]>
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 | |
8ae5e030 AV |
116 | static ssize_t sock_read_iter(struct kiocb *iocb, struct iov_iter *to); |
117 | static ssize_t sock_write_iter(struct kiocb *iocb, struct iov_iter *from); | |
89bddce5 | 118 | static int sock_mmap(struct file *file, struct vm_area_struct *vma); |
1da177e4 LT |
119 | |
120 | static int sock_close(struct inode *inode, struct file *file); | |
121 | static unsigned int sock_poll(struct file *file, | |
122 | struct poll_table_struct *wait); | |
89bddce5 | 123 | static long sock_ioctl(struct file *file, unsigned int cmd, unsigned long arg); |
89bbfc95 SP |
124 | #ifdef CONFIG_COMPAT |
125 | static long compat_sock_ioctl(struct file *file, | |
89bddce5 | 126 | unsigned int cmd, unsigned long arg); |
89bbfc95 | 127 | #endif |
1da177e4 | 128 | static int sock_fasync(int fd, struct file *filp, int on); |
1da177e4 LT |
129 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
130 | int offset, size_t size, loff_t *ppos, int more); | |
9c55e01c | 131 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 132 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c | 133 | unsigned int flags); |
1da177e4 | 134 | |
1da177e4 LT |
135 | /* |
136 | * Socket files have a set of 'special' operations as well as the generic file ones. These don't appear | |
137 | * in the operation structures but are done directly via the socketcall() multiplexor. | |
138 | */ | |
139 | ||
da7071d7 | 140 | static const struct file_operations socket_file_ops = { |
1da177e4 LT |
141 | .owner = THIS_MODULE, |
142 | .llseek = no_llseek, | |
8ae5e030 AV |
143 | .read_iter = sock_read_iter, |
144 | .write_iter = sock_write_iter, | |
1da177e4 LT |
145 | .poll = sock_poll, |
146 | .unlocked_ioctl = sock_ioctl, | |
89bbfc95 SP |
147 | #ifdef CONFIG_COMPAT |
148 | .compat_ioctl = compat_sock_ioctl, | |
149 | #endif | |
1da177e4 | 150 | .mmap = sock_mmap, |
1da177e4 LT |
151 | .release = sock_close, |
152 | .fasync = sock_fasync, | |
5274f052 JA |
153 | .sendpage = sock_sendpage, |
154 | .splice_write = generic_splice_sendpage, | |
9c55e01c | 155 | .splice_read = sock_splice_read, |
1da177e4 LT |
156 | }; |
157 | ||
158 | /* | |
159 | * The protocol list. Each protocol is registered in here. | |
160 | */ | |
161 | ||
1da177e4 | 162 | static DEFINE_SPINLOCK(net_family_lock); |
190683a9 | 163 | static const struct net_proto_family __rcu *net_families[NPROTO] __read_mostly; |
1da177e4 | 164 | |
1da177e4 LT |
165 | /* |
166 | * Statistics counters of the socket lists | |
167 | */ | |
168 | ||
c6d409cf | 169 | static DEFINE_PER_CPU(int, sockets_in_use); |
1da177e4 LT |
170 | |
171 | /* | |
89bddce5 SH |
172 | * Support routines. |
173 | * Move socket addresses back and forth across the kernel/user | |
174 | * divide and look after the messy bits. | |
1da177e4 LT |
175 | */ |
176 | ||
1da177e4 LT |
177 | /** |
178 | * move_addr_to_kernel - copy a socket address into kernel space | |
179 | * @uaddr: Address in user space | |
180 | * @kaddr: Address in kernel space | |
181 | * @ulen: Length in user space | |
182 | * | |
183 | * The address is copied into kernel space. If the provided address is | |
184 | * too long an error code of -EINVAL is returned. If the copy gives | |
185 | * invalid addresses -EFAULT is returned. On a success 0 is returned. | |
186 | */ | |
187 | ||
43db362d | 188 | int move_addr_to_kernel(void __user *uaddr, int ulen, struct sockaddr_storage *kaddr) |
1da177e4 | 189 | { |
230b1839 | 190 | if (ulen < 0 || ulen > sizeof(struct sockaddr_storage)) |
1da177e4 | 191 | return -EINVAL; |
89bddce5 | 192 | if (ulen == 0) |
1da177e4 | 193 | return 0; |
89bddce5 | 194 | if (copy_from_user(kaddr, uaddr, ulen)) |
1da177e4 | 195 | return -EFAULT; |
3ec3b2fb | 196 | return audit_sockaddr(ulen, kaddr); |
1da177e4 LT |
197 | } |
198 | ||
199 | /** | |
200 | * move_addr_to_user - copy an address to user space | |
201 | * @kaddr: kernel space address | |
202 | * @klen: length of address in kernel | |
203 | * @uaddr: user space address | |
204 | * @ulen: pointer to user length field | |
205 | * | |
206 | * The value pointed to by ulen on entry is the buffer length available. | |
207 | * This is overwritten with the buffer space used. -EINVAL is returned | |
208 | * if an overlong buffer is specified or a negative buffer size. -EFAULT | |
209 | * is returned if either the buffer or the length field are not | |
210 | * accessible. | |
211 | * After copying the data up to the limit the user specifies, the true | |
212 | * length of the data is written over the length limit the user | |
213 | * specified. Zero is returned for a success. | |
214 | */ | |
89bddce5 | 215 | |
43db362d | 216 | static int move_addr_to_user(struct sockaddr_storage *kaddr, int klen, |
11165f14 | 217 | void __user *uaddr, int __user *ulen) |
1da177e4 LT |
218 | { |
219 | int err; | |
220 | int len; | |
221 | ||
68c6beb3 | 222 | BUG_ON(klen > sizeof(struct sockaddr_storage)); |
89bddce5 SH |
223 | err = get_user(len, ulen); |
224 | if (err) | |
1da177e4 | 225 | return err; |
89bddce5 SH |
226 | if (len > klen) |
227 | len = klen; | |
68c6beb3 | 228 | if (len < 0) |
1da177e4 | 229 | return -EINVAL; |
89bddce5 | 230 | if (len) { |
d6fe3945 SG |
231 | if (audit_sockaddr(klen, kaddr)) |
232 | return -ENOMEM; | |
89bddce5 | 233 | if (copy_to_user(uaddr, kaddr, len)) |
1da177e4 LT |
234 | return -EFAULT; |
235 | } | |
236 | /* | |
89bddce5 SH |
237 | * "fromlen shall refer to the value before truncation.." |
238 | * 1003.1g | |
1da177e4 LT |
239 | */ |
240 | return __put_user(klen, ulen); | |
241 | } | |
242 | ||
e18b890b | 243 | static struct kmem_cache *sock_inode_cachep __read_mostly; |
1da177e4 LT |
244 | |
245 | static struct inode *sock_alloc_inode(struct super_block *sb) | |
246 | { | |
247 | struct socket_alloc *ei; | |
eaefd110 | 248 | struct socket_wq *wq; |
89bddce5 | 249 | |
e94b1766 | 250 | ei = kmem_cache_alloc(sock_inode_cachep, GFP_KERNEL); |
1da177e4 LT |
251 | if (!ei) |
252 | return NULL; | |
eaefd110 ED |
253 | wq = kmalloc(sizeof(*wq), GFP_KERNEL); |
254 | if (!wq) { | |
43815482 ED |
255 | kmem_cache_free(sock_inode_cachep, ei); |
256 | return NULL; | |
257 | } | |
eaefd110 ED |
258 | init_waitqueue_head(&wq->wait); |
259 | wq->fasync_list = NULL; | |
574aab1e | 260 | wq->flags = 0; |
eaefd110 | 261 | RCU_INIT_POINTER(ei->socket.wq, wq); |
89bddce5 | 262 | |
1da177e4 LT |
263 | ei->socket.state = SS_UNCONNECTED; |
264 | ei->socket.flags = 0; | |
265 | ei->socket.ops = NULL; | |
266 | ei->socket.sk = NULL; | |
267 | ei->socket.file = NULL; | |
1da177e4 LT |
268 | |
269 | return &ei->vfs_inode; | |
270 | } | |
271 | ||
272 | static void sock_destroy_inode(struct inode *inode) | |
273 | { | |
43815482 | 274 | struct socket_alloc *ei; |
eaefd110 | 275 | struct socket_wq *wq; |
43815482 ED |
276 | |
277 | ei = container_of(inode, struct socket_alloc, vfs_inode); | |
eaefd110 | 278 | wq = rcu_dereference_protected(ei->socket.wq, 1); |
61845220 | 279 | kfree_rcu(wq, rcu); |
43815482 | 280 | kmem_cache_free(sock_inode_cachep, ei); |
1da177e4 LT |
281 | } |
282 | ||
51cc5068 | 283 | static void init_once(void *foo) |
1da177e4 | 284 | { |
89bddce5 | 285 | struct socket_alloc *ei = (struct socket_alloc *)foo; |
1da177e4 | 286 | |
a35afb83 | 287 | inode_init_once(&ei->vfs_inode); |
1da177e4 | 288 | } |
89bddce5 | 289 | |
1da177e4 LT |
290 | static int init_inodecache(void) |
291 | { | |
292 | sock_inode_cachep = kmem_cache_create("sock_inode_cache", | |
89bddce5 SH |
293 | sizeof(struct socket_alloc), |
294 | 0, | |
295 | (SLAB_HWCACHE_ALIGN | | |
296 | SLAB_RECLAIM_ACCOUNT | | |
5d097056 | 297 | SLAB_MEM_SPREAD | SLAB_ACCOUNT), |
20c2df83 | 298 | init_once); |
1da177e4 LT |
299 | if (sock_inode_cachep == NULL) |
300 | return -ENOMEM; | |
301 | return 0; | |
302 | } | |
303 | ||
b87221de | 304 | static const struct super_operations sockfs_ops = { |
c6d409cf ED |
305 | .alloc_inode = sock_alloc_inode, |
306 | .destroy_inode = sock_destroy_inode, | |
307 | .statfs = simple_statfs, | |
1da177e4 LT |
308 | }; |
309 | ||
c23fbb6b ED |
310 | /* |
311 | * sockfs_dname() is called from d_path(). | |
312 | */ | |
313 | static char *sockfs_dname(struct dentry *dentry, char *buffer, int buflen) | |
314 | { | |
315 | return dynamic_dname(dentry, buffer, buflen, "socket:[%lu]", | |
c5ef6035 | 316 | d_inode(dentry)->i_ino); |
c23fbb6b ED |
317 | } |
318 | ||
3ba13d17 | 319 | static const struct dentry_operations sockfs_dentry_operations = { |
c23fbb6b | 320 | .d_dname = sockfs_dname, |
1da177e4 LT |
321 | }; |
322 | ||
c74a1cbb AV |
323 | static struct dentry *sockfs_mount(struct file_system_type *fs_type, |
324 | int flags, const char *dev_name, void *data) | |
325 | { | |
326 | return mount_pseudo(fs_type, "socket:", &sockfs_ops, | |
327 | &sockfs_dentry_operations, SOCKFS_MAGIC); | |
328 | } | |
329 | ||
330 | static struct vfsmount *sock_mnt __read_mostly; | |
331 | ||
332 | static struct file_system_type sock_fs_type = { | |
333 | .name = "sockfs", | |
334 | .mount = sockfs_mount, | |
335 | .kill_sb = kill_anon_super, | |
336 | }; | |
337 | ||
1da177e4 LT |
338 | /* |
339 | * Obtains the first available file descriptor and sets it up for use. | |
340 | * | |
39d8c1b6 DM |
341 | * These functions create file structures and maps them to fd space |
342 | * of the current process. On success it returns file descriptor | |
1da177e4 LT |
343 | * and file struct implicitly stored in sock->file. |
344 | * Note that another thread may close file descriptor before we return | |
345 | * from this function. We use the fact that now we do not refer | |
346 | * to socket after mapping. If one day we will need it, this | |
347 | * function will increment ref. count on file by 1. | |
348 | * | |
349 | * In any case returned fd MAY BE not valid! | |
350 | * This race condition is unavoidable | |
351 | * with shared fd spaces, we cannot solve it inside kernel, | |
352 | * but we take care of internal coherence yet. | |
353 | */ | |
354 | ||
aab174f0 | 355 | struct file *sock_alloc_file(struct socket *sock, int flags, const char *dname) |
1da177e4 | 356 | { |
7cbe66b6 | 357 | struct qstr name = { .name = "" }; |
2c48b9c4 | 358 | struct path path; |
7cbe66b6 | 359 | struct file *file; |
1da177e4 | 360 | |
600e1779 MY |
361 | if (dname) { |
362 | name.name = dname; | |
363 | name.len = strlen(name.name); | |
364 | } else if (sock->sk) { | |
365 | name.name = sock->sk->sk_prot_creator->name; | |
366 | name.len = strlen(name.name); | |
367 | } | |
4b936885 | 368 | path.dentry = d_alloc_pseudo(sock_mnt->mnt_sb, &name); |
28407630 AV |
369 | if (unlikely(!path.dentry)) |
370 | return ERR_PTR(-ENOMEM); | |
2c48b9c4 | 371 | path.mnt = mntget(sock_mnt); |
39d8c1b6 | 372 | |
2c48b9c4 | 373 | d_instantiate(path.dentry, SOCK_INODE(sock)); |
39d8c1b6 | 374 | |
2c48b9c4 | 375 | file = alloc_file(&path, FMODE_READ | FMODE_WRITE, |
ce8d2cdf | 376 | &socket_file_ops); |
b5ffe634 | 377 | if (IS_ERR(file)) { |
cc3808f8 | 378 | /* drop dentry, keep inode */ |
c5ef6035 | 379 | ihold(d_inode(path.dentry)); |
2c48b9c4 | 380 | path_put(&path); |
39b65252 | 381 | return file; |
cc3808f8 AV |
382 | } |
383 | ||
384 | sock->file = file; | |
77d27200 | 385 | file->f_flags = O_RDWR | (flags & O_NONBLOCK); |
39d8c1b6 | 386 | file->private_data = sock; |
28407630 | 387 | return file; |
39d8c1b6 | 388 | } |
56b31d1c | 389 | EXPORT_SYMBOL(sock_alloc_file); |
39d8c1b6 | 390 | |
56b31d1c | 391 | static int sock_map_fd(struct socket *sock, int flags) |
39d8c1b6 DM |
392 | { |
393 | struct file *newfile; | |
28407630 AV |
394 | int fd = get_unused_fd_flags(flags); |
395 | if (unlikely(fd < 0)) | |
396 | return fd; | |
39d8c1b6 | 397 | |
aab174f0 | 398 | newfile = sock_alloc_file(sock, flags, NULL); |
28407630 | 399 | if (likely(!IS_ERR(newfile))) { |
39d8c1b6 | 400 | fd_install(fd, newfile); |
28407630 AV |
401 | return fd; |
402 | } | |
7cbe66b6 | 403 | |
28407630 AV |
404 | put_unused_fd(fd); |
405 | return PTR_ERR(newfile); | |
1da177e4 LT |
406 | } |
407 | ||
406a3c63 | 408 | struct socket *sock_from_file(struct file *file, int *err) |
6cb153ca | 409 | { |
6cb153ca BL |
410 | if (file->f_op == &socket_file_ops) |
411 | return file->private_data; /* set in sock_map_fd */ | |
412 | ||
23bb80d2 ED |
413 | *err = -ENOTSOCK; |
414 | return NULL; | |
6cb153ca | 415 | } |
406a3c63 | 416 | EXPORT_SYMBOL(sock_from_file); |
6cb153ca | 417 | |
1da177e4 | 418 | /** |
c6d409cf | 419 | * sockfd_lookup - Go from a file number to its socket slot |
1da177e4 LT |
420 | * @fd: file handle |
421 | * @err: pointer to an error code return | |
422 | * | |
423 | * The file handle passed in is locked and the socket it is bound | |
424 | * too is returned. If an error occurs the err pointer is overwritten | |
425 | * with a negative errno code and NULL is returned. The function checks | |
426 | * for both invalid handles and passing a handle which is not a socket. | |
427 | * | |
428 | * On a success the socket object pointer is returned. | |
429 | */ | |
430 | ||
431 | struct socket *sockfd_lookup(int fd, int *err) | |
432 | { | |
433 | struct file *file; | |
1da177e4 LT |
434 | struct socket *sock; |
435 | ||
89bddce5 SH |
436 | file = fget(fd); |
437 | if (!file) { | |
1da177e4 LT |
438 | *err = -EBADF; |
439 | return NULL; | |
440 | } | |
89bddce5 | 441 | |
6cb153ca BL |
442 | sock = sock_from_file(file, err); |
443 | if (!sock) | |
1da177e4 | 444 | fput(file); |
6cb153ca BL |
445 | return sock; |
446 | } | |
c6d409cf | 447 | EXPORT_SYMBOL(sockfd_lookup); |
1da177e4 | 448 | |
6cb153ca BL |
449 | static struct socket *sockfd_lookup_light(int fd, int *err, int *fput_needed) |
450 | { | |
00e188ef | 451 | struct fd f = fdget(fd); |
6cb153ca BL |
452 | struct socket *sock; |
453 | ||
3672558c | 454 | *err = -EBADF; |
00e188ef AV |
455 | if (f.file) { |
456 | sock = sock_from_file(f.file, err); | |
457 | if (likely(sock)) { | |
458 | *fput_needed = f.flags; | |
6cb153ca | 459 | return sock; |
00e188ef AV |
460 | } |
461 | fdput(f); | |
1da177e4 | 462 | } |
6cb153ca | 463 | return NULL; |
1da177e4 LT |
464 | } |
465 | ||
600e1779 MY |
466 | #define XATTR_SOCKPROTONAME_SUFFIX "sockprotoname" |
467 | #define XATTR_NAME_SOCKPROTONAME (XATTR_SYSTEM_PREFIX XATTR_SOCKPROTONAME_SUFFIX) | |
468 | #define XATTR_NAME_SOCKPROTONAME_LEN (sizeof(XATTR_NAME_SOCKPROTONAME)-1) | |
ce23e640 | 469 | static ssize_t sockfs_getxattr(struct dentry *dentry, struct inode *inode, |
600e1779 MY |
470 | const char *name, void *value, size_t size) |
471 | { | |
971df15b | 472 | if (!strcmp(name, XATTR_NAME_SOCKPROTONAME)) { |
600e1779 | 473 | if (value) { |
971df15b AG |
474 | if (dentry->d_name.len + 1 > size) |
475 | return -ERANGE; | |
476 | memcpy(value, dentry->d_name.name, dentry->d_name.len + 1); | |
600e1779 | 477 | } |
971df15b | 478 | return dentry->d_name.len + 1; |
600e1779 | 479 | } |
971df15b | 480 | return -EOPNOTSUPP; |
600e1779 MY |
481 | } |
482 | ||
483 | static ssize_t sockfs_listxattr(struct dentry *dentry, char *buffer, | |
484 | size_t size) | |
485 | { | |
486 | ssize_t len; | |
487 | ssize_t used = 0; | |
488 | ||
c5ef6035 | 489 | len = security_inode_listsecurity(d_inode(dentry), buffer, size); |
600e1779 MY |
490 | if (len < 0) |
491 | return len; | |
492 | used += len; | |
493 | if (buffer) { | |
494 | if (size < used) | |
495 | return -ERANGE; | |
496 | buffer += len; | |
497 | } | |
498 | ||
499 | len = (XATTR_NAME_SOCKPROTONAME_LEN + 1); | |
500 | used += len; | |
501 | if (buffer) { | |
502 | if (size < used) | |
503 | return -ERANGE; | |
504 | memcpy(buffer, XATTR_NAME_SOCKPROTONAME, len); | |
505 | buffer += len; | |
506 | } | |
507 | ||
508 | return used; | |
509 | } | |
510 | ||
511 | static const struct inode_operations sockfs_inode_ops = { | |
512 | .getxattr = sockfs_getxattr, | |
513 | .listxattr = sockfs_listxattr, | |
514 | }; | |
515 | ||
1da177e4 LT |
516 | /** |
517 | * sock_alloc - allocate a socket | |
89bddce5 | 518 | * |
1da177e4 LT |
519 | * Allocate a new inode and socket object. The two are bound together |
520 | * and initialised. The socket is then returned. If we are out of inodes | |
521 | * NULL is returned. | |
522 | */ | |
523 | ||
f4a00aac | 524 | struct socket *sock_alloc(void) |
1da177e4 | 525 | { |
89bddce5 SH |
526 | struct inode *inode; |
527 | struct socket *sock; | |
1da177e4 | 528 | |
a209dfc7 | 529 | inode = new_inode_pseudo(sock_mnt->mnt_sb); |
1da177e4 LT |
530 | if (!inode) |
531 | return NULL; | |
532 | ||
533 | sock = SOCKET_I(inode); | |
534 | ||
29a020d3 | 535 | kmemcheck_annotate_bitfield(sock, type); |
85fe4025 | 536 | inode->i_ino = get_next_ino(); |
89bddce5 | 537 | inode->i_mode = S_IFSOCK | S_IRWXUGO; |
8192b0c4 DH |
538 | inode->i_uid = current_fsuid(); |
539 | inode->i_gid = current_fsgid(); | |
600e1779 | 540 | inode->i_op = &sockfs_inode_ops; |
1da177e4 | 541 | |
19e8d69c | 542 | this_cpu_add(sockets_in_use, 1); |
1da177e4 LT |
543 | return sock; |
544 | } | |
f4a00aac | 545 | EXPORT_SYMBOL(sock_alloc); |
1da177e4 | 546 | |
1da177e4 LT |
547 | /** |
548 | * sock_release - close a socket | |
549 | * @sock: socket to close | |
550 | * | |
551 | * The socket is released from the protocol stack if it has a release | |
552 | * callback, and the inode is then released if the socket is bound to | |
89bddce5 | 553 | * an inode not a file. |
1da177e4 | 554 | */ |
89bddce5 | 555 | |
1da177e4 LT |
556 | void sock_release(struct socket *sock) |
557 | { | |
558 | if (sock->ops) { | |
559 | struct module *owner = sock->ops->owner; | |
560 | ||
561 | sock->ops->release(sock); | |
562 | sock->ops = NULL; | |
563 | module_put(owner); | |
564 | } | |
565 | ||
eaefd110 | 566 | if (rcu_dereference_protected(sock->wq, 1)->fasync_list) |
3410f22e | 567 | pr_err("%s: fasync list not empty!\n", __func__); |
1da177e4 | 568 | |
19e8d69c | 569 | this_cpu_sub(sockets_in_use, 1); |
1da177e4 LT |
570 | if (!sock->file) { |
571 | iput(SOCK_INODE(sock)); | |
572 | return; | |
573 | } | |
89bddce5 | 574 | sock->file = NULL; |
1da177e4 | 575 | } |
c6d409cf | 576 | EXPORT_SYMBOL(sock_release); |
1da177e4 | 577 | |
c14ac945 | 578 | void __sock_tx_timestamp(__u16 tsflags, __u8 *tx_flags) |
20d49473 | 579 | { |
140c55d4 ED |
580 | u8 flags = *tx_flags; |
581 | ||
c14ac945 | 582 | if (tsflags & SOF_TIMESTAMPING_TX_HARDWARE) |
140c55d4 ED |
583 | flags |= SKBTX_HW_TSTAMP; |
584 | ||
c14ac945 | 585 | if (tsflags & SOF_TIMESTAMPING_TX_SOFTWARE) |
140c55d4 ED |
586 | flags |= SKBTX_SW_TSTAMP; |
587 | ||
c14ac945 | 588 | if (tsflags & SOF_TIMESTAMPING_TX_SCHED) |
140c55d4 ED |
589 | flags |= SKBTX_SCHED_TSTAMP; |
590 | ||
140c55d4 | 591 | *tx_flags = flags; |
20d49473 | 592 | } |
67cc0d40 | 593 | EXPORT_SYMBOL(__sock_tx_timestamp); |
20d49473 | 594 | |
d8725c86 | 595 | static inline int sock_sendmsg_nosec(struct socket *sock, struct msghdr *msg) |
1da177e4 | 596 | { |
01e97e65 | 597 | int ret = sock->ops->sendmsg(sock, msg, msg_data_left(msg)); |
d8725c86 AV |
598 | BUG_ON(ret == -EIOCBQUEUED); |
599 | return ret; | |
1da177e4 LT |
600 | } |
601 | ||
d8725c86 | 602 | int sock_sendmsg(struct socket *sock, struct msghdr *msg) |
228e548e | 603 | { |
d8725c86 | 604 | int err = security_socket_sendmsg(sock, msg, |
01e97e65 | 605 | msg_data_left(msg)); |
228e548e | 606 | |
d8725c86 | 607 | return err ?: sock_sendmsg_nosec(sock, msg); |
0cf00c6f | 608 | } |
c6d409cf | 609 | EXPORT_SYMBOL(sock_sendmsg); |
1da177e4 LT |
610 | |
611 | int kernel_sendmsg(struct socket *sock, struct msghdr *msg, | |
612 | struct kvec *vec, size_t num, size_t size) | |
613 | { | |
6aa24814 | 614 | iov_iter_kvec(&msg->msg_iter, WRITE | ITER_KVEC, vec, num, size); |
d8725c86 | 615 | return sock_sendmsg(sock, msg); |
1da177e4 | 616 | } |
c6d409cf | 617 | EXPORT_SYMBOL(kernel_sendmsg); |
1da177e4 | 618 | |
92f37fd2 ED |
619 | /* |
620 | * called from sock_recv_timestamp() if sock_flag(sk, SOCK_RCVTSTAMP) | |
621 | */ | |
622 | void __sock_recv_timestamp(struct msghdr *msg, struct sock *sk, | |
623 | struct sk_buff *skb) | |
624 | { | |
20d49473 | 625 | int need_software_tstamp = sock_flag(sk, SOCK_RCVTSTAMP); |
f24b9be5 | 626 | struct scm_timestamping tss; |
20d49473 PO |
627 | int empty = 1; |
628 | struct skb_shared_hwtstamps *shhwtstamps = | |
629 | skb_hwtstamps(skb); | |
630 | ||
631 | /* Race occurred between timestamp enabling and packet | |
632 | receiving. Fill in the current time for now. */ | |
633 | if (need_software_tstamp && skb->tstamp.tv64 == 0) | |
634 | __net_timestamp(skb); | |
635 | ||
636 | if (need_software_tstamp) { | |
637 | if (!sock_flag(sk, SOCK_RCVTSTAMPNS)) { | |
638 | struct timeval tv; | |
639 | skb_get_timestamp(skb, &tv); | |
640 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMP, | |
641 | sizeof(tv), &tv); | |
642 | } else { | |
f24b9be5 WB |
643 | struct timespec ts; |
644 | skb_get_timestampns(skb, &ts); | |
20d49473 | 645 | put_cmsg(msg, SOL_SOCKET, SCM_TIMESTAMPNS, |
f24b9be5 | 646 | sizeof(ts), &ts); |
20d49473 PO |
647 | } |
648 | } | |
649 | ||
f24b9be5 | 650 | memset(&tss, 0, sizeof(tss)); |
c199105d | 651 | if ((sk->sk_tsflags & SOF_TIMESTAMPING_SOFTWARE) && |
f24b9be5 | 652 | ktime_to_timespec_cond(skb->tstamp, tss.ts + 0)) |
20d49473 | 653 | empty = 0; |
4d276eb6 | 654 | if (shhwtstamps && |
b9f40e21 | 655 | (sk->sk_tsflags & SOF_TIMESTAMPING_RAW_HARDWARE) && |
f24b9be5 | 656 | ktime_to_timespec_cond(shhwtstamps->hwtstamp, tss.ts + 2)) |
4d276eb6 | 657 | empty = 0; |
20d49473 PO |
658 | if (!empty) |
659 | put_cmsg(msg, SOL_SOCKET, | |
f24b9be5 | 660 | SCM_TIMESTAMPING, sizeof(tss), &tss); |
92f37fd2 | 661 | } |
7c81fd8b ACM |
662 | EXPORT_SYMBOL_GPL(__sock_recv_timestamp); |
663 | ||
6e3e939f JB |
664 | void __sock_recv_wifi_status(struct msghdr *msg, struct sock *sk, |
665 | struct sk_buff *skb) | |
666 | { | |
667 | int ack; | |
668 | ||
669 | if (!sock_flag(sk, SOCK_WIFI_STATUS)) | |
670 | return; | |
671 | if (!skb->wifi_acked_valid) | |
672 | return; | |
673 | ||
674 | ack = skb->wifi_acked; | |
675 | ||
676 | put_cmsg(msg, SOL_SOCKET, SCM_WIFI_STATUS, sizeof(ack), &ack); | |
677 | } | |
678 | EXPORT_SYMBOL_GPL(__sock_recv_wifi_status); | |
679 | ||
11165f14 | 680 | static inline void sock_recv_drops(struct msghdr *msg, struct sock *sk, |
681 | struct sk_buff *skb) | |
3b885787 | 682 | { |
744d5a3e | 683 | if (sock_flag(sk, SOCK_RXQ_OVFL) && skb && SOCK_SKB_CB(skb)->dropcount) |
3b885787 | 684 | put_cmsg(msg, SOL_SOCKET, SO_RXQ_OVFL, |
744d5a3e | 685 | sizeof(__u32), &SOCK_SKB_CB(skb)->dropcount); |
3b885787 NH |
686 | } |
687 | ||
767dd033 | 688 | void __sock_recv_ts_and_drops(struct msghdr *msg, struct sock *sk, |
3b885787 NH |
689 | struct sk_buff *skb) |
690 | { | |
691 | sock_recv_timestamp(msg, sk, skb); | |
692 | sock_recv_drops(msg, sk, skb); | |
693 | } | |
767dd033 | 694 | EXPORT_SYMBOL_GPL(__sock_recv_ts_and_drops); |
3b885787 | 695 | |
1b784140 | 696 | static inline int sock_recvmsg_nosec(struct socket *sock, struct msghdr *msg, |
2da62906 | 697 | int flags) |
1da177e4 | 698 | { |
2da62906 | 699 | return sock->ops->recvmsg(sock, msg, msg_data_left(msg), flags); |
1da177e4 LT |
700 | } |
701 | ||
2da62906 | 702 | int sock_recvmsg(struct socket *sock, struct msghdr *msg, int flags) |
a2e27255 | 703 | { |
2da62906 | 704 | int err = security_socket_recvmsg(sock, msg, msg_data_left(msg), flags); |
a2e27255 | 705 | |
2da62906 | 706 | return err ?: sock_recvmsg_nosec(sock, msg, flags); |
1da177e4 | 707 | } |
c6d409cf | 708 | EXPORT_SYMBOL(sock_recvmsg); |
1da177e4 | 709 | |
c1249c0a ML |
710 | /** |
711 | * kernel_recvmsg - Receive a message from a socket (kernel space) | |
712 | * @sock: The socket to receive the message from | |
713 | * @msg: Received message | |
714 | * @vec: Input s/g array for message data | |
715 | * @num: Size of input s/g array | |
716 | * @size: Number of bytes to read | |
717 | * @flags: Message flags (MSG_DONTWAIT, etc...) | |
718 | * | |
719 | * On return the msg structure contains the scatter/gather array passed in the | |
720 | * vec argument. The array is modified so that it consists of the unfilled | |
721 | * portion of the original array. | |
722 | * | |
723 | * The returned value is the total number of bytes received, or an error. | |
724 | */ | |
89bddce5 SH |
725 | int kernel_recvmsg(struct socket *sock, struct msghdr *msg, |
726 | struct kvec *vec, size_t num, size_t size, int flags) | |
1da177e4 LT |
727 | { |
728 | mm_segment_t oldfs = get_fs(); | |
729 | int result; | |
730 | ||
6aa24814 | 731 | iov_iter_kvec(&msg->msg_iter, READ | ITER_KVEC, vec, num, size); |
1da177e4 | 732 | set_fs(KERNEL_DS); |
2da62906 | 733 | result = sock_recvmsg(sock, msg, flags); |
1da177e4 LT |
734 | set_fs(oldfs); |
735 | return result; | |
736 | } | |
c6d409cf | 737 | EXPORT_SYMBOL(kernel_recvmsg); |
1da177e4 | 738 | |
ce1d4d3e CH |
739 | static ssize_t sock_sendpage(struct file *file, struct page *page, |
740 | int offset, size_t size, loff_t *ppos, int more) | |
1da177e4 | 741 | { |
1da177e4 LT |
742 | struct socket *sock; |
743 | int flags; | |
744 | ||
ce1d4d3e CH |
745 | sock = file->private_data; |
746 | ||
35f9c09f ED |
747 | flags = (file->f_flags & O_NONBLOCK) ? MSG_DONTWAIT : 0; |
748 | /* more is a combination of MSG_MORE and MSG_SENDPAGE_NOTLAST */ | |
749 | flags |= more; | |
ce1d4d3e | 750 | |
e6949583 | 751 | return kernel_sendpage(sock, page, offset, size, flags); |
ce1d4d3e | 752 | } |
1da177e4 | 753 | |
9c55e01c | 754 | static ssize_t sock_splice_read(struct file *file, loff_t *ppos, |
c6d409cf | 755 | struct pipe_inode_info *pipe, size_t len, |
9c55e01c JA |
756 | unsigned int flags) |
757 | { | |
758 | struct socket *sock = file->private_data; | |
759 | ||
997b37da RDC |
760 | if (unlikely(!sock->ops->splice_read)) |
761 | return -EINVAL; | |
762 | ||
9c55e01c JA |
763 | return sock->ops->splice_read(sock, ppos, pipe, len, flags); |
764 | } | |
765 | ||
8ae5e030 | 766 | static ssize_t sock_read_iter(struct kiocb *iocb, struct iov_iter *to) |
ce1d4d3e | 767 | { |
6d652330 AV |
768 | struct file *file = iocb->ki_filp; |
769 | struct socket *sock = file->private_data; | |
0345f931 | 770 | struct msghdr msg = {.msg_iter = *to, |
771 | .msg_iocb = iocb}; | |
8ae5e030 | 772 | ssize_t res; |
ce1d4d3e | 773 | |
8ae5e030 AV |
774 | if (file->f_flags & O_NONBLOCK) |
775 | msg.msg_flags = MSG_DONTWAIT; | |
776 | ||
777 | if (iocb->ki_pos != 0) | |
1da177e4 | 778 | return -ESPIPE; |
027445c3 | 779 | |
66ee59af | 780 | if (!iov_iter_count(to)) /* Match SYS5 behaviour */ |
1da177e4 LT |
781 | return 0; |
782 | ||
2da62906 | 783 | res = sock_recvmsg(sock, &msg, msg.msg_flags); |
8ae5e030 AV |
784 | *to = msg.msg_iter; |
785 | return res; | |
1da177e4 LT |
786 | } |
787 | ||
8ae5e030 | 788 | static ssize_t sock_write_iter(struct kiocb *iocb, struct iov_iter *from) |
ce1d4d3e | 789 | { |
6d652330 AV |
790 | struct file *file = iocb->ki_filp; |
791 | struct socket *sock = file->private_data; | |
0345f931 | 792 | struct msghdr msg = {.msg_iter = *from, |
793 | .msg_iocb = iocb}; | |
8ae5e030 | 794 | ssize_t res; |
1da177e4 | 795 | |
8ae5e030 | 796 | if (iocb->ki_pos != 0) |
ce1d4d3e | 797 | return -ESPIPE; |
027445c3 | 798 | |
8ae5e030 AV |
799 | if (file->f_flags & O_NONBLOCK) |
800 | msg.msg_flags = MSG_DONTWAIT; | |
801 | ||
6d652330 AV |
802 | if (sock->type == SOCK_SEQPACKET) |
803 | msg.msg_flags |= MSG_EOR; | |
804 | ||
d8725c86 | 805 | res = sock_sendmsg(sock, &msg); |
8ae5e030 AV |
806 | *from = msg.msg_iter; |
807 | return res; | |
1da177e4 LT |
808 | } |
809 | ||
1da177e4 LT |
810 | /* |
811 | * Atomic setting of ioctl hooks to avoid race | |
812 | * with module unload. | |
813 | */ | |
814 | ||
4a3e2f71 | 815 | static DEFINE_MUTEX(br_ioctl_mutex); |
c6d409cf | 816 | static int (*br_ioctl_hook) (struct net *, unsigned int cmd, void __user *arg); |
1da177e4 | 817 | |
881d966b | 818 | void brioctl_set(int (*hook) (struct net *, unsigned int, void __user *)) |
1da177e4 | 819 | { |
4a3e2f71 | 820 | mutex_lock(&br_ioctl_mutex); |
1da177e4 | 821 | br_ioctl_hook = hook; |
4a3e2f71 | 822 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
823 | } |
824 | EXPORT_SYMBOL(brioctl_set); | |
825 | ||
4a3e2f71 | 826 | static DEFINE_MUTEX(vlan_ioctl_mutex); |
881d966b | 827 | static int (*vlan_ioctl_hook) (struct net *, void __user *arg); |
1da177e4 | 828 | |
881d966b | 829 | void vlan_ioctl_set(int (*hook) (struct net *, void __user *)) |
1da177e4 | 830 | { |
4a3e2f71 | 831 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 832 | vlan_ioctl_hook = hook; |
4a3e2f71 | 833 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 LT |
834 | } |
835 | EXPORT_SYMBOL(vlan_ioctl_set); | |
836 | ||
4a3e2f71 | 837 | static DEFINE_MUTEX(dlci_ioctl_mutex); |
89bddce5 | 838 | static int (*dlci_ioctl_hook) (unsigned int, void __user *); |
1da177e4 | 839 | |
89bddce5 | 840 | void dlci_ioctl_set(int (*hook) (unsigned int, void __user *)) |
1da177e4 | 841 | { |
4a3e2f71 | 842 | mutex_lock(&dlci_ioctl_mutex); |
1da177e4 | 843 | dlci_ioctl_hook = hook; |
4a3e2f71 | 844 | mutex_unlock(&dlci_ioctl_mutex); |
1da177e4 LT |
845 | } |
846 | EXPORT_SYMBOL(dlci_ioctl_set); | |
847 | ||
6b96018b AB |
848 | static long sock_do_ioctl(struct net *net, struct socket *sock, |
849 | unsigned int cmd, unsigned long arg) | |
850 | { | |
851 | int err; | |
852 | void __user *argp = (void __user *)arg; | |
853 | ||
854 | err = sock->ops->ioctl(sock, cmd, arg); | |
855 | ||
856 | /* | |
857 | * If this ioctl is unknown try to hand it down | |
858 | * to the NIC driver. | |
859 | */ | |
860 | if (err == -ENOIOCTLCMD) | |
861 | err = dev_ioctl(net, cmd, argp); | |
862 | ||
863 | return err; | |
864 | } | |
865 | ||
1da177e4 LT |
866 | /* |
867 | * With an ioctl, arg may well be a user mode pointer, but we don't know | |
868 | * what to do with it - that's up to the protocol still. | |
869 | */ | |
870 | ||
871 | static long sock_ioctl(struct file *file, unsigned cmd, unsigned long arg) | |
872 | { | |
873 | struct socket *sock; | |
881d966b | 874 | struct sock *sk; |
1da177e4 LT |
875 | void __user *argp = (void __user *)arg; |
876 | int pid, err; | |
881d966b | 877 | struct net *net; |
1da177e4 | 878 | |
b69aee04 | 879 | sock = file->private_data; |
881d966b | 880 | sk = sock->sk; |
3b1e0a65 | 881 | net = sock_net(sk); |
1da177e4 | 882 | if (cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15)) { |
881d966b | 883 | err = dev_ioctl(net, cmd, argp); |
1da177e4 | 884 | } else |
3d23e349 | 885 | #ifdef CONFIG_WEXT_CORE |
1da177e4 | 886 | if (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST) { |
881d966b | 887 | err = dev_ioctl(net, cmd, argp); |
1da177e4 | 888 | } else |
3d23e349 | 889 | #endif |
89bddce5 | 890 | switch (cmd) { |
1da177e4 LT |
891 | case FIOSETOWN: |
892 | case SIOCSPGRP: | |
893 | err = -EFAULT; | |
894 | if (get_user(pid, (int __user *)argp)) | |
895 | break; | |
e0b93edd JL |
896 | f_setown(sock->file, pid, 1); |
897 | err = 0; | |
1da177e4 LT |
898 | break; |
899 | case FIOGETOWN: | |
900 | case SIOCGPGRP: | |
609d7fa9 | 901 | err = put_user(f_getown(sock->file), |
89bddce5 | 902 | (int __user *)argp); |
1da177e4 LT |
903 | break; |
904 | case SIOCGIFBR: | |
905 | case SIOCSIFBR: | |
906 | case SIOCBRADDBR: | |
907 | case SIOCBRDELBR: | |
908 | err = -ENOPKG; | |
909 | if (!br_ioctl_hook) | |
910 | request_module("bridge"); | |
911 | ||
4a3e2f71 | 912 | mutex_lock(&br_ioctl_mutex); |
89bddce5 | 913 | if (br_ioctl_hook) |
881d966b | 914 | err = br_ioctl_hook(net, cmd, argp); |
4a3e2f71 | 915 | mutex_unlock(&br_ioctl_mutex); |
1da177e4 LT |
916 | break; |
917 | case SIOCGIFVLAN: | |
918 | case SIOCSIFVLAN: | |
919 | err = -ENOPKG; | |
920 | if (!vlan_ioctl_hook) | |
921 | request_module("8021q"); | |
922 | ||
4a3e2f71 | 923 | mutex_lock(&vlan_ioctl_mutex); |
1da177e4 | 924 | if (vlan_ioctl_hook) |
881d966b | 925 | err = vlan_ioctl_hook(net, argp); |
4a3e2f71 | 926 | mutex_unlock(&vlan_ioctl_mutex); |
1da177e4 | 927 | break; |
1da177e4 LT |
928 | case SIOCADDDLCI: |
929 | case SIOCDELDLCI: | |
930 | err = -ENOPKG; | |
931 | if (!dlci_ioctl_hook) | |
932 | request_module("dlci"); | |
933 | ||
7512cbf6 PE |
934 | mutex_lock(&dlci_ioctl_mutex); |
935 | if (dlci_ioctl_hook) | |
1da177e4 | 936 | err = dlci_ioctl_hook(cmd, argp); |
7512cbf6 | 937 | mutex_unlock(&dlci_ioctl_mutex); |
1da177e4 LT |
938 | break; |
939 | default: | |
6b96018b | 940 | err = sock_do_ioctl(net, sock, cmd, arg); |
1da177e4 | 941 | break; |
89bddce5 | 942 | } |
1da177e4 LT |
943 | return err; |
944 | } | |
945 | ||
946 | int sock_create_lite(int family, int type, int protocol, struct socket **res) | |
947 | { | |
948 | int err; | |
949 | struct socket *sock = NULL; | |
89bddce5 | 950 | |
1da177e4 LT |
951 | err = security_socket_create(family, type, protocol, 1); |
952 | if (err) | |
953 | goto out; | |
954 | ||
955 | sock = sock_alloc(); | |
956 | if (!sock) { | |
957 | err = -ENOMEM; | |
958 | goto out; | |
959 | } | |
960 | ||
1da177e4 | 961 | sock->type = type; |
7420ed23 VY |
962 | err = security_socket_post_create(sock, family, type, protocol, 1); |
963 | if (err) | |
964 | goto out_release; | |
965 | ||
1da177e4 LT |
966 | out: |
967 | *res = sock; | |
968 | return err; | |
7420ed23 VY |
969 | out_release: |
970 | sock_release(sock); | |
971 | sock = NULL; | |
972 | goto out; | |
1da177e4 | 973 | } |
c6d409cf | 974 | EXPORT_SYMBOL(sock_create_lite); |
1da177e4 LT |
975 | |
976 | /* No kernel lock held - perfect */ | |
89bddce5 | 977 | static unsigned int sock_poll(struct file *file, poll_table *wait) |
1da177e4 | 978 | { |
cbf55001 | 979 | unsigned int busy_flag = 0; |
1da177e4 LT |
980 | struct socket *sock; |
981 | ||
982 | /* | |
89bddce5 | 983 | * We can't return errors to poll, so it's either yes or no. |
1da177e4 | 984 | */ |
b69aee04 | 985 | sock = file->private_data; |
2d48d67f | 986 | |
cbf55001 | 987 | if (sk_can_busy_loop(sock->sk)) { |
2d48d67f | 988 | /* this socket can poll_ll so tell the system call */ |
cbf55001 | 989 | busy_flag = POLL_BUSY_LOOP; |
2d48d67f ET |
990 | |
991 | /* once, only if requested by syscall */ | |
cbf55001 ET |
992 | if (wait && (wait->_key & POLL_BUSY_LOOP)) |
993 | sk_busy_loop(sock->sk, 1); | |
2d48d67f ET |
994 | } |
995 | ||
cbf55001 | 996 | return busy_flag | sock->ops->poll(file, sock, wait); |
1da177e4 LT |
997 | } |
998 | ||
89bddce5 | 999 | static int sock_mmap(struct file *file, struct vm_area_struct *vma) |
1da177e4 | 1000 | { |
b69aee04 | 1001 | struct socket *sock = file->private_data; |
1da177e4 LT |
1002 | |
1003 | return sock->ops->mmap(file, sock, vma); | |
1004 | } | |
1005 | ||
20380731 | 1006 | static int sock_close(struct inode *inode, struct file *filp) |
1da177e4 | 1007 | { |
1da177e4 LT |
1008 | sock_release(SOCKET_I(inode)); |
1009 | return 0; | |
1010 | } | |
1011 | ||
1012 | /* | |
1013 | * Update the socket async list | |
1014 | * | |
1015 | * Fasync_list locking strategy. | |
1016 | * | |
1017 | * 1. fasync_list is modified only under process context socket lock | |
1018 | * i.e. under semaphore. | |
1019 | * 2. fasync_list is used under read_lock(&sk->sk_callback_lock) | |
989a2979 | 1020 | * or under socket lock |
1da177e4 LT |
1021 | */ |
1022 | ||
1023 | static int sock_fasync(int fd, struct file *filp, int on) | |
1024 | { | |
989a2979 ED |
1025 | struct socket *sock = filp->private_data; |
1026 | struct sock *sk = sock->sk; | |
eaefd110 | 1027 | struct socket_wq *wq; |
1da177e4 | 1028 | |
989a2979 | 1029 | if (sk == NULL) |
1da177e4 | 1030 | return -EINVAL; |
1da177e4 LT |
1031 | |
1032 | lock_sock(sk); | |
1e1d04e6 | 1033 | wq = rcu_dereference_protected(sock->wq, lockdep_sock_is_held(sk)); |
eaefd110 | 1034 | fasync_helper(fd, filp, on, &wq->fasync_list); |
1da177e4 | 1035 | |
eaefd110 | 1036 | if (!wq->fasync_list) |
989a2979 ED |
1037 | sock_reset_flag(sk, SOCK_FASYNC); |
1038 | else | |
bcdce719 | 1039 | sock_set_flag(sk, SOCK_FASYNC); |
1da177e4 | 1040 | |
989a2979 | 1041 | release_sock(sk); |
1da177e4 LT |
1042 | return 0; |
1043 | } | |
1044 | ||
ceb5d58b | 1045 | /* This function may be called only under rcu_lock */ |
1da177e4 | 1046 | |
ceb5d58b | 1047 | int sock_wake_async(struct socket_wq *wq, int how, int band) |
1da177e4 | 1048 | { |
ceb5d58b | 1049 | if (!wq || !wq->fasync_list) |
1da177e4 | 1050 | return -1; |
ceb5d58b | 1051 | |
89bddce5 | 1052 | switch (how) { |
8d8ad9d7 | 1053 | case SOCK_WAKE_WAITD: |
ceb5d58b | 1054 | if (test_bit(SOCKWQ_ASYNC_WAITDATA, &wq->flags)) |
1da177e4 LT |
1055 | break; |
1056 | goto call_kill; | |
8d8ad9d7 | 1057 | case SOCK_WAKE_SPACE: |
ceb5d58b | 1058 | if (!test_and_clear_bit(SOCKWQ_ASYNC_NOSPACE, &wq->flags)) |
1da177e4 LT |
1059 | break; |
1060 | /* fall through */ | |
8d8ad9d7 | 1061 | case SOCK_WAKE_IO: |
89bddce5 | 1062 | call_kill: |
43815482 | 1063 | kill_fasync(&wq->fasync_list, SIGIO, band); |
1da177e4 | 1064 | break; |
8d8ad9d7 | 1065 | case SOCK_WAKE_URG: |
43815482 | 1066 | kill_fasync(&wq->fasync_list, SIGURG, band); |
1da177e4 | 1067 | } |
ceb5d58b | 1068 | |
1da177e4 LT |
1069 | return 0; |
1070 | } | |
c6d409cf | 1071 | EXPORT_SYMBOL(sock_wake_async); |
1da177e4 | 1072 | |
721db93a | 1073 | int __sock_create(struct net *net, int family, int type, int protocol, |
89bddce5 | 1074 | struct socket **res, int kern) |
1da177e4 LT |
1075 | { |
1076 | int err; | |
1077 | struct socket *sock; | |
55737fda | 1078 | const struct net_proto_family *pf; |
1da177e4 LT |
1079 | |
1080 | /* | |
89bddce5 | 1081 | * Check protocol is in range |
1da177e4 LT |
1082 | */ |
1083 | if (family < 0 || family >= NPROTO) | |
1084 | return -EAFNOSUPPORT; | |
1085 | if (type < 0 || type >= SOCK_MAX) | |
1086 | return -EINVAL; | |
1087 | ||
1088 | /* Compatibility. | |
1089 | ||
1090 | This uglymoron is moved from INET layer to here to avoid | |
1091 | deadlock in module load. | |
1092 | */ | |
1093 | if (family == PF_INET && type == SOCK_PACKET) { | |
f3c98690 | 1094 | pr_info_once("%s uses obsolete (PF_INET,SOCK_PACKET)\n", |
1095 | current->comm); | |
1da177e4 LT |
1096 | family = PF_PACKET; |
1097 | } | |
1098 | ||
1099 | err = security_socket_create(family, type, protocol, kern); | |
1100 | if (err) | |
1101 | return err; | |
89bddce5 | 1102 | |
55737fda SH |
1103 | /* |
1104 | * Allocate the socket and allow the family to set things up. if | |
1105 | * the protocol is 0, the family is instructed to select an appropriate | |
1106 | * default. | |
1107 | */ | |
1108 | sock = sock_alloc(); | |
1109 | if (!sock) { | |
e87cc472 | 1110 | net_warn_ratelimited("socket: no more sockets\n"); |
55737fda SH |
1111 | return -ENFILE; /* Not exactly a match, but its the |
1112 | closest posix thing */ | |
1113 | } | |
1114 | ||
1115 | sock->type = type; | |
1116 | ||
95a5afca | 1117 | #ifdef CONFIG_MODULES |
89bddce5 SH |
1118 | /* Attempt to load a protocol module if the find failed. |
1119 | * | |
1120 | * 12/09/1996 Marcin: But! this makes REALLY only sense, if the user | |
1da177e4 LT |
1121 | * requested real, full-featured networking support upon configuration. |
1122 | * Otherwise module support will break! | |
1123 | */ | |
190683a9 | 1124 | if (rcu_access_pointer(net_families[family]) == NULL) |
89bddce5 | 1125 | request_module("net-pf-%d", family); |
1da177e4 LT |
1126 | #endif |
1127 | ||
55737fda SH |
1128 | rcu_read_lock(); |
1129 | pf = rcu_dereference(net_families[family]); | |
1130 | err = -EAFNOSUPPORT; | |
1131 | if (!pf) | |
1132 | goto out_release; | |
1da177e4 LT |
1133 | |
1134 | /* | |
1135 | * We will call the ->create function, that possibly is in a loadable | |
1136 | * module, so we have to bump that loadable module refcnt first. | |
1137 | */ | |
55737fda | 1138 | if (!try_module_get(pf->owner)) |
1da177e4 LT |
1139 | goto out_release; |
1140 | ||
55737fda SH |
1141 | /* Now protected by module ref count */ |
1142 | rcu_read_unlock(); | |
1143 | ||
3f378b68 | 1144 | err = pf->create(net, sock, protocol, kern); |
55737fda | 1145 | if (err < 0) |
1da177e4 | 1146 | goto out_module_put; |
a79af59e | 1147 | |
1da177e4 LT |
1148 | /* |
1149 | * Now to bump the refcnt of the [loadable] module that owns this | |
1150 | * socket at sock_release time we decrement its refcnt. | |
1151 | */ | |
55737fda SH |
1152 | if (!try_module_get(sock->ops->owner)) |
1153 | goto out_module_busy; | |
1154 | ||
1da177e4 LT |
1155 | /* |
1156 | * Now that we're done with the ->create function, the [loadable] | |
1157 | * module can have its refcnt decremented | |
1158 | */ | |
55737fda | 1159 | module_put(pf->owner); |
7420ed23 VY |
1160 | err = security_socket_post_create(sock, family, type, protocol, kern); |
1161 | if (err) | |
3b185525 | 1162 | goto out_sock_release; |
55737fda | 1163 | *res = sock; |
1da177e4 | 1164 | |
55737fda SH |
1165 | return 0; |
1166 | ||
1167 | out_module_busy: | |
1168 | err = -EAFNOSUPPORT; | |
1da177e4 | 1169 | out_module_put: |
55737fda SH |
1170 | sock->ops = NULL; |
1171 | module_put(pf->owner); | |
1172 | out_sock_release: | |
1da177e4 | 1173 | sock_release(sock); |
55737fda SH |
1174 | return err; |
1175 | ||
1176 | out_release: | |
1177 | rcu_read_unlock(); | |
1178 | goto out_sock_release; | |
1da177e4 | 1179 | } |
721db93a | 1180 | EXPORT_SYMBOL(__sock_create); |
1da177e4 LT |
1181 | |
1182 | int sock_create(int family, int type, int protocol, struct socket **res) | |
1183 | { | |
1b8d7ae4 | 1184 | return __sock_create(current->nsproxy->net_ns, family, type, protocol, res, 0); |
1da177e4 | 1185 | } |
c6d409cf | 1186 | EXPORT_SYMBOL(sock_create); |
1da177e4 | 1187 | |
eeb1bd5c | 1188 | int sock_create_kern(struct net *net, int family, int type, int protocol, struct socket **res) |
1da177e4 | 1189 | { |
eeb1bd5c | 1190 | return __sock_create(net, family, type, protocol, res, 1); |
1da177e4 | 1191 | } |
c6d409cf | 1192 | EXPORT_SYMBOL(sock_create_kern); |
1da177e4 | 1193 | |
3e0fa65f | 1194 | SYSCALL_DEFINE3(socket, int, family, int, type, int, protocol) |
1da177e4 LT |
1195 | { |
1196 | int retval; | |
1197 | struct socket *sock; | |
a677a039 UD |
1198 | int flags; |
1199 | ||
e38b36f3 UD |
1200 | /* Check the SOCK_* constants for consistency. */ |
1201 | BUILD_BUG_ON(SOCK_CLOEXEC != O_CLOEXEC); | |
1202 | BUILD_BUG_ON((SOCK_MAX | SOCK_TYPE_MASK) != SOCK_TYPE_MASK); | |
1203 | BUILD_BUG_ON(SOCK_CLOEXEC & SOCK_TYPE_MASK); | |
1204 | BUILD_BUG_ON(SOCK_NONBLOCK & SOCK_TYPE_MASK); | |
1205 | ||
a677a039 | 1206 | flags = type & ~SOCK_TYPE_MASK; |
77d27200 | 1207 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1208 | return -EINVAL; |
1209 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1210 | |
aaca0bdc UD |
1211 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1212 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1213 | ||
1da177e4 LT |
1214 | retval = sock_create(family, type, protocol, &sock); |
1215 | if (retval < 0) | |
1216 | goto out; | |
1217 | ||
77d27200 | 1218 | retval = sock_map_fd(sock, flags & (O_CLOEXEC | O_NONBLOCK)); |
1da177e4 LT |
1219 | if (retval < 0) |
1220 | goto out_release; | |
1221 | ||
1222 | out: | |
1223 | /* It may be already another descriptor 8) Not kernel problem. */ | |
1224 | return retval; | |
1225 | ||
1226 | out_release: | |
1227 | sock_release(sock); | |
1228 | return retval; | |
1229 | } | |
1230 | ||
1231 | /* | |
1232 | * Create a pair of connected sockets. | |
1233 | */ | |
1234 | ||
3e0fa65f HC |
1235 | SYSCALL_DEFINE4(socketpair, int, family, int, type, int, protocol, |
1236 | int __user *, usockvec) | |
1da177e4 LT |
1237 | { |
1238 | struct socket *sock1, *sock2; | |
1239 | int fd1, fd2, err; | |
db349509 | 1240 | struct file *newfile1, *newfile2; |
a677a039 UD |
1241 | int flags; |
1242 | ||
1243 | flags = type & ~SOCK_TYPE_MASK; | |
77d27200 | 1244 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
a677a039 UD |
1245 | return -EINVAL; |
1246 | type &= SOCK_TYPE_MASK; | |
1da177e4 | 1247 | |
aaca0bdc UD |
1248 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) |
1249 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1250 | ||
1da177e4 LT |
1251 | /* |
1252 | * Obtain the first socket and check if the underlying protocol | |
1253 | * supports the socketpair call. | |
1254 | */ | |
1255 | ||
1256 | err = sock_create(family, type, protocol, &sock1); | |
1257 | if (err < 0) | |
1258 | goto out; | |
1259 | ||
1260 | err = sock_create(family, type, protocol, &sock2); | |
1261 | if (err < 0) | |
1262 | goto out_release_1; | |
1263 | ||
1264 | err = sock1->ops->socketpair(sock1, sock2); | |
89bddce5 | 1265 | if (err < 0) |
1da177e4 LT |
1266 | goto out_release_both; |
1267 | ||
28407630 | 1268 | fd1 = get_unused_fd_flags(flags); |
bf3c23d1 DM |
1269 | if (unlikely(fd1 < 0)) { |
1270 | err = fd1; | |
db349509 | 1271 | goto out_release_both; |
bf3c23d1 | 1272 | } |
d73aa286 | 1273 | |
28407630 | 1274 | fd2 = get_unused_fd_flags(flags); |
198de4d7 AV |
1275 | if (unlikely(fd2 < 0)) { |
1276 | err = fd2; | |
d73aa286 | 1277 | goto out_put_unused_1; |
28407630 AV |
1278 | } |
1279 | ||
aab174f0 | 1280 | newfile1 = sock_alloc_file(sock1, flags, NULL); |
b5ffe634 | 1281 | if (IS_ERR(newfile1)) { |
28407630 | 1282 | err = PTR_ERR(newfile1); |
d73aa286 | 1283 | goto out_put_unused_both; |
28407630 AV |
1284 | } |
1285 | ||
aab174f0 | 1286 | newfile2 = sock_alloc_file(sock2, flags, NULL); |
28407630 AV |
1287 | if (IS_ERR(newfile2)) { |
1288 | err = PTR_ERR(newfile2); | |
d73aa286 | 1289 | goto out_fput_1; |
db349509 AV |
1290 | } |
1291 | ||
d73aa286 YD |
1292 | err = put_user(fd1, &usockvec[0]); |
1293 | if (err) | |
1294 | goto out_fput_both; | |
1295 | ||
1296 | err = put_user(fd2, &usockvec[1]); | |
1297 | if (err) | |
1298 | goto out_fput_both; | |
1299 | ||
157cf649 | 1300 | audit_fd_pair(fd1, fd2); |
d73aa286 | 1301 | |
db349509 AV |
1302 | fd_install(fd1, newfile1); |
1303 | fd_install(fd2, newfile2); | |
1da177e4 LT |
1304 | /* fd1 and fd2 may be already another descriptors. |
1305 | * Not kernel problem. | |
1306 | */ | |
1307 | ||
d73aa286 | 1308 | return 0; |
1da177e4 | 1309 | |
d73aa286 YD |
1310 | out_fput_both: |
1311 | fput(newfile2); | |
1312 | fput(newfile1); | |
1313 | put_unused_fd(fd2); | |
1314 | put_unused_fd(fd1); | |
1315 | goto out; | |
1316 | ||
1317 | out_fput_1: | |
1318 | fput(newfile1); | |
1319 | put_unused_fd(fd2); | |
1320 | put_unused_fd(fd1); | |
1321 | sock_release(sock2); | |
1322 | goto out; | |
1da177e4 | 1323 | |
d73aa286 YD |
1324 | out_put_unused_both: |
1325 | put_unused_fd(fd2); | |
1326 | out_put_unused_1: | |
1327 | put_unused_fd(fd1); | |
1da177e4 | 1328 | out_release_both: |
89bddce5 | 1329 | sock_release(sock2); |
1da177e4 | 1330 | out_release_1: |
89bddce5 | 1331 | sock_release(sock1); |
1da177e4 LT |
1332 | out: |
1333 | return err; | |
1334 | } | |
1335 | ||
1da177e4 LT |
1336 | /* |
1337 | * Bind a name to a socket. Nothing much to do here since it's | |
1338 | * the protocol's responsibility to handle the local address. | |
1339 | * | |
1340 | * We move the socket address to kernel space before we call | |
1341 | * the protocol layer (having also checked the address is ok). | |
1342 | */ | |
1343 | ||
20f37034 | 1344 | SYSCALL_DEFINE3(bind, int, fd, struct sockaddr __user *, umyaddr, int, addrlen) |
1da177e4 LT |
1345 | { |
1346 | struct socket *sock; | |
230b1839 | 1347 | struct sockaddr_storage address; |
6cb153ca | 1348 | int err, fput_needed; |
1da177e4 | 1349 | |
89bddce5 | 1350 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
e71a4783 | 1351 | if (sock) { |
43db362d | 1352 | err = move_addr_to_kernel(umyaddr, addrlen, &address); |
89bddce5 SH |
1353 | if (err >= 0) { |
1354 | err = security_socket_bind(sock, | |
230b1839 | 1355 | (struct sockaddr *)&address, |
89bddce5 | 1356 | addrlen); |
6cb153ca BL |
1357 | if (!err) |
1358 | err = sock->ops->bind(sock, | |
89bddce5 | 1359 | (struct sockaddr *) |
230b1839 | 1360 | &address, addrlen); |
1da177e4 | 1361 | } |
6cb153ca | 1362 | fput_light(sock->file, fput_needed); |
89bddce5 | 1363 | } |
1da177e4 LT |
1364 | return err; |
1365 | } | |
1366 | ||
1da177e4 LT |
1367 | /* |
1368 | * Perform a listen. Basically, we allow the protocol to do anything | |
1369 | * necessary for a listen, and if that works, we mark the socket as | |
1370 | * ready for listening. | |
1371 | */ | |
1372 | ||
3e0fa65f | 1373 | SYSCALL_DEFINE2(listen, int, fd, int, backlog) |
1da177e4 LT |
1374 | { |
1375 | struct socket *sock; | |
6cb153ca | 1376 | int err, fput_needed; |
b8e1f9b5 | 1377 | int somaxconn; |
89bddce5 SH |
1378 | |
1379 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
1380 | if (sock) { | |
8efa6e93 | 1381 | somaxconn = sock_net(sock->sk)->core.sysctl_somaxconn; |
95c96174 | 1382 | if ((unsigned int)backlog > somaxconn) |
b8e1f9b5 | 1383 | backlog = somaxconn; |
1da177e4 LT |
1384 | |
1385 | err = security_socket_listen(sock, backlog); | |
6cb153ca BL |
1386 | if (!err) |
1387 | err = sock->ops->listen(sock, backlog); | |
1da177e4 | 1388 | |
6cb153ca | 1389 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1390 | } |
1391 | return err; | |
1392 | } | |
1393 | ||
1da177e4 LT |
1394 | /* |
1395 | * For accept, we attempt to create a new socket, set up the link | |
1396 | * with the client, wake up the client, then return the new | |
1397 | * connected fd. We collect the address of the connector in kernel | |
1398 | * space and move it to user at the very end. This is unclean because | |
1399 | * we open the socket then return an error. | |
1400 | * | |
1401 | * 1003.1g adds the ability to recvmsg() to query connection pending | |
1402 | * status to recvmsg. We need to add that support in a way thats | |
1403 | * clean when we restucture accept also. | |
1404 | */ | |
1405 | ||
20f37034 HC |
1406 | SYSCALL_DEFINE4(accept4, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1407 | int __user *, upeer_addrlen, int, flags) | |
1da177e4 LT |
1408 | { |
1409 | struct socket *sock, *newsock; | |
39d8c1b6 | 1410 | struct file *newfile; |
6cb153ca | 1411 | int err, len, newfd, fput_needed; |
230b1839 | 1412 | struct sockaddr_storage address; |
1da177e4 | 1413 | |
77d27200 | 1414 | if (flags & ~(SOCK_CLOEXEC | SOCK_NONBLOCK)) |
aaca0bdc UD |
1415 | return -EINVAL; |
1416 | ||
1417 | if (SOCK_NONBLOCK != O_NONBLOCK && (flags & SOCK_NONBLOCK)) | |
1418 | flags = (flags & ~SOCK_NONBLOCK) | O_NONBLOCK; | |
1419 | ||
6cb153ca | 1420 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1421 | if (!sock) |
1422 | goto out; | |
1423 | ||
1424 | err = -ENFILE; | |
c6d409cf ED |
1425 | newsock = sock_alloc(); |
1426 | if (!newsock) | |
1da177e4 LT |
1427 | goto out_put; |
1428 | ||
1429 | newsock->type = sock->type; | |
1430 | newsock->ops = sock->ops; | |
1431 | ||
1da177e4 LT |
1432 | /* |
1433 | * We don't need try_module_get here, as the listening socket (sock) | |
1434 | * has the protocol module (sock->ops->owner) held. | |
1435 | */ | |
1436 | __module_get(newsock->ops->owner); | |
1437 | ||
28407630 | 1438 | newfd = get_unused_fd_flags(flags); |
39d8c1b6 DM |
1439 | if (unlikely(newfd < 0)) { |
1440 | err = newfd; | |
9a1875e6 DM |
1441 | sock_release(newsock); |
1442 | goto out_put; | |
39d8c1b6 | 1443 | } |
aab174f0 | 1444 | newfile = sock_alloc_file(newsock, flags, sock->sk->sk_prot_creator->name); |
b5ffe634 | 1445 | if (IS_ERR(newfile)) { |
28407630 AV |
1446 | err = PTR_ERR(newfile); |
1447 | put_unused_fd(newfd); | |
1448 | sock_release(newsock); | |
1449 | goto out_put; | |
1450 | } | |
39d8c1b6 | 1451 | |
a79af59e FF |
1452 | err = security_socket_accept(sock, newsock); |
1453 | if (err) | |
39d8c1b6 | 1454 | goto out_fd; |
a79af59e | 1455 | |
1da177e4 LT |
1456 | err = sock->ops->accept(sock, newsock, sock->file->f_flags); |
1457 | if (err < 0) | |
39d8c1b6 | 1458 | goto out_fd; |
1da177e4 LT |
1459 | |
1460 | if (upeer_sockaddr) { | |
230b1839 | 1461 | if (newsock->ops->getname(newsock, (struct sockaddr *)&address, |
89bddce5 | 1462 | &len, 2) < 0) { |
1da177e4 | 1463 | err = -ECONNABORTED; |
39d8c1b6 | 1464 | goto out_fd; |
1da177e4 | 1465 | } |
43db362d | 1466 | err = move_addr_to_user(&address, |
230b1839 | 1467 | len, upeer_sockaddr, upeer_addrlen); |
1da177e4 | 1468 | if (err < 0) |
39d8c1b6 | 1469 | goto out_fd; |
1da177e4 LT |
1470 | } |
1471 | ||
1472 | /* File flags are not inherited via accept() unlike another OSes. */ | |
1473 | ||
39d8c1b6 DM |
1474 | fd_install(newfd, newfile); |
1475 | err = newfd; | |
1da177e4 | 1476 | |
1da177e4 | 1477 | out_put: |
6cb153ca | 1478 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1479 | out: |
1480 | return err; | |
39d8c1b6 | 1481 | out_fd: |
9606a216 | 1482 | fput(newfile); |
39d8c1b6 | 1483 | put_unused_fd(newfd); |
1da177e4 LT |
1484 | goto out_put; |
1485 | } | |
1486 | ||
20f37034 HC |
1487 | SYSCALL_DEFINE3(accept, int, fd, struct sockaddr __user *, upeer_sockaddr, |
1488 | int __user *, upeer_addrlen) | |
aaca0bdc | 1489 | { |
de11defe | 1490 | return sys_accept4(fd, upeer_sockaddr, upeer_addrlen, 0); |
aaca0bdc UD |
1491 | } |
1492 | ||
1da177e4 LT |
1493 | /* |
1494 | * Attempt to connect to a socket with the server address. The address | |
1495 | * is in user space so we verify it is OK and move it to kernel space. | |
1496 | * | |
1497 | * For 1003.1g we need to add clean support for a bind to AF_UNSPEC to | |
1498 | * break bindings | |
1499 | * | |
1500 | * NOTE: 1003.1g draft 6.3 is broken with respect to AX.25/NetROM and | |
1501 | * other SEQPACKET protocols that take time to connect() as it doesn't | |
1502 | * include the -EINPROGRESS status for such sockets. | |
1503 | */ | |
1504 | ||
20f37034 HC |
1505 | SYSCALL_DEFINE3(connect, int, fd, struct sockaddr __user *, uservaddr, |
1506 | int, addrlen) | |
1da177e4 LT |
1507 | { |
1508 | struct socket *sock; | |
230b1839 | 1509 | struct sockaddr_storage address; |
6cb153ca | 1510 | int err, fput_needed; |
1da177e4 | 1511 | |
6cb153ca | 1512 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1513 | if (!sock) |
1514 | goto out; | |
43db362d | 1515 | err = move_addr_to_kernel(uservaddr, addrlen, &address); |
1da177e4 LT |
1516 | if (err < 0) |
1517 | goto out_put; | |
1518 | ||
89bddce5 | 1519 | err = |
230b1839 | 1520 | security_socket_connect(sock, (struct sockaddr *)&address, addrlen); |
1da177e4 LT |
1521 | if (err) |
1522 | goto out_put; | |
1523 | ||
230b1839 | 1524 | err = sock->ops->connect(sock, (struct sockaddr *)&address, addrlen, |
1da177e4 LT |
1525 | sock->file->f_flags); |
1526 | out_put: | |
6cb153ca | 1527 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1528 | out: |
1529 | return err; | |
1530 | } | |
1531 | ||
1532 | /* | |
1533 | * Get the local address ('name') of a socket object. Move the obtained | |
1534 | * name to user space. | |
1535 | */ | |
1536 | ||
20f37034 HC |
1537 | SYSCALL_DEFINE3(getsockname, int, fd, struct sockaddr __user *, usockaddr, |
1538 | int __user *, usockaddr_len) | |
1da177e4 LT |
1539 | { |
1540 | struct socket *sock; | |
230b1839 | 1541 | struct sockaddr_storage address; |
6cb153ca | 1542 | int len, err, fput_needed; |
89bddce5 | 1543 | |
6cb153ca | 1544 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 LT |
1545 | if (!sock) |
1546 | goto out; | |
1547 | ||
1548 | err = security_socket_getsockname(sock); | |
1549 | if (err) | |
1550 | goto out_put; | |
1551 | ||
230b1839 | 1552 | err = sock->ops->getname(sock, (struct sockaddr *)&address, &len, 0); |
1da177e4 LT |
1553 | if (err) |
1554 | goto out_put; | |
43db362d | 1555 | err = move_addr_to_user(&address, len, usockaddr, usockaddr_len); |
1da177e4 LT |
1556 | |
1557 | out_put: | |
6cb153ca | 1558 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1559 | out: |
1560 | return err; | |
1561 | } | |
1562 | ||
1563 | /* | |
1564 | * Get the remote address ('name') of a socket object. Move the obtained | |
1565 | * name to user space. | |
1566 | */ | |
1567 | ||
20f37034 HC |
1568 | SYSCALL_DEFINE3(getpeername, int, fd, struct sockaddr __user *, usockaddr, |
1569 | int __user *, usockaddr_len) | |
1da177e4 LT |
1570 | { |
1571 | struct socket *sock; | |
230b1839 | 1572 | struct sockaddr_storage address; |
6cb153ca | 1573 | int len, err, fput_needed; |
1da177e4 | 1574 | |
89bddce5 SH |
1575 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1576 | if (sock != NULL) { | |
1da177e4 LT |
1577 | err = security_socket_getpeername(sock); |
1578 | if (err) { | |
6cb153ca | 1579 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1580 | return err; |
1581 | } | |
1582 | ||
89bddce5 | 1583 | err = |
230b1839 | 1584 | sock->ops->getname(sock, (struct sockaddr *)&address, &len, |
89bddce5 | 1585 | 1); |
1da177e4 | 1586 | if (!err) |
43db362d | 1587 | err = move_addr_to_user(&address, len, usockaddr, |
89bddce5 | 1588 | usockaddr_len); |
6cb153ca | 1589 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
1590 | } |
1591 | return err; | |
1592 | } | |
1593 | ||
1594 | /* | |
1595 | * Send a datagram to a given address. We move the address into kernel | |
1596 | * space and check the user space data area is readable before invoking | |
1597 | * the protocol. | |
1598 | */ | |
1599 | ||
3e0fa65f | 1600 | SYSCALL_DEFINE6(sendto, int, fd, void __user *, buff, size_t, len, |
95c96174 | 1601 | unsigned int, flags, struct sockaddr __user *, addr, |
3e0fa65f | 1602 | int, addr_len) |
1da177e4 LT |
1603 | { |
1604 | struct socket *sock; | |
230b1839 | 1605 | struct sockaddr_storage address; |
1da177e4 LT |
1606 | int err; |
1607 | struct msghdr msg; | |
1608 | struct iovec iov; | |
6cb153ca | 1609 | int fput_needed; |
6cb153ca | 1610 | |
602bd0e9 AV |
1611 | err = import_single_range(WRITE, buff, len, &iov, &msg.msg_iter); |
1612 | if (unlikely(err)) | |
1613 | return err; | |
de0fa95c PE |
1614 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1615 | if (!sock) | |
4387ff75 | 1616 | goto out; |
6cb153ca | 1617 | |
89bddce5 | 1618 | msg.msg_name = NULL; |
89bddce5 SH |
1619 | msg.msg_control = NULL; |
1620 | msg.msg_controllen = 0; | |
1621 | msg.msg_namelen = 0; | |
6cb153ca | 1622 | if (addr) { |
43db362d | 1623 | err = move_addr_to_kernel(addr, addr_len, &address); |
1da177e4 LT |
1624 | if (err < 0) |
1625 | goto out_put; | |
230b1839 | 1626 | msg.msg_name = (struct sockaddr *)&address; |
89bddce5 | 1627 | msg.msg_namelen = addr_len; |
1da177e4 LT |
1628 | } |
1629 | if (sock->file->f_flags & O_NONBLOCK) | |
1630 | flags |= MSG_DONTWAIT; | |
1631 | msg.msg_flags = flags; | |
d8725c86 | 1632 | err = sock_sendmsg(sock, &msg); |
1da177e4 | 1633 | |
89bddce5 | 1634 | out_put: |
de0fa95c | 1635 | fput_light(sock->file, fput_needed); |
4387ff75 | 1636 | out: |
1da177e4 LT |
1637 | return err; |
1638 | } | |
1639 | ||
1640 | /* | |
89bddce5 | 1641 | * Send a datagram down a socket. |
1da177e4 LT |
1642 | */ |
1643 | ||
3e0fa65f | 1644 | SYSCALL_DEFINE4(send, int, fd, void __user *, buff, size_t, len, |
95c96174 | 1645 | unsigned int, flags) |
1da177e4 LT |
1646 | { |
1647 | return sys_sendto(fd, buff, len, flags, NULL, 0); | |
1648 | } | |
1649 | ||
1650 | /* | |
89bddce5 | 1651 | * Receive a frame from the socket and optionally record the address of the |
1da177e4 LT |
1652 | * sender. We verify the buffers are writable and if needed move the |
1653 | * sender address from kernel to user space. | |
1654 | */ | |
1655 | ||
3e0fa65f | 1656 | SYSCALL_DEFINE6(recvfrom, int, fd, void __user *, ubuf, size_t, size, |
95c96174 | 1657 | unsigned int, flags, struct sockaddr __user *, addr, |
3e0fa65f | 1658 | int __user *, addr_len) |
1da177e4 LT |
1659 | { |
1660 | struct socket *sock; | |
1661 | struct iovec iov; | |
1662 | struct msghdr msg; | |
230b1839 | 1663 | struct sockaddr_storage address; |
89bddce5 | 1664 | int err, err2; |
6cb153ca BL |
1665 | int fput_needed; |
1666 | ||
602bd0e9 AV |
1667 | err = import_single_range(READ, ubuf, size, &iov, &msg.msg_iter); |
1668 | if (unlikely(err)) | |
1669 | return err; | |
de0fa95c | 1670 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1da177e4 | 1671 | if (!sock) |
de0fa95c | 1672 | goto out; |
1da177e4 | 1673 | |
89bddce5 SH |
1674 | msg.msg_control = NULL; |
1675 | msg.msg_controllen = 0; | |
f3d33426 HFS |
1676 | /* Save some cycles and don't copy the address if not needed */ |
1677 | msg.msg_name = addr ? (struct sockaddr *)&address : NULL; | |
1678 | /* We assume all kernel code knows the size of sockaddr_storage */ | |
1679 | msg.msg_namelen = 0; | |
130ed5d1 | 1680 | msg.msg_iocb = NULL; |
1da177e4 LT |
1681 | if (sock->file->f_flags & O_NONBLOCK) |
1682 | flags |= MSG_DONTWAIT; | |
2da62906 | 1683 | err = sock_recvmsg(sock, &msg, flags); |
1da177e4 | 1684 | |
89bddce5 | 1685 | if (err >= 0 && addr != NULL) { |
43db362d | 1686 | err2 = move_addr_to_user(&address, |
230b1839 | 1687 | msg.msg_namelen, addr, addr_len); |
89bddce5 SH |
1688 | if (err2 < 0) |
1689 | err = err2; | |
1da177e4 | 1690 | } |
de0fa95c PE |
1691 | |
1692 | fput_light(sock->file, fput_needed); | |
4387ff75 | 1693 | out: |
1da177e4 LT |
1694 | return err; |
1695 | } | |
1696 | ||
1697 | /* | |
89bddce5 | 1698 | * Receive a datagram from a socket. |
1da177e4 LT |
1699 | */ |
1700 | ||
b7c0ddf5 JG |
1701 | SYSCALL_DEFINE4(recv, int, fd, void __user *, ubuf, size_t, size, |
1702 | unsigned int, flags) | |
1da177e4 LT |
1703 | { |
1704 | return sys_recvfrom(fd, ubuf, size, flags, NULL, NULL); | |
1705 | } | |
1706 | ||
1707 | /* | |
1708 | * Set a socket option. Because we don't know the option lengths we have | |
1709 | * to pass the user mode parameter for the protocols to sort out. | |
1710 | */ | |
1711 | ||
20f37034 HC |
1712 | SYSCALL_DEFINE5(setsockopt, int, fd, int, level, int, optname, |
1713 | char __user *, optval, int, optlen) | |
1da177e4 | 1714 | { |
6cb153ca | 1715 | int err, fput_needed; |
1da177e4 LT |
1716 | struct socket *sock; |
1717 | ||
1718 | if (optlen < 0) | |
1719 | return -EINVAL; | |
89bddce5 SH |
1720 | |
1721 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
1722 | if (sock != NULL) { | |
1723 | err = security_socket_setsockopt(sock, level, optname); | |
6cb153ca BL |
1724 | if (err) |
1725 | goto out_put; | |
1da177e4 LT |
1726 | |
1727 | if (level == SOL_SOCKET) | |
89bddce5 SH |
1728 | err = |
1729 | sock_setsockopt(sock, level, optname, optval, | |
1730 | optlen); | |
1da177e4 | 1731 | else |
89bddce5 SH |
1732 | err = |
1733 | sock->ops->setsockopt(sock, level, optname, optval, | |
1734 | optlen); | |
6cb153ca BL |
1735 | out_put: |
1736 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
1737 | } |
1738 | return err; | |
1739 | } | |
1740 | ||
1741 | /* | |
1742 | * Get a socket option. Because we don't know the option lengths we have | |
1743 | * to pass a user mode parameter for the protocols to sort out. | |
1744 | */ | |
1745 | ||
20f37034 HC |
1746 | SYSCALL_DEFINE5(getsockopt, int, fd, int, level, int, optname, |
1747 | char __user *, optval, int __user *, optlen) | |
1da177e4 | 1748 | { |
6cb153ca | 1749 | int err, fput_needed; |
1da177e4 LT |
1750 | struct socket *sock; |
1751 | ||
89bddce5 SH |
1752 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1753 | if (sock != NULL) { | |
6cb153ca BL |
1754 | err = security_socket_getsockopt(sock, level, optname); |
1755 | if (err) | |
1756 | goto out_put; | |
1da177e4 LT |
1757 | |
1758 | if (level == SOL_SOCKET) | |
89bddce5 SH |
1759 | err = |
1760 | sock_getsockopt(sock, level, optname, optval, | |
1761 | optlen); | |
1da177e4 | 1762 | else |
89bddce5 SH |
1763 | err = |
1764 | sock->ops->getsockopt(sock, level, optname, optval, | |
1765 | optlen); | |
6cb153ca BL |
1766 | out_put: |
1767 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
1768 | } |
1769 | return err; | |
1770 | } | |
1771 | ||
1da177e4 LT |
1772 | /* |
1773 | * Shutdown a socket. | |
1774 | */ | |
1775 | ||
754fe8d2 | 1776 | SYSCALL_DEFINE2(shutdown, int, fd, int, how) |
1da177e4 | 1777 | { |
6cb153ca | 1778 | int err, fput_needed; |
1da177e4 LT |
1779 | struct socket *sock; |
1780 | ||
89bddce5 SH |
1781 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
1782 | if (sock != NULL) { | |
1da177e4 | 1783 | err = security_socket_shutdown(sock, how); |
6cb153ca BL |
1784 | if (!err) |
1785 | err = sock->ops->shutdown(sock, how); | |
1786 | fput_light(sock->file, fput_needed); | |
1da177e4 LT |
1787 | } |
1788 | return err; | |
1789 | } | |
1790 | ||
89bddce5 | 1791 | /* A couple of helpful macros for getting the address of the 32/64 bit |
1da177e4 LT |
1792 | * fields which are the same type (int / unsigned) on our platforms. |
1793 | */ | |
1794 | #define COMPAT_MSG(msg, member) ((MSG_CMSG_COMPAT & flags) ? &msg##_compat->member : &msg->member) | |
1795 | #define COMPAT_NAMELEN(msg) COMPAT_MSG(msg, msg_namelen) | |
1796 | #define COMPAT_FLAGS(msg) COMPAT_MSG(msg, msg_flags) | |
1797 | ||
c71d8ebe TH |
1798 | struct used_address { |
1799 | struct sockaddr_storage name; | |
1800 | unsigned int name_len; | |
1801 | }; | |
1802 | ||
da184284 AV |
1803 | static int copy_msghdr_from_user(struct msghdr *kmsg, |
1804 | struct user_msghdr __user *umsg, | |
1805 | struct sockaddr __user **save_addr, | |
1806 | struct iovec **iov) | |
1661bf36 | 1807 | { |
08adb7da AV |
1808 | struct sockaddr __user *uaddr; |
1809 | struct iovec __user *uiov; | |
c0371da6 | 1810 | size_t nr_segs; |
08adb7da AV |
1811 | ssize_t err; |
1812 | ||
1813 | if (!access_ok(VERIFY_READ, umsg, sizeof(*umsg)) || | |
1814 | __get_user(uaddr, &umsg->msg_name) || | |
1815 | __get_user(kmsg->msg_namelen, &umsg->msg_namelen) || | |
1816 | __get_user(uiov, &umsg->msg_iov) || | |
c0371da6 | 1817 | __get_user(nr_segs, &umsg->msg_iovlen) || |
08adb7da AV |
1818 | __get_user(kmsg->msg_control, &umsg->msg_control) || |
1819 | __get_user(kmsg->msg_controllen, &umsg->msg_controllen) || | |
1820 | __get_user(kmsg->msg_flags, &umsg->msg_flags)) | |
1661bf36 | 1821 | return -EFAULT; |
dbb490b9 | 1822 | |
08adb7da | 1823 | if (!uaddr) |
6a2a2b3a AS |
1824 | kmsg->msg_namelen = 0; |
1825 | ||
dbb490b9 ML |
1826 | if (kmsg->msg_namelen < 0) |
1827 | return -EINVAL; | |
1828 | ||
1661bf36 | 1829 | if (kmsg->msg_namelen > sizeof(struct sockaddr_storage)) |
db31c55a | 1830 | kmsg->msg_namelen = sizeof(struct sockaddr_storage); |
08adb7da AV |
1831 | |
1832 | if (save_addr) | |
1833 | *save_addr = uaddr; | |
1834 | ||
1835 | if (uaddr && kmsg->msg_namelen) { | |
1836 | if (!save_addr) { | |
1837 | err = move_addr_to_kernel(uaddr, kmsg->msg_namelen, | |
1838 | kmsg->msg_name); | |
1839 | if (err < 0) | |
1840 | return err; | |
1841 | } | |
1842 | } else { | |
1843 | kmsg->msg_name = NULL; | |
1844 | kmsg->msg_namelen = 0; | |
1845 | } | |
1846 | ||
c0371da6 | 1847 | if (nr_segs > UIO_MAXIOV) |
08adb7da AV |
1848 | return -EMSGSIZE; |
1849 | ||
0345f931 | 1850 | kmsg->msg_iocb = NULL; |
1851 | ||
da184284 AV |
1852 | return import_iovec(save_addr ? READ : WRITE, uiov, nr_segs, |
1853 | UIO_FASTIOV, iov, &kmsg->msg_iter); | |
1661bf36 DC |
1854 | } |
1855 | ||
666547ff | 1856 | static int ___sys_sendmsg(struct socket *sock, struct user_msghdr __user *msg, |
95c96174 | 1857 | struct msghdr *msg_sys, unsigned int flags, |
28a94d8f TH |
1858 | struct used_address *used_address, |
1859 | unsigned int allowed_msghdr_flags) | |
1da177e4 | 1860 | { |
89bddce5 SH |
1861 | struct compat_msghdr __user *msg_compat = |
1862 | (struct compat_msghdr __user *)msg; | |
230b1839 | 1863 | struct sockaddr_storage address; |
1da177e4 | 1864 | struct iovec iovstack[UIO_FASTIOV], *iov = iovstack; |
b9d717a7 | 1865 | unsigned char ctl[sizeof(struct cmsghdr) + 20] |
89bddce5 SH |
1866 | __attribute__ ((aligned(sizeof(__kernel_size_t)))); |
1867 | /* 20 is size of ipv6_pktinfo */ | |
1da177e4 | 1868 | unsigned char *ctl_buf = ctl; |
d8725c86 | 1869 | int ctl_len; |
08adb7da | 1870 | ssize_t err; |
89bddce5 | 1871 | |
08adb7da | 1872 | msg_sys->msg_name = &address; |
1da177e4 | 1873 | |
08449320 | 1874 | if (MSG_CMSG_COMPAT & flags) |
08adb7da | 1875 | err = get_compat_msghdr(msg_sys, msg_compat, NULL, &iov); |
08449320 | 1876 | else |
08adb7da | 1877 | err = copy_msghdr_from_user(msg_sys, msg, NULL, &iov); |
89bddce5 | 1878 | if (err < 0) |
da184284 | 1879 | return err; |
1da177e4 LT |
1880 | |
1881 | err = -ENOBUFS; | |
1882 | ||
228e548e | 1883 | if (msg_sys->msg_controllen > INT_MAX) |
1da177e4 | 1884 | goto out_freeiov; |
28a94d8f | 1885 | flags |= (msg_sys->msg_flags & allowed_msghdr_flags); |
228e548e | 1886 | ctl_len = msg_sys->msg_controllen; |
1da177e4 | 1887 | if ((MSG_CMSG_COMPAT & flags) && ctl_len) { |
89bddce5 | 1888 | err = |
228e548e | 1889 | cmsghdr_from_user_compat_to_kern(msg_sys, sock->sk, ctl, |
89bddce5 | 1890 | sizeof(ctl)); |
1da177e4 LT |
1891 | if (err) |
1892 | goto out_freeiov; | |
228e548e AB |
1893 | ctl_buf = msg_sys->msg_control; |
1894 | ctl_len = msg_sys->msg_controllen; | |
1da177e4 | 1895 | } else if (ctl_len) { |
89bddce5 | 1896 | if (ctl_len > sizeof(ctl)) { |
1da177e4 | 1897 | ctl_buf = sock_kmalloc(sock->sk, ctl_len, GFP_KERNEL); |
89bddce5 | 1898 | if (ctl_buf == NULL) |
1da177e4 LT |
1899 | goto out_freeiov; |
1900 | } | |
1901 | err = -EFAULT; | |
1902 | /* | |
228e548e | 1903 | * Careful! Before this, msg_sys->msg_control contains a user pointer. |
1da177e4 LT |
1904 | * Afterwards, it will be a kernel pointer. Thus the compiler-assisted |
1905 | * checking falls down on this. | |
1906 | */ | |
fb8621bb | 1907 | if (copy_from_user(ctl_buf, |
228e548e | 1908 | (void __user __force *)msg_sys->msg_control, |
89bddce5 | 1909 | ctl_len)) |
1da177e4 | 1910 | goto out_freectl; |
228e548e | 1911 | msg_sys->msg_control = ctl_buf; |
1da177e4 | 1912 | } |
228e548e | 1913 | msg_sys->msg_flags = flags; |
1da177e4 LT |
1914 | |
1915 | if (sock->file->f_flags & O_NONBLOCK) | |
228e548e | 1916 | msg_sys->msg_flags |= MSG_DONTWAIT; |
c71d8ebe TH |
1917 | /* |
1918 | * If this is sendmmsg() and current destination address is same as | |
1919 | * previously succeeded address, omit asking LSM's decision. | |
1920 | * used_address->name_len is initialized to UINT_MAX so that the first | |
1921 | * destination address never matches. | |
1922 | */ | |
bc909d9d MD |
1923 | if (used_address && msg_sys->msg_name && |
1924 | used_address->name_len == msg_sys->msg_namelen && | |
1925 | !memcmp(&used_address->name, msg_sys->msg_name, | |
c71d8ebe | 1926 | used_address->name_len)) { |
d8725c86 | 1927 | err = sock_sendmsg_nosec(sock, msg_sys); |
c71d8ebe TH |
1928 | goto out_freectl; |
1929 | } | |
d8725c86 | 1930 | err = sock_sendmsg(sock, msg_sys); |
c71d8ebe TH |
1931 | /* |
1932 | * If this is sendmmsg() and sending to current destination address was | |
1933 | * successful, remember it. | |
1934 | */ | |
1935 | if (used_address && err >= 0) { | |
1936 | used_address->name_len = msg_sys->msg_namelen; | |
bc909d9d MD |
1937 | if (msg_sys->msg_name) |
1938 | memcpy(&used_address->name, msg_sys->msg_name, | |
1939 | used_address->name_len); | |
c71d8ebe | 1940 | } |
1da177e4 LT |
1941 | |
1942 | out_freectl: | |
89bddce5 | 1943 | if (ctl_buf != ctl) |
1da177e4 LT |
1944 | sock_kfree_s(sock->sk, ctl_buf, ctl_len); |
1945 | out_freeiov: | |
da184284 | 1946 | kfree(iov); |
228e548e AB |
1947 | return err; |
1948 | } | |
1949 | ||
1950 | /* | |
1951 | * BSD sendmsg interface | |
1952 | */ | |
1953 | ||
666547ff | 1954 | long __sys_sendmsg(int fd, struct user_msghdr __user *msg, unsigned flags) |
228e548e AB |
1955 | { |
1956 | int fput_needed, err; | |
1957 | struct msghdr msg_sys; | |
1be374a0 AL |
1958 | struct socket *sock; |
1959 | ||
1be374a0 | 1960 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
228e548e AB |
1961 | if (!sock) |
1962 | goto out; | |
1963 | ||
28a94d8f | 1964 | err = ___sys_sendmsg(sock, msg, &msg_sys, flags, NULL, 0); |
228e548e | 1965 | |
6cb153ca | 1966 | fput_light(sock->file, fput_needed); |
89bddce5 | 1967 | out: |
1da177e4 LT |
1968 | return err; |
1969 | } | |
1970 | ||
666547ff | 1971 | SYSCALL_DEFINE3(sendmsg, int, fd, struct user_msghdr __user *, msg, unsigned int, flags) |
a7526eb5 AL |
1972 | { |
1973 | if (flags & MSG_CMSG_COMPAT) | |
1974 | return -EINVAL; | |
1975 | return __sys_sendmsg(fd, msg, flags); | |
1976 | } | |
1977 | ||
228e548e AB |
1978 | /* |
1979 | * Linux sendmmsg interface | |
1980 | */ | |
1981 | ||
1982 | int __sys_sendmmsg(int fd, struct mmsghdr __user *mmsg, unsigned int vlen, | |
1983 | unsigned int flags) | |
1984 | { | |
1985 | int fput_needed, err, datagrams; | |
1986 | struct socket *sock; | |
1987 | struct mmsghdr __user *entry; | |
1988 | struct compat_mmsghdr __user *compat_entry; | |
1989 | struct msghdr msg_sys; | |
c71d8ebe | 1990 | struct used_address used_address; |
f092276d | 1991 | unsigned int oflags = flags; |
228e548e | 1992 | |
98382f41 AB |
1993 | if (vlen > UIO_MAXIOV) |
1994 | vlen = UIO_MAXIOV; | |
228e548e AB |
1995 | |
1996 | datagrams = 0; | |
1997 | ||
1998 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
1999 | if (!sock) | |
2000 | return err; | |
2001 | ||
c71d8ebe | 2002 | used_address.name_len = UINT_MAX; |
228e548e AB |
2003 | entry = mmsg; |
2004 | compat_entry = (struct compat_mmsghdr __user *)mmsg; | |
728ffb86 | 2005 | err = 0; |
f092276d | 2006 | flags |= MSG_BATCH; |
228e548e AB |
2007 | |
2008 | while (datagrams < vlen) { | |
f092276d TH |
2009 | if (datagrams == vlen - 1) |
2010 | flags = oflags; | |
2011 | ||
228e548e | 2012 | if (MSG_CMSG_COMPAT & flags) { |
666547ff | 2013 | err = ___sys_sendmsg(sock, (struct user_msghdr __user *)compat_entry, |
28a94d8f | 2014 | &msg_sys, flags, &used_address, MSG_EOR); |
228e548e AB |
2015 | if (err < 0) |
2016 | break; | |
2017 | err = __put_user(err, &compat_entry->msg_len); | |
2018 | ++compat_entry; | |
2019 | } else { | |
a7526eb5 | 2020 | err = ___sys_sendmsg(sock, |
666547ff | 2021 | (struct user_msghdr __user *)entry, |
28a94d8f | 2022 | &msg_sys, flags, &used_address, MSG_EOR); |
228e548e AB |
2023 | if (err < 0) |
2024 | break; | |
2025 | err = put_user(err, &entry->msg_len); | |
2026 | ++entry; | |
2027 | } | |
2028 | ||
2029 | if (err) | |
2030 | break; | |
2031 | ++datagrams; | |
a78cb84c | 2032 | cond_resched(); |
228e548e AB |
2033 | } |
2034 | ||
228e548e AB |
2035 | fput_light(sock->file, fput_needed); |
2036 | ||
728ffb86 AB |
2037 | /* We only return an error if no datagrams were able to be sent */ |
2038 | if (datagrams != 0) | |
228e548e AB |
2039 | return datagrams; |
2040 | ||
228e548e AB |
2041 | return err; |
2042 | } | |
2043 | ||
2044 | SYSCALL_DEFINE4(sendmmsg, int, fd, struct mmsghdr __user *, mmsg, | |
2045 | unsigned int, vlen, unsigned int, flags) | |
2046 | { | |
1be374a0 AL |
2047 | if (flags & MSG_CMSG_COMPAT) |
2048 | return -EINVAL; | |
228e548e AB |
2049 | return __sys_sendmmsg(fd, mmsg, vlen, flags); |
2050 | } | |
2051 | ||
666547ff | 2052 | static int ___sys_recvmsg(struct socket *sock, struct user_msghdr __user *msg, |
95c96174 | 2053 | struct msghdr *msg_sys, unsigned int flags, int nosec) |
1da177e4 | 2054 | { |
89bddce5 SH |
2055 | struct compat_msghdr __user *msg_compat = |
2056 | (struct compat_msghdr __user *)msg; | |
1da177e4 | 2057 | struct iovec iovstack[UIO_FASTIOV]; |
89bddce5 | 2058 | struct iovec *iov = iovstack; |
1da177e4 | 2059 | unsigned long cmsg_ptr; |
2da62906 | 2060 | int len; |
08adb7da | 2061 | ssize_t err; |
1da177e4 LT |
2062 | |
2063 | /* kernel mode address */ | |
230b1839 | 2064 | struct sockaddr_storage addr; |
1da177e4 LT |
2065 | |
2066 | /* user mode address pointers */ | |
2067 | struct sockaddr __user *uaddr; | |
08adb7da | 2068 | int __user *uaddr_len = COMPAT_NAMELEN(msg); |
89bddce5 | 2069 | |
08adb7da | 2070 | msg_sys->msg_name = &addr; |
1da177e4 | 2071 | |
f3d33426 | 2072 | if (MSG_CMSG_COMPAT & flags) |
08adb7da | 2073 | err = get_compat_msghdr(msg_sys, msg_compat, &uaddr, &iov); |
f3d33426 | 2074 | else |
08adb7da | 2075 | err = copy_msghdr_from_user(msg_sys, msg, &uaddr, &iov); |
1da177e4 | 2076 | if (err < 0) |
da184284 | 2077 | return err; |
1da177e4 | 2078 | |
a2e27255 ACM |
2079 | cmsg_ptr = (unsigned long)msg_sys->msg_control; |
2080 | msg_sys->msg_flags = flags & (MSG_CMSG_CLOEXEC|MSG_CMSG_COMPAT); | |
89bddce5 | 2081 | |
f3d33426 HFS |
2082 | /* We assume all kernel code knows the size of sockaddr_storage */ |
2083 | msg_sys->msg_namelen = 0; | |
2084 | ||
1da177e4 LT |
2085 | if (sock->file->f_flags & O_NONBLOCK) |
2086 | flags |= MSG_DONTWAIT; | |
2da62906 | 2087 | err = (nosec ? sock_recvmsg_nosec : sock_recvmsg)(sock, msg_sys, flags); |
1da177e4 LT |
2088 | if (err < 0) |
2089 | goto out_freeiov; | |
2090 | len = err; | |
2091 | ||
2092 | if (uaddr != NULL) { | |
43db362d | 2093 | err = move_addr_to_user(&addr, |
a2e27255 | 2094 | msg_sys->msg_namelen, uaddr, |
89bddce5 | 2095 | uaddr_len); |
1da177e4 LT |
2096 | if (err < 0) |
2097 | goto out_freeiov; | |
2098 | } | |
a2e27255 | 2099 | err = __put_user((msg_sys->msg_flags & ~MSG_CMSG_COMPAT), |
37f7f421 | 2100 | COMPAT_FLAGS(msg)); |
1da177e4 LT |
2101 | if (err) |
2102 | goto out_freeiov; | |
2103 | if (MSG_CMSG_COMPAT & flags) | |
a2e27255 | 2104 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2105 | &msg_compat->msg_controllen); |
2106 | else | |
a2e27255 | 2107 | err = __put_user((unsigned long)msg_sys->msg_control - cmsg_ptr, |
1da177e4 LT |
2108 | &msg->msg_controllen); |
2109 | if (err) | |
2110 | goto out_freeiov; | |
2111 | err = len; | |
2112 | ||
2113 | out_freeiov: | |
da184284 | 2114 | kfree(iov); |
a2e27255 ACM |
2115 | return err; |
2116 | } | |
2117 | ||
2118 | /* | |
2119 | * BSD recvmsg interface | |
2120 | */ | |
2121 | ||
666547ff | 2122 | long __sys_recvmsg(int fd, struct user_msghdr __user *msg, unsigned flags) |
a2e27255 ACM |
2123 | { |
2124 | int fput_needed, err; | |
2125 | struct msghdr msg_sys; | |
1be374a0 AL |
2126 | struct socket *sock; |
2127 | ||
1be374a0 | 2128 | sock = sockfd_lookup_light(fd, &err, &fput_needed); |
a2e27255 ACM |
2129 | if (!sock) |
2130 | goto out; | |
2131 | ||
a7526eb5 | 2132 | err = ___sys_recvmsg(sock, msg, &msg_sys, flags, 0); |
a2e27255 | 2133 | |
6cb153ca | 2134 | fput_light(sock->file, fput_needed); |
1da177e4 LT |
2135 | out: |
2136 | return err; | |
2137 | } | |
2138 | ||
666547ff | 2139 | SYSCALL_DEFINE3(recvmsg, int, fd, struct user_msghdr __user *, msg, |
a7526eb5 AL |
2140 | unsigned int, flags) |
2141 | { | |
2142 | if (flags & MSG_CMSG_COMPAT) | |
2143 | return -EINVAL; | |
2144 | return __sys_recvmsg(fd, msg, flags); | |
2145 | } | |
2146 | ||
a2e27255 ACM |
2147 | /* |
2148 | * Linux recvmmsg interface | |
2149 | */ | |
2150 | ||
2151 | int __sys_recvmmsg(int fd, struct mmsghdr __user *mmsg, unsigned int vlen, | |
2152 | unsigned int flags, struct timespec *timeout) | |
2153 | { | |
2154 | int fput_needed, err, datagrams; | |
2155 | struct socket *sock; | |
2156 | struct mmsghdr __user *entry; | |
d7256d0e | 2157 | struct compat_mmsghdr __user *compat_entry; |
a2e27255 | 2158 | struct msghdr msg_sys; |
766b9f92 DD |
2159 | struct timespec64 end_time; |
2160 | struct timespec64 timeout64; | |
a2e27255 ACM |
2161 | |
2162 | if (timeout && | |
2163 | poll_select_set_timeout(&end_time, timeout->tv_sec, | |
2164 | timeout->tv_nsec)) | |
2165 | return -EINVAL; | |
2166 | ||
2167 | datagrams = 0; | |
2168 | ||
2169 | sock = sockfd_lookup_light(fd, &err, &fput_needed); | |
2170 | if (!sock) | |
2171 | return err; | |
2172 | ||
2173 | err = sock_error(sock->sk); | |
2174 | if (err) | |
2175 | goto out_put; | |
2176 | ||
2177 | entry = mmsg; | |
d7256d0e | 2178 | compat_entry = (struct compat_mmsghdr __user *)mmsg; |
a2e27255 ACM |
2179 | |
2180 | while (datagrams < vlen) { | |
2181 | /* | |
2182 | * No need to ask LSM for more than the first datagram. | |
2183 | */ | |
d7256d0e | 2184 | if (MSG_CMSG_COMPAT & flags) { |
666547ff | 2185 | err = ___sys_recvmsg(sock, (struct user_msghdr __user *)compat_entry, |
a7526eb5 AL |
2186 | &msg_sys, flags & ~MSG_WAITFORONE, |
2187 | datagrams); | |
d7256d0e JMG |
2188 | if (err < 0) |
2189 | break; | |
2190 | err = __put_user(err, &compat_entry->msg_len); | |
2191 | ++compat_entry; | |
2192 | } else { | |
a7526eb5 | 2193 | err = ___sys_recvmsg(sock, |
666547ff | 2194 | (struct user_msghdr __user *)entry, |
a7526eb5 AL |
2195 | &msg_sys, flags & ~MSG_WAITFORONE, |
2196 | datagrams); | |
d7256d0e JMG |
2197 | if (err < 0) |
2198 | break; | |
2199 | err = put_user(err, &entry->msg_len); | |
2200 | ++entry; | |
2201 | } | |
2202 | ||
a2e27255 ACM |
2203 | if (err) |
2204 | break; | |
a2e27255 ACM |
2205 | ++datagrams; |
2206 | ||
71c5c159 BB |
2207 | /* MSG_WAITFORONE turns on MSG_DONTWAIT after one packet */ |
2208 | if (flags & MSG_WAITFORONE) | |
2209 | flags |= MSG_DONTWAIT; | |
2210 | ||
a2e27255 | 2211 | if (timeout) { |
766b9f92 DD |
2212 | ktime_get_ts64(&timeout64); |
2213 | *timeout = timespec64_to_timespec( | |
2214 | timespec64_sub(end_time, timeout64)); | |
a2e27255 ACM |
2215 | if (timeout->tv_sec < 0) { |
2216 | timeout->tv_sec = timeout->tv_nsec = 0; | |
2217 | break; | |
2218 | } | |
2219 | ||
2220 | /* Timeout, return less than vlen datagrams */ | |
2221 | if (timeout->tv_nsec == 0 && timeout->tv_sec == 0) | |
2222 | break; | |
2223 | } | |
2224 | ||
2225 | /* Out of band data, return right away */ | |
2226 | if (msg_sys.msg_flags & MSG_OOB) | |
2227 | break; | |
a78cb84c | 2228 | cond_resched(); |
a2e27255 ACM |
2229 | } |
2230 | ||
a2e27255 | 2231 | if (err == 0) |
34b88a68 ACM |
2232 | goto out_put; |
2233 | ||
2234 | if (datagrams == 0) { | |
2235 | datagrams = err; | |
2236 | goto out_put; | |
2237 | } | |
a2e27255 | 2238 | |
34b88a68 ACM |
2239 | /* |
2240 | * We may return less entries than requested (vlen) if the | |
2241 | * sock is non block and there aren't enough datagrams... | |
2242 | */ | |
2243 | if (err != -EAGAIN) { | |
a2e27255 | 2244 | /* |
34b88a68 ACM |
2245 | * ... or if recvmsg returns an error after we |
2246 | * received some datagrams, where we record the | |
2247 | * error to return on the next call or if the | |
2248 | * app asks about it using getsockopt(SO_ERROR). | |
a2e27255 | 2249 | */ |
34b88a68 | 2250 | sock->sk->sk_err = -err; |
a2e27255 | 2251 | } |
34b88a68 ACM |
2252 | out_put: |
2253 | fput_light(sock->file, fput_needed); | |
a2e27255 | 2254 | |
34b88a68 | 2255 | return datagrams; |
a2e27255 ACM |
2256 | } |
2257 | ||
2258 | SYSCALL_DEFINE5(recvmmsg, int, fd, struct mmsghdr __user *, mmsg, | |
2259 | unsigned int, vlen, unsigned int, flags, | |
2260 | struct timespec __user *, timeout) | |
2261 | { | |
2262 | int datagrams; | |
2263 | struct timespec timeout_sys; | |
2264 | ||
1be374a0 AL |
2265 | if (flags & MSG_CMSG_COMPAT) |
2266 | return -EINVAL; | |
2267 | ||
a2e27255 ACM |
2268 | if (!timeout) |
2269 | return __sys_recvmmsg(fd, mmsg, vlen, flags, NULL); | |
2270 | ||
2271 | if (copy_from_user(&timeout_sys, timeout, sizeof(timeout_sys))) | |
2272 | return -EFAULT; | |
2273 | ||
2274 | datagrams = __sys_recvmmsg(fd, mmsg, vlen, flags, &timeout_sys); | |
2275 | ||
2276 | if (datagrams > 0 && | |
2277 | copy_to_user(timeout, &timeout_sys, sizeof(timeout_sys))) | |
2278 | datagrams = -EFAULT; | |
2279 | ||
2280 | return datagrams; | |
2281 | } | |
2282 | ||
2283 | #ifdef __ARCH_WANT_SYS_SOCKETCALL | |
1da177e4 LT |
2284 | /* Argument list sizes for sys_socketcall */ |
2285 | #define AL(x) ((x) * sizeof(unsigned long)) | |
228e548e | 2286 | static const unsigned char nargs[21] = { |
c6d409cf ED |
2287 | AL(0), AL(3), AL(3), AL(3), AL(2), AL(3), |
2288 | AL(3), AL(3), AL(4), AL(4), AL(4), AL(6), | |
2289 | AL(6), AL(2), AL(5), AL(5), AL(3), AL(3), | |
228e548e | 2290 | AL(4), AL(5), AL(4) |
89bddce5 SH |
2291 | }; |
2292 | ||
1da177e4 LT |
2293 | #undef AL |
2294 | ||
2295 | /* | |
89bddce5 | 2296 | * System call vectors. |
1da177e4 LT |
2297 | * |
2298 | * Argument checking cleaned up. Saved 20% in size. | |
2299 | * This function doesn't need to set the kernel lock because | |
89bddce5 | 2300 | * it is set by the callees. |
1da177e4 LT |
2301 | */ |
2302 | ||
3e0fa65f | 2303 | SYSCALL_DEFINE2(socketcall, int, call, unsigned long __user *, args) |
1da177e4 | 2304 | { |
2950fa9d | 2305 | unsigned long a[AUDITSC_ARGS]; |
89bddce5 | 2306 | unsigned long a0, a1; |
1da177e4 | 2307 | int err; |
47379052 | 2308 | unsigned int len; |
1da177e4 | 2309 | |
228e548e | 2310 | if (call < 1 || call > SYS_SENDMMSG) |
1da177e4 LT |
2311 | return -EINVAL; |
2312 | ||
47379052 AV |
2313 | len = nargs[call]; |
2314 | if (len > sizeof(a)) | |
2315 | return -EINVAL; | |
2316 | ||
1da177e4 | 2317 | /* copy_from_user should be SMP safe. */ |
47379052 | 2318 | if (copy_from_user(a, args, len)) |
1da177e4 | 2319 | return -EFAULT; |
3ec3b2fb | 2320 | |
2950fa9d CG |
2321 | err = audit_socketcall(nargs[call] / sizeof(unsigned long), a); |
2322 | if (err) | |
2323 | return err; | |
3ec3b2fb | 2324 | |
89bddce5 SH |
2325 | a0 = a[0]; |
2326 | a1 = a[1]; | |
2327 | ||
2328 | switch (call) { | |
2329 | case SYS_SOCKET: | |
2330 | err = sys_socket(a0, a1, a[2]); | |
2331 | break; | |
2332 | case SYS_BIND: | |
2333 | err = sys_bind(a0, (struct sockaddr __user *)a1, a[2]); | |
2334 | break; | |
2335 | case SYS_CONNECT: | |
2336 | err = sys_connect(a0, (struct sockaddr __user *)a1, a[2]); | |
2337 | break; | |
2338 | case SYS_LISTEN: | |
2339 | err = sys_listen(a0, a1); | |
2340 | break; | |
2341 | case SYS_ACCEPT: | |
de11defe UD |
2342 | err = sys_accept4(a0, (struct sockaddr __user *)a1, |
2343 | (int __user *)a[2], 0); | |
89bddce5 SH |
2344 | break; |
2345 | case SYS_GETSOCKNAME: | |
2346 | err = | |
2347 | sys_getsockname(a0, (struct sockaddr __user *)a1, | |
2348 | (int __user *)a[2]); | |
2349 | break; | |
2350 | case SYS_GETPEERNAME: | |
2351 | err = | |
2352 | sys_getpeername(a0, (struct sockaddr __user *)a1, | |
2353 | (int __user *)a[2]); | |
2354 | break; | |
2355 | case SYS_SOCKETPAIR: | |
2356 | err = sys_socketpair(a0, a1, a[2], (int __user *)a[3]); | |
2357 | break; | |
2358 | case SYS_SEND: | |
2359 | err = sys_send(a0, (void __user *)a1, a[2], a[3]); | |
2360 | break; | |
2361 | case SYS_SENDTO: | |
2362 | err = sys_sendto(a0, (void __user *)a1, a[2], a[3], | |
2363 | (struct sockaddr __user *)a[4], a[5]); | |
2364 | break; | |
2365 | case SYS_RECV: | |
2366 | err = sys_recv(a0, (void __user *)a1, a[2], a[3]); | |
2367 | break; | |
2368 | case SYS_RECVFROM: | |
2369 | err = sys_recvfrom(a0, (void __user *)a1, a[2], a[3], | |
2370 | (struct sockaddr __user *)a[4], | |
2371 | (int __user *)a[5]); | |
2372 | break; | |
2373 | case SYS_SHUTDOWN: | |
2374 | err = sys_shutdown(a0, a1); | |
2375 | break; | |
2376 | case SYS_SETSOCKOPT: | |
2377 | err = sys_setsockopt(a0, a1, a[2], (char __user *)a[3], a[4]); | |
2378 | break; | |
2379 | case SYS_GETSOCKOPT: | |
2380 | err = | |
2381 | sys_getsockopt(a0, a1, a[2], (char __user *)a[3], | |
2382 | (int __user *)a[4]); | |
2383 | break; | |
2384 | case SYS_SENDMSG: | |
666547ff | 2385 | err = sys_sendmsg(a0, (struct user_msghdr __user *)a1, a[2]); |
89bddce5 | 2386 | break; |
228e548e AB |
2387 | case SYS_SENDMMSG: |
2388 | err = sys_sendmmsg(a0, (struct mmsghdr __user *)a1, a[2], a[3]); | |
2389 | break; | |
89bddce5 | 2390 | case SYS_RECVMSG: |
666547ff | 2391 | err = sys_recvmsg(a0, (struct user_msghdr __user *)a1, a[2]); |
89bddce5 | 2392 | break; |
a2e27255 ACM |
2393 | case SYS_RECVMMSG: |
2394 | err = sys_recvmmsg(a0, (struct mmsghdr __user *)a1, a[2], a[3], | |
2395 | (struct timespec __user *)a[4]); | |
2396 | break; | |
de11defe UD |
2397 | case SYS_ACCEPT4: |
2398 | err = sys_accept4(a0, (struct sockaddr __user *)a1, | |
2399 | (int __user *)a[2], a[3]); | |
aaca0bdc | 2400 | break; |
89bddce5 SH |
2401 | default: |
2402 | err = -EINVAL; | |
2403 | break; | |
1da177e4 LT |
2404 | } |
2405 | return err; | |
2406 | } | |
2407 | ||
89bddce5 | 2408 | #endif /* __ARCH_WANT_SYS_SOCKETCALL */ |
1da177e4 | 2409 | |
55737fda SH |
2410 | /** |
2411 | * sock_register - add a socket protocol handler | |
2412 | * @ops: description of protocol | |
2413 | * | |
1da177e4 LT |
2414 | * This function is called by a protocol handler that wants to |
2415 | * advertise its address family, and have it linked into the | |
e793c0f7 | 2416 | * socket interface. The value ops->family corresponds to the |
55737fda | 2417 | * socket system call protocol family. |
1da177e4 | 2418 | */ |
f0fd27d4 | 2419 | int sock_register(const struct net_proto_family *ops) |
1da177e4 LT |
2420 | { |
2421 | int err; | |
2422 | ||
2423 | if (ops->family >= NPROTO) { | |
3410f22e | 2424 | pr_crit("protocol %d >= NPROTO(%d)\n", ops->family, NPROTO); |
1da177e4 LT |
2425 | return -ENOBUFS; |
2426 | } | |
55737fda SH |
2427 | |
2428 | spin_lock(&net_family_lock); | |
190683a9 ED |
2429 | if (rcu_dereference_protected(net_families[ops->family], |
2430 | lockdep_is_held(&net_family_lock))) | |
55737fda SH |
2431 | err = -EEXIST; |
2432 | else { | |
cf778b00 | 2433 | rcu_assign_pointer(net_families[ops->family], ops); |
1da177e4 LT |
2434 | err = 0; |
2435 | } | |
55737fda SH |
2436 | spin_unlock(&net_family_lock); |
2437 | ||
3410f22e | 2438 | pr_info("NET: Registered protocol family %d\n", ops->family); |
1da177e4 LT |
2439 | return err; |
2440 | } | |
c6d409cf | 2441 | EXPORT_SYMBOL(sock_register); |
1da177e4 | 2442 | |
55737fda SH |
2443 | /** |
2444 | * sock_unregister - remove a protocol handler | |
2445 | * @family: protocol family to remove | |
2446 | * | |
1da177e4 LT |
2447 | * This function is called by a protocol handler that wants to |
2448 | * remove its address family, and have it unlinked from the | |
55737fda SH |
2449 | * new socket creation. |
2450 | * | |
2451 | * If protocol handler is a module, then it can use module reference | |
2452 | * counts to protect against new references. If protocol handler is not | |
2453 | * a module then it needs to provide its own protection in | |
2454 | * the ops->create routine. | |
1da177e4 | 2455 | */ |
f0fd27d4 | 2456 | void sock_unregister(int family) |
1da177e4 | 2457 | { |
f0fd27d4 | 2458 | BUG_ON(family < 0 || family >= NPROTO); |
1da177e4 | 2459 | |
55737fda | 2460 | spin_lock(&net_family_lock); |
a9b3cd7f | 2461 | RCU_INIT_POINTER(net_families[family], NULL); |
55737fda SH |
2462 | spin_unlock(&net_family_lock); |
2463 | ||
2464 | synchronize_rcu(); | |
2465 | ||
3410f22e | 2466 | pr_info("NET: Unregistered protocol family %d\n", family); |
1da177e4 | 2467 | } |
c6d409cf | 2468 | EXPORT_SYMBOL(sock_unregister); |
1da177e4 | 2469 | |
77d76ea3 | 2470 | static int __init sock_init(void) |
1da177e4 | 2471 | { |
b3e19d92 | 2472 | int err; |
2ca794e5 EB |
2473 | /* |
2474 | * Initialize the network sysctl infrastructure. | |
2475 | */ | |
2476 | err = net_sysctl_init(); | |
2477 | if (err) | |
2478 | goto out; | |
b3e19d92 | 2479 | |
1da177e4 | 2480 | /* |
89bddce5 | 2481 | * Initialize skbuff SLAB cache |
1da177e4 LT |
2482 | */ |
2483 | skb_init(); | |
1da177e4 LT |
2484 | |
2485 | /* | |
89bddce5 | 2486 | * Initialize the protocols module. |
1da177e4 LT |
2487 | */ |
2488 | ||
2489 | init_inodecache(); | |
b3e19d92 NP |
2490 | |
2491 | err = register_filesystem(&sock_fs_type); | |
2492 | if (err) | |
2493 | goto out_fs; | |
1da177e4 | 2494 | sock_mnt = kern_mount(&sock_fs_type); |
b3e19d92 NP |
2495 | if (IS_ERR(sock_mnt)) { |
2496 | err = PTR_ERR(sock_mnt); | |
2497 | goto out_mount; | |
2498 | } | |
77d76ea3 AK |
2499 | |
2500 | /* The real protocol initialization is performed in later initcalls. | |
1da177e4 LT |
2501 | */ |
2502 | ||
2503 | #ifdef CONFIG_NETFILTER | |
6d11cfdb PNA |
2504 | err = netfilter_init(); |
2505 | if (err) | |
2506 | goto out; | |
1da177e4 | 2507 | #endif |
cbeb321a | 2508 | |
408eccce | 2509 | ptp_classifier_init(); |
c1f19b51 | 2510 | |
b3e19d92 NP |
2511 | out: |
2512 | return err; | |
2513 | ||
2514 | out_mount: | |
2515 | unregister_filesystem(&sock_fs_type); | |
2516 | out_fs: | |
2517 | goto out; | |
1da177e4 LT |
2518 | } |
2519 | ||
77d76ea3 AK |
2520 | core_initcall(sock_init); /* early initcall */ |
2521 | ||
1da177e4 LT |
2522 | #ifdef CONFIG_PROC_FS |
2523 | void socket_seq_show(struct seq_file *seq) | |
2524 | { | |
2525 | int cpu; | |
2526 | int counter = 0; | |
2527 | ||
6f912042 | 2528 | for_each_possible_cpu(cpu) |
89bddce5 | 2529 | counter += per_cpu(sockets_in_use, cpu); |
1da177e4 LT |
2530 | |
2531 | /* It can be negative, by the way. 8) */ | |
2532 | if (counter < 0) | |
2533 | counter = 0; | |
2534 | ||
2535 | seq_printf(seq, "sockets: used %d\n", counter); | |
2536 | } | |
89bddce5 | 2537 | #endif /* CONFIG_PROC_FS */ |
1da177e4 | 2538 | |
89bbfc95 | 2539 | #ifdef CONFIG_COMPAT |
6b96018b | 2540 | static int do_siocgstamp(struct net *net, struct socket *sock, |
644595f8 | 2541 | unsigned int cmd, void __user *up) |
7a229387 | 2542 | { |
7a229387 AB |
2543 | mm_segment_t old_fs = get_fs(); |
2544 | struct timeval ktv; | |
2545 | int err; | |
2546 | ||
2547 | set_fs(KERNEL_DS); | |
6b96018b | 2548 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)&ktv); |
7a229387 | 2549 | set_fs(old_fs); |
644595f8 | 2550 | if (!err) |
ed6fe9d6 | 2551 | err = compat_put_timeval(&ktv, up); |
644595f8 | 2552 | |
7a229387 AB |
2553 | return err; |
2554 | } | |
2555 | ||
6b96018b | 2556 | static int do_siocgstampns(struct net *net, struct socket *sock, |
644595f8 | 2557 | unsigned int cmd, void __user *up) |
7a229387 | 2558 | { |
7a229387 AB |
2559 | mm_segment_t old_fs = get_fs(); |
2560 | struct timespec kts; | |
2561 | int err; | |
2562 | ||
2563 | set_fs(KERNEL_DS); | |
6b96018b | 2564 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)&kts); |
7a229387 | 2565 | set_fs(old_fs); |
644595f8 | 2566 | if (!err) |
ed6fe9d6 | 2567 | err = compat_put_timespec(&kts, up); |
644595f8 | 2568 | |
7a229387 AB |
2569 | return err; |
2570 | } | |
2571 | ||
6b96018b | 2572 | static int dev_ifname32(struct net *net, struct compat_ifreq __user *uifr32) |
7a229387 AB |
2573 | { |
2574 | struct ifreq __user *uifr; | |
2575 | int err; | |
2576 | ||
2577 | uifr = compat_alloc_user_space(sizeof(struct ifreq)); | |
6b96018b | 2578 | if (copy_in_user(uifr, uifr32, sizeof(struct compat_ifreq))) |
7a229387 AB |
2579 | return -EFAULT; |
2580 | ||
6b96018b | 2581 | err = dev_ioctl(net, SIOCGIFNAME, uifr); |
7a229387 AB |
2582 | if (err) |
2583 | return err; | |
2584 | ||
6b96018b | 2585 | if (copy_in_user(uifr32, uifr, sizeof(struct compat_ifreq))) |
7a229387 AB |
2586 | return -EFAULT; |
2587 | ||
2588 | return 0; | |
2589 | } | |
2590 | ||
6b96018b | 2591 | static int dev_ifconf(struct net *net, struct compat_ifconf __user *uifc32) |
7a229387 | 2592 | { |
6b96018b | 2593 | struct compat_ifconf ifc32; |
7a229387 AB |
2594 | struct ifconf ifc; |
2595 | struct ifconf __user *uifc; | |
6b96018b | 2596 | struct compat_ifreq __user *ifr32; |
7a229387 AB |
2597 | struct ifreq __user *ifr; |
2598 | unsigned int i, j; | |
2599 | int err; | |
2600 | ||
6b96018b | 2601 | if (copy_from_user(&ifc32, uifc32, sizeof(struct compat_ifconf))) |
7a229387 AB |
2602 | return -EFAULT; |
2603 | ||
43da5f2e | 2604 | memset(&ifc, 0, sizeof(ifc)); |
7a229387 AB |
2605 | if (ifc32.ifcbuf == 0) { |
2606 | ifc32.ifc_len = 0; | |
2607 | ifc.ifc_len = 0; | |
2608 | ifc.ifc_req = NULL; | |
2609 | uifc = compat_alloc_user_space(sizeof(struct ifconf)); | |
2610 | } else { | |
c6d409cf ED |
2611 | size_t len = ((ifc32.ifc_len / sizeof(struct compat_ifreq)) + 1) * |
2612 | sizeof(struct ifreq); | |
7a229387 AB |
2613 | uifc = compat_alloc_user_space(sizeof(struct ifconf) + len); |
2614 | ifc.ifc_len = len; | |
2615 | ifr = ifc.ifc_req = (void __user *)(uifc + 1); | |
2616 | ifr32 = compat_ptr(ifc32.ifcbuf); | |
c6d409cf | 2617 | for (i = 0; i < ifc32.ifc_len; i += sizeof(struct compat_ifreq)) { |
6b96018b | 2618 | if (copy_in_user(ifr, ifr32, sizeof(struct compat_ifreq))) |
7a229387 AB |
2619 | return -EFAULT; |
2620 | ifr++; | |
2621 | ifr32++; | |
2622 | } | |
2623 | } | |
2624 | if (copy_to_user(uifc, &ifc, sizeof(struct ifconf))) | |
2625 | return -EFAULT; | |
2626 | ||
6b96018b | 2627 | err = dev_ioctl(net, SIOCGIFCONF, uifc); |
7a229387 AB |
2628 | if (err) |
2629 | return err; | |
2630 | ||
2631 | if (copy_from_user(&ifc, uifc, sizeof(struct ifconf))) | |
2632 | return -EFAULT; | |
2633 | ||
2634 | ifr = ifc.ifc_req; | |
2635 | ifr32 = compat_ptr(ifc32.ifcbuf); | |
2636 | for (i = 0, j = 0; | |
c6d409cf ED |
2637 | i + sizeof(struct compat_ifreq) <= ifc32.ifc_len && j < ifc.ifc_len; |
2638 | i += sizeof(struct compat_ifreq), j += sizeof(struct ifreq)) { | |
2639 | if (copy_in_user(ifr32, ifr, sizeof(struct compat_ifreq))) | |
7a229387 AB |
2640 | return -EFAULT; |
2641 | ifr32++; | |
2642 | ifr++; | |
2643 | } | |
2644 | ||
2645 | if (ifc32.ifcbuf == 0) { | |
2646 | /* Translate from 64-bit structure multiple to | |
2647 | * a 32-bit one. | |
2648 | */ | |
2649 | i = ifc.ifc_len; | |
6b96018b | 2650 | i = ((i / sizeof(struct ifreq)) * sizeof(struct compat_ifreq)); |
7a229387 AB |
2651 | ifc32.ifc_len = i; |
2652 | } else { | |
2653 | ifc32.ifc_len = i; | |
2654 | } | |
6b96018b | 2655 | if (copy_to_user(uifc32, &ifc32, sizeof(struct compat_ifconf))) |
7a229387 AB |
2656 | return -EFAULT; |
2657 | ||
2658 | return 0; | |
2659 | } | |
2660 | ||
6b96018b | 2661 | static int ethtool_ioctl(struct net *net, struct compat_ifreq __user *ifr32) |
7a229387 | 2662 | { |
3a7da39d BH |
2663 | struct compat_ethtool_rxnfc __user *compat_rxnfc; |
2664 | bool convert_in = false, convert_out = false; | |
2665 | size_t buf_size = ALIGN(sizeof(struct ifreq), 8); | |
2666 | struct ethtool_rxnfc __user *rxnfc; | |
7a229387 | 2667 | struct ifreq __user *ifr; |
3a7da39d BH |
2668 | u32 rule_cnt = 0, actual_rule_cnt; |
2669 | u32 ethcmd; | |
7a229387 | 2670 | u32 data; |
3a7da39d | 2671 | int ret; |
7a229387 | 2672 | |
3a7da39d BH |
2673 | if (get_user(data, &ifr32->ifr_ifru.ifru_data)) |
2674 | return -EFAULT; | |
7a229387 | 2675 | |
3a7da39d BH |
2676 | compat_rxnfc = compat_ptr(data); |
2677 | ||
2678 | if (get_user(ethcmd, &compat_rxnfc->cmd)) | |
7a229387 AB |
2679 | return -EFAULT; |
2680 | ||
3a7da39d BH |
2681 | /* Most ethtool structures are defined without padding. |
2682 | * Unfortunately struct ethtool_rxnfc is an exception. | |
2683 | */ | |
2684 | switch (ethcmd) { | |
2685 | default: | |
2686 | break; | |
2687 | case ETHTOOL_GRXCLSRLALL: | |
2688 | /* Buffer size is variable */ | |
2689 | if (get_user(rule_cnt, &compat_rxnfc->rule_cnt)) | |
2690 | return -EFAULT; | |
2691 | if (rule_cnt > KMALLOC_MAX_SIZE / sizeof(u32)) | |
2692 | return -ENOMEM; | |
2693 | buf_size += rule_cnt * sizeof(u32); | |
2694 | /* fall through */ | |
2695 | case ETHTOOL_GRXRINGS: | |
2696 | case ETHTOOL_GRXCLSRLCNT: | |
2697 | case ETHTOOL_GRXCLSRULE: | |
55664f32 | 2698 | case ETHTOOL_SRXCLSRLINS: |
3a7da39d BH |
2699 | convert_out = true; |
2700 | /* fall through */ | |
2701 | case ETHTOOL_SRXCLSRLDEL: | |
3a7da39d BH |
2702 | buf_size += sizeof(struct ethtool_rxnfc); |
2703 | convert_in = true; | |
2704 | break; | |
2705 | } | |
2706 | ||
2707 | ifr = compat_alloc_user_space(buf_size); | |
954b1244 | 2708 | rxnfc = (void __user *)ifr + ALIGN(sizeof(struct ifreq), 8); |
3a7da39d BH |
2709 | |
2710 | if (copy_in_user(&ifr->ifr_name, &ifr32->ifr_name, IFNAMSIZ)) | |
7a229387 AB |
2711 | return -EFAULT; |
2712 | ||
3a7da39d BH |
2713 | if (put_user(convert_in ? rxnfc : compat_ptr(data), |
2714 | &ifr->ifr_ifru.ifru_data)) | |
7a229387 AB |
2715 | return -EFAULT; |
2716 | ||
3a7da39d | 2717 | if (convert_in) { |
127fe533 | 2718 | /* We expect there to be holes between fs.m_ext and |
3a7da39d BH |
2719 | * fs.ring_cookie and at the end of fs, but nowhere else. |
2720 | */ | |
127fe533 AD |
2721 | BUILD_BUG_ON(offsetof(struct compat_ethtool_rxnfc, fs.m_ext) + |
2722 | sizeof(compat_rxnfc->fs.m_ext) != | |
2723 | offsetof(struct ethtool_rxnfc, fs.m_ext) + | |
2724 | sizeof(rxnfc->fs.m_ext)); | |
3a7da39d BH |
2725 | BUILD_BUG_ON( |
2726 | offsetof(struct compat_ethtool_rxnfc, fs.location) - | |
2727 | offsetof(struct compat_ethtool_rxnfc, fs.ring_cookie) != | |
2728 | offsetof(struct ethtool_rxnfc, fs.location) - | |
2729 | offsetof(struct ethtool_rxnfc, fs.ring_cookie)); | |
2730 | ||
2731 | if (copy_in_user(rxnfc, compat_rxnfc, | |
954b1244 SH |
2732 | (void __user *)(&rxnfc->fs.m_ext + 1) - |
2733 | (void __user *)rxnfc) || | |
3a7da39d BH |
2734 | copy_in_user(&rxnfc->fs.ring_cookie, |
2735 | &compat_rxnfc->fs.ring_cookie, | |
954b1244 SH |
2736 | (void __user *)(&rxnfc->fs.location + 1) - |
2737 | (void __user *)&rxnfc->fs.ring_cookie) || | |
3a7da39d BH |
2738 | copy_in_user(&rxnfc->rule_cnt, &compat_rxnfc->rule_cnt, |
2739 | sizeof(rxnfc->rule_cnt))) | |
2740 | return -EFAULT; | |
2741 | } | |
2742 | ||
2743 | ret = dev_ioctl(net, SIOCETHTOOL, ifr); | |
2744 | if (ret) | |
2745 | return ret; | |
2746 | ||
2747 | if (convert_out) { | |
2748 | if (copy_in_user(compat_rxnfc, rxnfc, | |
954b1244 SH |
2749 | (const void __user *)(&rxnfc->fs.m_ext + 1) - |
2750 | (const void __user *)rxnfc) || | |
3a7da39d BH |
2751 | copy_in_user(&compat_rxnfc->fs.ring_cookie, |
2752 | &rxnfc->fs.ring_cookie, | |
954b1244 SH |
2753 | (const void __user *)(&rxnfc->fs.location + 1) - |
2754 | (const void __user *)&rxnfc->fs.ring_cookie) || | |
3a7da39d BH |
2755 | copy_in_user(&compat_rxnfc->rule_cnt, &rxnfc->rule_cnt, |
2756 | sizeof(rxnfc->rule_cnt))) | |
2757 | return -EFAULT; | |
2758 | ||
2759 | if (ethcmd == ETHTOOL_GRXCLSRLALL) { | |
2760 | /* As an optimisation, we only copy the actual | |
2761 | * number of rules that the underlying | |
2762 | * function returned. Since Mallory might | |
2763 | * change the rule count in user memory, we | |
2764 | * check that it is less than the rule count | |
2765 | * originally given (as the user buffer size), | |
2766 | * which has been range-checked. | |
2767 | */ | |
2768 | if (get_user(actual_rule_cnt, &rxnfc->rule_cnt)) | |
2769 | return -EFAULT; | |
2770 | if (actual_rule_cnt < rule_cnt) | |
2771 | rule_cnt = actual_rule_cnt; | |
2772 | if (copy_in_user(&compat_rxnfc->rule_locs[0], | |
2773 | &rxnfc->rule_locs[0], | |
2774 | rule_cnt * sizeof(u32))) | |
2775 | return -EFAULT; | |
2776 | } | |
2777 | } | |
2778 | ||
2779 | return 0; | |
7a229387 AB |
2780 | } |
2781 | ||
7a50a240 AB |
2782 | static int compat_siocwandev(struct net *net, struct compat_ifreq __user *uifr32) |
2783 | { | |
2784 | void __user *uptr; | |
2785 | compat_uptr_t uptr32; | |
2786 | struct ifreq __user *uifr; | |
2787 | ||
c6d409cf | 2788 | uifr = compat_alloc_user_space(sizeof(*uifr)); |
7a50a240 AB |
2789 | if (copy_in_user(uifr, uifr32, sizeof(struct compat_ifreq))) |
2790 | return -EFAULT; | |
2791 | ||
2792 | if (get_user(uptr32, &uifr32->ifr_settings.ifs_ifsu)) | |
2793 | return -EFAULT; | |
2794 | ||
2795 | uptr = compat_ptr(uptr32); | |
2796 | ||
2797 | if (put_user(uptr, &uifr->ifr_settings.ifs_ifsu.raw_hdlc)) | |
2798 | return -EFAULT; | |
2799 | ||
2800 | return dev_ioctl(net, SIOCWANDEV, uifr); | |
2801 | } | |
2802 | ||
6b96018b AB |
2803 | static int bond_ioctl(struct net *net, unsigned int cmd, |
2804 | struct compat_ifreq __user *ifr32) | |
7a229387 AB |
2805 | { |
2806 | struct ifreq kifr; | |
7a229387 AB |
2807 | mm_segment_t old_fs; |
2808 | int err; | |
7a229387 AB |
2809 | |
2810 | switch (cmd) { | |
2811 | case SIOCBONDENSLAVE: | |
2812 | case SIOCBONDRELEASE: | |
2813 | case SIOCBONDSETHWADDR: | |
2814 | case SIOCBONDCHANGEACTIVE: | |
6b96018b | 2815 | if (copy_from_user(&kifr, ifr32, sizeof(struct compat_ifreq))) |
7a229387 AB |
2816 | return -EFAULT; |
2817 | ||
2818 | old_fs = get_fs(); | |
c6d409cf | 2819 | set_fs(KERNEL_DS); |
c3f52ae6 | 2820 | err = dev_ioctl(net, cmd, |
2821 | (struct ifreq __user __force *) &kifr); | |
c6d409cf | 2822 | set_fs(old_fs); |
7a229387 AB |
2823 | |
2824 | return err; | |
7a229387 | 2825 | default: |
07d106d0 | 2826 | return -ENOIOCTLCMD; |
ccbd6a5a | 2827 | } |
7a229387 AB |
2828 | } |
2829 | ||
590d4693 BH |
2830 | /* Handle ioctls that use ifreq::ifr_data and just need struct ifreq converted */ |
2831 | static int compat_ifr_data_ioctl(struct net *net, unsigned int cmd, | |
6b96018b | 2832 | struct compat_ifreq __user *u_ifreq32) |
7a229387 AB |
2833 | { |
2834 | struct ifreq __user *u_ifreq64; | |
7a229387 AB |
2835 | char tmp_buf[IFNAMSIZ]; |
2836 | void __user *data64; | |
2837 | u32 data32; | |
2838 | ||
2839 | if (copy_from_user(&tmp_buf[0], &(u_ifreq32->ifr_ifrn.ifrn_name[0]), | |
2840 | IFNAMSIZ)) | |
2841 | return -EFAULT; | |
417c3522 | 2842 | if (get_user(data32, &u_ifreq32->ifr_ifru.ifru_data)) |
7a229387 AB |
2843 | return -EFAULT; |
2844 | data64 = compat_ptr(data32); | |
2845 | ||
2846 | u_ifreq64 = compat_alloc_user_space(sizeof(*u_ifreq64)); | |
2847 | ||
7a229387 AB |
2848 | if (copy_to_user(&u_ifreq64->ifr_ifrn.ifrn_name[0], &tmp_buf[0], |
2849 | IFNAMSIZ)) | |
2850 | return -EFAULT; | |
417c3522 | 2851 | if (put_user(data64, &u_ifreq64->ifr_ifru.ifru_data)) |
7a229387 AB |
2852 | return -EFAULT; |
2853 | ||
6b96018b | 2854 | return dev_ioctl(net, cmd, u_ifreq64); |
7a229387 AB |
2855 | } |
2856 | ||
6b96018b AB |
2857 | static int dev_ifsioc(struct net *net, struct socket *sock, |
2858 | unsigned int cmd, struct compat_ifreq __user *uifr32) | |
7a229387 | 2859 | { |
a2116ed2 | 2860 | struct ifreq __user *uifr; |
7a229387 AB |
2861 | int err; |
2862 | ||
a2116ed2 AB |
2863 | uifr = compat_alloc_user_space(sizeof(*uifr)); |
2864 | if (copy_in_user(uifr, uifr32, sizeof(*uifr32))) | |
2865 | return -EFAULT; | |
2866 | ||
2867 | err = sock_do_ioctl(net, sock, cmd, (unsigned long)uifr); | |
2868 | ||
7a229387 AB |
2869 | if (!err) { |
2870 | switch (cmd) { | |
2871 | case SIOCGIFFLAGS: | |
2872 | case SIOCGIFMETRIC: | |
2873 | case SIOCGIFMTU: | |
2874 | case SIOCGIFMEM: | |
2875 | case SIOCGIFHWADDR: | |
2876 | case SIOCGIFINDEX: | |
2877 | case SIOCGIFADDR: | |
2878 | case SIOCGIFBRDADDR: | |
2879 | case SIOCGIFDSTADDR: | |
2880 | case SIOCGIFNETMASK: | |
fab2532b | 2881 | case SIOCGIFPFLAGS: |
7a229387 | 2882 | case SIOCGIFTXQLEN: |
fab2532b AB |
2883 | case SIOCGMIIPHY: |
2884 | case SIOCGMIIREG: | |
a2116ed2 | 2885 | if (copy_in_user(uifr32, uifr, sizeof(*uifr32))) |
7a229387 AB |
2886 | err = -EFAULT; |
2887 | break; | |
2888 | } | |
2889 | } | |
2890 | return err; | |
2891 | } | |
2892 | ||
a2116ed2 AB |
2893 | static int compat_sioc_ifmap(struct net *net, unsigned int cmd, |
2894 | struct compat_ifreq __user *uifr32) | |
2895 | { | |
2896 | struct ifreq ifr; | |
2897 | struct compat_ifmap __user *uifmap32; | |
2898 | mm_segment_t old_fs; | |
2899 | int err; | |
2900 | ||
2901 | uifmap32 = &uifr32->ifr_ifru.ifru_map; | |
2902 | err = copy_from_user(&ifr, uifr32, sizeof(ifr.ifr_name)); | |
3ddc5b46 MD |
2903 | err |= get_user(ifr.ifr_map.mem_start, &uifmap32->mem_start); |
2904 | err |= get_user(ifr.ifr_map.mem_end, &uifmap32->mem_end); | |
2905 | err |= get_user(ifr.ifr_map.base_addr, &uifmap32->base_addr); | |
2906 | err |= get_user(ifr.ifr_map.irq, &uifmap32->irq); | |
2907 | err |= get_user(ifr.ifr_map.dma, &uifmap32->dma); | |
2908 | err |= get_user(ifr.ifr_map.port, &uifmap32->port); | |
a2116ed2 AB |
2909 | if (err) |
2910 | return -EFAULT; | |
2911 | ||
2912 | old_fs = get_fs(); | |
c6d409cf | 2913 | set_fs(KERNEL_DS); |
c3f52ae6 | 2914 | err = dev_ioctl(net, cmd, (void __user __force *)&ifr); |
c6d409cf | 2915 | set_fs(old_fs); |
a2116ed2 AB |
2916 | |
2917 | if (cmd == SIOCGIFMAP && !err) { | |
2918 | err = copy_to_user(uifr32, &ifr, sizeof(ifr.ifr_name)); | |
3ddc5b46 MD |
2919 | err |= put_user(ifr.ifr_map.mem_start, &uifmap32->mem_start); |
2920 | err |= put_user(ifr.ifr_map.mem_end, &uifmap32->mem_end); | |
2921 | err |= put_user(ifr.ifr_map.base_addr, &uifmap32->base_addr); | |
2922 | err |= put_user(ifr.ifr_map.irq, &uifmap32->irq); | |
2923 | err |= put_user(ifr.ifr_map.dma, &uifmap32->dma); | |
2924 | err |= put_user(ifr.ifr_map.port, &uifmap32->port); | |
a2116ed2 AB |
2925 | if (err) |
2926 | err = -EFAULT; | |
2927 | } | |
2928 | return err; | |
2929 | } | |
2930 | ||
7a229387 | 2931 | struct rtentry32 { |
c6d409cf | 2932 | u32 rt_pad1; |
7a229387 AB |
2933 | struct sockaddr rt_dst; /* target address */ |
2934 | struct sockaddr rt_gateway; /* gateway addr (RTF_GATEWAY) */ | |
2935 | struct sockaddr rt_genmask; /* target network mask (IP) */ | |
c6d409cf ED |
2936 | unsigned short rt_flags; |
2937 | short rt_pad2; | |
2938 | u32 rt_pad3; | |
2939 | unsigned char rt_tos; | |
2940 | unsigned char rt_class; | |
2941 | short rt_pad4; | |
2942 | short rt_metric; /* +1 for binary compatibility! */ | |
7a229387 | 2943 | /* char * */ u32 rt_dev; /* forcing the device at add */ |
c6d409cf ED |
2944 | u32 rt_mtu; /* per route MTU/Window */ |
2945 | u32 rt_window; /* Window clamping */ | |
7a229387 AB |
2946 | unsigned short rt_irtt; /* Initial RTT */ |
2947 | }; | |
2948 | ||
2949 | struct in6_rtmsg32 { | |
2950 | struct in6_addr rtmsg_dst; | |
2951 | struct in6_addr rtmsg_src; | |
2952 | struct in6_addr rtmsg_gateway; | |
2953 | u32 rtmsg_type; | |
2954 | u16 rtmsg_dst_len; | |
2955 | u16 rtmsg_src_len; | |
2956 | u32 rtmsg_metric; | |
2957 | u32 rtmsg_info; | |
2958 | u32 rtmsg_flags; | |
2959 | s32 rtmsg_ifindex; | |
2960 | }; | |
2961 | ||
6b96018b AB |
2962 | static int routing_ioctl(struct net *net, struct socket *sock, |
2963 | unsigned int cmd, void __user *argp) | |
7a229387 AB |
2964 | { |
2965 | int ret; | |
2966 | void *r = NULL; | |
2967 | struct in6_rtmsg r6; | |
2968 | struct rtentry r4; | |
2969 | char devname[16]; | |
2970 | u32 rtdev; | |
2971 | mm_segment_t old_fs = get_fs(); | |
2972 | ||
6b96018b AB |
2973 | if (sock && sock->sk && sock->sk->sk_family == AF_INET6) { /* ipv6 */ |
2974 | struct in6_rtmsg32 __user *ur6 = argp; | |
c6d409cf | 2975 | ret = copy_from_user(&r6.rtmsg_dst, &(ur6->rtmsg_dst), |
7a229387 | 2976 | 3 * sizeof(struct in6_addr)); |
3ddc5b46 MD |
2977 | ret |= get_user(r6.rtmsg_type, &(ur6->rtmsg_type)); |
2978 | ret |= get_user(r6.rtmsg_dst_len, &(ur6->rtmsg_dst_len)); | |
2979 | ret |= get_user(r6.rtmsg_src_len, &(ur6->rtmsg_src_len)); | |
2980 | ret |= get_user(r6.rtmsg_metric, &(ur6->rtmsg_metric)); | |
2981 | ret |= get_user(r6.rtmsg_info, &(ur6->rtmsg_info)); | |
2982 | ret |= get_user(r6.rtmsg_flags, &(ur6->rtmsg_flags)); | |
2983 | ret |= get_user(r6.rtmsg_ifindex, &(ur6->rtmsg_ifindex)); | |
7a229387 AB |
2984 | |
2985 | r = (void *) &r6; | |
2986 | } else { /* ipv4 */ | |
6b96018b | 2987 | struct rtentry32 __user *ur4 = argp; |
c6d409cf | 2988 | ret = copy_from_user(&r4.rt_dst, &(ur4->rt_dst), |
7a229387 | 2989 | 3 * sizeof(struct sockaddr)); |
3ddc5b46 MD |
2990 | ret |= get_user(r4.rt_flags, &(ur4->rt_flags)); |
2991 | ret |= get_user(r4.rt_metric, &(ur4->rt_metric)); | |
2992 | ret |= get_user(r4.rt_mtu, &(ur4->rt_mtu)); | |
2993 | ret |= get_user(r4.rt_window, &(ur4->rt_window)); | |
2994 | ret |= get_user(r4.rt_irtt, &(ur4->rt_irtt)); | |
2995 | ret |= get_user(rtdev, &(ur4->rt_dev)); | |
7a229387 | 2996 | if (rtdev) { |
c6d409cf | 2997 | ret |= copy_from_user(devname, compat_ptr(rtdev), 15); |
c3f52ae6 | 2998 | r4.rt_dev = (char __user __force *)devname; |
2999 | devname[15] = 0; | |
7a229387 AB |
3000 | } else |
3001 | r4.rt_dev = NULL; | |
3002 | ||
3003 | r = (void *) &r4; | |
3004 | } | |
3005 | ||
3006 | if (ret) { | |
3007 | ret = -EFAULT; | |
3008 | goto out; | |
3009 | } | |
3010 | ||
c6d409cf | 3011 | set_fs(KERNEL_DS); |
6b96018b | 3012 | ret = sock_do_ioctl(net, sock, cmd, (unsigned long) r); |
c6d409cf | 3013 | set_fs(old_fs); |
7a229387 AB |
3014 | |
3015 | out: | |
7a229387 AB |
3016 | return ret; |
3017 | } | |
3018 | ||
3019 | /* Since old style bridge ioctl's endup using SIOCDEVPRIVATE | |
3020 | * for some operations; this forces use of the newer bridge-utils that | |
25985edc | 3021 | * use compatible ioctls |
7a229387 | 3022 | */ |
6b96018b | 3023 | static int old_bridge_ioctl(compat_ulong_t __user *argp) |
7a229387 | 3024 | { |
6b96018b | 3025 | compat_ulong_t tmp; |
7a229387 | 3026 | |
6b96018b | 3027 | if (get_user(tmp, argp)) |
7a229387 AB |
3028 | return -EFAULT; |
3029 | if (tmp == BRCTL_GET_VERSION) | |
3030 | return BRCTL_VERSION + 1; | |
3031 | return -EINVAL; | |
3032 | } | |
3033 | ||
6b96018b AB |
3034 | static int compat_sock_ioctl_trans(struct file *file, struct socket *sock, |
3035 | unsigned int cmd, unsigned long arg) | |
3036 | { | |
3037 | void __user *argp = compat_ptr(arg); | |
3038 | struct sock *sk = sock->sk; | |
3039 | struct net *net = sock_net(sk); | |
7a229387 | 3040 | |
6b96018b | 3041 | if (cmd >= SIOCDEVPRIVATE && cmd <= (SIOCDEVPRIVATE + 15)) |
590d4693 | 3042 | return compat_ifr_data_ioctl(net, cmd, argp); |
6b96018b AB |
3043 | |
3044 | switch (cmd) { | |
3045 | case SIOCSIFBR: | |
3046 | case SIOCGIFBR: | |
3047 | return old_bridge_ioctl(argp); | |
3048 | case SIOCGIFNAME: | |
3049 | return dev_ifname32(net, argp); | |
3050 | case SIOCGIFCONF: | |
3051 | return dev_ifconf(net, argp); | |
3052 | case SIOCETHTOOL: | |
3053 | return ethtool_ioctl(net, argp); | |
7a50a240 AB |
3054 | case SIOCWANDEV: |
3055 | return compat_siocwandev(net, argp); | |
a2116ed2 AB |
3056 | case SIOCGIFMAP: |
3057 | case SIOCSIFMAP: | |
3058 | return compat_sioc_ifmap(net, cmd, argp); | |
6b96018b AB |
3059 | case SIOCBONDENSLAVE: |
3060 | case SIOCBONDRELEASE: | |
3061 | case SIOCBONDSETHWADDR: | |
6b96018b AB |
3062 | case SIOCBONDCHANGEACTIVE: |
3063 | return bond_ioctl(net, cmd, argp); | |
3064 | case SIOCADDRT: | |
3065 | case SIOCDELRT: | |
3066 | return routing_ioctl(net, sock, cmd, argp); | |
3067 | case SIOCGSTAMP: | |
3068 | return do_siocgstamp(net, sock, cmd, argp); | |
3069 | case SIOCGSTAMPNS: | |
3070 | return do_siocgstampns(net, sock, cmd, argp); | |
590d4693 BH |
3071 | case SIOCBONDSLAVEINFOQUERY: |
3072 | case SIOCBONDINFOQUERY: | |
a2116ed2 | 3073 | case SIOCSHWTSTAMP: |
fd468c74 | 3074 | case SIOCGHWTSTAMP: |
590d4693 | 3075 | return compat_ifr_data_ioctl(net, cmd, argp); |
6b96018b AB |
3076 | |
3077 | case FIOSETOWN: | |
3078 | case SIOCSPGRP: | |
3079 | case FIOGETOWN: | |
3080 | case SIOCGPGRP: | |
3081 | case SIOCBRADDBR: | |
3082 | case SIOCBRDELBR: | |
3083 | case SIOCGIFVLAN: | |
3084 | case SIOCSIFVLAN: | |
3085 | case SIOCADDDLCI: | |
3086 | case SIOCDELDLCI: | |
3087 | return sock_ioctl(file, cmd, arg); | |
3088 | ||
3089 | case SIOCGIFFLAGS: | |
3090 | case SIOCSIFFLAGS: | |
3091 | case SIOCGIFMETRIC: | |
3092 | case SIOCSIFMETRIC: | |
3093 | case SIOCGIFMTU: | |
3094 | case SIOCSIFMTU: | |
3095 | case SIOCGIFMEM: | |
3096 | case SIOCSIFMEM: | |
3097 | case SIOCGIFHWADDR: | |
3098 | case SIOCSIFHWADDR: | |
3099 | case SIOCADDMULTI: | |
3100 | case SIOCDELMULTI: | |
3101 | case SIOCGIFINDEX: | |
6b96018b AB |
3102 | case SIOCGIFADDR: |
3103 | case SIOCSIFADDR: | |
3104 | case SIOCSIFHWBROADCAST: | |
6b96018b | 3105 | case SIOCDIFADDR: |
6b96018b AB |
3106 | case SIOCGIFBRDADDR: |
3107 | case SIOCSIFBRDADDR: | |
3108 | case SIOCGIFDSTADDR: | |
3109 | case SIOCSIFDSTADDR: | |
3110 | case SIOCGIFNETMASK: | |
3111 | case SIOCSIFNETMASK: | |
3112 | case SIOCSIFPFLAGS: | |
3113 | case SIOCGIFPFLAGS: | |
3114 | case SIOCGIFTXQLEN: | |
3115 | case SIOCSIFTXQLEN: | |
3116 | case SIOCBRADDIF: | |
3117 | case SIOCBRDELIF: | |
9177efd3 AB |
3118 | case SIOCSIFNAME: |
3119 | case SIOCGMIIPHY: | |
3120 | case SIOCGMIIREG: | |
3121 | case SIOCSMIIREG: | |
6b96018b | 3122 | return dev_ifsioc(net, sock, cmd, argp); |
9177efd3 | 3123 | |
6b96018b AB |
3124 | case SIOCSARP: |
3125 | case SIOCGARP: | |
3126 | case SIOCDARP: | |
6b96018b | 3127 | case SIOCATMARK: |
9177efd3 AB |
3128 | return sock_do_ioctl(net, sock, cmd, arg); |
3129 | } | |
3130 | ||
6b96018b AB |
3131 | return -ENOIOCTLCMD; |
3132 | } | |
7a229387 | 3133 | |
95c96174 | 3134 | static long compat_sock_ioctl(struct file *file, unsigned int cmd, |
89bddce5 | 3135 | unsigned long arg) |
89bbfc95 SP |
3136 | { |
3137 | struct socket *sock = file->private_data; | |
3138 | int ret = -ENOIOCTLCMD; | |
87de87d5 DM |
3139 | struct sock *sk; |
3140 | struct net *net; | |
3141 | ||
3142 | sk = sock->sk; | |
3143 | net = sock_net(sk); | |
89bbfc95 SP |
3144 | |
3145 | if (sock->ops->compat_ioctl) | |
3146 | ret = sock->ops->compat_ioctl(sock, cmd, arg); | |
3147 | ||
87de87d5 DM |
3148 | if (ret == -ENOIOCTLCMD && |
3149 | (cmd >= SIOCIWFIRST && cmd <= SIOCIWLAST)) | |
3150 | ret = compat_wext_handle_ioctl(net, cmd, arg); | |
3151 | ||
6b96018b AB |
3152 | if (ret == -ENOIOCTLCMD) |
3153 | ret = compat_sock_ioctl_trans(file, sock, cmd, arg); | |
3154 | ||
89bbfc95 SP |
3155 | return ret; |
3156 | } | |
3157 | #endif | |
3158 | ||
ac5a488e SS |
3159 | int kernel_bind(struct socket *sock, struct sockaddr *addr, int addrlen) |
3160 | { | |
3161 | return sock->ops->bind(sock, addr, addrlen); | |
3162 | } | |
c6d409cf | 3163 | EXPORT_SYMBOL(kernel_bind); |
ac5a488e SS |
3164 | |
3165 | int kernel_listen(struct socket *sock, int backlog) | |
3166 | { | |
3167 | return sock->ops->listen(sock, backlog); | |
3168 | } | |
c6d409cf | 3169 | EXPORT_SYMBOL(kernel_listen); |
ac5a488e SS |
3170 | |
3171 | int kernel_accept(struct socket *sock, struct socket **newsock, int flags) | |
3172 | { | |
3173 | struct sock *sk = sock->sk; | |
3174 | int err; | |
3175 | ||
3176 | err = sock_create_lite(sk->sk_family, sk->sk_type, sk->sk_protocol, | |
3177 | newsock); | |
3178 | if (err < 0) | |
3179 | goto done; | |
3180 | ||
3181 | err = sock->ops->accept(sock, *newsock, flags); | |
3182 | if (err < 0) { | |
3183 | sock_release(*newsock); | |
fa8705b0 | 3184 | *newsock = NULL; |
ac5a488e SS |
3185 | goto done; |
3186 | } | |
3187 | ||
3188 | (*newsock)->ops = sock->ops; | |
1b08534e | 3189 | __module_get((*newsock)->ops->owner); |
ac5a488e SS |
3190 | |
3191 | done: | |
3192 | return err; | |
3193 | } | |
c6d409cf | 3194 | EXPORT_SYMBOL(kernel_accept); |
ac5a488e SS |
3195 | |
3196 | int kernel_connect(struct socket *sock, struct sockaddr *addr, int addrlen, | |
4768fbcb | 3197 | int flags) |
ac5a488e SS |
3198 | { |
3199 | return sock->ops->connect(sock, addr, addrlen, flags); | |
3200 | } | |
c6d409cf | 3201 | EXPORT_SYMBOL(kernel_connect); |
ac5a488e SS |
3202 | |
3203 | int kernel_getsockname(struct socket *sock, struct sockaddr *addr, | |
3204 | int *addrlen) | |
3205 | { | |
3206 | return sock->ops->getname(sock, addr, addrlen, 0); | |
3207 | } | |
c6d409cf | 3208 | EXPORT_SYMBOL(kernel_getsockname); |
ac5a488e SS |
3209 | |
3210 | int kernel_getpeername(struct socket *sock, struct sockaddr *addr, | |
3211 | int *addrlen) | |
3212 | { | |
3213 | return sock->ops->getname(sock, addr, addrlen, 1); | |
3214 | } | |
c6d409cf | 3215 | EXPORT_SYMBOL(kernel_getpeername); |
ac5a488e SS |
3216 | |
3217 | int kernel_getsockopt(struct socket *sock, int level, int optname, | |
3218 | char *optval, int *optlen) | |
3219 | { | |
3220 | mm_segment_t oldfs = get_fs(); | |
fb8621bb NK |
3221 | char __user *uoptval; |
3222 | int __user *uoptlen; | |
ac5a488e SS |
3223 | int err; |
3224 | ||
fb8621bb NK |
3225 | uoptval = (char __user __force *) optval; |
3226 | uoptlen = (int __user __force *) optlen; | |
3227 | ||
ac5a488e SS |
3228 | set_fs(KERNEL_DS); |
3229 | if (level == SOL_SOCKET) | |
fb8621bb | 3230 | err = sock_getsockopt(sock, level, optname, uoptval, uoptlen); |
ac5a488e | 3231 | else |
fb8621bb NK |
3232 | err = sock->ops->getsockopt(sock, level, optname, uoptval, |
3233 | uoptlen); | |
ac5a488e SS |
3234 | set_fs(oldfs); |
3235 | return err; | |
3236 | } | |
c6d409cf | 3237 | EXPORT_SYMBOL(kernel_getsockopt); |
ac5a488e SS |
3238 | |
3239 | int kernel_setsockopt(struct socket *sock, int level, int optname, | |
b7058842 | 3240 | char *optval, unsigned int optlen) |
ac5a488e SS |
3241 | { |
3242 | mm_segment_t oldfs = get_fs(); | |
fb8621bb | 3243 | char __user *uoptval; |
ac5a488e SS |
3244 | int err; |
3245 | ||
fb8621bb NK |
3246 | uoptval = (char __user __force *) optval; |
3247 | ||
ac5a488e SS |
3248 | set_fs(KERNEL_DS); |
3249 | if (level == SOL_SOCKET) | |
fb8621bb | 3250 | err = sock_setsockopt(sock, level, optname, uoptval, optlen); |
ac5a488e | 3251 | else |
fb8621bb | 3252 | err = sock->ops->setsockopt(sock, level, optname, uoptval, |
ac5a488e SS |
3253 | optlen); |
3254 | set_fs(oldfs); | |
3255 | return err; | |
3256 | } | |
c6d409cf | 3257 | EXPORT_SYMBOL(kernel_setsockopt); |
ac5a488e SS |
3258 | |
3259 | int kernel_sendpage(struct socket *sock, struct page *page, int offset, | |
3260 | size_t size, int flags) | |
3261 | { | |
3262 | if (sock->ops->sendpage) | |
3263 | return sock->ops->sendpage(sock, page, offset, size, flags); | |
3264 | ||
3265 | return sock_no_sendpage(sock, page, offset, size, flags); | |
3266 | } | |
c6d409cf | 3267 | EXPORT_SYMBOL(kernel_sendpage); |
ac5a488e SS |
3268 | |
3269 | int kernel_sock_ioctl(struct socket *sock, int cmd, unsigned long arg) | |
3270 | { | |
3271 | mm_segment_t oldfs = get_fs(); | |
3272 | int err; | |
3273 | ||
3274 | set_fs(KERNEL_DS); | |
3275 | err = sock->ops->ioctl(sock, cmd, arg); | |
3276 | set_fs(oldfs); | |
3277 | ||
3278 | return err; | |
3279 | } | |
c6d409cf | 3280 | EXPORT_SYMBOL(kernel_sock_ioctl); |
ac5a488e | 3281 | |
91cf45f0 TM |
3282 | int kernel_sock_shutdown(struct socket *sock, enum sock_shutdown_cmd how) |
3283 | { | |
3284 | return sock->ops->shutdown(sock, how); | |
3285 | } | |
91cf45f0 | 3286 | EXPORT_SYMBOL(kernel_sock_shutdown); |