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> | |
408eccce | 75 | #include <linux/ptp_classify.h> |
1da177e4 LT |
76 | #include <linux/init.h> |
77 | #include <linux/poll.h> | |
78 | #include <linux/cache.h> | |
79 | #include <linux/module.h> | |
80 | #include <linux/highmem.h> | |
1da177e4 LT |
81 | #include <linux/mount.h> |
82 | #include <linux/security.h> | |
83 | #include <linux/syscalls.h> | |
84 | #include <linux/compat.h> | |
85 | #include <linux/kmod.h> | |
3ec3b2fb | 86 | #include <linux/audit.h> |
d86b5e0e | 87 | #include <linux/wireless.h> |
1b8d7ae4 | 88 | #include <linux/nsproxy.h> |
1fd7317d | 89 | #include <linux/magic.h> |
5a0e3ad6 | 90 | #include <linux/slab.h> |
600e1779 | 91 | #include <linux/xattr.h> |
1da177e4 LT |
92 | |
93 | #include <asm/uaccess.h> | |
94 | #include <asm/unistd.h> | |
95 | ||
96 | #include <net/compat.h> | |
87de87d5 | 97 | #include <net/wext.h> |
f8451725 | 98 | #include <net/cls_cgroup.h> |
1da177e4 LT |
99 | |
100 | #include <net/sock.h> | |
101 | #include <linux/netfilter.h> | |
102 | ||
6b96018b AB |
103 | #include <linux/if_tun.h> |
104 | #include <linux/ipv6_route.h> | |
105 | #include <linux/route.h> | |
6b96018b AB |
106 | #include <linux/sockios.h> |
107 | #include <linux/atalk.h> | |
076bb0c8 | 108 | #include <net/busy_poll.h> |
f24b9be5 | 109 | #include <linux/errqueue.h> |
06021292 | 110 | |
e0d1095a | 111 | #ifdef CONFIG_NET_RX_BUSY_POLL |
64b0dc51 ET |
112 | unsigned int sysctl_net_busy_read __read_mostly; |
113 | unsigned int sysctl_net_busy_poll __read_mostly; | |
06021292 | 114 | #endif |
6b96018b | 115 | |
1da177e4 | 116 | static int sock_no_open(struct inode *irrelevant, struct file *dontcare); |
027445c3 BP |
117 | static ssize_t sock_aio_read(struct kiocb *iocb, const struct iovec *iov, |
118 | unsigned long nr_segs, loff_t pos); | |
119 | static ssize_t sock_aio_write(struct kiocb *iocb, const struct iovec *iov, | |
120 | unsigned long nr_segs, loff_t pos); | |
89bddce5 | 121 | static int sock_mmap(struct file *file, struct vm_area_struct *vma); |
1da177e4 LT |
122 | |
123 | static int sock_close(struct inode *inode, struct file *file); | |
124 | static unsigned int sock_poll(struct file *file, | |
125 | struct poll_table_struct *wait); | |
89bddce5 | 126 | static long sock_ioctl(struct file *file, unsigned int cmd, unsigned long arg); |
89bbfc95 SP |
127 | #ifdef CONFIG_COMPAT |
128 | static long compat_sock_ioctl(struct file *file, | |
89bddce5 | 129 | unsigned int cmd, unsigned long arg); |
89bbfc95 | 130 | #endif |
1da177e4 | 131 | static int sock_fasync(int fd, struct file *filp, int on); |
1da177e4 LT |
132 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
133 | int offset, size_t size, loff_t *ppos, int more); | |
9c55e01c | 134 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 135 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c | 136 | unsigned int flags); |
1da177e4 | 137 | |
1da177e4 LT |
138 | /* |
139 | * Socket files have a set of 'special' operations as well as the generic file ones. These don't appear | |
140 | * in the operation structures but are done directly via the socketcall() multiplexor. | |
141 | */ | |
142 | ||
da7071d7 | 143 | static const struct file_operations socket_file_ops = { |
1da177e4 LT |
144 | .owner = THIS_MODULE, |
145 | .llseek = no_llseek, | |
146 | .aio_read = sock_aio_read, | |
147 | .aio_write = sock_aio_write, | |
148 | .poll = sock_poll, | |
149 | .unlocked_ioctl = sock_ioctl, | |
89bbfc95 SP |
150 | #ifdef CONFIG_COMPAT |
151 | .compat_ioctl = compat_sock_ioctl, | |
152 | #endif | |
1da177e4 LT |
153 | .mmap = sock_mmap, |
154 | .open = sock_no_open, /* special open code to disallow open via /proc */ | |
155 | .release = sock_close, | |
156 | .fasync = sock_fasync, | |
5274f052 JA |
157 | .sendpage = sock_sendpage, |
158 | .splice_write = generic_splice_sendpage, | |
9c55e01c | 159 | .splice_read = sock_splice_read, |
1da177e4 LT |
160 | }; |
161 | ||
162 | /* | |
163 | * The protocol list. Each protocol is registered in here. | |
164 | */ | |
165 | ||
1da177e4 | 166 | static DEFINE_SPINLOCK(net_family_lock); |
190683a9 | 167 | static const struct net_proto_family __rcu *net_families[NPROTO] __read_mostly; |
1da177e4 | 168 | |
1da177e4 LT |
169 | /* |
170 | * Statistics counters of the socket lists | |
171 | */ | |
172 | ||
c6d409cf | 173 | static DEFINE_PER_CPU(int, sockets_in_use); |
1da177e4 LT |
174 | |
175 | /* | |
89bddce5 SH |
176 | * Support routines. |
177 | * Move socket addresses back and forth across the kernel/user | |
178 | * divide and look after the messy bits. | |
1da177e4 LT |
179 | */ |
180 | ||
1da177e4 LT |
181 | /** |
182 | * move_addr_to_kernel - copy a socket address into kernel space | |
183 | * @uaddr: Address in user space | |
184 | * @kaddr: Address in kernel space | |
185 | * @ulen: Length in user space | |
186 | * | |
187 | * The address is copied into kernel space. If the provided address is | |
188 | * too long an error code of -EINVAL is returned. If the copy gives | |
189 | * invalid addresses -EFAULT is returned. On a success 0 is returned. | |
190 | */ | |
191 | ||
43db362d | 192 | int move_addr_to_kernel(void __user *uaddr, int ulen, struct sockaddr_storage *kaddr) |
1da177e4 | 193 | { |
230b1839 | 194 | if (ulen < 0 || ulen > sizeof(struct sockaddr_storage)) |
1da177e4 | 195 | return -EINVAL; |
89bddce5 | 196 | if (ulen == 0) |
1da177e4 | 197 | return 0; |
89bddce5 | 198 | if (copy_from_user(kaddr, uaddr, ulen)) |
1da177e4 | 199 | return -EFAULT; |
3ec3b2fb | 200 | return audit_sockaddr(ulen, kaddr); |
1da177e4 LT |
201 | } |
202 | ||
203 | /** | |
204 | * move_addr_to_user - copy an address to user space | |
205 | * @kaddr: kernel space address | |
206 | * @klen: length of address in kernel | |
207 | * @uaddr: user space address | |
208 | * @ulen: pointer to user length field | |
209 | * | |
210 | * The value pointed to by ulen on entry is the buffer length available. | |
211 | * This is overwritten with the buffer space used. -EINVAL is returned | |
212 | * if an overlong buffer is specified or a negative buffer size. -EFAULT | |
213 | * is returned if either the buffer or the length field are not | |
214 | * accessible. | |
215 | * After copying the data up to the limit the user specifies, the true | |
216 | * length of the data is written over the length limit the user | |
217 | * specified. Zero is returned for a success. | |
218 | */ | |
89bddce5 | 219 | |
43db362d | 220 | static int move_addr_to_user(struct sockaddr_storage *kaddr, int klen, |
11165f14 | 221 | void __user *uaddr, int __user *ulen) |
1da177e4 LT |
222 | { |
223 | int err; | |
224 | int len; | |
225 | ||
68c6beb3 | 226 | BUG_ON(klen > sizeof(struct sockaddr_storage)); |
89bddce5 SH |
227 | err = get_user(len, ulen); |
228 | if (err) | |
1da177e4 | 229 | return err; |
89bddce5 SH |
230 | if (len > klen) |
231 | len = klen; | |
68c6beb3 | 232 | if (len < 0) |
1da177e4 | 233 | return -EINVAL; |
89bddce5 | 234 | if (len) { |
d6fe3945 SG |
235 | if (audit_sockaddr(klen, kaddr)) |
236 | return -ENOMEM; | |
89bddce5 | 237 | if (copy_to_user(uaddr, kaddr, len)) |
1da177e4 LT |
238 | return -EFAULT; |
239 | } | |
240 | /* | |
89bddce5 SH |
241 | * "fromlen shall refer to the value before truncation.." |
242 | * 1003.1g | |
1da177e4 LT |
243 | */ |
244 | return __put_user(klen, ulen); | |
245 | } | |
246 | ||
e18b890b | 247 | static struct kmem_cache *sock_inode_cachep __read_mostly; |
1da177e4 LT |
248 | |
249 | static struct inode *sock_alloc_inode(struct super_block *sb) | |
250 | { | |
251 | struct socket_alloc *ei; | |
eaefd110 | 252 | struct socket_wq *wq; |
89bddce5 | 253 | |
e94b1766 | 254 | ei = kmem_cache_alloc(sock_inode_cachep, GFP_KERNEL); |
1da177e4 LT |
255 | if (!ei) |
256 | return NULL; | |
eaefd110 ED |
257 | wq = kmalloc(sizeof(*wq), GFP_KERNEL); |
258 | if (!wq) { | |
43815482 ED |
259 | kmem_cache_free(sock_inode_cachep, ei); |
260 | return NULL; | |
261 | } | |
eaefd110 ED |
262 | init_waitqueue_head(&wq->wait); |
263 | wq->fasync_list = NULL; | |
264 | RCU_INIT_POINTER(ei->socket.wq, wq); | |
89bddce5 | 265 | |
1da177e4 LT |
266 | ei->socket.state = SS_UNCONNECTED; |
267 | ei->socket.flags = 0; | |
268 | ei->socket.ops = NULL; | |
269 | ei->socket.sk = NULL; | |
270 | ei->socket.file = NULL; | |
1da177e4 LT |
271 | |
272 | return &ei->vfs_inode; | |
273 | } | |
274 | ||
275 | static void sock_destroy_inode(struct inode *inode) | |
276 | { | |
43815482 | 277 | struct socket_alloc *ei; |
eaefd110 | 278 | struct socket_wq *wq; |
43815482 ED |
279 | |
280 | ei = container_of(inode, struct socket_alloc, vfs_inode); | |
eaefd110 | 281 | wq = rcu_dereference_protected(ei->socket.wq, 1); |
61845220 | 282 | kfree_rcu(wq, rcu); |
43815482 | 283 | kmem_cache_free(sock_inode_cachep, ei); |
1da177e4 LT |
284 | } |
285 | ||
51cc5068 | 286 | static void init_once(void *foo) |
1da177e4 | 287 | { |
89bddce5 | 288 | struct socket_alloc *ei = (struct socket_alloc *)foo; |
1da177e4 | 289 | |
a35afb83 | 290 | inode_init_once(&ei->vfs_inode); |
1da177e4 | 291 | } |
89bddce5 | 292 | |
1da177e4 LT |
293 | static int init_inodecache(void) |
294 | { | |
295 | sock_inode_cachep = kmem_cache_create("sock_inode_cache", | |
89bddce5 SH |
296 | sizeof(struct socket_alloc), |
297 | 0, | |
298 | (SLAB_HWCACHE_ALIGN | | |
299 | SLAB_RECLAIM_ACCOUNT | | |
300 | SLAB_MEM_SPREAD), | |
20c2df83 | 301 | init_once); |
1da177e4 LT |
302 | if (sock_inode_cachep == NULL) |
303 | return -ENOMEM; | |
304 | return 0; | |
305 | } | |
306 | ||
b87221de | 307 | static const struct super_operations sockfs_ops = { |
c6d409cf ED |
308 | .alloc_inode = sock_alloc_inode, |
309 | .destroy_inode = sock_destroy_inode, | |
310 | .statfs = simple_statfs, | |
1da177e4 LT |
311 | }; |
312 | ||
c23fbb6b ED |
313 | /* |
314 | * sockfs_dname() is called from d_path(). | |
315 | */ | |
316 | static char *sockfs_dname(struct dentry *dentry, char *buffer, int buflen) | |
317 | { | |
318 | return dynamic_dname(dentry, buffer, buflen, "socket:[%lu]", | |
319 | dentry->d_inode->i_ino); | |
320 | } | |
321 | ||
3ba13d17 | 322 | static const struct dentry_operations sockfs_dentry_operations = { |
c23fbb6b | 323 | .d_dname = sockfs_dname, |
1da177e4 LT |
324 | }; |
325 | ||
c74a1cbb AV |
326 | static struct dentry *sockfs_mount(struct file_system_type *fs_type, |
327 | int flags, const char *dev_name, void *data) | |
328 | { | |
329 | return mount_pseudo(fs_type, "socket:", &sockfs_ops, | |
330 | &sockfs_dentry_operations, SOCKFS_MAGIC); | |
331 | } | |
332 | ||
333 | static struct vfsmount *sock_mnt __read_mostly; | |
334 | ||
335 | static struct file_system_type sock_fs_type = { | |
336 | .name = "sockfs", | |
337 | .mount = sockfs_mount, | |
338 | .kill_sb = kill_anon_super, | |
339 | }; | |
340 | ||
1da177e4 LT |
341 | /* |
342 | * Obtains the first available file descriptor and sets it up for use. | |
343 | * | |
39d8c1b6 DM |
344 | * These functions create file structures and maps them to fd space |
345 | * of the current process. On success it returns file descriptor | |
1da177e4 LT |
346 | * and file struct implicitly stored in sock->file. |
347 | * Note that another thread may close file descriptor before we return | |
348 | * from this function. We use the fact that now we do not refer | |
349 | * to socket after mapping. If one day we will need it, this | |
350 | * function will increment ref. count on file by 1. | |
351 | * | |
352 | * In any case returned fd MAY BE not valid! | |
353 | * This race condition is unavoidable | |
354 | * with shared fd spaces, we cannot solve it inside kernel, | |
355 | * but we take care of internal coherence yet. | |
356 | */ | |
357 | ||
aab174f0 | 358 | struct file *sock_alloc_file(struct socket *sock, int flags, const char *dname) |
1da177e4 | 359 | { |
7cbe66b6 | 360 | struct qstr name = { .name = "" }; |
2c48b9c4 | 361 | struct path path; |
7cbe66b6 | 362 | struct file *file; |
1da177e4 | 363 | |
600e1779 MY |
364 | if (dname) { |
365 | name.name = dname; | |
366 | name.len = strlen(name.name); | |
367 | } else if (sock->sk) { | |
368 | name.name = sock->sk->sk_prot_creator->name; | |
369 | name.len = strlen(name.name); | |
370 | } | |
4b936885 | 371 | path.dentry = d_alloc_pseudo(sock_mnt->mnt_sb, &name); |
28407630 AV |
372 | if (unlikely(!path.dentry)) |
373 | return ERR_PTR(-ENOMEM); | |
2c48b9c4 | 374 | path.mnt = mntget(sock_mnt); |
39d8c1b6 | 375 | |
2c48b9c4 | 376 | d_instantiate(path.dentry, SOCK_INODE(sock)); |
cc3808f8 | 377 | SOCK_INODE(sock)->i_fop = &socket_file_ops; |
39d8c1b6 | 378 | |
2c48b9c4 | 379 | file = alloc_file(&path, FMODE_READ | FMODE_WRITE, |
ce8d2cdf | 380 | &socket_file_ops); |
39b65252 | 381 | if (unlikely(IS_ERR(file))) { |
cc3808f8 | 382 | /* drop dentry, keep inode */ |
7de9c6ee | 383 | ihold(path.dentry->d_inode); |
2c48b9c4 | 384 | path_put(&path); |
39b65252 | 385 | return file; |
cc3808f8 AV |
386 | } |
387 | ||
388 | sock->file = file; | |
77d27200 | 389 | file->f_flags = O_RDWR | (flags & O_NONBLOCK); |
39d8c1b6 | 390 | file->private_data = sock; |
28407630 | 391 | return file; |
39d8c1b6 | 392 | } |
56b31d1c | 393 | EXPORT_SYMBOL(sock_alloc_file); |
39d8c1b6 | 394 | |
56b31d1c | 395 | static int sock_map_fd(struct socket *sock, int flags) |
39d8c1b6 DM |
396 | { |
397 | struct file *newfile; | |
28407630 AV |
398 | int fd = get_unused_fd_flags(flags); |
399 | if (unlikely(fd < 0)) | |
400 | return fd; | |
39d8c1b6 | 401 | |
aab174f0 | 402 | newfile = sock_alloc_file(sock, flags, NULL); |
28407630 | 403 | if (likely(!IS_ERR(newfile))) { |
39d8c1b6 | 404 | fd_install(fd, newfile); |
28407630 AV |
405 | return fd; |
406 | } | |
7cbe66b6 | 407 | |
28407630 AV |
408 | put_unused_fd(fd); |
409 | return PTR_ERR(newfile); | |
1da177e4 LT |
410 | } |
411 | ||
406a3c63 | 412 | struct socket *sock_from_file(struct file *file, int *err) |
6cb153ca | 413 | { |
6cb153ca BL |
414 | if (file->f_op == &socket_file_ops) |
415 | return file->private_data; /* set in sock_map_fd */ | |
416 | ||
23bb80d2 ED |
417 | *err = -ENOTSOCK; |
418 | return NULL; | |
6cb153ca | 419 | } |
406a3c63 | 420 | EXPORT_SYMBOL(sock_from_file); |
6cb153ca | 421 | |
1da177e4 | 422 | /** |
c6d409cf | 423 | * sockfd_lookup - Go from a file number to its socket slot |
1da177e4 LT |
424 | * @fd: file handle |
425 | * @err: pointer to an error code return | |
426 | * | |
427 | * The file handle passed in is locked and the socket it is bound | |
428 | * too is returned. If an error occurs the err pointer is overwritten | |
429 | * with a negative errno code and NULL is returned. The function checks | |
430 | * for both invalid handles and passing a handle which is not a socket. | |
431 | * | |
432 | * On a success the socket object pointer is returned. | |
433 | */ | |
434 | ||
435 | struct socket *sockfd_lookup(int fd, int *err) | |
436 | { | |
437 | struct file *file; | |
1da177e4 LT |
438 | struct socket *sock; |
439 | ||
89bddce5 SH |
440 | file = fget(fd); |
441 | if (!file) { | |
1da177e4 LT |
442 | *err = -EBADF; |
443 | return NULL; | |
444 | } | |
89bddce5 | 445 | |
6cb153ca BL |
446 | sock = sock_from_file(file, err); |
447 | if (!sock) | |
1da177e4 | 448 | fput(file); |
6cb153ca BL |
449 | return sock; |
450 | } | |
c6d409cf | 451 | EXPORT_SYMBOL(sockfd_lookup); |
1da177e4 | 452 | |
6cb153ca BL |
453 | static struct socket *sockfd_lookup_light(int fd, int *err, int *fput_needed) |
454 | { | |
00e188ef | 455 | struct fd f = fdget(fd); |
6cb153ca BL |
456 | struct socket *sock; |
457 | ||
3672558c | 458 | *err = -EBADF; |
00e188ef AV |
459 | if (f.file) { |
460 | sock = sock_from_file(f.file, err); | |
461 | if (likely(sock)) { | |
462 | *fput_needed = f.flags; | |
6cb153ca | 463 | return sock; |
00e188ef AV |
464 | } |
465 | fdput(f); | |
1da177e4 | 466 | } |
6cb153ca | 467 | return NULL; |
1da177e4 LT |
468 | } |
469 | ||
600e1779 MY |
470 | #define XATTR_SOCKPROTONAME_SUFFIX "sockprotoname" |
471 | #define XATTR_NAME_SOCKPROTONAME (XATTR_SYSTEM_PREFIX XATTR_SOCKPROTONAME_SUFFIX) | |
472 | #define XATTR_NAME_SOCKPROTONAME_LEN (sizeof(XATTR_NAME_SOCKPROTONAME)-1) | |
473 | static ssize_t sockfs_getxattr(struct dentry *dentry, | |
474 | const char *name, void *value, size_t size) | |
475 | { | |
476 | const char *proto_name; | |
477 | size_t proto_size; | |
478 | int error; | |
479 | ||
480 | error = -ENODATA; | |
481 | if (!strncmp(name, XATTR_NAME_SOCKPROTONAME, XATTR_NAME_SOCKPROTONAME_LEN)) { | |
482 | proto_name = dentry->d_name.name; | |
483 | proto_size = strlen(proto_name); | |
484 | ||
485 | if (value) { | |
486 | error = -ERANGE; | |
487 | if (proto_size + 1 > size) | |
488 | goto out; | |
489 | ||
490 | strncpy(value, proto_name, proto_size + 1); | |
491 | } | |
492 | error = proto_size + 1; | |
493 | } | |
494 | ||
495 | out: | |
496 | return error; | |
497 | } | |
498 | ||
499 | static ssize_t sockfs_listxattr(struct dentry *dentry, char *buffer, | |
500 | size_t size) | |
501 | { | |
502 | ssize_t len; | |
503 | ssize_t used = 0; | |
504 | ||
505 | len = security_inode_listsecurity(dentry->d_inode, buffer, size); | |
506 | if (len < 0) | |
507 | return len; | |
508 | used += len; | |
509 | if (buffer) { | |
510 | if (size < used) | |
511 | return -ERANGE; | |
512 | buffer += len; | |
513 | } | |
514 | ||
515 | len = (XATTR_NAME_SOCKPROTONAME_LEN + 1); | |
516 | used += len; | |
517 | if (buffer) { | |
518 | if (size < used) | |
519 | return -ERANGE; | |
520 | memcpy(buffer, XATTR_NAME_SOCKPROTONAME, len); | |
521 | buffer += len; | |
522 | } | |
523 | ||
524 | return used; | |
525 | } | |
526 | ||
527 | static const struct inode_operations sockfs_inode_ops = { | |
528 | .getxattr = sockfs_getxattr, | |
529 | .listxattr = sockfs_listxattr, | |
530 | }; | |
531 | ||
1da177e4 LT |
532 | /** |
533 | * sock_alloc - allocate a socket | |
89bddce5 | 534 | * |
1da177e4 LT |
535 | * Allocate a new inode and socket object. The two are bound together |
536 | * and initialised. The socket is then returned. If we are out of inodes | |
537 | * NULL is returned. | |
538 | */ | |
539 | ||
540 | static struct socket *sock_alloc(void) | |
541 | { | |
89bddce5 SH |
542 | struct inode *inode; |
543 | struct socket *sock; | |
1da177e4 | 544 | |
a209dfc7 | 545 | inode = new_inode_pseudo(sock_mnt->mnt_sb); |
1da177e4 LT |
546 | if (!inode) |
547 | return NULL; | |
548 | ||
549 | sock = SOCKET_I(inode); | |
550 | ||
29a020d3 | 551 | kmemcheck_annotate_bitfield(sock, type); |
85fe4025 | 552 | inode->i_ino = get_next_ino(); |
89bddce5 | 553 | inode->i_mode = S_IFSOCK | S_IRWXUGO; |
8192b0c4 DH |
554 | inode->i_uid = current_fsuid(); |
555 | inode->i_gid = current_fsgid(); | |
600e1779 | 556 | inode->i_op = &sockfs_inode_ops; |
1da177e4 | 557 | |
19e8d69c | 558 | this_cpu_add(sockets_in_use, 1); |
1da177e4 LT |
559 | return sock; |
560 | } | |
561 | ||
562 | /* | |
563 | * In theory you can't get an open on this inode, but /proc provides | |
564 | * a back door. Remember to keep it shut otherwise you'll let the | |
565 | * creepy crawlies in. | |
566 | */ | |
89bddce5 | 567 | |
1da177e4 LT |
568 | static int sock_no_open(struct inode *irrelevant, struct file *dontcare) |
569 | { | |
570 | return -ENXIO; | |
571 | } | |
572 | ||
4b6f5d20 | 573 | const struct file_operations bad_sock_fops = { |
1da177e4 LT |
574 | .owner = THIS_MODULE, |
575 | .open = sock_no_open, | |
6038f373 | 576 | .llseek = noop_llseek, |
1da177e4 LT |
577 | }; |
578 | ||
579 | /** | |
580 | * sock_release - close a socket | |
581 | * @sock: socket to close | |
582 | * | |
583 | * The socket is released from the protocol stack if it has a release | |
584 | * callback, and the inode is then released if the socket is bound to | |
89bddce5 | 585 | * an inode not a file. |
1da177e4 | 586 | */ |
89bddce5 | 587 | |
1da177e4 LT |
588 | void sock_release(struct socket *sock) |
589 | { | |
590 | if (sock->ops) { | |
591 | struct module *owner = sock->ops->owner; | |
592 | ||
593 | sock->ops->release(sock); | |
594 | sock->ops = NULL; | |
595 | module_put(owner); | |
596 | } | |
597 | ||
eaefd110 | 598 | if (rcu_dereference_protected(sock->wq, 1)->fasync_list) |
3410f22e | 599 | pr_err("%s: fasync list not empty!\n", __func__); |
1da177e4 | 600 | |
b09e786b MP |
601 | if (test_bit(SOCK_EXTERNALLY_ALLOCATED, &sock->flags)) |
602 | return; | |
603 | ||
19e8d69c | 604 | this_cpu_sub(sockets_in_use, 1); |
1da177e4 LT |
605 | if (!sock->file) { |
606 | iput(SOCK_INODE(sock)); | |
607 | return; | |
608 | } | |
89bddce5 | 609 | sock->file = NULL; |
1da177e4 | 610 | } |
c6d409cf | 611 | EXPORT_SYMBOL(sock_release); |
1da177e4 | 612 | |
67cc0d40 | 613 | void __sock_tx_timestamp(const struct sock *sk, __u8 *tx_flags) |
20d49473 | 614 | { |
140c55d4 ED |
615 | u8 flags = *tx_flags; |
616 | ||
b9f40e21 | 617 | if (sk->sk_tsflags & SOF_TIMESTAMPING_TX_HARDWARE) |
140c55d4 ED |
618 | flags |= SKBTX_HW_TSTAMP; |
619 | ||
b9f40e21 | 620 | if (sk->sk_tsflags & SOF_TIMESTAMPING_TX_SOFTWARE) |
140c55d4 ED |
621 | flags |= SKBTX_SW_TSTAMP; |
622 | ||
e7fd2885 | 623 | if (sk->sk_tsflags & SOF_TIMESTAMPING_TX_SCHED) |
140c55d4 ED |
624 | flags |= SKBTX_SCHED_TSTAMP; |
625 | ||
e1c8a607 | 626 | if (sk->sk_tsflags & SOF_TIMESTAMPING_TX_ACK) |
140c55d4 | 627 | flags |= SKBTX_ACK_TSTAMP; |
e7fd2885 | 628 | |
140c55d4 | 629 | *tx_flags = flags; |
20d49473 | 630 | } |
67cc0d40 | 631 | EXPORT_SYMBOL(__sock_tx_timestamp); |
20d49473 | 632 | |
228e548e AB |
633 | static inline int __sock_sendmsg_nosec(struct kiocb *iocb, struct socket *sock, |
634 | struct msghdr *msg, size_t size) | |
1da177e4 LT |
635 | { |
636 | struct sock_iocb *si = kiocb_to_siocb(iocb); | |
1da177e4 LT |
637 | |
638 | si->sock = sock; | |
639 | si->scm = NULL; | |
640 | si->msg = msg; | |
641 | si->size = size; | |
642 | ||
1da177e4 LT |
643 | return sock->ops->sendmsg(iocb, sock, msg, size); |
644 | } | |
645 | ||
228e548e AB |
646 | static inline int __sock_sendmsg(struct kiocb *iocb, struct socket *sock, |
647 | struct msghdr *msg, size_t size) | |
648 | { | |
649 | int err = security_socket_sendmsg(sock, msg, size); | |
650 | ||
651 | return err ?: __sock_sendmsg_nosec(iocb, sock, msg, size); | |
652 | } | |
653 | ||
0cf00c6f GZ |
654 | static int do_sock_sendmsg(struct socket *sock, struct msghdr *msg, |
655 | size_t size, bool nosec) | |
1da177e4 LT |
656 | { |
657 | struct kiocb iocb; | |
658 | struct sock_iocb siocb; | |
659 | int ret; | |
660 | ||
661 | init_sync_kiocb(&iocb, NULL); | |
662 | iocb.private = &siocb; | |
0cf00c6f GZ |
663 | ret = nosec ? __sock_sendmsg_nosec(&iocb, sock, msg, size) : |
664 | __sock_sendmsg(&iocb, sock, msg, size); | |
1da177e4 LT |
665 | if (-EIOCBQUEUED == ret) |
666 | ret = wait_on_sync_kiocb(&iocb); | |
667 | return ret; | |
668 | } | |
0cf00c6f GZ |
669 | |
670 | int sock_sendmsg(struct socket *sock, struct msghdr *msg, size_t size) | |
671 | { | |
672 | return do_sock_sendmsg(sock, msg, size, false); | |
673 | } | |
c6d409cf | 674 | EXPORT_SYMBOL(sock_sendmsg); |
1da177e4 | 675 | |
894dc24c | 676 | static int sock_sendmsg_nosec(struct socket *sock, struct msghdr *msg, size_t size) |
228e548e | 677 | { |
0cf00c6f | 678 | return do_sock_sendmsg(sock, msg, size, true); |
228e548e AB |
679 | } |
680 | ||
1da177e4 LT |
681 | int kernel_sendmsg(struct socket *sock, struct msghdr *msg, |
682 | struct kvec *vec, size_t num, size_t size) | |
683 | { | |
684 | mm_segment_t oldfs = get_fs(); | |
685 | int result; | |
686 | ||
687 | set_fs(KERNEL_DS); | |
688 | /* | |
689 | * the following is safe, since for compiler definitions of kvec and | |
690 | * iovec are identical, yielding the same in-core layout and alignment | |
691 | */ | |
89bddce5 | 692 | msg->msg_iov = (struct iovec *)vec; |
1da177e4 LT |
693 | msg->msg_iovlen = num; |
694 | result = sock_sendmsg(sock, msg, size); | |
695 | set_fs(oldfs); | |
696 | return result; | |
697 | } | |
c6d409cf | 698 | EXPORT_SYMBOL(kernel_sendmsg); |
1da177e4 | 699 | |
92f37fd2 ED |
700 | /* |
701 | * called from sock_recv_timestamp() if sock_flag(sk, SOCK_RCVTSTAMP) | |
702 | */ | |
703 | void __sock_recv_timestamp(struct msghdr *msg, struct sock *sk, | |
704 | struct sk_buff *skb) | |
705 | { | |
20d49473 | 706 | int need_software_tstamp = sock_flag(sk, SOCK_RCVTSTAMP); |
f24b9be5 | 707 | struct scm_timestamping tss; |
20d49473 PO |
708 | int empty = 1; |
709 | struct skb_shared_hwtstamps *shhwtstamps = | |
710 | skb_hwtstamps(skb); | |
711 | ||
712 | /* Race occurred between timestamp enabling and packet | |
713 | receiving. Fill in the current time for now. */ | |
714 | if (need_software_tstamp && skb->tstamp.tv64 == 0) | |
715 | __net_timestamp(skb); | |
716 | ||
717 | if (need_software_tstamp) { | |
718 | if (!sock_flag(sk, SOCK_RCVTSTAMPNS)) { | |
719 | struct timeval tv; | |
720 | skb_get_timestamp(skb, &tv); | |
721 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMP, | |
722 | sizeof(tv), &tv); | |
723 | } else { | |
f24b9be5 WB |
724 | struct timespec ts; |
725 | skb_get_timestampns(skb, &ts); | |
20d49473 | 726 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPNS, |
f24b9be5 | 727 | sizeof(ts), &ts); |
20d49473 PO |
728 | } |
729 | } | |
730 | ||
f24b9be5 | 731 | memset(&tss, 0, sizeof(tss)); |
c199105d | 732 | if ((sk->sk_tsflags & SOF_TIMESTAMPING_SOFTWARE) && |
f24b9be5 | 733 | ktime_to_timespec_cond(skb->tstamp, tss.ts + 0)) |
20d49473 | 734 | empty = 0; |
4d276eb6 | 735 | if (shhwtstamps && |
b9f40e21 | 736 | (sk->sk_tsflags & SOF_TIMESTAMPING_RAW_HARDWARE) && |
f24b9be5 | 737 | ktime_to_timespec_cond(shhwtstamps->hwtstamp, tss.ts + 2)) |
4d276eb6 | 738 | empty = 0; |
20d49473 PO |
739 | if (!empty) |
740 | put_cmsg(msg, SOL_SOCKET, | |
f24b9be5 | 741 | SCM_TIMESTAMPING, sizeof(tss), &tss); |
92f37fd2 | 742 | } |
7c81fd8b ACM |
743 | EXPORT_SYMBOL_GPL(__sock_recv_timestamp); |
744 | ||
6e3e939f JB |
745 | void __sock_recv_wifi_status(struct msghdr *msg, struct sock *sk, |
746 | struct sk_buff *skb) | |
747 | { | |
748 | int ack; | |
749 | ||
750 | if (!sock_flag(sk, SOCK_WIFI_STATUS)) | |
751 | return; | |
752 | if (!skb->wifi_acked_valid) | |
753 | return; | |
754 | ||
755 | ack = skb->wifi_acked; | |
756 | ||
757 | put_cmsg(msg, SOL_SOCKET, SCM_WIFI_STATUS, sizeof(ack), &ack); | |
758 | } | |
759 | EXPORT_SYMBOL_GPL(__sock_recv_wifi_status); | |
760 | ||
11165f14 | 761 | static inline void sock_recv_drops(struct msghdr *msg, struct sock *sk, |
762 | struct sk_buff *skb) | |
3b885787 NH |
763 | { |
764 | if (sock_flag(sk, SOCK_RXQ_OVFL) && skb && skb->dropcount) | |
765 | put_cmsg(msg, SOL_SOCKET, SO_RXQ_OVFL, | |
766 | sizeof(__u32), &skb->dropcount); | |
767 | } | |
768 | ||
767dd033 | 769 | void __sock_recv_ts_and_drops(struct msghdr *msg, struct sock *sk, |
3b885787 NH |
770 | struct sk_buff *skb) |
771 | { | |
772 | sock_recv_timestamp(msg, sk, skb); | |
773 | sock_recv_drops(msg, sk, skb); | |
774 | } | |
767dd033 | 775 | EXPORT_SYMBOL_GPL(__sock_recv_ts_and_drops); |
3b885787 | 776 | |
a2e27255 ACM |
777 | static inline int __sock_recvmsg_nosec(struct kiocb *iocb, struct socket *sock, |
778 | struct msghdr *msg, size_t size, int flags) | |
1da177e4 | 779 | { |
1da177e4 LT |
780 | struct sock_iocb *si = kiocb_to_siocb(iocb); |
781 | ||
782 | si->sock = sock; | |
783 | si->scm = NULL; | |
784 | si->msg = msg; | |
785 | si->size = size; | |
786 | si->flags = flags; | |
787 | ||
1da177e4 LT |
788 | return sock->ops->recvmsg(iocb, sock, msg, size, flags); |
789 | } | |
790 | ||
a2e27255 ACM |
791 | static inline int __sock_recvmsg(struct kiocb *iocb, struct socket *sock, |
792 | struct msghdr *msg, size_t size, int flags) | |
793 | { | |
794 | int err = security_socket_recvmsg(sock, msg, size, flags); | |
795 | ||
796 | return err ?: __sock_recvmsg_nosec(iocb, sock, msg, size, flags); | |
797 | } | |
798 | ||
89bddce5 | 799 | int sock_recvmsg(struct socket *sock, struct msghdr *msg, |
1da177e4 LT |
800 | size_t size, int flags) |
801 | { | |
802 | struct kiocb iocb; | |
803 | struct sock_iocb siocb; | |
804 | int ret; | |
805 | ||
89bddce5 | 806 | init_sync_kiocb(&iocb, NULL); |
1da177e4 LT |
807 | iocb.private = &siocb; |
808 | ret = __sock_recvmsg(&iocb, sock, msg, size, flags); | |
809 | if (-EIOCBQUEUED == ret) | |
810 | ret = wait_on_sync_kiocb(&iocb); | |
811 | return ret; | |
812 | } | |
c6d409cf | 813 | EXPORT_SYMBOL(sock_recvmsg); |
1da177e4 | 814 | |
a2e27255 ACM |
815 | static int sock_recvmsg_nosec(struct socket *sock, struct msghdr *msg, |
816 | size_t size, int flags) | |
817 | { | |
818 | struct kiocb iocb; | |
819 | struct sock_iocb siocb; | |
820 | int ret; | |
821 | ||
822 | init_sync_kiocb(&iocb, NULL); | |
823 | iocb.private = &siocb; | |
824 | ret = __sock_recvmsg_nosec(&iocb, sock, msg, size, flags); | |
825 | if (-EIOCBQUEUED == ret) | |
826 | ret = wait_on_sync_kiocb(&iocb); | |
827 | return ret; | |
828 | } | |
829 | ||
c1249c0a ML |
830 | /** |
831 | * kernel_recvmsg - Receive a message from a socket (kernel space) | |
832 | * @sock: The socket to receive the message from | |
833 | * @msg: Received message | |
834 | * @vec: Input s/g array for message data | |
835 | * @num: Size of input s/g array | |
836 | * @size: Number of bytes to read | |
837 | * @flags: Message flags (MSG_DONTWAIT, etc...) | |
838 | * | |
839 | * On return the msg structure contains the scatter/gather array passed in the | |
840 | * vec argument. The array is modified so that it consists of the unfilled | |
841 | * portion of the original array. | |
842 | * | |
843 | * The returned value is the total number of bytes received, or an error. | |
844 | */ | |
89bddce5 SH |
845 | int kernel_recvmsg(struct socket *sock, struct msghdr *msg, |
846 | struct kvec *vec, size_t num, size_t size, int flags) | |
1da177e4 LT |
847 | { |
848 | mm_segment_t oldfs = get_fs(); | |
849 | int result; | |
850 | ||
851 | set_fs(KERNEL_DS); | |
852 | /* | |
853 | * the following is safe, since for compiler definitions of kvec and | |
854 | * iovec are identical, yielding the same in-core layout and alignment | |
855 | */ | |
89bddce5 | 856 | msg->msg_iov = (struct iovec *)vec, msg->msg_iovlen = num; |
1da177e4 LT |
857 | result = sock_recvmsg(sock, msg, size, flags); |
858 | set_fs(oldfs); | |
859 | return result; | |
860 | } | |
c6d409cf | 861 | EXPORT_SYMBOL(kernel_recvmsg); |
1da177e4 | 862 | |
ce1d4d3e CH |
863 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
864 | int offset, size_t size, loff_t *ppos, int more) | |
1da177e4 | 865 | { |
1da177e4 LT |
866 | struct socket *sock; |
867 | int flags; | |
868 | ||
ce1d4d3e CH |
869 | sock = file->private_data; |
870 | ||
35f9c09f ED |
871 | flags = (file->f_flags & O_NONBLOCK) ? MSG_DONTWAIT : 0; |
872 | /* more is a combination of MSG_MORE and MSG_SENDPAGE_NOTLAST */ | |
873 | flags |= more; | |
ce1d4d3e | 874 | |
e6949583 | 875 | return kernel_sendpage(sock, page, offset, size, flags); |
ce1d4d3e | 876 | } |
1da177e4 | 877 | |
9c55e01c | 878 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 879 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c JA |
880 | unsigned int flags) |
881 | { | |
882 | struct socket *sock = file->private_data; | |
883 | ||
997b37da RDC |
884 | if (unlikely(!sock->ops->splice_read)) |
885 | return -EINVAL; | |
886 | ||
9c55e01c JA |
887 | return sock->ops->splice_read(sock, ppos, pipe, len, flags); |
888 | } | |
889 | ||
ce1d4d3e | 890 | static struct sock_iocb *alloc_sock_iocb(struct kiocb *iocb, |
89bddce5 | 891 | struct sock_iocb *siocb) |
ce1d4d3e | 892 | { |
d29c445b KO |
893 | if (!is_sync_kiocb(iocb)) |
894 | BUG(); | |
1da177e4 | 895 | |
ce1d4d3e | 896 | siocb->kiocb = iocb; |
ce1d4d3e CH |
897 | iocb->private = siocb; |
898 | return siocb; | |
1da177e4 LT |
899 | } |
900 | ||
ce1d4d3e | 901 | static ssize_t do_sock_read(struct msghdr *msg, struct kiocb *iocb, |
027445c3 BP |
902 | struct file *file, const struct iovec *iov, |
903 | unsigned long nr_segs) | |
ce1d4d3e CH |
904 | { |
905 | struct socket *sock = file->private_data; | |
906 | size_t size = 0; | |
907 | int i; | |
1da177e4 | 908 | |
89bddce5 SH |
909 | for (i = 0; i < nr_segs; i++) |
910 | size += iov[i].iov_len; | |
1da177e4 | 911 | |
ce1d4d3e CH |
912 | msg->msg_name = NULL; |
913 | msg->msg_namelen = 0; | |
914 | msg->msg_control = NULL; | |
915 | msg->msg_controllen = 0; | |
89bddce5 | 916 | msg->msg_iov = (struct iovec *)iov; |
ce1d4d3e CH |
917 | msg->msg_iovlen = nr_segs; |
918 | msg->msg_flags = (file->f_flags & O_NONBLOCK) ? MSG_DONTWAIT : 0; | |
919 | ||
920 | return __sock_recvmsg(iocb, sock, msg, size, msg->msg_flags); | |
921 | } | |
922 | ||
027445c3 BP |
923 | static ssize_t sock_aio_read(struct kiocb *iocb, const struct iovec *iov, |
924 | unsigned long nr_segs, loff_t pos) | |
ce1d4d3e CH |
925 | { |
926 | struct sock_iocb siocb, *x; | |
927 | ||
1da177e4 LT |
928 | if (pos != 0) |
929 | return -ESPIPE; | |
027445c3 | 930 | |
73a7075e | 931 | if (iocb->ki_nbytes == 0) /* Match SYS5 behaviour */ |
1da177e4 LT |
932 | return 0; |
933 | ||
027445c3 BP |
934 | |
935 | x = alloc_sock_iocb(iocb, &siocb); | |
ce1d4d3e CH |
936 | if (!x) |
937 | return -ENOMEM; | |
027445c3 | 938 | return do_sock_read(&x->async_msg, iocb, iocb->ki_filp, iov, nr_segs); |
1da177e4 LT |
939 | } |
940 | ||
ce1d4d3e | 941 | static ssize_t do_sock_write(struct msghdr *msg, struct kiocb *iocb, |
027445c3 BP |
942 | struct file *file, const struct iovec *iov, |
943 | unsigned long nr_segs) | |
1da177e4 | 944 | { |
ce1d4d3e CH |
945 | struct socket *sock = file->private_data; |
946 | size_t size = 0; | |
947 | int i; | |
1da177e4 | 948 | |
89bddce5 SH |
949 | for (i = 0; i < nr_segs; i++) |
950 | size += iov[i].iov_len; | |
1da177e4 | 951 | |
ce1d4d3e CH |
952 | msg->msg_name = NULL; |
953 | msg->msg_namelen = 0; | |
954 | msg->msg_control = NULL; | |
955 | msg->msg_controllen = 0; | |
89bddce5 | 956 | msg->msg_iov = (struct iovec *)iov; |
ce1d4d3e CH |
957 | msg->msg_iovlen = nr_segs; |
958 | msg->msg_flags = (file->f_flags & O_NONBLOCK) ? MSG_DONTWAIT : 0; | |
959 | if (sock->type == SOCK_SEQPACKET) | |
960 | msg->msg_flags |= MSG_EOR; | |
1da177e4 | 961 | |
ce1d4d3e | 962 | return __sock_sendmsg(iocb, sock, msg, size); |
1da177e4 LT |
963 | } |
964 | ||
027445c3 BP |
965 | static ssize_t sock_aio_write(struct kiocb *iocb, const struct iovec *iov, |
966 | unsigned long nr_segs, loff_t pos) | |
ce1d4d3e CH |
967 | { |
968 | struct sock_iocb siocb, *x; | |
1da177e4 | 969 | |
ce1d4d3e CH |
970 | if (pos != 0) |
971 | return -ESPIPE; | |
027445c3 | 972 | |
027445c3 | 973 | x = alloc_sock_iocb(iocb, &siocb); |
ce1d4d3e CH |
974 | if (!x) |
975 | return -ENOMEM; | |
1da177e4 | 976 | |
027445c3 | 977 | return do_sock_write(&x->async_msg, iocb, iocb->ki_filp, iov, nr_segs); |
1da177e4 LT |
978 | } |
979 | ||
1da177e4 LT |
980 | /* |
981 | * Atomic setting of ioctl hooks to avoid race | |
982 | * with module unload. | |
983 | */ | |
984 | ||
4a3e2f71 | 985 | static DEFINE_MUTEX(br_ioctl_mutex); |
c6d409cf | 986 | static int (*br_ioctl_hook) (struct net *, unsigned int cmd, void __user *arg); |
1da177e4 | 987 | |
881d966b | 988 | void brioctl_set(int (*hook) (struct net *, unsigned int, void __user *)) |
1da177e4 | 989 | { |
4a3e2f71 | 990 | mutex_lock(&br_ioctl_mutex); |
1da177e4 | 991 | br_ioctl_hook = hook; |
4a3e2f71 | 992 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
993 | } |
994 | EXPORT_SYMBOL(brioctl_set); | |
995 | ||
4a3e2f71 | 996 | static DEFINE_MUTEX(vlan_ioctl_mutex); |
881d966b | 997 | static int (*vlan_ioctl_hook) (struct net *, void __user *arg); |
1da177e4 | 998 | |
881d966b | 999 | void vlan_ioctl_set(int (*hook) (struct net *, void __user *)) |
1da177e4 | 1000 | { |
4a3e2f71 | 1001 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 1002 | vlan_ioctl_hook = hook; |
4a3e2f71 | 1003 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 LT |
1004 | } |
1005 | EXPORT_SYMBOL(vlan_ioctl_set); | |
1006 | ||
4a3e2f71 | 1007 | static DEFINE_MUTEX(dlci_ioctl_mutex); |
89bddce5 | 1008 | static int (*dlci_ioctl_hook) (unsigned int, void __user *); |
1da177e4 | 1009 | |
89bddce5 | 1010 | void dlci_ioctl_set(int (*hook) (unsigned int, void __user *)) |
1da177e4 | 1011 | { |
4a3e2f71 | 1012 | mutex_lock(&dlci_ioctl_mutex); |
1da177e4 | 1013 | dlci_ioctl_hook = hook; |
4a3e2f71 | 1014 | mutex_unlock(&dlci_ioctl_mutex); |
1da177e4 LT |
1015 | } |
1016 | EXPORT_SYMBOL(dlci_ioctl_set); | |
1017 | ||
6b96018b AB |
1018 | static long sock_do_ioctl(struct net *net, struct socket *sock, |
1019 | unsigned int cmd, unsigned long arg) | |
1020 | { | |
1021 | int err; | |
1022 | void __user *argp = (void __user *)arg; | |
1023 | ||
1024 | err = sock->ops->ioctl(sock, cmd, arg); | |
1025 | ||
1026 | /* | |
1027 | * If this ioctl is unknown try to hand it down | |
1028 | * to the NIC driver. | |
1029 | */ | |
1030 | if (err == -ENOIOCTLCMD) | |
1031 | err = dev_ioctl(net, cmd, argp); | |
1032 | ||
1033 | return err; | |
1034 | } | |
1035 | ||
1da177e4 LT |
1036 | /* |
1037 | * With an ioctl, arg may well be a user mode pointer, but we don't know | |
1038 | * what to do with it - that's up to the protocol still. | |
1039 | */ | |
1040 | ||
1041 | static long sock_ioctl(struct file *file, unsigned cmd, unsigned long arg) | |
1042 | { | |
1043 | struct socket *sock; | |
881d966b | 1044 | struct sock *sk; |
1da177e4 LT |
1045 | void __user *argp = (void __user *)arg; |
1046 | int pid, err; | |
881d966b | 1047 | struct net *net; |
1da177e4 | 1048 | |
b69aee04 | 1049 | sock = file->private_data; |
881d966b | 1050 | sk = sock->sk; |
3b1e0a65 | 1051 | net = sock_net(sk); |
1da177e4 | 1052 | if (cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15)) { |
881d966b | 1053 | err = dev_ioctl(net, cmd, argp); |
1da177e4 | 1054 | } else |
3d23e349 | 1055 | #ifdef CONFIG_WEXT_CORE |
1da177e4 | 1056 | if (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST) { |
881d966b | 1057 | err = dev_ioctl(net, cmd, argp); |
1da177e4 | 1058 | } else |
3d23e349 | 1059 | #endif |
89bddce5 | 1060 | switch (cmd) { |
1da177e4 LT |
1061 | case FIOSETOWN: |
1062 | case SIOCSPGRP: | |
1063 | err = -EFAULT; | |
1064 | if (get_user(pid, (int __user *)argp)) | |
1065 | break; | |
e0b93edd JL |
1066 | f_setown(sock->file, pid, 1); |
1067 | err = 0; | |
1da177e4 LT |
1068 | break; |
1069 | case FIOGETOWN: | |
1070 | case SIOCGPGRP: | |
609d7fa9 | 1071 | err = put_user(f_getown(sock->file), |
89bddce5 | 1072 | (int __user *)argp); |
1da177e4 LT |
1073 | break; |
1074 | case SIOCGIFBR: | |
1075 | case SIOCSIFBR: | |
1076 | case SIOCBRADDBR: | |
1077 | case SIOCBRDELBR: | |
1078 | err = -ENOPKG; | |
1079 | if (!br_ioctl_hook) | |
1080 | request_module("bridge"); | |
1081 | ||
4a3e2f71 | 1082 | mutex_lock(&br_ioctl_mutex); |
89bddce5 | 1083 | if (br_ioctl_hook) |
881d966b | 1084 | err = br_ioctl_hook(net, cmd, argp); |
4a3e2f71 | 1085 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
1086 | break; |
1087 | case SIOCGIFVLAN: | |
1088 | case SIOCSIFVLAN: | |
1089 | err = -ENOPKG; | |
1090 | if (!vlan_ioctl_hook) | |
1091 | request_module("8021q"); | |
1092 | ||
4a3e2f71 | 1093 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 1094 | if (vlan_ioctl_hook) |
881d966b | 1095 | err = vlan_ioctl_hook(net, argp); |
4a3e2f71 | 1096 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 | 1097 | break; |
1da177e4 LT |
1098 | case SIOCADDDLCI: |
1099 | case SIOCDELDLCI: | |
1100 | err = -ENOPKG; | |
1101 | if (!dlci_ioctl_hook) | |
1102 | request_module("dlci"); | |
1103 | ||
7512cbf6 PE |
1104 | mutex_lock(&dlci_ioctl_mutex); |
1105 | if (dlci_ioctl_hook) | |
1da177e4 | 1106 | err = dlci_ioctl_hook(cmd, argp); |
7512cbf6 | 1107 | mutex_unlock(&dlci_ioctl_mutex); |
1da177e4 LT |
1108 | break; |
1109 | default: | |
6b96018b | 1110 | err = sock_do_ioctl(net, sock, cmd, arg); |
1da177e4 | 1111 | break; |
89bddce5 | 1112 | } |
1da177e4 LT |
1113 | return err; |
1114 | } | |
1115 | ||
1116 | int sock_create_lite(int family, int type, int protocol, struct socket **res) | |
1117 | { | |
1118 | int err; | |
1119 | struct socket *sock = NULL; | |
89bddce5 | 1120 | |
1da177e4 LT |
1121 | err = security_socket_create(family, type, protocol, 1); |
1122 | if (err) | |
1123 | goto out; | |
1124 | ||
1125 | sock = sock_alloc(); | |
1126 | if (!sock) { | |
1127 | err = -ENOMEM; | |
1128 | goto out; | |
1129 | } | |
1130 | ||
1da177e4 | 1131 | sock->type = type; |
7420ed23 VY |
1132 | err = security_socket_post_create(sock, family, type, protocol, 1); |
1133 | if (err) | |
1134 | goto out_release; | |
1135 | ||
1da177e4 LT |
1136 | out: |
1137 | *res = sock; | |
1138 | return err; | |
7420ed23 VY |
1139 | out_release: |
1140 | sock_release(sock); | |
1141 | sock = NULL; | |
1142 | goto out; | |
1da177e4 | 1143 | } |
c6d409cf | 1144 | EXPORT_SYMBOL(sock_create_lite); |
1da177e4 LT |
1145 | |
1146 | /* No kernel lock held - perfect */ | |
89bddce5 | 1147 | static unsigned int sock_poll(struct file *file, poll_table *wait) |
1da177e4 | 1148 | { |
cbf55001 | 1149 | unsigned int busy_flag = 0; |
1da177e4 LT |
1150 | struct socket *sock; |
1151 | ||
1152 | /* | |
89bddce5 | 1153 | * We can't return errors to poll, so it's either yes or no. |
1da177e4 | 1154 | */ |
b69aee04 | 1155 | sock = file->private_data; |
2d48d67f | 1156 | |
cbf55001 | 1157 | if (sk_can_busy_loop(sock->sk)) { |
2d48d67f | 1158 | /* this socket can poll_ll so tell the system call */ |
cbf55001 | 1159 | busy_flag = POLL_BUSY_LOOP; |
2d48d67f ET |
1160 | |
1161 | /* once, only if requested by syscall */ | |
cbf55001 ET |
1162 | if (wait && (wait->_key & POLL_BUSY_LOOP)) |
1163 | sk_busy_loop(sock->sk, 1); | |
2d48d67f ET |
1164 | } |
1165 | ||
cbf55001 | 1166 | return busy_flag | sock->ops->poll(file, sock, wait); |
1da177e4 LT |
1167 | } |
1168 | ||
89bddce5 | 1169 | static int sock_mmap(struct file *file, struct vm_area_struct *vma) |
1da177e4 | 1170 | { |
b69aee04 | 1171 | struct socket *sock = file->private_data; |
1da177e4 LT |
1172 | |
1173 | return sock->ops->mmap(file, sock, vma); | |
1174 | } | |
1175 | ||
20380731 | 1176 | static int sock_close(struct inode *inode, struct file *filp) |
1da177e4 | 1177 | { |
1da177e4 LT |
1178 | sock_release(SOCKET_I(inode)); |
1179 | return 0; | |
1180 | } | |
1181 | ||
1182 | /* | |
1183 | * Update the socket async list | |
1184 | * | |
1185 | * Fasync_list locking strategy. | |
1186 | * | |
1187 | * 1. fasync_list is modified only under process context socket lock | |
1188 | * i.e. under semaphore. | |
1189 | * 2. fasync_list is used under read_lock(&sk->sk_callback_lock) | |
989a2979 | 1190 | * or under socket lock |
1da177e4 LT |
1191 | */ |
1192 | ||
1193 | static int sock_fasync(int fd, struct file *filp, int on) | |
1194 | { | |
989a2979 ED |
1195 | struct socket *sock = filp->private_data; |
1196 | struct sock *sk = sock->sk; | |
eaefd110 | 1197 | struct socket_wq *wq; |
1da177e4 | 1198 | |
989a2979 | 1199 | if (sk == NULL) |
1da177e4 | 1200 | return -EINVAL; |
1da177e4 LT |
1201 | |
1202 | lock_sock(sk); | |
eaefd110 ED |
1203 | wq = rcu_dereference_protected(sock->wq, sock_owned_by_user(sk)); |
1204 | fasync_helper(fd, filp, on, &wq->fasync_list); | |
1da177e4 | 1205 | |
eaefd110 | 1206 | if (!wq->fasync_list) |
989a2979 ED |
1207 | sock_reset_flag(sk, SOCK_FASYNC); |
1208 | else | |
bcdce719 | 1209 | sock_set_flag(sk, SOCK_FASYNC); |
1da177e4 | 1210 | |
989a2979 | 1211 | release_sock(sk); |
1da177e4 LT |
1212 | return 0; |
1213 | } | |
1214 | ||
43815482 | 1215 | /* This function may be called only under socket lock or callback_lock or rcu_lock */ |
1da177e4 LT |
1216 | |
1217 | int sock_wake_async(struct socket *sock, int how, int band) | |
1218 | { | |
43815482 ED |
1219 | struct socket_wq *wq; |
1220 | ||
1221 | if (!sock) | |
1222 | return -1; | |
1223 | rcu_read_lock(); | |
1224 | wq = rcu_dereference(sock->wq); | |
1225 | if (!wq || !wq->fasync_list) { | |
1226 | rcu_read_unlock(); | |
1da177e4 | 1227 | return -1; |
43815482 | 1228 | } |
89bddce5 | 1229 | switch (how) { |
8d8ad9d7 | 1230 | case SOCK_WAKE_WAITD: |
1da177e4 LT |
1231 | if (test_bit(SOCK_ASYNC_WAITDATA, &sock->flags)) |
1232 | break; | |
1233 | goto call_kill; | |
8d8ad9d7 | 1234 | case SOCK_WAKE_SPACE: |
1da177e4 LT |
1235 | if (!test_and_clear_bit(SOCK_ASYNC_NOSPACE, &sock->flags)) |
1236 | break; | |
1237 | /* fall through */ | |
8d8ad9d7 | 1238 | case SOCK_WAKE_IO: |
89bddce5 | 1239 | call_kill: |
43815482 | 1240 | kill_fasync(&wq->fasync_list, SIGIO, band); |
1da177e4 | 1241 | break; |
8d8ad9d7 | 1242 | case SOCK_WAKE_URG: |
43815482 | 1243 | kill_fasync(&wq->fasync_list, SIGURG, band); |
1da177e4 | 1244 | } |
43815482 | 1245 | rcu_read_unlock(); |
1da177e4 LT |
1246 | return 0; |
1247 | } | |
c6d409cf | 1248 | EXPORT_SYMBOL(sock_wake_async); |
1da177e4 | 1249 | |
721db93a | 1250 | int __sock_create(struct net *net, int family, int type, int protocol, |
89bddce5 | 1251 | struct socket **res, int kern) |
1da177e4 LT |
1252 | { |
1253 | int err; | |
1254 | struct socket *sock; | |
55737fda | 1255 | const struct net_proto_family *pf; |
1da177e4 LT |
1256 | |
1257 | /* | |
89bddce5 | 1258 | * Check protocol is in range |
1da177e4 LT |
1259 | */ |
1260 | if (family < 0 || family >= NPROTO) | |
1261 | return -EAFNOSUPPORT; | |
1262 | if (type < 0 || type >= SOCK_MAX) | |
1263 | return -EINVAL; | |
1264 | ||
1265 | /* Compatibility. | |
1266 | ||
1267 | This uglymoron is moved from INET layer to here to avoid | |
1268 | deadlock in module load. | |
1269 | */ | |
1270 | if (family == PF_INET && type == SOCK_PACKET) { | |
89bddce5 | 1271 | static int warned; |
1da177e4 LT |
1272 | if (!warned) { |
1273 | warned = 1; | |
3410f22e YY |
1274 | pr_info("%s uses obsolete (PF_INET,SOCK_PACKET)\n", |
1275 | current->comm); | |
1da177e4 LT |
1276 | } |
1277 | family = PF_PACKET; | |
1278 | } | |
1279 | ||
1280 | err = security_socket_create(family, type, protocol, kern); | |
1281 | if (err) | |
1282 | return err; | |
89bddce5 | 1283 | |
55737fda SH |
1284 | /* |
1285 | * Allocate the socket and allow the family to set things up. if | |
1286 | * the protocol is 0, the family is instructed to select an appropriate | |
1287 | * default. | |
1288 | */ | |
1289 | sock = sock_alloc(); | |
1290 | if (!sock) { | |
e87cc472 | 1291 | net_warn_ratelimited("socket: no more sockets\n"); |
55737fda SH |
1292 | return -ENFILE; /* Not exactly a match, but its the |
1293 | closest posix thing */ | |
1294 | } | |
1295 | ||
1296 | sock->type = type; | |
1297 | ||
95a5afca | 1298 | #ifdef CONFIG_MODULES |
89bddce5 SH |
1299 | /* Attempt to load a protocol module if the find failed. |
1300 | * | |
1301 | * 12/09/1996 Marcin: But! this makes REALLY only sense, if the user | |
1da177e4 LT |
1302 | * requested real, full-featured networking support upon configuration. |
1303 | * Otherwise module support will break! | |
1304 | */ | |
190683a9 | 1305 | if (rcu_access_pointer(net_families[family]) == NULL) |
89bddce5 | 1306 | request_module("net-pf-%d", family); |
1da177e4 LT |
1307 | #endif |
1308 | ||
55737fda SH |
1309 | rcu_read_lock(); |
1310 | pf = rcu_dereference(net_families[family]); | |
1311 | err = -EAFNOSUPPORT; | |
1312 | if (!pf) | |
1313 | goto out_release; | |
1da177e4 LT |
1314 | |
1315 | /* | |
1316 | * We will call the ->create function, that possibly is in a loadable | |
1317 | * module, so we have to bump that loadable module refcnt first. | |
1318 | */ | |
55737fda | 1319 | if (!try_module_get(pf->owner)) |
1da177e4 LT |
1320 | goto out_release; |
1321 | ||
55737fda SH |
1322 | /* Now protected by module ref count */ |
1323 | rcu_read_unlock(); | |
1324 | ||
3f378b68 | 1325 | err = pf->create(net, sock, protocol, kern); |
55737fda | 1326 | if (err < 0) |
1da177e4 | 1327 | goto out_module_put; |
a79af59e | 1328 | |
1da177e4 LT |
1329 | /* |
1330 | * Now to bump the refcnt of the [loadable] module that owns this | |
1331 | * socket at sock_release time we decrement its refcnt. | |
1332 | */ | |
55737fda SH |
1333 | if (!try_module_get(sock->ops->owner)) |
1334 | goto out_module_busy; | |
1335 | ||
1da177e4 LT |
1336 | /* |
1337 | * Now that we're done with the ->create function, the [loadable] | |
1338 | * module can have its refcnt decremented | |
1339 | */ | |
55737fda | 1340 | module_put(pf->owner); |
7420ed23 VY |
1341 | err = security_socket_post_create(sock, family, type, protocol, kern); |
1342 | if (err) | |
3b185525 | 1343 | goto out_sock_release; |
55737fda | 1344 | *res = sock; |
1da177e4 | 1345 | |
55737fda SH |
1346 | return 0; |
1347 | ||
1348 | out_module_busy: | |
1349 | err = -EAFNOSUPPORT; | |
1da177e4 | 1350 | out_module_put: |
55737fda SH |
1351 | sock->ops = NULL; |
1352 | module_put(pf->owner); | |
1353 | out_sock_release: | |
1da177e4 | 1354 | sock_release(sock); |
55737fda SH |
1355 | return err; |
1356 | ||
1357 | out_release: | |
1358 | rcu_read_unlock(); | |
1359 | goto out_sock_release; | |
1da177e4 | 1360 | } |
721db93a | 1361 | EXPORT_SYMBOL(__sock_create); |
1da177e4 LT |
1362 | |
1363 | int sock_create(int family, int type, int protocol, struct socket **res) | |
1364 | { | |
1b8d7ae4 | 1365 | return __sock_create(current->nsproxy->net_ns, family, type, protocol, res, 0); |
1da177e4 | 1366 | } |
c6d409cf | 1367 | EXPORT_SYMBOL(sock_create); |
1da177e4 LT |
1368 | |
1369 | int sock_create_kern(int family, int type, int protocol, struct socket **res) | |
1370 | { | |
1b8d7ae4 | 1371 | return __sock_create(&init_net, family, type, protocol, res, 1); |
1da177e4 | 1372 | } |
c6d409cf | 1373 | EXPORT_SYMBOL(sock_create_kern); |
1da177e4 | 1374 | |
3e0fa65f | 1375 | SYSCALL_DEFINE3(socket, int, family, int, type, int, protocol) |
1da177e4 LT |
1376 | { |
1377 | int retval; | |
1378 | struct socket *sock; | |
a677a039 UD |
1379 | int flags; |
1380 | ||
e38b36f3 UD |
1381 | /* Check the SOCK_* constants for consistency. */ |
1382 | BUILD_BUG_ON(SOCK_CLOEXEC != O_CLOEXEC); | |
1383 | BUILD_BUG_ON((SOCK_MAX | SOCK_TYPE_MASK) != SOCK_TYPE_MASK); | |
1384 | BUILD_BUG_ON(SOCK_CLOEXEC & SOCK_TYPE_MASK); | |
1385 | BUILD_BUG_ON(SOCK_NONBLOCK & SOCK_TYPE_MASK); | |
1386 | ||
a677a039 | 1387 | flags = type & ~SOCK_TYPE_MASK; |
77d27200 | 1388 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1389 | return -EINVAL; |
1390 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1391 | |
aaca0bdc UD |
1392 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1393 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1394 | ||
1da177e4 LT |
1395 | retval = sock_create(family, type, protocol, &sock); |
1396 | if (retval < 0) | |
1397 | goto out; | |
1398 | ||
77d27200 | 1399 | retval = sock_map_fd(sock, flags & (O_CLOEXEC | O_NONBLOCK)); |
1da177e4 LT |
1400 | if (retval < 0) |
1401 | goto out_release; | |
1402 | ||
1403 | out: | |
1404 | /* It may be already another descriptor 8) Not kernel problem. */ | |
1405 | return retval; | |
1406 | ||
1407 | out_release: | |
1408 | sock_release(sock); | |
1409 | return retval; | |
1410 | } | |
1411 | ||
1412 | /* | |
1413 | * Create a pair of connected sockets. | |
1414 | */ | |
1415 | ||
3e0fa65f HC |
1416 | SYSCALL_DEFINE4(socketpair, int, family, int, type, int, protocol, |
1417 | int __user *, usockvec) | |
1da177e4 LT |
1418 | { |
1419 | struct socket *sock1, *sock2; | |
1420 | int fd1, fd2, err; | |
db349509 | 1421 | struct file *newfile1, *newfile2; |
a677a039 UD |
1422 | int flags; |
1423 | ||
1424 | flags = type & ~SOCK_TYPE_MASK; | |
77d27200 | 1425 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1426 | return -EINVAL; |
1427 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1428 | |
aaca0bdc UD |
1429 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1430 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1431 | ||
1da177e4 LT |
1432 | /* |
1433 | * Obtain the first socket and check if the underlying protocol | |
1434 | * supports the socketpair call. | |
1435 | */ | |
1436 | ||
1437 | err = sock_create(family, type, protocol, &sock1); | |
1438 | if (err < 0) | |
1439 | goto out; | |
1440 | ||
1441 | err = sock_create(family, type, protocol, &sock2); | |
1442 | if (err < 0) | |
1443 | goto out_release_1; | |
1444 | ||
1445 | err = sock1->ops->socketpair(sock1, sock2); | |
89bddce5 | 1446 | if (err < 0) |
1da177e4 LT |
1447 | goto out_release_both; |
1448 | ||
28407630 | 1449 | fd1 = get_unused_fd_flags(flags); |
bf3c23d1 DM |
1450 | if (unlikely(fd1 < 0)) { |
1451 | err = fd1; | |
db349509 | 1452 | goto out_release_both; |
bf3c23d1 | 1453 | } |
d73aa286 | 1454 | |
28407630 | 1455 | fd2 = get_unused_fd_flags(flags); |
198de4d7 AV |
1456 | if (unlikely(fd2 < 0)) { |
1457 | err = fd2; | |
d73aa286 | 1458 | goto out_put_unused_1; |
28407630 AV |
1459 | } |
1460 | ||
aab174f0 | 1461 | newfile1 = sock_alloc_file(sock1, flags, NULL); |
28407630 AV |
1462 | if (unlikely(IS_ERR(newfile1))) { |
1463 | err = PTR_ERR(newfile1); | |
d73aa286 | 1464 | goto out_put_unused_both; |
28407630 AV |
1465 | } |
1466 | ||
aab174f0 | 1467 | newfile2 = sock_alloc_file(sock2, flags, NULL); |
28407630 AV |
1468 | if (IS_ERR(newfile2)) { |
1469 | err = PTR_ERR(newfile2); | |
d73aa286 | 1470 | goto out_fput_1; |
db349509 AV |
1471 | } |
1472 | ||
d73aa286 YD |
1473 | err = put_user(fd1, &usockvec[0]); |
1474 | if (err) | |
1475 | goto out_fput_both; | |
1476 | ||
1477 | err = put_user(fd2, &usockvec[1]); | |
1478 | if (err) | |
1479 | goto out_fput_both; | |
1480 | ||
157cf649 | 1481 | audit_fd_pair(fd1, fd2); |
d73aa286 | 1482 | |
db349509 AV |
1483 | fd_install(fd1, newfile1); |
1484 | fd_install(fd2, newfile2); | |
1da177e4 LT |
1485 | /* fd1 and fd2 may be already another descriptors. |
1486 | * Not kernel problem. | |
1487 | */ | |
1488 | ||
d73aa286 | 1489 | return 0; |
1da177e4 | 1490 | |
d73aa286 YD |
1491 | out_fput_both: |
1492 | fput(newfile2); | |
1493 | fput(newfile1); | |
1494 | put_unused_fd(fd2); | |
1495 | put_unused_fd(fd1); | |
1496 | goto out; | |
1497 | ||
1498 | out_fput_1: | |
1499 | fput(newfile1); | |
1500 | put_unused_fd(fd2); | |
1501 | put_unused_fd(fd1); | |
1502 | sock_release(sock2); | |
1503 | goto out; | |
1da177e4 | 1504 | |
d73aa286 YD |
1505 | out_put_unused_both: |
1506 | put_unused_fd(fd2); | |
1507 | out_put_unused_1: | |
1508 | put_unused_fd(fd1); | |
1da177e4 | 1509 | out_release_both: |
89bddce5 | 1510 | sock_release(sock2); |
1da177e4 | 1511 | out_release_1: |
89bddce5 | 1512 | sock_release(sock1); |
1da177e4 LT |
1513 | out: |
1514 | return err; | |
1515 | } | |
1516 | ||
1da177e4 LT |
1517 | /* |
1518 | * Bind a name to a socket. Nothing much to do here since it's | |
1519 | * the protocol's responsibility to handle the local address. | |
1520 | * | |
1521 | * We move the socket address to kernel space before we call | |
1522 | * the protocol layer (having also checked the address is ok). | |
1523 | */ | |
1524 | ||
20f37034 | 1525 | SYSCALL_DEFINE3(bind, int, fd, struct sockaddr __user *, umyaddr, int, addrlen) |
1da177e4 LT |
1526 | { |
1527 | struct socket *sock; | |
230b1839 | 1528 | struct sockaddr_storage address; |
6cb153ca | 1529 | int err, fput_needed; |
1da177e4 | 1530 | |
89bddce5 | 1531 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
e71a4783 | 1532 | if (sock) { |
43db362d | 1533 | err = move_addr_to_kernel(umyaddr, addrlen, &address); |
89bddce5 SH |
1534 | if (err >= 0) { |
1535 | err = security_socket_bind(sock, | |
230b1839 | 1536 | (struct sockaddr *)&address, |
89bddce5 | 1537 | addrlen); |
6cb153ca BL |
1538 | if (!err) |
1539 | err = sock->ops->bind(sock, | |
89bddce5 | 1540 | (struct sockaddr *) |
230b1839 | 1541 | &address, addrlen); |
1da177e4 | 1542 | } |
6cb153ca | 1543 | fput_light(sock->file, fput_needed); |
89bddce5 | 1544 | } |
1da177e4 LT |
1545 | return err; |
1546 | } | |
1547 | ||
1da177e4 LT |
1548 | /* |
1549 | * Perform a listen. Basically, we allow the protocol to do anything | |
1550 | * necessary for a listen, and if that works, we mark the socket as | |
1551 | * ready for listening. | |
1552 | */ | |
1553 | ||
3e0fa65f | 1554 | SYSCALL_DEFINE2(listen, int, fd, int, backlog) |
1da177e4 LT |
1555 | { |
1556 | struct socket *sock; | |
6cb153ca | 1557 | int err, fput_needed; |
b8e1f9b5 | 1558 | int somaxconn; |
89bddce5 SH |
1559 | |
1560 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
1561 | if (sock) { | |
8efa6e93 | 1562 | somaxconn = sock_net(sock->sk)->core.sysctl_somaxconn; |
95c96174 | 1563 | if ((unsigned int)backlog > somaxconn) |
b8e1f9b5 | 1564 | backlog = somaxconn; |
1da177e4 LT |
1565 | |
1566 | err = security_socket_listen(sock, backlog); | |
6cb153ca BL |
1567 | if (!err) |
1568 | err = sock->ops->listen(sock, backlog); | |
1da177e4 | 1569 | |
6cb153ca | 1570 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1571 | } |
1572 | return err; | |
1573 | } | |
1574 | ||
1da177e4 LT |
1575 | /* |
1576 | * For accept, we attempt to create a new socket, set up the link | |
1577 | * with the client, wake up the client, then return the new | |
1578 | * connected fd. We collect the address of the connector in kernel | |
1579 | * space and move it to user at the very end. This is unclean because | |
1580 | * we open the socket then return an error. | |
1581 | * | |
1582 | * 1003.1g adds the ability to recvmsg() to query connection pending | |
1583 | * status to recvmsg. We need to add that support in a way thats | |
1584 | * clean when we restucture accept also. | |
1585 | */ | |
1586 | ||
20f37034 HC |
1587 | SYSCALL_DEFINE4(accept4, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1588 | int __user *, upeer_addrlen, int, flags) | |
1da177e4 LT |
1589 | { |
1590 | struct socket *sock, *newsock; | |
39d8c1b6 | 1591 | struct file *newfile; |
6cb153ca | 1592 | int err, len, newfd, fput_needed; |
230b1839 | 1593 | struct sockaddr_storage address; |
1da177e4 | 1594 | |
77d27200 | 1595 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
aaca0bdc UD |
1596 | return -EINVAL; |
1597 | ||
1598 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) | |
1599 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1600 | ||
6cb153ca | 1601 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1602 | if (!sock) |
1603 | goto out; | |
1604 | ||
1605 | err = -ENFILE; | |
c6d409cf ED |
1606 | newsock = sock_alloc(); |
1607 | if (!newsock) | |
1da177e4 LT |
1608 | goto out_put; |
1609 | ||
1610 | newsock->type = sock->type; | |
1611 | newsock->ops = sock->ops; | |
1612 | ||
1da177e4 LT |
1613 | /* |
1614 | * We don't need try_module_get here, as the listening socket (sock) | |
1615 | * has the protocol module (sock->ops->owner) held. | |
1616 | */ | |
1617 | __module_get(newsock->ops->owner); | |
1618 | ||
28407630 | 1619 | newfd = get_unused_fd_flags(flags); |
39d8c1b6 DM |
1620 | if (unlikely(newfd < 0)) { |
1621 | err = newfd; | |
9a1875e6 DM |
1622 | sock_release(newsock); |
1623 | goto out_put; | |
39d8c1b6 | 1624 | } |
aab174f0 | 1625 | newfile = sock_alloc_file(newsock, flags, sock->sk->sk_prot_creator->name); |
28407630 AV |
1626 | if (unlikely(IS_ERR(newfile))) { |
1627 | err = PTR_ERR(newfile); | |
1628 | put_unused_fd(newfd); | |
1629 | sock_release(newsock); | |
1630 | goto out_put; | |
1631 | } | |
39d8c1b6 | 1632 | |
a79af59e FF |
1633 | err = security_socket_accept(sock, newsock); |
1634 | if (err) | |
39d8c1b6 | 1635 | goto out_fd; |
a79af59e | 1636 | |
1da177e4 LT |
1637 | err = sock->ops->accept(sock, newsock, sock->file->f_flags); |
1638 | if (err < 0) | |
39d8c1b6 | 1639 | goto out_fd; |
1da177e4 LT |
1640 | |
1641 | if (upeer_sockaddr) { | |
230b1839 | 1642 | if (newsock->ops->getname(newsock, (struct sockaddr *)&address, |
89bddce5 | 1643 | &len, 2) < 0) { |
1da177e4 | 1644 | err = -ECONNABORTED; |
39d8c1b6 | 1645 | goto out_fd; |
1da177e4 | 1646 | } |
43db362d | 1647 | err = move_addr_to_user(&address, |
230b1839 | 1648 | len, upeer_sockaddr, upeer_addrlen); |
1da177e4 | 1649 | if (err < 0) |
39d8c1b6 | 1650 | goto out_fd; |
1da177e4 LT |
1651 | } |
1652 | ||
1653 | /* File flags are not inherited via accept() unlike another OSes. */ | |
1654 | ||
39d8c1b6 DM |
1655 | fd_install(newfd, newfile); |
1656 | err = newfd; | |
1da177e4 | 1657 | |
1da177e4 | 1658 | out_put: |
6cb153ca | 1659 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1660 | out: |
1661 | return err; | |
39d8c1b6 | 1662 | out_fd: |
9606a216 | 1663 | fput(newfile); |
39d8c1b6 | 1664 | put_unused_fd(newfd); |
1da177e4 LT |
1665 | goto out_put; |
1666 | } | |
1667 | ||
20f37034 HC |
1668 | SYSCALL_DEFINE3(accept, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1669 | int __user *, upeer_addrlen) | |
aaca0bdc | 1670 | { |
de11defe | 1671 | return sys_accept4(fd, upeer_sockaddr, upeer_addrlen, 0); |
aaca0bdc UD |
1672 | } |
1673 | ||
1da177e4 LT |
1674 | /* |
1675 | * Attempt to connect to a socket with the server address. The address | |
1676 | * is in user space so we verify it is OK and move it to kernel space. | |
1677 | * | |
1678 | * For 1003.1g we need to add clean support for a bind to AF_UNSPEC to | |
1679 | * break bindings | |
1680 | * | |
1681 | * NOTE: 1003.1g draft 6.3 is broken with respect to AX.25/NetROM and | |
1682 | * other SEQPACKET protocols that take time to connect() as it doesn't | |
1683 | * include the -EINPROGRESS status for such sockets. | |
1684 | */ | |
1685 | ||
20f37034 HC |
1686 | SYSCALL_DEFINE3(connect, int, fd, struct sockaddr __user *, uservaddr, |
1687 | int, addrlen) | |
1da177e4 LT |
1688 | { |
1689 | struct socket *sock; | |
230b1839 | 1690 | struct sockaddr_storage address; |
6cb153ca | 1691 | int err, fput_needed; |
1da177e4 | 1692 | |
6cb153ca | 1693 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1694 | if (!sock) |
1695 | goto out; | |
43db362d | 1696 | err = move_addr_to_kernel(uservaddr, addrlen, &address); |
1da177e4 LT |
1697 | if (err < 0) |
1698 | goto out_put; | |
1699 | ||
89bddce5 | 1700 | err = |
230b1839 | 1701 | security_socket_connect(sock, (struct sockaddr *)&address, addrlen); |
1da177e4 LT |
1702 | if (err) |
1703 | goto out_put; | |
1704 | ||
230b1839 | 1705 | err = sock->ops->connect(sock, (struct sockaddr *)&address, addrlen, |
1da177e4 LT |
1706 | sock->file->f_flags); |
1707 | out_put: | |
6cb153ca | 1708 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1709 | out: |
1710 | return err; | |
1711 | } | |
1712 | ||
1713 | /* | |
1714 | * Get the local address ('name') of a socket object. Move the obtained | |
1715 | * name to user space. | |
1716 | */ | |
1717 | ||
20f37034 HC |
1718 | SYSCALL_DEFINE3(getsockname, int, fd, struct sockaddr __user *, usockaddr, |
1719 | int __user *, usockaddr_len) | |
1da177e4 LT |
1720 | { |
1721 | struct socket *sock; | |
230b1839 | 1722 | struct sockaddr_storage address; |
6cb153ca | 1723 | int len, err, fput_needed; |
89bddce5 | 1724 | |
6cb153ca | 1725 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1726 | if (!sock) |
1727 | goto out; | |
1728 | ||
1729 | err = security_socket_getsockname(sock); | |
1730 | if (err) | |
1731 | goto out_put; | |
1732 | ||
230b1839 | 1733 | err = sock->ops->getname(sock, (struct sockaddr *)&address, &len, 0); |
1da177e4 LT |
1734 | if (err) |
1735 | goto out_put; | |
43db362d | 1736 | err = move_addr_to_user(&address, len, usockaddr, usockaddr_len); |
1da177e4 LT |
1737 | |
1738 | out_put: | |
6cb153ca | 1739 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1740 | out: |
1741 | return err; | |
1742 | } | |
1743 | ||
1744 | /* | |
1745 | * Get the remote address ('name') of a socket object. Move the obtained | |
1746 | * name to user space. | |
1747 | */ | |
1748 | ||
20f37034 HC |
1749 | SYSCALL_DEFINE3(getpeername, int, fd, struct sockaddr __user *, usockaddr, |
1750 | int __user *, usockaddr_len) | |
1da177e4 LT |
1751 | { |
1752 | struct socket *sock; | |
230b1839 | 1753 | struct sockaddr_storage address; |
6cb153ca | 1754 | int len, err, fput_needed; |
1da177e4 | 1755 | |
89bddce5 SH |
1756 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1757 | if (sock != NULL) { | |
1da177e4 LT |
1758 | err = security_socket_getpeername(sock); |
1759 | if (err) { | |
6cb153ca | 1760 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1761 | return err; |
1762 | } | |
1763 | ||
89bddce5 | 1764 | err = |
230b1839 | 1765 | sock->ops->getname(sock, (struct sockaddr *)&address, &len, |
89bddce5 | 1766 | 1); |
1da177e4 | 1767 | if (!err) |
43db362d | 1768 | err = move_addr_to_user(&address, len, usockaddr, |
89bddce5 | 1769 | usockaddr_len); |
6cb153ca | 1770 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1771 | } |
1772 | return err; | |
1773 | } | |
1774 | ||
1775 | /* | |
1776 | * Send a datagram to a given address. We move the address into kernel | |
1777 | * space and check the user space data area is readable before invoking | |
1778 | * the protocol. | |
1779 | */ | |
1780 | ||
3e0fa65f | 1781 | SYSCALL_DEFINE6(sendto, int, fd, void __user *, buff, size_t, len, |
95c96174 | 1782 | unsigned int, flags, struct sockaddr __user *, addr, |
3e0fa65f | 1783 | int, addr_len) |
1da177e4 LT |
1784 | { |
1785 | struct socket *sock; | |
230b1839 | 1786 | struct sockaddr_storage address; |
1da177e4 LT |
1787 | int err; |
1788 | struct msghdr msg; | |
1789 | struct iovec iov; | |
6cb153ca | 1790 | int fput_needed; |
6cb153ca | 1791 | |
253eacc0 LT |
1792 | if (len > INT_MAX) |
1793 | len = INT_MAX; | |
de0fa95c PE |
1794 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1795 | if (!sock) | |
4387ff75 | 1796 | goto out; |
6cb153ca | 1797 | |
89bddce5 SH |
1798 | iov.iov_base = buff; |
1799 | iov.iov_len = len; | |
1800 | msg.msg_name = NULL; | |
1801 | msg.msg_iov = &iov; | |
1802 | msg.msg_iovlen = 1; | |
1803 | msg.msg_control = NULL; | |
1804 | msg.msg_controllen = 0; | |
1805 | msg.msg_namelen = 0; | |
6cb153ca | 1806 | if (addr) { |
43db362d | 1807 | err = move_addr_to_kernel(addr, addr_len, &address); |
1da177e4 LT |
1808 | if (err < 0) |
1809 | goto out_put; | |
230b1839 | 1810 | msg.msg_name = (struct sockaddr *)&address; |
89bddce5 | 1811 | msg.msg_namelen = addr_len; |
1da177e4 LT |
1812 | } |
1813 | if (sock->file->f_flags & O_NONBLOCK) | |
1814 | flags |= MSG_DONTWAIT; | |
1815 | msg.msg_flags = flags; | |
1816 | err = sock_sendmsg(sock, &msg, len); | |
1817 | ||
89bddce5 | 1818 | out_put: |
de0fa95c | 1819 | fput_light(sock->file, fput_needed); |
4387ff75 | 1820 | out: |
1da177e4 LT |
1821 | return err; |
1822 | } | |
1823 | ||
1824 | /* | |
89bddce5 | 1825 | * Send a datagram down a socket. |
1da177e4 LT |
1826 | */ |
1827 | ||
3e0fa65f | 1828 | SYSCALL_DEFINE4(send, int, fd, void __user *, buff, size_t, len, |
95c96174 | 1829 | unsigned int, flags) |
1da177e4 LT |
1830 | { |
1831 | return sys_sendto(fd, buff, len, flags, NULL, 0); | |
1832 | } | |
1833 | ||
1834 | /* | |
89bddce5 | 1835 | * Receive a frame from the socket and optionally record the address of the |
1da177e4 LT |
1836 | * sender. We verify the buffers are writable and if needed move the |
1837 | * sender address from kernel to user space. | |
1838 | */ | |
1839 | ||
3e0fa65f | 1840 | SYSCALL_DEFINE6(recvfrom, int, fd, void __user *, ubuf, size_t, size, |
95c96174 | 1841 | unsigned int, flags, struct sockaddr __user *, addr, |
3e0fa65f | 1842 | int __user *, addr_len) |
1da177e4 LT |
1843 | { |
1844 | struct socket *sock; | |
1845 | struct iovec iov; | |
1846 | struct msghdr msg; | |
230b1839 | 1847 | struct sockaddr_storage address; |
89bddce5 | 1848 | int err, err2; |
6cb153ca BL |
1849 | int fput_needed; |
1850 | ||
253eacc0 LT |
1851 | if (size > INT_MAX) |
1852 | size = INT_MAX; | |
de0fa95c | 1853 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 | 1854 | if (!sock) |
de0fa95c | 1855 | goto out; |
1da177e4 | 1856 | |
89bddce5 SH |
1857 | msg.msg_control = NULL; |
1858 | msg.msg_controllen = 0; | |
1859 | msg.msg_iovlen = 1; | |
1860 | msg.msg_iov = &iov; | |
1861 | iov.iov_len = size; | |
1862 | iov.iov_base = ubuf; | |
f3d33426 HFS |
1863 | /* Save some cycles and don't copy the address if not needed */ |
1864 | msg.msg_name = addr ? (struct sockaddr *)&address : NULL; | |
1865 | /* We assume all kernel code knows the size of sockaddr_storage */ | |
1866 | msg.msg_namelen = 0; | |
1da177e4 LT |
1867 | if (sock->file->f_flags & O_NONBLOCK) |
1868 | flags |= MSG_DONTWAIT; | |
89bddce5 | 1869 | err = sock_recvmsg(sock, &msg, size, flags); |
1da177e4 | 1870 | |
89bddce5 | 1871 | if (err >= 0 && addr != NULL) { |
43db362d | 1872 | err2 = move_addr_to_user(&address, |
230b1839 | 1873 | msg.msg_namelen, addr, addr_len); |
89bddce5 SH |
1874 | if (err2 < 0) |
1875 | err = err2; | |
1da177e4 | 1876 | } |
de0fa95c PE |
1877 | |
1878 | fput_light(sock->file, fput_needed); | |
4387ff75 | 1879 | out: |
1da177e4 LT |
1880 | return err; |
1881 | } | |
1882 | ||
1883 | /* | |
89bddce5 | 1884 | * Receive a datagram from a socket. |
1da177e4 LT |
1885 | */ |
1886 | ||
b7c0ddf5 JG |
1887 | SYSCALL_DEFINE4(recv, int, fd, void __user *, ubuf, size_t, size, |
1888 | unsigned int, flags) | |
1da177e4 LT |
1889 | { |
1890 | return sys_recvfrom(fd, ubuf, size, flags, NULL, NULL); | |
1891 | } | |
1892 | ||
1893 | /* | |
1894 | * Set a socket option. Because we don't know the option lengths we have | |
1895 | * to pass the user mode parameter for the protocols to sort out. | |
1896 | */ | |
1897 | ||
20f37034 HC |
1898 | SYSCALL_DEFINE5(setsockopt, int, fd, int, level, int, optname, |
1899 | char __user *, optval, int, optlen) | |
1da177e4 | 1900 | { |
6cb153ca | 1901 | int err, fput_needed; |
1da177e4 LT |
1902 | struct socket *sock; |
1903 | ||
1904 | if (optlen < 0) | |
1905 | return -EINVAL; | |
89bddce5 SH |
1906 | |
1907 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
1908 | if (sock != NULL) { | |
1909 | err = security_socket_setsockopt(sock, level, optname); | |
6cb153ca BL |
1910 | if (err) |
1911 | goto out_put; | |
1da177e4 LT |
1912 | |
1913 | if (level == SOL_SOCKET) | |
89bddce5 SH |
1914 | err = |
1915 | sock_setsockopt(sock, level, optname, optval, | |
1916 | optlen); | |
1da177e4 | 1917 | else |
89bddce5 SH |
1918 | err = |
1919 | sock->ops->setsockopt(sock, level, optname, optval, | |
1920 | optlen); | |
6cb153ca BL |
1921 | out_put: |
1922 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
1923 | } |
1924 | return err; | |
1925 | } | |
1926 | ||
1927 | /* | |
1928 | * Get a socket option. Because we don't know the option lengths we have | |
1929 | * to pass a user mode parameter for the protocols to sort out. | |
1930 | */ | |
1931 | ||
20f37034 HC |
1932 | SYSCALL_DEFINE5(getsockopt, int, fd, int, level, int, optname, |
1933 | char __user *, optval, int __user *, optlen) | |
1da177e4 | 1934 | { |
6cb153ca | 1935 | int err, fput_needed; |
1da177e4 LT |
1936 | struct socket *sock; |
1937 | ||
89bddce5 SH |
1938 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1939 | if (sock != NULL) { | |
6cb153ca BL |
1940 | err = security_socket_getsockopt(sock, level, optname); |
1941 | if (err) | |
1942 | goto out_put; | |
1da177e4 LT |
1943 | |
1944 | if (level == SOL_SOCKET) | |
89bddce5 SH |
1945 | err = |
1946 | sock_getsockopt(sock, level, optname, optval, | |
1947 | optlen); | |
1da177e4 | 1948 | else |
89bddce5 SH |
1949 | err = |
1950 | sock->ops->getsockopt(sock, level, optname, optval, | |
1951 | optlen); | |
6cb153ca BL |
1952 | out_put: |
1953 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
1954 | } |
1955 | return err; | |
1956 | } | |
1957 | ||
1da177e4 LT |
1958 | /* |
1959 | * Shutdown a socket. | |
1960 | */ | |
1961 | ||
754fe8d2 | 1962 | SYSCALL_DEFINE2(shutdown, int, fd, int, how) |
1da177e4 | 1963 | { |
6cb153ca | 1964 | int err, fput_needed; |
1da177e4 LT |
1965 | struct socket *sock; |
1966 | ||
89bddce5 SH |
1967 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1968 | if (sock != NULL) { | |
1da177e4 | 1969 | err = security_socket_shutdown(sock, how); |
6cb153ca BL |
1970 | if (!err) |
1971 | err = sock->ops->shutdown(sock, how); | |
1972 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
1973 | } |
1974 | return err; | |
1975 | } | |
1976 | ||
89bddce5 | 1977 | /* A couple of helpful macros for getting the address of the 32/64 bit |
1da177e4 LT |
1978 | * fields which are the same type (int / unsigned) on our platforms. |
1979 | */ | |
1980 | #define COMPAT_MSG(msg, member) ((MSG_CMSG_COMPAT & flags) ? &msg##_compat->member : &msg->member) | |
1981 | #define COMPAT_NAMELEN(msg) COMPAT_MSG(msg, msg_namelen) | |
1982 | #define COMPAT_FLAGS(msg) COMPAT_MSG(msg, msg_flags) | |
1983 | ||
c71d8ebe TH |
1984 | struct used_address { |
1985 | struct sockaddr_storage name; | |
1986 | unsigned int name_len; | |
1987 | }; | |
1988 | ||
08adb7da AV |
1989 | static ssize_t copy_msghdr_from_user(struct msghdr *kmsg, |
1990 | struct user_msghdr __user *umsg, | |
1991 | struct sockaddr __user **save_addr, | |
1992 | struct iovec **iov) | |
1661bf36 | 1993 | { |
08adb7da AV |
1994 | struct sockaddr __user *uaddr; |
1995 | struct iovec __user *uiov; | |
1996 | ssize_t err; | |
1997 | ||
1998 | if (!access_ok(VERIFY_READ, umsg, sizeof(*umsg)) || | |
1999 | __get_user(uaddr, &umsg->msg_name) || | |
2000 | __get_user(kmsg->msg_namelen, &umsg->msg_namelen) || | |
2001 | __get_user(uiov, &umsg->msg_iov) || | |
2002 | __get_user(kmsg->msg_iovlen, &umsg->msg_iovlen) || | |
2003 | __get_user(kmsg->msg_control, &umsg->msg_control) || | |
2004 | __get_user(kmsg->msg_controllen, &umsg->msg_controllen) || | |
2005 | __get_user(kmsg->msg_flags, &umsg->msg_flags)) | |
1661bf36 | 2006 | return -EFAULT; |
dbb490b9 | 2007 | |
08adb7da | 2008 | if (!uaddr) |
6a2a2b3a AS |
2009 | kmsg->msg_namelen = 0; |
2010 | ||
dbb490b9 ML |
2011 | if (kmsg->msg_namelen < 0) |
2012 | return -EINVAL; | |
2013 | ||
1661bf36 | 2014 | if (kmsg->msg_namelen > sizeof(struct sockaddr_storage)) |
db31c55a | 2015 | kmsg->msg_namelen = sizeof(struct sockaddr_storage); |
08adb7da AV |
2016 | |
2017 | if (save_addr) | |
2018 | *save_addr = uaddr; | |
2019 | ||
2020 | if (uaddr && kmsg->msg_namelen) { | |
2021 | if (!save_addr) { | |
2022 | err = move_addr_to_kernel(uaddr, kmsg->msg_namelen, | |
2023 | kmsg->msg_name); | |
2024 | if (err < 0) | |
2025 | return err; | |
2026 | } | |
2027 | } else { | |
2028 | kmsg->msg_name = NULL; | |
2029 | kmsg->msg_namelen = 0; | |
2030 | } | |
2031 | ||
2032 | if (kmsg->msg_iovlen > UIO_MAXIOV) | |
2033 | return -EMSGSIZE; | |
2034 | ||
2035 | err = rw_copy_check_uvector(save_addr ? READ : WRITE, | |
2036 | uiov, kmsg->msg_iovlen, | |
2037 | UIO_FASTIOV, *iov, iov); | |
2038 | if (err >= 0) | |
2039 | kmsg->msg_iov = *iov; | |
2040 | return err; | |
1661bf36 DC |
2041 | } |
2042 | ||
666547ff | 2043 | static int ___sys_sendmsg(struct socket *sock, struct user_msghdr __user *msg, |
95c96174 | 2044 | struct msghdr *msg_sys, unsigned int flags, |
c71d8ebe | 2045 | struct used_address *used_address) |
1da177e4 | 2046 | { |
89bddce5 SH |
2047 | struct compat_msghdr __user *msg_compat = |
2048 | (struct compat_msghdr __user *)msg; | |
230b1839 | 2049 | struct sockaddr_storage address; |
1da177e4 | 2050 | struct iovec iovstack[UIO_FASTIOV], *iov = iovstack; |
b9d717a7 | 2051 | unsigned char ctl[sizeof(struct cmsghdr) + 20] |
89bddce5 SH |
2052 | __attribute__ ((aligned(sizeof(__kernel_size_t)))); |
2053 | /* 20 is size of ipv6_pktinfo */ | |
1da177e4 | 2054 | unsigned char *ctl_buf = ctl; |
08adb7da AV |
2055 | int ctl_len, total_len; |
2056 | ssize_t err; | |
89bddce5 | 2057 | |
08adb7da | 2058 | msg_sys->msg_name = &address; |
1da177e4 | 2059 | |
08449320 | 2060 | if (MSG_CMSG_COMPAT & flags) |
08adb7da | 2061 | err = get_compat_msghdr(msg_sys, msg_compat, NULL, &iov); |
08449320 | 2062 | else |
08adb7da | 2063 | err = copy_msghdr_from_user(msg_sys, msg, NULL, &iov); |
89bddce5 | 2064 | if (err < 0) |
1da177e4 LT |
2065 | goto out_freeiov; |
2066 | total_len = err; | |
2067 | ||
2068 | err = -ENOBUFS; | |
2069 | ||
228e548e | 2070 | if (msg_sys->msg_controllen > INT_MAX) |
1da177e4 | 2071 | goto out_freeiov; |
228e548e | 2072 | ctl_len = msg_sys->msg_controllen; |
1da177e4 | 2073 | if ((MSG_CMSG_COMPAT & flags) && ctl_len) { |
89bddce5 | 2074 | err = |
228e548e | 2075 | cmsghdr_from_user_compat_to_kern(msg_sys, sock->sk, ctl, |
89bddce5 | 2076 | sizeof(ctl)); |
1da177e4 LT |
2077 | if (err) |
2078 | goto out_freeiov; | |
228e548e AB |
2079 | ctl_buf = msg_sys->msg_control; |
2080 | ctl_len = msg_sys->msg_controllen; | |
1da177e4 | 2081 | } else if (ctl_len) { |
89bddce5 | 2082 | if (ctl_len > sizeof(ctl)) { |
1da177e4 | 2083 | ctl_buf = sock_kmalloc(sock->sk, ctl_len, GFP_KERNEL); |
89bddce5 | 2084 | if (ctl_buf == NULL) |
1da177e4 LT |
2085 | goto out_freeiov; |
2086 | } | |
2087 | err = -EFAULT; | |
2088 | /* | |
228e548e | 2089 | * Careful! Before this, msg_sys->msg_control contains a user pointer. |
1da177e4 LT |
2090 | * Afterwards, it will be a kernel pointer. Thus the compiler-assisted |
2091 | * checking falls down on this. | |
2092 | */ | |
fb8621bb | 2093 | if (copy_from_user(ctl_buf, |
228e548e | 2094 | (void __user __force *)msg_sys->msg_control, |
89bddce5 | 2095 | ctl_len)) |
1da177e4 | 2096 | goto out_freectl; |
228e548e | 2097 | msg_sys->msg_control = ctl_buf; |
1da177e4 | 2098 | } |
228e548e | 2099 | msg_sys->msg_flags = flags; |
1da177e4 LT |
2100 | |
2101 | if (sock->file->f_flags & O_NONBLOCK) | |
228e548e | 2102 | msg_sys->msg_flags |= MSG_DONTWAIT; |
c71d8ebe TH |
2103 | /* |
2104 | * If this is sendmmsg() and current destination address is same as | |
2105 | * previously succeeded address, omit asking LSM's decision. | |
2106 | * used_address->name_len is initialized to UINT_MAX so that the first | |
2107 | * destination address never matches. | |
2108 | */ | |
bc909d9d MD |
2109 | if (used_address && msg_sys->msg_name && |
2110 | used_address->name_len == msg_sys->msg_namelen && | |
2111 | !memcmp(&used_address->name, msg_sys->msg_name, | |
c71d8ebe TH |
2112 | used_address->name_len)) { |
2113 | err = sock_sendmsg_nosec(sock, msg_sys, total_len); | |
2114 | goto out_freectl; | |
2115 | } | |
2116 | err = sock_sendmsg(sock, msg_sys, total_len); | |
2117 | /* | |
2118 | * If this is sendmmsg() and sending to current destination address was | |
2119 | * successful, remember it. | |
2120 | */ | |
2121 | if (used_address && err >= 0) { | |
2122 | used_address->name_len = msg_sys->msg_namelen; | |
bc909d9d MD |
2123 | if (msg_sys->msg_name) |
2124 | memcpy(&used_address->name, msg_sys->msg_name, | |
2125 | used_address->name_len); | |
c71d8ebe | 2126 | } |
1da177e4 LT |
2127 | |
2128 | out_freectl: | |
89bddce5 | 2129 | if (ctl_buf != ctl) |
1da177e4 LT |
2130 | sock_kfree_s(sock->sk, ctl_buf, ctl_len); |
2131 | out_freeiov: | |
2132 | if (iov != iovstack) | |
a74e9106 | 2133 | kfree(iov); |
228e548e AB |
2134 | return err; |
2135 | } | |
2136 | ||
2137 | /* | |
2138 | * BSD sendmsg interface | |
2139 | */ | |
2140 | ||
666547ff | 2141 | long __sys_sendmsg(int fd, struct user_msghdr __user *msg, unsigned flags) |
228e548e AB |
2142 | { |
2143 | int fput_needed, err; | |
2144 | struct msghdr msg_sys; | |
1be374a0 AL |
2145 | struct socket *sock; |
2146 | ||
1be374a0 | 2147 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
228e548e AB |
2148 | if (!sock) |
2149 | goto out; | |
2150 | ||
a7526eb5 | 2151 | err = ___sys_sendmsg(sock, msg, &msg_sys, flags, NULL); |
228e548e | 2152 | |
6cb153ca | 2153 | fput_light(sock->file, fput_needed); |
89bddce5 | 2154 | out: |
1da177e4 LT |
2155 | return err; |
2156 | } | |
2157 | ||
666547ff | 2158 | SYSCALL_DEFINE3(sendmsg, int, fd, struct user_msghdr __user *, msg, unsigned int, flags) |
a7526eb5 AL |
2159 | { |
2160 | if (flags & MSG_CMSG_COMPAT) | |
2161 | return -EINVAL; | |
2162 | return __sys_sendmsg(fd, msg, flags); | |
2163 | } | |
2164 | ||
228e548e AB |
2165 | /* |
2166 | * Linux sendmmsg interface | |
2167 | */ | |
2168 | ||
2169 | int __sys_sendmmsg(int fd, struct mmsghdr __user *mmsg, unsigned int vlen, | |
2170 | unsigned int flags) | |
2171 | { | |
2172 | int fput_needed, err, datagrams; | |
2173 | struct socket *sock; | |
2174 | struct mmsghdr __user *entry; | |
2175 | struct compat_mmsghdr __user *compat_entry; | |
2176 | struct msghdr msg_sys; | |
c71d8ebe | 2177 | struct used_address used_address; |
228e548e | 2178 | |
98382f41 AB |
2179 | if (vlen > UIO_MAXIOV) |
2180 | vlen = UIO_MAXIOV; | |
228e548e AB |
2181 | |
2182 | datagrams = 0; | |
2183 | ||
2184 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2185 | if (!sock) | |
2186 | return err; | |
2187 | ||
c71d8ebe | 2188 | used_address.name_len = UINT_MAX; |
228e548e AB |
2189 | entry = mmsg; |
2190 | compat_entry = (struct compat_mmsghdr __user *)mmsg; | |
728ffb86 | 2191 | err = 0; |
228e548e AB |
2192 | |
2193 | while (datagrams < vlen) { | |
228e548e | 2194 | if (MSG_CMSG_COMPAT & flags) { |
666547ff | 2195 | err = ___sys_sendmsg(sock, (struct user_msghdr __user *)compat_entry, |
a7526eb5 | 2196 | &msg_sys, flags, &used_address); |
228e548e AB |
2197 | if (err < 0) |
2198 | break; | |
2199 | err = __put_user(err, &compat_entry->msg_len); | |
2200 | ++compat_entry; | |
2201 | } else { | |
a7526eb5 | 2202 | err = ___sys_sendmsg(sock, |
666547ff | 2203 | (struct user_msghdr __user *)entry, |
a7526eb5 | 2204 | &msg_sys, flags, &used_address); |
228e548e AB |
2205 | if (err < 0) |
2206 | break; | |
2207 | err = put_user(err, &entry->msg_len); | |
2208 | ++entry; | |
2209 | } | |
2210 | ||
2211 | if (err) | |
2212 | break; | |
2213 | ++datagrams; | |
2214 | } | |
2215 | ||
228e548e AB |
2216 | fput_light(sock->file, fput_needed); |
2217 | ||
728ffb86 AB |
2218 | /* We only return an error if no datagrams were able to be sent */ |
2219 | if (datagrams != 0) | |
228e548e AB |
2220 | return datagrams; |
2221 | ||
228e548e AB |
2222 | return err; |
2223 | } | |
2224 | ||
2225 | SYSCALL_DEFINE4(sendmmsg, int, fd, struct mmsghdr __user *, mmsg, | |
2226 | unsigned int, vlen, unsigned int, flags) | |
2227 | { | |
1be374a0 AL |
2228 | if (flags & MSG_CMSG_COMPAT) |
2229 | return -EINVAL; | |
228e548e AB |
2230 | return __sys_sendmmsg(fd, mmsg, vlen, flags); |
2231 | } | |
2232 | ||
666547ff | 2233 | static int ___sys_recvmsg(struct socket *sock, struct user_msghdr __user *msg, |
95c96174 | 2234 | struct msghdr *msg_sys, unsigned int flags, int nosec) |
1da177e4 | 2235 | { |
89bddce5 SH |
2236 | struct compat_msghdr __user *msg_compat = |
2237 | (struct compat_msghdr __user *)msg; | |
1da177e4 | 2238 | struct iovec iovstack[UIO_FASTIOV]; |
89bddce5 | 2239 | struct iovec *iov = iovstack; |
1da177e4 | 2240 | unsigned long cmsg_ptr; |
08adb7da AV |
2241 | int total_len, len; |
2242 | ssize_t err; | |
1da177e4 LT |
2243 | |
2244 | /* kernel mode address */ | |
230b1839 | 2245 | struct sockaddr_storage addr; |
1da177e4 LT |
2246 | |
2247 | /* user mode address pointers */ | |
2248 | struct sockaddr __user *uaddr; | |
08adb7da | 2249 | int __user *uaddr_len = COMPAT_NAMELEN(msg); |
89bddce5 | 2250 | |
08adb7da | 2251 | msg_sys->msg_name = &addr; |
1da177e4 | 2252 | |
f3d33426 | 2253 | if (MSG_CMSG_COMPAT & flags) |
08adb7da | 2254 | err = get_compat_msghdr(msg_sys, msg_compat, &uaddr, &iov); |
f3d33426 | 2255 | else |
08adb7da | 2256 | err = copy_msghdr_from_user(msg_sys, msg, &uaddr, &iov); |
1da177e4 LT |
2257 | if (err < 0) |
2258 | goto out_freeiov; | |
89bddce5 | 2259 | total_len = err; |
1da177e4 | 2260 | |
a2e27255 ACM |
2261 | cmsg_ptr = (unsigned long)msg_sys->msg_control; |
2262 | msg_sys->msg_flags = flags & (MSG_CMSG_CLOEXEC|MSG_CMSG_COMPAT); | |
89bddce5 | 2263 | |
f3d33426 HFS |
2264 | /* We assume all kernel code knows the size of sockaddr_storage */ |
2265 | msg_sys->msg_namelen = 0; | |
2266 | ||
1da177e4 LT |
2267 | if (sock->file->f_flags & O_NONBLOCK) |
2268 | flags |= MSG_DONTWAIT; | |
a2e27255 ACM |
2269 | err = (nosec ? sock_recvmsg_nosec : sock_recvmsg)(sock, msg_sys, |
2270 | total_len, flags); | |
1da177e4 LT |
2271 | if (err < 0) |
2272 | goto out_freeiov; | |
2273 | len = err; | |
2274 | ||
2275 | if (uaddr != NULL) { | |
43db362d | 2276 | err = move_addr_to_user(&addr, |
a2e27255 | 2277 | msg_sys->msg_namelen, uaddr, |
89bddce5 | 2278 | uaddr_len); |
1da177e4 LT |
2279 | if (err < 0) |
2280 | goto out_freeiov; | |
2281 | } | |
a2e27255 | 2282 | err = __put_user((msg_sys->msg_flags & ~MSG_CMSG_COMPAT), |
37f7f421 | 2283 | COMPAT_FLAGS(msg)); |
1da177e4 LT |
2284 | if (err) |
2285 | goto out_freeiov; | |
2286 | if (MSG_CMSG_COMPAT & flags) | |
a2e27255 | 2287 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2288 | &msg_compat->msg_controllen); |
2289 | else | |
a2e27255 | 2290 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2291 | &msg->msg_controllen); |
2292 | if (err) | |
2293 | goto out_freeiov; | |
2294 | err = len; | |
2295 | ||
2296 | out_freeiov: | |
2297 | if (iov != iovstack) | |
a74e9106 | 2298 | kfree(iov); |
a2e27255 ACM |
2299 | return err; |
2300 | } | |
2301 | ||
2302 | /* | |
2303 | * BSD recvmsg interface | |
2304 | */ | |
2305 | ||
666547ff | 2306 | long __sys_recvmsg(int fd, struct user_msghdr __user *msg, unsigned flags) |
a2e27255 ACM |
2307 | { |
2308 | int fput_needed, err; | |
2309 | struct msghdr msg_sys; | |
1be374a0 AL |
2310 | struct socket *sock; |
2311 | ||
1be374a0 | 2312 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
a2e27255 ACM |
2313 | if (!sock) |
2314 | goto out; | |
2315 | ||
a7526eb5 | 2316 | err = ___sys_recvmsg(sock, msg, &msg_sys, flags, 0); |
a2e27255 | 2317 | |
6cb153ca | 2318 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
2319 | out: |
2320 | return err; | |
2321 | } | |
2322 | ||
666547ff | 2323 | SYSCALL_DEFINE3(recvmsg, int, fd, struct user_msghdr __user *, msg, |
a7526eb5 AL |
2324 | unsigned int, flags) |
2325 | { | |
2326 | if (flags & MSG_CMSG_COMPAT) | |
2327 | return -EINVAL; | |
2328 | return __sys_recvmsg(fd, msg, flags); | |
2329 | } | |
2330 | ||
a2e27255 ACM |
2331 | /* |
2332 | * Linux recvmmsg interface | |
2333 | */ | |
2334 | ||
2335 | int __sys_recvmmsg(int fd, struct mmsghdr __user *mmsg, unsigned int vlen, | |
2336 | unsigned int flags, struct timespec *timeout) | |
2337 | { | |
2338 | int fput_needed, err, datagrams; | |
2339 | struct socket *sock; | |
2340 | struct mmsghdr __user *entry; | |
d7256d0e | 2341 | struct compat_mmsghdr __user *compat_entry; |
a2e27255 ACM |
2342 | struct msghdr msg_sys; |
2343 | struct timespec end_time; | |
2344 | ||
2345 | if (timeout && | |
2346 | poll_select_set_timeout(&end_time, timeout->tv_sec, | |
2347 | timeout->tv_nsec)) | |
2348 | return -EINVAL; | |
2349 | ||
2350 | datagrams = 0; | |
2351 | ||
2352 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2353 | if (!sock) | |
2354 | return err; | |
2355 | ||
2356 | err = sock_error(sock->sk); | |
2357 | if (err) | |
2358 | goto out_put; | |
2359 | ||
2360 | entry = mmsg; | |
d7256d0e | 2361 | compat_entry = (struct compat_mmsghdr __user *)mmsg; |
a2e27255 ACM |
2362 | |
2363 | while (datagrams < vlen) { | |
2364 | /* | |
2365 | * No need to ask LSM for more than the first datagram. | |
2366 | */ | |
d7256d0e | 2367 | if (MSG_CMSG_COMPAT & flags) { |
666547ff | 2368 | err = ___sys_recvmsg(sock, (struct user_msghdr __user *)compat_entry, |
a7526eb5 AL |
2369 | &msg_sys, flags & ~MSG_WAITFORONE, |
2370 | datagrams); | |
d7256d0e JMG |
2371 | if (err < 0) |
2372 | break; | |
2373 | err = __put_user(err, &compat_entry->msg_len); | |
2374 | ++compat_entry; | |
2375 | } else { | |
a7526eb5 | 2376 | err = ___sys_recvmsg(sock, |
666547ff | 2377 | (struct user_msghdr __user *)entry, |
a7526eb5 AL |
2378 | &msg_sys, flags & ~MSG_WAITFORONE, |
2379 | datagrams); | |
d7256d0e JMG |
2380 | if (err < 0) |
2381 | break; | |
2382 | err = put_user(err, &entry->msg_len); | |
2383 | ++entry; | |
2384 | } | |
2385 | ||
a2e27255 ACM |
2386 | if (err) |
2387 | break; | |
a2e27255 ACM |
2388 | ++datagrams; |
2389 | ||
71c5c159 BB |
2390 | /* MSG_WAITFORONE turns on MSG_DONTWAIT after one packet */ |
2391 | if (flags & MSG_WAITFORONE) | |
2392 | flags |= MSG_DONTWAIT; | |
2393 | ||
a2e27255 ACM |
2394 | if (timeout) { |
2395 | ktime_get_ts(timeout); | |
2396 | *timeout = timespec_sub(end_time, *timeout); | |
2397 | if (timeout->tv_sec < 0) { | |
2398 | timeout->tv_sec = timeout->tv_nsec = 0; | |
2399 | break; | |
2400 | } | |
2401 | ||
2402 | /* Timeout, return less than vlen datagrams */ | |
2403 | if (timeout->tv_nsec == 0 && timeout->tv_sec == 0) | |
2404 | break; | |
2405 | } | |
2406 | ||
2407 | /* Out of band data, return right away */ | |
2408 | if (msg_sys.msg_flags & MSG_OOB) | |
2409 | break; | |
2410 | } | |
2411 | ||
2412 | out_put: | |
2413 | fput_light(sock->file, fput_needed); | |
1da177e4 | 2414 | |
a2e27255 ACM |
2415 | if (err == 0) |
2416 | return datagrams; | |
2417 | ||
2418 | if (datagrams != 0) { | |
2419 | /* | |
2420 | * We may return less entries than requested (vlen) if the | |
2421 | * sock is non block and there aren't enough datagrams... | |
2422 | */ | |
2423 | if (err != -EAGAIN) { | |
2424 | /* | |
2425 | * ... or if recvmsg returns an error after we | |
2426 | * received some datagrams, where we record the | |
2427 | * error to return on the next call or if the | |
2428 | * app asks about it using getsockopt(SO_ERROR). | |
2429 | */ | |
2430 | sock->sk->sk_err = -err; | |
2431 | } | |
2432 | ||
2433 | return datagrams; | |
2434 | } | |
2435 | ||
2436 | return err; | |
2437 | } | |
2438 | ||
2439 | SYSCALL_DEFINE5(recvmmsg, int, fd, struct mmsghdr __user *, mmsg, | |
2440 | unsigned int, vlen, unsigned int, flags, | |
2441 | struct timespec __user *, timeout) | |
2442 | { | |
2443 | int datagrams; | |
2444 | struct timespec timeout_sys; | |
2445 | ||
1be374a0 AL |
2446 | if (flags & MSG_CMSG_COMPAT) |
2447 | return -EINVAL; | |
2448 | ||
a2e27255 ACM |
2449 | if (!timeout) |
2450 | return __sys_recvmmsg(fd, mmsg, vlen, flags, NULL); | |
2451 | ||
2452 | if (copy_from_user(&timeout_sys, timeout, sizeof(timeout_sys))) | |
2453 | return -EFAULT; | |
2454 | ||
2455 | datagrams = __sys_recvmmsg(fd, mmsg, vlen, flags, &timeout_sys); | |
2456 | ||
2457 | if (datagrams > 0 && | |
2458 | copy_to_user(timeout, &timeout_sys, sizeof(timeout_sys))) | |
2459 | datagrams = -EFAULT; | |
2460 | ||
2461 | return datagrams; | |
2462 | } | |
2463 | ||
2464 | #ifdef __ARCH_WANT_SYS_SOCKETCALL | |
1da177e4 LT |
2465 | /* Argument list sizes for sys_socketcall */ |
2466 | #define AL(x) ((x) * sizeof(unsigned long)) | |
228e548e | 2467 | static const unsigned char nargs[21] = { |
c6d409cf ED |
2468 | AL(0), AL(3), AL(3), AL(3), AL(2), AL(3), |
2469 | AL(3), AL(3), AL(4), AL(4), AL(4), AL(6), | |
2470 | AL(6), AL(2), AL(5), AL(5), AL(3), AL(3), | |
228e548e | 2471 | AL(4), AL(5), AL(4) |
89bddce5 SH |
2472 | }; |
2473 | ||
1da177e4 LT |
2474 | #undef AL |
2475 | ||
2476 | /* | |
89bddce5 | 2477 | * System call vectors. |
1da177e4 LT |
2478 | * |
2479 | * Argument checking cleaned up. Saved 20% in size. | |
2480 | * This function doesn't need to set the kernel lock because | |
89bddce5 | 2481 | * it is set by the callees. |
1da177e4 LT |
2482 | */ |
2483 | ||
3e0fa65f | 2484 | SYSCALL_DEFINE2(socketcall, int, call, unsigned long __user *, args) |
1da177e4 | 2485 | { |
2950fa9d | 2486 | unsigned long a[AUDITSC_ARGS]; |
89bddce5 | 2487 | unsigned long a0, a1; |
1da177e4 | 2488 | int err; |
47379052 | 2489 | unsigned int len; |
1da177e4 | 2490 | |
228e548e | 2491 | if (call < 1 || call > SYS_SENDMMSG) |
1da177e4 LT |
2492 | return -EINVAL; |
2493 | ||
47379052 AV |
2494 | len = nargs[call]; |
2495 | if (len > sizeof(a)) | |
2496 | return -EINVAL; | |
2497 | ||
1da177e4 | 2498 | /* copy_from_user should be SMP safe. */ |
47379052 | 2499 | if (copy_from_user(a, args, len)) |
1da177e4 | 2500 | return -EFAULT; |
3ec3b2fb | 2501 | |
2950fa9d CG |
2502 | err = audit_socketcall(nargs[call] / sizeof(unsigned long), a); |
2503 | if (err) | |
2504 | return err; | |
3ec3b2fb | 2505 | |
89bddce5 SH |
2506 | a0 = a[0]; |
2507 | a1 = a[1]; | |
2508 | ||
2509 | switch (call) { | |
2510 | case SYS_SOCKET: | |
2511 | err = sys_socket(a0, a1, a[2]); | |
2512 | break; | |
2513 | case SYS_BIND: | |
2514 | err = sys_bind(a0, (struct sockaddr __user *)a1, a[2]); | |
2515 | break; | |
2516 | case SYS_CONNECT: | |
2517 | err = sys_connect(a0, (struct sockaddr __user *)a1, a[2]); | |
2518 | break; | |
2519 | case SYS_LISTEN: | |
2520 | err = sys_listen(a0, a1); | |
2521 | break; | |
2522 | case SYS_ACCEPT: | |
de11defe UD |
2523 | err = sys_accept4(a0, (struct sockaddr __user *)a1, |
2524 | (int __user *)a[2], 0); | |
89bddce5 SH |
2525 | break; |
2526 | case SYS_GETSOCKNAME: | |
2527 | err = | |
2528 | sys_getsockname(a0, (struct sockaddr __user *)a1, | |
2529 | (int __user *)a[2]); | |
2530 | break; | |
2531 | case SYS_GETPEERNAME: | |
2532 | err = | |
2533 | sys_getpeername(a0, (struct sockaddr __user *)a1, | |
2534 | (int __user *)a[2]); | |
2535 | break; | |
2536 | case SYS_SOCKETPAIR: | |
2537 | err = sys_socketpair(a0, a1, a[2], (int __user *)a[3]); | |
2538 | break; | |
2539 | case SYS_SEND: | |
2540 | err = sys_send(a0, (void __user *)a1, a[2], a[3]); | |
2541 | break; | |
2542 | case SYS_SENDTO: | |
2543 | err = sys_sendto(a0, (void __user *)a1, a[2], a[3], | |
2544 | (struct sockaddr __user *)a[4], a[5]); | |
2545 | break; | |
2546 | case SYS_RECV: | |
2547 | err = sys_recv(a0, (void __user *)a1, a[2], a[3]); | |
2548 | break; | |
2549 | case SYS_RECVFROM: | |
2550 | err = sys_recvfrom(a0, (void __user *)a1, a[2], a[3], | |
2551 | (struct sockaddr __user *)a[4], | |
2552 | (int __user *)a[5]); | |
2553 | break; | |
2554 | case SYS_SHUTDOWN: | |
2555 | err = sys_shutdown(a0, a1); | |
2556 | break; | |
2557 | case SYS_SETSOCKOPT: | |
2558 | err = sys_setsockopt(a0, a1, a[2], (char __user *)a[3], a[4]); | |
2559 | break; | |
2560 | case SYS_GETSOCKOPT: | |
2561 | err = | |
2562 | sys_getsockopt(a0, a1, a[2], (char __user *)a[3], | |
2563 | (int __user *)a[4]); | |
2564 | break; | |
2565 | case SYS_SENDMSG: | |
666547ff | 2566 | err = sys_sendmsg(a0, (struct user_msghdr __user *)a1, a[2]); |
89bddce5 | 2567 | break; |
228e548e AB |
2568 | case SYS_SENDMMSG: |
2569 | err = sys_sendmmsg(a0, (struct mmsghdr __user *)a1, a[2], a[3]); | |
2570 | break; | |
89bddce5 | 2571 | case SYS_RECVMSG: |
666547ff | 2572 | err = sys_recvmsg(a0, (struct user_msghdr __user *)a1, a[2]); |
89bddce5 | 2573 | break; |
a2e27255 ACM |
2574 | case SYS_RECVMMSG: |
2575 | err = sys_recvmmsg(a0, (struct mmsghdr __user *)a1, a[2], a[3], | |
2576 | (struct timespec __user *)a[4]); | |
2577 | break; | |
de11defe UD |
2578 | case SYS_ACCEPT4: |
2579 | err = sys_accept4(a0, (struct sockaddr __user *)a1, | |
2580 | (int __user *)a[2], a[3]); | |
aaca0bdc | 2581 | break; |
89bddce5 SH |
2582 | default: |
2583 | err = -EINVAL; | |
2584 | break; | |
1da177e4 LT |
2585 | } |
2586 | return err; | |
2587 | } | |
2588 | ||
89bddce5 | 2589 | #endif /* __ARCH_WANT_SYS_SOCKETCALL */ |
1da177e4 | 2590 | |
55737fda SH |
2591 | /** |
2592 | * sock_register - add a socket protocol handler | |
2593 | * @ops: description of protocol | |
2594 | * | |
1da177e4 LT |
2595 | * This function is called by a protocol handler that wants to |
2596 | * advertise its address family, and have it linked into the | |
e793c0f7 | 2597 | * socket interface. The value ops->family corresponds to the |
55737fda | 2598 | * socket system call protocol family. |
1da177e4 | 2599 | */ |
f0fd27d4 | 2600 | int sock_register(const struct net_proto_family *ops) |
1da177e4 LT |
2601 | { |
2602 | int err; | |
2603 | ||
2604 | if (ops->family >= NPROTO) { | |
3410f22e | 2605 | pr_crit("protocol %d >= NPROTO(%d)\n", ops->family, NPROTO); |
1da177e4 LT |
2606 | return -ENOBUFS; |
2607 | } | |
55737fda SH |
2608 | |
2609 | spin_lock(&net_family_lock); | |
190683a9 ED |
2610 | if (rcu_dereference_protected(net_families[ops->family], |
2611 | lockdep_is_held(&net_family_lock))) | |
55737fda SH |
2612 | err = -EEXIST; |
2613 | else { | |
cf778b00 | 2614 | rcu_assign_pointer(net_families[ops->family], ops); |
1da177e4 LT |
2615 | err = 0; |
2616 | } | |
55737fda SH |
2617 | spin_unlock(&net_family_lock); |
2618 | ||
3410f22e | 2619 | pr_info("NET: Registered protocol family %d\n", ops->family); |
1da177e4 LT |
2620 | return err; |
2621 | } | |
c6d409cf | 2622 | EXPORT_SYMBOL(sock_register); |
1da177e4 | 2623 | |
55737fda SH |
2624 | /** |
2625 | * sock_unregister - remove a protocol handler | |
2626 | * @family: protocol family to remove | |
2627 | * | |
1da177e4 LT |
2628 | * This function is called by a protocol handler that wants to |
2629 | * remove its address family, and have it unlinked from the | |
55737fda SH |
2630 | * new socket creation. |
2631 | * | |
2632 | * If protocol handler is a module, then it can use module reference | |
2633 | * counts to protect against new references. If protocol handler is not | |
2634 | * a module then it needs to provide its own protection in | |
2635 | * the ops->create routine. | |
1da177e4 | 2636 | */ |
f0fd27d4 | 2637 | void sock_unregister(int family) |
1da177e4 | 2638 | { |
f0fd27d4 | 2639 | BUG_ON(family < 0 || family >= NPROTO); |
1da177e4 | 2640 | |
55737fda | 2641 | spin_lock(&net_family_lock); |
a9b3cd7f | 2642 | RCU_INIT_POINTER(net_families[family], NULL); |
55737fda SH |
2643 | spin_unlock(&net_family_lock); |
2644 | ||
2645 | synchronize_rcu(); | |
2646 | ||
3410f22e | 2647 | pr_info("NET: Unregistered protocol family %d\n", family); |
1da177e4 | 2648 | } |
c6d409cf | 2649 | EXPORT_SYMBOL(sock_unregister); |
1da177e4 | 2650 | |
77d76ea3 | 2651 | static int __init sock_init(void) |
1da177e4 | 2652 | { |
b3e19d92 | 2653 | int err; |
2ca794e5 EB |
2654 | /* |
2655 | * Initialize the network sysctl infrastructure. | |
2656 | */ | |
2657 | err = net_sysctl_init(); | |
2658 | if (err) | |
2659 | goto out; | |
b3e19d92 | 2660 | |
1da177e4 | 2661 | /* |
89bddce5 | 2662 | * Initialize skbuff SLAB cache |
1da177e4 LT |
2663 | */ |
2664 | skb_init(); | |
1da177e4 LT |
2665 | |
2666 | /* | |
89bddce5 | 2667 | * Initialize the protocols module. |
1da177e4 LT |
2668 | */ |
2669 | ||
2670 | init_inodecache(); | |
b3e19d92 NP |
2671 | |
2672 | err = register_filesystem(&sock_fs_type); | |
2673 | if (err) | |
2674 | goto out_fs; | |
1da177e4 | 2675 | sock_mnt = kern_mount(&sock_fs_type); |
b3e19d92 NP |
2676 | if (IS_ERR(sock_mnt)) { |
2677 | err = PTR_ERR(sock_mnt); | |
2678 | goto out_mount; | |
2679 | } | |
77d76ea3 AK |
2680 | |
2681 | /* The real protocol initialization is performed in later initcalls. | |
1da177e4 LT |
2682 | */ |
2683 | ||
2684 | #ifdef CONFIG_NETFILTER | |
6d11cfdb PNA |
2685 | err = netfilter_init(); |
2686 | if (err) | |
2687 | goto out; | |
1da177e4 | 2688 | #endif |
cbeb321a | 2689 | |
408eccce | 2690 | ptp_classifier_init(); |
c1f19b51 | 2691 | |
b3e19d92 NP |
2692 | out: |
2693 | return err; | |
2694 | ||
2695 | out_mount: | |
2696 | unregister_filesystem(&sock_fs_type); | |
2697 | out_fs: | |
2698 | goto out; | |
1da177e4 LT |
2699 | } |
2700 | ||
77d76ea3 AK |
2701 | core_initcall(sock_init); /* early initcall */ |
2702 | ||
1da177e4 LT |
2703 | #ifdef CONFIG_PROC_FS |
2704 | void socket_seq_show(struct seq_file *seq) | |
2705 | { | |
2706 | int cpu; | |
2707 | int counter = 0; | |
2708 | ||
6f912042 | 2709 | for_each_possible_cpu(cpu) |
89bddce5 | 2710 | counter += per_cpu(sockets_in_use, cpu); |
1da177e4 LT |
2711 | |
2712 | /* It can be negative, by the way. 8) */ | |
2713 | if (counter < 0) | |
2714 | counter = 0; | |
2715 | ||
2716 | seq_printf(seq, "sockets: used %d\n", counter); | |
2717 | } | |
89bddce5 | 2718 | #endif /* CONFIG_PROC_FS */ |
1da177e4 | 2719 | |
89bbfc95 | 2720 | #ifdef CONFIG_COMPAT |
6b96018b | 2721 | static int do_siocgstamp(struct net *net, struct socket *sock, |
644595f8 | 2722 | unsigned int cmd, void __user *up) |
7a229387 | 2723 | { |
7a229387 AB |
2724 | mm_segment_t old_fs = get_fs(); |
2725 | struct timeval ktv; | |
2726 | int err; | |
2727 | ||
2728 | set_fs(KERNEL_DS); | |
6b96018b | 2729 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)&ktv); |
7a229387 | 2730 | set_fs(old_fs); |
644595f8 | 2731 | if (!err) |
ed6fe9d6 | 2732 | err = compat_put_timeval(&ktv, up); |
644595f8 | 2733 | |
7a229387 AB |
2734 | return err; |
2735 | } | |
2736 | ||
6b96018b | 2737 | static int do_siocgstampns(struct net *net, struct socket *sock, |
644595f8 | 2738 | unsigned int cmd, void __user *up) |
7a229387 | 2739 | { |
7a229387 AB |
2740 | mm_segment_t old_fs = get_fs(); |
2741 | struct timespec kts; | |
2742 | int err; | |
2743 | ||
2744 | set_fs(KERNEL_DS); | |
6b96018b | 2745 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)&kts); |
7a229387 | 2746 | set_fs(old_fs); |
644595f8 | 2747 | if (!err) |
ed6fe9d6 | 2748 | err = compat_put_timespec(&kts, up); |
644595f8 | 2749 | |
7a229387 AB |
2750 | return err; |
2751 | } | |
2752 | ||
6b96018b | 2753 | static int dev_ifname32(struct net *net, struct compat_ifreq __user *uifr32) |
7a229387 AB |
2754 | { |
2755 | struct ifreq __user *uifr; | |
2756 | int err; | |
2757 | ||
2758 | uifr = compat_alloc_user_space(sizeof(struct ifreq)); | |
6b96018b | 2759 | if (copy_in_user(uifr, uifr32, sizeof(struct compat_ifreq))) |
7a229387 AB |
2760 | return -EFAULT; |
2761 | ||
6b96018b | 2762 | err = dev_ioctl(net, SIOCGIFNAME, uifr); |
7a229387 AB |
2763 | if (err) |
2764 | return err; | |
2765 | ||
6b96018b | 2766 | if (copy_in_user(uifr32, uifr, sizeof(struct compat_ifreq))) |
7a229387 AB |
2767 | return -EFAULT; |
2768 | ||
2769 | return 0; | |
2770 | } | |
2771 | ||
6b96018b | 2772 | static int dev_ifconf(struct net *net, struct compat_ifconf __user *uifc32) |
7a229387 | 2773 | { |
6b96018b | 2774 | struct compat_ifconf ifc32; |
7a229387 AB |
2775 | struct ifconf ifc; |
2776 | struct ifconf __user *uifc; | |
6b96018b | 2777 | struct compat_ifreq __user *ifr32; |
7a229387 AB |
2778 | struct ifreq __user *ifr; |
2779 | unsigned int i, j; | |
2780 | int err; | |
2781 | ||
6b96018b | 2782 | if (copy_from_user(&ifc32, uifc32, sizeof(struct compat_ifconf))) |
7a229387 AB |
2783 | return -EFAULT; |
2784 | ||
43da5f2e | 2785 | memset(&ifc, 0, sizeof(ifc)); |
7a229387 AB |
2786 | if (ifc32.ifcbuf == 0) { |
2787 | ifc32.ifc_len = 0; | |
2788 | ifc.ifc_len = 0; | |
2789 | ifc.ifc_req = NULL; | |
2790 | uifc = compat_alloc_user_space(sizeof(struct ifconf)); | |
2791 | } else { | |
c6d409cf ED |
2792 | size_t len = ((ifc32.ifc_len / sizeof(struct compat_ifreq)) + 1) * |
2793 | sizeof(struct ifreq); | |
7a229387 AB |
2794 | uifc = compat_alloc_user_space(sizeof(struct ifconf) + len); |
2795 | ifc.ifc_len = len; | |
2796 | ifr = ifc.ifc_req = (void __user *)(uifc + 1); | |
2797 | ifr32 = compat_ptr(ifc32.ifcbuf); | |
c6d409cf | 2798 | for (i = 0; i < ifc32.ifc_len; i += sizeof(struct compat_ifreq)) { |
6b96018b | 2799 | if (copy_in_user(ifr, ifr32, sizeof(struct compat_ifreq))) |
7a229387 AB |
2800 | return -EFAULT; |
2801 | ifr++; | |
2802 | ifr32++; | |
2803 | } | |
2804 | } | |
2805 | if (copy_to_user(uifc, &ifc, sizeof(struct ifconf))) | |
2806 | return -EFAULT; | |
2807 | ||
6b96018b | 2808 | err = dev_ioctl(net, SIOCGIFCONF, uifc); |
7a229387 AB |
2809 | if (err) |
2810 | return err; | |
2811 | ||
2812 | if (copy_from_user(&ifc, uifc, sizeof(struct ifconf))) | |
2813 | return -EFAULT; | |
2814 | ||
2815 | ifr = ifc.ifc_req; | |
2816 | ifr32 = compat_ptr(ifc32.ifcbuf); | |
2817 | for (i = 0, j = 0; | |
c6d409cf ED |
2818 | i + sizeof(struct compat_ifreq) <= ifc32.ifc_len && j < ifc.ifc_len; |
2819 | i += sizeof(struct compat_ifreq), j += sizeof(struct ifreq)) { | |
2820 | if (copy_in_user(ifr32, ifr, sizeof(struct compat_ifreq))) | |
7a229387 AB |
2821 | return -EFAULT; |
2822 | ifr32++; | |
2823 | ifr++; | |
2824 | } | |
2825 | ||
2826 | if (ifc32.ifcbuf == 0) { | |
2827 | /* Translate from 64-bit structure multiple to | |
2828 | * a 32-bit one. | |
2829 | */ | |
2830 | i = ifc.ifc_len; | |
6b96018b | 2831 | i = ((i / sizeof(struct ifreq)) * sizeof(struct compat_ifreq)); |
7a229387 AB |
2832 | ifc32.ifc_len = i; |
2833 | } else { | |
2834 | ifc32.ifc_len = i; | |
2835 | } | |
6b96018b | 2836 | if (copy_to_user(uifc32, &ifc32, sizeof(struct compat_ifconf))) |
7a229387 AB |
2837 | return -EFAULT; |
2838 | ||
2839 | return 0; | |
2840 | } | |
2841 | ||
6b96018b | 2842 | static int ethtool_ioctl(struct net *net, struct compat_ifreq __user *ifr32) |
7a229387 | 2843 | { |
3a7da39d BH |
2844 | struct compat_ethtool_rxnfc __user *compat_rxnfc; |
2845 | bool convert_in = false, convert_out = false; | |
2846 | size_t buf_size = ALIGN(sizeof(struct ifreq), 8); | |
2847 | struct ethtool_rxnfc __user *rxnfc; | |
7a229387 | 2848 | struct ifreq __user *ifr; |
3a7da39d BH |
2849 | u32 rule_cnt = 0, actual_rule_cnt; |
2850 | u32 ethcmd; | |
7a229387 | 2851 | u32 data; |
3a7da39d | 2852 | int ret; |
7a229387 | 2853 | |
3a7da39d BH |
2854 | if (get_user(data, &ifr32->ifr_ifru.ifru_data)) |
2855 | return -EFAULT; | |
7a229387 | 2856 | |
3a7da39d BH |
2857 | compat_rxnfc = compat_ptr(data); |
2858 | ||
2859 | if (get_user(ethcmd, &compat_rxnfc->cmd)) | |
7a229387 AB |
2860 | return -EFAULT; |
2861 | ||
3a7da39d BH |
2862 | /* Most ethtool structures are defined without padding. |
2863 | * Unfortunately struct ethtool_rxnfc is an exception. | |
2864 | */ | |
2865 | switch (ethcmd) { | |
2866 | default: | |
2867 | break; | |
2868 | case ETHTOOL_GRXCLSRLALL: | |
2869 | /* Buffer size is variable */ | |
2870 | if (get_user(rule_cnt, &compat_rxnfc->rule_cnt)) | |
2871 | return -EFAULT; | |
2872 | if (rule_cnt > KMALLOC_MAX_SIZE / sizeof(u32)) | |
2873 | return -ENOMEM; | |
2874 | buf_size += rule_cnt * sizeof(u32); | |
2875 | /* fall through */ | |
2876 | case ETHTOOL_GRXRINGS: | |
2877 | case ETHTOOL_GRXCLSRLCNT: | |
2878 | case ETHTOOL_GRXCLSRULE: | |
55664f32 | 2879 | case ETHTOOL_SRXCLSRLINS: |
3a7da39d BH |
2880 | convert_out = true; |
2881 | /* fall through */ | |
2882 | case ETHTOOL_SRXCLSRLDEL: | |
3a7da39d BH |
2883 | buf_size += sizeof(struct ethtool_rxnfc); |
2884 | convert_in = true; | |
2885 | break; | |
2886 | } | |
2887 | ||
2888 | ifr = compat_alloc_user_space(buf_size); | |
954b1244 | 2889 | rxnfc = (void __user *)ifr + ALIGN(sizeof(struct ifreq), 8); |
3a7da39d BH |
2890 | |
2891 | if (copy_in_user(&ifr->ifr_name, &ifr32->ifr_name, IFNAMSIZ)) | |
7a229387 AB |
2892 | return -EFAULT; |
2893 | ||
3a7da39d BH |
2894 | if (put_user(convert_in ? rxnfc : compat_ptr(data), |
2895 | &ifr->ifr_ifru.ifru_data)) | |
7a229387 AB |
2896 | return -EFAULT; |
2897 | ||
3a7da39d | 2898 | if (convert_in) { |
127fe533 | 2899 | /* We expect there to be holes between fs.m_ext and |
3a7da39d BH |
2900 | * fs.ring_cookie and at the end of fs, but nowhere else. |
2901 | */ | |
127fe533 AD |
2902 | BUILD_BUG_ON(offsetof(struct compat_ethtool_rxnfc, fs.m_ext) + |
2903 | sizeof(compat_rxnfc->fs.m_ext) != | |
2904 | offsetof(struct ethtool_rxnfc, fs.m_ext) + | |
2905 | sizeof(rxnfc->fs.m_ext)); | |
3a7da39d BH |
2906 | BUILD_BUG_ON( |
2907 | offsetof(struct compat_ethtool_rxnfc, fs.location) - | |
2908 | offsetof(struct compat_ethtool_rxnfc, fs.ring_cookie) != | |
2909 | offsetof(struct ethtool_rxnfc, fs.location) - | |
2910 | offsetof(struct ethtool_rxnfc, fs.ring_cookie)); | |
2911 | ||
2912 | if (copy_in_user(rxnfc, compat_rxnfc, | |
954b1244 SH |
2913 | (void __user *)(&rxnfc->fs.m_ext + 1) - |
2914 | (void __user *)rxnfc) || | |
3a7da39d BH |
2915 | copy_in_user(&rxnfc->fs.ring_cookie, |
2916 | &compat_rxnfc->fs.ring_cookie, | |
954b1244 SH |
2917 | (void __user *)(&rxnfc->fs.location + 1) - |
2918 | (void __user *)&rxnfc->fs.ring_cookie) || | |
3a7da39d BH |
2919 | copy_in_user(&rxnfc->rule_cnt, &compat_rxnfc->rule_cnt, |
2920 | sizeof(rxnfc->rule_cnt))) | |
2921 | return -EFAULT; | |
2922 | } | |
2923 | ||
2924 | ret = dev_ioctl(net, SIOCETHTOOL, ifr); | |
2925 | if (ret) | |
2926 | return ret; | |
2927 | ||
2928 | if (convert_out) { | |
2929 | if (copy_in_user(compat_rxnfc, rxnfc, | |
954b1244 SH |
2930 | (const void __user *)(&rxnfc->fs.m_ext + 1) - |
2931 | (const void __user *)rxnfc) || | |
3a7da39d BH |
2932 | copy_in_user(&compat_rxnfc->fs.ring_cookie, |
2933 | &rxnfc->fs.ring_cookie, | |
954b1244 SH |
2934 | (const void __user *)(&rxnfc->fs.location + 1) - |
2935 | (const void __user *)&rxnfc->fs.ring_cookie) || | |
3a7da39d BH |
2936 | copy_in_user(&compat_rxnfc->rule_cnt, &rxnfc->rule_cnt, |
2937 | sizeof(rxnfc->rule_cnt))) | |
2938 | return -EFAULT; | |
2939 | ||
2940 | if (ethcmd == ETHTOOL_GRXCLSRLALL) { | |
2941 | /* As an optimisation, we only copy the actual | |
2942 | * number of rules that the underlying | |
2943 | * function returned. Since Mallory might | |
2944 | * change the rule count in user memory, we | |
2945 | * check that it is less than the rule count | |
2946 | * originally given (as the user buffer size), | |
2947 | * which has been range-checked. | |
2948 | */ | |
2949 | if (get_user(actual_rule_cnt, &rxnfc->rule_cnt)) | |
2950 | return -EFAULT; | |
2951 | if (actual_rule_cnt < rule_cnt) | |
2952 | rule_cnt = actual_rule_cnt; | |
2953 | if (copy_in_user(&compat_rxnfc->rule_locs[0], | |
2954 | &rxnfc->rule_locs[0], | |
2955 | rule_cnt * sizeof(u32))) | |
2956 | return -EFAULT; | |
2957 | } | |
2958 | } | |
2959 | ||
2960 | return 0; | |
7a229387 AB |
2961 | } |
2962 | ||
7a50a240 AB |
2963 | static int compat_siocwandev(struct net *net, struct compat_ifreq __user *uifr32) |
2964 | { | |
2965 | void __user *uptr; | |
2966 | compat_uptr_t uptr32; | |
2967 | struct ifreq __user *uifr; | |
2968 | ||
c6d409cf | 2969 | uifr = compat_alloc_user_space(sizeof(*uifr)); |
7a50a240 AB |
2970 | if (copy_in_user(uifr, uifr32, sizeof(struct compat_ifreq))) |
2971 | return -EFAULT; | |
2972 | ||
2973 | if (get_user(uptr32, &uifr32->ifr_settings.ifs_ifsu)) | |
2974 | return -EFAULT; | |
2975 | ||
2976 | uptr = compat_ptr(uptr32); | |
2977 | ||
2978 | if (put_user(uptr, &uifr->ifr_settings.ifs_ifsu.raw_hdlc)) | |
2979 | return -EFAULT; | |
2980 | ||
2981 | return dev_ioctl(net, SIOCWANDEV, uifr); | |
2982 | } | |
2983 | ||
6b96018b AB |
2984 | static int bond_ioctl(struct net *net, unsigned int cmd, |
2985 | struct compat_ifreq __user *ifr32) | |
7a229387 AB |
2986 | { |
2987 | struct ifreq kifr; | |
7a229387 AB |
2988 | mm_segment_t old_fs; |
2989 | int err; | |
7a229387 AB |
2990 | |
2991 | switch (cmd) { | |
2992 | case SIOCBONDENSLAVE: | |
2993 | case SIOCBONDRELEASE: | |
2994 | case SIOCBONDSETHWADDR: | |
2995 | case SIOCBONDCHANGEACTIVE: | |
6b96018b | 2996 | if (copy_from_user(&kifr, ifr32, sizeof(struct compat_ifreq))) |
7a229387 AB |
2997 | return -EFAULT; |
2998 | ||
2999 | old_fs = get_fs(); | |
c6d409cf | 3000 | set_fs(KERNEL_DS); |
c3f52ae6 | 3001 | err = dev_ioctl(net, cmd, |
3002 | (struct ifreq __user __force *) &kifr); | |
c6d409cf | 3003 | set_fs(old_fs); |
7a229387 AB |
3004 | |
3005 | return err; | |
7a229387 | 3006 | default: |
07d106d0 | 3007 | return -ENOIOCTLCMD; |
ccbd6a5a | 3008 | } |
7a229387 AB |
3009 | } |
3010 | ||
590d4693 BH |
3011 | /* Handle ioctls that use ifreq::ifr_data and just need struct ifreq converted */ |
3012 | static int compat_ifr_data_ioctl(struct net *net, unsigned int cmd, | |
6b96018b | 3013 | struct compat_ifreq __user *u_ifreq32) |
7a229387 AB |
3014 | { |
3015 | struct ifreq __user *u_ifreq64; | |
7a229387 AB |
3016 | char tmp_buf[IFNAMSIZ]; |
3017 | void __user *data64; | |
3018 | u32 data32; | |
3019 | ||
3020 | if (copy_from_user(&tmp_buf[0], &(u_ifreq32->ifr_ifrn.ifrn_name[0]), | |
3021 | IFNAMSIZ)) | |
3022 | return -EFAULT; | |
417c3522 | 3023 | if (get_user(data32, &u_ifreq32->ifr_ifru.ifru_data)) |
7a229387 AB |
3024 | return -EFAULT; |
3025 | data64 = compat_ptr(data32); | |
3026 | ||
3027 | u_ifreq64 = compat_alloc_user_space(sizeof(*u_ifreq64)); | |
3028 | ||
7a229387 AB |
3029 | if (copy_to_user(&u_ifreq64->ifr_ifrn.ifrn_name[0], &tmp_buf[0], |
3030 | IFNAMSIZ)) | |
3031 | return -EFAULT; | |
417c3522 | 3032 | if (put_user(data64, &u_ifreq64->ifr_ifru.ifru_data)) |
7a229387 AB |
3033 | return -EFAULT; |
3034 | ||
6b96018b | 3035 | return dev_ioctl(net, cmd, u_ifreq64); |
7a229387 AB |
3036 | } |
3037 | ||
6b96018b AB |
3038 | static int dev_ifsioc(struct net *net, struct socket *sock, |
3039 | unsigned int cmd, struct compat_ifreq __user *uifr32) | |
7a229387 | 3040 | { |
a2116ed2 | 3041 | struct ifreq __user *uifr; |
7a229387 AB |
3042 | int err; |
3043 | ||
a2116ed2 AB |
3044 | uifr = compat_alloc_user_space(sizeof(*uifr)); |
3045 | if (copy_in_user(uifr, uifr32, sizeof(*uifr32))) | |
3046 | return -EFAULT; | |
3047 | ||
3048 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)uifr); | |
3049 | ||
7a229387 AB |
3050 | if (!err) { |
3051 | switch (cmd) { | |
3052 | case SIOCGIFFLAGS: | |
3053 | case SIOCGIFMETRIC: | |
3054 | case SIOCGIFMTU: | |
3055 | case SIOCGIFMEM: | |
3056 | case SIOCGIFHWADDR: | |
3057 | case SIOCGIFINDEX: | |
3058 | case SIOCGIFADDR: | |
3059 | case SIOCGIFBRDADDR: | |
3060 | case SIOCGIFDSTADDR: | |
3061 | case SIOCGIFNETMASK: | |
fab2532b | 3062 | case SIOCGIFPFLAGS: |
7a229387 | 3063 | case SIOCGIFTXQLEN: |
fab2532b AB |
3064 | case SIOCGMIIPHY: |
3065 | case SIOCGMIIREG: | |
a2116ed2 | 3066 | if (copy_in_user(uifr32, uifr, sizeof(*uifr32))) |
7a229387 AB |
3067 | err = -EFAULT; |
3068 | break; | |
3069 | } | |
3070 | } | |
3071 | return err; | |
3072 | } | |
3073 | ||
a2116ed2 AB |
3074 | static int compat_sioc_ifmap(struct net *net, unsigned int cmd, |
3075 | struct compat_ifreq __user *uifr32) | |
3076 | { | |
3077 | struct ifreq ifr; | |
3078 | struct compat_ifmap __user *uifmap32; | |
3079 | mm_segment_t old_fs; | |
3080 | int err; | |
3081 | ||
3082 | uifmap32 = &uifr32->ifr_ifru.ifru_map; | |
3083 | err = copy_from_user(&ifr, uifr32, sizeof(ifr.ifr_name)); | |
3ddc5b46 MD |
3084 | err |= get_user(ifr.ifr_map.mem_start, &uifmap32->mem_start); |
3085 | err |= get_user(ifr.ifr_map.mem_end, &uifmap32->mem_end); | |
3086 | err |= get_user(ifr.ifr_map.base_addr, &uifmap32->base_addr); | |
3087 | err |= get_user(ifr.ifr_map.irq, &uifmap32->irq); | |
3088 | err |= get_user(ifr.ifr_map.dma, &uifmap32->dma); | |
3089 | err |= get_user(ifr.ifr_map.port, &uifmap32->port); | |
a2116ed2 AB |
3090 | if (err) |
3091 | return -EFAULT; | |
3092 | ||
3093 | old_fs = get_fs(); | |
c6d409cf | 3094 | set_fs(KERNEL_DS); |
c3f52ae6 | 3095 | err = dev_ioctl(net, cmd, (void __user __force *)&ifr); |
c6d409cf | 3096 | set_fs(old_fs); |
a2116ed2 AB |
3097 | |
3098 | if (cmd == SIOCGIFMAP && !err) { | |
3099 | err = copy_to_user(uifr32, &ifr, sizeof(ifr.ifr_name)); | |
3ddc5b46 MD |
3100 | err |= put_user(ifr.ifr_map.mem_start, &uifmap32->mem_start); |
3101 | err |= put_user(ifr.ifr_map.mem_end, &uifmap32->mem_end); | |
3102 | err |= put_user(ifr.ifr_map.base_addr, &uifmap32->base_addr); | |
3103 | err |= put_user(ifr.ifr_map.irq, &uifmap32->irq); | |
3104 | err |= put_user(ifr.ifr_map.dma, &uifmap32->dma); | |
3105 | err |= put_user(ifr.ifr_map.port, &uifmap32->port); | |
a2116ed2 AB |
3106 | if (err) |
3107 | err = -EFAULT; | |
3108 | } | |
3109 | return err; | |
3110 | } | |
3111 | ||
7a229387 | 3112 | struct rtentry32 { |
c6d409cf | 3113 | u32 rt_pad1; |
7a229387 AB |
3114 | struct sockaddr rt_dst; /* target address */ |
3115 | struct sockaddr rt_gateway; /* gateway addr (RTF_GATEWAY) */ | |
3116 | struct sockaddr rt_genmask; /* target network mask (IP) */ | |
c6d409cf ED |
3117 | unsigned short rt_flags; |
3118 | short rt_pad2; | |
3119 | u32 rt_pad3; | |
3120 | unsigned char rt_tos; | |
3121 | unsigned char rt_class; | |
3122 | short rt_pad4; | |
3123 | short rt_metric; /* +1 for binary compatibility! */ | |
7a229387 | 3124 | /* char * */ u32 rt_dev; /* forcing the device at add */ |
c6d409cf ED |
3125 | u32 rt_mtu; /* per route MTU/Window */ |
3126 | u32 rt_window; /* Window clamping */ | |
7a229387 AB |
3127 | unsigned short rt_irtt; /* Initial RTT */ |
3128 | }; | |
3129 | ||
3130 | struct in6_rtmsg32 { | |
3131 | struct in6_addr rtmsg_dst; | |
3132 | struct in6_addr rtmsg_src; | |
3133 | struct in6_addr rtmsg_gateway; | |
3134 | u32 rtmsg_type; | |
3135 | u16 rtmsg_dst_len; | |
3136 | u16 rtmsg_src_len; | |
3137 | u32 rtmsg_metric; | |
3138 | u32 rtmsg_info; | |
3139 | u32 rtmsg_flags; | |
3140 | s32 rtmsg_ifindex; | |
3141 | }; | |
3142 | ||
6b96018b AB |
3143 | static int routing_ioctl(struct net *net, struct socket *sock, |
3144 | unsigned int cmd, void __user *argp) | |
7a229387 AB |
3145 | { |
3146 | int ret; | |
3147 | void *r = NULL; | |
3148 | struct in6_rtmsg r6; | |
3149 | struct rtentry r4; | |
3150 | char devname[16]; | |
3151 | u32 rtdev; | |
3152 | mm_segment_t old_fs = get_fs(); | |
3153 | ||
6b96018b AB |
3154 | if (sock && sock->sk && sock->sk->sk_family == AF_INET6) { /* ipv6 */ |
3155 | struct in6_rtmsg32 __user *ur6 = argp; | |
c6d409cf | 3156 | ret = copy_from_user(&r6.rtmsg_dst, &(ur6->rtmsg_dst), |
7a229387 | 3157 | 3 * sizeof(struct in6_addr)); |
3ddc5b46 MD |
3158 | ret |= get_user(r6.rtmsg_type, &(ur6->rtmsg_type)); |
3159 | ret |= get_user(r6.rtmsg_dst_len, &(ur6->rtmsg_dst_len)); | |
3160 | ret |= get_user(r6.rtmsg_src_len, &(ur6->rtmsg_src_len)); | |
3161 | ret |= get_user(r6.rtmsg_metric, &(ur6->rtmsg_metric)); | |
3162 | ret |= get_user(r6.rtmsg_info, &(ur6->rtmsg_info)); | |
3163 | ret |= get_user(r6.rtmsg_flags, &(ur6->rtmsg_flags)); | |
3164 | ret |= get_user(r6.rtmsg_ifindex, &(ur6->rtmsg_ifindex)); | |
7a229387 AB |
3165 | |
3166 | r = (void *) &r6; | |
3167 | } else { /* ipv4 */ | |
6b96018b | 3168 | struct rtentry32 __user *ur4 = argp; |
c6d409cf | 3169 | ret = copy_from_user(&r4.rt_dst, &(ur4->rt_dst), |
7a229387 | 3170 | 3 * sizeof(struct sockaddr)); |
3ddc5b46 MD |
3171 | ret |= get_user(r4.rt_flags, &(ur4->rt_flags)); |
3172 | ret |= get_user(r4.rt_metric, &(ur4->rt_metric)); | |
3173 | ret |= get_user(r4.rt_mtu, &(ur4->rt_mtu)); | |
3174 | ret |= get_user(r4.rt_window, &(ur4->rt_window)); | |
3175 | ret |= get_user(r4.rt_irtt, &(ur4->rt_irtt)); | |
3176 | ret |= get_user(rtdev, &(ur4->rt_dev)); | |
7a229387 | 3177 | if (rtdev) { |
c6d409cf | 3178 | ret |= copy_from_user(devname, compat_ptr(rtdev), 15); |
c3f52ae6 | 3179 | r4.rt_dev = (char __user __force *)devname; |
3180 | devname[15] = 0; | |
7a229387 AB |
3181 | } else |
3182 | r4.rt_dev = NULL; | |
3183 | ||
3184 | r = (void *) &r4; | |
3185 | } | |
3186 | ||
3187 | if (ret) { | |
3188 | ret = -EFAULT; | |
3189 | goto out; | |
3190 | } | |
3191 | ||
c6d409cf | 3192 | set_fs(KERNEL_DS); |
6b96018b | 3193 | ret = sock_do_ioctl(net, sock, cmd, (unsigned long) r); |
c6d409cf | 3194 | set_fs(old_fs); |
7a229387 AB |
3195 | |
3196 | out: | |
7a229387 AB |
3197 | return ret; |
3198 | } | |
3199 | ||
3200 | /* Since old style bridge ioctl's endup using SIOCDEVPRIVATE | |
3201 | * for some operations; this forces use of the newer bridge-utils that | |
25985edc | 3202 | * use compatible ioctls |
7a229387 | 3203 | */ |
6b96018b | 3204 | static int old_bridge_ioctl(compat_ulong_t __user *argp) |
7a229387 | 3205 | { |
6b96018b | 3206 | compat_ulong_t tmp; |
7a229387 | 3207 | |
6b96018b | 3208 | if (get_user(tmp, argp)) |
7a229387 AB |
3209 | return -EFAULT; |
3210 | if (tmp == BRCTL_GET_VERSION) | |
3211 | return BRCTL_VERSION + 1; | |
3212 | return -EINVAL; | |
3213 | } | |
3214 | ||
6b96018b AB |
3215 | static int compat_sock_ioctl_trans(struct file *file, struct socket *sock, |
3216 | unsigned int cmd, unsigned long arg) | |
3217 | { | |
3218 | void __user *argp = compat_ptr(arg); | |
3219 | struct sock *sk = sock->sk; | |
3220 | struct net *net = sock_net(sk); | |
7a229387 | 3221 | |
6b96018b | 3222 | if (cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15)) |
590d4693 | 3223 | return compat_ifr_data_ioctl(net, cmd, argp); |
6b96018b AB |
3224 | |
3225 | switch (cmd) { | |
3226 | case SIOCSIFBR: | |
3227 | case SIOCGIFBR: | |
3228 | return old_bridge_ioctl(argp); | |
3229 | case SIOCGIFNAME: | |
3230 | return dev_ifname32(net, argp); | |
3231 | case SIOCGIFCONF: | |
3232 | return dev_ifconf(net, argp); | |
3233 | case SIOCETHTOOL: | |
3234 | return ethtool_ioctl(net, argp); | |
7a50a240 AB |
3235 | case SIOCWANDEV: |
3236 | return compat_siocwandev(net, argp); | |
a2116ed2 AB |
3237 | case SIOCGIFMAP: |
3238 | case SIOCSIFMAP: | |
3239 | return compat_sioc_ifmap(net, cmd, argp); | |
6b96018b AB |
3240 | case SIOCBONDENSLAVE: |
3241 | case SIOCBONDRELEASE: | |
3242 | case SIOCBONDSETHWADDR: | |
6b96018b AB |
3243 | case SIOCBONDCHANGEACTIVE: |
3244 | return bond_ioctl(net, cmd, argp); | |
3245 | case SIOCADDRT: | |
3246 | case SIOCDELRT: | |
3247 | return routing_ioctl(net, sock, cmd, argp); | |
3248 | case SIOCGSTAMP: | |
3249 | return do_siocgstamp(net, sock, cmd, argp); | |
3250 | case SIOCGSTAMPNS: | |
3251 | return do_siocgstampns(net, sock, cmd, argp); | |
590d4693 BH |
3252 | case SIOCBONDSLAVEINFOQUERY: |
3253 | case SIOCBONDINFOQUERY: | |
a2116ed2 | 3254 | case SIOCSHWTSTAMP: |
fd468c74 | 3255 | case SIOCGHWTSTAMP: |
590d4693 | 3256 | return compat_ifr_data_ioctl(net, cmd, argp); |
6b96018b AB |
3257 | |
3258 | case FIOSETOWN: | |
3259 | case SIOCSPGRP: | |
3260 | case FIOGETOWN: | |
3261 | case SIOCGPGRP: | |
3262 | case SIOCBRADDBR: | |
3263 | case SIOCBRDELBR: | |
3264 | case SIOCGIFVLAN: | |
3265 | case SIOCSIFVLAN: | |
3266 | case SIOCADDDLCI: | |
3267 | case SIOCDELDLCI: | |
3268 | return sock_ioctl(file, cmd, arg); | |
3269 | ||
3270 | case SIOCGIFFLAGS: | |
3271 | case SIOCSIFFLAGS: | |
3272 | case SIOCGIFMETRIC: | |
3273 | case SIOCSIFMETRIC: | |
3274 | case SIOCGIFMTU: | |
3275 | case SIOCSIFMTU: | |
3276 | case SIOCGIFMEM: | |
3277 | case SIOCSIFMEM: | |
3278 | case SIOCGIFHWADDR: | |
3279 | case SIOCSIFHWADDR: | |
3280 | case SIOCADDMULTI: | |
3281 | case SIOCDELMULTI: | |
3282 | case SIOCGIFINDEX: | |
6b96018b AB |
3283 | case SIOCGIFADDR: |
3284 | case SIOCSIFADDR: | |
3285 | case SIOCSIFHWBROADCAST: | |
6b96018b | 3286 | case SIOCDIFADDR: |
6b96018b AB |
3287 | case SIOCGIFBRDADDR: |
3288 | case SIOCSIFBRDADDR: | |
3289 | case SIOCGIFDSTADDR: | |
3290 | case SIOCSIFDSTADDR: | |
3291 | case SIOCGIFNETMASK: | |
3292 | case SIOCSIFNETMASK: | |
3293 | case SIOCSIFPFLAGS: | |
3294 | case SIOCGIFPFLAGS: | |
3295 | case SIOCGIFTXQLEN: | |
3296 | case SIOCSIFTXQLEN: | |
3297 | case SIOCBRADDIF: | |
3298 | case SIOCBRDELIF: | |
9177efd3 AB |
3299 | case SIOCSIFNAME: |
3300 | case SIOCGMIIPHY: | |
3301 | case SIOCGMIIREG: | |
3302 | case SIOCSMIIREG: | |
6b96018b | 3303 | return dev_ifsioc(net, sock, cmd, argp); |
9177efd3 | 3304 | |
6b96018b AB |
3305 | case SIOCSARP: |
3306 | case SIOCGARP: | |
3307 | case SIOCDARP: | |
6b96018b | 3308 | case SIOCATMARK: |
9177efd3 AB |
3309 | return sock_do_ioctl(net, sock, cmd, arg); |
3310 | } | |
3311 | ||
6b96018b AB |
3312 | return -ENOIOCTLCMD; |
3313 | } | |
7a229387 | 3314 | |
95c96174 | 3315 | static long compat_sock_ioctl(struct file *file, unsigned int cmd, |
89bddce5 | 3316 | unsigned long arg) |
89bbfc95 SP |
3317 | { |
3318 | struct socket *sock = file->private_data; | |
3319 | int ret = -ENOIOCTLCMD; | |
87de87d5 DM |
3320 | struct sock *sk; |
3321 | struct net *net; | |
3322 | ||
3323 | sk = sock->sk; | |
3324 | net = sock_net(sk); | |
89bbfc95 SP |
3325 | |
3326 | if (sock->ops->compat_ioctl) | |
3327 | ret = sock->ops->compat_ioctl(sock, cmd, arg); | |
3328 | ||
87de87d5 DM |
3329 | if (ret == -ENOIOCTLCMD && |
3330 | (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST)) | |
3331 | ret = compat_wext_handle_ioctl(net, cmd, arg); | |
3332 | ||
6b96018b AB |
3333 | if (ret == -ENOIOCTLCMD) |
3334 | ret = compat_sock_ioctl_trans(file, sock, cmd, arg); | |
3335 | ||
89bbfc95 SP |
3336 | return ret; |
3337 | } | |
3338 | #endif | |
3339 | ||
ac5a488e SS |
3340 | int kernel_bind(struct socket *sock, struct sockaddr *addr, int addrlen) |
3341 | { | |
3342 | return sock->ops->bind(sock, addr, addrlen); | |
3343 | } | |
c6d409cf | 3344 | EXPORT_SYMBOL(kernel_bind); |
ac5a488e SS |
3345 | |
3346 | int kernel_listen(struct socket *sock, int backlog) | |
3347 | { | |
3348 | return sock->ops->listen(sock, backlog); | |
3349 | } | |
c6d409cf | 3350 | EXPORT_SYMBOL(kernel_listen); |
ac5a488e SS |
3351 | |
3352 | int kernel_accept(struct socket *sock, struct socket **newsock, int flags) | |
3353 | { | |
3354 | struct sock *sk = sock->sk; | |
3355 | int err; | |
3356 | ||
3357 | err = sock_create_lite(sk->sk_family, sk->sk_type, sk->sk_protocol, | |
3358 | newsock); | |
3359 | if (err < 0) | |
3360 | goto done; | |
3361 | ||
3362 | err = sock->ops->accept(sock, *newsock, flags); | |
3363 | if (err < 0) { | |
3364 | sock_release(*newsock); | |
fa8705b0 | 3365 | *newsock = NULL; |
ac5a488e SS |
3366 | goto done; |
3367 | } | |
3368 | ||
3369 | (*newsock)->ops = sock->ops; | |
1b08534e | 3370 | __module_get((*newsock)->ops->owner); |
ac5a488e SS |
3371 | |
3372 | done: | |
3373 | return err; | |
3374 | } | |
c6d409cf | 3375 | EXPORT_SYMBOL(kernel_accept); |
ac5a488e SS |
3376 | |
3377 | int kernel_connect(struct socket *sock, struct sockaddr *addr, int addrlen, | |
4768fbcb | 3378 | int flags) |
ac5a488e SS |
3379 | { |
3380 | return sock->ops->connect(sock, addr, addrlen, flags); | |
3381 | } | |
c6d409cf | 3382 | EXPORT_SYMBOL(kernel_connect); |
ac5a488e SS |
3383 | |
3384 | int kernel_getsockname(struct socket *sock, struct sockaddr *addr, | |
3385 | int *addrlen) | |
3386 | { | |
3387 | return sock->ops->getname(sock, addr, addrlen, 0); | |
3388 | } | |
c6d409cf | 3389 | EXPORT_SYMBOL(kernel_getsockname); |
ac5a488e SS |
3390 | |
3391 | int kernel_getpeername(struct socket *sock, struct sockaddr *addr, | |
3392 | int *addrlen) | |
3393 | { | |
3394 | return sock->ops->getname(sock, addr, addrlen, 1); | |
3395 | } | |
c6d409cf | 3396 | EXPORT_SYMBOL(kernel_getpeername); |
ac5a488e SS |
3397 | |
3398 | int kernel_getsockopt(struct socket *sock, int level, int optname, | |
3399 | char *optval, int *optlen) | |
3400 | { | |
3401 | mm_segment_t oldfs = get_fs(); | |
fb8621bb NK |
3402 | char __user *uoptval; |
3403 | int __user *uoptlen; | |
ac5a488e SS |
3404 | int err; |
3405 | ||
fb8621bb NK |
3406 | uoptval = (char __user __force *) optval; |
3407 | uoptlen = (int __user __force *) optlen; | |
3408 | ||
ac5a488e SS |
3409 | set_fs(KERNEL_DS); |
3410 | if (level == SOL_SOCKET) | |
fb8621bb | 3411 | err = sock_getsockopt(sock, level, optname, uoptval, uoptlen); |
ac5a488e | 3412 | else |
fb8621bb NK |
3413 | err = sock->ops->getsockopt(sock, level, optname, uoptval, |
3414 | uoptlen); | |
ac5a488e SS |
3415 | set_fs(oldfs); |
3416 | return err; | |
3417 | } | |
c6d409cf | 3418 | EXPORT_SYMBOL(kernel_getsockopt); |
ac5a488e SS |
3419 | |
3420 | int kernel_setsockopt(struct socket *sock, int level, int optname, | |
b7058842 | 3421 | char *optval, unsigned int optlen) |
ac5a488e SS |
3422 | { |
3423 | mm_segment_t oldfs = get_fs(); | |
fb8621bb | 3424 | char __user *uoptval; |
ac5a488e SS |
3425 | int err; |
3426 | ||
fb8621bb NK |
3427 | uoptval = (char __user __force *) optval; |
3428 | ||
ac5a488e SS |
3429 | set_fs(KERNEL_DS); |
3430 | if (level == SOL_SOCKET) | |
fb8621bb | 3431 | err = sock_setsockopt(sock, level, optname, uoptval, optlen); |
ac5a488e | 3432 | else |
fb8621bb | 3433 | err = sock->ops->setsockopt(sock, level, optname, uoptval, |
ac5a488e SS |
3434 | optlen); |
3435 | set_fs(oldfs); | |
3436 | return err; | |
3437 | } | |
c6d409cf | 3438 | EXPORT_SYMBOL(kernel_setsockopt); |
ac5a488e SS |
3439 | |
3440 | int kernel_sendpage(struct socket *sock, struct page *page, int offset, | |
3441 | size_t size, int flags) | |
3442 | { | |
3443 | if (sock->ops->sendpage) | |
3444 | return sock->ops->sendpage(sock, page, offset, size, flags); | |
3445 | ||
3446 | return sock_no_sendpage(sock, page, offset, size, flags); | |
3447 | } | |
c6d409cf | 3448 | EXPORT_SYMBOL(kernel_sendpage); |
ac5a488e SS |
3449 | |
3450 | int kernel_sock_ioctl(struct socket *sock, int cmd, unsigned long arg) | |
3451 | { | |
3452 | mm_segment_t oldfs = get_fs(); | |
3453 | int err; | |
3454 | ||
3455 | set_fs(KERNEL_DS); | |
3456 | err = sock->ops->ioctl(sock, cmd, arg); | |
3457 | set_fs(oldfs); | |
3458 | ||
3459 | return err; | |
3460 | } | |
c6d409cf | 3461 | EXPORT_SYMBOL(kernel_sock_ioctl); |
ac5a488e | 3462 | |
91cf45f0 TM |
3463 | int kernel_sock_shutdown(struct socket *sock, enum sock_shutdown_cmd how) |
3464 | { | |
3465 | return sock->ops->shutdown(sock, how); | |
3466 | } | |
91cf45f0 | 3467 | EXPORT_SYMBOL(kernel_sock_shutdown); |