Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | /* |
2 | * NET An implementation of the SOCKET network access protocol. | |
3 | * | |
4 | * Version: @(#)socket.c 1.1.93 18/02/95 | |
5 | * | |
6 | * Authors: Orest Zborowski, <obz@Kodak.COM> | |
02c30a84 | 7 | * Ross Biro |
1da177e4 LT |
8 | * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG> |
9 | * | |
10 | * Fixes: | |
11 | * Anonymous : NOTSOCK/BADF cleanup. Error fix in | |
12 | * shutdown() | |
13 | * Alan Cox : verify_area() fixes | |
14 | * Alan Cox : Removed DDI | |
15 | * Jonathan Kamens : SOCK_DGRAM reconnect bug | |
16 | * Alan Cox : Moved a load of checks to the very | |
17 | * top level. | |
18 | * Alan Cox : Move address structures to/from user | |
19 | * mode above the protocol layers. | |
20 | * Rob Janssen : Allow 0 length sends. | |
21 | * Alan Cox : Asynchronous I/O support (cribbed from the | |
22 | * tty drivers). | |
23 | * Niibe Yutaka : Asynchronous I/O for writes (4.4BSD style) | |
24 | * Jeff Uphoff : Made max number of sockets command-line | |
25 | * configurable. | |
26 | * Matti Aarnio : Made the number of sockets dynamic, | |
27 | * to be allocated when needed, and mr. | |
28 | * Uphoff's max is used as max to be | |
29 | * allowed to allocate. | |
30 | * Linus : Argh. removed all the socket allocation | |
31 | * altogether: it's in the inode now. | |
32 | * Alan Cox : Made sock_alloc()/sock_release() public | |
33 | * for NetROM and future kernel nfsd type | |
34 | * stuff. | |
35 | * Alan Cox : sendmsg/recvmsg basics. | |
36 | * Tom Dyas : Export net symbols. | |
37 | * Marcin Dalecki : Fixed problems with CONFIG_NET="n". | |
38 | * Alan Cox : Added thread locking to sys_* calls | |
39 | * for sockets. May have errors at the | |
40 | * moment. | |
41 | * Kevin Buhr : Fixed the dumb errors in the above. | |
42 | * Andi Kleen : Some small cleanups, optimizations, | |
43 | * and fixed a copy_from_user() bug. | |
44 | * Tigran Aivazian : sys_send(args) calls sys_sendto(args, NULL, 0) | |
89bddce5 | 45 | * Tigran Aivazian : Made listen(2) backlog sanity checks |
1da177e4 LT |
46 | * protocol-independent |
47 | * | |
48 | * | |
49 | * This program is free software; you can redistribute it and/or | |
50 | * modify it under the terms of the GNU General Public License | |
51 | * as published by the Free Software Foundation; either version | |
52 | * 2 of the License, or (at your option) any later version. | |
53 | * | |
54 | * | |
55 | * This module is effectively the top level interface to the BSD socket | |
89bddce5 | 56 | * paradigm. |
1da177e4 LT |
57 | * |
58 | * Based upon Swansea University Computer Society NET3.039 | |
59 | */ | |
60 | ||
1da177e4 | 61 | #include <linux/mm.h> |
1da177e4 LT |
62 | #include <linux/socket.h> |
63 | #include <linux/file.h> | |
64 | #include <linux/net.h> | |
65 | #include <linux/interrupt.h> | |
aaca0bdc | 66 | #include <linux/thread_info.h> |
55737fda | 67 | #include <linux/rcupdate.h> |
1da177e4 LT |
68 | #include <linux/netdevice.h> |
69 | #include <linux/proc_fs.h> | |
70 | #include <linux/seq_file.h> | |
4a3e2f71 | 71 | #include <linux/mutex.h> |
1da177e4 | 72 | #include <linux/if_bridge.h> |
20380731 ACM |
73 | #include <linux/if_frad.h> |
74 | #include <linux/if_vlan.h> | |
1da177e4 LT |
75 | #include <linux/init.h> |
76 | #include <linux/poll.h> | |
77 | #include <linux/cache.h> | |
78 | #include <linux/module.h> | |
79 | #include <linux/highmem.h> | |
1da177e4 LT |
80 | #include <linux/mount.h> |
81 | #include <linux/security.h> | |
82 | #include <linux/syscalls.h> | |
83 | #include <linux/compat.h> | |
84 | #include <linux/kmod.h> | |
3ec3b2fb | 85 | #include <linux/audit.h> |
d86b5e0e | 86 | #include <linux/wireless.h> |
1b8d7ae4 | 87 | #include <linux/nsproxy.h> |
1fd7317d | 88 | #include <linux/magic.h> |
5a0e3ad6 | 89 | #include <linux/slab.h> |
600e1779 | 90 | #include <linux/xattr.h> |
1da177e4 LT |
91 | |
92 | #include <asm/uaccess.h> | |
93 | #include <asm/unistd.h> | |
94 | ||
95 | #include <net/compat.h> | |
87de87d5 | 96 | #include <net/wext.h> |
f8451725 | 97 | #include <net/cls_cgroup.h> |
1da177e4 LT |
98 | |
99 | #include <net/sock.h> | |
100 | #include <linux/netfilter.h> | |
101 | ||
6b96018b AB |
102 | #include <linux/if_tun.h> |
103 | #include <linux/ipv6_route.h> | |
104 | #include <linux/route.h> | |
6b96018b AB |
105 | #include <linux/sockios.h> |
106 | #include <linux/atalk.h> | |
107 | ||
1da177e4 | 108 | static int sock_no_open(struct inode *irrelevant, struct file *dontcare); |
027445c3 BP |
109 | static ssize_t sock_aio_read(struct kiocb *iocb, const struct iovec *iov, |
110 | unsigned long nr_segs, loff_t pos); | |
111 | static ssize_t sock_aio_write(struct kiocb *iocb, const struct iovec *iov, | |
112 | unsigned long nr_segs, loff_t pos); | |
89bddce5 | 113 | static int sock_mmap(struct file *file, struct vm_area_struct *vma); |
1da177e4 LT |
114 | |
115 | static int sock_close(struct inode *inode, struct file *file); | |
116 | static unsigned int sock_poll(struct file *file, | |
117 | struct poll_table_struct *wait); | |
89bddce5 | 118 | static long sock_ioctl(struct file *file, unsigned int cmd, unsigned long arg); |
89bbfc95 SP |
119 | #ifdef CONFIG_COMPAT |
120 | static long compat_sock_ioctl(struct file *file, | |
89bddce5 | 121 | unsigned int cmd, unsigned long arg); |
89bbfc95 | 122 | #endif |
1da177e4 | 123 | static int sock_fasync(int fd, struct file *filp, int on); |
1da177e4 LT |
124 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
125 | int offset, size_t size, loff_t *ppos, int more); | |
9c55e01c | 126 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 127 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c | 128 | unsigned int flags); |
1da177e4 | 129 | |
1da177e4 LT |
130 | /* |
131 | * Socket files have a set of 'special' operations as well as the generic file ones. These don't appear | |
132 | * in the operation structures but are done directly via the socketcall() multiplexor. | |
133 | */ | |
134 | ||
da7071d7 | 135 | static const struct file_operations socket_file_ops = { |
1da177e4 LT |
136 | .owner = THIS_MODULE, |
137 | .llseek = no_llseek, | |
138 | .aio_read = sock_aio_read, | |
139 | .aio_write = sock_aio_write, | |
140 | .poll = sock_poll, | |
141 | .unlocked_ioctl = sock_ioctl, | |
89bbfc95 SP |
142 | #ifdef CONFIG_COMPAT |
143 | .compat_ioctl = compat_sock_ioctl, | |
144 | #endif | |
1da177e4 LT |
145 | .mmap = sock_mmap, |
146 | .open = sock_no_open, /* special open code to disallow open via /proc */ | |
147 | .release = sock_close, | |
148 | .fasync = sock_fasync, | |
5274f052 JA |
149 | .sendpage = sock_sendpage, |
150 | .splice_write = generic_splice_sendpage, | |
9c55e01c | 151 | .splice_read = sock_splice_read, |
1da177e4 LT |
152 | }; |
153 | ||
154 | /* | |
155 | * The protocol list. Each protocol is registered in here. | |
156 | */ | |
157 | ||
1da177e4 | 158 | static DEFINE_SPINLOCK(net_family_lock); |
190683a9 | 159 | static const struct net_proto_family __rcu *net_families[NPROTO] __read_mostly; |
1da177e4 | 160 | |
1da177e4 LT |
161 | /* |
162 | * Statistics counters of the socket lists | |
163 | */ | |
164 | ||
c6d409cf | 165 | static DEFINE_PER_CPU(int, sockets_in_use); |
1da177e4 LT |
166 | |
167 | /* | |
89bddce5 SH |
168 | * Support routines. |
169 | * Move socket addresses back and forth across the kernel/user | |
170 | * divide and look after the messy bits. | |
1da177e4 LT |
171 | */ |
172 | ||
1da177e4 LT |
173 | /** |
174 | * move_addr_to_kernel - copy a socket address into kernel space | |
175 | * @uaddr: Address in user space | |
176 | * @kaddr: Address in kernel space | |
177 | * @ulen: Length in user space | |
178 | * | |
179 | * The address is copied into kernel space. If the provided address is | |
180 | * too long an error code of -EINVAL is returned. If the copy gives | |
181 | * invalid addresses -EFAULT is returned. On a success 0 is returned. | |
182 | */ | |
183 | ||
43db362d | 184 | int move_addr_to_kernel(void __user *uaddr, int ulen, struct sockaddr_storage *kaddr) |
1da177e4 | 185 | { |
230b1839 | 186 | if (ulen < 0 || ulen > sizeof(struct sockaddr_storage)) |
1da177e4 | 187 | return -EINVAL; |
89bddce5 | 188 | if (ulen == 0) |
1da177e4 | 189 | return 0; |
89bddce5 | 190 | if (copy_from_user(kaddr, uaddr, ulen)) |
1da177e4 | 191 | return -EFAULT; |
3ec3b2fb | 192 | return audit_sockaddr(ulen, kaddr); |
1da177e4 LT |
193 | } |
194 | ||
195 | /** | |
196 | * move_addr_to_user - copy an address to user space | |
197 | * @kaddr: kernel space address | |
198 | * @klen: length of address in kernel | |
199 | * @uaddr: user space address | |
200 | * @ulen: pointer to user length field | |
201 | * | |
202 | * The value pointed to by ulen on entry is the buffer length available. | |
203 | * This is overwritten with the buffer space used. -EINVAL is returned | |
204 | * if an overlong buffer is specified or a negative buffer size. -EFAULT | |
205 | * is returned if either the buffer or the length field are not | |
206 | * accessible. | |
207 | * After copying the data up to the limit the user specifies, the true | |
208 | * length of the data is written over the length limit the user | |
209 | * specified. Zero is returned for a success. | |
210 | */ | |
89bddce5 | 211 | |
43db362d | 212 | static int move_addr_to_user(struct sockaddr_storage *kaddr, int klen, |
11165f14 | 213 | void __user *uaddr, int __user *ulen) |
1da177e4 LT |
214 | { |
215 | int err; | |
216 | int len; | |
217 | ||
89bddce5 SH |
218 | err = get_user(len, ulen); |
219 | if (err) | |
1da177e4 | 220 | return err; |
89bddce5 SH |
221 | if (len > klen) |
222 | len = klen; | |
230b1839 | 223 | if (len < 0 || len > sizeof(struct sockaddr_storage)) |
1da177e4 | 224 | return -EINVAL; |
89bddce5 | 225 | if (len) { |
d6fe3945 SG |
226 | if (audit_sockaddr(klen, kaddr)) |
227 | return -ENOMEM; | |
89bddce5 | 228 | if (copy_to_user(uaddr, kaddr, len)) |
1da177e4 LT |
229 | return -EFAULT; |
230 | } | |
231 | /* | |
89bddce5 SH |
232 | * "fromlen shall refer to the value before truncation.." |
233 | * 1003.1g | |
1da177e4 LT |
234 | */ |
235 | return __put_user(klen, ulen); | |
236 | } | |
237 | ||
e18b890b | 238 | static struct kmem_cache *sock_inode_cachep __read_mostly; |
1da177e4 LT |
239 | |
240 | static struct inode *sock_alloc_inode(struct super_block *sb) | |
241 | { | |
242 | struct socket_alloc *ei; | |
eaefd110 | 243 | struct socket_wq *wq; |
89bddce5 | 244 | |
e94b1766 | 245 | ei = kmem_cache_alloc(sock_inode_cachep, GFP_KERNEL); |
1da177e4 LT |
246 | if (!ei) |
247 | return NULL; | |
eaefd110 ED |
248 | wq = kmalloc(sizeof(*wq), GFP_KERNEL); |
249 | if (!wq) { | |
43815482 ED |
250 | kmem_cache_free(sock_inode_cachep, ei); |
251 | return NULL; | |
252 | } | |
eaefd110 ED |
253 | init_waitqueue_head(&wq->wait); |
254 | wq->fasync_list = NULL; | |
255 | RCU_INIT_POINTER(ei->socket.wq, wq); | |
89bddce5 | 256 | |
1da177e4 LT |
257 | ei->socket.state = SS_UNCONNECTED; |
258 | ei->socket.flags = 0; | |
259 | ei->socket.ops = NULL; | |
260 | ei->socket.sk = NULL; | |
261 | ei->socket.file = NULL; | |
1da177e4 LT |
262 | |
263 | return &ei->vfs_inode; | |
264 | } | |
265 | ||
266 | static void sock_destroy_inode(struct inode *inode) | |
267 | { | |
43815482 | 268 | struct socket_alloc *ei; |
eaefd110 | 269 | struct socket_wq *wq; |
43815482 ED |
270 | |
271 | ei = container_of(inode, struct socket_alloc, vfs_inode); | |
eaefd110 | 272 | wq = rcu_dereference_protected(ei->socket.wq, 1); |
61845220 | 273 | kfree_rcu(wq, rcu); |
43815482 | 274 | kmem_cache_free(sock_inode_cachep, ei); |
1da177e4 LT |
275 | } |
276 | ||
51cc5068 | 277 | static void init_once(void *foo) |
1da177e4 | 278 | { |
89bddce5 | 279 | struct socket_alloc *ei = (struct socket_alloc *)foo; |
1da177e4 | 280 | |
a35afb83 | 281 | inode_init_once(&ei->vfs_inode); |
1da177e4 | 282 | } |
89bddce5 | 283 | |
1da177e4 LT |
284 | static int init_inodecache(void) |
285 | { | |
286 | sock_inode_cachep = kmem_cache_create("sock_inode_cache", | |
89bddce5 SH |
287 | sizeof(struct socket_alloc), |
288 | 0, | |
289 | (SLAB_HWCACHE_ALIGN | | |
290 | SLAB_RECLAIM_ACCOUNT | | |
291 | SLAB_MEM_SPREAD), | |
20c2df83 | 292 | init_once); |
1da177e4 LT |
293 | if (sock_inode_cachep == NULL) |
294 | return -ENOMEM; | |
295 | return 0; | |
296 | } | |
297 | ||
b87221de | 298 | static const struct super_operations sockfs_ops = { |
c6d409cf ED |
299 | .alloc_inode = sock_alloc_inode, |
300 | .destroy_inode = sock_destroy_inode, | |
301 | .statfs = simple_statfs, | |
1da177e4 LT |
302 | }; |
303 | ||
c23fbb6b ED |
304 | /* |
305 | * sockfs_dname() is called from d_path(). | |
306 | */ | |
307 | static char *sockfs_dname(struct dentry *dentry, char *buffer, int buflen) | |
308 | { | |
309 | return dynamic_dname(dentry, buffer, buflen, "socket:[%lu]", | |
310 | dentry->d_inode->i_ino); | |
311 | } | |
312 | ||
3ba13d17 | 313 | static const struct dentry_operations sockfs_dentry_operations = { |
c23fbb6b | 314 | .d_dname = sockfs_dname, |
1da177e4 LT |
315 | }; |
316 | ||
c74a1cbb AV |
317 | static struct dentry *sockfs_mount(struct file_system_type *fs_type, |
318 | int flags, const char *dev_name, void *data) | |
319 | { | |
320 | return mount_pseudo(fs_type, "socket:", &sockfs_ops, | |
321 | &sockfs_dentry_operations, SOCKFS_MAGIC); | |
322 | } | |
323 | ||
324 | static struct vfsmount *sock_mnt __read_mostly; | |
325 | ||
326 | static struct file_system_type sock_fs_type = { | |
327 | .name = "sockfs", | |
328 | .mount = sockfs_mount, | |
329 | .kill_sb = kill_anon_super, | |
330 | }; | |
331 | ||
1da177e4 LT |
332 | /* |
333 | * Obtains the first available file descriptor and sets it up for use. | |
334 | * | |
39d8c1b6 DM |
335 | * These functions create file structures and maps them to fd space |
336 | * of the current process. On success it returns file descriptor | |
1da177e4 LT |
337 | * and file struct implicitly stored in sock->file. |
338 | * Note that another thread may close file descriptor before we return | |
339 | * from this function. We use the fact that now we do not refer | |
340 | * to socket after mapping. If one day we will need it, this | |
341 | * function will increment ref. count on file by 1. | |
342 | * | |
343 | * In any case returned fd MAY BE not valid! | |
344 | * This race condition is unavoidable | |
345 | * with shared fd spaces, we cannot solve it inside kernel, | |
346 | * but we take care of internal coherence yet. | |
347 | */ | |
348 | ||
aab174f0 | 349 | struct file *sock_alloc_file(struct socket *sock, int flags, const char *dname) |
1da177e4 | 350 | { |
7cbe66b6 | 351 | struct qstr name = { .name = "" }; |
2c48b9c4 | 352 | struct path path; |
7cbe66b6 | 353 | struct file *file; |
1da177e4 | 354 | |
600e1779 MY |
355 | if (dname) { |
356 | name.name = dname; | |
357 | name.len = strlen(name.name); | |
358 | } else if (sock->sk) { | |
359 | name.name = sock->sk->sk_prot_creator->name; | |
360 | name.len = strlen(name.name); | |
361 | } | |
4b936885 | 362 | path.dentry = d_alloc_pseudo(sock_mnt->mnt_sb, &name); |
28407630 AV |
363 | if (unlikely(!path.dentry)) |
364 | return ERR_PTR(-ENOMEM); | |
2c48b9c4 | 365 | path.mnt = mntget(sock_mnt); |
39d8c1b6 | 366 | |
2c48b9c4 | 367 | d_instantiate(path.dentry, SOCK_INODE(sock)); |
cc3808f8 | 368 | SOCK_INODE(sock)->i_fop = &socket_file_ops; |
39d8c1b6 | 369 | |
2c48b9c4 | 370 | file = alloc_file(&path, FMODE_READ | FMODE_WRITE, |
ce8d2cdf | 371 | &socket_file_ops); |
cc3808f8 AV |
372 | if (unlikely(!file)) { |
373 | /* drop dentry, keep inode */ | |
7de9c6ee | 374 | ihold(path.dentry->d_inode); |
2c48b9c4 | 375 | path_put(&path); |
28407630 | 376 | return ERR_PTR(-ENFILE); |
cc3808f8 AV |
377 | } |
378 | ||
379 | sock->file = file; | |
77d27200 | 380 | file->f_flags = O_RDWR | (flags & O_NONBLOCK); |
39d8c1b6 DM |
381 | file->f_pos = 0; |
382 | file->private_data = sock; | |
28407630 | 383 | return file; |
39d8c1b6 | 384 | } |
56b31d1c | 385 | EXPORT_SYMBOL(sock_alloc_file); |
39d8c1b6 | 386 | |
56b31d1c | 387 | static int sock_map_fd(struct socket *sock, int flags) |
39d8c1b6 DM |
388 | { |
389 | struct file *newfile; | |
28407630 AV |
390 | int fd = get_unused_fd_flags(flags); |
391 | if (unlikely(fd < 0)) | |
392 | return fd; | |
39d8c1b6 | 393 | |
aab174f0 | 394 | newfile = sock_alloc_file(sock, flags, NULL); |
28407630 | 395 | if (likely(!IS_ERR(newfile))) { |
39d8c1b6 | 396 | fd_install(fd, newfile); |
28407630 AV |
397 | return fd; |
398 | } | |
7cbe66b6 | 399 | |
28407630 AV |
400 | put_unused_fd(fd); |
401 | return PTR_ERR(newfile); | |
1da177e4 LT |
402 | } |
403 | ||
406a3c63 | 404 | struct socket *sock_from_file(struct file *file, int *err) |
6cb153ca | 405 | { |
6cb153ca BL |
406 | if (file->f_op == &socket_file_ops) |
407 | return file->private_data; /* set in sock_map_fd */ | |
408 | ||
23bb80d2 ED |
409 | *err = -ENOTSOCK; |
410 | return NULL; | |
6cb153ca | 411 | } |
406a3c63 | 412 | EXPORT_SYMBOL(sock_from_file); |
6cb153ca | 413 | |
1da177e4 | 414 | /** |
c6d409cf | 415 | * sockfd_lookup - Go from a file number to its socket slot |
1da177e4 LT |
416 | * @fd: file handle |
417 | * @err: pointer to an error code return | |
418 | * | |
419 | * The file handle passed in is locked and the socket it is bound | |
420 | * too is returned. If an error occurs the err pointer is overwritten | |
421 | * with a negative errno code and NULL is returned. The function checks | |
422 | * for both invalid handles and passing a handle which is not a socket. | |
423 | * | |
424 | * On a success the socket object pointer is returned. | |
425 | */ | |
426 | ||
427 | struct socket *sockfd_lookup(int fd, int *err) | |
428 | { | |
429 | struct file *file; | |
1da177e4 LT |
430 | struct socket *sock; |
431 | ||
89bddce5 SH |
432 | file = fget(fd); |
433 | if (!file) { | |
1da177e4 LT |
434 | *err = -EBADF; |
435 | return NULL; | |
436 | } | |
89bddce5 | 437 | |
6cb153ca BL |
438 | sock = sock_from_file(file, err); |
439 | if (!sock) | |
1da177e4 | 440 | fput(file); |
6cb153ca BL |
441 | return sock; |
442 | } | |
c6d409cf | 443 | EXPORT_SYMBOL(sockfd_lookup); |
1da177e4 | 444 | |
6cb153ca BL |
445 | static struct socket *sockfd_lookup_light(int fd, int *err, int *fput_needed) |
446 | { | |
447 | struct file *file; | |
448 | struct socket *sock; | |
449 | ||
3672558c | 450 | *err = -EBADF; |
6cb153ca BL |
451 | file = fget_light(fd, fput_needed); |
452 | if (file) { | |
453 | sock = sock_from_file(file, err); | |
454 | if (sock) | |
455 | return sock; | |
456 | fput_light(file, *fput_needed); | |
1da177e4 | 457 | } |
6cb153ca | 458 | return NULL; |
1da177e4 LT |
459 | } |
460 | ||
600e1779 MY |
461 | #define XATTR_SOCKPROTONAME_SUFFIX "sockprotoname" |
462 | #define XATTR_NAME_SOCKPROTONAME (XATTR_SYSTEM_PREFIX XATTR_SOCKPROTONAME_SUFFIX) | |
463 | #define XATTR_NAME_SOCKPROTONAME_LEN (sizeof(XATTR_NAME_SOCKPROTONAME)-1) | |
464 | static ssize_t sockfs_getxattr(struct dentry *dentry, | |
465 | const char *name, void *value, size_t size) | |
466 | { | |
467 | const char *proto_name; | |
468 | size_t proto_size; | |
469 | int error; | |
470 | ||
471 | error = -ENODATA; | |
472 | if (!strncmp(name, XATTR_NAME_SOCKPROTONAME, XATTR_NAME_SOCKPROTONAME_LEN)) { | |
473 | proto_name = dentry->d_name.name; | |
474 | proto_size = strlen(proto_name); | |
475 | ||
476 | if (value) { | |
477 | error = -ERANGE; | |
478 | if (proto_size + 1 > size) | |
479 | goto out; | |
480 | ||
481 | strncpy(value, proto_name, proto_size + 1); | |
482 | } | |
483 | error = proto_size + 1; | |
484 | } | |
485 | ||
486 | out: | |
487 | return error; | |
488 | } | |
489 | ||
490 | static ssize_t sockfs_listxattr(struct dentry *dentry, char *buffer, | |
491 | size_t size) | |
492 | { | |
493 | ssize_t len; | |
494 | ssize_t used = 0; | |
495 | ||
496 | len = security_inode_listsecurity(dentry->d_inode, buffer, size); | |
497 | if (len < 0) | |
498 | return len; | |
499 | used += len; | |
500 | if (buffer) { | |
501 | if (size < used) | |
502 | return -ERANGE; | |
503 | buffer += len; | |
504 | } | |
505 | ||
506 | len = (XATTR_NAME_SOCKPROTONAME_LEN + 1); | |
507 | used += len; | |
508 | if (buffer) { | |
509 | if (size < used) | |
510 | return -ERANGE; | |
511 | memcpy(buffer, XATTR_NAME_SOCKPROTONAME, len); | |
512 | buffer += len; | |
513 | } | |
514 | ||
515 | return used; | |
516 | } | |
517 | ||
518 | static const struct inode_operations sockfs_inode_ops = { | |
519 | .getxattr = sockfs_getxattr, | |
520 | .listxattr = sockfs_listxattr, | |
521 | }; | |
522 | ||
1da177e4 LT |
523 | /** |
524 | * sock_alloc - allocate a socket | |
89bddce5 | 525 | * |
1da177e4 LT |
526 | * Allocate a new inode and socket object. The two are bound together |
527 | * and initialised. The socket is then returned. If we are out of inodes | |
528 | * NULL is returned. | |
529 | */ | |
530 | ||
531 | static struct socket *sock_alloc(void) | |
532 | { | |
89bddce5 SH |
533 | struct inode *inode; |
534 | struct socket *sock; | |
1da177e4 | 535 | |
a209dfc7 | 536 | inode = new_inode_pseudo(sock_mnt->mnt_sb); |
1da177e4 LT |
537 | if (!inode) |
538 | return NULL; | |
539 | ||
540 | sock = SOCKET_I(inode); | |
541 | ||
29a020d3 | 542 | kmemcheck_annotate_bitfield(sock, type); |
85fe4025 | 543 | inode->i_ino = get_next_ino(); |
89bddce5 | 544 | inode->i_mode = S_IFSOCK | S_IRWXUGO; |
8192b0c4 DH |
545 | inode->i_uid = current_fsuid(); |
546 | inode->i_gid = current_fsgid(); | |
600e1779 | 547 | inode->i_op = &sockfs_inode_ops; |
1da177e4 | 548 | |
19e8d69c | 549 | this_cpu_add(sockets_in_use, 1); |
1da177e4 LT |
550 | return sock; |
551 | } | |
552 | ||
553 | /* | |
554 | * In theory you can't get an open on this inode, but /proc provides | |
555 | * a back door. Remember to keep it shut otherwise you'll let the | |
556 | * creepy crawlies in. | |
557 | */ | |
89bddce5 | 558 | |
1da177e4 LT |
559 | static int sock_no_open(struct inode *irrelevant, struct file *dontcare) |
560 | { | |
561 | return -ENXIO; | |
562 | } | |
563 | ||
4b6f5d20 | 564 | const struct file_operations bad_sock_fops = { |
1da177e4 LT |
565 | .owner = THIS_MODULE, |
566 | .open = sock_no_open, | |
6038f373 | 567 | .llseek = noop_llseek, |
1da177e4 LT |
568 | }; |
569 | ||
570 | /** | |
571 | * sock_release - close a socket | |
572 | * @sock: socket to close | |
573 | * | |
574 | * The socket is released from the protocol stack if it has a release | |
575 | * callback, and the inode is then released if the socket is bound to | |
89bddce5 | 576 | * an inode not a file. |
1da177e4 | 577 | */ |
89bddce5 | 578 | |
1da177e4 LT |
579 | void sock_release(struct socket *sock) |
580 | { | |
581 | if (sock->ops) { | |
582 | struct module *owner = sock->ops->owner; | |
583 | ||
584 | sock->ops->release(sock); | |
585 | sock->ops = NULL; | |
586 | module_put(owner); | |
587 | } | |
588 | ||
eaefd110 | 589 | if (rcu_dereference_protected(sock->wq, 1)->fasync_list) |
1da177e4 LT |
590 | printk(KERN_ERR "sock_release: fasync list not empty!\n"); |
591 | ||
b09e786b MP |
592 | if (test_bit(SOCK_EXTERNALLY_ALLOCATED, &sock->flags)) |
593 | return; | |
594 | ||
19e8d69c | 595 | this_cpu_sub(sockets_in_use, 1); |
1da177e4 LT |
596 | if (!sock->file) { |
597 | iput(SOCK_INODE(sock)); | |
598 | return; | |
599 | } | |
89bddce5 | 600 | sock->file = NULL; |
1da177e4 | 601 | } |
c6d409cf | 602 | EXPORT_SYMBOL(sock_release); |
1da177e4 | 603 | |
2244d07b | 604 | int sock_tx_timestamp(struct sock *sk, __u8 *tx_flags) |
20d49473 | 605 | { |
2244d07b | 606 | *tx_flags = 0; |
20d49473 | 607 | if (sock_flag(sk, SOCK_TIMESTAMPING_TX_HARDWARE)) |
2244d07b | 608 | *tx_flags |= SKBTX_HW_TSTAMP; |
20d49473 | 609 | if (sock_flag(sk, SOCK_TIMESTAMPING_TX_SOFTWARE)) |
2244d07b | 610 | *tx_flags |= SKBTX_SW_TSTAMP; |
6e3e939f JB |
611 | if (sock_flag(sk, SOCK_WIFI_STATUS)) |
612 | *tx_flags |= SKBTX_WIFI_STATUS; | |
20d49473 PO |
613 | return 0; |
614 | } | |
615 | EXPORT_SYMBOL(sock_tx_timestamp); | |
616 | ||
228e548e AB |
617 | static inline int __sock_sendmsg_nosec(struct kiocb *iocb, struct socket *sock, |
618 | struct msghdr *msg, size_t size) | |
1da177e4 LT |
619 | { |
620 | struct sock_iocb *si = kiocb_to_siocb(iocb); | |
1da177e4 LT |
621 | |
622 | si->sock = sock; | |
623 | si->scm = NULL; | |
624 | si->msg = msg; | |
625 | si->size = size; | |
626 | ||
1da177e4 LT |
627 | return sock->ops->sendmsg(iocb, sock, msg, size); |
628 | } | |
629 | ||
228e548e AB |
630 | static inline int __sock_sendmsg(struct kiocb *iocb, struct socket *sock, |
631 | struct msghdr *msg, size_t size) | |
632 | { | |
633 | int err = security_socket_sendmsg(sock, msg, size); | |
634 | ||
635 | return err ?: __sock_sendmsg_nosec(iocb, sock, msg, size); | |
636 | } | |
637 | ||
1da177e4 LT |
638 | int sock_sendmsg(struct socket *sock, struct msghdr *msg, size_t size) |
639 | { | |
640 | struct kiocb iocb; | |
641 | struct sock_iocb siocb; | |
642 | int ret; | |
643 | ||
644 | init_sync_kiocb(&iocb, NULL); | |
645 | iocb.private = &siocb; | |
646 | ret = __sock_sendmsg(&iocb, sock, msg, size); | |
647 | if (-EIOCBQUEUED == ret) | |
648 | ret = wait_on_sync_kiocb(&iocb); | |
649 | return ret; | |
650 | } | |
c6d409cf | 651 | EXPORT_SYMBOL(sock_sendmsg); |
1da177e4 | 652 | |
894dc24c | 653 | static int sock_sendmsg_nosec(struct socket *sock, struct msghdr *msg, size_t size) |
228e548e AB |
654 | { |
655 | struct kiocb iocb; | |
656 | struct sock_iocb siocb; | |
657 | int ret; | |
658 | ||
659 | init_sync_kiocb(&iocb, NULL); | |
660 | iocb.private = &siocb; | |
661 | ret = __sock_sendmsg_nosec(&iocb, sock, msg, size); | |
662 | if (-EIOCBQUEUED == ret) | |
663 | ret = wait_on_sync_kiocb(&iocb); | |
664 | return ret; | |
665 | } | |
666 | ||
1da177e4 LT |
667 | int kernel_sendmsg(struct socket *sock, struct msghdr *msg, |
668 | struct kvec *vec, size_t num, size_t size) | |
669 | { | |
670 | mm_segment_t oldfs = get_fs(); | |
671 | int result; | |
672 | ||
673 | set_fs(KERNEL_DS); | |
674 | /* | |
675 | * the following is safe, since for compiler definitions of kvec and | |
676 | * iovec are identical, yielding the same in-core layout and alignment | |
677 | */ | |
89bddce5 | 678 | msg->msg_iov = (struct iovec *)vec; |
1da177e4 LT |
679 | msg->msg_iovlen = num; |
680 | result = sock_sendmsg(sock, msg, size); | |
681 | set_fs(oldfs); | |
682 | return result; | |
683 | } | |
c6d409cf | 684 | EXPORT_SYMBOL(kernel_sendmsg); |
1da177e4 | 685 | |
20d49473 PO |
686 | static int ktime2ts(ktime_t kt, struct timespec *ts) |
687 | { | |
688 | if (kt.tv64) { | |
689 | *ts = ktime_to_timespec(kt); | |
690 | return 1; | |
691 | } else { | |
692 | return 0; | |
693 | } | |
694 | } | |
695 | ||
92f37fd2 ED |
696 | /* |
697 | * called from sock_recv_timestamp() if sock_flag(sk, SOCK_RCVTSTAMP) | |
698 | */ | |
699 | void __sock_recv_timestamp(struct msghdr *msg, struct sock *sk, | |
700 | struct sk_buff *skb) | |
701 | { | |
20d49473 PO |
702 | int need_software_tstamp = sock_flag(sk, SOCK_RCVTSTAMP); |
703 | struct timespec ts[3]; | |
704 | int empty = 1; | |
705 | struct skb_shared_hwtstamps *shhwtstamps = | |
706 | skb_hwtstamps(skb); | |
707 | ||
708 | /* Race occurred between timestamp enabling and packet | |
709 | receiving. Fill in the current time for now. */ | |
710 | if (need_software_tstamp && skb->tstamp.tv64 == 0) | |
711 | __net_timestamp(skb); | |
712 | ||
713 | if (need_software_tstamp) { | |
714 | if (!sock_flag(sk, SOCK_RCVTSTAMPNS)) { | |
715 | struct timeval tv; | |
716 | skb_get_timestamp(skb, &tv); | |
717 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMP, | |
718 | sizeof(tv), &tv); | |
719 | } else { | |
842509b8 | 720 | skb_get_timestampns(skb, &ts[0]); |
20d49473 | 721 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPNS, |
842509b8 | 722 | sizeof(ts[0]), &ts[0]); |
20d49473 PO |
723 | } |
724 | } | |
725 | ||
726 | ||
727 | memset(ts, 0, sizeof(ts)); | |
728 | if (skb->tstamp.tv64 && | |
729 | sock_flag(sk, SOCK_TIMESTAMPING_SOFTWARE)) { | |
730 | skb_get_timestampns(skb, ts + 0); | |
731 | empty = 0; | |
732 | } | |
733 | if (shhwtstamps) { | |
734 | if (sock_flag(sk, SOCK_TIMESTAMPING_SYS_HARDWARE) && | |
735 | ktime2ts(shhwtstamps->syststamp, ts + 1)) | |
736 | empty = 0; | |
737 | if (sock_flag(sk, SOCK_TIMESTAMPING_RAW_HARDWARE) && | |
738 | ktime2ts(shhwtstamps->hwtstamp, ts + 2)) | |
739 | empty = 0; | |
92f37fd2 | 740 | } |
20d49473 PO |
741 | if (!empty) |
742 | put_cmsg(msg, SOL_SOCKET, | |
743 | SCM_TIMESTAMPING, sizeof(ts), &ts); | |
92f37fd2 | 744 | } |
7c81fd8b ACM |
745 | EXPORT_SYMBOL_GPL(__sock_recv_timestamp); |
746 | ||
6e3e939f JB |
747 | void __sock_recv_wifi_status(struct msghdr *msg, struct sock *sk, |
748 | struct sk_buff *skb) | |
749 | { | |
750 | int ack; | |
751 | ||
752 | if (!sock_flag(sk, SOCK_WIFI_STATUS)) | |
753 | return; | |
754 | if (!skb->wifi_acked_valid) | |
755 | return; | |
756 | ||
757 | ack = skb->wifi_acked; | |
758 | ||
759 | put_cmsg(msg, SOL_SOCKET, SCM_WIFI_STATUS, sizeof(ack), &ack); | |
760 | } | |
761 | EXPORT_SYMBOL_GPL(__sock_recv_wifi_status); | |
762 | ||
11165f14 | 763 | static inline void sock_recv_drops(struct msghdr *msg, struct sock *sk, |
764 | struct sk_buff *skb) | |
3b885787 NH |
765 | { |
766 | if (sock_flag(sk, SOCK_RXQ_OVFL) && skb && skb->dropcount) | |
767 | put_cmsg(msg, SOL_SOCKET, SO_RXQ_OVFL, | |
768 | sizeof(__u32), &skb->dropcount); | |
769 | } | |
770 | ||
767dd033 | 771 | void __sock_recv_ts_and_drops(struct msghdr *msg, struct sock *sk, |
3b885787 NH |
772 | struct sk_buff *skb) |
773 | { | |
774 | sock_recv_timestamp(msg, sk, skb); | |
775 | sock_recv_drops(msg, sk, skb); | |
776 | } | |
767dd033 | 777 | EXPORT_SYMBOL_GPL(__sock_recv_ts_and_drops); |
3b885787 | 778 | |
a2e27255 ACM |
779 | static inline int __sock_recvmsg_nosec(struct kiocb *iocb, struct socket *sock, |
780 | struct msghdr *msg, size_t size, int flags) | |
1da177e4 | 781 | { |
1da177e4 LT |
782 | struct sock_iocb *si = kiocb_to_siocb(iocb); |
783 | ||
784 | si->sock = sock; | |
785 | si->scm = NULL; | |
786 | si->msg = msg; | |
787 | si->size = size; | |
788 | si->flags = flags; | |
789 | ||
1da177e4 LT |
790 | return sock->ops->recvmsg(iocb, sock, msg, size, flags); |
791 | } | |
792 | ||
a2e27255 ACM |
793 | static inline int __sock_recvmsg(struct kiocb *iocb, struct socket *sock, |
794 | struct msghdr *msg, size_t size, int flags) | |
795 | { | |
796 | int err = security_socket_recvmsg(sock, msg, size, flags); | |
797 | ||
798 | return err ?: __sock_recvmsg_nosec(iocb, sock, msg, size, flags); | |
799 | } | |
800 | ||
89bddce5 | 801 | int sock_recvmsg(struct socket *sock, struct msghdr *msg, |
1da177e4 LT |
802 | size_t size, int flags) |
803 | { | |
804 | struct kiocb iocb; | |
805 | struct sock_iocb siocb; | |
806 | int ret; | |
807 | ||
89bddce5 | 808 | init_sync_kiocb(&iocb, NULL); |
1da177e4 LT |
809 | iocb.private = &siocb; |
810 | ret = __sock_recvmsg(&iocb, sock, msg, size, flags); | |
811 | if (-EIOCBQUEUED == ret) | |
812 | ret = wait_on_sync_kiocb(&iocb); | |
813 | return ret; | |
814 | } | |
c6d409cf | 815 | EXPORT_SYMBOL(sock_recvmsg); |
1da177e4 | 816 | |
a2e27255 ACM |
817 | static int sock_recvmsg_nosec(struct socket *sock, struct msghdr *msg, |
818 | size_t size, int flags) | |
819 | { | |
820 | struct kiocb iocb; | |
821 | struct sock_iocb siocb; | |
822 | int ret; | |
823 | ||
824 | init_sync_kiocb(&iocb, NULL); | |
825 | iocb.private = &siocb; | |
826 | ret = __sock_recvmsg_nosec(&iocb, sock, msg, size, flags); | |
827 | if (-EIOCBQUEUED == ret) | |
828 | ret = wait_on_sync_kiocb(&iocb); | |
829 | return ret; | |
830 | } | |
831 | ||
c1249c0a ML |
832 | /** |
833 | * kernel_recvmsg - Receive a message from a socket (kernel space) | |
834 | * @sock: The socket to receive the message from | |
835 | * @msg: Received message | |
836 | * @vec: Input s/g array for message data | |
837 | * @num: Size of input s/g array | |
838 | * @size: Number of bytes to read | |
839 | * @flags: Message flags (MSG_DONTWAIT, etc...) | |
840 | * | |
841 | * On return the msg structure contains the scatter/gather array passed in the | |
842 | * vec argument. The array is modified so that it consists of the unfilled | |
843 | * portion of the original array. | |
844 | * | |
845 | * The returned value is the total number of bytes received, or an error. | |
846 | */ | |
89bddce5 SH |
847 | int kernel_recvmsg(struct socket *sock, struct msghdr *msg, |
848 | struct kvec *vec, size_t num, size_t size, int flags) | |
1da177e4 LT |
849 | { |
850 | mm_segment_t oldfs = get_fs(); | |
851 | int result; | |
852 | ||
853 | set_fs(KERNEL_DS); | |
854 | /* | |
855 | * the following is safe, since for compiler definitions of kvec and | |
856 | * iovec are identical, yielding the same in-core layout and alignment | |
857 | */ | |
89bddce5 | 858 | msg->msg_iov = (struct iovec *)vec, msg->msg_iovlen = num; |
1da177e4 LT |
859 | result = sock_recvmsg(sock, msg, size, flags); |
860 | set_fs(oldfs); | |
861 | return result; | |
862 | } | |
c6d409cf | 863 | EXPORT_SYMBOL(kernel_recvmsg); |
1da177e4 LT |
864 | |
865 | static void sock_aio_dtor(struct kiocb *iocb) | |
866 | { | |
867 | kfree(iocb->private); | |
868 | } | |
869 | ||
ce1d4d3e CH |
870 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
871 | int offset, size_t size, loff_t *ppos, int more) | |
1da177e4 | 872 | { |
1da177e4 LT |
873 | struct socket *sock; |
874 | int flags; | |
875 | ||
ce1d4d3e CH |
876 | sock = file->private_data; |
877 | ||
35f9c09f ED |
878 | flags = (file->f_flags & O_NONBLOCK) ? MSG_DONTWAIT : 0; |
879 | /* more is a combination of MSG_MORE and MSG_SENDPAGE_NOTLAST */ | |
880 | flags |= more; | |
ce1d4d3e | 881 | |
e6949583 | 882 | return kernel_sendpage(sock, page, offset, size, flags); |
ce1d4d3e | 883 | } |
1da177e4 | 884 | |
9c55e01c | 885 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 886 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c JA |
887 | unsigned int flags) |
888 | { | |
889 | struct socket *sock = file->private_data; | |
890 | ||
997b37da RDC |
891 | if (unlikely(!sock->ops->splice_read)) |
892 | return -EINVAL; | |
893 | ||
9c55e01c JA |
894 | return sock->ops->splice_read(sock, ppos, pipe, len, flags); |
895 | } | |
896 | ||
ce1d4d3e | 897 | static struct sock_iocb *alloc_sock_iocb(struct kiocb *iocb, |
89bddce5 | 898 | struct sock_iocb *siocb) |
ce1d4d3e CH |
899 | { |
900 | if (!is_sync_kiocb(iocb)) { | |
901 | siocb = kmalloc(sizeof(*siocb), GFP_KERNEL); | |
902 | if (!siocb) | |
903 | return NULL; | |
1da177e4 LT |
904 | iocb->ki_dtor = sock_aio_dtor; |
905 | } | |
1da177e4 | 906 | |
ce1d4d3e | 907 | siocb->kiocb = iocb; |
ce1d4d3e CH |
908 | iocb->private = siocb; |
909 | return siocb; | |
1da177e4 LT |
910 | } |
911 | ||
ce1d4d3e | 912 | static ssize_t do_sock_read(struct msghdr *msg, struct kiocb *iocb, |
027445c3 BP |
913 | struct file *file, const struct iovec *iov, |
914 | unsigned long nr_segs) | |
ce1d4d3e CH |
915 | { |
916 | struct socket *sock = file->private_data; | |
917 | size_t size = 0; | |
918 | int i; | |
1da177e4 | 919 | |
89bddce5 SH |
920 | for (i = 0; i < nr_segs; i++) |
921 | size += iov[i].iov_len; | |
1da177e4 | 922 | |
ce1d4d3e CH |
923 | msg->msg_name = NULL; |
924 | msg->msg_namelen = 0; | |
925 | msg->msg_control = NULL; | |
926 | msg->msg_controllen = 0; | |
89bddce5 | 927 | msg->msg_iov = (struct iovec *)iov; |
ce1d4d3e CH |
928 | msg->msg_iovlen = nr_segs; |
929 | msg->msg_flags = (file->f_flags & O_NONBLOCK) ? MSG_DONTWAIT : 0; | |
930 | ||
931 | return __sock_recvmsg(iocb, sock, msg, size, msg->msg_flags); | |
932 | } | |
933 | ||
027445c3 BP |
934 | static ssize_t sock_aio_read(struct kiocb *iocb, const struct iovec *iov, |
935 | unsigned long nr_segs, loff_t pos) | |
ce1d4d3e CH |
936 | { |
937 | struct sock_iocb siocb, *x; | |
938 | ||
1da177e4 LT |
939 | if (pos != 0) |
940 | return -ESPIPE; | |
027445c3 BP |
941 | |
942 | if (iocb->ki_left == 0) /* Match SYS5 behaviour */ | |
1da177e4 LT |
943 | return 0; |
944 | ||
027445c3 BP |
945 | |
946 | x = alloc_sock_iocb(iocb, &siocb); | |
ce1d4d3e CH |
947 | if (!x) |
948 | return -ENOMEM; | |
027445c3 | 949 | return do_sock_read(&x->async_msg, iocb, iocb->ki_filp, iov, nr_segs); |
1da177e4 LT |
950 | } |
951 | ||
ce1d4d3e | 952 | static ssize_t do_sock_write(struct msghdr *msg, struct kiocb *iocb, |
027445c3 BP |
953 | struct file *file, const struct iovec *iov, |
954 | unsigned long nr_segs) | |
1da177e4 | 955 | { |
ce1d4d3e CH |
956 | struct socket *sock = file->private_data; |
957 | size_t size = 0; | |
958 | int i; | |
1da177e4 | 959 | |
89bddce5 SH |
960 | for (i = 0; i < nr_segs; i++) |
961 | size += iov[i].iov_len; | |
1da177e4 | 962 | |
ce1d4d3e CH |
963 | msg->msg_name = NULL; |
964 | msg->msg_namelen = 0; | |
965 | msg->msg_control = NULL; | |
966 | msg->msg_controllen = 0; | |
89bddce5 | 967 | msg->msg_iov = (struct iovec *)iov; |
ce1d4d3e CH |
968 | msg->msg_iovlen = nr_segs; |
969 | msg->msg_flags = (file->f_flags & O_NONBLOCK) ? MSG_DONTWAIT : 0; | |
970 | if (sock->type == SOCK_SEQPACKET) | |
971 | msg->msg_flags |= MSG_EOR; | |
1da177e4 | 972 | |
ce1d4d3e | 973 | return __sock_sendmsg(iocb, sock, msg, size); |
1da177e4 LT |
974 | } |
975 | ||
027445c3 BP |
976 | static ssize_t sock_aio_write(struct kiocb *iocb, const struct iovec *iov, |
977 | unsigned long nr_segs, loff_t pos) | |
ce1d4d3e CH |
978 | { |
979 | struct sock_iocb siocb, *x; | |
1da177e4 | 980 | |
ce1d4d3e CH |
981 | if (pos != 0) |
982 | return -ESPIPE; | |
027445c3 | 983 | |
027445c3 | 984 | x = alloc_sock_iocb(iocb, &siocb); |
ce1d4d3e CH |
985 | if (!x) |
986 | return -ENOMEM; | |
1da177e4 | 987 | |
027445c3 | 988 | return do_sock_write(&x->async_msg, iocb, iocb->ki_filp, iov, nr_segs); |
1da177e4 LT |
989 | } |
990 | ||
1da177e4 LT |
991 | /* |
992 | * Atomic setting of ioctl hooks to avoid race | |
993 | * with module unload. | |
994 | */ | |
995 | ||
4a3e2f71 | 996 | static DEFINE_MUTEX(br_ioctl_mutex); |
c6d409cf | 997 | static int (*br_ioctl_hook) (struct net *, unsigned int cmd, void __user *arg); |
1da177e4 | 998 | |
881d966b | 999 | void brioctl_set(int (*hook) (struct net *, unsigned int, void __user *)) |
1da177e4 | 1000 | { |
4a3e2f71 | 1001 | mutex_lock(&br_ioctl_mutex); |
1da177e4 | 1002 | br_ioctl_hook = hook; |
4a3e2f71 | 1003 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
1004 | } |
1005 | EXPORT_SYMBOL(brioctl_set); | |
1006 | ||
4a3e2f71 | 1007 | static DEFINE_MUTEX(vlan_ioctl_mutex); |
881d966b | 1008 | static int (*vlan_ioctl_hook) (struct net *, void __user *arg); |
1da177e4 | 1009 | |
881d966b | 1010 | void vlan_ioctl_set(int (*hook) (struct net *, void __user *)) |
1da177e4 | 1011 | { |
4a3e2f71 | 1012 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 1013 | vlan_ioctl_hook = hook; |
4a3e2f71 | 1014 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 LT |
1015 | } |
1016 | EXPORT_SYMBOL(vlan_ioctl_set); | |
1017 | ||
4a3e2f71 | 1018 | static DEFINE_MUTEX(dlci_ioctl_mutex); |
89bddce5 | 1019 | static int (*dlci_ioctl_hook) (unsigned int, void __user *); |
1da177e4 | 1020 | |
89bddce5 | 1021 | void dlci_ioctl_set(int (*hook) (unsigned int, void __user *)) |
1da177e4 | 1022 | { |
4a3e2f71 | 1023 | mutex_lock(&dlci_ioctl_mutex); |
1da177e4 | 1024 | dlci_ioctl_hook = hook; |
4a3e2f71 | 1025 | mutex_unlock(&dlci_ioctl_mutex); |
1da177e4 LT |
1026 | } |
1027 | EXPORT_SYMBOL(dlci_ioctl_set); | |
1028 | ||
6b96018b AB |
1029 | static long sock_do_ioctl(struct net *net, struct socket *sock, |
1030 | unsigned int cmd, unsigned long arg) | |
1031 | { | |
1032 | int err; | |
1033 | void __user *argp = (void __user *)arg; | |
1034 | ||
1035 | err = sock->ops->ioctl(sock, cmd, arg); | |
1036 | ||
1037 | /* | |
1038 | * If this ioctl is unknown try to hand it down | |
1039 | * to the NIC driver. | |
1040 | */ | |
1041 | if (err == -ENOIOCTLCMD) | |
1042 | err = dev_ioctl(net, cmd, argp); | |
1043 | ||
1044 | return err; | |
1045 | } | |
1046 | ||
1da177e4 LT |
1047 | /* |
1048 | * With an ioctl, arg may well be a user mode pointer, but we don't know | |
1049 | * what to do with it - that's up to the protocol still. | |
1050 | */ | |
1051 | ||
1052 | static long sock_ioctl(struct file *file, unsigned cmd, unsigned long arg) | |
1053 | { | |
1054 | struct socket *sock; | |
881d966b | 1055 | struct sock *sk; |
1da177e4 LT |
1056 | void __user *argp = (void __user *)arg; |
1057 | int pid, err; | |
881d966b | 1058 | struct net *net; |
1da177e4 | 1059 | |
b69aee04 | 1060 | sock = file->private_data; |
881d966b | 1061 | sk = sock->sk; |
3b1e0a65 | 1062 | net = sock_net(sk); |
1da177e4 | 1063 | if (cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15)) { |
881d966b | 1064 | err = dev_ioctl(net, cmd, argp); |
1da177e4 | 1065 | } else |
3d23e349 | 1066 | #ifdef CONFIG_WEXT_CORE |
1da177e4 | 1067 | if (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST) { |
881d966b | 1068 | err = dev_ioctl(net, cmd, argp); |
1da177e4 | 1069 | } else |
3d23e349 | 1070 | #endif |
89bddce5 | 1071 | switch (cmd) { |
1da177e4 LT |
1072 | case FIOSETOWN: |
1073 | case SIOCSPGRP: | |
1074 | err = -EFAULT; | |
1075 | if (get_user(pid, (int __user *)argp)) | |
1076 | break; | |
1077 | err = f_setown(sock->file, pid, 1); | |
1078 | break; | |
1079 | case FIOGETOWN: | |
1080 | case SIOCGPGRP: | |
609d7fa9 | 1081 | err = put_user(f_getown(sock->file), |
89bddce5 | 1082 | (int __user *)argp); |
1da177e4 LT |
1083 | break; |
1084 | case SIOCGIFBR: | |
1085 | case SIOCSIFBR: | |
1086 | case SIOCBRADDBR: | |
1087 | case SIOCBRDELBR: | |
1088 | err = -ENOPKG; | |
1089 | if (!br_ioctl_hook) | |
1090 | request_module("bridge"); | |
1091 | ||
4a3e2f71 | 1092 | mutex_lock(&br_ioctl_mutex); |
89bddce5 | 1093 | if (br_ioctl_hook) |
881d966b | 1094 | err = br_ioctl_hook(net, cmd, argp); |
4a3e2f71 | 1095 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
1096 | break; |
1097 | case SIOCGIFVLAN: | |
1098 | case SIOCSIFVLAN: | |
1099 | err = -ENOPKG; | |
1100 | if (!vlan_ioctl_hook) | |
1101 | request_module("8021q"); | |
1102 | ||
4a3e2f71 | 1103 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 1104 | if (vlan_ioctl_hook) |
881d966b | 1105 | err = vlan_ioctl_hook(net, argp); |
4a3e2f71 | 1106 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 | 1107 | break; |
1da177e4 LT |
1108 | case SIOCADDDLCI: |
1109 | case SIOCDELDLCI: | |
1110 | err = -ENOPKG; | |
1111 | if (!dlci_ioctl_hook) | |
1112 | request_module("dlci"); | |
1113 | ||
7512cbf6 PE |
1114 | mutex_lock(&dlci_ioctl_mutex); |
1115 | if (dlci_ioctl_hook) | |
1da177e4 | 1116 | err = dlci_ioctl_hook(cmd, argp); |
7512cbf6 | 1117 | mutex_unlock(&dlci_ioctl_mutex); |
1da177e4 LT |
1118 | break; |
1119 | default: | |
6b96018b | 1120 | err = sock_do_ioctl(net, sock, cmd, arg); |
1da177e4 | 1121 | break; |
89bddce5 | 1122 | } |
1da177e4 LT |
1123 | return err; |
1124 | } | |
1125 | ||
1126 | int sock_create_lite(int family, int type, int protocol, struct socket **res) | |
1127 | { | |
1128 | int err; | |
1129 | struct socket *sock = NULL; | |
89bddce5 | 1130 | |
1da177e4 LT |
1131 | err = security_socket_create(family, type, protocol, 1); |
1132 | if (err) | |
1133 | goto out; | |
1134 | ||
1135 | sock = sock_alloc(); | |
1136 | if (!sock) { | |
1137 | err = -ENOMEM; | |
1138 | goto out; | |
1139 | } | |
1140 | ||
1da177e4 | 1141 | sock->type = type; |
7420ed23 VY |
1142 | err = security_socket_post_create(sock, family, type, protocol, 1); |
1143 | if (err) | |
1144 | goto out_release; | |
1145 | ||
1da177e4 LT |
1146 | out: |
1147 | *res = sock; | |
1148 | return err; | |
7420ed23 VY |
1149 | out_release: |
1150 | sock_release(sock); | |
1151 | sock = NULL; | |
1152 | goto out; | |
1da177e4 | 1153 | } |
c6d409cf | 1154 | EXPORT_SYMBOL(sock_create_lite); |
1da177e4 LT |
1155 | |
1156 | /* No kernel lock held - perfect */ | |
89bddce5 | 1157 | static unsigned int sock_poll(struct file *file, poll_table *wait) |
1da177e4 LT |
1158 | { |
1159 | struct socket *sock; | |
1160 | ||
1161 | /* | |
89bddce5 | 1162 | * We can't return errors to poll, so it's either yes or no. |
1da177e4 | 1163 | */ |
b69aee04 | 1164 | sock = file->private_data; |
1da177e4 LT |
1165 | return sock->ops->poll(file, sock, wait); |
1166 | } | |
1167 | ||
89bddce5 | 1168 | static int sock_mmap(struct file *file, struct vm_area_struct *vma) |
1da177e4 | 1169 | { |
b69aee04 | 1170 | struct socket *sock = file->private_data; |
1da177e4 LT |
1171 | |
1172 | return sock->ops->mmap(file, sock, vma); | |
1173 | } | |
1174 | ||
20380731 | 1175 | static int sock_close(struct inode *inode, struct file *filp) |
1da177e4 LT |
1176 | { |
1177 | /* | |
89bddce5 SH |
1178 | * It was possible the inode is NULL we were |
1179 | * closing an unfinished socket. | |
1da177e4 LT |
1180 | */ |
1181 | ||
89bddce5 | 1182 | if (!inode) { |
1da177e4 LT |
1183 | printk(KERN_DEBUG "sock_close: NULL inode\n"); |
1184 | return 0; | |
1185 | } | |
1da177e4 LT |
1186 | sock_release(SOCKET_I(inode)); |
1187 | return 0; | |
1188 | } | |
1189 | ||
1190 | /* | |
1191 | * Update the socket async list | |
1192 | * | |
1193 | * Fasync_list locking strategy. | |
1194 | * | |
1195 | * 1. fasync_list is modified only under process context socket lock | |
1196 | * i.e. under semaphore. | |
1197 | * 2. fasync_list is used under read_lock(&sk->sk_callback_lock) | |
989a2979 | 1198 | * or under socket lock |
1da177e4 LT |
1199 | */ |
1200 | ||
1201 | static int sock_fasync(int fd, struct file *filp, int on) | |
1202 | { | |
989a2979 ED |
1203 | struct socket *sock = filp->private_data; |
1204 | struct sock *sk = sock->sk; | |
eaefd110 | 1205 | struct socket_wq *wq; |
1da177e4 | 1206 | |
989a2979 | 1207 | if (sk == NULL) |
1da177e4 | 1208 | return -EINVAL; |
1da177e4 LT |
1209 | |
1210 | lock_sock(sk); | |
eaefd110 ED |
1211 | wq = rcu_dereference_protected(sock->wq, sock_owned_by_user(sk)); |
1212 | fasync_helper(fd, filp, on, &wq->fasync_list); | |
1da177e4 | 1213 | |
eaefd110 | 1214 | if (!wq->fasync_list) |
989a2979 ED |
1215 | sock_reset_flag(sk, SOCK_FASYNC); |
1216 | else | |
bcdce719 | 1217 | sock_set_flag(sk, SOCK_FASYNC); |
1da177e4 | 1218 | |
989a2979 | 1219 | release_sock(sk); |
1da177e4 LT |
1220 | return 0; |
1221 | } | |
1222 | ||
43815482 | 1223 | /* This function may be called only under socket lock or callback_lock or rcu_lock */ |
1da177e4 LT |
1224 | |
1225 | int sock_wake_async(struct socket *sock, int how, int band) | |
1226 | { | |
43815482 ED |
1227 | struct socket_wq *wq; |
1228 | ||
1229 | if (!sock) | |
1230 | return -1; | |
1231 | rcu_read_lock(); | |
1232 | wq = rcu_dereference(sock->wq); | |
1233 | if (!wq || !wq->fasync_list) { | |
1234 | rcu_read_unlock(); | |
1da177e4 | 1235 | return -1; |
43815482 | 1236 | } |
89bddce5 | 1237 | switch (how) { |
8d8ad9d7 | 1238 | case SOCK_WAKE_WAITD: |
1da177e4 LT |
1239 | if (test_bit(SOCK_ASYNC_WAITDATA, &sock->flags)) |
1240 | break; | |
1241 | goto call_kill; | |
8d8ad9d7 | 1242 | case SOCK_WAKE_SPACE: |
1da177e4 LT |
1243 | if (!test_and_clear_bit(SOCK_ASYNC_NOSPACE, &sock->flags)) |
1244 | break; | |
1245 | /* fall through */ | |
8d8ad9d7 | 1246 | case SOCK_WAKE_IO: |
89bddce5 | 1247 | call_kill: |
43815482 | 1248 | kill_fasync(&wq->fasync_list, SIGIO, band); |
1da177e4 | 1249 | break; |
8d8ad9d7 | 1250 | case SOCK_WAKE_URG: |
43815482 | 1251 | kill_fasync(&wq->fasync_list, SIGURG, band); |
1da177e4 | 1252 | } |
43815482 | 1253 | rcu_read_unlock(); |
1da177e4 LT |
1254 | return 0; |
1255 | } | |
c6d409cf | 1256 | EXPORT_SYMBOL(sock_wake_async); |
1da177e4 | 1257 | |
721db93a | 1258 | int __sock_create(struct net *net, int family, int type, int protocol, |
89bddce5 | 1259 | struct socket **res, int kern) |
1da177e4 LT |
1260 | { |
1261 | int err; | |
1262 | struct socket *sock; | |
55737fda | 1263 | const struct net_proto_family *pf; |
1da177e4 LT |
1264 | |
1265 | /* | |
89bddce5 | 1266 | * Check protocol is in range |
1da177e4 LT |
1267 | */ |
1268 | if (family < 0 || family >= NPROTO) | |
1269 | return -EAFNOSUPPORT; | |
1270 | if (type < 0 || type >= SOCK_MAX) | |
1271 | return -EINVAL; | |
1272 | ||
1273 | /* Compatibility. | |
1274 | ||
1275 | This uglymoron is moved from INET layer to here to avoid | |
1276 | deadlock in module load. | |
1277 | */ | |
1278 | if (family == PF_INET && type == SOCK_PACKET) { | |
89bddce5 | 1279 | static int warned; |
1da177e4 LT |
1280 | if (!warned) { |
1281 | warned = 1; | |
89bddce5 SH |
1282 | printk(KERN_INFO "%s uses obsolete (PF_INET,SOCK_PACKET)\n", |
1283 | current->comm); | |
1da177e4 LT |
1284 | } |
1285 | family = PF_PACKET; | |
1286 | } | |
1287 | ||
1288 | err = security_socket_create(family, type, protocol, kern); | |
1289 | if (err) | |
1290 | return err; | |
89bddce5 | 1291 | |
55737fda SH |
1292 | /* |
1293 | * Allocate the socket and allow the family to set things up. if | |
1294 | * the protocol is 0, the family is instructed to select an appropriate | |
1295 | * default. | |
1296 | */ | |
1297 | sock = sock_alloc(); | |
1298 | if (!sock) { | |
e87cc472 | 1299 | net_warn_ratelimited("socket: no more sockets\n"); |
55737fda SH |
1300 | return -ENFILE; /* Not exactly a match, but its the |
1301 | closest posix thing */ | |
1302 | } | |
1303 | ||
1304 | sock->type = type; | |
1305 | ||
95a5afca | 1306 | #ifdef CONFIG_MODULES |
89bddce5 SH |
1307 | /* Attempt to load a protocol module if the find failed. |
1308 | * | |
1309 | * 12/09/1996 Marcin: But! this makes REALLY only sense, if the user | |
1da177e4 LT |
1310 | * requested real, full-featured networking support upon configuration. |
1311 | * Otherwise module support will break! | |
1312 | */ | |
190683a9 | 1313 | if (rcu_access_pointer(net_families[family]) == NULL) |
89bddce5 | 1314 | request_module("net-pf-%d", family); |
1da177e4 LT |
1315 | #endif |
1316 | ||
55737fda SH |
1317 | rcu_read_lock(); |
1318 | pf = rcu_dereference(net_families[family]); | |
1319 | err = -EAFNOSUPPORT; | |
1320 | if (!pf) | |
1321 | goto out_release; | |
1da177e4 LT |
1322 | |
1323 | /* | |
1324 | * We will call the ->create function, that possibly is in a loadable | |
1325 | * module, so we have to bump that loadable module refcnt first. | |
1326 | */ | |
55737fda | 1327 | if (!try_module_get(pf->owner)) |
1da177e4 LT |
1328 | goto out_release; |
1329 | ||
55737fda SH |
1330 | /* Now protected by module ref count */ |
1331 | rcu_read_unlock(); | |
1332 | ||
3f378b68 | 1333 | err = pf->create(net, sock, protocol, kern); |
55737fda | 1334 | if (err < 0) |
1da177e4 | 1335 | goto out_module_put; |
a79af59e | 1336 | |
1da177e4 LT |
1337 | /* |
1338 | * Now to bump the refcnt of the [loadable] module that owns this | |
1339 | * socket at sock_release time we decrement its refcnt. | |
1340 | */ | |
55737fda SH |
1341 | if (!try_module_get(sock->ops->owner)) |
1342 | goto out_module_busy; | |
1343 | ||
1da177e4 LT |
1344 | /* |
1345 | * Now that we're done with the ->create function, the [loadable] | |
1346 | * module can have its refcnt decremented | |
1347 | */ | |
55737fda | 1348 | module_put(pf->owner); |
7420ed23 VY |
1349 | err = security_socket_post_create(sock, family, type, protocol, kern); |
1350 | if (err) | |
3b185525 | 1351 | goto out_sock_release; |
55737fda | 1352 | *res = sock; |
1da177e4 | 1353 | |
55737fda SH |
1354 | return 0; |
1355 | ||
1356 | out_module_busy: | |
1357 | err = -EAFNOSUPPORT; | |
1da177e4 | 1358 | out_module_put: |
55737fda SH |
1359 | sock->ops = NULL; |
1360 | module_put(pf->owner); | |
1361 | out_sock_release: | |
1da177e4 | 1362 | sock_release(sock); |
55737fda SH |
1363 | return err; |
1364 | ||
1365 | out_release: | |
1366 | rcu_read_unlock(); | |
1367 | goto out_sock_release; | |
1da177e4 | 1368 | } |
721db93a | 1369 | EXPORT_SYMBOL(__sock_create); |
1da177e4 LT |
1370 | |
1371 | int sock_create(int family, int type, int protocol, struct socket **res) | |
1372 | { | |
1b8d7ae4 | 1373 | return __sock_create(current->nsproxy->net_ns, family, type, protocol, res, 0); |
1da177e4 | 1374 | } |
c6d409cf | 1375 | EXPORT_SYMBOL(sock_create); |
1da177e4 LT |
1376 | |
1377 | int sock_create_kern(int family, int type, int protocol, struct socket **res) | |
1378 | { | |
1b8d7ae4 | 1379 | return __sock_create(&init_net, family, type, protocol, res, 1); |
1da177e4 | 1380 | } |
c6d409cf | 1381 | EXPORT_SYMBOL(sock_create_kern); |
1da177e4 | 1382 | |
3e0fa65f | 1383 | SYSCALL_DEFINE3(socket, int, family, int, type, int, protocol) |
1da177e4 LT |
1384 | { |
1385 | int retval; | |
1386 | struct socket *sock; | |
a677a039 UD |
1387 | int flags; |
1388 | ||
e38b36f3 UD |
1389 | /* Check the SOCK_* constants for consistency. */ |
1390 | BUILD_BUG_ON(SOCK_CLOEXEC != O_CLOEXEC); | |
1391 | BUILD_BUG_ON((SOCK_MAX | SOCK_TYPE_MASK) != SOCK_TYPE_MASK); | |
1392 | BUILD_BUG_ON(SOCK_CLOEXEC & SOCK_TYPE_MASK); | |
1393 | BUILD_BUG_ON(SOCK_NONBLOCK & SOCK_TYPE_MASK); | |
1394 | ||
a677a039 | 1395 | flags = type & ~SOCK_TYPE_MASK; |
77d27200 | 1396 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1397 | return -EINVAL; |
1398 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1399 | |
aaca0bdc UD |
1400 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1401 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1402 | ||
1da177e4 LT |
1403 | retval = sock_create(family, type, protocol, &sock); |
1404 | if (retval < 0) | |
1405 | goto out; | |
1406 | ||
77d27200 | 1407 | retval = sock_map_fd(sock, flags & (O_CLOEXEC | O_NONBLOCK)); |
1da177e4 LT |
1408 | if (retval < 0) |
1409 | goto out_release; | |
1410 | ||
1411 | out: | |
1412 | /* It may be already another descriptor 8) Not kernel problem. */ | |
1413 | return retval; | |
1414 | ||
1415 | out_release: | |
1416 | sock_release(sock); | |
1417 | return retval; | |
1418 | } | |
1419 | ||
1420 | /* | |
1421 | * Create a pair of connected sockets. | |
1422 | */ | |
1423 | ||
3e0fa65f HC |
1424 | SYSCALL_DEFINE4(socketpair, int, family, int, type, int, protocol, |
1425 | int __user *, usockvec) | |
1da177e4 LT |
1426 | { |
1427 | struct socket *sock1, *sock2; | |
1428 | int fd1, fd2, err; | |
db349509 | 1429 | struct file *newfile1, *newfile2; |
a677a039 UD |
1430 | int flags; |
1431 | ||
1432 | flags = type & ~SOCK_TYPE_MASK; | |
77d27200 | 1433 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1434 | return -EINVAL; |
1435 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1436 | |
aaca0bdc UD |
1437 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1438 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1439 | ||
1da177e4 LT |
1440 | /* |
1441 | * Obtain the first socket and check if the underlying protocol | |
1442 | * supports the socketpair call. | |
1443 | */ | |
1444 | ||
1445 | err = sock_create(family, type, protocol, &sock1); | |
1446 | if (err < 0) | |
1447 | goto out; | |
1448 | ||
1449 | err = sock_create(family, type, protocol, &sock2); | |
1450 | if (err < 0) | |
1451 | goto out_release_1; | |
1452 | ||
1453 | err = sock1->ops->socketpair(sock1, sock2); | |
89bddce5 | 1454 | if (err < 0) |
1da177e4 LT |
1455 | goto out_release_both; |
1456 | ||
28407630 | 1457 | fd1 = get_unused_fd_flags(flags); |
bf3c23d1 DM |
1458 | if (unlikely(fd1 < 0)) { |
1459 | err = fd1; | |
db349509 | 1460 | goto out_release_both; |
bf3c23d1 | 1461 | } |
28407630 | 1462 | fd2 = get_unused_fd_flags(flags); |
198de4d7 AV |
1463 | if (unlikely(fd2 < 0)) { |
1464 | err = fd2; | |
28407630 AV |
1465 | put_unused_fd(fd1); |
1466 | goto out_release_both; | |
1467 | } | |
1468 | ||
aab174f0 | 1469 | newfile1 = sock_alloc_file(sock1, flags, NULL); |
28407630 AV |
1470 | if (unlikely(IS_ERR(newfile1))) { |
1471 | err = PTR_ERR(newfile1); | |
1472 | put_unused_fd(fd1); | |
1473 | put_unused_fd(fd2); | |
1474 | goto out_release_both; | |
1475 | } | |
1476 | ||
aab174f0 | 1477 | newfile2 = sock_alloc_file(sock2, flags, NULL); |
28407630 AV |
1478 | if (IS_ERR(newfile2)) { |
1479 | err = PTR_ERR(newfile2); | |
198de4d7 AV |
1480 | fput(newfile1); |
1481 | put_unused_fd(fd1); | |
28407630 | 1482 | put_unused_fd(fd2); |
198de4d7 AV |
1483 | sock_release(sock2); |
1484 | goto out; | |
db349509 AV |
1485 | } |
1486 | ||
157cf649 | 1487 | audit_fd_pair(fd1, fd2); |
db349509 AV |
1488 | fd_install(fd1, newfile1); |
1489 | fd_install(fd2, newfile2); | |
1da177e4 LT |
1490 | /* fd1 and fd2 may be already another descriptors. |
1491 | * Not kernel problem. | |
1492 | */ | |
1493 | ||
89bddce5 | 1494 | err = put_user(fd1, &usockvec[0]); |
1da177e4 LT |
1495 | if (!err) |
1496 | err = put_user(fd2, &usockvec[1]); | |
1497 | if (!err) | |
1498 | return 0; | |
1499 | ||
1500 | sys_close(fd2); | |
1501 | sys_close(fd1); | |
1502 | return err; | |
1503 | ||
1da177e4 | 1504 | out_release_both: |
89bddce5 | 1505 | sock_release(sock2); |
1da177e4 | 1506 | out_release_1: |
89bddce5 | 1507 | sock_release(sock1); |
1da177e4 LT |
1508 | out: |
1509 | return err; | |
1510 | } | |
1511 | ||
1da177e4 LT |
1512 | /* |
1513 | * Bind a name to a socket. Nothing much to do here since it's | |
1514 | * the protocol's responsibility to handle the local address. | |
1515 | * | |
1516 | * We move the socket address to kernel space before we call | |
1517 | * the protocol layer (having also checked the address is ok). | |
1518 | */ | |
1519 | ||
20f37034 | 1520 | SYSCALL_DEFINE3(bind, int, fd, struct sockaddr __user *, umyaddr, int, addrlen) |
1da177e4 LT |
1521 | { |
1522 | struct socket *sock; | |
230b1839 | 1523 | struct sockaddr_storage address; |
6cb153ca | 1524 | int err, fput_needed; |
1da177e4 | 1525 | |
89bddce5 | 1526 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
e71a4783 | 1527 | if (sock) { |
43db362d | 1528 | err = move_addr_to_kernel(umyaddr, addrlen, &address); |
89bddce5 SH |
1529 | if (err >= 0) { |
1530 | err = security_socket_bind(sock, | |
230b1839 | 1531 | (struct sockaddr *)&address, |
89bddce5 | 1532 | addrlen); |
6cb153ca BL |
1533 | if (!err) |
1534 | err = sock->ops->bind(sock, | |
89bddce5 | 1535 | (struct sockaddr *) |
230b1839 | 1536 | &address, addrlen); |
1da177e4 | 1537 | } |
6cb153ca | 1538 | fput_light(sock->file, fput_needed); |
89bddce5 | 1539 | } |
1da177e4 LT |
1540 | return err; |
1541 | } | |
1542 | ||
1da177e4 LT |
1543 | /* |
1544 | * Perform a listen. Basically, we allow the protocol to do anything | |
1545 | * necessary for a listen, and if that works, we mark the socket as | |
1546 | * ready for listening. | |
1547 | */ | |
1548 | ||
3e0fa65f | 1549 | SYSCALL_DEFINE2(listen, int, fd, int, backlog) |
1da177e4 LT |
1550 | { |
1551 | struct socket *sock; | |
6cb153ca | 1552 | int err, fput_needed; |
b8e1f9b5 | 1553 | int somaxconn; |
89bddce5 SH |
1554 | |
1555 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
1556 | if (sock) { | |
8efa6e93 | 1557 | somaxconn = sock_net(sock->sk)->core.sysctl_somaxconn; |
95c96174 | 1558 | if ((unsigned int)backlog > somaxconn) |
b8e1f9b5 | 1559 | backlog = somaxconn; |
1da177e4 LT |
1560 | |
1561 | err = security_socket_listen(sock, backlog); | |
6cb153ca BL |
1562 | if (!err) |
1563 | err = sock->ops->listen(sock, backlog); | |
1da177e4 | 1564 | |
6cb153ca | 1565 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1566 | } |
1567 | return err; | |
1568 | } | |
1569 | ||
1da177e4 LT |
1570 | /* |
1571 | * For accept, we attempt to create a new socket, set up the link | |
1572 | * with the client, wake up the client, then return the new | |
1573 | * connected fd. We collect the address of the connector in kernel | |
1574 | * space and move it to user at the very end. This is unclean because | |
1575 | * we open the socket then return an error. | |
1576 | * | |
1577 | * 1003.1g adds the ability to recvmsg() to query connection pending | |
1578 | * status to recvmsg. We need to add that support in a way thats | |
1579 | * clean when we restucture accept also. | |
1580 | */ | |
1581 | ||
20f37034 HC |
1582 | SYSCALL_DEFINE4(accept4, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1583 | int __user *, upeer_addrlen, int, flags) | |
1da177e4 LT |
1584 | { |
1585 | struct socket *sock, *newsock; | |
39d8c1b6 | 1586 | struct file *newfile; |
6cb153ca | 1587 | int err, len, newfd, fput_needed; |
230b1839 | 1588 | struct sockaddr_storage address; |
1da177e4 | 1589 | |
77d27200 | 1590 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
aaca0bdc UD |
1591 | return -EINVAL; |
1592 | ||
1593 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) | |
1594 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1595 | ||
6cb153ca | 1596 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1597 | if (!sock) |
1598 | goto out; | |
1599 | ||
1600 | err = -ENFILE; | |
c6d409cf ED |
1601 | newsock = sock_alloc(); |
1602 | if (!newsock) | |
1da177e4 LT |
1603 | goto out_put; |
1604 | ||
1605 | newsock->type = sock->type; | |
1606 | newsock->ops = sock->ops; | |
1607 | ||
1da177e4 LT |
1608 | /* |
1609 | * We don't need try_module_get here, as the listening socket (sock) | |
1610 | * has the protocol module (sock->ops->owner) held. | |
1611 | */ | |
1612 | __module_get(newsock->ops->owner); | |
1613 | ||
28407630 | 1614 | newfd = get_unused_fd_flags(flags); |
39d8c1b6 DM |
1615 | if (unlikely(newfd < 0)) { |
1616 | err = newfd; | |
9a1875e6 DM |
1617 | sock_release(newsock); |
1618 | goto out_put; | |
39d8c1b6 | 1619 | } |
aab174f0 | 1620 | newfile = sock_alloc_file(newsock, flags, sock->sk->sk_prot_creator->name); |
28407630 AV |
1621 | if (unlikely(IS_ERR(newfile))) { |
1622 | err = PTR_ERR(newfile); | |
1623 | put_unused_fd(newfd); | |
1624 | sock_release(newsock); | |
1625 | goto out_put; | |
1626 | } | |
39d8c1b6 | 1627 | |
a79af59e FF |
1628 | err = security_socket_accept(sock, newsock); |
1629 | if (err) | |
39d8c1b6 | 1630 | goto out_fd; |
a79af59e | 1631 | |
1da177e4 LT |
1632 | err = sock->ops->accept(sock, newsock, sock->file->f_flags); |
1633 | if (err < 0) | |
39d8c1b6 | 1634 | goto out_fd; |
1da177e4 LT |
1635 | |
1636 | if (upeer_sockaddr) { | |
230b1839 | 1637 | if (newsock->ops->getname(newsock, (struct sockaddr *)&address, |
89bddce5 | 1638 | &len, 2) < 0) { |
1da177e4 | 1639 | err = -ECONNABORTED; |
39d8c1b6 | 1640 | goto out_fd; |
1da177e4 | 1641 | } |
43db362d | 1642 | err = move_addr_to_user(&address, |
230b1839 | 1643 | len, upeer_sockaddr, upeer_addrlen); |
1da177e4 | 1644 | if (err < 0) |
39d8c1b6 | 1645 | goto out_fd; |
1da177e4 LT |
1646 | } |
1647 | ||
1648 | /* File flags are not inherited via accept() unlike another OSes. */ | |
1649 | ||
39d8c1b6 DM |
1650 | fd_install(newfd, newfile); |
1651 | err = newfd; | |
1da177e4 | 1652 | |
1da177e4 | 1653 | out_put: |
6cb153ca | 1654 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1655 | out: |
1656 | return err; | |
39d8c1b6 | 1657 | out_fd: |
9606a216 | 1658 | fput(newfile); |
39d8c1b6 | 1659 | put_unused_fd(newfd); |
1da177e4 LT |
1660 | goto out_put; |
1661 | } | |
1662 | ||
20f37034 HC |
1663 | SYSCALL_DEFINE3(accept, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1664 | int __user *, upeer_addrlen) | |
aaca0bdc | 1665 | { |
de11defe | 1666 | return sys_accept4(fd, upeer_sockaddr, upeer_addrlen, 0); |
aaca0bdc UD |
1667 | } |
1668 | ||
1da177e4 LT |
1669 | /* |
1670 | * Attempt to connect to a socket with the server address. The address | |
1671 | * is in user space so we verify it is OK and move it to kernel space. | |
1672 | * | |
1673 | * For 1003.1g we need to add clean support for a bind to AF_UNSPEC to | |
1674 | * break bindings | |
1675 | * | |
1676 | * NOTE: 1003.1g draft 6.3 is broken with respect to AX.25/NetROM and | |
1677 | * other SEQPACKET protocols that take time to connect() as it doesn't | |
1678 | * include the -EINPROGRESS status for such sockets. | |
1679 | */ | |
1680 | ||
20f37034 HC |
1681 | SYSCALL_DEFINE3(connect, int, fd, struct sockaddr __user *, uservaddr, |
1682 | int, addrlen) | |
1da177e4 LT |
1683 | { |
1684 | struct socket *sock; | |
230b1839 | 1685 | struct sockaddr_storage address; |
6cb153ca | 1686 | int err, fput_needed; |
1da177e4 | 1687 | |
6cb153ca | 1688 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1689 | if (!sock) |
1690 | goto out; | |
43db362d | 1691 | err = move_addr_to_kernel(uservaddr, addrlen, &address); |
1da177e4 LT |
1692 | if (err < 0) |
1693 | goto out_put; | |
1694 | ||
89bddce5 | 1695 | err = |
230b1839 | 1696 | security_socket_connect(sock, (struct sockaddr *)&address, addrlen); |
1da177e4 LT |
1697 | if (err) |
1698 | goto out_put; | |
1699 | ||
230b1839 | 1700 | err = sock->ops->connect(sock, (struct sockaddr *)&address, addrlen, |
1da177e4 LT |
1701 | sock->file->f_flags); |
1702 | out_put: | |
6cb153ca | 1703 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1704 | out: |
1705 | return err; | |
1706 | } | |
1707 | ||
1708 | /* | |
1709 | * Get the local address ('name') of a socket object. Move the obtained | |
1710 | * name to user space. | |
1711 | */ | |
1712 | ||
20f37034 HC |
1713 | SYSCALL_DEFINE3(getsockname, int, fd, struct sockaddr __user *, usockaddr, |
1714 | int __user *, usockaddr_len) | |
1da177e4 LT |
1715 | { |
1716 | struct socket *sock; | |
230b1839 | 1717 | struct sockaddr_storage address; |
6cb153ca | 1718 | int len, err, fput_needed; |
89bddce5 | 1719 | |
6cb153ca | 1720 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1721 | if (!sock) |
1722 | goto out; | |
1723 | ||
1724 | err = security_socket_getsockname(sock); | |
1725 | if (err) | |
1726 | goto out_put; | |
1727 | ||
230b1839 | 1728 | err = sock->ops->getname(sock, (struct sockaddr *)&address, &len, 0); |
1da177e4 LT |
1729 | if (err) |
1730 | goto out_put; | |
43db362d | 1731 | err = move_addr_to_user(&address, len, usockaddr, usockaddr_len); |
1da177e4 LT |
1732 | |
1733 | out_put: | |
6cb153ca | 1734 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1735 | out: |
1736 | return err; | |
1737 | } | |
1738 | ||
1739 | /* | |
1740 | * Get the remote address ('name') of a socket object. Move the obtained | |
1741 | * name to user space. | |
1742 | */ | |
1743 | ||
20f37034 HC |
1744 | SYSCALL_DEFINE3(getpeername, int, fd, struct sockaddr __user *, usockaddr, |
1745 | int __user *, usockaddr_len) | |
1da177e4 LT |
1746 | { |
1747 | struct socket *sock; | |
230b1839 | 1748 | struct sockaddr_storage address; |
6cb153ca | 1749 | int len, err, fput_needed; |
1da177e4 | 1750 | |
89bddce5 SH |
1751 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1752 | if (sock != NULL) { | |
1da177e4 LT |
1753 | err = security_socket_getpeername(sock); |
1754 | if (err) { | |
6cb153ca | 1755 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1756 | return err; |
1757 | } | |
1758 | ||
89bddce5 | 1759 | err = |
230b1839 | 1760 | sock->ops->getname(sock, (struct sockaddr *)&address, &len, |
89bddce5 | 1761 | 1); |
1da177e4 | 1762 | if (!err) |
43db362d | 1763 | err = move_addr_to_user(&address, len, usockaddr, |
89bddce5 | 1764 | usockaddr_len); |
6cb153ca | 1765 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1766 | } |
1767 | return err; | |
1768 | } | |
1769 | ||
1770 | /* | |
1771 | * Send a datagram to a given address. We move the address into kernel | |
1772 | * space and check the user space data area is readable before invoking | |
1773 | * the protocol. | |
1774 | */ | |
1775 | ||
3e0fa65f | 1776 | SYSCALL_DEFINE6(sendto, int, fd, void __user *, buff, size_t, len, |
95c96174 | 1777 | unsigned int, flags, struct sockaddr __user *, addr, |
3e0fa65f | 1778 | int, addr_len) |
1da177e4 LT |
1779 | { |
1780 | struct socket *sock; | |
230b1839 | 1781 | struct sockaddr_storage address; |
1da177e4 LT |
1782 | int err; |
1783 | struct msghdr msg; | |
1784 | struct iovec iov; | |
6cb153ca | 1785 | int fput_needed; |
6cb153ca | 1786 | |
253eacc0 LT |
1787 | if (len > INT_MAX) |
1788 | len = INT_MAX; | |
de0fa95c PE |
1789 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1790 | if (!sock) | |
4387ff75 | 1791 | goto out; |
6cb153ca | 1792 | |
89bddce5 SH |
1793 | iov.iov_base = buff; |
1794 | iov.iov_len = len; | |
1795 | msg.msg_name = NULL; | |
1796 | msg.msg_iov = &iov; | |
1797 | msg.msg_iovlen = 1; | |
1798 | msg.msg_control = NULL; | |
1799 | msg.msg_controllen = 0; | |
1800 | msg.msg_namelen = 0; | |
6cb153ca | 1801 | if (addr) { |
43db362d | 1802 | err = move_addr_to_kernel(addr, addr_len, &address); |
1da177e4 LT |
1803 | if (err < 0) |
1804 | goto out_put; | |
230b1839 | 1805 | msg.msg_name = (struct sockaddr *)&address; |
89bddce5 | 1806 | msg.msg_namelen = addr_len; |
1da177e4 LT |
1807 | } |
1808 | if (sock->file->f_flags & O_NONBLOCK) | |
1809 | flags |= MSG_DONTWAIT; | |
1810 | msg.msg_flags = flags; | |
1811 | err = sock_sendmsg(sock, &msg, len); | |
1812 | ||
89bddce5 | 1813 | out_put: |
de0fa95c | 1814 | fput_light(sock->file, fput_needed); |
4387ff75 | 1815 | out: |
1da177e4 LT |
1816 | return err; |
1817 | } | |
1818 | ||
1819 | /* | |
89bddce5 | 1820 | * Send a datagram down a socket. |
1da177e4 LT |
1821 | */ |
1822 | ||
3e0fa65f | 1823 | SYSCALL_DEFINE4(send, int, fd, void __user *, buff, size_t, len, |
95c96174 | 1824 | unsigned int, flags) |
1da177e4 LT |
1825 | { |
1826 | return sys_sendto(fd, buff, len, flags, NULL, 0); | |
1827 | } | |
1828 | ||
1829 | /* | |
89bddce5 | 1830 | * Receive a frame from the socket and optionally record the address of the |
1da177e4 LT |
1831 | * sender. We verify the buffers are writable and if needed move the |
1832 | * sender address from kernel to user space. | |
1833 | */ | |
1834 | ||
3e0fa65f | 1835 | SYSCALL_DEFINE6(recvfrom, int, fd, void __user *, ubuf, size_t, size, |
95c96174 | 1836 | unsigned int, flags, struct sockaddr __user *, addr, |
3e0fa65f | 1837 | int __user *, addr_len) |
1da177e4 LT |
1838 | { |
1839 | struct socket *sock; | |
1840 | struct iovec iov; | |
1841 | struct msghdr msg; | |
230b1839 | 1842 | struct sockaddr_storage address; |
89bddce5 | 1843 | int err, err2; |
6cb153ca BL |
1844 | int fput_needed; |
1845 | ||
253eacc0 LT |
1846 | if (size > INT_MAX) |
1847 | size = INT_MAX; | |
de0fa95c | 1848 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 | 1849 | if (!sock) |
de0fa95c | 1850 | goto out; |
1da177e4 | 1851 | |
89bddce5 SH |
1852 | msg.msg_control = NULL; |
1853 | msg.msg_controllen = 0; | |
1854 | msg.msg_iovlen = 1; | |
1855 | msg.msg_iov = &iov; | |
1856 | iov.iov_len = size; | |
1857 | iov.iov_base = ubuf; | |
230b1839 YH |
1858 | msg.msg_name = (struct sockaddr *)&address; |
1859 | msg.msg_namelen = sizeof(address); | |
1da177e4 LT |
1860 | if (sock->file->f_flags & O_NONBLOCK) |
1861 | flags |= MSG_DONTWAIT; | |
89bddce5 | 1862 | err = sock_recvmsg(sock, &msg, size, flags); |
1da177e4 | 1863 | |
89bddce5 | 1864 | if (err >= 0 && addr != NULL) { |
43db362d | 1865 | err2 = move_addr_to_user(&address, |
230b1839 | 1866 | msg.msg_namelen, addr, addr_len); |
89bddce5 SH |
1867 | if (err2 < 0) |
1868 | err = err2; | |
1da177e4 | 1869 | } |
de0fa95c PE |
1870 | |
1871 | fput_light(sock->file, fput_needed); | |
4387ff75 | 1872 | out: |
1da177e4 LT |
1873 | return err; |
1874 | } | |
1875 | ||
1876 | /* | |
89bddce5 | 1877 | * Receive a datagram from a socket. |
1da177e4 LT |
1878 | */ |
1879 | ||
89bddce5 | 1880 | asmlinkage long sys_recv(int fd, void __user *ubuf, size_t size, |
95c96174 | 1881 | unsigned int flags) |
1da177e4 LT |
1882 | { |
1883 | return sys_recvfrom(fd, ubuf, size, flags, NULL, NULL); | |
1884 | } | |
1885 | ||
1886 | /* | |
1887 | * Set a socket option. Because we don't know the option lengths we have | |
1888 | * to pass the user mode parameter for the protocols to sort out. | |
1889 | */ | |
1890 | ||
20f37034 HC |
1891 | SYSCALL_DEFINE5(setsockopt, int, fd, int, level, int, optname, |
1892 | char __user *, optval, int, optlen) | |
1da177e4 | 1893 | { |
6cb153ca | 1894 | int err, fput_needed; |
1da177e4 LT |
1895 | struct socket *sock; |
1896 | ||
1897 | if (optlen < 0) | |
1898 | return -EINVAL; | |
89bddce5 SH |
1899 | |
1900 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
1901 | if (sock != NULL) { | |
1902 | err = security_socket_setsockopt(sock, level, optname); | |
6cb153ca BL |
1903 | if (err) |
1904 | goto out_put; | |
1da177e4 LT |
1905 | |
1906 | if (level == SOL_SOCKET) | |
89bddce5 SH |
1907 | err = |
1908 | sock_setsockopt(sock, level, optname, optval, | |
1909 | optlen); | |
1da177e4 | 1910 | else |
89bddce5 SH |
1911 | err = |
1912 | sock->ops->setsockopt(sock, level, optname, optval, | |
1913 | optlen); | |
6cb153ca BL |
1914 | out_put: |
1915 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
1916 | } |
1917 | return err; | |
1918 | } | |
1919 | ||
1920 | /* | |
1921 | * Get a socket option. Because we don't know the option lengths we have | |
1922 | * to pass a user mode parameter for the protocols to sort out. | |
1923 | */ | |
1924 | ||
20f37034 HC |
1925 | SYSCALL_DEFINE5(getsockopt, int, fd, int, level, int, optname, |
1926 | char __user *, optval, int __user *, optlen) | |
1da177e4 | 1927 | { |
6cb153ca | 1928 | int err, fput_needed; |
1da177e4 LT |
1929 | struct socket *sock; |
1930 | ||
89bddce5 SH |
1931 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1932 | if (sock != NULL) { | |
6cb153ca BL |
1933 | err = security_socket_getsockopt(sock, level, optname); |
1934 | if (err) | |
1935 | goto out_put; | |
1da177e4 LT |
1936 | |
1937 | if (level == SOL_SOCKET) | |
89bddce5 SH |
1938 | err = |
1939 | sock_getsockopt(sock, level, optname, optval, | |
1940 | optlen); | |
1da177e4 | 1941 | else |
89bddce5 SH |
1942 | err = |
1943 | sock->ops->getsockopt(sock, level, optname, optval, | |
1944 | optlen); | |
6cb153ca BL |
1945 | out_put: |
1946 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
1947 | } |
1948 | return err; | |
1949 | } | |
1950 | ||
1da177e4 LT |
1951 | /* |
1952 | * Shutdown a socket. | |
1953 | */ | |
1954 | ||
754fe8d2 | 1955 | SYSCALL_DEFINE2(shutdown, int, fd, int, how) |
1da177e4 | 1956 | { |
6cb153ca | 1957 | int err, fput_needed; |
1da177e4 LT |
1958 | struct socket *sock; |
1959 | ||
89bddce5 SH |
1960 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1961 | if (sock != NULL) { | |
1da177e4 | 1962 | err = security_socket_shutdown(sock, how); |
6cb153ca BL |
1963 | if (!err) |
1964 | err = sock->ops->shutdown(sock, how); | |
1965 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
1966 | } |
1967 | return err; | |
1968 | } | |
1969 | ||
89bddce5 | 1970 | /* A couple of helpful macros for getting the address of the 32/64 bit |
1da177e4 LT |
1971 | * fields which are the same type (int / unsigned) on our platforms. |
1972 | */ | |
1973 | #define COMPAT_MSG(msg, member) ((MSG_CMSG_COMPAT & flags) ? &msg##_compat->member : &msg->member) | |
1974 | #define COMPAT_NAMELEN(msg) COMPAT_MSG(msg, msg_namelen) | |
1975 | #define COMPAT_FLAGS(msg) COMPAT_MSG(msg, msg_flags) | |
1976 | ||
c71d8ebe TH |
1977 | struct used_address { |
1978 | struct sockaddr_storage name; | |
1979 | unsigned int name_len; | |
1980 | }; | |
1981 | ||
228e548e | 1982 | static int __sys_sendmsg(struct socket *sock, struct msghdr __user *msg, |
95c96174 | 1983 | struct msghdr *msg_sys, unsigned int flags, |
c71d8ebe | 1984 | struct used_address *used_address) |
1da177e4 | 1985 | { |
89bddce5 SH |
1986 | struct compat_msghdr __user *msg_compat = |
1987 | (struct compat_msghdr __user *)msg; | |
230b1839 | 1988 | struct sockaddr_storage address; |
1da177e4 | 1989 | struct iovec iovstack[UIO_FASTIOV], *iov = iovstack; |
b9d717a7 | 1990 | unsigned char ctl[sizeof(struct cmsghdr) + 20] |
89bddce5 SH |
1991 | __attribute__ ((aligned(sizeof(__kernel_size_t)))); |
1992 | /* 20 is size of ipv6_pktinfo */ | |
1da177e4 | 1993 | unsigned char *ctl_buf = ctl; |
a74e9106 | 1994 | int err, ctl_len, total_len; |
89bddce5 | 1995 | |
1da177e4 LT |
1996 | err = -EFAULT; |
1997 | if (MSG_CMSG_COMPAT & flags) { | |
228e548e | 1998 | if (get_compat_msghdr(msg_sys, msg_compat)) |
1da177e4 | 1999 | return -EFAULT; |
228e548e | 2000 | } else if (copy_from_user(msg_sys, msg, sizeof(struct msghdr))) |
1da177e4 LT |
2001 | return -EFAULT; |
2002 | ||
228e548e | 2003 | if (msg_sys->msg_iovlen > UIO_FASTIOV) { |
a74e9106 ED |
2004 | err = -EMSGSIZE; |
2005 | if (msg_sys->msg_iovlen > UIO_MAXIOV) | |
2006 | goto out; | |
2007 | err = -ENOMEM; | |
2008 | iov = kmalloc(msg_sys->msg_iovlen * sizeof(struct iovec), | |
2009 | GFP_KERNEL); | |
1da177e4 | 2010 | if (!iov) |
228e548e | 2011 | goto out; |
1da177e4 LT |
2012 | } |
2013 | ||
2014 | /* This will also move the address data into kernel space */ | |
2015 | if (MSG_CMSG_COMPAT & flags) { | |
43db362d | 2016 | err = verify_compat_iovec(msg_sys, iov, &address, VERIFY_READ); |
1da177e4 | 2017 | } else |
43db362d | 2018 | err = verify_iovec(msg_sys, iov, &address, VERIFY_READ); |
89bddce5 | 2019 | if (err < 0) |
1da177e4 LT |
2020 | goto out_freeiov; |
2021 | total_len = err; | |
2022 | ||
2023 | err = -ENOBUFS; | |
2024 | ||
228e548e | 2025 | if (msg_sys->msg_controllen > INT_MAX) |
1da177e4 | 2026 | goto out_freeiov; |
228e548e | 2027 | ctl_len = msg_sys->msg_controllen; |
1da177e4 | 2028 | if ((MSG_CMSG_COMPAT & flags) && ctl_len) { |
89bddce5 | 2029 | err = |
228e548e | 2030 | cmsghdr_from_user_compat_to_kern(msg_sys, sock->sk, ctl, |
89bddce5 | 2031 | sizeof(ctl)); |
1da177e4 LT |
2032 | if (err) |
2033 | goto out_freeiov; | |
228e548e AB |
2034 | ctl_buf = msg_sys->msg_control; |
2035 | ctl_len = msg_sys->msg_controllen; | |
1da177e4 | 2036 | } else if (ctl_len) { |
89bddce5 | 2037 | if (ctl_len > sizeof(ctl)) { |
1da177e4 | 2038 | ctl_buf = sock_kmalloc(sock->sk, ctl_len, GFP_KERNEL); |
89bddce5 | 2039 | if (ctl_buf == NULL) |
1da177e4 LT |
2040 | goto out_freeiov; |
2041 | } | |
2042 | err = -EFAULT; | |
2043 | /* | |
228e548e | 2044 | * Careful! Before this, msg_sys->msg_control contains a user pointer. |
1da177e4 LT |
2045 | * Afterwards, it will be a kernel pointer. Thus the compiler-assisted |
2046 | * checking falls down on this. | |
2047 | */ | |
fb8621bb | 2048 | if (copy_from_user(ctl_buf, |
228e548e | 2049 | (void __user __force *)msg_sys->msg_control, |
89bddce5 | 2050 | ctl_len)) |
1da177e4 | 2051 | goto out_freectl; |
228e548e | 2052 | msg_sys->msg_control = ctl_buf; |
1da177e4 | 2053 | } |
228e548e | 2054 | msg_sys->msg_flags = flags; |
1da177e4 LT |
2055 | |
2056 | if (sock->file->f_flags & O_NONBLOCK) | |
228e548e | 2057 | msg_sys->msg_flags |= MSG_DONTWAIT; |
c71d8ebe TH |
2058 | /* |
2059 | * If this is sendmmsg() and current destination address is same as | |
2060 | * previously succeeded address, omit asking LSM's decision. | |
2061 | * used_address->name_len is initialized to UINT_MAX so that the first | |
2062 | * destination address never matches. | |
2063 | */ | |
bc909d9d MD |
2064 | if (used_address && msg_sys->msg_name && |
2065 | used_address->name_len == msg_sys->msg_namelen && | |
2066 | !memcmp(&used_address->name, msg_sys->msg_name, | |
c71d8ebe TH |
2067 | used_address->name_len)) { |
2068 | err = sock_sendmsg_nosec(sock, msg_sys, total_len); | |
2069 | goto out_freectl; | |
2070 | } | |
2071 | err = sock_sendmsg(sock, msg_sys, total_len); | |
2072 | /* | |
2073 | * If this is sendmmsg() and sending to current destination address was | |
2074 | * successful, remember it. | |
2075 | */ | |
2076 | if (used_address && err >= 0) { | |
2077 | used_address->name_len = msg_sys->msg_namelen; | |
bc909d9d MD |
2078 | if (msg_sys->msg_name) |
2079 | memcpy(&used_address->name, msg_sys->msg_name, | |
2080 | used_address->name_len); | |
c71d8ebe | 2081 | } |
1da177e4 LT |
2082 | |
2083 | out_freectl: | |
89bddce5 | 2084 | if (ctl_buf != ctl) |
1da177e4 LT |
2085 | sock_kfree_s(sock->sk, ctl_buf, ctl_len); |
2086 | out_freeiov: | |
2087 | if (iov != iovstack) | |
a74e9106 | 2088 | kfree(iov); |
228e548e AB |
2089 | out: |
2090 | return err; | |
2091 | } | |
2092 | ||
2093 | /* | |
2094 | * BSD sendmsg interface | |
2095 | */ | |
2096 | ||
95c96174 | 2097 | SYSCALL_DEFINE3(sendmsg, int, fd, struct msghdr __user *, msg, unsigned int, flags) |
228e548e AB |
2098 | { |
2099 | int fput_needed, err; | |
2100 | struct msghdr msg_sys; | |
2101 | struct socket *sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2102 | ||
2103 | if (!sock) | |
2104 | goto out; | |
2105 | ||
c71d8ebe | 2106 | err = __sys_sendmsg(sock, msg, &msg_sys, flags, NULL); |
228e548e | 2107 | |
6cb153ca | 2108 | fput_light(sock->file, fput_needed); |
89bddce5 | 2109 | out: |
1da177e4 LT |
2110 | return err; |
2111 | } | |
2112 | ||
228e548e AB |
2113 | /* |
2114 | * Linux sendmmsg interface | |
2115 | */ | |
2116 | ||
2117 | int __sys_sendmmsg(int fd, struct mmsghdr __user *mmsg, unsigned int vlen, | |
2118 | unsigned int flags) | |
2119 | { | |
2120 | int fput_needed, err, datagrams; | |
2121 | struct socket *sock; | |
2122 | struct mmsghdr __user *entry; | |
2123 | struct compat_mmsghdr __user *compat_entry; | |
2124 | struct msghdr msg_sys; | |
c71d8ebe | 2125 | struct used_address used_address; |
228e548e | 2126 | |
98382f41 AB |
2127 | if (vlen > UIO_MAXIOV) |
2128 | vlen = UIO_MAXIOV; | |
228e548e AB |
2129 | |
2130 | datagrams = 0; | |
2131 | ||
2132 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2133 | if (!sock) | |
2134 | return err; | |
2135 | ||
c71d8ebe | 2136 | used_address.name_len = UINT_MAX; |
228e548e AB |
2137 | entry = mmsg; |
2138 | compat_entry = (struct compat_mmsghdr __user *)mmsg; | |
728ffb86 | 2139 | err = 0; |
228e548e AB |
2140 | |
2141 | while (datagrams < vlen) { | |
228e548e AB |
2142 | if (MSG_CMSG_COMPAT & flags) { |
2143 | err = __sys_sendmsg(sock, (struct msghdr __user *)compat_entry, | |
c71d8ebe | 2144 | &msg_sys, flags, &used_address); |
228e548e AB |
2145 | if (err < 0) |
2146 | break; | |
2147 | err = __put_user(err, &compat_entry->msg_len); | |
2148 | ++compat_entry; | |
2149 | } else { | |
2150 | err = __sys_sendmsg(sock, (struct msghdr __user *)entry, | |
c71d8ebe | 2151 | &msg_sys, flags, &used_address); |
228e548e AB |
2152 | if (err < 0) |
2153 | break; | |
2154 | err = put_user(err, &entry->msg_len); | |
2155 | ++entry; | |
2156 | } | |
2157 | ||
2158 | if (err) | |
2159 | break; | |
2160 | ++datagrams; | |
2161 | } | |
2162 | ||
228e548e AB |
2163 | fput_light(sock->file, fput_needed); |
2164 | ||
728ffb86 AB |
2165 | /* We only return an error if no datagrams were able to be sent */ |
2166 | if (datagrams != 0) | |
228e548e AB |
2167 | return datagrams; |
2168 | ||
228e548e AB |
2169 | return err; |
2170 | } | |
2171 | ||
2172 | SYSCALL_DEFINE4(sendmmsg, int, fd, struct mmsghdr __user *, mmsg, | |
2173 | unsigned int, vlen, unsigned int, flags) | |
2174 | { | |
2175 | return __sys_sendmmsg(fd, mmsg, vlen, flags); | |
2176 | } | |
2177 | ||
a2e27255 | 2178 | static int __sys_recvmsg(struct socket *sock, struct msghdr __user *msg, |
95c96174 | 2179 | struct msghdr *msg_sys, unsigned int flags, int nosec) |
1da177e4 | 2180 | { |
89bddce5 SH |
2181 | struct compat_msghdr __user *msg_compat = |
2182 | (struct compat_msghdr __user *)msg; | |
1da177e4 | 2183 | struct iovec iovstack[UIO_FASTIOV]; |
89bddce5 | 2184 | struct iovec *iov = iovstack; |
1da177e4 | 2185 | unsigned long cmsg_ptr; |
a74e9106 | 2186 | int err, total_len, len; |
1da177e4 LT |
2187 | |
2188 | /* kernel mode address */ | |
230b1839 | 2189 | struct sockaddr_storage addr; |
1da177e4 LT |
2190 | |
2191 | /* user mode address pointers */ | |
2192 | struct sockaddr __user *uaddr; | |
2193 | int __user *uaddr_len; | |
89bddce5 | 2194 | |
1da177e4 | 2195 | if (MSG_CMSG_COMPAT & flags) { |
a2e27255 | 2196 | if (get_compat_msghdr(msg_sys, msg_compat)) |
1da177e4 | 2197 | return -EFAULT; |
c6d409cf | 2198 | } else if (copy_from_user(msg_sys, msg, sizeof(struct msghdr))) |
89bddce5 | 2199 | return -EFAULT; |
1da177e4 | 2200 | |
a2e27255 | 2201 | if (msg_sys->msg_iovlen > UIO_FASTIOV) { |
a74e9106 ED |
2202 | err = -EMSGSIZE; |
2203 | if (msg_sys->msg_iovlen > UIO_MAXIOV) | |
2204 | goto out; | |
2205 | err = -ENOMEM; | |
2206 | iov = kmalloc(msg_sys->msg_iovlen * sizeof(struct iovec), | |
2207 | GFP_KERNEL); | |
1da177e4 | 2208 | if (!iov) |
a2e27255 | 2209 | goto out; |
1da177e4 LT |
2210 | } |
2211 | ||
2212 | /* | |
89bddce5 SH |
2213 | * Save the user-mode address (verify_iovec will change the |
2214 | * kernel msghdr to use the kernel address space) | |
1da177e4 | 2215 | */ |
89bddce5 | 2216 | |
a2e27255 | 2217 | uaddr = (__force void __user *)msg_sys->msg_name; |
1da177e4 LT |
2218 | uaddr_len = COMPAT_NAMELEN(msg); |
2219 | if (MSG_CMSG_COMPAT & flags) { | |
43db362d | 2220 | err = verify_compat_iovec(msg_sys, iov, &addr, VERIFY_WRITE); |
1da177e4 | 2221 | } else |
43db362d | 2222 | err = verify_iovec(msg_sys, iov, &addr, VERIFY_WRITE); |
1da177e4 LT |
2223 | if (err < 0) |
2224 | goto out_freeiov; | |
89bddce5 | 2225 | total_len = err; |
1da177e4 | 2226 | |
a2e27255 ACM |
2227 | cmsg_ptr = (unsigned long)msg_sys->msg_control; |
2228 | msg_sys->msg_flags = flags & (MSG_CMSG_CLOEXEC|MSG_CMSG_COMPAT); | |
89bddce5 | 2229 | |
1da177e4 LT |
2230 | if (sock->file->f_flags & O_NONBLOCK) |
2231 | flags |= MSG_DONTWAIT; | |
a2e27255 ACM |
2232 | err = (nosec ? sock_recvmsg_nosec : sock_recvmsg)(sock, msg_sys, |
2233 | total_len, flags); | |
1da177e4 LT |
2234 | if (err < 0) |
2235 | goto out_freeiov; | |
2236 | len = err; | |
2237 | ||
2238 | if (uaddr != NULL) { | |
43db362d | 2239 | err = move_addr_to_user(&addr, |
a2e27255 | 2240 | msg_sys->msg_namelen, uaddr, |
89bddce5 | 2241 | uaddr_len); |
1da177e4 LT |
2242 | if (err < 0) |
2243 | goto out_freeiov; | |
2244 | } | |
a2e27255 | 2245 | err = __put_user((msg_sys->msg_flags & ~MSG_CMSG_COMPAT), |
37f7f421 | 2246 | COMPAT_FLAGS(msg)); |
1da177e4 LT |
2247 | if (err) |
2248 | goto out_freeiov; | |
2249 | if (MSG_CMSG_COMPAT & flags) | |
a2e27255 | 2250 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2251 | &msg_compat->msg_controllen); |
2252 | else | |
a2e27255 | 2253 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2254 | &msg->msg_controllen); |
2255 | if (err) | |
2256 | goto out_freeiov; | |
2257 | err = len; | |
2258 | ||
2259 | out_freeiov: | |
2260 | if (iov != iovstack) | |
a74e9106 | 2261 | kfree(iov); |
a2e27255 ACM |
2262 | out: |
2263 | return err; | |
2264 | } | |
2265 | ||
2266 | /* | |
2267 | * BSD recvmsg interface | |
2268 | */ | |
2269 | ||
2270 | SYSCALL_DEFINE3(recvmsg, int, fd, struct msghdr __user *, msg, | |
2271 | unsigned int, flags) | |
2272 | { | |
2273 | int fput_needed, err; | |
2274 | struct msghdr msg_sys; | |
2275 | struct socket *sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2276 | ||
2277 | if (!sock) | |
2278 | goto out; | |
2279 | ||
2280 | err = __sys_recvmsg(sock, msg, &msg_sys, flags, 0); | |
2281 | ||
6cb153ca | 2282 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
2283 | out: |
2284 | return err; | |
2285 | } | |
2286 | ||
a2e27255 ACM |
2287 | /* |
2288 | * Linux recvmmsg interface | |
2289 | */ | |
2290 | ||
2291 | int __sys_recvmmsg(int fd, struct mmsghdr __user *mmsg, unsigned int vlen, | |
2292 | unsigned int flags, struct timespec *timeout) | |
2293 | { | |
2294 | int fput_needed, err, datagrams; | |
2295 | struct socket *sock; | |
2296 | struct mmsghdr __user *entry; | |
d7256d0e | 2297 | struct compat_mmsghdr __user *compat_entry; |
a2e27255 ACM |
2298 | struct msghdr msg_sys; |
2299 | struct timespec end_time; | |
2300 | ||
2301 | if (timeout && | |
2302 | poll_select_set_timeout(&end_time, timeout->tv_sec, | |
2303 | timeout->tv_nsec)) | |
2304 | return -EINVAL; | |
2305 | ||
2306 | datagrams = 0; | |
2307 | ||
2308 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2309 | if (!sock) | |
2310 | return err; | |
2311 | ||
2312 | err = sock_error(sock->sk); | |
2313 | if (err) | |
2314 | goto out_put; | |
2315 | ||
2316 | entry = mmsg; | |
d7256d0e | 2317 | compat_entry = (struct compat_mmsghdr __user *)mmsg; |
a2e27255 ACM |
2318 | |
2319 | while (datagrams < vlen) { | |
2320 | /* | |
2321 | * No need to ask LSM for more than the first datagram. | |
2322 | */ | |
d7256d0e JMG |
2323 | if (MSG_CMSG_COMPAT & flags) { |
2324 | err = __sys_recvmsg(sock, (struct msghdr __user *)compat_entry, | |
b9eb8b87 AB |
2325 | &msg_sys, flags & ~MSG_WAITFORONE, |
2326 | datagrams); | |
d7256d0e JMG |
2327 | if (err < 0) |
2328 | break; | |
2329 | err = __put_user(err, &compat_entry->msg_len); | |
2330 | ++compat_entry; | |
2331 | } else { | |
2332 | err = __sys_recvmsg(sock, (struct msghdr __user *)entry, | |
b9eb8b87 AB |
2333 | &msg_sys, flags & ~MSG_WAITFORONE, |
2334 | datagrams); | |
d7256d0e JMG |
2335 | if (err < 0) |
2336 | break; | |
2337 | err = put_user(err, &entry->msg_len); | |
2338 | ++entry; | |
2339 | } | |
2340 | ||
a2e27255 ACM |
2341 | if (err) |
2342 | break; | |
a2e27255 ACM |
2343 | ++datagrams; |
2344 | ||
71c5c159 BB |
2345 | /* MSG_WAITFORONE turns on MSG_DONTWAIT after one packet */ |
2346 | if (flags & MSG_WAITFORONE) | |
2347 | flags |= MSG_DONTWAIT; | |
2348 | ||
a2e27255 ACM |
2349 | if (timeout) { |
2350 | ktime_get_ts(timeout); | |
2351 | *timeout = timespec_sub(end_time, *timeout); | |
2352 | if (timeout->tv_sec < 0) { | |
2353 | timeout->tv_sec = timeout->tv_nsec = 0; | |
2354 | break; | |
2355 | } | |
2356 | ||
2357 | /* Timeout, return less than vlen datagrams */ | |
2358 | if (timeout->tv_nsec == 0 && timeout->tv_sec == 0) | |
2359 | break; | |
2360 | } | |
2361 | ||
2362 | /* Out of band data, return right away */ | |
2363 | if (msg_sys.msg_flags & MSG_OOB) | |
2364 | break; | |
2365 | } | |
2366 | ||
2367 | out_put: | |
2368 | fput_light(sock->file, fput_needed); | |
1da177e4 | 2369 | |
a2e27255 ACM |
2370 | if (err == 0) |
2371 | return datagrams; | |
2372 | ||
2373 | if (datagrams != 0) { | |
2374 | /* | |
2375 | * We may return less entries than requested (vlen) if the | |
2376 | * sock is non block and there aren't enough datagrams... | |
2377 | */ | |
2378 | if (err != -EAGAIN) { | |
2379 | /* | |
2380 | * ... or if recvmsg returns an error after we | |
2381 | * received some datagrams, where we record the | |
2382 | * error to return on the next call or if the | |
2383 | * app asks about it using getsockopt(SO_ERROR). | |
2384 | */ | |
2385 | sock->sk->sk_err = -err; | |
2386 | } | |
2387 | ||
2388 | return datagrams; | |
2389 | } | |
2390 | ||
2391 | return err; | |
2392 | } | |
2393 | ||
2394 | SYSCALL_DEFINE5(recvmmsg, int, fd, struct mmsghdr __user *, mmsg, | |
2395 | unsigned int, vlen, unsigned int, flags, | |
2396 | struct timespec __user *, timeout) | |
2397 | { | |
2398 | int datagrams; | |
2399 | struct timespec timeout_sys; | |
2400 | ||
2401 | if (!timeout) | |
2402 | return __sys_recvmmsg(fd, mmsg, vlen, flags, NULL); | |
2403 | ||
2404 | if (copy_from_user(&timeout_sys, timeout, sizeof(timeout_sys))) | |
2405 | return -EFAULT; | |
2406 | ||
2407 | datagrams = __sys_recvmmsg(fd, mmsg, vlen, flags, &timeout_sys); | |
2408 | ||
2409 | if (datagrams > 0 && | |
2410 | copy_to_user(timeout, &timeout_sys, sizeof(timeout_sys))) | |
2411 | datagrams = -EFAULT; | |
2412 | ||
2413 | return datagrams; | |
2414 | } | |
2415 | ||
2416 | #ifdef __ARCH_WANT_SYS_SOCKETCALL | |
1da177e4 LT |
2417 | /* Argument list sizes for sys_socketcall */ |
2418 | #define AL(x) ((x) * sizeof(unsigned long)) | |
228e548e | 2419 | static const unsigned char nargs[21] = { |
c6d409cf ED |
2420 | AL(0), AL(3), AL(3), AL(3), AL(2), AL(3), |
2421 | AL(3), AL(3), AL(4), AL(4), AL(4), AL(6), | |
2422 | AL(6), AL(2), AL(5), AL(5), AL(3), AL(3), | |
228e548e | 2423 | AL(4), AL(5), AL(4) |
89bddce5 SH |
2424 | }; |
2425 | ||
1da177e4 LT |
2426 | #undef AL |
2427 | ||
2428 | /* | |
89bddce5 | 2429 | * System call vectors. |
1da177e4 LT |
2430 | * |
2431 | * Argument checking cleaned up. Saved 20% in size. | |
2432 | * This function doesn't need to set the kernel lock because | |
89bddce5 | 2433 | * it is set by the callees. |
1da177e4 LT |
2434 | */ |
2435 | ||
3e0fa65f | 2436 | SYSCALL_DEFINE2(socketcall, int, call, unsigned long __user *, args) |
1da177e4 LT |
2437 | { |
2438 | unsigned long a[6]; | |
89bddce5 | 2439 | unsigned long a0, a1; |
1da177e4 | 2440 | int err; |
47379052 | 2441 | unsigned int len; |
1da177e4 | 2442 | |
228e548e | 2443 | if (call < 1 || call > SYS_SENDMMSG) |
1da177e4 LT |
2444 | return -EINVAL; |
2445 | ||
47379052 AV |
2446 | len = nargs[call]; |
2447 | if (len > sizeof(a)) | |
2448 | return -EINVAL; | |
2449 | ||
1da177e4 | 2450 | /* copy_from_user should be SMP safe. */ |
47379052 | 2451 | if (copy_from_user(a, args, len)) |
1da177e4 | 2452 | return -EFAULT; |
3ec3b2fb | 2453 | |
f3298dc4 | 2454 | audit_socketcall(nargs[call] / sizeof(unsigned long), a); |
3ec3b2fb | 2455 | |
89bddce5 SH |
2456 | a0 = a[0]; |
2457 | a1 = a[1]; | |
2458 | ||
2459 | switch (call) { | |
2460 | case SYS_SOCKET: | |
2461 | err = sys_socket(a0, a1, a[2]); | |
2462 | break; | |
2463 | case SYS_BIND: | |
2464 | err = sys_bind(a0, (struct sockaddr __user *)a1, a[2]); | |
2465 | break; | |
2466 | case SYS_CONNECT: | |
2467 | err = sys_connect(a0, (struct sockaddr __user *)a1, a[2]); | |
2468 | break; | |
2469 | case SYS_LISTEN: | |
2470 | err = sys_listen(a0, a1); | |
2471 | break; | |
2472 | case SYS_ACCEPT: | |
de11defe UD |
2473 | err = sys_accept4(a0, (struct sockaddr __user *)a1, |
2474 | (int __user *)a[2], 0); | |
89bddce5 SH |
2475 | break; |
2476 | case SYS_GETSOCKNAME: | |
2477 | err = | |
2478 | sys_getsockname(a0, (struct sockaddr __user *)a1, | |
2479 | (int __user *)a[2]); | |
2480 | break; | |
2481 | case SYS_GETPEERNAME: | |
2482 | err = | |
2483 | sys_getpeername(a0, (struct sockaddr __user *)a1, | |
2484 | (int __user *)a[2]); | |
2485 | break; | |
2486 | case SYS_SOCKETPAIR: | |
2487 | err = sys_socketpair(a0, a1, a[2], (int __user *)a[3]); | |
2488 | break; | |
2489 | case SYS_SEND: | |
2490 | err = sys_send(a0, (void __user *)a1, a[2], a[3]); | |
2491 | break; | |
2492 | case SYS_SENDTO: | |
2493 | err = sys_sendto(a0, (void __user *)a1, a[2], a[3], | |
2494 | (struct sockaddr __user *)a[4], a[5]); | |
2495 | break; | |
2496 | case SYS_RECV: | |
2497 | err = sys_recv(a0, (void __user *)a1, a[2], a[3]); | |
2498 | break; | |
2499 | case SYS_RECVFROM: | |
2500 | err = sys_recvfrom(a0, (void __user *)a1, a[2], a[3], | |
2501 | (struct sockaddr __user *)a[4], | |
2502 | (int __user *)a[5]); | |
2503 | break; | |
2504 | case SYS_SHUTDOWN: | |
2505 | err = sys_shutdown(a0, a1); | |
2506 | break; | |
2507 | case SYS_SETSOCKOPT: | |
2508 | err = sys_setsockopt(a0, a1, a[2], (char __user *)a[3], a[4]); | |
2509 | break; | |
2510 | case SYS_GETSOCKOPT: | |
2511 | err = | |
2512 | sys_getsockopt(a0, a1, a[2], (char __user *)a[3], | |
2513 | (int __user *)a[4]); | |
2514 | break; | |
2515 | case SYS_SENDMSG: | |
2516 | err = sys_sendmsg(a0, (struct msghdr __user *)a1, a[2]); | |
2517 | break; | |
228e548e AB |
2518 | case SYS_SENDMMSG: |
2519 | err = sys_sendmmsg(a0, (struct mmsghdr __user *)a1, a[2], a[3]); | |
2520 | break; | |
89bddce5 SH |
2521 | case SYS_RECVMSG: |
2522 | err = sys_recvmsg(a0, (struct msghdr __user *)a1, a[2]); | |
2523 | break; | |
a2e27255 ACM |
2524 | case SYS_RECVMMSG: |
2525 | err = sys_recvmmsg(a0, (struct mmsghdr __user *)a1, a[2], a[3], | |
2526 | (struct timespec __user *)a[4]); | |
2527 | break; | |
de11defe UD |
2528 | case SYS_ACCEPT4: |
2529 | err = sys_accept4(a0, (struct sockaddr __user *)a1, | |
2530 | (int __user *)a[2], a[3]); | |
aaca0bdc | 2531 | break; |
89bddce5 SH |
2532 | default: |
2533 | err = -EINVAL; | |
2534 | break; | |
1da177e4 LT |
2535 | } |
2536 | return err; | |
2537 | } | |
2538 | ||
89bddce5 | 2539 | #endif /* __ARCH_WANT_SYS_SOCKETCALL */ |
1da177e4 | 2540 | |
55737fda SH |
2541 | /** |
2542 | * sock_register - add a socket protocol handler | |
2543 | * @ops: description of protocol | |
2544 | * | |
1da177e4 LT |
2545 | * This function is called by a protocol handler that wants to |
2546 | * advertise its address family, and have it linked into the | |
55737fda SH |
2547 | * socket interface. The value ops->family coresponds to the |
2548 | * socket system call protocol family. | |
1da177e4 | 2549 | */ |
f0fd27d4 | 2550 | int sock_register(const struct net_proto_family *ops) |
1da177e4 LT |
2551 | { |
2552 | int err; | |
2553 | ||
2554 | if (ops->family >= NPROTO) { | |
89bddce5 SH |
2555 | printk(KERN_CRIT "protocol %d >= NPROTO(%d)\n", ops->family, |
2556 | NPROTO); | |
1da177e4 LT |
2557 | return -ENOBUFS; |
2558 | } | |
55737fda SH |
2559 | |
2560 | spin_lock(&net_family_lock); | |
190683a9 ED |
2561 | if (rcu_dereference_protected(net_families[ops->family], |
2562 | lockdep_is_held(&net_family_lock))) | |
55737fda SH |
2563 | err = -EEXIST; |
2564 | else { | |
cf778b00 | 2565 | rcu_assign_pointer(net_families[ops->family], ops); |
1da177e4 LT |
2566 | err = 0; |
2567 | } | |
55737fda SH |
2568 | spin_unlock(&net_family_lock); |
2569 | ||
89bddce5 | 2570 | printk(KERN_INFO "NET: Registered protocol family %d\n", ops->family); |
1da177e4 LT |
2571 | return err; |
2572 | } | |
c6d409cf | 2573 | EXPORT_SYMBOL(sock_register); |
1da177e4 | 2574 | |
55737fda SH |
2575 | /** |
2576 | * sock_unregister - remove a protocol handler | |
2577 | * @family: protocol family to remove | |
2578 | * | |
1da177e4 LT |
2579 | * This function is called by a protocol handler that wants to |
2580 | * remove its address family, and have it unlinked from the | |
55737fda SH |
2581 | * new socket creation. |
2582 | * | |
2583 | * If protocol handler is a module, then it can use module reference | |
2584 | * counts to protect against new references. If protocol handler is not | |
2585 | * a module then it needs to provide its own protection in | |
2586 | * the ops->create routine. | |
1da177e4 | 2587 | */ |
f0fd27d4 | 2588 | void sock_unregister(int family) |
1da177e4 | 2589 | { |
f0fd27d4 | 2590 | BUG_ON(family < 0 || family >= NPROTO); |
1da177e4 | 2591 | |
55737fda | 2592 | spin_lock(&net_family_lock); |
a9b3cd7f | 2593 | RCU_INIT_POINTER(net_families[family], NULL); |
55737fda SH |
2594 | spin_unlock(&net_family_lock); |
2595 | ||
2596 | synchronize_rcu(); | |
2597 | ||
89bddce5 | 2598 | printk(KERN_INFO "NET: Unregistered protocol family %d\n", family); |
1da177e4 | 2599 | } |
c6d409cf | 2600 | EXPORT_SYMBOL(sock_unregister); |
1da177e4 | 2601 | |
77d76ea3 | 2602 | static int __init sock_init(void) |
1da177e4 | 2603 | { |
b3e19d92 | 2604 | int err; |
2ca794e5 EB |
2605 | /* |
2606 | * Initialize the network sysctl infrastructure. | |
2607 | */ | |
2608 | err = net_sysctl_init(); | |
2609 | if (err) | |
2610 | goto out; | |
b3e19d92 | 2611 | |
1da177e4 | 2612 | /* |
89bddce5 | 2613 | * Initialize skbuff SLAB cache |
1da177e4 LT |
2614 | */ |
2615 | skb_init(); | |
1da177e4 LT |
2616 | |
2617 | /* | |
89bddce5 | 2618 | * Initialize the protocols module. |
1da177e4 LT |
2619 | */ |
2620 | ||
2621 | init_inodecache(); | |
b3e19d92 NP |
2622 | |
2623 | err = register_filesystem(&sock_fs_type); | |
2624 | if (err) | |
2625 | goto out_fs; | |
1da177e4 | 2626 | sock_mnt = kern_mount(&sock_fs_type); |
b3e19d92 NP |
2627 | if (IS_ERR(sock_mnt)) { |
2628 | err = PTR_ERR(sock_mnt); | |
2629 | goto out_mount; | |
2630 | } | |
77d76ea3 AK |
2631 | |
2632 | /* The real protocol initialization is performed in later initcalls. | |
1da177e4 LT |
2633 | */ |
2634 | ||
2635 | #ifdef CONFIG_NETFILTER | |
2636 | netfilter_init(); | |
2637 | #endif | |
cbeb321a | 2638 | |
c1f19b51 RC |
2639 | #ifdef CONFIG_NETWORK_PHY_TIMESTAMPING |
2640 | skb_timestamping_init(); | |
2641 | #endif | |
2642 | ||
b3e19d92 NP |
2643 | out: |
2644 | return err; | |
2645 | ||
2646 | out_mount: | |
2647 | unregister_filesystem(&sock_fs_type); | |
2648 | out_fs: | |
2649 | goto out; | |
1da177e4 LT |
2650 | } |
2651 | ||
77d76ea3 AK |
2652 | core_initcall(sock_init); /* early initcall */ |
2653 | ||
1da177e4 LT |
2654 | #ifdef CONFIG_PROC_FS |
2655 | void socket_seq_show(struct seq_file *seq) | |
2656 | { | |
2657 | int cpu; | |
2658 | int counter = 0; | |
2659 | ||
6f912042 | 2660 | for_each_possible_cpu(cpu) |
89bddce5 | 2661 | counter += per_cpu(sockets_in_use, cpu); |
1da177e4 LT |
2662 | |
2663 | /* It can be negative, by the way. 8) */ | |
2664 | if (counter < 0) | |
2665 | counter = 0; | |
2666 | ||
2667 | seq_printf(seq, "sockets: used %d\n", counter); | |
2668 | } | |
89bddce5 | 2669 | #endif /* CONFIG_PROC_FS */ |
1da177e4 | 2670 | |
89bbfc95 | 2671 | #ifdef CONFIG_COMPAT |
6b96018b | 2672 | static int do_siocgstamp(struct net *net, struct socket *sock, |
644595f8 | 2673 | unsigned int cmd, void __user *up) |
7a229387 | 2674 | { |
7a229387 AB |
2675 | mm_segment_t old_fs = get_fs(); |
2676 | struct timeval ktv; | |
2677 | int err; | |
2678 | ||
2679 | set_fs(KERNEL_DS); | |
6b96018b | 2680 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)&ktv); |
7a229387 | 2681 | set_fs(old_fs); |
644595f8 | 2682 | if (!err) |
ed6fe9d6 | 2683 | err = compat_put_timeval(&ktv, up); |
644595f8 | 2684 | |
7a229387 AB |
2685 | return err; |
2686 | } | |
2687 | ||
6b96018b | 2688 | static int do_siocgstampns(struct net *net, struct socket *sock, |
644595f8 | 2689 | unsigned int cmd, void __user *up) |
7a229387 | 2690 | { |
7a229387 AB |
2691 | mm_segment_t old_fs = get_fs(); |
2692 | struct timespec kts; | |
2693 | int err; | |
2694 | ||
2695 | set_fs(KERNEL_DS); | |
6b96018b | 2696 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)&kts); |
7a229387 | 2697 | set_fs(old_fs); |
644595f8 | 2698 | if (!err) |
ed6fe9d6 | 2699 | err = compat_put_timespec(&kts, up); |
644595f8 | 2700 | |
7a229387 AB |
2701 | return err; |
2702 | } | |
2703 | ||
6b96018b | 2704 | static int dev_ifname32(struct net *net, struct compat_ifreq __user *uifr32) |
7a229387 AB |
2705 | { |
2706 | struct ifreq __user *uifr; | |
2707 | int err; | |
2708 | ||
2709 | uifr = compat_alloc_user_space(sizeof(struct ifreq)); | |
6b96018b | 2710 | if (copy_in_user(uifr, uifr32, sizeof(struct compat_ifreq))) |
7a229387 AB |
2711 | return -EFAULT; |
2712 | ||
6b96018b | 2713 | err = dev_ioctl(net, SIOCGIFNAME, uifr); |
7a229387 AB |
2714 | if (err) |
2715 | return err; | |
2716 | ||
6b96018b | 2717 | if (copy_in_user(uifr32, uifr, sizeof(struct compat_ifreq))) |
7a229387 AB |
2718 | return -EFAULT; |
2719 | ||
2720 | return 0; | |
2721 | } | |
2722 | ||
6b96018b | 2723 | static int dev_ifconf(struct net *net, struct compat_ifconf __user *uifc32) |
7a229387 | 2724 | { |
6b96018b | 2725 | struct compat_ifconf ifc32; |
7a229387 AB |
2726 | struct ifconf ifc; |
2727 | struct ifconf __user *uifc; | |
6b96018b | 2728 | struct compat_ifreq __user *ifr32; |
7a229387 AB |
2729 | struct ifreq __user *ifr; |
2730 | unsigned int i, j; | |
2731 | int err; | |
2732 | ||
6b96018b | 2733 | if (copy_from_user(&ifc32, uifc32, sizeof(struct compat_ifconf))) |
7a229387 AB |
2734 | return -EFAULT; |
2735 | ||
43da5f2e | 2736 | memset(&ifc, 0, sizeof(ifc)); |
7a229387 AB |
2737 | if (ifc32.ifcbuf == 0) { |
2738 | ifc32.ifc_len = 0; | |
2739 | ifc.ifc_len = 0; | |
2740 | ifc.ifc_req = NULL; | |
2741 | uifc = compat_alloc_user_space(sizeof(struct ifconf)); | |
2742 | } else { | |
c6d409cf ED |
2743 | size_t len = ((ifc32.ifc_len / sizeof(struct compat_ifreq)) + 1) * |
2744 | sizeof(struct ifreq); | |
7a229387 AB |
2745 | uifc = compat_alloc_user_space(sizeof(struct ifconf) + len); |
2746 | ifc.ifc_len = len; | |
2747 | ifr = ifc.ifc_req = (void __user *)(uifc + 1); | |
2748 | ifr32 = compat_ptr(ifc32.ifcbuf); | |
c6d409cf | 2749 | for (i = 0; i < ifc32.ifc_len; i += sizeof(struct compat_ifreq)) { |
6b96018b | 2750 | if (copy_in_user(ifr, ifr32, sizeof(struct compat_ifreq))) |
7a229387 AB |
2751 | return -EFAULT; |
2752 | ifr++; | |
2753 | ifr32++; | |
2754 | } | |
2755 | } | |
2756 | if (copy_to_user(uifc, &ifc, sizeof(struct ifconf))) | |
2757 | return -EFAULT; | |
2758 | ||
6b96018b | 2759 | err = dev_ioctl(net, SIOCGIFCONF, uifc); |
7a229387 AB |
2760 | if (err) |
2761 | return err; | |
2762 | ||
2763 | if (copy_from_user(&ifc, uifc, sizeof(struct ifconf))) | |
2764 | return -EFAULT; | |
2765 | ||
2766 | ifr = ifc.ifc_req; | |
2767 | ifr32 = compat_ptr(ifc32.ifcbuf); | |
2768 | for (i = 0, j = 0; | |
c6d409cf ED |
2769 | i + sizeof(struct compat_ifreq) <= ifc32.ifc_len && j < ifc.ifc_len; |
2770 | i += sizeof(struct compat_ifreq), j += sizeof(struct ifreq)) { | |
2771 | if (copy_in_user(ifr32, ifr, sizeof(struct compat_ifreq))) | |
7a229387 AB |
2772 | return -EFAULT; |
2773 | ifr32++; | |
2774 | ifr++; | |
2775 | } | |
2776 | ||
2777 | if (ifc32.ifcbuf == 0) { | |
2778 | /* Translate from 64-bit structure multiple to | |
2779 | * a 32-bit one. | |
2780 | */ | |
2781 | i = ifc.ifc_len; | |
6b96018b | 2782 | i = ((i / sizeof(struct ifreq)) * sizeof(struct compat_ifreq)); |
7a229387 AB |
2783 | ifc32.ifc_len = i; |
2784 | } else { | |
2785 | ifc32.ifc_len = i; | |
2786 | } | |
6b96018b | 2787 | if (copy_to_user(uifc32, &ifc32, sizeof(struct compat_ifconf))) |
7a229387 AB |
2788 | return -EFAULT; |
2789 | ||
2790 | return 0; | |
2791 | } | |
2792 | ||
6b96018b | 2793 | static int ethtool_ioctl(struct net *net, struct compat_ifreq __user *ifr32) |
7a229387 | 2794 | { |
3a7da39d BH |
2795 | struct compat_ethtool_rxnfc __user *compat_rxnfc; |
2796 | bool convert_in = false, convert_out = false; | |
2797 | size_t buf_size = ALIGN(sizeof(struct ifreq), 8); | |
2798 | struct ethtool_rxnfc __user *rxnfc; | |
7a229387 | 2799 | struct ifreq __user *ifr; |
3a7da39d BH |
2800 | u32 rule_cnt = 0, actual_rule_cnt; |
2801 | u32 ethcmd; | |
7a229387 | 2802 | u32 data; |
3a7da39d | 2803 | int ret; |
7a229387 | 2804 | |
3a7da39d BH |
2805 | if (get_user(data, &ifr32->ifr_ifru.ifru_data)) |
2806 | return -EFAULT; | |
7a229387 | 2807 | |
3a7da39d BH |
2808 | compat_rxnfc = compat_ptr(data); |
2809 | ||
2810 | if (get_user(ethcmd, &compat_rxnfc->cmd)) | |
7a229387 AB |
2811 | return -EFAULT; |
2812 | ||
3a7da39d BH |
2813 | /* Most ethtool structures are defined without padding. |
2814 | * Unfortunately struct ethtool_rxnfc is an exception. | |
2815 | */ | |
2816 | switch (ethcmd) { | |
2817 | default: | |
2818 | break; | |
2819 | case ETHTOOL_GRXCLSRLALL: | |
2820 | /* Buffer size is variable */ | |
2821 | if (get_user(rule_cnt, &compat_rxnfc->rule_cnt)) | |
2822 | return -EFAULT; | |
2823 | if (rule_cnt > KMALLOC_MAX_SIZE / sizeof(u32)) | |
2824 | return -ENOMEM; | |
2825 | buf_size += rule_cnt * sizeof(u32); | |
2826 | /* fall through */ | |
2827 | case ETHTOOL_GRXRINGS: | |
2828 | case ETHTOOL_GRXCLSRLCNT: | |
2829 | case ETHTOOL_GRXCLSRULE: | |
55664f32 | 2830 | case ETHTOOL_SRXCLSRLINS: |
3a7da39d BH |
2831 | convert_out = true; |
2832 | /* fall through */ | |
2833 | case ETHTOOL_SRXCLSRLDEL: | |
3a7da39d BH |
2834 | buf_size += sizeof(struct ethtool_rxnfc); |
2835 | convert_in = true; | |
2836 | break; | |
2837 | } | |
2838 | ||
2839 | ifr = compat_alloc_user_space(buf_size); | |
2840 | rxnfc = (void *)ifr + ALIGN(sizeof(struct ifreq), 8); | |
2841 | ||
2842 | if (copy_in_user(&ifr->ifr_name, &ifr32->ifr_name, IFNAMSIZ)) | |
7a229387 AB |
2843 | return -EFAULT; |
2844 | ||
3a7da39d BH |
2845 | if (put_user(convert_in ? rxnfc : compat_ptr(data), |
2846 | &ifr->ifr_ifru.ifru_data)) | |
7a229387 AB |
2847 | return -EFAULT; |
2848 | ||
3a7da39d | 2849 | if (convert_in) { |
127fe533 | 2850 | /* We expect there to be holes between fs.m_ext and |
3a7da39d BH |
2851 | * fs.ring_cookie and at the end of fs, but nowhere else. |
2852 | */ | |
127fe533 AD |
2853 | BUILD_BUG_ON(offsetof(struct compat_ethtool_rxnfc, fs.m_ext) + |
2854 | sizeof(compat_rxnfc->fs.m_ext) != | |
2855 | offsetof(struct ethtool_rxnfc, fs.m_ext) + | |
2856 | sizeof(rxnfc->fs.m_ext)); | |
3a7da39d BH |
2857 | BUILD_BUG_ON( |
2858 | offsetof(struct compat_ethtool_rxnfc, fs.location) - | |
2859 | offsetof(struct compat_ethtool_rxnfc, fs.ring_cookie) != | |
2860 | offsetof(struct ethtool_rxnfc, fs.location) - | |
2861 | offsetof(struct ethtool_rxnfc, fs.ring_cookie)); | |
2862 | ||
2863 | if (copy_in_user(rxnfc, compat_rxnfc, | |
127fe533 | 2864 | (void *)(&rxnfc->fs.m_ext + 1) - |
3a7da39d BH |
2865 | (void *)rxnfc) || |
2866 | copy_in_user(&rxnfc->fs.ring_cookie, | |
2867 | &compat_rxnfc->fs.ring_cookie, | |
2868 | (void *)(&rxnfc->fs.location + 1) - | |
2869 | (void *)&rxnfc->fs.ring_cookie) || | |
2870 | copy_in_user(&rxnfc->rule_cnt, &compat_rxnfc->rule_cnt, | |
2871 | sizeof(rxnfc->rule_cnt))) | |
2872 | return -EFAULT; | |
2873 | } | |
2874 | ||
2875 | ret = dev_ioctl(net, SIOCETHTOOL, ifr); | |
2876 | if (ret) | |
2877 | return ret; | |
2878 | ||
2879 | if (convert_out) { | |
2880 | if (copy_in_user(compat_rxnfc, rxnfc, | |
127fe533 | 2881 | (const void *)(&rxnfc->fs.m_ext + 1) - |
3a7da39d BH |
2882 | (const void *)rxnfc) || |
2883 | copy_in_user(&compat_rxnfc->fs.ring_cookie, | |
2884 | &rxnfc->fs.ring_cookie, | |
2885 | (const void *)(&rxnfc->fs.location + 1) - | |
2886 | (const void *)&rxnfc->fs.ring_cookie) || | |
2887 | copy_in_user(&compat_rxnfc->rule_cnt, &rxnfc->rule_cnt, | |
2888 | sizeof(rxnfc->rule_cnt))) | |
2889 | return -EFAULT; | |
2890 | ||
2891 | if (ethcmd == ETHTOOL_GRXCLSRLALL) { | |
2892 | /* As an optimisation, we only copy the actual | |
2893 | * number of rules that the underlying | |
2894 | * function returned. Since Mallory might | |
2895 | * change the rule count in user memory, we | |
2896 | * check that it is less than the rule count | |
2897 | * originally given (as the user buffer size), | |
2898 | * which has been range-checked. | |
2899 | */ | |
2900 | if (get_user(actual_rule_cnt, &rxnfc->rule_cnt)) | |
2901 | return -EFAULT; | |
2902 | if (actual_rule_cnt < rule_cnt) | |
2903 | rule_cnt = actual_rule_cnt; | |
2904 | if (copy_in_user(&compat_rxnfc->rule_locs[0], | |
2905 | &rxnfc->rule_locs[0], | |
2906 | rule_cnt * sizeof(u32))) | |
2907 | return -EFAULT; | |
2908 | } | |
2909 | } | |
2910 | ||
2911 | return 0; | |
7a229387 AB |
2912 | } |
2913 | ||
7a50a240 AB |
2914 | static int compat_siocwandev(struct net *net, struct compat_ifreq __user *uifr32) |
2915 | { | |
2916 | void __user *uptr; | |
2917 | compat_uptr_t uptr32; | |
2918 | struct ifreq __user *uifr; | |
2919 | ||
c6d409cf | 2920 | uifr = compat_alloc_user_space(sizeof(*uifr)); |
7a50a240 AB |
2921 | if (copy_in_user(uifr, uifr32, sizeof(struct compat_ifreq))) |
2922 | return -EFAULT; | |
2923 | ||
2924 | if (get_user(uptr32, &uifr32->ifr_settings.ifs_ifsu)) | |
2925 | return -EFAULT; | |
2926 | ||
2927 | uptr = compat_ptr(uptr32); | |
2928 | ||
2929 | if (put_user(uptr, &uifr->ifr_settings.ifs_ifsu.raw_hdlc)) | |
2930 | return -EFAULT; | |
2931 | ||
2932 | return dev_ioctl(net, SIOCWANDEV, uifr); | |
2933 | } | |
2934 | ||
6b96018b AB |
2935 | static int bond_ioctl(struct net *net, unsigned int cmd, |
2936 | struct compat_ifreq __user *ifr32) | |
7a229387 AB |
2937 | { |
2938 | struct ifreq kifr; | |
2939 | struct ifreq __user *uifr; | |
7a229387 AB |
2940 | mm_segment_t old_fs; |
2941 | int err; | |
2942 | u32 data; | |
2943 | void __user *datap; | |
2944 | ||
2945 | switch (cmd) { | |
2946 | case SIOCBONDENSLAVE: | |
2947 | case SIOCBONDRELEASE: | |
2948 | case SIOCBONDSETHWADDR: | |
2949 | case SIOCBONDCHANGEACTIVE: | |
6b96018b | 2950 | if (copy_from_user(&kifr, ifr32, sizeof(struct compat_ifreq))) |
7a229387 AB |
2951 | return -EFAULT; |
2952 | ||
2953 | old_fs = get_fs(); | |
c6d409cf | 2954 | set_fs(KERNEL_DS); |
c3f52ae6 | 2955 | err = dev_ioctl(net, cmd, |
2956 | (struct ifreq __user __force *) &kifr); | |
c6d409cf | 2957 | set_fs(old_fs); |
7a229387 AB |
2958 | |
2959 | return err; | |
2960 | case SIOCBONDSLAVEINFOQUERY: | |
2961 | case SIOCBONDINFOQUERY: | |
2962 | uifr = compat_alloc_user_space(sizeof(*uifr)); | |
2963 | if (copy_in_user(&uifr->ifr_name, &ifr32->ifr_name, IFNAMSIZ)) | |
2964 | return -EFAULT; | |
2965 | ||
2966 | if (get_user(data, &ifr32->ifr_ifru.ifru_data)) | |
2967 | return -EFAULT; | |
2968 | ||
2969 | datap = compat_ptr(data); | |
2970 | if (put_user(datap, &uifr->ifr_ifru.ifru_data)) | |
2971 | return -EFAULT; | |
2972 | ||
6b96018b | 2973 | return dev_ioctl(net, cmd, uifr); |
7a229387 | 2974 | default: |
07d106d0 | 2975 | return -ENOIOCTLCMD; |
ccbd6a5a | 2976 | } |
7a229387 AB |
2977 | } |
2978 | ||
6b96018b AB |
2979 | static int siocdevprivate_ioctl(struct net *net, unsigned int cmd, |
2980 | struct compat_ifreq __user *u_ifreq32) | |
7a229387 AB |
2981 | { |
2982 | struct ifreq __user *u_ifreq64; | |
7a229387 AB |
2983 | char tmp_buf[IFNAMSIZ]; |
2984 | void __user *data64; | |
2985 | u32 data32; | |
2986 | ||
2987 | if (copy_from_user(&tmp_buf[0], &(u_ifreq32->ifr_ifrn.ifrn_name[0]), | |
2988 | IFNAMSIZ)) | |
2989 | return -EFAULT; | |
2990 | if (__get_user(data32, &u_ifreq32->ifr_ifru.ifru_data)) | |
2991 | return -EFAULT; | |
2992 | data64 = compat_ptr(data32); | |
2993 | ||
2994 | u_ifreq64 = compat_alloc_user_space(sizeof(*u_ifreq64)); | |
2995 | ||
2996 | /* Don't check these user accesses, just let that get trapped | |
2997 | * in the ioctl handler instead. | |
2998 | */ | |
2999 | if (copy_to_user(&u_ifreq64->ifr_ifrn.ifrn_name[0], &tmp_buf[0], | |
3000 | IFNAMSIZ)) | |
3001 | return -EFAULT; | |
3002 | if (__put_user(data64, &u_ifreq64->ifr_ifru.ifru_data)) | |
3003 | return -EFAULT; | |
3004 | ||
6b96018b | 3005 | return dev_ioctl(net, cmd, u_ifreq64); |
7a229387 AB |
3006 | } |
3007 | ||
6b96018b AB |
3008 | static int dev_ifsioc(struct net *net, struct socket *sock, |
3009 | unsigned int cmd, struct compat_ifreq __user *uifr32) | |
7a229387 | 3010 | { |
a2116ed2 | 3011 | struct ifreq __user *uifr; |
7a229387 AB |
3012 | int err; |
3013 | ||
a2116ed2 AB |
3014 | uifr = compat_alloc_user_space(sizeof(*uifr)); |
3015 | if (copy_in_user(uifr, uifr32, sizeof(*uifr32))) | |
3016 | return -EFAULT; | |
3017 | ||
3018 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)uifr); | |
3019 | ||
7a229387 AB |
3020 | if (!err) { |
3021 | switch (cmd) { | |
3022 | case SIOCGIFFLAGS: | |
3023 | case SIOCGIFMETRIC: | |
3024 | case SIOCGIFMTU: | |
3025 | case SIOCGIFMEM: | |
3026 | case SIOCGIFHWADDR: | |
3027 | case SIOCGIFINDEX: | |
3028 | case SIOCGIFADDR: | |
3029 | case SIOCGIFBRDADDR: | |
3030 | case SIOCGIFDSTADDR: | |
3031 | case SIOCGIFNETMASK: | |
fab2532b | 3032 | case SIOCGIFPFLAGS: |
7a229387 | 3033 | case SIOCGIFTXQLEN: |
fab2532b AB |
3034 | case SIOCGMIIPHY: |
3035 | case SIOCGMIIREG: | |
a2116ed2 | 3036 | if (copy_in_user(uifr32, uifr, sizeof(*uifr32))) |
7a229387 AB |
3037 | err = -EFAULT; |
3038 | break; | |
3039 | } | |
3040 | } | |
3041 | return err; | |
3042 | } | |
3043 | ||
a2116ed2 AB |
3044 | static int compat_sioc_ifmap(struct net *net, unsigned int cmd, |
3045 | struct compat_ifreq __user *uifr32) | |
3046 | { | |
3047 | struct ifreq ifr; | |
3048 | struct compat_ifmap __user *uifmap32; | |
3049 | mm_segment_t old_fs; | |
3050 | int err; | |
3051 | ||
3052 | uifmap32 = &uifr32->ifr_ifru.ifru_map; | |
3053 | err = copy_from_user(&ifr, uifr32, sizeof(ifr.ifr_name)); | |
3054 | err |= __get_user(ifr.ifr_map.mem_start, &uifmap32->mem_start); | |
3055 | err |= __get_user(ifr.ifr_map.mem_end, &uifmap32->mem_end); | |
3056 | err |= __get_user(ifr.ifr_map.base_addr, &uifmap32->base_addr); | |
3057 | err |= __get_user(ifr.ifr_map.irq, &uifmap32->irq); | |
3058 | err |= __get_user(ifr.ifr_map.dma, &uifmap32->dma); | |
3059 | err |= __get_user(ifr.ifr_map.port, &uifmap32->port); | |
3060 | if (err) | |
3061 | return -EFAULT; | |
3062 | ||
3063 | old_fs = get_fs(); | |
c6d409cf | 3064 | set_fs(KERNEL_DS); |
c3f52ae6 | 3065 | err = dev_ioctl(net, cmd, (void __user __force *)&ifr); |
c6d409cf | 3066 | set_fs(old_fs); |
a2116ed2 AB |
3067 | |
3068 | if (cmd == SIOCGIFMAP && !err) { | |
3069 | err = copy_to_user(uifr32, &ifr, sizeof(ifr.ifr_name)); | |
3070 | err |= __put_user(ifr.ifr_map.mem_start, &uifmap32->mem_start); | |
3071 | err |= __put_user(ifr.ifr_map.mem_end, &uifmap32->mem_end); | |
3072 | err |= __put_user(ifr.ifr_map.base_addr, &uifmap32->base_addr); | |
3073 | err |= __put_user(ifr.ifr_map.irq, &uifmap32->irq); | |
3074 | err |= __put_user(ifr.ifr_map.dma, &uifmap32->dma); | |
3075 | err |= __put_user(ifr.ifr_map.port, &uifmap32->port); | |
3076 | if (err) | |
3077 | err = -EFAULT; | |
3078 | } | |
3079 | return err; | |
3080 | } | |
3081 | ||
3082 | static int compat_siocshwtstamp(struct net *net, struct compat_ifreq __user *uifr32) | |
3083 | { | |
3084 | void __user *uptr; | |
3085 | compat_uptr_t uptr32; | |
3086 | struct ifreq __user *uifr; | |
3087 | ||
c6d409cf | 3088 | uifr = compat_alloc_user_space(sizeof(*uifr)); |
a2116ed2 AB |
3089 | if (copy_in_user(uifr, uifr32, sizeof(struct compat_ifreq))) |
3090 | return -EFAULT; | |
3091 | ||
3092 | if (get_user(uptr32, &uifr32->ifr_data)) | |
3093 | return -EFAULT; | |
3094 | ||
3095 | uptr = compat_ptr(uptr32); | |
3096 | ||
3097 | if (put_user(uptr, &uifr->ifr_data)) | |
3098 | return -EFAULT; | |
3099 | ||
3100 | return dev_ioctl(net, SIOCSHWTSTAMP, uifr); | |
3101 | } | |
3102 | ||
7a229387 | 3103 | struct rtentry32 { |
c6d409cf | 3104 | u32 rt_pad1; |
7a229387 AB |
3105 | struct sockaddr rt_dst; /* target address */ |
3106 | struct sockaddr rt_gateway; /* gateway addr (RTF_GATEWAY) */ | |
3107 | struct sockaddr rt_genmask; /* target network mask (IP) */ | |
c6d409cf ED |
3108 | unsigned short rt_flags; |
3109 | short rt_pad2; | |
3110 | u32 rt_pad3; | |
3111 | unsigned char rt_tos; | |
3112 | unsigned char rt_class; | |
3113 | short rt_pad4; | |
3114 | short rt_metric; /* +1 for binary compatibility! */ | |
7a229387 | 3115 | /* char * */ u32 rt_dev; /* forcing the device at add */ |
c6d409cf ED |
3116 | u32 rt_mtu; /* per route MTU/Window */ |
3117 | u32 rt_window; /* Window clamping */ | |
7a229387 AB |
3118 | unsigned short rt_irtt; /* Initial RTT */ |
3119 | }; | |
3120 | ||
3121 | struct in6_rtmsg32 { | |
3122 | struct in6_addr rtmsg_dst; | |
3123 | struct in6_addr rtmsg_src; | |
3124 | struct in6_addr rtmsg_gateway; | |
3125 | u32 rtmsg_type; | |
3126 | u16 rtmsg_dst_len; | |
3127 | u16 rtmsg_src_len; | |
3128 | u32 rtmsg_metric; | |
3129 | u32 rtmsg_info; | |
3130 | u32 rtmsg_flags; | |
3131 | s32 rtmsg_ifindex; | |
3132 | }; | |
3133 | ||
6b96018b AB |
3134 | static int routing_ioctl(struct net *net, struct socket *sock, |
3135 | unsigned int cmd, void __user *argp) | |
7a229387 AB |
3136 | { |
3137 | int ret; | |
3138 | void *r = NULL; | |
3139 | struct in6_rtmsg r6; | |
3140 | struct rtentry r4; | |
3141 | char devname[16]; | |
3142 | u32 rtdev; | |
3143 | mm_segment_t old_fs = get_fs(); | |
3144 | ||
6b96018b AB |
3145 | if (sock && sock->sk && sock->sk->sk_family == AF_INET6) { /* ipv6 */ |
3146 | struct in6_rtmsg32 __user *ur6 = argp; | |
c6d409cf | 3147 | ret = copy_from_user(&r6.rtmsg_dst, &(ur6->rtmsg_dst), |
7a229387 | 3148 | 3 * sizeof(struct in6_addr)); |
c6d409cf ED |
3149 | ret |= __get_user(r6.rtmsg_type, &(ur6->rtmsg_type)); |
3150 | ret |= __get_user(r6.rtmsg_dst_len, &(ur6->rtmsg_dst_len)); | |
3151 | ret |= __get_user(r6.rtmsg_src_len, &(ur6->rtmsg_src_len)); | |
3152 | ret |= __get_user(r6.rtmsg_metric, &(ur6->rtmsg_metric)); | |
3153 | ret |= __get_user(r6.rtmsg_info, &(ur6->rtmsg_info)); | |
3154 | ret |= __get_user(r6.rtmsg_flags, &(ur6->rtmsg_flags)); | |
3155 | ret |= __get_user(r6.rtmsg_ifindex, &(ur6->rtmsg_ifindex)); | |
7a229387 AB |
3156 | |
3157 | r = (void *) &r6; | |
3158 | } else { /* ipv4 */ | |
6b96018b | 3159 | struct rtentry32 __user *ur4 = argp; |
c6d409cf | 3160 | ret = copy_from_user(&r4.rt_dst, &(ur4->rt_dst), |
7a229387 | 3161 | 3 * sizeof(struct sockaddr)); |
c6d409cf ED |
3162 | ret |= __get_user(r4.rt_flags, &(ur4->rt_flags)); |
3163 | ret |= __get_user(r4.rt_metric, &(ur4->rt_metric)); | |
3164 | ret |= __get_user(r4.rt_mtu, &(ur4->rt_mtu)); | |
3165 | ret |= __get_user(r4.rt_window, &(ur4->rt_window)); | |
3166 | ret |= __get_user(r4.rt_irtt, &(ur4->rt_irtt)); | |
3167 | ret |= __get_user(rtdev, &(ur4->rt_dev)); | |
7a229387 | 3168 | if (rtdev) { |
c6d409cf | 3169 | ret |= copy_from_user(devname, compat_ptr(rtdev), 15); |
c3f52ae6 | 3170 | r4.rt_dev = (char __user __force *)devname; |
3171 | devname[15] = 0; | |
7a229387 AB |
3172 | } else |
3173 | r4.rt_dev = NULL; | |
3174 | ||
3175 | r = (void *) &r4; | |
3176 | } | |
3177 | ||
3178 | if (ret) { | |
3179 | ret = -EFAULT; | |
3180 | goto out; | |
3181 | } | |
3182 | ||
c6d409cf | 3183 | set_fs(KERNEL_DS); |
6b96018b | 3184 | ret = sock_do_ioctl(net, sock, cmd, (unsigned long) r); |
c6d409cf | 3185 | set_fs(old_fs); |
7a229387 AB |
3186 | |
3187 | out: | |
7a229387 AB |
3188 | return ret; |
3189 | } | |
3190 | ||
3191 | /* Since old style bridge ioctl's endup using SIOCDEVPRIVATE | |
3192 | * for some operations; this forces use of the newer bridge-utils that | |
25985edc | 3193 | * use compatible ioctls |
7a229387 | 3194 | */ |
6b96018b | 3195 | static int old_bridge_ioctl(compat_ulong_t __user *argp) |
7a229387 | 3196 | { |
6b96018b | 3197 | compat_ulong_t tmp; |
7a229387 | 3198 | |
6b96018b | 3199 | if (get_user(tmp, argp)) |
7a229387 AB |
3200 | return -EFAULT; |
3201 | if (tmp == BRCTL_GET_VERSION) | |
3202 | return BRCTL_VERSION + 1; | |
3203 | return -EINVAL; | |
3204 | } | |
3205 | ||
6b96018b AB |
3206 | static int compat_sock_ioctl_trans(struct file *file, struct socket *sock, |
3207 | unsigned int cmd, unsigned long arg) | |
3208 | { | |
3209 | void __user *argp = compat_ptr(arg); | |
3210 | struct sock *sk = sock->sk; | |
3211 | struct net *net = sock_net(sk); | |
7a229387 | 3212 | |
6b96018b AB |
3213 | if (cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15)) |
3214 | return siocdevprivate_ioctl(net, cmd, argp); | |
3215 | ||
3216 | switch (cmd) { | |
3217 | case SIOCSIFBR: | |
3218 | case SIOCGIFBR: | |
3219 | return old_bridge_ioctl(argp); | |
3220 | case SIOCGIFNAME: | |
3221 | return dev_ifname32(net, argp); | |
3222 | case SIOCGIFCONF: | |
3223 | return dev_ifconf(net, argp); | |
3224 | case SIOCETHTOOL: | |
3225 | return ethtool_ioctl(net, argp); | |
7a50a240 AB |
3226 | case SIOCWANDEV: |
3227 | return compat_siocwandev(net, argp); | |
a2116ed2 AB |
3228 | case SIOCGIFMAP: |
3229 | case SIOCSIFMAP: | |
3230 | return compat_sioc_ifmap(net, cmd, argp); | |
6b96018b AB |
3231 | case SIOCBONDENSLAVE: |
3232 | case SIOCBONDRELEASE: | |
3233 | case SIOCBONDSETHWADDR: | |
3234 | case SIOCBONDSLAVEINFOQUERY: | |
3235 | case SIOCBONDINFOQUERY: | |
3236 | case SIOCBONDCHANGEACTIVE: | |
3237 | return bond_ioctl(net, cmd, argp); | |
3238 | case SIOCADDRT: | |
3239 | case SIOCDELRT: | |
3240 | return routing_ioctl(net, sock, cmd, argp); | |
3241 | case SIOCGSTAMP: | |
3242 | return do_siocgstamp(net, sock, cmd, argp); | |
3243 | case SIOCGSTAMPNS: | |
3244 | return do_siocgstampns(net, sock, cmd, argp); | |
a2116ed2 AB |
3245 | case SIOCSHWTSTAMP: |
3246 | return compat_siocshwtstamp(net, argp); | |
6b96018b AB |
3247 | |
3248 | case FIOSETOWN: | |
3249 | case SIOCSPGRP: | |
3250 | case FIOGETOWN: | |
3251 | case SIOCGPGRP: | |
3252 | case SIOCBRADDBR: | |
3253 | case SIOCBRDELBR: | |
3254 | case SIOCGIFVLAN: | |
3255 | case SIOCSIFVLAN: | |
3256 | case SIOCADDDLCI: | |
3257 | case SIOCDELDLCI: | |
3258 | return sock_ioctl(file, cmd, arg); | |
3259 | ||
3260 | case SIOCGIFFLAGS: | |
3261 | case SIOCSIFFLAGS: | |
3262 | case SIOCGIFMETRIC: | |
3263 | case SIOCSIFMETRIC: | |
3264 | case SIOCGIFMTU: | |
3265 | case SIOCSIFMTU: | |
3266 | case SIOCGIFMEM: | |
3267 | case SIOCSIFMEM: | |
3268 | case SIOCGIFHWADDR: | |
3269 | case SIOCSIFHWADDR: | |
3270 | case SIOCADDMULTI: | |
3271 | case SIOCDELMULTI: | |
3272 | case SIOCGIFINDEX: | |
6b96018b AB |
3273 | case SIOCGIFADDR: |
3274 | case SIOCSIFADDR: | |
3275 | case SIOCSIFHWBROADCAST: | |
6b96018b | 3276 | case SIOCDIFADDR: |
6b96018b AB |
3277 | case SIOCGIFBRDADDR: |
3278 | case SIOCSIFBRDADDR: | |
3279 | case SIOCGIFDSTADDR: | |
3280 | case SIOCSIFDSTADDR: | |
3281 | case SIOCGIFNETMASK: | |
3282 | case SIOCSIFNETMASK: | |
3283 | case SIOCSIFPFLAGS: | |
3284 | case SIOCGIFPFLAGS: | |
3285 | case SIOCGIFTXQLEN: | |
3286 | case SIOCSIFTXQLEN: | |
3287 | case SIOCBRADDIF: | |
3288 | case SIOCBRDELIF: | |
9177efd3 AB |
3289 | case SIOCSIFNAME: |
3290 | case SIOCGMIIPHY: | |
3291 | case SIOCGMIIREG: | |
3292 | case SIOCSMIIREG: | |
6b96018b | 3293 | return dev_ifsioc(net, sock, cmd, argp); |
9177efd3 | 3294 | |
6b96018b AB |
3295 | case SIOCSARP: |
3296 | case SIOCGARP: | |
3297 | case SIOCDARP: | |
6b96018b | 3298 | case SIOCATMARK: |
9177efd3 AB |
3299 | return sock_do_ioctl(net, sock, cmd, arg); |
3300 | } | |
3301 | ||
6b96018b AB |
3302 | return -ENOIOCTLCMD; |
3303 | } | |
7a229387 | 3304 | |
95c96174 | 3305 | static long compat_sock_ioctl(struct file *file, unsigned int cmd, |
89bddce5 | 3306 | unsigned long arg) |
89bbfc95 SP |
3307 | { |
3308 | struct socket *sock = file->private_data; | |
3309 | int ret = -ENOIOCTLCMD; | |
87de87d5 DM |
3310 | struct sock *sk; |
3311 | struct net *net; | |
3312 | ||
3313 | sk = sock->sk; | |
3314 | net = sock_net(sk); | |
89bbfc95 SP |
3315 | |
3316 | if (sock->ops->compat_ioctl) | |
3317 | ret = sock->ops->compat_ioctl(sock, cmd, arg); | |
3318 | ||
87de87d5 DM |
3319 | if (ret == -ENOIOCTLCMD && |
3320 | (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST)) | |
3321 | ret = compat_wext_handle_ioctl(net, cmd, arg); | |
3322 | ||
6b96018b AB |
3323 | if (ret == -ENOIOCTLCMD) |
3324 | ret = compat_sock_ioctl_trans(file, sock, cmd, arg); | |
3325 | ||
89bbfc95 SP |
3326 | return ret; |
3327 | } | |
3328 | #endif | |
3329 | ||
ac5a488e SS |
3330 | int kernel_bind(struct socket *sock, struct sockaddr *addr, int addrlen) |
3331 | { | |
3332 | return sock->ops->bind(sock, addr, addrlen); | |
3333 | } | |
c6d409cf | 3334 | EXPORT_SYMBOL(kernel_bind); |
ac5a488e SS |
3335 | |
3336 | int kernel_listen(struct socket *sock, int backlog) | |
3337 | { | |
3338 | return sock->ops->listen(sock, backlog); | |
3339 | } | |
c6d409cf | 3340 | EXPORT_SYMBOL(kernel_listen); |
ac5a488e SS |
3341 | |
3342 | int kernel_accept(struct socket *sock, struct socket **newsock, int flags) | |
3343 | { | |
3344 | struct sock *sk = sock->sk; | |
3345 | int err; | |
3346 | ||
3347 | err = sock_create_lite(sk->sk_family, sk->sk_type, sk->sk_protocol, | |
3348 | newsock); | |
3349 | if (err < 0) | |
3350 | goto done; | |
3351 | ||
3352 | err = sock->ops->accept(sock, *newsock, flags); | |
3353 | if (err < 0) { | |
3354 | sock_release(*newsock); | |
fa8705b0 | 3355 | *newsock = NULL; |
ac5a488e SS |
3356 | goto done; |
3357 | } | |
3358 | ||
3359 | (*newsock)->ops = sock->ops; | |
1b08534e | 3360 | __module_get((*newsock)->ops->owner); |
ac5a488e SS |
3361 | |
3362 | done: | |
3363 | return err; | |
3364 | } | |
c6d409cf | 3365 | EXPORT_SYMBOL(kernel_accept); |
ac5a488e SS |
3366 | |
3367 | int kernel_connect(struct socket *sock, struct sockaddr *addr, int addrlen, | |
4768fbcb | 3368 | int flags) |
ac5a488e SS |
3369 | { |
3370 | return sock->ops->connect(sock, addr, addrlen, flags); | |
3371 | } | |
c6d409cf | 3372 | EXPORT_SYMBOL(kernel_connect); |
ac5a488e SS |
3373 | |
3374 | int kernel_getsockname(struct socket *sock, struct sockaddr *addr, | |
3375 | int *addrlen) | |
3376 | { | |
3377 | return sock->ops->getname(sock, addr, addrlen, 0); | |
3378 | } | |
c6d409cf | 3379 | EXPORT_SYMBOL(kernel_getsockname); |
ac5a488e SS |
3380 | |
3381 | int kernel_getpeername(struct socket *sock, struct sockaddr *addr, | |
3382 | int *addrlen) | |
3383 | { | |
3384 | return sock->ops->getname(sock, addr, addrlen, 1); | |
3385 | } | |
c6d409cf | 3386 | EXPORT_SYMBOL(kernel_getpeername); |
ac5a488e SS |
3387 | |
3388 | int kernel_getsockopt(struct socket *sock, int level, int optname, | |
3389 | char *optval, int *optlen) | |
3390 | { | |
3391 | mm_segment_t oldfs = get_fs(); | |
fb8621bb NK |
3392 | char __user *uoptval; |
3393 | int __user *uoptlen; | |
ac5a488e SS |
3394 | int err; |
3395 | ||
fb8621bb NK |
3396 | uoptval = (char __user __force *) optval; |
3397 | uoptlen = (int __user __force *) optlen; | |
3398 | ||
ac5a488e SS |
3399 | set_fs(KERNEL_DS); |
3400 | if (level == SOL_SOCKET) | |
fb8621bb | 3401 | err = sock_getsockopt(sock, level, optname, uoptval, uoptlen); |
ac5a488e | 3402 | else |
fb8621bb NK |
3403 | err = sock->ops->getsockopt(sock, level, optname, uoptval, |
3404 | uoptlen); | |
ac5a488e SS |
3405 | set_fs(oldfs); |
3406 | return err; | |
3407 | } | |
c6d409cf | 3408 | EXPORT_SYMBOL(kernel_getsockopt); |
ac5a488e SS |
3409 | |
3410 | int kernel_setsockopt(struct socket *sock, int level, int optname, | |
b7058842 | 3411 | char *optval, unsigned int optlen) |
ac5a488e SS |
3412 | { |
3413 | mm_segment_t oldfs = get_fs(); | |
fb8621bb | 3414 | char __user *uoptval; |
ac5a488e SS |
3415 | int err; |
3416 | ||
fb8621bb NK |
3417 | uoptval = (char __user __force *) optval; |
3418 | ||
ac5a488e SS |
3419 | set_fs(KERNEL_DS); |
3420 | if (level == SOL_SOCKET) | |
fb8621bb | 3421 | err = sock_setsockopt(sock, level, optname, uoptval, optlen); |
ac5a488e | 3422 | else |
fb8621bb | 3423 | err = sock->ops->setsockopt(sock, level, optname, uoptval, |
ac5a488e SS |
3424 | optlen); |
3425 | set_fs(oldfs); | |
3426 | return err; | |
3427 | } | |
c6d409cf | 3428 | EXPORT_SYMBOL(kernel_setsockopt); |
ac5a488e SS |
3429 | |
3430 | int kernel_sendpage(struct socket *sock, struct page *page, int offset, | |
3431 | size_t size, int flags) | |
3432 | { | |
3433 | if (sock->ops->sendpage) | |
3434 | return sock->ops->sendpage(sock, page, offset, size, flags); | |
3435 | ||
3436 | return sock_no_sendpage(sock, page, offset, size, flags); | |
3437 | } | |
c6d409cf | 3438 | EXPORT_SYMBOL(kernel_sendpage); |
ac5a488e SS |
3439 | |
3440 | int kernel_sock_ioctl(struct socket *sock, int cmd, unsigned long arg) | |
3441 | { | |
3442 | mm_segment_t oldfs = get_fs(); | |
3443 | int err; | |
3444 | ||
3445 | set_fs(KERNEL_DS); | |
3446 | err = sock->ops->ioctl(sock, cmd, arg); | |
3447 | set_fs(oldfs); | |
3448 | ||
3449 | return err; | |
3450 | } | |
c6d409cf | 3451 | EXPORT_SYMBOL(kernel_sock_ioctl); |
ac5a488e | 3452 | |
91cf45f0 TM |
3453 | int kernel_sock_shutdown(struct socket *sock, enum sock_shutdown_cmd how) |
3454 | { | |
3455 | return sock->ops->shutdown(sock, how); | |
3456 | } | |
91cf45f0 | 3457 | EXPORT_SYMBOL(kernel_sock_shutdown); |