c_n->file_mode;
		n->gid = c_n->gid;
		n->uid = c_n->uid;
		memmove (n->mounted_vol, c_n->mounted_vol, (sizeof (c_n->mounted_vol) + 3 * sizeof (unsigned int)));
		n->wdog_pid = c_n->wdog_pid;
		n->mounted_uid = c_n->mounted_uid;
	} else if (version == 4) {
		struct compat_ncp_mount_data_v4 *c_n = raw_data;
		struct ncp_mount_data_v4 *n = raw_data;

		n->dir_mode = c_n->dir_mode;
		n->file_mode = c_n->file_mode;
		n->gid = c_n->gid;
		n->uid = c_n->uid;
		n->retry_count = c_n->retry_count;
		n->time_out = c_n->time_out;
		n->ncp_fd = c_n->ncp_fd;
		n->wdog_pid = c_n->wdog_pid;
		n->mounted_uid = c_n->mounted_uid;
		n->flags = c_n->flags;
	} else if (version != 5) {
		return NULL;
	}

	return raw_data;
}


721 722
struct compat_nfs_string {
	compat_uint_t len;
723
	compat_uptr_t data;
724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747
};

static inline void compat_nfs_string(struct nfs_string *dst,
				     struct compat_nfs_string *src)
{
	dst->data = compat_ptr(src->data);
	dst->len = src->len;
}

struct compat_nfs4_mount_data_v1 {
	compat_int_t version;
	compat_int_t flags;
	compat_int_t rsize;
	compat_int_t wsize;
	compat_int_t timeo;
	compat_int_t retrans;
	compat_int_t acregmin;
	compat_int_t acregmax;
	compat_int_t acdirmin;
	compat_int_t acdirmax;
	struct compat_nfs_string client_addr;
	struct compat_nfs_string mnt_path;
	struct compat_nfs_string hostname;
	compat_uint_t host_addrlen;
748
	compat_uptr_t host_addr;
749 750
	compat_int_t proto;
	compat_int_t auth_flavourlen;
751
	compat_uptr_t auth_flavours;
752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785
};

static int do_nfs4_super_data_conv(void *raw_data)
{
	int version = *(compat_uint_t *) raw_data;

	if (version == 1) {
		struct compat_nfs4_mount_data_v1 *raw = raw_data;
		struct nfs4_mount_data *real = raw_data;

		/* copy the fields backwards */
		real->auth_flavours = compat_ptr(raw->auth_flavours);
		real->auth_flavourlen = raw->auth_flavourlen;
		real->proto = raw->proto;
		real->host_addr = compat_ptr(raw->host_addr);
		real->host_addrlen = raw->host_addrlen;
		compat_nfs_string(&real->hostname, &raw->hostname);
		compat_nfs_string(&real->mnt_path, &raw->mnt_path);
		compat_nfs_string(&real->client_addr, &raw->client_addr);
		real->acdirmax = raw->acdirmax;
		real->acdirmin = raw->acdirmin;
		real->acregmax = raw->acregmax;
		real->acregmin = raw->acregmin;
		real->retrans = raw->retrans;
		real->timeo = raw->timeo;
		real->wsize = raw->wsize;
		real->rsize = raw->rsize;
		real->flags = raw->flags;
		real->version = raw->version;
	}

	return 0;
}

L
Linus Torvalds 已提交
786
#define NCPFS_NAME      "ncpfs"
787
#define NFS4_NAME	"nfs4"
L
Linus Torvalds 已提交
788

789 790 791 792
COMPAT_SYSCALL_DEFINE5(mount, const char __user *, dev_name,
		       const char __user *, dir_name,
		       const char __user *, type, compat_ulong_t, flags,
		       const void __user *, data)
L
Linus Torvalds 已提交
793
{
794
	char *kernel_type;
L
Linus Torvalds 已提交
795
	unsigned long data_page;
796
	char *kernel_dev;
797
	struct filename *dir;
L
Linus Torvalds 已提交
798 799
	int retval;

800
	retval = copy_mount_string(type, &kernel_type);
L
Linus Torvalds 已提交
801 802 803
	if (retval < 0)
		goto out;

804 805 806
	dir = getname(dir_name);
	retval = PTR_ERR(dir);
	if (IS_ERR(dir))
L
Linus Torvalds 已提交
807 808
		goto out1;

809
	retval = copy_mount_string(dev_name, &kernel_dev);
L
Linus Torvalds 已提交
810 811 812
	if (retval < 0)
		goto out2;

813
	retval = copy_mount_options(data, &data_page);
L
Linus Torvalds 已提交
814 815 816 817 818
	if (retval < 0)
		goto out3;

	retval = -EINVAL;

819
	if (kernel_type && data_page) {
A
Arnd Bergmann 已提交
820
		if (!strcmp(kernel_type, NCPFS_NAME)) {
L
Linus Torvalds 已提交
821
			do_ncp_super_data_conv((void *)data_page);
822
		} else if (!strcmp(kernel_type, NFS4_NAME)) {
823 824
			if (do_nfs4_super_data_conv((void *) data_page))
				goto out4;
L
Linus Torvalds 已提交
825 826 827
		}
	}

828
	retval = do_mount(kernel_dev, dir->name, kernel_type,
L
Linus Torvalds 已提交
829 830
			flags, (void*)data_page);

831
 out4:
L
Linus Torvalds 已提交
832 833
	free_page(data_page);
 out3:
834
	kfree(kernel_dev);
L
Linus Torvalds 已提交
835
 out2:
836
	putname(dir);
L
Linus Torvalds 已提交
837
 out1:
838
	kfree(kernel_type);
L
Linus Torvalds 已提交
839 840 841 842 843 844 845 846 847 848 849 850
 out:
	return retval;
}

struct compat_old_linux_dirent {
	compat_ulong_t	d_ino;
	compat_ulong_t	d_offset;
	unsigned short	d_namlen;
	char		d_name[1];
};

struct compat_readdir_callback {
851
	struct dir_context ctx;
L
Linus Torvalds 已提交
852 853 854 855 856
	struct compat_old_linux_dirent __user *dirent;
	int result;
};

static int compat_fillonedir(void *__buf, const char *name, int namlen,
857
			loff_t offset, u64 ino, unsigned int d_type)
L
Linus Torvalds 已提交
858 859 860
{
	struct compat_readdir_callback *buf = __buf;
	struct compat_old_linux_dirent __user *dirent;
861
	compat_ulong_t d_ino;
L
Linus Torvalds 已提交
862 863 864

	if (buf->result)
		return -EINVAL;
865
	d_ino = ino;
866 867
	if (sizeof(d_ino) < sizeof(ino) && d_ino != ino) {
		buf->result = -EOVERFLOW;
868
		return -EOVERFLOW;
869
	}
L
Linus Torvalds 已提交
870 871 872 873 874 875
	buf->result++;
	dirent = buf->dirent;
	if (!access_ok(VERIFY_WRITE, dirent,
			(unsigned long)(dirent->d_name + namlen + 1) -
				(unsigned long)dirent))
		goto efault;
876
	if (	__put_user(d_ino, &dirent->d_ino) ||
L
Linus Torvalds 已提交
877 878 879 880 881 882 883 884 885 886 887
		__put_user(offset, &dirent->d_offset) ||
		__put_user(namlen, &dirent->d_namlen) ||
		__copy_to_user(dirent->d_name, name, namlen) ||
		__put_user(0, dirent->d_name + namlen))
		goto efault;
	return 0;
efault:
	buf->result = -EFAULT;
	return -EFAULT;
}

888 889
COMPAT_SYSCALL_DEFINE3(old_readdir, unsigned int, fd,
		struct compat_old_linux_dirent __user *, dirent, unsigned int, count)
L
Linus Torvalds 已提交
890 891
{
	int error;
892
	struct fd f = fdget(fd);
A
Al Viro 已提交
893 894 895 896
	struct compat_readdir_callback buf = {
		.ctx.actor = compat_fillonedir,
		.dirent = dirent
	};
L
Linus Torvalds 已提交
897

898
	if (!f.file)
899
		return -EBADF;
L
Linus Torvalds 已提交
900

901
	error = iterate_dir(f.file, &buf.ctx);
902
	if (buf.result)
L
Linus Torvalds 已提交
903 904
		error = buf.result;

905
	fdput(f);
L
Linus Torvalds 已提交
906 907 908 909 910 911 912 913 914 915 916
	return error;
}

struct compat_linux_dirent {
	compat_ulong_t	d_ino;
	compat_ulong_t	d_off;
	unsigned short	d_reclen;
	char		d_name[1];
};

struct compat_getdents_callback {
917
	struct dir_context ctx;
L
Linus Torvalds 已提交
918 919 920 921 922 923 924
	struct compat_linux_dirent __user *current_dir;
	struct compat_linux_dirent __user *previous;
	int count;
	int error;
};

static int compat_filldir(void *__buf, const char *name, int namlen,
925
		loff_t offset, u64 ino, unsigned int d_type)
L
Linus Torvalds 已提交
926 927 928
{
	struct compat_linux_dirent __user * dirent;
	struct compat_getdents_callback *buf = __buf;
929
	compat_ulong_t d_ino;
930 931
	int reclen = ALIGN(offsetof(struct compat_linux_dirent, d_name) +
		namlen + 2, sizeof(compat_long_t));
L
Linus Torvalds 已提交
932 933 934 935

	buf->error = -EINVAL;	/* only used if we fail.. */
	if (reclen > buf->count)
		return -EINVAL;
936
	d_ino = ino;
937 938
	if (sizeof(d_ino) < sizeof(ino) && d_ino != ino) {
		buf->error = -EOVERFLOW;
939
		return -EOVERFLOW;
940
	}
L
Linus Torvalds 已提交
941 942 943 944 945 946
	dirent = buf->previous;
	if (dirent) {
		if (__put_user(offset, &dirent->d_off))
			goto efault;
	}
	dirent = buf->current_dir;
947
	if (__put_user(d_ino, &dirent->d_ino))
L
Linus Torvalds 已提交
948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966
		goto efault;
	if (__put_user(reclen, &dirent->d_reclen))
		goto efault;
	if (copy_to_user(dirent->d_name, name, namlen))
		goto efault;
	if (__put_user(0, dirent->d_name + namlen))
		goto efault;
	if (__put_user(d_type, (char  __user *) dirent + reclen - 1))
		goto efault;
	buf->previous = dirent;
	dirent = (void __user *)dirent + reclen;
	buf->current_dir = dirent;
	buf->count -= reclen;
	return 0;
efault:
	buf->error = -EFAULT;
	return -EFAULT;
}

967 968
COMPAT_SYSCALL_DEFINE3(getdents, unsigned int, fd,
		struct compat_linux_dirent __user *, dirent, unsigned int, count)
L
Linus Torvalds 已提交
969
{
970
	struct fd f;
L
Linus Torvalds 已提交
971
	struct compat_linux_dirent __user * lastdirent;
A
Al Viro 已提交
972 973 974 975 976
	struct compat_getdents_callback buf = {
		.ctx.actor = compat_filldir,
		.current_dir = dirent,
		.count = count
	};
L
Linus Torvalds 已提交
977 978 979
	int error;

	if (!access_ok(VERIFY_WRITE, dirent, count))
980
		return -EFAULT;
L
Linus Torvalds 已提交
981

982 983
	f = fdget(fd);
	if (!f.file)
984
		return -EBADF;
L
Linus Torvalds 已提交
985

986
	error = iterate_dir(f.file, &buf.ctx);
987 988
	if (error >= 0)
		error = buf.error;
L
Linus Torvalds 已提交
989 990
	lastdirent = buf.previous;
	if (lastdirent) {
991
		if (put_user(buf.ctx.pos, &lastdirent->d_off))
L
Linus Torvalds 已提交
992 993 994 995
			error = -EFAULT;
		else
			error = count - buf.count;
	}
996
	fdput(f);
L
Linus Torvalds 已提交
997 998 999
	return error;
}

1000
#ifdef __ARCH_WANT_COMPAT_SYS_GETDENTS64
L
Linus Torvalds 已提交
1001 1002

struct compat_getdents_callback64 {
1003
	struct dir_context ctx;
L
Linus Torvalds 已提交
1004 1005 1006 1007 1008 1009 1010
	struct linux_dirent64 __user *current_dir;
	struct linux_dirent64 __user *previous;
	int count;
	int error;
};

static int compat_filldir64(void * __buf, const char * name, int namlen, loff_t offset,
1011
		     u64 ino, unsigned int d_type)
L
Linus Torvalds 已提交
1012 1013 1014
{
	struct linux_dirent64 __user *dirent;
	struct compat_getdents_callback64 *buf = __buf;
1015 1016
	int reclen = ALIGN(offsetof(struct linux_dirent64, d_name) + namlen + 1,
		sizeof(u64));
L
Linus Torvalds 已提交
1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051
	u64 off;

	buf->error = -EINVAL;	/* only used if we fail.. */
	if (reclen > buf->count)
		return -EINVAL;
	dirent = buf->previous;

	if (dirent) {
		if (__put_user_unaligned(offset, &dirent->d_off))
			goto efault;
	}
	dirent = buf->current_dir;
	if (__put_user_unaligned(ino, &dirent->d_ino))
		goto efault;
	off = 0;
	if (__put_user_unaligned(off, &dirent->d_off))
		goto efault;
	if (__put_user(reclen, &dirent->d_reclen))
		goto efault;
	if (__put_user(d_type, &dirent->d_type))
		goto efault;
	if (copy_to_user(dirent->d_name, name, namlen))
		goto efault;
	if (__put_user(0, dirent->d_name + namlen))
		goto efault;
	buf->previous = dirent;
	dirent = (void __user *)dirent + reclen;
	buf->current_dir = dirent;
	buf->count -= reclen;
	return 0;
efault:
	buf->error = -EFAULT;
	return -EFAULT;
}

1052 1053
COMPAT_SYSCALL_DEFINE3(getdents64, unsigned int, fd,
		struct linux_dirent64 __user *, dirent, unsigned int, count)
L
Linus Torvalds 已提交
1054
{
1055
	struct fd f;
L
Linus Torvalds 已提交
1056
	struct linux_dirent64 __user * lastdirent;
A
Al Viro 已提交
1057 1058 1059 1060 1061
	struct compat_getdents_callback64 buf = {
		.ctx.actor = compat_filldir64,
		.current_dir = dirent,
		.count = count
	};
L
Linus Torvalds 已提交
1062 1063 1064
	int error;

	if (!access_ok(VERIFY_WRITE, dirent, count))
1065
		return -EFAULT;
L
Linus Torvalds 已提交
1066

1067 1068
	f = fdget(fd);
	if (!f.file)
1069
		return -EBADF;
L
Linus Torvalds 已提交
1070

1071
	error = iterate_dir(f.file, &buf.ctx);
1072 1073
	if (error >= 0)
		error = buf.error;
L
Linus Torvalds 已提交
1074 1075
	lastdirent = buf.previous;
	if (lastdirent) {
1076
		typeof(lastdirent->d_off) d_off = buf.ctx.pos;
1077
		if (__put_user_unaligned(d_off, &lastdirent->d_off))
1078 1079 1080
			error = -EFAULT;
		else
			error = count - buf.count;
L
Linus Torvalds 已提交
1081
	}
1082
	fdput(f);
L
Linus Torvalds 已提交
1083 1084
	return error;
}
1085
#endif /* __ARCH_WANT_COMPAT_SYS_GETDENTS64 */
L
Linus Torvalds 已提交
1086

1087 1088 1089 1090
/*
 * Exactly like fs/open.c:sys_open(), except that it doesn't set the
 * O_LARGEFILE flag.
 */
1091
COMPAT_SYSCALL_DEFINE3(open, const char __user *, filename, int, flags, umode_t, mode)
1092
{
1093 1094 1095 1096 1097 1098 1099
	return do_sys_open(AT_FDCWD, filename, flags, mode);
}

/*
 * Exactly like fs/open.c:sys_openat(), except that it doesn't set the
 * O_LARGEFILE flag.
 */
1100
COMPAT_SYSCALL_DEFINE4(openat, int, dfd, const char __user *, filename, int, flags, umode_t, mode)
1101 1102
{
	return do_sys_open(dfd, filename, flags, mode);
1103 1104
}

L
Linus Torvalds 已提交
1105 1106
#define __COMPAT_NFDBITS       (8 * sizeof(compat_ulong_t))

1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157
static int poll_select_copy_remaining(struct timespec *end_time, void __user *p,
				      int timeval, int ret)
{
	struct timespec ts;

	if (!p)
		return ret;

	if (current->personality & STICKY_TIMEOUTS)
		goto sticky;

	/* No update for zero timeout */
	if (!end_time->tv_sec && !end_time->tv_nsec)
		return ret;

	ktime_get_ts(&ts);
	ts = timespec_sub(*end_time, ts);
	if (ts.tv_sec < 0)
		ts.tv_sec = ts.tv_nsec = 0;

	if (timeval) {
		struct compat_timeval rtv;

		rtv.tv_sec = ts.tv_sec;
		rtv.tv_usec = ts.tv_nsec / NSEC_PER_USEC;

		if (!copy_to_user(p, &rtv, sizeof(rtv)))
			return ret;
	} else {
		struct compat_timespec rts;

		rts.tv_sec = ts.tv_sec;
		rts.tv_nsec = ts.tv_nsec;

		if (!copy_to_user(p, &rts, sizeof(rts)))
			return ret;
	}
	/*
	 * If an application puts its timeval in read-only memory, we
	 * don't want the Linux-specific update to the timeval to
	 * cause a fault after the select has completed
	 * successfully. However, because we're not updating the
	 * timeval, we can't restart the system call.
	 */

sticky:
	if (ret == -ERESTARTNOHAND)
		ret = -EINTR;
	return ret;
}

L
Linus Torvalds 已提交
1158 1159 1160 1161
/*
 * Ooo, nasty.  We need here to frob 32-bit unsigned longs to
 * 64-bit unsigned longs.
 */
1162
static
L
Linus Torvalds 已提交
1163 1164 1165
int compat_get_fd_set(unsigned long nr, compat_ulong_t __user *ufdset,
			unsigned long *fdset)
{
1166
	nr = DIV_ROUND_UP(nr, __COMPAT_NFDBITS);
L
Linus Torvalds 已提交
1167 1168 1169 1170 1171 1172 1173 1174 1175 1176
	if (ufdset) {
		unsigned long odd;

		if (!access_ok(VERIFY_WRITE, ufdset, nr*sizeof(compat_ulong_t)))
			return -EFAULT;

		odd = nr & 1UL;
		nr &= ~1UL;
		while (nr) {
			unsigned long h, l;
1177 1178
			if (__get_user(l, ufdset) || __get_user(h, ufdset+1))
				return -EFAULT;
L
Linus Torvalds 已提交
1179 1180 1181 1182
			ufdset += 2;
			*fdset++ = h << 32 | l;
			nr -= 2;
		}
1183 1184
		if (odd && __get_user(*fdset, ufdset))
			return -EFAULT;
L
Linus Torvalds 已提交
1185 1186 1187 1188 1189 1190 1191 1192 1193 1194
	} else {
		/* Tricky, must clear full unsigned long in the
		 * kernel fdset at the end, this makes sure that
		 * actually happens.
		 */
		memset(fdset, 0, ((nr + 1) & ~1)*sizeof(compat_ulong_t));
	}
	return 0;
}

1195
static
1196 1197
int compat_set_fd_set(unsigned long nr, compat_ulong_t __user *ufdset,
		      unsigned long *fdset)
L
Linus Torvalds 已提交
1198 1199
{
	unsigned long odd;
1200
	nr = DIV_ROUND_UP(nr, __COMPAT_NFDBITS);
L
Linus Torvalds 已提交
1201 1202

	if (!ufdset)
1203
		return 0;
L
Linus Torvalds 已提交
1204 1205 1206 1207 1208 1209 1210

	odd = nr & 1UL;
	nr &= ~1UL;
	while (nr) {
		unsigned long h, l;
		l = *fdset++;
		h = l >> 32;
1211 1212
		if (__put_user(l, ufdset) || __put_user(h, ufdset+1))
			return -EFAULT;
L
Linus Torvalds 已提交
1213 1214 1215
		ufdset += 2;
		nr -= 2;
	}
1216 1217 1218
	if (odd && __put_user(*fdset, ufdset))
		return -EFAULT;
	return 0;
L
Linus Torvalds 已提交
1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234
}


/*
 * This is a virtual copy of sys_select from fs/select.c and probably
 * should be compared to it from time to time
 */

/*
 * We can actually return ERESTARTSYS instead of EINTR, but I'd
 * like to be certain this leads to no problems. So I return
 * EINTR just for safety.
 *
 * Update: ERESTARTSYS breaks at least the xview clock binary, so
 * I'm trying ERESTARTNOHAND which restart only when you want to.
 */
1235
int compat_core_sys_select(int n, compat_ulong_t __user *inp,
1236 1237
	compat_ulong_t __user *outp, compat_ulong_t __user *exp,
	struct timespec *end_time)
L
Linus Torvalds 已提交
1238 1239
{
	fd_set_bits fds;
1240
	void *bits;
1241
	int size, max_fds, ret = -EINVAL;
1242
	struct fdtable *fdt;
1243
	long stack_fds[SELECT_STACK_ALLOC/sizeof(long)];
L
Linus Torvalds 已提交
1244 1245 1246 1247

	if (n < 0)
		goto out_nofds;

1248
	/* max_fds can increase, so grab it once to avoid race */
L
Linus Torvalds 已提交
1249
	rcu_read_lock();
1250
	fdt = files_fdtable(current->files);
1251
	max_fds = fdt->max_fds;
L
Linus Torvalds 已提交
1252
	rcu_read_unlock();
1253 1254
	if (n > max_fds)
		n = max_fds;
L
Linus Torvalds 已提交
1255 1256 1257 1258 1259 1260 1261

	/*
	 * We need 6 bitmaps (in/out/ex for both incoming and outgoing),
	 * since we used fdset we need to allocate memory in units of
	 * long-words.
	 */
	size = FDS_BYTES(n);
1262 1263 1264 1265 1266 1267 1268
	bits = stack_fds;
	if (size > sizeof(stack_fds) / 6) {
		bits = kmalloc(6 * size, GFP_KERNEL);
		ret = -ENOMEM;
		if (!bits)
			goto out_nofds;
	}
L
Linus Torvalds 已提交
1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283
	fds.in      = (unsigned long *)  bits;
	fds.out     = (unsigned long *) (bits +   size);
	fds.ex      = (unsigned long *) (bits + 2*size);
	fds.res_in  = (unsigned long *) (bits + 3*size);
	fds.res_out = (unsigned long *) (bits + 4*size);
	fds.res_ex  = (unsigned long *) (bits + 5*size);

	if ((ret = compat_get_fd_set(n, inp, fds.in)) ||
	    (ret = compat_get_fd_set(n, outp, fds.out)) ||
	    (ret = compat_get_fd_set(n, exp, fds.ex)))
		goto out;
	zero_fd_set(n, fds.res_in);
	zero_fd_set(n, fds.res_out);
	zero_fd_set(n, fds.res_ex);

1284
	ret = do_select(n, &fds, end_time);
L
Linus Torvalds 已提交
1285 1286 1287 1288 1289 1290 1291 1292 1293 1294

	if (ret < 0)
		goto out;
	if (!ret) {
		ret = -ERESTARTNOHAND;
		if (signal_pending(current))
			goto out;
		ret = 0;
	}

1295 1296 1297 1298
	if (compat_set_fd_set(n, inp, fds.res_in) ||
	    compat_set_fd_set(n, outp, fds.res_out) ||
	    compat_set_fd_set(n, exp, fds.res_ex))
		ret = -EFAULT;
L
Linus Torvalds 已提交
1299
out:
1300 1301
	if (bits != stack_fds)
		kfree(bits);
L
Linus Torvalds 已提交
1302 1303 1304 1305
out_nofds:
	return ret;
}

1306 1307 1308
COMPAT_SYSCALL_DEFINE5(select, int, n, compat_ulong_t __user *, inp,
	compat_ulong_t __user *, outp, compat_ulong_t __user *, exp,
	struct compat_timeval __user *, tvp)
1309
{
1310
	struct timespec end_time, *to = NULL;
1311 1312 1313 1314 1315 1316 1317
	struct compat_timeval tv;
	int ret;

	if (tvp) {
		if (copy_from_user(&tv, tvp, sizeof(tv)))
			return -EFAULT;

1318
		to = &end_time;
1319 1320 1321
		if (poll_select_set_timeout(to,
				tv.tv_sec + (tv.tv_usec / USEC_PER_SEC),
				(tv.tv_usec % USEC_PER_SEC) * NSEC_PER_USEC))
1322 1323 1324
			return -EINVAL;
	}

1325 1326
	ret = compat_core_sys_select(n, inp, outp, exp, to);
	ret = poll_select_copy_remaining(&end_time, tvp, 1, ret);
1327 1328 1329 1330

	return ret;
}

C
Christoph Hellwig 已提交
1331 1332 1333 1334 1335 1336 1337 1338
struct compat_sel_arg_struct {
	compat_ulong_t n;
	compat_uptr_t inp;
	compat_uptr_t outp;
	compat_uptr_t exp;
	compat_uptr_t tvp;
};

1339
COMPAT_SYSCALL_DEFINE1(old_select, struct compat_sel_arg_struct __user *, arg)
C
Christoph Hellwig 已提交
1340 1341 1342 1343 1344 1345 1346 1347 1348
{
	struct compat_sel_arg_struct a;

	if (copy_from_user(&a, arg, sizeof(a)))
		return -EFAULT;
	return compat_sys_select(a.n, compat_ptr(a.inp), compat_ptr(a.outp),
				 compat_ptr(a.exp), compat_ptr(a.tvp));
}

1349
static long do_compat_pselect(int n, compat_ulong_t __user *inp,
1350 1351 1352 1353 1354 1355 1356
	compat_ulong_t __user *outp, compat_ulong_t __user *exp,
	struct compat_timespec __user *tsp, compat_sigset_t __user *sigmask,
	compat_size_t sigsetsize)
{
	compat_sigset_t ss32;
	sigset_t ksigmask, sigsaved;
	struct compat_timespec ts;
1357
	struct timespec end_time, *to = NULL;
1358 1359 1360 1361 1362 1363
	int ret;

	if (tsp) {
		if (copy_from_user(&ts, tsp, sizeof(ts)))
			return -EFAULT;

1364 1365
		to = &end_time;
		if (poll_select_set_timeout(to, ts.tv_sec, ts.tv_nsec))
1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379
			return -EINVAL;
	}

	if (sigmask) {
		if (sigsetsize != sizeof(compat_sigset_t))
			return -EINVAL;
		if (copy_from_user(&ss32, sigmask, sizeof(ss32)))
			return -EFAULT;
		sigset_from_compat(&ksigmask, &ss32);

		sigdelsetmask(&ksigmask, sigmask(SIGKILL)|sigmask(SIGSTOP));
		sigprocmask(SIG_SETMASK, &ksigmask, &sigsaved);
	}

1380 1381
	ret = compat_core_sys_select(n, inp, outp, exp, to);
	ret = poll_select_copy_remaining(&end_time, tsp, 0, ret);
1382 1383 1384 1385 1386 1387 1388 1389 1390 1391

	if (ret == -ERESTARTNOHAND) {
		/*
		 * Don't restore the signal mask yet. Let do_signal() deliver
		 * the signal on the way back to userspace, before the signal
		 * mask is restored.
		 */
		if (sigmask) {
			memcpy(&current->saved_sigmask, &sigsaved,
					sizeof(sigsaved));
1392
			set_restore_sigmask();
1393 1394 1395 1396 1397 1398 1399
		}
	} else if (sigmask)
		sigprocmask(SIG_SETMASK, &sigsaved, NULL);

	return ret;
}

1400 1401 1402
COMPAT_SYSCALL_DEFINE6(pselect6, int, n, compat_ulong_t __user *, inp,
	compat_ulong_t __user *, outp, compat_ulong_t __user *, exp,
	struct compat_timespec __user *, tsp, void __user *, sig)
1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414
{
	compat_size_t sigsetsize = 0;
	compat_uptr_t up = 0;

	if (sig) {
		if (!access_ok(VERIFY_READ, sig,
				sizeof(compat_uptr_t)+sizeof(compat_size_t)) ||
		    	__get_user(up, (compat_uptr_t __user *)sig) ||
		    	__get_user(sigsetsize,
				(compat_size_t __user *)(sig+sizeof(up))))
			return -EFAULT;
	}
1415 1416
	return do_compat_pselect(n, inp, outp, exp, tsp, compat_ptr(up),
				 sigsetsize);
1417 1418
}

1419 1420 1421
COMPAT_SYSCALL_DEFINE5(ppoll, struct pollfd __user *, ufds,
	unsigned int,  nfds, struct compat_timespec __user *, tsp,
	const compat_sigset_t __user *, sigmask, compat_size_t, sigsetsize)
1422 1423 1424 1425
{
	compat_sigset_t ss32;
	sigset_t ksigmask, sigsaved;
	struct compat_timespec ts;
1426
	struct timespec end_time, *to = NULL;
1427 1428 1429 1430 1431 1432
	int ret;

	if (tsp) {
		if (copy_from_user(&ts, tsp, sizeof(ts)))
			return -EFAULT;

1433 1434 1435
		to = &end_time;
		if (poll_select_set_timeout(to, ts.tv_sec, ts.tv_nsec))
			return -EINVAL;
1436 1437 1438
	}

	if (sigmask) {
1439
		if (sigsetsize != sizeof(compat_sigset_t))
1440 1441 1442 1443 1444 1445 1446 1447 1448
			return -EINVAL;
		if (copy_from_user(&ss32, sigmask, sizeof(ss32)))
			return -EFAULT;
		sigset_from_compat(&ksigmask, &ss32);

		sigdelsetmask(&ksigmask, sigmask(SIGKILL)|sigmask(SIGSTOP));
		sigprocmask(SIG_SETMASK, &ksigmask, &sigsaved);
	}

1449
	ret = do_sys_poll(ufds, nfds, to);
1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460

	/* We can restart this syscall, usually */
	if (ret == -EINTR) {
		/*
		 * Don't restore the signal mask yet. Let do_signal() deliver
		 * the signal on the way back to userspace, before the signal
		 * mask is restored.
		 */
		if (sigmask) {
			memcpy(&current->saved_sigmask, &sigsaved,
				sizeof(sigsaved));
1461
			set_restore_sigmask();
1462 1463 1464 1465 1466
		}
		ret = -ERESTARTNOHAND;
	} else if (sigmask)
		sigprocmask(SIG_SETMASK, &sigsaved, NULL);

1467
	ret = poll_select_copy_remaining(&end_time, tsp, 0, ret);
1468 1469 1470 1471

	return ret;
}

1472 1473 1474 1475 1476
#ifdef CONFIG_FHANDLE
/*
 * Exactly like fs/open.c:sys_open_by_handle_at(), except that it
 * doesn't set the O_LARGEFILE flag.
 */
1477 1478
COMPAT_SYSCALL_DEFINE3(open_by_handle_at, int, mountdirfd,
			     struct file_handle __user *, handle, int, flags)
1479 1480 1481 1482
{
	return do_handle_open(mountdirfd, handle, flags);
}
#endif